United States      Science Advisory Board   EPA-SAB-lAQC-94-QG9b
Environmental              July 94
Protection Agency     Washington, DC
REVIEW OF DRAFT
"ADDENDUM TO THE
METHODOLOGY FOR
ASSESSING HEALTH RISKS
ASSOCIATED WITH
INDIRECT EXPOSURE TO
COMBUSTOR EMISSIONS."

PREPARED  BY THE INDOOR AIR
QUALITY/TOTAL HUMAN
EXPOSURE  COMMITTEE OF THE
SCIENCE ADVISORY BOARD

-------
                 UNITED STATES ENVIRONMENTAL PROTECTION AGENCY
                               WASHINGTON, D.C. 20460
                                                                    OFFICE OF TH6 ADMINISTRATOR
                                                                     SCIENCE ADVISORY BOARD
EPA~SAB-IAQC-94-009b
July 29,  1994

Honorable Carol M. Browner
Administrator
U.S. Environmental Protection Agency
401 M Street, SW
Washington, DC  20460

RE:    Draft "Addendum to the Methodology for Assessing Health Risks Associated with
       Indirect Exposure to Combustor Emissions"

Dear Mrs, Browner:

       On December 3,  1993, the Indoor Air Quality/Total Human Exposure Committee (the
Committee) of the Science Advisory Board reviewed the draft document "Addendum to the
Methodology for Assessing Health Risks Associated with Indirect Exposure to Combustor
Emissions" (the Addendum), In view of pressing EPA and public concerns about
incinerators, the Committee sent you in interim letter (EPA-SAB-lAQC-94-OQ9a) on
February  15,  1994 to provide you with some of the major findings of the Committee.  The
attached is a more detailed report of our conclusions and recommendations.

       The assessment of risks  from stationary combustors entails a complex range of issues,
including the existence of many different kinds of combustion devices and  raw materials,
both direct and indirect exposure routes, concerns regarding transportation and disposal of
raw materials and combustion ash, and the need to  account for cumulative impacts of
multiple combustor sources at the regional, national and international levels.  We preface our
comments by emphasizing, however, that this report is primarily focused on the questions
surrounding indirect exposure assessment which are the subject of the Addendum,  For
lipophillic contaminants,  such as dioxins,  furans} polychlorinated biphenyls, certain
pesticides, and for metals such as lead and mercury, indirect  exposures through food have
been demonstrated to be  dominant contributors to total dose within non-occupationally
                                                                           Prints sn Qtptnn
                                                                           at ttia 7S% nicyei

-------
exposed populations.  It is also likely that atmospheric pollution from combustors and other
thermal processes significantly contribute to the ubiquitous presence of. some of the highly
persistent lipophillic compounds.

       To grapple with the complex indirect exposure pathway issues, the Agency needs to
be able to estimate the environmental fate of combustor emissions and their consequent
potential for human exposures.  This task requires the development of models to predict
accumulations of chemical contaminants in the environment and identification of the
chemicals, environmental compartments, and exposure pathways most likely to be of concern
so that appropriate actions can be taken before there is widespread and/or irreversible
damage.

       The Addendum the Committee reviewed is a critical part of the Agency's effort to
deal with these very difficult challenges.  The effort has significant merit because it is the
first attempt by the Agency to put all the models relevant to indirect exposure together in a
coherent fashion. The overall exposure model presented in the Addendum represents a major
improvement over earlier incinerator risk assessment models, with much potential practical
value.

       The model is an effort that pushes at the very edge of our current scientific
knowledge.  This is the source of its considerable merit, but it is also the source of Us most
serious limitations.  The Committee's major findings, therefore, concern the proper use, and
the potential misuses,  of the Addendum as a tool by the Agency and others.   During the
review, the Committee was left with the strong impression that the Addendum would be the
primary or even the sole tool to be used by the Agency to carry out routine evaluations of
risks from combustors. This impression caused serious concern in the Committee for two
main reasons.

       First, the evaluation of indirect exposures from single sites is too narrow a basis for
decisions regarding  stationary combustor risks.  The potential risks of combustors are
multiple, and they include direct and indirect risks.  In addition, while a single new facility
may not result in a significant risk, the cumulative effect of the addition of a facility to an
area with a number of existing combustors may well result in aggregate health risks that
reach levels of concern.  While the full scope of these issues is beyond a narrow review of
the Addendum, the Committee feels strongly that any national exposure assessment model
framework for large combustor operations must address both local (i.e., within 50 km)
impacts on human health and the environment, and also the contribution of the operations of
any single combustor to the more distant (>50 km) regional impacts.  This consideration
affects both direct and indirect exposure pathways.

-------
       The second and most directly relevant reason for concern is that the Committee has
many serious reservations about the possible routine use of the methodology in the Addendum
as a detailed quantitative exposure model for combustors.  These reservations are discussed
in some detail in the attached report, and, incidentally,  many of them have been noted by
previous SAB committees in their review of hazardous waste and domestic waste
incineration.

       The Committee's principal conclusion is that the Addendum is not ready for release
as an "EPA Methodology" for routine regulatory assessment of indirect exposures from
stationary combustors due to the substantial scientific uncertainties in the models and the
absence of information in the Addendum concerning those uncertainties and limitations.  The
major scientific concerns were as follows:

       1.      Lack of validation and reliance on default input values for many of the inter-
              media transfer factors, for many chemicals, leaving  the Committee with many
              reservations about widespread application of this document as an  EPA
              methodology for all types of combustors and chemicals.

       2.      Lack of information on incinerator upset conditions, which may contribute
              significantly to the total emissions from combustors.

       3,      Insufficient attention to the chemical nature of the emissions and the frequency
              of upset conditions for the full range of combustors addressed by the
              Addendum. In particular,  species whose  exposure potential is determined by
              multiple chemical forms and physical states, such as mercury, should be
              explicitly considered.

      4.      Requirements for the use of site-specific human exposure data which may be
              impractical and excessively costly to obtain, A  balance must be struck
             between needs for site-specific data with very strong impact on the exposure
             results and data with less impact where defaults may be applicable.

      5.     Possibility of the violation of the laws of  Conservation of Mass and of
             chemical thermodynamics in two sub-components of the methodology, that is,
             in the models, for volatilization from soil and plant surfaces and for transfer
             from air to both plants and soils. This should be checked and corrected if
             necessary.

      6.     Insufficient treatment and exposition of uncertainty and variability throughout
             the Addendum.

-------
       Therefore, the Committee makes the following recommendations:

       1.    Use the Addendum as an analytical tool to identify the chemicals most
             likely to accumulate in the environment, the environmental compartments
             most at risk of excessive accumulations, and the exposure pathways most
             likely to lead to aggregate risks of concern.  Such analyses will provide
             strategic guidance in utilizing environmental sampling to obtain actual data on
             indirect exposure to humans and to ecosystems. We do not recommend
             release of the Addendum as an  "EPA Methodology" for routine,  quantitative,
             site-specific risk assessments  for incinerators and other combustors,

       2.    Develop and implement a strategic plan to collect critical input data for
             the models and to validate the methodology. For this purpose, take
             advantage of the re-permitting process to collect relevant data to  help validate
             the Addendum's  models,

       3.    Establish a framework to ensure that the entire range of potential risks
             from stationary combustors are addressed holistieally. This must include
             both direct and indirect risks, as well as local, regional, national  and
             international concerns.

       In closing, we wish to emphasize that the Committee is keenly aware of the
difficulties inherent in the "state of the science"  nature of the work which the Addendum
effort entails, especially when the work must be done under the combined pressures of
severely limited resources and public demands for "something" to  be done quickly.

       We urge you to implement these recommendations and would be happy to review a
revised Addendum in the future.
                                       Sincerely,
  r    enevieve M. Matanoski, Chair
Executive Committee
Science Advisory Board
                                                             u*u
                                               r. Joan M. DaiseyyChair
                                             Indoor Air Quality/Total
                                              Human Exposure Committee
                                             Science Advisory Board
                                         4

-------
                                   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 balanced, expert assessment of
scientific matters related to problems facing the Agency,  This report has not been
reviewed for approval by the Agency and, hence, the contents of this report  do not
necessarily represent the views  and policies of the Environmental Protection
Agency, nor of other agencies in the  Executive Branch of the Federal government,
nor does mention of trade names or  commercial products constitute a
recommendation for use.

-------
                                 ABSTRACT

      On December 3, 1993, the Indoor Air Quality/Total Human Exposure
Committee  (the Committee) of the Science Advisory Board reviewed the draft
document "Addendum to the Methodology for Assessing Health Risks Associated
with Indirect Exposure to Combustor Emissions" (the Addendum),

      Although the multi-media model of the Addendum is not yet fully developed,
the Committee found merit in the model and recommended its use as an analytical
tool to identify the chemicals most likely to  accumulate in the environment, the
environmental compartments most at risk of unacceptable accumulations, and the
exposure pathways and chemicals most likely to result in aggregate health risks
that reach levels of concern. Such analyses will provide  strategic guidance for
environmental sampling to obtain data on indirect exposure to humans and to
ecosystems.  However, they did not recommend the release of the Addendum as an
"EPA Methodology" for routine, quantitative, site-specific risk assessments for
incinerators because of substantial scientific uncertainties in the model and the
absence of many important model parameters.

      In addition to their general findings regarding the use  and possible misuse
of the methodology in the Addendum, the Committee addressed numerous specific
issues concerning:  1) air emissions and modeling; 2) soil impacts and the food
chain; 3) water impacts and modeling; and 4) exposure.  Finally, the Committee
stressed the need to establish a framework to ensure that the entire  range of
potential risks from stationary combustors are  addressed holistically, including
both direct and indirect risks, as well as local,  regional, national and international
concerns.
                                      11

-------
                ENVIRONMENTAL PROTECTION AGENCY
                       SCIENCE ADVISORY BOARD
                          INDOOR AIR QUALITY/
                 TOTAL HUMAN EXPOSURE COMMITTEE

                   Review of Indirect Exposure Addendum
Chair

Dr. Joan Daisey, Indoor Environment Program, Lawrence Berkeley Laboratory,
Berkeley, CA

Members

Dr. Paul Bailey, Mobil, Environmental Health and Sciences Laboratory, Princeton,
NJ.

Dr. Robert Hazen, State of New Jersey, Department of Environmental Protection
and Energy, Trenton, NJ.

Dr. Timothy Larson, Environmental Science and Engineering Program,
Department of Civil Engineering, University of Washington, Seattle, WA.

Dr. Brian Leaderer, John B, Pierce Foundation Laboratory, and Division of
Environmental Health Sciences, Department of Epidemiology and Public Health,
Yale School of Medicine, New Haven,  CT.

