UNITED STATES ENVIRONMENTAL PROTECTION AGENCY

                      '        WASHINGTON. D.C. 20460


                                                   EPA-SAB'EEC-90-009
                                                                         OFFICE OF
       27, 1990                                                      THE ADMINISTRATOR
Honorable William K. ReUly
Administrator
U.S. Environmental Protection Agency
401 M Street, S.W.
Washington, D.C. 20460

Dear Mr. Reiily:

     The Science Advisory Boaid has completed its review of the  Office of Solid Waste's
(OSW) Combined Analytical-Numerical Saturated Zone (CANSAZ) flow and transport model
for use in the EPA Composite Model for Surface Impoundments (HPACMS), and is pleased to
submit its final report.  This report resulted from a Saturated Zone Model Subcommittee review
on May 30-31.

     The charge  to the Subcommittee identified two major areas for SAB review. The first
relates to the  appropriateness  of the assumptions  underlying CANSAZ when  used in a
nationwide, Monte Carlo assessment, including which parameters should be estimated on a site-
specific versus a nationwide basis. The second relates to the adequacy of the code for simulating
conditions beneath surface impoundments.

     The Subcommittee found  that the proposed  CANSAZ  model represents a significant
advance over the current  Vertical Horizontal Spread (VHS)  model, but that a  number  of
difficulties remain in the model components, inputs and intended mode of application.  The
following describes the major issues considered by the Subcommittee and the principal findings.

     1) Proper mathematical formulation:  - The numerical and solution  methods are properly
formulated and solved. The technique employed is innovative and computationally efficient, and
the numerical accuracy of  the method is supported by rigorous comparisons with analytical
solutions.

     2) Adequacy of physical  assumptions: - The CANSAZ  model includes  the  basic
assumptions and processes incorporated in current management-oriented ground water models,
but these models neglect  several  important processes now  known to impact contaminant
transport and fate in actual fiejd sites. A number of assumptions restrict the applicability of the
model, including:
                                                             *
     a) Dimensionality Of the model (failure to include transverse flow from mounding),
     b) Limitation to steady-state prediction,
     c) Assumption of homogeneous aquifer media,
     d) The inclusion of limited hydrodynamic, chemical and biological processes.

-------
 Particular processes which are not incorporated, but could be important at particular sites, are
 discussed in our report.

     3) Adequacy of Monte Carlo Approach: - The mechanics of the Monte Carlo procedure
 appear to be properly designed and implemented.  However, determining the adequacy of the
 input distributions is more problematic.  These  input data are critical to  accurate model
 assessment.  The current input data-set is not adequately documented or supported by field data.
 It is  thus recommended that  a  panel  of hydrogeologists, soil  scientists, and engineers  be
 convened to review the proposed model input values and documentation,

     4) Adequacy of field testing of the model: - The EPA has made a good start at validating
 the EPACMS model by confirming the numerical accuracy and presenting a preliminary field
 validation study.  The field validation study  should  be documented, and extended in more
 detailed studies at other sites, A special effort is needed to validate the EPACMS model if it is
 used  in a nationwide assessment.  An extensive nationwide data collection  and monitoring
 program will be needed to accomplish  this.

     5)  Overall modeling  approach:  -  The  Subcommittee strongly   prefers  site-specific
determinations.  The Agency could consider the use of a provisional delisting, whereby a waste
is delisted only if it is disposed at the site which is analyzed. The Subcommittee recognizes drat
 the policy criteria of the Agency may preclude this, indicating instead the use of the generic
 nationwide evaluation.  Should  the  Agency decide  to utilize the  nationwide Monte  Carlo
approach, the  assessment would be improved by incorporating regional variations  in the
assessment and explicitly banning the disposal  of delisted wastes in vulnerable hydrogeologic
settings so that these sites may be excluded from the model assessment.

     These findings and  recommendations are made for the use of the CANSAZ and EPACMS
models  in the  limited fashion described in the report.  We are pleased to have had the
opportunity to be of service to the Agency, and look forward to your response to this report.
                                        Sincerely,
       .	.	         _               	***>*MsL-
 i- i	,ji1...M^^^^^^.i..iiiiiii-iiiiiiiiiii ii                 MM 	 ~>ii^	r-JOB—1	1	«,,un,,,,im,	lmm^^^^11    .•niFHiwi	"Jr 	
Raymond C, Loehr, Chairman             Richard A. Conway, Chairmaii a
Executive Committee                     Environmental Engineering Committee
Science Advisoty Board                  Science Advisory Board
                                        Mitchell J. Small, Chairman
                                        Saturated Zone Model Subcommittee
                                        Science Advisory Board *

-------
                U.S. Environmental        Waahlnf ton, DC
                Protection Agency        EPA-SAB-EEC-90-009
          Report of the Saturated Zone

          Model Subcommittee
          Review of the CANSAZ Flow and

          Transport Model for Use in EPACMS
A SCIENCE ADVISORY BOARD REPORT                  March 1990

-------
                                           NOTICE

This report has been written as a part of the activities of the Science Advisory Board, a public advisory
group providing extramural scientific information and advice to the Administrator and other officials of the
Environmental Protection Agency,  The Board is structured to provide a balanced, expert assessment of
scientific matters related to problems facing the Agency,  This report has not been reviewed for approval
by the Agency; hence, the comments of this report do  not necessarily represent the views and policies of
the Environmental  Protection Agency or of other Federal agencies.  Any mention of trade names or
commercial products does not constitute endorsement  or recommendation for use.

-------
                                        ABSTRACT

      The Saturated Zone Model Subcommittee of the Environmental Engineering Committee of the EPA
Science  Advisory  Board (SAB)  has  prepared a  report on  the Agency's CANSAZ (Combined
Analytical-Numerical Saturated Zone) flow and transport model for use in EPACMS (the EPA Composite
Mode! for Surface impoundments). The Subcommittee examined the appropriateness of the assumptions
underlying CANSAZ for use  in a nationwide Monte Carlo assessment, as well as the adequacy of the
code for simulating conditions beneath surface impoundments. Specifically, the Subcommittee evaluated
five topics which were  1J proper  mathematical formulation, 2) adequacy of physical assumptions. 3)
adequacy of the Monte Carlo approach, 4) adequacy of field testing of the model, and S) the overall
modeling approach.

      While the mathematical equations  and the numerical solution methods are property formulated and
solved, the Subcommittee suggests improvements to the physical assumptions, improvements to the
current input data-set documentation and validation, and improvements needed to validate the  EPACMS
model for use in either a site-specific evaluation or a nationwide assessment.  The Subcommittee highly
prefers site-specific evaluation, but recommendations are made to improve the assessment should the
Agency choose to utilize the nationwide Monte Carlo approach. These suggestions are to incorporate
regional variations  in  the assessment, explicitly  ban the disposal of delisted  wastes in  extreme
hydrogeoiogtc settings, and provide a mechanism for all constituents in a  waste to be evaluated in a
listing or delisting decision.
Key Words: modeling, saturated zone modeling, surface impoundment models.

-------
                       SATURATED ZONE MODEL SUBCOMMITTEE
                      ENVIRONMENTAL ENGINEERING COMMITTEE
                                         of the
                              SCIENCE ADVISORY BOARD
Chairman
     Dr. Mitchell J. Small
     Department of Civil Engineering
     Carnegie Mellon University

     Dr. Mary P. Anderson
     Department of Geology and Geophysics
     University of Wisconsin - Madison

     Mr. George Carpenter
     Environmental Response Division
     Michigan Department of Natural Resources

     Dr. Keros Cartwright
     Illinois State Geological Survey

     Or, Ishwar P. Murarka
     Electric Power Research Institute

     Dr. Paul V, Roberts
     Department of Civil Engineering
     Stanford University

     Executive Secretary:
     Dr. K. Jack Kooyoomjlan

     Staff Secretary
     Mrs.  Marie Miller

     Director, Science Advisory Board:
     Or, Donaid G. Barnes
     Science Advisory Board (A101F)
     U.S.  Environmental Protection Agency
     401 M Street, S.W.
     Washington, D.C. 20460

     Chairman, Executive  Committee, Science Advisory Board:
     Dr. Raymond C. Loahr
     Civil Engineering Department
     University df Texas

     Chairman, invironmental Engineering Committee:
     Mr. Richard A, Conway
     Senior Corporate Fellow
     Union Carbide Corporatism.

