Modeling and Low-Level
Waste Management:
An Interagency Workshop
December 1-4, '  D
Denver, Color?
                Sponsored by
                U.S. Department of Energy
                U.S. Environmental Protection Agency
                U.S. Geological Survey
                U.S. Nuclear Regulatory Commission

-------
     Printed in the United States of America. Available from
             National Technical Information Service
                  U.S. Department of Commerce
        5285 Port Royal Road, Springfield, Virginia 22161
     NTIS price codes—Printed Copy:  A17  Microfiche   A01
This report was prepared as an account of work sponsored by an agency of the
United States Government. Neither the United StatesGovernment nor any agency
thereof, nor any of their employees, makes any warranty, express or implied, or
assumes any legal liability or responsibility for the accuracy, completeness, or
usefulness of any information, apparatus, product, or process disclosed, or
represents that its use would not infringe privately owned rights. Reference herein
to any specific commercial product, process, or service by trade name, trademark,
manufacturer,  or otherwise,  does not necessarily  constitute or imply  its
endorsement, recommendation, or favoring by the United States Government or
any agency thereof. The views and opinions of authors expressed herein do not
necessarily state or reflect those of the United States Government or any agency
thereof.

-------
Modeling and  Low-Level
Waste  Management:
An  Interagency Workshop
December 1-4.  1980
Denver. Colorado
                                       Compiled by

                                      Craig A. Little
                             Health and Safety Research Division
                                           and

                                     Leroy E. Stratton
                               Environmental Sciences Division

                               Oak Ridge National Laboratory
                                 Oak Ridge, Tennessee 37830
                                          for the

       INTERAGENCY LOW-LEVEL WASTE MODELING COMMITTEE:

       Robert S. Lowrie (Chairman), Program Manager, Low-Level Waste Management Program,
       Oak Ridge National Laboratory, Oak Ridge, Tennessee

       Edward F. Hawkins, Low-Level Waste Licensing Branch, Division of Waste Management,
       U.S. Nuclear Regulatory Commission, Washington, D.C.
       Dewey E. Large, Manager, ORO Low-Level Radioactive Waste Management Program,
       U.S. Department of Energy, Oak Ridge, Tennessee
       G. Lewis Meyer, Criterion and Standards Division, Office of Radiation Programs,
       U.S. Environmental Protection Agency, Washington, D.C.

       John B. Robertson, Office of Radiohydrology,
       U.S. Geological Survey, Reston, Virginia

       Craig A. Little (Scientific Secretary), Health and Safety Research Division,
       Oak Ridge National Laboratory, Oak Ridge, Tennessee
       Workshop  sponsored jointly by U.S.  Environmental Protection Agency, U.S. Nuclear
       Regulatory Commission, U.S. Geological Survey, and the Oak Ridge Operations Low-Level
       Radioactive Waste Management Program, U.S. Department of Energy, under contract W-7405-
       eng-26 with the Union Carbide Corporation.
                                  Date Published: June 1981

-------
                                  Preface

    This workshop was  organized  by the Interagency Low-Level  Waste Modeling
Committee and was held December 1-4,1980 in Denver, Colorado. Committee membership
is listed on the title page. Each of the four federal agencies represented on the committee
contributed equal funding to support the meeting.
    The paramount objective was to foster cooperation between the agencies having
various responsibilities and involvements in modeling for low-level waste applications.
Another aim of the workshop was to begin a dialogue between model users, such as
regulators and assessors, and modelers. These Proceedings are testimony that progress has
been made toward  attainment of these goals.
    The meeting had three distinct sessions. First, each agency presented its point of view
concerning modeling and the need for models in low-level radioactive waste applications.
Second, a larger group of more technical papers was presented by persons actively involved
in model  development or applications. Each of these  papers came from one of the
sponsoring agencies or from an agency contractor. Last, four workshops—two running
concurrently—were held to attempt to reach a consensus among participants regarding
numerous waste modeling topics. Each of the three-sections is identified in the Contents.
    We are grateful to many people  for their assistance  in organizing the meeting  and
compiling the  Proceedings:  speakers  and  authors  who promptly delivered their
manuscripts; workshop chairmen and secretaries who recorded and summarized their
sessions; Bonnie Reesor of the ORNL Conference Office; and particularly Thelma Patton
and Brenda Baber of the ORNL Low-Level Waste Management Programs Office. Thanks.
                                    Craig A. Little
                                    Secretary for Interagency Low-Level
                                    Waste Modeling Committee
June 1981
                                       111

-------
                                Contents


                                                                       Page

Preface	 iii

                             Agency Papers

USDOE Needs in Modeling for Low-Level Radioactive
  Waste Management	  3
    Dewey E. Large

Modeling and Analyses to Support EPA's Low-Level
  Radioactive Waste Standards	  9
    G. Lewis Meyer

Modeling in Low-Level Radioactive Waste Management
  from the U.S. Geological Survey Perspective	 21
    John B. Robertson

Low-Level Radioactive Waste Modeling Needs from the
  USNRC Perspective	 31
    R. Dale Smith and Edward F. Hawkins


                            Technical Papers

Modeling Contaminant Transport in Porous Media in
  Relation to Nuclear-Waste Disposal: A Review	 43
    David B. Grove and Kenneth L. Kipp

An Optimum Model to Predict Radionuclide Transport in
  an Aquifer for the Application to Health Effects Evaluations	 65
    Cheng Y. Hung

Modeling of Water and Solute Transport Under Variably
  Saturated Conditions: State of the Art	 81
    E. G. Lappala

Verification of Partially Saturated Zone Moisture
  Migration: Model and Field Facilities	139
    S. J. Phillips and T. L. Jones

Radionuclide Transport Under Changing Soil
  Salinity Conditions	155
    T. L. Jones, G. W. Gee, and P. J. Wierenga

On the Computation of the Velocity Field and Mass
  Balance in the Finite-Element Modeling of Groundwater Flow	165
    Gour-Tsyh Yeh

-------
 Fracture Flow Modeling for Low-Level Nuclear Waste Disposal	187
    Brian Y. Kanehiro,  Varuthamadhina Guvanasen,
    Charles R. Wilson, and Paul A. Witherspoon

 A Conceptual Analysis  of the Groundwater Flow System
   at the Maxey Flats Radioactive Waste Burial Site,
   Fleming County, Kentucky	197
    D. W. Pollock and H. H. Zehner

 Computer Simulation of Groundwater Flow at a Commercial
   Radioactive-Waste Landfill Near West Valley, Cattaraugus
   County, New York	215
    David E. Prudic

 State-of-the-Art in Modeling Solute and Sediment
  Transport in Rivers	249
    William W. Sayre

 An Overview of Resuspension Models: Application to
  Low-Level Waste Management	273
    J. W. Healy

 An Overview of EPA's Health-Risk Assessment Model for
  the Shallow Land Disposal of LLW	283
    G. Lewis Meyer and Cheng Y. Hung

 Risk Assessment and Radioactive Waste Management	303
    Jerry J. Cohen and  Craig F. Smith

 Modeling in the Determination of Use Conditions for a
  Decommissioned Shallow-Land Burial Site	315
    E. S. Murphy

 A Systems Model for Evaluating Population Dose from
  Low-Level Waste Activities	327
    J. A. Stoddard and D. H. Lester

 Models and Criteria for  LLW Disposal Performance	333
    Craig F. Smith and Jerry J. Cohen


                          Workshop Summaries

 Release Mechanisms Workshop 	343
    Les Dole and David E. Fields

 Overall Systems Modeling Workshop	351
    James Steger and Leroy E. Stratton

Environmental Transport Parameters and Processes Workshop	357
    Edward L. Albenesius and Craig A. Little

Model Verification/Validation Workshop	373
    T. L. Jones and Donald Jacobs
List of Attendees	 379
                                      VI

-------
Agency Papers

-------
  USDOE NEEDS IN MODELING FOR LOW-LEVEL RADIOACTIVE WASTE MANAGEMENT
                            Dewey E. Large

              National  Low-Level  Waste  Management Program
                       U.S.  Department  of Energy
                      Oak Ridge, Tennessee 37830
     Greetings from John Peel who is Manager of the National
Low-Level Waste Management Program (NLLWMP) for the Department of
Energy Idaho Operations.  Speaking for him and as Associate Program
Manager of the NLLWMP for Technology Development for Oak Ridge
Operations, we will give you this Program's perspective of the DOE's
requirements for modeling in Low-Level Radioactive Waste Management.

     On the agenda, this presentation is entitled, "USDOE Needs in
Modeling for Low-Level Waste Management."  At the outset, we want to
emphasize that the Department of Energy does not need additional
models as entities.  We do need reliable and credible tools for
decision making which involve identification, selection, adaptation,
and implementation of existing models.  This is an important
distinction and is the basis for the Low-Level Waste Management
Program's efforts  in environmental modeling.  Recently, the Program
issued a strategy  document which endorses states being allowed to
assume and implement their responsibilities for managing their
respective low-level wastes;  it identifies areas of assistance that
federal agencies can provide to states.  In summary, the states must
decide upon a course of action regarding siting and selection of
disposal sites; making that decision often requires and is expected
to use environmental models as decision-making tools.

     We are not being trite when we emphasize that an environmental
model is only a tool that can be used in a decision making process
and should not be  used as a replacement for management judgement.
The value of a model is to focus complex technical considerations to
provide management with the ability to examine various options,
compare impacts, and to explore the mitigative effects of
alternatives.  If  and when it becomes necessary, models are used to
predict events and to aid in resolving the "What if?" questions.
The reliability for prediction is, of course, a function of the
model itself, but  as importantly, the accuracy of the input data.
Because of the inherent inaccuracies and uncertainties of data, or
simplifying assumptions used when data do not exist, the reliability
of model prediction will always be subject to question.  For this
reason, decision making cannot and should not be based solely on
model prediction.

     In July, many of you attended a meeting on Low-Level Waste
Modeling held in Bethesda, Maryland.  At that meeting, several key
issues common to all modeling groups were identified.  Tonight, let

-------
us examine some conclusions reached at that meeting and how the
Program intends to proceed in its modeling efforts.  We will also
present some  issues which we believe can only be resolved through
interagency cooperation.

     An important conclusion at our July meeting was that many
models lack credibility.  Consequently, a "Credibility Gap" does
exist.  This  is partly due to ineffective use of models;  but, as
importantly,  it is due to a lack of understanding by model
developers of what is required for decision making.  Are models
really telling us what we need to know?  Models are used to predict,
but results are often presented as absolute numbers.  When looking
at those numbers, how often have you seen model results that attempt
questionable accuracy and predict one "FEMTO-REM" per billion
population?  These numbers do not deal with a real world situation;
and, based on cause-effect relationships, such a degree of precision
is neither necessary nor desirable.

     Clearly, there is a need for model developers to state what
their models can do and provide some bounds or limitations on model
results.   Concurrently, the user must be educated concerning the
real function of that model and not attempt to push it beyond the
limits of credibility.  This brings up another point concerning
credibility.  How often have we seen "worst-case analysis" end up as
"absolute truth"?  Decision making is an interactive process between
government and public.  A report analyzing some worst case scenario
may be so prohibitive that reasonable alternatives may be ignored.
As our colleague Jerry Cohen says, "Give me a worst case and I'll
develop yet a worse one."  We must reemphasize the use of
engineering and management judgement in defining model scenarios.
Our principal short term goal is to provide a siting model by
September 1981.  Until this task is completed, the program will not
be able to identify and support other areas of model development
work.  We have neither the time nor the funds to conduct extensive
new model development, and we do not believe that such effort and
expense is necessary since many adequate models already exist.

     The siting model is needed by decision makers to examine the
adequacy of a site, explore various alternatives, and develop
preliminary operating and design criteria.

     The basic guidance for our model development has been that the
model should:

     1.  Estimate health effects and maximum dose to any individual for
         the  population at risk;

     2.  Estimate the health  effects with reasonable accuracy;

     3.  Estimate the health  effects within a reasonable cost of
         computer execution;

-------
4.  Be as simple as possible while giving adequate detail for
    standards development; and

5.  Be comprised from existing sub-models; when adequate
    sub-models are available, thereby reducing development time
    and costs.

Our approach to model development has been to:

1.  Review existing models;

2.  Select viable models;

3.  Utilize viable sub-models from selected models; and

4.  Fill in gaps with new sub-models.


Our consideration and treatment of models include:

1.  Verification and validation

    The credibility of a model is dependent upon  its ability to
    simulate and predict.  A verified model reasonably simulates
    physical transports and pathways; whereas, a  validated model
    predicts (within limitations) actual transport of a substance.
    There are many models that cannot be validated because of
    simplifying assumptions used in their development.  But, any
    model to be used in decision making must, as  a minimum,
    provide a reasonable simulation of the real world.  The
    majority of our laboratory and field efforts  will be directed
    toward validation and verification.

2.  Improved characterization

    Unreliable or inadequate data will always reduce the
    credibility of model results.  Field measurements and uniform
    site characterization procedures are necessary to improve
    input data.

3.  Bounding the credibility of model results

    Sensitivity analyses, error bars, and probabilities are
    important.  Errors in input data should be propagated through
    the model to provide estimates of result error.   Sensitivity
    analyses identify limitations of model  boundary conditions.
    Analysis of probability allows the user to identify the
    dominant pathways and events.  All  of these  serve  to bound  the
    credibility of the results and provide  the user  with  an
    understanding of the model's  limitations  in  decision  making.

-------
     At this time, let us turn our attention to a discussion of
criteria.  A model should be used to determine how effectively a site
will meet criteria.  At the present time, we do not have recognized and
approved criteria for reference of implementation.  The federal
agencies must come to an agreement on what we are expecting models to
provide;  until we have such criteria, it may be pointless to continue
modeling.

     The most basic criteria should be expressed in terms of the
maximum allowable doses to the public from routine and accidental
releases.  In determining these criteria, we must take into
consideration the economic impact on industry and the principles of as
low as reasonably achievable (ALARA).

     The 500 mrem maximum individual exposure limit is based on sound
scientific evidence and reasoning.  It is a regulation and industry
standard.  However, do we select and operate sites based solely on the
500 mrem limit or do we establish some lower limit?  If we establish a
lower limit, is it scientifically defensible, and can we justify the
economic impact on industry?  For purposes of discussion, let us
examine the following "straw man" criteria:

     1.  The maximum allowable dose from any accident or non-routine
         release is 500 mrem under normal conditions.

     2.  The physical properties of that site should not permit
         exposure of individuals to more than 50 mrem under normal
         conditions.  This does not take into account any properties
         for waste form or special retardation barriers but is based
         solely on waste inventory and the site characteristics.

     3.  The dose from a release event should be inversely proportional
         to the probability of the event.  This, of course, means the
         higher the probability, the lower the dose.

     4.  Sites are to be selected, designed, and operated so that
         during operational periods the maximum dose will not exceed
         500 mrem; for the 25 years of post-operational monitoring, the
         maximum dose does not exceed 100 mrem; and after 25 years the
         maximum dose does not exceed 50 mrem.  This last "straw man"
         criteria  is our obligation not to pass on to future
         generations a burden for maintaining these sites.

     One way for a decision maker to determine if his site meets these
"straw-man" criteria  is  through  the  use of a model.  Based upon our
experience, it would seem that such  criteria would not only be easy to
meet but also difficult  to violate  under any reasonable management
system.  Given the guidelines we have suggested,  it might be that
sufficient modeling capabilities already exist to provide reasonable
assurance of compliance.

-------
     The following are a few major items we believe can only be
resolved through the effective interaction of the federal agencies:

     1.  Reference sets of source term and dose conversion factors

         There is a need to establish a set of reference source terms
         to describe a low-level radioactive waste site.  One of the
         major complaints of model users and developers is that too
         much time and effort is spent in developing source terms, and
         there is no guarantee that those terms will be considered
         reasonable.  There already exist several excellent detailed
         source term documents; all that is required is for the
         interagency group to agree on terms and compile these into a
         single reference document.

     2.  Standard documentation format

         A model to be transferred must be in a form easily
         understandable to the user.  There is a need to develop
         procedures and formats for documenting and transferring models
         from developers to users.

     3.  Standard site characterization procedures

         Collection of data for the characterization of a site are not
         uniform procedures and result in large variances in data.  A
         set of reference characterization procedures should be
         developed to identify data to be collected and acceptable
         collection methods.

     4.  Validation experiments

         This was an idea proposed at our July meeting.  In essence, we
         would develop experiments that would test various scale model
         shallow land disposal facilities to points of failure.  The
         data obtained would be used to define the conservativeness of
         our models.

     5.  Reference validation data sets

         Data from field tests should be developed into a reference
         document which would allow model users to test their models
         against real world data.

     6.  Model categorization system

         Model users often have little to guide them in selecting  a
         model.  By providing a categorization system, models would be
         grouped according to function, application, and boundary
         condition.  This categorization system should be developed
         concurrently with the model documentation procedures.

-------
     In conclusion, we believe that by working together with common
objectives the several agencies and supporting organizations can
successfully find adequate tools and use them to make appropriate
decisions for managing the Nation's low-level radioactive waste.

-------
                MODELING AND ANALYSES TO SUPPORT EPA's
                LOW-LEVEL  RADIOACTIVE WASTE  STANDARDS

                            G. Lewis Meyer

                    Criteria  and Standards Division
                     Office of Radiation Programs
                U.  S. Environmental  Protection  Agency
                       Washington, D.C.  20460
                               ABSTRACT

     EPA is developing a generally applicable standard for the
management and disposal of LLW as part of the U.S. National
Radioactive Waste Management Program.  It will issue a proposed
standard in September, 1981, and a final standard in early 1984.
EPA will use predetermined models to estimate the health risks and
costs/benefits from disposing of LLW by a number of land and sea
disposal alternatives.  These estimates will be used to support
EPA's decision on the proposed standard and for the supporting EIS
and Regulatory Analysis.  This paper describes why EPA is using a
generic "systems" approach to modeling the impact of LLW disposal,
the types of disposal alternatives which will be evaluated, its
approach to cost/benefit analysis, the current status of model
development, general input data requirements for the model, and a
general schedule for development of the models and the standard.

               THE NEED FOR A STANDARD 	 AND MODELING

     Before 1975, there was no demonstrated need for an
environmental standard for the disposal of low-level radioactive
waste (LLW) because it was believed that there would be no release
of radioactive from the disposal sites.d)  Since then, studies at
the Maxey Flats (KY), West Valley (NY), and Oak Ridge (TN) disposal
sites have shown that unplanned releases to the uncontrolled
environment can occur if LLW are not properly disposed and that
the impact of these releases could be substantial.  For example,
more than 24.6 million liters of radioactive leachate containing
appreciable concentrations of H-3, Co-60, Sr-90, Cs-137, Pu-238,
and Pu-239/240 have been pumped from the trenches at Maxey Flats and
treated.(2)  Pumping operations are continuing.  If stopped, these
radioactive leachates would be released by overflow directly at land
surface or to the groundwater.  The experience at West Valley has
been similar.(3)

     The public has become increasingly concerned about the safe
disposal of all radioactive wastes including LLW.  Generators of LLW
have become very concerned about the lack of adequate disposal space

-------
                                  10
in the future since three of  six operating commercial LLW disposal
facilities have closed.  State regulators have become concerned
about the quality and the quantity of LLW being shipped to the
disposal sites in their states.  These occurances established the
need for EPA to act pursuant  to its charge to identify radiation
hazards to the general environment and to set generally applicable
environmental standards where needed.(4,5)  On February 12, 1980,
the President directed EPA to develop the necessary radioactive
waste standards on an accelerated basis.(6)

     Consequently, EPA is developing a generally applicable
environmental standard for the management and disposal of LLW.
It has guidance on how to develop this standard from a number of
sources.  These include NEPA (7),  EPA's "Improving Government
Regulations" (8), BEIR II (9), "Decision Making in EPA" (10), and
ICRP-26 (11).  This guidance,  which clearly establishes EPA's needs
for modeling and models,  has been summarized quite well in EPA's
proposed "Criteria for Radioactive Wastes" (12) as follows,

     "...Radiation protection requirements for radioactive wastes
should be based primarily on an assessment of risk to individuals
and populations;  such assessments should be based on predetermined
models..." and

     "...Any risks due to radioactive waste management or disposal
activities should be deemed unacceptable unless it has been
justified that the further reduction in risk that could be achieved
by more complete isolation is impracticable on the basis of
technical and social considerationd..."

     Before going on to discuss EPA's general modeling needs, the
role of modeling in EPA's standards development process should be
put into perspective.  MODELING IS ONLY A TOOL!  Developing and
setting environmental standards is less than an exact science.
In most cases, arrival at the standard involves a combination of
public acceptance, political accord,  good judgment stemming from
long experience in public health protection, and expert technical
judgement.  The models will support our judgement and standard.
The results from our modeling efforts will not be the basis for
EPA's standard.

            EPA'S APPROACH - A GENERIC "SYSTEMS" APPROACH

     The primary purpose of EPA's modeling efforts is to assess the
potential impacts from managing and disposing of LLW by a number of
realistic alternatives.  This includes estimating the potential
releases to the environment, doses, health risks, and costs.

-------
                                  11
A number of factors have influenced EPA's approach to doing this.
These factors, which include EPA's authorizing legislation and
mandates, the character of the wastes, themselves, and the realistic
means by which they may be managed or disposed, are discussed
briefly below.

     The guidance given to EPA for developing the standard is broad,
general, and sometimes contradictory.  The environment should not
be degraded.(7)  We should assess the impact of all reasonable
alternatives.(8)  Even the smallest amount of radiation is
harmful.(9)  The risk should be reduced as low as reasonably
achievable (ALARA) but we should take social and economic factors
into account.(9,11,12)

     The radioactive wastes for which this standard is being
developed are an extremely heterogeneous group of wastes.
They include all wastes which have been categorized as "low-level"
radioactive wastes and, in addition, those transuranic-contaminated
wastes  (TRU) which do not fall under the High-Level Radioactive
Waste (HLW) Standard.(13)  The term "low-level" does not adequately
reflect the potential hazard of wastes currently classified as LLW.
LLW are, in fact, a large diverse class of wastes in which some
types may contain tens of thousands of curies per cubic meter
(14,15) and remain radioactive for millions of years.  Other LLW may
contain only  traces of radioactive material.  They may be solid,
mushy,  liquid,  or gaseous; may range from insoluble to completely
soluble; and may be animal, vegetable, or mineral.

     Disposal in the  ground or in the ocean are the disposal methods
which are most  readily available at this time.  Any land or ocean
disposal facility is  a complex "system" which includes the waste,
the site geology, hydrology or oceanography, and meteorology, the
emplacement method, site engineering, and all of the pathways from
it 	  and the  performance of the system will change if changes are
made to any of  its parts.  Analysis of a single pathway would not
provide an adequate analysis of the impact of a "disposal system."
Thus, the disposal facility and the wastes in it must be regarded as
a disposal system and analyzed as a whole if we are to assess its
overall impact.

     Because of the above factors, our model(s) will be used to
evaluate the effectiveness of each disposal alternative using a
systems approach —  to examine each alternative as a disposal
system  and to compare it with alternative disposal systems.
Sufficient detail will be included in the models to permit
evaluation of the effectiveness of proposed changes in the system
design  for improving  the retention capability of the disposal system

-------
                                  12


 (i.e., change  in waste  form  or  packaging, engineering barrier,
 etc.).  Changes in the  design of  the disposal system may also affect
 the cost  of disposal.   These costs  will be assessed or modeled
 separately.

     There  are a number of land and sea disposal methods which are
 suitable  for the disposal of some types of LLW.  However, no one
 disposal  method is suitable  for all types of LLW.  Therefore, the
 model(s)  should be applicable to a  variety of disposal methods,
 types of  wastes, and hydrogeological settings; the results of which
 should be useful in comparing the potential impacts and costs of the
 different disposal alternatives.  Thus, it is also clear that a
 generic systems model(s) is  needed.

     The  natural characteristics of a disposal site or "disposal
 system" (i.e., geology, hydrology, and meteorology) are determined
 and relatively fixed once a site is selected.  Normally, they can be
 changed by selecting another site.  However, emplacing wastes in the
 site, the interaction of the wastes with the site, and engineering
 changes to the site can significantly alter the functioning of the
 hydrogeologic  portion of the "disposal system" and directly affect
 its performance.   The wastes, their form, the method of their
 emplacement, and site engineering also affect the performance of the
 "disposal system" but are mainly determined by man.  The point being
made is — the natural characteristics of the "disposal system" are
mainly determined in the site selection process; man can control the
 performance and costs of the rest of the system.

               TYPES OF ENVIRONMENTAL ASSESSMENT MODELS

     We have reviewed a number of alternative methods for disposing
 of LLW to identify those which could be implemented quickly without
 delay for technical hardware or development.  So far, we have
 identified eight land and two ocean disposal methods which could be
 used for  the disposal of one or more types of LLW.  Our review has
 shown, however, that not one of these methods is suitable for the
 disposal  of  all types of LLW.

     In five land disposal methods, the wastes would generally be
 disposed  of  at or near  land  surface and generally not below 50
 meters depth.  These shallow land disposal (SLD) methods include
 engineered surface storage (SLD-SS), sanitary landfill (SLD-LF),
 shallow land disposal as presently  practiced (SLD-P), improved
 shallow land disposal (SLS-I), and  intermediate depth disposal
 (SLD-I).  In three land disposal methods, the wastes would generally
 be disposed  of at depths greater than 50 meters.  The deep land
 disposal  (DLD) methods  include deep geological disposal in a mined
 cavity (DLD-MC), hydrofracturing  (DLD-HF), and deep-well injection
 (DLD-DWI).   The two ocean disposal  methods include disposal on the

-------
                                 13


sea floor (OD-OSF) and beneath the sea floor (OD-BSF).  We are,
therefore, developing environmental assessment models for shallow
land disposal (SLD), deep land disposal (DLD), and ocean disposal
(OD).

     The SLD assessment model will be used to analyze the potential
health risk from disposing of LLW at or near land surface to a depth
of 50 meters but generally above the water table.  Experience has
shown that near surface geology, hydrology, and meteorology affect
the performance and retention capability of a SLD site very much.
Therefore, the SLD model is being designed with sufficient
flexibility to cover at least three types of wide-spread
hydrogeologic/climatic settings which we believe will cover the most
sensitive parameters that affect radionuclide retention and site
performance.  These settings include:  (1) an arid zone site for
which no particular regard is given to the permeability of the
disposal medium (because of a lack of water as a driving force);
(2) a humid zone site with a low-permeability disposal medium (which
would fill like a "bathtub" in the event the trench cap leaked); and
(3) a humid zone site with a moderately permeable disposal medium
(which would allow water infiltrating through the trench cap to leak
out the bottom of the trench like a "sieve").

     Additional settings will be developed as necessary by altering
in the input parameters.  The water pathways would certainly be
less important for  conventional SLD and improved SLD in the arid
climates, whereas,  all of the pathways would be important for these
methods in the two  humid climate scenarios.  For intermediate depth
SLD  (10-l8m deep),  the groundwater pathway would be considered most
important.

     The DLD assessment model will closely parallel the SLD
assessment model in design, pathways, etc.  However, it will
basically be used to analyze the impact of disposing of LLW at
depths below 50 meters and generally below the water table, although
there are some areas in the western United States where water tables
are as deep as 700 meters.  Near surface hydrogeology and meteor
ology normally will not have much affect on the performance of the
DLD alternatives.  More emphasis will be given to hydrodynamics and
groundwater transport.

     There is not sufficient information on the ocean disposal
assessment model to discuss it at this time.

                         BENEFIT/COST MODELING

     The benefit/cost analysis program to support the LLW waste
standard will be different and more extensive than the one conducted

-------
                                  14
for the HLW  standard.  To  begin with, we have extensive experience
and impact and  cost  data from operating SLD facilities.  These data
can be used  to  develop a base case against which other disposal
methods can  be  compared.   For HLW, there was no such experience  or
data.  There are also at least ten alternatives for disposing of LLW
for which the benefits and costs can realistically be estimated  and
compared.  No comparison was made for HLW.  Finally, the length  of
time which LLW  must  be retained is generally much less than for  HLW;
in most cases,  only  several hundred years or less rather than
thousands of years.  Thus, the period of analysis can be much
shorter for  LLW.

     The benefit gained in waste disposal, as used herein, is the
reduction in health  effects gained from taking a control action.
The greater  the reduction  in health effects, the greater is the
benefit.  The benefits, or reduction in health effects, which may
be gained from  each  disposal alternative will be calculated using
one of the SLD, DLD, or OD assessment models discussed earlier.
To help the  user, the benefits will be expressed, so far as
possible, in several forms including radionuclides released to
the environment and  doses,  health effects and health risks to
individuals  and populations.

     On the  cost side of the analysis, one, and possibly two,
analyses will be made.  The projected direct costs for controlling
or disposing of a unit of waste,  whether it be in specific activity
or unit volume, will be estimated for each disposal alternative
considered.  These costs will be useful for the tradiational
"cost-effectiveness" types of analysis.  They will also furnish
guidance on  how much is being spent on the control of LLW.
A model(s) or method(s) will be developed to calculate and compare
these direct costs for the different disposal methods.

     A second,  broader type of economic analysis may be required,
depending upon whether EPA internal triggering criteria relating
to total dollars for costs of compliance, increases in energy
consumption, etc., are met.(8)  The purpose of these economic
analyses would  be to fully identify, on a national basis, the
overall benefits and costs of the regulations; not just the direct
costs.  They would include preparation of a profile of the affected
sectors of the  economy including producers, consumers, and
industry.  The  following impacts would be analyzed: price effects;
production effects;  industry growth; profitability and capital
availability effects; employment effects; balance of trade effects;
and energy effects.

-------
                                 15

              CURRENT  STATUS  OF  MODEL  DEVELOPMENT PROGRAM

     As noted earlier, our general modeling program is directed at
making realistic technical analyses of the health risks and costs of
disposing of LLW by a number of practical disposal alternatives.
The major steps and flow of achieving this goal includes: defining
our modeling needs and approach; developing the models; compiling
data on the wastes, disposal alternatives and actual disposal sites
to use in the models; and conducting the analyses.

     Definition of our modeling needs and approaches to environ-
mental assessment modeling is being done in-house.  The concept and
design of our SLD model is complete.(16)  Design is in progress on
our DLD and cost models.  Design of the OD model has not begun yet.

     Specific guidance for development of the infiltration, leaching,
and groundwater transport submodels for the main SLD assessment
model release was developed in-house by Dr. C. Y. Hung.(16,17,18,19)•
The submodels for the atmospheric and terrestrial pathways submodels
and dose and health risk which will be used in the SLD model are
models which were or  are being developed for EPA for use in the
Clean Air Act.(20,21)  They were developed by Oak Ridge National
Laboratory  (ORNL).  ORNL is currently developing our SLD assessment
model under an interagency agreement with DOE.  This includes
development of several additional submodels and, as necessary,
modifying and coding  all of the submodels into an overall main
assessment model.

     The overall design and submodels of the DLD assessment model
will closely parallel the SLD model.  The major difference is that
the DLD model will be used to analyze the impact of disposing of LLW
at depths below 50 meters and generally below the water table.
The DLD model will probably include the modification and combination
of submodels from our HLW model and the SLD model currently being
developed.  However, the dose/health risk model used will be the one
from the SLD model, which is different from the one currently used
in our HLW model.  Work on the DLD model has not begun yet, nor has
work on the cost model.

     Disposal of LLW on and beneath the ocean floor, which includes
two of ten disposal options currently considered practical, is
regulated under different authorities than land disposal.  The ocean
disposal program is moving at a different pace than the LLW
program.  At present, evaluations of former ocean disposal sites,
development of siting criteria, and evaluations of ocean disposal
technology are under way.  However, development of an environmental
assessment model for ocean disposal appears to be several years in
the future.

-------
                                  16
                       INPUT DATA FOR THE MODELS

     With an estimated 20 million dollars or more per year being
spent by other Federal Agencies on research and development for LLW
management, we believe that most of the data needed to support EPA's
models is either already available or is being collected by others.
Therefore, most of EPA's efforts will be directed at collecting,
collating and critically evaluating data from others.  We are
presently carefully reviewing the information available on
radioactive and hazardous waste-related projects funded by other
Offices within EPA, other Federal agencies, the States, and
industry, to facilitate rapid collection of the needed data.
Gaps in our data needs will be filled as necessary.

     Field data will be collected from three existing LLW disposal
sites to provide actual or, at least, realistic data for input into
the SLD assessment model and for making the cost/risk analysis for
disposing of LLW by this method.  The sites are Barnwell (SC), West
Valley (NY), and Beatty (NV) which represent the three general
regional hydrogeological and climatic settings we intend to model.

     Two categories of site data will be collected; one from
existing data, including published and unpublished data, and the
other from field observations.  Existing data will be used whenever
possible.  ORNL is currently preparing a list of data needs for the
SLD model and is compiling some of the existing data.  Collection of
additional data to fill in the gaps between the model data needs and
the existing site data will start shortly.

     We currently have an in-house project for compiling and
evaluating information on waste characteristics and on improved
methods for treating, packaging, and managing LLW.  This information
will be used to estimate the potential costs and reductions in
health effects which can be achieved through waste management
alternatives which already exist.  Information is being collected
on:  the character and quantities of primary wastes being produced
by typical waste generating facilities before treatment and
packaging; what kinds of segregation and elimlination of wastes
may be done within each facility; the treatment, volume reduction,
solidification, and packaging options that are available for the
primary wastes and the characteristics and quantities of the
resulting wastes; the hazard potential of the various wastes; and
projections of future wastes based on several optimum management
and treatment modes.

     We will soon begin to collect information and develop the
general characteristics of the  natural (geological), engineering,
and institutional barriers which are currently being used, or could

-------
                                    17
 be used in  the  near  future to retain LLW when disposed  of by  shallow
 land disposal and  other selected land disposal alternatives.   The
 types of  land disposal methods to be considered was  discussed
 earlier.  Information to be collected includes; technical and cost
 data on SLD as  presently practiced, improved SLD, and the other land
 disposal  alternatives; site selection criteria for all  land disposal
 alternatives; and  post-operational controls for all  land  disposal
 alternatives.

            TIMING FOR  DEVELOPMENT  OF  LLW MODELS AND STANDARD

      The  schedule  for development of the LLW standard,  development
 of the models,  technical, environmental and cost assessments, and
 collection  of data to support the assessments is shown  in Figure 1.
     PROPOSED

     FINAL

  SLD MODEL

  OLD MODEL
1980     I    1981    I    1982    j    1983     |   1984

       I

       K

       h_
  STANDARD                      .                    n
       I     A	y
  OD MODEL                      I  No Schedule at This Tin

  COST/BENEFIT ANALYSES            ,      A_
     OF LAND DISPOSAL
Figure 1.   General schedule for the development of EPA's LLW assessment
models, analysis  of the risks, costs and  benefits of different disposal
alternatives,  and development of the LLW standard.

-------
                                 18
                              REFERENCES

1.    U.S. Atomic Energy Commission,  Draft Generic Environmental
     Statement for Mixed Oxide Fuel  (GESMO),  WASH-1337 (1974).

2.    Dames and Moore, Inc.,  "Review  of Low-Level  Radioactive Waste
     Disposal History," an unpublished report for USNRC on LLW
     Disposal Technology (1980).

3.    U.S. Environmental Protection Agency, Summary Report on the
     Low-Levej^Radioactive Waste  Burial Site,  West Valley, New York
     (1963-1975), EPA-902/4-77-010  (Feb. 1977).

4.    U.S. Code of Federal Regulations, Reorganization Plan No.  3,  of
     1970, (1970).

5.    U.S. Code of Federal Regulations, Atomic Energy Act of 1954,  as
     amended, (1970).

6.    Office of the U.S. President, Comprehensive  Radioactive Waste
     Management Program - Message from the President of the United
     States, (Feb. 12,  1980).

7.    U.S. Code of Federal Regulations, National Environmental Policy
     Act of 1969, (1969).

8.    U.S. Environmental Protection Agency, Improving Government
     Regulations, (Jul. 1978).

9.    U.S. Environmental Protection Agency, Considerations of Health
     Benefit - Cost Analysis for  Activities Involving Ionizing
     Radiation Exposure and Alternatives,  EPA 520/4-77-003, (1977).

10.  U.S. National Research Council, Volume II, Decision Making in
     the Enviromental Protection  Agency,  A National Research
     Council Report to EPA on Environmental Decision Making, (1977).

11.  International Commission on  Radiation Protection, Publication
     26:  Recommendations of the  International Commission on
     Radiation Protection, (Jan.  1977).

12.  U.S. Enviornmental Protection Agency, Criteria for Radioactive
     Wastes, Recommendations for  Federal Radiation Guidance,
     (Nov. 15, 1978).

13.  U.S. Environmental Protection Agency, Generally Applicable
     Standard for the Management and Disposal of High-Level
     Radioactive Wastes, unpublished draft (1980).

-------
                                 19
14.   U.S.  Nuclear  Regulatory Commission,  Draft  Programmatic
     Environmental Impact  Statement  Related  to  Decontamination  and
     Disposal  of Radioactive Wastes  Resulting from March 28,  1979,'
     accident.  Three Mile  Island Nuclear  Station,  Unit 2,
     NUREG-0683,  (Jul.  1980).

15.   Atomic Industrial  Forum,  A Survey and Evaluation of Handling
     and Disposing of Solid Low-Level  Nuclear Fuel Cycle Wastes,
     AIF/NESP-008ES, (1977).

16.   U.S.  Environmental Protection Agency, Technical  Specifications
     for Environmental  Assessment Model Development,  unpublished
     TAD/ORP document,  (1979).

17.   G.L.  Meyer and C.Y. Hung,  "An Overview  of  EPA's  Health Risk
     Models for Shallow Land Disposal  of  Low-Level Radioactive
     Waste," Proceedings of DOE/EPA/NRC/USGS Interagency Workshop on
     Modeling and  Low-Level Waste Management, Denver, Colorado,
     (Dec. 1980).

18.   C.Y. Hung, "Prediction of Long-Term  Leachability of a
     Solidified Radioactive Waste from a  Short-Term Test by a
     Similitude Law for Leaching Systems," Proceedings of ORNL
     Conference on the Leachability  of Radioactive Solids,
     Gatlingburg,  Tennessee, (Dec.  1980).

19.   C.Y. Hung, "An Optimum Model  to Predict Radionuclide Transport
     in an Aquifer for Application  to Health Risk Evaluations,"
     Proceedings of DOE/EPA/NRC/USGS Interagency Workshop on
     Modeling and Low-Level Waste Management, Denver, Colorado,
     (Dec. 1980).

20.   U.S. Environmental Protection  Agency,  NEMOS (Atmospheric)  and
     TERRA (Terrestrial) Transport  Codes, under development by  ORNL,
     (1980).

21.   U.S. Environmental Protection Agency,  ANDROS, DARTAB and
     RADRISK (Dose and Risk Factor)  Codes,  under development by
     ORNL, (1979).

-------
           MODELING IN LOW-LEVEL RADIOACTIVE WASTE  MANAGEMENT
              FROM THE U.S. GEOLOGICAL  SURVEY  PERSPECTIVE*

                            John B. Robertson
                         U.S. Geological Survey
                            Reston, Virginia
                                ABSTRACT

          The United States Geological Survey (USGS) is a  long-
     standing proponent of using models as tools in geohydrologic
     investigations.  These models vary from maps and core samples
     to elaborate digital computer algorithms, depending on the
     needed application and resources available.  Being a non-
     regulatory scientific agency, the USGS uses models primarily
     for:  improving modeling technology, testing hypotheses,
     management of water resources, providing technical  advice to
     other agencies, parameter sensitivity analysis, and deter-
     mination of parameter values  (inverse problems).   At low-level
     radioactive waste disposal sites, we are most interested in
     developing better capabilities for understanding the ground-
     water flow regime within and  away from burial trenches,  geo-
     chemical factors affecting nuclide concentration and mobility
     in groundwater, and the effects  that various changes in  the
     geohydrologic conditions have on groundwater flow and nuclide
     migration.  Although the Geological Survey has modeling
     capabilities in a variety of  complex problems, significant
     deficiencies and limitations  remain in certain areas, such as
     fracture flow conditions and  solute transport  in the unsatu-
     rated zone.  However, even more  serious  are the deficiencies
     in measuring or estimating adequate input  data for models and
     verification of model utility on real problems.  Flow and
     transport models are being used  by the USGS in several low-
     level disposal site studies,  with varying  degrees of success.
                              INTRODUCTION

     A model generally means a simulator of a thing or process that
cannot easily be observed directly.  This broad definition is applicable
to use of the term in the U.S. Geological Survey (USGS) and in this
presentation.  Everyone uses models, whether they be analogue, mathemati-
cal, or conceptual.  Our favorite photographs of people and scenes are
*
 This paper is prepared for inclusion in  the  program of the  Interagency
 Workshop on Modeling and Low-Level  Waste Management,  December  1-4,  1980,
 in Denver, Colorado, sponsored  by  the  DOE-NRC-EPA-USGS Interagency  Low-
 Level  Radioactive Waste Committee.

                                  21

-------
                                    22

simplified models  of more  illusive and complex real-world conditions.
What  driver  hasn't used a  road map as a scale model of the real highway
system?   (That  example alone quickly teaches us lessons in the problems
of model  inaccuracies.)  Nearly all of us use our car radio or home
stereo to simulate our favorite live-music performances.

      Scientific models are often thought of as mysterious, electro-
mathematical  devices fully understood by no one and worthy of consider-
able  distrust.  However, the most commonly and confidently used models
in science and  technology are similar to those used by most everyone
else  — photographs, drawings, diagrams, and simple equations.  The types
of models chosen,  and the ways in which they are used, simply depend on
the desired  output and available resources.  Types and applications of
models used  by  the USGS are those appropriate to the Agency's mission
and its resources.  The same is generally true of other agencies and
organizations.  Before elaborating on the USGS's use of models in low-
level waste management technology, it would seem appropriate first to
describe  the mission and philosophy of the Survey as it relates to
radioactive waste  disposal.  This should help clarify the differences in
modeling  philosophies and approaches between the Survey and various
other Federal or State agencies.   These differences often appear confus-
ing,  conflicting,  or even contradictory, though in reality they are
usually complimentary.
                            WHAT IS THE USGS?

     The Geological Survey is an old-line Federal agency established by
an act of Congress in 1879 and charged with the responsibility for
"classification of the public lands and examination of geological struc-
ture, mineral resources, and products of the national domain."  It is the
Nation's principal source of information about its physical resources -
the configuration and character of the land surface; the composition and
structure of the underlying rocks; the quantity, extent, and distribution
of water and mineral resources.  In contrast to many agencies, the
Survey's mission has changed very little in more than 100 years; it
remains principally a scientific research and fact-finding agency,
rather than an action or regulatory organization.

     Most of the USGS activities in waste disposal stem from its respon-
sibility to provide data on surface and groundwater quality essential to
the development and conservation of water resources.  Fundamental in
this mission is the development of appropriate technology and tools to
better assess physical and chemical behavior of hydro!ogic systems and
their response to stress.

     The Water Resources Division of the Survey has a line item in its
appropriation to conduct investigations and research aimed at establish-
ing a technical basis upon which earth science criteria can be developed,
tested, and enforced by other agencies for the selecting and operating
of low-level waste disposal sites.  This program includes several field

-------
                                  23


investigations of existing commercial  disposal  sites and more fundamental
research into such fields  as  groundwater geochemistry and modeling
techniques.

     As the Federal Government's  principal  earth-science research agency,
a fundamental part of the  USGS role  is that of  technical  advisor and
consultant to other Federal agencies and states.   The Survey  is  often
mandated by the President  or Congress  to provide  specific technical
services to Federal and state programs.  A  recent example is  our Presi-
dentially specified participation in the National high-level  waste
management program.  Our role as  technical  consultant to other agencies
includes critical review of proposed regulations  and criteria, investiga-
tions of Federal waste disposal sites, review of  proposed and ongoing
programs, cooperative investigation at the requests of states, and  other
related investigations, generally of nonregulatory nature.  It is
generally the policy of the USGS to avoid specific disposal site selec-
tion and characterization studies, preferring instead to provide technical
advice and tools for other groups to conduct such investigations properly.

     Specific states, for example, have come to  us for help in searching
for a low-level waste burial site.  Rather than  suggest  specific sites,
we recommend approaches based on our  knowledge of positive and negative
factors that appear important in site  performance.   In addition, we
offer any regional, areal, or local data which we have gathered that
could be pertinent to their considerations.  Our role is restricted
solely to earth  science considerations, whereas  other agencies such as
EPA and NRC must consider much broader social, political,  legal, and
technical issues.  Therein lies the fundamental  difference between USGS
and the other, more action and regulation-orientated, agencies.


                    MODELING APPLICATIONS  IN THE USGS

     The Geological Survey has long held a keen  interest in development
and application of models as tools for quantitative  analysis of complex
hydrologic problems.  Models have always been regarded by the Survey as
means, rather than ends, to problem solving.  The principal use  of
models is to aid in gaining a more quantitative  understanding of complex
hydrologic systems involving many interdependent physical and chemical
processes and boundary conditions.  Waste disposal and contamination
problems represent some of the most complex problems investigated by the
Survey.  For many of these problems, modeling is  the only means  we have
of examining all major components of a system together.

     Another key application of models is to test and evaluate our
concepts of system dynamics.  Interpretations of  field conditions gener-
ally must be based on very limited field data.  The integrated picture
developed from a few point-observations can often turn out like  the
blind men and the elephant.  If one set of data indicates the elephant
is like a snake, a model can be constructed to  simulate  the snake-like
attributes.  If the-model  is good, it  would soon  indicate that a snake's

-------
                                   24


characteristics and  behavior are different from those of the elephant's
trunk, suggesting  that our  initial impression was wrong and that more
data are  required.   Eventually we might obtain data from enough parts of
the elephant  to construct a model that confirms our blind interpretation
of the beast.

     Models have been used  extensively to help interpret hydrology and
decipher  performance of existing low-level radioactive waste burial
sites.  These sites  represent long-term complex experiments whose results
we are currently observing  and recording.  The unknowns in these experi-
ments are the experimental  conditions.  The challenge then, is to deter-
mine the  experimental conditions that produced the results currently
observable.   Models  are one of the few tools that can allow us to test
various combinations of conjectured conditions.  One of the most important
outputs of this type of modeling is a sensitivity analysis - a ranking
of processes  and parameters according to the magnitude of their influence
on the result of interest (perhaps the concentration distribution of a
particular nuclide).  These results then serve as guidelines for more
detailed  studies.  Thus, the model output (combined with all of our
other tools)  can be  invaluable in selecting which segments of a system
deserve more  attention.

     Our  philosophy  is to engage the use of modeling early in any complex
study as  a feedback  tool — repeatedly adjusting the model, in an iterative
procedure, as  new  information is gathered, accompanied by corresponding
adjustments in data-gathering and testing priorities, according to
modeling  results.  This application of modeling is immensely helpful in
establishing  earth-science guidelines for disposal site selection and
operation.

     Another  example of USGS model employment is in resource and problem
management.   We are  often requested to assess impacts of various develop-
ment options  on an aquifer so that a management agency can make sound
development decisions.  This may involve geochemical and hydraulic
impacts of a  potentially degrading activity such as a waste disposal
facility;  or  it could involve assessment of various remedial approaches
to a contamination problem.

     Finally,  USGS modeling capabilities can be used as an independent
test and  verification of modeling results of other agencies used for
judging the conformance of a particular site to regulatory criteria.
This can  be done directly or by providing the appropriate models to
other agencies for their use.  We view this as one of our most important
roles — developing more accurate, more versatile, and more efficient
models for use by other government organizations.  We maintain several
research  projects dedicated to advancing model technology and teach
courses on modeling  open to some non-USGS personnel.

-------
                                   25


                 WHAT KIND OF MODELS DOES THE USGS USE?

     As mentioned previously, the Geological Survey uses many types of
models, ranging from core samples to elaborate room-sized physical
models.  Although the types of models generally depend on the need and
available resources, USGS modeling can be classified into two general
approaches.

     The first approach is to use a small sample or portion of the real
field system to represent the total system.   A core sample is an example.
In some studies, a core sample might be used for measuring hydraulic
properties which might then be extrapolated  to represent those properties
over a much larger field area.  Water samples similarly are often used
as models to represent chemical characteristics of a larger zone of an
aquifer.  These models generally have very limited use and, in fact,
have been commonly misused.  Hydraulic conductivity measurements on core
samples, for instance, are often orders of magnitude different from
average areal conductivities in many formations.

     The second approach is use of models that simulate the entire
system of interest, generally in some simplified way.  One of the first,
and still the most commonly used model of this type in groundwater
hydrology, is Darcy's Law:
where

     q is the water flux rate, L/T

      K = hydraulic conductivity, L/T

     g— = hydraulic head gradient, dimensionless

     This empirically based model is the basis for nearly all other more
sophisticated groundwater flow and solute transport models.  One of the
more elaborate of these (but still simple) is the Theis equation (Theis,
1935), which is perhaps the most widely used model in the world for
interpreting the hydraulic characteristics of aquifers.  Although very
few actual field conditions meet the strict qualifying conditions for
application of the Theis equation, it has nonetheless enjoyed widespread
success as an approximation.

     Models receiving most attention in recent years are, of course,
ones which approximate solutions to complex partial differential equations
of flow,  solute, and energy transport.   These models range from elaborate
physical  and electrical analogues to ingenious mathematical schemes to
solve huge matrices of simultaneous equation.  These mathematical
approaches include a variety of finite  difference schemes, such as
iterative-alternating-direction-implicit procedure, and various finite

-------
                                   26


element algorithms such as the Galerkin procedure.  All of these methods
and a number of others have advantages and disadvantages for different
applications.  It is not the purpose of this paper to describe or compare
these models.  Anderson (1979) made an excellent review of model use for
simulating contaminant movement in groundwater.

     One of the few examples of the development of application and
verification of a numerical model to a low-level waste problem was by
Robertson (1974).  In that case, the field-observed distributions of
chloride, tritium, and strontium-90 in a large aquifer were successfully
simulated by a two-dimension transient flow and transport model which
included effects of dispersion, radioactive decay, and cation exchange.
Figure 1 indicates one of the results of that modeling effort for tritium.

     We have recently attempted to model groundwater flow (and in some
cases, solute transport) at four of the commercial low-level radioactive
waste disposal sites with varying degrees of success.  Two of these
investigations are described in other presentations of this meeting.


                       USGS MODELING CAPABILITIES

     Models and modelers developed in the Geological Survey are capable
of modeling groundwater flow in stimulated, three-dimensional, hetero-
geneous, anisotropic, transient conditions, on a fairly routine basis.
Non-reactive solute transport, including dispersion, can be modeled
under similar conditions.  Transport of reactive contaminants can be
modeled in two dimensions for simple linear sorption and first-order
rate reactions.  Multi-component reactive solutes can generally be
modeled only in one dimension.  Multi-component chemical equilibrium
models, such as WATEQ, are used routinely for complex solutions in a
non-transport mode.  Flow in the unsaturated zone can be simulated for
fairly simple transient non-uniform conditions with no hysteresis;
however, solute transport modeling in the unsaturated zone is in its
infancy.  Other papers in these proceedings review the state-of-the-art in
the USGS as well as other models in saturated and unsaturated ground-
water conditions and in streams.

     We do not yet have satisfactory modeling capabilities in the fol-
lowing areas:

     •  fractured rock systems, such as Maxey Flats, Kentucky, except
        for very high and/or very uniform fractures systems;

     •  flow in very dry, heterogeneous unsaturated zones, such as Beatty,
        Nevada;

     •  unsaturated flow with hysteresis or reactive solute transport;

     •  flow in any media involving complex multicomponent reactions or
        non-first order rate reactions;

-------
                                       27
 Solid waste
burial
               Explanation
           Map  adapted from Robertson  (1974)
     Equal  Tritium Concentration in pico curies per milliliter for 1968
     ^-50  Well  samples
      '"—50  Digital  model
      INEL  = Idaho National Engineering Laboratory
      ICPP  = Idaho Chemical Processing Plant (nuclear fuel reprocessing)
       TRA  = Test  Reactor Area
     EBR-1  = Experimental Breeder Reactor 1 site (inactive)
       CFA  = Central  Facilities Area

      INEL  boundary
    Figure  1.  Comparison of  ICPP-TRA waste tritium plumes in the Snake River
              Plain aquifer  for 1968 based on well sample data and computer
              model.

-------
                                  28


     •  flow involving transport and biologically-controlled reactions.

Although these limitations are severe, we have research proceeding on
all these subjects.

     For most situations amenable to modeling, our models are better
than our field data.  Field data on dispersivity, chemical reaction
parameters, and other inputs are grossly inadequate at most sites of
concern.  Therefore, in order to build new and better models or even to
apply existing models, development of field and laboratory testing
technology must proceed at comparable or faster rates.  Deficiencies in
our ability to measure and estimate model input parameters are actually
more serious than model limitations and currently are deserving of more
research attention.  Related to this is a continuing need to enhance the
credibility of models with many more applications and demonstrations on
well-documented field problems.
                               CONCLUSIONS

     The USGS is a strong proponent of the use of appropriate models in
all hydro!ogic problems.  The more complex the problems such as those
involving waste disposal and groundwater contamination, the greater the
need for modeling and the more complex are the models needed.  Models
are only tools to be used in conjunction with, not a substitute for,
investigative analytical tools.

     In waste disposal studies, the USGS needs and uses models for the
following main purposes:

     •  to learn how to build better models;

     •  to test and confirm theories and concepts;

     •  to aid in managing water resources and stresses upon those
        resources;

     •  for intellectual stimulation;

     •  to gain competence and experience in order to advise other groups
        on model application;

     •  sensitivity analysis to screen or rank interacting processes and
        parameters;

     •  to "back-out" values of parameters such as dispersivity, distribu-
        tion coefficients, and hydraulic conductivity (inverse problems);

     •  to gain better general insight into complex multi-component
        systems.

-------
                                   29


    Although  significant deficiencies exist in current models,  the
greater  deficiency  is  in our ability to provide adequate input informa-
tion to  models and  to  demonstrate model accuracy and reliability on real
field  problems.

-------
                                  30


                               REFERENCES

Anderson, Mary P., 1979, Using models to simulate the movement of contam-
     inants through groundwater flow systems:  CRC Critical Reviews in
     Environmental Control, November 1979, pp.  97-155.

Robertson, J. B., 1974, Digital modeling of radioactive and chemical waste
     transport in the Snake River Plain Aquifer of the National Reactor
     Testing Station, Idaho:  U.S. Geological Survey open-file rept.
     ID022054.

Theis, C. V., 1935, The relation between the lowering of the piezometric
     surface and the rate and duration of discharge of a well usins
     groundwater storage:  Trans. Amer. Geophysical Union, vol. 16,
     pp. 519-524.

-------
                LOW-LEVEL RADIOACTIVE WASTE MODELING NEEDS
                        FROM THE USNRC PERSPECTIVE

                               R. Dale Smith
                             Edward F. Hawkins

                     Low-Level  Waste Licensing Branch
                       Division of Waste Management
                    U.S. Nuclear Regulatory Commission
                          Washington, D.C.   20555
                                 ABSTRACT

     The NRC is  developing a low-level  waste regulation to be issued
for public  comment  by April  30,  1981.  Publication of the final  rule
and final EIS is expected in 1982.   This regulation, 10 CFR Part 61
will  apply  only  to  near surface  disposal.  Two basic approaches  have
been considered  in  developing specific  regulations:  a prescriptive
approach and a performance objective approach.

     To accomplish  the licensing and regulation of low-level waste
disposal  facilities,  the information that will be needed and the
methodologies to be used to assess  performance are being established.
Regulatory  guides will cover site selection and characterization,
facility design  and operations,  waste classification, waste form
performance, site closure, post-operational surveillance, funding
and monitoring.   In developing licensing review procedures, initial
efforts are to establish analytical  modeling capabilities.  We are
establishing our modeling needs, determining models availability,
obtaining potentially useful models and adopting those that meet
our needs.   As a test case,  the  Barnwell facility will be modeled
and its performance assessed. Our  analyses will then be compared
with those  of our contractor, ORNL  and  the Licensee.
     The NRC schedule for promulgation of a low-level waste regulation
calls for issuance of the notice of proposed rulemaking along with a
supporting draft EIS by April  30, 1981.  Publication of a final  rule and
final EIS is expected in 1982  and will depend on the nature and extent
of comments received and whether a formal hearing is held.
                                 31

-------
                                 32
     Part 61 will apply to the near surface disposal  of waste by such
means as shallow-burial, engineered structures,  and deeper burial.
The disposal of waste by other methods (e.g., deep mined cavity or
other deep burial) will not be dealt with in this rulemaking action
and will be considered at a later time through a separate rulemaking.

     In developing specific regulations, we have considered two basic
approaches:  a prescriptive approach and a performance objective
approach.   In the former approach, specific detailed requirements for
design and  operation of a waste disposal facility would be set out in
the regulations.  Prescriptive standards would specify the particular
practices,  designs, or methods which are to be employed—for example,
the thickness of the cover material (the cap) over a shallow land
burial disposal trench, or the maximum slope of the trench walls.

     Setting of  prescriptive standards requires a considerable amount
of detailed knowledge  about potential designs, techniques, and proce-
dures  for disposing of waste in order to prescribe which designs,
techniques, and  procedures are best and also assumes that the state-
of-art  in waste  disposal  is developed to the point where there are
clear  choices to  be made  among all the potential approaches.

     A regulation oriented toward  performance objectives, on the other
hand,  would establish  the overall  objectives to be achieved  in waste
disposal and would  leave  flexibility  in how the objectives would be
achieved.   The  performance standards  would specify levels of impact
which  should not be exceeded at a  disposal facility  in  order to
provide protection  against radiological  hazards.

     NRC staff  believes  that  neither a  purely prescriptive  nor  a
performance objective approach will  be  satisfactory  in  Part  61.  A
combination of  both approaches is  needed  such that performance  ob-
jectives are established to  define the  overall  performance  expected
in  disposal while at  the same  time,  certain  specific minimum tech-
nical  criteria  would  be established.

      NRC's overall  goal  is  to assure protection of the  public health
and safety.  In considering  radioactive waste disposal, this goal
would  appear to fall  into two time frames.  That  is, we must be
concerned  with  protection of workers and the public  during  the short-
term operational phase and protection of the public  over the long

-------
                                 33
term after operations cease.  Thus, the near-term performance objective
will be to assure that the disposal facility will be operated in com-
pliance with existing standards as set out in Part 20 of our regula-
tions.   Although routine operational releases are not expected at a
disposal facility, they might occur in those cases where waste pro-
cessing or volume reduction operations are conducted at a disposal
site.  In such cases we would propose to apply the same limits that
would apply if these operations were conducted at fuel cycle facili-
ties (40 CFR Part 190).

     Assuring safety over the long term involves two considerations:
1) protection of an individual who might unknowingly contact the
waste at some point in the future; and 2) protection of the general
public from potential releases to the environment.  We believe that
intentional intrusion into the waste—e.g., an archaeologist
reclaiming artifacts-cannot reasonably be protected against.  After
active institutional controls cease, however, one or a few individuals
could inadvertently disturb the waste through such activities as
construction or farming.  It is important to note that consideration
of intrusion involves hypothetical exposure pathways which are assumed
to occur.  Actual intrusion into the waste may never occur.  But,
for purposes of Part 61, intrusion is considered such that if it
should occur the one or few individuals contacting the waste should
not receive an unacceptable exposure.  With respect to the unknowing
intruder, we are presently considering use of the Part 20 maximum
individual dose limit of 500 mrem/yr to assure protection of such an
intruder.

     With respect to migration and protection of the ground water,
we are evaluating a range of exposure guidelines from 1 to 25 mrem/yr
that would be applied at the site boundary.  We expect our analyses
will result in a limit around 4-5 mrem/yr.

     I'd like to review what these performance objectives mean
in terms establishing minimum technical requirements; that is, the
prescriptive part of the regulation.  These requirements would in-
volve placing controls on the various parts or barriers of the
overall disposal "system" and will determine which wastes are ac-
ceptable for near-surface disposal and which wastes are not and must
therefore be disposed of by other methods providing much greater
confinement.  The principal parts or barriers of an overall LLW
disposal system that are readily identifiable and will be addressed

-------
                                 34
in the prescriptive requirements are:

—  the characteristics of the waste and its packaging;

—  the characteristics of the site into which the waste is placed;

—  the methods by which the site is utilized, the waste emplaced and
    the site closed; and

—  the degree and length of institutional  control, surveillance,
    and monitoring of the site after closure.

     It is instructive to start with the intruder performance objec-
tive since although it deals with a hypothetical  pathway, it provides
a basis for several limiting requirements and establishes two
"classes" of waste suitable for near-surface disposal.

     There are two principal means of controlling potential exposures
to an intruder—use of institutional controls and use of natural or
engineered barriers which would make it more difficult for a potential
intruder to contact the waste.  Institutional controls can be further
separated into two types:  (1) "active" controls, which require per-
formance of some action by a person or agency-e.g., controlled access
to the site, including barries to entry (fences)  and periodic inspec-
tion and monitoring of the disposal site by a regulatory or other
governmental agency; and (2) "passive" controls,  which are not
specifically dependent upon human actions-e.g., government ownership
of land, redundant records and restrictions on land use.  Natural
or engineered barriers could include deeper burial or use of caissons
backfilled with concrete.

     Neither institutional controls nor engineered barriers can be
relied upon to completely prevent human intrusion.  Although we have
not completed our final analyses we expect to allow reliance on
active institutional controls to prevent intrusion for about 100
years.  This will mean that certain wastes (e.g., those with short
half-lives) will not present an undue risk to an intruder when
active institutional controls are removed.  The active controls will
be followed by several hundred additional years of passive controls.

-------
                                 35
Passive controls,  although certainly minimizing potential  for inad-
vertent intrusion  may not be completely effective in preventing it.
A second class  of  waste would then be that which may present an undue
risk to an intruder when active controls are removed.  For this waste
additional  measures must be taken to reduce this risk.  This can be
done by introducing an additional barrier to intrusion, such as
deeper burial or engineered barriers.  These additional measures
will not be relied on for more than 500 years; thus waste which
could potentially  result in an exposure to an intruder greater than
500 mrem after  500 years is not acceptable for near surface disposal.

     A similar  type of analysis for migration cannot be performed in
as direct a manner.  While intruder exposures are relatively non-site
specific, potential ground-water releases are very site specific and
are more difficult to treat in as generic a fashion as the intruder.
In order to carry  out a site specific analysis, the variables beyond
the control of  the site operator need to be fixed to some extent.
Of principal importance is the long-term stability of the waste and
disposal facility.  Unless the waste and the disposal site are stable
over time, there is no way to predict the long-term radiological
impacts of disposal, or the activities (maintenance, monitoring, etc.)
and associated  costs required to maintain potential impacts to low
levels.  This would suggest that all waste should be placed in a
solid, non-compressible form for disposal in the near surface.
However, many wastes—particularly trash waste streams—contain very
low levels of activity and present little or no concern from the
standpoint of long-term migration.  Thus, a more reasonable require-
ment would be that compressible waste containing low-levels of activity
should be segregated and disposed of separately.  All other waste
would be placed in a solid, non-compressible form (e.g., solidifica-
tion in the case of resins and filter sludges) and the disposal
facility should be designed and operated to enhance stability over the
long-term.  Certain nuclides, concentrations of nuclides or waste
forms may also  require individual consideration from the standpoint
of migration on a  site specific basis.  This might include, for example,
a large one-time shipment of a very mobile radionuclide or a large
quantity of chelating agent in a single shipment.  Thus, migration
considerations  establish "classes" of wastes that require special
disposal considerations.

-------
                                  36
     We expect to establish a common sense base of siting requirements
that could be consistently applied throughout the country.  The require-
ments would essentially eliminate certain limited areas from consider-
ation due to undesirable characteristics leaving large areas in each
region where acceptable sites could be found.  The kinds of require-
ments we would expect to establish include:

     1)  Avoid areas of fractured bedrock;

     2)  Avoid siting in wetlands, high hazard coastal areas, flood-
         plains, swamps or other types of very wet or potentially
         very wet terrain;

     3)  The land surface should be relatively stable structurally
         and geomorphically;

     4)  The hydrogeologic conditions of the site should be simple
         enough for reliable residence time prediction to be made; and

     5)  The site should provide long travel times for radionuclides
         from disposal trenches to the biosphere.

     The completion of Part 61 does not, however, mean that our job
is done.  To accomplish the licensing and regulation of low-level
waste disposal facilities, we are also establishing the information
needed and the methodologies we will use to assess the performance of
low-level waste disposal facilities.  These efforts are concentrated in
two activities—the development of regulatory guides to support the
regulation and the development of licensing review procedures.  The
regulatory guides are being initiated more or less in parallel with
our efforts on licensing review procedures.  As we now envision it,
regulatory guides will be developed in the areas of site selection,
site characterization, facility design and operations, waste classi-
fication, waste form performance, site closure, post-operational
surveillance, funding and monitoring.  As with the regulation, these
guides will be directed at disposal by means of shallow and inter-
mediate depth land burial.  Future guides, as needed, will be
directed at other methods of disposal.

-------
                                  37
     To  develop our  licensing review procedures,  that is, the way
we will  review applications for low-level  waste disposal, we are
initially concentrating  on methodologies we will  use to assess per-
formance.  The first step in this  development is  to establish our
capability to perform analytical modeling.  Although NRC has extensive
experience in modeling and some modeling has been done of low-level
waste disposal in the past, a systematic approach is now being taken
to establish a program to assure that our analytical capabilities
are in place as soon as  possible.   To accomplish  this, we have em-
barked on  an effort  to establish our modeling needs, determine the
availability of models to meet these needs, obtain prospective models,
develop  our operational  capability of the selected models, and test
the models on existing or proposed sites.   Let me make it clear that
it is not  our intention  to develop a new set of models.  Rather, we
will investigate models  that are already in use and adapt, or adopt
them for our use.  If our investigations reveal  that there are no
acceptable models to meet a specific need, we will then initiate a
program  of model development.

     Our first task  will  be to develop screening  criteria to be used
to select  potential  models that can be used to simulate low-level
waste sites and the  effects of construction, operation and ultimate
closure.  Models will  need to be able to simulate the geological
environment, ground  and  surface water pathways, air pathways, biotic
and vegetation pathways, radionuclide transport and the assessment
of potential radiological and non-radiological impacts.  Models
will also be screened in terms of  their level of  detail and our need
for such detail.  In some cases site screening and gross evaluation
type of models may be all that are required; whereas for other cases
very detailed, highly accurate models may be needed to fully assess
impacts.  Perhaps in a few cases a model may need to be developed
that is  applicable to a  specific site.  The screening criteria must
also take into account the need for models in a wide variety of
geologic and climatic settings, e.g., the obvious differences be-
tween arid and humid sites.  This  screening will  also be done in the
context  of the time  and  computer systems available to us.

-------
                                  38
     Through our own efforts and those of our contractors and consul-
tants, our search for available and acceptable models will  begin with
a review of those that NRC already has.  We will  also investigate what
other agencies are doing and what they have available.  Very specifical-
ly, this is one of the primary reasons we are here at this  workshop.   We
are very anxious to find out what you are doing.   Our efforts will
also include the work going on at the National Laboratories, the
universities by private industry and any other source that  we can find.

     As we accumulate models for potential use, we will  evaluate and
categorize them for our particular needs in terms of their usefulness
to us for making licensing decisions.  Key items in our  evaluations
will be:

—  their applicability (humid vs. arid sites, saturated and unsaturated
    flow, single and multiple aquifers, solute transport, dose assess-
    ment, etc.);

—  the details of the model (data requirements;  simple, conservative
    vs. detailed, more realistic; difficulty of use, model  options,
    computer requirements, etc.);

—  availability of documentation (users and programmers manuals,
    sample runs, examples, etc.); and

—  validation and verification of models (theoretical proofs, pre-
    vious applications to field conditions, reproducibility, re-
    peatability, etc.).

     It is obvious that this is a major undertaking both in terms of
time and staff effort.  However, it is necessary if we are to be able
to make knowledgeable licensing decisions in a timely manner.  To enable
us to focus on a specific goal, we have elected to concentrate our
initial efforts on performing an assessment of the Barnwell Low-Level
Waste Disposal Facility in South Carolina as  a test case.  There are
several reasons why we made this choice.  First, we have agreed to
assist the state of South Carolina this next year in the environmen-
tal assessment of the Barnwell site as a  part of their review for the
renewal of the state license for the  faciltiy.  Second,  we have con-
tracted with Oak Ridge National Laboratory to assist us in this ef-
fort, including performance assessment of the facility.   Theirs and the
licensees analyses should provide a good  comparison to our efforts.
Last, since the Barnwell facility has been operating for some time

-------
                                 39
and there are other nuclear facilities nearby that have been studied
extensively, we feel  that data requirements for the selected models
should probably not be a limiting factor.   This should provide us more
flexbility in model selection.

     Once we have completed one test case  (Barnwell)  we will then turn
over efforts to expanding our capabilities to other types of sites
and situations.  Hopefully, we will  not have to go back to the
beginning of the exercise since we should  have learned a great deal
the first time through and should have accumulated some applicable
information.

     In conclusion, I would like to summarize the modeling needs from
our perspective:

     1)  We have identified the need for models that  are directed at
         enabling us  to make licensing decisions.

     2)  We have determined that the ability to perform assessments
         must be contained within our own  staff.

     3)  Our modeling capabilities must be able, eventually, to
         evaluate the full range of potential sites,  facilities,
         designs and  operations that may occur.  We have, however,
         elected to first concentrate on models that  are poten-
         tially applicable to the Barnwell facility,  which will
         be used as a test case.

     4)  We have determined that the best  course for us is to
         adopt, or adapt, existing models  to our use to the fullest
         extent possible.

-------
Technical Papers

-------
            MODELING  CONTAMINANT TRANSPORT  IN  POROUS  MEDIA
           IN  RELATION  TO  NUCLEAR-WASTE  DISPOSAL:   A  REVIEW

                  David B.  Grove and  Kenneth L.  Kipp

                        U.S.  Geological  Survey
                   P.  0.  Box 25046,  Mail Stop 413
                            Federal Center
                        Denver, Colorado 80225
                               ABSTRACT

          The modeling  of  solute transport  in  saturated  porous
     media is reviewed  as  it  is applied  to  the movement  of radio-
     active waste  in  the subsurface.  Those processes, both
     physical and chemical, that affect radionuclide  movement  are
     discussed  and the  references that best illustrate these  pro-
     cesses listed.   Movement is separated  into  convection,
     convection-dispersion, and convection-dispersion and  chemical
     reactions.  Solutions of equations  describing such  movement
     are divided into one-, two-, and three-dimensional  analytical
     and numerical examples.   Discussions of recent  work in the
     area of stochastic modeling are followed  by discussions  of
     applications  of  the models to  selected field sites.
                             INTRODUCTION

     An important consideration  in  the  evaluation  of existing or pro-
posed nuclear waste-disposal  proposals  is  the  impact on the environment
and the exposure to man  of waste material  that may escape containment
and be transported out of the engineered repository.  For the purpose
of this paper, presently available  solute-transport models will  be
reviewed with respect to the  transport  of  radionuclides by ground-water
flow under saturated conditions  from such  a  repository.

     Mathematical modeling of radionuclide transport in porous media is
the only way to quantify transport  rates and discharges.   The steps in
modeling involve:  (1) Formulation  of mathematical expressions that
represent the primary physical  and  chemical  mechanisms involved;
(2) determination of the parameters that appear in these expressions
using measured laboratory and field data;  (3)  finding an acceptable
method of solving these  mathematical expressions;  (4) evaluation of the
acceptability of the formulated mathematical model in reproducing
observed physical and chemical  phenomena;  and  (5)  the application of
the model to the prediction of the  rates of  radionuclide transport and
to evaluate various waste-disposal  proposals.   Examples and references
have been selected that the authors feel  best  illustrate the various
factors that affect such transport.  The  interested reader is referred
to a comprehensive review by Anderson (1979) for more detailed infor-
mation.

                                   43

-------
                                  44
                         THE GENERAL EQUATIONS

     Because the primary mechanism of solute transport in ground water
depends on the direction and magnitude of the ground-water flow, the
equation describing such flow must be formulated and solved with the
appropriate boundary and initial conditions.

     One form of the flow equation for a compressible fluid in a satu-
rated nonhomogeneous anisotropic porous medium is that given by
Konikow and Grove (1977) as:
                                          4.   no
                                    P« at + pone at

                                        s  3m.    + +
                                    v5- E  gT+Wp                (1)
                                    vo *=1  3t
where
                                                                 2
     k.. is the intrinsic permeability (a second-order tensor), L ;
     p   is the fluid density, ML"3;

     y   is the dynamic viscosity, ML  T  ;

     P   is the fluid pressure, ML"1!  ;
                                                        2
     g   is the gravitational acceleration constant,  LT  ;
      *
     z   is the elevation of the reference point  above a standard
         datum, L;
      *    *
     W  = W  (x,y,z,t) is the volume flux per unit volume (positive sign
         for outflow and negative for inflow), T"';
      *                                            -3
     p   is the density of the source/sink fluid, ML   ;

     a   is the vertical compressibility coefficient  of the medium,


     p   is the fluid density at a reference pressure, temperature, and
         concentration, ML"3;

     n   is the total porosity for compressive storage (dimensionless);

     e   is the effective porosity (dimensionless);
                                                             1  2
     3   is the compressibility coefficient  of the fluid, LM T ;
                                               o
     v   is the reference volume of the fluid, L  ;

     m,   is the mass of species k in the reference volume v , M;
      K                                     .               0
     s   is the number of species, (dimensionless);

     x.  are the cartesian coordinates, L; and
     t   is time,  T.

-------
                                  45


     The equation for transport of a dissolved chemical species can be
obtained by combining the principle of the conservation of mass with
Pick's law of diffusion.   Assumptions that result in the following
equation can be found in  Konikow and Grove (1977):


       8C    3
     e "at" ~ "iw~~
       dt   cX •
              T,  \       d /      £-   \

                             - C'W* + £  Rk                        (2)


where
     C   is the concentration of the solute species in the flowing
         phase, ML"3;
     e   is the effective porosity of the medium;
                                                                 2 -1
     D.. is the hydrodynamic dispersion coefficient (a tensor), L T  ;

     V.  is the interstitial  velocity,  LT"1;
     W   is the volumetric flow rate/unit volume of porous medium of a
         source/sink, T~^;
     C1  is the concentration of the species in the source/sink flow,
         ML"3; and
     R,  is the rate of production of solute species in the kth
         reaction T~lL~3.

     Once the set of reactions described by R, is specified, equation
(2) can be written for each chemical species of a system and the set of
equations solved simultaneously with the flow equation.  The boundary
conditions for the solute-transport equation include boundaries along
which the solute concentration is specified, and those along which
continuity of the flux of the solute species is required.  Initial
conditions specifying the solute-concentration distributions must be
given for all species.

     The primary coupling of the flow equation to the solute-transport
equation is through the interstitial velocity obtained from Darcy's law.
The back-coupling or effect of the transport equation on the flow
equation is through the dependence of fluid density and viscosity on
solute concentration.

     For instances where the solute concentrations are very dilute it
commonly is assumed that the fluid density and viscosity are constant.
The flow equation can then be solved independently from the transport
equation.  In essence, the presence of the chemical-solute species does
not affect the flow field, but the flow field is still the primary
mechanism for solute transport.

-------
                                  46


     For studies where a two-dimensional areal model is appropriate,
the above equation can be integrated in the vertical direction through-
out the saturated thickness of the aquifer.  Such situations occur when
horizontal ground-water movement is much larger than vertical movement.

                         TRANSPORT MECHANISMS

     The primary solute-transport mechanism is the flow of ground water
resulting in convection of the solute species, and is represented by
the second term on the right hand side of equation (2).

     Hydrodynamic dispersion, a secondary transport mechanism, des-
cribes the net macroscopic transport effect of movements of the
individual solute particles through the porous medium, Bear (1972).
Hydrodynamic dispersion consists of two processes.  One is diffusion
of solute on the molecular scale described by a Fickian diffusion
term with a molecular-diffusion coefficient.  The other is mechanical
dispersion that describes transport due to the variations of velocity
across individual pores, and the mixing of fluid at pore junctions.
The two coefficients usually are lumped together and the mechanical
dispersion dominates in all but the slowest of flow systems.

     Scheidegger (1961) relates the dispersion coefficient to the
interstitial velocity by the equation:

                 V  V
     D.. = a..   -*-JL                                              (3)
      ^J    i3im  I y I

where
     a..    is the dispersivity or characteristic length of the porous
      •z-jroi  med-jym (a fourth-order tensor), L;

     V  and V- are the components of the flow velocity of the fluid in
      m     the m and « direction, respectively, LT~1; and
     |V|    is the magnitude of the velocity vector, LT~ .

     When the coordinate system is aligned with the average velocity
in a steady, uniform flow field in an isotropic medium, the dispersion
coefficient tensor reduces to two components, a longitudinal and a
transverse coefficient (Scheidegger, 1961, and Bachmat and Bear, 1964,
and Bredehoeft, 1969).

     Recent analyses (Smith and Schwartz, 1980) of the dispersive
effects caused by macroscale heterogeneities in the permeability of a
porous medium, have shown that dispersion of solute at this scale is
not described by a Fickian diffusion term.

     Transport and the associated dispersion and diffusion through
fractures,where there may be a flowing and a stagnant fluid phase, can
be analyzed using the concept of dead-end pores and is discussed by
De Smedt and Wierenga (1979).

-------
                                  47


                          CHEMICAL REACTIONS

     The reaction term includes all  chemical processes that can occur
during transport (Back and Cherry, 1976).  These may include homo-
geneous reactions that occur within the fluid phase and heterogeneous
reactions that take place between the solid and liquid phases.

     An example of a common homogeneous reaction is an irreversible
rate reaction such as radioactive decay.  In this instance the reaction
rate term R in equation 2 becomes:

     R = - x(ec + pb c)                                             (4)

where

     X is the radioactive-decay constant (equals In 2/half life), T  ;
       and

     c is the concentration of the species adsorbed on the solid (mass
       of solute/mass of sediment), ML"3.

Heterogeneous reactions include sorption ion exchange, diffusion into a
stagnant fluid phase with possible sorption or reaction, precipitation
and dissolution.  When heterogeneous reactions are being considered, a
separate mass balance equation for the sorbed or stagnant phase must
be solved simultaneously with equation (2).  For the simple hetero-
geneous reaction such as equilibrium-controlled sorption or ion exchange
with a linear adsorption isotherm the reaction rate term for a single
species can be written as:

     R = - K. p.  T£                                                 (5)
            Q  D dt
where

     K. is the distribution coefficient that defines the relationship
        between the dissolved and adsorbed ions.

Grove (1976) and Enfield and others (1976) discuss basic forms of
sorption and ion-exchange mechanisms.  The sorption mechanisms that
have been simulated to date include kinetically controlled sorption
and equilibrium sorption.

                MATHEMATICAL MODELS OF SOLUTE TRANSPORT

     The solutions to the governing equations of flow and solute
transport for a given system geometry, boundary conditions, initial
conditions, and reaction-mechanism procedure form the mathematical
model of the system.  The following sections summarize transport
models that have been developed which are relevant to nuclear-waste
disposal.  Analytical and numerical  solutions to the models are dis-
cussed separately with these being further subdivided under the
headings of (1) convection, (2) convection-dispersion, and (3)  convection-
dispersion and chemical reactions.  When appropriate, the types of
models are further divided according to dimensionality.

-------
                                  48
Analytical Models
     Convection.  The transport model for convection only in one
dimension is trivial as the output matches the input with only a time
delay for travel.  For two dimensions, in a horizontal plane, Nelson
(1978a, 1978b) gives several examples for travel times and concentra-
tion changes during solute movement from a source to an outflow boundary.

     Convection-dispersion.  For the convection-dispersion model in one
dimension the flow field is usually taken as uniform and steady, thus
eliminating the need to solve the flow equation.  Bear (1972) presents
some solutions for a variety of boundary conditions and source terms
for this case.  Other solutions include those of Brenner (1962),
Lindstrom and Boersma (1971), Gelhar and Collins (1971), and Al-Niami
and Rushton (1977).  For two dimensions, solutions are presented by
Bear (1972), Sauty (1980) and Fried and Combarnous (1971).  For the
cases of radial transport and equation transformation of equations to
simpler forms, results are available from Bear and Jacobs (1965), Bear
(1972), Hoopes and Harleman (1967), and Phillips and Gelhar (1978).

     Convection-dispersion-reaction.  Examples of one-dimensional
solutions that involve linear homogeneous reactions are given by
Bear (1972), Marino (1974a), and Parlange and Starr (1978).  Two-
dimensional examples with similar reactions have been presented by
Hunt (1973, 1974).  Many one-dimensional transport models have been
developed for various heterogeneous reactions.  Examples for linear
equilibrium sorption and linear kinetically-controlled sorption are
given by Lapidus and Amundson (1952), Ogata (1964), Girshon and Nir
(1969), Marino (1974a, 1974b) and Enfield and others (1976).  An
analytical transport model for a three-member radionuclide decay chain
with linear equilibrium sorption and an impulse or decaying-slug source
terms has been derived by Lester and others (1975).  The cases of both
dispersive and nondispersive transport were treated.  Nondispersive
transport was examined in a similar manner by Rosinger and Tremaine
(1978).  Details of the numerical implementation of the model are
given by Kipp (1979) and DeMier and others (1979).

     Examples of analytical solutions for two-dimensional transport
including kinetic sorption and decay are presented by Eldor and Dagan
(1972) and Wilson and Miller (1978).

Numerical Models

     Convection.  The one-dimensional convective-transport case with
constant interstitial velocity can be solved analytically and is of no
interest numerically.  For two dimensions Nelson (1978c, 1978d) gives
several good examples of modeling convective-transport and transient-
flow systems.