Dr. Paul Lioy, Department of Environmental and Occupational Health Sciences
and Community Medicine, Eobert Wood Johnson School of Medicine, Piscataway,
NJ,

Dr. Morton Lippmann, Institute of Environmental Medicine, New York University
Medical  Center, Tuxedo, NY,

Dr. Maria Morandi, University of Texas Health Science Center at Houston, School
of Public Health, Houston, TX. (absent from meeting),

Mr. Roger Morse, ENTEK, Environmental and Technical Services, Troy, NY
12180. (absent from meeting)
                                   in

-------
Dr. Jonathan M. Samet, The University of New Mexico School of Medicine, New
Mexico Tumor Registry, Albuquerque, NM. (absent from meeting)

Mr. Ron WHte, American Lung Association, Washington, DC.

Dr, Jo Ann Lighty, University of Utah, Department of Chemical and Fuels
Engineering, Salt Lake City, UT, (liaison from Environmental Engineering
Committee),

Dr. Ann B. McElroy, New York Sea Grant Institute, State University of New
York, Stoney Brook, NY, (liaison from Ecological Processes and Effects
Committee).

Federal Liaison^

Dr. Rayford P. Hosker, Jr.,  National Oceanic and Atmospheric Administration,
ATDD, Atlanta, GA.

Dr. Maureen Lichtveld, M.D., Agency  for Toxic Substances and Disease Registry,
Office of the Assistant Administrator, Atlanta, GA.

Invited Partjcipantg

Thomas McKone, Pb~D., Lawrence Livermore Laboratory, Livermore, CA,

Mr. Thomas Webster, Boston University School of Public Health, Boston, MA.

Science Advisory Board Staff

Mr. Manuel R. Gomez, Designated Federal Official, Science Advisory Board, U,S,
EPA, Washington, DC

Mrs. Dorothy Clark, Secretary, Science Advisory Board, U.S. EPA,
      Washington, DC
                                     IV

-------
                         TABLE OF CONTENTS
 1,    EXECUTIVE SUMMARY  	,	,	. ,   1
      1.1   General Findings	,	   1
      1.2   Air Emissions and Modeling Issues  . , ,	, ,	   3
      1.3   Soil Impact and Food Chain Issues  	.	   4
      1.4   Water Impact Modeling Issues	,	 .   4
      1.5   Exposure Issues ,,...,,,,.,	,	,	   4

 2.    BACKGROUND AND CHARGE	,	,	   6

 3.    FINDINGS  .	,  , ,	,	  . , , ,	   7
      3.1   Introduction.  . .	,	,	   7
      3,2   General Findings	.,.....,   8
      3.3   Responses to  Charge to the Committee  .  ,	.,.,,,   14
           3.3.1  General Issues	   14
           3.3.2  Air Emissions and Modeling Issues	,	   16
           3.3.3  Soil Impact  and Food Chain Issues	   23
           3.3.4  Water Impact Modeling Issues ,	   29
           3.3.5  Exposure Issues	   30
      3,4   Summary of Major Recommendations	   37

APPENDIX A.     Charge to the Committee  	   38

APPENDIX B,     Interim Letter to Administrator (EPA^SAB-IAQC-94-
                 009a)	 ,	,		   39

APPENDIX C     Miscellaneous Comments and Specific Corrections and
                 Suggestions  for Clarification of Text	   40

APPENDIX D     Earlier  SAB Reports Concerning Hazardous and
                 Domestic Waste Incineration,	   47

-------
1.    EXECUTIVE SUMMARY   ,

1.1   General Findings

      The assessment of risks from stationary eombustors entails a complex range
of issues, including the existence of many different kinds of combustion devices
and raw materials, both direct and indirect exposure routes, concerns regarding
transportation and disposal of raw materials and combustion ash, and the need to
account for cumulative impacts of multiple combustor sources at the regional,
national and international levels.  This report, however, is primarily focused on
the questions surrounding indirect exposure assessment which are the subject of
the Addendum. Indirect exposures are those that occur after transfer of airborne
contaminants into water, soil and the food chain.

      To grapple with indirect exposure pathway issues, the Agency must be able
to develop and validate models to estimate accumulations of chemical contaminants
in the environment and specifically to identify the chemicals, environmental
compartments, and exposure pathways most likely to be of concern so that
appropriate actions can be taken before there is widespread and/or irreversible
damage.  The Addendum is a critical part of the Agency's effort to deal with these
very difficult challenges. The effort has significant merit because it is the first
attempt by the Agency to put all the models relevant to indirect exposure together
in a coherent fashion,

      The Committee's principal conclusion is that the Addendum is  not ready for
release as an "EPA Methodology" for routine regulatory assessment of indirect
exposures from stationary combustors due to the substantial scientific uncertainties
in the models and the absence of information in the Addendum concerning those
uncertainties and limitations.  The major scientific concerns were as follows:

      1.    Lack of validation and reliance on default input values for many
            intermedia transfer factors for many chemicals, leaving the Committee
            with many reservations about widespread application of this document
            as an EPA methodology for all types of combustors and  chemicals.

      2,     Lack of information on incinerator upset conditions, which may
            contribute significantly to the total emissions from combustors.

-------
       3,     Insufficient attention to the chemical nature of the emissions and the
             frequency of upset conditions for the full range of combustors
             addressed by the Addendum. In particular, species whose exposure
             potential is determined by multiple chemical forms and physical
             states, such as mercury s should be explicitly considered.

       4.     Requirements for the use of site-specific exposure data which may be
             impractical and excessively costly to obtain. A balance must be struck
             between needs for site-specific data with very strong impact on the
             exposure results and data with less impact where defaults may be
             applicable.

       5.     Possbility of violation of the Law of Conservation  of Mass and the
             laws of chemical thermodynamics in two sub-components of the
             methodology, that is,  in the models for volatilization from soil and
             plant surfaces and for transfer from air to  both plants and soils.

       6.     Insufficient treatment and exposition of uncertainty and variability
             throughout the Addendum.

       The Committee found, however, that the Addendum has  significant  practical
potential value.  They recommended the  use of the Addendum as  an analytical too!
to identify the chemicals from combustors that are most likely to  accumulate in
the environment, the environmental compartments most at risk of unacceptable
accumulations, and the  exposure pathways most likely to result in aggregate
health risks that reach  levels of concern, Such analyses will provide strategic
guidance in designing environmental sampling to obtain actual data on indirect
exposure to humans and to ecosystems,

      In addition,  the Committee stressed that the evaluation of indirect
exposures from single sites is too narrow a basis for decisions regarding stationary
combustor risks.  The potential risks of combustors include direct and indirect
risks from single and multiple facilities.  While a single  new facility may not pose
a significant  risk, the cumulative effect of the  addition of a facility to an area with
a number of existing combustors may  well result in aggregate health risks that
reach levels of concern.   The Committee  feels strongly that any national model
framework to guide exposure assessments for large combustor operations must
address both local  (i.e., within 50 km) impacts on human health and the

-------
environment, and also the contribution of the operations of any single combustor
to the more distant (>50 km) regional impacts.

       Finally, if the Addendum is released for use as an analytical tool, the
Committee recommended the addition of an integrating introductory chapter to
provide additional information which would be necessary to conduct an indirect
exposure and risk assessment with the Addendum.  This chapter should include  a
decision tree to assist the user in deciding what models to use and when, guidance
regarding how good the models are for which chemicals and which
situations/eombustor types, and some additional  screening procedures to help avoid
spending time modeling exposures for pathway-chemical combinations that  are
likely to lead to extremely low exposures and risks,

1.2   Air Emissions and Modeling Issues

      The chemicals identified as of concern from combustor emissions would be
more usefully presented if grouped by chemical class and combustor type and/or
fuel or waste composition.  The grouping by chemical class will provide a more
rational framework for both modeling and model validation.

      The hierarchical approach for estimating emissions needs some clarification,
and the role of upset conditions on total emissions needs to be more carefully
defined. Additional measured data describing the emission factors during start-
ups, shut-downs, upsets and poor operation  are necessary.  The re-permitting
process for incinerators offers a unique opportunity to obtain such, data and the
Agency should take advantage of it.

      The net effect of using the recommended procedure to estimate the vapor
phase/particle phase partitioning of semivolatile organic compounds is to
underestimate wet deposition near the source and to overestimate it  further from
the source.  The Addendum could benefit from incorporating more of the recent
literature  on the issue  of adsorption of organics.

      Some clarification  is needed regarding which particles are being addressed
in the particle size distribution of particulate matter from combustors, combustor
particle emissions or outdoor aerosols upon which combustor vapor emissions
condense.  The reasons for selecting a default distribution need to be more clearly
stated.

-------
       There are shortcomings in the basic transport and dispersion model used in
 the Addendum—related to calms and terrain-forced changes in wind direction.
 Explicit guidelines are therefore needed to avoid misuse  of the model under certain
 circumstances.  Finally, certain modifications of the COMDEP model  (based on the
 Shulman  and Shire model) are needed to correct underpredictions of  wake
 concentrations for short stacks.

       Finally, the atmospheric lifetime and environmental fate of different forms
 of mercury are among the species of most concern and should receive increased
 attention.

 1.3    Soil Impact and Pood Chain Issues

       It is very difficult to make judgements concerning the recommendations to
 account for vapor phase  impacts to soils  because there is such a paucity of
 environmental measurements available to validate any approaches.

       The assumptions for mixing soil depths for tillage and non-tillage situations
 seemed reasonable. The 30% assumption for the percent of contaminants in wet
 deposition that is retained on plant surfaces is reasonable.  The concentrations
 resulting from the re-suspension from soils of emisiion particles should not be
 extensively modeled if they are orders of magnitude smaller than concentrations
 resulting from stack emissions.

       One of the most significant issues  in the methodology that can and should
 be addressed is the inherent lack of reliability associated with both measured data
 and models used to determine inter-media transfer factors (ITFs), which are
 factors that define the partitioning- and mass transfer between environmental
 media.

 1.4    Water  Impact Modeling Issues

      A tiered approach  for the evaluation of aquatic impacts is the more sensible
approach than the use of alternatives that involve more complex models and
 numerous inputs, many of which are not available,  Consideration should be given
to the  impacts of multiple combustors impacting coastal regions, where much of
the U.S. population resides.

 1.5    Exposure Issues

-------
      The use of geographically-defined boundaries based on dispersion/deposition
models is reasonable for estimating concentrations in soil, water, and air, and
direct exposures to these media.  However, it may not be sufficient for defining
the population at risk from indirect exposure.  The recommendation to develop
distributions of exposures is a substantial improvement over the use of worst-case
scenarios and is commended.