-------
                                 TABLE OF CONTENTS
1.0    EXECUTIVE SUMMARY..,,	,„......,.	„	..	1

2.0    INTRODUCTION AND BACKGROUND,-..	.	.	B
       2,1   Proposed Uses for CANSAZ and EPACMS	,,5

3.0    ISSUES ADDRESSED IN REVIEW..,,,	„..	,	7
       3,1   Mathematical Formulation	,	,	....	,7
       3.2   Physical Assumptions,,,,,	.,.„	.,	8
       3,3   Monte Carlo Analysis	,	13
             3.3.1   Monte Carto Method	,	.	14
             3,3,2   Monti Carlo Inputs	 15
       3.4   Model Validation	,„......	 16
       3,5   Overall Approach	...	,,.......19

APPENDIX A - THE CHARGE TO THE SUBCOMMITTEE	„„„....	,.,22

APPENDIX 8 - GLOSSARY OF TERMS	I...	25

APPENDIX C * RESOURCE MATERIAL AND REFERENCES CITED,.,,,,,	26

-------
                                10  EXECUTIVE SUMMARY

      This report presents the EPA Science Advisory Board's (SAB) review of the Office of Solid Waste's
 (OSW) CANSAZ (Combined Analytical- Numerical Saturated Zone) flow and transport model for use in
 the EPACMS (EPA Composite Model for Surface  Impoundments).   This review is based upon  two
 documents, CANSAZ, published in October 1988 (Ref. 5) and EPACMS, published in April 1989 (Ref. 9}
 and briefings by the OSW to the Saturated Zone Model (SZM) Subcommittee at a meeting on May 30-31,
 1989.

      The use of CANSAZ and EPACMS distinguishes the OSW model for surface impoundments
 (EPACMS) from the OSW model for landfills (EPACML). which uses a  simpler flow and transport model
 for the saturated zone which is unable to account for mounding effects. The CANSAZ model is the
 saturated zone  component of the EPACMS, which  also includes a  module far ttie saturated zone,
 FECTUZ (Finite Element Code for Simulating Row and Transport In the Unsaturated Zone).  FECTUZ
 was reviewed previously by the Environmental Engineering Committee of tie Science Advisory Board
 (Ref. 20).

      The primary intended use of EPACMS is in the RCRA delisting program, where it will be used to
 establish ttie relationship between constituted concentrations  in surface impoundment leachate and
 concentrations at downgradient welt locations (i.e., the dilution  attenuation factor).  The model will be
 used in a Monte Carlo mode to perform a nationwide assessment of dilution attenuation factors for the
 delisting decision.  The Subcommittee review focused on the model in this use, but because CANSAZ
 and EPACMS could also be used for other, possibly site-specific applications, use in this mode was
considered as well.

     To assess the adequacy of CANSAZ and its use in EPACMS, the  Subcommittee considered issues
related to the technical validity of the model, and broader issues  of model use and implementation in the
proposed regulatory approach.  The five  areas that were examined  by the  Subcommittee  are briefly
stated as follows:
    .   1)   Proper Mathematical^orrnulation.
       2)   Adequacy of Physical Assumtions,
       3)   Adequacy of Monte Carlo Approach,                        *
       4)   Adequacy of Field Testing of the Model, and
       5)   Overall Modeling Approach.
     Findings of the Subcommittee on each of the above areas are summarized as follows:

-------
        1)   Proper Mathematical Formulation:
      Given the assumptions of the model, are the numerical formulation and solution methods correct?
The CANSAZ model consists of  an analytical flow module coupled with  a numerical solute transport
module which uses a new method known as a the Laplace Transform Galerkin (LTG) method.  The
equations,  boundary conditions and  solution  methods appear  to be  properly formulated given  the
assumptions of the  model.  The technique employed is innovative and computationally efficient  The
accuracy of the method is supported by rigorous comparisons with known analytical solutions and other
numerical methods.   The Subcommittee thus concludes that  the mathematical equations and  the
numerical solution methods are properly formulated and solved,
        2)  Adequacy of Physical Assumptions:
      Does the model include the important  processes  (ground water flow, pollutant transport and
transformation) which determine contaminant fate?  Is it properly formulated in terms of dimensionality,
spatial and temporal aggregation?  Are there omitted processes and what are their potential implications
for this assessment?

      The  CANSAZ model  includes the basic  assumptions  and  processes employed in  current
management-oriented  ground water models.  It  represents a  significant advance  over the  Vertical
Horizontal Spread (VHS) model currently  used for delisting. However, management-oriented models of
this type are seriously limited in their assumptions and neglect or oversimplify many important processes
now known to Impact contaminant transport and fate at actual field sites.  These assumptions involve the
dimensionality of the model (failure to include transverse flow from mounding), t)e limitation to steady-
state prediction, the assumption of homogeneous aquifer media, the inclusion of limited  hydrodynamic,
chemical and biological processes, and assumptions concerning ieachate and source characteristics-
Many of these limitations are noted explicitly in the CANSAZ reports and documentation.

      Particular procesess which are not incorporated in the  model include the transport of immiscible
organic phases, density-dependent vertical transport of the plume, vertical movement of the water table,
nonequilibrium adsorption of contaminants on the soil matrix, biodegradatfon,  multispecies chemical
reactions, cosolvent effects,  moWteation of chemical conditions (e.g., pH, Eh,  DO) in the receiving
aquifer, and the effects of background concentrations.  As a result of these limitations, tie CANSA2-
EPACMS  model is  not adequate for rigorous site evaluations which need to  consider site-specific
processes and phenomena not incorporated in the CANSAZ-EPACMS model. The model can, however,

-------
 be used  for a broad  national  assessment of regulatory impacts provided adequate  input data are
 obtained,  The development of a capability to incorporate the processes discussed above in regulatory
 models will require extensive, long-term data collection and model development research,
        3)   Adequacy of Monte Carlo Approach:
      Is the probabilistic method technically valid and supported by adequate input data?  Trie purpose of
 the Monte Carlo approach is to predict the distribution of the dilution attenuation factor (DAF) that would
 occur between surface impoundment leaehate and downgradient well points at surface impoundment
 sites throughout the United States,  To perform this analysis,  a distribution of inputs for the EPACMS
 model is developed and sampled to represent the distribution of meteorological conditions, soil properties,
 impoundment size and geometry, and welt location at existing impoundments.

      The mechanics of the Monte Carlo procedure appear  to be designed and implemented properly.
 The Monte Carlo module includes an impressive range of available distributions and is well integrated
 with the transport code.  There are, however, some significant problems In the method presented-  In
 particular, methods are needed to  incorporate correlation among  parameters currently  assumed to be
 independent, such as the hydraulic conductivity of the soil and the hydraulic gradient at the site, the soil
 porosity and bulk density, and soil properties and temperature which may covary on a regional basis.

      While the mechanics of  the Monte Carlo  method are  generally acceptable, determining the
 adequacy of the  input distributions is more problematic.  A  property formulated model can  provide
 accurate and meaningful predictions only if its inputs and parameters are correctly estimated.  The current
 input  data set is not  adequately  documented or  supported  by field data.   To correct this,  it  is
 recommended  that a panel of hydrogeoiogists, soil scientists, and  engineers be convened to review the
proposed model input values and documentation, and consideration be given to subjecting the resulting
data set to public review and comment prior to implementing the EPACMS delisting procedure.
       4)   Adequacy  of Field Testing of the Model:

      Has the  model been tested  (calibrated or validated) at particular field sites?  What methods are
appropriate for  validating the nationwide Monte Carlo approach?  The EPA has made  a good start at
validating the EPACMS model by confirming the numerical accuracy and presenting a preliminary  field
                                                                     #
validation study. This field validation study should be documented, and extended in more  detailed studies
at other sites.   A special effort is needed to validate the EPACMS modal if it is used  in a nationwide
assessment.   An  extensive nationwide data collection  and  monitoring program will be  needed  to

-------
 accomplish this. This will provide information on the proper inputs to the model and indicate the degree of
 confidence  and  conservatism  in the  predicted  distribution  of the  dilution  attenuation  factors.
 Improvements in tnt available data base for site characterization are needed to match the improvements
 which have occurred in modeling technology.
        5)   Overall Modeling Approach:
      Is the nationwide Monte Carlo approach appropriate for the intended uses (e.g., delating a waste,
 or closure  of a site}?  Which inputs require site-specific  data versus national distributions?  Are there
 alternative approaches?

      The nationwide assessment  using EPACMS applies a generic model to all potential impoundment
 sites in the United States, with  the Monte Carlo evaluation used to capture the site-to-site variability in
 model parameters.  Site-specific hydrogeologie parameters are not used to evaluate the deiisting petition
 of a particular facility.