     Convection-dispersion.  The numerical solution of the solute-
transport equation with dispersion can be quite difficult because of
numerical-diffusion errors and solution oscillations caused by the

-------
                                  49


particular technique.  Because of these problems a variety of numerical
techniques have been used.  These include finite-difference and finite-
element methods and the method of characteristics (MOC).  Examples
illustrating these problems and the use of various methods to minimize
numerical dispersion (Lantz, 1971) and oscillations have been proposed
by Stone and Brian (1963), Chaudari (1971), and Chhatwal and others
(1973).  Examples of finite-element methods used to overcome such
difficulties for one-dimensional problems are given by Price and
others (1968), Pinder and Shapiro (1979) and Van Genuchten and Gray
(1978).  Huyakorn (1977) proposed using an upwind finite-element method
to overcome the oscillation problem in convection-dominated transport.
Pinder and Cooper (1970) and Reddell and Sunada (1970) describe the
method of characteristics/particle in cell model for transport cal-
culations that overcomes numerical-dispersion and overshoot/undershoot
problems.  A description of the U.S. Geological Survey's MOC model is
given by Konikow and Bredehoeft (1978).

     Peaceman and Rachford (1955, 1962) presented one of the first
finite-difference numerical techniques for two-dimensional transport.
Several of the finite-difference formulations cited previously have
been extended to two-dimensional problems with convective and dis-
persive transport.

     Many two-dimensional finite-element based solute-transport models
have been developed.  Guymon (1970) used a variational principle;
Nalluswami and others (1972) included the mixed partial-derivative
terms; Pinder (1973) used the Galerkin technique; and Smith and others
(1973) demonstrated the greater generality of the Galerkin technique
over variational formulations.  Lee and Cheng (1974) and Segol and
others (1975) used a finite-element method to solve the density
dependent coupled flow and solute-transport equations.

     Huyakorn and Taylor  (1977) compared the relative merits of the
velocity-pressure formulation, the stream-function formulation, and
the hydraulic-potential formulation for the coupled flow and solute-
transport problem and found the first method to be the most accurate,
yet the most expensive.

     The three-dimensional solute-transport models with dispersion and
convection include Khaleel and Reddell's  (1977) and Reddell and Sunada's
(1970) method of characteristics model and Gupta and others (1975)
finite-element model.

     Convection-dispersion-reaction.  The numerical simulation of
convection, dispersion, and reaction of chemical species through
porous media in most instances simply involves the addition of the
reaction term on the convection-dispersion equation.  While most
applications of this are  for one-dimensional column experiments the
extension of the principle to two- and three-dimensional systems is
quite  straight forward.

-------
                                  50


     The one-dimensional PERCOL model of Routson and Serne (1972)
treats convective transport with equilibrium reactions of a set of
"macroions" (ions present in high concentrations).  It then computes
the transport of microions (ions present in low concentrations), which
are radionuclides present in small concentrations and undergo equilib-
rium sorption.  The sorption coefficients are a function of the macroion
concentrations.

     The MMT1D radionuclide-transport model of Washburn and others
(1980) computes the transport of N-member radionuclide decay chains and
includes linear equilibrium-sorption with constant sorption coefficients.
The dispersive transport is simulated by a discrete parcel random-walk
technique described by Ahlstrom and others (1977).  Hill and Grimwood
(1978) also developed a one-dimensional model of radionuclide-chain
transport with linear equilibrium sorption and radioactive decay.

     Finite-difference techniques are illustrated by Lai and Jurinak
(1972) and finite-element methods by Rubin and James (1973).  Both
solve the general differential equation with linear or nonlinear
reactions.

     An early two-dimensional transport model with chemical reactions
was that of Schwartz and Domenico (1973); it used finite differences,
neglected dispersion, but treated a multiplicity of chemical reactions
including both equilibrium and kinetic.  The REFQS model of Shaffer
and Ribbens (1974) also calculated the transport of several chemical
species using the same techniques as Schwartz and Domenico, but
included dispersive transport.

     Schwartz (1975) developed a radionuclide transport method-of-
characteristies model which included binary cation exchange and linear
decay.  Finite-element techniques were used by Duguid and Reeves (1976)
to simulate transport with equilibrium sorption and linear decay; by
Prakash (1976) for transport in a bounded radial-flow system for a
partially penetrating well; by Cabrera and Marino (1976) for two-
dimensional flow and transport in a stream-aquifer system; by Pickens
and others (1979) for transport to tile drains; and by Pickins and
Lennox (1976) for a steady free surface-flow system with constant
equilibrium sorption.

     Ahlstrom and others (1977) presented the Discrete Parcel Random
Walk transport model, MMT-DPRW, which is an extension of the particle-
in-cell method in that the dispersive transport is simulated by a
statistical random-walk computation.  The idea of a numerical random-
walk simulation for dispersive transport in porous media also was used
by Todorovic (1975).  A cell grid is used and parcels of fluid are
tracked during the simulations.  The model can incorporate the chemical
reaction computations performed by Routson and Seme's (1972) PERCOL.

     The case of flow in a fractured medium with diffusion into adjacent
porous blocks, including linear equilibrium sorption and linear decay
mechanisms, has been investigated using finite elements by Grisak and
Pickens (1980).

-------
                                  51


     Three-dimensional flow systems with solute transport are difficult
to model numerically because of the large number of computational nodes
involved, even for systems of moderate size.  An approximate method of
treating this problem was given by Ross and Koplik (1979) who perform
the one-dimensional transport calculations with equilibrium sorption
and linear decay along steady-state streamtubes generated by an appro-
priate three-dimensional flow model.  A convolution integration with
the analytical Green's function solution is used to obtain the con-
centration field.

     The INTERA (Intercomp Resource Development and Engineering, Inc.,
1976, INTERA Environmental Consultants, Inc., 1979) three-dimensional
flow, solute- and energy-transport model is a fully coupled flow and
transport simulation code using finite-difference techniques.  Linear
equilibrium sorption and linear decay for single-species transport is
included.  This model was extended to handle radionuclide-decay chains
and named SWIFT (Dillon and others, 1978).  The radionuclide species
are assumed to be in dilute solution and thus do not affect the flow
field.

          COMPUTATIONAL PROBLEMS IN SOLUTE TRANSPORT MODELING

     With analytical-solution models, the only difficulties that may
arise are roundoff errors, overflows in certain function evaluations,
and insufficient accuracy in numerical quadrature.  The first two
problems can be resolved by combining terms in the proper order.  This
is particularly necessary in the radionuclide-chain calculations.
Accurate numerical quadrature can usually be achieved by using an
adaptive method and isolating any significant peaks of the integrand.

     With finite-difference and finite-element models there is the
possibility of excessive numerical dispersion (Oster and others, 1970;
Lantz, 1971) oscillation in the solution due to the space- or time-
discretization method, (Price and others, 1966), and overshoot/undershoot
near a steep concentration gradient (Gray and Pinder, 1976).  Oscillation
and overshoot/undershoot are probably related if not the same thing.  As
mentioned above, various correction terms have been used to minimize
numerical dispersion.  The method of characteristics was developed to
avoid this problem.

     Oscillation problems can be avoided by proper selection of time-
and space-step sizes (Price and others, 1966; Siemieniuch and Gladwell,
1978) but restrictions may make a given computation quite impractical.
Higher order time-integration methods have been proposed to avoid
oscillation problems, for example, Norsett methods (Smith and others,
1977).  Overshoot/undershoot occurs because a finite-difference or
finite-element representation cannot propagate all frequencies contained
in the solution-function representation at the same phase speed or
without attenuation (Gray and Pinder, 1976).  One of the advantages
of finite-element methods is that the use of higher-order basis functions
gives better propagation behavior over a wider range of frequencies.
Higher-order basis functions minimize numerical  instabilities but cost
more in computer-solution time (Grove, 1977).

-------
                                   52


     Discrete-parcel tracking or Lagrangian-simulation techniques are
oscillation free and nearly free from numerical dispersion and are mass
conservative at the cost of a slower computation and the presence of
some random noise errors from the dispersion calculation (Ahlstrom and
others, 1977).

     The main problems with three-dimensional models are the computation-
time and computer-storage requirements.  This is particularly true if
transport of multiple species are being simulated.  Fast and efficient
equation solvers for the matrix equations generated have been developed
and are being improved.  Out-of-core matrix storage is another technique
being used to minimize core-storage requirements.  At the present time
however, many of the larger three-dimensional problems can only be run
on large, very fast computers.

           STOCHASTIC CONCEPTS IN SOLUTE TRANSPORT MODELING

     It has long been realized that there is uncertainty in the des-
cription of the physical and chemical properties of the porous medium.
To collect sufficient data on a real system for a precise deterministic
model is an impossible task.  Thus, some effort has been made to attempt
to quantify the uncertainty in transport-model prediction and correlate
it with the uncertainty in the various model parameters.  Only a few
results have been achieved to date.  A group of studies involved the
modeling of large-scale dispersion by using macroscopically variable
hydraulic-conductivity fields (Warren and Skiba, 1964; Heller, 1972;
Schwartz, 1977; Smith and Schwartz, 1980).  The models used tracer-
particle tracking techniques with a one-dimensional mean flow field.
Results of these studies showed that a unique dispersivity cannot be
defined when the low-permeability inclusions are not arranged homo-
geneously.  Smith and Schwartz (1980) found that even with statistically
homogeneous conductivity fields, the tracer particles do not take on
a normal distribution and a constant dispersivity does not exist.  Also,
a large uncertainty is associated with the predicted solute elution
curve even when the statistical features of the porous medium are known.

     Gelhar and others (1979) looked at longitudinal dispersion due to
variations in hydraulic conductivity in a stratified porous medium.
They developed a stochastic differential equation describing the system
and used spectral techniques to solve it.  The transient nature of the
dispersion process and non-Fickian dispersion effects were identified.

     All of the above studies show that, unless sufficient transport
time has elapsed, dispersion due to large scale heterogeneities of
the porous medium does not obey the deterministic Fickian type
mechanism.

                 FIELD APPLICATION OF TRANSPORT MODELS

     Low-level radioactive waste-disposal sites are being thoroughly
studied by the U.S. Geological Survey as well as other Federal and State
agencies.  Although a qualitative  knowledge as to what is going on at

-------
                                   53


these sites is usually known,a quantitative understanding of contaminant
movement, both within and off site, generally is lacking.  Such movement
can be studied through the use of the ground-water solute-transport
models discussed earlier.  As mentioned, a wide variety of models are
available for use.  However, examples of field applications using
such models available in the general literature are somewhat limited.

     There are, however, several situations where modeling has been
applied to the movement of chemical species in ground water with good
results.  Although those discussed in the following paragraphs are not
exhaustive they are representative of "state-of-the-art" applications.
Discussions of each field site will be brief and interested readers are
referred to the references for more detail.

     As was mentioned previously, solution of the solute-transport
equation presents significant mathematical problems that must be over-
come by fairly sophisticated solution methods.  For this reason the
first studies of  field applications used a particle-tracking technique
that was entitled the method of characteristics (HOC).  The technique
was used by Bredehoeft and Finder  (1973) who studied the salt-water
contamination of  a limestone aquifer at Brunswick, Georgia.  The source
of the contamination was  underlying brackish water migrating upward
through natural conduits  in  response to lowered hydraulic  heads in the
overlying aquifers.  Simulation studies showed  interceptor wells to  be
the most feasible method  to  protect the well field.  This  reference  also
provides an extensive theoretical  background to the partial differential
equations that describe  solute  transport.

      Konikow  and  Bredehoeft  (1974)  investigated salinity increases in
ground water  and  surface  water  in  an alluvial  valley  in  southeastern
Colorado as related to irrigation  practices.   The  MOC  model was used.
Dissolved-solids  concentrations  calculated by  the  model  were within
10 percent of the measured values  for  both the aquifer and stream
approximately 80  percent  of  the  time.

      Konikow  (1976) used  this same model  to reproduce  the 30-year
history  of chloride contamination  at the  Rocky Mountain  Arsenal,
Colorado.  Here,  liquid  industrial  waste  had seeped out  of unlined
disposal ponds  and spread over  many square miles  in an alluvial
aquifer.

      Robson  (1974) used  the  MOC to model  the water quality of  a shallow
alluvial aquifer  near  Barstow,  California.  The aquifer had been
polluted by  percolation  of wastes  and  sewage from industrial  and
municipal  sources for  about  60  years.   Several strategies to alleviate
 the  pollution problem  and to protect downgradient domestic wells  were
 investigated by model  simulation.   Robson (1978)  also modeled  the same
 system in  a  vertical  section.   This model  has the advantage of being
 able  to simulate vertical  flow and water quality changes in a single-
 or multiple-aquifer  system.

-------
                                   54


     Robertson  (1974) used a MOC water-quality model to predict the
migration of industrial and radioactive liquid wastes throughout a
15-square-mile  area at the Idaho National Engineering Laboratories
(INEL) in Idaho.  The wastes were disposed of in the Snake River Plain
aquifer.  Chloride and tritium were found to have moved as far as
5 miles downgradient from the injection point.  Strontium-90 and
cesium-137 were significantly retarded due to sorption.  The strontium
was detected no more than 1.5 miles downgradient and cesium migration
was not detected.  All of the above species were adequately simulated
using the MOC model and estimates made of the long-term effects of
these contaminants on aquifer usage.  Robertson (1977) also modeled
solute transport from waste seepage ponds at INEL.  A three-segment
numerical model, incorporating analytical and multidimensional
numerical solutions, was used in this study.

     Grove (1977) used a two-dimensional finite-element solute-
transport code to simulate the same data from Robertson's (1974) study.
Virtually the same results were obtained using this numerical technique.

     Finder used the finite-element method to simulate several field
situations of water-quality change.  A chromate contamination problem
at Long Island, New York was well simulated by Pinder (1973).  He also
simulated the two-dimensional movement of a saltwater  front in the
Cutler area of the Biscayne aquifer near Miami, Florida (Pinder, 1975).
Due to the density effect caused by the saltwater concentrations, a set
of non-linear partial differential equations had to be solved simul-
taneously.

     Cheng (1975) used a finite-element method to study the movement of
variable-density fluids through a causeway dividing the Great Salt Lake
in Utah.  The model did not match concentration variables measured in
the field, but it did give some insight into the hydrology of the system.

     Ahlstrom and others (1977) documented the theoretical foundation
and numerical-solution procedure for the Hanford MMT-DPRW model.  A
simulation of the tritium plume beneath the Hanford, Washington site
was included to illustrate the use of the model  in a typical application.

                              CONCLUSION

     It is clear from the review of available ground-water solute-
transport models for saturated flow systems that many of the basic
mechanisms that apply to real site situations can be simulated with
certain simplifying assumptions.  Analytical models are adequate for
parameter-sensitivity analyses and preliminary investigations of
transport phenomena, particularly when there is a gross lack of field
data for model-parameter determination.  The simplifying assumptions
necessary for analytical solutions restrict their ability to realisti-
cally model  a real system.  Numerical models have the ability to treat
the more complicated systems such as heterogeneous porous media.
However, determining the parameter distributions for a real system from
a sparse data set can be fraught with uncertainties that limit the
quality of the model predictions.

-------
                                   55
     Areas where the modeling technology need further development
include flow and transport in fractured/porous media, transport of
multiple chemical species with complex chemical and sorptive inter-
actions, and transport modeling using stochastic principles in order
to handle the uncertainties in real systems.
                              REFERENCES

Ahlstrom, S. W., Foote, H.  P., Arnett, R.  C. ,  Cole, C.  R., and Serne,
     R. J., 1977, Multicomponent mass transport model:   Theory and
     numerical implementation (Discrete Parcel  Random Walk Version),
     Pacific Northwest Laboratories, Richland,  Washington, BNWL-2127,
     107 p.

Al-Miami, A. N. S., and Rushton, K. R., 1977,  Analysis  of  flow against
     dispersion in porous media:  Journal  of Hydrology,  v.  33, p.  87-97.

	1979,  Dispersion  in stratified porous media:   Analytical  solutions:
     Water  Resources Research, v. 15, no.  5, p. 1044-1048.
Anderson, M. P., 1979, Using models to simulate the movement of
     contaminants through groundwater flow systems:  Critical  Reviews
     in Environmental Control, v. 9, no. 2, p. 97-156.

Bachmat, Y., and Bear, J., 1964, The general equations of hydrodynamic
     dispersion in homogeneous, isotropic, porous mediums:  Journal
     of Geophysical Research, v. 69, no. 12, p. 2561-2567.

Back, W., and Cherry, J. A., 1976, Chemical aspects of present and
     future hydrogeologic problems ij^Saleem, S. A., ed., Advances in
     Groundwater Hydrology:  Minneapolis, American Water Research
     Association, p. 153-165.

Bear, J., 1972, Dynamics of fluids in porous media:  New York, American
     Elsevier, 764 p.

Bear, J.s and Jacobs, M., 1965, On the movement of water bodies
     injected into aquifers:  Journal of Hydrology, v. 3, p. 37-57.

Bredehoeft, J. D., 1969, Finite difference approximations to the
     equations of ground-water flow:  Water Resources Research, v. 5,
     no. 2, p. 531-534.

Bredehoeft, J. D., and Pinder, G. F., 1973, Mass transport in flowing
     groundwater:  Water Resources Research, v. 9, no. 1, p. 194-210.

Brenner, H., 1962, The diffusion model of longitudinal mixing in beds
     of finite length:   Chemical Engineering Science, v. 17, p.  229-243.

-------
                                  56


Cabrera, G., and Marino,  M.  A.,  1976,  A finite  element model of
     contaminant-movement in groimdwater:   Water Resources Bulletin,
     v.  12,  no.  2, p.  317-335.

Chaudhari, N.  M., 1971, An improved numerical technique for solving
     multidimensional  miscible displacement equations:  Society
     Petroleum Engineers  Journal,  v. 11, no. 3, p.  277-284.

Cheng, R.  T.,  1975, Study of fluid movements through  causeway:  Journal
     of the  Hydraulics Division, HY1,  p. 155-165.

Chhatwal,  S.  S., Cox,  R.  L., Green, D.  W.,  and  Ghandi, B., 1973,
     Experimental and  mathematical modeling of  liquid-liquid miscible
     displacement in porous  media:  Water  Resources Research, v. 9,
     no. 5,  p.  1369-1377.

DeMier,  W. V., Cloninger, M. 0., Burkholder, H. C., and Liddell, P. J.,
     1979, GET OUT - A computer  program for predicting radionuclide
     decay chain transport through geologic media:  Pacific Northwest
     Laboratory, Richland, Washington,  PNL-2970, 321  p.

DeSmedt, F., and Wierenga, P. J.,  1979, A  generalized solution for flow
     in soils  with mobile and immobile  water:   Water  Resources Research,
     v.  15,  no.  5, p.  1137-1141.

Dillon, R. T., Lantz,  R.  B., and Pahwa, S.  B.,  1978,  Risk methodology
     for geologic disposal of radioactive  waste:  The Sandia waste
     isolation flow and transport  (SWIFT)  model:  Sandia  Laboratories,
     Albuquerque, New  Mexico, SAND 78-2267.

Duguid, J. 0., and Reeves, M., 1976, Material  transport through porous
     media:   A finite  element Galerkin model:   Oak Ridge  National
     Laboratory, Oak Ridge, Tennessee, ORNL-4928, 201 p.

Eldor, M., and Dagan,  G., 1972,  Solutions  of hydrodynamic dispersion
     in porous media:  Water Resources Research, v. 8, no. 5, p. 1316-
     1331.

Enfield, C.  G., Harlin,  C. C., Jr., and Bledsoe, B. E., 1976, Comparison
     of five kinetic models for  orthophosphate reactions  in mineral
     soils:   Soil Science Society  America Journal, v. 40, p. 243-249.

Fried, J.  J., and Combarnous, M. A., 1971, Dispersion in  porous media  iji
     Advances in Hydrosciences:   New York, Academic Press, v. 7,
     p. 169-282.

Gelhar, L. W., and Collins, M.  A., 1971, General analysis of longitudinal
     dispersion in nonuniform flow:  Water Resources  Research, v.  7,
     no.  6, p. 1511-1521.

-------
                                   57


Gelhar, L. W., Gutjahr, A. L., and Naff, R. L., 1979, Stochastic
     analysis of macrodispersion in a stratified aquifer:  Water
     Resources Research, v. 15, no. 6, p. 1387-1397.

Gershon, N.  D., and Nir, A., 1969, Effects of boundary conditions of
     models  on tracer distribution in flow through porous mediums:
     Water Resources Research, v. 5, no. 4, p. 830-839.

Gray, W. G., and Finder, G. F., 1976, An analysis of the numerical
     solution of the transport equation:  Water Resources Research,
     v. 12,  no. 3, p. 547-555.

Grisak, G. E., and Pickens, J. E., 1980, Solute transport through
     fractured media 1. The effect of matrix diffusion:  Water Resources
     Research, v. 16, no. 4, p. 719-730.

Grove, D. B., 1976, Ion exchange reactions important in groundwater
     quality models:  Advances in Groundwater Hydrology, American
     Water Resources Association, p. 144-152.

	1977, The use of Galerkin finite element methods to solve mass
     transport equations:  U.S. Geological Survey Water Resources
     Investigations 77-49, 55 p.

Gupta, S. K., Tanji, K. K., and Luthin, J. N., 1975, A three-dimensional
     finite element groundwater model:  California Water Resource
     Center, University of California, Davis, Contribution no. 152,
     119 p.

Guymon, G. L. 4 1970, A finite element solution of the one-dimensional
     diffusion-convection equation:  Water Resources Research, v. 6,
     no. 1, p, 204-210.

Heller, J. P., 1972, Observations of mixing and diffusion in porous
     media in^ Proceedings of the second International Symposium on
     Fundamentals of Transport Phenomena in Porous Media, International
     Association of Hydraulic Research, Guelph, Canada, p. 1-26.

Hill, M. D.,  and Grimwood, P.  D., 1978, Preliminary assessment of the
     radiological  protection aspects of disposal of high-level waste
     in geologic formations:  National Radiological Protection Board,
     Chi 1 ton, Oxfordshire, United Kingdom, NRPB-R69, 53 p.

Hunt, B. W.,  1973, Dispersion  from pit in uniform seepage:  Proceedings
     American Society Chemical  Engineers, v.  99 HYI, p. 13-21.

     1974, Radial  dispersion in ground-water flow:  Proceedings
     American Society Chemical Engineers, v.  100, EM 6, p.  117-1127.

Hoopes, J. A., and Harleman, D. R. F., 1967,  Dispersion in  radial  flow
     from a recharge well:  Journal of Geophysical  Research, v.  72,
     no. 14, p. 3595-3607.

-------
                                  58


Huyakorn, P. S., 1977, Solution of steady-state convective transport
     equations using an upwind finite element scheme:  Applied
     Mathematical Modelling, v. 1, p. 187-195.

Huyakorn, P. S., and Taylor, C., 1977, Finite element models  for
     coupled groundwater flow and convective dispersion jr^Gray, W.  G.
     et al., eds.:  Finite Elements in Water Resources, Plymouth,
     Pentech Press, FE1, p.  1.131-1.151.

INTERA Environmental Consultants, Inc., 1979, Revision of the documenta-
     tion for a model for calculating effects of liquid waste disposal
     in deep saline aquifers:  U.S.  Geological  Survey Water Resources
     Investigations 79-96, 72 p.

Intercomp Resource Development and Engineering, Inc., 1976, A model
     for calculating effects of liquid waste disposal in deep saline
     aquifers, Part 1 - Development, Part 2 - Documentation:   U.S.
     Geological Survey Water Resources Investigations 76-61,  249 p.

Khaleel, R., and Reddell, D. L., 1977, Simulation of pollutant movement
     in groundwater aquifers:  Texas Water Resources Institute,  Texas
     A & M University, College Station, Texas,  Technical  Report, no. 81,
     248 p.

Kipp, K. L., 1979, Numerical implementation of  the three-member  nuclide
     decay chain transport equations:  AERE Harwell, Didcot,  Oxfordshire,
     United Kingdom, AERE-R-9849.

Konikow, L.  F., 1976, Modeling solute transport in ground water:
     International Conference on Environmental  Sensing and Assessment,
     Las Vegas, Nevada, article 20-3, 11  p.

Konikow, L.  F., and Bredehoeft, J.  D., 1974, Modeling flow and chemical
     quality changes in an irrigated stream-aquifer system:   Water
     Resources Research, v.  10, no.  3, p.  546-562.

     1978, Computer model of two-dimensional solute transport and
     dispersion in ground water:   Techniques  of Water Resources
     Investigations of the U.S.  Geological  Survey, Book 7,  Automated
     Data Processing and Computations,  v.  C2, 90 p.

Konikow, L.  F., and Grove, D.  B.,  1977, Derivation of equations
     describing solute transport in ground water:  U.S.  Geological
     Survey  Water Resources Investigations 77-19, 30 p.

Lai, S., and Jurinak, J.  J., 1972, Cation  adsorption in one-dimensional
     flow through soils:   A numerical  solution:  Water Resources
     Research,  v. 8, no.  1, p.  99-107.

Lantz, R. B., 1971, Quantitative evaluation of numerical  diffusion
     (Truncation error):   Society Petroleum Engineers Journal, v. 11,
     no. 3,  p.  315-320.

-------
                                  59


Lapidus, L., and Amundson, N. R., 1952, Mathematics of adsorption  in
     beds VI:  The effect of longitudinal diffusion in ion exchange and
     chromatographic columns:  Journal of Physical  Chemistry,  v. 56,
     p. 984-988.

Lee, C., and Cheng, R.  T., 1974, On seawater encroachment in coastal
     aquifers:  Water Resources Research, v. 10, no.  5, p. 1039-1044.

Lester, p.  H., Jansen,  G., and Burkholder, H. C., 1975, Migration  of
     radionuclide chains through an adsorbing medium:  American
     Institute Chemical Engineers Symposium Series, v. 71, no.  152,
     p. 202-213.

Lindstrom, F. T., and Boersma, L., 1971, A theory on the mass  transport
     of previously distributed chemicals in a water saturated  sorbing
     porous medium:  Soil Science, v.  Ill, no. 3, p.  192-199.

Marino, M.  A., 1974a, Distribution of contaminants  in porous media
     flow:  Water Resources Research,  v. 10, no. 5, p. 1013-1018.

     _1974b, Models of dispersion in a granular medium:  Journal  of
     Hydrology, v. 36, p. 261-277.

Nalluswami, M. , Logenbaugh, R. A., and Sunada, D.  K.,  1972, Finite
     element method for the hydrodynamic dispersion equation with
     mixed partial derivatives:  Water Resources Research,  v. 8,
     no. 5, p. 1247-1250.

Nelson, R. W. , 1978a, Evaluating the environmental consequences  of
     groundwater  contamination 1. An overview of contaminant arrival
     distributions as general evaluation requirements:  Water Resources
     Research, v. 14, no. 3, p. 409-415.

     _1978b, Evaluating  the environmental consequences of groundwater
     contamination 2. Obtaining location/arrival time and location/
     outflow quantity distributions for steady flow:  Water Resources
     Research, v. 14, no. 3, p. 416-428.

	1978c, Evaluating the environmental consequences of groundwater
     contamination 3. Obtaining contaminant arrival  distributions  for
     steady flow in heterogeneous systems:  Water Resources Research,
     v. 14, no. 3, p. 429-440.

	1978d, Evaluating the environmental consequences of groundwater
     contamination 4.  Obtaining and utilizing contaminant arrival
     distributions in 'transient flow systems:  Water Resources
     Research, v. 14, no. 3, p. 441-450.

Ogata, A., 1964, Mathematics of dispersion with a linear isotherm:
     U.S. Geological Survey Professional Paper 411-H, 9 p.

-------
                                   60


 Oster,  C. A., Sonnichsen, J. C., and Jaske, R. T. , 1970, Numerical
     solution to the convective diffusion equation:  Water Resources
     Research, v. 6, no. 6, p. 1746-1752.

 Parlange, J. Y., and Starr, J. L., 1978, Dispersion in soil columns:
     Effect of boundary conditions and irreversible reactions:  Soil
     Science Society America Journal, v. 42, no. 1, p. 15-18.

 Peaceman, D. W., and Rachford, H. H., 1955, The numerical  solution  of
     parabolic and elliptic differential equations:  Journal  Society
     Industrial Applied Mathematics, v. 3, no. 1, p.  28-41

	1962, Numerical Calculation of multidimensional  miscible  dis-
     placement:  Society Petroleum Engineering Journal, v.  237,  no.  4,
     p. 327-338.

 Phillips, K. J., and Gelhar, L. VI., 1978, Contaminant transport  to
     deep wells:   Proceedings American Society Chemical Engineers,
     v. 104, HY2, p. 1565-1575.

Pickens, J.  F., Gillham, R. W., and Cameron, D.  R.,  1979,  Finite
     element analysis of the transport of water and solutes in tile-
     drained soils:  Journal of Hydrology, v.  40, p.  243-264.

Pickens, J.  F., and Lennox, W. C., 1976, Numerical simulation  of waste
     movement in  steady groundwater flow systems:  Water Resources
     Research, v. 12, no.  2, p. 171-180.

Pinder, 6. F., 1973, A Galerkin-finite element simulation  of groundwater
     contamination on Long Island, New York:  Water Resources  Research,
     v. 9, no. 6, p. 1657-1670.

	1975, Numerical simulation of salt-water  intrusion in  coastal
     aquifers, Part 1:  Water Resources Program, Report WRP-75-2,
     50 p., Office of Water Research and Technology.

Pinder, G. F., and Cooper, H. H., Jr., 1970, A numerical  technique  for
     calculating the transient position of the saltwater front:  Water
     Resources Research, v. 6, no. 3, p. 875-882.

Pinder, G. F., and Shapiro, A., 1979, A new collocation method for  the
     solution of the convection dominated transport equation:  Water
     Resources Research, v. 15, no. 5, p. 1177-1182.

Prakash, A., 1976, Radial dispersion through adsorbing porous media:
     Proceedings American Society Chemical Engineers, v.  102, p. 379-396.

Price, H.  S., Cavendish, J. C., and Varga, R.  S.,  1968, Numerical
     methods of higher-order accuracy for diffusion-convection equations:
     Society Petroleum Engineers  Journal, v. 8, no.  3, p.  295-303.

-------
                                  61


Price, H. S., Varga, R.  S., and Warren, F.  E., 1966, Application  of
     oscillation matrices to diffusion-convection equations:   Journal
     of Mathematics and physics, v.  XLV, no.  3, p.  301-311.

Reddell, D. L., and Sunada, D. K., 1970, Numerical  simulation  of
     dispersion in groundwater aquifers:  Hydrology Paper no.  41,
     Colorado State University, Fort Collins, Colorado,  79 p.

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 1DO-22054, 41 p.

	1977, Numerical modeling of subsurface radioactive  solute transport
     from waste-seepage ponds at the Idaho National Engineering
     Laboratory:  U.S. Geological Survey Open-File Report 76-717, 68  p.

Robson, S. G., 1974, Feasibility of digital water-quality modeling
     illustrated by application at Barstow, California:   U.S.  Geological
     Survey Water Resources Investigations 46-73, 66 p.

     1978, Application of digital profile modeling techniques  to
     groundwater solute transport at Barstow, California:   U.S.
     Geological Survey Water-Supply Paper 2020, 28 p.

Rosinger, E. L. J., and Tremaine, K. K. R., 1978, GARD-A computer program
     for the geochemical assessment of radionuclide disposal:  Whiteshell
     Nuclear Research Establishment, Pinawa, Manitoba, Canada,
     AECL-6318, 49 p.

Ross, B., and Koplik, C. M., 1979, A new numerical method for solving
     the solute transport equation:  Water Resources Research, v. 15,
     no. 4, p. 949-955.

Routson, R. C., and Serne, R. J., 1972, One-dimensional model of the
     movement of trace radioactive solute through soil columns:   The
     Percol model:  Pacific Northwest Laboratories, Richland, Washington,
     BNWL-1718, 60 p.

Rubin, J., and James, R. V., 1973, Dispersion-affected transport of
     reacting solutes in saturated porous media:  Galerkin method
     applied to equilibrium-controlled exchange in unidirectional
     steady water flow:  Water Resources Research, v.  9, no. 5,
     p. 1332-1356.

Sauty, J. P., 1980, An analysis of hydrodispersive transfer in aquifers:
     Water Resources Research, v. 16, no. 1, p. 145-158.

Scheidegger, A. E., 1961, General theory of dispersion in porous media:
     Journal Geophysical Research, v. 66, no. 10, p. 3273-3278.

-------
                                  62
Schwartz, F. W., 1975, On radioactive waste management:   An  analysis
     of the parameters controlling subsurface contaminant transfer:
     Journal of Hydrology, v.  27, p.  51-71.

     J977, Macroscopic dispersion in  porous media:  The  controlling
     factors:  Water Resources Research, v.  13, no.  4,  p.  743-752.

Schwartz, F. W., and Domenico, P.  A., 1973,  Simulation  of  hydro-
     chemical patterns in regional  groundwater flow: Water Resources
     Research, v. 9, no.  3, p. 707-720.

Segol, G., Pinder, G. F., and Gray, W.  G., 1975, A Galerkin-finite
     element technique for calculating  the transient position  of  the
     saltwater front:  Water Resources  Research, v.  11, no.  2, p. 343-
     347.

Shaffer, M.  J., and Ribbens, R.  W., 1974,  Generalized description of
     return  flow quality  simulation model:  Engineering and Research
     Center, Bureau of Reclamation, Denver,  Colorado, 78 p.

Siemieniuch, J. L., and Gladwell,  I., 1978,  Analysis of explicity
     difference methods for a diffusion-convection equation:
     International Journal Numerical Methods in Engineering, v. 12,
     no. 6,  p. 899-916.

Smith, I. M., Farraday, R. B., and O'Connor, B. A.,  1973,  Rayleigh-
     Ritz and Galerkin finite elements  for diffusion-convection
     problems:  Water Resources  Research,  v. 9, no.  3,  p.  593-606.

Smith, I. M., Siemieniuch, J. L.,  and Gladwell, I.,  1977,  Evaluation
     of Norsett methods for integrating differential equations in
     time:  International Journal  for Numerical Methods in Engineering,
     v. 11,  p. 57-74.

Smith, L., and Schwartz,  F. W.,  1980, Mass Transport 1. A  stochastic
     analysis of macroscopic dispersion:  Water Resources  Research,
     v. 16,  no. 2, p. 303-313.

Stone, H. L., and Brian,  P. L. T., 1963, Numerical solution of
     convective transport problems:  American Institution  of Chemical
     Engineers Journal, v. 9, no.  5, p.  681-688.

Todorovic, P., 1975, A stochastic  model  of dispersion in a porous
     medium:  Water Resources Research, v. 11, no. 2, p. 348-354.

Van Genuchten, M. Th., and Gray, W. G., 1978, Analysis  of  some
     dispersion corrected numerical schemes  for solution of the
     transport equation:   International  Journal of Numerical Methods
     in Engineering, v. 12, no.  3, p. 387-404.

-------
                                  63
Warren, J. E., and Skiba, F. F., 1964, Macroscopic dispersion:
     Society Petroleum Engineers Journal, p.  215-230.
Washburn, J. F., Kaszeta, F. E., Simmons, C.  S.,
     1980, Multicomponent mass transport model:
     migration of radionuclides in groundwater:
     Laboratory, Rich!and, Washington, PNL-3179,
and Cole, C.  R.,
A model  for simulating
Pacific  Northwest
130 p.
Wilson, J. L., and Miller, P. J., 1978, Two-dimensional  plume in
     uniform ground-water flow:  Proceedings American Society Chemical
     Engineers, v. 104, HY4, p. 503-514.

-------
   AN  OPTIMUM MODEL TO  PREDICT  RADIONUCLIDE  TRANSPORT  IN AN AQUIFER
           FOR THE APPLICATION  TO  HEALTH  EFFECTS  EVALUATIONS
                             Cheng  Y.  Hung

                    Criteria and Standards  Division
                     Office of Radiation  Programs
                 U.S. Environmental  Protection  Agency
                        Washington,  D.C.  20460
                               ABSTRACT

           This paper presents an optimum groundwater  transport
      model for simulating radionuclide transport in an  aquifer
      using an approximate solution of the basic transport
      equation.  The model is designed to avoid (1) relatively
      high computer simulation costs normally experienced in
      numerical models and (2) the large errors which  are
      sometimes introduced when the physical  boundary  conditions
      are converted to a mathematical form suitable for  an
      analytical model.  The model neglects the effect of
      radionuclide transport through dispersion first  and then
      compensates for this effect later with a health  effects
      correction factor.  This correction factor is found to  be
      a function of the Peclete number and the "transport
      number," which has been defined, and can be determined  by
      using the parameters of the groundwater transport  system.
      The model has a low cost of simulation and yet maintain-
      ing reasonable accuracy for the peak and cumulative radio-
      nuclide discharges, which are of primary interest  for
      risk assessments.  Preliminary analyses indicated  that
      the model can be integrated into most of the risk  assess-
      ment models for health effects evaluations.
                             INTRODUCTION

     Comprehensive studies on the evaluation of the optimum method of
disposal of radioactive wastes have been conducted by various government
agencies [1,2] and by an Interagency Review Group [3].  All of these
studies have unanimously concluded that the geological disposal method
is one of the most viable alternatives.  Because transport through an
aquifer is a primary pathway of transporting radionuclides to the
biosphere, a groundwater model which simulates the migration of radio-
nuclides in an aquifer is one of the important transport submodels
employed for health effects evaluation.

     To date, there are more than one hundred and eight groundwater
transport models [4].  However, some of the models fail to consider the
process of radioactive decay and/or sorption and are, therefore, not

                                   65

-------
                                   66


suitable for application to health effects evaluations.  The rest of the
models which may be suitable for health effects evaluation can be sub-
divided into two groups:  the analytical model and the numerical model.
In general, analytical models are limited by specific mathematical bound-
ary conditions which require some approximation of actual physical condi-
tions [5, 6, 7].  As a result, these models suffer considerable error in
simulation due to these approximations.

     The application of a numerical model requires, in general, many more
tedious computations than the analytical approach does.  Besides, the
accuracy of simulation depends greatly on the adjustments of time and
space increments; severe errors may result if they are improperly adjusted.
On the other hand, a proper adjustment of these increments, in some
cases, may result in excessive computer time required for the simulation.
The cost may become prohibitive when the model is applied to long-term
simulations such as health effects evaluations.

     The purpose of this study is to present a groundwater transport
model which could overcome the shortcomings of existing models and to
characterize the model when it is applied in health effects evaluations.


      APPLICATION OF EXISTING MODEL IN HEALTH EFFECTS EVALUATIONS

     The basic equations for a groundwater transport system include the
momentum, the energy and the continuity equations for the hydrodynamic
system, and the continuity equation for the solute and the constituent
radionuclide in a decay chain.  Using tensor notations, the equations
take the form: [12]

     Y. = -(k/y)(vp - pgvz)                                             (1)

     V-[(pk/y)H(vp - pgvz)] + V-j^-VT - q,  - q'H
              = (3/3t)[npU + (1 -nMpCpTj]                            (2)

     7Py_ + q- = -(a/3t)(np)                                            (3)
     v-[pC(k/u)(vp - pgvz)] + nV-pD^-vC - q'C

              = (3/at)(pnRC)                                           (4)

in which C is the concentration of the radionuclide in the fluid phase;
y^ is the velocity vector; k is the permeability; v is the viscosity of
the fluid; p is the pressure; p is the mass density; g is the gravita-
tional acceleration; D is the dispersivity tensor; T is the temperature;
qL is the rate of heat loss; H is the fluid enthalpy; n is the porosity;
z is the height above reference plane; \A is the decay constant; R is
the retardation factor; U is the internal energy; Cp is the specific
heat; q' is the rate of fluid withdrawal; and subscripts h and c are
the heat energy and component of mass respectively.

-------
                                   67


     The above non-linear equations characterize the transport of radio-
nuclides in the groundwater system.  Since each of the above equations
are related through dependent variables, the direct solution of the
system equation for any boundary and initial conditions is extremely
difficult and not practical.  A commonly used practice in solving the
system equation is to assume that the flow of groundwater is steady and
that there is no heat energy being generated or absorbed in the system.
The system equation then reduces to a single equation.

     R(9C/3t) - v-(jl'VC) + VvC + AdRC = 0                              (5)

in which V is the interstitial velocity (V = v/n).  Equation (5) has
been further simplified and solved by numerous authors [7, 8, 9] employ-
ing numerical or analytical approaches.  The difficulties encountered in
applying these models for health effects assessments are described in the
following sections.


Numerical Models

     The existing multidimensional models are solved either by the finite
difference method [8] or by the finite-element method [9].  Although a
model employing the finite-element method is, in general, found to be
more efficient than one employing the finite difference method [10], the
time required to execute a computer model employing the finite element
method is still far beyond the limitations of a normal project budget for
risk assessment when a number of cases must be considered.

     The simulation of the vertical migration of tritium from a burial
trench toward the groundwater table conducted for the West Valley, New
York burial site [11] used finite element model developed by Duguid and
Reeves [9].  Approximately seven minutes of computer process time was
required on UNIVAC-110 computer for simulating 200 years of real time.
The analysis also indicated that the half strength concentration
(C/Co = 0.5) wave front moved 0.6 m in 20 years.  Assuming that 1000
years of real time is required to simulate the system and that the time
required for the simulation of the radionuclide migration in a ground-
water stream is the same as that for the vertical migration, then the
total time required for simulating the West Valley or similar groundwater
system would be 700 min for completing simulation of ten critical radio-
nuclides.  This is equivalent to $10,500 per run, a cost that is unques-
tionably excessive.

     A one-dimensional INTERA model [12] developed by INTERA Environ-
mental Consultants, Inc., employs a finite difference method.  It has a
computer processing cost of $200 for simulating the migration of a single
radionuclide in a one mile long aquifer for a 100,000 years real time
simulation.  This is also too costly when many other radionuclides and
pathways are to be considered for a complete health effects evaluation.

-------
                                   68


Analytical Models

     Lester, Jansen, and Buckholder developed an analytical groundwater
transport model for a one-dimensional, semi -infinite aquifer system
which has impulse release and decay step release boundary conditions [6].
These boundary conditions at x = 0 are expressed mathematically as

     C = C0fi(t)                                                        (6)

and

     C = (C0/td)exp(-Xdt)                                              (7)

for the impulse release and the decaying step release, respectively.  In
the above equation, x is the space coordinate, t is the time, td is the
duration of radionuclide leaching, and CQ is the concentration of radio-
nuclides at t = 0.

     Ford, Bacon, and Davis, Inc., developed an analytical model which
assumed that the rate of radionuclide release at any time is propor-
tional to the inventory of radionuclides remaining at the source point.
Mathematically, it is expressed as
     C = (\yQw)exPL"-Ad + XL)t]                                      (8)

in which AL is the leaching constant, Im is the initial inventory of
the radionuclides in the waste depository, and Qw is the rate of ground-
water flow.

     These  analytical models have made considerable contributions in
groundwater transport modeling by simplifying the procedures of simula-
tion.  However, the application of these models to health effects
assessments is limited because it requires the approximation of con-
verting the physical boundary condition to a form meeting the require-
ments of an analytical model.  This may result in considerable error
of simulation in some cases.

     The above discussion implies that existing numerical and analytical
groundwater models may either be too costly to compute or may introduce
large errors when applied to health effects assessments.  Therefore, a
more accurate and more economic groundwater transport model has been
developed for health risk assessment and is presented herein.


   THEORETICAL BACKGROUND OF THE OPTIMUM GROUNDWATER TRANSPORT MODEL
                    FOR HEALTH EFFECTS EVALUATIONS


Derivation of Basic Equation

     This transport model is intended to simulate a uniform, one-
dimensional flow in an aquifer.  It receives radionuclides released from

-------
                                  69
the waste disposal site and transports them to the biosphere for bio-
logical uptake.  This includes wells,  springs, streams,  and other use
and discharge points.  It is assumed that the dissolved  raionuclides
are in sorption equilibrium with the aquifer formation and  that decay
is in progress for both dissolved and sorbed radionuclides.

     The basic one-dimensional groundwater transport  equation  for the
model simulating radionuclide migration in an aquifer may be reduced
from Eq. (4) to

     D(32C/3x2) - V(9C/3x) - R(3C/3t) - AdRc = 0                       (9)

and should be solved by using the following initial  and  boundary
conditions:

     C = 0,              at all x,           when t = 0,              (10)

     C = Cn(t),          at all x,           when t > 0,
          °                                                           (ID
     C = finite,         at x = °°            when t > 0

For the convenience of health effects analysis, one may  convert the
radionuclide concentration, C, into the rate of radionuclide transport
by multiplying Eqs. (9), (10), and  (11) by the rate of groundwater flow.
When this transformation is completed, Eqs. (10) and (11)  become:

     D(32Q/3x2) - V(3Q/3x) - R(3Q/3t) - RXdQ = 0                      (12)

     Q = 0,              at all x,           when t = 0,

     Q = QQ(t),          at x = 0,           when t > 0, and          (13)

     Q finite, CyH       at x = °°,           when t > 0
      •
where Q denotes the rate of radionuclide transport.

     Since Eq. (13) is an undefined boundary condition,  the analytical
solution for Eq. (12) cannot be obtained.  However, its  solution can
be expressed in an integral form by employing the method of Green's
function, that is:

     Q(t) = \(t - T)ll(T)dT                                         (14)
            0
in which u denotes the radionuclide release rate at the  discharge end,
x = L, which responds to the unit release of a radionuclide at x = 0
and T is a dummy variable of time.

     The response of unit release in Eq. (14), U(T),  has been  thoroughly
studied by Burkholder et al. [6] with the following  results:
     U(T) = (V/2L)V(RP/"03)Exp{-Nd0 - (P0/4R)[(R/0) - I]2}           (15)

-------
                                  70


in which P is the Peclet number (= VL/D); 0 is the dimension! ess time
(= TV/L); Nd is the decay number (= xrfL/V).  By substituting Eq. (15)
into Eq. (14) one obtains.
     Q(t) = Q/oU-r) (V/2L)V(RP/*e3)

            Exp{-Nde-(Pe/4R)[(R/e) - l]

     Equation (16) can not be integrated because Q0(t-r) is undefined.
However, if the dispersion term, D(92Q/3x2), in Eq. (12) is neglected,
then this response of unit release, U(T) can be simplified and becomes [6]

     U'(T) = Exp(-RLXd/V)6(T - RL/V)                                  (17)

As a consequence, Eq. (14) can be integrated and the result is

     
-------
                                   71


If this is the case, there is a linear relationship between the cumulative
radionuclides released and the expected total health effects which is
governed by the health effects conversion factors [13].  Thus, Eq. (21)
can be rewritten as:

    n  =  [HECFo/°Q(t)dt]/[HECF /°0/(t)dt]                              (22)

where HECF represents the health effects conversion factor.  Subsequent
substitution of Eq.  (14) into Eq.  (22) yields

            00  t«                    oo  t

                                  U  U   U

Since no radionuclides are being released at x = 0, when t<0, the upper
limit of the integration in Eq. (23) may be converted from time t to
infinity without changing the results of the integration.  Thus, Eq. (23)
may be rewritten as:
            00  „).                    m  ^
     ^  =  '-n   ^ Qfp^~T'^*TjdTdt]/[ /  / Qn(t-t)u"(T)dTdt1             (?4^


Now, since the time  integral of the functions Qo(t--r) and U(T) converge
to finite numbers when t is infinite, Eq. (24) can be transformed to


    Tl  =  [  /"U(T)dT /\(t-T)dt]/[ /Y(T)dTn/"Qn(t-T)dt]
                        co                                             (25)
Equation (25) implies that the health effects correction factor n is
independent of environmental time and, hence, can be characterized
independent of time.  Once the health effects correction factor is deter-
mined, the rate of radionuclide release at the discharge end may be
approximated by:

     Q(t) =  nExp(-RLXd/V)Q0(t - RL/V)                                 (26)

Equation (26) is the basic equation of this groundwater transport model.


Characterization of Health Effects Correction Factor

     To determine the characteristics of the health effects correction
factor, one may substitute Eqs. (15) and (17) into Eq. (25) and then
complete the integration of the denominator.  This yields
         y°0/2)J(RP/pe3)Exp[-N.e  - (Pe/4R)(R/e - l)2]de
     n = 0	1	d	           (27)
                             Exp(-RLxd/V)

A series of computations were made for the numerator and the denominator
of Eq. (27), using various values for P, Nd, and R.   The error between
the results of analysis obtained for the denominator and for the numera-
tor were also computed and extended to evaluate the relative error.

-------
                                   72


     Careful examination of the results reveal that the relative error
 is a primary function of the Peclet number and the parameter expressed
 by RxdL/V.  This parameter represents the ratio of the radioactive decay
 constant, x^, and the "transport constant," V/RL, and is designated as
 "transport number" for this study.  The above results were plotted on a
 Peclet number vs. transport number plane as shown in Fig. 1.  This figure
 indicates that the relative error, e, increases with the increase in
 transport number and decreases with the increase in Peclet number.

     Based on the definition of relative error, the health effects
 correction factor can be computed by

     n = 1 + e                                                        (28)

 Since the relative error is always greater than 0, the health effects
 correction factor is always greater than 1.


 Proposed Groundwater Transport Model for Health Effects Assessment

     Based on the previous discussions a groundwater system with initial
 and boundary conditions as shown in Eq. (13) may be simulated by Eq. (26),
 which is duplicated as;

     Q(t) = nExp(-RLx /V)6 (t - RL/V)                                  (29)
                     d    0
 Equation (29) represents an algebraic equation, in which, constant n is
 determined from Fig. 1 and Eq. (28); and the terms, RL/V and xd are also
 known constants determined from the characteristics of the groundwater
 system.  Therefore, the proposed model described by Eq.  (26) is the
 simplest and most economical to process when it is integrated into a
 health effects assessment model.

     It should be noted that when this model is employed in a health
 effects assessment model, there may be some error in the health effects
 for each generation.  The size of this error depends on the characteris-
 tics of the system, the nature of the radionuclide being considered, and
 the boundary conditions.  The nature of this error is characterized in the
 following section.  Nevertheless, the cumulative health effects evaluated
 for each generation, based on this model, is expected to be a insignifi-
 cant error relative to the true solution obtained from the one dimensional
model expressed in Eqs.  (9), (10), and (11).  This fact is also discussed
 in the following section.


     CHARACTERIZATION OF THE POTENTIAL ERROR OF THE PROPOSED MODEL
Hypothetical Groundwater System

     In order to characterize the nature of error incurred by the pro-
posed model, a hypothetical groundwater transport system was established.

-------
                                   73
In the hypothetical groundwater system, it was assumed that:  the veloc-
ity of groundwater flow is 100 m/yr; the location of radionuclide dis-
charge is 500 m down stream from the source point where the radionuclide
is released; the coefficient of dispersion is 300 m2/yr; the rate of
radionuclide release at the source point, x = 0, varies with time and
is represented by

     Q0(t) = S1n[(ir/td)t]          for  0 < t  <  td

     Qo(t) = 0                     for      t  >  td                    (30)

where td is the duration of radionuclide release.

     Three cases are assumed for the analysis.  Case I represents a
fast release, t^j = 200 yr, and long radionuclide half-life, Xd =
0.000693.  Case II represents a fast release, td = 200 yr,  and short
radionuclide half-life xd = 0.0693.  Case III represents a  slower
release, td = 400 yr, and long radionuclide half-life, Xd = 0.000693.
Each case is also subdivided into three subcases with the retardation
factors equal to 1, 10, and 100, respectively.


Method and Results of Analysis

     Two models were developed for the analysis.  One of them was
developed based on a numerical integration of Eq. (16).and  was desig-
nated the "Exact Model."  The other model was developed based on
Eq. (29) and was designated as the "Optimum Model."  Each of these
models were designated to compute the radionuclide discharge rate at
the downstream end of the aquifer and the cumulative radionuclides
released at the discharging point over infinite time.  The  results of
computer analyses using these two models are presented in Figs. 2,
3, and 4.
Discussion of Results

     In case of long radionuclide half-life, Cases I and III, the rate
of radionuclide release and the cumulative radionuclide release com-
puted from the exact model and the optimum model agree very weTT for
all three subcases assumed.  However, the results of the Case II
analysis, where the radionuclide half-life is short, indicate, that
the rate of radionuclide discharge obtained from the optimum model
deviates more and more from that obtained from the exact model as
the retardation factor increases.  Nevertheless, the major deviations
are merely in the time-lag of radionuclide discharge, whereas the peak
discharge showed little deviation (Fig. 3).  The results of this latter
analysis also indicated that the above deviation will be mitigated  for
those radionuclides with longer half-life, i.e., with smaller transport
number (Fig. 2).  Fortunately, a system with larger transport numbers
results in a limited quantity of radionuclide discharge at its dis-
charging point as can be seen by comparing case II-3 versus cases II-2

-------
                                  74
or II-1.  This result implies that the discharge of radionuclides  under
a high transport number, such as Case II-3, would not be a major concern
from the view point of health effects assessment.  Besides, this
deviation should further decrease with the increase in the duration  of
radionuclide release at the radionuclide source point, x = 0,  as shown
in Figs. 2 and 4.

     Nevertheless, the cumulative radioactive discharge, which is  the
primary parameter in determining the total health effects from a radio-
active waste disposal site, remains the same for values obtained from
the exact model and from the optimum model.  This agreement implies
that the proposed optimum groundwater transport model can be integrated
into the health effects assessment model without introducing any signifi-
cant error in the assessment of health effects.
                              CONCLUSION

     An optimum groundwater transport model  for health effects assess-
ment has been derived and characterized.  The processing time required
to simulate this model is significantly less than for other comparable
models because it only requires algebraic calculations.

     The error from simulating the peak discharge of radionuclides
using the "optimum" model, as compared with the error from use of the
"exact" model [i.e., the numerical model represented by Eqs. (16) and
(30)], may be noticeable when the groundwater transport number is greater
than 5.  However, the cumulative radionuclides released under the same
conditions is normally only a small fraction of the cumulative radio-
nuclides released at the source point.  Therefore, the effect of this
error is mitigated.

     The error in assessment of health effects from the proposed optimum
model, as compared to that obtained from an exact model, has been
found to be is insignificant in practical applications.

-------
                                  75


                             REFERENCES


 1.  The Mitre Corporation, Alternative Disposal Concepts for High-level
     and Transuranic Radioactive Waste Disposal, report prepared for
     USEPA, ORP/CSD 79-1 (1979).	

 2.  Ford, Bacon, & Davis Utah, Inc., Screening of Alternative Methods
     for the Disposal of Low-Level Radioactive Wastes, report prepared
     for USNRC, NUREG/CR-0308, UC 209-02 (1978).

 3.  Interagency Review Group, Report to the President by the Interagenc.y
     Review Group on Nuclear Waste Management. TID-29442 (March 1979).

 4.  Science Applications, Inc., A Compilation of Models and Monitoring
     Techniques, report prepared for USDOE, ORNL/SUB-79/13617/2,
     SAI/OR-565-2.

 5.  C. B. Smith, D. J. Egan, W. A. Williams, J. M. Gruhlke, C. Y.  Hung,
     and B. Serini, Population Risks from Disposal of High-level
     Radioactive Wastes in Geological Repositories. USEPA report,
     520/3-80-006.	

 6.  D. H. Lester, George Jansen, and H. C. Burkholder, "Migration  of
     Radionuclide Chains through an Adsorbing Medium," AICHE Symp.
     Series 7j_: 202 (1975).

 7.  Ford, Bacon, & Davis Utah, Inc., A Classification System for
     Radioactive Waste Disposal —What Waste Goes Where, report prepared
     for USNRC, NUREG-0456, FBDU-224-10 (1978).

 8.  INTERA Environmental Consultants, Inc., Development of Radioactive
     Waste Migration Model, report prepared for Sandia Laboratories
     (August 1977).

 9.  J. 0. Duguid and M. Reeves, Material Transport Through Porous  Media:
     A Finite-Element Galerkin Model, Oak Ridge National Laboratory
     report, ORNL-4929 (1976).

10.  George F.  Finder and William G. Gray, Finite Element Simulation in
     Surface and Subsurface Hydrology, Academic Press (1977).

11.  New York State Geological Survey, Computer Modeling of Radionuclide
     Migration Pathway at a Low-Level Waste Burial Ground.   West Valley,
     New York,  report prepared for USEPA (1980).

12.  INTERA Environmental Consultants, Inc., Development of Radioactive
     Waste Migration Model, report prepared for Sandia Laboratories
     (August 1977).

13.  J. M. Smith, T. W. Fowler, and A. S. Goldin, Environmental  Pathway
     Models for Evaluating Population Risks for Disposal of High-Level
     Radioactive Waste in Geologic Repositories, USEPA report, 520/5-
     80-002 (1980).

-------
                                   76
           o.i
0.5   1.0
5.0  10.0
                                                         50.0  100.0
                        Transport Number, RA L/V
                                           d
     Fig. 1.   Results of the Health  Effects  Correction Factor Analysis

Using Eqs.  (27)  and  (28).

-------
                                   77
      CJ 
-------
                                       78
            10
                                               -1
                              Cumulative Release, Case II-l
                               Rate  of  Release,  Case  II-l
                              Cumulative Release, Case II-2
                                       Rate of Release at x=0
                                    ^— Result  of Analysis, Optimum Model
                                    ——Result  of Analysis, Exact Model
                                   <-Rate Of Release, Case II-2
                                              ^Cumulative Release,
                                                          Case II-3
                                                      te of Release,
                                                          Case II-3
700
                                                                   800
      Fl9M ?',  Con;Parison of the  Results  of Analyses Obtained from  the
Optimum Model  and the  Exact Model  Case  II, *d =0.0693,  td  = 200 yrs

-------
                                   79
           L:
                             umulatrve Release, Case IIl-l
                            Cumulative Release,
                            Case III-2
                    ino
                          200   300
500   600
700
                                  Time,  yr
800
     Fig. 4.   Comparison of the  Results of Analyses Obtained from the
Optimum Model  and the Exact Model,  Case II, xd = 0.000693,  td = 400 yrs,

-------
     MODELING OF WATER AND SOLUTE TRANSPORT UNDER VARIABLY
            SATURATED CONDITIONS:  STATE OF THE ART

                         E. G. Lappala

                U.S. Department of the Interior
                      Geological Survey
                     Denver Federal Center
                       Denver, Colorado
                           ABSTRACT

      This paper reviews  the  equations used in deterministic
models of mass and energy transport in variably saturated
porous media.  Analytic,  quasi-analytic, and numerical solution
methods to the nonlinear  forms of transport equations are
discussed with respect to their advantages and limitations.

      The factors that influence the selection of a modeling
method are discussed in this  paper; they include the following:
      1. The degree of coupling required among the equations
         describing the transport of liquids, gases, solutes,
         and energy;
      2. The inclusion of an  advection term in the equations;
      3. The existence of sharp fronts;
      4. The degree of nonlinearity and hysteresis in the
         transport coefficients and boundary conditions;
      5. The existence of complex boundaries; and
      6. The availability and reliability of data required
         by the models.

      Analytic solutions  are  available for a variety of steady-
state nonlinear multidimensional problems such as infiltration
of liquid water from point or line sources.  Analytic treat-
ment of heterogeneity and anisotropy is limited to spatial
variations in material properties (such as exponential depend-
ence of hydraulic conductivity with depth) that are not
generally realistic.  Some quasi-analytic methods are numeri-
cally unstable, leading to oscillatory solutions, and in a few
cases, convergence to the wrong solution.  The principal ad-
vantage of existing analytic  and quasi-analytic methods is in
deriving general conclusions  about the nature of solutions
that may be used to check the validity of numeric methods based
on finite differences or  finite elements.

      Numeric methods are reviewed as they have been applied
to variably saturated transport problems.  The preferred
method with the most flexibility for extremely nonlinear water-
flow problems is a finite difference scheme applied to the
balance equation,- written in  terms of pressure or total head,
with upstream weighting on the relative hydraulic conductivity.

                              81

-------
                                   82
      In addition,  nonlinear coefficients should be evaulated
      implicitly and  improved by iteration.   Galerkin finite element
      schemes may result  in oscillatory and  non-convergent  solutions
      to these problems.   Finite element methods based on orthogonal
      collocation somewhat reduce these problems.   However,  the ad-
      vantage of increased accuracy of finite element schemes over
      finite difference schemes is reduced by the large computer
      times  required  to formulate and solve  the resulting algebraic
      equations.   Finite  element schemes appear to be preferred
      for linear and  mildly nonlinear forms  of the transport equa-
      tions  when sharp  fronts are involved.   Verified,  documented
      computer codes  that use both finite element and finite differ-
      ence schemes  are  available for a variety of realistic  prob-
      lems that arise in  designing and managing waste disposal sites.
      However,  use  of these models is not routine because of numeric
      problems of solving the descriptive equations.   Consequently,
      most problems should be analyzed with  more than one approach
      or model to assure  consistency.
                              INTRODUCTION

     Quantitative description of movement of water and  solutes  through
variably saturated porous media is required to  enable adequate  design and
management of waste-disposal  sites.  Water originating  as precipitation
or irrigation water percolates through overburden and the waste itself.
This water may dissolve pollutants which may reach underlying aquifers
or streams and lakes by downward and lateral movement.  In addition,
gaseous waste forms may escape to the biosphere via upward diffusion.
Thus, the timing and location of pollutant loadings to  potable  water
supplies and the biosphere require quantitative description of  the
movement of recharge waters,  dissolved solutes, and gaseous pollutants.

     Since this movement occurs through porous  and fractured media that
may or may not be totally saturated with water, a three-phase system
comprising moist air, water,  and solids must be analyzed.  The  equations
describing water and solute movement in these systems are complicated;
their solution usually requires some type of numerical  approximation.
These approximations, the required transport coefficients, and  initial
and boundary conditions comprise a model of a particular field  problem.
Field problems generally include heterogeneity  in the transport coef-
ficients and boundary conditions that vary in both time and space.

     Until recently, solutions to these equations for realistic field
problems were impossible; however, with the development of modern computers,
many realistic problems can be analyzed.  It is the purpose of  this paper
to:  (1) review the equations involved in variably saturated mass and
energy transport and the principal modeling methods that have been used for
the solution of these equations; (2) present problems encountered in the
principal solution methods; and (3) present needs for future work in the
application of models incorporating these methods.  About 330 reports on
solutions to the equations for transport of mass and energy in  variably

-------
                                    83
saturated systems were reviewed.  Primary emphasis of the review was on
problems involving flow of liquid water and dissolved solutes.  However,
some discussion of problems involving combined movement of heat and
moisture in both liquid and vapor phases, and gaseous diffusion is in-
cluded.  Some of the references  through 1971 include comprehensive reviews
by other authors (Braester and others, 1971).  Problems of transport in
systems that are saturated at all times will be covered in another paper
at this symposium (Grove and Kipp, 1980).
                       BASIC TRANSPORT EQUATIONS

     Physics-based models of movement of water, solutes, and heat through
variably saturated porous or fractured media are derived by combining
mathematical statements of the conservation of mass and energy with
constitutive equations relating these statements to measurable variables,
such as pressure, temperature, and concentration.  These equations are
generally written to apply over some finite space in which the transport
properties are constant or over which they can be assumed to vary in
some simple manner.  The continuity of mass equation is:
                          /V
where
     S  => a general mass storage coefficient,

     p  = the mass density,

     t  = time,

     V  = volume of the region considered,

     S  = surface of the region considered,

     q  = the mass flux density orthogonal  to  the  surface S.

A similar equation for the balance of energy is:
where

     Su = a general enthalpy storage coefficient,
      H
     h  « energy content or enthalpy,

     q.  => enthalpy flux.

-------
                                   84
     Constitutive equations that relate mass and energy flux to measurable
variables state that the flux is proportional to the first power of a
driving force which is expressed as a gradient in the measurable variable.
For the flux of liquid water (q ), a generalized Darcy's law is the
appropriate equation:



where
     p.  =* mass density of liquid water,

     K   =* liquid hydraulic conductivity coefficient,

     K.  « conductivity for thermally induced liquid flux,

     K   = conductivity for osmotically induced liquid flux,

     $ > Q > $i.» aiu* $  are pressure, gravitational, thermal, and osmotic
      p  Tg   h       o

             potentials,

     7   = gradient operator,

     i   = subscript to indicate liquid water.

     In most applications, the only potentials of significance are  the
first three.  Osmotic potential may be important where materials are
present that act as semipenneable membranes.

     A similar equation for the flux of moist air (qfl), neglecting  all
driving forces except gradients in pressure and thermal potentials, is:

                                                     a'


where all terms are as described above with the subscript a  for air
replacing £. for water.

     The flux of gases, including water vapor, and  solutes dissolved  in
liquid water  CO in response  to a concentration  gradient is described  by
Pick's law:
                                a  = - L Vpc                          (5)

where
     L  is a  general coefficient  that accounts  for  the effects of diffusion
          and hydrodynamic dispersion  (Bear,  1972), and

     p  is the mass concentration of  the  gas  or  solute.

-------
                                   85
     This formulation of Pick' s law using mass concentration is valid for
diffusion of a constituent of a low enough concentration that the mass of
the solvent is not appreciably affected by the presence of the solute.
For most cases relative to waste disposal, this is true; one exception is
in the treatment of coupled heat and water-vapor flow under high tempera-
tures.  In this case, when water vapor moves due to a concentration
gradient, the concentration should be expressed as a mass fraction (mass
of water vapor as a ratio to mass of moist air).

     When a moving fluid is involved, an  advection term is added to the
righthand side of equation 5:

                              ^ = - L Vpc + qfpc                     (6)

where

     q- = volumetric flux rate of the solvent  fluid.

     The flux of enthalpy (q^) due to heat conduction  and  advection of
enthalpy in a moving fluid is described by adding  an advection  term to
Fourier's law:
                          qh = - X VT + CpPf  T qf                     (7)

where

     X  is a general thermal conductivity,

     T  is the temperature,

     C  is the mass specific heat, and
      P
     p  is fluid density.

     If phase changes are involved, the  latent heat  of vaporization or
the heat of fusion must be accounted  for  by  incorporating  them  in the
general thermal conductivity, and in  the  advection term (in the case
of latent heat) .

     Combining equations 3-7 with equations  1 and  2  and adding  source/sink
terms results in the following balance equations  for liquid water, moist
air, water vapor, or gaseous  pollutants,  solutes,  and  enthalpy,  that  are
applicable to most field problems in waste disposal.

                                Liquid water
  'v  S*dV + 'sP*Kp{V(f>P  + VM*+ V*h   dS + vQ,elV=0   (8)

-------
                                    86
                               Moist air
                  dp
/v  Sa -d  dV + I  P  K   V*  + V*    dS + /  ^  dV =
            v   a           B  a  a    P     ha       v

                   Water vapor or gaseous pollutants
                   dp
            /   S  — j£ dV+f  t L  Vp  -q  p  IdS+f  Q  dV = 0   (10)
            'v   g  dt      Js t  g   g    a Kg f     ;v xg

                   Solutes dissolved in liquid water
'v
s
s
dps
dt
dV
*/.
\
L
s
7ps ~ \
f.\
dS
+ 'v
Q,
                                                          dV = 0    (11)

                            Flux of energy
             */v  SHl  IV + /g {X VT - CpPf Tqf\dS + /vQH = 0    (12)

     The source/sink term in equations 8-12 is generally a function of
space, time, and perhaps temperature, pressure, or concentration.
Equations 8-12 along with appropriate boundary and initial conditions
and measured or derived values of the transport coefficients comprise
the basis of models applicable to field problems.  Boundary conditions
are required to specify the solution variable (s) or the flux of mass
or energy across the boundary of the modeled region as a function of
time and space.  Examples are specified concentration at a contaminant
source, rainfall rate, and heat flux from a canister containing spent
nuclear fuel.

     Equations 8-12 are partial differential equations that may be non-
linear owing to the dependence of the transport coefficients, boundary
conditions, and (or) source/sink terms on the solution variable.  It is
principally the nonlinear nature of these equations when they are
applied to variably saturated flow problems that requires extension of
the solution methods and applicable models from those used for linear
problems.  The nonlinearity in the transport coefficients is exemplified
by the general hydraulic conductivity in equation 8, which is a function
of properties of fluid, medium, and pressure potential:


                               V ' k Kr
where_                                               2
     K  = the intrinsic permeability of the medium, L  ,

     y  = dynamic viscosity of water, "ML  T  ,
      Xr
     K  = a relative conductivity which depends upon the capillary
      r     pressure potential 0|») dimensionless,
     4)  = the difference between air pressure  (P ) and liquid pressure
            (P  ) energy /unit mass.

Figure 1 shows typical functions of K for a sand and clay.

-------
                                   87
     The general  storage  coefficient  in  the water  flux equation  is also
highly nonlinear  and depends  upon  the capillary pressure potential  Cfig. 2)
The transport coefficients  in the  energy flux  equation are also  nonlinear
because of  their  temperature  dependence.  Nonlinearity in the  transport
coefficients is further exacerbated when coupled equations are used.
For example, in a coupled water-flux  solute-transport problem, hydraulic
conductivity of the medium  may be  affected by  the  solute concentration;
as, for example,  when waters  high  in  sodium infiltrate soils containing
clay minerals CFrankel and  others, 1978;  Pupiski and Shainberg,  1979).
Another example is the strong temperature dependence of dynamic  viscosity
in the water flux equation  for coupled heat and moisture flow  problems.

     In addition  to nonlinearity in transport  coefficients, certain
boundary conditions may cause the  equations being  solved to be nonlinear.
An example would  be isothermal evaporation, where  the evaporative flux
depends upon the  unknown  pressure  head in the  soil immediately below the
land surface.  Source/sink  terms may  be  nonlinear  also; examples would
be the rate of water uptake by a plant root system, which depends on the
unknown pressure  potential  in the  soil surrounding the root, and sorption
of a contaminant  species  which depends upon the solute concentration.
Table 1 shows the degree  of nonlinearity generally associated  with
equations 8-10 as they might  be applied  to variably saturated  conditions
at a waste-disposal site.
                             MODELING APPROACHES

     Depending upon the problem being analyzed, any subset of equations
8-12 may be required to describe the physical system.

     The solution of the equations that are part of a model of a given
field problem may be accomplished in one of three ways:   (1) By use of
analytic or quasi-analytic methods; (2) by use of an analog model such as
resistance-capacitance networks or deformable membranes;  (3) by the use of
approximate numerical techniques.  Because of the nonlinearities in
equations 8-14, analog models have found very limited application.  Since
only one reference to the use of electric analog model to analyze variably
saturated flow problems was found (Bouwer and Little, 1959), further
discussion of analog models will be omitted.

     Although analytic and numeric solution methods will be discussed
separately, this subdivision is arbitrary.  In general, analytic methods
utilize direct integration or series solutions to equations 8-12, and
treat time and space as continuous; numeric methods involve discretization
of the solution domain and the time span of interest into finite intervals.
A section is also included on parametric or input-output models that may
incorporate some features of both analytic and numeric methods.

-------
o
g
u
i
o
>-
1
h-
i
o
8
      -2
       01234
            LOG OF PRESSURE HEAD. IN CENTIMETERS OF WATER

       Figure 1.  — Relative hydraulic  conductivity to
               liquid water for a sand and clay.
M
i
o
       0123
            LOG OF PRESSURE HEAD, IN CENTIMETERS OF WATER
       Figure 2.  — Nonlinear storage coefficient for
                        a sand and clay.

-------
Table 1.—Degree of nonlinearity found in moat formulation® of equations  8-12


   [V » very nonlinear, S » slightly nonlinear, L » can be assumed  linear]




(8)
(9)
(10)

(11)
(12)

Equation


Liquid water 	
Moist air 	
Water vapor and


Solutes in liquid water —
Enthalpy 	
Transport coefficients


Storage Dispersion, diffusion,
or conductance
V V
s v

V V
L L
S S

Source and Boundary
sink terms conditions

s-v s-v
s s

s-v s-v
s s
v s-v
                                                                                                00
                                                                                                vo

-------
                                   90
                 FACTORS THAT DETERMINE A MODELING APPROACH

     There are several factors that determine  the modeling method used to
analyze a given field problem:

     1.  The degree of nonlinearity and hysteresis in transport co-
           efficients, boundary conditions, and source/sink terms (see
           table 11;
     2.  Inclusion of an advection or gravity  term;
     3.  Existence of sharp fronts;
     4.  Heterogeneity and anisotropy;
     5.  Existence of complex boundaries;
     6.  Data availability;
     7.  Time and funding available;
     8.  Necessity to couple two or more of equations 8-12.

     Factor 8, coupling of equations, will be  discussed first, with
       factors 1 through 7 discussed in following sections.

     Equations 8-12 are all coupled to some degree.  The following two
examples illustrate coupling.  The liquid and  solute flux equations are
coupled, because the liquid flux is required to evaluate the advection
term in the solute flux equation.  Hydraulic conductivity in the liquid
flux equation may also depend on the solute concentration (as discussed
previously).  The thermal potential appearing  in the liquid flux equation
(equation 8) depends upon the temperature distribution derived from the
energy flux equation (equation 12).

     The degree of coupling required in solutions to the pertinent
equations depends principally upon the difference in time scales in the
phenomena being modeled.  For example, air pressure response is generally
very rapid compared to liquid response, so that implicit consideration
of simultaneous air and water flow due to pressure gradients is not
required unless air pockets are entrapped.  Large differences in time
scales may be used to advantage in solving field problems by specifying
the values of a rapid response variable as boundary conditions for a
variable that responds less rapidly.  Sometimes the sensitivity to time
variations in a secondary variable in an equation is low with respect to
the response of the principle variable.  For example, it has been shown
by Wierenga (1977) and Beese and Wierenga (1980) that, under certain
conditions, the pollutant concentration at depth in a soil profile
can be simulated as well using a time-averaged water-flux rate (q_) (in
equation 11), as with a fully transient treatment yielding the time
variations in q- that are caused by cyclic variations in rainfall and
evaporation.

     When it is required to couple two or more of equations 8-12 for a
given application, care must be taken in proper boundary-condition
specifications with respect to all coupled equations.

-------
                                    91
     Coupled equations are complex enough to generally preclude any ana-
lytical or quasi-analytical approach.  The numeric methods couple the
equations by either solving them sequentially or simultaneously.  Coupling
of equations using a sequential solution is usually accomplished through
the use of source and sink terms.  For example, in a problem of combined
heat and liquid flow (equations 8 and 12) , the liquid movement due to
thermal gradients is incorporated as a source or sink in equation 8, and
the movement of heat by moving fluid may be incorporated as a sink in the
enthalpy flux equation (equation 12).  Sequential solutions are exem-
plified by water and solute flow problems where the water flow equation
(equation 8) is solved first, and the resultant fluid velocities are used
in the solute flow equation (equation 11).  If the degree of coupling is
slight, this two-step sequential procedure is generally adequate.  For
strongly coupled problems such as coupled heat and moisture flow problems,
however, the solutions from the two-step process must be improved by
iteration.  This iteration can be accomplished either separately or
combined with the iteration required to linearize nonlinear forms of the
equations, and the iteration  that may be used to solve the linearized
algebraic equations.  Table 2 summarizes the degree of coupling of equa-
tions 8-12 for most problems  relative to waste-disposal studies.
    Table 2.—Degree of coupling  of equations  8-12  exhibited in problems

                   relating  to radioactive  waste disposal

      [S = strongly coupled,  M = moderately  coupled, W =  weakly coupled]
                                              Equation number

                                               10         11
                         12
           Equation
                                 Moist

                                  air
Water vapor   Solute

and gaseous  dissolved

 pollutants     in
              liquid
Enthalpy
(8) Liquid water 	 M

(9) Moist air 	 — ~
(10) Water vapor and
gaseous pollutants — 	
(11) Solute dissolved in
Tln,,4A 	 	
W-M
W-M
*+

—

— —
S
W


M

••— —
M-S
M


S

W-M

-------
                                   92
                      Analytic and Quasi-analytic Methods

     When the problem to be solved can be cast in simple form, a closed
form analytic or quasi-analytic method may be available.  These can be
classified as direct, similarity, quasilinear, and integral methods.

     These categories are attributed to a review by Philip (1974).  The
discussion that follows is written with emphasis on solutions to the non-
linear form of equation 8, the water-flow problem.  However, the methods
apply equally well to nonlinear forms of equations 9-12.  Analytic solu-
tion methods for linear forms of equations 8-12 will not be explicitly
addressed in this paper; solutions to linear problems are, in general,
special cases of the nonlinear methods presented.  A review of analytic
solutions to the solute flux equation (equation 11) will be given by
another paper at this symposium (Grove and Kipp, 1980).

                               Direct Methods

     Direct methods involve direct integration of some form of the
descriptive equation, solving for the constants of integration using
boundary and initial conditions.  A simple example would be the problem
of isothermal isohaline steady upward flux from a shallow water table at
a depth I.  The governing equation is Darcy's law (equation 3), because
the time derivative appearing in equation 8 is zero.  If the desired flux
rate is q, then the problem can be written as:
/
                                      q + K  0(0

                               *.
                                                d*.                  (14)
which for certain analytic forms of K  (iji) can be  integrated directly.
Gardner (1958, 1959) and Ripple and others  (1970) detail this solution
method for both uniform and layered media.  The integral in equation 14
may not be integrable in closed form because of the nature of the K  (i|0
function.  In this case, numerical integration methods such as Gaussian
quadrature must be used, but this solution  type can still be termed a
direct method.  An application of direct  solution methods that is
important in transient infiltration problems is the assumption of a step
function shape for the wetting front.  This has been  shown to be equiva-
lent to the assumption to a step function for hydraulic diffusivity,
D - K/S (Philip, 1973).  The resultant formulation, which is attributed
to Green and Ampt (1911), has been shown  to accurately represent the
infiltration rate under conditions of  ponded water on the surface.  Mein
and Larson (1973) have extended the method  to a flux  boundary condition
imposed by constant rainfall rates.

-------
                                    93


     Direct integration methods  require  the descriptive equation to be a
function of only one independent variable; thus, transient problems are
not immediately amenable  to  this solution method.  However, the use of a
similarity variable that  combines  two independent variables may render
the problem tractable by  direct  methods.  The most common of these trans-
formations is the so-called  Boltzman  transformation  (Ames, 1965):

                                   * - xt*2                           (15)
where
     $ is the transformed variable,
     x is a spatial coordinate,  and
     t is time.

     Even though a similarity transformation reduces the number of inde-
pendent variables by one, direct solution still requires the transport
coefficients to be constants or  to have  simple linear or step function
dependence on the transformed independent variable.

                                 Linearization

     When the transport coefficients  cannot be expressed as simple
functions the descriptive equations must be linearized.  The most common
linearization method used in the reviewed reports consists of using an
exponential function to describe hydraulic conductivity for steady-flow
problems, and hydraulic diffusivity* (D)  for transient problems.

     The exponential dependence  of hydraulic conductivity on pressure
potential for example can be described by
                              K = K  exp  (ct  ^)                       (16)
                                   s
where
     K  = the hydraulic conductivity at  saturation, and
      S
     a  • a constant.

The second step in the linearization process  is  to introduce a new variable
termed the matric flux potential, by using  the Kirchoff  transformation
(Ames, 1965):
J
                                      .

                                       K (a)  da                      (17)
These steps result in a linear  form of  equation 8 written with v as the
dependent variable, and render  essentially all  steady-flow problems
solvable using linear methods  (Philip,  19741 .   It should be noted that
for a wide range of soils, the  £(.*)  function  is very well represented
by equation 16 (Laliberte and  others, 1966; Mualem, 1976).

-------
                        Approximate Analytic Methods

     When a linearized form of the descriptive equations cannot be solved
by direct integration, approximate methods must be used.  These can be
categorized as series expansion methods and integral methods.  Series
expansion methods involve expressing the solution as a series with unknown
coefficients that usually involve integrals of the transport coefficients.
The form of the series is often based on empirical observation.  Philip
(1955) was the first to apply this method of solution to nonlinear
equations described by equation 8.  Philip's solution is of the form:
                               i<          "Ml
                         X = <|>t^ + xt + cat '  + . . .              (18)
                                                                         %
where , x> w» are t^ie unknown coefficients.  The expansion in terms of t
is baaed upon the observed square root of time dependence of the advance
of a wetting front in the absence of gravity.  The coefficients in equation
18 can be determined by an iterative trial and error process proposed by
Philip for arbitrary forms of the transport coefficients.  When this
process Is followed, the method is essentially a numerical rather than
an analytic method.  Brutsaert (1968a, b, 1977) has derived exact
expressions for the first two coefficients in the above equation for
one-dimensional absorption (gravitational potential ignored), and infil-
tration (gravitational potential included), by using a flux potential
formulation and assuming a simple form for the nonlinear storage
coefficient.  The principal drawback to Philip's method is its divergence
from the true solution at long times.  The maximum applicable time may be
a few days or less (Brutsaert, 1977).

     Series solutions may also be derived by expressing the series in
terms of a perturbation variable, or some variable that is assumed to vary
a small amount over the range of the solution (Ames, 1965).  Babu
(1976a, b, c, and 1977), has used this solution method, with the terms
in the series involving integrals of the hydraulic diffusivity function.

     The other general category of approximate analytic methods that can
be applied to transient as well as steady-state problems involves an
iterative evaluation of an integral form of equation 8.  The method is
useful in reducing the number of independent variables in problems that
cannot meet the similarity requirements (Philip, 1974).  The method in-
volves making a first approximation to the form of the solution to
equation 8.  This may be accomplished by ignoring the time-dependent
first term and solving the resultant steady-state problem by use of the
previously discussed methods.  This first approximation is inserted in
the integral of the transient equation with respect to the variable to be
eliminated.  The resultant equation is solved by a direct or series
solution method.

     These steps may be repeated to iteratively improve the solution.
Ames  (1965) gives a further explanation of  these iterative methods.

-------
                                    95
     Parlange, in a series of papers  (1971a, t>, c; 1972a, b, c; 1973a)
has applied the iterative method to a variety of problems ranging from
one-dimensional transient absorption  to axisymetric transient infiltration
problems.  Cisler (1974) derived a correction to Parlange's method which
improved the accuracy significantly.  These iterative methods are very
sensitive to the initial estimate of  the solution.  Knight and Philip
(1973) have shown that these methods  can fail to converge if the initial
estimate is very far from the solution.  Parlange and Braddock (1978)
have shown that iterative methods may converge  to the wrong solution
after a finite number of iterations.