      The use of Monte Carlo simulations to estimate the  effects of uncertainty is
not sufficiently clear because the distinction between variability and uncertainty is
blurred.  The guidance for the use of Monte  Carlo analysis is not sufficiently
detailed.

      The choice of a model for estimating breast milk concentrations on the basis
of maternal intake requires validation by comparison with  actual data. For TCDD,
the Smith approach appears preferable to the Travis et al.  model.

-------
2.    BACKGROUND AND CHARGE

      On December 3, 1993, the Indoor Air Quality/Total Human Exposure
Committee (the Committee) of the Science Advisory Board reviewed the draft
document "Addendum to the Methodology for Assessing Health Risks Associated
with Indirect Exposure to Combustor Emissions" (the Addendum).  The charge for
this review can be found in Appendix A.  In view of pressing EPA and public
concerns about incinerators, the Committee prepared an interim letter to the
Administrator to  provide her with preliminary information on some of the major
findings of the Committee (EPA-SAB-IAQC-94-QQ9a). The interim letter can be
found in Appendix B. This report is the Committee's detailed review of the
Addendum.

-------
3.    FINDINGS

3.1   Introduction.

      The assessment of risks from stationary combustors, henceforth called
combustors for simplicity,  entails t complex range of issues.  A complete risk
assessment for an individual combustor must, in principle, include direct exposures
from both combustion emissions and from accidental releases of fuels (including
hazardous waste) and waste ash during transportation and handling at the site as
well as disposal of the waste ash.  The population density in the area of such a
facility, as well as the presence and contributions of other stationary and mobile
combustion sources must also be considered.  It is thus important to emphasize at
the outset that this report primarily addresses only one  of these aspects, namely
the questions surrounding indirect exposure assessment  which are the subject of
the Addendum.

      Indirect exposures from combustors are those that occur via a transfer of
airborne  contaminants from a  combustor into water, soil, and the food chain.  For
lipophilic contaminants, such as dioxins, furans, polychlorinated biphenylSj
pesticides, and for metals  such as lead and mercury, indirect exposures through
food have been demonstrated to be dominant contributors to total dose within
non-occupationally exposed populations. It is also likely that atmospheric pollution
from combustors  and other thermal processes significantly contribute to the
ubiquitous presence of some of the highly persistent lipopfaillic compounds.

      To estimate indirect exposures to the population,  the Agency needs to be
able to estimate the  environmental fate of combustor emissions and their
consequent potential for human exposures. This effort requires the development
of models to predict  accumulations of chemical contaminants in environmental
media and identification of the chemicals,  environmental compartments, and
exposure pathways most likely to be of concern, so that  appropriate actions can be
taken before there is widespread and/or irreversible damage.  The Addendum the
Committee reviewed is a critical part of the Agency's effort to deal with the
difficult challenges of indirect exposure pathways.

      The remainder of this Chapter describes the Committee's general fmdinp
(Section 3,2), the  detailed  responses to the questions raised in the  Charge for  the
review (Section 3.3), and a summary of the Committee's major recommendations
(Section 3,4).  Appendix A contains the charge for the review.  Appendix B

-------
 contains a copy of the earlier interim letter to the Administrator,  Appendix C
 contains miscellaneous comments and specific corrections and recommendations for
 clarification of text in the Addendum.  Appendix D lists earlier reports by the SAB
 dealing with hazardous and domestic waste incineration.

 3.2    General Findiop

       In general, the Committee found that the multi-media model of the
 Addendum has merit and practical potential value because  it is the first attempt
 by the Agency to put all the models relevant to  indirect exposure together in a
 coherent fashion.  The overall exposure model presented in the Addendum is
 comprehensive and includes most plausible exposure pathways, with a few
 exceptions that are discussed later in this report.  The integration of surface
 water, soil, and atmosphere in the proposed water impact model is well done and
 represents a  major  improvement over earlier incinerator  risk assessment models.
 The inclusion of the net effect of gaseous diffusion from  air to water and
 volatilization from water to air increases the credibility of the model and should
 improve its reliability.

      The Committee is also very aware of the difficulties  inherent in the "state of
 the science" nature of the work  which the Addendum effort entails, especially
 when the  work must be done tinder the combined  pressures of severely limited
 resources  and public demands for "something" to be done quickly.  The fact that
 the Addendum is working at the edge  of our scientific knowledge is the source of
 its considerable merit, but it is also the source of its most serious  limitations.  The
 Committee's major  findings, therefore, concern the proper use, and the potential
 misuses, of the Addendum as a tool by the Agency and others.

      Our principal conclusion is that the Addendum is  not ready for release as an
 "EPA Methodology" for routine, quantitative, site-specific risk assessments for
combustors.  We do not believe that the Addendum can be used for calculating
absolute values of risks from indirect exposures that can  be used with confidence.
 However,  as described below, we also concluded that the Addendum, with a
paradigm  shift, could be very valuable  to the Agency and to others.

      During the review, the Committee was left with the strong impression that
the Addendum would be the primary or even the sole tool to be used by the
Agency to carry out routine, site-specific evaluations of risks from  combustors,
This impression caused serious concern in the Committee for two main reasons,

                                      8

-------
First, the Committee has many reservations about the possible routine use of this
specific methodology as a detailed quantitative exposure model for combustors
because of its early stage of development.  Secondly, in general terms, the
evaluation of indirect  exposures from single sites is too narrow a basis for
decisions regarding combustor risks.  These concerns, incidentally, have been noted
by previous SAB committees in their review of hazardous waste and domestic
waste incineration.

      With regard to  the first concern, which is the principal subject of this
report, the major reservations concerning the Addendum as a tool for  quantitative
indirect exposure modeling for combustors are as follows:

      a)    Lack of Validation and Reliance on Default Input Values,  There is a
            general lack of measured data to estimate input parameters and very
            little validation of the exposure models.  The general reliance on
            default input data and the large number of assumptions left the
            Committee with many reservations about widespread application of
            this document as a universal EPA methodology for all types of
            combustors and chemicals.  Sensitivity analysis of the model would
            help to identify which input parameters and exposure pathways are
            likely to be most important. For some of the indirect pathways, the
            exposures are likely to be trivial For example, how large must
            emissions from an incinerator be before surface waters are sufficiently
            contaminated to be of concern with, respect to  exposure due to eating
            contaminated fish?  And what is the time scale?

                  The Committee recommends that the Agency develop and
            implement a strategic plan to collect the most  critical input data for
            the model and to validate the methodology.  Sensitivity analysis of
            the model should be used to help identify the most important data
            gaps for various classes of chemicals and to develop a strategic
            research plan. The results of such an analysis should also be
            compared to experimental measurements wherever possible, as a
            reality check.  This application of the model may also alert EPA to
            potential accumulations of certain chemicals in key environmental
            compartments.

                  There is a related need to have a feedback loop built into the
            methodology for indirect and direct exposure assessment,  including

-------
      emissions measurements and a second found of modeling after a
      facility has been built.  Other important measurements would include
      contaminant deposition and air measurements.  These measurements
      should be conducted for the compounds of greatest concern, and used
      to validate the initial predictive modeling results.

            In this context, we note a  short news Item in the Journal of
      the Air and Waste Management Association^. Vol 42, No, 7, p. 950,
      1992, reporting a study done by  the Texas Air Control Board (TACB)
      in Midlothian, Texas, site of the  highest concentration of cement
      plants burning hazardous waste  in the U.S.  The TACB conducted an
      extensive environmental monitoring study involving 955 samples of
      air, soil, water and other materials, Samples  were analyzed for about
      40 contaminants.  All levels were below state  thresholds of potential
      health concern.   EPA should investigate the possibility of using these
      available date in a model validation exercise.

b)    Lack of Information on Incinerator Upset Conditions. The Committee
      is concerned about the lack of information on the frequency and
      impact of upset conditions (i.e., upsets in the  combustion process that
      result in incomplete combustion), as well as breakdowns of control
      devices such as precipitators, on  both the chemical composition and
      the quantity of emissions from incinerators. The model relies  on
      measured data from four California incinerators. These limited data
      suggest that upset conditions may contribute significantly to the total
      annual emissions from an incinerator. However, the frequency of
      upset conditions has not been determined for  sufficient numbers and
      types of incinerators to be reasonably confident of the adequacy of the
      default values recommended for general use.  The Committee noted
      that the EPA's  re-permitting process for incinerators offers a unique
      opportunity to obtain existing data on the frequency and duration of
      upset conditions for various types of incinerators in the U.S.  Other
      useful data may be available that could be required  as part of the
      re-permitting process.  In addition, some incineration studies and
      emission data sets already exist in other countries. It is strongly
      recommended that the Agency compile and review these previous
      efforts as a way of focusing its future directions.
                                10

-------
            A related issue is the definition of an "upset condition," Is this
      a failure to meet carbon monoxide (CO) permit conditions?  A recent
      article by Dempsey and Oppelt (J. Air Waste Manage. Assoc., 43:25,
      1993) notes that detection of process failure is an important research
      need.

c)    Multiple Combustor Sources.  The Committee is concerned that the
      Addendum does not require the risk assessor to account for the
      impact of multiple combustor sources. While a single new facility
      may not result in a significant risk, the cumulative effect of the
      addition of a facility to an area with  a number of existing combustors
      may well result in aggregate health risks that reach levels of concern.
      Furthermore, in many areas dominated by mobile combustion
      processes,  stationary combustor sources may be a relatively  minor
      source of total emissions.  There is a need for a more regional and
      multi-source approach to evaluating risks from indirect  exposures
      from combustion sources.  There are  also concerns that the impact
      beyond 50 km (the maximum distance considered in the Addendum)
      may be of concern for areas with significant upwind sources.

d)    Combustors Addressed  by the Addendum.  Although the Addendum
      nominally  addresses all stationary combustors (incinerator, fossil fuel,
      etc.), the document as now written appears to place more emphasis
      on incinerators and does not adequately address  all combustors.  It is
      known that the chemical nature of the emissions and the frequency of
      upset conditions will differ substantially among various  types of
      combustors but the document does not reflect this body of
      information,   For example, there is a substantial body of information
      on emissions  from coal combustors that should be referenced if the
      document is  to be all inclusive.

e)    Use of Site-Specific Data.  Throughout the document thare are
      repeated statements that site-specific data should be used whenever
      possible.  In reality, these recommendations may seldom be  followed
      because of the costs inherent in obtaining site specific information,
      particularly emission rates and contaminant concentrations  in media.
      There is a need to consider if there should be a requirement (instead
      of recommendation or preference) for site-specific data only for
      variables which could have a very strong impact  on the exposure