      Site-specific determinations  of waste  disposal impacts are highly  preferred based  on  scientific
 criteria. The variations between sites resulting from variations in hydrogeologic conditions are known to
 be so great that the particular conditions of storage or disposal must be specified to allow for scientifically
 credible evaluation. The Subcommittee recognizes that the use of a site-specific model to make deiisting
 decisions would require a new  administrative approach to the regulation, as a waste could be delisted
 only for  the site  that was analyzed.  To implement this,  a provisional deiisting approach could be
 developed, whereby a waste would revert to its hazardous status if it is disposed of at any site other than
 that approved.  Because the EPA could decide that their administrative and policy objectives dictate that
this, or other site-specific approaches, art inappropriate or infeasible, the Agency may choose to utilize
the generic, nationwide approach. If this is the case, the Subcommittee believes the nationwide approach
could be improved by incorporating the following features:
       a)   Incorporating regional variations in the assessment;
       b)   ExpJteKty banning the disposal of delisted wastes in vulnerable hydrogeologic settings,
            thereby allowing these to be excluded from the model assessment; and,
       c)   Providing a mechanism for all significant constituents in a waste to be evaluated in a listing
            or deiisting deeisiorTShd ensuring that all significant constituents that must be
            quantitatively evaluated are analyzed with the EPACMS model. -
Furthermore, the Subcommittee would encourage any evolution in the Agency programs which would
improve the capability to assess groundwater contamination on a site-specific basis.

-------
                          2.0   INTRODUCTION AND BACKGROUND

      In April 1989, Devereaux Barnes, Director of the Characterization and Assessment Division of the
 Office of Solid Waste (OSW), requested that the Environmental Engineering Committee (EEC) of the SAB
 conduct a review of the Combined Analytical-Numerical Saturated Zone (CANSAZ) flow and transport
 model for use in the EPA Composite Model for Surface Impoundments (EPACMS). Trie formal request is
 included as Appendix A of this report.  Trie CANSAZ code was developed to simulate the flow and
 migration of contaminants beneath surface impoundments, where hydraulic mounding could occur. The
 CANSAZ module for the saturated zone, together with the unsaturated zone code (FECTUZ), comprise
 the OSW fate and transport module for surface impoundments (EPACMS). The unsaturated zone code
 FECTUZ was reviewed previously  by the SAi (SAB-6EC-88-Q3Q), The use of CANSAZ distinguishes trie
 OSW model  for surface impoundments (EPACMS) from trie OSW model for  landfills (EPACML).
 EPACML uses a simpler flow  and transport model for the saturated zone that does not account  for
 mounding effects.

     This review is based upon two documents received by the Saturated Zone Model Subcommittee:
     1, CANSAZ:  Combined Analytical-Numerical  Code tor Simulating Flow and Contaminant
      Transport in the Saturated Zone, Prepared by E.A. Sudicky, University of Waterloo and
      HydroGeologic, Inc.  for  U.S. EPA OSW, October  1i88  (ANM/123D/FT, ID No.  89-017).
      (Reference 5).
     2. EPACMS:    Composite  Model  for  Simulating   Leachate  Migration  from  Surface
      Impoundments and Monte Carlo Uncertainty Analysis, User's GukJe, Prepared by Hydro-
      Geologic Inc. for U.S. EPA OSW, September 1988; Revised April 1989 (ANM/123D/FT, !0
      No. 89-030). (Reference 9),

     In addition, the Subcommittee met and was briefed by the OSW and its consultants on May 30-31,
 1989. The Subcommittee review  encompasses both  the written reports and the oral presentation and
discussion which ensued at this  meeting.
2.1   Proposed Uses tor CANSAZ and EPACMS
     An important consideration in the Subcommittee review of CANSAZ ajti EPACMS is the intended
uses of the model.  Considerable discussion occurred on this issue at the Subcommittee meeting, and
additional documentation was provided by OSW.  The following, excerpted from this documentation,
describes the intended uses:

-------
      OSW is considering using the coupled CANSAZ and FECTUZ modules (EPACMS) in the fiCRA Dfrlfeting
     Program, In this use, the Agency intends to specify the modal input parameters, to the extent possible, as
     distributions based on nationwide data. The use of  the model in the Delisting Program along with other
     potential uses are described below:

      Dellstlng Program

      Under  the RCRA  Delisting  Program, individual waste generators cart petition  th« Agency to exclude
     ("dslisf)  their wastes from the  lists of hazardous wastes in the Federal Code of Regulations (40 CFR
     261.32).  An integral part of the delisting evaluation is the use of fate and transport models as  predictive
     tools to estimate dilution/attenuation of chemical constituents leaching from waste sttes to nearby drinking
     water sources.

      The model in currant use in the Delisting Program (the VMS model) is a simplified one, The Agency has
     stated that the simplified model will  be replaced by comprehensive ones when they become available in a
     form which is appropriate for use in the Delisting Program.

      The CANSAZ is of interest to the Delisting Program because a large number of wastes which are the
     subject of delisting petitions are managed in surface impoundments. There is no model currently in use for
     delisting  that directly accounts for the specific differences between landfills and surface impoundments, a
     factor which has raised comments in the review of numerous petitions. The surface impoundment code (if
     adopted) would provide additional flexibility for the Deiisting Program and wouW help increase th* efficiency
     of the review process.  Since delisting decisions are rulemakings and require Federal Register notice and
     promulgation, any specific uses of the model would be proposed for public comment.

      Other Uses

      The OSW does not  have any other specific uses planned for CANSAZ at this time.  However,  it is
     anticipated that any  other uses,  if identified, would  be for the development of regulations under  RCRA for
     the identification of hazardous wastes, The potential  uses would be limited to rnisefcle flows (non-oily
     wastes),  and the code would be implemented  with most of the input parameters as the  Monte Carlo
     variables.


      As indicated, the primary intended  use of EPACMS is in the  RCRA delisfing program, where a

nationwide assessment of dilution attenuation factors will be performed to determine maximum allowable

contaminant  concentrations for delisting impounded wastes. In this  application, the EPACMS will replace

the  current  VHS  model.   The Subcommittee  review focused on this  use  and  evaluated  the

appropriateness of CANSAZ and EPACMS for use in developing a nationwide rule. Because the uses of

CANSAZ and EPACMS could include other, possibly  site-specific  applications,  the Subcommittee also

considered the adequacy of CANSAZ and EPACMS tor use in site-specific evaluations.

-------
                             3.0  ISSUES ADDRESSED IN REVIEW


      The request  from  OSW identified  two  major areas for SAB  review.  The  first relates to the

 appropriateness of  the assumptions underlying CANSAZ when  used in  a nationwide, Monte Carlo

 assessment, including which parameters should be estimated on a site-specific vs. a nationwide basis.

 The second relates to the adequacy of the code for simulating conditions beneath surface impoundments.

 The Subcommittee focused on these issues by evaluating five topics:

     1. Proper Mathematical Fgimujation:  Given the assumptions of the model, are the numerical
       formulation and solution methods correct?

     2- Adequacy of Physical Assumptions:  Does the model include the  important processes
       (ground water flow, pollutant transport and transformation) which determine contaminant
       fate? Is it properly formulated in terms of dimensionality, spatial and temporal aggregation?
       Are there omitted processes and what are their potential implications for the assessment?

     3. Adequacy of Monte Carlo Approach:  Is  the probabilistic method technically valid  and
       supported by adequate input data?

     4. Adequacy of Field Testing of  the Model:   Has the model  been tested (calibrated or
       validated)  at particular field sites?  What  methods are  appropriate  for validating the
       nationwide Monte Carlo approach?

     5. Overall Modeling Approach:  Is the nationwide, Monte Carlo approach appropriate tor the
       intended uses (e.g., delisting a waste, or closure of  a site)?   Which inputs require site-
       specific  data versus national distributions?   Are there alternative approaches?  Can the
       Subcommittee propose a data collection program to support the assessment?


      The Subcommittee  review thus begins with more narrow questions related to the technical validity

of the model, and moves to broader issues of model use and implementation in the regulatory setting.

Each of the above areas is now addressed.
3.1   Mathematical Formulation


      The CANSAZ model consists of two major components;  A flow model and a solute transport

model.  The flow model is solved analytically.  The solute transport model is solved with a new method

known as LTG that combines the Laplace Transform (LT)  and a finite element Galerkin (G) method.

Details of the new solution techjtiflue are contained in a paper that has been published in a refereed

journal (Reference 17).                                                *


      The LTG solution approximately simulates dispersion  transverse to the cross section,  rather than

rigorously.  Tests suggest that  this quasi-3-D solution is  relatively accurate when compared  with a

-------
                                               8

 rigorous 3-D dispersion solution.  Extensive comparisons of LTG solutions with analytical solutions and
 conventional finite element solutions suggest that the numerical trrors associated with LTG are of the
 same magnitude or smaller than other numerical solutions. CANSAZ also provides an analytical solution
 option for solute transport.  The analytical solution has more restrictive assumptions than the numerical
 solution option, except that the analytical solution treats three-dimensional dispersion rigorously.