                     Availability of Analytic Solutions

     The length restrictions for a symposium paper preclude listing all
the analytic solutions found in this  review.  An excellent compilation of
the available solutions through 1971  is given by Braester and others  (1971).
Solutions since 1971 are summarized in Babu  (1976a, b, c, and 1977); Philip
(1974); Parlange and Babu (1976a, b); Raats  (1971, 1972, and 1977);
Lomen and Warrick (1976a, b, and 1978a, b); Warrick  (1974); and Warrick
and Lomen (1977).  A summary of the types  of problems that have been
solved for the nonlinear water flow equation is given in table 3.  One-
dimensional problems have been solved for  a variety  of realistic boundary
conditions.  The principal restrictions are:   (1) The requirement to use
simple forms for the functional dependence of  the transport coefficients
upon the solution variable;  (2) the limiting of heterogeneous media in
transient cases to simple functional  dependence upon depth; and  (3) the
failure to include source and sink terms of a  general nature.

     For multidimensional cases, the  geometry  is restricted to two-
dimensional planar and three-dimensional cylindrical or  spherical symmetry.
Few transient multidimensional problems have been analyzed, and no
solutions to multidimensional transient problems that involve sources and
sinks were found.

     While it may appear that the  application  of analytic  solutions  to
field problems that relate  to waste disposal  is somewhat  limited, there
are three main advantages to  these methods.  First,  they give closed-form
solutions that allow general  conclusions  to  be drawn regarding the move-
ment of mass and  (or) energy  in variably  saturated  systems.  Second,  these
conclusions can be used in  a  qualitative  sense to  check  the results  of
approximate numeric methods.  Third,  analytic  solutions  that exist  for
simple problems can be used quantitatively to  check approximate  numeric
methods applied to the same simple problems.   Examples  of these  general-
ities are the square root of  time  dependence of horizontal absorption
and the initial stages of vertical infiltration,  and acceptable  forms
of moisture and potential profiles for simple systems  (Childs,  1969).

-------
                                   96





     Table 3.—Types of variably saturated flow problems for which.



        analytic and quasi-analytic solutions are available







                                    Number of reported solutions
           Factor
One-dimensional
                             Steady    Transient
  Two and three-




   dimensional






Steady    Transient
No gravity term:
Homogeneous :
Source/ sink 	
No source/ sink 	
Heterogeneous :
Source/sink 	
No source/ sink 	


1 0
1 24

0 0
0 0


0 0
0 5

0 0
0 0
Gravity term:



     Homogeneous:



         Source/sink—	     2



         No source/sink——     2



     Heterogeneous:



         Source/sink	     0



         No source/sink	     4
           1




          21








           0




           2
   3




  11








   0




   2
1




3








0




0

-------
                                   97
                                Numeric Methods

     Numeric methods require  subdividing  the spatial and  temporal domains
into discrete intervals over  which  the variables  such as  pressure,
temperature and concentration are assumed to be either constant or to
vary in a simple manner.  This discretization results in  a set of non-
linear algebraic equations that must  first be linearized  and then solved
simultaneously, usually using high-speed  digital  computers.

     The emphasis of the following  discussion is  on solutions to the
liquid flow equation Cequation 8);  however,  the methods are applicable
to nonlinear forms of any of  equations 8-12.

     The principal advantage  of numeric methods is their  flexibility.  In
principle, methods that are suitable  for  one dimension can be simply
extended to two and three dimensions.  Conversely, computer codes incor-
porating these methods for multi-dimensional problems can collapse to
handle one-dimensional cases.   Complex boundaries, nonlinear transport
coefficients and boundary conditions, heterogeneity and anisotropy, and
spatial and temporal source/sink variations  can all be handled.  The
principal disadvantages of numeric  methods for nonlinear  problems are the
large computer core and time  requirements and the complexity of computer
codes.

     The following discussion is pertinent to the solution of the water
flow equation.  By suitable rearrangement,  this equation  can be written
in terms of one of three variables.   If the  system under  consideration
remains unsaturated at all times, the flow equation can be written in
terms of the volumetric moisture content  ($)•  This is referred to as
the '0-based1 approach, in which the  driving forces are gradients in
moisture-content concentration.  The  0-based equation is  identical in
appearance with the solute-flux equation  Cequation 11), with the
gravity-driven flux term in equation  8 being equivalent to the advective
term in equation 11.  When part of  the system is  saturated, this approach
fails, because the volumetric moisture content remains constant at satura-
tion, but flow still occurs due to  gradients in pressure, gravitational,
and thermal potentials.  For  partially saturated  systems, equation 8 can
be written in terms of the matric flux potential  Cequation 17), the
pressure potential  , or the sum of  the  pressure and gravitational
potentials (<|>  +  )?  The pressure-potential approach can also be refer-
red to as the pressure-head Ch) approach  by  using units of length for the

pressure potential:  h = —^-,  where  g  » gravitational acceleration.
Similarly, the combined prlssure and  gravitational potential approach is

often termed the total head CH) approach,  where H =» — C4>  + 
-------
                                   98
                        Spatial Discretization Methods

     There are two basic methods of subdividing the spatial domain of a
problem described by equations 8-12:  the use of finite differences or
finite elements.

     Finite-difference methods require subdivision of the region considered
into subregions that have a regular geometry.  These regions are usually
rectangular, but arbitrary polygonal subregions have also been used
(Narasimhan and Witherspoon, 1977).  In finite-difference methods, spatial
derivatives are approximated by assuming that the solution variable varies
linearly between the geometric centers of each of the subregions, and the
storage coefficients are assumed to be constant over each subregion.  Any
sources or sinks present are assumed to contribute uniformly over each
subregion in which they are present.

     Because the gradients in the solution variable are assumed to be
linear between adjacent subregions, the use of finite-difference methods
may require many subdivisions in problems that involve rapidly changing
gradients; many problems in variably saturated flow described by
equations 8-12 are this type.  Examples are wetting fronts in infiltration
problems; sharp solute concentration fronts resulting from a step or pulse
input of contaminant; and evaporation-condensation fronts in combined
heat and moisture flow from a buried heat source.  The existence of sharp
fronts can result in oscillatory solutions even if a fine grid spacing is
used CFinlayson and others, 1978).   For these types of problems, a
numerical dispersion term is often added.  For the water-flow problem,
the use of upstream weighting on relative hydraulic conductivity is
commonly used (Brutsaert, 1971; Finlayson and others, 1978).  Similar
terms must often be used in finite difference solutions to equations
with a strong advection term (equations 11 and 12) to prevent oscillations.

     The requirement for fine subdivision to accurately represent sharp
fronts may result in a large number of simultaneous equations that must
be solved.  This restriction, when combined with restrictions on time-step
length to prevent oscillation and nonconvergence caused by nonlinearities
and (or) presence of an advection term, can result in formulations of
field problems that are very expensive to solve.

     Another disadvantage of finite-difference methods is that, due to
the regular shape of the subregions, many subregions may be required to
accurately represent the geometry of a given field problem.

     Accurate representation of complex geometries is an advantage of the
finite-element method of spatial discretization.  In this method, the
region of interest is subdivided into subregions or elements that may
have irregular shapes.  These may be simple linear quadralaterals,
triangles, or complex quadralaterals with curved sides.

-------
                                   99
     In solving the balance equations 8-12 for these irregular subregions
with a finite-element method, the surface integrals are converted to
volume Integrals using Gauss' divergence theorem.  The resultant equations
are evaluated at a finite number of points or nodes that coincide with the
vertices of and points interior to the subregions or elements.  For
example, the solute transport equation Cequation 10) would become:
                              30r
                       e " Sc IT + 7Cqf(5c) ~ V'LV^c-               (19)
where
     e   is a residual between the approximate and true solutions, and
     p"   is the approximate value of p .
      c                               c*
There are two types of finite element methods:  those derived using vari-
ational calculus, and those based on the method of weighted residuals.
Only the latter will be discussed in thia paper as most of the published
solutions to equations 8-12 use them, and because the resultant equations
from both methods are the same in cases where a variational formulation
exists.

     The weighted-residual methods multiply the residual, e, by a weight-
ing function, which is a polynomial in the spatial coordinates only.
To minimize the error of approximation, it is then required that:

                         /ew. dV = 0, 1-1,  	m                  (20)
                         V
where there are m weighting functions over the solution domain.  This
step in the finite-element method results in  a powerful tool  to control
the accuracy of the approximate solution by choosing suitable weighting
functions (w.).  This has been an important advance in simulating some
sharp front problems  CBender and others, 1975).

     The next step in weighted-residual finite-element methods is to
approximate the solution variable with a series  of  the form:
                                M  .
                           h =  I  h   (t) a)   (x)                    (21)
                               j=>l  J      3
where^
     h  (t) is an unknown coefficient which becomes  the desired solution
      •*       variable, and
     a)  (x) are basis functions that are a function of space  only.

The choice of basis  functions defines  further the  type of  finite-element
method used.  If the basis functions are chosen  to  be the  same as the
weighting functions  (w. in equation 20) , then the method Is a Galerkin
method  (Finlayson and others, 1978).   If the  basis  functions  are chosen
to be proportional to dirac  delta  functions,  then  the method  is an
orthogonal collocation method  CFinlayson and  others,  1978).

     Galerkin methods applied to forms  of equations 8-12 that include a
nonlinear storage term can result  in oscillations when sharp  fronts are
involved  (Mercer and Faust,  1977).  Since the solution of  the algebraic

-------
                                  100


equations resulting from either finite-difference or finite-element
methods often requires iteration due to nonlinearity of the problem,
these oscillations very often result in a failure of the iterative
process to converge.  The oscillations are caused by the evaluation of
the nonlinear storage term at points adjacent to the point of interest.
These oscillations can be minimized by a Galerkin formulation that
evaluates the storage terms as an Integrated average over each element
(Neuman, 1973).  This is sometimes called capacitance lumping.  Ortho-
gonal collocation on finite elements results in a formulation that does
not include the extra storage terms, and hence, is free of this type of
oscillations (Finlayson and others, 1978).

     The following conclusions regarding the preferability of finite-
difference or finite-element methods are adapted from Finlayson and
others (1978, p. 72) in a study of numerical methods applied to very
dry soils.
     1.  "***finite elements*** (Galerkin and orthogonal collocation)***
           are capable of tracking steeper fronts than finite difference
           methods***."
     2.  Most predictive simulations in hydrology using finite element
           methods have not involved fronts as steep as those encountered
           in infiltration into extremely dry soils.
     3.  "***for linear problems with steep fronts, finite element
           methods are much faster for equivalent accuracy (than finite
           difference methods).  For nonlinear problems, the Galerkin
           finite element methods are slower than finite difference
           methods***."
     A.  "***for both finite difference and finite element methods,***
           very steep fronts can only be handled by including numerical
           dispersion in the form of averaged  (or upstream)(conductivities),
           special weighting functions or some other techniques***"
           "***As the front becomes steeper, the number of spatial nodes
           or grid points increases and the time step  (required to mini-
           mize oscillations) correspondingly decreases***."
     5.  "***The preferred method for dry soils is a finite difference
           method with averaged (conductivities) because it introduces
           numerical dispersion and allows calculations without a
           prohibitively large number of nodes***."

                          Time Integration Schemes

     Once the spatial derivatives in equations 8-12 have been approximated
by either a  finite-difference or finite-element scheme, time derivatives
must be approximated In some fashion.  For both finite-difference and
finite-element  spatial discretizations, this is usually done using  finite
differences  in  time.  It is possible to use the Galerkin approximation
 (similar to  equation 21) for the time derivative, as described by Gray
and Finder  (1974) and Zyvoloski  (1973).  However, it has been shown that
no significant  accuracy Improvement is obtained by  this method and  that
additional  computation is required  to formulate and solve  the resulting
 algebraic equations  (Yoon and Yeh,  1975).

-------
                                  101
     Commonly used time-differencing schemes evaluate the spatial deriva-
tive approximations at the previous time-step  (explicit method), the
current time-step (implicit method) or as a weighted average of these
(Crank-Nicolson method).  The explicit method  results in the simplest set
of equations to solve, but is subject to stability restrictions on the
step length that often make its use infeasible (Haverkamp and others,
1977).  The Crank-Nicholson method is the most accurate, but Jeppson
(1974a) has shown that, if the nonlinear transport coefficients are used
with a Crank-Nicholson method that is valid for constant coefficients,
the solution is grossly in error.  If a Crank-Nicholson scheme  that is
valid for varying coefficients is used, Jeppson (1972) also showed that,
if the storage coefficient is divided into the conductivity terms
before the equations are solved, the resulting system of equations is
difficult if not impossible to solve, due to rapid oscillations in the
neighborhood of the true solution.  In addition to these problems,
Neuman (1975b} showed that a Crank-Nicholson approach cannot be used in
solution of equation 8 for problems in which part of the system is
saturated and boundary conditions change in the saturated zone.

     The fully implicit method, which evaluates the spatial derivative
approximations at the current time-step, appears to be the preferred
method (Jeppson, 1974a; Neuman, 1975b; Haverkamp and others, 1977).
While it is slightly less accurate than the Crank-Nicholson approximation,
it permits solution of the widest range of problems likely to be encoun-
tered in waste-disposal problems.

     Recently, Narasimhan and Witherspoon  (1977, 1978a and b), and
Narasimhan and others (1978) developed models  that utilize a so-called
mixed explicit-implicit time-integration scheme.  This scheme is based on
the following:  stability criteria for explicit methods depend  on the
rate of change of the solution variable, being more  stringent for small
grid spacings and for rapid changes in the solution  variable.   Many
problems may involve subregions over which changes are slow due to a large
ratio between the storage coefficient and  the  conductance or to large
elements or finite-difference cells.  The  equations  for  these subregions
can o.ften be evaluated explicitly, requiring implicit evaluation only
where rapid changes are occurring.  However, in many cases,  the computer
time required to evaluate the status of all points  in the system may be
greater than if the entire region had been evaluated implicitly.

     Another time-integration scheme takes advantage of  the  fact that
after the spatial derivatives have been approximated by  finite  differences
or finite elements, the result  is a  set of ordinary  differential
equations of the form:            ^ _
                                 S dt ~ Ax  h>

where A2 h  is  the approximation to  the spatial derivatives.  This set of
equations can be integrated  in  time using methods developed for ordinary
differential equations.  The most common of these methods are fourth-order
Runge-Kutta and fifth-order Milne (James and others, 1968; International
Business Machines, 1975).  These'are essentially explicit methods and are

-------
                                   102
subject to time step restrictions.  Haverkamp and others (1977) found
that 5 to 10 times more computer time was required to solve the nonlinear
one-dimensional water-flow problem with these methods than with a fully
implicit scheme.

     The principal advantage of these methods of time integration is that
they have been incorporated into an applications-oriented language by
International Business Machines (1975) entitled: "Continuous System
Modeling Program" (CSMP).  This language makes it very simple to code a
model for a particular application.  Hillel  (1977) gives sample programs
using CSMP for solving one-dimensional finite differenced forms of the
water flow equation (8); water and solute flow equations (8 and 11); and
moisture and heat flow equations (8, 10, and 12).  CSMP solutions to
equation 8 have also been reported by Van der Ploeg and others (1974) and
Bhuiyan and others (1971).  Combined one-dimensional solute and water-
flow problems have been solved by CSMP and reported by Wierenga (1977),
Beese and Wierenga Q.980), and Siddle and others (1977).

                            Linearization Methods

     Any or all of equations 8-12 may exhibit some degree of nonlinearity
in the transport coefficients or boundary conditions.  Since the equations
resulting from any spatial and temporal discretization schemes must be
linear to solve, some form of linearization  is required.  Three main
linearization methods commonly have been used:  explicit methods, pre-
dictor-corrector methods, and fully implicit methods.

     Explicit linearization involves evaluation of the nonlinear co-
efficients and boundary conditions using values of the solution variable
from the previous time step.  As would be expected,  this method may
require very small time steps to achieve acceptable  accuracy (Jeppson,
1974a; Rubin, 1968; Haverkamp and others, 1977).  For infiltration prob-
lems, the general effect of explicit linearization with  large time steps
is for the solution to lag behind the true solution  (Haverkamp and others,
1977).

     Predictor-corrector methods involve two Iterations  for each time
step.  The first step solves the equations for  one-half  the step length
using the coefficients evaluated at the previous time step or by extrapo-
lating to one-half the time step.  Extrapolation may use the previous
two or three time steps (Rubin, 1967).  The  solution Is  then repeated
for the full time step, using the values  of  the solution variable at one-
half the time step to evaluate the coefficients.   Jeppson  (1974a) has
shown that oscillations in sharp fronts  can result from this method.
The accuracy of the method is dependent  on the accuracy  of the extrapo-
lation; since Iteration is not continued  until convergence, instability
can result  (Jeppson, 1974a; Braester  and  others, 1971),  particularly
with highly nonlinear problems.

-------
                                   103
     Fully implicit schemes involve evaluating the nonlinear co-
efficients and boundary conditions at the current time step by
iteratively solving the equations until convergence occurs.  The value
of the solution variable to evaluate the coefficient for the first
iteration may be obtained by extrapolation from previous time steps
(Freeze, 1971b; Neuman, 1975b) .  While implicit linearization may be the
most time-consuming method, it results in the most stable, accurate
scheme with the least computational effort (Haverkamp and others, 1977).
In addition, the iterative process gives a means of controlling the
accuracy of the solution by specifying the convergence tolerance.

     Although all numeric schemes require a large number of computations,
iterative linearization requires an even larger number.  This would be
only of passing interest, except the earliest reference this author
found to the solution of a form of equation 8 describing vertical one-
dimensional water flow was accomplished by Klute in 1952, using hand
calculations for a fully implicit iterative scheme.

     Convergence of fully implicit iterative linearization schemes is not
guaranteed, and may be slow.  The most effective means of accelerating
convergence is to apply a Newton-Rap hs on scheme to the matrix equations
(Jeppson, 1974b; Brutsaert, 1971; Mercer and Faust, 1977; Finnemore,
1970) .  This scheme may be applied to both the conductivity and storage
terms, or to the more nonlinear of the two.  This author has found it
sufficient for many highly nonlinear problems to apply this linearization
technique to the storage term only.

     The iteration required to solve nonlinear equations often fails to
converge when the solution is approached.  The convergence failure takes
the form of oscillations around the true solution, and often occurs when
static equilibrium is approached.  There is no panacea for this problem;
however, the use of an adjustable damping factor often will result in
convergent solutions.  This is an ad hoc procedure; a more rigorous
approach to optimize the iterative process for nonlinear problems is
currently being investigated by R. L. Cooley, U. S. Geological Survey
(oral commun. , 1980) .

                           Matrix Solution Methods

     All numerical schemes except those involving explicit time and co-
efficient evaluation result in a set of algebraic equations that must
be solved simultaneously.  These equations can be written  in matrix
form as:                       HVl
     Direct methods  involve  some  form  of Gaussian elimination  and  take
advantage of  the  sparse nature of the  coefficient matrices  IA]  and [B].
These methods generally require large  computer  storage  and  time require-
ments.  To obtain accurate solutions,  roundoff  and truncation  error must
be minimized  by the  proper use of single and  double-precision  arithmetic,

-------
                                   104


and such  features as pivoting  to assure diagonal dominance of  the
matrices  {A] and [B] (Nobel, 1969).  Recent  advances in equation
ordering  (Price and Coats, 1973) have significantly optimized  the storage
and time  requirements for using direct solution methods, thus  making
their use more feasible for large  problems.

     The  most commonly used iterative matrix solution methods  are the
iterative alternating direction implicit  (IADI), line successive over
relaxation  (LSOR), and the strongly implicit (SIP) methods.  These methods
solve sets  of simultaneous equations by iteratively improving  an initial
guess at  the solution.  The initial guess uses values of the solution
variable  from the previous time step.  As Braester and others  (1971,
P- 113) point out, "***unfortunately however, none of the methods
(LSOR, IADI) will guarantee convergence of the unsaturated flow solution
in all cases***."

     It should be noted that if the matrix equations are solved by an
iterative procedure, this procedure must be  embedded in the iteration
required  to handle the nonlinearity of the equations.  Some authors
(Freeze,  1971; Brutsaert, 1971) have accomplished this requirement with
an "inner iteration loop" to solve the matrix equations and an "outer
iteration loop" to handle the  nonlinearity.  This author and others have
found it  efficient to combine  the  two iterative loops.

     The  principal advantages  of iterative-solution methods over direct
methods are:  (1) smaller computer core requirements; and (2)  ability to
control the accuracy of the solution (and hence its cost) by specifying
the convergence criteria.  For example, this author has found  it computa-
tionally more efficient to solve strongly nonlinear problems to the same
degree of accuracy with LSOR with  a very restrictive convergence criteria
rather than a direct solution  with a less stringent criteria.

                   Availability of Verified  Numeric Models

     Over 280 references to numeric solutions to nonlinear forms of
equations 8-12 were found in the literature.  The earliest solutions were
for simple  one-dimensional liquid-water absorption problems.   Present
capability  includes solutions  of complex three-dimensional water and
solute-flow problems and coupled-heat and moisture-flow problems.  Figures
3 through 6 summarize chronological development of models incorporating
these solutions.

     Although many real-world  problems can be solved by making the one-
dimensional approximation to the required equations, general codes should
have multidimensional capabilities.  Many of these codes have  been applied
to a variety of field problems.  Table 4 lists the capabilities, applica-
tions, and  limitations of using some of these codes.  While the codes in
table 4 are documented and verified, their use is not necessarily
straightforward because of the many numerical tricks that may  be required
to obtain a solution.  Extreme nonlinearities can preclude solutions in
many cases.  These problems are inherent to  the nature of the  nonlinear

-------
                       105
2
5
65
60
55
50
45
40
35
30
25
20
15
10
 5
                               ONE-DIMENSIONAL
 1960
                  1965
 1970
YEAR
1975
1980
   Figure 3.  — Development  of variably saturated
        finite difference  models since 1960.
 1970
                                                  1980
  Figure 4.  — Development  of variably saturated
          finite element models since 1970.

-------
                     106
 3 -
 1970
                                                 1980
  Figure  5.  — Development of variably  saturated
   water  and solute transport models  since 1970.
10
               (ONE-DIMENSIONAL MODELS ONLY)
                          I
 1970
1975
YEAR
                                                 1980
  Figure 6. — Development of variably saturated
   water and heat transport  models since 1970.

-------
Table t.—Ftaturti ef a"kilobit gmtrol eenputtr codtt far nntrla tolution of variably tatimttd
       vattr, tolutt, and hat trantpert probltnt rttating to ndloaetiv* HWt« dttpotal

Hodtl
BUM Author!
Baca, King*
and Norton
U97J)

Brandt and
othtri (1971)


Brealer
(1»7S)



CSU trutaacrr
(1971)


Brutaaert,
BrcUenbach and
Sunada (1970,
1971, WJ»
Cauaeade.
and other*
(197»)


Cool*?
(1981)



Soluta
Haxlxn trane-
dtmnalom port

1 No


2
(3-»»liT»- Mo
•atrle)

2
(3-axlajm- Taa
metric)



I No



2 Ho
(J-axlajna-
•etrie)

2 Mo



2
(3-utayB- No
Mtrle)


Haat Tlaw
trana- Spatial Inta-
port dliccetiittlon (ntton
CaUrkin-
T» ilntt* alcMnt Illicit


Tinlea
Mo dlffarmci


rinlt* Crank-
No dlffaraau KUhoUon



Flnlt*
No diffartnco Implicit



No Tlniti bpllclt
dlffarane*


Mo rinlti Explicit
dltfaranc*


Orthogonal
Ho collocation Implicit
on Plnlta
•laawato



Llnttdutton Solution
Implicit Cauaalaa
•itti Mavtoa- oliaiination
Hapliaon

lapllclt
with Mavtoo- ADI
lUphaon
tBn1i«<*
inpl ic 1C
vlth Ncvton- ADI
kaphion


lapllclt Una SOI
with Navton- with
Raphaon cornctloo

Inpllclt
with NMton- Llna sot
Raphaon

Alternating
linear
oolutlon with Mono
nonlinear
correction
Implicit SI?




Sourco/ Boundary
alnk condition
typ«» tjrpa*

bpliclt All



Nona All


Explicit All



All,
Explicit Including
aaepago
facia

Explicit All




Nona All


Explicit All.
Including
aaapaga
facaa



Verification
field data, lao-
therMl coluam
data, aulytie
eolution

Unknown


Coluin data



Ub. field data.
other aodela


Ub. field data.
analytic
aolutlona


Analytic
aolut ion


Ub. field data.
other aodela





Ritnitrka
Log tranaforrution of
prenaure head •
tinaAturat<*d
conditions c
-------
TakU t.—ftatuiva of mafoblt amerat otfuUf oodtt for nmnto tolutitm of variably tuturatad *jt«T,
        •oluti, and Awl tfmtfart protbmi Mtattnff to radtoaatl* *ut*
tolttte Halt Tim*
Modal Kaxlmum trana- trana- Spatial Intt-
•a«M Authora dimenolen* port pert diecrotlutlon (ration
Finlayaon,
XeUon. and 1 Ma Toe Finite Implicit
aaca (l«7«) difference
Fre«o
(19/1) 1 Mo Mo Finite Implicit
difference
Ftlod, Cnletkin-
Clllham, ana J Xo po finite a lament Implicit
Flclunu (H77) vith lumped
capacitance



Frlnd and Calerkln-
Verja (U7») 1 -Mo Mo finite element Implicit
vith lumped
capacitance



Hamnel Finite Creak-
Cm*) 1 Mo Tea difference Micholno*
Khaleol end
Redden 1 Ten Mo Finite Implicit
(1»76) difference

Uppala 2
(1M1) (J-axiar»>- Wo Mo Finite Implicit
metric) difference




! Llixnrliattom (olutian
Implicit
vlth Itettto*)* ttakmovn
Kaphaon
Implicit
vlth Ume tot
extraaolatla*
Implicit
vlth Oauatia*
functional «lt»im»tiom
axpanaln
in apace
and time
extrapolation
Implicit
•1th Cauailaa
functional olimlnatlom
expanalon
in apace
and time
extrepolatlo*
Implicit Cauaalaa
elimination
Implicit Cauaalan
elimlnatiom


Explicit or
Implicit UOK. tit.
vtth full Ceueelan
or partial ellmlBatiom
Kevton -
Source/ ioundarjr
aink condition
txpea typea Veriflcetlon

(xpliclt All Field data

All, Uh, field data,
Explicit Includlni analytic
aatpaia aolutloM
(one All On«-dlmen«lon
column dralnaea
data




Column data.
bcpliclt All Finite
difference
•olut lone


Explicit All Uh ami
field data
Analytic and
Explicit All numeric
•olut lone,
experimental
data

Explicit plua All Other modala.
extraction analytic
»y note eolutlona



Kcaarkii
Temperature ecuation
aolvfd analytjcall))-
heat conduction cnljr
Autcmatlc tine ftej*
control.
hyiteretil
Gravity term
trenri*d explicitly*
waate dtapoeal
application
alnulata*



Application to
radioactive vavce
manacement are.i





Tranigwrt >olv
-------
Tabu ^,~r*aturl• of availabU gmtral eonputtr oodff for mtmria tolutlon of variably taturattd aittr,
         aolute, and h*at transport problmt rtlatinff to radiaaativ* aattt dfajxwal--Continued
Solute Neat
Model Maximum trana- Irani- Epatlal
•aw Authore dimenelona port port dlacretliatlon
Lappala and 2
rolled O-axla/v Mo Ye* Finite
{1981) Mtrlc) difference

Marlao CalerkU-
(197S) 2 Tea Ko finite element
with lumped
capacitance



T»BST Rerailmhan 2 Integrated
(197)) O-axleym- Ko »» finite
•«trlc) difference


and Jeppeon 2 No No Finite
(w,j) dlffarenc.

Xaruimhan 2 finite element
(1977) O-axleym)- «0 «o with lumped
Htrlc) capacitance


IMSAT Heuman 2 Calerkln-
2 (1972, 1»7J (J-»xiay»- do »o finite element
197*. 197$) metric) with lumped
ferrena Unit*

Time
Inte-
gration

Implicit



Implicit





Mixed
explicit-
Implicit


Tine
centered
Mixed
explicit-
Implicit,
implicit-
explicit, at
time centered
Implicit
or time
TlM



Llnearltetlom,
Implicit
with Newton-
taphaon

Implicit
with
functional
expanelon
in apace and
extrapolation
In time
Explicit or
Implicit
with
extrapolation
Implicit
with Hevton-
Upheon
Implicit
or
explicit



Implicit
with
extrapolation
Implicit



Solution
LSOd, SIP,
Gauaalaa
elimination


Cauiela*
elimination





rolnt
sot


rolnt
SO*

rolnt
>0x



Cauaalan
elimination
IADI

Source/
elnk
typea
Explicit
plua note
and
evaporation

Explicit






Explicit



•one


Explicit




evapo-
ration,
implicit
explicit

Boundary
condition
typea

All



All,
Including
aeepage
faeee



All



All


All




All,
Including
aeepaga
face*
All



Verification

Analytic
aolutiona
lab data

Unknown






other modela.
analytic
aolutiona
Held
experimental
data

data, analytic
aolutieni, other
modela



Ub, field data,
othrr numeric
modela, analytic
aolutlona




Remarka

Unpubllahed research
model









matlc time ate?
contrnl, defornuble
media, may be alow

Sloping eoordtnJtea

Automatic tli* met
control, mar or
a lew



reported aprlic'.ition
to very dry aeiln.
well bore calculation*



-------
Tabla t.—Ttatunt of availablt gmvnt eatputtr aodti for ntvrto tvtution ff variably taturatid water,
        •olutt, and tuat traniport pmhtfre rtlating to nxHoaetit* itutt iUtpo«tl~Cmtl



Linoarliatlo* folutloa
Inpllcit
vttti CaviatM
functional •llnlaatlo*
•xponilM
In »p«c«
and tlM
axtrapoUtlon
lofllclt Oautlan
altei nation


Inpllclt
with C«u««l««
•xtrapolatlo* allnlnatlon



txpllelt ADI


Inpllclt Ron



Iipllclt
with *0«
Nevtoo.-
tathaon
Ixpllelt Hone

tourc*/ Boundary
• ink condition
typea typaa Verification Xeaarka

Mnt All, Coluan data,
including finite
aeepaga difference
facaa oodcli.
annlylic
aolutiona
txpllclt All, finite (me r«r»Tta r< n»
including dKferrnca obtainable aolvllni
aeepaga nnlrla, for very *on-
facaa (laid data linear problem*
one-JIrwneion
txpllcit All coluum, lab
data. Brealera
finite
dlfforenc*
•ndel
Explicit All Unknown


Nona All, lab tank
including data
aoepaga
facea
Coiw«Tgcnce
Explicit Ail Unknown problem for vtry
nonlinear problcna


Explicit All Other -odola, e-b..ed -
analytic unaaturated
eolutloni condition! only.
Ko gravity tent.
very long CrT tlm

-------
T«bl« ^.—•f^atUf»^ of avallablt gmaral oofitttr oodit far nmHe lolutlon of variably taturatcJ uater,
         toluti, and htat trantpart proMewe nlatlng to radtoaaHvt Halt* oK*poea{--Contlnued

Model
aaae Auchora
Vauclln
and othera
(1975)

W.1 and
Jeppaon
(1971)



Zyvoloikl
anil Iruch
(M73)


.Solute Heat Tine
Maxima; trana- trane- Spatial Inte-
elMMlona port port dlicretliatlon gratlon

2 No Ho Finite Implicit
difference

2
O-axU]W Mo Ro Finite (taady
•atrlc) difference etata




2 Mo Ho Calerkln- TlM
finite element centered




Llnearlaatlon Solution
Implicit IADI



Mwton-
lapheon *O*




tBpllclt
«lth *hepa Cauaalan
function ellmioatloo.
expanalon
In aptfca
Source/ Boundary
• ink condition
typea typea
Ron* All,
Including
aiepaga
facea

Ron* Dirlchlet






ROM M.1




Verification
tab tank
data



None
reported




Finite
difference
awdela



Reaarka




Im
-------
                                   112
forms of equations 8-12 and the methods of numeric approximation used
for their solution.  A thorough understanding of the physical phenomena
described by the equations themselves, the numeric methods involved, and
the definitions of valid solutions (in a qualitative or semi-quantitative
sense) are requisite for the responsible use of these models.

     One check of the validity of numerical solutions, when the results
cannot be anticipated a priori, was suggested by the National Academy of
Sciences in a report on risk assessment in radioactive waste disposal
studies (National Academy of Sciences, 1979).  Their suggestion was the
application of several models to the same problem to give a consistency
check.  This check is necessary because of uncertainties in a numerical
model owing to roundoff and convergence errors as well as uncertainties
in material properties and transport coefficients for a given field
problem.  Such duplicate simulation might be an affront to the authors
of particular computer codes, and might appear a duplication of effort
to some involved in the management of scientific studies.  However, in
the light of the stakes involved in proper selection and management of
disposal sites for extremely hazardous wastes for very long times, such
checking seems mandatory.

                              Parametric Models

     The use of solutions to equations 8-12 discussed above depends upon
several factors:   (1) the availability of an analytic solution or compu-
ter code using a particular numerical solution; (2) the availability of
measured values of the nonlinear transport coefficients as a function of
the solution variable as well as a function of space; (3) the availability
of data to calibrate the models over the range of expected conditions dur-
ing the life of the disposal site; and (4) funding to adequately simulate
nonlinear transport processes over periods that may exceed 1,000 years.
Restrictions imposed by the last three factors may preclude the use of
detailed physics-based models.  In these cases, adequate predictive
capability may be met by the use of parametric or input-output models.

     Parametric models simulate the movement of mass or energy by consid-
ering the modeled  system to consist of a series of empirical functions
that transform an  input into an output:

               Input    -    Transfer        ^
               Aupuu         functions                 v

      Cfor example, rainfall)           (for example, infiltration rate)

Transfer functions may be purely empirical or based on some physical
principle.  Often, transfer functions are derived  from analytic solutions
to simple formulations of one part of the simulated system.  As an  example,
the translation of rainfall into infiltration can  be accomplished by using
the Green-Ampt infiltration equation  (see p. 19) as the  transfer  function.

-------
                                   113


     Parametric models are most applicable to systems in which the
physical and chemical processes simulated are so weakly coupled that they
can be considered independent.  However, some feedback mechanisms can be
incorporated in these models to allow some degree of coupling.

     Eagleson, in a series of papers (1978a, b, c, d, e, f, and g) gives
a comprehensive overview of parametric models applied to rainfall-
infiltration-runoff modeling.  Comprehensive parametric models that in-
clude the movement of water in variably saturated soils are described by
Skaggs (1978), Afonda (1978), Crow and others (1977), Fitzpatric and Nix
(1969), Neibling (1976), and Neibling and others  (1977),

     Parametric models of solute movement in variably saturated systems
have been described by King and Hanks (1973), and Davidson and others
(1978).  Dutt and others (1972) combine a numeric solution to the liquid-
water flow problem with a parametric representation of the chemical
transformations for an unsaturated system.

     The principal advantages of parametric models are: (1) the reduced
number of dependent variables and parameters; and (2) that they are
usually structured to give directly the desired output in terms of mass
and energy flux rates.  Most analytic or numeric methods solve for the
spatial and temporal variation in pressure, concentration, or temperature.
Flux rates must then be computed as an additional step.  Hence, para-
metric models directly provide the desired fluxes with less effort than
numeric models.  Another advantage of parametric models is that because
their mathematical structure is relatively simple, the effects of uncer-
tainties in the input can be quantitatively related to uncertainties
in the output.

     Parametric models usually are given some physical basis by fitting
transfer functions to observed data from field experiments; this practice
often limits the generality of these models.  The principal disadvantage
of this approach is that data collection and analysis must be pursued
over long time periods to assure an adequate model.

     An approach that minimizes this expensive process consists of using
detailed numeric models as the "experimental facility" rather than a field
site.  Once a numeric model is verified, if it is sufficiently general, it
may be used cheaply to conduct numerical experiments to derive the appro-
priate transfer functions.  An example of this use of numeric models is
the work of Smith (1972) in deriving rainfall-infiltration transfer
functions.  This seems a particularly promising area for future work.
                                    SUMMARY

     Physics-based or deterministic models  that describe mass and energy
transport through variably saturated porous media  can be expressed by one
or more nonlinear partial differential equations.  These equations are
derived by combining equations  for  continuity  of mass and energy with
constitutive relationships.  The  latter express mass or energy flux as

-------
                                  114


proportional to a driving gradient in measurable variables, such as tem-
perature, pressure, or concentration.  The proportionality functions are
defined as general transport coefficients.  The equations can be nonlinear
because of dependence of the transport coefficients, boundary conditions
and sink strengths upon temperature, pressure, or concentration.

     The most flexible formulation of the nonlinear liquid-water flow
problem is obtained by writing the equation in terms of pressure or total
potential rather than as a concentration-dependent problem with volumetric
moisture content as the solution variable.  The pressure or total poten-
tial formulation allows simple unified treatment of both saturated and
unsaturated conditions.

     Analytic, quasi-analytic, and numeric methods are available for
solving the descriptive equations for a given waste-disposal problem.
Analytic and quasi-analytic solutions are generally available for most
multidimensional steady-state water-flow problems, if the hydraulic con-
ductivity can be described by an exponential function of head or moisture
content.  Analytic solutions that consider heterogeneity in the transport
coefficients are available if the heterogeneity is restricted to layering
for one-dimensional, steady-state problems, and to linear or exponential
dependence on the spatial coordinates in multidimensional transient
problems.  Analytic solutions that include source and sink terms that are
simple functions of spatial coordinates and the solution variable are
available for some simple geometries.

     If the assumptions required for a particular analytic solution are
reasonable for analyzing waste-disposal problems, it should be used.
However, most waste-disposal problems are sufficiently complex that
analytic solutions are not flexible enough to provide reasonable answers.
Few, if any, analytic approaches are available when two or more coupled
equations are needed.

     The complexities of most waste-disposal problems can usually be
handled by approximate numeric methods.  These methods consist of spatial
discretization, using finite differences or finite elements combined with
suitable time integration, linearization, and matrix solution schemes.

     For problems with very mild nonlinearities in the storage term,
Galerkin finite-element methods are the most flexible.  These methods are
particularly accurate for tracking sharp fronts.  When the storage term
is more than mildly nonlinear, the Galerkin finite-element schemes must
be modified by lumping the storage term over each element.  Orthogonal
collocation on finite elements and finite differences are the most
flexible solution methods for the nonlinear transport equations.  The
application of upstream weighting on the conductivity that may be
required to prevent numerical oscillation in some problems is more
straightforward in finite-difference methods than in finite-element
methods.

-------
                                  115
     Time-integration schemes that are fully implicit in the solution
variable are preferred for nonlinear problems, and required for the vari-
ably saturated water-flow equation, when boundary conditions change in
the saturated zone.  Crank-Nicholson methods are the most accurate, but
a scheme that is valid for nonconstant transport coefficients must be
used.  Explicit time-integration schemes for nonlinear problems have such
stringent time-step restrictions that their use is generally not feasible
for simulating large systems over long time periods.

     Linearization methods that evaluate nonlinear terms using the current
values of the solution variable (implicit linearization) with iterative
improvement are the most accurate and flexible.  Evaluating the nonlinear
terms using values of the solution variables from the previous time step,
and extrapolation using previous time steps without iterative improvement,
can result in erroneous solutions.  Implicit Newton-Raphson linearization
of the storage and Cor) conductance terms often speeds convergence, and
is required for a convergent solution in many extremely nonlinear prob-
lems.  Ad hoc damping coefficients may have to be applied to obtain con-
vergence in many problems as they approach steady state.  The matrix
equations that result from the foregoing steps can be solved by some form
of direct Gaussian elimination or by an iterative scheme.  Direct solution
methods are the most accurate if they include matrix conditioning and the
proper use of single- and double-precision arithmetic.  The major dis-
advantage of direct methods are large computer core and time requirements.
Iterative matrix solution methods require less core and often less time
to achieve the same accuracy.  Two additional advantages of iterative
methods are:  (1) accuracy of the solution can be controlled by specifying
convergence criteria; and (2) the methods can be efficiently embedded in
the iteration required for linearization and coupling of equations.

     Documented computer codes are available to handle the following
types of variably saturated transport problems relative to waste
disposal:
     1.  Three-dimensional movement of liquid water and solutes in
           unfractured, heterogeneous, anisotropic media;
     2.  Two-dimensional planar and three-dimensional axisymmetric fully
           coupled movement of liquid water, moist air, water vapor, and
           heat in unfractured heterogeneous anisotropic media;
     3.  Three-dimensional uncoupled movement of liquid water, moist air,
           and solutes in unfractured heterogeneous, anisotropic media.

     Some of the multidimensional, multiphase, multicomponent codes are
so general that their application to a simple problem may be inefficient.
However, several codes are available that are efficient for simpler
problems.  The capabilities of the most flexible documented codes are
summarized in table 4.

     Although there are many existing computer codes with capability to
simulate many realistic problems  in radioactive waste disposal, there are
several areas in which considerable effort  should be expended to assure
adequate ability to predict  the fate of radionuclides and other high-level
wastes emplaced in variably  saturated media.  These  include the following:

-------
                                  116
     1.  The physical mechanisms of flow of water and solutes through
variably saturated, fractured, and structured media need to be incorpo-
rated in simulation models.  Movement through large openings in these
media dominate infiltration and solute movement.

     2.   The direct use of detailed physics-based models for long-term
(on the order of 100 to 1,000+ years) prediction of the fate of radio-
active pollutants may not be defendable on the following grounds:
     a.  Uncertainties in data due to spatial variation of transport
           properties and boundary conditions may be so large that
           collection of the required data for adequate simulation with a
           detailed model would be prohibitively expensive.  This data
           uncertainty is further compounded by the necessity to determine
           many transport properties in the laboratory at a generally
           smaller scale from the field problem; the transferability of
           laboratory determined data to field conditions must be
           established.
     b.  Insufficient data are available to adequately calibrate detailed
           models for particular sites.
     c.  Computer costs for running realistic problems may be prohibitive.

     The need for usable, reliable prediction tools for waste-disposal
problems can perhaps be better met by the use of simplified parametric or
input-output models, in which the functions that translate input into
output have some physical basis.  These functions can often be determined
by numerical experimentation with physics-based models.  The physics-based
models used to train the parametric models need to be verified by
simulating physical and chemical transport from controlled laboratory and
small-scale field experiments.

     3.  For both physics-based and parametric models, the uncertainty in
any prediction of the fate of pollutants should be quantified, based on
all the assumptions in the model and uncertainties in the data.  This
area of modeling of ground-water systems is in its infancy.  However, it
will be required so that confidence limits can be placed on any predictive
results and incorporated in risk analyses used in design and management
decisions.

     Whether physics-based models are used directly to simulate waste-
disposal problems or used to train parametric models, their adequacy
should be checked by duplicate simulation of the same problems with
another model or models.  This step seems mandatory in the light of the
complexity of most waste-disposal problems, uncertainty in model
parameters, and the necessity to use numerical tricks to obtain a
solution.

-------
                                   117


                          SELECTED REFERENCES

                               General

Ames, W. F., 1965, Nonlinear partial  differential  equations  in
      engineering:  Academic Press 511 p.
Arnett, R.  C., and others, 1976,  Conceptual  and mathematical modeling  of
     the Hanford ground-water  flow regime:   Atlantic  Richfield Hanford Co.,
     Report no. ARH-ST-140.
Bear, J., 1972, Dynamics of fluids in porous media:   American'Elsevier,
     764 p.
Cooley, R.  L., 1974, Finite element solutions  for  the equations of
     groundwater flow:  Technical Report  Series HW—publication no.  18,
     Center Water Resources Research, Desert Research Institute,
     University of Nevada, 134 p.
Gray, W. G., and Finder, G. F., 1974, Galerkin approximation of the  time
     derivative in the finite element analysis of  groundwater flow:  Water
     Resources Research, v. 10, p. 821-828.
Grove, D. B., and Kipp, K. L., 1980, Modeling  contaminant  transport  in
     porous media in relation to nuclear waste disposal—a review:   Paper
     prepared for the Interagency Conference on Radioactive  Waste Modeling,
     December 1980, 30 p.
International Business Machines,  1975, Continuous  system modeling program
     III (CSMP III) program reference manual:  International Business
     Machines Corporation, 206 p.
James, M. L., Smith, G. M., and Wolford,  J. C., 1968,  Applied  numerical
     methods for digital computation with FORTRAN:   International Text
     Book Co., 514 p.
Laliberte,  G. E., Corey, A. T., and Brooks, R.  H.,  1966,  Properties of
     unsaturated porous media:  Colorado State University Hydrology
     Paper  17, 40 p.
Mualem, Y., 1976, A new model for predicting the hydraulic  conductivity
     of unsaturated porous media:  Water Resources  Research,  v. 12,
     p. 513-522.
National Academy of Sciences, 1979, Implementation  of  long  term environ-
     mental radiation standards.   The issue of verification:   Report
     prepared by the panel on The Implementation Requirements of
     Environmental Radiation Standards, Committee on Radioactive Waste
     Management, 65 p.
Nobel, B.,  1969, Applied linear algebra:  Prentice  Hall,  Inc., 523  p.
Narasimhan, T. N., and Witherspoon, P. A., 1976, An integrated finite
     difference method for analyzing fluid flow in porous media: Water
     Resources Research, v. 12, p. 57-64.
Price, H. S., and Coats, K. H., 1973, Direct methods in reservoir
     simulation:  Society of Petroleum Engineers of the American Institute
     of Mining Engineers, Paper No. 4278, 20 p.
United States Environmental Protection Agency,  1978,  Utilization of
     numerical groundwater models for water resource management: U.S.
     Environmental Protection Agency,  R.  S. Kerr Environmental Research
     Laboratory.

-------
                                  118


                              REFERENCES

                          Analytic Solutions

Babu, D. K., 1976a, Infiltration analysis and perturbation methods 1.
     Absorption with exponential diffusivity:  Water Resources Research,
     v. 12, p. 89-93.
	1976b, Infiltration analysis and perturbation methods 2.  Horizontal
     absorption:  Water Resources Research, v. 12, p. 1031-1018.
	1976c, Infiltration analysis and perturbation methods 3.  Vertical
     infiltration:  Water Resources Research, v. 12, p. 1014-1024.
     	1977, Perturbation solutions for vertical infiltration problems.
      A uniformly valid solution:  Advances in Water Resources, v. 1,
     p. 111-116.
      1979, Steady state moisture flows in unsaturated soil containing a
     plane, isolated, impermeable barrier:  Advances in Water Resources.
     v. 2, p. 35-45.
Batu, V. M., 1977, Steady infiltration from a ditch:  Theory and
     experiment:  Soil Science Society of America Journal, v. 41,
     p. 677-682.
	1978, Steady infiltration from single and periodic strip sources:
     Soil Science Society of America Journal, v. 42, p. 544-549.
      1980, Flow net for unsaturated infiltration from periodic strip
     sources:  Water Resources Research, v. 16, p. 284-288.
Batu, V. M., and Gardner, W. R., 1978, Steady state solute convection in
     two dimensions with nonuniform infiltration:  Soil Science Society
     of America Journal, v. 42, p. 18-22.
Ben Asher, J., Lomen,vD. 0., and Warrick, A. W., 1978, Linear and nonlinear
     models of infiltration from a point source:   Soil Science Society of
     America Journal, v. 42, p. 3-6.
Braester, C., and others, 1971, A survey of the equations and solutions
     of unsaturated flow in porous media:  First annual report, project
     no. A105WC77, Hydraulic Engineering Laboratory, Technion, Haifa,
     Israel, 176 p.
Braester, C. N., 1973, Moisture variation at the soil surface and the
     advance of the wetting front during infiltration at constant flux:
     Water Resources Research, v. 9, p. 697-994.
Brustkern, R. L., and Morel-Seytoux, H. J., 1970,  Analytical treatment of
     two-phase infiltration:  Journal of Hydraulics Division, American
     Society of Civil Engineers, v. HY12, p. 2535-2548.
Brutsaert, Wilfried, 1968a, The adaptability of an exact solution to
     horizontal infiltration:  Water Resources  Research, v. 4, p. 785-789.
	1968b, A solution for vertical infiltration  into a dry porous medium:
     Water Resources Research, v. 4, p. 1031-1038.
	1974, More on an approximate solution for non-linear diffusion:
     Water Resources Research, v. 10, p. 1251-1252.
	1976, The concise formulation of diffusive sorption of water in a
     dry soil:  Water Resources Research, v. 12, p. 1118-1124.
	1977, Vertical infiltration in dry soil:  Water Resources Research,
     v. 13, p. 363-368.

-------
                                  119


Brutsaert, Wilfried,  and Weisman, R. N.,  1970,  Comparison  of  solutions
     of a nonlinear diffusion equation:   Water  Resources Research  v. 6
     p. 642-644.
Childs, E. C.,  1969,  An introduction to  the  physical basis of soil water
     phenomena:   John Wiley  and  Sons,  Ltd.,  493 p.
Childs, E. C.,  and Bybordi,  M. M.,  1969,  The vertical movement  of water
     in stratified porous media,  1.  Infiltration:  Water  Resources
     Research,  v. 5,  p. 446-459.
Cisler, J.,  1974, Note on the Parlange method for the numerical solution
     of horizontal infiltration  of  water  in  soil:   Soil Science, v.  117
     p. 70-73.
Collis-George,  N., 1977, Infiltration  equations for simple soil systems:
     Water Resources  Research, v. 13,  p.  395-403.
de Laat, P.  J. M., 1976, A pseudo steady  state  solution of water movement
     in the  unsaturated zone of  the soil:  Journal  of Hydrology, v.  30,
     p. 11-27.
Dlrksen, C., 1978, Transient and  steady flow from subsurface  line sources
     at constant  hydraulic head  in  anisotropic  soil:  American  Society
     of Agricultural  Engineers Transactions,  v.  21.
Elrick, D. E.,  and Laryea, K. B., 1979, Sorption of water  in  soils:  A
     comparison of techniques  for solving the diffusion equation:  Soil
     Science, v.  128, p. 369-374.
Freyberg, D. L.,  and  others,  1980,  Application  of the Green-Ampt model
     of infiltration  under time-dependent  surface water depths:  Water
     Resources Research, v.  16, p.  517-528.
Fujioka, Y., and Kitamura, T., 1964, Approximate solution  of  a  vertical
     drainage problem:  Journal of  Geophysical  Research, v. 69,
     p. 5249-5255.
Garber, M.,  and Zaslavsky, D., 1977, Flow in a  soil underlined  by an
     impermeable membrane:   Soil Science,  v.  123, p. 1-9.
Gardner, W.  R., 1958, Some steady state solutions of the unsaturated
     moisture flow equation  with application to evaporation from a water
     table:  Soil Science, v. 85, p. 228-232.
	1959,  Solution  of the  flow equation for the drying of soils and
     other porous media:  Soil Science Society  of America  Journal,
     v. 23,  p. 183-187.
Gardner, W.  R., 1962, Approximate solution of a nonsteady  state drainage
     problem.  Soil Science  Society of America  Journal, v.  26,  p. 129-132.
Gardner, W.  R., and Mayhugh, M.  S., 1958,  Solutions and tests of the
     diffusion equation for  the movement  of  water in soil:  Soil Science
     Society of America Journal, v. 22, p. 197-201.
Gilley, J. R., and Allred, E. R., 1974a,  Irrigation and root  extraction
     from subsurface  irrigation  laterals:  American Society of
     Agricultural Engineers  Transactions,  v.  17, p. 927-933.
Green, W. H., and Ampt, G. A., 1911, Studies on soil physics, 1, The flow
     of air  and water through soils:   Journal of Agricultural Science,
     v. 4, p. 1-24.
Hayhoe, H. W., 1978,  Numerical study of quasi-analytic and finite
     difference solutions of the  soil  water  transfer equation:  Soil
     Science, v.  125, p. 68-74.

-------
                                   120
 Heaslet, M.  A.,  and Alksne,  A.,  1961,  Diffusion from a fixed surface
     with  concentration dependent coefficients:  Journal of  Society
     of Industrial Applied Math,  v.  9,  p.  584-596.
 Irmay,  S.,  1966,  Solutions of the nonlinear diffusion equation with  a
     gravity term in hydrology,  in P.  E. Rijtema and H.  Wassink  (eds.),
     Water  in the unsaturated zone:  Wageningen Symposium, UNESCO, Paris,
     France,  p.  478-499.
 Knight,  J. H., and Philip, J.  R.,  1973, On solving  the unsaturated flow
     equation, 2.   Critique  of Parlanges method:  Soil Science,  v. 116,
     p.  407-416.
 Kobayashi, H., 1966,  A theoretical analysis and numerical solutions  of
     unsaturated  flow in soils,  in P.  E. Rijtema and H.  Wassink  (eds.),
     Water  in the unsaturated zone:  Proceedings of the  Wageningen
     Symposium, UNESCO,  Paris, France.
 Larson, N. M., and Reeves, M., 1976, Analytical analysis of  soil moisture
     and trace contaminant transport:   Oak Ridge National Laboratory
     Report  ORNL  NSF/EATC 12,  178  p.
 Liu, P. F.,  1976,  A perturbation  solution  for  a nonlinear diffusion
     equation:  Water Resources Research,  v. 12,  p.  1235-1240.
 Lomen, D. 0., and Warrick, A.  W.,  1976a, Solution of the one dimensional
     linear  moisture flow equation with implicit water extraction
     functions:   Soil Science Society  of America Journal, v.  40, p.342-344.
	1976b,  Time-dependent  linearized  infiltration II;  Line sources: Soil
     Science  Society of  America Journal, v.  38,  p.  568-572.
	1978a,  Linearized  moisture  flow with loss at  the soil  surface:  Soil
     Science Society of America Journal, v. 42, p. 396-399.
     _1978b, Time-dependent solutions to the one-dimensional linearized
     moisture flow equation with water extraction:  Journal of Hydrology,
     v. 39, p. 59-67.
Mein, R. G., and Larson, C. L., 1973, Modeling  infiltration during a steady
     rain:  Water Resources Research, v. 9, p.  384-394.
Merrill, S. D., Raats, P. A. C., and Dirksen, C., 1978, Laterally confined
     flow from a point source at the surface of an inhomogeneous soil
     column:  Soil Science Society of America Journal, v. 42, p. 851-857.
Morel-Seytoux, H. J., 1978, Derivation of equations for variable rainfall
     infiltration:  Water Resources Research, v. 14, p. 561-568.
Morel-Seytoux, H. J., and Khanji, J. J., 1974,  Derivation of an equation of
     infiltration:  Water Resources Research, v. 10, p. 795-800.
	1975, Prediction of inhibition in a horizontal column:  Soil Science
     Society of America Journal, v. 39, p. 613-617.
Nakano, Y., 1980, Traveling wave solutions of saturated and unsaturated
     flow through porous media:  Water Resources Research, v. 16, p.
     117-122.
Neilsen, D. R., Kirkham, D., and Van Wijk, W. R., 1961, Diffusion equation
     calculations of field soil water infiltration profiles:  Soil Science
     Society of America Journal, v. 25, p. 5-8.
Parlange, J. Y., 1971a, Theory of water movement in soils, 1, one
     dimensional absorption:  Soil Science, v.  Ill, p. 134-137.
	1971b, Theory of water movement in soils, 2, one dimensional
     infiltration:  Soil Science, v. Ill, p. 170-174.

-------
                                   121


Parlange,  J.  Y.,  1971c,  Theory of water movement in soils,  3,  two and
      three dimensional absorption:  Soil Science,  v.  112,  p.  313-317.
	1972a,  Theory of water movement  in soils,  4,  two and  three
      dimensional  steady infiltration:   Soil Science,  v.  113,  p.  96-100.
	1972b,  Theory of water movement  in soils,  5,  unsteady infiltration
      from  spherical cavities:  Soil Science,  v.  113,  p.  156-161.
       1972c,  Theory of water movement  in soils,  6,  effect  of  water depth
      over  soil:   Soil Science,  v.  113,  p.  308-312.
     _1972d,  Theory of water movement  in soils,  7,  multidimensional
      cavities under pressure:  Soil Science,  v.  113,  p.  379-382.
     	1972e,  Theory of water movement  in soils,  8,  one dimensional
      infiltration with constant flux at the  surface:  Soil  Science,
      v.  114,  p.  1-4.
     _1973a,  Theory of water movement  in soils,  10, cavities with constant
      flux:   Soil Science, v.  116,  p. 1-7.
     __1973b,  A note on three parameter  soil water diffusion function.
      Application to the horizontal infiltration  of  water:   Soil Science
      Society of  America Journal, v.  37,  p. 318-319.
     _1973c,  Horizontal infiltration of water in soils.  A  theoretical
      interpretation of recent experiments:   Soil Science Society  of
      America Journal,  v. 37,  p.  329-330.
     _1975a,  Theory of water movement  in soils,  11, conclusion and
      discussion  of  some recent  developments:   Soil  Science  119, p. 158-161.
     _1975b,  Convergence and validation of time  expansion solutions.  A
      comparison  to  exact and approximate solutions:   Soil Science Society
      of America  Journal, v.  39,  p.  3-6.
      _1975c,  On  solving the  flow equation in  unsaturated soils by
     optimization.  Horizontal  infiltration:   Soil  Science  Society of
     America Journal, v. 39, p. 415-418.
Parlange, J. Y., and Aylor, D., 1972, Theory of water movement  in soils,
     9, the dynamics of capillary rise:   Soil  Science, v. 114,  p. 79-81.
	1975, Response of an unsaturated soil to forest transpiration:
     Water Resources Research,  v. 11, p.  319-323.
Parlange, J. Y., and Babu, D. K., 1976a,  A comparison of  techniques  for
     solving the diffusion equation with  an exponential diffusivity:
     Water Resources Research,  v. 12, p.  1317-1318.
	1976b, On solving the infiltration  equation—a comparison of
     perturbation and iterative techniques:  Water  Resources Research,
     v. 12, p. 1315-1316.
Parlange, J. Y., and Braddock,  R. D., 1978, Comment on the  paper "Numerical
     study of quasi analytic and finite difference  solutions of the  soil
     water transfer equation" by H. N. Hayhoe: Soil Science, v. 127,
     p. 187-198.
	1980, An application of Brutsaerts  and optimization techniques to
     the nonlinear diffusion equation.  The influence of  tailing:  Soil
     Science, v. 129, p. 145-149.
Philip, J. R., 1954, An infiltration equation  with  physical significance:
     Soil Science, v. 77, p. 153-157.
	1955, Numerical solution  of equations of the  diffusion type with
     diffusivity concentration  dependent:  Transactions of  the  Faraday
     Society v. 51, p. 885-892.

-------
                                  122
Philip, J. R., 1957a, Numerical solution of equations of the diffusion
     type with diffusivity concentration dependent II:  Australian
     Journal of Physics, v. 10, 29-42.
	1957b, The theory of infiltration, 1, the infiltration equation and
     its solution:  Soil Science, v. 83, p. 345-357.
	1966a, Absorption and infiltration in two and three dimensional
     systems, Jji P. E. Rijtema and H. Wassink  (eds.), Water in the
     unsaturated  zone:  Wageningen Symposium, UNESCO, Paris, France,
     p. 503-516.
	1966b, A linearization technique  for the study of infiltration, in
     P. E. Rijtema and H. Wassink (eds.), Water in the unsaturated zone:
     Wageningen Symposium, UNESCO, Paris, France, p. 471-475.
      1967, Theory of infiltration, in  Advances in hydroscience, v. 5,
     ed. V. T. Chow:  Academic Press, New York, p. 216-296.
     	1968a, Absorption and infiltration in  two and  three dimensional
     systems, ^n P. E. Rijtema and H. Wassink  (eds.) Water in  the
     unsaturated zone:  Wageningen Symposium,  UNESCO, Paris, France,
     p. 503-515 and 523-525.
     	1968b, Steady infiltration  from buried point sources and spherical
     cavities:  Water Resources Research, v. 4, p. 1039-1047.
     	1971, General theorem on steady infiltration from  surface
     sources, with application to point and  line  sources:  Soil Science
     Society of America Journal,  v.  35, p. 867-871.
     	1972, Steady infiltration from buried, surface and perched point
     and line sources in heterogeneous soils,  1,  analysis:  Soil Science
     Society of America Journal,  v.  36, p. 268-273.
     	1973, On solving unsaturated flow equation, 1, the flux-concentration
     relation:  Soil Science, v.  116, p. 328-335.
      1974, Recent progress in the solution  of nonlinear diffusion
     equations:  Soil Science, v.  117, p.  257-264.
Philip, J. R., and Forrester, R. I.,  1975,  Steady  infiltration  from buried,
     surface  and perched point, and line  sources in  heterogeneous  soils,
     II, flow details discussion:  Soil Science Society  of America Journal,
     v. 39, p. 408-414.
Philip, J. R., and Knight, J. H.,  1974, On solving the unsaturated flow
     equation, 3, new quasi  analytic  technique:  Soil Science,  v.  117,
     p. 1-13.
Raats, P. A.  C., 1970, Steady infiltration from line sources  and furrows:
     Soil Science of.America Journal, v.  34,  p. 709-714.
	1971a,  Some properties of flows in  unsaturated soils with an
     exponential dependence  of the hydraulic  conductivity upon  the
     pressure head:  Journal of Hydrology,  v. 14,  p. 129-138.
	1971b,  Steady  infiltration  from point sources, cavities,  and basins:
     Soil Science Society  of America  Journal, v. 35, p.  689-694.
	1972, Steady infiltration from  sources  at arbitrary depth:   Soil
     Science  Society of America Journal,  v. 36, p. 399-401.
      1973, Steady upward  and downward  flows  in a  class  of
     unsaturated soils:  Soil Science,  v.  115, p.  409-413.
	1977, Laterally confined flows  of water from sources and  to  sinks in
     unsaturated soils:  Soil Science Society of America Journal,  v.  41,
     p.  294-304.

-------
                                  123
Raats, P. A. C., and Gardner, W. R., 1974, Movement of water in the
     unsaturated zone near a water table in Jan Van Schilfgaarde (ed.)
     Drainage for agriculture:  Agronomy 17, p. 311-357.  American
     Society of Agronomy, Madison, Wisconsin.
Reichardt, K., Nielsen, D. R., and Bigger, J. W., 1972, Scaling of
     horizontal infiltration into homogeneous soils:  Soil Science Society
     of America Journal, v. 36, p. 241-245.
Ripple, C. D., Rubin, J., and van Hylckama, T. E. A., 1972, Estimating
     steady-state evaporation rates from bare soils under conditions of
     high water table:  U.S. Geological Survey Water-Supply Paper 2019A,
     39 p.
Scott, E. J., and others, 1962, Power series solutions of the one
     dimensional flow equations for exponential and linear diffusivity
     functions:  Agricultural Research Service, U.S. Department of
     Agriculture, Washington, D.C., report no. 41-64, 39 p.
Shampine, L. F., 1973, Concentration—dependent diffusion, 2, singular
     problems:  Quarterly Applied Mathematics, v. 31, p. 287-293.
Skopp, J., and Warrick, A. W., 1974, A two phase model for the miscible
     displacement of reactive solutes in soils:  Soil Science Society of
     America Journal, v. 38, p. 545-580.
Smith, R. E., and Parlange, J. Y., 1978, A parameter-efficient hydrologic
     infiltration model:  Water Resources Research, v. 14, p. 533-536.
Talsma, T., and Parlange, J. Y.,  1972, One dimensional infiltration:
     Australian Journal of Soil Research, v. 10, p. 143-150.
Thomas, A. W., and others, 1976,  Comparison of calculated and measured
     capillary potentials from line sources:  Soil Science Society of
     America Journal, v. 40, p. 10-14.
Thomas, A. W., Kruse, E. G., and  Duke, H. R., 1974, Steady infiltration
     from line sources buried in  soil:  American Society of Agricultural
     Engineers Transactions, v. 17, p. 115-128.
Turner, N. C., and Parlange, J. Y., 1974, Lateral movement at the periphery
     of a one dimensional flow of water:  Soil Science, v. 118, p. 70-77.
Wagner, C.,  1952, On  the solution to diffusion problems involving
     concentration-dependent diffusion coefficients:  Journal of
     Metallurgy Transactions, v.  4, p. 91-96.
Warrick, A.  W., 1974, Time dependent linerized  infiltration  I.  Point
     Source:   Soil Science Society  of America Journal, v.  38, p. 383-386.
Warrick, A.  W., and Lomen, D. 0., 1977, Time dependent  linearized
     infiltration III.   Strip and disc sources:   Soil  Science Society of
     America Journal, v. 40, p.  639-643.
	1977,  Flow from a  line  source above  a shallow water  table:   Soil
     Science Society  of  America  Journal, v.  41,  p.  849-852.
Wooding, R.  A.,  1968,  Steady  infiltration  from a shallow  circular  pond:
     Water  Resources  Research, v. 4, p.  1259-1273.
Zachmann, D. W., 1978, A mathematical  treatment of infiltration from a
     line source into an inclined porous  medium:   Soil Science  Society
     of  America Journal,  v.  42,  p.  685-688.
Zachmann, D. W., and Thomas,  W.  E., 1973,  A mathematical  investigation  of
     steady infiltration from line sources:  Soil Science  Society  of
     America Journal, v.  37,  p.  495-500.

-------
                                  124


                              REFERENCES

                  Numeric Solution  (Water Flow Only)

Afshar, A., and Marino, M. A., 1978, Model for simulating soil-water
     content considering evapotranspiration:  Journal of Hydrology,
     v. 37, May 1970, p. 314.
Ahuja, L. R., 1973, A numerical and  similarity analysis of infiltration
     into crusted soils:  Water Resources Research, v. 9, p. 987-994.
Ashcroft, G., and others, 1962, Numerical method for solving the diffusion
     equation, I, Horizontal flow in semi infinite media:  Soil Science
     Society of America Journal, v.  26, p. 522-525.
Beese, F., Van der Ploeg, R. R., and Richter, W., 1977, Test of a soil
     water model under field conditions:  Soil Science of America Journal,
     v. 41, p. 979-984.
Belmans, C., Feyen, J., and Hillel,  D., 1978, Comparison between simulated
     and measured moisture profiles  in a sandy soil during extraction by
     a growing root system in Modeling, identification, and control in
     environmental systems, (ed.) G. C. Vansteenkiste:  North Holland
     Publishing Co.
Bhuiyan, S. I., and others, 1971, Dynamic simulation at vertical
     infiltration into unsaturated  soils:  Water Resources Research,
     v. 7, p. 1597-1606.
Bouwer, H., and Little, W. C., 1959, A unifying numerical solution for two-
     dimensional steady flow problems in porous media with an electrical
     resistance network:  Soil Science Society of America Journal, v. 23,
     p. 91-96.
Braester, C., and others, 1971, A survey of the equations and solutions of
     unsaturated flow in porous media:  First annual report, project no.
     A10SWC77, Hydraulic Engineering Laboratory, Technion, Haifa, Israel,
     176 p.
Brandt, A., and others, 1971, Infiltration from a trickle source, I.
     Mathematical model:  Soil Science Society of America Journal, v. 35,
     p. 675-682.
Brooks, R. H., and others, 1971, Drainage of soil profiles:  American
     Society of Civil Engineers Proceedings, v. IR 3, p. 455-467.
Bruch, J. C., Jr., 1975, Finite element solutions for unsteady and
     unsaturated flow in porous media:  California Water Resources Center,
     University of California, Davis, California, contribution no. 151.
	1977, Two-dimensional unsteady water flow in unsaturated porous
     media, ^.n Finite elements in water resources  (eds.) Gray, Finder,
     and Brebia:  Pentech Press, p.  3.3-3.19.
Brutsaert, W. F., 1970, Immiscible  multiphase flow in ground water
     hydrology.  A computer analysis of the well flow problem:  PhD  thesis,
     Colorado State University, Fort Collins, Colorado, 119 p.
	1971, A functional iteration  technique for solving the Richards
     equation applied to two dimensional Infiltration problems:  Water
     Resources Research, v. 7, p. 1583-1516.
	1973, Numerical solution of multiphase well  flow:  American Society
     of Civil Engineers Transactions, v. HYll, p.  1981-2001.

-------
                                  125
Brutsaert, W. F., Breitenback, E. A., and Sunada, D. K., 1971, Computer
     analysis of free surface well flow:  American Society of Civil
     Engineers Journal, v. 97(IR1), p. 405-420.
Caussade, B. H., Dournes, G., and Renard, G., 1979, A new numerical
     solution of unsteady two dimensional flow in unsaturated porous
     media:  Soil Science, v. 127, p. 193-201.
Cheng, E. D. H., 1975, Numerical analysis of one-dimensional water
     infiltration:  University of Hawaii Water Resources Research Center
     Technical Report No. 92, 44 p.
Cooley, R. L., 1970, A finite difference method for analyzing liquid flow
     in variably saturated porous media:  Hydrologic Engineering Center
     Technical Paper 27, U.S. Army Corps of Engineers, Sacramento,
     California, 37 p.
	1971, A finite difference method for unsteady flow in variably
     saturated porous media.  Application to a single pumping well:  Water
     Resources Research, v.  7, p. 1607, 1625.
Cooley, R. L., and Dunohoe,  D. A. T., 1969, Numerical simulation of
     unconfined flow into a  single pumping water well using two phase
     flow theory:  Technical Completion Report, Institute for Res. on
     Land and Water Resources, Pennsylvania State University State
     College, Pennsylvania,  75 p.
Dabiri, H., 1970, Digital simulation  of unsaturated ground water flow
     considering evapotranspiration:  Doctor of Engineering Dissertion,
     Universtiy of Kansas, Lawrence,  Kansas.
Dane, J. H., and Wierenga, P. J., 1975, Effect of hystresis on the
     prediction of infiltration, redistribution, and drainage of water
     in a layered soil:  Journal of Hydrology, v. 25, p. 229-242.
Day, P. R., and Luthin, J. N., 1956,  A numerical solution of the
     differential equation of flow for a vertical drainage problem:  Soil
     Science Society of America Journal, v. 20, p. 443-447.
Doherty, P. C., 1972, Unsaturated darcian flow by the Galerkin method:
     U.S. Geological Survey  Computer  contribution, program no. C938.
Feddes, R. A., Bressler, E., and Neuman, S. P., 1974, Field test of a
     modified numerical model for water uptake by root systems:  Water
     Resources Research, v.  10, p. 1199-1206.
Feddes, R. A., Kowalik, P. J., and Zaradny, H., 1973, Simulation of field
     water use and crop yield:  Halstead Press, J. Wiley and Sons,
     New York, 188 p.
Feddes, R. A., Neuman, S. P., and Bressler, E., 1975, Finite element
     analysis  of two dimensional flow  in  soils considering water  uptake
     by roots, II Field applications:  Soil Science  Society of America
     Journal, v. 39, p. 231-237.
Feddes, R. A., and others, 1976,  Simulation of  field  water uptake  by
     plants using a  soil water dependent  root extraction function:
     Journal of Hydrology, v.  31, p.  13-26.
Feddes, R. A., and Zaradny,  H.,  1977, Numerical model for  transient water
     flow in nonhomogeneous  soil-root systems with ground water  influence:
     Proceedings  of  IFIP  Conference on Modeling and  Simulation  of  Land,
     Air, and Water  Resources, Ghent, Belgium.

-------
                                   126
Feddes, R. A., and Zaradny, H.,  1977, Numerical  model  for  transient
     water soil-root  systems with  ground water influence,  in Modeling,
     identification and control, environmental systems,  (ed.) G. C.
     Vansteenkiste:   North Holland Publishing Company.
Finlayson, B. A., 1974, A comparison of orthagonal  collocation  on  finite
     elements with finite difference methods for partially saturated  one-
     dimensional flow in porous  media:  Consulting  report  prepared for
     Computer Sciences Corporation, Northwest Operations,  Rlchland,
     Washington.
	1975a, MPHASE-A-A computer model simulating air-water movement  in
     dessicated soils:  Consulting report under  agreement  CCA-103  with
     Computer Sciences Corporation, Richland, Washington.
	1975b, Water movement in dessicated soils, in Finite elements in
     water resources  (eds.) Gray, Finder, and Brebia:  Pentech Press,
     p. 3.91-3.107.
Finlayson, B. A., Nelson, R. W., and Baca, R. G., 1978, A preliminary
     investigation into the theory and  techniques of modeling the natural
     moisture movement in unsaturated sediments:  Rockwell International,
     Inc., Rockwell Hanford Operations, Energy Systems Group, Richland,
     Washington, 144  p.
Finnemore, E. J., 1970, Moisture movement above  a varying water table:
     Department of Civil Engineering, Stanford University, Technical
     Report No. 132,  127 p.
Fletcher, J. E., and  El-Shafei, Y. Z.,  1970, Theoretical study of
     infiltration into range and forest soils:   Utah Water Research
     Laboratory, College of Agriculture, Report  PRWG60-1, 36 p.
Freeze, R. A., 1969,  The mechanism of natural groundwater recharge and
     discharge 1, One dimensional, vertical, unsteady, unsaturated flow
     above a recharging or discharging  groundwater flow system:  Water
     Resources Research, v. 5, p. 153-171.
	1971a, Influence of the unsaturated flow domain on seepage through
     earth dams:  Water Resources Research, v. 7, p. 929-941.
     _1971b, Three-dimensional, transient saturated-unsaturated flow in a
     groundwater basin:  Water Resources Research, v. 7, p. 347-366.
Freeze, R. A., 1972a, A physics-based approach to hydrologic response
     modeling.  Phase 1, Model development:  Completion Report for OWRR
     Contract No. 14-31-001-3694, U.S. Department of Interior, 119 p.
	1972b, The role of subsurface flow in generating surface runoff,
     1.  Baseflow contribution to channel flow:  Water Resources Research,
     v. 8, p. 609-623.
	1972c, Role of subsurface flow in generating surface runoff 2.
     Upstream source areas:  Water Resources Research, v. 8, p. 1272-1283.
Frind, E. 0., Gillham, R. W., and Pickens, J. F., 1977, Application of
     unsaturated flow properties in the design of geologic environments
     forradioactive waste storage facilities, in Finite elements in water
     resources (eds.) Brebia, Finder, Gray:  Pentech Press.
Frind, E. 0., and Verge, M. J., 1978, Three dimensional modeling of
     groundwater flow systems:  Water Resources  Research, v. 14, p. 844-856.
Geisel, W., Renger, M., and Strebel, 0., 1973, Numerical treatment of the
     unsaturated water flow equation, comparison of experimental and
     computed results:  Water Resources Research, v. 9, p. 174-177.

-------
                                  127
Gilham, R. W., Klute, A., and Heerman, P. F., 1979, Measurement numerical
     simulation of hysteretic flow in a heterogeneous porous medium:
     Soil Science Society of America Journal, v. 43, p. 1061-1067.
Green, D. W., Dabiri, H., and Khare, J. D., 1972, Numerical modeling of
     unsaturated groundwater flow including effects of evapotranspiration:
     Project completion report, Kansas Water Resources Research Institute,
     University of Kansas, 185 p.
Green, D. W., and others, 1970, Numerical modeling of unsaturated
     groundwater flow and comparison of the model to a field experiment:
     Water Resources Research, v. 6, p. 862-874.
Green, R. E., Hanks, R. J., and Larson, W. E., 1964, Estimates of field
     infiltrtion by numerical solution of the moisture flow equation:
     Soil Science Society of America Journal, v. 28, p. 15-19.
Green, W. H., and Ampt, G. A., 1911, Studies in soil physics I.  The flow
     of air  and water through soils:  Journal of Agricultural Science,
     v. 4, p. 1-24.
Gupta, S. K., and others, 1978, Field simulation of soil water movement
     with crop water extraction:  Water Science and Engineering Paper
     No. 4013, University of California, Department of Air, Land, and
     Water Resources, 121 p.
Hanks, R. J., and Bowers, S. A., 1962, Numerical solution of the moisture
     flow equation for infiltration into layered soils:  Soil Science
     Society of America Journal, v. 26, p. 520-534.
Hanks, R. J., and Klute, A., 1968, A numerical method for estimating
     infiltration, redistribution, drainage and evaporation of water from
     soil:   American Society of Agricultural Engineers, Annual Meeting,
     Paper No. 68-714, Logan, Utah, 16 p.
Hanks, R. J., Klute, A., and Bresler, E., 1969, A numeric method for
     estimating infiltration, redistribution, drainage and evaporation  of
     water from soil:  Water Resources Research, v. 5, p. 1064-1069.
Haverkamp, R., and others, 1977, A comparison of numerical simulation
     models  for one-dimensional filtration:  Soil Science Society of
     America Journal, v. 41, p. 785-794.
Haverkamp, R., and Vauclin, M., 1979, A note on estimating finite
     difference interblock hydraulic conductivity values for  transient
     unsaturated flow problems:  Water Resources Research, v. 15,
     p. 181-187.
Hayhoe, H. N., 1978, Numerical study of  quasi-analytic and finite
     difference solutions of the soil water  transfer  equation:   Soil
     Science, v. 175, p. 68-74.
Hillel, D.,  1977, Simulation of  soil water dynamics,  a compendium of recent
     work:   International Development Research  Centre, Ottawa,  Canada,
     v. 214.
Hillel, D.,  and Talpaz, H., 1976,  Simulation of root  growth and its effects
     on patterns of  soil water uptake by nonuniform root systems:   Soil
     Science, v. 121, p. 307-312.
Hoa, N. T.,  Gaydu, R.,  and Thirriot,  C.,  1977,  Influence of the hystresis
     effect  on transient  flows  in  unsaturated porous  media:   Water
     Resources Research, v.  13,  p.  992-996.

-------
                                  128
Hornberger, G. M., Fungaroli, A. A., and Remson, I., 1966, A computer
     program for the numerical analysis of soil moisture systems:
     Drexel Institute of Technology, College of Engineering, Science
     Series No. 1, 14 p.
Hornberger, G. M., and Remson, I., 1970, A moving boundary model of a
     one-dimensional saturated-unsaturated transient porous flow system:
     Water Resources Research, v. 6, p. 898-905.
Hornberger, G. M., Remson, I., and Fugaroli, A. A., 1969, Numeric studies
     of a composite soil moisture groundwater system:  Water Resources
     Research, v. 5, p. 797-802.
Hornung, U., 1977, A numerical method for the simulation of unsteady
     ground-water flow in both saturated and unsaturated soils:  Soil
     Science, v. 124, p. 140-141.
Hromadka, T. V., II, and Guymon, G. L., 1980, Some effects of
     linearizing the unsaturated soil moisture transfer diffusivity model:
     Water Resources Research, v. 16, p. 643-650.
Ibrahim, H. I., and Brutsaert, W. F., 1963, Intermittent infiltration
     into soils with hystresis:  Journal of Hydraulics Division,
     American Society of Civil Engineers, v. HY94, p. 113-137.
Jensen, M. E., and Hanks, R. J., 1967, Nonsteady state drainage from porous
     media:  American Society of Civil Engineers, Irrigation and Drainage
     Division, v. 93, p. 209-231.
Jeppsen, R. W., 1968a, Solution to axisymmetric seepage from ponds through
     homogeneous and nonhomogeneous media:  Utah Water Research Laboratory,
     Utah State University, Logan, Utah, Report No. WE 52-2, 69 p.
	1968b, Seepage from channels through porous mediums:  Water Resources
     Research, v. 4, p. 435-445.
	1968c, Seepage through dams in the complex potential plane: American
     Society of Civil Engineers, Irrigation and Drainage Division, v. 94,
     p. 23-29.
	1968d, Seepage from ditches, solution by finite differences: American
     Society of Civil Engineers, Hydraulics Division, v. 94, p. 759-783.
      1970a» Transient flow of water from infiltrometers—formulation of
     mathematical model and preliminary numerical solutions and analyses of
     results:  Utah Water Research Laboratory Report PRWG59C2.
	1970b, Solution to transient vertical moisture movement based upon
     saturation-capillary pressure data and a modified Burdine theory:
     Utah Water Research Laboratory, Utah State University, Logan, Utah,
     Report PRWG-59c-5, 46 p.
	1970c, Formulation and solution of transient  flow of water from an
     Tnfiltrometer using the Kirchoff transformation:  Utah Water Research
     Laboratory, Utah State University, Logan, Utah, Report No. WG 59-C-3.
	1972, Limitations of some finite difference methods in solving the
     strongly nonlinear equation of unsaturated  flow in soils:  Utah Water
     Research Laboratory, Utah State University, Report PDWG 59c-8.
Jeppson, R. W.,  1974a, Axisymmetric infiltration into soils, I, Numerical
     techniques  of  solution:  Journal of Hydrology, v. 23, p.  111-130.
	1974b, Axisymmetric Infiltration  into  soils,  II, Summary of
     infiltration characteristics related  to problem specifications:
     Journal of  Hydrology, v. 23, p.  191-202.

-------
                                  129
Jeppson, R. W., and Nelson, W., 1970, Inverse  formulation and finite
     difference solution to partially saturated seepage from canals:
     Soil Science Society of America Journal,  v. 34, p. 9-14.
Jeppson, R. W., and others, 1975, Use of axisymmetric infiltration model
     and field data to determine hydraulic properties of soils:  Water
     Resources Research, v. 11, p. 127-138.
Kafri, V., and Ben-Asher, J.,  1978, Computer estimates of recharge through
     soils in  southern Arizona, USA:  Journal  of Hydrology, v. 38,
     p. 125-138.
Kastanck, F.,  1971, Numerical  simulation technique for vertical drainage
     from a soil column:  Journal of Hydrology, v. 14, p. 213-232.
Khare, J. D.,  1971, Test and application of a  computer model describing
     unsaturated groundwater flow in the presence of evapotranspiration:
     M.S. Thesis, University of Kansas, 1971.
Klute, A., 1952, A numerical method for solving the flow equations for
     water in unsaturated materials:  Soil Science, v. 73, p. 105-116.
Klute, A., and Heerman, D. F., 1974, Soil water profile development
     under a periodic boundary condition:  Soil Science, v. 117,
      p. 265-271.
Krishnamurthi, N., 1975, Simulation of gravitational water movement in
     soil:  PhD Dissertation,  Department of Civil Engineering, Colorado
     State University, Fort Collins, Colorado.
Krishnamurthi, N., Sunada, D.  K., and Longenbaugh, R. A., 1978, On
     oscillation of numerical  solution of a modified Richard's equation:
     Water Resources Research, v. 14, p. 52-54.
Kroszynski, U., 1975, Flow in  a vertical porous column drained at its
     bottom at constant flux:  Journal of Hydrology, v. 74, p. 135-153.
Lees, S. J., and Watson, K. K., 1975, The use  of dependent domain models
     of hystresis in numerical soil water studies:  Water Resources
     Research, v. 11, p. 943-948.
Liakopoulos, A. C., 1964a, Theoretical aspects of the flow of water through
     anisotropic unsaturated soils:  Bulletin  of the International
     Association of Scientific Hydrology, v. 9, p. 67-70.
	1964b, Transient flow through unsaturated porous media:  Doctor
     of Engineering thesis, University of California, Berkeley, California.
	1965, Theoretical solutions of the unsteady unsaturated flow
     problems  in soils:  Bulletin of the International Association of
     Scientific Hydrology, v.  10, p. 5-39.
      1966, Theoretical prediction of evaporation losses from groundwater:
     Water Resources Research, v. 2, p.  227-240.
Lin, S. H., 1978, Transient water infiltration with  constant  surface  flux:
     Journal of Hydrology, v.  37, p. 387-392.
Longenbaugh, R. A., 1975, Computer  estimates of natural  recharge  from soil
     moisture data, High Plains of  Colorado:   Colorado State  University
     Environmental Resources Center Completion Report Series  No.  64.
Luthin, J. W., Orhun, A., and  Taylor, G.  S., 1975, Coupled  saturated-
     unsaturated transient flow in  porous media;  Experimental and numeric
     model:  Water Resources Research, v.  11,  p.  975-980.
Mein, R. G., and Larson, C. L., 1973, Modeling infiltration during a
     steady rain:  Water Resources  Research, v. 9, p. 384-394.