                               11

-------
      results (e.g., atmospheric conditions, emission data, soil
      characteristics) as compared to those which may not be as crucial and
      where defaults may be applicable. Obviously, such recommendations
      would require that the Agency perform  validation studies and
      sensitivity analysis prior to releasing thtt .Addendum.  At a minimum,
      a set of criteria should be provided to indicate tor which variables,
      parameters, and conditions site-specific data are indispensable.  A
      balance must be struck between needs for site-Apeciflc data with very
      strong impact on the exposure results and  data with less impact
      where defaults may be applicable.

f)    Conservation of Mass. Another issue that  should be addressed is the
      potential violation of the law of conservation of mass and the laws of
      chemical thermodynamics for some components of the model.  In
      general, the proposed transport models are mass conserving.
      However, it wag noted that contaminants are allowed to volatilize
      from soil and plant surfaces, but it is not clear that the lost chemical
      is then treated as an input to  the atmospheric transport model, thus
      violating conservation of mass. With regard to chemical
      thermodynamics,  it should be recognized that soils, surface waters,
      sediments, and {in particular)  plant tissues are all secondary  to the
      atmospheric source of contaminants and, thus, are not likely  to have
      an annual average chemical potential (that is, "fugaeity") that exceeds
      the chemical mass potential in the atmosphere.  Because of the way
      contaminants are allowed to transfer from  air to both plants  and soils
      by a combination of deposition and diffusion and be lost by a removal
      process that is  defined by a rate constant that is  not clearly linked to
      the mass potential in these compartments,  there is a possibility that
      these compartments  could become chemical traps that receive
      chemicals but do  not exchange them in a way that maintains chemical
      equilibrium. This problem is easy to fix by placing some limits on
      the chemical potential of any compartment relative to the air. The
      Committee notes  that the model for the surface water is carefully
      constructed to maintain both conservation of mass and comply with
      rules of chemical  thermodynamics.  This care should be extended to
      the soil and plants compartments.

g)    Uncertainty and Variability.  The treatment of uncertainty and the
      distinction between uncertainty and population and environmental

                               12

-------
            variability should be integrated as much as possible into all aspects of
            the method and explicitly presented.  Numbers that have large
            statistical variances should not be reported as single values,
            Regression equations for biotransfer factors should not be presented
            without including the standard error of the estimator in these
            equations. The user(s) should be provided with some very explicit
            guidance on chemicals and compartments for which there are large
            uncertainties and/or insufficient information to even estimate
            uncertainties.  In addition, the results of any risk assessment
            involving this model should always include estimates of the associated
            uncertainties.

      The Committee's second major concern involves the need to evaluate  the
full range of potentially important risks associated with combustors, including
indirect exposures, as part, of any comprehensive regulatory strategy,   A full
discussion of this topic is outside the scope of this report, but the Committee
wishes to emphasize one aspect because of its critical importance.  Any national
exposure assessment model  framework for large combustor operations must
address both local (Le,, within 50 km) impacts on human health and the
environment, and also the contribution of the operations of the combustor to the
more distant (>50 km) regional impacts.  This consideration affects both direct
and indirect exposure pathways.

      For local permitting purposes, it is both practical and necessary to measure
and/or model exposures within a local area (e.g., 50 km).  However, for the
purpose of protecting the public health and welfare it is also necessary for EPA
and the Nation to consider the cumulative impacts of combustor emissions on
exposure on a regional, national and international scale. The Addendum does not
address the latter issue at all, and it may not be feasible or desirable to  revise it
to do so.  On the other hand, it is essential that  EPA's overall risk assessment for
combustors address the contributions to both direct and indirect exposures of
effluents undergoing longer-range transport, especially of longer-lived trace metals
and air toxics  in fine particles and vapors. Thus the cumulative exposures of
people and ecosystems to combustor effluents need to be examined on both a local
and national scale and applied in an integrated risk assessment framework in
order to protect the public health and welfare in a responsible and  cost-effective
manner.
                                      13

-------
      In summary, for the reasons described above, the Committee does not
support the release of this Addendum as an "EPA Methodology" for routine,
quantitative, site-specific risk assessments for combustors.  We do not believe that
the Addendum can be used for calculating absolute values of risks from indirect
exposures that can be used with confidence.  The Committee agrees, however, that
knowledge of the important pathways underlying "indirect risks" from combustors
is  important, and we concluded that the methodology in the Addendum, with a
paradigm shift, may have considerable value for analytical rather than site-specific
regulatory purposes.  That is, the methodology may be valuable for its ability to
identify chemicals, environmental compartments and exposure pathways most
likely to be significantly impacted by combustor sources and therefore to  provide
strategic guidance in utilizing environmental sampling to obtain actual data on
indirect exposure to humans and to ecosystems.

      Specifically, the Committee strongly recommends that EPA consider using
the indirect exposure model to identify  the environmental compartments  in which
particular contaminants are likely to accumulate and reach levels of concern.  This
would provide the Agency with a strategic basis for designing monitoring plans for
these compartments in particular regions.  The targetted monitoring data, in turn,
would provide a means of validating and improving the indirect exposure model, as
well as early warning of any accumulations of hazardous contaminants in the
environment which may be of concern.  Such strategic monitoring might  be co-
funded by a regional consortium of stationary combustor companies and operated
with EPA oversight.

3,3   Responses to Charge to the Committee

      The following sections (3,3,1  to 3.3.5) address the questions (highlighted in
bold) posed by the Charge to the Committee, which is  reproduced in Appendix A.

3,3.1  General Issues

      The EPA Working Group recognizes the awkwardness of needing two
reference documents to determine procedures for conducting indirect exposure
assessments, and has recommended that the 1990 IED [Indirect Exposure
Document] be updated with this addendum as a single methodology.   In the
interim, is the presentation of material  in the Addendum clear? And  can the SAB
make specific recommendations for further clarity and completeness?
                                     14

-------
      At the outset, and consistent with the emphasis in the earlier section and
throughout this document, the Committee is of the opinion that the Addendum
should be used for analysis rather than prediction, because the model has not been
validated and lacks information about many important parameters.  If the
document is to be released in any official fashion for such uses, therefore, the
Committee strongly recommends a revision to make this limitation unequivocally
clear to the user(s).

      For use in this manner, however, the Addendum is not easy to read or
follow nor does it provide the kind of decision-making information necessary to
conduct an indirect exposure and risk assessment for combustors in an efficient
and effective manner.  An integrating introductory chapter should be provided for
the reader for this purpose and to place indirect exposure in perspective with
direct routes of exposure,  A clear definition of indirect exposure is also needed in
this chapter, with some reference to EPA documents that are to be used for direct
exposure assessment. In the draft Addendum, there is no overall guidance or
decision tree to assist one in deciding what models to use  and when. This
introductory chapter should  provide some guidance on this and  on how useful the
models are for which chemicals and which situations/combustor types.  There may
be more confidence in their  use, for example, for dioxins, than for many other
classes.  Further, because the IED and the Addendum are to be applied to
facilities other than municipal incinerators, a "Practical Decision Tree" is required
to define the boundary conditions, complete parameterization, and to identify the
chemicals of greatest concern.  A decision tree could also provide some  indication
of the importance of various exposure pathways for key  chemicals and scenarios.
There is also a need to provide some guidance on the sensitivity of the results to
important input parameters  and on the overall "errors" which can be expected.

      The EPA Working Group recognizes that implementation of the full
methodology is very resource intensive and that screening procedures are needed
to narrow the scope and level of detail where appropriate.  The current draft does
provide some general guidance on narrowing the list of compounds of concern.
However, in the current form, most decisions about how to establish screening
procedures would need to be made within the Agency Programs as part of their
implementation guidance. Does the SAB believe this is appropriate! or should
further screening procedures be included here?  If so can such procedures be
recommended?
                                      15

-------
       Since the integrated model has many components and many input values
 (often default), some additional screening procedures would be very useful to avoid
 spending time modeling exposures for pathway-chemical combinations that are
 likely to lead to extremely low exposures and risks.  Addition of a Decision Tree
 which includes some guidance based  on sensitivity analysis would be very helpful
 If the decisions  on screening are expected to be made at the Program level, the
 Addendum  should state this explicitly.  Furthermore, it should be noted that such
 an approach could lead to inconsistencies across Programs,

 3.3,2  Air Emissions and Modeling Issues

       Tables  3-1 to 3-3 present a listing of chemicals that have been measured in
 the stack gas emissions of a variety of combustion, sources and that may be subject
 to direct/indirect exposure/risk assessment.  Is this sufficient in terms of an initial
 listing and  a first step  in the process?

       Tables  3-1 to 3-3 present over 90 chemicals which have been identified in
 combustor emissions. For incinerators, the list  is probably too long and too
 detailed; for addressing combustors in general, it is probably not sufficiently
 inclusive. It would be more useful to present these data grouped by chemical class
 and combustor type and/or fuel  or waste composition, rather than a single listing,
 This might  eliminate looking at all chemicals  for each type of combustor or feed
 composition*  For example, the specific metal  emissions from coal combustors are
 probably quite different from those from hazardous waste incinerators.  The
 grouping of chemicals by chemical class will provide  a more rational framework for
 both modeling and model validation, i.e., default parameters for chemicals within a
 class can often be approximated based on known parameters for  one or two
 chemicals within the class.  The criteria for listing compounds for various
 combustor types should also be explicitly presented,  The issue of relative amounts
 of compounds also needs to  be addressed on some common basis , e.g., BTUs
 generated, as part of the criteria for listing compounds.

      For indirect exposures, it  is also essential to consider the chemical stability
 of the chemicals in the  environment.  For example, 1,3-butadiene emitted from an
 incinerator into the atmosphere  will be very short-lived and, therefore, is not likely
to be significant  with respect to indirect exposures.  It is suggested that  the
 lifetimes of the chemicals in, the list be examined and short-lived ones eliminated
 from consideration for indirect exposure assessments.
                                      16

-------
      With respect to missing compounds, there are some concerns about omission
 of trace species such as vanadium, from the lists, particularly if the Addendum is
 to apply to  all types of combustors.  How about nitrogenous organic compounds?
 Are the waste streams for incinerators ever high in nitrogen? If so, amines and
 N-nitrosamines, many of which are carcinogens, may be emitted. Also, cobalt and
 beryllium are cited as being emitted by sewage sludge incinerators on page 3-12,
 but are not included in Table 3-3.

      Finally, one of the species  of most concern is mercury,  The atmospheric
 lifetime and environmental fate of mercury are determined by the chemical form
 (i.e., water soluble Hg   or insoluble Hg(O)) and the physical state (i.e., gas or
 particulate).  Unfortunately, data of this kind are sparse and often facility-specific.
 Consequently, measurements to address this critical issue will be required.