      The equations and boundary conditions appear to be properly formulated for the assumptions of
 the model.  The  model's developers are among the foremost practitioners of hydrogeologic transport
 model development   The  technique employed  is  innovative,  computationally efficient, and  highly
 accurate, judging from  the comparisons  with  analytical  solutions  presented in  the  background
 documentation.    The  LTG  method  appears  to  represent a  significant  advance  over  previous
 computational methods in terms of efficiency and robustness. However, the LTG method incorporated in
 CANSAZ is intrinsically limited in  its ability to deal with heterogeneity and three-dimensional transport.
 The implications of these and other limiting model assumptions are addressed in tie following sections.


 3.2  Physical Assumptions

      The CANSAZ model includes only a limited set of ground water flow and contaminant transport and
 fate processes. These include the processes generally incorporated in the current generation of ground
water  management  models.  The  assumptions are internally consistent  and represent a distinct
improvement over the VHS model. However, given the rapid evolution of ground water science, it is dear
that in many  respects the model does  not  adhere rigorously to currently understood and emerging
principles of flow and  transport.  In most respects, the deviations  represent simplifications  that are
justifiable in terms of computational expediency.  However, these simplifications inherently  limit the
accuracy of the model's predictions,  and will particularly limit or preclude its use tor detailed site-specific
applications and decisions,

      The major assumptions inlRe CANSAZ model concern the dimensionality of tie model formulation,
                                                                     *
the aggregation  in  time  (i.e.,  dynamic  vs.  steady-state simulation),  aggregation  in  space  (the
consideration  of  spatial  heterogeneity)  and  the inclusion  of contaminant  flow and transformation
processes.  '

-------
      Dimensionality:  CANSAZ is restricted to two-dimensional flow in  the vertical and longitudinal
 directions, ignoring the horizontal transverse flow resulting from mounding.  Formation of a mound at the
 water table  caused by seepage from the impoundment will cause radial flow away from the mound,
 including  a horizontal component of flow transverse to the principal axis.  If attention is focused on the
 centerline of the plume, the effect of neglecting the component of flow transverse to the cross section is
 conservative in that concentrations will be higher in CANSAZ simulations than in a three-dimensional flow
 simulation.   Indeed,  it  is  generally  the case  tfiat  two-dimensional flow  models predict  higher
 concentrations along the longitudinal axis than appropriate for three-dimensional ftow fields  (Reference 4,
 14),  (Note:  Although CANSAZ neglects  flow and advection of contaminants transverse to the cross
 section, it does simulate dispersion of contaminants transverse  to the cross section.)  While the two-
 dimensional  flow  field  assumption is  generally  conservative,  ignoring  transverse  flow  beneath
 impoundments can seriously misrepresent the shape of the plume, particularly when the regional ground
 water flow is low compared to the impoundment infiltration rate. The CANSAZ model is not appropriate
 for use at sites where this  is the  case.   Moreover,  if the  receptor well  location is assumed to vary
 randomly  between o" and 45° relative to  the  plume for the Monte Carlo analysis, as described  in the
 EPACMS  User's Guide (Ref, 9, pp 101-103),  then the two-dimensional model will not be conservative.
 Rather it will underestimate concentrations at observation points that are far removed from the principal
 plume axis.

      Steady-State .Conditions:  The flow field is assumed to be steady-state.  Also,  infiltration from the
 impoundment is assumed to continue indefinitely.  High infiltration  rates cause high velocities  in the
 saturated  zone. Under the  steady-state  assumption,  the high velocities  will persist indefinitely.  The
 model does not allow for the possibility of dilution and  attenuation  of the plume if seepage from the
 impoundment  should  cease.   The assumption  of  steady-state  flow  (and  steady-state  input  of
contaminants) wil resutt in conservative predictions of concentration.

     The assumption of steady-state  conditions is  a major  simplification  that  greatly facilitates
computation and communication of the results.  However, such  a steady-state must be  considered
 hypothetical  as well as conservative, as it  has yet to be  observed in real situations,  and,  therefore, the
                                                                     *
corresponding predictions are not amenable to field verification. The steady-state condition is acceptable
as a benchmark for Monte Carlo analysis of policy decisions, recognizing that any site-specific analysis,
including  comparisons with  field  data for purposes  of verification, should  be based on  transient

-------
                                              10

 simulations.  Once a verification of this type is performed, a simpler model formulation, limited to only the
 steady-state condition, could be used for regulatory application, A simpler formulation would be easier to
 understand and follow by those in the EPA and the regulated community likely to use the model,
                                                                      V
      Homogeneous Aquifer:  The flow component of CANSAZ assumes that the aquifer is relatively
 homogeneous. Because the solute transport component of the model utilizes output from the flow model,
 it too is necessarily restricted to homogeneous aquifers. Aquifers, of course, are never homogeneous.
 While the assumption of homogeneous aquifer material has been frequently used, recent research makes
 it clear that the assumption is inappropriate for simulating solute transport in ground water. Most disposal
 sites are  highly non-uniform, with high-permeability zones that constitute conduits  of rapid transport.
 Paths of high hydraulic conductivity are important avenues for contaminants in fractured rock and day,
 and also exist in continuous porous media (Ref. 2,16).

      There  is currently much discussion in the literature over the appropriate way to incorporate
 heterogeneities into solute transport models,  in one approach, the heterogeneities are accounted for in
 part by adjusting  the dispersion coefficient, but this strategy cannot adequately represent  the extreme
 cases of heterogeneity such as fractured media, as these have yet to be studied adequately and deviate
 significantly from the advtctjon-dispersion model. Many go beyond  this viewpoint  and suggest that the
 advection-dispersion equation is simply not valid for application to heterogeneous  aquifers  of any type.
 However,  at present, no consensus has emerged for a  practical alternative to the advection-dispersion
 equation which m used  in CANSAZ.. Nor  is there consensus over  the appropriate way to incorporate
 heterogeneities.  In view of these doubts about the validity of the assumption of homogenetty, it is critical
 that EPA  remain  flexible and be  willing to replace CANSAZ with another approach  when there is a
 consensus.   Also, it would be prudent at  this time to establish strict criteria to delimit the amount of
 heterogeneity that can be tolerated in applying CANSAZ  in  site-specific applications. This criterion might
 be stated  in terms of a permissible range  of hydraulic conductivity variation at a given site:  e.g., four
 orders of magnitudes or  less.   The framing  of  this  criterion should  be included in the scope of
 deliberations of an a
-------
                                              11

 on the contrary, that many impoundments and landfills contain significant amounts of immiscible organic
 phases,  and that migrations of such phases can be an important avenue of transport.  This issue was
 also  raised in the SAB's  review  of  the  unsaturated  zone transport model (FECTUZ),   Failure  to
 incorporate immiscible transport constitutes a serious  limitation, especially in site-specific  applications
 where immiscible contaminant phases  are believed to exist.  While the OSW indicates that CANSAZ will
 only be applied for misctble flows (non-oily wastes), multiphase conditions are likely to be present in many
 situations, even at Subtitle D facilities.

      A  second hydrodynamic limitation is that CANSAZ presumes contaminants are  dilute  and that
 density-dependent transport plays no significant role.   Even  slight  density differences can exert a
 significant influence on vertical plume movement For example, two well-documented transport studies
 (at the Borden and Cape Cod sites) have revealed pronounced vertical dips in plume  movement with
 solute concentrations as low as 1 g/liter. Such amplification of vertical plume movement may not be of
 great consequence in Monte Carlo simulations conducted with a  presumed  uniform  distribution  of
 monitoring points over the depth. However, in any site-specific  application with explicit positions for the
 monitoring points, it is essential to include this phenomenon.

      in  terms of the hydraulic conditions represented by EPACMS, the linkage between the saturated
 model CANSAZ and the unsaturated model FECTUZ does not allow for movement of the water table that
 would shorten the length of  the  unsaturated column represented in FECTUZ,  This assumption is not
 likely to affect the proposed regulatory  application, but could be  an important effect in other  applications
 of the linked mode!.

      Chemica(_ Prpcessgs_:  The CANSAZ model assumes that all chemical transformations can be
 represented by simulating two processes; adsorption and first-order decay due to hydrolysis. Adsorption
 is represented using the  linear equilibrium model.  As such, adsorption has no effect on steady-state
 predictions, unless the first-order decay is assumed to occur in tie adsorbed as well as the liquid phase.
 Adsorption does affect transient predictions, however, delaying the arrival of the plume.

      The assumption of linear equilibrium adsorption appears to be approximately valid tor most organic
contaminants at low concentration.  However, there is a general consensus that linear equilibrium does
not apply to metals. CANSAZ should not be applied in its current form to simulate metal transport.