-------
                                  130
Mercer, J. W., and Faust, C. R.,  1977,  The  application  of  finite  element
      techniques  to imiscible flow in  porous media,  in Finite elements  in
      water resources  (eds.) Gray,  Finder, and  Brehia:   Pentech Press,
      p. 1.21-1.50.
Molz, F. J.,  and others, 1968,  Soil moisture availability  for
      transpiration: Water Resources Research,  v.  4, p.  1161-1169.
Molz, F. J.,  and Remson, I., 1979, Extraction-term  models  of soil moisture
      use by transpiring plants:   Water  Resources  Research, v. 6,
      p. 1346-1356.
Narasimhan, T. N., 1975, A unified numerical model  for  saturated-
     unsaturated groundwater flows:   PhD dissertation,  Department of
     Civil Engineering, University of California, Berkeley, California.
Narasimhan, T. N., Neuman, S. P.,  and Witherspoon,  P. A.,  1978, Finite
     element method for subsurface hydrology using  a mixed explicit-
      implicit scheme:  Water Resources  Research,  v. 14, p. 813-817.
Narasimhan, T. N., and Witherspoon, P.  A.,  1977,  Numerical model  for
     saturated-unsaturated flow in deformable  porous media, 1, Theory:
     Water Resources Research,  v.  13, p. 657-664.
	1978, Numerical model for saturated-unsaturated flow in deformable
     porous media, 3, Applications:  Water  Resources Research, v. 14,
     p. 1017-1034.
Narasimhan, T. N., Witherspoon, P. A.,  and  Edwards, D.  L., 1978,  Numerical
     model for saturated-unsaturated  flow in deformable porous media,
     2, The algorithum:  Water  Resources Research,  v. 14, p. 255-261.
Natur, F. S., King, L. G., and  Jeppson, R.  W., 1975, Unsaturated-saturated
     flow through heterogeneous sloping lands:  Utah Water Research
     Laboratory, College of Engineering, Utah  State University, Logan,
     Utah.
Nelson, R. W., 1962, Steady Darcian transport  of  fluids in heterogeneous,
     partially saturated porous media,  Part 1, Mathematical and numerical
     formulation:  Hanford Atomic  Products  Operation, Atomic Energy
     Commission  Research Development Report HW 72335, part 1.
Neuman, S. P., 1972, Finite element computer programs for flow in
     saturated-unsaturated porous  media:  Volcani Institute of Agriculture,
     Israel Institute of Technology, Haifa, Israel.
	1973, Saturated-unsaturated  seepage by finite elements:  Journal of
     the Hydraulics Division, American  Society of Civil Engineers,
     v. Hy 12, p. 2233-2250.
      1974, Galerkin approach to  unsaturated flow in soils, in Finite
     element methods in flow problems:  University of Alabama Press,
     Huntsville, Alabama, p. 517-521.
     __1975a, Galerkin method of analyzing unsteady flow in saturated-
     unsaturated porous media, in Chapter 10, Finite elements in fluids,
     1, Ceds.) R. W. Gallagher, J. J. Oden, C. Taylor, and 0. C.
     Zienkiewicz:  John Wiley, London, England.
     _1975b, Galerkin method of simulating water uptake in plants, in
     Modeling and simulation of water resources systems (ed.) G. C.
     Vansteenklste:  North Holland Publishing Co.

-------
                                  131
Neuman, S. P., Feddes, R. A., and Bresler, E., 1974, Development of
     methods tools, and solutions for unsaturated flow—third annual
     report, part 1:  Hydrodynamics and Hydraulic Engineering Laboratory,
     Technion, Haifa, Israel.
	1975, Finite element analysis of two dimensional flow in soils
     considering water uptake by roots, I, Theory:  Soil Science Society
     of America Journal, v. 59, p. 224-230.
Neuman, S. P., Narasimhan, T. N., and Witherspoon, P. A., 1977, Application
     of mixed explicit-implicit finite element method to nonlinear
     diffusion type problems, jLn Finite elements in water resources  (eds.)
     Gray, Finder, and Brebia:  Pentech Press.
Ng., B. L., 1968, A predictor-corrector method for solving the diffusion
     equation for water movement in unsaturated soil:  M.S. thesis, Oregon
     State University, Corvallis, Oregon.
Nimah, M. N., 1972, Model for estimating  soil water flow, water content,
     evapotranspiration and root extraction:  Unpublished PhD dissertation,
     Utah State University, Logan, Utah,  131 p.
Nimah, M. N*, and Hanks, R. J., 1973a, Model  for estimating soil water,
     plant and atmosphere interrelations,  1, Description and sensitivity:
     Soil Science Society of America Journal, v. 37, p. 522-527.
	1973b, Model for estimating soil water, plant and atmospheric
     Interrelations,  2, Field test of model:  Soil Science Society of
     America Journal, v. 37, p. 528-532.
Perrens, S. J., and Watson, K. K., 1977,  Numerical analysis of two
     dimensional infiltration and redistribution:  Water Resources
     Research, v. 13, p. 781-790.
Pikul, M. F., Street, R. L., and Remson,  I.,  1974, A numerical model based
     on coupled one-dimensional Richard's and Boussinesq equations:  Water
     Resources Research, v. 10, p. 295-302.
Reeves, M., and Duguid, J.  0., 1975, Water movement through saturated-
     unsaturated porous media; A finite element Galerkin model:   Oak
     Ridge National Laboratory, Oak Ridge, Tennessee.
Reeves, M., and Miller, E.  E., 1975, Estimating  infiltration  from erratic
     rainfall:  Water Resources Research,  v.  11, p. 102-110.
Reisenauer, A. E., 1963, Methods for solving problems  of partially
     saturated steady flow  in soils:   Journal of Geophysical  Research,
     v. 68, p. 5725-5733.
Reisenauer, A. E., Nelson,  R. W., and  Knudsen,  C.  N.,  1963,  Steady Darcian
     transport of  fluids  in heterogeneous partially  saturated porous
     media, Part  2, The computer program: Atomic  Energy  Commission,
     Research and Development Report No.  HW72335,  part 2,  Hanford Atomic
     Products Operation, Richland, Washington.
Remson, I., and  others,  1965, Vertical drainage of an unsaturated soil:
     Journal of  the Hydraulics Division,  American Society of Civil
     Engineers,  v.  91, Proceedings Paper  No. 4196, p.  55-74.
Remson, J., Fungaroli, A. A., and Hornberger, G.  M.,  1967, Numerical
     analysisof  soil  moisture  systems:   Journal of the Irrigation and
     Drainage Division,  American Society  of  Civil Engineers,  v.  93,
     p.  153-166.
Rubin,  J.,  1966a,  Numerical analysis of ponded  rainfall infiltration,  in
     Water in the unsaturated  zone:   Proceedings of the Wageningen
      Symposium,  UNESCO,  Paris,  France.

-------
                                  132


Rubin, J., 1966b, Theory of rainfall uptake by soils initially drier
     than their field capacity and its applications:  Water Resources
     Research, v. 2, p. 739-749.
	1967, Numerical method for analyzing hystresis-affected, post
     infiltration redistribution of soil moisture:  Soil Society of
     Americo Journal, v. 31, p. 13-20.
	1968, Theoretical analysis of two dimensional transient flow of water
     in unsaturated and partly unsaturated soils:  Soil Science Society of
     America Journal, v. 32, p. 607-615.
Rubin, J., and Steinhardt, R., 1963, Soil water relations during rain
     infiltration, I, Theory:  Soil Science Society of America Journal,
     v. 27, p. 240-251.
Rubin, J., Steinhardt, R., and Reinger, P., 1964, Soil water relations
     during rain infiltration, II, Moisture content profiles during rains
     of low intensities:  Soil Science Society of America Journal,
     v. 28, p. 1-5.
Sawhnen, B. L., and Parlange, J. Y.., 1974, Two dimensional water
     infiltration  from a trench in unsaturated soils:  Soil Science
     Society of America Journal, v. 38, p. 867-871.
Selim, H. M.,  and  Kirkham, D., 1973, Unsteady two dimensional flow of
     water in  unsaturated soils above an impervious barrier:  Soil
     Science Society of America Journal, v. 37, p. 489-495.
Staple, W. J., 1969, Comparison of computed and measured moisture
     redistribution following infiltration:  Soil Science Society
     of America Journal v. 33, p. 840-847.
Stephenson, G. R., and Freeze, R. A., 1974, Mathematical simulation of
     subsurface flow contributions to snowmelt runoff, Reynolds Creek
     Watershed, Idaho:  Water Resources Research, v. 10, p. 284-294.
Taylor, G. S., and Luthin, J. W., 1969, Computer methods for transient
     analysis  of water table aquifers:  Water Resources Research,
     v. 5, p.  144-152.
Todsen, M., 1973,  Numerical studies of two dimensional saturated-
     unsaturated drainage models:  Journal of Hydrology, v. 20,
     p. 311-330.
Van der Ploeg, R.  R., and Benecke, P., 1974, Unsteady, unsaturated-
     n-dimensional moisture flow in soil; A computer simulation program:
     Soil Science  Society of America Journal, v. 38, p. 881-885.
Vauclin, M., and others, 1977, Note:  A comparison of exact and numerical
     solutions of  the diffusion equation near singularities:  Soil
     Science v. 124, p. 181.
Vauclin, M., Vachand, G., and Khanji, J., 1975, Two dimensional
     numerical analysis of  transient water transfer in saturated-
     unsaturated soils, in Computer simulation of water resources
     systems  (ed.) G. C. Vansteenkiste:  North Holland Publishing Co.,
     Amsterdam.
Verma, R. D.,  1969, Physical analysis of  the  outflow  from an unconfined
     aquifer:  Thesis presented  to Cornell University at Ithaca, New York.
Verma, R. D.,  Brutsaert, W., 1970, Unconfined aquifer seepage by capillary
     flow theory:  Journal  of  the Hydraulics  Division, American  Society
     of Civil  Engineers, v. 96, p.  1331-1344.

-------
                                  133
Wang, F. C., 1963, An approach to the solution of unsteady, unsaturated
     flow problems in soils:  Department of Civil Engineering, Stanford
     University, Stanford, California, Technical Report 19, 134 p.
Wang, F. C., Hassan, N. A., and Franzini, J. B., 1964, A method of
     analyzing unsteady unsaturated flow in soils:  Journal of Geophysical
     Research, v. 69, p. 7569-7579.
Wang, F. C., and Lakshminarayana, V., 1968, Mathematical simulation of
     water movement  through unsaturated nonhomogeneous soils:  Soil
     Science Society of America Journal, v. 37, p. 139-334.
Watson, K. K., 1967, Experimental and numerical study of column drainage:
     Journal of the Hydraulics Division, American Society of Civil
     Engineers, v. 93, p. 1-15.
Watson, K. K., and Whisler, F. 0., 1978, The use of dynamic soil water
     characteristics in a numerical desorption model:  Soil Science,
     v. 175, p. 83-91.
Watson, K. K., Whisler, F. D., and Curtis, A. A., 1978, Numerical analysis
     of one dimensional decay of a perched water table:  Journal of
     Hydrology, v. 36, p. 109-119.
Wei, C. Y., and Jeppson, R. W., 1971, Finite difference solutions of
     axisymmetric infiltration through partially saturated porous medium:
     Utah Water Research Laboratory, Report PRG59C6, 62 p.
Whisler, F. D., and Bouwer, H., 1970, Comparison of methods for calculating
     vertical drainage and infiltration for soils:  Journal of.Hydrology,
     v. 10, p. 1-19.
Whisler, G. F., and Klute, A., 1966, Analysis of infiltration  into
     stratified soil columns, jin (eds.) P. E. Rijtema and H. Wassink,
     Water in the unsaturated zone:  Proceedings of Wageningen
     Symposiums: UNESCO, Paris, France.
Whistler, F. D., Klute, A., and Millington, R. J., 1968, Analysis of steady
     state evaporation from a soil column:  Soil Science Society of America
     Journal, v. 32, p. 167-174.
Whisler, F. D., Watson, K. K., 1969, Analysis of infiltration  into draining
     porous media:  American Society of Civil Engineers, v. 95, paper 6946,
     p. 481-491.
Wind, G. P., and Van Doorne, W., 1975, A numerical model for the simulation
     of unsaturated vertical flow of moisture in soils:  Journal of
     Hydrology, v. 24, p. 1-20.
Yoon, Y. S., and Yeh, W. W. G., 1975, The Galerkin method for  nonlinear
     parabolic equations of unsteady groundwater flow:  Water  Resources
     Research, v. 11, p. 751-754.
Zaradny, H., 1978, Boundary conditions in modeling water flow  in
     unsaturated soils:  Soil Science, v. 125, p. 75-82.
Zyvoloski, G., 1973, Finite element weighted residual solution of one
     dimensional unsteady and unsaturated flows in porous media:  M.S.
     Thesis, Department of Mechanical Engineering, University  of California
     Santa Barbara, California.
Zyvoloski, G., and Bruch, J. C., 1973, Finite element weighted residual
     solution of one dimensional, unsteady and unsaturated flows in
     porous media:  University of California Report UCSB-ME-73-74,
     Santa Barbara, California.

-------
                                  134
Zyvoloski, G., Bruch, J. C., Jr., and Sloss, J. M., 1975, Finite element
     solution of the two dimensional unsaturated flow equation:
     University of California Technical Report UCSB-ME-75-2, 100 p.,
     Santa Barbara, California.
	1976, Solution of the equation for two-dimensional infiltration
     problems:  Soil Science, v. 122, p. 65-70.

                              REFERENCES

                  Numeric Solutions (Water and Solute Flow)

Beese, F., and Wierenga, P. J., 1980, Solute transport through soil with
     adsorption and root water uptake computed with a transient and a
     constant flux model:  Soil Science, v. 129, p. 245-252.
Bender, S. J., Finder, G. F., and Gray, W. G., 1975, A comparison of
     numerical approximations to the one dimensional convective-
     diffusive equation:  Princeton University Department of Civil
     Engineering Report, 52 p.
Bresler, E., 1973a, Simultaneous transport of solutes and water under
     transient unsaturated flow conditions:  Water Resources Research,
     v. 9, p. 975-986.
	1973b, Anion exclusion and coupling effects  in nonsteady transport
     through unsaturated soils—Theory:  Soil Science Society of America
     Journal, v. 37, p. 663-669.
	1975, Two dimensional transport of solutes during nonsteady
     infiltration from a trickle source:  Soil Science Society of America
     Journal, v. 39, p. 604-613.
      1978, Theoretical modeling of mixed electrolyte solution flows for
     unsaturated soils:  Soil Science, v. 125, p. 196-203.
Bresler, E., and Hanks, R. J., 1969, Numerical method of estimating
     simultaneous flow of water and salts in unsaturated soils:  Soil
     Science Society of America Journal, v. 33, p. 827-832.
Bresler, E., and Russo, D., 1975, Two dimensional solute transfer during
     nonsteady infiltration; Laboratory test of tasthematical model:
     Soil Science Society of America Journal, v. 39, p. 585-587.
Childs, S. W., and Hanks, R. J., 1976, Model of soil salinity effects on
     crop growth:  Soil Science Society of America Journal, v. 40,
     P- 	•
Davidson, J. M., Baker, C. R., and Brusewitz, G. H., 1975, Simultaneous
     transport of water and adsorbed solutes through soil under transient
     flow conditions:  Transactions, American Society of Agricultural
     Engineers, v. 18, p. 535-539.
Davidson, J. M., and others, 1975, Use of soil parameters for describing
     pesticide movement through soils:  Environmental Protection
     Technology Series, CPA 660/2-75-009.
Davidson, J. M., and others, 1978, Simulation of nitrogen movement,
     transformation and uptake in plant root zone:  Environmental
     Protection Agency, Ecological Research Series 600/3-78-079, 105 p.
Duguid, J. O., and Reeves, M., 1976, A comparison of mass transport using
     average and transient rainfall boundary conditions in Finite elements
     in water resources (eds.) Gray, Pinder, and Brebia:  Pentech Press,
     p. 2.25-2.36.

-------
                                  135
Dutt, G. R., Shafer, M. J., and Moore, W. J., 1972, Computer simulation
     model of dynamic bio-physio-chemical processes in soils:  University
     of Arizona, Agricultural Experiment Station, Technical Bulletin 196,
     101 p.
Frankel, H., Goertzen, J. 0., and Rhodes, J. D., 1978, Effects of clay type
     and content, exchangeable sodium presentage and electrolyte
     concentration  on clay dispersion and soil hydraulic conductivity:
     Soil Science Society of America Proceedings, v. 47, p. 32-37.
Gureghian, A. B., Ward, D. S., and Cleary, R. W., 1979, Simultaneous
     transport of water and reacting solutes through multilayered soils
     under transient unsaturated flow conditions:  Journal of Hydrology,
     v. 41, p. 253-278.
Hillel, D., 1977, Simulation of soil water dynamics; A compendium of recent
     work:  International Development Research Centre, Ottawa, 214 p.
Hillel, D., Talpaz, H., and von Keylen, H., 1976, A macrosocopic-scale
     model of water uptake by a nonuniform root  system and of water and
     salt movement in the soil profile:  Soil Science, v. 121, p. 242-255.
Hillel, D., van Beek, G. E. M., and Talpaz, H.,  1975, A microscopic-scale
     model of soil water uptake and salt movement to plant roots:  Soil
     Science, v. 120, p. 385-399.
Khaleel, R., and Reddell, D. L., 1976, Simulation of pollutent movement  in
     ground water aquifers:  Texas A and M University, College Station,
     Texas, report 81, 248 p.
King, L. G., and Hanks, R. J., 1973, Irrigation management for control of
     quality of irrigation return flow; Prepared for USEPA:  Report No.
     EPA RZ-73-765, 307 p.
Marino, M. A., 1978, Solute transport in saturated-unsaturated porous
     medium in Modeling, identification and control in environmental
     systems (ed.) G. C. Vansteenkiste:  North Holland Publishing Co.
Pickins, J. F., Gillham, R. W., and Cameron, D.  R., 1979, Finite element
     analysis of the transport of water and solutes in tile drained soils:
     Journal of Hydrology, v. 40, p. 743-764.
Pupiski, H., and Shainberg, J., 1979, Salt effects on the hydraulic
     conductivity of a sandy soil:  Soil Science Society of America
     Journal, v. 43, p. 429-433.
Reeves, M., and Duguid, J. 0., 1976, Material transport through porous
     media; A finite element Galerkin model:  Oak Ridge National Laboratory,
     Oak Ridge, Tennessee,;v!98 p.
Segol, G., 1976, A three-dimensional Galerkin finite element model  for  the
     analysis of contaminant transport in variably  saturated porous media,
     users guide:  Department of Earth Sciences, Waterloo University,
     Ontario, Canada.
Selim, H. M., 1978, Transport of reactive solutes during transient,
     unsaturated water flow in multilayered  soils:  Soil Science, v.  126,
     p. 127-135.
Selim, H. M., Davidson, J. M., and Rao, P.  S. C.,  1977, Transport of
     reactive solutes  through multi-layered  soils:  Soil Science  Society
     of America Journal, v. 41, p. 3-10.

-------
                                  136


Sidle, R. C., Kardos, L. T., and Th. vanGenuchten, M., 1977, Heavy metals
     transport model in a sludge-treated soil:  Journal of Environmental
     Quality, v. 6, p. 	.
Stuff, R. G., and Dale, D. F., 1978, A soil moisture model accounting for
     shallow water table influences:  Soil Science Society of America
     Journal, v. 42, p. 637-643.
Watts, D. G., 1975, A soil-water-nitrogen-plant model for irrigated corn
     on course textured soils:  PhD dissertation, Utah State University,
     197 p.
Wierenga, P. J., 1977, Solute distribution profiles computed with steady-
     state and transient water movement models:  Soil Science Society of
     America Journal, v. 41, p. 1050-1055.
Wierenga, P. J., and others, 1975, Predicting ionic distributions in a
     large soil column:  Soil Science Society of America Journal, v. 59,
     p. 1080-1084.

                              REFERENCES

                Numeric Solutions (Water and Heat Flow)

Baca, R. J., and King, I. P., 1978, Finite element models for simultaneous
     heat and moisture transport in unsaturated soils:  Rockwell
     International, Hanford Operations, RHO-SA-31.
Finlayson, B. A., Nelson, R. W., and Baca, R. G., 1978, A preliminary
     Investigation into the theory and techniques of modeling the nautral
     moisture movement in unsaturated sediments:  Rockwell International,
     Inc., Rockwell Hanford Operations, Energy Systems Group, Richland,
     Washington, 144 p.
Guymon, G. L., and Berg, R. L., 1976, Galerkin finite element analog of
     coupled moisture and heat transport in algid soils, in Finite elements
     in water resources feds.) Gray, Pinder, Brebia:  Pentech Press,
     p. 3.107-3.114.
Guymon, G. L., and Luthin, J. N., 1974, A coupled heat and moisture
     transport model for artic soils:  Water Resources Research, v. 10,
     p. 995-1001.
Hammel, J. E., 1979, Modeling tillage effects on evaporation and seed zone
     water content during fallow in eastern Washington:  PhD Thesis,
     Washington State University, Pullman, Washington, 65 p.
Harlan, R. L., 1973, Analysis of coupled heat and fluid transport in
     partially frozen soil:  Water Resources Research, v. 9, p. 1321-1323.
Hillel, D., 1977, Simulation of soil water dynamics; A compendium of recent
     work:  Interntional Development Research Centre, Ottawa, Canada, 714 p.

                              REFERENCES

                           Parametric Models

Afonda, A. A., 1978, A unified approach to watershed modeling:  Nordic
     Hydrology, v. 9, p. 161-172.
Crow, F. R., and others, 1977, Evaluating components of the USDAHL
     Hydrology Model applied to grass land watersheds:  American Society
     of Agricultural Engineers Transcript, v. 20, p. 692.

-------
                                  137


Davidson, J. M., and others, 1978, Simulation of nitrogen movement,
     transformation and uptake in plant root zone:  U.S. Environmental
     Protection Agency Reprint, EPA 60013-78-029, 105 p.
Dutt, G. R., Shaffer, M. J., and Moore, W. R., 1972, Computer simulation
     model of dynamic bio-physiochemical processes in soils:  University
     of Arizona, Department of Soil, Water, and Engineering, Technical
     Bulletin 196, 101 p.
Eagleson, P. S., 1978a, Climate, soil and vegetation, 1, Introduction to
     water balance dynamics:  Water Resources Research, v. 14, p. 705-712.
	1978b, Climate, soil and vegetation, 2, The distribution of annual
     precipitation derived from observed storm sequences:  Water Resources
     Research, v. 14, p. 713-721.
	1978c, Climate, soil and vegetation, 3, A simplified model of soil
     moisture movement in the liquid phase:  Water Resources Research,
     v. 14, p.  727-730.
      1978d, Climate, soil and vegetation, 4, The expected value of annual
     evapotranspiration:  Water Resources Research, v. 14, p. 731-739.
      1978e, Climate, soil and vegetation, 5, A derived distribution of
     storm surface runoff:  Water Resources Research, v. 14, p. 741-743.
Eagleson, P. S., 1978f, Climate, soil and vegetation, 6, Dynamics of the
     annual water balance:  Water Resources Research, v. 14, p. 749-764.
	1978g, Climate,  soil and vegetation, 7, A derived distribution of
     annual water yield:  Water Resources Research, v. 14, p.  765-776.
Fitzpatrick, E. A., and Nix, H. A., 1969, A model for simulating soil water
     regime in alternating follow-crop systems:  Agricultural  Meteorology,
     v.  6, p. 303-319.
King, L. G., and Hanks, R. J., 1973, Irrigation management for control of
     quality of irrigation return,flow:  U.S. Environmental Protection
     Agency Report. EPAR-7-73-765, 307 p.
Neibling, W. H., 1976, Estimating groundwater recharge from conservation
     bench terraces:   M.S. Thesis, Department of Agricultural  Engineering,
     Kansas State University, Manhattan, Kansas.
Neibling, W. H., Koelliker, J. H., and Ohmes, F. E.,  1977, A continuous
     water budget model for western Kansas:  Presented at annual 1977
     meeting, American Society of Agricultural Engineers, Paper No.  77-7056.
Reisenaur, A. E., and  others, 1975, Partially  saturated  transient
     groundwater flow  model theory and numerical implementation:   Battelle
     Pacific Northwest Laboratories, Richland, Washington,  31  p.
Skaggs,  R. W., 1978, A water management model  for  shallow water table
     soils:  Water Resources Research Institute, University of North
     Carolina, Report  No. 134, 178 p.
Smith, R. E., 1972, The infiltration envelope; Results  from a  theoretical
     infiltrometer:  Journal  of Hydrology, v.  17,  p.  1-21.

-------
             VERIFICATION OF PARTIALLY SATURATED ZONE MOISTURE

                   MIGRATION: MODEL AND FIELD FACILITIES
                               S.  J.  Phillips

                                T. L. Jones
                          Rockwell International
                       Rockwell  Hanford  Operations
                           Energy Systems Group
                       Rich!and,  Washington   99352


                       Battelle  Memorial  Institute
                       Pacific Northwest Laboratory
                       Richland,  Washington   99352


                                 ABSTRACT

     Two large field-experimental facilities, designed to evaluate
moisture migration under partially saturated geohydrologic conditions,
have been constructed on the Hanford Site.  The function of these
facilities is to monitor the rate and magnitude of moisture migration.
Models have been used to simulate and predict one dimensional liquid and
vapor phase water migration.  Parametric data from these facilities are
being used for verification of the UNSAT and MPHASE codes.  To date,
these models have shown only limited success in the simulation of:
volumetric water content, water storage, water drainage, liquid flux and
vapor flux.  Continued studies will enhance the capability to simulate,
predict and verify moisture migration and radionuclide transport under
field conditions at the Hanford Site.
                               139

-------
                                 140
                                INTRODUCTION

     The Hanford Site located in south central Washington state was
established in December, 1942.  The Site was constructed for production
of nuclear defense materials.  For nearly four decades, nuclear fuels
fabrication plants, reactors, chemical separations and processing plants,
laboratories, and waste storage and waste disposal facilities have been
operated.  These facilities have produced substantial quantities of low-
level radioactive waste materials that have been stored or permanently
inturned, i.e., disposed within the partially saturated geohydrologic
system at the site.  Nuclear waste materials from other defense and
commercial facilities have also been sent to the Hanford Site for storage
or disposal.
     Low-level wastes have been released within the partially saturated
geologic media of the site by several disposal methods, e.g., shallow
land burial.  These methods rely on the geochemical sorption of long-
lived radionuclides and the attenuation of  liquid- and vapor-phase flux
of radionuclides in arid geohydrological environments.  Theoretical
investigations of radionuclide and moisture transport in partially
saturated systems have suggested a potential for loss of containment
and migration of radionuclides away from disposal facilities.  This
may result in wastes entering the biosphere.  As a result, numerous
laboratory and field studies have been conducted to evaluate the rate,
magnitude, and consequences of radionuclide and moisture migration at
the Hanford Site.
     Following is a generalized review of specific numerical modeling
efforts and field experimental facilities that have been used in part,
for the validation of these models.  The purpose of this report is to
show the initial results and complexities of modeling and experimental
studies to understand and verify migration  of liquid- and vapor-phase
materials in the geohydrologic system, relative to low-level waste
disposal.  The results of this report cannot be considered conclusive
relative to validation of the models.  Results are given only to
enumerate progress toward this goal.

-------
                                141
     Much of the information reported here has been taken directly from
published reports originating from the Hanford Site.

                           METHODS AND MATERIALS

SHALLOW CAISSON, LYSIMETER, MICROMETEOROLOGICAL FACILITY
     A Shallow Caisson, Lysimeter, Micrometeorological experimental facil-
ity was constructed on the Hanford Site  in 1978. U-»2»3J  This facility
was expressly designed for:   (1)  development  and calibration of monitoring
instruments and  (2) development of methods to evaluate mass and energy
balance factors  that control  radionuclide transport in and around
shallow, partially saturated, geohydrologic systems.  In addition, the
Shallow Facility was designed to  function analogous to a one-dimensional,
shallow land, waste burial structure.  Experimentation and monitoring
activities were  also conducted that  were amenable  to model validation,
i.e., one-dimensional, nonisothermal, diphasic,  partially saturated water
and radionuclide transport.
     The Shallow Facility, Figure 1,  shows micrometeorological and hydro-
geological monitoring instrumentation as it exists at present.
     The Shallow Facility consists of three buried caissons 8.1 m by
2.7 m, and four  buried caissons 7.5  m by 0.6  m.  These caissons are
attached to form a seven-caisson  array.  The  caissons are  interconnected
with steel tubing to provide  access  to the outer caissons from the cen-
tral caisson.  These tubes are installed at 15-cm  vertical  intervals
between the central caisson  and large caissons  and between the central
caisson and small caissons at 30  cm  intervals.   The tubes provide access
for instruments  and destructive sampling of radiocontaminants  and other
materials.
     Weighing lysimeters to  determine mass balance are  installed below
grade and attached to the central caisson.  Uniform sand was backfilled
into the six peripheral caissons  of  the  array between 8.1 m  and 6.3 m
deep.  A layer of coarse quartz sand was then introduced  into  the cais-
sons between 6.0 m and 6.3 m.  This  layer provided for evaluation of

-------
                                                                                          I-
                                                                                          ro
     FIGURE 1.   Oblique View Schematic of Shallow Facility Showing

Caissons, Lysimeters,  Micrometeorologic Instrument Sensors,  etc.

-------
                                 143
layering effects.  The remaining 6.0 m of the caisson were then back-
filled with uniform sand.  The four small caissons were excavated to a
depth of 60.5 cm and a 1.0 cm layer of either cobalt-60 (60Co), -
ethylenediaminetetracetic acid (EDTA) or tritium  (3H) tagged backfill
material was emplaced in alternating small caissons.  The concentrations
of respective radionuclides that were designed to function as  liquid and
vapor phase migration tracers were 6.4 yCi Kg   and 6.37 m Ci  Kg"1
at volumetric moisture contents equal to that of  the larger caissons at
the 60.0 cm depth.  Excavated backfill material was then emplaced over
the tagged layer.
     During and subsequent to backfilling operations, active and passive
monitoring access tubes, sensors and transducers  were installed for:
(1) neutron thermalization and sodium-iodide well logging, (2) geohydro-
logical thermal and matric potential measurement, and (3) micrometero-
logical thermal relative humidity, wind, precipitation, solar  radiation,
evaporation, etc.  Passive monitoring procedures  were developed and
active data collected for computer processing.
     Moisture migration and radionuclide transport occurs as a function
of moisture content induced by changes  in precipitation.  In order to
evaluate this function one-half of the facility field,  in bilateral
symmetrical configuration, was irrigated at twice the ambient  precipita-
tion rate at nearly ambient frequency.

MODIFIED UNSAT MODEL APPLICATION
     Soon after construction of the Shallow Facility, modeling activities
were completed to produce an initial prediction of water migration.  A
modified one-dimensional, isothermal, partially saturated, finite dif-
ference model was used'-4'5-'.  Equations for total water flux were
developed to account for thermal and matric potential gradients.  Using  a
-1.0 cm cm"1 water potential gradient and equations for thermally
induced water flux, a nonisothermal water flux  in response to  thermal
gradients was calculated for various water contents  in  geologic media.

-------
                                  144
This suggested that simulation of  isothermal  conditions  could  adequately
describe the flux of liquid water  within the  geologic media.
     The following model simulating conditions  in  the caisson  facility
were used:   (1) an 8.0 m profile of uniform sand with a  coarse quartz
sand layer,  (2) a static water table at 8.0 m depth, (3) measured
retentivity  and saturated hydraulic conductivity values  for the  uniform
and coarse quartz sand, (4) partially saturated hydraulic conductivity
values as determined by the Millington and Quirk method"- , (5)  the
vapor flux from the surface was taken as 0.001 cm  day  , (6) measured
atmospheric temperature, wind, solar radiation, cloud cover, and precip-
itation, calculated potential and  actual evaporation and (7) standard
initial conditions of depth, suction head, volumetric water content,
initial storage, final storage, infiltration, evaporation, and drainage.
A model sensitivity analysis was also conducted to evaluate such factors
as hydraulic conductivity, total precipitation, precipitation  distribu-
tion, and layering.

DEEP CAISSON/DEEP WELL FACILITY
     In 1971 a Deep Caisson/Deep Well Facility was designed to evaluate
the migration of naturally occurring moisture in the partially saturated
zone'-     .  This Deep Caisson Facility has served as a  one-dimensional
structure amenable to partially saturated zone model simulation  and
verification.  The Caisson Facility, Figure 2, shows geohydrological
monitoring instrumentation.  Near  the Deep Caisson Facility is a deep
instrumented well designed to monitor matric  potential and temperature in
geologic media between the surface and the water table.  This
instrumented well is also illustrated in Figure 2.
     The Deep Caisson Facility consists of two thermally insulated cais-
sons 3.0 m in diameter by 17.0 m in length buried  below  grade.  Centrally
located between these caissons is  an instrument access caisson.  Neutron
thermalization access tubes and matric potential sensors were  installed
vertically through the two outer caissons.  One caisson was sealed at its
lowest extremity to form a liquid  barrier,while the other remained open

-------
                                     145
                                                                 RCP8012 120
     FIGURE 2.  Oblique View Schematic of Deep  Caisson/Deep Well  Facility
Showing Caissons and Geohydrological Instrument Sensors,  etc.

-------
                                 146
so that direct communication between the ambient geohydrologic system
would result.  Uniform sand excavated during caisson construction  and
emplacement was backfilled into the two caissons.
     A deep instrumented well was constructed to a depth of 97.0 m.  Ma-
tric potential and thermal sensors were affixed to a cable and lowered
into the well.  The well was then backfilled with geologic media removed
during construction.

MPHASE MODEL APPLICATION
     In 1975 a computer code, MPHASE, was developed to simulate water
migration through the Deep Caisson Facility.  A one-dimensional, steady
state, nonisothermal, liquid and vapor phase, partially saturated, finite
element model formulation was used.'- ^  Equations were developed  and
modified for determination of saturation, total water flux and liquid and
vapor phase fluxes.  These equations accounted for thermal and diphasic
water potential gradients.  Dimensionless equations were formulated for:
(1) water balance, (2) dry air balance, (3) energy balance, (4) liquid
mass flux, (5) wet air mass flux and (6) dry air mass flux.  Depth and
time were also made dimensionless.  A variable pressure, nonispthermal
gradient for liquid and vapor phase flux was determined for all depths
from the ground surface to the water table.
     The MPHASE model used the following conditions:  (1) 97° m deep pro-
file of homogeneous geologic media (2) static water table, (3) measured
liquid and vapor hydraulic conductivities,  (4) measured temperature,
(5) measured capillary pressure and (6) fixed and/or time varying boundary
conditions.

                           RESULTS AND DISCUSSION

MODIFIED UNSAT MODEL SIMULATION RESULTS
     Sensitivity analyses of geologic media sand dependence on hydraulic
conductivity to moisture migration at the shallow facility were performed

-------
                                  147

using ambient, 0.1 and 0.01 factors.  Simulations demonstrated that:
(1) water penetration into the geologic media was reduced by reducing the
hydraulic conductivity by 0.1 or  0.01, (2) water storage and seepage
below 2.0 m was not significantly affected,  (3) a reduction in hydraulic
conductivity reduced the water flow into the Shallow Facility so that it
tended to remain closer to the ground surface and (4) a reduction in
hydraulic conductivity resulted in reduced evaporation.  Table 1 shows
the results of varying hydraulic  conductivity.

                  TABLE 1.  Simulation Results  of Varying
                     Hydraulic Conductivity Values.a>&
Variable
Ambient
Ambient/ 10
Ambient/ 100
Infiltration
(cm)
24.1
24.5
24.5
Evaporation
(cm)
26.3
23.3
22.7
Drainage
(cm)
-0.4
-0.2
-0.1
Storage
(cm)
42.6
45.7
46.2
          Modified after Gee and Simmons,  1979
           Initial storage - 44.6 cm

      Model simulations indicate that by varying the hydraulic conductiv-
ity there can be a pronounced effect on water  content  and flux near the
ground surface, indicating these effects are reduced with depth.  In
addition, modification of hydraulic conductivity can,  over time, influ-
ence the migration of water and nonsorbed radionuclides within geohydro-
logic systems.
     The effects of annual precipitation in the form of rainfall were also
evaluated by sensitivity simulations.  The model,  as anticipated, predic-
ted that an increase in precipitation would increase drainage to the water
table.  The model results indicated:  (1) changes  of precipitation from
one year to the next are reflected in the migration of moisture through
and from the geologic media, (2) a significant decrease in annual precipi-
tation below mean annual precipitation will propably result in negative

-------
                                 148
drainage of moisture to the water table whereas an increase will result
in positive drainage, (3) moisture stored in the profile will not be dras-
tically affected. The effect of total precipitation over six years using
a one standard precipitation year value are shown in Table 2.

            TABLE 2.  Simulation of Total Precipitation Effects on
                  Six Consecutive Years Based  on Multiples  of
                     One  Standard Precipitation Year.a»b,c
Year
1
2
3
4
5
6
Infiltration
(cm)
49.1 (2.0 x)
36.8 (1.5 x)
8.1 (0.33 x)
24.5 (1.0 x)
36.8 (1.5 x)
18.5 (0.75 x)
Evaporation
(cm)
34.9
31.5
13.1
23.2
30.9
21.1
Drainage
(cm)
7.5
6.8
0.6
-0.3
1.9
1.0
Storage
(cm)
51.0
49.3
43.9
45.5
49.3
45.6
            Modified after Gee and Simmons, 1979
            "Initial storage - 44.6 cm
            Potential evaporation - 131.0 cm

     Results of this simulation demonstrate that moisture migration is
significantly affected by changes in total annual precipitation.  Fur-
thermore, annual precipitation history additionally affects moisture
migration.
     The sensitivity of the temporal distribution of precipitation on an
annual basis was simulated in addition to the total annual precipitation.
Two years with different temporal precipitation were selected, i.e., one
year with precipitation occurring predominantly in the fall, year a, and
the other with predominantly spring and winter precipitation, year b, as
in Table 3.  Model simulation results have indicated that by alternating
years a and b variable increases or decreases of moisture evaporation,
drainage and storage result depending on the alternation sequence.  After
using developed standard initial conditions and sequentially repeating

-------
                                 149
year a, moisture, evaporation, drainage and storage all  increase.   Model
simulation results for alternative and repetitive years  are shown  in
Table 3 and 4 respectively.

                     TABLE 3.  Simulation of a Three-Year
                         Precipitation Sequence.a,b,c
Year
a
b
c
Infiltration
(cm)
23.5
24.6
24.0
Evaporation
(cm)
20.1
25.0
24.6
Drainage
(cm)
-0.1
0.7
0.2
Storage
(cm)
48.0
46.7
46.4
          Modified after Gee and Simmons
          "Initial storage - 44.6 cm
          Potential evaporation - 147.7 cm, year a; 131.2 cm, year b
                     TABLE 4.  Simulation of a Three-Year
                        Repetitive Precipitation.a,b,c
Year
a
a
a
Infiltration
(cm)
23.5
23.5
23.5
Evaporation
(cm)
20.1
22.0
22.1
Drainage
(cm)
-0.1
1.2
1.3
Storage
(cm)
48.0
48.2
48.2
          Modified after Gee and Simmons
          blnitial storage - 44.6 cm
          Potential evaporation - 147.7 cm, year  a; 131.2 cm, year b

     The  simulation indicates that the  temporal  distribution of precipi-
tation  significantly affects the hydrogeologic  system.  Furthermore, it
illustrates the  complexity of predictive modeling  of systems with
variable  upper boundary  layer conditions.
     Sensitivity simulations were also  completed to evaluate the effect
of  layering,  relative to partially saturated moisture migration.  Hetero-
geneity of geologic media in texturally alternating layers was found to

-------
                                  150
have a pronounced effect on moisture flux and drainage.  Layering reduces
the flux of liquid phase moisture across lateral layer boundaries during
drainage, until the water saturation in the upper layer reaches some
critical valve wherein the water will freely drain into the next lower
vertical layer.
     Model results indicate:  (1) the interface between course and fine
textured geologic media at a depth of 6.0 m significantly perturbates the
moisture content profile, (2) drainage is reduced where a coarse layer
exists in the profile while drainage is increased when this layer is
omitted from the profile (3) storage of moisture in the profile is
increased where the coarse layer is present; conversely, storage is
reduced with omission of the coarse layer.
     Simulation results of layering cases are shown in Table 5 for a non-
homogeneous coarse geologic media layer between 6.0 m and 6.5 m in depth,
(a) and a homogeneous profile (b).

               TABLE  5.  Simulations  of Layering Effects.a»b»c,d
Water
Table
(a)
(b)
Infiltration
(cm)
23.5
23.5
Evaporation
(cm)
20.1
20.4
Drainage
(cm)
-0.01
0.3
Storage
Initial
44.6
50.0
(cm)
Final
48.0
52.7
      Modified after Gee and Simmons, 1979
       Potential evaporation - 147.7 cm
      cCoarse sand layer 600-650 cm
      dSand layer

      Results of this sensitivity simulation demonstrate some of the
effects of layering on the flux of moisture through, and the retention
of moisture within, geologic media.  These results may also provide
insight into the utility of design of engineered barriers relative to
the disposal of nuclear materials.

-------
                                 151

     The UNSAT model simulations, using the configuration and data from
the Shallow Facility, were used to provide an initial understanding and
prediction of moisture migration and nonsorbed radionuclide transport in
partially saturated geologic media.
     Field data collected at the Shallow Facility over the past two years
have shown:  (1) the gravity potential dominates the flux of moisture
through geologic media under both ambient and accelerated precipitation
conditions, (2) low distribution coefficient 60Co-EDTA has not been
transported under ambient or accelerated precipitation conditions as
                o
calculated, (3)  H has been transported both in the vapor and liquid
                                                                  3
phase within the partially saturated profile of the facility, (4)  H
has been transported in the liquid phase (data that suggests  H has
also been transported in the vapor phase) within the partially saturated
profile of the facility, (5) moisture in the liquid phase has accumulated
in the bottom of the caissons receiving ambient plus two times ambient
precipitation, where no accumulation of moisture has been found in the
ambient precipitation caissons.
     These simulation results are not intended to be conclusive.  The
results were derived using actual and assumed data from the Shallow
Facility and adjacent experimental sites soon after construction.  The
purpose of modeling this system was to provide an approximation of mois-
ture migration and nonsorbed radionuclide transport at the Shallow
Facility.  These data could then be used to develop monitoring systems.
The model results also increased our  understanding of the mechanism of
flow in geologic media under various  conditions.

MPHASE MODEL SIMULATION RESULTS
     Steady state and transient  simulations of liquid and vapor phase
moisture flux were performed for both the open and closed bottom  caissons
of the Deep Caisson Facility.  Model  solutions for the open  bottom cais-
son during steady state  conditions yielded  applicable results while solu-
tions for the closed bottom  caisson were found to require further model

-------
                                  152

development.  Steady state simulation results indicated:  (1) near the
ground surface moisture migration is predominantly in the vapor phase,
(2) liquid phase moisture flux in the caisson is upward but is extremely
small, (3) vapor phase migration throughout the profile is upward at a
nearly constant rate, liquid phase flux below the caisson is downward at
a rate lower than that of the vapor phase, and (5) the total flux is
upward.
     Table 6 shows dimensionless values for liquid and vapor fluxes
versus saturation at various depths.  The caissons extend to a dimension-
less depth of 0.16.
          TABLE 6.  Vapor and Liquid Phase Flux vs. Saturation at
                Various  Depths  in  Deep Open  Bottom Caisson
                         (Dimensionless Values).a»b
Depth
0.0000
0.1154
0.3846
0.5000
0.6154
0.8846
1.0000
Saturation
0.0600
0.0600
0.0600
0.0600
0.0600
0.0611
1.0000
Liquid Flux
-1.843 (-12)
-8.670 (-12)
2.019 (-10)
1.186 (-9)
2.997 (-9)
1.091 (-8)
-1.307 (-7)
Vapor Flux
-1.307 (-7)
-1.307 (-7)
-1.309 (-7)
-1.319 (-7)
-1.314 (-7)
-1.393 (-7)
0.000
Total Flux
-1.307 (-7)
-1.307 (-7)
-1.307 (-7)
-1.307 (-7)
-1.284 (-7)
-i;282 (-7)
-1.307 (-7)
        Modified after Finlayson, 1975
        bDimensionless flux x 5.5 x 10"4 = g sec -1

    The total flux was found to be quite dependent on the pressure gra-
dient that determines the direction of liquid mass flux.  This relates
whether the pressure gradient is greater or less than the gravitational
force acting on the liquid.
     Continued monitoring of variables controlling moisture migration at
the Deep Caisson Facility could yield results amenable to model calibra-
tion and verification.

-------
                                 153

                                CONCLUSIONS

     Field monitoring and model simulations of moisture migration and
radionuclide transport and controlling factors are exceedingly complex.
This is indicated by the variability of model simulations and model
sensitivity simulations reported herein.  Presently only limited simu-
lation and prediction estimates of migration and transport can be made.
     Further field, laboratory and model development efforts will enhance
the capability to simulate, predict, error bound and verify moisture
migration and radionuclide transport under field conditions at the
Hanford Site.

-------
                                  154


                                REFERENCES
 1.   S.  J.  Phillips,  L.  L. Armes, R. E. Fitzner, G. W. Gee, G. A.
     Sandness,  C.  S.  Simmons, Characterization  of the Hanford 300 Area
     Burial Grounds:  Final  Report Decontamination and Decomissloning.
     PNL-2557  (January,  1980).

 2.   S.  0.  Phillips,  A.  E. Reisenauer, W. H. Richard, and G. A. Sandness,
     Initial Site  Characterization and Evaluation of Radionuclide
     Contaminated  Solid  Waste Burial Grounds, BNWL-2184 (February. 1977).

 3.   S.  J.  Phillips,  A.  C. Campbell, M. 0. Campbell, G. W. Gee, H. H.
     Hooper, and K. 0. Schwarzmiller, A Test Facility for Monitoring
     Water/Radionuclide  Transport Through Partially Saturated Geologic
     Media: Design,  Construction, and Preliminary Description, PNL-3226
     and PNL-3227  (November  1979).

 4.   G.  W.  Gee and C. S. Simmons, Characterization of the Hanford 300
     Area Burial Grounds:  Fluid Transport and  Modeling, PNL-2921
     (August,  1979).

 5.   S.  K.  Gupta,  K.  K.  Tanji,  D. R. Nielsen, J. W. Biggar, C. S. Simmons
     and J. L.  Maclntyre, Field Simulation of Soil-Water Movement with
     Crop Water Extraction,  Water Science and Engineering Papers No. 4013.
     University of California  at Davis (January, 1978).

 6   A.  E.  Reisner, Calculations of Soil Hydraulic Conductivity From Soil
     Water  Retention  Relationships, BNWL-1710.  1973.

 7.   T.  L.  Jones,  Sediment Moisture Relations:  Lysimeter Project
     1976-1977 Water  Year, RHO-ST-15  (June, 1978).

 8.   J.  J.  C.  Hsieh,  A.  E. Reisenauer, and L. E. Brownell, A Study of
     Soil Water Potential  and Temperature in Hanford Soils. BNWL-1712
     (1973).

 9.   R.  E.  Isaacson,  L.  E. Brownell, R. W. Nelson and E. L. Roetman, Soil
     Moisture  Transport  in Arid Site Vadose Zones. ARH-SA-169, 1974.

10.   J.  J.  C.  Hsieh,  L.  E. Brownell  and D. E. Reisenauer, Lysimeter
     Experiments.  Description  and Progress Report on Neutron
     Measurements. BNWL-1711,  1973.

11.   B.  A.  Finlayson, MPHASE -  A Computer Model Simulating Air-Water
     Movement  in Desiccated  Soils. RHO-SD-31. 1978.

-------
    RADIONUCLIDE TRANSPORT UNDER CHANGING SOIL SALINITY CONDITIONS*
                              T.  L.  Jones
                               G. W.. Gee
                     Pacific Northwest Laboratory
                      Richland, Washington  99352

                                  and

                            P. J. Wierenga
                      New Mexico State University
                     Las Cruces,  New Mexico  88001
                                SUMMARY

     This paper presents experimental  measurements and computer simula-
tions of strontium-85 movement through laboratory soil columns.  There
are two primary objectives of the work.  The first is to examine the
enhanced mobility of strontium-85 when the soil salinity is raised.   The
second is to compare the ability of two solute transport models to
describe this phenomena.  The experiment consisted of injecting a pulse
of what will be termed a "low salt" solution spiked with strontium-85
into a soil column.  The column was then leached with over eighteen  pore
volumes of the low salt solution resulting in no leaching of the
strontium-85 from the column.  Subsequently, the column was leached  with
a "high salt" solution, causing the rapid release of the strontium-85
within five to six pore volumes.

     These data were analyzed using models based on two sets of equations.
The first used the traditional convective dispersive equation.  This
equation contains the three parameters; water velocity V, dispersion
coefficient D, and the distribution coefficient Kd.  The first two may
be combined to form the dimensionless Peclet number (P)' and the Kd  is
often expressed as a dimensionless retardation factor (R).  This equation
in dimensionless form is a two parameter solute transport equation.   The
second model uses the mobile-immobile water equation described by Van
Genuchten and Wierenga (1976a).  This model uses a set of two equations
which contain six parameters which may be combined into four dimensionless
parameters.

     These models involve a large number of parameters that must be
evaluated.  The techniques used in this work used the curve fitting
routine described by van Genuchten (1980), as well as some batch Kd
measurements described by Gee and Campbell (1980).  The curve fitting
was done on breakthrough curves obtained from separate column Teachings
with strontium and tritium using the high salt solution.  The parameter
     *Prepared for the U.S; Department of Energy under contract DE-A-
C06-76RLO-1830.

                                 155

-------
                                  156
estimates were then used in the low salt-high salt experiment described
above.  This follows the reasoning outlined by van Genuchten and Wierenga
(1977a, 1977b).

     The results demonstrate that strontium-85 is subject to greater
mobility when the soil salinity is increased.  This confirms the neces-
sity of evaluating the soil salinity very carefully in modeling efforts.
The results also indicate that batch Kd values, combined with column Kd
values and parameters obtained by curve fitting procedures do a fairly
good job in describing the "snowplow effect."  The comparison of the two
models reveal insignificant differences when describing the movement of
tritium and strontium-85 under high salinity conditions.  There was a
significant improvement, however, when the mobile-immobile water equations
were used to describe the combined low salt-high salt experiment.


                             INTRODUCTION

     The characterization of sorption reactions that remove radionuclides
from solution in soils and sediments is a very active area of research.
Among the diverse processes being studied are oxide and carbonate pre-
cipitation, precipitation of hydroxide compounds, coprecipitation with
amorphous hydrous oxides, lattice replacement, and ion exchange.  In
recent years the relative emphasis placed on ion exchange has decreased,
however, it still is an important process for certain radionuclides.  A
fundamental property of ion exchange reactions is that they are dependent
on the solution concentration of the exchanging ion as well as the ionic
strength and composition of the bulk solution.  The general trend of
strontium-85 observed by Gee and Campbell (1980) was for the sorption
(i.e., Kd value) to increase with increasing ionic strength of the bulk
solution.  This would indicate that ion exchange plays some role in the
sorption of strontium-85.

     The implication for shallow land burial sites is that errors can be
made in predicting the movement of some radionuclides through sandy soils
when only average salt concentrations are used in predictive equations
rather than actual time dependent salt concentration.  Burial ground
soils with relatively high concentrations of salt may, under increased
rainfall, flush sorbed radionuclides from the soil exchange sites creat-
ing a much higher effluent concentration than predicted.  This flushing
action by a high salt concentration influent solution has been called the
"snowplow effect" (Starr and Parlange 1979).  Included in this report is
verification of the "snowplow effect" for strontium-85.  Conventional
solute transport equations and those that include the mobile and immobile
water concept (Van Genuchten and Wierenga 1976b), are used to analyze
the data.

Materials and Soil

     The soil used in this experiment was a coarse sand found at the
Hanford site.  This soil is classified as a typic torripsament.  The
clay content is <4%, the cation exchange capacity is approximately
5 meq/100 g and the soil pH is 8.2.

-------
                                157
     The solutions used in the tests were two mixed salt solutions
designated as low (L) and high (H) salt.  The two solutions had the
following compositions:

     Low salt (L)  -0.15M NaN03, 0.01M KN03> 0.002M Ca (N03)2

     High salt (H) -0.15M NaN03, 0.15M KN03> 0.01M Ca (N03)2

     The experimental equipment used to demonstrate the snow plow effect
was designed after that used by Van Genuchten and Wierenga (1977a,b).
This is the same design described in Gee and Campbell  (1980).  Influent
solutions were applied at the upper end of soil  columns packed in plexi-
glass cylinders of 5.4 cm inside diameter.  Columns 27.5 cm long of
Rupert sand were packed to a dry bulk density of 1.73  g/cm3.  At the
lower end of the vertically positioned column, a porous plate for con-
trolling vacuum and supporting the soil was attached.   The column was
connected to a vacuum chamber enclosing an automatic fraction collector.
Figure 1 shows the apparatus used for measuring radionuclide transport
in unsaturated soil.

     The low salt solution containing a 50 nci/m£ spike of strontium-85
was pumped through the column.  After 4.3 pore volumes were injected,
the strontium-85 was removed from the influent solution.  The column was
maintained at a moisture content of 0.17 cm3/cm3, and  the flow rate was
6.8 cm/day.  After 18.25 pore volumes of effluent were collected, the
relative concentration of strontium-85 (C/C ) was <0.01.  The high salt
solution containing no strontium-85 was then added to  the column at a
flow rate of 7.0 cm/day.  The effluent was collected and the break-
through curve determined.

     The results of the low-salt, high-salt experiment are shown in
Figure 4.  The data show quite clearly the rapid release of strontium-85
after the addition of the high salt solution at 18.25  pore volumes. The
rapid rise in effluent concentration and the C/C0 value >1 are manifes-
tations of this showplow effect.  The two curves shown represent the
curves predicted by the two and four parameter models.  The parameters
used in the simulations are discussed in the next section.  Figure 4
shows a distinct improvement in describing the snowplow effect by using
the four parameter model.  Neither model predicts the  fast rise of the
curve or the C/C0 > 1.  The four parameter model does  however do a good
job of describing the tailing.
                         COMPUTER SIMULATION

     The computer models used for this simulation are described in
detail by Van Genuchten and Wierenga (1974, 1976b) and Van Genuchten
(1980).  They were written using the IBM system 360 Continuous Systems
Modeling Program (S/360 CSMP), simulation language.  The first model
used the convective dispersive equation:

-------
             NEGATIVE PRESSURE
                  SOURCE
  FLUID
MANOMETER'
DIFFERENTIAL
PRESSURE
REGULATOR     PRESSURE
 C-,          REGULATOR
              REGULATED
              NEGATIVE '
              PRESSURE
                                                    POSITIVE
                                                    PRESSURE
                                                     SOURCE
         PARTIALLY
         SATURATED
         SEDIMENT
         COLUMNS
                                FRACTION
                                COLLECTOR
                                                                                                               on
                                                                                                               c»
                                                                    TAGGED SOLUTION
STEPF
SYRING
C

-X.

ULSE
EPUMP


O
                                                T
                                             VACUUM
                                             CHAMBER
     Fig.  1.   Apparatus for measuring radionuclide movement in unsaturated  soil  columns.

-------
                                159
 o
fc>
O
     0.70
     0.60
     0.50
     0.40
     0.30
     0.20
     0.10  -
STRONTIUM -85

HIGH SALT
                              O   DATA


                              	2 PARAMETER
                                       	,	4 PARAMETER
                                                       I
        3.00      6.00      9.00     12.00     15.00     18.00


                            PORE VOLUMES
       Fig. 2.   High salt  strontium  breakthrough curve.

-------
                                   160
          o
         O
              1.00
              0.80  -
              0.60  -
              0.40  -
              0.20  -
                   0        1.00     2.00      3.00     4.00      5.00
                                   PORE VOLUMES
            Fig. 3.   High salt tritium  breakthrough  curve.
                                                  o    DATA
                                               	4 PARAMETER
                                                       2 PARAMETER
          17   18    19    20   21    22   23    24    25    26    27    28    29
                                    PORE VOLUME

Fig. 4.   Comparison of two and  four parameter  fit for snowplow effect.

-------
                                 161
                                       ax
where
     C = concentration of solute
     t = time
     x = vertical distance from surface
     D = dispersion coefficient
     v = pore water velocity
     P = bulk density
     0 = volumetric water content
    Kd = radionuclide distribution  coefficient
in dimensionless form this becomes
                          D aC   1 a C   aC
                                                                      (2)
                                   m+  d-f )p« ^ - «m D 1^ - vra «m ^
and
                                TT ' «  (Cm - Ci»,'                   (8)

-------
                                 162


where

     e  = mobile water fraction

      f = fraction of exchange sites in contact with mobile water

C , C.  = solute concentration in mobile and immobile water phase

     V  = mobile water velocity
      m

      a = mass transfer coefficient

in dimension!ess form these equations become:
                                                                       (9)
                      (1-B) R — = a (Cm - C.J                      (1Q)




                                                                      (ID
                                                                      (12)
                                 * + 
-------
                                 163


of the bulk solution and also the radionuclide solution.  As the soil
salinity increased from low to high as the high salt solution moved
through the column, the Kd value was assumed to vary linearly from
6 to 0.6.

     The parameter values for the mobile-immobile water equations were
obtained in a similar way.  An analytical solution (Van Genuchten and
Wierenga 1976b), of Equations 9 and 10 was curve fit to the data in
Figures 2 and 3.  The CSMP model used Equations 7 and 8 so a conversion
using equations 3, 4, 11, and 13 was done.  The only conversion which is
not straightforward is the use of Equation 13.  The value of B and R from
the tritium experiment was used first.  Assuming 0 * f * 1, we calculate
a range of 0 of 0.84 to 0.95.  Assuming 0 is the same for tritium and
strontium, we then use B and R from the high salt strontium data.  Assum-
ing 0.84 < 0 * 0.95, we obtain a range for f of 0.69 to 0.7.  The simula-
tion of the snowplow experiment used 0 = 0.9 and f = 0.7.
                              CONCLUSIONS

     Radionuclide transport under changing soil salinity conditions was
investigated by using two solute transport equations to describe the
transport of strontium-85 through laboratory columns.  The two parameter
convective dispersive equation was compared to the four parameter mobile-
immobile water equations of Van Genuchten and Wierenga (1976a).

     Both models were relatively successful in describing the rapid flush
of strontium-85 from a column with a high salt solution, a phenomena
called the snow plow effect.  This effect was simulated by using a Con-
tinuous System Modeling Program (CSMP) simulation model to solve the two
equations.  The four parameter mobile-immobile water model predicted the
release of the strontium more accurately, but neither model predicted
the peak effluent concentration well.  Both models confirm enhanced
mobility of strontium-85 with increased salt concentration of leaching
waters.
        Table 1.  Parameter Estimates for Computer Simulations
                          R     3     a     D     a     Kd    0m
Tritium            106   1.1   0.9   0.4    -     -     -     0.9   0.5

Strontium-85
     2 parameter    19   7.2    -          58     -    0.6
     4 parameter    29   8.4   0.7   0.5   41   0.12   0.74   0.9   0.7

-------
                                 164


                              REFERENCES
Gee, G. W., and A. C. Campbell.  1980.  Monitoring and Physical Charac-
     terization of Unsaturated Zone Transport-Laboratory Analysis.   PNL-
     3304.  Pacific Northwest Laboratory Richland, Washington.

Starr, J. L., and J. Y. Parlange.  1979.  "Dispersion in Soil  Columns:
     The Snowplow Effect."  Soil Soi. Soo. Am.  J.  43: 448-450.

Van Genuchten, M. Th., and P. J. Wierenga.  1974.   "Simulation of One-
     Dimensional Solute Transfer in Porous Media."  Agricultural Experi-
     ment Station Bulletin 628.  New Mexico State University,  Las
     Cruces, New Mexico.

Van Genuchten, M. Th., and P. J. Wierenga.  1976a.  "Mass Transfer Studies
     in Sorbing Porous Media.  I.  Analytical Solutions."  Soil Soi.  Soo.
     Am. J. 40: 473-480.

Van Genuchten, M. Th., and P. J. Wierenga.  1976b.  "Numerical Solution
     for Convective Dispersion with Intra-Aggregate Diffusion and Non-
     Linear Adsorption."  In System Simulation in Water Resources
     (G. C. Vansteenkiste, ed.).  North Holland Publishing Company,
     Oxford, New York.

Van Genuchten, M. Th., and P. J. Wierenga.  1977a.  "Mass Transfer Studies
     in Sorbing Porous Media.  II.  Experimental Evaluation with Tritium
     (3H20)."  Soil Soi. Soo. Am. J. 41: 272-278.

Van Genuchten, M. Th., and P. J. Wierenga.  1977b.  "Mass Transfer Studies
     in Sorbing Porous Media.  III.  Experimental  Evaluation with 2, 4,
     5-T."  Soil Sci. Soo. Am. J. 41: 278-285.

Van Genuchten, M. Th.  1980.  "Determining Transport Parameters from
     Solute Displacement Experiments.  U.S. Salinity Laboratory Research
     Report No. 118.  USDA/SEA, Riverside, California.

-------
       ON  THE  COMPUTATION OF THE VELOCITY FIELD AND MASS BALANCE
           IN THE FINITE-ELEMENT MODELING OF GROUNDWATER  FLOW
                             Gour-Tsyh Yeh

                     Environmental Sciences Division
                     Oak Ridge National Laboratory
                      Oak Ridge, Tennessee  37830
                                ABSTRACT
          Darcian  velocity has been conventionally calculated in
     the finite-element modeling  of  groundwater flow by taking the
     derivatives  of the computed pressure  field.   This results in
     discontinuities  in  the velocity  field at  nodal  points  and
     element boundaries.  Discontinuities  become enormous when the
     computed pressure  field  is far  from  a  linear  distribution.
     It is proposed  in this paper that the finite element proce-
     dure that  is  used to  simulate  the  pressure  field  or  the
     moisture content  field  also be applied  to Darcy's  law with
     the derivatives of the computed  pressure field  as  the load
     function.   The  problem  of discontinuity  is  then  eliminated,
     and the error of  mass balance  over  the region of  interest is
     much reduced.   The reduction  is from 23.8  to  2.?%  by  one
     numerical  scheme  and  from  29.7  to   -3.6%  by  another  for a
     transient problem.


                            INTRODUCTION


     To study  the transport of dissolved  constituents in a subsurface
flow  system, the velocity  field   therein must  be  determined first.
Several  finite-element Galerkin  models  have   been  developed  to obtain
the  flow  field [1-8].   The  continuity equation of  water-mass  governing
the  distribution  of   pressure   head  was  solved  by  the   Galerkin
finite-element method,  subject to appropriate  boundary  and  initial  con-
ditions.  The  flow field was computed with Darcy's  law  by taking  the
derivatives  of the calculated  pressure field  [3-6].   Inherent in  that
approach, however, was the  resulting discontinuity  in the velocity  at
nodal points and  element boundaries, which unfortunately  lead  to a  vio-
lation of the conservation of mass  in  a  local  sense.   Results  from two
hypothetical sample  problems  showed that  the  discontinuity in the  ve-
locity  field obtained   by  the  conventional  approach  ranges  from  very
small  to several  hundred  percent  depending  on  the  location  in  the
region.   Furthermore,  the overall mass balance is  not preserved.   When
spatial  distribution  of  the  velocity  is  significant,  inputting  this
discontinuous flow field to the contaminant transport computation  could
conceivably produce large error.

                                  165

-------
                                   166
      It  is proposed  to solve  Darcy's law  for the  velocity  field  at
nodal  points  by the  same finite-element  method  used for  the pressure
field  rather  than  simply  to  take the  derivatives  of  the  approximate
pressure   field.   This  approach  is  consistent  with  the  spirit  of
finite-element  modeling of  groundwater  flow.   It also  yields a  con-
tinuous  velocity field  over  the  region  of  interest  including  nodal
points and element  boundaries.  Accordingly,
use of the finite-element  method,  as applied
the discontinuity described  above.   The  two
reexami ned
presented.
                                   this  paper describes the
                                   to Darcy's law, to  remove
                                   hypothetical  problems are
in the light  of the proposed approach,  and  the results are
                          MATHEMATICAL  STATEMENT
     The governing equations to  describe  the  water movement in variably
saturated-unsaturated porous media have been  derived  in  detail [4]in the
following form:
                       F    -
- Q = 0
                                                          (1)
and
                                                                     (2)
where
                                                                     (3)
and
                               H - h -4-
                                                          (4)
in which  h is the  pressure  head, 0 is the  moisture  content, n  is  the
effective  porosity,  « '   and  3' are  the modified  coefficients of  com-
pressibility  of  the medium  frame and water,  respectively,  Kfj  is  the
hydraulic  conductivity  tensor, Vj  is the  Darcian velocity vector,  t
is the time,  Q is  the artificial  recharge or withdrawal,  z  is vertical
coordinate, and H is the total head.   In Eqs. (1) and  (2),

-------
                                   167
             the  partial  differentiation  with  respect to  the spatial
              -j; and  2^x3  is  the  vertical  coordinate.   In general,
 8-j  denotes
coordinate,
Eq. (1) is nonlinear  because both the soil property,
Eq.  (3)),  and  hydraulic   conductivity   tensor,   K,-
functions of the pressure head, h.
                                                         (represented by
                                                          are  nonlinear
     The initial condition of Eq.  (1) may take the following form:
                                          in  R
                                                                      (5)
where  h0 is  a
                prescribed function  of  the  spatial  coordinate,  Xj,
        region  of interest  enclosed  by the  boundary,  B(xi,  x?,  x
                                          h,        '
   R
is  the
=  0 as shown  in Fig.  1.   The function,  h0, may  also be  obtained  by
simulating  the  steady-state  version  of  Eq.  (1)  with  time-invariant
boundary  conditions.    In  the  groundwater  flow model,  three  types  of
boundary  conditions  are  generally encountered.   On  the  first  type
(Dirichlet) boundary, the pressure head is prescribed:
                                            on  B,
                                                                      (6)
where  B]  is  a  portion  of  the boundary  B  and  h-|  is  a given  input
function  of time  and  spatial  coordinate  on  B-).   On  the second  type
(Neumann) boundary, the flux is prescribed as:
                                            on  B,
                                                                     (7)
where n-  is the directional  cosine of the  outward unit  vector normal
       -j
        B
to the  B£ portion of  the  boundary B.  The  third type is  the variable
boundary  in  the  sense  that  either  the  Dirichlet  or  the  Neumann
conditions may prevail :
                      h = h3(xi,t)
                                           on   B,
(8a)
or
                           = q
                                           on
(8b)

-------
                               168
                                          ORNL-DWG 78-20110R
  t
                  B
                             B
           B= B,   U  B2  U
                                                B-
     Fig.  1.   Spatial  boundaries,  BI ,  83,  83,  and  B-|,  of  flow
region R.  (B.C.  = Boundary Conditions).

-------
                                  169
where  113  and  Q3 are  two known  input functions  of  time  and  spatial
coordinate  x-j  on  the  83  portion  of  B.   The  boundaries,   B-j,   Bj>,
83,   and    the  impervious   boundary,  Bj,   constitute   the   entire
boundary,  B(XT,  X2,  ^3) =  0.   Initially  Eq.  (8a)  is   applied   to
the  boundary  83 when  the exact boundary  conditions cannot in  general
be predicted  a priori.   Such  a case would  arise at the ground  surface
where  either  ponding  (Dirichlet)  or  infiltration  (Neumann) conditions
could  prevail  [6].   This can only be  determined in the cyclic  process
of solving Eqs.  (1) and (?).

     Applying  the  Galerkin  finite-element procedure  to Eq.  (1),  one
obtains the following matrix equation:
                                                                     (9)
where the temporal derivative of the head, hj is given by


                                     dh.
 The matrix equation coefficients are defined as:
                         = /„ N.FN.  dR  ,                            (ID
                            »  1  J
                      D4 = /_ [K  -O N )  - QN ]  dR,   and            (13)
                       i    R   mJ   mi      i



                      Q± = /   N±q2 dB + /   N±q3 dB  ,               (1A)

                            B2            B3



 where Nj is the basis function at the nodal  point i.

      Eq.  (9)  can be  solved by  any of  the  six alternative  numerical
 schemes listed  in  Table 1 [8].  They  are dependent on  the method of
 time  marching  and  the  treatment  of  the  mass matrix,  L?1jJ-    For
 example;  schlne 1  uses  the Central  difference time marching with no
 mass lumping for the matrix,

-------
                                  170
     After the pressure field, h, is obtained,  the  Darcian  velocity  can
be obtained  by  Darcy's law,  Eq.  (2).   Conventionally,  it  is  evaluated
numerically by taking the derivative of the  approximate pressure  field,
h = hN, as follows:
                     V1  =  -K1j  [3j(hkNk'  + V                       (15)
This approach  naturally  yields the discontinuity  in  the velocity,  V-j,
at  nodal  points  and  element boundaries.   If  continuity  is  to  be
achieved, the  nodal  gradient must also  be treated as  an  unknown [6].
This in  turn will increase,  by a  factor of 3  in the  two-dimensional
problems  or  4 in  the  three-dimensional  problems,  the  number of  equa-
tions and the  bandwidth  of the system and thereby  increase  drastically
the  computing  time  and  computer storage  requirements.  To  circumvent
this problem,  it is  proposed that the finite-element method  be  applied
to Eq. (2) to yield:
            [Sijl{Vnj} = {Dni}            n - 1, 2, or 3   ,           (16)
where
                            S'   - /  N N. dR                        (17a)
                             ij    R 1  J
and

                  D^,- =  -    /  N-[K mOm(N.h.) +  6mo>  ] dR            (17b)
Eqs. (9) and  (16) can be  solved  simultaneously by iteration.   This will
not complicate the problem,  since Eq. (9) has  to be  solved iteratively
because  of  its nonlinearity for  each time  step.   It will not  require
any additional storage  either.   However, it will  increase  the CPU time
by a factor of less than  3  or  4  for two- and three-dimensional problems
respectively.  If  the  storage  for the  matrix [sfj ] is  available,  the
CPU time wijl  remain  practically  the  same as the conventional  approach,
because [Sjj] needs to be evaluated and triangular!zed only once.

-------
                                  171
     The mass balance over the  whole region of interest  is  obtained  by
integrating Eq. (1):
                              i
                        /  F ^ dR = / F  dB  ,                       (18)
                         R  8t       B  n
where  Fn  is the  normal  flux through  the global  boundary  B(x,z) ~  0.
In fact, Fn denotes:
                            F  - n. K..  3.H  .                        (19)
                             n    i  ij   j


Having  obtained the  pressure  head  field,  h,  one could  integrate  the
rightand  left-hand  sides of  Eq.   (18)  independently.   If  the  solution
for h is  free  of error,  one would expect the equality of  the two inte-
grals.   In  this paper, the  integral  of  the  right-hand  side  is  broken
into five components:
                                      dB   .                          (20
                          F  =  /   F  dB  ,                         (2D
                           N    J    n
                                f    F  dB   ,                        (22)

                                 B3S  n
                          F  -  /    F  dB  , and
                           K.     _    n
                                 B3R
                                    Fn dB  ,

                                 B
(23)

-------
                                   172
where  Fp,  FN,  FS,  FR,  and  Fj  represent  the  fluxes  through  the
Dirichlet  boundary,   B];   the   Neumann   boundary,  82;   the   seepage
boundary,  635;   the   rainfall-infiltration  boundary,   83^;   and  the
impervious  Neumann boundary,  Bj;  respectively.   On the  other  hand,
the integral  on the left-hand side of Eq. 18,
                                        dR  ,
(25)
represents  the volumetric increasing  rate of  the  moisture  content  in
the  region.   For  exact  solution,  the  net  flux  across  the  whole
boundary, defined by
                    Fnet = FD + FN + FS + FR + FI
(26)
should satisfy the following equation:
                              F    = F
                               net    V   *
(27)
In  addition, FI  should theoretically  be equal  to zero.   However,  in
any practical  numerical  simulation, Eq.  (27) will  not be satisfied and
FI  will  be  nonzero.    Nevertheless,   the  mass  balance  computation
should provide a means  to  check the numerical scheme and consistency in
the computer code.
                                 RESULTS
     Two  sample  problems  are  used  to compare  the  results  from  the
models  represented by  Eqs.  (15)  and  (16),  respectively.   The  first
example  is  a hypothetical  seepage  pond problem described earlier [9].
The  second   one  is the  Freeze's transient  problem  reported elsewhere
[4,10].   In addition, results by all  six  alternative numerical schemes
(Table 1) are compared in both examples.
Seepage Pond Problems

     A seepage  pond is assumed  to  be situated entirely  in the unsatu-
rated  zone  above  a water table  (Fig. 2a).   This  pond provides a source

-------
                       173
Table 1.  Listing of Alternative Numerical  Schemes
Numerical
schemes
1
2
3
4
5
6
Time -marching
Central Backward Mid-
difference difference difference
X
X
X
X
X
X
Mass
Without
lumping
X
X


X

matrix
With
lumping


X
X

X

-------
                                                                    ORNL-DWG 79-15951R2
                             NEUMANN
      NODE NO. 179
 ELEMENT NO. 153
ELEMENT NO.
 30m
                                       ELEMENT NO. 160
                                       ELEMENT NO. 159
                                                                      ELEMENT NO. 522
                                                                      NODE NO. 587

                                                                      ELEMENT NO. 528
                                                                     ELEMENT  NO. 521
                                                                                    j
                                                                                   P»
                               180m
ELEMENT NO. 1
NODE NO. 2
ELEMENT NO. 2
             Fig.  2.   Spatial discretization of the seepage pond problem,

-------
                                  175
of  water which   infiltrates  into  subsurface  aquifers.   After  water
reaches  the  water  table,  it  flows  toward  a  stream.    Is  is  further
assumed that the  system  is composed of highly permeable  sand  with  soil
properties shown  in Fig. 3.   For  the  finite-element computation,  the
entire  region  is discretized  by  595  nodal  points and 528  elements
(Fig. 2b).  Seven nodal  points  on the stream-soil interface are  desig-
nated  as  Dirichlet nodes.   Seven  nodal  points  on  the  bottom of  the
seepage  pond  are considered  as  constant  Neumann flux  points and  are
assigned  a  constant infiltration rate of  4.0 x 10~4  cm/s.   The  top
sides of  all elements  on the sloping surface, except the  two elements
immediately to the  right of the  seepage  pond, are  considered variable
boundary surface, i.e.,  seepage-rainfall boundary.   In other words,  the
nodal  points  on  this  surface  are  either Dirichlet  or  Neumann  points
with the infiltration rate equal to the excess rainfall  rate.

     The  Darcian  velocity field  computed  by  model  Eq.   (15)   shows  the
discontinuity at  every  nodal  point as indicated  by multivectors  at any
point  in  Fig.  4.   The  severity  of this  discontinuity  depends on  the
location.   This  discontinuity  is  completely  eliminated with  model
Eq.  (16)  as can  be seen  from  Fig.  5, which  shows the  unique velocity
vector  at all  nodal  points.   Table 2  shows the  comparison  of  the
computed  Darcian  velocity components  simulated  by model  Eq.   (15)  and
(16), respectively, for  the three selective  nodal  points 2, 179 and 587
(Fig. 2).  These  three  sample  points are taken from computer  output to
illustrate the difference between the two models.   It is seen  that at
nodal  point  no.  2, the  vertical  velocity  component  as  computed  from
element  no.  2 is about 2.58 times that  computed from  element  no.  1.
The  values of  the horizontal  component, Vx,  at  nodal  point no.  179 as
computed  from  element  nos.  159  and  160  are about  1.41  times  those
computed  from element nos. 152  and  153; while the values of  the verti-
cal  component,  Yz,  at  the  same  point as  computed  from element  nos.
153  and  160  are   about  4.69  times those computed  from  element nos. 152
and  159.  At  nodal point  no.  587,  the  vertical  velocity  component,
Vz,  as  computed  from  element  no.  522  is  about  1.27  times  that
computed  from  element  nos.  521  and  528.   On  the other  hand,  results
from model  Eq.  (16)  show that  the values  of velocity  components  are
identical at the  same point  as  is expected.   However, the CPU times on
IBM  360/91  by  model Eq.  (15)  and  (Ifi) are  0.92 and 2.51  minutes,  re-
spectively.  A saving of computer storage of about 150 K is achieved by
model  Eq.  (16)  though.   Since the  steady-state solution is sought, nu-
merical  scheme  nos. 1  through  6 (Table 1)  yield identical  results as
expected.


Freeze Transient  Problem

     A very small (6  x  3 m)  laboratory-sized watershed was monitored by
Freeze  [10] to test his finite  difference  computer code.  The same data
were also  used   by Reeves  and  Duguid  [4]  to  debug  and test their
finite-element  model.   These   watershed  data are  again  used  in  the
present  paper to  compare model  Eqs.  (15) with  (16).

-------
                           176
                                        ORNL-DWG 74-3908
   -400   -300    -200    -100      0

          A, PRESSURE HEAD (cm of water)
100
    Fig. 3.  Hydraulic conductivity and soil  moisture characteristics
of a hypothetical sandy soil.

-------
                                                                   ORNL-DWG 79-13061R


                                                            (a] VELOCITY VECTOR PLOT
     Fig.  4.   Flow  velocity  plot  as  simulated by model  Eq,
seepage pond  problem.
(15)  for the

-------
                                                                 ORNL-DWG 79-13060R

                                                          (a) VELOCITY VECTOR PLOT
                                                                                          oo
     Fig.  5.   Flow  velocity plot  as  simulated by Model Eq.  (16) for the
seepage pond  problem.

-------
                              179
Table 2.  Comparison of Velocity (cm/sec) Components Simulated by
Model Eqs.  (15) and  (16), Respectively,  at Three Selected Points
Node Element
number number
1
2
2
152
153
179
159
160
521
587 522
528
Model Eq.
Vx
2.38E-8
2.33E-8
2.26E-5
2.26E-5
3.31E-5
3.31E-5
6.28E-5
6.28E-5
7.89E-10
(15)
vz
-2.253E-8
-6.54E-8
-9.15E-5
-4.31E-4
-9.15E-5
-4.31E-4
1.85E-4
2.36E-4
1 .85E-4
Model Eq.
Vx
1.24E-8
1.24E-8
3.03E-5
3.03E-5
3.03E-5
3.03E-5
1.23E-5
1.23E-5
1.23E-5
(16)
vz
-4.44E-8
-4.44E-8
-2.94E-4
-2.94E-4
-2.94E-4
-2.94E-4
1 .84E-4
1.84E-4
1 .84E-4

-------
                                   180
     The flow system is shown in Fig. 6.   It  is  composed  of highly per-
meable sand,  the  unsaturated  properties of which  are shown in Fig.  3.
To  obtain  initial conditions,  pressure-head values  were pre-  scribed
along  the  stream channel,  part of  the  slope,   and  the  upper  plateau.
Taking all other  boundaries to be impermeable,   a  steady-state  solution
was  determined  which  was  the initial  condition  for  the  transient
calculation.

     Using Freeze's transient boundary condition (Figure  6b) and  Reeves
and Duguid's  finite-element discretization  #2 (Figure 6c),  results  ob-
tained by models  Eqs.  (15)  and  (16), respectively, are  shown in Figs. 7
and 8.  Model  Eq.  (15)  again  displays the  discontinuity of  velocity
vectors at all  nodal  points, while model Eq.  (16)  has completely  elimi-
nated  this  inconsistency.  Furthermore,  Table   3  shows  that  the  mass
balance has not been  satisfied  by model  Eq. (15).  At the  end  of about
3-h simulation  time,  the  total  net mass through all  boundaries is only
about  76.2% of  the mass accumulated  in  the  media as  computed by numeri-
cal scheme no.  1  of model  Eq.  (15).   In other words, 23.8%  of  the mass
has not  been  accounted for,  i.e.,  has been  lost through  boundaries.
Reeves and Duguid [4] speculated that this large loss of mass  might be
eliminated by  adding  triangular elements.   However,  without using tri-
angular elements, model Eq. (16) only yields  a  2.?% mass  loss  by  elimi-
nating the discontinuity  of the velocity and  using Eq.  (25) to evaluate
the  moisture  rate change.  An  even  larger mass   loss of  29.7%  is
obtained by numerical  scheme no.  2  of model Eq.  (15).  Model  Eq. (16),
on  the other  hand,   renders  a 3.6% mass  gain.  Thus,  error of  mass
balance  (positive for  loss,  negative for  gain) by  model  Eq. (16)  is
much smaller  than that by model Eq.  (15).   The  CUP time on IBM  360/91
for models Eq.  (15) and Eq. (16) are 2.3 and 4.F minutes, respectively,
but a  saving  of  computer  storage  of about  140 K  is  obtained  by  the
latter.