      A general procedure is presented that, if applied, on a site-specific basis,
 could reduce the munber and kinds of contaminants subject to review.  Is the
 procedure adequate for this purpose, or should the Agency continue to develop a
 more defied/refined screening procedure? If a detailed procedure is
 recommended, can you provide guidance or suggestions as to a screening method
 that would be appropriate?

      The Committee recommends that the Addendum provide lists of chemicals
 grouped by chemical class and by combustor type, with chemical selection criteria
 explicitly stated, e«g,, toxieity,  concentrations in emissions, likelihood of significant
 exposures, lifetime in the environment, and potential for bioaceumulation.  The
 recommendation would  then be to use the sub-set of chemicals by combustor type
 on this list PLUS any new ones that meet these same criteria.

      The Committee also has concerns about the recommendation to eliminate
 from consideration those compounds "which lack verified dose-response
relationships for specific health endpoints."  In fact, this phrase  kcks any criteria
 for  "verified dose-response  relationships,"  The proposal for ranking is also
problematic.  What is the absolute framework in which the ranking would be
accomplished?  How are different types  of health outcomes (e.g., cancer,
reproductive effects) to be comparatively ranked? There are probably other sources
of dose-response information other than RfD and RfC data in, the IRIS  that meet
criteria that define a verified dose-response relationship.  If not, then at a
minimum, some uncertainty factor for risk assessment should be developed.

-------
      A hierarchical approach for estimating emission rates of contaminants is
presented. Is this procedure adequate in terms of providing estimates of average
emission estimates. A more difficult issue is the magnitude, frequency and
duration of increased emissions that may occur during temporary upset conditions,
during start-up and shut-down procedures and during emergency events.  The
Agency presents default assumptions that address these occurrences based on
information developed by the State  of California Air Resources Board, Are these
default assumptions adequate in terms of reflecting short-term and longer-term
increases  in emissions when Dial-operation occur? Can the SAB recommend
additional assumptions, alternative approaches, and/or databases that may be used?

      The hierarchical approach for estimating emissions needs some clarification.
The data  from trial burn conditions should not be used to represent  normal
operating conditions (temperature, feed, etc.).  Furthermore, the Addendum does
not reference other studies on emissions from stationary combustion  sources which
have been completed.  This information would be useful to those trying to use this
approach.  Will the information in AP-42, which "will be available" for MWC
(municipal waste combustors), sewage sludge incinerators,  and medical waste
incinerators, take into account different types of incinerators, air pollution control
devices, etc.?

      The basis of the default assumptions on incinerator upsets is not clear from
the document.  For example, how many incinerators were  monitored, and for how
long? The basis for the assumption of a one-hour upset duration as  well as
CARB's selection of default values for emissions under poor operation conditions
need to be further documented.  An hour upset seems quite long. Normally the
waste feed would shut-off immediately and the upset time  would depend upon the
gas and solid residence times in the incinerators as  well as how  long the
contaminants were volatilizing from the solid. In the case of liquids, the  time
would be only the gas residence time. How  well does one  hour represent actual
conditions?  Are there national data on  the nature,  frequency, and duration of less
than optimum operating conditions for different types of incinerators which could
be used as a more "realistic" guideline?

      The default assumptions that are recommended suggest that upset
conditions may have  significant impact on the total  emissions, If the frequency of
upsets is substantially greater than  the  California data indicate, then the impact of
upsets on  average emissions could be very significant. If this is the case, then
EPA needs to  get some real data on what the emission factors actually are during

                                      18

-------
start-ups, shut-downs, upsets, and poor operation.  Normally, incinerators are not
allowed to operate with waste materials during start-up (natural gas or fuel oil are
normally used).  Calculations based just on good operating conditions may be quite
misleading.  In addition, various incineration studies and emission data sets exist
in other countries. It is strongly recommended that the Agency compile and review
previous efforts as a way of focusing its future  directions.  The Committee also
notes that the EPA's re-permitting process for incinerators offers a unique
opportunity to obtain existing data on the frequency and duration of upset
conditions for various types of incinerators in the U.S.  Other useful data may be
available that could be required as part of the re-permitting process.  Data on the
frequency of upsets for  ECRA units should be available.

      A specific procedure (based primarily on adsorption theory) is presented to
estimate the vapor phase/particle phase partitioning of semivolatile organic
compounds in the ambient air. This V/P ratio is then applied to the air modeling
of titie contaminant emissions.  Specifically, the V/P ratio expected in ambient air
is assumed to apply at the stack.  Is this method appropriate or can alternatives
be recommended?

      Some clarification of the underlying assumption about which, particles are
being addressed—those from the incinerator or those in the ambient atmosphere—is
needed  in the discussions on pp. 3-30 and 3-31.  The  assumption appears to be
that particle emissions from incinerators are generally removed by control devices
and that the combustor vapors that condense the ambient particulate matter are
being addressed here.  If this is the case, then it should be clearly stated,

      The net effect of using the recommended procedure is to underestimate wet
deposition near the source and to overestimate it further from the source.  This is
because p is high near the source and low downwind  and therefore f is higher
near the source than downwind.

      On p. 3-35, Equation 3-12  says that for p « (cSt), 4> is independent of p.
This only applies for adsorption,   If the molecules are water-soluble, then
absorption is important and 4> is  not independent of p.  Similar considerations
apply for organics being absorbed into an organic matrix (rather than simply
adsorbed on a surface).   In this latter  case,  the weight fraction of organic matter
in the aerosol is probably important.  Equation 3-12 has the correct limits in
either case,  i.e., that $ approaches 1 at large p and zero at small p. However, to
the extent that absorption  is important, it does not have the correct form between

                                      19

-------
the limits. The examples cited in the documents apply co very low values of 4>, on
the order of  10"12 atmospheres in some cases.  At thest; low values of p, absorption
may be an important process even for very slightly soluble  vapors.

      In addition, there has been much more work done on adsorption of organics
since Junge's publication.  This section does not reflect the more recent (1982 to
current) literature on this  issue.  Specifically, temperature has long been known to
affect the distribution between vapor and particle  phases (See Yamasaki et al,,
Environ. Sci. TechnoL,  16: 189-194,  1982; Pankow, Almas. Environ., 25A:
2229-2239, 1991; Baek et al., Chemosphere, 22: 503-520, 1991).  Summer-winter
differences between the percentages  in the vapor phase can differ by a factor of
three.  Just recently,  Pankow et al. (Environ. Set. &  TechnoL, 27: 2220-2226,
1993) reported some evidence that relative humidity affects the partitioning.  This
effect appears to be smaller than that due to temperature but EPA should be
aware of this work.

      Also, Whitby's data  for surface to volume ratios (p.3-35) are actually based
on measurements of particle number concentration.  Surface and volume were
estimated assuming spherical droplets. Therefore,  these data do not reflect actual
measurements of surface-to-volume ratio.  The actual surface-to-volume ratio in
the context of Equation 3-12 could be very different, given  that some  fraction of
the particle mass contains  materials with very high adsorptive capacities.  Because
this  fraction  is usually small, the overall surface to volume  ratio could be highly
variable.

      Since the vapor/particle partitioning for semi-volatile organic compounds is
an input parameter that is used throughout the modeling on atmospheric and
inter-media transport, it is likely to  have a significant impact on final outcomes of
the modeling. Thus, some representative measured data  are critically needed.
Measurements made out the stack should be made with a dilution sampler and
should utilize the more  advanced denuder-filter samplers for determining the
partitioning wherever possible.

      In order to evaluate the potential deposition flux of particles near the
source, a particle-size distribution must be known,  A  default assumption of the
particle size distribution of particulate matter in eombustor emissions is presented
in Table 3-7.  Is this default distribution of particulate matter appropriate for use
in those situations where site  specific information is not available? The emission
rate  of the particle-bound portion of the contaminant is apportioned to the particle

                                      20

-------
ske distribution based on the assumption, that the contaminant will be adsorbed
on the surface of the particle. Is the procedure for assigning the portion of
emissions to the partieulate array logical and technically defensible?

      The reason for selecting the size distribution presented in Table 3-7 as the
default distribution needs to be justified.  How does it compare with whatever size
distribution data are currently available? (some of these data may be available at
the State level).  How sensitive are the exposure predictions to the particle size
distribution? If exposure predictions are sensitive to this parameter, then why not
simply require some direct measurement? Compared to the cost of building and
operating an incinerator, this cost is generally trivial.  Also, if combustor
partieulate emissions are being addressed (see comments above) the method of
measuring the particle size distributions from the incinerator should be specified,
A dilution sampler that allows for condensation and coagulation should be used as
it provides a better measurement of what the size-distribution will look like in the
environment.

      The Addendum recommends running the COMDEP model twice in order to
isolate the ambient air concentration and the wet/dry deposition flux of the
particle-bound portion of the contaminant. Is this approach appropriate and
sufficient for purposes of providing  concentration terms to be used in estimating
inhalation exposures, for estimating media concentrations, and for estimating
contamination of the human food chain?

      It  should be noted (p. 3-16) that MPTER-DS is only one of a number of
model frameworks for estimating deposition  from a point source.  To the extent
that the  vapor/particle partitioning depends upon airborne concentration and,
therefore, changes with distance from the source, the MPTER-DS  model does not
describe  this phenomenon. An alternative framework is the segmented plume
model approach (e.g., Zanetti and co-workers). This framework can include both
deposition (wet and dry) and changes in the vapor-particle partitioning. Another
addition  worth considering is the adjustment of the sigma-z value  below the plume
centerline due to  shear induced dispersion (see, e.g., Venkatram et fll.). This
phenomenon is important under weakly stable conditions and can result in ground
level concentrations that are several orders of magnitude greater than those
predicted by symmetry models, Because weakly stable air is a frequent nighttime
phenomenon, this adjustment could  be important to the predictions of dry
deposition  on a long-term basis.
                                      21

-------
      The scavenging coefficients in Table 3-4 are not very useful because they are
too general.  The current version of the model allows the user to specify the
coefficients as a function of particle size.  A more useful set of suggested default
values should be provided here as a  function of particle size.

      With respect to the vapor deposition, there has been a good deal of work on
predicting air-surface exchange during the last decade or so, generally using the
multiple resistance analogy, under the National Acid Precipitation Assessment
Program; in future generations of this model, perhaps some of that effort could be
adopted, so that dry deposition is better treated.