-------
                                                 12

       Biodegradation is not currently included in the CANSAZ implementation, although it is recognized
 as being important in many situations, and can, in principle, be incorporated by adjusting the first-order
 decay rate in the model.   The implications of omitting biodegradation were discussed In detail in the
 SAB-EEC  review of FECTUZ (SAB-EEC-8S-Q3Q).  There it was noted that:
      ...consensus  is lacking for generalized prediction of transformation rate constants, as these depend
     strongly on conditions such as organism adaptations and concentrations, pH, and the presence or absence
     of electron acceptors (oxygen under aerobic conditions), toxicants,  essential nutrients, etc.  which are
     site-specific. Site-specific applications of the FECTUZ (or, in general, the EPACMS) model package can
     lead to over-estimates of solute transport since site-specific biotransformation analyses generally result in
     biodegradation being a primary process influencing chemical fate. Hence, estimates of chemical transport
     made  without considerations of biotransformation are almost always  so overly conservative as to affect
     regulatory decisions.  Generalized chemical transport predictions will necessarily suffer dye to lack of
     generally applicable biotransf or matron rate constants; however, site-specific analyses should include ail of
     the fate processes for which specific data can be reasonably obtained,
      Implementation of the full  range of  transformation possibilities, including the uncertainties In condftiorts
     that influence  the rate constants,  would  magnify enormously the  uncertainty spectrum of  predicted
     outcomes in Monte Carlo simulation.
 The report further notes  that any  inclusion of biological transformation  must  explicitly  consider the
 formation of possibly hazardous byproducts.

      The  CANSAZ and EPACMS  models  do not allow for chemical reactions between  two or more
 chemical species. As such, CANSAZ is generally restricted to the simulation of groups of contaminants in
 the dilute range.  The documentation, however, does not specify that range. EPA should state specific
 criteria for the upper limit of concentration.   At  higher concentrations,  a variety  of phenomena not
 considered in CANSAZ may play a  role,  including the following: facilitated  transport, cosolvem effects,
 and competition for sorption sites, Most of these factors would tend to increase contaminant mobility, and
 thus to increase the potential health threat,

      An additional chemical  assumption in CANSAZ is that contaminants exert no effect on ambient
chemical conditions.  The impact of the  waste on the receiving environment can in fact be  important
Concentrated leachates often contain sufficient  acidity or alkalinity to change the pH  in  the plume
substantially, even where  the native ground water possesses moderate buffer capacity.   A change in
 leachate pH, particularly acidification, would enhance the mobilization of  many metal species, while
concentrated solvent exposure rgay dry and fracture clays.  Many solvents are not contained by clays and
could  also enhance other contaminant  transport.   Similar considerations .apply to dissolved oxygen
concentration and redox state (Eh),  It is probabiy  infeasibte to take this  coupling into  account  on  a
generalized, nationwide basis. However, these interactions should  be accounted tor in any site-specific
application in which  the important processes (e.g., sorption, transformation  or hydraulic  stability of

-------
                                              13

 confining layer) depend importantly on geochemical conditions (pH, Eh, DO, presence of solvents),

      Leachate and Source Characteristics: The discharge rate from an impoundment Is determined by
 she impoundment size (cross-sectional area) and infiltration rate. The infiltration rate can be input directly
 to the model or computed from the depth of the fluid in the impoundment and the thickness and hydraulic
 conductivity of the impeding layer at the base of the unit.  It is unclear from the reports, however, as to
 how the impoundment sizes and tht amounts of wastes are generated  for the nationwide, Monte Carlo
 assessment.  It should be noted that the resulting leachate composition and flux could have a large effect
 on the dilution attenuation factors and the resulting decisions.

      The CANSAZ model evaluates the impact of the impoundment without considering other facilities or
 background contamination which  may be present  Few impoundments  are located away  from other
 facilities, Other landfills or lagoons, agricultural leachate, or process or potable water withdrawals may be
 present and alter site conditions. In the CANSAZ simulations, the saturated zone is assumed to be free
 from contaminants initially.  In areas with industrial or agricultural  pollution, or with  naturally occurring
 sources of certain constituents, contaminants may be present in background concentrations. The effects
 of facilities may also limit applicability of individual replications or runs with fixed flow rates or infiltration.
 Again, these  limitations  are most important when considering site-specific evaluations.  The CANSAZ
 model needs  to have  the ability to be  initialized for existing conditions  and incorporate other flow-field
 modifications  when used in a site-specific application.

      fn summary, the CANSAZ model incorporates a number of simplifications  concerning the modet
dimensionality,  temporal representation, assumption of  homogeneity,  omission  of  hydrodynamic,
chemical and microbiological  processes, and representation  of leachate and source  characteristics.
These assumptions preclude the use of CANSAZ for rigorous site-specific evaluations.  However, the
model is formulated at the proper level of detail for  nationwide assessments, so long as proper and
representative rnodrt inputs can be determined.
3.3  Monte Carlo Analysis .

      The purpose of the Monte Carlo approach is to predict the distribution of the dilution attenuation
factor (DAF) that would occur between surface impoundment leachate and downgradient well points at

-------
                                               14

 surface impoundment sites throughout the United States, A smaller value of the DAF implies less dilution
 and attenuation by the aquifer, resulting in a greater impact at the receptor well. To perform the analysis,
 a distribution of inputs for  the EPACMS model is developed to be representative of the distribution of
 meteorological  conditions, soil properties, impoundment size and geometry, and well location at  existing
 impoundments. The joint input distribution is sampled many times in the Monte Carlo analysis,  and the
 model is evaluated with  a  specified waste  quantity for each sample, resulting in a calculated national
 distribution for  the DAF conditioned on the waste quantity. The DAF for which only a small percentage
 (e.g., 15 percent) of the simulated sites are less than the given value is selected as the design DAF which
 is assumed to occur for the given waste and waste quantity.  Because the DAF is affected by chemical
 properties, including hydrolysis rates and adsorption coefficients, the analysis must be performed for each
 chemical,
     •>S'
      There are three issues that  the Subcommittee addressed to determine the validity of the Monte
 Carlo approach:
     1, Is the nationwide, Monte Carlo approach appropriate for delisting decisions?
     2, Are the  mechanics of the Monte Carlo method properly formulated?
     3. Are the  input distributions developed  for the Monte Carlo method adequately supported?

      The first  question is addressed in detail later in this report in Section 3.5, Overall Approach.  The
 general conclusion is that site-specific  evaluations are  highly preferred to  the use of a nationwide
 assessment but that if a nationwide assessment is performed because of administrative or other policy
 constraints, then the second and third issues must be property addressed. These are now considered,
 3.3.1   Monte Carlo Method


      The mechanics of the Monte Carlo procedure appear to be designed and implemented property,
 The Monte Carlo moduli includes  an  impressive range of available distributions  and is well  integrated
with the transport code. There are,  however, some particular problems in the method presented.

      The major problem is that covariance between model parameters is not incorporated.  This is a
particular problem for soil properties such as the hydraulic conductivity, hydraulic gradient, porosity and
bulk density which are likely to exhibit  a high degree  of correlation (in the  case of the hydraulic
conductivity and hydraulic gradient, a negative correlation is expected).  Other  examples include the

-------
                                               15

 aquifer geometry, soil properties, and temperature, which may covary on a regional basis,

      The  one area where covariance  is considered is in the simulation of soil characteristic curve
 parameters for the  unsaturated zone  component of the  model (FECTUZ).   Incorporating  a similar
 procedure  for the saturated zone parameters appears to be necessary, but could  result in significant
 computational difficulties.   Perhaps the most straightforward  way  to  represent the covarianee which
 occurs at actual field sites is to measure the soil parameters at many sites, and use the input vector for
 each site as a single, joint input to the model. The Monte Carlo analysis is then equivalent to running the
 model  for  each of  tie  sites sampled, assuming they provide a representative sample of the target
 population.  The implementation of a nationwide site sampling and characterization program, discussed
 later, would thus provide the most direct solution to the problem of identifying and incorporating input
 variable covariance.
 3-3.2  Monte Carlo Inputs

      While the mechanics of the Monte Carlo method  are generally  acceptable, determining the
 adequacy of the  input distributions is  more problematic.   A property formulated model  can  provide
 accurate and meaningful predictions only if its inputs and parameters are correctly estimated.

     The Subcommittee was not able to judge whether the proposed nationwide data base reported in
 the documentation to EPAGMS is appropriate for the intended regulatory use.  As such, the model inputs
 should be carefully  reviewed before EPACMS is disseminated  tor use. The estimation  procedures for
 EPACMS are in many cases undocumented, in other cases incomplete, and in some cases inappropriate.
 For  example,  methods  proposed  for  calculating porosity  and  hydraulic  conductivity  are wholly
 inappropriate and do not conform to standard hydrogeologic methods.   Estimation of the  hydraulic
conductivity from mean grain size via the Carman-Kozeny equation (Ref.  9, Eq. 8.2,14.)  is invalid for
 heterogeneous media, and in  any  case, the relevant grain size is d10, not the mean.  The hydraulic
conductivity and porosity should be estimated directly from representative field data, rather than indirectly.
 In other cases (e.g., for dispersivity, the ratio of horizontal  to vertical hydraulic conductivity, the aquifer
thickness, length of aquifer and distribution coefficient, K^). the documentation is inadequate  to judge
whether  the proposed distributions are reasonable.  Another  example is  the proposed distribution for
ground water recharge.  This distribution was generated using data from 100 cities analyzed with the
HELP model.  The  committee  was not able to review the  input data,  nor the HELP modtl itself.  The

-------
                                                16


 resulting recharge .distribution appears to be biased toward the Great Plains and Midwest.  It is unclear

 how such bias, if present in the recharge distribution, and other parameters, will affect the outcome of the

 Monte Carlo analysis.