     Table 3  also shows the percentage  of mass change by  all  alterna-
tive  numerical  schemes.    It  is   noted  that   the  central  difference
standard  Gale rid n scheme  in  model  Eq.  (16) yields  the best  results.
This  is  not  surprising  since  the  water   transport  equation  does  not
contain advection terms.
                                CONCLUSION
     A  continuous  velocity  field  is required  for  the simulation  of
waste  transport  in the  subsurface aquifer  system.   The  conventional
approach  used in  finite-element modeling,  that is,  taking  the  deri-
vative  of the approximately  computed pressure  head  field,  results  in
discontinuity  of velocity at  element  boundaries and nodal  points.   It
may  also  yield large error in overall mass  balance computation.  Thus,
it  is  imperative that  the  same finite-element method that is employed
to  simulate  the  pressure  field be applied to Darcy's law to obtain the
velocity  field.   This will   ensure  a  Continuous  velocity  field  and
greatly  reduce the overall  mass balance  error.   The  central difference
Galerkin  scheme  yields the lowest  error, since  the  governing equation
is  basically the Laplacian type  in  spatial coordinates.

-------
                                    181
                                                   ORNL-DWG 79-13083R
                         • Comparison with FD Solution
            (a) STEADY STATE
               BOUNDARY CONDITIONS
IMPERMEABLE
                                    IMPERMEABLE
            (b) TRANSIENT BOUNDARY
               CONDITIONS
            RAINFALL
                                    IMPERMEABLE
            (c) SPATIAL MESH DISCRETIZATION

           T
           70cm

10
13C

)cm
) cm







































UlOO cm-


X


*•






f
/ /
*£''''
'
' / /[ /
s / /
*.
s "
----
. - • - -
—100 cm—
     Fig.   6.    Configuration   of  Freeze's   experimental   watershed:
(a)  steady  state  boundary condition,  (b)  transient boundary  condition,
(c)  spatial finite element discretization  after  Reeves and Duguid (197F),

-------
                                                             ORNL-DWG 79-13082R
(a) VELOCITY VECTOR PLOT
               Fig. 7.  Flow velocity plot  at  time equal  to  2.96 h  of Freeze's
          transient problem as simulated by Model  Eq. (15).
                                                                                         00
                                                                                         N>

-------
                                                               ORNL-DWG 79-43067R
(a) VELOCITY VECTOR PLOT
                Fig. 8.   Flow velocity plot at time  equal to 2.96  \\  of  Freeze's
            transient problem  as simulated by Model Eq. (16).
                                                                                           00
                                                                                           CO

-------
                            184
Table 3.  Comparison of Percentage of Mass Change of Freeze's
  Transient Problem as Simulated by Model Eq. (15) and (16)
Scheme
Model
Eq.
Eq.
(15)
(16)
1 2 3
23.8 29.7 N/A
2.2 -3.6 8.9
4
N/A
3.0
5
N/A
-3.2
6
N/A
-3.3

-------
                                   185


                                REFERENCES
 1.  S.  P.  Neumann and P.  A.  Witherspoon, "Analysis  of  nonsteady flow
     with  a  free  surface  using  the  finite  element  method,"  Water
     Resour. Res., 7(3). 611-623  (1971).                           -

 2.  G.  F.  Pinder and  E.  0.  Frind, "Application  of Galerkin procedure
     to aquifer analysis, Water Resour. Res.. £(1), 108-120 (1972).
 3.  G. F. Pinder,  "A  galerkin  finite  element simulation of groundwater
     contamination on  Long  Island,  New York," Water Resour. Res., 9(6),
     1657-1669 (1973).                         - —  ~

 4.  M.   Reeves   and   J.   0.    Duguid,    Water   Movement   Through
     Saturated -Unsatu rated  Porous  Media:   A Finite  Element  Galerkin
     Model,  ORNL-4927.  Oak  Ridge  National   Laboratory,   Oak  Ridge,
     Tennessee (1975).

 5.  S. K.  Gupta,  K.  K.  Tanji, and J.  N.  Luthin,  A Three-dimensional
     Finite Element Groundwater Model,  University  of California,  Davis,
     Water Resources Center Contribution Series #1F2 (1975).

 6.  G. Segol ,  A  Three-dimensional  Galerkin  Finite-Element Model  for
     the    Analysis    of    Contaminant    Transport    in    Variably
     Saturated -Unsatu rated  Porous  Media,  Dept.   of  Earth  Sciences,
     University of Waterloo, Canada (1976).

 7.  Y. H. Huang and  S. J. Wu,  "Comparison of Three-dimensional  Finite
     Elements  for  Aquifer  Simulation,   Finite   Element  in   Water
     Resources," Proceedings  of the  First  International Conference  on
     Finite  Elements   in  Water  Resources,   edited  by W.  IT.   Gray,
     G. F. Pinder   and   (H   A~T   Brebbia,    Pentech   Press,   London,
     pp. 2.87-2.105 (1977).

 8.  G. T.  Yeh and  D. S. Ward,  FEMWATER:    A Finite Element  Model  of
     Water Flow through Saturated-Unsaturated Porous  Media, ORNL-5567,
     Oak Ridge National Laboratory, Oak Ridge, Tennessee (1980).

 9.  J. Duguid  and M.  Reeves,  Material Transport in Porous Media:   A
     Finite  Element  Galerkin  Model,  ORNL-4928,  Oak  Ridge  National
     Laboratory,  Oak Ridge, Tennessee (1976).
10.  R.  A.  Freeze,  A  Physics-based  Approach  to  Hydro!ogic  Response
     Modeling;  PI
     Contract  NoT
     D.C. (197217
Modeling;  Phase"I - Model  Development,  Completion Report for OWRR
Contract  No.  14-31-001-3694,  Pept.  of  the  Interior,  Washington,

-------
     FRACTURE FLOW MODELING FOR LOW-LEVEL NUCLEAR WASTE DISPOSAL
          Brian Y. Kanehiro      Varuthamadhina Ruvanasen
          Charles R. Wilson      Paul A. Witherspoon

                    Lawrence  Berkeley Laboratory
                      University of  California
                     Berkeley, California  94720


                               ABSTRACT

          Fracture flow  modeling as applied to  low level  radio-
     active  waste  disposal  sites is discussed.   Low level  waste
     disposal  sites presently  exist in fractured basalts,  crys-
     talline igneous  rocks, and limestones.  The primary  objec-
     tive of forecasting the migration of radionuclides from
     these sites requires the  consideration of  fracture flow
     modeling.  Such  modeling  is also an integral  part of the
     planning and  analysis of  field work at the site.

          Fundamental  mathematical  models describing discrete
     fracture flow and equivalent anisotropic porous medium
     flow are reviewed.   Consideration  is given to the theoret-
     ical and field aspects of the problem of when and how to
     treat a fractured medium  as an equivalent  anisotropic
     porous  medium.   The input parameters required for each
     type of model are discussed.

          Finally  a brief description of some of the models
     developed at  the Lawrence Berkeley  Laboratory is given.
     The models include  those  based on the integrated finite
     difference, polygonal finite difference, and finite-
     element schemes.
                             INTRODUCTION

     Fracture flow modeling is an attempt to consider the effects of
discrete fractures in approximating the nature of fluid movement in a
fractured medium.  Fracture flow predominates whenever the fluid flux
through the fractures is appreciably greater than that through the
unfractured, porous matrix.  This condition implies that the hydraulic
conductivity of the fractures, whether completely open or partially
filled, is substantially greater than that of the matrix.  Fracture
flow strongly influences groundwater particle flow velocities, flowpath
directions and the mechanisms of sorption-desorption of migrating radio-
nuclides, and thus is of considerable importance to the safe disposal
of low-level nuclear waste within media where an interconnected network
of open fractures exists.  This paper discusses the fundamental prin-
ciples of fracture flow and the circumstances under which fracture flow

                                 187

-------
                                  188
must be considered in modeling low-level waste disposal sites.   Brief
descriptions of the various types of fracture flow codes available at
LBL are given.

     The most important fractured media in terms of existing low-level
waste disposal sites are basalt, crystalline igneous rocks, and lime-
stone.  This is not to suggest that these rock types are the best for
disposal or the most commonly occurring types of fractured media, but
rather that many low-level waste sites have been constructed in them.
Indeed, basalt and limestone are generally poor choices for containment
and problems requiring remedial efforts are quite likely to occur at
such sites unless they are properly designed for the existing fracture
system.  Matrix flow in each of these rock types is likely to be rela-
tively small in comparison to the flow in the fractures.

     The most obvious reason for the utilization of a model in studying
a low-level waste site is to forecast the future migration of radionu-
clides from the site.  There are, however, other reasons for employing
a model that are just as important to the full assessment of the site.
Because of the geologic and hydrologic complexities of most sites, the
input parameters for forecasting the future of a site are generally
difficult and costly to obtain.  The availability of a model during the
field phase of site assessment, when data for input parameters are being
collected, can insure that the most useful information is collected in
the most efficient manner.  In particular, parameter sensitivity studies
can be used to plan field work and provide assurance that characteriza-
tion of some aspects is not being overemphasized to the neglect of others.
Further, the availability of a model permits analysis of field test data,
such as complex tracer tests, to an extent that would not be practical  if
only more conventional analytical techniques were available.


                    FUNDAMENTAL MATHEMATICAL  MODELS

     Flow through a fractured medium is generally mathematically described
in one of three ways:  (1) by considering each fracture as a discrete
hydraulic conduit; (2) by assuming a hydraulically equivalent anisotropic
porous medium and using the appropriate porous medium model; or  (3) by a
combination of (1) and (2).

     Flow through a single fracture is normally represented by a parallel
plate analogy  [1].  If the flow is assumed to be laminar, the fluid single-
phase and Newtonian, and the parallel plates smooth, the steady  flux from
a single fracture may be described by:
where q = flux per unit length of fracture  (L/T)
      b = fracture half-aperture            (L)
      p = fluid density                     (M/L3)
      g = acceleration due to gravity       (L/T2)

-------
                                  189


      y = dynamic viscosity of the fluid  (M/LT)
      h = piezometric head                (L]
      I = length along the fracture plane (L]

     Utilizing the parallel  plate analogy, it can be seen that the factor
(2b)2/12 in eq. (1) directly corresponds to the intrinsic permeability  of
a porous medium (k^).  The extent of this correspondence is  demonstrated
in the following parallel  development of the fundamental  equations govern-
ing fluid flow and nonreactive solute transport in both porous media  and
fractures.

     The equations governing flow in an anisotropic porous medium can be
derived from continuity and Darcy's law and may be tensorially written  as:

           3  i/    oh  ,  n _ c-  on
          ^r   i\ • •  ^r   I  \j — j  ^ .
          ^v   T t  jv    "    C *JT
          8x.  ij  3x.j        s 3t


where K.. = hydraulic conductivity tensor
            k-Hpg
          = -^—                                               (L/T)

        Q = flow rate of source or sink per unit volume of rock (1/T)
       S$ = specific storage                                    (1/L)

        t = time                                                (T)
      k.. = intrinsic permeability tensor                       (L2)


     Along a fracture plane, consideration of continuity and eq.  (1)
requires that:
                            u
           3   0  3h

                            X/
- Ss %
where ?. = rectangular coordinates along the fracture plane

      K  = fracture hydraulic conductance
                  ,                                             o/n
       g = fracture specific storage coefficient                (1/L)
      q. = flow rate per unit volume of fracture across the
       1   fracture walls                                       (1/T)
      m. = component of the unit vector outward normal to the
       1   fracture plane

     The second term in the above equation refers to the difference
between influx and efflux at the upper and lower fracture walls.

     The equation governing the transport of a nonreactive solute in
porous media is tensorially written  [2]:
                                                                     (4)

-------
                                 190


where   C = solute concentration                      (M/L3)
       u. = seepage velocity in the ith direction     (L/T)
      Dii = hydrodynamic dispersion tensor            (L2/T)
      S'n = so^ute concentration at sources or sinks  (M/L3)

     The velocities and coefficients of hydrodynamic dispersion are
calculated from
                e  ax.
                     J

where a^i = convective dispersivity tensor          (L)

        ]u| = seepage velocity magnitude              (L/T)
         D. = molecular diffusiyity coefficient       (L2/T)
        T.. = tensor of tortuosity
        6^. = Kronecker delta

          9 = effective porosity

     Along a fracture plane the nonreactive solute transport equation can
be written             ,
                     3u.C
" = 2b -^ D^. ^-    (7)
I        i       .1
in which the third term refers to solute influx and efflux across the
fracture walls.  At present there exists no experimental  data for DV..
the hydrodynamic dispersion coefficient tensor.  For approximation  J
purposes, however, D^j can be considered to vary from D^ii for a clean
fracture to D^j for a fully-filled fracture.  Transport of  reactive
constituents can be treated by the source/sink terms in eqs. (4) or (7).


                  THE EQUIVALENT POROUS MEDIUM MODEL

     A model based on only discrete fractures provides a theoretically
correct flow pattern if all fractures and connections between fractures
are considered.  But, while such a model may be relatively easy to con-
struct from a numerical point of view, the mass of input data required,
the computational effort, and the difficulty of obtaining data for many
real-world problems makes it generally difficult, if not impossible to
apply to regional analysis.

     One way to circumvent this problem is to represent the fracture
system by an equivalent porous medium.  An equivalent porous medium is  a
fictitious porous medium with material properties so selected that, under
specific hydrologic and geometric conditions, the hydraulic behavior of
the porous medium is in some respects the same as the hydraulic behavior

-------
                                  191


of the fractured rock.  The validity of the equivalent porous medium concept
is thus very site-specific and flowpath-specific.   If it is assumed that
for the purposes of the problem at hand, the domain being considered can
indeed be represented by an anisotropic porous medium, then the major
problem becomes that of determining the permeability tensor.  The mathe-
matics of calculating directional permeabilities from fracture orientation
and aperture data based on the assumption that permeability is a function
of aperture squared were first discussed by Romm and Pozinenko [3].  Exten-
sive work in this area has been done by several others [1, 4, 5].

     Under a general field gradient, flow in a single conduit may be des-
     d as [1]:
cribed as
                              - V
where  q- = flow rate per unit width of the fracture plane
            in the jth direction                             (L2/T)
      6.. = Kronecker delta
      M.J. = 3x3 matrix formed by direction cosines of the
       1J   unit normal to the fracture plane
       x. = Cartesian coordinates                            (L)
        j

     It is further assumed that for a large number of fractures intersect
ing a sample line, each fracture has its image at a distance L equal to
the length of the sample line and in the direction of the sample line.
Thus, the spacing between the fracture and its repetition is given by [6]:

          w=L|niD.|                                                (9)


where  w = spacing between fractures                         (L)
       L = length of sample line                             (L)
      n- = component of unit normal to fracture plane
      D! = component of sampling line direction
     The effective intrinsic porous medium permeability of a fracture is
then given by:
     The equivalent permeability of the medium at a given sampling station
is obtained by summing the contributions  from the individual fractures.
Where more than one station  is used, the  average of all stations is calcu-
lated.  Knowing the permeability tensor,  its principal components and dir-
ections may then be calculated.

     This approach assumes that each fracture completely penetrates the
volume being characterized,  without changes in aperture or orientation.

-------
                                  192


Since in nature many fractures are observed to terminate over relatively
short distances, this model would generally be expected to overestimate
rock mass permeability.

     If a similar procedure is followed to determine Dji based on field
data, then the anisotropic porous medium model describes by eqs. (2) and
(4) may be used directly.

     In some cases where a relatively small number of large fractures
exist together with a large number of small fractures with different
orientation, it may be useful to employ both the discrete fracture and
porous medium models.  The equivalent tensorial properties would be
applied to that portion of the medium dominated by the smaller fractures
and the problem would be treated as a fractured anisotropic porous medium.
Eqs. (2), (3), (4), and (7), and consideration of continuity would govern
the system.


                       MODELING OF A FIELD SITE

     Selection of a model for a particular field application requires two
primary considerations.  First, the type of output required of the model
must be clearly defined, and second, the type of input available for the
model from the field site must be identified.  Put in another way, the
data needs for decision-making that are to be supplied to the model must
be compatible with the capabilities of the model, which in turn must be
compatible with the physical and economic realities of obtaining the
necessary data from the field and operating the model.  Clearly, the out-
put needs, the choice of model, and the input needs are all interdependent.

     The types of output parameters required of models for safe disposal
of nuclear waste will involve dose rates and contaminant concentrations
under environmental conditions to be prescribed by the Environmental
Protection Agency.  Our purview as earth scientists will be to recommend
the most appropriate approaches to obtaining the required information.
This means balancing the applicability of models against the availability
of input parameters to best achieve the output requirements.

     Groundwater movement and nuclide transport in fractured media can be
modeled with either a discrete fracture model or an equivalent porous
medium model.  The types of input parameters that are required for these
two models are not the same.  The fracture model requires fracture geometry
data such as fracture orientation, spacing, and aperture while the porous
medium model requires the large-scale averaged parameters of permeability
and dispersion.

     The question of which model to use for a given field site is, in
general, not a simple one.  The limitations of each model must be evaluated,
and its applicability with regard to the model output requirements must be
determined.  From a theoretical point of view, the choice of models is
generally a question of scale.  If the representative elementary volume  [2]
concept is valid for the site, and if both the size of the site and the

-------
                                  193


scale of required output parameters are larger than the representative
elementary volume, then the anisotropic porous medium equivalent can be
employed.  The other extreme of a case where there are a rather small  num-
ber of identifiable distinct fractures within the entire site,  calls for a
discrete-fracture model.  Both types of models suffer from difficulties in
acquisition of input data.  Parameters suitable for equivalent  porous
medium models must be obtained from field tests designed to average the
individual effects of many fractures over volumes larger than the repre-
sentative elementary volume.  On the other hand, parameters suitable for
discrete-fracture models must accurately identify the specific  geometry
of the principal flowpaths.  Unfortunately, many near-surface low-level
waste field sites will probably not clearly fit into either category and
a hybrid model may be required.

     In principle, the most straightforward method for determining the
size of the representative elementary volume is to perform hydro!ogic
tests in different portions of the site starting with small-scale tests
(i.e., closely-spaced wells) and expanding the scale of each of the tests
in increments.  As the tensors of permeability and dispersion,  computed
from tests of progressively increasing scale, approach some type of con-
sensus in magnitude and orientation at a particular test site,  it could
be assumed that the scale of the individual test at that particular loca-
tion is approaching that of the representative elementary volume at that
location.  Further, if the results of the individual tests at different
locations are seen to converge, then it could be concluded that the site
as a whole can be described as a single anisotropic porous medium with
similar properties throughout.
                             s
     While this procedure appears to be straightforward, it may be imprac-
tical for many real-world problems.  First, the procedure might be costly
and time-consuming.   Second, most field sites are likely to be heterogen-
eous, implying that this procedure would have to be performed for each
fractured geologic formation at the site.  Finally, determination of ten-
sorial properties in  the field, particularly for a three-dimensional case,
would be  quite difficult.   Despite these difficulties, the alternative of
measuring the geometry of all  significant discrete fractures is likely to
be impossible over any reasonably-sized field site, leaving us with little
choice but to employ  an equivalent porous medium model or a hybrid model
at many  sites.

     An  alternative to  the  direct large-scale field measurement approach
described above  is to compute  equivalent porous medium properties based
on many  individual measurements of  single  fractures,  usinn a procedure
similar  to that  proposed  by Snow  [1]  and discussed above.  An extensive
field evaluation of this  approach is  now being  conducted by LBL at.the
Stripa test  station  in Sweden, and  the results  will be compared with
those of a large-scale test involving 105  to  106 m3 of rock  [7].

     For the modeling of  a real  field site,  it  may be best to not decide
on the type  of  mathematical  model  to  be used  until substantial  field  data
are  available.   Information on discrete fractures  will  be  required  in  the
design of field tests and in evaluating the applicability  of the equivalent

-------
                                   194


 porous medium concept,  thus the field  work will  of recessity  provide  input
 parameters for both equivalent  porous  medium and discrete  fracture models.
 Such an approach will provide more flexibility and will  ultimately result
 in a better understanding  of the site.   After, and to  a  certain  degree
 during the field-work phase of  the study,  both types of  models,  together
 with a hybrid having both  discrete fractures and equivalent porous media,
 can be tried and the best  model  identified.

     ^From a practical point of  view the  need for fracture  data will
 require the addition of a  limited number of activities to  those  required
 for the standard porous medium  type characterization and assessment of a
 low-level  site.   In particular,  fracture mapping and fracture core logging
 will  have to be  added to the geologic  activities.   Hydraulic  and tracer
 testing will  have to be performed in a somewhat  more complicated fashion
 so as to test individual fractures.  Finally,  preparation  of  two or three
 types of mathematical models will  be required.


                 MODELS  AT  LAWRENCE BERKELEY  LABORATORY
                              %

      The Lawrence Berkeley Laboratory  is presently using three types of
 numerical  models that have application to  the  modeling of  low-level waste
 deposit sites.   The first  group  of models  is  based on the  integrated
 finite difference scheme.   Examples  of these  programs are  TERZAGHI [8],
 CCC  [9],  and  SHAFT79 [10].   Of these, TERZAGHI may be the  most useful for
 low-level  site modeling.   It solves  a nontensorial  form  of eq. (2) in
 three dimensions.   It has  been used  to model discrete fractures  and some
 types of anisotropic porous  media.   It also  has  the capability to deal
 with  one-dimensional consolidation.

      The  second  group of models  is  based on  the  polygonal  finite differ-
 ence  scheme.  Examples  are  BIFDA  and GENIE  [11].   These  models solve a
 three-dimensional form  of  eqs.  (2)  and (4).  They  can be used to deal with
 both  discrete fractures and  anisotropic porous media.  In  addition, they
 can handle  heat  transport.

      The third group of models is  based on the finite-element scheme.
 Examples of these models are ROCMAS  [12]. FRACFLOW [13], and  LINE ELEMENT
 [14].   These programs solve  various  forms of eqs.  (2), (3), (4), and/or
 (7),  for two-dimensional cases.  The first two can  be used for either
 discrete fractures or porous media.  In addition,  the first and  second
 can deal with the effects of stress  and the second  can also deal with
 heat  and anisotropic problems.  The  third program  is a steady state line
 element code presently  being used  in research on the definition  of the
 permeability tensor  in  random systems of discrete  fractures.

      The first group of models evolved from a  coupled fluid-heat flow
model.  The second group was developed to provide  good chemical   transport
 and anisotropic media capabilities while maintaining three-dimensional
 capabilities.  The third group was developed to deal specifically with
 discrete fractures with stress-dependent properties.

-------
                                  195


                           ACKNOWLEDGEMENTS

     Work performed under the auspices of the U. S. Department of Energy
through Lawrence Berkeley Laboratory, University of California, under
contract W-7405-ENG-48.
                              REFERENCES

1.   D. T. Snow, "A Parallel Plate Model of Fractured Permeable Media,"
     Ph.D. Thesis, University of California, Berkeley, California (1965).

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

3.   E. S. Romm and B. V. Pozinenko, "Investigation of Seepage in Frac-
     tured Rocks," Trudy VNIGRI, 214 pp (in Russian), (1963).

4.   C. Louis and M.  Parnot, "Three-Dimensional  Investigation  of Flow
     Conditions and Grand Mai son Damsite," Proc.  Sym. Percolation Through
     Fissured Rocks,  Stuttgart, Paper T4-F, (1972).

5.   M. A. Caldwell,  "The Theoretical Determination of the Permeability
     Tensor for Jointed Rocks," Proc. Sym. Percolation Through Fissured
     Rocks. Stuttgart, Paper Tl-C, (1972).

6.   L. Bianchi and D. T. Snow, "Permeability of Crystalline Rock Inter-
     preted from Measured Orientations and Apertures of Fractures,"
     Annals of Arid Zones. 8^(2), (1969).

7.   J. Gale and P. A. Witherspoon, "An Approach to the Fracture Hydro-
     logy at Stripa—Preliminary Results," Sem.  on In-Situ Heating
     Experiments in Geologic Formations.  Org.  Econ. Coop. Dev. Ludvika,
     Sweden.  Also Lawrence Berkeley Laboratory report LBL-7079, SAC-15,
     Berkeley, California (1978).

8.   T. N. Narasimhan and P. A. Witherspoon, "Numerical Model  for Saturated-
     Unsaturated Flow in Deformable Porous Media.  1.  Theory," Water
     Resources Research, 13_(3), 657-664, (June 1977).

9.   M. J. Lippmann,  C. F. Tsang, and P. A. Witherspoon, "Analysis of the
     Response of Geothermal Reservoirs under Injection and Production
     Procedures," SPE 6537, presented at the 47th Annual California
     Regional Meeting SPE-AIME, Bakersfield, California, (April 1977).

10.  K. Pruess, R. C. Schroeder, P. A. Witherspoon, and J. M.  Zerzan,
     "SHAFT78, a Two-Phase Multidimensional Computer Program for Geo-
     thermal Reservoir Simulation," Lawrence Berkeley Laboratory report
     LBL-8264, Berkeley, California  94720 (1979).

-------
                                  196


11.  B. Y. Kanehiro,  "A Three-Dimensional  Polygonal  Solution  of Fluid Flow
     and Chemical  Transport in Anisotropic Porous Media,"  (in preparation).

12.  J. Noorishad, M. S. Ayatollahi,  and P.  A.  Witherspoon, "A Finite-
     Element Method for Stress and Stress and Fluid-Flow Analysis  in
     Fractured Rock Masses," submitted to ASCE  Journal  of  Engineering
     Mechanics Division, (1980).

13.  V. Guvanasen, "Finite-Element Solution of  Fluid Flow, Heat and Solute
     Transport, and Thennomechanical  Deformation in  Fractured Porous Media,"
     (in preparation).

14.  C. R. Wilson and P. A. Witherspoon, "An Investigation of Laminar Flow
     in Fractured Porous Rock," Dept. of Civil  Engineering, Pub. #70-6,
     University of California, Berkeley (1970).

-------
         A CONCEPTUAL ANALYSIS OF THE GROUNDWATER FLOW SYSTEM
                 AT THE MAXEY FLATS RADIOACTIVE WASTE
                 BURIAL SITE, FLEMING COUNTY, KENTUCKY


                             D. W.  Pollock
                             H. H.  Zehner

                        U.S.  Geological  Survey
                           Reston,  Virginia
                         Louisville,  Kentucky


                               ABSTRACT

          The complex groundwater system at the Maxey Flats
     radioactive waste burial site in Fleming County, Kentucky
     was investigated with the aid of a  highly simplified,  steady
     state, groundwater flow model.  An  analysis of the local
     hydrogeology suggests that at least two saturated groundwater
     flow systems are present at the  site:  a lower system  in the
     bottom part of the Ohio Shale, and  an upper, perched system
     extending from the base of the Bedford Shale up to a water
     table in the weathered part of the  Nancy Member of the Bordon
     Formation.   Unsaturated conditions  may also exist in the
     Sunbury Shale.  Simulations indicate that most of the  water
     which recharges the upland surface  of Maxey Flats is even-
     tually discharged as seepage along  the hillside through the
     upper part of the Farmers Member of the Bordon Formation and
     the Sunbury Shale.


                             INTRODUCTION

     This report discusses the groundwater flow system at the Maxey Flats
radioactive waste burial site in Fleming County, Kentucky.   The ground-
water system at Maxey Flats is complex;  flow occurs mainly  through
fractures in hydraulically "tight"  shales and sandstones.  In addition,
water levels from wells at the site indicate the possibility of unsaturated
conditions at depth.  At the present  time the available data do not
warrant a detailed model of groundwater  flow; however, careful examination
of the hydrogeology and the hydro!ogic setting of the site  can help
provide a conceptual understanding  of the groundwater system.

     Maxey Flats is topographically an isolated knob located in the
Appalachian plateau region of eastern Kentucky.  The Maxey  site is
located on one part of this knob and  is  bounded on three sides by streams
(fig. 1).  Relief between the upland  surface and bordering  valley bottoms
is about 300 feet.

     Maxey Flats is underlain by fractured shales and sandstones which
dip gently (25 to 40 feet per mile) toward the southeast.  The rock units

                                   197

-------
                                   198
 38° 16'
                                    Topographic Divide
                                               83° 34'
                                              Trench Area
 38° 15'
               \    Jock Lick Creek
     Fig. 1
Kentucky.
                  SCALE
       CONTOUR INTERVAL = 100 FEET
Generalized topographic map of Maxey Flats, Fleming County,

-------
                                  199


directly underlying the Maxey Flats site are the Nancy and Farmers Members
(predominantly a shale and sandstone respectively) of the Mississippian
Bordon Formation, the Henley Bed (shale) of the Farmers Member, the
Mississippian Sunbury Shale, the Mississippian and Devonian Bedford Shale,
and the Devonian Ohio Shale, plus the upper part of the Silurian Crab
Orchard Formation (shale).  In this report, the Nancy and Farmers Members
will be referred to as the Nancy shale and Farmers sandstone, and the
Henley Bed will be referred to as the Henley shale.

     The weathered and unweathered Nancy shale are separated by a thin
(1 to 2 foot thick) sandstone bed, which is present over most of the site.
This bed will be referred to as the sandstone marker bed.  The upper part
of the Farmers sandstone  is a sequence of alternating, thin  (1 to 3 foot
thick) sandstone and shale beds, whereas the lower part of the unit is
predominantly sandstone.  The two parts probably have different hydraulic
characteristics so, in this report, they are separated into  the upper and
lower Farmers sandstone.  Figure 2 is a generalized geologic section taken
along line A-B, shown in  figure 1.  Radioactive wastes were  buried in
trenches excavated in the Nancy shale on the upland surface.

     The distribution of wells at Maxey Flats is shown in figure 3.  E-wells
were drilled by Emcon Associates.  UA^-wells, northeast of the burial area,
and. UB-wells, in the trench area, were drilled by  the U.S. Geological Survey
in 1976 and 1977.  Well construction and water-level data are summarized
in figures 4 and 5.


                               ANALYSIS

                           Conceptual Model

     Eleven hydrogeologic units are exposed at Maxey Flats.  The sequence
of  units, from top to bottom, and their thicknesses  in feet, are:

     1.  Colluvium, which partially covers the hillsides  (0  to 5)

     2.  Weathered Nancy shale, which covers  the  hilltops  (1 to 25, with
         an average of about 20)

     3.  Sandstone marker bed, which is usually at the base  of the
         weathered Nancy  shale (1 to 2)

     4.  Unweathered Nancy shale (20 to 40, with  an  average  of about 25)

     5.  Upper Farmers sandstone (15 to 20)

     6.  Lower Fanners sandstone (15 to 20)

     7.  Henley  shale  (7)

     8.  Sunbury shale (20)

-------
 HI
 u.
 UJ
   1100
   1000
•ft
 CO
 UJ  900

 O
 ffl
 <

 g  soo


 1
Alluvium
                                              Ohio Shale
                                                                                      Henley Shale


                                                                                           Sunbury Shale


                                                                                            Colluvium
 UJ
    700
                                                  Crab Orchard Shale
                                            0

                                            I
                                                     I
I
 1000

	I FEET
                                                        SCALE



                           Fig.  2.   Geologic  section along  line  A-B in figure  1
                                                                                                                 B

-------
         201
         SCALE
Fig.  3.   Location of wells.

-------
   1050
   1000
    950
u_
z
of
Q
                               6    8
                 3'/'
<   800

    750
    700 -
                                                                -Weathered
                                                               --Nancy Shale

                                                                 Unweathered
                                                                 Nancy Shale
                                                                 Upper Farmers
                                                                 Sandstone
                                                                "Lower Farmers
                                                                 Sandstone
                                                           Henley Shale
                                           Sunbury Shale
                                Bedford Shale
                                Ohio Shale
                                                          • Cemented Interval
                                                           Open Interval
                                                            Water Level
                                Crab Orchard Formation
          Fig. 4.   Well  construction  information and water level data  for
     wells drilled  by Emcon Associates.
                                                                                                    ro
                                                                                                    o
                                                                                                    t\j

-------
    1050
    1000
s
"-    950
Z

LJ
o
D
<    900
          UA-1 UA-2 UA-3 UA-4
        s
     800
     750
UB-1 UB-1AUB-2 UB-3 UB-4



                        Weathered Nancy Shale






                        Unweathered Nancy Shale




                        Upper Farmers Sandstone




                        Lower Farmers Sandstone



                       ^Henley Shale


                        Sunbury Shale




                        Bedford Shale


1


^


1




•\


1


















(_

1


V
I
I
L
h
i ^
E

                                               s
               'Ohio Shale
                                                                  Cemented Interval
                                                                  Open Interval
                                                                    Water Level
         Fig.  5.   Well construction information  and water level  data for wells

    drilled by U.S.G.S.
                                                                                                    ro
                                                                                                    o
                                                                                                    UJ

-------
                                   204
      9.   Bedford shale (20)

     10.   Ohio Shale (185)

     11.   Crab Orchard Formation, with only the upper few feet exposed
          in the valley bottoms  (approximately 120)

      No  quantitative data are available regarding hydraulic properties
 of these units.  Qualitative, visual observations of some properties
 (including porosity, fracture density, and groundwater discharge) were
 made at  outcrops.   The hydrogeologic units above the valley bottoms are
 grouped  into four categories, based on visual observations and water-level
 recovery rates in wells finished in some of the units (Table 1).  The
 colluvium is not  considered in this study.  It undoubtedly plays a role
 in controlling seepage along the hillsides; however, because there are no
 wells on the hillsides, the groundwater regime in the colluvium is unknown.
 The hydrogeology  of  Maxey Flats can be thought of as a series of extremely
 poorly permeable  rock units alternating with a few layers of somewhat higher
 permeability.   The upper part of the Crab Orchard Formation, which crops
 out in the  valley bottoms, is a clayey shale with very few fractures; in
 this report,  it is considered to be the "impermeable" base of the flow
 system.

      The only significant groundwater flow at Maxey Flats occurs through
 fractures spaced from 1 to 40 feet  apart.   One way of modeling flow through
 fractured rock is to treat the fractured rock mass as a porous medium.
That  is, instead of distinguishing  individual fracture-flow pathways, each
 point  in the rock mass is assigned  an hydraulic conductivity which repre-
 sents a measure of the "average" water-transmitting ability of the fractures
 in the vicinity of that point.   Fractured rocks appear to behave as porous
media on a scale where the distance between fractures is small compared
with the dimensions  of the flow system being considered.  Fractured rocks
are usually treated  as porous media for regional  flow studies, but for
 some small-scale studies (such as dam site investigations) it is sometimes
necessary to consider flow through  a network of discrete fractures (e.g.,
Wilson and Witherspoon, 1970).   The major drawback with fracture-network
flow models is that  they require detailed information about the spatial
distribution and hydraulic properties of individual  fractures which is
 rarely available.  In the case of Maxey Flats, it is probably possible to
obtain an adequate representation of the head distribution in the fracture
network by assuming  porous media flow;  however, it may be very difficult
to measure the head  distribution in the fracture  network.  In fractured
rocks that have extremely low intergranular permeability, the water level
 in a well may represent the head in a single fracture at the point where
the fracture intersects the open interval.  If more than one fracture
intersects the well, the water level  will  represent a composite of the
heads in the individual fractures at the points of intersection.  Water
level data from fractured rocks  is  therefore often difficult to interpret
because the location of fracture intersections is usually not known.

     Water levels from wells at Maxey Flats indicate a large decrease in
head with depth.  This fact has  led previous workers to suggest that

-------
                                  205
    Table 1.  Qualitative classification of the hydrogeologic units
Medium
         Hydraulic Conductivity
     Low                 Very Low
                     Extremely Low
Colluvium
Weathered Nancy
shale, including
the sandstone
marker bed
Sunbury Shale
Ohio Shale
Unweathered Nancy
shale
Lower Farmers
sandstone
Upper part of
Crab Orchard
Formation
       Table 2.  Distribution of discharge for the simulations
                    illustrated in figures 7 and 8.
                            Percent  of Total Discharge

        Upper Farmers  sandstone      Sunbury Shale       Bedford Shale
 Figure  7
 Figure  8
      60
      66
     37
     23
   3
   11

-------
                                   206
 unsaturated conditions may  exist within as many as four hydrogeologic
 units:   the Nancy  shale,  Farmers sandstone, Sunbury Shale, and the Ohio
 Shale  (Papadopulos and Winograd, 1974).   Isolated lenses of saturation
 bounded above  and  below by  unsaturated conditions can occur in sequences
 of rock which  have layers of contrasting  hydraulic conductivity.  These
 lenses  of saturation, often referred to as "perched" groundwater, form
 whenever a layer of relatively  low conductivity (a confining bed) is
 unable  to transmit enough water to maintain complete saturation throughout
 an underlying  layer of higher conductivity which drains laterally.  Seepage
 outflow along  the  hillsides tends to drain the more permeable layers and
 create  a favorable situation  for the occurrence of perched ground water
 zones,  separated by drained unsaturated zones.

      In theory,  saturated-unsaturated flow models (e.g., Freeze, 1971)
 can accurately simulate the groundwater flow field for a hydrogeologic
 system  containing  perched groundwater.   At Maxey Flats, however, the lack
 of data, especially with  regard to the  hydraulic properties of the
 unsaturated zones,  precludes the use of a  complicated saturated-
 unsaturated flow model.  An obvious alternative is to treat each perched
 groundwater lense  as a distinct, saturated flow system.  The entire
 groundwater system  can then be modeled  by  determining the flow field for
 each of the perched  systems sequentially,  from top to bottom, using the
 discharge from the  basal  confining  bed  in  one system as recharge to the
 underlying system.

     It is  assumed  here that the transition from saturated to unsaturated
 conditions  occurs at the  base of the confining bed,  where the pressure is
 assumed to  be atmospheric.  It can  be shown (Zaslavsky, 1964) that the
 pressure  in the  lower part of the confining bed will  actually be slightly
 less than atmospheric,  and that the lowermost part of the confining bed
 can even  be unsaturated in certain  situations.  For the purposes of this
 study,  however, the error  due to neglecting these additional  complications
 is small compared to the other uncertainties  in the analysis.

     It should  also be  noted that what  is  interpreted as an unsaturated
zone in a confined  layer may actually be a zone saturated at less-than-
atmospheric pressure.  In  such a case,  a well  completed in a rock unit
saturated partly  with water  at less-than-atmospheric pressure and partly
with water at greater-than-atmospheric  pressure would accumulate water only
from that part  of the unit at greater-than-atmospheric pressure.   The
water level in  the  well could therefore be far below the top of the
 saturated unit.  For purposes of brevity,  the upper part of such a unit
will be referred  to in  this  report  as unsaturated, but it is understood
that the  "unsaturated"  zone  may actually be saturated, but partly at less-
 than-atmospheric  pressure.

     For an isolated hill  such as Maxey Flats, the potential  for a perched
water table to  exist within  a confined  layer  increases as the contract
 in hydraulic conductivity  between layers increases.   Thick layers are also
more likely to  contain  unsaturated  zones simply because, they require a
greater inflow from the overlying confining bed to stay completely

-------
                                   207


 saturated.   Furthermore,  because discharge may occur through seepage
 faces  in each  layer,  inflow  from above to successively deeper layers
 decreases;  consequently,  deep  layers are more likely to contain unsaturated
 zones  than  are shallow layers  of equivalent conductivity and thickness.

     The Ohio  Shale possesses  many of the characteristics important for
 the  development of unsaturated conditions; it is thick, has adequate
 permeability,  and is  located at a low stratigraphic position in the hill.
 There  is, in fact, evidence  that water table conditions probably do exist
 within the  Ohio Shale.  Three  wells open to nearly the entire 185-foot
 thickness contain less than  50 feet of water.  If the shale were completely
 saturated at greater  than atmospheric pressure, these low water levels
 would  indicate a large vertical head gradient, and therefore a large com-
 ponent of vertical flow.  However, the Ohio Shale is underlain by the
 poorly permeable upper part of the Crab Orchard Formation; consequently,
 flow in the Ohio Shale (especially the lower part) should be nearly hori-
 zontal due  to  the high contrast in hydraulic conductivity.  The most
 plausible explanation for the  low water levels appears to be that the upper
 part of the Ohio Shale is unsaturated.  At least two wells drilled in the
 Ohio Shale  encountered isolated zones of water some 30 to 50 feet above
 the ultimate static water levels, which could indicate that there is
 actually more  than one saturated zone within the Ohio Shale.  There is no
 way to check this hypothesis, however, because all of the wells in the Ohio
 Shale are open  to the entire 185-foot thickness.

     The existence of perched water tables within the upper Farmers
 sandstone and  the Sunbury Shale is problematical.  Dry wells finished in
 these units probably fail to intersect water-bearing fractures, and others
 with extremely  low water levels may intersect fractures only at deep levels.
 A number of wells do not have relatively high water levels, which suggests
 that the upper  Farmers sandstone and the Sunbury Shale could be completely
 saturated.   Due to its lower stratigraphic position in the hill, the Sun-
 bury Shale is  the more likely of these two units to contain an unsaturated
 zone.

     Based on the preceding analysis, Maxey Flats could be considered to
 consist, conceptually, of two saturated flow systems:  an upper system
 which extends from just above the sandstone marker bed down to the base of
 the Bedford Shale, and a lower system which occupies the bottom part of the
 Ohio Shale.   Only the upper flow system is considered in detail because
 water level  data is insufficient to warrant a model analysis of groundwater
 flow in the Ohio Shale.
                          Mathematical Model

     Idealized, two-dimensional vertical cross-sectional flow models are
useful  tools for studying general conceptual features of complex flow
systems.  The governing equation for two-dimensional, steady-state ground-
water flow is:

-------
                                  208
     *\       J4 h    Ji       H h
     aF * zz aT'   ay * yy ay'


where  K  and KZZ are the hydraulic conductivities in the horizontal and

vertical directions, respectively, and h is the hydraulic head.  For this
report, equation 1 was solved numerically using a finite difference
approximation.

     A representative vertical cross section was taken through the central
part of the hill and oriented perpendicular to the topographic divide
(line  B'-B in figure 1).   Boundary conditions for the upper flow system
are summarized in figure  6.   The left boundary is considered to be a
no-flow boundary corresponding to a groundwater divide near the center of
the hill.  Heads along the lower boundary are set equal to the elevation
of the  base of the Bedford Shale.  Heads along the water table in the
weathered Nancy shale are assumed to vary linearly from a value cor-
responding to the water level  in well UB-1A near the center of the hill to
a value equal  to the elevation of the top of the sandstone marker bed at
the edge of the upland surface.   The seepage boundary along the hillside
is slightly more complicated.   As a first approximation, the head at each
point along the hillside  was  set equal to its elevation.  Equation 1 was
then solved for the  head  distribution.  Occasionally, some sections of
the hillside showed  an anomolous horizontal component of flow back into
the hill, which indicates that the flow conditions within the system are
unable to suport the heads specified along those parts of the hillside.
These "back flow"  regions were changed to be no-flow boundaries, and
equation 1  was  solved again.   This procedure was repeated until no more
anomolous "back flow" regions  occurred.  An iterative approach of this
type is frequently used to model steady-state flow in systems which contain
seepage faces  (Freeze and Cherry, 1979).


                               DISCUSSION

     Figures 7  and 8 present  the results of simulations for two different
conductivity distributions which reproduce the important features of the
head distribution in the  upper flow system at Maxey Flats.  The flow system
is characterized by  nearly vertical flow in the three major zones of low
hydraulic conductivity (the unweathered Nancy shale, the lower Farmers
sandstone and Henley shale, and the Bedford Shale) and essentially hori-
zontal flow in  the upper  Farmers sandstone and the Sunbury Shale.  The
presence of the unsaturated zone below the Bedford shale assures a large
average vertical head loss in  the upper flow system.  In addition, the
simulations demonstrate how a  well such as 14E, located near the center of
the hill and finished primarily in the upper Farmers sandstone, can have
a water level  far above the top of the upper Farmers sandstone, whereas
similar wells located around  the perimeter of the hill often have much lower
water levels.

-------
                         SPECIFIED HEAD, h = h (y)
                         	A	
   -c -c
   *O|iO



     o
    O
    z
z
I,
                                                                                      SEEPAGE FACE

                                                                                        BOUNDARY
ro
o
                                          SPECIFIED HEAD

                                               h = 0

                   Fig.  6.   Schematic diagram illustrating boundary conditions for the

              groundwater flow model.

-------
      500
      1000
       100
      500
                                                Weathered Nancy Shale
                                                                Sandstone Marker Bed
Q W
ZQ
Otr
oo
  Ul
  m
li
cc
LU
  10
  10
1000
                                                                                Unweathered Nancy Shale
                                                                                    Upper Farmers Sandstone
Lower Farmers
  Sandstone
Henley Shale




    Sunbury Shale






     Bedford Shale
                    ro

                    o
                                                   SCALE

                                        VERTICAL EXAGGERATION - 10X
                                        CONTOUR INTERVAL = 10 FEET
                                      DATUM = BASE OF BEDFORD SHALE

                       Fig.  7.   Simulated head distribution for  the relative conductivity

                  distribution, 500: 1000: 100:  500: 10:  10:  1000:  1.

-------
                                 Weathered Nancy Shale
                                                Sandstone Marker Bed
                                                         Unweathered Nancy Shale
                                                                 Upper Farmers Sandstone
                                                                   Lower Farmers Sandstone
                                                                                            i
                                                                     Henley Shale



                                                                          Sunbury Shale
                                                                          Bedford Shale
                                               FEET
                               SCALE
                    VERTICAL EXAGGERATION = 10X
                     CONTOUR INTERVAL = 10 FEET
                  DATUM = BASE OF BEDFORD SHALE

     Fig.  8.   Simulated head distribution for the relative conductivity
distribution, 125: 250: 25: 250: 2.5:  2.5:  250: 1.

-------
                                 212
     The shaded areas in figures 7 and 8 indicate regions  where the
predicted head is less than the elevation head;  that is,  regions where
the pressure is less than atmospheric.  Because  rocks tend to become
unsaturated when the pressure falls more than a  few feet  below atmospheric
pressure, regions of sub-atmospheric pressure predicted by these simula-
tions can be interpreted as an indication that unsaturated conditions
could exist even though the model used in this analysis has no way of
accounting for unsaturated conditions.

     Figures 7 and 8 also illustrate two of many possible distributions
of unsaturated zones.  In figure 7, the upper flow system is completely
saturated except, perhaps, for a small part of the Sunbury Shale and the
overlying confining layers near the hillside.  It is probably reasonable
to assume that most of the more permeable units are unsaturated to some
extent near the edge of the hill.  In figure 8, the head everywhere in
the Sunbury Shale is far below the top of the unit, which implies  that
water table conditions could exist throughout the Sunbury Shale.   All that
can be concluded, however, is that it is at least possible that water
table conditions exist in the Sunbury Shale beneath the entire area of  the
hill.

     Most of the water which enters as recharge to  the upland  surface of
Maxey Flats in the  two simulations discussed above  is  discharged along
the hillside through either the upper Farmers sandstone or the Sunbury
Shale (Table 2).  It  should be  noted  that the discharge distribution  is
a  strong function of  the anisotropy in hydraulic  conductivity.   For most
of the  simulations  made  during  this study it was  assumed  that  the  hori-
zontal  conductivity was  100  times  larger than the vertical conductivity
due to  the  fact  that  fractures  appear to be much  more  continuous  in  the
horizontal  direction  than  in  the  vertical.  A few simulations  were made
which assumed  isotropic conditions (K   = KZZ).   In those cases,  a much
 larger  fraction  of  the groundwater flow  (over 75 percent) discharged  across
 the base of the  Bedford Shale.   However, the  isotropic simulations were
 not able to account for the  large heads  in  the  upper Farmers sandstone.

      The sandstone  marker bed, one of the most  permeable  units at the
 site,  occurs  directly beneath the trenches;  consequently, the possibility
 of radionuclide movement in  the sandstone  marker bed is  of major concern.
 The models presented in this report cannot  provide good,  quantitative
 estimates of flow in the sandstone marker bed,  but they  do emphasize that
 the horizontal component of the hydraulic gradient in the sandstone marker
 bed should be essentially equal to the slope of the upper water table.
 Thus, in order to minimize horizontal flow in the sandstone marker bed
 it is extremely important to maintain low water levels in the trenches over
 the central part of the hill.  In the simulations presented here most of
 the flow in the sandstone marker bed is downward into the underlying
 unweathered Nancy shale.  Field data indicates that there is a significant
 component of horizontal flow of contaminated trench water in the  sandstone
 marker  bed.

-------
                                  213


     This investigation illustrates the difficulty and complexity of
defining the subsurface flow system in a layered,  fractured-rock system
of extremely low permeability.   The basic constraint  in modeling the
system is the shortage of reliable field-measured  parameters  such as the
horizontal and vertical hydraulic conductivities in all  the major layers
as well as the lack of adequate hydraulic head measurements.   The model
itself is capable of incorporating many different  combinations of input
parameters and boundary conditions.  The results presented in  this report
do not necessarily approximate "actual" conditions in  the field;  they
merely typify some possibilities based on qualitative  estimates  of
hydraulic conditions at the site.
                              REFERENCES

Freeze, R. A. 1971, Three-dimensional, transient, saturated-unsaturated
     flow in a groundwater basin, Water Resources Res., v.  7,  pp.  347-366.

Freeze, R. A., and Cherry, J. A., 1979, Groundwater, Prentice-Hall,  Inc.,
     Englewood Cliffs, New Jersey.

Papadopulos, S.  S., and Winograd, I.  S., 1974, Storage of low-level  radio-
     active wastes in the ground:  Hydrogeologic and hydrochemical factors,
     U.S.  Geological Survey Open-File Report 74-344.

Wilson, C. R., and Witherspoon, P. A., 1970, An Investigation of laminar
     flow in fractured rocks, Geotechnical Report No. 70-6, University
     of California, Berkeley.-

Zaslavski, D. A., 1964, Saturation-unsaturation transition in infiltration
     to a non-uniform soil profile, 8th Congr. Int. Soc. Soil Sci.
     Bucharest.

-------
             COMPUTER SIMULATION OF GROUNDWATER  FLOW AT A

              COMMERCIAL RADIOACTIVE-WASTE  LANDFILL NEAR

               WEST VALLEY, CATTARAUGUS  COUNTY,  NEW YORK
                           David E. Prudic*
                        U.S. Geological Survey
                             P.O. Box  1350
                          Albany, N.Y.  12201
                               ABSTRACT

        Commercial low-level radioactive wastes were  buried  in
    trenches excavated in a clay-rich till  of  low  hydraulic  conduc-
    tivity near West Valley, N.Y., from 1963 to 1975.   Groundwater
    from the trenches flows out laterally and  downward  through
    approximately 25 m of till.

        A two-dimensional finite-element computer  model was  used  to
    evaluate factors controlling groundwater flow  in  two  vertical
    sections.  Measured heads in 24 piezometers along these  sections
    were most accurately reproduced by simulating  four  geologic
    units, each internally isotropic but differing  in hydraulic con-
    ductivity to reflect fracturing near land  surface and increased
    consolidation with depth.  The model accurately simulated the
    changes in head that followed removal of water  from one  of the
    trenches.  Specific-storage values required to  calibrate the
    model for transient conditions agreed well with values derived
    from four consolidation tests of till samples.

        Infiltration rates ranging from 3.8 centimeters per  year  near
    swampy areas to 1.5 cm/yr along scraped or smooth,  sloping sur-
    faces were used to simulate observed area! variation  of  head  in
    the till.  Simulations indicated that water flowing laterally
    out of a trench would not intersect nearby intermittent  streams
    as long as the water level in that trench  remained  below the
    trench cover.
*  Present address  U.S. Geological Survey
                    Room 227, Federal  Bldg.
                    705 North Plaza St.
                    Carson City, NV  89701
                                  215

-------
                                 216


                              INTRODUCTION
Location and Physical  Setting

    A State-licensed  landfill  for commercial  radioactive wastes is among
the facilities at  the  Western  New York Nuclear  Service Center, which
occupies 13.5 km2  (3,350 acres)  in  the northern part of Cattaraugus
County, about 50 km south of Buffalo, N.Y.  (fig.  1).   The center is
in a sparsely populated region of the glaciated Allegheny plateau (1)
that is characterized  by rounded hills of Devonian  shale capped with
thin deposits of till  and separated by bedrock  valleys containing as
much as 150 m of drift.

    The climate at the center  is humid content!al with an annual average
temperature of 7°C.  Temperature extremes range from 38°C during summer
to -26°C during winter.  Precipitation averages 100  cm/yr without
distinctly wet or dry  seasons.  During winter,  precipitation is normally
in the form of snow.

    Operation of the landfill began in November 1963  and continued until
May 1975.  The 4-ha (11-acre) landfill contains a series of trenches
excavated in silty-clay till of low hydraulic conductivity and con-
taining scattered pods of silt and sand.  Radioactive wastes were buried
in trenches that are generally 6 m deep, 10 m wide  at the surface, and
180 m long (fig. 1).  Trenches 6 and 7 were set aside for burial of spe-
cial material of high  specific activity  (2).
Purpose and Scope

    This  study was done as part of a national  study  by the U.S.
Geological Survey to determine the principal  factors that control the
subsurface movement of radioisotopes at  several  radioactive-waste land-
fills.  Results of this study have been  used  in  an  investigation begun
in 1975 by the New York State Geological Survey,  as  lead agency under
contract  with the U.S. Environmental Protection  Agency,  and the U.S.
Nuclear Regulatory Agency, to evaluate all  processes of radioisotope
migration at the site.  This report describes  results of computer
modeling  of groundwater flow near the north trenches (fig.  1).

    Test  holes drilled or augered near the  landfill  during 1975-78 were
designed  to provide information about the flow of groundwater as well
as the geology.  Cores were collected and analyzed  to determine the
distribution of radioisotopes and to detect possible movement of water
away from the burial trenches (3,4).  Several  laboratory tests were  run
on selected core samples of the till to  determine hydraulic properties.
Test holes were assigned a letter, and in the  few instances when two or
more test holes were drilled in proximity (1  to  3 m of one another), a
sequential number followed the letter; for  example,  test holes I, 12,
and 13.   (See fig. 1.)

-------
                              217


^
/
{
~J
^n

,
J. NEW YORK
/n
Location map \





















West •
Valley


Cattaraugus
County






•A
fo,.
\




V















—



%,, >..
^^^-" >v
/ i '"v /
f x ,..-(
R •'' *' ***
s^JZ> --,--• T(Z S : Swamp
7 vVr,, / ~*
X «LO ^ -R '•-..-•
/ 2^C
f
^"\
\

1
n) E,
Swamp f*]
FT'_ . .
A-

Ld I*
.JH • D«
\
!/
i
F^
1 r-
Fence / i i
./Special ^*l
/' burial areas i
\n:

Security fence




• '












(








1



Lagoon
i
j i
•u '+L








L

•a


South
trenches












o
o
14
'7'i6
r









0
13
—


1






°
b
—



>

0










5




|





°
b
1 —
B1














0

-





O




0
10
—
o

















3
— —



f





o
c
m"

0








2







1










0
8


A R '
** /
':
•-.
s
>.
1 J4^^J

TJ2 I A
North S V.
trenches •.
f)
$/
jf

I/
."
r "
\ ^
•%>,
n
EXPLANATION
^E
Test hole drilled 1975-78
Q2C
Test hole augered 1973
* 0
Sump pipe in trench
10
Trench number
A A1

Cross-section profile

0 ^5 50 76 METERS


Base from U.S. Geological Survey
	 	 _J Ashford Hoi low, 1 964, 1 :24,000
NOTE: Map based on plane-table survey by R.J. Martin, U.S. Geological Survey, 1975-78

 Figure 1.  Layout of trenches  and  location  of test  holes
             and wells used to  monitor  water-level trends.

-------
                                218
    To determine head  distribution  in  three  dimensions,  each test hole
was designed to contain from 1 to 4 piezometers  3  cm  in  diameter and
having a 30-cm-long screen.  Each screen was  finished in a sand envelope
about 45 cm thick.  At least 3 m of grout  separated piezometers when
more than one was finished in a test hole.   Two  small-diameter tubes (1
cm and 0.5 cm) extended from the screen to land  surface.  The smallest
diameter tubing ended near the bottom  of the  screen,  whereas the larger
ended near the top of the screen.  The small-diameter tubing was used to
minimize the amount of water needed to fill  it and thus  reduce the lag
time between head change in the till and response  in  the piezometer.
The double-tube system was used to  obtain  periodic water samples.

    Piezometers were designated according  to  test-hole number followed
by a number preceded by a hyphen.  The numbers designate individual
piezometers within a test hole, and piezometers  were  numbered sequen-
tially from shallow to deep; for example,  G-l, G-2, and  G-3.

    During this investigation, hydrologic  studies  were limited to the
till for the following reasons:  (a) the trenches  were excavated only in
this till;  (b)  trench cover consisted  of reworked  till,  and (c)
radioisotope migration was confined to the till  (3,4).

    A two-dimensional finite-element model developed  by  Reeves and
Duguld (5)  was  used to study factors controlling groundwater flow in
the till.  The  objective was to simulate groundwater  flow in a cross
section perpendicular to the north  trenches  (nos.  2-5) from test hole G
eastward to Franks Creek (fig. 1) because  these  trenches had had high
Initial  water levels in February 1976  that were  lowered  by pumping in
July 1976.   This lowering produced  a stress  that normally would not have
occurred and allowed for better calibration  of the model.
Acknowledgments

    Mark Reeves (formerly with Oak Ridge  National  Laboratory) assisted
with the initial computer runs and helped adapt the program to the
conditions at West Valley.  Nuclear Fuels Services, the site operator,
provided personnel to monitor radiation levels  whenever Survey personnel
and their contractors were near or on  the burial  trenches; also their
staff answered many questions about burial  procedures and practices.
Stephen Mollelo (New York State Geological  Survey) and Robert Wozniak
(New York State Department of Environmental  Conservation) assisted with
the data collection.
                   GEOLOGIC CONDITIONS  NEAR  LANDFILL

    Surflcial  geologic mapping  (6)  near the  landfill defined the nature
and extent of  the  till in which  the burial trenches are excavated.  The
till, composed predominantly  of  silt and clay,  is a valley fades that

-------
                                219


is widely distributed  in  the  Cattaraugus Creek basin.  It is found below
the level of saddles in the  drainage  divide to the south and must have
been deposited by an ice  sheet  partly suspended in ponded water.

    The till is commonly  covered  by thin gravel  deposits (6) that, at
the landfill, are largely absent.  The till  extends to about 30 m below
land surface near the  trenches  and overlies a bedded lacustrine unit 10
to 20 m thick consisting  of  fine  sand and silt and locally capped by
gravel.  The upper part of the  lacustrine unit is unsaturated (fig. 2).

    Bedrock is as much as 150 m below land surface near the trenches.
Although the geology and  hydrology of the thick  glacial  deposits beneath
the lacustrine unit are largely unknown,  the lacustrine unit probably
provides an avenue for slow  lateral flow to points of discharge in the
bluff along Buttermilk Creek, 550 m northeast of the landfill (3).

    Fourteen till samples from  test holes near the landfill  were ana-
lyzed for particle size.  On  the  average, the samples contained 50 per-
cent clay, 27 percent  silt,  10  percent sand, and 13 percent gravel;
LaFleur (6) reported that 10  to 20 percent pebbles and cobbles is
characteristic at most exposures.  Typically the unsorted till inter-
fingers with a similar till  having a  pebble qontent of less than 5 per-
cent and a matrix containing  tiny blebs  and torn, deformed wisps of
quartz silt; this latter  till constitutes between 20 and 25 percent of
the till thickness (6).  The  unsorted till  also  includes thin layers of
silt, sand, and rarely coarser  sand,  which together constitute about 7
percent of the core footage  logged (3).   Both subunits were encountered
in all test holes (fig. 2).   However,  stratified materials penetrated by
nearby test holes commonly differ appreciably in altitude and(or) litho-
logy, and even where altitude and lithology are  similar, the steep dips
indicated by many of the cores  and exposures seem to negate the possibil-
ity of a continuous subhorizontal layer  that could indicate a widespread
inplace glaciofluvial  deposit.


Distribution of Fractures in  Till

    A network of intersecting horizontal  and vertical fractures charac-
terizes the oxidized upper 2  m  of the till  near the landfill.  Fractures
and root tubes whose surfaces are bordered by firm oxidized till were
recognizable to depths of3mto4.5min cores  from several test holes
next to the landfill (3).  Some vertical  fractures extend downward into
the unweathered till.  Dana  and others (7) noted that the number of
fractures and their degree of openness decreased with depth in excava-
tions near the landfill.  Both  manganese and calcite deposits have been
observed along several fractures  in excavations  and also in cores from
test holes next to the landfill.  Results from a laboratory study (8)
suggests that the theoretical limit to which a fracture could penetrate
in the till is about 15 m.

-------
425-1
Waste-burial
  trenches
D
                                             November 1976-April 1980
                                              	    Piezometer^	I	
375
           LOCATION MAP
           rfL
                           <



                           •

                        LS.LY
                        SG.etc.

                        LY/L/S


                        T(L3%)
Gravel

Silt

Sand
                                                                       EXPLANATION
Y   Clay

T   Till:50%clay,27%silt,
    10% sand, 13% grave I
                                                                                                                                 ro
                                                                                                                                 i i
                                                                                                                                 •• •
    Base from U.S. Geological Survey
    Ashford Hollow, N.Y., 1 964
Mixed materials with increasing abundance
from left to right (example - LS is a si Ity sand)

Interbedded layers; commonly deformed and
dipping at angle to core

Till incorporating torn, discontinuous deformed
layers of silt in percentage  indicated

Dip of interbedded layers as identified in cores;
orientation not known

Log inferred from geophysical data and other holes

Unsaturated; based on examination of core samples
and from neutron-moisture logs
                                                                                         D Test hole (See figure 1;
                                                                                           Interval not cored, or hole
                                                                                           penetrated reworked material
                                                                                           (cover or backfill)
                                                                                           Continuous or frequent cores
                                                                                           of inplace material
                                                                                           Infrequent cores of inplace
                                                                                           material
       Figure 2.   Generalized cross  section from  test hole  G to  Buttermilk  Creek.    Spring
                     may  be  perched  water  of  local  origin;  flow is  probably  less  than
                     1  liter  per minute.    (Modified  from Prudic and Randall,  1979.)

-------
                                   221


                GROUNDWATER FLOW NEAR THE NORTH TRENCHES

    Much  of  the precipitation on the landfill either runs off to nearby
 streams and  gullies or returns to the atmosphere by evapotranspiration
 (3).   The small  amount that infiltrates below the roots of plants moves
 generally downward, even beneath small  valleys bordering the site, as
 indicated by  the distribution of hydraulic head along a section perpen-
 dicular to the  north burial  trenches in February 1976 (fig. 3).  At that
 time,  most of the north trenches were nearly filled with water (3) that
 was generally moving outward and downward from the trenches.  A small
 flow component  toward trench 5 may be inferred from the hydraulic head
 near land surface within 10  m west of the trench (fig.  3).

     Pressure head within the till ranged from near zero to more than 5 m.
 Low-pressure  heads generally occurred beneath steep and(or) smooth
 slopes or where the upper layers of soil  were scraped to allow for
 rapid  and complete runoff.   Pressure heads in the till  were generally
 greater where water had accumulated above the till,  as  in the trenches,
 in natural depressions in land surface  (as near holes J and G in fig.
 1), and in remnants of late-glacial  fluvial  gravel  atop the till  (3).


 Response  of Water Levels in  Piezometers to
 Pumpout of Water from Trenches 2-5

    Water was pumped from trenches 3, 4,  and 5 on several  dates between
July and  November 1976 to lower .the water levels and was pumped again
 between August  and November  1977,  including  trench 2.  The piezometers
 next to trenches 5 and 2 responded to the pumpout (figs. 4 and 5), but
 those  farther away showed no response (fig.  6).

    The decline  of water levels in piezometers near trench 5 was most
 rapid  in  those  finished in  undisturbed  material near the reported alti-
 tude of the trench floor (D-l,  E-2,  and F-2) or in a sandy backfill next
to the trench (E-l)  (3).

    The sudden water-level  increases in piezometers D-2 and F-3 (fig. 4)
 in May and June  1976  were caused by attempts to sample  the piezometers,
whereby nitrogen was  injected at pressures as high as 4.2 kg/cm2 down
the larger of two tubes that extended to  the piezometer screen* which
forced water  up  the smaller  tube to the surface.   Apparently pressures
were large enough  to  deform  the till and  spread it away from the column
of hardened cement grout above the piezometer, in effect increasing the
length of  piezometer.   Because total head in the till declines with depth,
the water  level  rose  when water from shallower depth entered the piezo-
meter.   The subsequent decline in  water level reflects  partial  resealing
of till against  the grout column as well  as  a response  to trench  pumpout.
    Other piezometers were  sampled by suction lift in combination
    with gas pressures  less  than 0.5 kg/cm2.

-------
430-i
         r\\)  Backfill
               Weathered till
         V/X  Till with oxidized fractures
               Unweathered till
    • • • • • * • • •
Assumed water-table
Water level m trench
       414
                                       Equipotential line, altitude, in meters
                                                                                                                    ro
                                                                                                                    ro
390
        Figure 3.   Cross  section A-A1  through north trenches  showing head distribution,
                     February 1976.   (Location of  section is  indicated in  fig.  1.)

-------
 420
        I   I   I   I   I  I   I   I   I   I  |  I   I   I  I   I   I   I   I   1  I   I   I   I  I   I   I
I_J _ L_l _ \ — I
                             I  i   i   I   i  I   I   I _ 1 _ I _ L_l
        FMAMJ  JASONDJ   FMAMJJASONDJ  F  M A  M
                                                                                                   ro
                                                                                                   ro
                                                                                                   CO
 406
Figure 4.  Water-level trends  in  piezometers  finished next to trench 5.
           (Letter and number  identifies  piezometer,  sump refers to well  in trench.)

-------
    418
    416
u
  <412
  O

-J Q
UJ O
cc
LJJ

<   408
    406
              "II   I   '  Trenches 3-5 j'I  '''   '   I   I	1	1	1	1	1	1	1	1	1	T
                         —  pumped  _J       _——•	                    '  ^—•



                                                                Base of piezometer 14-1 (dry)'
                                                                                   (dry)
                                                                                      \
                                                                                         2-1A
         I   I   I   I   I   I   I   I   I   I   I   I  I   I   I   I    I   I   I   I   1   I   I  I   I   I
       JFMAMJJASONDJ   FMAMJJASONDj  FMAM

                       1976                                 1977                       1978
  Figure 5.   Water-level  trends  in piezometers finished next  to trench  2.
               (Letter  and  number  identifies  piezometer,  sump refers to well in  trench.)
                                                                                                           ro
                                                                                                           ro

-------

    420
    418
          I  I	1	1	1	1	1	1	1	1
                                                     I   I	1	1	1	1	1	1	1
   CM

 LLJ O5
 LLJ Q
 LJJ
 +
 DC
    410
    408
                                J2-1
                          G-3
J	I	I	I	I	I	I   I   I   I  I   I  I   I   I   I   I   I   I   I   I  I   I   I   I
        J   FMAMJ  JASONDJ  FMAMJ  JAS   ONDJ  FMAM

                      1976                               1977                      1978
                                                                                                         ro
                                                                                                         ro
                                                                                                         en
Figure 6.   Water-level trends  1n piezometers  finished distant  from trenches  2  through 5.
            (Letter and number  identifies piezometers.)

-------
                                   226


Although the two effects cannot be separately measured,  for purposes of
model calibration, the water-level trends in piezometers  D-2 and F-3
caused by the pumpouts were estimated from measurements  in E-3, which
was  finished at similar depth along the west side  of  trench 5.

     The decline in water level in piezometers next to trench 2  was less
pronounced than that next to trench 5.  Because water level in  trench 2
was  lower than in trench 5,'the decrease  in water  level  was less.  Of
the  piezometers finished next to trench 2, A2-1, finished in a  lens of
layered silt and fine sand that dips from piezometer  R-l toward trench 2,
responded most rapidly (fig.  5).  Piezometers in test hole R, 6 m east
of those in test holes A and A2 (fig. 1), did not  show a decline beyond
the  normal yearly fluctuations (fig. 4).

     Test holes I,  12, 13, and 14 are near the south end  of trench 2,
about 70 m from the sump from which water was pumped  out of trench 2
(fig. 1).  Well  2-1A, driven into trench  2 in June 1976  near test hole
I, had nearly the same water level as the sump until  the sump water was
lowered below an altitude of 414.5 m (fig. 4); however,  continued
pumping from the sump did not cause a decrease in  well  2-1A.  After the
pumpout of trench 2,  the difference between water-level  trends  in piezo-
meters at the north end and those at the  south end probably reflects
differences in water level  within trench  2 itself.
Hydraulic Properties of Saturated and Unsaturated  Till

    Hydraulic conductivity of the saturated  unfractured till  was
obtained from both field and laboratory  studies.   Field tests were done
by withdrawing 200 to 500 ml of water out  of piezometers finished at
depths from 5 to 16 m, then measuring the  water-level  rise in the piezo-
meters until they had stabilized.

    Two methods were used to analyze the data.  The first assumed hori-
zontal flow to the piezometer (9); the second assumed spherical flow
(10).  Average hydraulic conductivity calculated  from 12 tests was 6 x
10'8 cm/s by the first method and 2 x 10'8 cm/s by the second.
Hydraulic conductivity measured in laboratory tests at confining
pressures on seven core samples from depths  of 5  to 16 m produced an
average vertical hydraulic conductivity  of 2 x 1Q-8 cm/s, which compares
well with values obtained from field tests that assumed spherical flow.
Fickies and others (8) reported anistropy  with 40  times greater horizon-
tal  than vertical conductivity in two core samples, although laboratory
tests of nine core samples suggest little  or no anistropy.  Analyses of
thin  sections of 18 till samples do not  suggest preferential  horizontal
layering of the clay  grains within the till.

     Consolidation tests were run on four samples  to determine the
effects of  increased  confining pressures on  hydraulic conductivity.  In
general, the vertical  hydraulic conductivity decreased by about 40 per-
cent as confining pressures increased from near atmospheric to 7
kg/cm2 (equivalent to pressure at a depth  of 30 m), which suggests that

-------
                                 227
increased  overburden pressures may  reduce  hydraulic conductivity of  the
till  (fig.  7).   However, much of this  decrease occurred at confining
pressures  between near atmospheric  and 1 kg/cm2.   A decrease of less
than  10 percent was observed between pressures of 1 to 3.5 kg/cm2
(equivalent to  pressure at a depth  of  5 to 16 m), and at pressures from
3.5 to 7 kg/cm2,  the decrease averaged 13  percent.

    Field  tests of seven piezometers finished in  small lenses of silt
and sand yielded an average hydraulic  conductivity  of 2 x 10'6 cm/s;
their range of  values (7 x 10~8 to  6 x 10-6  cm/s) was considerably
greater than the range for till.  (Weighted  average hydraulic conduc-
tivity of  the till  and sorted materials was  calculated to be 8 x
10-8 cm/s.j

    Porosity of the till was determined by measuring dry unit weight and
specific gravity  of 28 core samples as  described  by Morris and Johnson
(11).  Average  porosity was 32.4 percent with a standard deviation of
4.7 percent.
   85
   LU
   cc
   EC.
   £10
   CO
   3
   co
   CO.
   LU
   £20
    25
                                          o

                                          a
            A

           O
                                        Hole E - 10.35to 10.5m
                                         Hole M- 15.8to 16.0m
      Hole D - 5.8to6.0m
                                        Hole C2- 12.35to 12.5m
                          J.
J_
      0246          8        10

       VERTICAL HYDRAULIC CONDUCTIVITY, IN CENTIMETERS PER SECOND X 10
                              12
                             ,-8
  Figure  7.   Relationship of hydraulic  conductivity to confining
              pressures as determined  from consolidation tests.

-------
                                 228
      Because the model simulates unsaturated as well  as  saturated con-
  ditions, functions relating volumetric moisture  content and hydraulic
  conductivity to soil-moisture tension were needed.   Both functions were
  assumed to be nonhysteretic and were calculated  from pore-size-
  distribution curves derived from mercury porosimeter tests (12) of seven
  core samples of till.  The curves relating soil-moisture content to
  soil-moisture tension were calculated on the assumption that the
  capillary pressure of mercury is 5.2 times that  of water in air (12).
  Curves relating hydraulic conductivity to soil-moisture tension were
  calculated from the soil-moisture curves with an  equation by Millington
  and Quirk (13);  both  curves used in the model are depicted in figure 8.
  Saturated hydraulic-conductivity values calculated from the soil-
  moisture tension curve and an equation by Marshall (14) averaged
  6 x 10-8 cm/s,  which  is reasonably close to measured values.
                                                              125
                                                                 u
               -20.000      -15.000      -10.000      -5000
                 PRESSURE HEAD, IN CENTIMETERS OF WATER
Figure 8.  Relationship of hydraulic conductivity  and volumetric soil
           moisture content to soil-moisture  tension, as calculated
           from seven mercury porosimeter  tests  of the till.

-------
                                  229


It was not considered necessary  to  then  relate the computed curves to
measured values, as is often  required,  because the till  was nearly
saturated, even with soil-moisture  tensions  exceeding 1,000 cm.

    To simulate transient conditions of groundwater flow at the burial
site, data on storage within  the  geologic  materials are  needed.  The
storage equation used in the  model  uses  the  modified terms of compressi-
bility for water and the solid matrix  (5); the modified  term of
compressibility of water was  obtained  by multiplying the compressibility
of water at 10°C (15) by the  density of  water  and by gravity.  The value
used in model simulations was 4.3 x 10~8 cnr*.

    Specific storage, which is the  sum  of  the  modified terms of
compressibility of the water  and  the till, was  calculated from the four
consolidation tests described previously.  The  specific  storage value of
the till, which averaged 8 x  10-° cm'1,  consistently decreased with
increasing confining pressures, similar  to the  hydraulic-conductivity
values of the till.  Specific-storage  values ranged from 16 x 10~6 cm'1
at a depth of 5.8 m to 2 x 10~° cnr* at  a  depth  of 16 m.   These values
are essentially the modified  term of compressibility of  the matrix
because the value of water compressibility is  much lower.

    The specific storage values obtained from  the consolidation tests
are in the lower range of values  typical of  a  medium-hard clay as pre-
sented by Domenico and Mifflin (16) and  an order of magnitude less than
intergranular values presented by Grisak and Cherry (17)  for a clay-rich
till in Manitoba, Canada.
         COMPUTER MODELING OF GROUND-WATER  FLOW  PERPENDICULAR
                           TO NORTH TRENCHES

    The groundwater flow model developed  by Reeves and Duguid (5) was
chosen for the study at West Valley because it is  capable of simulating
(a) saturated and unsaturated flow with a moving water table, in cross
sections, (b) groundwater outflow to  land surface, and (c) temporal
variation in inflow from precipitation.   Other reasons include (d) no
restrictions on the number of units with  varying hydraulic properties;
(e) no maximum difference between hydraulic-conductivity values of model
units; (f) use of constant infiltration rate  from  land surface during
steady-state simulations; and (g) availability of  a solute model that
was developed and coupled to the flow  model (18).

    Reeves and Duguid (5) stated that  the equations representing flow in
saturated and unsaturated porous media consist of  equations of (a) con-
tinuity of the water; (b) continuity  of the matrix; (c)  motion of the
water; (d) consolidation of the matrix, and (e)  compressibility of the
water and combined these equations into one:

                                                                      (1)

-------
                                  230
Where   e  =  volumetric  moisture content,  dimension!ess  ratio;
        n  =  porosity, dimensionless ratio;
       3'  =  modified coefficient of compressibility  of  the water, L"1;
       a1  =  modified coefficient of compressibility  of  medium,  L'1;
      •4&  =  change in moisture content with  respect  to  pressure head, L"1
       3J1  =  change in pressure head with  respect  to  time,  LI"*;
       3t
       ](  =  hydraulic conductivity tensor, LT'1;
       h  =  pressure head, L;
       z  =  vertical  distance, L;
       x  =  horizontal  distance, L.
This equation reduces to Richard's equation  (19)  for unsaturated flow
and reduces to the elastic storage equation  for  saturated flow (5).
Because the model solves for total head,  both  equations are automati-
cally satisfied.
    The hydraulic-conductivity tensor is  dependent  on  pressure head, as
shown in  the following equation:
              K = Ks •  Kr(h)                                          (2)
Where K     = hydraulic-conductivity tensor, IT'1;
      Ks    = saturated-hydraulic conductivity tensor, LT'1;
      Kr(h) = relative hydraulic conductivity  as function of
              pressure head, dimensionless.

Steady-State Simulations
    A finite-element grid was developed  to  represent the  cross section
A-A' from hole G east to Franks Creek  (fig.  1).   The following condi-
tions were  established for steady-state  simulation  of  flow in  February
1976 and  February 1978:
  (a) No-flow boundaries were assigned  to both sides of the section
      because hydraulic gradient  is  predominantly vertical.

-------
                                  231
  (b) The silt was modeled at  a  pressure  head of zero at the base of
      the till because the silt  underlying  the till  is generally
      unsaturated.

  (c) Infiltration from precipitation  was at a steady rate sufficient
      to saturate the till to  land  surface;  however,  the rate was
      decreased in some model  runs  to  create local  unsaturated conditions.

  (d) A horizontal hydraulic conductivity of 8 x 10~8 cm/s was assigned
      to the unweathered till  unit;  this  is  a weighted average hydraulic
      conductivity of the till and  the  sorted material.

  (e) Constant head within the trenches was  made equal  to observed head
      in February 1976 or February  1978.  (This assumption of constant
      head in the trenches is  reasonable  for periods  of  2 to 3 months
      because the water level  in these  trenches did  not  generally change
      more than 5 cm in a month.)

    Several  steady-state simulations were run to evaluate the relative
significance of am'sotropy, variations  in hydraulic  conductivity caused
by fracturing and increased confining  pressures within the till, and
infiltration from precipitation.  Simulations became  progressively more
complicated, starting with a single  isotropic unit  but eventually con-
sidering as many as five units with  varying  degrees  of am'sotropy.  The
mean absolute departure of simulated heads  from heads observed at 17
piezometers in the plane of the  section was  calculated for each model
run as a general  index of degree of ,fit;  the results  are given in table 1.

    Simulations that incorporated a  single  isotropic  unit yielded the
same pressure heads regardless of hydraulic  conductivity (table 1, case
1); however, changes in computed inflow and  outflow  rates were directly
proportional to any change in  hydraulic conductivity.  Simulations that
incorporated a single anisotropic unit  produced better fits (decreased
the mean absolute departure),  and best  results were  obtained when hori-
zontal hydraulic conductivity was 100  times  the vertical value (table 1,
case 2).  Simulating several  shallow isotropic layer(s)  at a hydraulic
conductivity greater than 8 x  10~° cm/s to  represent  the abundantly
fractured weathered till  and the fractured  unweathered till also pro-
duced better fits than a single  isotropic unit, and  best results were
obtained when the hydraulic conductivity  was between  10  and 100 times
higher in the weathered till  than in the  unweathered  till (table 1, cases
3 and 7).  Simulating am'sotropy of  the layered units reduced the mean
absolute departures (table 1,  cases 4,  5, 6, and 8).

    Most simulations incorporated an infiltration rate large enough to
saturate the till  to land surface,  but  infiltration  rates in a few simu-
lations were reduced enough to cause the  uppermost  nodes of the model to
become unsaturated.   When am'sotropy was  included,  the result was a
slight lowering of heads in most nodes  in the section.  In isotropic
simulations, large changes in  heads  occurred in the  nodes beneath points
where infiltration was the only  source  of water. For example, reducing
infiltration rate near model  holes  D and  I  did not  greatly affect heads

-------
                                 232
     Table 1.   Differences  between  heads  observed  in  February  1976
               and simulated  heads  for  assumed conditions.
                          "                              Mean absolute
                                                         departure from
                                                         observed heads
                                                        of February 1976
Case              Conditions                             (centimeters)
      one isotropic  unit
          (a)   Kx =  Kz  =  3.0x10-8  cm/s                        201
          (b)   Kx =  Kz  =  3.0x10-7  Cm/s                        201

      one unit  with  anisotropy
          (a)   Kz =  0.1 Kx                                     161
          (b)   Kz =  0.01  Kx                                    114
          (c)   Kz =  0.001 Kx                                   116

      two isotropic  units (weathered  and
            unweathered till)
          (a)   K  (weathered) =  10  K  (unweathered)              123
          (b)   K  (weathered) =  100 K  (unweathered)            127

      two anisotropic units  (weathered and
            unweathered till)
          Kx (weathered)  =  10 Kx (unweathered)
          (a)   Kz =  0.1 Kx  in each unit                        93
          (b)   Kz =  0.01  Kx  in  each  unit                       74

      two anisotropic units  (weathered and
            unweathered till)
          Kx (weathered)  =  100  Kx (unweathered)
          (a)   Kz =  0.1 Kx  in each unit                        90
          (b)   Kz =  0.1 Kx  (weathered)
               Kz =  Kx  (unweathered)                           109
          (c)   Kz =  0.01  Kx (weathered)
               Kz =  0.01  Kx (unweathered)                      92

      two anisotropic units (weathered and
            unweathered till)
          Kx (weathered)  = 10 Kx  {unweathered)
          Kz = 0.01  Kx in each  unit
          (a)  simulated sand and silt lenses  near
               D, I, and J                                      65
          (b)  as above  plus increased heads  at
               base of model near G                            60

-------
                                  233
     Table  1.   Differences between heads observed in February 1976
               and simulated heads for assumed conditions. (Continued)
                                                         Mean absolute
                                                         departure from
                                                         observed heads
                                                        of February 1976
Case               Conditions                            (centimeters)
7   Three isotropic units (weathered, unweathered
      with fractures and unweathered till)
      (a)  K (weathered) = 10 K  (unweathered)
           K (unweathered w/fractures) = 5 K  (unweathered)     116
      (b)  K (weathered) = 100 K (unweathered)
           K (unweathered w/fractures) = 10 K  (unweathered)    121

8   Three anisotropic units  (weathered, unweathered
      with fractures and unweathered till) with Kx the
      same as noted in case  7(a) above
      (a)  Kz = 0.1  Kx  in each  unit                             96
      (b)  Kz = 0.01 Kx  in each  unit                             93
      (c)  Kz = 0.01 Kx  in each  unit plus  simulated
           sand and silt lenses  near D, I,  and J;
           increased head at base  of model  near G,
           and lower infiltration  near holes  D and  I             57
      (d)  same as 8(c)  except Kz  = 0.1 Kx for each  unit        53

 9   Four  isotropic units (weathered,  unweathered  with
      fractures,  and 2  unweathered till units to
      compensate  for overburden  pressures)
      K  (weathered) =  10 K  (unweathered)
      K  (unweathered w/fractures)  = 5 K  (unweathered)
      (a)  K (lowest unweathered unit)  =  0.5  K (unweathered)   120
      (b)  K (lowest unweathered unit)  =  0.7  K (unweathered)   104
      (c)  K (lowest unweathered unit)  =  0.85 K  (unweathered)  107
      (d)  K (lowest unweathered unit)  =  0.7  K (unweathered)
           plus  simulated  sand and silt  lenses near D, I
           and J; and  reduced infiltration near D and I         31

           KSHowest unweathered unit)  =  0.85 K (unweathered)
            except beneath  hole G where
           K (lowest unweathered unit) =  0.7 K (unweathered)    21

-------
                                  234
 In  nodes  below  the  bottom of  the  trenches,  whereas  it did near holes H
 and J.   Infiltration  rates  near swampy  areas  were adjusted to saturate
 the till  barely to  land  surface.  The final  infiltration rate used in
 the computer  simulations  was  about 3.8  cm/yr  near swampy areas, whereas
 the model  infiltration rate near  hole I was 1.5  cm/yr.