      In addition, it has been pointed out that the basic transport and dispersion
model is a Gaussian plume type, adjusted to account for some of the  inherent
shortcominp of that model, and drawing  on decades of EPA and air  quality
community experience with such models.  But the model still has two central
problems:

      a)    It just cannot deal with calms (the algorithm has the wind  speed U in
            the denominator, a consequence of the model's basic assumption that
            along-wind diffusion is negligible compared to transport  with the
            wind, and so the computations "blow up" as U approaches zero).  In
            some  areas of the U, S, (e.g., the southeast), calms are very common
            - up to 20% of the time or more - especially  at night.  The model
            simply ignores these periods, which unfortunately may account for the
            highest concentration levels.  This can lead to underestimates of dose.

      b)    The model  uses "straight-line" winds from a single location  to move
            the plume.  This  is adequate  for relatively  flat terrain, especially
            when  averaged over long periods, but it cannot account for
            terrain-forced changes in wind direction, which are often very
            persistent.  A simple example is the  wind within a curving valley,
            which often simply follows the valley; however, a computed
            straight-line plume will exit the valley, and produce erroneous
            concentration predictions, both short- and long-term.  If  these are put
            together with actual population patterns (population centers are often
            within valleys), there is a risk of underestimating the dose.  A coastal
            zone which experiences frequent and repeatable sea/lake  breeze
            patterns is  another area where a straight-line  model will not perform
            well.  Note: the authors do a good job of emphasizing that on-site

                                     22

-------
            data are highly recommended for the calculations; in a complex
            terrain site, such data are essential.  Data from some airport 50-100
            km away are simply irrelevant.

            Because of these two shortcomings, it is important that some explicit
            guidelines be established for use of the model, to ensure that it is not
            applied  in inappropriate circumstances.

      It is also recommended that the EPA consider taking into account the
substantial progress that has been made in atmospheric transport modeling, and
move on in its regulatory models to codgs which can deal  with calms, and with
spatially non-uniform wind fields.  This may require (at least in the short term)
an increased amount of site-specific wind data to drive the model, but greatly
increased realism should result.  For the  purposes of the combustor document, the
Gaussian model is adequate as long as its application is limited to conditions
where its inherent limitations are not exceeded.   This means that it may not be
possible to do a reasonable assessment of all sites with this particular tool.

      The Bowers et al, algorithm for building  wake effects (p. 3-17) has been
shown to be flawed in a significant way (Shulman and Scire, J. Air and Waste
Manage, Assoc,, p. 1122-1127, 1993),  The algorithm has nonsensical mathematical
limits such that a tall, skinny building can generate wake  concentrations greater
than the stack gas concentrations! It also misses the fact that a portion of the
plume can be trapped in the wake (because it has an "all or nothing" feature) and
thus will significantly underpredict the wake concentrations when the stack is
short.  This latter feature was verified via wind tunnel testing in the paper by
Shulman and Scire.  It is recommended that the Shulman and Scire model be
incorporated into COMDEP.  It is a  straightforward fix,

      There also needs  to be some discussion of the methods used to estimate the
turbulence parameters, u* and L (p.  3-22 to 3-23).  There  are a number of
approaches that could be used, ranging from direct methods based upon cloud
cover and wind speed to a simple look-up table  based upon stability class. The
latter method is the  least accurate, especially under stable conditions. If one
wants to use other methods, there should be a list of those methods.

3.3,3  Soil Impact and Food Chain Issues
                                      23

-------
       Impacts to soils have been modeled as a function of contaminants depositing
 onto soils in the particle-phase; deposition has included wet plus dry deposition of
 particle-bound contaminants,  A common shortcoming identified is the lack of
 consideration of vapor phase impacts to soils.  This Addendum recommends the
 inclusion of a vapor-phase diffusive flux term to soils, based on the gas-phase mass
 transfer coefficient described in the IED and used there to  estimate a volatilization
 rate from soils. Is this approach accurate, appropriate, and sufficiently described?

       With regard to deposition, the  Committee is pleased  that the Agency no
 longer recommends the use of the ISC deposition model.  The new recommended
 procedure is also troubling, however.  Given the present state of knowledge, the
 recommended procedure is not superior to simply assuming a total particulate
 deposition  velocity on the order of 1  cm/sec (together with  a presentation of the
 uncertainty around this value),  For additional information, see Webster and
 Gannett, 1989a.

       The Sehmel-Hodgson model is  semi-empirical.  One of the resistance
 integrals (box 3) is evaluated using wind-tunnel data.  Dry  deposition onto  plants
 is of considerable interest for indirect exposure assessment.  It is  not clear  that
 the surfaces  used in the wind tunnel  experiments  (including artificial grass)
 adequately model plant surfaces (e.g., waxy hairs).

       The  boundary conditions of the model may  not match.  Many risk assessors
 estimate dry deposition by multiplying a deposition velocity by an air
 concentration at a reference height (calculated fay a plume model).  As Sehmel and
 Hodpon state, their deposition velocity model assumes that "particles diffuse at a
 constant flux from  a uniform concentration of particles.,,"   Although this does not
 precisely match the conditions in a plume, it is unclear how much  difference this
 discrepancy would make.  When a surface depletion model is available, Sehmel and
 Hodgson note that the box 3 resistance integral can be used directly (Sehmel and
 Hodgson, 1980),

      How well has the Sehmel-Hodgson model been validated with field data?
Travis and Yambert (1991) compare the predictions of several such models  with
various field  data.  All of the models  they analyzed tend to underestimate mean
 deposition velocities for particles with diameters of 0,05-1 ^m (an important range
of particle sizes for incinerators).  There was great variation in  the field data, with
deposition velocities as high as about 1  cm/sec for all particle sizes from 0,02 jim
to 10  )^m.  Unfortunately, as is often the case with field data, detailed

                                      24

-------
meteorological data are not provided.  We are left with a difficult problem:  how
do we validate less than ideal models  with problematic field data?  Has anybody
conducted a more rigorous evaluation of the Sehmel-Hodgson model?

      Koester and Hites (1992) measured wet and dry deposition of ambient
particulate-bound PCDD and PCDF in Bloomington and Indianapolis,  The average
dry deposition velocity was about 0,2  cm/sec; wet deposition flux was about the
same as dry deposition.  McVeety and Hites (1988) estimated dry deposition of
PAHs at about 1 cm/sec to water.  A number of factors may account for this four-
fold difference,  including the type of surface, water vs.  "frisbee"  collectors (neither
of which behave like plants).

      Mixing soil depths of 20 cm for tillage operations (farming, e.g.) and 1 cm
for non-tillage situations (pasture, lawns at residences) were recommended in the
1990 EED, and are also recommended here with additional verbiage on their use
and interpretation. Are these depths  appropriate, and  is the SAB aware of other
methodologies to determine more appropriate soil mixing depths?

      These assumptions seemed reasonable to  the Committee.

      A differentiation between below and above ground vegetation is
recommended in this document, and a procedure supplied for below ground
vegetation is described.  Is the differentiation appropriate, and is the procedure  for
the underground vegetation appropriate?

      The Addendum recommends separating the estimation of transfer to above
ground and below ground parts of plants.  This seems reasonable.  However, the
proposed model should be evaluated with respect to factors such as; the theoretical
basis of the model, the reliability of empirical data, the range of validity, the
similarity of modelled situation to experimental set-up, and the validation with
field data.

      As an example of concerns about data reliability, Travis and  Arms' (1988)
regression equation used results on TCDD from Helling et al.t originally reported
in Isensee and Jones (1971).  In actuality, no TCDD was detected in the plants,
making the point an upper bound.  Examination of the regressed data suggests
that it may depend to a large degree on one extreme data point for PBB,
                                      25

-------
       The 1990 IED assumed that wet deposition (i.e., deposition of particle-bound
 contaminants via rainfall) would not impact above ground vegetation
 (vegetables/fruits, and animal feeds).  The reference to this assumption was an
 internal memorandum which did not include a technical justification. The
 Working Group recognized that for some compounds evidence exists that wet
 deposition is roughly equal in magnitude to dry deposition, so that zeroing out wet
 deposition halves the impact of depositions to plants.  Subsequent to the
 completion of the Addendum document, the Working Group was made aware of a
 key publication by Hoffman,  et al., (1992, Quantification of the interception and
 initial retention of radioactive contaminants deposited on pasture grass by
 simulated rainfall.  Atmospheric Environment 26(A) (18) 3313-3321).  Data from
 this article suggests that the retention of particles on the vegetation tested ranged
 between 24 and 37% of all depositing,  depending on size of particle and vegetation,
 Without this information, the Working Group recommended the following
 assumptions for the percent of contaminants in wet deposition that is retained on
 the plant surface: 1)  100% for contaminants with high sorptive tendencies and
 10% for contaminants with low sorptive tendencies. The Hoffman article suggests
 that an assumption of 30% retention of particle-bound contaminants, regardless of
 sorptive tendencies of the contaminant is a more supportable assumption.  Which
 of the two approaches, or other approaches, would the SAB recommend? And is
 the SAB aware of other information that would assist in developing values for the
 retention of particle-bound contaminants on plants during rainfall?

      The 30% assumption seems more supportable.  The  previous assumption
 that the sorptive properties of the contaminant on the particle determine retention
 does not seem reasonable since it is the interaction between the plant surface and
 the particle as a whole that is more likely to be the determinant of retention.

      After contaminants from stack emission deposit onto soils, they can be
resuspended as particles or volatilise into the air.  Model testing has shown that
resulting air concentrations associated with these soil emissions are significantly
smaller in magnitude compared to concentrations resulting from the stack
emissions.  Accordingly, the Addendum does not recommend increasing modeled
air concentrations from stack emissions to reflect additional emissions from soil.
Does the SAB concur with this  recommendation or can alternative approaches be
recommended?

      Does "significantly smaller" mean an order of magnitude or two or a factor
of less than an order  of magnitude? There should be some statement that

                                      26

-------
provides this information.  If it is "order of magnitude" smaller, then the
Committee agrees that there is  little point in extensive modeling.

      Recent literature has emphasized the importance of vapor-phase transfers to
vegetation, particularly for diosin-Mke compounds.  This Addendum recommends
substantial changes for estimating vegetative impacts by vapors. These changes
include:

      1)     Neglect soil volatilization as an additional reservoir for vapor
             transfers.  The argument was presented that the soil-plant empirical
             transfer factor already includes all routes of soil to plant transfers.

      2)     Do not include the vapor fraction term, since the air modeling will
             directly yield a vapor-phase concentration.

      3)     In external model testing, the Bacci azalea leaf model was shown to
             overestimate the transfer of vapors to plants.  An empirical correction
            reducing the value of the Bacci transfer factor by 90% was
            recommended based on external validation exercises with dioain.