      Some additional concerns were noted with regards to the Monte Carlo analysis. These include:

              a.   More documentation  is  needed  to  support  the Gelhar-distributton for longitudinal
        dispersivity (Ref. 9, pp 98-99),

              b.  Citation and documentation is needed for the EPIC and REA data on well distances
        (Ref, 9. p, 102). The report does not indicate what these acronyms represent

              c.  The  sensitivity analysis which was presented orally to illustrate the effect of different
        model parameters  should be included as part of the written report. Given the assumptions of the
        model on vertical mixing, and the screening depths generally used for wells, careful consideration
        should be given to  interpreting the effect of well depth.

              d.  "Monte Carlo" is not a verb and parameters are not "Monte Cartoed."  They art sampled
        or generated in a Monte Carlo analysis.

      In summary, it appears that ranges and distributions for many of the important inputs for £PACMS

have  been  estimated  without  the support of  adequate  field studies and documentation,  it is  thus

recommended that additional work  be done in this area, and  that a  panel  of  hydrogeologists. soil

scientists  and engineers be convened to review the  proposed or modified model input  values and

documentation.  Consideration should be given to subjecting the resulting data set to public review prior

to implementing the EPACMS delisting procedure. The gathering of fiild data and documentation for the

model inputs is an important part of the model validation exercise discussed in the following section.
3.4  Model Validation



      There are a number of steps that can be taken to validate models for use in regulatory decisions

(e.g., Reference 3, 6).   This issue was an important part of  the recent SAB-EEC  resolution on

mathematical models (Reference 21).  The resolution states that


      as a preliminary step, th« elemaatt of the basic aquations and the computational procedures employed to
    solve them  should toe tested to ensure that the  model generates results consistent  with its  underlying
    theory. Tha confirmed model should then be calibrated with field data and subsequently validated with
    additional data collected under varying environmental conditions. After the particular regulatory program
    has bean implemented, field surveys and long-term monitoring should be conducted for comparison with
    model projections. The stepwise procedure of checking the numerical consistency of a model, followed by
    field calibration, validation and a posteriori evaluation should bs an established protocol tor environmental
    quality models in all media, recognizing that the  particular implementation of this may differ for surface
    water, air and ground water quality models.  It is also recognized that the degree and extent to which the

-------
                                              17

     process of validation is conducted for a modal depends on the significance of the environmental issua and
     the consequsfica o< an erroneous decision concerning the problem.

      The  OSW and its contractors have taken steps to  verity and validate the EPACMS mode).  The
 numerical  consistency and  accuracy of the model have  been verified by comparisons with known
 analytical solutions and other numerical models.  An oral presentation was provided to the Subcommittee
 on the application of EPACMS to model the migration of aWiearb through the unsaturated zone and the
 underlying  aquifer at a field site on Long Island, New York.  Site parameters were estimated based on
 previous modeling studies and entered as input to the EPACMS model. Reasonable, ordeMjf-magnityde
 agreement between the model and observations was demonstrated, although the model did somewhat
 underestimate downgradient concentrations. The OSW personnel  indicated that a better  agreement
 couid be obtained by adjusting model parameters, but that an order-of-magnitude agreement is what they
 expect to be representative of field test conditions.

      The study presented by Dr. Saleem at the Subcommittee meeting represents a good start at model
 validation, and should be formally documented-  It does not, however, constitute a thorough validation.
 For example, the aldicarb site does not include  the flow dynamics of a leaking impoundment. Stilt, it is
 representative of the type of field  study that the Subcommittee  wishes to encourage.  Because the
 EPACMS is proposed for use in a nationwide assessment,  the Subcommittee identified a further set of
 validation studies for the Monte Carlo approach, considering both model inputs and outputs.  These are
 presented in a hierarchical manner, reflecting different levels of effort and resource commitment.

      The first approach consists of "limited validation" for one or more actual sites (such as the aldicarb
 study),  tn this approach, actual field measurements from  a site are utilized to conduct sensitivity and/or
 site-specific Monte Carlo analysis.  The statistical properties are defined by the measurements at the site.
 Once the model is exercised arid the distribution of outputs is obtained, the results are compared with the
 real-world observations for that site. If the model predictions are consistently higher than those observed,
 an estimate of the degree of conservatism can be obtained. Use of several sites in  this manner will
 increase the understanding of the expected level of bias and precision in the model results.

      In addition to the aldicarb site already analyzed, the EPA program  staff has been provided by the
 Subcommittee with  EPRf data (Reference  11)  for a 30-year-old site in New York State  where the
migration of organic compounds in the saturated  zone has been measured.  The  hydrogeological,
geochemical, and microbial biodegradation properties have also been investigated.  The CANSAZ model

-------
                                               18
 should  be applied in both a transient and steady-state mode and the results compared to measured
 concentrations* "This comparison should provide an additional validation of the model.

      The second approach in the hierarchy requires that a more extensive field validation effort be
 launched where several sites are rigorously characterized to generate the input data for the model.  The
 model is then exercised to predict steady-state concentrations tor each of the validation sites.  Reid
 measurements of the plume concentrations at each of the sites are statistically compared ID the model
 generated concentrations.   Further information on the consistency, precision and  degree of under- or
 overpredictton are obtained from this validation effort.

      The use of EPACMS in a nationwide assessment requires a consistent set of input parameters and
 boundary conditions.  The third level of validation requires a critical analysis of the input and output data
 sets generated by the Monte Carlo scheme. To determine the actual distribution of model input at sites !n
 the U.S., a broad data-gathering program  is needed. Following this, the  input data generated by the
 Monte Carlo analysis can be examined to establish which types of the monitored sites are represented
 and whether the proper geographical weighting is achieved. This type of validation should assure that trie
 distribution of inputs for the model is in close proximity to the distribution of sites in the U.S.

      The Subcommittee recognizes that a nationwide monitoring program represents a major effort, and
 would require  a special study and significant funding.  If, however, a nationwide evaluation is  used for
 developing regulations, it is necessary to support this effort with adequate nationwide data.  The first step
 in such  a study would be a systematic organization and cataloging of tie studies thus far conducted and
 the data collected at  present waste disposal sites, including land disposal and impoundment facilities.
 The information  already collected In site investigation studies should provide a good start in the national
 characterization  program, and will help Inform the selection of additional  sites and monitoring efforts.
Considerable cart will be  needed to ensure that  the data base  provides an unbiased sample of the
 national distribution of geologic and hydroiogic conditions at sites, as particular locations  may currently
 have  more information available due  to special  characteristics or public or political concerns.   A
 representative, national  characterization program can provide  the data necessary to estimate input
                              <•*•'•
parameters for  the nationwide  model, provide  information on downgradierit well concentrations for
                                                                       *
validation of the OAF distribution predicted by the national model, and provide insights on the problems
and processes which are most important it real disposal sites.

-------
                                                19

       It is clear that all three approaches will require resources and time to develop the data and conduct
 the validations. In the near term, we recommend that the Agency, through the QRD, conduct a limited
 validation  along ttie  lines of the first approach outlined  above.   Plans should then be developed for
 long-term  studies of  the type necessary for the second and third levels of validation, particularly if the
 generic, nationwide approach to the regulation is maintained.

 3.5  Overall Approach

 The nationwide assessment using EPACMS applies a generic model to all potential impoundment sites in
 the United States, with the Monte Carlo evaluation used  to capture the site to site variability in model
 parameters.  The Subcommittee believes, considering only scientific criteria, that the use of a site-specific
 assessment for making delisting decisions  is 'clearly preferred to the proposed generic -  Monte Carlo
 approach.  This viewpoint is based on the belief that facility-specific decisions on delisting should be
 based on  a site-specific evaluation of the facility where the waste is disposed.   Site-specific decisions
 require site-specific data.  The variation among sites resulting from variations in hydrogeotogie conditions
 is known to be so great that the site-specific conditions of  storage or disposal must be specified to allow
 for a scientifically credible evaluation.

      The  Subcommittee recognizes that  this perspective calls into question the entire  approach  to
 delisting currently espoused by the Agency,  The  current delisting procedure  applies to the waste,
 regardless of the facility used for treatment  or disposal, in evaluating delisting petitions for wastes, the
 Agency assumes that the waste could be placed in a Subtitle D land disposal unit anywhere in the United
 States. The Agency has therefore adopted  a national analysts approach in determining whether or not to
 reclassify a waste from hazardous to nonhazardous.  If  a waste is  redassified as  nonhazardous, its
 resulting disposal location is unspecified and the waste is no longer subject to Subtitle C regulation.