     In  all  model  runs, simulated  heads  in  the deepest piezometer at hole
 G were  lower  than observed heads.  This would be expected if the
 lacustrine  silt or  sand beneath the till  south and  east of the trenches
 became  thinner  to the west, pinching out  near hole  G,  or that the unit
 became  finer  toward the west.  Because  this unit seems  to be unsaturated
 beneath the trenches, such a  thinning out  or  fining of  the sediments
 would indicate  saturated  conditions near  hole G  at  the  altitude of 391
 m.   In  this case the model boundary below  hole G would  be represented by
 pressure heads  greater than zero.  Although this hypothesis is unproved,
 it  seems plausible from records of test holes in the area.  Alternatively,
 the till at depth beneath hole G  could  have a slightly  lower hydraulic
 conductivity.  Pressure head  in the piezometers  in  hole G was reasonably
 simulated assuming isotropic  conditions and assuming hydraulic conduc-
 tivity of the till below 400  m altitude beneath  hole G  to be 30 percent
 less than that of the unweathered till.  Although neither assumption has
 been proved, the latter seems more reasonable than  assuming slightly
 positive heads at the base of the till  near hole G.

    The best simulations with anisotropy  included three units (the
weathered zone,  the unweathered till with  fractures,  and the unweathered
 till), and also  included localized silt and sand lenses near piezometers
D-l, J-l, and 12-1.   In these tests the weathered till  unit had a horizon-
 tal  hydraulic conductivity 10 times higher  than  the unweathered till
with oxidized fractures,  whereas  the localized silt and sand units had
 values 5 times higher than the unweathered  till.  The  horizontal  hydrau-
 lic conductivity in all three till units was  10  times  higher than the
 vertical; the localized silts and sand  lenses were  assumed to be isotro-
pic.  The mean absolute departure in this  run was 53  cm (table 1, case 8d).

    The best simulations assumed  four isotropic  till  units with the
 horizontal hydraulic conductivities described above,  plus an additional
 unit below the 400-m altitude beneath the  trenches  and  396 m beneath
 Franks Creek.  The hydraulic  conductivity  of  this unit  was 6.9 x 1Q-8
 cm/s, or 15 percent less than that of the  unweathered  till  except below
 hole G, where the value was 5.6 x 10~8 cm/s.   The purpose of this lower
 unit was to represent the reduction in hydraulic  conductivity with
 depth, either from greater overburden pressure or the lower percentage
 of  disturbed silt and sand layers below 400 m, as suggested by some of
 the  test borings.  A four-unit model was also hypothesized by field
 observations in test trenches dug near the  burial ground (7).

    The above simulations included localized  silt and sand lenses near
 piezometers D-l, J-l, and 12-1.   The mean  absolute  departure in this run
was 21 cm.  Figure 9 depicts  the  nodes, boundary  conditions,  and  the
 relative hydraulic-conductivitiy  values used  in  the  best-fit model  simu-
 lations.

-------
430—1
                             Trenches: Fixed head for nodes below
                                    trench water level; seepage
                                    nodes above water level
                                                                             Constant infiltration rate
                                                                              along land surface
                                        {  1    i   j    |  LLK=5      0-00-0	O
                                        '  1    I   !    ! 90   o     o oo    _
                                      _-d  o-__A   6—6                  •   °
!>—A  o-
   o  o   o  o
                           oo   oo
                           oooo
                              Node associated with piezometer9
                            •"o   oo   oo  o   o'ooo
                                                                                        ro
                                                                                        to
                                                                                        en
390
        Figure 9.   Cross  section A-A' through north trenches showing arrangement of  nodes,
                     boundary  conditions, and relative hydraulic  conductivities  used to  best-fit
                     model  simulation.

-------
                               236
    Table 2 compares simulated values for best computer  simulations
with observed values for both anisotropic and isotropic  conditions for
both February 1976 and February 1978.  The isotropic  four-layer simula-
tions produced smaller average departures than the  anisotropic three-
layer simulations and also were qualitatively more  reasonable in at
least two respects:  (a) they yielded negative heads  at  piezometers 1-1
and 14-1, which were dry on both dates, whereas  the best anisotropic
simulations yielded positive heads, and (b) they  also yielded about the
same heads on both dates in the piezometers at hole G,  in agreement with
observed conditions, whereas the best anisotropic simulations yielded
significantly lower heads at hole G in February  1978  in  response to the
lower water level simulated in the trenches.

    The distribution of heads within the till during  February 1976 and
1978, based on the best steady-state computer simulations,  are shown in
figures 10 and 11.  From the simulation of February 1976, water moving
from trench 5 would flow outward only 2 to 3 m before it would move
downward; the same pattern Is indicated east of  trench  2.  In February
1978, the outward movement of water was less than 2 cm/yr.   Average com-
puted vertical  flow rate beneath the trenches was 3 cm/yr in February
1976 and less than 2 cm/yr in February 1978.  These model simulations
indicated a groundwater discharge to land surface west of hole G and on
the lowest part of the slope near the east side  of  the model; then model
discharges averaged 100 cm^/d (0.05 gallons per  day)  per m^ of surface
but occurred only over small areas and were derived largely from flow
through the weathered till.  These rates could not  be verified in the
field because the streams bordering the landfill  are  small  and during
summer are dry much of the time.  In addition, streamflow could not be
measured to the accuracy required to determine such small seepage rates.
However, near the computed areas of discharge are marshy areas that
could be the result of shallow groundwater seepage.

    Although the steady-state model simulations  reproduced observed
heads in the till from hydraulic properties obtained  from field and
laboratory measurements, the simulations could not  independently define
the hydraulic conductivity of the till because the  same  heads could be
simulated by changing all values of hydraulic conductivity and infiltra-
tion by a proportionate amount.
Transient-State Simulations

    Transient conditions were simulated  in an  attempt to reproduce the
observed drawdowns in piezometers finished in  the  till  after the pumpout
of water from trenches 3-5 in the summer of 1976.   Initial  calibration
of the flow model was obtained by simulating the drawdown in piezometer
D-l after the pumpout of trench 5 from July 7-17,  1976.

    Figure 12 compares observed water levels in piezometer D-l with
simulated values resulting from different specific-storage values.  The
best fit was obtained from a specific-storage  value  of 16 x 10~6 cm~l.

-------
                            237
Table 2.  Heads observed in February 1976 and February 1978
          compared with results of best-fitting computer
          simulations for isotropic and anisotropic conditions
               [All values are in centimeters]






Piezometers
February 1976-
6-3
6-2
G-l
H-l
D2-1
0-2
D-l
13-1
1-2
12-1
1-1
14-1
J4-1
J5-1
J-l
J2-1
Sl-6
S2-1
February 19/8-
G-3
6-2
6-1
H-l
D2-1
D-2
0-1
13-1
1-2
12-1
1-1
14-1
J4-1
J5-1
J-l
J2-1
Sl-1
S2-1
Mean absolute





Observed
heads

382.
374.
320.
178.
47.
348.
562.
68.
169.
248.
dry
dry
214.
84.
545.
235.
260.
178.
387.
386.
327.
225.
47.
246.
285.
39.
137.
224.
dry
dry
221.
75.
561.
260.
261.
181.
departure
Calculated
heads from
best
computer
simulation
with
anisotropy
(table 1:
case 8,
part d)

254.
276.
291.
187.
46.
308.
629.
142.
218.
220.
91.
144.
202.
45.
537.
166.
221.
140.
179.
221.
277.
177.
26.
172.
322.
94.
155.
152.
67.
32.
200.
45.
526.
158.
221.
138.

Standard deviation





Absolute
departure
from
observed
heads

128.
98.
29.
9.
1.
40.
67
74.
49.
28.
>91.
>144.
12.
39.
8.
69.
39.
38.
208.
165.
50.
48.
21.
74.
37.
55.
18.
72.
>67.
>32.
21.
30.
35.
102.
45.
43.
1976=53':
1978=62.
1976=41.
1978=51.
Calculated
heads from
best
computer
simulation
with
isotropic
conditions
(table 1:
case 9,
part e)

359.
330.
307.
187.
47.
359.
560.
89.
173.
180.
-79.
-96.
219.
46.
524.
288.
262.
152.
360.
329.
304.
188.
15.
220.
268.
73.
152.
158.
-22.
-154.
220.
48.
520.
286.
262.
152.





Absol ute
departure
from
observed
heads

23.
44.
13.
9.
0.
11.
2.
21.
4.
68.
™ •
-.
5.
38.
13.
57.
4.
25.
27.
57.
23.
37.
32.
26.
17.
34.
15.
66.
—
-
1.
27.
41.
26.
1.
29.
1976=21.
1978=29.
1976=27.
1978=17.

-------
    430-i
WEST

A





-3.8-
                                                                                                              EAST

                                                                                                                 A1
< 05 420-
5/2 < 410-
   -
Q LU
DQ
I- O
  O400 —
    390-
                 -2.9-
           -1.5-t
                                                                  Infiltration rate, in centimeters per year


                                                            -1.5	\-3.Q4	3.8	1	1.5-
                                        Trenches
                                      I
                                   I   I       I
                              VL-J   i	I  i—
                                                 • 414.0
                                       Observed head in piezometer
                                                   D
      Backfill


EvX-1  Weathered till


V/J  Till with oxidized fractures

I    I  Unweathered till
                                                       • • • • •% • • *
                                                   Computed water-table
                                                   Water  level in trench

                                                       	414	

                                             Equipotential  line, altitude, in meters
                                                                                               50 METERS
           Figure  10.  Cross  section  A-A'  through  north  trenches  showing  computer-simulated
                         distribution  of  head  in February  1976.

-------
         WEST

         A
                                                                                                   EAST

                                                                                                      A1
    430-1
         -3.8-
< 0,420-

02

I- LL
<£<410-
uj U
LLI JZ
Qoi
DQ
to
l_ UJ
-JO400-
    390-
   -2.9-
           -1.5H
                   Infiltration rate, in centimeters per year

              -1.5	H3.0H	3.8	1	1.5-
                                                                                                              -3.8-
                                        Trenches
                                  -J   I	I   L..J   L..J
          -406

           -404

               -402

      Backfill

1<1^  Weathered till

V/\  Till with oxidized fractures

1    |  Unweathered till
          • 414.0
Observed head in piezometer
            D

        Test hole
        • • • • y* • • • •
   Computed water-table
-402-
                                                    Water level in trench

                                                        	414	

                                              Equipotential line, altitude, in meters
                                           2.5
      50 METERS
                                                                                                                                  PC
                                                                                                                                  u
                                                                                                                                  u
          Figure  11.   Cross section A-A1  through  north  trenches showing computer-simulated
                          distribution  of  head in  February  1978.

-------
                               240
However, the simulation assumed that water was  instantaneously withdrawn
from trench 5 to lower the water-surface altitude  from 418.5 m to 416.5 m,
whereas this decrease actually occurred over  a  10-day  period.  An
instantaneous drop in trench 5 water level would have  caused the water
level in piezometer D-l to decline more rapidly than  it did (see curve
in fig. 8)); thus, a more representative value  of  specific storage would
be between 8 x 10'6 and 12 x 10'6 cnr1, which agrees well  with the
average value of specific storage obtained from laboratory tests.  In
simulations over longer periods, the best fit to the water-level de-
clines in the deeper piezometers (D-2, F-3, E-3) were  obtained when the
specific-storage value of the till was near 12  x 10~6  cm-1.  Simulated
water levels in piezometers D2-1, D-l, and D-2, based  on a specific-
storage value of 12 x 10~6 cm"1, are compared in figure 13 to values
observed during the 8 months after the pumpout  of  trenches 3-5.   Con-
sistent with field observations, water levels in piezometers distant
from trench 5 showed no substantial declines.

    As discussed previously, more than one combination of hydraulic con-
ductivity, specific storage, and infiltration value can be used to
correctly simulate head in the till after the pumpout  of the trenches.
However, the simulations must approximate actual flows within the till
because (a) simulated and observed drawdowns  in piezometers next to
trench 5 are controlled more by the trench water level than by infiltra-
tion from land surface,  and (b) the values of hydraulic conductivity and
specific storage tested by model calibration  are close to the values
determined in the field and laboratory.
                                              i—i—i—i—i—i   i   i
                       Observed water level in piezometer D-1
        6  8
10  12 14 16  18 20 22 24 26 28 30  1   3

         JULY          DATE' 1976
7  9  11  13  15 17 19 21  23

   AUGUST
  Figure  12.  Comparison  of observed  water level  in piezometer D-l,
              July-August 1976,  with  computer-generated water levels
              produced by changing values of specific storage.

-------
                              241


    No attempt was made to calibrate  the unsaturated functions of
hydraulic conductivity and volumetric soil-moisture content in relation
to soil-moisture tension because  the  entire section is simulated with
saturated or nearly saturated conditions.

    After the pumpout of water  from  trench  5 in September 1976, water
level resumed the rising trend  observed before the pumpout (fig.13).
Transient-state model results of  the  situation 25 days after the pumpout
was complete showed flow into trench  5 through the till  to be 0.2 cm3/s
and flow out of the trench to be  0.4  cm-Vs.  (As a verification, flow
out of trench 5 was manually calculated from Darcy's law to be 0.8 cm3/s
based on a hydraulic conductivity of  8 x 10~8 cm/s, a trench-floor area
of 1 x 10? cm^, and a gradient  of 1:1.)  The transient-state simulation
also indicated that, 6 months after  the pumpout, flow from the till into
trench 5 had decreased to 0.1 cm3/s.   Thus, the continued rise in trench
5 water level after the pumpout was  probably not caused by flow of water
from the surrounding till.   If  the rise were caused by the slow drainage
of water from less permeable zones within trench 5 that were saturated
before the pumpout, the rise should  have slowed with time, which was not
the case after the first few days (fig. 4).  Therefore,  it seems likely
that the rise was caused by  continued infiltration of precipitation
through the cover.

       COMPUTER MODELING OF  GROUNDWATER FLOW TO NEAREST STREAM

    In general, water from the  trenches moves downward through the till
beneath the trench floor until  it reaches the lacustrine unit 30 m below
land surface; from there it  should move eastward to Buttermilk Creek.
However, if some of the trench  water could flow to any of Buttermilk
Creek's tributaries that surround the trenches, both the length of the
flow path and travel time of  trench water to reach land surface would be
much less.  The shortest distance between a trench and an adjacent
stream, and also the greatest  vertical relief,  is at the north end of
trenches 3 and 4  (fig.  1).   To  evaluate this possible flow path, piezo-
meters were installed along  the slope from trench 4 to the small stream
north of trench 4, in line with holes C, C2, and an older test hole  (2C)
that had been installed by the  site operator (fig. 1).

    A finite-element grid  (fig. 14)  was designed to represent  the  cross
section through these piezometers, and the model was used to  simulate
conditions observed  in  February 1978.  The boundary conditions were  the
same as those used for  the  section A-A1, perpendicular to the  north
trenches  (fig. 9), and  values  of hydraulic conductivity and specific
storage were obtained  from  the  same  section.  The best simulations used
infiltration rates of 2 and  2.5 cm/yr  along the  steep slope and  the
smooth  surface near  hole  2C  and C, respectively, and these rates
approximate  those  rates used for scraped or sloping surfaces  in  the  other
section.  The average  absolute deviation from observed heads  in  seven
piezometers was 14 cm.  The  shallow  piezometer  in  test hole C  (C-l)  was
simulated as dry.  Computed  average  head value  near hole 2C compares
well with the  observed  head  value, although the exact screened interval
in hole  2C  is  uncertain.

-------
<
O
  420




  419




  418 -




  417
65
> 416

y


§415

2
C3
_i 414
i= 413




> 412


<
                                                                      Observed water level in trench 5
oi


Z


LU



I
  411




  410




  409
< 408
                       D2-1
                    Observed water level in piezometer

                    Simulated water level in piezometer
                                                                                                                   ro
                                                                                                                   4*
                                                                                                                   ro
                              D-2
JUNE
                JULY
                        AUG
                                 SEPT

                                 1976
                                         OCT
                                          NOV
                                                          DEC
                                                                  JAN
                                                                          FEB      MAR

                                                                                  1977
                                                                                          APR
                                                                                                  MAY
   Figure 13.
                 Comparison of computer-generated  water  levels with observed water levels
                 1n piezometers  D-l, D-2,  and D2-1 during  8 months  after  pumpout  of water

                 from trench 5.

-------
E
    425-i
    420-
  0)415-
00
-
S410-


'•
Q

<405-
 -y400-
 yj P
 D uj
 r; LU
 LJ0395H
    390-
     385
                                                                                           Fixed head for nodes below trench
                                                                                           water level, seepage nodes above
                                                                                           water level
              Constant infiltration rate
                along land surface

          OO O
                       o   o
          o-o-o-
                                                                                    associated with piezometer

                                                                                                 o        o       o       o
                                      o    o
                                                                                                         -O-
                                                            K- 0.85

                                         Node
                                                                                           -O	O-
                                               Fixed head (zero pressure)
                                                EXPLANATION

                                      Backfill           I     I

                                      Weathered till

                                      Unweathered till with
                                      oxidized fractures
                                                                                                                 -o-
                                                           Unweathered till

                                                           Alluvium
                                                                                                       10 METERS
              Figure  14.
                          Cross  section B-B1  from trench 4  to small  stream  showing  arrangement of  nodes,
                          boundary conditions, and relative  hydraulic  conductivities used in  the best-fit
                          model  simulations.   (Location of  section  is  indicated in  fig.  1.)
                                                                                                                                    i

-------
                                  244


    Simulation of heads for February 1976, when water level was  near the
top of trench 4, produced slightly positive heads in piezometer  C-l,
although it had been dry at this time, and simulated heads  in  piezome-
ters C2-1 and C2-2 increased to almost 110 cm, or about 70  cm  more than
the observed increase.  Increasing the hydraulic conductivity  of the
weathered unit in the model lowered the model heads near  test  holes C
and C2 and increased them along the slope, but, when the  lower trench-
water levels of February 1978 were simulated, the decrease  in  piezome-
ters C2-1 and C2-2 was still 70 cm more than the observed change.

    The change in simulated heads in piezometers C2-1 and C2-2 that
resulted from the lowering of model trench-water level approximated the
observed change when the model trench boundary was moved  to 4  m south of
the monument that reportedly marks the end of the trench.   (Exact loca-
tion of the trench boundaries is not known; the monuments that mark the
ends of trenches 1-4 were not placed immediately after completion of the
trenches but after the cover over trenches 1-4 had been modified into
individual mounds.  Also, the ends of the trenches are sloped  to allow
for bulldozer access (20), but the slopes may differ from their design.)
Moving the model boundary away from test holes C and C2 seemed logical
1n that the observed head decline in piezometers C2-1 and C2-2 was
considerably less than that observed in piezometers next  to trench 5.
Piezometers in test hole B, finished 3 m north of the monuments on
trenches 2 and 3 (fig. 1), did not show any appreciable decline in water
levels after the pumpout of water from the sumps in those trenches.

    The change in trench 4 water level between February  1976 and 1978
was reflected In wells 80 m and 175 m south of the north  end;  it is
possible that the water level in the extreme  north end of trench 4  did
not fully respond to the pumpout of water from the sump  135 m to the
south.  A similar lack of response was noted  In  the north end of trench
5  (18), although Kelleher  (2) reported that water had  seeped out along
the north end of trench 4 when the water level was at  an  altitude of 421
m.  A simulation of the October 1975 water level  in trench  4  (altitude
419 m) produced a slight pressure head (13 cm) in piezometer C-l, which
is consistent with Indications during the drilling of  test  holes C  and C2
that a sand interval 2 cm  thick, encountered  at  the depth of piezometer
C-l, was saturated.

    Simulations with the trench boundary moved 4 m  south  produced an
average absolute deviation  of 11 cm for February 1978  and 14 cm for
February 1976.  In both  simulations,  heads did not change along the
slope  toward  the  stream.   The simulation  of  February  1978 suggests  that
water  leaving trench 4 at  that time would not move  to  the nearby  stream
and  that  there  was a  slight gradient  in the  water  table  toward  trench  4
from  holes  2C,  C,  and C2.   Even when  the  water  level  approached the  top
of trench 4  (altitude 418  m)  as  In February  1976, water  from  the  trench
would  move  only about 8  m through  the weathered  till  before it would
move  downward through the  unweathered till  (fig. 15).

    Discharge Into the  stream was  about  100  (cnP/di/m2,  which is  similar
to the  values calculated for section  A-A1,  perpendicular to the north

-------
       B
  425-i
:<>
  400
Infiltration rate, in centimeters per year

                       2.5
                                                                                        .............
                Backfill

                Weathered till

                Unyveathered till with
                oxidized fractures

                Unweathered till

                Alluvium
                                               EXPLANATION
               Observed head in piezometer


                       Test hole
Water level in trench

        /
     Flow path
                  Computed water-table
                     - 414 -
           Equipotential line,  altitude, in meters
                                                                                       10 METERS
                                                                                                                                      i. ,
                                                                                                                                      .! ,
                                                                                                                                      • n
                        Figure 15.   Computer-simulated  heads along  section B-B1  from
                                       trench 4  to  a small  stream  during  February 1976.

-------
                              246
burial  trenches.  Most of this flow was from infiltration of  precipita-
tion  along  the  slope.  The discharge rate  increased by  only  0.25
(cm-Vdi/np when the simulated water level  in trench 4 was  increased to
match  the level for February 1976; this increase  is attributed to the
slightly higher water levels near test hole 2C.
                           MODEL LIMITATIONS

    In general, simulated heads closely matched observed  heads when silt
and sand lenses of limited extent were represented.  Although  no con-
tinuous stratigraphic unit within the till is known that  is  more per-
meable than the till, a lens of silt and sand a few meters  in  length in
one direction would locally alter the flow system, as conceptualized
from simple modeling based on average till properties.

    The simulations were based on the assumption  (a) that the  unweathered
till has a uniform hydraulic conductivity along the sections described
except for variations caused by known sand and silt lenses  and by con-
solidation of the till  from overburden, and (b) infiltration rates of
swamp areas differ from those of smooth, sloping  surfaces.   Although the
modeling results would be the same if infiltration rates  were  made
constant along the entire section and hydraulic conductivity were
increased throughout areas of lower heads.  Such  a change in hydraulic
conductivity does not seem correct, though, because cores from swampy
areas did not differ significantly from those taken from  smooth, sloping
surfaces and because piezometers along smooth, sloping  surfaces did not
consistently show higher hydraulic conductivity than those  near swamps.
                                SUMMARY

    Groundwater flow at the burial  site is predominantly downward  at
an average rate of generally less than 3 cm/yr.  Results of  computer
simulations support conclusions from laboratory tests and  visual  exami-
nation of cores that the till is generally isotropic but becomes  less
permeable with depth as a result of two factors: (a) fractures  caused by
weathering and a corresponding increase in hydraulic conductivity  of the
till near land surface (these fractures become less abundant with  depth
and are absent below 5 m), and (b)  consolidation of the till  by overbur-
den pressure, which reduces intrinsic permeability progressively with
depth.  Simulation of four isotropic layers of differing hydraulic con-
ductivity was an adequate representation of these factors.   Hydraulic
conductivity of the upper two layers (fractured till) was  determined by
model calibration only.

    The specific storage value of the till derived from computer  simula-
tions of water-level trends in piezometers next to trench  5  was about 12
x 10-6 cm-l.  Th^ is a factor of only 1.5 more than the average  value
obtained from four consolidation tests of core samples.

-------
                                  247
    Some of the local  variations in pressure head can be explained  by
areal  differences in infiltration rates.  Adequate simulations were
obtained by assuming that infiltration rates on areas of smooth,  sloping
surfaces were about 50 percent less than the rate of 3.8 cm/yr near
swampy areas.

    A principal conclusion from the computer model runs is that water
leaving the trenches will not intersect the nearby streams unless the
trench-water levels rise to intersect the reworked till  that forms the
trench covers.  This conclusion assumes that none of the pods of  layered
sediments encountered at various depths within the till  are large enough
to extend from a trench to a nearby stream.
                           REFERENCES CITED

 1. N. M. Fenneman, Physiography of Eastern United States (1st ed.), New
    York, McGraw-Hill Book Company, Inc., 714 p. (1938).

 2. W. J. Kelleher, "Water Problems at the West Valley Site," in M. W.
    Carter, Bernd Kahn, and A. A. Moghissi, (eds.), Management of
    Low-Level Radioactive Waste. V. 2, New York, Pergamon Press, p.
    843-851 (1979).

 3. D. E. Prudic, and A. D. Randall, "Groundwater Hydrology and
    Subsurface Migration of Radioisotopes at a Low-Level, Solid
    Radioactive-Waste Disposal Site, West Valley, New York, in M.  W.
    Carter, Bernd Kahn, and A. A. Moghissi, (eds.), Management of
    Radioactive Waste. V. 2, New York, Pergamon Press,  p. 853-882  (1979)

 4. D. E. Prudic, Core Sampling Beneath Low-Level Radioactive-Waste
    Burial Trenches 11-14, West Valley, Cattaraugus County, New  York,
    U.S. Geological Survey Open-File Report 79-1532, 55  p.  (1979).

 5. Mark Reeves, and J. 0. Duguid, Water Movement Through Saturated-
    Unsaturated Porous Media—A Finite Element Galerkin  Model, Oak
    Ridge, Tenn., Oak Ridge Natl. Lab, ORNL-4927, 232 p.  (1975).

 6. R. G. LaFleur, Glacial Geology and Stratigraphy of  Western New York
    Nuclear Service~Center and Vicinity, Cattaraugus and Erie Counties,~
    New York:U.S. Geological Survey Open-File Report  79-789,  17  p.
    (1979).

 7. R. H. Dana, Jr., R. H. Fakundiny, R. G. LaFleur, S.  A.  Molello, and
    P. R. Whitney, Geologic Study of the Burial Medium  at a Low-Level
    Radioactive Waste~Burial Site at West Valley, New York.  New  York
    State Geological Survey Open-File Report 79-2411, 70 p.  (1979).

 8. R. H. Fickies, R. H. Fakundiny, and E. T. Mosley, Geotechnical
    Analysis of Soil Samples from Test Trench at Western New  York
    Nuclear Service Center, West Valley, New York, New  York State
    Geological Survey, NUREG/CR-0644, 21 p. (1979).

-------
                             248


                     REFERENCES  CITED  (Continued)


 9. H. H. Cooper,  Jr.,  J.  D.  Bredehoeft,  and  S.  S.  Papadopulos,
    "Response of a Finite-Diameter Well  to  an Instantaneous Change Of
    Water," Water Resources Research,  v.  3, p.  263-269 (1967).

10. M. J. Hvorslev,  Time Lag  and Soil  Permeability  in Groundwater
    Observations,  Vicksburg,  Miss.,  U.S.  Army Corps Of Engrs.,  Exp.
    Station, Bull.  36,  50  p.  (1951).

11. D. A. Morris,  and A. T. Johnson, Summary  of Hydrologic and Physical
    Properties of Rock  and Soil  Materials,  as Analyzed by the Hydrologic
    Laboratory of the U.S. Geological  Survey, 1948-60, U.S. Geological
    Survey Water Supply Paper 1839-D,  42  p.  (1966).

12. W. R. Purcell,  "Capillary Pressures—Their Measurement Using Mercury
    and the Calculation of Permeability," Am. Inst. of Mining and
    Metalurgical Engineers, Transactions, v.  186, Petroleum Development
    and Technology,  p.  39-48  (1949).

13. R. J. Millington, and  J.  P.  Quirk,  "Permeability of Porous Solids,"
    Trans. Faraday Soc., v. 57,  p. 1200-1207  (1961).

14. T. J. Marshall,  "A  Relation  Between Permeability and Size
    Distribution of  Pores," Journal  of Soil  Science, v. 9, no. 1, p.  1-8
    (1958).

15. S. N. Davis,  and R.  J. DeWiest,  Hydrogeology,  New York, John Wiley
    and Sons,  Inc.,  463 p. (1966).

16. P. A. Domenico,  and M. D. Mifflin, "Water from Low-Permeability
    Sediments and Land  Subsidence,"  Water Resources Research, v. 1, p.
    536-576 (1965).

17. G. E. Grisak,  and J. A. Cherry,  "Hydrologic Characteristics and
    Response of Fractured  Till  and  Clay Confining   a Shallow Aquifer,"
    Canadian Geotech. Jour.,  v.  12,  no. 1,  p. 23-43 (1975).

18. J. 0. Duguid,  and Mark Reeves,  Material Transport Through Porous
    Media—A Finite-Element Galerkin Model, Oak Ridge, TennT, Oak Ridge
    Natl. Lab., ORNL-4928, 201  p.  11976).

19. R. A. Freeze, and J. A.  Cherry,  Groundwater. Englewood Cliffs, N.J.,
    Prentice-Hall, Inc., 604  p. (197571

20. D. E. Prudic, Installation of Water-and Gas-Sampling Wells into
    Low-Level Radioactive-Waste Burial Trenches. West Valley, New York,
    U.S.  Geological Survey Open-File  Report  78-718, 70 p.  (1978).

-------
                STATE-OF-THE-ART IN MODELING SOLUTE AND
                     SEDIMENT TRANSPORT IN RIVERS
                           William W.  Sayre

                        U.S.  Geological  Survey
                       Box 25046, Mail  Stop 413
                         Denver Federal  Center
                        Denver, Colorado  80225
                              ABSTRACT

          This overview is structured around a comprehensive
     general model based on the conservation of mass principle
     as applied to dissolved and particulate constituents in
     rivers, with a few restricted but more specific examples
     that illustrate the state-of-the-art in modeling typical
     physical, chemical, and biological processes undergone by
     selected constituents in rivers.  These examples include:
     simplified one- and two-dimensional formulations focusing
     on the hydrodynamic advection and dispersion mechanisms; a
     two-dimensional biochemical oxygen demand-dissolved oxygen
     model; a one-dimensional polychlorinated biphenyl  model that
     includes uptake and release of constituent by suspended sedi-
     ment, and deposition and erosion of contaminated particles;
     and a one-dimensional sediment transport model that accounts
     for interactions between the flow and the bed, and is capable
     of tracking dispersing slugs of sediment through cycles of
     erosion, entrainment, transport in suspension and as bed
     load, and burial and storage in the bed.
                             INTRODUCTION

     There are many models in various stages of development that simu-
late one aspect or another of solute and sediment transport in rivers.
I will not attempt to discuss them all or even a representative sample.
There are a number of summaries, symposia, and bibliographies in the
literature that do this, for example [1, 2, 3, 4, 5, 6, 7].  Rather, I
will present a somewhat subjective overview that is structured first
around a fairly comprehensive but nonspecific general model, and then
some restricted but more specific examples that illustrate the state-of-
the-art in modeling typical  physical, chemical, and biological processes
that are undergone by selected constituents in rivers.

     Almost all mathematical models of river-transport processes are
essentially accounting procedures.  Nearly always they are based on the
principle of conservation of one or more of the following entities:
                                 249

-------
                                  250
 (1)  Mass of dissolved or participate constituents,  and  transporting
 media (water,  sediment particles!; C2) momentum  of  the  flowing  water,
 which fixes the basic flow properties, including the velocity,  shear
 stress,  and pressure distributions;  C3)  energy,  for example,  thermal
 energy in waste-heat loading and ice-related  phenomena,  or  kinetic  and
 potential energy associated with fluid motion; and  (4)  probability, for
 example, accounting for the sums of  probabilities in a  stochastic model.
                            A GENERAL MODEL

     The starting point for modeling solute and sediment transport
 processes in rivers is usually the conservation of mass principle.   The
 other conservation prtnctples are more likely to enter in supporting
 roles.  The basic three-dimensional p-D) conservation of mass  equation
 for a dissolved or particulate constituent that is being transported by
 the flowing stream is:

                       fed*)  - fcUx £} - fefcy |£) - fj(.z jf) -  is,

          Advection                   Turbulent diffusion       Internal
            ••*——^————^——        sources
                            Hydrodynamic                         and
                                                                sinks
                                                     	CD
where

           C =  concentration of solute or particulate constituent,

     x,  y,  z =  distances in  longitudinal, vertical and transverse
               directions,
     u,  v,  w =  time-averaged  local velocities of solute or particulate
               constituent  in  x, y, z directions,

  e , e  ,  EZ =  local coefficients of turbulent diffusion for solute
      y       or particulate  constituent in x, y, z directions,

         ZS. =  sum of  internal source and sink terms,
           t =  time.

     The advection and  turbulent diffusion terms on the left-hand side
of Eq. 0)  represent the hydrodynamic transport and mixing mechanisms
that occur within the  river channel.  They must be known if Eq.  (1)  is
to be solved.  Usually they are estimated by some empirically based
computational method or determined by measurement.  It should be noted
that particulate constituents, unless they consist of or behave  like
fine suspended  sediments, cannot be assumed to have the same local
velocities and  turbulent diffusion coefficients as the transporting
fluid.  In particular,  if deposition, burial, scour and re-entrainment
of particulate  constituents occur, an additional conservation equation

-------
                                  251
should be introduced for the material stored in the bed.  Except for
determination of water-surface elevation and the cross-sectional average
velocity in the downstream direction, methods for solving the complete
momentum and energy equations to obtain local velocities and diffusion
coefficients in natural rivers are still in their infancy.  This is
mainly because of complications associated with irregularities of
channel shape and alinement, and with the nonlinearity of the equations.

     The internal source and sink terms on the right-hand side represent
time rates of increase Cor decrease) of solute or particulate concen-
tration due to physical, chemical, or biological reactions and transfer
processes occurring within the flow cross section.  Examples of these
mechanisms are radioactive decay, uptake and release of constituent by
suspended sediment, photolysis, hydrolysis, oxidation-reduction reac-
tions, bacterial degradation, and accumulation and depuration by
biological organisms.  The rates of these processes depend on factors
such as concentrations of the constituent and substances with which it
reacts, solar radiation, and variables which define the state of the
aquatic environment, such as temperature, turbidity, pH, and turbulence
level.  External source and sink mechanisms, such as evaporative trans-
fer, uptake and release of constituents by bottom sediments, scour and
deposition of contaminated sediment, are represented as boundary con-
ditions at the water surface or river bed and banks.  The rates of these
processes may depend on variables that define the state of adjacent
environments, for example, wind velocity, air temperature and humidity,
dynamics of flow-sediment interaction at the bed, in addition to the
factors listed previously for the internal sources and sinks.  Fortu-
nately, many source/sink mechanisms can be simulated acceptably by
first-order-reaction kinetic models.

     If there are reactions between different constituents, or if
exchange of a constituent occurs between different phases of the aquatic
environment (for example, uptake and release of dissolved constituent by
suspended or bottom sediments), a set of coupled  equations, with one
equation for each constituent and (or)  phase, may be formulated.  These
equations, each  similar to Eq. 1, are coupled by  source/sink terms that
specify the rates of reaction or exchange.  To obtain spatial and tem-
poral  concentration distributions for the different constituents and
phases, the entire system of equations  must be  solved.   Formulations of
coupled-system models  for biochemical oxygen demand (BOD) and dissolved
oxygen (DO), and polychlorinated biphenyls  (PCB's) are  presented in  the
next section.


            SELECTED RESTRICTED  FORMS OF GENERAL  MODEL

Two-Dimensional  and One-Dimensional  Formulations

     Except in  the  initial mixing region just  downstream from a  con-
centrated  source of constituents, for  example,  a  simple pipe  outfall,
adoption  of 3-D  models like  Eq.  (1)  for use in rivers  is rarely  either
justified  or feasible.  Firstly,  substances transported by  rivers  tend

-------
                                  252


to become mixed, first over the depth of flow Cor possibly across the
channel  1n situations where there is persistent density stratification),
and then throughout the entfre channel  cross  section.   Thus,  after
distances that are typically on the order of  50 to 100 times  the depth
and 100  to 500 times the vridth, the mixing can be modeled  first as a
two-dimensional (2-D) and eventually as a one-dimensional  (1-D) process.
Secondly, neither 3-D modeling of the momentum or energy equations nor
any other computational method 1s far enough  advanced  to provide suf-
ficient  information about the local  velocities, particularly  the com-
ponents  normal to the mean flow,  and diffusion coefficients.   Thirdly,
even with advances in modeling and increased  computer  storage capacity,
complete solutions for 3-D models are likely  to remain prohibitively
expensive and time-consuming, except for the  most idealized and
simplified cases,  such as uniform flow  in a prismatic  channel.  Using
integral transform methods, Cleary and  Adrian 18]  have obtained analyt-
ical solutions for the 3-D advection-diffusion equation with  constant
longitudinal  velocity, zero secondary velocities,  and  constant diffusion
coefficients, for  a solute in a rectangular channel with uniform flow.

     In reducing general  3-D conservation equations to simplified 2-D
and 1-D forms, integrating the equations over the depth of flow and (or)
across the width of the channel ,  with the use of appropriate  boundary
conditions,  is preferred  rather than simply neglecting the advection and
diffusion terms for whatever dimension  is being eliminated, and assuming
for convenience that the  remaining velocities and  diffusion coefficients
are constant  throughout the depth or cross section.  Following the
appropriate  integration procedure, one  finds  that some terms  are elim-
inated naturally due to boundary  conditions,  and that  other terms may
adopt a new form 1n the reduced equation.   Also, one is more  likely to
obtain correctly formulated external  source/sink terms when they arise
naturally from applying boundary  conditions to the integrated 3-D
equations.   Following  the concepts of Fischer 19], integration of Eq.
0) over the  depth of flow leads  to  the 2-D conservation of mass equation:
           Advection                    Dispersion

        * S5*   +  rS*  .                                            (2)

      Internal    External
       Sources and sinks

Similarly, Integration over the  cross- sectional  area  leads  to the 1-D
formulation:


                                    •*      *       •               <3>
         Advection    Dispersion    Internal    External
                                     Sources  and sinks

-------
                                   253

                     —d      —A
In Eqs. C2) and (3) C  ) and (  } represent averages taken over the
depth, d, and the cross-sectional area, A, respectively.  Also, the new
coefficients E , E , and K  no longer represent turbulent diffusion
alone.  They now represent longitudinal and transverse dispersion from
the combined action of vertical and cross-sectional variation of the
local velocities u and w and turbulent diffusion.

     Formulating Eq. (2) and the depth-integrated continuity equation
for water in a 2-D orthogonal curvilinear coordinate system, and re-
placing the transverse distance z by the transverse cumulative discharge
measured from one bank

          qc = /  d Od mz dz                                       (4)
               ZL
leads to the stream-tube version 110] of the 2-D equation:
     3t    m
            A
A definition sketch for the coordinate system is shown in Fig. 1. Eq.
(5) is more suitable than Eq. (2) for solute-transport modeling appli-
cations in natural rivers.  The transverse advection term, which does
not appear in Eq. (5), has been formally eliminated by introducing the
continuity equation, and not simply neglected.  Another advantage of Eq.
(5) is that the new transverse coordinate variable, q , varies within
a fixed range from zero to the total river discharge, Q, irrespective
of whether the river meanders, or contracts and expands in width.

     In Eqs. (4) and (5), m  and m  are metric coefficients to correct
for differences in transverse and longitudinal distances resulting from
curvature in channel alinement.  They can be evaluated from the stream
tube geometry.  Except in abruptly converging or diverging sections and
in sharply curved reaches, little error results from assuming that they
are equal to unity.  The longitudinal dispersion term containing E  is
not included In Eq. (5), because it contributes little to longitudinal
dispersion in comparison to the advection term.  However, there is some
evidence [11, 12] that retaining the term and assigning small values of
E  helps to control troublesome overshoot problems in numerical solu-
tions of Eqs. (2) and (5).

     Finite-difference models based on Eq. (2) [11] and Eq. 5 [12]
have been used successfully to simulate both unsteady and steady-state
dispersion of dye and waste heat in the Missouri River.  The alter-
nating-direction implicit (ADI) method was used in both cases.  Analyt-
ical solutions of a simplified steady-state form of Eq. (5) for a con-
                                          O J
servative solute, wherein the quantity m d u E  is assumed to be a
                                        J\     £
constant or longitudinally varying diffusion factor, have been used with
good results In a number of cases [10, 13, 14].  Analytical solutions of

-------
                                          254
Horizontal distances
along coordinate surfaces

  LOA= x
   OB
    z axis
     Transverse
  coordinate surfaces
                                                                                  x axis
                                                                            Longitudinal
                                                                          coordinate surfaces
                                                                  Right bank
                                  PLAN VIEW
           Longitudinal
        coordinate surfaces
            y = VB (x. z. t)
                                                     y = ys (x, z. t)
                                                               y =0
                                 SECTION D - D

              Fig.  1.    Coordinate system  for  natural  channels,

-------
                                  255
Eq. (5) also have been obtained for transient cases that have been
simplified still further by assuming the channel to be straight and
prismatic, and that u  and E  are constants 115].

     Analytical and numerical solutions of the 1-D model, both with and
without source/sink terms, have been widely and successfully used, espe-
cially in thermal energy (temperature} and BOD-DO models 116, 17, 18, 19,
20].  However, in modeling instantaneous concentrated-source dye experi-
ments that simulate accidental releases of slugs of conservative soluble
pollutants, significantly improved results commonly are obtained by cou-
pling the 1-D advection-dlffusion equation (Eq. (3) with no source/sink
terms) for the main stream to a dead-zone equation 121, 22] as follows:


       m . nA   m
     at

     aC,
The subscripts m and d denote main stream and dead zone respectively;
K = coefficient for mass transfer between main stream and dead zone (in
units of velocity); and P = average length of interface, in plane of
cross section, between the subareas A_ and Ad.  Coupling with the dead-
zone equation simulates the effect of^separation zones caused by bank
and bed irregularities, which trap some of the dispersing material and
release it slowly back into the main stream, adding to the tails, and
consequently the skewness of the concentration-distribution curves, as
observed in many dye experiments 123, 24].  In thermal energy and BOD-DO
applications, temperature and concentrations usually change slowly
enough that addition of the dead-zone equation, and often even inclusion
of the longitudinal dispersion term, Is unnecessary.

Two-Dimensional Biochemical Oxygen Demand-Dissolved Oxygen Model

     The BOD-DO model is selected for illustration because it exem-
plifies a coupled-system model that includes both internal and external
source/sink terms.  Also, its 1-D form has a long history of successful
application dating back more than 50 years to the Streeter-Phelps equa-
tions [25], the most rudimentary form of the model.  The stream-tube
version of the 2-D model formulated here has not yet been applied.
However, the ADI finite-difference techniques used in solving Eqs. (2)
and (.5) should work, after modification to include the source/sink
terms.  The 2-D model would be appropriate for use in wide rivers, where
the time scale for mixing across the channel equals or exceeds the time
scales for BOD assimilation and DO replenishment.

     The model consists of a coupled set of three equations for first-
stage carbonaceous BOD  (CBOD), nitrogenous BOD  CNBOD), and dissolved
oxygen (DO),   in most respects, it follows Thomann's  126] 1-D formu-
lation.  The dependent variables are:

-------
                                   256
      L = depth-averaged CBOD concentration, in mg/A;
      L  = depth-averaged NBOD concentration, in mg/A;
      C = depth-averaged DO concentration,  in mg/Ji.

The three equations are:
                                                                     (8)
                         Bacterial  oxidation;
                      Sedimentation,  Adsorption
3t
      — t  qu _ Q
m
                3X
                             }  =  -K
                                Bacterial
                                oxidation
                                                          (9)
                                 Bacterial
                                 oxidation
                                    Surface
                                   reaeration
                                                               (10)
                           SpCx,p,t)    -    SR(x,p,t)
          Removal  by
           benthal
           demand
              Production  by
                  plant
             photosynthesi s
                                     Removal by
                                       plant
                                     respiration
     On the left-hand  side  of  the  equations,  p = qc/Q is the cumulative
discharge normalized by  the total  river discharge Q,  and D is the
                                                              p J    p
transverse diffusion factor, taken as  the average value of m d u E /Q
                                                            /»     £
and assumed to be constant  across, but not  necessarily along, the
channel.  On the right-hand side,  KJ,  K  , 1C,  and K  are all first-
order rate coefficients,  K. and K.,  are biological-rate coefficients
that depend on the properties  of tne oxygen-demanding pollutants and the
river water.  K  relates to settling (or resuspension) and adsorption,
and, consequently, to  the properties of the pollutant and the turbulence
in the river.  The reaeration  coefficient,  K , depends on the rate of
surface renewal and water temperature; it hai  been related to river
depth and velocity, as well as temperature, by several empirical equa-
tions.  C  is the saturation concentration  of DO.  The functional forms
of the soDrce/sink terms SB, Sp, and SR are not well  understood.  The
benthal oxygen-demand  rate, SB» relates to  the aerobic decomposition of
organic matter in the  surface  layer  of bottom deposits.  The net photo-
synthetic oxygen production rate,  Sp-SR,  is highly variable and depends
primarily on the amount  of  solar raaiant  energy received, and its

-------
                                  257
distribution over the river depth, nutrient concentrations, dissolved
oxygen, and water temperature.  Point and distributed sources of BOD,
from waste outfalls, tributary Inflows, runoff, and so forth, can be
represented by boundary conditions at and along the banks (p=0,l).

     As yet there are no operational  2-D BOD-DO computer models.  How-
ever, some of the better-known computer models that are available for
1-D modeling of BOD and DO In rivers  are DOSAG-1  127]; SNOSCI and SRMSCI
[28, 29]; QUAL-I 130, 31]; QUAL-II 132]; and RECEIV [33].  With the
exception of DOSAG-1, these programs  also can model other selected
water-quality constituents.

One-Dimensional Polychlorinated Biphenyl Model

     Although this model has not yet  been developed beyond the Initial
formulation stage, It 1s presented here because it includes uptake and
release of constituent by suspended sediment, and deposition and erosion
of contaminated sediment particles.  Further investigation may well
indicate that some revisions, additions, or deletion of terms, and so
forth, should be made.  However, for  the purpose of this presentation, a
hypothetical chemical, say ABC, would serve about as well as PCB.

     Because PCB in both dissolved -and adsorbed form, together with
suspended and bottom sediments, are known to be important in determining
the distribution of PCB in the aquatic environment, it is appropriate to
formulate a coupled-system model consisting of four equations.  The
dependent variables for the four equations are:

          C(x,t) = concentration of dissolved PCB, yg/Jl, assumed to be
                   constant throughout the cross section;

          M(x,t) = concentration of suspended sediment, mg/*,, assumed
                   to be constant throughout the cross section;

    rM = C (x,t) = concentration of PCB adsorbed by suspended
          p        sediment, ng/fc, assumed to be constant throughout
                   the cross section;
    rhM.=C.(x,t) = concentration of PCB adsorbed by bottom sediment,
                   vg/l, assumed to be constant throughout the active
                   bottom layer of thickness, h(x), that is involved  in
                   the erosion-deposition process;

where
          r(x,t) e adsorption ratio for suspended  sediment, (iig adsorbed
                   PCB)/(nig suspended sediment), assumed to be constant
                   throughout the cross section;
              Mh = concentration of bottom sediment in active bottom
                   layer g/A, assumed to be constant;
         rh(x,t) = adsorption ratio for bottom sediment,  (yg adsorbed
                   PCB)/(g bottom sediment),  assumed  to  be constant
                   throughout active bottom layer.

-------
                                  258
The governing equations are:

          iC + n l£ _ 1 3_ r*v iCi
          3t     3x ~ A~ 3X
               = -K,  CM(rc-r)  +  l^rM + W - Sb                       (11)


             (Hi) - U |^rM) - Jfe [AK, fj (rM)J
                                           V.        .
               - K,  CM(rc-r) - K2rM + Wp -  / rM + Y £ rbMb        (12)
          ft W=M-^bV                              (14)

The variables t, x, U, A, and K , were defined previously.  The
remaining quantities to be defined are:
          K,  = adsorption coefficient, Ji/pg PCB-day;
          r  = limiting adsorption ratio at full adsorptive capacity,
              tug adsorbed PCB)/(mg suspended sediment);

          Kp  = desorption coefficient, day" ;
      W(x,t)  = rate of addition of dissolved PCB from nonpoint external
              sources (for example, runoff and atmospheric fallout),
              ug/ji-day, assumed to be instantaneously mixed throughout
              the cross section;
     Sb(x,t)  = net rate of removal of dissolved PCB due to accumulation
              and depuration by biological organisms, "«/»-^"-
     W_(x,t) • rate of addition of particulate PCB from nonpoint
      "       external sources, vg/fc-day, assumed to be instantaneously
              mixed throughout the cross section;

          YS = settling velocity of suspended sediment, m/day;

        d(x) = mean depth of channel, m;

           Y = entrainment rate coefficient for bottom sediment, day"  ;
        h(x) = mean thickness of active bottom layer involved  in
              erosion-deposition process, m.

A definition  sketch for the model is shown in Fig. 2.

     Eqs. (11) through  (14) are written for a particular size  fraction
of fine sediment, for which it is assumed that:   (1) Concentration
distribution  of  suspended sediment throughout the cross section is

-------
11
h
            U
                                    W. W,
                                          (Additions from nonpoint
                                             external sources)
C -	+~rM
 (Release and uptake
by suspended sediment)
                            Vs
                           	 rM (Settling)
                            d
                                        ^(Temporary storage in the bed)

            Fig.  2.    Definition sketch for  one-dimensional  PCB model.
                                                                                            U
                                         (Entrainment)
                                                                                                                ro
                                                                                                                un

-------
                                   260


 uniform;  C2)  mean  velocity and longitudinal dispersion coefficient are
 the same  as for water; and (3) settling velocity, V_, limiting adsorption
 coefficient,  r  , and rate coefficients (KI.KO.Y) are all constants.   If
 all  size  fractions are to be considered, Eqs. (12)-(14) are written
 separately for  each size fraction, and the terms, K-,CM(r -r) and K2rM, in
 Eq.  (11)  are  summed over all  size fractions.  No provisions were made in
 Eqs.  (11) through  04) for either adsorption or desorption to occur
 directly  between the water column and the bottom sediment, or for bed-
 load  transport.  Also, an additional  set of equations could be written
 to  simulate exchange mechanisms for PCB within the bed.  Such equations
 would be  required to forecast long-term Cseveral years) changes in PCB
 concentration following curtailment of its use.

      Modeling of coupled sediment-contaminant systems (such as the PCB
 model) Is still very new.   The only operating models of which the writer
 is aware  are a group of models developed by Onishi and his colleagues
 134,  35,  36] at BatteH e Pacific  Northwest Laboratories.*  These consist
 of 1-D and  2-D models that include equations for dissolved constituent,
 suspended  sediment, adsorbed  constituent, and exchange mechanisms that
 differ somewhat from those postulated in the PCB model.  These models,
 which include both  finite-element and finite-difference methods, have
 been used for radionuclides In the Columbia and  Clinch Rivers, and for
 Kepone in the James River.  The development of this class of models is
 hindered by the lack of reliable  physically-based relationships and
 coefficients for computing rates  of exchange of  sediment between the
 active bottom layer,  the bedload  zone,  and suspension in the flow.   In
 addition,  existing  bedload transport  formulas are not sufficiently
 reliable,  except possibly  for steady  flow in straight uniform channels.

 One-Dimensional  Sediment Transport Model

     The model formulated  1n  this section exemplifies the type of
 sediment-transport  model that needs to  be developed further before the
modeling of sediment-contaminant  systems  can come of age.   In addition
 to predicting  transport  rates and hydraulic resistance, such a model
would account  for Interactions between  the flow and the bed for non-
 uniform and slowly-varying unsteady flow, including the development of
 bed forms; erosion  and deposition by  particle size class,  possibly
 leading to degradation or  aggradation of the bed; and changes in sed-
 iment size gradation and armoring of  the bed.  The model  also needs to
 be capable of tracking dispersing slugs of sediment through cycles  of
 erosion, entrainment, transport in suspension and as bedload, deposi-
 tion, and  burial and storage  in the bed.
      The use of trade name is  for descriptive  purposes  only and does
not imply endorsement by the U.S.  Geological  Survey.

-------
                                   261


     A definition sketch for a model that meets the above requirements
is shown in Fig. 3.  The model consists of the following system of
equations.

 Equation of motion for water:

-------
 u
QM
                            Suspended load zone
 Qb
                 Inactive deposition layer
                     Original bed material
                                                                                                          Datum
             Fig.  3.   Definition  sketch for one-dimensional  sediment  transport  model
                                                                                                                   ro
                                                                                                                   en
                                                                                                                   IN3

-------
                                   263



Continuity equation for water:


            H^ v     ^iv                                              \  /
            0 A     0 A

Continuity equation for sediment in suspended load zone:


     1M. + n — - ! — /A"  aMx -   W " M  ^ KW
     at     ax   A ax
                                 Settling    Diffusive             (17)
                                             transfer
Continuity equation for sediment in bedload layer:

        aMD.     an.
     Wa -^  + -=§. = WVJ^ -
         o U      pA     b O,
                    Settling   Diffusive
                               transfer
                               YWhM                                (18)
                     Settling  Entrainment


Continuity equation for sediment in active bed layer:

        Wr\FYi
        C/IH _ i ii| ij
     P,
                    - YWhMb           ,                             (19)
           Settling   Entrainment

Bedload transport:

     Qb = f (U,d,D,...)                                            (20)

Eqs. (17) -(20), the sediment equations, are all written separately for
each size fraction; following solution, the results are summed over all
size fractions.  The sum of the rates of change of masses, m . , of the
different size ranges of sediment in the active bed layer from Eq. (19),
is related to the rate of change of bed elevation, z^, by the equation:

            az.       -  am.
where n = porosity of the bed, and p.. = mass density of particles in the
i-th size range.

-------
                                   264
     Quantities in Eqs. (15)-(20) not heretofore defined, or with new
definitions, are:

     g = acceleration of gravity;   a_
                                    eJZu
    S  = longitudinal bed slope -   ^-;
    Sf = friction slope determined from a  hydraulic  resistance
         equation;
     M = average volumetric sediment concentration;  and  the subscripts
         a, BL and b denote, respectively:   at  the level  a above the
         bed, in the bedload layer,  and in  the  active bed layer;

     W = active bed width,  over which deposition or  entrainment
         1s occurring;
     K = coefficient governing  rate  of diffusive transfer of sediment
         between suspended  and  bedload zones, in units of velocity;

     Y = coefficient governing  rate  of entrainment of sediment from
         active bed layer  into  bedload layer, in units of reciprocal
         time;

     D = characteristic  sediment diameter.

     Eq.  (20)  is a symbolically stated formula  for bedload discharge.
Any one of a number of formulas 137], or even an empirical rating rela-
tionship developed for the  reach being modeled, could be used.

     In addition to the  system  of equations (15)-(21), the model also
needs to include an accounting  procedure for computing and updating the
size distribution  of the bed material. Selective net removal  of finer
size fractions  by  erosion  tends to coarsen  the  bed,  leading to an
increased resistance to  erosion, or  armoring, of the bed.

     Probably the  weakest  link  in the model  described in the preceding
paragraphs is the  lack of  Information on the transfer and entrainment
coefficients,  K and y   Capability for predicting a  and  h, the thick-
nesses of the bedload and active layers, 1s not much better.

     The model  as  proposed  in Eqs.  (15)-(21), with the bed material
accounting procedure, is a  modification of  the  sediment  transport and
armoring model  developed by Bennett  and Nordln  [38], which has been
applied to reaches of the  East  Fork  River  1n Wyoming, the East Fork of
the Carson River In Nevada  139], and the Rio Grande  River downstream
from Cochiti Dam in New Mexico  [40].   Significant modifications include
the replacement of a single bedload-zone equation by the bedload-layer
and active bed-layer equations, and  modifications of the settling,
diffusive transfer, and  entrainment  terms on the righthand sides of Eqs,
(17)-(19).  The Bennett-Nordin  model  is in  some ways similar to the U.S.
Army Corps of Engineers  HEC-6 model  [41] for simulating  scour and depo-
sition in reservoirs and rivers.  However,  HEC-6 neither explicitly
separates the bedload from the  suspended-load transport  nor simulates

-------
                                  265


the exchange of sediment between suspension in the flow and storage in
the bed, so that dispersing slugs of sediment can be tracked through
repeated transport-deposition-burial-resuspension cycles.  The Bennett-
Nordin and HEC-6 models apply only to cohesionless sediment; erosion and
deposition mechanisms for cohesive sediments are somewhat different.
Transport models for cohesive soils have been developed by Ariathurai
and Krone 142] , and Odd and Owen [43] .

     In most instances, significant simplifications of the sediment
transport model are possible.  For gradually varying subcritical  flow,
bed transients are known to propagate at a much slower rate than either
water-surface transients 12] or concentration transients for suspended
or dissolved matter.  If one's interest is restricted to long-term
phenomena, a quasi-steady state for the liquid phase, suspended-load
zone, and bedload layer can be assumed, wherein A, U, M, and MB,  are
taken to be invariant throughout the time interval At = x/U, wnere At
and AX are the time and distance steps in the numerical computation.
Eqs. 05} and 06) then reduce to the steady-state non-uniform flow
equation for computing water surface profile 144]:

               dd _  o  f
                  -
where Fr is the Froude number, and the time derivative terms in Eqs. (17)
and 08) are eliminated.  Also, in long-term studies, the longitudinal
dispersion term in Eq. (17) can be eliminated, because longitudinal
dispersion in the suspended load zone is negligible in comparison to
that due to storage in and resuspension from the stream bed.  In sit-
uations where the changing bed geometry is of much more concern than
exchanges between and distinguishing among the modes of transport in the
different zones, further simplification can be achieved by combining Eq.
(18) with Eq. (17) and (or) Eq. (19).  This results in an equation
involving transport of the total sediment load:  Qs = QM + Qfa.


                           CONCLUSIONS

     This review of solute- and sediment- transport modeling in rivers
concludes with an attempt at rating the state-of-the-art for various
classes of models.  State-of-the-art is taken here to mean mainly
overall reliability in application.  A steady or quasi-steady flow, for
a period that is not less than the duration of one time step, throughout
the reach that is being modeled, is assumed in all cases.  State-of-the-
art ratings on a scale of 1 to 10 (in order of improving state),
assigned subjectively by the writer to various classes of models, are
listed in Table 1.  The classes are defined by groups of the categories
of characteristics and phenomena, Identified by capital letter in the
listing beneath the table.

-------
                                266
 Table 1 .—Subjective Rating of State-of-the-Art, on 1 to 10 Scale
       (in order of improving state), for Various Classes of
                      River Transport Models
Categories^ of Characteristics
and Phenomena included in Model
A, C, E
B, C, E
B, C, E, G
B, C, F, G
B, C, F, H
B, C, F, G, H
A, D, E
B, D, E
B, D, E, G
B, D, F, G
B, D, F, H
B, D, F, G, H
A, D, F, J
B, D, F, I, J
B, D, F, H, I, J
B, D, F, G, H, I, J
1-D
Model
8
8
7
6-7
6-7
5-7
6
6
5
4-5
4-5
3-5
5
3-4
2-3
2
2-D
Model
7
7
6
4-5
3-5
3-5
4
4
3
2-4
2-4
2-3
3
1-2
1-2
1
Listing of categories
   A.  Conservative  constituents
   B.  Non-conservative  constituents
   C.  No density (buoyancy)  effects
   D.  Density (buoyancy)  effects
   E.  Single system
   F.  Coupled system
   G.  Gas or heat transfer at water  surface
   H.  Chemical/biological  reactions
   I.  Adsorption/desorption  by  sediment
   J.  Deposition and  entrainment  of  sediment

-------
                                  267


     In the listing of categories, the designations single system and
coupled system are used in the sense defined by Thomann 126].  In a
single system, usually simulated by a differential equation for a
single constituent, there is a functional relationship between an input
(discharge of constituent at source) and a single output (concentration
of constituent in river).  Coupled systems are simulated by two or more
differential equations, with interacting source/sink terms.

     Neither the categories nor the groupings listed in the table
purport to be exhaustive.  However, they do represent a large proportion
of present river-transport models.


                                REFERENCES

1.   R. A. Baker Ced.), Contaminants and Sediments, Vol. 1:  Fate
     and Transport, Case Studies. Modeling, Toxicity, Ann Arbor
     Science Publishers, Inc. (.1980):

          Ch. 15, by B. H. Johnson, p. 331-348
          Ch. 16, by L. J. Thibodeaux, L.-K. Chang and D. J. Lewis,
               p. 349-371
          Ch. 18, by Y. Onishi, D. L. Schreiber and R. B. Codell,
               p. 393-406
          Ch. 19, by C. G. Uchrin and W. J. Weber, Jr., p. 407-425.

2.   J. A. Cunge, F. M. Holly, Jr. and A. Verwey, Practical Aspects
     of Computational River Hydraulics, Pitman Advanced Publishing
     Program 0980).
     G. P. Grimsrud, E. J. Finnemore and H. J. Owen, Evaluation of
     Water Quality Models:  A Mi
     600/5-76-004 (July, 1976).
Water Quality Models:  A Management Guide for Planners, EPA-
•']•_•-•-•^^^^^^^^* u."j*"f^ j L.m^ '"'^*—'— "uri ' '  "    ' * ' •—• • —- • i ii i ••-••-- --in
4.   M. E. Jennings and N. Yotsukura, Status of Surface-water
     Modeling in the U.S. Geological Survey, U.S. Geological
     Survey Circular 809  0979).

5.   Oak Ridge National Laboratory, Proceedings of a Workshop on
     the Evaluation of Models Used for  the  Environmental Assessment
     of Radionuclide Releases, Report on Hydro!ogic Transport of
     Radionuclides, p. 33-72, CONF 770901 (April, 1978).

6.   Task Committee on Computational Hydraulics of the  Hydromechanics
     Committee of the Hydraulics  Division,  "Sources of  Computer
     Programs in Hydraulics," Journal of the Hydraulics Division,
     ASCE, Vol. 106, No.  HY5, p.  915-922 CMay, 1980).

-------
                                  268
7.   H. W. Shen (ed.), Modeling of Rivers,  Wiley-Interscience (1979):

          Ch. 1, by H. W. Shen
          Ch. 8, by W. A. Thomas
          Ch. 10, by Y. H. Chen
          Ch. 13, by B. C. Yen
          Ch. 14, by 6. D. Ashton
          Ch. 15, by W. W. Sayre and  R.  Caro-Cordero
          Ch. 17, by P. A. Krenkel  and V. Novotny
          Ch. 18, by P. A. Krenkel  and R. J.  Ruane

8.   R. W. Cleary and D. D.  Adrian, "New Analytical  Solutions for
     Dye Diffusion Equations," Journal of the Environmental
     Engineering Division, ASCE, Vol. 99, No.  EE3,  p.  213-227
     (June, 1973).

9.   H. B. Fischer,  "Longitudinal  Dispersion  and  Turbulent Mixing
     in Open-Channel  Flow,"  Annual  Review of  Fluid  Mechanics. Vol.  5,
     p. 57-78 Q973).

10.  N. Yotsukura and  W.  N.  Sayre,  "Transverse Mixing  in  Natural
     Channels,"  Water  Resources  Research, Vol. 12,  No.  4,  p.  695-704
     CAugust,
11.  U.S.  Army Engineer  District, Omaha,  Corps of  Engineers,  Omaha,
     Nebraska, and  Sutron Corporation, Arlington,  Virginia, Missouri
     River Mathematical  Model of the Lateral and Longitudinal  Mixin
     Processes Tn  Open Channels, MRD Sediment  Series,  No.  16  (May
ing
7T979),
12.   T.  0.  Harden  and H. T. Shen,  "Numerical  Simulation  of  Mixing in
     Natural  Rivers," Journal of the Hydraulics  Division, ASCE,  Vol.
     105,  No. HY4, p. 393-408 (.April, 1979).

13.   Y.  L.  Lau and B. 6. Krishnappan, "Modeling  Transverse  Mixing
     in  Natural  Streams," Journal  of the Hydraulics  Division,  ASCE,
     Vol.  107, No. HY2, p. 209-226 (February,  1981).

14.   N.  Yotsukura  and E. D. Cobb,  "Transverse  Diffusion  of  Solutes
     in  Natural  Streams," U.S. Geological Survey Professional  Paper
     582-C 0972).

15.   H.  T.  Shen, "Transient Mixing in River Channels," Journal of
     the Environmental  Engineering Division,  ASCE, Vol.  104, No.  EE3,
     p.  445-459 (.June,  1978).

16.   D.  R.  F. Harleman, D. N. Brocard, and T.  0. Najarian,  A Predictive
     Model  for Transient Temperature Distributions in Unsteady Flows,
     Technical Report 175, R. M. Parsons Laboratory  for  Water  Resources
     and Hydrodynamics, MIT  C1973).

-------
                                   269


 17.   P.  P.  Pally,  T.-Y.  Su, A.  R. Giaquinta,  and  J.  F.  Kennedy,  The
      Thermal  Regimes  of  the Upper Mississippi and Missouri  RiversT"
      IIHR  Report No.  182,  Iowa  Institute of Hydraulic Research,  U.
      of  Iowa  [October, 1976).

 18.   H.  E.  Jobson  and T. N. Keefer,  "Modeling Highly Transient Flow,
      Mass,  and Heat Transport in the Chattahoochee River near Atlanta,
      Georgia," U.S. Geological  Survey Professional Paper 1136, (1979).

 19.   D.  A.  Bella and  W.  E. Dobbins,  "Difference Modeling of Stream
      Pollution," Journal of the Sanitary Engineering Division. ASCE,
      Vol.  94, No.  SA5, p.  995-1016 COctober, 1968')".

 20.   R.  Dresnack and  W.  E. Dobbins,  "Numerical Analysis of BOD and
      DO  Profiles," Journal of the Sanitary Engineering Division.
      ASCE,  Vol. 94, No.  SA5, p. 789-807 COctober, 1968).

 21.   J.  R.  Hays, P. A. Krenkel  and K. B. Schnelle, Jr., Mass Transport
      Mechanisms in Open-Channel Flow. Technical  Report 8, Sanitary and
      Water  Resources  Engineering, Vanderbilt University 0966).

 22.   F.  B.  Pedersen,  Prediction of Longitudinal  Dispersion in Natural
      Streams, Series  Paper 14,  Institute of Hydrodynamics and Hydraulic
      Engineering, Technical University of Denmark (February, 1977).

 23.   C.  F.  Nordin, Jr. and G. V. Sabol, Empirical  Data on Longitudinal
      Dispersion in Rtvers, Water Resources Investigations 20-74, U.S.
      Geological Survey (1974).

 24.   T.  J.  Day, "Longitudinal Dispersion in Natural  Channels," Water
      Resources Research, Vol. 11, No. 6, p. 090-918 (December, 1975).

 25.   H. W.  Streeter and E. B. Phelps, A Study of the Pollution and
      Natural Purification of the Ohio River, III.  Factors Concerned in
      the Phenomena of Oxidation and Aeration,  Public Health Bulletin
      No. 146, U.S.  Public Health Service (February,  1925).   (Reprinted
      by U.S., DHEW, PHS, 1958).

26.   R. V. Thomann, Systems Analysis and Water Quality Management,
     McGraw-Htll  Book Co. (1972).

27.  Texas Water Development Board, DOSAG-I, Simulation of Water Quality
      in Streams and Canals:  Program Documentation and User's ManuaT
     NTIS, Springfield, Virginia (.PB 202 974)  CSeptember 1970).

28.  Systems Control, Inc.  and Snohomish County  Planning Dept., Water
     Quality Management Plan for the Snohomish River Basin and tfii
          riagc  	             	
Stillaguamisn- River Basin, Vol. IV-Computer Program Documentation,
Part A:  Steady State Stream Model  (SNOSCI), and Part B;   Dynamic
Estuary Model ISRMSCI), Snohomish County Planning Dept.,  Everett,

-------
                                   270
29,  Systems Control, Inc. and Snohomish County Planning Dept.,
     Water Quality Management Plan for the Snohomish River Basin and
     the Stjllaguamisii River Basin;  Volume Ill-Methodology (Chapter 4),
     Snohomish County Planning Dept., Everett,  Washington (1974).

30.  Frank D. Masch and Associates, and the Texas  Water Development
     Board, Simulation of Water Quality in Streams and Canals:   Theory
     and Description of the QUAL-I Mathematical  Modeling System, Texas
               •f
               Ml
     Water Development Board,  Austin,  Texas,  Report 128 (May, 1971 ) .

31.  Texas Water Development Board,  QUAL- I, Simulation of Water Quality
     in Streams and Canals:  Program Documentation  and User's Manual.
     Texas Water Development Board,  Austin, Texas  (September, 1970).

32.  L. A. Roesner, J.  R.  Monser  and D.  E. Eyenson, Computer Program
     Documentation for  the Stream Water  Quality Model, QUAL-II, Water
     Resources Engineers,  Inc., Walnut Creek,  California  (May, 1973) .

33.  Metcalf & Eddy,  Inc., University  of Florida, and  Water Resources
     Engineers, Inc., Storm  Water Management Model, Vols. 1-4, EPA
     Report No. 11024DOC,  CU.S. Govt.  Printing Office, Stock Nos. 5501-
     0109, 0108, 0107,  0105) (July-October, 1971).

34.  Y. Onishi, P.  A. Johanson, R. G.  Baca, and E.  L.  Hilty, Studies
     of Columbia River  Water Quality--Deyelopment of Mathematical Models
     for Sediment and Radionuclide Transport Analysis, BNWL-B-452,
     Battelle Pacific Northwest Laboratories,  Richland, Washington
     (January, 1976).

35.  Y. Onishi, Finite  Element Models  for Sediment  and Contaminant
     Transport in Surface  Waters-'- Transport of Sediment and Radionuclides
     in the Clinch River,  BNWL-2227, Battelle  Pacific  Northwest
     Laboratories,  Richland, Washington  (July, 1977).

36.  Y. Onishi and R. M. Ecker, "The Movement  of Kepone in the James
     River" (Chapter VII), and "Mathematical Simulation of Sediment
     and Kepone Transport  in the  Tidal James River" (Appendix M), in
     The Feasibility of Mitigating Kepone Contamination in the James
     River Basin, Appendix A to the  EPA  Kepone Mitigation Project Report ,
     Battell e Pacific Northwest Laboratories,  Richland, Washington
     (June, 1978).

37.  V. A. Vanoni (ed.), Sedimentation Engineering, Ch. II-H, Sediment
     Discharge Formulas, p.  190-230, ASCE Task Committee  for the
     Preparation of the Manual on Sedimentation of  the Sedimentation
     Committee of the Hydraulics  Division, ASCE-Manuals and Reports on
     Engineering Practice-No.  54  (1975).
38.  J. P. Bennett and C.  F.  Nordin,  "Simulation  of  Sediment Transport
         Armouring," Hydro! ogtcal
        555-569 (December, 1977).
     and Armouring," Hydro! ogtcal  Sciences-Bulletin,  Vol.  22,  No.  4,
                    ember,

-------
                                  271
39.  T. Katzer and J. P. Bennett, Sediment Transport Model for the East
     Fork of the Carson River, Carson Valley, Nevada, Open File Report
     80-160, U.S. Geological Survey, Carson City, Nevada (1980).

40.  R. C. Mengis, Modeling of a Transient Streambed in the Rio Grande,
     M.S. Thesis, Colorado School of Mines (May, 1980).

41.  Hydrologic Engineering Center, Generalized Computer Program HEC-6,
     Scour and Deposition in Rivers and Reservoirs, Users Manual, U.S.
     Army Corps of Engineers, 723-G2-L2470, Davis, California (March,
     1977).

42.  R. Ariathurai and R. B. Krone, "Finite Element Model for Cohesive
     Sediment Transport," Journal of the Hydraulics Division, ASCE,
     Vol. 102, No. HY3, p. 323-338 [March, 1976}.

43.  N. V. M. Odd and M. W. Owen, "A Two-Layer Model for Mud Transport
     in the Thames Estuary," Proceedings, Institution of Civil
     Engineers, London, England, Supplement Ox), Paper 7517S (1972).

44.  F. M. Henderson, Open Channel Flow, The Macmillan Co. (1966).

-------
                   AN OVERVIEW OF RESUSPENSION MODELS:
                APPLICATION TO LOW-LEVEL WASTE MANAGEMENT

                               J. W.  Healy

                    Los Alamos Scientific Laboratory
                             P. 0. Box 1663
                      Los Alamos, New Mexico 87545
                                ABSTRACT

          Resuspension is one of the potential  pathways to man  for
     radioactive  or  chemical  contaminants  that  are  in  the
     biosphere.    In waste  management,  spills or  other surface
     contamination can serve  as a  source for  resuspension  during
     the  operational  phase.    After the  low-level  waste disposal
     area is  closed, radioactive materials can be  brought  to  the
     surface  by  animals  or  insects  or,  in  the  long  term,   the
     surface  can  be  removed  by  erosion.   Any of  these methods
     expose  the  material  to  resuspension  in  the atmosphere.
     Intrusion  into  the  waste  mass can  produce  resuspension of
     potential  hazard to  the  intruder.   Removal of items from  the
     waste  mass  by   scavengers  or  archeologists  can   result in
     potential  resuspension exposure to  others  handling or working
     with the object.  The ways  in which resuspension  can  occur  are
     wind   resuspension, mechanical   resuspension  and  local
     resuspension.   While methods of predicting  exposure are  not
     accurate, they include the  use'of the resuspension  factor,  the
     resuspension rate and mass  loading  of the air.
                              INTRODUCTION

     Resuspension  and   inhalation  of  the  resuspended  material   is  a
potential pathway whereby radioactive materials can be transferred  from
the ground,  or  other surface, to  man.    In general,  it  is of  greatest
importance  for  those   radionuclides  that  are  strongly  discriminated
against in the food pathway.  If  the  radionuclides  are strongly  taken  up
by foods, and food from the contaminated  area  is used, the contribution
to the dose from resuspension  and inhalation is usually small.

     In this  paper,  we discuss the  application  of this  pathway to the
management of low level wastes and describe briefly some of the methods
of  estimating  air  concentrations   from  resuspended  material.    More
detailed  discussions  of  the  technical  details are  available  in the
literature [1,2].
                                273

-------
                                274
                              RESUSPENSION

     Before entertaining the question of how resuspension is involved in
 low-level  waste  management,  let  us  consider  what  resuspension  is.
 Technically,  it  is the suspension of  a  material  in a  fluid  after this
 material  has once been  deposited  on a  surface from a  previous
 suspension.   Thus, j^esuspension.   This fluid can  be either air or water
 and  resuspension  in both can  be of  importance  in  analyzing  potential
 problems  from low-level  wastes.   Air  suspension  is used  in  estimating
 quantities inhaled and water suspension  is important  in the movement of
 radioactive particles in streams  or other bodies  of water.   However, for
 this  discussion,   we will  confine  our   attention  to  the  problem  of
 suspension in air.

     It  should  be  apparent  that  not  all  sources  meet  the  technical
 definition of resuspension given  above.   For example,  if  a radioactive
 material is spilled on  the surface or if  it  moves  upward to the surface,
 it has not been deposited by a  previous suspension and should  not really
 be  considered  as  £esuspension.   However,  because  the mechanisms  for
 suspension are similar  regardless of the method of contamination,  it is
 convenient to change the definition  to  include all  cases  in  which the
material becomes  airborne  from  the surface.

     It  should  also be made  clear  that the  suspensions  that  we  are
concerned with are not  the true colloidal suspensions that  the chemists
are  used to  but.  rather,  are  particles kept   in  suspension  by  the
turbulent movement  of  the  fluid  in  which they are carried.   Thus, an
aerosol  resulting  from  resuspended  materials  can  contain much  larger
particles than could be indefinitely  suspended  in still air.   In  fact,
those of you  who  have been  caught in sand storms  with strong  winds will
realize that  large particles that  can sting when  they hit  can  be  carried
under conditions of highly turbulent  winds.
                    RESUSPENSION  AND  LOW-LEVEL WASTES

      Now, how  does resuspension  fit  into  the  analysis  of  low-level
waste?   Obviously,  if we  do  our  jobs  perfectly  it  will  not  become a
factor.   There  are. however,  complications  such as  accidents  or spills
that must be considered and  accounted  for  as  well  as a  few  potential
complications that  could  arise  with  the  long-term  behaviour of  the
burial grounds.

     In figure 1, we find an elementary schematic  drawing  of a few ways
that people can be exposed  from low level  waste operations.  A number of
the  pathways  to man have  no  connection  with  resuspension  but  I  have
included  some  of  these as  dotted  boxes  and lines  for the  sake  of
completeness.

     The  transport  accidents  and spills  can  result  in  contaminated
surfaces  that   can give  rise   to  resuspension and  inhalation  of

-------
                            Direct
                             radiation
                                                              • —••• <^^W —••• ^"W ~~H


                                                               Milk    I
                                                            •— r —
                                                                       I
                                                            ^ Animal
                                                             L-i-
                                                Suseensionj  Plant   i
                                                   I	 	I	I
                                                  	ICQD sport _,
    Fig.  1.  Schematic drawing of pathways to man  from low-level waste
operations.  Solid lines are related  to resuspension while dotted lines
are other  pathways.
                                                                             ro
                                                                             ^j
                                                                             en

-------
                                276


radioactive materials by man.   These,  however,  should be of  relatively
minor  concern,  from the resuspension  point  of view,  because they  are
usually  noted  and  the  area  cleaned  before  serious  resuspension  can
occur.  There may be some long-term build-up of activity at places  such
as  loading docks  where  the  material  could  be  resuspended,  but  this
problem can be controlled by  monitoring.   These  potential  problems  are
encountered in the  operational  phase of the  disposal area and  can  be
handled  by "institutional"  methods  because  the  workers  are  already
there.

     However,   there  are   later  opportunities   for  movement   of
radioactivity  to  surfaces where  resuspension could be  a factor.   Some of
these  are  shown  in  Fig.  1.   For example, transport  of  the activity  to
the surface by water, ants, burrowing mammals  or other means can  produce
a source exposed to winds  or other disturbances, that  could  result  in
resuspension.   The possibility that  could  expose  the largest number of
people to  small  quantities  starts  with transport  to an aquifer.   The
water  from this  aquifer  could be  withdrawn from  wells  and used  for
irrigation, with  the radioactive materials adsorbing  on  clays or  humus
to gradually produce large contaminated areas.   If the aquifer  empties
into a stream, radioactive  materials will  accumulate  in sediments  which
can be deposited  on  flood plains or  the water  can be  used for  irrigation
with  the   same  result   as  the  use  of  well   water.    One   aspect  of
resuspension in  these  cases  which  leads  to the  food  pathway  is  the
resuspension and deposition on  food plants.   Experience with nuclides
that are only  poorly taken  up by plants indicates  that  this  can  be  an
important  factor.

     Another  series of pathways   involving  resuspension  occurs  in
disposal  areas  containing  long-lived  radionuclides.   Here  we  can
postulate  eventual  removal  of the cover  material  by erosion, either by
wind or water.   When this  occurs,  we are left with a contaminated area
to depths  of several meters,  augmented by  surface contamination  in  the
surrounding area  caused  by the materials removed and  carried downwind or
downstream. This radioactive  material  can  be  resuspended by winds or by
movement in the area.   This is a futuristic scenario  depending upon the
particular site of  disposal.   With average  erosion  rates, it could occur
some thousands or  tens  of thousands  of  years from now although  gully
erosion could  occur much faster  but  on a  limited scale.  In some areas,
the  net erosion  pattern  is  such that  the  disposal  area will  be  simply
buried deeper, but  the present tendency to  look for  areas that are "high
and dry" to site  disposal areas  could eliminate this  possibility  because
the  high sites with a long  distance to the water  table  are  just  those
that will   have a negative-gain erosion  potential.

     The  intrusion  case,  i.e. where  an  individual  digs  into an  area,
either  knowingly or  unknowingly,  results  in resuspension caused  by his
activities.  In an enclosed space  this  can  result  in high concentrations
of  dust and  associated  radioactive materials with  the  individual
inhaling these materials.  However,  this  exposure  will  be limited to the

-------
                                 277


individual or  individuals concerned  unless  contaminated  items  or
materials are removed and taken to a place where others can be involved.

     This  possibility brings  up  the  case  in  which  an  archeologist.
either  amateur  or professional,  investigates  these disposal  areas  for
"treasures" of the past.   Old  burial  grounds are the  source  of  much  of
our knowledge of the  prehistoric  past  and they serve as  magnets  to draw
those interested in history.  We hope that the level of knowledge in the
future  will  be  such that  proper  precautions are taken  when  this  event
does occur,  but  this is  just  the type  of  point that we  cannot  assure
without  surveillance of  disposal  areas until  the radionuclides  have
decayed.  Obviously any contamination on any objects removed could  serve
as a source of resuspension and inhalation by any individual who  handles
or works with these objects.

     One special  case in this category results from the current practice
of  using  non-degredable  plastics  to  contain  the  contamination  during
handling  of  the  wastes   en   route  to   the  disposal  area.    Current
information gives little  reason to  believe  that  these  plastics will  not
preserve their contents and remain intact under the conditions of burial
in the  soil.   This raises the  possibility  of  recovery of such packages
with consequent exposure when the package is opened.