Are these changes technically accurate, complete, and sufficiently described?

      One of the most significant issues in the methodology that can and should
be addressed is the inherent lack of reliability associated with both measured data
and models used to determine inter-media  transfer factors (ITFs).  There are a
number of ITFs proposed for use in the Addendum ineluding-oetanol-water
partition coefficients (Kow); organic-carbon partition coefficients (K^; soil-water
and sediment-water partition coefficients (Kd); Henry's law constant; mass-transfer
coefficients in water and air; steady-state bioconcentration factors for plant root
concentrations relative to soil concentration, plant leaf concentration relative to air
concentration, and fish concentration relative to water/or sediment concentrations;
steady-state biotransfer factors for milk or dairy-product concentration relative to
contaminant  intake by cattle; meat concentration relative to contaminant intake by
animals;  egg  concentration relative to contaminant intake by chickens;  and breast
milk concentrations relative to  contaminant intake by mothers;  and contaminant
biodegradation factors in soil.  Recent work on the reliability of methods for
measuring and estimating these types  of ITFs reveals that the reliability for
determining ITF values has an associated error factor in  the range of 1.5 to 10
depending in part on whether the ITFs are measured or estimated.  McKone has

                                      27

-------
found that overall variance in quantitative estimates of ITFs comes from several
factors including (1) variability among experiments; (2) our ignorance regarding
the processes of metabolism and chemical partitioning; and (3) the reliability with
which we can measure both the outcome (biotransfer or partition factor) and the
explanatory variable (i.e., Kow),(SAE and QSAR, in Environmental Research 1,41-
51, 1993)  It is likely that the lack of reliability for determining these ITFs can he
a major contribution to the overall variance in the exposure estimates.

      With respect to vapor plant transfer, the Committee is pleased that EPA
has considered  this potentially important pathway.  A thorough evaluation would
discuss (as noted earlier) the theoretical  basis of the model, reliability of empirical
data, range of validity, similarity  of modelled situation to  experimental set-up and
validation with field data.

      A  comparison between the  results  of Bacci et al, and McCrady et at.
indicates two problems with current use  of the  Bacci model;

      a)  plants differ in their uptake  (kl) and volatility (part of k2)  rates;

      b)  photodegradation appears to be a real phenomenon that should be taken
      into account.

      EPA suggests using the Bacci model with a common correction factor of 0.1
for all compounds.  EPA bases this correction factor of modelled and measured
average "dioxin" environmental concentrations.  There are some problems with this
approach.  First, the Addendum does not present enough data for proper
valuation.  Second, even if the data were appropriate, the comparisons should take
into account the variance of the data, not just mean values,  Third, the sensitivity
of various compounds to photodegradation varies  (even within the dioxin family).

      On the other hand, comparison of the  grass and azalea results in Table  III
of MCCready et al. suggest that application, of the Bacci model overestimates
2,3,7,8-TCDd vapor transfer to grass by about an order of magnitude or  so (about
40 using the expected value for 2,3,7,8-TCDD from Bacci's regression model).  The
results of McCrady et al. seem superior for estimating vapor deposition of
2.3,7,8-TCDD onto grass, but the  generalizability of this ratio is unclear.  To do so
we would need  to know more about photodegradation of other compounds and the
differences between plants.  For instance, Bacci's results may be a better
approximation for certain other plants,

                                      28

-------
      Table III of McCrady et al, indicates that inclusion of photodegradation
reduces the Bv of 2,3,7,8-TCDD onto grass by about a factor of four (relative to
only including volatilization in k2). The discrepancy between the results of
McCrady et al, and Bacci is also due to the use of different plants and isoraers.
Note that the equilibrium "bioconcentration" factor (Bv) depends on the ratio of kl
to k2. Based on  Table III of McCrady et al, kl for grass (2,3,7,8-TCDD) is about
9 times larger than for azaleas (1,2,3,4-TCDD) while the k2 for volatilization is
about factor of 26 larger for grass.

      The Addendum's discussion of the time required to reach steady state
(p,5-9,10)  is somewhat confusing.  If vapor to plant transfer is modeled as a first
order system,

      dC/dt = kl*Ca-k2*C
      C(t) = kl/k2*Ca*(l-exp(-k2*t))

then the time required to reach equilibrium depends only on k2, not kl.  In Table
III of McCrady et al., the k2 value for  grass is larger  than the k2 value for
azaleas, accounting for the  more rapid equilibration.

3,3.4 Water  Impact Modeling Issues

      A three-tier approach to evaluating aquatic impacts, from screening level,
Tier 1, to site specific and more complex modeling, Tier 3, was promoted in the
Indirect Exposure Document.  This has been replaced by a steady-state model for
long-term average risks and a storm event model for short-term acute risks.  Is
the replacement of the three-tier system with these two models appropriate?

      A tiered approach  would seem to be a more sensible approach since the
aquatic modeling  is quite complex and  requires many  inputs. At a minimum,  some
guidance should be provided concerning the conditions under which one might
undertake such detailed calculations.An additional point is that the Addendum
appears to suggest lakes  or ponds as the bodies of water of concern. Much of the
U.S. population resides in coastal cities.  Has any thought been given to impacts of
multiple incinerators impacting coastal regions, e.g., Atlantic seacoast, and possible
impacts of seafood production  and health of the ecosystems in these areas?

      The previous IED  long-term average water model was quite simplistic,
accounting for loading and  dilution only.  The steady-state water model proposed

                                      29

-------
 in the Addendum accounts for several water body loss processes, ,,,  It also allows
 for formation of a reaction product, thereby maMtaining reservoirs of more than
 one contaminant in the aquatic ecosystem.  These processes, of course, require
 more input data. Does the addition of these processes make the simple screening
 level approach more appropriate for assessing health risks?  And can the SAB
 provide comments on specific algorithms?

       The Committee did not have sufficient information to address this question.

       The proposed watershed loading model calculates average soil concentrations
 in a manner similar to the old IED.  One additional process modeled is the
 atmospheric diffusion loading to the soil.  Long-term average erosion of chemical
 from the watershed is multiplied by sediment delivery ratio and a  pollutant
 enrichment factor before loading to the  waterbody. Are the watershed calculations
 appropriate for calculating long-term  average loadings via erosion to the
 waterbody?

       The Committee did not have sufficient information to address this question,

       The proposed storm event water model, like the previous  version, accounts
 for loading and dilution only. There  are two significant differences.  The proposed
 version includes a watershed sediment delivery ratio, so that predicted loading of
 sediment and chemical is reduced by typically over 70%.  Second, the proposed
 version calculates a peak instream storm flow that includes storm  runoff volume
 added  to the average annual  stream flow. The old version diluted storm loads
 into base flow only.  The additional dilution further reduces the calculated peak
 sediment and chemical concentrations. Is the proposed storm  event model
 appropriate for calculating peak  concentrations for assessment of acute or
 threshold health risks?

      The  Committee did not have sufficient information to address this question.
3.3.5  Exposure Issues

      The Addendum recommends a procedure to define the extent of the study
area on the basis of an isopleth corresponding to a de minimus risk level. Is this
appropriate; if not could alternatives be suggested?
                                      30

-------
       The use of geographically-defined boundaries based on dispersion/deposition
 models is reasonable for estimating concentrations in soil, water, and air, and
 direct exposures to these media. However, it may not be sufficient for defining
 the population at risk from indirect exposure.  The Working Group has partly
 recognized this by including in the population at risk those individuals who live in
 areas of negligible risk but spend time within areas of concern. However,  indirect
 exposure to people living outside the areas of concern could also occur because the
 medium of exposure (e.g., food) is transported from the areas at increased risk to
 individuals who neither live nor spend time in those areas.  It would not be
 unreasonable for populations residing and working beyond the fifty mile boundary
 to consume foods such as beef or milk produced within the fifty miles.  The
 criteria for defining the population at risk should consider the complete indirect
 exposure pathways (e.g., consumption of beef and/or milk from cattle range-fed
 within the  50 Km) on a site-specific basis. (This approach is apparently
 suggested in method #2 for estimating population risk on page 15-2-3),

      The idea of "isopleth rings" as  presented in Figure 2-1 (page 2-5)  and Table
 2-1 (page 2-7) should be refined. The shape and area of the isopleths derived
 from the dispersion/deposition models could be quite different from those
 represented by the "rings," and will depend on combustor-specific characteristics
 (e.g., stack height), as well as local topography and prevailing atmospheric
 conditions,  A "ring", as represented,  could encompass both areas of high and low
 media contamination* depending on prevailing wind direction, for example.
 Further, a high (or low) media concentration  isopleth could extend beyond a "ring11.
 Therefore, within any given ring" one could expect to find a distribution of
 exposures (and risk) depending not only on population activities but  also  on
 media concentrations.

      Why was 200m selected as the minimum distance  for the first  "ring" of
 receptors under all conditions?  Is this distance reasonable in situations when
 downwash/building wake  effects occur?

      The recommendation to develop distributions of exposures is a substantial
improvement over the use of worst case scenarios and is to be commended.  The
two procedures described  are generally clear and reasonable,  One clarification -
 for Method 1 (series of point estimates), how does one calculate an average since
the modeling is based on  several scenarios? Is the average assumed  to be the
model outcome when average values are used  as input?
                                      31

-------
      The Addendum recommends a procedure to develop distributions of
exposure on the basis of either a series of either point estimates of exposure or a
Monte Carlo assessment.  Both would involve identifying isopleths defining areas
of equal media contamination levels and characterizing behavior patterns in these
areas* Are these procedures clear and reasonable?

      The Addendum provides some general  guidance for conducting a Monte
Carlo assessment in the areas of separating uncertainty and variability, addressing
dependence among variables and selecting distributions.  Is this guidance
appropriate and sufficiently detailed?

      The use  of Monte Carlo simulations to estimate  the effects of  uncertainty is
not sufficiently clear. The distinction between "variability"  in exposures and risks
(due to variability in the location and activities so the  population), and
"uncertainty" due to  uncertainties in measurements and parameterization need to
be made clear.

      Monte Carlo Analysis is only one approach used to examine variance.  It
can yield large  quantities of data that are not particularly useful.  One must also
be concerned with the form of the exposure distributions, and examine the
high/median/low features of a distribution at  a site.  These  distributional analyses
can help define the types of actual measurements which should be taken after the
plant is operational,  and  how they can be used in a follow up assessment,

      The Addendum provides two procedures for estimating breast  milk
concentrations  on the basis of maternal intake. Are these the best models
available for this purpose?  Are the assumptions sufficiently well stated?