      The  viewpoint that only site-specific evaluations are acceptable leads to administrative difficulties,
 as it would require that the disposal site for the deiisted waste be specified.  However,  the entire purpose
 of the delisting procedure is jgj-emove the burden of hazardous waste regulations  from the waste  in
 question; requiring a  specification of the waste disposal location and requiring an analysis of potential
 impacts at that site would dictate a level of effort similar  to that tor facility permitting.  Solution of this
dilemma of an apparent conflict between the administrative and scientific objectives of the Agency would
 require a new approach to the regulation.  One possible suggestion is to conduct a site-specific analysis

-------
                                              20

 to implement a provisional delisting.  That is, if a given waste is disposed of at a given approved site, then
 it is delisted.  Any disposition at other than the given site makes the waste hazardous. The burden of
 gathering the site and waste data  and conducting  the model analysis {e.g.,  using  a model such  as
 CANSAZ with appropriate modifications for the particular site),  would  be placed fully upon the waste
 generator. The generator must demonstrate that the waste can be safely disposed of at the subject
 surface impoundment for the delisting petition to be acceptable.

      The Subcommittee recognizes that, based on administrative or other policy considerations {e.g.,
 the desire to facilitate rapid delisting without undue administrative requirements and delay), the Agency
 may elect to maintain the proposed nationwide framework. The question then arises as to how tits type
 of assessment can best be performed, in particular, whether rtgionalization is appropriate, and whether a
 very extreme or conservative decision threshold is necessary to be protective of the environment.

      The proposal for regionalization  of the ground water impact assessment has been made in a
 previous SAB-EEC report on the RCRA land-ban proposals (Reference 19). Without regionaftzatton, the
 same Monte Carlo data set is  used to represent all sites  across  the country.  Yet there am vast
 differences in the hydrogeology and environmental sensitivity  across the country. Sites in the arid west
 with hundreds of feet to the water table and (Me rain, sites in the southeast karst regions, sites in the Gulf
 Coastal Plain, and sites in the  High Plains of the Dakotas are  all represented by the same range of
 national data.  This places severe restrictions on facilities located in environmentally sound sites, and may
 allow poorly located facilities to be dellsted.  The Subcommittee thus suggests that some sort of regional
 approach be considered by the Agency. Regional ranges of data inputs for the hydrotogte parameters
could reduce the uncertainty and reduce both false positives and false negatives. The  regional approach
could use either geographic regions or hydrogeologic regions. The EPA has developed a ground water
flow assessment model, DRASTIC, based on hydrogeologic regions (Ref, 22).   A similar approach could
be considered for the current application, however, further analysis would be required to determine the
suitability of DRASTIC, or any other framework, for this purpose.  (The Subcommittee  did not review the
 DRASTIC model.)  Once  a basis for regionalization is determined, model input parameters such  as
                          4M»*
recharge, hydraulic conductivity, porosity, hydraulic gradient distributions and ground water temperature
can be selected to be  more representative of the possible range of aquifer conditions within regions.
However, as  conditions still vary greatly even within regions, the use of site-specific analysis is stilt
preferable to trie regionalization approach proposed.

-------
                                               21

      Another issue related to the consideration of different hydrogeoiogic regions is whether highly
 conservative delistrng thresholds should be required in the use of a nationwide model.  While this is in
 many respects an  issue of Agency policy,  it includes some scientific issues as well.  If all possible
 disposal sites are permitted for delisted wastes, then a nationwide demonstration must consider worst
 case conditions, or at least the minimal requirements of Subtitle 0 facilities (because the waste must be
 demonstrated safe for that lower level of waste management),  A Monte Carte analysts of the entire range
 of environmental conditions would not be appropriate because disposal in high trinsmissivity conditions,
 like  fractured rock or karst, will  certainly lead to problems ar»d should  be prohibited.  If Monte Carlo
 analysis is used with all sites considered, a very high level of facility protection is required to account for
 disposal in these extreme settings.  Certainly, an 85% threshold is unacceptable because this would allow
 a 15% failure rate and result in a substantial number of contamination sites requiring remediation,  A way
 to address  this  dilemma would  be to ban the disposal of  delisted  wastes at sites with known  high
 transmissivify conditions, such as fractured rock or karst.  This would impose some degree of regulation
 on delisted wastes, but much  less than that required for a Subtitle C waste, and the simple restriction
 should be straightforward to implement-  Once such vulnerable hydrogeoiogic settings are excluded from
 the set of potential disposal sites for delisted wastes, a decision based on Monte Carto analysis of We
 remaining sites is more likely to provide adequate protection.

      One final problem identified by the Subcommittee is that only a limited number of contaminants will
 be simulated by  the model.   In the current application, unless a waste stream contains only  those
 simulated compounds, the waste could not be delisted. Few wastes are so simply limited to a  few
 compounds- The full range of compounds in a waste must be considered  in a listing or delisttng decision,
 Once all are identified, toxicotogtaaf and other evidence could be presented to demonstrate that certain
compounds are not hazardous, perhaps at some predetermined threshold level, and these would not be
subject  to  further evaluation.   Then, only compounds which are hazardous would be subjected to
migration potential review before delisting.   Thus, all  significant constituents of the waste would be
evaluated in some manner before a delisting decision is made.  It is clear, however, that the CANSAZ
model and data base will not be ready for use until they can handle ail significant constituents that must
be quantitatively addressed in a delisting petition.

-------
APPENDIX A - THE CHARGE TO THE SUBCOMMITTEE

-------
             UNITED STATES ENVIRONMENTAL PROTECTION AGENCY
                         WASHINGTON, o.c. 20410
                              AW   d 1969

                                                S0UO
MEMORANDUM

SUBJECT: Science Advisory Board Review of th« Surface
          Impoundment Groundwat*r Code (CAHSAZV

FROM:    Devereaux Barnes, Director  ///£[     r*>
         Characterization and Aaaeasawnt Division (OS-330)

TO:      Dr. Donald Barnes, Director
         Science Adviaory Board  (A-101)

     The purpose of this memorandum is to request your review of
the Combined-numerical Saturated 2,on*  (CANSAZ) flow and transport
module for the simulation of flow   nd transport of contaminants
in the saturated zone.  The code was developed to better simulate
the migration of contaminants beneath those surface impoundment*
where hydraulic mounding occurs.  The incorporation of CANSA1
represents the major difference between the osw fate and
transport models for landfills  (EPACHL) and for surface
impoundments (EPACMS) .  Both IPACMS and IPACML are applicable to
aqueous wastes and are generally implemented on a nation-wide
basis using the Monte Carlo techniques.

SPECIFIC AREAS OF REVIEW

     The two major areas listed below are identified for SAB
review.  However, there may be other concerns about the CAKSAZ
module.  Some of these concern* may be generic to groundwater
models in general i they include the biodegradation of
contaminants, heterogeneities and fractures in the subsurface
materials, multiphase transport, and th* quality and quantity of
the data.  Th* OSW is aware of these concern* and welcomes  the
SAB's view* on th«»*  However, at this time we are particularly
interested in comments specific  to  the CANSA2 module.

     1} Assumptions trader! yiag  the  CAHSAJ

     CANSAZ was developed for possible us* in  th* development  of
regulation* in th«Hiazmrdous waste  identification progrta.   The
cod* possibly could be used in  th*  Del is ting .Program because a
larg* portion of p*tition*d wastes  ar« managed in surface
impoundnents ,  In thl* program,  th* cod* would fa* iapl*»*nted on
a nation-wid* basis uaing th* Mont* Carlo t*chniqu**,  although

-------
certain parameters related to th* dimensions of the surface
impoundment and the volume of the vast* may b* fixed based on
sita-specific conditions.

     Two important questions concerning assumptions arei  1) Are
the assumptions made in the development of the code appropriate,
considering the intended us* and the limitations of the available
data?  and 2) Which parameters should be used only as part of a
Monte Carlo Analysis and which ones could be set to site-specific
conditions?

     2) Adequacy of CAMS*]

     The code was developed to account for the effects of
mounding beneath surface impoundments on the transport of
contaminants.  Thm mounding creates a variable velocity field
which requires that both th* horizontal and the vertical
components of the velocity be ct -sidered  in si-   .ting transport
of contaminants.  A critical review question is whether the code
is adequate for simulating the transport of contaminants beneath
surface impoundments containing aqueous wastes, keeping in mind
the intended regulatory uses of the code.

     ThanX you for your help.  Please contact me, Alec McBride
(382-4761) or Dr. Zubair Saleem  (3S2-4767), if we can b* of any
assistance on this project.