                         CLASSES OF RESUSPENSION

     We have divided the overall resuspension problem into three  classes
defined by both  their characteristics  and the information  needed  to
solve them.

     The first class  is  wind resuspension.   This occurs  when the  wind
blowing over a field resuspends particles and moves them  downwind.   Much
is known about the mechanics of this class through  the  many  studies  of
wind  erosion  of  agricultural   fields.   In  essence,  there   is  little
suspension  of soil  particles   unless  a  phenomenon called  "saltation"
occurs.   Here,  particles on the  order of  100-200 ym  in diameter  start
rolling over the surface  and then jump into the  air under the pressure
of the  wind.   They,  then, travel  forward  with  the  wind, gaining energy
and meanwhile falling under  the force of gravity,  until  they  impact  on
the  ground.   This  impact  dislodges  other  smaller particles that  are
suspended in the air.  Such a process  will  start at the  edge  of  a  field
and  increase  in  intensity  downwind  until  a maximum  is  reached.   One
other process for wind resuspension, that has not been well investigated
could   result from  the  deposition  of an  aerosol  on vegetation.
Presumably  any  of this material  that  is  not  bound  to  the  foliage  by
adsorption or absorption  could  be dislodged by  movement  of the  foliage
under the action of the wind.

     The  second  class  of  resuspension   is  that  caused  by  mechanical
disturbance of the soil  with the  resuspended  particles  moving downwind
to expose people.  This could be,  for  example,  a farmer  plowing  a field

-------
                                27S
or  people  walking or working and disturbing the soil.  The familiar dust
cloud  raised  by such  activities  in  dry  weather  is a  sign  of such
resuspension.   Here  the forces required for  resuspension  come from the
activity but the movement, once the material is in the air. is dependent
upon the winds.   Many  such  disturbances are point  sources  at the point
of  disturbance  rather than  a  broad  area  source as   with  wind
resuspension.  Another difference between these two classes is the depth
to  which the disturbance reaches.  In wind  resuspension  the depth from
which  resuspension can occur is dependent upon  the  depth  reached by the
seltating  particle  or  fairly  shallow.   In mechanical disturbance the
depth  can  be inches or even  feet depending upon the type of disturbance.
Thus  a source  at greater  depth can  be resuspended  than  occurs with
winds.

     The final  class  is local   resuspension.   This  is  the resuspension
leading to a cloud in the  immediate  vicinity of an individual  causing
the disturbance.  It can apply  to an  individual  handling  a contaminated
object, to  an  individual  digging  a  hole in  contaminated  soil or to the
dust cloud  surrounding  a farmer plowing a  field.    Whereas  the  concern
with the previous two  classes  was with  people  downwind from the source
of the resuspension. the concern  with  this  class  is with  the individual
causing the disturbance.
                       EVALUATION OF RESUSPENSION

     Now  that  we  have  pointed  out  some of  the  ways  in which  the
resuspension  pathway  to  men  can  be  of  interest  to  low-level  waste
disposal, how  can  we  evaluate  its  seriousness?   In other  words, if we
can define a source term how can we evaluate the  potential  exposure to
man?   The  only  proper answer  at  this  time  is  that  we  do  it  with
considerable uncertainty and with many assumptions.   Unfortunately, the
studies  required  to  better  understand  these  problems  and to obtain
adequate coefficients  for use are not well  funded.

     There are  three  methods,  or models,  for  estimating resuspension:
the resuspension factor, the resuspension  rate and  the  mass loading of
dust in the air.

     The  resuspension  factor  is  defined  as  the  ratio   of   the  air
concentration resulting from resuspension  to the quantity of contaminant
per unit area at the location where the ay sample was taken.   Usually,
the air concentration  is in activity per m  and the ground contamination
is.in activity per nr.  Thus, the units of the resuspension  factor are
nf .  There are many values of the resuspension factor in the literature
because  it  has been  the  most  widely used  method  for  many years  and
measurements  are relatively  simple.   Typical  of these values are those
in a table by Mishina  [3\   In this table, the values vary over  about 10
order  of magnitude.    However,  closer examination of this  table shows
that wind resuspensiqn  with  freshly deposited material  is  in the range
of 1C~B  to  2xlO"b  m'1 and for aged deposits  from  6xlO~10 to 10~13 m~r.

-------
                                 279


Fo£ mechanical disturbance the  range  seems  to be from 2xlO~6 to  7xlO~5
m~ .  However, many of these tests were done in arid country  and may  not
be typical of other areas.

     A decrease in the resuspension with time  can be  expected following
deposition from the atmosphere.   This  decrease  is caused by  penetration
of the contaminant into the soil or by other means  of fixation, such as
adsorption  on  large  soil   particles.    Initially,   rather  crude
measurements at the  Nevada  Test Site  indicated  a decrease  with  a half
life of  about  one  month  [4j.    However, as  information has accumulated,
it is apparent that any decrease with time  is  over  a much longer  period
of time than one month.  Anspaugh-CSjjhas  formulated a relation in which
the resuspension is j^nitially 10"  m~  with a  decrease over  a period of
several years to 10"  m"  .   It  should  be  stressed,  however,  that  such a
relation  is  based  upon meager  data obtained in  only one  area.   It may
also be noted that such a decrease with time will probably not occur for
other types  of deposition such  as  liquid spills  or  contaminants brought
to the  surface  from a below grade disposal site because these actions
result in an initial  mixing of  the contaminant  with  the soil.

     The resuspension factor has conceptual  problems when  it  is used for
describing wind or mechanical  resuspension.   For example,  it does not
account  for  resuspension  that  occurs  upwind.     It  is  primarily  an
empirical concept with  little hope  of  eventually  describing  the
important variables involved in resuspension.   However, from  a practical
standpoint,  most  of the  past   measurements  have been made  using this
method  of  interpretation  so  that  some  data  are available.   The
resuspension  factor,  or   some  variation of  it.  may  also  be  a   likely
choice  in describing  local  resuspension  because the receptor  and the
source of contamination in the  air are at  the  same place.

     The second method of describing resuspension is through the use of
the resuspension  rate.   The resuspension  rate  is  the  fraction  of the
contaminant  in the ground that  is  resuspended  per unit time.  This rate
describes a source term that can be used with  the known correlations for
meterological  dispersion  to  estimate  the  air  concentrations of  the
contaminant  downwind.    There  are  not   many measurements  of  the
resuspension  rate available.   Sehemel,,^] using, a  nonradioactive
contaminant,  measured  values  of  10~     to  10"  sec"   for  wind
resuspension in,&  lightly vegetated area.   Anspaugh et al. [7] measured
values of 2x10    to 10   for wind resuspension in an  area,at  the  Nevad^
Test, Site.   For mechanical  resuspension  values of lxlO~   to 1x10"
sec"  for  an individual  walking in a contaminated  asphalt highway have
been inferred  [2] from measurements  made  by  Sehmel  [8].   For a truck
driving through a tracer  ongan  asphalt highway Sehmel  [9j  reports  values
equivalent to rates  of 1C   to 8xlC   sec   with the resuspension rate
varying with the speed of the vehicle.

     The  resuspension rate  is  plagued with  the  difficulty that too  few
measurements have been made to  allow use for types of soils and in areas
different  than  those in  which  measurements have been  made.   It also

-------
                                280
requires knowledge of the  pattern  of contamination  on  or  in  the  soils  of
the contaminated  area.   However,  it does  provide  additional detail  on
the meaning of the measurements and. as information increases,  may  well
be the method of choice for specific areas.   It  s  value  in  studying the
phenomenon of resuspension  is  illustrated  by the  findings  in  the  past
few years that the rate of resuspension  increases with the wind  speed  at
rates up to the 8th  power of the  wind  speed  depending upon the  type  of
soil [10].

     The third method  for calculating  resuspension is the mass  loading
approach.   Here   it  is assumed that the  dust in  the air has  the  same
concentration of contaminants  as occurs in the  soils.  Anspaugh  et al
[11] have collected information from a number of sources and  have shown
that reasonable  agreement  between  measured  and calculated concentrations
can be had  if one assumes,  for the Calculation, that  the concentration
of dust  in  the  air  is  100  ug  per m .   Since it  is  doubtful  that the
actual  concentration  is this high, this finding  should be regarded  as a
correlation  rather  than  as  a  true  relation  between actual  dust and
concentration of contaminant.

     There are problems with this  concept also in that the measured  mass
loading at a given point  depends  upon  the dust  which  could  have arisen
from a  large distance upwind and would  serve  to dilute the dust  from the
contaminated  area,   particularly  if  the   contaminated area  is  small.
There have also  been  questions  raised  as to the  source of the dust  from
the soil.   That is,   if the  soil  contains a  large fraction  of smaller
particles which contain all of the  activity, the  concentration  in  this
fraction   is  the  one that  should be  used  in  the calculations.  One
investigator [12]  has gone so far as  to destroy all  of  the  natural
aggregates in the soil and to use the concentration in the remaining
small  particles   which do  not exist  in  nature,  as  the basis  of the
calculation.   However, all correction factors suggested for  this effect
using  the  particle  size  distribution   in  natural  soil   have  been
comparatively small,  less  than a  factor of  two.  The use  of  the  mass
loading  has  certain  value,   however, particularly in  view of the
correlation mentioned earlier.   It should be  noted  that this  correlation
was obtained on  ambient air with no  disturbances  in the vicinity.  Thus,
if it is to be used,  the mass  loading should  be  increased by some factor
to allow for the possibility of mechanical  disturbance.  We  believe that
the mass loading may  be the best method for  generic studies  not tied  to
a specific site.
                               DISCUSSION

     This has been a  very  brief  discussion of the  role  of resuspension
in low level waste management.  Other  references  are available that will
provide  additional  detail   [1,2].   However,  it  is  apparent  that  the
information  available  is  not really  adequate  to  do more  than make  a
crude estimate of the  inhalation  problem  from any  of the  three classes
of   resuspension.     Additional   studies,   providing   both   basic

-------
                                281
understandings  of the mechanisms of wind resuspension and empirical  data
tied to specific types of areas and actions are needed to provide  data
for better estimates.
                              REFERENCES

 ].  G.  A.   Sehmel.  "Transuranic   and  Tracer  Simulant  Resuspension."
     Transuram'c  Elements in the Environment,  W.  C.  Hanson Ed.,  Tech.
     Info.  Center,  U.S. Dept. of Energy (1980).

 2.  J. W.  Healy, "Review of  Resuspension  Models." Trensuram'c Elements
     in the Environment.  W.  C.  Hanson  Ed.,  Tech.  Info.  Center,  U.S.
     Dept.  of Energy  (1980].

 2.  J.  Mishima.  "A  Review of  Research  on  Plutonium  Releases  During
     Overheating and  Fires."  Hanford Laboratories Report HW-83668
     (1964).

 4.  R.  H.  Wilson. R.  G.  Thomas,  and J.  N.  Stannard,  "Biomedical  and
     Aerosol  Studies  Associated with a  Field  Release  of Plutonium."
     University  of Rochester Atomic Energy  Project Report  WT-1511
     (November I960).

 5.  L. R.  Anspaugh.  J. H.  Shinn,  and D. W. Wilson. "Evaluation  of the
     Resuspension Pathway  Toward  Protective  Guidelines  for  Soil
     Contamination with  Radioactivity,"  Population Dose Evaluation and
     Standards for Man and  his Environment.  Symposium Proceedings,
     Pcrtcroz. STI/PUB/375.  International  Atomic Energy Agency,  Vienna
     T1974).

 6.  G.  A.  Sehmel, "Plutonium  and  Tracer Particle  Resuspension:   An
     Overview of Selected Battelle-Northwest  Experiments." Transuranics
     in Natural Environments. Symposium Proceedings,  M.  G.  White  and P.
     B.  Dunaway,  Eds.. ERbA  Report NVO-m, pp.  181-210,  Nevada
     Operations Office, NT1S.

 7.  L.  R. Anspaugh,  P. L.  Phelps,  N.  C. Kennedy, J.  H.  Shinn, and J. M.
     Reichmann.  "Experimental Studies on  the  Resuspension of Plutonium
     from  Aged' Sources at  the  Nevada  Test  Site, Atmosphere  - Surface
     Exchange of  Particulate  and Gaseous  Pollutants (1974)," ERDA
     Symposium Series.  No.  36. R.  J.  Enclemann  and G.  A. Sehmel
     Coordinators, pp. 'M-M,  CONF-740921,NHS (19/b).

 8.  G.  A.  Sehmel, "Tracer  Particle  Resuspension Caused by Wind  Forces
     Upon  an Asphalt  Surface." Pacific  Northwest Laboratory  Annual
     Report for 1976  to the USAEC Division of Biolocy and Medicine,  Part
     1   Atmospheric  Sciences. Battelle  Pacific Northwest  Laboratory
     Report BNKl-1651  (Ptl]; pp. 136-138,  NllS  (19/ZJ.

-------
                                282
 9.  G. A. Sehmel,  "Particle  Resuspension  from an Asphalt Road Caused by
     Car and Truck  Traffic,"  Atmos. Environ..  7:291-309 (1973).

10.  D. A. Gillette, and  I.  H.  Blifford,  Jr., "Measurements  of Aerosol
     Size Distribution and Vertical  Fluxes  of Aerosols  on  Land Subject
     to Wind Erosion," J.  Appl.  Meteorol., 11, 977-987 (1972}.

11.  L.  R.  Anspaugh,  J.  H.  Shinn,  P.  L.  Philps,  and  N.  C.  Kennedy,
     "Resuspension  and  Redistribution  of  Plutonium  in  Soils."  Health
     Phys., 29,  571-582  (1975).

12.  C. J. Johnson. R. R.  Tidball,  and R.  C.  Severson. "Plutonium Hazard
     in  Respirable  Dust  on  the  Surface of  Soil,"  Science.  193,  488
     (1970).                                                  '

-------
          AN OVERVIEW OF  EPA'S  HEALTH RISK ASSESSMENT MODEL
                 FOR THE SHALLOW LAND DISPOSAL OF LLW
                            G. Lewis Meyer
                            Cheng Y.  Hung

                    Criteria and  Standards Division
                     Office of Radiation Programs
                 U.  S.  Environmental  Protection Agency
                        Washington,  D.  C.  20460
                               ABSTRACT

    EPA is developing a generally applicable environmental standard
for managing and disposing of LLW as part of the National Radioactive
Waste Management Program.  To support the LLW standard, EPA will
assess potential impacts from disposing of LLW by a number of
alternatives including estimating potential releases to the
environment, doses, health risks and costs from eight land disposal
methods.  This paper presents an overview of EPA's assessment model
for shallow land disposal (SLD) of LLW.  It describes model design
considerations} conceptual design of model and approach to developing
model.  The SLD model has data preparation and main calculational
sections.  The main calculational section has release, transport and
dose/health submodels including leaching, release, groundwater
transport, surface water transport, atmospheric transport,
resuspension, irrigation, operational spillage, human intrusion,
biological uptake by man, dose and health risk submodels.  Where
possible, existing models/submodels were used.  However, new
submodels were developed for infiltration, leaching, release and
groundwater transport.  Basic system equations and brief explanations
of these four submodels are given.

                             INTRODUCTION

    The Environmental Protection Agency  (EPA)  is developing a
generally applicable environmental standard for the management and
disposal of low-level radioactive wastes  (LLW) as part of  the U.S.
National Radioactive Waste Management Program.(1,2,3)  The President
has directed EPA to accelerate development of  its standard.(3)   The
current schedule calls for EPA to issue  a proposed LLW standard  in
September,  1982, and  to publish a final  LLW standard  in  January,  1984.


                                283

-------
                               284
    The EPA's program to develop the  LLW standard includes assessing
the health risks from disposing of LLW by shallow land disposal (SLD)
and by other alternative disposal methods.   The potential costs and
benefits (reductions in health risks) from using each of these
alternatives will also be compared.   This information will provide
input to and support for EPA's decisions concerning the standard.
The -analyses will also be used to fulfill the requirements of NEPA,
the Environmental Impact Statement and the Regulatory Analysis.

    The EPA has guidance from a number of sources on how to develop
the LLW standard.  These include NEPA (4), EPA's Improving Government
Regulations (5), BIER II (6),  Decision Making in EPA (7), ICRP 26
(8), and EPA's proposed Criteria for  Radioactive Waste Management
(9).  These guidelines advise EPA to:

    o    compare the health risk and  cost of reasonable alternatives;

    o    ensure that the health risk  from the activity is acceptable;

    o    ensure that the health risk  is as low as reasonably
         achievable, taking into account social and economic factors
         (ALMA); and

    o    use predetermined model(s) to make the above assessments.

    The above guides clearly establish EPA's use of models to support
the development of standards.  Models for estimating the health
risks,  making the benefit/cost analyses and comparing the
alternatives will be necessary for the LLW standard.  The purpose of
these models is to assess and predict the potential future impacts
from managing and disposing of LLW by a number of realistic
alternatives.  These assessments will include estimating the
potential releases of radionuclides to the environment and the
subsequent doses, health risks and associated costs.

    The focus of this paper is on EPA's environmental assessment
model to simulate the health risks from disposing of LLW by shallow
land disposal methods (SLD model).  The general design and layout of
EPA's SLD model was done in-house and a preliminary report on it was
presented in February, 1979.(10)  This report included a general
description of the model's data input requirements, the pathways,
scenarios and land disposal alternatives to be considered, and its
potential applications.  Specific guidance for development of the SLD
model, with particular attention to the infiltration, leaching,
release and groundwater transport submodels, were developed in-house
by Hung and other ORP staff. (11)  Since then, Hung has developed a
groundwater transport submodel for the SLD model (12), and a method
for characterizing the long term leachability of solidified

-------
                                285
radioactive waste.(13)  The SLD model is currently being developed by
the Oak Ridge National Laboratory under an interagency agreement with
the U.S. Department of Energy.  It is expected that the assessment
model for other land disposal alternatives can be adapted from the
basic SLD model without major modifications.

                CONCEPTUAL MODEL DESIGN AND DEVELOPMENT

Model Design Considerations

    The following considerations influenced the design of EPA's SLD
assessment model and its submodels.

    o    A generic model is required because EPA will be analyzing
         the potential impact of disposing of a wide variety of waste
         types by different disposal methods under sharply different
         hydrogeologic and climatic conditions.

    o    A land disposal facility  should be modeled as a complete
         disposal system which  includes the wastes, the site geology,
         hydrology, and climate, and all the pathways  from it because
         the performance of the system will change if  changes are
         made  to  any of its parts.

    o    A systems approach is  needed  to evaluate  the  effectiveness
         of  each  disposal  system and proposed changes  in its design
         (i.e., modifying  the waste  form or packaging).

     o    The model  should be  modular  in construction  to  allow
         substitution or  modification  of  submodels when  it is used to
         analyze  different disposal  systems and sufficiently  flexible
         to  analyze  disposal  systems with  a wide range of
         hydrogeologic and climatic  conditions.

     o   A simple,  one-dimensional analytical approach would be
          adequate and, in fact, preferred for EPA's  risk assessment
          modeling needs to keep computer time and costs  within
          reasonable bounds, to avoid potential errors in the
          numerical approach, to stay within the quality of the input
          data currently available, and to utilize and stay within the
          state-of-the-art modeling for many pathways.

 Conceptual Model Design

     The conceptual model  for the  environmental assessment of SLD
 consists of a group of preparation submodels group and the main
 calculational model.  The main calculational model includes the
 leaching, release and environmental transport  submodels and

-------
                                286
simplified individual and population dose and health risk submodels.
A block diagram of the general  layout of  the  SLD model is given in
Figure 1.

    The preparation submodels prepare secondary input data for use in
the main calculational model.   These secondary input data are
independent of the time of simulation and can be processed prior to
being integrated into the main  calculational  model (i.e., annual rate
of infiltration, health effects conversion factors, unit response of
region to a release).  The purpose  of the preparation submodels is to
prepare files of these data (i.e.,  annualized synthetic hydrological
or meteorological conditions or other appropriate basis) as
efficiently as possible, for input  into the calculational submodels,
thus reducing central processing unit time and costs.

    The submodels within the main calculational model simulate the
release and transport of radionuclides from the waste by a driving
force due either to natural processes or  to human activities.  The
transport submodels include: leaching; release; groundwater
transport; surface water transport; atmospheric transport;
resuspension; irrigation; operational spillage; human intrusion,
including farming on the site;  and  biological uptake by man including
food chain, drinking water, inhalation, and direct irradiation.

    Water is considered to be the most important driving force for
releasing radionuclides from the waste.  The predominant release
modes are expected to be leaching,  erosion, gas generation, and human
intrusion.  When the radionculides  are released from a burial trench,
they will be available for transport to environmental receptors.  The
receptors considered are groundwater, surface water, air, and land
surface.  The main model simulates  the transport of the radionuclides
released among these receptors.  The resultant accumulation of
radionuclides in the receptors  are  then available for biological
uptake by humans through the food chain,  drinking water, inhalation,
and direct irradiation pathways.

    The results of these simulations furnish input into the dose and
health risk submodels.  The dose submodel integrates the total doses
from the  total radionculide uptake  by humans from all critical
pathways  to provide an estimate of  potential individual dose.  The
health risk submodel evaluates  the  corresponding health effects and
integrates them over the affected population to obtain the potential
total fatal health effects and  number of years of life lost.

Approach  to Model Development

    Development of our conceptual SLD assessment model into a working
model was  guided by the considerations discussed earlier and the need
to develop it within a relatively short time.  To meet our schedule,

-------
                           287



READ INPUT




«HML onuMn JS5/-S-I SUB ^TiS*
1






DO A I - I, 1000


LEACHING


1 OPERATIONAL _..«,
1 SPILLAGE RELEASE
.
ATMS
TRAMS

1,
J

\ \Ji. 	

WRIC QCUND IWTO SURFACE WATER
PORT TUHSKKt TRAfSPORT


rRESuspENSKM j-t
1
^


^ IRRIGATIGN
1
/ ATMBPHERE \ / GROUND SURFACE \ / QUUND HATER \ / SURFACE MATER \



BIOUKICAL


IfOIVUJUAL
DOSE


POPULATION DOSE,
H3ALTH EFFECTS 1
HEALTH RISK


/





a'
o
•H
-P
(fl

M
UNIT RESPONSE Q.
[ INDIVIOML DOSE (jj



H
n_i

r*
IUWN INTRUSION T
REdAMATION C

	 ' £
i~
fl
t
(
T
4
«i
r*
^
c
r"
t
BIOLOGICAL C
PAHMAY

.
" Submodels H
^
D
3
^
5
14
5
H
J
3
H
3
J
H
0
^
.*
4
INDIVIDUAL *&
DOSE S



                           PRINT OUTPUTS
Figure 1.
Block Diagram  of EPA Environmental Assessment
Model for  the  Shallow Land Disposal  of  LLW.

-------
                                288
we took the following steps; we,   (1)  reviewed  existing assessment
models, (2) selected viable models from those reviewed, (3) utilized
viable submodels from the models  selected,  and  (4)  filled in any gaps
with new or improved submodels.

    A recent review sponsored  by  ORNL identified over 350
codes or models that are available which might  have application to
LLW management and disposal.(14)   More recently, R. W. Nelson
identified more than one hundred  groundwater models which might be
useful for LLW management.(15) These reviews clearly point out the
size of the task of reviewing  "all available" models and of culling
and selecting appropriate models  and submodels.  We selected the
following models for in depth  review:

    o    NRG model for waste  classification (NRC-FBD)(16);

    o    NRG model for LLW EIS (NRC-D&M)(17) ;

    o    DOE model for SLD disposal site at Savannah River
         (DOE-SRLX18)

    o    AIF SLD assessment model (AIF-D&M) (19) ;

    o    EPA model for assessing  deep geological disposal of HLW
         (EPA-LOGX20);

    o    EPA model for HLW standard (EPA-HLW)(21); and

    o    EPA models for Clean Air Act (EPA-CAA)(22).

    We are continuing to review new models coming on the scene  to
take advantage of new and improved approaches.   For example, the new
NRG model for systems analysis of SLD (23) looks very  promising and
useful.  It has several submodels which could be very useful to our
SLD model when they are developed.  However, the approach of NRC's
systems is different than EPA's on several points.  Also, it may not
be completed in time for use in developing our  standard.   In any
case, we tried not to duplicate the modeling efforts of others  and,
where  possible, to use the best (for EPA's purposes) of others.

Findings of Model Review

    From our review of the above models, we  found  that none of  them
met EPA's requirements without additions or modifications.  We  did
 find,  however,  that some  existing  submodels  developed  by EPA and
others could be reasonable combined together without major
modifications.  The infiltration,  leaching,  release,  and groundwater
 transport submodels were  an exception.   In  some cases, EPA models and

-------
                                289
submodels were available for adaption and EPA models were selected
for the SLD model to maintain consistency with current EPA policies
and calculational practices.  Figure 2 summarizes the comparison of
the risk assessment models which were reviewed in detail prior to
design of EPA's LLW model.  An estimate of the relative suitability
of submodels within these models for use in EPA's risk assessment
model for SLD are indicated.  A description of key differences
between EPA submodels and approaches and other assessment models
follows.

    Among the risk assessment models reviewed, the infiltration,
leaching and release processes had been lumped together into a
combined simple model represented by a leaching rate which is
proportional to the remaining inventory of radionuclides.  This is
too simple for realistic health risk and benefit/cost assessments.
Such a combined model assumes a constant leaching rate.  This rate
must be guessed at over an infinite period of time.

    A significant problem with this type of model is that it can only
be validated through extensive field and laboratory studies.  It is
doubtful that even this approach can successfully accomodate all of
the boundary conditions and factors involved.  For example, the
leaching rate of radionuclides from the wastes is affected by many
parameters such as trench cap conditons, the amount of infiltration,
the accumulation of water in the trenches, waste form
characteristics, and the properties of the radionuclides.  Therefore,
the reliability of estimates of leaching rates will be extremely
variable, even if they can be determined with field data.

    In addition, leaching and release represent two different
processes.  Predominant factors governing these processes include the
rate of infiltration, accumulation of water in the trench, waste
form, and disposal site characteristics.  The processes may be
controlled through changes in waste preparation and disposal
operations such as trench engineering, waste pretreatment and site
selection.  Use of the existing combined models for reasonably
accurate predictions at a site would require expensive and time
consuming pilot studies before this type of model would have a valid
application.

    A more realistic yet  simple way of modeling these  processes  is  to
subdivide them into three separate components, using dynamic
simulation models  for  the infiltration,  leaching and release
processes.  A major advantage of using individual dynamic simulation
models  for each of the processes  is that  it allows one to simulate
changes in waste treatment and disposal  procedures, which further
allows  an analysis of  the costs and benefits  of  these  changes.
Another advantage  is that the basic input parameters  for  these

-------
                                      SUBMODELS
UKHna — 1 ROEME 1 	 r"5JT
MODELS 1 1 L_^
O3MPAREP ' 	 	 ™"
NRC(fBiD) 1 1

NRC(DiMj "1 1 1

rr-tcp,, - h

EPA(LOGAN)

EPA (HIM) •] [-


EPA(CAA)

1
EPA(LLW) 1
SUBMODEL USED itm-uio IEPA-UN) INK
IN EPA(LLW) MODEL, *"
KMM.I 	 IOBCUMJMWCT _| mfwiauc _|WH
u. \ I tMHBtoBt j nwcmxir | n

1 	 1 1 	 1 	

H H -

—

-• h

H I-

\ \ \
\ 1 1

1 II 1 1
1 II 11
OR (EM-UM) (EM-CM (EP
EDI ORtemni
EF
•""«4- Hs
MWTOKr ] 1 ***

[--



[•-

ZHZ

H

I
1



»-CM (CVK-OI
OR a>
*-MUI| mm.
^IW.lGWIn. - luw _
irm.Ei<»

H H h-



-j j — i i —

H h

D-ID-

I I
J I



* icm-CM fMc.sn.,
OH OR
HI cm mxi Bcrrrai
"*" I nis': ~ ni»:

I H h

-\ 	 1-

I 	 1 I 	

I H H |

H -\

—4 I
I I

l
I
(FTA-CAA lO-A-CM) (CTA-CM)
an
                                                                                                     rvs
                                                                                                     vo
                                                                                                     o
Figure 2.  The availability of submodels from existing risk assessment models and those

           selected  for  use in the EPA risk assessment model.

-------
                                 291
dynamic models can be obtained from simple field measurements and
laboratory tests without great delays or expense for lengthy field
and laboratory studies.

    Some of the groundwater transport which were models reviewed
utilized an analytical approach which is economical to compute but is
limited by a specific boundary condition.  The boundary condition
commonly used for these models, except the D&M model, is an
exponentially depleted release rate.  However, the output function of
radionuclides released from a realistic physical model is normally an
irregular function of time and may not reasonably be approximated by
the above boundary condition.  Therefore, those existing models which
used an analytical approach, were not applicable to our needs.

    The groundwater transport models, which used a numerical
approach, are generally not economical to operate if the numerical
error is to be controlled to an acceptable level.  This is because
improper division of time and space increments may introduce an
unacceptable level of numerical error.  Therefore, the numerical
groundwater transport model is, in general, considered to be
undesirable for use in a risk assessment model.  In addition
computational times and costs of numerical models are generally too
high for risk assessment models where it intended to make a large
number of calcualtions.

             INFILTRATION, LEACHING AND RELEASE SUBMODELS

    The release of radionuclides from the waste disposal trench to
the geosphere is a complicated series of physical processes which
includes infiltration, leaching and release from the trenches.  The
rate of infiltration of water through the trench cover is a function
of meteorological conditions such as precipitation, temperature, wind
speed and air humidity; the characteristics of the trench cover; and
drainage conditions at the site.  The leaching of radionuclides from
the waste is a function of the form and sorption characteristics of
the waste and the duration of its inundation in trench water.  The
rate of release of radionuclides from the trench is a function of the
transmisivity of the host formation, its natural sorptive and
retentive properties and the pressure head of water in the trench.

    The basic system equations for each of these processes are
briefly described in the following sections.   It should be noted that
the basic system equations presented herein cannot be directly
employed as basic equations for risk assessment modeling applications
because they are too sophisticated mathematically and will be
simplified further during development of the EPA model.

-------
                                292
Infiltration Submodel

    The infiltration submodel  is  intended  to  simulate the average
annual rate of infiltration  through  the  trench  cover during and
subsequent to a period of precipitation.   Figure 3 is a schematic
diagram of the hydrological  system associated with the earthen trench
cover.  It includes overland flow and  subsurface flow systems.  The
momentum and continuity equations for  each of these subsystems are
                  ° *S" s(t)  " E(t> " v(t>                   w

for the overland flow subsystem and

       v . -K »<* * y>                                        (3)
                 *y

       3s _   30                                              ,,x
       3t    ^F

for the subsurface flow system.  In the above equations, u and v are
the velocities  of water along  and perpendicular to the slope of the
trench cover, respectively;  h  is the depth of overland flow; D is the
thickness of the trench cover; K is the permeability of the trench
cover; n is the Manning coefficient of roughness of the surface of
the trench cover; R is the rate of rainfall:  E is the rate of
evaporation; ¥ is the soil moisture tension potential of the trench
cover; S is the water content  of the trench cover: x and y are the
coordinates along and perpendicular to the slope of the trench cover
repectively; and t is the time.

    Since the above system equations are extremely difficult to solve
and our primary interest is  determining the average infiltration rate
v, one may simplify the momentum equation by neglecting the
convective acceleration and  local acceleration terms appearing in
Equation 1.  When this is done, Equation 1 can be simplified and it
yields
      q - JL_JtSiSH_                                        (5)


and Equation 2, the continuity equation, can be rewritten as


           R(t) - E(t) - v(t) —2-                           {6)

-------
                         293
                      Free  surface of water
                                                Diversion
               Radioactive waste
               in a LLW trench
Figure 3.   Schematic Drawing of a Two-Dimensional
           Infiltration Model for a LLW Burial Trench

-------
                                294
where q is the overland flow and H is  the  average depth of flow.

    Equations 5, 6, 3 and 4 are the basic  equations for solving the
infiltration of precipitation through  the  trench cover having
overland runoff q, depth of overland flow  H and the moisture content
of the trench cover S as dependent variables.   These system equations
can be further simplified and solved by a  linearized equation for a
tank model.(24)

Leaching and Release Submodels

    Some LLW, such as bulky solids and trash,  are disposed of
directly into the trench in their original form.   Other LLW, which
were originally liquids or slurries, are solidified to immobilize the
radionuclides in suspension or solution before disposing of them into
the trenches.  EPA's risk assessment model is  designed to simulate
the leaching and release of both types of  waste.  The following
discussion uses as its general case, a 55-gallon drum (or larger
container) filled with solidified liquid waste such as evaporator
concentrates solidified in cement and  is applicable for the leaching
and release of one radionuclide.

    The basic mass balance equation governing  the radionuclide
concentration within the solidified liquid waste has been derived by
Hung (13) as

           D_ 32C,
where C^ is the radionuclide concentration in the solidified waste;
De is the effective dispersivity:  R is the retardation factor; £ is
the porosity: X^ is the decay constant; x is a coordinate; and t is
time.  The initial conditions used to solve this equation are
                                                            (8)
    Cl * C0,      for  0 < x £ X.

    c2-o,


and the boundary conditions are

    CL - C2(t)    at x - 0

                                                           (9)
    dCl
      — -  "       at x = L

-------
                                 295
In these equations, C2 is the concentration of the radionuclide in
the trench water: L is the leaching length, which is computed from
the ratio of the volume of the waste container to its surface area;
and C0 is the initial radionuclide concentration in the solidified
waste.

    The computations for solving Equation 7 for solidified wastes are
too complicated to integrate into EPA's risk assessment model.   It
was determined, however, that the processes described by this
submodel can further be approximated and simplified by eliminating
the space independent variable.  Thus Equation 7 can be simplified
from a partial differential equation into an ordinary differential
equation.

    Despite the appearance of adding complexity  to modeling  the
infiltration, leaching and release processes  (i.e.,  going  from a
simple declining percentage model to three  dynamic  simulation
models), in our opinion, it is a necessary  step  if  the  effects of
changes and improvements in radionuclide retention  due  to  waste
solidification and engineering changes to the trench cap are to be
evaluated.  Further, such detailed analyses are  necessary  to provide
data for our cost/benefit analyses to support our standard.
    The dynamic equation  for  the exfiltration of water  from  the
trench can be approximated by

     Qe » 0,       for Hw
Where  At  is  the  bottom area  of  the  trench;  K^  is  the hydraulic
conductivity of  the  host  formation;  and H,^  is  the depth of water  in
trench.   The rate  of radionuclide release can  thereafter be computed
from the  rate of water leaving  the  trench and  the radionuclide
concentration G£ in  the trench  water.

     The above basic  equations consider the  water  which has been
solidified  to improve radionuclide  retention in the waste.  For
wastes which were  originally solid  or are  liquid, the diffusivity
process is  sufficiently large to maintain C]^=C2 at all times.
Therefore,  computation of leaching  and release for these types  of
waste  is  much simpler than for  wastes which have  been solidified.

                     GROUNDWATER TRANSPORT SUBMODEL

     To date, more  than one hundred  and eight groundwater transport
models have been developed.(15) However, some of these models  fail

-------
                                296
to consider the process of radioactive decay and/or sorption and are,
therefore, not suitable for application to health effects
evaluations.  The rest of the models which may be suitable for health
effects evaluation can be subdivided into  two groups, the analytical
model and the numerical model.  In general, analytical models are
limited by specific mathematical boundary  conditions which require
some approximation of actual physical conditions.  As a result, these
models suffer considerable error in simulation due to these
approximations.

    A numerical model requires,  in general, many more computations
than the analytical model does.   In addition the accuracy of
simulation depends greatly on the adjustments of time and space
increments.  Severe errors may result if they are not properly
adjusted.  On the other hand, obtaining proper adjustment of these
increments may,  in some cases, result in excessive computer time for
making the simulation.  Thus, calculational costs may become
prohibitive when the model is applied to long-term simulations such
as health effects evaluations.

    For example, Hung (12) reviewed the use of a finite element model
developed by Duguid and Reeves (25) to simulate the vertical
migration of tritium from a burial trench towards the groundwater
table at the West Valley (NY) disposal site.(26)  This application
required seven minutes of central computer process time on a
UNIVAC-110 to simulate 200 years of real time.  If it is further
assumed that 1000 years of real  time would be required to simulate
the vertical migration of a radionuclide to the nearest underlying
aquifer and its horizontal migration in the groundwater to the
nearest discharge point, 700 minutes of processing time would be
required for simulating ten radionuclides.  This would be equivalant
to $10,000 per run on a commercial computer, a cost which is
unquestionably excessive when a  number of runs may be required to
test the effects of changing several parameters.

    A one-dimensional numerical  model developed for EPA by INTERA
Environmental Consultants, Inc., (27) employs the finite difference
method.  In a review of this model, Hung (12) found that it had a
computer processing cost of $200 for simulating the migration of a
single radionuclide in an aquifer one mile long for a real time of
100,000 years.  This, also, is too costly when many other
radionuclides and pathways must  be considered.

    Several analytical models, for example those by Lester, Jansen,
and Burkholder (28), Ford Bacon, and Davis, Utah, Inc. (16), and
Dames and Moore (17), have made  considerable contributions to
groundwater transport modeling by simplifying the procedures for
simulation.  However, the application of these models to health

-------
                                 297
effects assessments is limited because it requires converting
approximations of physical boundary conditions into mathematical
forms which meet the requirements of an analytical model.  This may
result in considerable error in some cases.

    For the reasons outlined above, the existing numerical and
analytical models seemed either too costly to process or contained
the possibility of introducing unexpected errors of analysis when
applied to health risk calculations.  A more economical and, we
believe, more suitable groundwater transport model for health risk
assessment has been developed by Hung. (12)

    The basic equation for the groundwater transport model is
expressed by
     Q(t>
in which Q and QQ are the rate of radionuclide release at the
discharge end and at the trench; r| is the correction  factor for
health effects; R is the retardation factor: L is the length of the
aquifer; A^ is the decay constant for the radionuclide; and V is the
interstitial velocity of groundwater flow.  Because this is an
algebraic equation, it  is simple and economical to process when
integrated into a health effects assessment model.

    The correction factor   closely approximates and  corrects for the
effects of dispersion.  The correction "factor »; was expressed as


           /" (1/2) /(RP/ir 93 ) Exp{ -Nd 9- {P 9/4R) (R/6-1)2 }dO
       »  -fl - - —      (12)
Derivation  of  17  is  described  in  detail  in a separate paper by
Hung. (12)-

                                SUMMARY

     EPA is  developing an environmental assessment model to support
its  environmental standard for the management and disposal of LLW.
This model  will be used to analyze the potential transport and
release of  radionuclides to the biosphere and to estimate the

-------
                                298
subsequent doses,  health  risks,  and  costs  from disposing of LLW by
shallow land disposal.

    The SLD model  is  a  generic,  analytical systems  model for
analyzing a disposal  facility as a complete "disposal  system" which
includes the waste,  the site geology, hydrology and climate, and all
of the pathways from it.   It can be  used for analyzing the
performance of planned  and existing  disposal facilities, and is
particularly suited  to  evaluating the effects of changes to a
disposal facility  (i.e.,  changes in  waste  form and  packaging, trench
cap design or water  control engineering).   The SLD  model is
sufficiently flexible to  use for analyzing sites having disposal
media with a wide  range of permeabilities  and located  in either humid
or arid climates.

    Improvements have been made  in EPA's SLD model  for modeling the
influx of water into  the  trenches, the  leaching of  radionuclides from
the waste, and the release of radionuclides from the trenches.  These
features are particularly important  for evaluating  LLW management and
disposal practices because they  are  the principle parts of the SLD
disposal system which man can change to increase the retention
capability of a disposal  facility.

    Individual dynamic  simulation models have been  developed for
infiltration, leaching  and release which can use information from
simple laboratory  tests and field measurements for  input data.  These
separate submodels allow  one to  simulate changes in the performance
of a disposal facility  caused by changes in waste treatment and
packaging, trench  cover modification and site water control
engineering.  This further allows one to analyze the costs and
benefits of any changes.   The SLD model also has a new groundwater
transport submodel which  includes a  correction factor for
dispersivity, even though the model  is  an  analytical, one-dimensional
model.  The basic  transport equations for  these four submodels are
presented.  These  basic equations will  be  simplified further during
coding of the model.  The other  submodels  are, for the most part,
being adapted from existing models with minor modifications.

    It is believed that the basic  features of EPA's SLD model, such
as economical operation,  realistic  simulation, use of generally
available data, and  capability of evaluating changes to a disposal
facility, will make  it  a potentially useful model for other
government agencies  and industry who are  interested in assessing the
impact of proposed or existing  disposal facilities.  When completed,
coding and documentation on it will  be made available to interested
parties.

-------
                                 299
                              REFERENCES

1.  U.S. Department of Energy, Report to the President by the
    Interagency Review Group on Nuclear Waste Management, TID-29442,
    (March 1979).

2.  U.S. Department of Energy, National Plan for Radioactive Waste
    Management, Draft No. 4 (Jan. 1981).

3.  Office of the U.S. President, Comprehensive Radioactive Waste
    Management Program - Message from the President of the United
    States. (Feb. 12, 1980).

4.  U.S. Code of Federal Regulations, National Environmental Policy
    Act of 1969. (1969).

5.  U.S. Environmental Protection Agency, Improving Government
    Regulations, Final Report Implementing Executive Order 12044,
    Federal Register, Vol. 44, No. 104, pp. 30988-30998, May 29, 1979.

6.  U.S. Environmental Protection Agency, Considerations of Health
    Benefit - Cost Analysis for Activities Involving Ionizing
    Radiation Exposure and Alternatives, EPA 520/4-77-003, (1977).

7.  U.S. National Research Council, Volume II Decision Making in the
    Environmental PRotection Agency,  A National Research Council
    Report to EPA on Environmental Decision Making, (1977).

8.  International Commission on Radiation Protection, Publication
    26;  Recommendations of the International Commission on Radiation
    Protection,  (Jan. 1977).

9.  U.S. Environmental Protection Agency, Criteria for Radioactive
    Wastes, Recommendations for Federal Radiation  Guidance,
    (Nov.  15  ,  1978).

10. G.L. Meyer,  S.T.  Bard,  C.Y. Hung  and J. Neiheisel, "Modeling and
    Environmental Assessment of Land  Disposal Methods for Low-Level
    Radioactive Waste,"  Proceedings of  Health Physics Society Twelvth
    Midyear Topical  Symposium on Low-Level Radioactive Waste
    Management,  Williamsburg, Viriginia, Feb. 11-15, 1979.

11. U.S. Environmental Protection Agency, Technical Specifications
    for Environmental Asssessment Model^ Development, unpublished
    TAD/ORP document, (1979).

12. C.Y.  Hung,  "An  Optimum Model  to Predict Radionuclide  Transport  in
    an Aquifer  for  Application  to Health Risk Evaluations,"

-------
                                300
    Proceedings of DOE/EPA/NRC/USGS  Interagency  Workshop on Modeling
    and Low-Level Waste Management,  Denver, Colorado,  (Dec. 1980).

13. C.Y. Hung, "Prediction of  Long-Term Leachability of a Solidified
    Radioactive Waste  from a Short-Term Test by  a Similitude Law for
    Leaching Systems," Proceedings of  ORNL Conference on the
    Leachability of Radioactive  Solids, Gatlinburg,  Tennessee,
    (Dec. 1980).

14. Science Applications,  Inc.,  A Compilation  of Models and
    Monitoring Techniques,  Report prepared for DOE,
    ORNL/SUB-79/13617/2, SAI/OR-56-2.

15. R.W. Nelson, "Subsurface Modeling:  Goals,  Expectations and
    Performance",  U.S. Department of Energy Workshop on Low-Level
    Waste Management Environmental Modeling, Bethesda, Maryland
    (July 30-31, 1980).

16. Ford, Bacon, & Davis Utah,  Inc., Screening of Alternative Methods
    for the Disposal of Low-Level Radioactive  Waste, Report prepared
    for U.  S.  NRC, NUREG/CR-0308, UC 209-02, 1978.

17. U.S. Nuclear Regulatory Commission, Report on Waste
    Classification Methodology,  unpublished report by Dames and
    Moore,  Inc., 1980.

18. E.L. Wilhite,  Dose to  Man  Model  for Buried Solid Waste,
    DPST-78-296, Savannah  River Laboratory (June 1978).

19. Atomic Industrial  Forum, Generic Methodology for Assessment of
    Radiation Doses from Groundwater Migration of Radionuclides in
    LWR Wastes at Shallow  Land Burial  Trenches,  a report by Dames &
    Moore,  Inc., June, 1978.

20. U.S. Environmental Protection Agency, Development and Application
    of a Risk Assessment Method for  Radioactive Waste Management,
    EPA-520/6-78-005 (June 1978).

21. U.S. Environmental Protection Agency, Population Risks  from
    Disposal of High-Level Radioactive Wastes  in Geologic
    Repositories. EPA 520/3-80-006  (December,  1980).

22. U.S. Environmental Protection Agency, Nemos  (Atmospheric), TERRA
    (Terrestrial) and ANDROS,  DARTAB and RADRISK (Dose and Risk
    Factor) Codes, under development by ORNL.

23. U.S. Nuclear Regulatory Commission, Systems Analysis of Shallow
    Land Burial, unpublished  report  by Science Applications,  Inc.
    (1980).

-------
                                 301
24. M.H. Diskin., A Basic Study of the Linearity of the
    Rainfall-Runoff Process in Watersheds, Ph.D. Thieses Univ. of
    111. (1964).

25. J. 0. Duguid and M. Reeves, "Material Transport Through Porous
    Media:   A Finite-Element Galerkin Model: Oak Ridge National
    Laboratory Report, ORNL-4929, (1976).

26. INTERA Environmental Consultants, Inc., "Development of
    Radioactive Waste Migration MOdel," Report prepared for Sandia
    Laboratories, (Aug, 1977).

27. D. H. Lester, George Jansen, and H.C. Burkholder, "Migration of
    Radionuclide Chains through an Adsorbing Medium," AICHE Symp.
    Series, 71, 202, (1975).

-------
             RISK ASSESSMENT AND RADIOACTIVE WASTE MANAGEMENT*

                              Jerry J. Cohen
                              Craig F. Smith

     Lawrence Livermore National Laboratory, University of California
                        Livermore,  California 94550
                                 ABSTRACT

     In order to evaluate the hazard of toxic material  buried under-
ground, the concept of geotoxicity is presented.   This  concept per-
mits the development of a tool for quantifying the degree of hazard
of such material, and this tool is formulated as  the geotoxicity
hazard index.  The components of the geotoxicity  hazard index are
described and discussed.  Finally,  some sample results  are offered.
A more detailed description can be found in the report  "A Hazard
Index for Underground Toxic Material," UCRL-52889, June 1980.
                                INTRODUCTION

     The waste management hazard assessment program at LLNL  has  as  its
objective the development and assessment of rationale for radioactive
waste management programs.  The bulk of previous research in waste
management programs has been devoted to the development of methodologies
for handling and disposing of waste.  Our program deals with the develop-
ment of rationale.  We are more concerned with the "Why?", than  the "How
to" aspects of the problem.  Obviously, predictive modeling  plays an
important role in this work.  In this paper we will cover our overall
program and in particular, our efforts in developing a hazard index for
underground toxic material which can perhaps provide an alternative
approach toward defining risks.  A hazard index might also provide  some
insight for waste management programs in determining "how good is good
enough."


                                GEOTOXICITY

     The overall approach to problem definition which we apply is the
geotoxicity study.  This study involves the characterization and asses-
sment of the harmful effects of hazardous material buried in the earth(s
*This work was performed under the auspices of the U. S. Department of
Energy by the Lawrence Livermore Laboratory under contract No.
W-7405-Eng-48.

                                 303

-------
                                 304
 crust  by  either man or nature.  It should be recognized that incorpora-
 tion of toxic materials in the earths crust via radioactive or toxic
 waste  burial does not present unique or unprecedented risks to health and
 the environment.  Nature has done essentially the same throughout  the
 entire history of the earth in toxic mineral deposits.  The geotoxicity
 concept utilizes this perspective to provide useful insights on the fate
 of these  materials in the biosphere.  B. L. Cohen has characterized the
 approach  as using the earth itself as a large analog computer.   In any
 case,  it  is most useful to have a measure or "yardstick" with which one
 may characterize the hazard of toxic material buried underground.   With
 this objective in mind we are developing the Geotoxicity Hazard Index
 (GHI).  This index may be used:

     • to provide an easy to use comparative tool
     •  to gain perspective on geotoxicity
     •  to enable screening of concepts for the disposal of toxic  material
     •  to provide insight on the relative hazard of toxic  components
     •  to permit examination of the source of toxic hazard

     It should  be noted that the GHI is not intended to substitute for
detailed systems modeling used for prediction of specific consequences
and their magnitude.   It can however be used to provide additional
insight on the  severity of the general problem (or lack of  it).  Also,
the results of  systems  analysis  can provide valuable input  to hazard
index determinations.
                         REVIEW OF PREVIOUS WORK

     The approach that was taken in our work was  to  review and evaluate
previous efforts in the  safety index  area  to either  select an existing
index, or identify the requirements for a  new index  if, as it turned out,
a new index would be required.

     Evaluation of previous work resulted  in a classification of indices
as shown on the first portion of SLIDE fl.  The first category consists
of the simple indices which are measures of quantity such as mass,  number
of curies, volume,  heat  output, and so on.  Somewhat more complicated are
the toxic indices which  incorporate specific toxicity in addition to the
quantity of material.  Some examples  of this second  category are the
dilution volume hazard index or the volume of water  required in dilution
to drinking water levels, and the number of lethal doses index.  These
indices are widely used  in the literature.  However, they are measures of
toxicity and not hazard.  Clearly, a  material can be toxic but not  par-
ticularly hazardous if there does not exist a means  for exposure to the
material.  The next level of complexity involves  indices which, in  addi-
tion to toxicity, incorporate availability of the toxic material at the
end of the transport process.  Such indices may incorporate transport and
radioactive decay effects.  The final category consists of complex
indices which include value judgement or decision analysis.  This ap-
proach was rejected for  the geotoxicity hazard index.  The incorporation

-------
                             Slide  1

              Our Review of Previous Work Included:

•    Evaluation of previous indices for radioactive and stable toxic
     materials

               simple indices (mass,  volume,  heat,  etc.)
               toxic indices (toxicity and quantity)
               complex indices (incorporating availability)
               complex indices (including value judgment/decision theory)

•    Determination of desirable features for our index

               simple, easy to use
               containing enough information to be meaningful
               applicable to both radioactive and stable toxic materials
               emplaced by man or nature

•    Identification of factors that could be included

-------
                                 306
of value judgement greatly increases  the level of  subjectivity and  there-
fore, such indices would be difficult to defend.

     Since it was determined from our review that  none of  the previous
indices met our objectives for a  geotoxicity index, we used the review to
focus on the desirable features and factors that could be  included  in the
new index.  The desirable features include simplicity  in form to promote
easy use, and, as a corresponding trade-off, sufficient complexity  to
provide for a meaningful measure  of hazard.  Other desired features are
the capability to handle radioactive  as well as stable toxic material,
and natural deposition as well as emplacement by man.

     There are several factors which  could be included in  an index  of
hazard.  Some of these are shown  on SLIDE #2.  Of  the  parameters listed,
the four on the left were considered  basic requirements for our index;
toxicity, availability,  longevity or  persistence,  and  transformation
product effects are key parameters and have been explicitly incorporated
into the Geotoxicity Hazard Index.

     Chemical and physical form,  and  the effects of biological concen-
tration are factors which are implicitly included  in the previous four
factors.  Perceived harm as opposed to actual detriment could be included
in an index,  but,  as stated earlier,  this approach was rejected.
                       THE GEOTOXICITY HAZARD INDEX

     Based upon the  previously discussed considerations,  a  Geotoxicity
Hazard Index (GHI) was  developed  in  the following form.
     GHI =TIi •  Pi

where

     TI = toxicity index
     P  = persistence
     A  = availability
     C  = buildup correction
     i  = material index

     For a given material i,  GHI  is  determined as a product of four
factors which represent toxicity, persistence,  availability,  and decay
product buildup.  The index can be viewed as a toxicity index which is
modified by three dimensionless parameters to successively incorporate
time, transport, and decay effects.   For a mixture of toxic materials,
the total GHI is obtained through summation of the individual indices for
the toxic components.   Each of the factors of the GHI will be discussed
in turn.

-------
                       Slide 2








    POTENTIAL PARAMETERS FOR A HAZARD MEASURE INCLUDE:








•   TOXICITY                     •   CHEMICAL FORM



•   AVAILABILITY (TRANSPORT)     •   PHYSICAL FORM



•   LONGEVITY      x              •   BIOAGCUMULATION



•   TRANSFORMATION PRODUCTS      •   PERCEIVED HARM

-------
                                 308
Toxicity Index

     The first term in GHI is the toxicity  index,  TI (SLIDE #3).   The
basic toxicity index which was chosen  is the  public  drinking water dilu-
tion volume.  For radionuclides,  it  is the  volume  of water required for
dilution to MPC level, while  for  stable toxic materials,  it is  the
analagous volume relative to  the  EPA drinking water  standards,  or  DWS.
Both apply to public drinking water  supplies.  This  choice of a toxicity
index permits the intercomparison of radioactive and nonradioactive
materials.  While not based on a  rigorous equivalence of  effects,  the
comparability of MPC and DWS  has  been  studied previously  and they  are
probably the best available measures of relative toxicity,  for  chronic,
low level exposures.  SLIDE #3 shows some examples of specific  toxicity
values for several materials.   To produce a toxicity index,  the specific
toxicity values would be multiplied  by the  appropriate material inven-
tory.  It is interesting to note  the values for  1-129 versus elemental
mercury.  On a unit mass basis, these  materials  are  of comparable
toxicity.
Persistence

     The second factor  in the GHI  is  the  persistence factor,  P (SLIDE #4),
For stable materials, the persistence factor has a  value of one;  indica-
tive of maximum persistence.  For  materials which decay, P has a  value
less than one.   Persistence is determined as the average fraction of the
original material which is present in undecayed form,  over a  reference
300 year time frame.  The lower part  of this slide  illustrates the  rela-
tionship between P and  half-life.   Also shown  on SLIDE #4  are some
examples of materials with half-lives running  from  infinity down  to the
24 day half-life of Th-234.


Availability

     The third term in  the GHI is  the availability  factor, A  (SLIDE #5),
that relates the ingestion of a material  to its concentration in  the
earth's crust.   The availability factor,  A, is durther defined in terms
of a modification factor, m, and AQ which is the average availability
of the natural analog of the material under consideration. Ag is the
ratio of the ingestion  rate of material i to its crustal abundance,
divided by the same ratio for naturally occurring radium-226.   The  modi-
fication factor, m,  is  intended to account for differences between  the
actual burial conditions and the average  conditions for the natural
analog.  For example, if a material is significantly less  available than
its natural analog because of isolation barriers, siting,  waste form,
depth, or other condition, the m factor would  be significantly less than
one.  For availability  no better or worse than the  average natural  con-
ditions, m would be given a value  of  one.  Some examples of the
natural availability factor for several elements are also  shown SLIDE #5.
Note that radium, by definition, has  a value of one.   Other elements are

-------
                                Slide  3





                         GHI = TI • P • A  • C






          TI = TOXICITY INDEX:  A MEASURE  OF INNATE TOXICITY



   •   FOR RADIONUCLIDES:           TI = Q/MPC



   •   FOR STABLE MATERIALS:        TI = M/DWS



   •   PERMITS INTERCOMPARISON OF RADIOACTIVE AND NONRADIOACTIVE MATERIALS



   •   EXAMPLES:



MATERIAL             SPECIFIC TOXICITY                SPECIFIC  TOXICITY

                        (per Curie)                        (per  gram)


                     70                               !      9  3
 SR-90             10' m  per Curie                   1.4 x Kr  m   per  gram



 PU-239            2 x lO'' m3 per Curie               1.2 x 10  m3  per  gram

                         O  o                                 /£  *j

 H-3               3 x 10  nr per Curie               2.9 x 10  nT  per  gram


                           6  °i                               23
 1-129             2.5 x 10  nr per Curie             ^.^ x 10  nr  per  gram


                                                              2  3
 MERCURY                  	                         5.0 x 10  m   per  gram



 LEAD                     	                         2.0 x 10  m   per  gram

-------
                                        Slide 4




                                  GIU = TI * P * A .  C






                                 P » PERSISTENCE FACTOR




                    •    FOR STABLE MATERIALS, P  = 1.0



                    •    FOR DECAYING MATERIALS,  P^l.O



                    •    PERSISTENCE IS JUDGED RELATIVE TO A 300 YEAR TIME PERIOD
                                                         EXAMPLES:
 1,0







 0.8









: 0.6
I








 0,4








 0.2
               MATERIAL
   10''   1.0     10    102    ID3


                 Half-Life, years
104    105
                                                                    HALF-LIFE
STABLE LEAD
U -
PU
SR
TH
238
- 238
- 90
- 231*
00
4.5 x 109
88 YEARS
28 YEARS
2k DAYS
1.0
1.0
.38
.14
3.2 x 10"1*
                                                                                                         CO


                                                                                                         o

-------
                      Slide 5


              GHI = TI •  P • A • C



             A = AVAILABILITY FACTOR



A IS THE FACTOR THAT RELATES THE INGESTION RATE OF A MATERIAL


TO ITS CONCENTRATION IN THE EARTH'S CRUST
INGESTION RATE. / CRUSTAL ABUNDANCE-

             :_  / Vsf\uo i ni. nuuiiunnuEr,
              Kfi                    K3
                      INGESTION RATE.. / CKUblAL

A * m AQ , WHERE AQ = INGESTION RATED  / CRUSTAL
   - AQ IS THE AVAILABILITY OF THE NATURAL ANALOG


   - m IS A MODIFICATION FACTOR THAT ACCOUNTS FOR DIFFERENCES


     BETWEEN ACTUAL BURIAL CONDITIONS AND THE NATURAL ANALOG


   - EXAMPLES:
ELEMENT
RADIUM
LEAD
URANIUM
IODINE
A0
1.0
9.7
.27
54.8

-------
                                     312
either more or  less available in nature  than radium, and  these differ-
ences  are reflected in their values for  AQ.
Buildup Correction

      The last component of  GHI is C, the buildup correction factor
(SLIDE #6).  This  factor is used to accommodate  the buildup of decay
progeny more toxic than the parent.  Two examples are the  decay of U-238
through Ra-226 to  stable lead,  and the  transformation of elemental
mercury into the more toxic methylated  form.  SLIDE #6 shows graphically
how  C is determined.   On the  right side of SLIDE #6 is a table which
gives some examples of the buildup correction factors for  several
materials.  It should be noted that in  the majority of cases, C is equal
to one.  Materials with a significant buildup of decay toxicity are the
exception rather than the rule.   U-238  is the most extreme case which we
have identified.
                                 SAMPLE RESULTS

      In order to demonstrate the use of  the Geotoxicity Hazard Index,
several representative toxic material deposits  were considered.   SLIDE #7
shows the results of the GHI calculations for these sample  cases.   While
they  involve several assumptions and simplifications, they  are shown to
illustrate the nature of the calculational results and to indicate  the
types of applications for which the GHI  could be used.  Future effort on
the GHI will include converting the index into  dimensionless form,  and
normalization to avoid the large powers  of ten  seen in these results.

      Another goal toward which future effort will be devoted is improve-
ment  in estimation  of the availability factor (A).  This work is  seen as
an iterative process by incorporating new information into  improved esti-
mates.  Further refinement of  the calculational parameters  will provide
increased confidence in the results.
                                   DISCLAIMER


                This document was prepared as an account of work sponsored by an agency of
                the United States Government. Neither the United States Government nor the
                University of California nor any of their employees, makes any warranty, ex-
                press or implied, or assumes any legal liability or responsibility  for the ac-
                curacy, completeness, or usefulness of any information, apparatus, product, or
                process disclosed, or represents that its use would not infringe privately owned
                rights. Reference herein to any specific commercial products, process, or service
                by trade name, trademark, manufacturer, or otherwise, does not necessarily
                constitute or imply its endorsement, recommendation, or favoring by the United
                States Government or the University of California. The views and opinions of
                authors expressed herein do not necessarily state or reflect those of the United
                Stales Government thereof, and shall not be used for advertising or product en-
                dorsement purposes.

-------
                                           Slide 6

                                      GHI = TI • P • A • C
                                C = BUILDUP CORRECTION FACTOR



                    C IS THE CORRECTION FACTOR TO ACCOUNT FOR THE BUILDUP OF DECAY

                    PROGENY MORE TOXIC THAN THE PARENT
,226
                                                          •PB
                                                            206
                         HG
                    EXAMPLES:
                                                                                                       CO

                                                                                                       CO
          TI
            max
TOTAL
TOXICITY
INDEX
                                     TI
                                 C =
                                       max
                                      TI.
                                                                 MATERIAL
                        U - 238

                        NP - 23T

                        PU - 239

                        ARSENIC

                        LEAD
>  x ]

T9

1.0

1.0

1.0
                             TIME

-------
                    Slide  7
      Results of Sample GHI Applications
     Site                              GHI
Low Level Waste Facility           2,7 v 1011
                                           13
High Level Waste Repository        4.4 x 10
                                           12
Uranium Ore Deposit                3.0 x 10
                                           13
Mercury Ore Deposit                1.3 x 10
                                           12
Lead Ore Deposit                   9.7 x 10

-------
           MODELING IN THE DETERMINATION OF USE CONDITIONS FOR  A
                   DECOMMISSIONED SHALLOW-LAND BURIAL SITE

                                E. S.  Murphy

                          Energy Systems Department
                        Pacific Northwest Laboratory
                        Richland, Washington  99352
                                  ABSTRACT

          Because site stabilization results in smaller dollar costs
     and lower occupational doses than does waste removal,  it is  the
     preferred alternative for the decommissioning of shallow-land
     burial grounds.  Release of a burial  ground for public use fol-
     lowing site stabilization would probably be on a conditional-use
     basis.  A pathway modeling methodology has been developed for
     the evaluation of use conditions for a decommissioned  site.
     The methodology consists of estimating the maximum annual dose
     to the maximum-exposed individual resulting from a specific  use
     scenario.  The methodology is sensitive to site parameters,  to
     radionuclide inventories, and to uncertainties that exist in
     nuclide transport models and in the parameters used in the
     models to represent specific sites.  Several research  needs
     are identified that would result in improved predictions of
     radionuclide migration from shallow-land burial grounds.


                                INTRODUCTION

     A study,  ' sponsored by the Nuclear Regulatory Commission,  was  per-
formed to estimate dollar costs and public and occupational radiation  doses
resulting from decommissioning reference shallow-land burial grounds  by
the decommissioning alternatives of 1) site stabilization plus long-term
care and 2) waste removal.  On the basis of minimum cost and occupational
radiation dose, site stabilization is the preferred decommissioning alter-
native.

     Site stabilization implies that all or some portion of the waste  is
left in place, and release of the site would probably be on a conditional-
use basis.  A methodology was developed for evaluating scenarios  for  the
conditional or unrestricted release of a burial ground after burial opera-
tions cease.  The methodology uses pathway modeling techniques to estimate
the maximum annual dose to the maximum-exposed individual for various  site
use scenarios.  The maximum annual doses for the given use  scenarios  are
compared to determine which scenario is more appropriate.

     Results and conclusions derived from this pathway analysis approach
depend on the site characteristics, on the burial ground radionuclide
inventory, on the mathematical models used to evaluate potential  exposure
pathways and estimate radiation doses, and on the parameters used in  the
models.  Uncertainties exist in the models used to estimate radionuclide

                                   315

-------
                                  316
transport and in the parameters and boundary conditions used in the models
to represent specific sites.  These uncertainties can affect the validity
and seriously limit the usefulness of the pathway methodology approach.


                        DECOMMISSIONING ALTERNATIVES

     Decommissioning is defined as the measures taken at the end of a
facility's operating life to ensure that future risk to public safety from
the facility is within acceptable bounds.   For a shallow-land burial site,
the basic decommissioning alternatives are 1) site stabilization followed
by a period of long-term care and 2) waste removal.

     Site stabilization involves the use of engineered procedures to reduce
the rate and extent of radionuclide migration from buried wastes left in
place in a decommissioned shallow-land burial ground.  Potential site stabi-
lization activities include:

  •  engineered routing/flow control of ground and surface water

  •  modification of trench  caps to minimize water infiltration into the
     trenches

  •  stabilization of the land surface and erosion control

  •  grouting and/or use of  chemical additives to reduce the mobility of
     the waste

  •  control  of plants and animals that might disrupt surface stabilization
     measures or transport radioactivity from the trenches

  •  erection of physical  barriers to control human activities at the site.

     Site stabilization implies that all  or some portion of the waste is
left in place, and onsite public activities are restricted to land uses
that do not result in excessive public exposure to radiation and that do
not compromise stabilization procedures or waste confinement.

     Site stabilization is followed by a period of long-term care of the
site.  Long-term care activities include site surveillance and maintenance,
environmental monitoring, and administrative procedures for control of the
site.  Long-term care continues until  it is determined that the radioactivity
at the site has decayed to the point where the waste no longer poses a
significant radiological hazard, or until  additional actions are taken to
reach this point.

     Waste removal involves  the exhumation of buried waste and contaminated
soil, packaging of the waste and soil, transportation, and reburial of the
waste at another disposal  site.  Because of the potential for significant
radiation dose to decommissioning workers and the high dollar costs, waste

-------
                                   317
removal would likely be considered only in situations where site stabi-
lization and long-term care are insufficient to ensure the continued
capability of the site to provide adequate containment of the buried waste.

     At a particular site, combinations of decommissioning activities may
be necessary to ensure that future risk from buried radioactive material
is within acceptable bounds.  Combinations of decommissioning alternatives
may be necessary when individual burial trenches are known to contain high
concentrations of transuranic waste or waste mixed with organic complexing
agents.  In these instances, the removal of the waste from part or all of
a particular trench or trenches may be a requirement in conjunction with
stabilization of the rest of the site.  Partial waste relocation may also
be required if burial trenches are located in an area with geologic or
hydrologic characteristics that make stabilization a costly or technically
unfeasible alternative.  In this case, the waste removed from one trench
might be reburied in another onsite trench or it might be transported to
another disposal site.


                     DECOMMISSIONING OF REFERENCE SITES

     Two generic burial grounds, one located on an arid western site and
the other located on a humid eastern site, were used as reference facilities.
The two burial grounds were assumed to have the same site capacity for waste,
the same radioactive waste inventory, and similar trench characteristics and
operating procedures.  The climate, geology, and hydrology of the two sites
were chosen to be typical of actual western and eastern sites.  Each site
description provided a basis for evaluating decommissioning methods and
costs, and for estimating possible environmental impacts.

     Costs of stabilization of the reference sites were estimated to range
from $0.5 million to $8 million depending on the site and on the stabiliza-
tion option chosen.  Annual long-term care costs were estimated to range
from $80,000 to $400,000.

     Stabilization of a burial ground involves modification of and addition
to surface soils, but no intentional uncovering or exhumation of buried
waste.  There is no transportation of radioactive waste.  Therefore, routine
site stabilization operations are not expected to result in any significant
radiation dose to the general population.  Modest exposure of decommission-
ing workers is expected; however, the dose rates would be less than those
experienced during waste burial operations.

     Waste removal was estimated to cost about $1.4 billion and to require
approximately 1000 man-years of effort extending over a period of more than
20 years.  Approximately 93% of the cost of waste removal was found to be
associated with the packaging, shipment, and offsite disposal of the
exhumed waste.

     The 50-year committed dose equivalent to the affected population from
routine waste removal operations was estimated to be very small compared to

-------
                                  318


the 50-year dose from natural background radiation.  However, the occupa-
tional dose from waste removal was estimated to be approximately 30,000
man-rem,  indicating that this decommissioning alternative can be very costly
in terms  of worker exposure to radiation.


                       ANALYSIS OF RELEASE CONDITIONS

     Waste removal  implies the complete removal of the buried waste and
contaminated soil and can result in release of the site for unrestricted
public use.  Site stabilization implies that all or some portion of the
buried waste is left in place.  Release of a stabilized site would presumably
be on a conditional-use basis, with public activities restricted to those
that limit exposure to radiation and that do not compromise stabilization
procedures or waste confinement.  An important factor in determining the
appropriate decommissioning alternative for a particular site is the esti-
mated risk to the public from the decommissioned site.

     A methodology  was developed for predicting conditions for the conditional
or unrestricted release of a burial  ground after burial operations cease.
The methodology consists of estimating the calculated maximum annual dose
to the maximum-exposed individual resulting from a specific release scenario.
The maximum annual  doses from different scenarios are compared to determine
which scenario is more appropriate.   Doses to the maximum-exposed individual
are calculated for  all  important exposure pathways using state-of-the-art
modeling techniques.   For the western site the important pathways are
inhalation, direct  external  exposure, and ingestion of foods grown on the
released decommissioned burial ground.  For the eastern site, additional
exposure pathways of importance are ingestion of aquatic foods from a nearby
river and ingestion of water from a well drilled into an aquifer beneath the
site.

     Land-use scenarios are evaluated to determine release conditions
following stabilization of the reference burial grounds.  Examples of path-
way analysis results are shown in Table 1.  The dose numbers in the table
are calculated maximum annual total  body doses to an individual who lives
and works on a conditionally released burial site having various use
restrictions.

     The first two  values in the table are estimated maximum annual total
body doses at the western and eastern sites during the first 50 years after
site release assuming that individuals living on a site do not engage in
excavation or drink water from onsite wells.  The major contributors to
dose are 63ni  (assumed to be present in reactor decommissioning waste
buried on the site) and 210Pb (a radioactive daughter of 226Ra).  The third
value is the estimated maximum annual total body dose at the western site
assuming that excavation is permitted.  Most of the dose from site excavation
is from external exposure to the gamma radiation from 137cs.

-------
                                  319
       TABLE 1.  Maximum Annual Total Body Doses for Conditional
                 Release of the Reference Burial Grounds With
                 Different Use Restrictions(a)

                                                  Maximum Annual
                                                  Total Body Dose
        	Release Scenario	      (mrem)	


          No excavation and no drinking of water             4
            from onsite wells - Western Site

          No excavation and no drinking of water            23
            from onsite wells - Eastern Site

          Excavation permitted - Western Site              380

          Total erosion of trench overburden^ '          1 300
            - Western Site

          Drinking permitted from contaminated well    120 000
            - Eastern Site
          (a) Conditional release assumed to occur 200 years after
              site closure.
          (b) Calculated at 450 years after site release (650 years
              after site closure).
     The fourth value in the table is the estimated maximum annual total
body dose at the western site assuming total erosion of the trench over-
burden.  Total erosion is estimated to occur approximately 450 years after
site release, based on the assumed erosion rate at the western site.
Radionuclides that contribute most of the dose are 230Th, 226Ra, 210pD, 235u,
238u, 239pu, 240pu and 24lAm.  While it is unlikely that site conditions
would result in total overburden removal, this scenario demonstrates the
importance of maintaining an adequate layer of overburden to prevent both
crop root penetration into the waste and radionuclide dispersal by human
activities or wind action.  Institutional controls may be necessary to
restrict human activities at a conditionally released site and to maintain
an adequate depth of overburden.

     The last value in the table is the estimated maximum annual total body
dose at the eastern site assuming that the site resident obtains all of his
drinking water from a shallow well drilled into the contaminated aquifer
beneath the site.  The dominant radionuclides contributing to the large
annual dose are 14C, 63N1, 90$r, 137Cs, and 226Ra.

-------
                                  320
      For the reference burial sites, the pathway modeling approach indicates
 some  restrictions that should be placed on a conditionally released site.
 Conditional release of the western site following site stabilization would
 be  possible provided that the following actions are enforced:

  •  stabilize the ground surface to minimize surface erosion

  •  control the type of farming or other land use to prevent the growth
      of deep-rooted plants

  •  restrict activities that result in excavation of the site.

 The conditional release of the decommissioned eastern site would include
 enforcement of the restrictions described for the western site,  plus the
 following additional  restrictions:

  •  prohibit the use of water from shallow wells drilled on or  near the
      site

  •  maintain site drainage features to control surface water runoff and
     to prevent inundation of burial trenches with water

  •  stabilize the waste to minimize leaching to the aquifer, or control
     the use of aquatic organisms and water from nearby streams.

     For the site stabilization alternative, unconditional release of the
reference sites,  even after a long-term care period of 200 years, would
not be feasible because of the significant inventory of long-lived radio-
nuclides in the buried waste.

     The methodology  developed for defining site use conditions  is both
site- and inventory-specific.   The use restrictions listed above for the
reference sites are given only as examples of results to be expected from
the application of the modeling technique.  The methodology must be
reapplied and the radiation doses recalculated for each burial ground having
a different radionuclide inventory and different site characteristics.

     In addition  to a dependence on site characteristics and radionuclide
inventories, the  results and conclusions of the pathway-modeling approach
depend on the mathematical  models used and on the parameters and boundary
conditions employed with the models to represent specific site character-
istics.  Many uncertainties exist in the radionuclide transport  models and
in the parameters (e.g., distribution coefficients, dispersion coefficients,
leach rates) used with the models.   Because of these uncertainties, a
generally conservative approach was attempted that may result in conservative
(high) estimates  of doses to the maximum-exposed individual.

-------
                                   321


                        TRANSPORT MODEL UNCERTAINTIES

      Radionuclide migration via the groundwater pathway was simulated by
 use of  the MMT  (Multicomponent Mass Transport) model.(2)  There are no
 established models for treating overland flow.  To account for overland flow,
 the conservative approach is taken that the burial trenches are saturated
 with  water.  All of the water flowing through the trenches arrives at the
 surface and flows overland to the nearby river.  No significant sorption is
 assumed during  overland flow.  These factors result in a conservatively
 high  estimate of the radioactivity leached to the river by this pathway.

      Although there are some uncertainties in nuclide transport models, the
 major uncertainties are in the parameters and boundary conditions fed into
 the models to represent specific sites.  Order of magnitude uncertainties
 exist in values for soil permeability, dispersion coefficients, distribution
 coefficients (Kd), and leach rates.  Some of these parameter values
 (e.g.,  Kd's and leach rates) are relatively easy to measure in the labora-
 tory, but extremely difficult to measure under field conditions.

      Because of the great difficulty of direct field measurement of Kd,
 values  determined by laboratory measurement are normally used to calculate
water transport of radionuclides.  However, the validity of applying labora-
 tory  values to field situations is questionable because of the strong
 dependence of measured values of Kd on the physical and chemical conditions
 of measurement.  Variables such as mineralogy, particle size, nature of
 solution, and chemical nature of the radioactive species can strongly affect
 the measured values.  Table 2 gives examples of published values of Kd used
 in several studies of radionuclide migration.  It is apparent that, for some
radionuclides, significant differences exist between values reported by dif-
 ferent  authors.

      Leach rates are influenced by many factors, including the character-
 istics  of the radionuclide and of the waste material, the properties of the
leachant, frequency of leachant changing, leaching time, and temperature.
Specific field data on the Teachability of radionuclides from waste buried
 in shallow-land burial sites are not available.  The effects of chelates on
the mobility of the radionuclides in buried waste are just beginning to be
studied.  Published leach-rate data come mainly from laboratory measurements
in which small  samples are leached by distilled water or by actual  or simula-
ted disposal-environment water.   Laboratory leach-rate data are summarized
in a recent Brookhaven National  Laboratory report.(3)  This report  gives
leach-rate values for cement monoliths that range from 10-1 to TO"9 g/cm2-day.

     An example of the effect of a change in leach  rate on radionuclide con-
centration in a surface stream located 1  km from the reference eastern burial
ground  is shown in Figure 1  for a long-lived radionuclide with a small  Kd
value.  The nuclide chosen is "Tc (Kd = 0).   Increasing the assumed leach
time* by an order of magnitude results in almost an order of magnitude decrease
in the maximum radionuclide  concentration in the ground water at the point
  The leach time is the time required for the radioactive material  to  migrate
  from the burial ground at a constant leach rate.

-------
         TABLE  2.   Reported Values  of Distribution  Coefficients for  Selected Elements


Element
Co
Sr
I
Cs
Pu
Am


. .
ARHCO

5-38

12-200
200
1200


NECO NECO/ \
May(b) June^ '
138-593 700-800
3-17 3-6

4000-9000 27-50




Dames. &.
Moore^'
75
20
0
200
2000
2000
Kd U/kg)
Leddicotte^e;
Dry
Site
1000
10
5
1000
1000
1000


Leddicottetej ,^
Humid
Site
2500
50
25
2500
2500
2500
Staleyv"
Sand
100
2
0.1
20
200
70

Staley^t;
Silt &
Clay
1000
20
1
200
2000
700
(a)  R.  C.  Arnett,  D.  J.  Brown  and  R.  G.  Baca,  Hanford  Grpundwater Transport  Estimates for Hypo-
    thetical  Radioactive Waste Incidents.  ARH-LD-162,  Atlantic  Richfield Hanford Company,
    Richland, Washington, June 1977.
(b)  Sheffield. Low-Level Radioactive  Waste Disposal  Site  -  Report of Additional Investigations
    Conducted During  February  and  March. 1978.  Nuclear Engineering  Company.  Inc. Louisville.
    Kentucky, Revised May 5.  1978.(Distribution  coefficients  for  sands.)
(c)  Sheffield. Illinois  Low-Level  Radioactive  Waste  Disposal  Site - Responses  to Questions,
    Nuclear Engineering  Company,  Inc.,  Washington, DC, June 29,  1978.(Distribution coefficients
    for loess, till,  and shale.)
(d)  A.  E.  Aikens,  Jr., R. E.  Berlin,  J.  Clancy and 0.  I.  Oztunali,  Generic Methodology for Assess-
    ment of Radiation Doses from Groundwater Migration of Radionuclides in LWR Wastes in Shallow-^
    Land Burial Trenches, AIF/NESP-013,  Prepared by  Dames and Moore for Atomic Industrial Forum.
    Inc.,  January 1979.   (Estimated values for a typical  desert soil.)
(e)  G.  W.  Leddicotte  and W. A. Rodger,  "Suggested  Quantity and  Concentration Limits to be Applied to
    Key Isotopes in Shallow Land Burial," in M. W. Carter,  et al.,  Management  of Low-Level Radioactive
    Waste, Vol. 2, New York,  Pergamon Press, 1979.
(f)  G.  B.  Staley, G.  P.  Turi  and D. L.  Schreiber,  "Radionuclide Migration from Low-Level Waste:  A
    Generic Overview," in M.  W. Carter, et al., Management of Low-Level Radioactive Waste. Vol. 2,
    New York, Pergamon Press,  1979.  (Distribution coefficients for sand were  selected to be  near the
    low end of the range of values reported in NUREG-0140.   Kd  values were increased by a factor of 10
    for slit, clay, loess, and till.)
CO
ro
r>o

-------
    8.000E-05
UJ
ce.
=)

g  6.000E-05
o
     4.000E-05
     2.000E-05
                                                       100 YEAR LEACH TIME
                                                  1000 YEAR LEACH TIME.
                                                                  — --^	— —
            300
        FIGURE 1.
 320
340
360
380      400      420

    TIME (YEARS)
440
460      480      1500
Predicted Technetium-99 Concentration  in  the Ground Water  at  the Point of
Discharge to  the  Surface Stream
                                                                                                               CO
                                                                                                               ro
                                                                                                               CO

-------
                                   324
of discharge to the surface stream.   It also results in a significant post-
ponement of the time when this concentration attains its maximum value.  For
a radionuclide with a large K^ value,  such as 230ih, neither the maximum
radionuclide concentration nor the time when the concentration attains its
maximum value are significantly affected by an order of magnitude increase
in leach time.

     To improve the reliability of modeling studies of radionuclide trans-
port, the need exists to demonstrate (verify) models on actual field sites
and to develop better methods for measuring critical parameters such as
distribution coefficients and leach rates under actual  field conditions.


                                 CONCLUSIONS

     Decommissioning of reference shallow-land burial grounds by the alter-
native of waste removal  is estimated to be much more expensive and to
result in a much larger dose to decommissioning workers than site stabili-
zation.  Therefore, site stabilization is assumed to be the preferred decom-
missioning alternative.   Site stabilization would probably result in the
conditional release of a decommissioned site with use restrictions that
would prevent excessive public exposure to radiation and that would ensure
the adequate confinement of the buried waste.

     A methodology has been developed  for evaluating use restrictions for
the conditional  use of a burial ground after burial operations cease.  The
methodology uses pathway modeling techniques to estimate the calculated
maximum annual  dose to the maximum-exposed individual.   Results and con-
clusions of the pathway modeling approach depend on the site characteristics,
on the radionuclide inventories, and on the mathematical models and the
parameters employed with the models to represent specific site character-
istics.

     Some uncertainties exist in nuclide transport models and major uncer-
tainties exist in parameters and boundary conditions used in the models to
represent a specific site.  Several  research needs can be identified that
would result in improved predictions of radionuclide migration from
shallow-land burial grounds.  These research needs include:

 1.  development of transport models for overland flow

 2.  verification of models by comparison of predicted results with experi-
     mental results for real sites

 3.  identification and quantification of differences between field- and
     laboratory-measured values of distribution coefficients

 4.  determination of leach rates for specific radionuclides under field
     conditions

 5.  quantification of the effect of complexing agents on the mobility and
     transport of radionuclides from shallow-land burial grounds.

-------
                                 325


                                REFERENCES


1.  E. S. Murphy and G.  M. Holter, Technology,  Safety and  Costs  of Decom-
    missioning a Reference Low-Level Waste Burial  Ground.  NUREG/CR-0570,
    Pacific Northwest Laboratory for U.S.  Nuclear  Regulatory  Commission,
    June 1980.

2.  S. W. Ahlstrom,  et al.,  Multicomponent Mass Transport  Model:   Theory
    and Numerical  Implementation (Discrete-Parcel-Random-Walk Version).
    BNWL-2127, Pacific Northwest Laboratory,  Richland,  Washington,
    May 1977.