Attachment

cc: Matt Straus
    Alec McBride
    or. Jack Kooyoo»jian
    Or. Zubelr Saleem

-------
                                   25
                          APPENDIX B - ACRONYMS

ASTM -      AMERICAN SOCIETY OF TESTING MATERIALS
CANSAZ -    COMBINED ANALYTICAL-NUMERICAL SATURATED
           ZONE FLOW AND TRANSPORT MODEL
OAF -       DILUTION ATTENUATION FACTOR
DO -        DISSOLVED OXYGEN
DRASTIC -   DEPTH TO GROUNOWATER, NET RECHARGE, AQUIFER MEDIA,
           SOIL MEDIA, TOPOGRAPHY, IMPACT OF VADOSE ZONE, HYDRAULIC
           CONDUCTIVITY OF AQUIFER"
EEC -       ENVIRONMENTAL ENGINEERING COMMITTEE OF THE SCIENCE ADVISORY
           BOARD
Eh -        REDOX STATE
EPA -       U.S. ENVIRONMENTAL PROTECTION AGENCY (ALSO USEPA)
EPACML -    EPA COMPOSITE MODEL FOR LANDFILLS
EPACMS -    EPA COMPOSITE MODEL FOR fURFACE IMPOUNDMENTS
EPRI -       ELECTRIC POWER RESEARCH INSTITUTE
FECTUZ -    FINITE ELEMENT CODE FOR SIMULATING FLOW AND
           TRANSPORT IN THE UNSATURATED ZONE
G -         GALERKIN FINITE ELEMENT METHOD
HELP -      HYDROLOGIC EVALUATION OF LANDFILL PERFORMANCE
IAHS -       INTERNATIONAL ASSOCIATION OF HYDROLOGICAL SCIENCES
Kd -         DISTRIBUTION COEFFICIENT
LT -         LAPLACE TRANFORM
LTG -       COMBINED LAPLACE TRANSFORM (LT) AND FINITE ELEMENT GALERKIN
           (G) METHOD
NWWA -     NATIONAL WATER-WELL ASSOCIATION
OSW -       OFFICE OF SOLID WASTE OF THE U.S. ENVIRONMENTAL PROTECTION AGENCY
pH -        NEGATIVE LOG OF HYDROGEN ION CONCENTRATION
RCRA -      RESOURCE CONSERVATION AND RECOVERY ACT
SAB -       SCIENCE ADVISORY BOARD OF THE U.S. ENVIRONMENTAL PROTECTION
           AGENCY

Subtitle D
Facilities -    NON-HAZARDOUS WASTE LAND TREATMENT, DISPOSAL OR STORAGE
           FACILITIES AS SPECIFIED BY THE RCRA LEGISLATION AND IMPLEMENTING
           REGULATIONS "
SZM -       SATURATED ZONE MODEL
SZMS -      SATURATED ZONE MODEL SUBCOMMITTEE
VHS -       VERTICAL, HORIZONTAL SATURATED ZONE MODEL

-------
                                            26


                APPENDIX C • RESOURCE MATERIAL AND REFERENCES CITED

 1)                Ambrose, Robert 8., Jr. and Thomas Q, Bamweli, Jr., "Environmental Software at •
                  U.S. Environmental Protection Agency's Center for Exposure Assessment Modelin
                  (undated reprint), 18 pages.
 2)                Anderson,  M.P., 'Hydrogeologie Fades  Modes to Delineate  Large-Scale Spa
                  Trends  in  Glacial  and  Glaciofluvial  Sediments,*  Geological  Society of Amer
                  Bulletin. 101(4). 1989. PP. 501-511.

 3)                ASTM, "Standard Practice for Evaluating Environmental Fate Models of Chemica
                  American Society for Testing and Materials (ASTM), Committee E«47 on Biotogi
                  Effects and Environmental Fate, Designation ES78-84, Philadelphia, PA. 1984.
 4)                Burnett.  R.D.  and  i.p.  Frind, "Simulation of Contaminant  Transport in Tni
                  Dimensions, 2,  Dimensionality effects," Water Resources Research. 23{4), 1987.
                  695-705,
 5)                CANSA2:   Combined  Analytical-Numerical   Code  for  Simulating  Flow  z
                  Contaminant Transport in the Saturated Zone, Prepared by E,A, Sudicky. J.B.  K
                  and RS. Huyakorn, University of Waterloo and HydroGeologic, Inc. tor the U.S. E
                  Office of Solid Waste,(ANM/l23D/FTt ID No. S9-017). October 1988.
 6)                Chapra,  S.L.  and  K.H. Reekhow, 1983,  Engineering  Approaches  far  y
                  Management, Volume 2:  Mechanistic Modeling. Ann  Arbor Science, Butterwo
                  Boston, 1983.
 7)                Carson, Alan S., U.S. EPA Office of Solid Waste Memo to Harry Tomo, Liaison to
                  Science  Advisory Board, entitled  'Response  to  the Science  Advisory  Boar
                  Comments on OSWs Development of the RCRA Hazardous Waste Identification ?
                  Land Disposal Restrictions Groundwater Screening Procedure," February 13.1S85
8)                EPACML:  Background Document lor EPA's Composite Landfill Model, Woodws
                  Clyde Consultants, 1988.
9)                EPACMS:   Composite  Model for Simulating  Leachate  Migration from  Surf;
                  Impoundments and Monte Carlo Uncertainty Analysis, Users  Guide, Prepared
                  HydroGeologic Inc. tor U.S.  EPA Office of Solid  Waste, (ANM/123D/FT, ID I
                  89-030) September 1988; Revised April 1989,
 10}               EPRI, A Review of Field Scale Physical Solute Transport Processes in Saturated «
                  Unsaturated Porous Media, Electric Power Research institute (EPRI), Palo Alto, (
                  198S.
11)               IPRl, Personal Communication, I.P, Murarka.
12)               FECTUZ: Finite Element and Semi-Analytical Coda tor Simulating One-Dimensional
                  Flow and Solute Transport in tie Unsaturateci Zone, U.S. EPA Office of Solid Waste,
                  Washington, D.C,, 1988.
13)               Freeze, R.A. and J. Cherry, Groundwater. Prentice Hall, Englewood Cliffs, N.J,. 19'
14)               Frind, E.G., E.A. Sudicky and J.W.  Molson,  Three-Dimensional Simulation
                  Organic Transport with Aerobic BiodegradattorV In:  Groundwater Contaminati
                  IAHS PubHcation No, 185,1989, pp. 89-96.
15}               Looney, Brian  B. et ai,  "Analysis  of the Validity  of Analytical  Models Used
                  Assessjpnt of FoTty-'Five Waste Site Areas:   Subsurface  Flow and  Chemi
                1  Transport,"  Pioceedings  of  the  NWWA Conference  and Exposition, on  Spiv
                  Ground  Water  Problems with Models. Vol. 2, Denver. Colorado, February 10-
                  1987, pp. 954-982.
16)               Siiman,  S.E. and  A.L. Wright, "Stochastic Analysis of Paths of High Hydra
                  Conductivity in  Porous  Media,'  Water Resources Research. 24<11). 1988.
                  1901-1910,

-------
                                            27


17)              Sudieky, E.A.,  "The Laplace Transform Gaferkin Technique:  A Continuous Time-
                 Continuous  Finite  Element  Theory  and  Application  to  Mass  Transport  in
                 Groundwater." Water Resources Research. 25(8), 1989, pp. 1833-1846.
18}              U,S. EPA,  Hazardous Waste Management  System;  Land Disposal Restrictions,
                 Proposed Rule. Federal Register. Vol. 51, No. 9, January 14,1986, pp. 1602-1766.
19)              U.S. £PA/SAB, Report of the Environmental  Engineering Committee of the Science
                 Advisory Board, "On the Review of the RCRA Hazardous Waste identification and
                 Land Disposal Restrictions Ground Water Screen Procedure," April 1985.

20)              U.S. EPA/SAB,  Report of the  Unsaturated  Zone  Code Subcommittee of  the
                 Environmental Engineering Committee of the Science Advisory Board, "Review of the
                 Office of Solid  Waste's Unsaturated Zone Code (FECTUZ) for the OSW Fate and
                 Transport Model" SAB-EEC-88-030, July 12,1i88.
21)              U.S. EPA/SAB, Report of the Environmental  Engineering Committee of the Science
                 Advisory  Board,  "Resolution  on the  Use of Mathematical  Models  by  EPA  for
                 Regulatory  Assessment  and  Decision-Making.'  EPA-SAB-EEC-89-Q12,  January
                 1989.
22)              U.S. EPA, "DRASTIC: A Standardized System for Evaluating Ground Water Pollution
                 Potential Using Hydrogeologic Settings,' EPA/6QO/2-85/Q18.1985.

-------