3.  P. Colombo and R.  M.  Neilson,  Critical  Review  of Properties  of Solidified
    Radioactive Waste Packages  Generated at Nuclear Power  Stations.  BNL-
    NUREG-50591, Brookhaven  National  Laboratory, Upton, New York,  1976.

-------
             A SYSTEMS MODEL FOR EVALUATING POPULATION DOSE
                    FROM LOW-LEVEL WASTE ACTIVITIES
                  J. A. Stoddard          D.  H.  Lester

                      Nuclear Technology Division
                       Science Applications,  Inc.
                           1200 Prospect St.
                       LaJolla, California 92038
                                ABSTRACT

               This  paper  briefly  describes  a  computer   code
     developed  to  evaluate population dose or maximum individual
     dose from the various activities associated with shallow land
     burial   of low-level wastes.  The code, called BURYIT, treats
     a wide variety of  scenarios  to  determine  the  effects  of
     various  parameters  associated  with  siting,  packaging and
     burial  procedures  on  potential  public  exposure.    In  the
     analysis of radionuclide dispersal, BURYIT treats unsaturated
     zone seepage, aquifer transport, wind erosion and atmospheric
     transport.   Population  exposure pathways considered include
     direct exposure to undispersed  wastes,  direct  exposure  to
     contaminated  air,  direct  exposure  to contaminated ground,
     inhalation of contaminated air, and ingestion of contaminated
     food, water and milk.
                              INTRODUCTION

          As  part  of  an  effort to  develop analysis tools  to  support
rulemaking  and  evaluation   of  license  alternatives,   the   Nuclear
Regulatory Commission* has sponsored the development of a systems  level
computer code (BURYIT) for use in assessing potential   population   dose
from  the  various activities associated with low-level waste disposal.
This code  is currently under development by Science Applications,  Inc.
(SAI).   When  completed, it  will   derive  either  population  dose  or
maximum individual dose for  a  wide   variety  of  shallow  land  burial
problems  and  conditions   and will be  used to evaluate the  effects  of
several  parameters  including  site  location,   site   design,   package
integrity and operational procedures.
*NMSS, Low-Level Waste Branch,  J.   D.  Thomas, Project  Officer.
                                  327

-------
                                  328
                          PROBLEM DESCRIPTION

           The  problems  which must be analyzed are varied and broad in
 scope.  Many different release scenarios, ranging from  short  term  to
 long  term  and involving different pathways, must be analyzed.

           The  scope  of the problems was established early in the code
 development project.  At that time, we visited five shallow-land burial
 sites   (Maxey  Flats,  Barnwell,   Rich!and  (NECO),  Han ford  (DOE) and
 Beatty) and reviewed available literature to determine  the  nature  of
 low-level  wastes,  their  sources,  and the most likely pathways which
 would lead to  public  exposure.    We  identified  over  fifty  release
 scenarios  for  the  various phases of low-level  waste disposal.  These
 scenarios ranged widely in characteristics  and indicated the need  for
 a  very  versatile code.   For example, we wanted to be able to evaluate
 direct radiation exposure from wastes in transport or in an  open  pit,
 population  exposure  from contamination released by explosion or fire,
 and long-term exposure from nuclides leached from the site over periods
 of thousands of years.
                            CODE DESCRIPTION

          Our  approach to solving the variety of release scenarios was
to  develop  a  modular  code  which  is  software  linked  within  the
executive.    This   software   linking  enables  BURYIT  to  call  the
appropriate analysis modules in the proper  order  for  the  particular
scenario.   For  example, for a scenario involving leaching of nuclides
to the surface followed by wind erosion and atmospheric transport,  the
scenario-related  input  data  includes  a specification for successive
calls to the transport module for unsaturated soil,  the  wind  erosion
module, the air transport module and the dose evaluation module.

          The  executive, in solving a problem, may call  some or all  of
the following modules:

PREDOS - This module  loads  required  data  for  the  nuclides  to  be
analyzed.   These  data  include half lives, dose factors, and transfer
coefficients (for example, from grass to milk via cow).  The  data  are
extracted  by  direct access reads from a disk file containing data for
about 350 radionuclides.

UNSAT - This module, which was derived from the HYDRO code,  calculates
the  transport of radionuclides in the unsaturated soil zone.  Given an
isotope distribution in layered soil  of specified  characteristics,  it
calculates  the  nuclide  distribution  as a function of time, treating
nuclide discharge to an aquifer and  surfacing  by  evapotranspiration.
The  HYDRO code, from which UNSAT was derived, was developed in 1979 by
SAI[1] based on an irrigation control   model  developed  for  the  U.S.
Environmental Protection Agency.[2]

-------
                                  329
AQUIFR - This module, which was developed from the PNL GETOUT  code,[3]
calculates nuclide transport in the saturated soil zone.  An aquifer is
treated as an  one-dimensional  flowing  body  with  axial  dispersion,
sorption  in the soil, and nuclide decay.  The calculated output is the
aquifer discharge rate for each radionuclide.

EROSIO - The EROSIO module calculates the rate at which soil is  eroded
by   wind  action.   When  the  soil  is  contaminated,  perhaps  by  a
combination of leaching with evapotranspiration,  EROSIO  provides  the
quantity  of  radioisotopes released for atmospheric transport.  EROSIO
is a modification of the U.S.  Department of  Agriculture  WEROS  (wind
erosion) code.[4]

ATMOS  -  The  ATMOS  module  calculates  the  atmospheric transport of
contaminants by use of standard Gaussian  plume  techniques.   For  the
calculation   of  population  dose,  sector  averaged  air  and  ground
radionuclide concentrations are calculated.  Annual average results can
be  calculated based on an array of input weather conditions or results
can be calculated for a specified input weather  condition.   Both  wet
and dry deposition and plume depletion are taken into account.

DOSE   -  This  module  calculates  the  maximum  individual  dose  and
population  dose  for  the  following  exposure  pathways:  (a)  direct
exposure  to  an  airborne  cloud of contamination, (b) direct exposure
from ground contamination, (c) inhalation of airborne contamination and
(d)  ingestion  of  contamination.   The  inhalation  pathway  includes
resuspended contamination as well  as nuclides  in  the  initial  plume.
The  ingestion  pathway  includes water via contaminated aquifer, leafy
vegetables contaminated by atmospheric deposition, produce contaminated
by  root  uptake, and beef and milk contaminated by animal ingestion of
contaminated pasture grass.

          For the maximum individual dose, the food and milk  ingestion
are  based  on quantities specified by NRC Regulatory Guide 1.109.  For
calculation of the population  dose,  it  is  assumed  that  all  foods
produced  are  eaten  and thus contribute to population exposure.  This
production  based  (rather  than  consumption  based)  ingestion  model
eliminates  the  need for concern about foods exported from or imported
into the analysis area.

DIRECT - This subroutine uses formulas from Rockwell[5] and Foderaro[6]
to  calculate  the  direct  gamma  radiation  exposure from undispersed
wastes.  Example problems are direct radiation from wastes in the  open
pit   or  shielded  or  unshielded  individual  waste  packages  during
transportation or processing.

-------
                                  330
                               DATA BASES

          BURYIT extracts data from three data bases:  a  scenario  data
base,   a  radionuclide  inventory  data   base,  and  a  data  base  of
radionuclide properties.

          The scenario data base contains release  pathway  descriptors
for  over  fifty  different release scenarios.  These  release scenarios
were derived for the following phases of  operation and  post-operation:
waste  packaging,  waste transportation,  waste arrival  phase (pre-entry
inspection), on-site handling and trench   emplacement,   storage  in  an
uncovered  trench,  waste  burial   (including backfill  and compaction),
post-burial  (early time period,  <100 years)  and post-burial  (later time
period,  >100  years).    The  100-year breakpoint  corresponds  to  an
anticipated change in post-burial  institutional controls.

          The radionuclide inventory data base  contains  source  terms
for five sets of wastes: control  rods and high activity components from
light water  reactors (LWRs);  miscellaneous   wastes  (resins,  sludges,
trash, etc.)  from LWRs;  LWR decomissioning wastes; miscellaneous wastes
from  hospitals   and   similar    institutions;   and    typical   waste
concentrations  in a low-level  waste shallow-land burial trench.

          The  radionuclide  properties   data file contains half lives,
dose factors  and gamma  emission  data (by  gamma energy  group)  for  over
350 radionuclides.
                            INPUT AND  OUTPUT

          Input  data  falls  into the  following categories:  scenario and
inventory selections,  weather  data, geologic data,  demographic data and
option  switches.    Of  these,  the   geologic data  tends to be the most
bulky.  We may, in  the  future,  develop   a   geologic   data  file  with
element  and  soil   type  dependent   coefficients from which BURYIT can
extract the required soil  data.

          The output is  the  population or  maximum individual  dose given
by  organ,  pathway, isotope,  population age group, and radial distance
from the source. Additional detail from the analysis   process  can  be
output on option.
                       CODE  EXECUTION  EXPERIENCE

          BURYIT  has  been   used   to  generate  some sample cases and to
perform some sensitivity studies,  but  has  not yet been used to  analyze
any   existing   or   proposed  low-level   waste   disposal  facilities.
Execution times vary from a  few CPU seconds on  a   DEC-10  computer  for
simple problems involving the calculation  of only direct gamma exposure

-------
                                 331
to nearly an hour for complex problems  involving  many  radionuclides,
many  soil  layers  and hundreds of repeated wet/dry cycles for chronic
release problems involving groundwater transport and  time  periods  of
thousands  of  years.   Typical problems require less than five minutes
CPU.

          In the  sensitivity  study,  major  variables  affecting  the
result  for  a given airborne release scenario were the population, the
agriculture and the weather pattern.  Major water  pathway  sensitivity
parameters   were   aquifer   travel  time  and  turnover  and  nuclide
solubility.

          Test cases for generic facilities indicated  relatively  high
population  doses  from  incinerator  accidents  involving the airborne
release of respirable contamination.  Chronic releases via  soil  water
pathways tended to be small because of nuclide binding in the soil.
                               REFERENCES
1.   B.   Amirijafari  and  B.   Cheney,  "Simulation  of  Radionuclide
     Transport in the Unsaturated Zone Beneath  the  Radioactive  Waste
     Management   Complex,"   report   submitted  to  EG&G  by  Science
     Applications, Inc.  (June 1979).

2.   L.  G.  King and R.  J.  Hanks, "Irrigation Management for Control
     of  Quality  of  Irrigation  Return  Flow,"  prepared  for  USEPA,
     EPA-R273-265 (June 1973).

3.   W.  V.  DeMier, M.  0.  Cloninger, H.  C.  Burkholder, and P.   J.
     Liddell,  "GETOUT - A Computer Program for Predicting Radionuclide
     Decay Chain Transport Through Geologic Media,"  Pacific  Northwest
     Laboratory, PNL-2970 (August 1979).

4.   N.  P.  Woodruff, and F.  H.  Siddoway, "A Wind Erosion Equation,"
     Soil.  Sci.  Soc.  Amer.  Proc. 29 602-608 (1965).

5.   Rockwell, Reactor Shielding Design Manual. TID-7004 (March 1956).

6.   A.  Foderaro, The Photon  Shielding  Manual,  Pennsylvannia  State
     University (1975T.

-------
                       MODELS AND CRITERIA FOR LLW
                          DISPOSAL PERFORMANCE

                             Craig F. Smith
                             Jerry J. Cohen

    Lawrence Livermore National Laboratory, University of California
                          Livermore, CA  94550
                                ABSTRACT

    A primary objective of the Low Level Waste (LLW) Management
Program is to assure that public health is protected.  Predictive
modeling, to some extent, will play a role in meeting this objective.
This paper considers the requirements and limitations of predictive
modeling in providing useful inputs to waste management decision
making.  In addition, criteria development needs and the relation
between criteria and models are discussed.
                                 PREFACE

    This paper is a compilation and summary of informal remarks
delivered to the LLW Modeling Workshop at Denver, CO in December, 1980.
                              INTRODUCTION

    Calculational models for the prediction of the consequences of
waste management activites should be consistent with applicable
criteria; however, definitive criteria for judging the effectiveness
of low-level waste management activities have not, as yet, been
established.  Despite the absence of official criteria, there has been
a considerable amount of work on model development.  Aside from the
development of definitive criteria, there are several other areas that
should be defined before beginning model development.  Model
development requires consideration of what is to be predicted, how
accurate it must be, and how it can be validated.
    Over three hundred existing models applicable to low-level
radioactive waste have been identified.!  Yet, an extensive degree
of effort in model development continues.  Given this situation, one
might logically ask the questions:  Why isn't what we have good
enough?;  What are the specific gaps that need to be filled?; and How
can we recognize an acceptable model should one evolve?  These
questions should be resolved before more extensive model development
takes place.  Failure to do so could result in an endless quest for
some unattainable state of perfection.
*This work was performed under the auspices of the U.S. Department of
Energy by the Lawrence Livermore National Laboratory under contract
No. W-7405-Eng-48.

                                  333

-------
                                 334
                          PATHWAYS  FOR  EXPOSURE

    The  determination of what processes are to be modeled requires the
 identification of pathways for human exposure to low-level waste
 materials.  Figure 1 illustrates the major pathways of concern.
    Water pathways are generally believed to be the most important
 pathways for human exposure.  The two forms of water pathways are
 surface  runoff and groundwater transport.  Climatic conditions  (arid
 versus humid) strongly influence the importance of these pathways.
    The  airborne pathways can result from operational release during
 waste disposal activities, release to the air through erosion,  and
 release  in conjuction with the evolution of gases in the waste
 material.  The erosional  release pathway is a consideration primarily
 during the post-closure phase of the disposal operation.
    Intrusion is another  potential  pathway for exposure to LLW
 materials.  Human intrustion can be either intentional or
 inadvertent.  In addition, plant and animal intrusion present
 potential exposure pathways.
    Other pathways include combinations of the previously discussed
modes.  For example,  surface runoff followed by subsequent wind
 transport could be considered.
    An important objective should be the identification of pathways so
trivial  that attention can be more reasonably directed to those which
have a greater likelihood of transporting significant quantities of
hazardous material to the biosphere.
                          MODELING APPROACHES

    Considering the many potential pathways for human exposure, a
review of previous modeling studies indicates that perhaps an
inordinate amount of effort has been devoted to the groundwater
pathways.  This is particularly true in light of previous studies2»3
which indicate that in terms of maximum individual exposures, the
groundwater pathways constitute a relatively minor threat.  Figure 2,
taken from reference 2, shows the results of such a study.
    There are several major approaches to the modeling of
environmental transport.  These include:
    •    Physical (Analytical)
    •    Empirical (Analog)
    t    Statistical (Probabilistic)
    t    Combinations
    Physical models are determined by an analytical or theoretical
description of the process being modeled based on first principles.
Due to the infinite complexity of real world processes, physical
models can never be exact predictive tools.
    Empirical models are based on experimental data without the
necessity for a detailed underlying theoretical basis.  They represent
an acceptance of the inability to thoroughly understand the process
while making the best use of the data that we do have.
    Statistical models use probabilistic inference to model the
process.  In some cases this is because the underlying theory is

-------
           Figure 1
Evnlronmental Pathways for LLW


1
[WATER |
T~~
i
SURFACE
RUNOFF

1 i
IGROUNDWATERI OPERATIOI
RELEASE

| PATHWAYS |
1
|
IAIR | IN-

i i
WL lEROSIONl GAS (INTENTION/
EVOLUTION


1 1
FRUSION | | OTHER
|
]
\L| INADVERTENT!

. 1 	 ! .
                                         HUMAN I IPLANTIIANIMALI
                                                                         CO
                                                                         CO
                                                                         en

-------
 (0


TJ
T3




1
O
CL

8
 X
 CO



 Q)

 *-<


 S

 Qi


 O



'ro


 c
 c
      10-
                     10
    T       10     102     103      104


Pu-239 interface concentration - juCi/cm3
                                                                             10s
                                                                                                  CO

                                                                                                  CO
     FIG.  2.  Annual  individual dose vs HLW/LLW interface concentrations
         239
     for    Pu calculated for six exposure  scenarios.

-------
                                  337
probabilistic in nature, and in other cases it is a reflection of the
uncertainty about input parameters or the detailed underlying
physics.  Models may also use a combination of these approaches.
                            MODEL  LIMITATIONS

    There are several well known limitations to predictive modeling.
These include data limitations (where the complexity of the model may
outrun the available data), computational limitations (where the
complexity of the model outruns the computational capabilities), and
developmental limitations (where time, economic, or other factors
limit the desirability or capability for model development).  In
addition, the apparent search for perfection in predictive modeling is
not only unnecessary, but may even be counterproductive.
    There appears to be a tendency in model development to equate the
level of complexity and detail with the credibility of the results.
Modelers should be cautioned to keep in mind the famous GIGO
admonition (garbage in, garbage out).  Model complexity should never
be used to disguise either lack of understanding of phenomenology, or
deficiency in quality of data input.  No degree of model complexity or
precision can compensate for a lack of either understanding or
information in the analytical process.
                               VALIDATION

    In light of these limitations, It is important to consider the
question of model validiation.  In the absence of experience
(particularly the required observations over prolonged time periods)
the process of validation is tenuous at best.  How can we tell if a
model works or not?  One possible approach is to use natural analogs
for the purpose.4
    An application of this approach may be found in predictive
modeling for the effects of Iodine-129 in radioactive waste.  This
radionuclide has been a source of concern because of its extremely
long  ( -lO? yr) half-life.5  A recent study6 indicates that, as
a result of groundwater leaching of high level waste, a maximum
radiation dose of 3.3 rem/yr to the thyroid could result.  Although we
are unfamiliar with the details of the model producing this result,
the prediction can be checked using the natural analog approach.
    Assuming the high-level waste repository contained the waste from
    GWe-yr of power production, it would have a total inventory of
about  10 gm of 1-129.  Further, assuming the repository was 1000 m
deep  and sited in a watershed of 104 km2, the total natural iodine
inventory in the top 1000 m layer of earth would be about 8 x
1Q12  gm.  if the iodine in the waste leaches at a rate no more or
less  than that in the soil, then, at equilibrium, the average ratio of
I-129/total iodine is 1.3 x 10'6 at the point of groundwater
discharge.  Because of its low specific activity, it can be shown that
if all the iodine in the human thyroid were 1-129, the maximum dose to
that  organ would be ~19 rem/yr.  Therefore an individual receiving his

-------
                                 338
 total  iodine  intake from that watershed could receive a maximum
 thyroid  dose  of only (1.3 x 10~6) x 19 = 2.4 x lO'5 rem/yr, or
 about  five  orders of magnitude less than the predicted 3.3 rem/yr.
 Although such predictive models are based on admittedly conservative
 assumptions,  it might be in order to question the degree of
 conservatism.  Excessive conservatism, sometimes bordering on the
 absurd,  can call the entire modeling procedure into question.
                                CRITERIA

    As indicated previously, the development of models in the absence
of definitive criteria is problematic.   Previously proposed guidelines
have tended to be vague.   Such guidelines are not very useful in
planning the development  of models.
    While radiological protection should be the overriding concern,
the approaches to the required analysis is sometimes confusing.
Emphasis on the maximum individual  dose results in "worst case
analysis".  Two deficiencies of the  "worst case" approach are: (1) it
is highly dependent on the imagination  of the modeler in selecting his
assumptions, and (2) to the non-professional, the result may appear to
reflect reality instead of an extreme and unlikely situation.  An
approach toward eliminating this problem would be to incorporate the
concept of probability into the modeling process.  This approach
requires the determination of probabilistic criteria.7
    In any case, definitive standards are vital.  If you don't know
what you are looking for, the chances are that you won't find it.  In
addtion,  predictive modeling capabilities should not dictate
criteria - it should be the other way around.
                            SUMMARY COMMENTS

    Definitive criteria are essential  in the planning of waste
management activities.   In particular,  the strategy for model
development in support  of the LLW program would logically depend on
the prior establishment of criteria in  order to ensure reasonableness
and consistency.
    Model development should take into  account the question "how good
is good enough?"  in addition to pathways (what is to be modeled) and
approach (how to  do it).   Recognizing the limitations of the modeling
process, it should also be noted that models are tools to assist in
the decision making process.  They should not be considered to be an
end in themselves.

-------
                                       339

                                    REFERENCES
1.   Mosier,  J.  E., et  al.,  "Low  Level  Waste  Management:   A Compilation
     of Models  and Monitoring Techniques,"  ORNL/SUB-79/13617/2,  April
     1980.

2.   Cohen,  J.  J.  and W. C.  King,  "Determination of  a Radioactive Waste
     Classification System," UCRL-52535.  March 1978.
3.   Adam,  J. A.,  and V. L.  Rogers,  "A  Classification System for
     Radioactive Waste  Disposal -  What  Waste  Goes Where?"  NUREG-0456,
     June  1978.

4.   Cohen,  B.  L.  "A Super Analog  Computer for Evaluating  the Safety of
     Buried  Radioactive Waste," Waste Managetnent-81,  Tucson, 1981.
5.   Kocher,  D.  C., "A  Dynamic Model of the Global  Iodine  Cycle for the
     Estimation  of Dose to the World Population from Releases of
     Iodine-129  to the  Environment," ORNL/NUREG-59,  1979.

6.   Croff,  A.  G., J. 0. Blomeke  and B.  C. Finney,  "Actinide
     Partitioning - Transmutation  Program Final Report," ORNL-5566,
     June  1980.
7.   Cohen,  J.  J., "Suggested Nuclear Waste Management Radiological
     Performance Objectives", NUREG/CR-0579.  UCRL-52626. December 1978.
                                   DISCLAIMER

               This document was prepared as an account of work sponsored by an agency of
               the United States Government. Neither the United States Government nor the
               University of California nor any of their employees, makes any warranty, ex-
               press or implied, or assumes any legal liability or responsibility for the ac-
               curacy, completeness, or usefulness of any information, apparatus, product, or
               process disclosed, or represents that its use would not infringe privately owned
               rights. Reference herein to any specific commercial products, process, or service
               by trade name, trademark, manufacturer, or otherwise, does not necessarily
               constitute or imply its endorsement, recommendation, or favoring by the United
               States Government or the University of California. The views and opinions of
               authors expressed herein do not necessarily state or reflect those of the United
               States Government thereof, and shali not be used for advertising or product en-
               dorsement purposes.

-------
Workshop Summaries

-------
                SUMMARY OF RELEASE MECHANISMS WORKSHOP
                         L. R. Dole,  Chairman
                       D. E. Fields,* Secretary

                 4,    Chemical Technology Division
                  Health and Safety Research Division
                     Oak Ridge National Laboratory
                      Oak Ridge, Tennessee  37830
T
     The purpose of this workshop was to identify the parameters that
control the release rates of radionuclides from low-level  waste (LLW)
shallow land burial sites in which radioactive waste has been and will
be placed.  This panel (for participants, see Appendix A)  searched for
deterministic relationships describing source terms for the low-level
waste trenches whose components are waste forms, containers, backfills,
and other "engineered" bariers.

     The need for this release mechanisms workshop was established in
a previous group of workshops [1], during the special Interagency Meeting
on Low-Level Waste Environmental Modeling in Bethesda, Maryland,
July 30-31, 1980.  All three July 1980 workshops, (1) Model Status and
Development, (2) Validation and Verification, and (3) Criteria for
Selecting and Applying, declared the inadequacy of the present descrip-
tion of the low-level waste disposal site source terms, and the lack of
models describing the leach mechanisms and the chemical forms of poten-
tial contaminant releases, resulting from shallow land burial sites
containing LLW.

     Therefore, the goal  of this panel was to discuss the past experience
and ongoing research concerning the chemical, biological, and mechanical
processes summarized in Table 1.

     Due to the makeup of the panel, discussions focused most sharply on
Mechanical Processes.  Based on several field studies at existing burial
sites and operating experiences, the discussions summarized below
(1) described the three common site geologies, (2) identified subsidence-
driven scenarios, (3) identified site surveillance requirements,
(4) determined the present "state-of-the-art" of in situ source term
characterization, and (5) outlined the course of action to effectively
improve our quantitative understanding of release from LLW burial
trenches.

     The panel  was skeptical  that one "standard site model" could deal
with the range of geological  and climatological  settings in which LLW
                                 343

-------
                         344
 Table 1.   Processes Related  to  Radionuclide Releases
       from Low-Level  Shallow Land Burial  Sites
1.  Chemical  Processes
    a.  Leaching  of waste forms
    b.  Canister  corrosion and degradation
    c.  Nuclide speciation, complexation, etc.
    d.  Nuclide retardation in backfill systems
    e.  Others
2.  Biological Processes
    a.  Barrier degradation
    b.  Mobilization of fission  products and actinides
    c.  Gas generation
    d.  Others
3.  Mechanical Processes
    a.  Compaction and  subsidence
    b.  Failure of barriers
    c.  Water penetration of capillary barriers
    d.  Others

-------
                                 345


burial sites are placed.  A minimum of three typical settings were con-
sidered.  First most sites west of longitude 100°W are located in porous
host rocks and in semiarid to arid climates.  Since these sites gener-
ally lose more moisture through transpiration than is received from
precipitation, there is little or no hydraulic gradient under normal
conditions.

     Conversely, the two classes of eastern sites receive up to forty
inches of annual precipitation.  The permeable eastern sites are located
in a range of hosts from sandy soil to fractured shales.  On the other
extreme, the third site classification includes eastern sites located in
clay sequences in which essentially no permeability can be measured.

     Burial site operators from all of the above classes of burial sites
reported subsidence in their trenches.  The impact of this subsidence on
the trench cover has ranged from minor fissures to gaps that a man could
fall into.  In isolated cases, the trench contents were visible.  These
cases have ranged from potential to actual short circuits of the pathway
to the biosphere.

     The panel considered that the probabilities and consequences of sub-
sidence features generally represented a more significant risk than
ground water scenarios.  Subsidence short-circuit scenarios include:
(1) funnel ing run off through the trench, accelerating ground water
contamination, (2) flooding and overflowing of the trenches, delivering
activity to the ground surface, and (3) exposing the waste to weathering.

     Operating practices have been developed to mitigate some of the
causes of subsidence listed in Table 2.  While vibratory compaction tech-
niques have proven effective in filling voids and providing dense back-
fills, there are problem wastes and some subsidence is inevitable.
          Table 2.  Causes of Waste Burial Trench Subsidence
        1.  Compaction of the waste forms

        2.  Failure of backfill to fill voids between
            containers

        3.  Compaction of loose backfill material

        4.  Degradation of the waste by corrosion, biological
            decay, and dissolution

-------
                                 346


      Therefore, a surveillance and trench cap repair program is required
 for at least the first ten years after the trenches are closed.  It is
 conceivable that a trench subsidence surveillance program should be
 carried on for upwards of 50 years.

      Panel discussion explored the possibility that trench cover fissuring
 may have more severe consequences in sites in impermeable formations.  At
 first glance, the "mother nature" permeable systems, which rely on long
 travel  times into and through the ground water to the biosphere, appeared
 less vulnerable to the short-circuit delivery to the ground surface from
 an  overflowing trench —the so-called "bath tub effect."  However, the
 permeable sites are still subject to flash flooding and are not exempt
 from this scenario.   Also, disposal in an impermeable host is highly
 desirable since travel times can be measured in geological eras.

      The potential  "bath tub" scenario does not preclude sites in imper-
 meable  hosts.   It does demand that the trench covers be designed care-
 fully,  that waste forms that do not contribute to subsidence are used,
 and  that there is a  rigorous surveillance program.  Nevertheless, imper-
 meable  sites are good choices for the very long-lived nuclides.

      In an attempt to coyer the processes in Table 1, the panel explored
 the results of site  studies, which included ground and trench water
 sampling and trench  exhumation.  In spite of the diversity in waste forms
and sites, there were some consistent observations.

     Shallow burial  trenches generally are deep enough to insure anoxic
reducing conditions.   With the exceptions of rare geochemistries and
aggressive reagents  inside drums, corrosion of normal carbon steel  can
be slow.  There are  cases of intact drums older than 10 years.   Even the
biodegradation  of clothing and cardboard at a South Carolina sites was
small after 10 years.

     The inhomogeneity of the waste is an insurmountable problem.  Most
of the trench  contents are there because they were suspect and may have
less activity  than the dirt that was moved to bury them.  It has been
observed that  90% of the activity in a particular trench was in one small
group of boxes, which represented a negligible contribution to the volume
of the waste buried.

     Regional  ground water sampling, because of long travel times,  is a
very long-range effort.   It is difficult to obtain representative trench
water samples  because of capillary holdup in the waste host, which
results in poor flow distribution through coarse sump materials.  While
a "bath tub" trench  might make leachate more readily accessible, the
 inhomogeneity  of the trench contents makes an overall quantitative
 accounting questionable.

     Nevertheless, attempts have been made to correlate the radionuclide
 inventories from hopeless shipping records and the activity in  trench
waters.  Table 3 presents the guesstimates of yearly fractional release
 (FR).

-------
                                 347
         Table 3.  Estimates of Yearly Fractional Release (FR)
                 from LLW Shallow Land Burial Trenches
              Site                                     FR


         Savannah River                               10~8

         Oak Ridge                                    10"6

         West Valley                                2.5 x
     The uncertainty range in the FR of four orders of magnitude cannot
be reduced in the present situation.  In the case of the existing trenches
as they are found, both the inhomogeneity of the contents and the uncer-
tain nuclide inventories preclude the possibility of refining our quan-
titative understanding of releases.

     There are two solutions to this dilemma.  The first requires two
assumptions.  These are (1) that the nuclide retention in the trenches
is insignificant in respect to the travel times to the biosphere, and
(2) that the chemical buffer capacity of the host is sufficient to over-
come any adverse chemical effect on the nuclide retention from the trench
contents.  These assumptions reduce the LLW trench source term problem
to estimating inventories.

     The second course of action involves a ten-year research program.
This program entails selecting sites and waste categories, segregating
the waste, assaying the nuclide content, and preparing study trenches.
Because the processes controlling the source terms are slow, the in situ
study requires at least ten years.  This field data will require an elab-
orate laboratory back-up study in order to interpret the results in terms
of quantitative deterministic algorithms which are based on specific
release mechanisms.

     In summary, the decision to institute a lengthy, costly, and complex
study balances against the potential impact of a regulatory decision
requiring expensive waste conditioning and, possibly, extensive remedial
treatment of buried waste.

     In the previous workshops, the premise was stated that a model must
first simulate before it can predict.  This workshop indicated that a
data base to successfully test a simulation does not exist in the case
of shallow land burial.  Furthermore, this workshop concluded that such
a data base is not likely to be generated from current burial sites.

-------
                                 348


                             REFERENCES
1.  R. S. Lowrie et al., Summary of the Meeting  on  Low-Level  Waste
    Environmental Modeling, July 30-31, 1980,  Bethesda,  MD,
    ORNL/NFW-80/26.

-------
                                 349
                              APPENDIX A

                           List of Attendees
                      RELEASE MECHANISMS WORKSHOP
Name

E. L. Albenesius
Jim Bennett
Clinton M. Case
Don A. Clark
Richard Codell
Leslie R. Dole
Ken Erickson
David E. Fields
Jesse Freeman

Gerry Grisak

Cheng Y. Hung
Donald Jacobs
Thomas Johnson
Craig A. Little
Steve Phillips

David E. Prudic
Andrew E. Reisenauer
George Saulnier
Jacob Sedlet
Robert W. Thomas
Edwin P. Weeks
Ken Whitaker
Charles R. Wilson
Location

Savannah River Laboratory
U.S. Geological Survey
Desert Research Institute
U.S. Environmental Protection Agency
U.S. Nuclear Regulatory Commission
Oak Ridge National Laboratory
Sandia Laboratories
Oak Ridge National Laboratory
U.S. Nuclear Regulatory Commission,
WM Uranium Recovery
Environmental Canada,
National Hydrology Research Institute
U.S. Environmental Protection Agency
Evaluation Research Corporation
Illinois State Geological Survey
Oak Ridge National Laboratory
Rockwell Hanford,
Research and Engineering
U.S. Geological Survey
Pacific Northwest Laboratory
TRW, Inc.
Argonne National Laboratory
Los Alamos Scientific Laboratory
U.S. Geological Survey
Chem Nuclear Systems, Inc.
Lawrence Berkeley Laboratory

-------
             SUMMARY OF WORKSHOP ON OVERALL SYSTEMS MODELING

                   James 6. Steger, Workshop Chairman
                     Los Alamos National Laboratory
                          Los Alamos, NM  87545

                  Leroy E. Stratton, Workshop Secretary
                      Oak Ridge National Laboratory
                          Oak Ridge, TN  37830


     This workshop was tasked to look into the need for and status of
overall systems modeling for low-level waste.  In attendance were 15
modelers, 6 managers/users, and 5 people who claimed to be neither of
the above.

     The approach consisted of providing the attendees with a written
list of 10 questions that were to be discussed in sequence.  A consensus
position was to be obtained on all of the questions.  The following is a
brief summary of the results.


Ql.  What needs to be modeled?  What submodels are needed?

     This question led to a discussion that lasted in excess of one hour.
     It was concluded that there need to be three different levels of
     models.

       I.  Screening Model.  A need exists for a single overall screening
           model to make a first-order assessment.  It should be able to
           be expanded and improved upon as experience is gained in its
           use.  Probabilistic models should be considered as well as
           deterministic models.

      II.  Site Evaluation.  A more detailed level of models to evaluate
           sites that pass the screening test is necessary.  These models
           should have submodels for source term, airborne pathway, sur-
           face pathway, subsurface pathway, biological pathway, and human
           intrusion.

     III.  Process/Mechanisms.  Researchers trying to understand the many
           processes/mechanisms involved in low-level waste disposal will
           have needs for many models in their work to test theories,
           explain experimental results and communicate with other
           researchers.  These models will be quite specific to the prob-
           lem being investigated.

                When applying models as a predictive tool for evaluation,
           the three kinds of models listed above should be used in a
           hierarchal sequence.  If the simple screening model with con-
           servative parameters shows that there are no problems, then
           the modeling effort should stop.  However, if there is an

                                   351

-------
                                  352


           indication that limits  are  being  approached,  then a more
           detailed analysis should  be performed.

                The group did not  feel  that  the status of modeling was
           adequate and did not agree  with the DOE comment (see Large,
           this volume) that no more models  are needed.   The group also
           cautioned that there seems  to  be  an over-emphasis placed on
           groundwater pathway modeling.

Q2.  Should the coupling of models proceed,  and if so, how?

          Coupling of models requires  that they are capable of interact-
     ing.  In the case of Screening  Models,  they can probably be coupled,
     but this level  of detail  is not necessary.  Simple input/output rela-
     tionships are adequate.  For  Site Evaluation Models, the pathway
     models are not sufficiently developed to permit complete coupling.
     For Process/Mechanism Models, coupling  is problem dependent and
     desirable in some cases.

Q3.  Are different systems models  needed  for arid vs. humid sites?  For
     every site?

          A single model  can probably  be  developed for screening purposes
     for both arid and humid sites.  However, there are enough differences
     between the arid and humid sites  that a single site evaluation model
     does not seem likely.   It is  desirable  though that effort should be
     made to keep the form and scope of these models as similar as possible
     to facilitate comparison of the results.  For process/mechanism model-
     ing, specific models will  be  required.

Q4.  What determines when models for a specific pathway are adequate?

          This question generated  considerable discussion, with the final
     conclusion that the user must determine when a model was adequate.
     A model is a tooT to help managers make decisions.  The adequacy of
     the tool will usually be proportional to the amount of time that the
     modeler and user spend communicating.   Once the user specifies what
     he needs and the modeler understands these needs, it is up to the
     modeler to assure model internal  consistency and the proper balance
     between components.

Q5.  What level of accuracy should we  target for?

          As in the previous question, this  question can only be answered
     by the user.  In any event, the status  of modeling today will only
     permit making relative comparisons.   Obtaining absolute values are
     beyond present capabilities.

Q6.  Can enough data be obtained on  an economical basis?

          For Screening Models, enough data  are already available or
     relatively easy to obtain. The answer  is not so clear for the Site

-------
                                   353


     Evaluation Model.  Data will usually be obtainable on an economical
     basis, but in some cases the data will be very difficult to acquire.
     Process/Mechanism Models are usually designed to achieve complete
     understanding and consequently usually require more data than is
     easily obtainable.  One caution was discussed at this time, and that
     is that care must be used in taking data from other sources to use
     in a model.  Unless it is clearly understood how and why the data
     were obtained, it is very easy to misuse others' data.

Q7.  What is the major purpose of an overall systems model(site evalua-
     tion, evaluation of operations, licensing, design of monitoring
     systems)?

          This question is probably not appropriate for the group's con-
     sideration.  Models contribute to all aspects of the problem, and
     it is not clear how one judges what the major purpose is to be.

Q8.  Can post-operational performance be modeled and predicted using
     preoperational data?

          The group felt that the answer to this question was yes with
     the present level of experience and data, but recommended that moni-
     toring data be used to adjust the model as necessary to keep it as
     accurate as possible.  Man-made modifications can be modeled better
     than making predictions of nature's actions in the long-term
     (300 years).

Q9.  Do the models have the capability to handle different fuel cycles?

          It was concluded that a different fuel cycle would not make
     the problem any more difficult and therefore the answer to this ques-
     tion is yes.  The main problem at this time is obtaining adequate
     information on waste properties and release mechanisms.

Q10. Are there any models that propose to include such processes as trans-
     portation, regional siting, etc., for optimization of sites?  Are
     such models desirable?

          The group did not know of any total systems models (generation
     to dose-to-man), and felt that it would be unrealistic to try to
     include all of these factors in a single model.  For additional
     information, several independent efforts are underway, which con-
     sider parts of the overall system.  EG&G is developing a waste
     generator system model.  Sandia is developing a transportation
     model.  USNRC is preparing an environmental impact statement to
     assess the impact of the requirements of the proposed 10 CFR 61.

-------
                                  354


                                 SUMMARY

     The group was very enthusiastic about the exchange which took
 place.  It is very rare when a working modeler can have a direct conver-
 sation with the policy setters in the DOE, NRC, EPA, and USGS, as well
 as his peers at the same time.  There were many comments concerning how
 valuable the workshop participants felt that the exchange this meeting
 facilitated was.

     There was surprisingly little difference of opinion or views remain-
 ing after discussion and resolution of terminology.  Possibly a glossary
 of terms would be useful.   Communication was a constant theme on almost
 every question.   Everyone agreed that it needs to be improved.  The data
 gatherers, model developers, and manager-users must fully understand
 each other's needs and capabilities.

     The issues  of model  adequacy and accuracy cannot be determined by
 either the modelers or the users alone.   There is a limit to what a
 modeler can do,  and the user must identify the minimum needs.  Close
 interaction is the only sure way to obtain satisfactory results.

     The most serious limitation to overall system modeling is the lack
 of information about waste properties and release mechanisms.  There are
many different waste types and many different disposal environments.
 This situation makes for a very complex series of possible actions and
 reactions,  which are impossible to model with existing data.  Possibly
 the only solution is to pretreat the waste and to tighten the specifica-
 tion on disposal practices.

     The final comment and the unanimous opinion of the group is that
 there is a critical need for standards to be used as targets.  Until one
 knows what needs to be accomplished, it is difficult to determine how
 best to get there or whether or not success has been achieved.

-------
                                   355
                               APPENDIX A
             Overall Systems Modeling Workshop Participants
     Name
         Affiliation
Richard Codell
Paul Dickman
David Grove
Ed Hawkins
Richard Healy
Richard Heystee
Preston Hunter
Tim Jones
Dan Kapsch
Ken Kipp
Eric Lappala
D. H. McKenzie
Jim Mercer
Lew Meyer

Bruce Napier
Yasuo Onishi
Fred Pail let
Dave Pollock
Jack Robertson
John C. Rodgers
Bob Root
Dale Smith
J. C. Sonnichsen
Jim Steger
Jim Stoddard
Leroy Stratton
Burnell Vincent

G. T. Yeh
U.S. Nuclear Regulatory Commission
EG&G Idaho
U.S. Geological Survey
U.S. Nuclear Regulatory Commission
U.S. Geological Survey
Ontario Hydro
Ford, Bacon & Davis
Battelle Pacific Northwest Laboratory
Monsanto Research Corporation
U.S. Geological Survey
U.S. Geological Survey
Battelle Pacific Northwest Laboratory
GeoTrans, Inc.
U.S. Environmental Protection Agency,
Office of Radiation Programs
Battelle Pacific Northwest Laboratory
Battelle Pacific Northwest Laboratory
U.S. Geological Survey
U.S. Geological Survey
U.S. Geological Survey
Los Alamos National Laboratory
Savannah River Laboratory
U.S. Nuclear Regulatory Commission
Westinghouse Hanford
Los Alamos National Laboratory
Science Applications, Inc.
Oak Ridge National Laboratory
U.S. Environmental Protection Agency,
Solid Waste Office
Oak Ridge National Laboratory

-------
                          SUMMARY  OF  WORKSHOP  ON
            ENVIRONMENTAL TRANSPORT  PARAMETERS AND  PROCESSES
                   E. L. Albenesius, Workshop Chairman
                        Savannah River Laboratory
                      Aiken,  South Carolina  29801
                    C. A.  Little,  Workshop Secretary
                      Oak Ridge National Laboratory
                       Oak Ridge, Tennessee  37830
                             ABSTRACT

          This workshop was held on the afternoon of
     December 3, 1980 in Denver, Colorado.  Nearly thirty
     participants discussed aspects of modeling the
     environmental transport of radionuclides from low-level
     radioactive waste shallow land burial grounds to the
     receiving human populations or individuals.  Pathways of
     transport considered included groundwater, surface water,
     the atmosphere, and food chains".  The ability to model
     transport in these pathways was discussed.  Several
     modeling and data needs were brought to light and several
     recommendations were made to regulatory and funding
     agencies.
                               INTRODUCTION

     The Workshop on Environmental Transport Parameters and Processes
was held on the afternoon of December 3, 1980 at the Denver Marina
Hotel, Denver, Colorado.  Approximately thirty individuals participated
in the discussion;  their names and affiliations are listed in
Table 1.  The topics of discussion were limited to those that
considered only transport from the trench boundary to some human
receptor.  Processes occuring within the trench were discussed in the
Workshop on Release Mechanisms.  Questions of accuracy and precision
were largely discussed in the Workshop on Model Verification and
Validation.
                         PURPOSE OF THE WORKSHOP

     The ultimate purpose of this workshop, as well as the complete
meeting, was to foster communication and cooperation between the
various regulatory and funding agencies, model users, and model
builders.  In more specific terms, the purpose of the workshop was to

                                    357

-------
                                  358
         Table 1.   List  of Attendees of Environmental Transport
                          Processes Workshop
      NAME
      AFF RATION
 Ed  L. Albenesius
 Clinton M. Case
 Don Clark
 Ken Erickson
 David E. Fields
 Jesse Freeman
 Richard Heystee
 Cheng Y. Hung
 Craig A. Little
 James Mercer
 Bruce Napier
 Yasuo Onishi
 David Pollock
 Andy Reisenauer
 Bob Root
 George Saulnier
 Jacob Sedlet
 James Stoddard
 R. 6. Thomas
 Edwin P. Weeks
 Charles Wilson
 Ken Whitaker
 G. T. Yeh
Savannah River Laboratory
Desert Research Institute
USEPA-Kerr Environmental Res.  Lab.
Sandia National Laboratory
Oak Ridge National Laboratory
USNRC
Ontario Hydro
USEPA
Oak Ridge National Laboratory
Geo Trans. Inc.
Battelle Pacific Northwest Lab.
Battelle Pacific Northwest Lab.
USGS
Battelle Pacific Northwest Lab.
Savannah River Laboratory
TRW, Energy Systems
Argonne National Laboratory
Science Applications,  Inc.
Los Alamos National Laboratory
USGS
Lawrence Berkeley Laboratory
Chem-Nuclear Systems,  Inc.
Oak Ridge National Laboratory
discuss the role in  modeling  of groundwater,  surface water, and
atmospheric transport  processes and  assess the relative importance of
each of these processes,  the  adequacy of existing models and data
bases, and the needs for  further development.  Table 2 lists the
processes and pathways considered by the workshop participants.  As
mentioned earlier, processes  occuring within  the boundary of the
trench were specifically  excluded from discussion by this workshop,
but were dealt with  in the Release Mechanisms Workshop.
                           SURVEY OF MODELS

     In an attempt to ascertain the state-of-the-art of environmental
transport modeling for low-level waste management, a pre-workshop
survey form was sent to each potential participant several weeks
prior to the Denver meeting.  The survey form is appended to this
summary (Appendix A).  For each pathway or process, respondents were
asked to submit names and characteristics of environmental transport

-------
                                  359
            Table 2.  Pathways From Burial Ground Site to Man
           and Processes That Were Considered by the Workshop
                       Groundwater Transport

                       Surface Water Transport

                       Resuspension

                       Atmospheric Transport

                       Irrigation

                       Food Chain Transfer

                       Direct Exposure
models with which they were familiar or had experience.  This
exercise resulted in only a dozen model replies.  Models were
identified for only three categories:  groundwater, atmospheric
transport, and dose-to-man or "systems" models.  These replies are
tabulated in Appendix B, Tables B.I - B.3.  The groundwater model
capabilities ranged from one-dimensional to multidimensional
simulations of either water flow or transport of radionuclide decay
chains.  The atmospheric codes listed included both chronic release
annual average models (AIRDOS EPA) and accidental release models
(CRAC).  The dose-to-man models all included some representation of
the transport simulated by the groundwater and atmospheric models.
None of the models in Table B.3 are presently documented in public
literature.  This is in contrast to the groundwater and atmospheric
codes listed in Tables B.I and B.2 which for the most part are
documented in the available literature.  This brief exercise suggests
that, for low-level waste, dose-to-man modeling is much less well
developed than either of the other types of modeling.
                          WORKSHOP DISCUSSION

     Several questions were discussed during the course of the
workshop.  These included:

     1.   What is the relative ability to model each of the above
          pathways or processes?
     2.   What needs to be done to improve existing or future models
          of these processes?
     3.   If you were sponsoring the low-level waste modeling program
          for the entire nation, on what research would you spend
          your program budget?

-------
                                   360
 These  questions were posed to generate discussion among the  attendees
 and  for  the  most part were not answered directly or specifically.
 Nevertheless,  answers were discerned for each model type during  the
 ensuing  discussion.

     With  regard to groundwater (GW) models, the point was made  that
 GW flow  and  transport codes were often so detailed and numerically
 complicated  that they have little utility for many regulatory  uses.
 The  SWIFT  code, for example, may cost as much as $200 per run.   Uses
 such as  those  by EPA where numerous sites must be screened would seem
 to preclude  using a code that is expensive to run.  It was agreed
 that,  in terms of code development, the computer time was a  small
 expense  compared to either development manpower or data collection.
 The  participants also agreed that, in general, the quality of
 groundwater models greatly exceeds that of the data used for input
 parameters (e.g.,  kj, storage coefficients).

     Surface water flow and transport models were also discussed.
 Onishi and his coworkers at PNL have recently reviewed 28 surface
 water models and their characteristics (Onishi et al., 1981)   The
 group generally agreed that, with the exception of sediment
 transport, surface water models are collectively adequate or better.
 Sediment transport however, is not well developed.  It was stated
 that models, especially assessment models, usually strive for
 description of average conditions and discount the effects of  low
 probability events.  For example, the James River model developed for
 the Kepone pollution case (Onishi and Wise 1978) has a deficiency in
 the  inadequate treatment of pulsed input such as wash from severe
 flooding.  In the  case of a sediment transported pollutant,  such as
 Kepone, the greater human hazard may be neglected if sediment
 transport resulting from low-probability events is not well modeled.
 The relatively long time frames of low-level waste modeling may  make
 detailed modeling  of either dissolved or sediment-attached pollutants
 unnecessary; dilution over total  surface water flow may be acceptable
 with much less effort.

     Consensus regarding atmospheric transport modeling was  that the
 Gaussian Plume model (Gifford 1968) is the best known and most widely
 used model.  The Gaussian Plume is reasonably accurate to distances
 out  to perhaps 80 km from the source except for situations of  complex
 meteorology.  For modeling potential airborne impacts on humans  from
 low-level waste burial  sites, resuspension will be the most  difficult
 process  to model.   Resuspension likely provides a large portion  of
 the material input to the transported plume.  However, resuspension
 models are at this point conceptually crude and for the most part
 only empirical models (see Healy, this volume).  Data to parameterize
 resuspension models are also likely to be highly site-specific.

     Food chain transfer models were seen to be a weak link  in the
modeling of transport from burial grounds to humans.  This is  the

-------
                                   361
case for two reasons.  First, the models used to "transport"
radionuclides through food chains are generally simplistic
multiplicative chain or equilibrium models.  For certain applications
such as integrated doses to a static population over a long time
period (several generations) these models may be adequate.  However,
for any application resulting from short term releases the models are
crude.  The second difficulty with food chain models is the lack of
quality element-specific data for most of the transfer factors in the
food chain (e.g., forage-to-milk transfer within cattle, gut-to-organ
transfer within humans).

     A consensus among the modelers was that credible scenarios of
exposure need to be developed.  So-called worst-case scenarios may
need to be discarded for the simple reason that they are not always
conservative; virtually any "worst-case" scenario may be made more
damaging by employing less realistic assumptions.  Scenarios for
impact estimation from trench intrusion, site reclamation, on-site
spillage, etc. are needed for modelers to construct reasonable
modeling approaches.  Many of the participants felt that some
important regulatory decisions were being made by modelers by default
when dose calculation scenarios were designed by modelers.

     The environmental transport of radionuclides from a shallow  land
burial ground is very complex.  Most of the workshop attendees agreed
that the best use of modeling of low-level waste burial is as a
screening device for the selection of potential new disposal sites.
In this context, dose or risk estimates from models would be used  in
a relative sense and would not be considered estimates of the
absolute radiological exposure or risk.  Site selection and screening
processes should utilize not only models but surveys of geology,
hydrology, intuition, etc.


                       CONSENSUS OF THE WORKSHOP

     The participants reached consensus on  a number of questions  and
issues.  Five statements of agreement are  summarized  in this section.

     Modeling expertise is generally adequate for the perceived needs
of modeling  low-level waste environmental  transport.  However, some
of the submodels need improvement.   In particular,  submodels of
resuspension, sediment transport by  surface water,  and  intra-trench
processes that  lead  to radionuclides leaching out of  the trench need
conceptual improvement.

     Even though modeling expertise  is adequate, the  predictive
capability of the models perse  is not satisfactory.  The  lack of
faith  in predictability arises from  an awareness that the models  do  a
poor job of  estimating absolute values.  The models  are of great
value  in generating  relative numbers.  One  such model application
would  be ranking the suitability of  several alternative sites.

-------
                                  362
     The participants agreed that to improve the ability to simulate
a given site the greatest need is for a more complete data base.  An
example of needed types of data not readily accessible is given in
Table 3.  Operators interested in predicting the performance of a
chosen new site would be advised to make an effort to acquire the
listed data.  Many of these parameters require site specific
measurements and none of these data are readily available for
presently unidentified sites.   Some data, such as radionuclide
inventory and physical form may be acquired through accurate record
keeping.  Others, such as dispersivity values are not even measurable
quantities in the field, but are usually back-calculated from the
results of a hydrologic flow model.

     The workshop agrees that  of the numerous models listed in
several recent model  compendia (e.g., Hosier et al., 1979), only a
few codes, probably less than  fifty, were adequately documented.
Adequate documentation includes not only internal code comments, but
also a comprehensive  user's manual describing theory, operation, and
a sample problem to check model results.  It may also be a useful
task for some group to specify the use of given models in given
situations.   In some  ways this could be similar to the USNRC
specifying its Regulatory Guide models.  What group would be
responsible  for pursuing such  an examination of models for low-level
waste is unclear.

     A final consensus of the  workshop is that credible scenarios of
human exposure to low-level waste buried in shallow trenches need to
be developed.  Definition of allowed human activities under such
headings as  intrusion, reclamation, etc. are, at present, not
standardized.  This lack of definition makes difficult the
intercomparison of results from several models.  Again, the location
of the responsibility for developing scenarios is not clear.

-------
                                  363
       Table 3.  Necessary, But Not Readily Available, Data for
           Modeling Low-Level Waste Impacts at Hypothetical
                            Burial Grounds
Trench Contents
     Radionuclide inventory by activity
     Chemical form of elements
     Physical form of elements
     Presence of extraneous material (pesticides, herbicides, etc.)
     Distribution coefficients (k^)
     Waste form interaction with biosphere

Geologic Characteristics
     Geochemistry
     Stratigraphy
     Fracture distribution

Hydrologic Characteristics
     Head distribution
     Sorption coefficients (in presence of expected groundwater
     chemistry)
     Conductivity (for both saturated and unsaturated conditions and
     in presence of expected groundwater chemistry).
     Porosity
     Dispersivity values
     Boundary conditions
     Water storage coefficients

Biological interactions (such as burrowing animals)

-------
                                 364


                             REFERENCES
 1.  DeMier, W. V., M. 0.  Cloninger,  H.  C.  Burkholder,  and
    P. J. Liddell.  1979.  GETOUT -  A computer program for
    predicting radionuclide decay chain transport through geologic
    media.  PNL-2970.  Battelle Pacific Northwest Laboratories,
    Richland, Washington.

 2.  Dillon, R. T., R. B.  Lantz, and  S.  B.  Pahwa.   1978.  Risk
    methodology for geologic disposal of radioactive waste:   The
    Sandia Waste Isolation  Flow and  Transport (SWIFT)  Model.
    SAND 78-1267, NUREG/CR-0424.   Sandia Laboratories, Albuquerque,
    New Mexico.

 3.  Gifford, F. A.  1968.  An outline of theories of diffusion in
    the lower layers of the atmosphere   Chapter 3.  In D. H. Slade
    (ed.), Meteorology and  Atomic Energy 1968  TID-24190.  U. S.
    Atomic Energy Commission/Division of Technical Information,
    Washington, D.C.

4.  Moore, R. E., C. F. Baes III, L. M. McDowell-Boyer,
    A. P.  Watson, F. 0. Hoffman,  J.  C.  Pleasant,  and C. W. Miller.
    1979.   AIRDOS-EPA:  A computerized  methodology for estimating
    environmental concentrations  and dose to man  from airborne
    releases of radionuclides.   ORNL-5532.   Oak Ridge National
    Laboratory, Oak Ridge,  Tennessee.

5.  Mosier,  J. E., J. R.  Fowler,  C.  J.  Barton, W. W. Tolbert,
    S. C.  Meyers, J. E. Vancil, H. A. Price, M. J. R.  Vasko,
    E. E.  Rutz, T. X. Wendeln,  and L. D. Rikerstsen.  1980.
    Low-level waste management:  A compilation of models and
    monitoring techniques.   ORNL/SUB-79/13617/2.   Prepared by
    Science Applications, Inc., Oak  Ridge Tennessee for Oak Ridge
    National Laboratory,  Oak Ridge,  Tennessee.

6.  Nuclear Regulatory Commission.   1975.   Reactor safety study:  An
    assessment of accident  risks  in  U.  S.  commercial nuclear power
    plants.   WASH-1400 NUREG-75/014.  U. S.  Nuclear Regulatory
    Commission, Washington, D.C.

7.  Onishi,  Y., R. J. Seme, E. M. Arnold,  C. E.  Cowan, and
    F. L.  Thompson.  1980.   Critical review:  Radionuclide
    transport, sediment transport and water quality management
    modeling; and radionuclide sorption/desorption mechanisms.
    NUREG/CR-1322, PNL-2901.  Pacific Northwest Laboratory,
    Richland, Washington.

8.  Onishi, Y. and S. E.  Wise.   1978.  Mathematical simulation of
    transport of sediment and kepone in the James River Estuary."
    PNL-2731.  Pacific Northwest  Laboratory, Richland.

-------
                                 365
 9.  Sagendorf, J. R. and J.  T.  Gall.   1977.   XOQDOQ program for the
     meteorological evaluation of routing effluent releases at
     nuclear power stations.   NUREG-0324.  U.S.  Nuclear Regulatory
     Commission, Washington,  D.C.

10.  Trescott, P. C., G. F. Pinder, and S. P.  Larson.  1976.
     Finite-difference model  for aquifer simulation in two dimensions
     with results of numerical experiments.  U.  S. Geological  Survey
     Techniques of Water Resources Investigations, Book 7, Chap. 1
     USGS, Reston, Virginia.   116 pp.

-------
                                  366
Model Output

     What is the model output?  For a groundwater flow model  it might
well head distribution; for atmospheric transport, Chi/Q.   Be as specific
as possible.
Data

     What input data are necessary for model/code operation?  Are these
data generic or site-specific?   Are they  presently available,  included in
the model/code documentation?  Are the included data credible?  Again,
be as specific as possible.
Complexity

     Is the model  simple  or complex?   For example,  solute transport models
for groundwater may be  1-,  2-, or  3-dimensional  and may include constant
or changing chemical  conditions.   How about  ease of operation?


Documentation

     Is the model  well  documented? Are  the  assumptions clearly and com-
pletely stated?  If the model has  been coded for computer use, is the
necessary machine-intrinsic information  such as  JCL explained?  Is there
an adequate user's manual?   Is a sample  problem  included?
Reference

     Give a complete  reference to  the model  or code.   If the work is
ongoing, please list  the  site and  principal  investigator or contact
persons.
Thank you for your cooperation.

-------
                                  367


                              APPENDIX A

                        PREWORKSHOP SURVEY FORM
     Due to the shortness of time and the potentially large  number of
participants, it is imperative that this workshop be structured  and
focused.  The ultimate goal of this and the other workshops  is to foster
cooperation and communication between the represented agencies and their
contractors.  A more specific goal is the determination  of the state-of-
the-art of and needs for environmental transport models  useful for low-
level waste management.  As preparation for the workshop, would  you  please
participate in the following exercise before arriving in Denver? The
tables we produce will become an important source of information.

     Please complete one of the following forms for as many  applicable
models as you wish (make additional copies if you need them).  List  those
models which you have used or know of; they may be in development, well-
known, for a single pathway, or "comprehensive."

     You need not include every known model on your lists; we are interested
in specific examples of the existing or developing environmental transport
models so we can gauge where work is needed.

     Please leave copies of your forms at the meeting registration desk
in the hotel.  The data you provide will be compiled before  the  workshop
begins.

     The following guidelines may be helpful:


Pathway or Process

     Choose one of these (or add others):  groundwater transport, surface
water transport, resuspension, atmospheric transport, irrigation, food
chain transport to humans, direct exposure, erosion, intrusion.
Model
     Name a specific model or code of a model.
Model Type

     Describe the type of model you have named.  Possible entries might
include Gaussian Plume model, linear compartment model, equilibrium model,
time-dependent model, finite-element or finite difference solution
technique.  Many other entries are possible.

-------
                                   368
Pathway or Process:
Model:
Model Type:
Model Output:
Data:
Complexity:
 Documentation:
 Reference:

-------
             369
         APPENDIX B



RESULTS OF PREWORKSHOP SURVEY

-------
                                              Table B.  1.  Results of Pre-workshop Survey:  Groundwater Models.
Name
GETOUT
Model lype
One-dimensional
transport;
axial dispersion
Output
C1/yr discharged 1n
time and space
Data Input
Sorption equilibrium
constants; groundwater
velocities; dispersion
coefficient; height and
width of Input release
band
Complexity
Complex in dealing
with numerical
accuracy and decay
chains; simple with
regard to transport
Documentation
Fair; well commented
code; W. V. DeMler et al.
(1979) PNL-2970
HYDRO
SWIFT
TRESCOTT
One-dimensional flow;
finite element
solution
Multi-dimensional
transport; finite
difference solution.
Groundwater flow;
3-dimenslonal finite
difference solution
Radionucllde concentration
in soil 1n time and space;
amount released to aquifer
in time;
Pressure, mass concent.
or temperature in space
and time;
Hyraullc head and well
draw down in time;
Site specific kd;
sorption coefficients;
multi-layer geological
structure; water
conductivities

Permeability, porosity
etc., boundary and
initial conditions
Initial head; storage
coefficient;
hydraulic conductivity;
transm1ssiv1ty;
recharge; grid
spacing.
Crank-Nicholson Finite
Element solution;
constant chemical
conditions; constant
solubilities

1-3 dimensions;
radioactive decay
chains;  fairly
difficult to use

Constant input;
changing input parms
for transient runs
requires sequence of
runs.
Mo formal document
available;
Bahrain Amirijafari
SAI
Well documented;
Dillon, Lantz &
Pahwa (1978)
Well documented
Including theory,
input description;
Trescott, Pinder &
Larson (1976)
                                                                                                                                                                   CO
                                                                                                                                                                   -J
                                                                                                                                                                   o

-------
Table B.2  Results of Pre-workshop Survey:  Atmospheric Transport Models
Name
AIRDOS - II
AIRDOS - EPA
CRAC
XOQDOQ
Model Type
Gaussian Plume;
analytical;
chronic releases
Analytical;
reactor accident
releases
Gaussian Plume;
analytical
Output
Annual population and
maximum individual
dose
Population doses, latent
and chronic; population
fatalities and injuries;
accident costs
X/Q or D/Q for either
annual average or
Data Input
Extensive; includes
population dist.,
annual average
meteoro logy; agricul.
data
Population and agricul.
data; reactor releases
inventory and release
fraction; health effects
parameters
Site specific weather
data, stack height
Complexity
Fairly simple to
operate
Fairly large, but
straight forward
Fairly simple
model
Documentation
Good; sample problem
included; user's manual;
Moore et al. (1979)
Well -commented;
documented in WASH-1400
Appendix VI.
NRC (1975)
Rough draft document;
Sagendorf and Gall
                                                                                                                        CO
     probabilistic
(1977)

-------
                                                Table B.3.  Results of Pre-workshop Survey: Dose-to-Man Models
Name
BURYIT
DITTY1
DOSTOMAN
Model Type
Analytical;
finite-element
Multiplicative
chain, equilibrium
model
Finite difference
and linear
compartment
Output
Population doses
Integrated population
dose
Compartment
Inventories vs
time
Data Input
Site-specific geology,
meteorology, pop.
distribution
Scenarios; releases;
population dlst.
Intercompartmental
transfer coefficients
Complexity
One-dimensional
fairly simple
Many pathways;
easy to use
Easy to run
Documentation
Early 1981; Science
Applications, Inc.;
David Lester
Early 1981; Battelle-
Pac1f1c Northwest Lab.;
Bruce Napier
In-house documentation
only; Savannah River
Lab.; Bob Root


10
>j
ro

MAX I
PRESTO
Multiplicative
chain; equilibrium
model
Analytical and
multiplicative
chain combination
Organ doses
Integrated pop.
dose and Individual
dose; population
health effects
Pathways; food
consumption rates,etc.
Site specific geology,
hydrology, meteorology,
source Inventory, pop.
distribution; generic
agricultural parameters
Easy after first run
Easy to run; I.e. One
scenario per run
Preliminary user's
mannual; Battelle
Pacific Northwest Lab.;
Bruce Napier

1981; Oak Ridge
National Laboratory;
Craig Little

-------
            SUMMARY  OF MODEL VERIFICATION/VALIDATION WORKSHOP

                    T. L. Jones, Workshop Chairman

                     Pacific Northwest Laboratory
                      Richland, Washington 99352

                                  and

                    D. G. Jacobs, Workshop Secretary

                    Evaluation Research Corporation
                      Oak Ridge, Tennessee 37830


                             INTRODUCTION

     The goal of the Model Validation/Verification Workshop was to deter-
mine how the various agencies could best cooperate in establishing the
credibility of models used in support of low-level waste management.

     This workshop  seemed particularly susceptible to problems of semantics.
Definitions for such widely used terms such as verification, validation
and calibration are not available.  The group had the option of establish-
ing a set of definitions or of finding a way of avoiding them altogether.
The latter approach seemed to be the only practical option for such a
short workshop.  The discussion shifted from rigorous definitions to
discussions of what each of us felt constituted "good modeling."  The dis-
cussion focused on  detailed accounts of actual modeling projects conducted
by various participants.  There was remarkable agreement among the group
on methods and techniques used.  It was soon realized that although each
modeler may have a  different name for each step in the modeling process,
there was an underlying methodology common to all efforts discussed.  The
common concern of all modelers seemed to be to have their models and model
results accepted.  The basic ingredients needed to give a model credibility
were identified as:  (1) documentation, (2) reasonable estimation of
important parameters, (3) mathematical checks, and (4) applications to
independent data sets.

     Participants (see Appendix A) felt that proper documentation of models
was not currently available.  Descriptions of models should include the
usual information such as the equations solved together with the basic
computer techniques used; however, more information is necessary.  Explana-
tions of the basic physical or empirical assumptions used to formulate
the theory, examples of intended applications, and possibly subjective
estimates of the overall value of the model are very valuable to model
users.  Documentation should include minimum computer requirements necessary
to run the code, estimates of run time, and a detailed list of input data
requirements.  The discussion on making reasonable estimates of parameters
brought up the subject of interaction with experimental and field personnel.
The need for such interaction is so obvious it is difficult to discuss.
There were no new ideas on how to foster more communication, but it was
agreed that no modeling effort is complete without strong input from people
with the proper physical insight and experimental data,

                                  373

-------
                                  374
     The last two items of the methodology list were handled together and
 resulted in the major recommendation of the workshop.   It was recommended
 that the mathematical checks (number 3 above)  and applications to
 independent'data sets (number 4)  should be accomplished by using standard
 test cases to cross-check most, if not all, waste management models.  A
 good example cited for this type of activity was a recent Gordon confer-
 ence sponsored by the petroleum industry where several  models were applied
 to the same test cases so that model  results could be compared.   Members
 of the workshop who had attended  this conference fel.t the technique was
 successful and could be applied to models used for low-level waste manage-
 ment.  It was felt that test cases could be developed within major disci-
 plines such as saturated and unsaturated flow, leach models, atmospheric
 transport, etc.  A conference similar to the present interagency meeting
 could provide a forum for presenting the results of applying many models
 to the same test data.   The expertise of most  workshop participants was in
 the area of saturated moisture and solute transport.  A detailed list of
 specifications that test cases should meet was developed and is  presented
 below.
                 TEST CASES FOR MODEL  INTERCOMPARISON

     The description of a  test  case was  divided into four areas:  (1) data
sources; (2)  data specifications;  (3)  simulation specifications; and
(4) points of comparison.


                             Data Sources

     A primary source of data for  making mathematical checks are the
analytical solutions to the transport  equations.  While these data sets do
not usually include complicated boundary conditions, they do provide
valuable data on the ability of the computer code to solve the fundamental
equations of interest.   For more realistic  data sets, it was felt that there
was nothing wrong with hypothetical test data.   If people experienced in
the area of transport were given a chance,  very realistic test cases could
be produced which would closely resemble real  world applications.  This
method seemed very useful  for making relative comparisons between models
while not being too useful for  any absolute evaluation of a model's
accuracy.  The primary source of data  for test cases is laboratory and field
experiments.  These data not only  provide the most useful information for
making absolute evaluations of  models, but  the constant use of field
information in model development will  also  help prevent the development of
models whose data requirements  are unrealistic in terms of what is possible
to do in the field.
                          Data Specifications

     For the area of saturated moisture transport it was decided that the
test cases should be three-dimensional  (3-D)  wherever possible.  There

-------
                                   375
are sufficient 3-D models available to warrant this, and also 3-D data
could be used in 1- and 2-D models by extracting the appropriate-subset
of data.  The general consensus was that there is no need to worry about
the non-isothermal cases for low-level waste.  The test cases could be
isothermal experiments.  There were two fundamental data types which
should be included.  In hydrology models,'it makes a difference if the
data is cross sectional in nature or planometric.  Different models are
required for each type of data and test cases involving each type should
be developed.


                       Simulation Specifications

     In discussing test cases, it was most difficult for the participants
to agree on the details of what scenarios and the specifics of what "pro-
cesses to include.

     It was recommended that the specific radionuclides tritium,
strontium-90, carbon-14, and uranium-238 should be included.  It was felt
that this list included sufficiently broad chemical and transport
characteristics to be appropriate.  An important feature of the simula-
tion specifications is the form of output required.  At a minimum, the
output time should be fixed.  If models are going to be compared to one
another, they must produce equivalent output for the same time period.
One cannot compare heads with flux or water level at two years with water
level at two years and six months.  The time scale of the test cases was
discussed briefly and 300 years seemed reasonable.  Obviously, no experi-
mental data will be available for that time period, but test cases involving
experimental data should be extended beyond the range of available data.


                         Points of Comparison

     It is necessary to define exactly how the model outputs are going to
be compared.  This is not a trivial matter and can be the key to the
success of the effort.  One basic prediction of flow models is the direction
of flow.  Models that don't agree on anything else should still predict
the same direction of flow.  Mass balance is also a fundamental parameter
of interest.  While preserving mass balance is not a sufficient condition
to accept a model result, -it is necessary.  If dispersion is being pre-
dicted, there are at least three measures to'examine.  The breakthrough time
for the'solute, the peak concentration over some, time period, and an
estimate of the overall fit to the concentration-time graph or concentration-
space graph.  A test procedure such as least-squares analysis may be
appropriate.  It should be clear from the outset what basis of comparisons
will be.  Other areas of comparison involve the computer characteristics
such as language, CPU time, memory allocation, and the size of the time
of a space step used in the simulation.  There was some discussion about
making a standard time and space step part of the simulation specification
but no agreement was reached.

-------
                                 376
     The important concept was  that test cases could and  should  be  developed
to aid in the evaluation of models used in waste management,  and it was
recommended that there  should be  interagency cooperation  in this effort.
The discussion presented here is  just an example of what  the  test cases
for saturated water and solute  flow might contain.  A serious attempt
at designing test cases would involve much more time and  experts from
many fields.

-------
                                  377
                              APPENDIX A

                           LIST OF ATTENDEES
                   VERIFICATION/VALIDATION WORKSHOP
        Name
             Affiliation
John A. Adam
Bahram Amirijafari
Thomas S. Baer
Jim Bennett
Richard Codell
Jerry Cohen
Paul Dickman
Gerry Grisak
David Grove
Joanna Hamilton
Ed Hawkins
Preston Hunter
Don Jacobs
Thomas Johnson
Tim Jones
Dan Kapsch
Kenneth Kipp
Eric Lappala
Linda Lehman
Dan McKenzie
G. Lewis Meyer
E. S. Murphy
Fred Paillet
Steve Phillips
John Pickens
Matthew Pope
David Prudic
Jack Robertson
John C. Rodgers
William Sayre
Dale Smith
Jack Sonnichsen
James Steger
Ford, Bacon and Davis,  Utah
Science Applications,  Inc.
Nuclear Experience Co., Inc.
U.S.G.S.
USNRC
Lawrence Livermore National Lab
EG&G - Idaho
Environment Canada
U.S.G.S.
Dames and Moore
USNRC
Ford Bacon and Davis,  Utah
Evaluation Research Corp.
Illinois State Geological Survey
Battelle-Pacific Northwest Laboratory
Monsanto Research Corp.
U.S.G.S.
U.S.G.S.
USNRC
Battelle-Pacific Northwest Laboratory
USEPA
Battelle-Pacific Northwest Laboratory
U.S.G.S.
Rockwell Hanford
Environment Canada
EG&G Idaho
U.S.G.S.
U.S.G.S.
LASL
U.S.G.S.
USNRC
Westinghouse Hanford
LASL

-------
               MODELING AND LOW-LEVEL WASTE MANAGEMENT:
                        AN INTERAGENCY WORKSHOP
                           Denver, Colorado
                           LIST OF ATTENDEES
John A. Adam
Senior Engineer
Ford, Bacon & Davis, Utah
2110 Craig Drive
Colorado Springs, CO  80908
(303) 488-2499

E. L. Albenesius
Research Manager
E. I. DuPont de Nemours
Savannah River Laboratory
Aiken, SC  29808
FTS: 239-2482
(803) 450-6211, Ext. 2482

Bahrain Amirijafari
Director of Petroleum Engineering
Science Applications, Inc.
1726 Cole Blvd.
Suite 350
Golden, CO  80401
(303) 279-0701

Thomas S. Baer
Vice President
Nuclear Engineering Co., Inc.
P.O. Box 7246
Louisville, KY  40222
(502) 426-7160

Jim Bennett
Hydro!ogist
U.S. Geological Survey
MS 430
USGS National Headquarters
Reston, VA  22092
(FTS) 928-6947
(703) 860-6947
Clinton M. Case
Research Professor
Desert Research Institute
Water Resources Center
P.O. Box 60220
Reno, Nevada  89506
(702) 673-7375

Hans Claassen
Research Chemist
USGS
Denver Federal Center
Denver, CO  80225
(FTS) 234-2115
(303) 234-2115

Don A. Clark
Research Chemist
EPA
RSKERL
P.O. Box 1198
Ada, OK  74820
(FTS) 743-2300
(405) 332-8800

Richard Codell
Senior Hydraulic Engineer
U.S. NRC
Washington, DC  20555
(FTS) 492-8473
(301) 492-8473

Jerry Cohen
Lawrence Livermore National
L-262
P.O. Box 808
Livermore, CA  94550
(FTS) 532-6449
(415) 422-6449
Lab.
                                  379

-------
                                 380
Paul Dickman
Senior Engineer
Low-Level Waste Program
EG&G - Idaho
P.O. Box 1625
Idaho Falls, ID  83401
(FTS) 583-0610
(208) 526-0610

Leslie R. Dole
Physical Chemist
Oak Ridge National Laboratory
P.O. Box X/Y-12
Building 9204-3
Oak Ridge, TN  37830
(FTS) 626-7421
(615) 576-7421

Ken Erickson
Sandia Laboratories
P.O. Box 5800
Albuquerque, NM  87185
(FTS) 844-4133
(505) 844-4133

David E.  Fields
Oak Ridge National Laboratory
Health and Safety Research Division
P.O. Box X
Building 7509
Oak Ridge, TN  37830
(FTS) 624-5435
(615) 574-5435

Jesse Freeman
Radio!ogic Transport Analyst
USNRC, WM Uranium Recovery
MS SS-483
Bethesda, MD  20555
(FTS) 427-4543
(301) 427-4543

Gerry Grisak
Hydrogeologist
Environment Canada
National Hydrology Research Institute
562 Booth Street
Ottawa, Ontario, Canada
K1A OE7
(613) 733-6476
David B. Grove
Chemical Engineer
USGS. MS 413
Denver Federal Center
Denver, CO  80225
(FTS) 234-2404
(303) 234-2404

Joanna Hamilton
Engineer
Dames & Moore
1626 Cole Blvd.
Golden, CO  80401
(303) 232-6262

Ed Hawkins
Section Chief
U.S. NRC
Washington, DC  20555
(FTS) 427-4533

Richard W. Healy
Hydrologist
U.S. Geological Survey
P.O. Box 1026
Champaign, IL  61820
(FTS) 958-5365
(217) 398-5365

Richard Heystee
Geotechnical Engineer
Ontario Hydro
700 University Avenue
HI 5 D26
Toronto, Ontario
Canada, M5G 1X6
(416) 469-3786

Cheng Y. Hung
Hydrologist
U.S. EPA
401 M Street SW
ANR 460
Washington, DC  20460
(FTS) 557-8978

-------
                                  381
Preston H. Hunter
Program Manager
Ford Bacon & Davis, Utah
P.O. Box 8009
375 Chipeta Way
Salt Lake City, Utah  84108
(801) 583-3773, Ext 287

Donald Jacobs
Manager, Oak Ridge Operations
Evaluation Research Corporation
800 Oak Ridge Turnpike
Oak Ridge, TN  37830
(615) 482-7973

Thomas Johnson
Geologist
Illinois State Geological Survey
514 E. Peabody
Natural Resources Bldg.
Champaign, IL 61820
(217) 344-1481

Tim Jones
Research Scientist
Pacific Northwest Laboratory
Water and Land Resources Dept.
P.O. Box 999
Rich!and, WA  99352
(FTS) 375-7511
(509) 375-2611

Dan Kapsch
Senior Research Chemist
Monsanto Research Corp.
Mound Facility
Miamisburg, OH  45342
(FTS) 774-4207
(513) 865-4207

Kenneth Kipp
Hydro!ogist
USGS
MS 413
Denver Federal Center
Box 25046
Denver, CO  80225
(FTS) 234-2404
(303) 234-2404
Dewey E. Large
Manager, ORO Low-Level Waste
  Management Program
P.O. Box E
Oak Ridge, TN  37830
(FTS) 626-0715
(615) 576-0715

Eric G. Lappala
Hydrologist
USGS
MS 413, Bldg. 53
Denver Federal Center
Lakewood, CO  80228
(FTS) 234-2404
(303) 234-2404

Linda Lehman
Hydrogeologist
USNRC
Division of Waste Management
MS 905 SS
Washington, DC  20555
(FTS) 427-4177
(301) 427-4177

Craig Little
Research Associate
Oak Ridge National Laboratory
Health & Safety Research Division
P.O. Box X
Oak Ridge, TN  37830
(FTS) 626-2106
(615) 576-2106

Robert S. Lowrie
Program Manager
National Low-Level Waste Management
  Program
Oak Ridge National Laboratory
P.O. Box X
Oak Ridge, TN  37830
(FTS) 624-7259
(615) 574-7259

-------
                                  382
Dan McKenzie
Battelle Northwest
Robert Young Bldg.
Rich!and, WA  99352
(FTS) 444-6573
(509) 376-6573

James W. Mercer
Hydrologist
GeoTrans, Inc.
P.O. Box 2550
Reston, VA  22090
(703) 435-4400

G. Lewis Meyer
LLW Standard Program
USEPA
Criteria & Standards Division
Office of Radiation Programs
Washington,  DC  20460
(FTS) 557-8977
(702) 557-8977

E. S. Murphy
Senior Research Scientist
Battelle Northwest
P.O. Box 999
Rich!and, WA  99352
(FTS) 555-4321
(509) 376-4321

Bruce Napier
Environmental  Scientist
Battelle, Pacific Northwest
P.O. Box 999
Rich!and, WA  99352
(FTS) 444-4217
(509) 376-4217

Yasuo Onishi
Staff Engineer
Pacific Northwest Laboratories
P.O. Box 999
Rich!and, WA  99352
(509) 375-2425
Fred Pail let
Geophysicist
Borehole Geophysics, USGS
MS 403
Denver Federal Center
Lakewood, CO  80225
(FTS) 234-2617

Steve Phillips
Senior Engineer
Rockwell Hanford
Research & Engineering
Richland, WA  99352
(FTS) 373-3468
(509) 373-3468

John Pickens
Hydrogeologist
Environment Canada
National Hydrology Research  Inst.
Inland Waters Directorate
562 Booth Street
Ottawa, Ontario
Canada K1A OE7
(613) 995-4045

David W. Pollock
Hydrologist
USGS
MS 431
National Center
Reston, VA  22090
(FTS) 928-6892
(703) 860-6892

Matthew Pope
Scientist
EG&G
P.O. Box 1625
Idaho Falls,  ID  83401
(208) 526-0688

David E. Prudic
Hydrologist
USGS
Rm. 225, Federal Bldg,
705 N. Plaza  St.
Carson City,  NV  89701
(702) 882-1388

-------
                                  383
Andrew E. Reisenauer
Research Scientist
Pacific Northwest Laboratory
Richland, WA  99352
(FTS) 444-7511, Ext. 375-2513
(509) 375-2513

John (Jack) Robertson
Hydrologist
USGS
National Center, MS 410
Reston, VA  22092
(FTS) 928-6976
(203) 860-6976

John C. Rodgers
Los Alamos Scientific Laboratory
P.O. Box 1663, MS 495
Los Alamos, NM  87545
(FTS) 843-3167
(505) 662-3167

Robert Root
Geologist
du Pont - Savanna River Laboratory
Bldg. 773-A
Aiken, SC  29801
(FTS) 239-2612
(803) 725-2612

George Saulnier
TRW, Inc.
405 Urban S C
Suite 212
Lakewood, CO  80206
(303) 986-5533

William W. Sayre
Research Hydrologist
USGS, WRD
Box 25046, MS 413, DFC
Lakewood, CO  80225
(FTS) 234-2404
(303) 234-2404
Jacob Sedlet
Section Head
Argonne National Laboratory
9700 S. Cass Ave.
Argonne, IL  60439
(FTS) 972-5644
(312) 972-5646

Dale Smith
Chief, LLW Licensing Branch
USNRC
Washington, DC  20555
(FTS) 427-4433
(301) 427-4433

Jack Sonnichsen
Principal Engineer
Westinghouse Hanford
Federal Bldg. 420
P.O. Box 1970
Richand, WA  99352
(FTS) 444-6802
(509) 376-6802

James G. Steger
Alternate Group Leader
Los Alamos Scientific Laboratory
LS-6, MS-495
P.O. Box 1663
Los Alamos, NM  87545
(FTS) 843-3331
(505) 667-3331

James A. Stoddard
Science Applications, Inc.
P.O. Box 2351
1200 Prospect St.
La Jolla, CA  92038
(714) 454-3811, Ext. 2294

Leroy E. Stratton
National Low-Level Waste Manage-
  ment Program
Oak Ridge National Laboratory
P.O. Box X
Bldg. 1505
Oak Ridge, TN   37830
(FTS) 626-0504
(615) 576-0504

-------
                                       384
    Robert G. Thomas
    Los Alamos Scientific Laboratory
    P.O. Box 1663, MS-400
    Los Alamos, NM  87545
    (FTS) 843-3318
    (505) 667-3318

    Burnell W. Vincent
    Environmental Engineer
    EPA
    Office of Solid Waste
    Washington, DC  20460
    (FTS) 755-9116
    (202) 755-9116

    Edwin P. Weeks
    Hydrologist
    USGS, MS 413
    Denver Federal Center
    Denver, CO  80225
    (FTS) 234-2404
    (303) 234-2404

    Ken Whitaker
    Project Engineer
    Chem Nuclear Systems, Inc.
    P.O. Box 726
    Barnwell, SC  29812
    (803) 259-1781

    Charles R. Wilson
    Staff Scientist
    Lawrence Berkeley Laboratory
    c/o University of California
    Berkeley, CA  94720
    (FTS) 451-6433
    (415) 486-6433

    George Yeh
    Research Staff
    Oak Ridge National Laboratory
    P.O. Box X
    Oak Ridge, TN  37830
    (FTS) 624-7285
    (615) 574-7285
*U.S. GOVERNMENT PRINTING OFFICE: 1981-740-062/102

-------