      Accuracy can only be determined by comparing estimates with actual data.
For TCDD, the Smith approach would appear to be more accurate than the  Travis
et al. model, as presented.  Since the Travis et aL model was largely based on
estimates rather than direct measurements of organic compound concentrations in
human milk.  Can any of their estimates be compared  to actual data for those
same compounds, if those data are now available?  This may explain part of the
discrepancy and perhaps assist in adjusting their equation.  Anyway, if application
of a  simpler model results in estimates which approximate actual data  better, use
of other more complex and apparently less reliable alternatives appears to be
unnecessary.
                                      32

-------
      The various approaches-bioeoneentration, biotransfer, fraction transferee!
(McLachlan et al,, 199Q)-all have an underlying commonality based on
pharmaeokinetics. Among the factors that should be taken into account in
modeling bioeoncentration of lipophillic, low volatility compounds  are;  absorption,
metabolism, growth, various routes of elimination (lactation, metabolism, excretion
of parent compound across the gut).

      Some of the best data on bioaccumulation of dioxin and similar compounds
into cow's milk comes from McLachlan (1990, 1993) who examined (presumably)
steady-state input and outpout of xenobioties from a dairy cow eating feed grown
in a relatively unpolluted area. This case provides a closer match to many real-
world "indirect exposure" situations than extrapolation from spiking experiments
(requiring additional assumptions about the bioavailability to the  cow),
McLachlan's data art preferrable the TCDD estimate  of Fries and Pastenbaeh
(1990) which was not at steady state (Fries and Pastenbaueh justified the latter on
the basis of similarity to BCFs for certain other xenobiotics.  However, some of
the latter were not persistent; for instance, arochlors  contain both readily
metabolizable and non-metabolizable congeners). Although some claim that the
BCF for TCDD should be the same for beef fat and milk fat, this appears unlikely.
Correction for experimental values to steady-state results in higher
bioconcentration in beef (Jensen et al., 1981).  Lactation provides  an efficient
means of excreting lipophilic compounds.

      For  most compounds, many model parameters for bioconcentration are not
readily available and structure-activity relationships are recommended in the
Addendum as a means of estimation,  EPA should make clear that good
experimental data are preferrable to use of a structure-activity relationship.
Second, the use of several of the SABs in the Addendum that estimate
bioaccumulation from food or feed to breast milk, cow's milk or beef based on
logKow (a measure of the lipophilicity of a compound) cannot be recommended for
several reasons.

      These SARs assume a direct proportionality between bioaccumulation and
1°£KOW, There is some theoretical basis for this relationship for fish based on
partitioning between water and fish lipid. But there are  several well known-
limitations to even this situation. It does not work so well for compounds which
are easily metabolized.  It does not properly take into account biomagnification
(which leads to concentrations above equilibrium with water). It  does not work
well for highly lipophilic compounds. Indeed, the relationship between

                                      33

-------
 bioeoneentration, bioaccumulation and Kow seems to plateau and then decrease for
 these compounds.  This may be a function of poor absorption.

       For terrestrial animals, partitioning occurs primarily between body lipid and
 gut contents.  The other limitations apply as well,  A decrease of bioaccumulation
 at high log KQw is usually observed.  The ability of an organism to metabolize a
 compound has no necessary relationship to Hpophilicity.  For example, some
 PCDDs are relatively easy to metabolize while others are not; it depends on the
 number and arrangement of the chlorine atoms. As a result, compounds with
 similar log KQWs may bioaecumulate to very different degrees (unfortunately, the
 more toxic PCDDs are the same ones that bioaccumulate).

       The data used in such SAHs must be carefully scrutinized.  Inclusion of
 both readily metabolized and poorly metabolized compounds in the SARs—as was
 done in the models in the Addendum-will bias estimates, generally downward.
 The data  should normally represent steady-state conditions (unless the time
 required for steady-state is excessively long), Empirical data not corrected to
 steady state will also bias estimates downward.

       Concentrations in beef cattle will be diluted by growth, but a "balance" of
 these two factors would be coincidence. Finally, the diet of beef and dairy cattle
 may vary considerably by location and type of farm. For instances the cattle of
 "subsistence" or small farmers may consume more pasture than larger commercial
 herds.

      The Addendum describes a procedure for estimating population risk for
 cancer on the  basis of local food production.  Is this approach clear and
 reasonable?

      The Committee did not have sufficient information to address this question.

      The 1990 IED document used the 1978 USDA food consumption! survey to
 estimate diet fractions for locally grown foods,  Some of these estimates do not
 appear reasonable.  The Working Group strongly recommends trying to base these
estimates  on local surveys, but could not resolve how to make these estimates  if
 this was not feasible. Can the SAB provide any further guidance on how to
address this issue?

      The Committee did not  have sufficient information to address this question.

                                      34

-------
       Indirect pathways are defined as aU pathways other than inhalation.  This
 document addresses food chain, water, soil, and breast roilk exposures.  Should
 other pathways be considered?

       A somewhat revised and expanded definition of indirect exposure is needed
 that includes the idea of transfer to another medium, for example, deposition of
 aerosols followed by resuspension and inhalation or ingestion.

       Penetration of the incinerator-produced aerosols into buildings and
 subsequent deposition onto surfaces is mentioned as an exposure pathway (p. 4-1)
 but is not treated adequately.  This pathway may be more significant than
 suggested in this document for populations living near incinerators. When
 windows are open, there is quite a  rapid accumulation of dust on indoor surfaces*
 For homes near incinerators, this dust is likely to be relatively enriched in
 incinerator contaminants (i.e;, there will be less dilution by other outdoor aerosols
 or by soil).  Children would be expected to pick up such dusts on their hands and
 carry them to their mouths.  The recommendation that the estimated
 concentration in the top 1  cm of soil be used as the estimate for this deposited
 dust from air  seems inappropriate as the soil itself will cause a dilution.

 Some Additional Exposure Issues

      Section 7,, Determining Dermal Exposure from Dermal Absorption via Soil,
 p. 7-1,  The conclusion/recommendation for the issue of dermal risks is to apply
 the USEPA guidance document, "Dermal  Exposure Assessment; Principles and
 Applications."  One of our  Committee members reviewed this document and noted
 that the application is not  entirely straightforward.  Parameters for its
 implementation may not be available for many chemicals.  The equations have not
 been provided in the Addendum and it is therefore impossible to evaluate effects
 of uncertainties in these parameters on the outcome of an exposure and risk
 assessment.

      Section 8,  Dust Resuspension, p. 8-1.  The conclusion/recommendation on
 estimating exposures to fugitive dusts from tilling and  vehicle traffic is to add
 algorithms  for estimating this exposure. As in Section 7, however, documents are
 cited but no equations or example calculations are provided as the reader is led to
 expect from the "Example Calculations" heading. Vehicle generated fugitive dust
may be 100 times as great  as wind-generated fugitive dusts.  This exposure
pathway is  particularly important for chemicals that are carcinogens by inhalation

                                     35

-------
but not ingestion.  For chemicals that are carcinogens by both routes, the larger
mass of ingested soil will be the more important pathway,

      Section 11.  Determining Exposure form Pish Intake, Bioconcentration
Factor, p. 11,1,  The conclusion/recommendation for this section is confusing.  In,
the first  paragraph, it is stated that "The use of the Biota Sediment Accumulation
factor (BASF) is recommended for dioxin-like compounds,"  In the second
paragraph appears: "While this recommendation is made here, it should be  noted
that water column approaches for lipophilic compounds, including dioxins, are
currently being used in the Agency and used appropriately."  Which is it? BASFs
may he hard to come by.  It is not a good idea to incorporate terms for exposure
assessments for which input values are not available.   The models are simply not
usable.  Parameter values should be provided for at least the top 50 contaminants.

      For the second issue on this page.  "Alternate approaches for estimation of
aquatic bioconcentration and bioaeeumulation may also be appropriate,"  it is
suggested that the Food and Gill Exchange of Toxic Substances Model of ERL
Athens, GA be consulted.  Where is it and how does its output compare to the
other suggested methods? What is the overall recommendation?  Exposure to
toxics by the consumption of contaminated fish is a very important route of
exposure. The Methodology for Assessing Health Risks Associated  with Exposure
to Combustor Emissions will not be very useable until some of these issues have
been addressed  and examples for many chemicals have been worked through the
model.

      Section 12.  Determining Exposure from Dermal Absorption from Water, p,
12-1.  The conclusions/recommendations for the issue  of dermal absorption from
water are:  "This chapter should be replaced  by the appropriate sections of the
report "Dermal  Exposure Assessment: Principles and Applications." Which  are
the appropriate sections?  Where are the equations? Are the parameter values
available for a significant number of chemicals?
                                     36

-------
3.4   Summary of Major Recommendations

      The following are the major recommendations arising from this review.  The
bases for these recommendations are summarized in Sections  3.2 and 3.3 above.

      1,     Use the Addendum as an analytical tool rather than a site-specific
            regulatory methodology. That is, use the addendum to identify
            chemicals, environmental compartments and exposure pathways
            associated with these  sources and thus provide strategic guidance in
            utilizing environmental sampling to obtain actual data on indirect
            exposure to humans and to ecosystems,

      2.     Do not release the Addendum as an "EPA Methodology" for routine,
            quantitative, site-specific risk assessments for incinerators.

      3,     Develop and implement a strategic plan to collect critical input data
            that can be used to refine the models and to validate the
            methodology. Take advantage of the re-permitting process  as an
            oppotunity for the collection of such relevant data,

      4     Establish a framework to ensure that the entire  range of potential
            risks from combustors are addressed holistieally.  This must include
            both direct and indirect risks, as well as local,  regional, national and
            international concerns for both waste combustors and fossil fuel
            combustors.
                                      37

-------
APPENDIX A.    Charge to the Committee
                                 38

-------
                 UNITED STATES ENVIRONMENTAL PROTECTION AGENCY
                             WASHINGTON, D»e.  20460
 MEMORANDUM
 SUBJECT1.
 FROM:
TO:
                                                                 OFFICE 0?
                                                           RESEARCH AND DEVELOPMENT
Science Advisory Board Review of the Draft Report
"Addendum:  Methodology for Assessing Health Risks
Associated With Indirect Exposure To Cornbustor Emissions
Working Group Recommendations"

William H. Fariand, Ph.D.
Director
Office of Health and Environmental
  Assessment (8601)

Donald G. Barnes, Ph.D.
Director
Science Advisory Board {1400}
      The purpose of this memorandum is to outline our charge to the Science
Advisory Board (SA8) for review of the subject draft report.  This report was
developed by a Working Group which began deliberations in February, 1993. The
purpose of the report was to. evaluate and make recommendations concerning the
state of the science with regard to indirect exposure methodologies relating to
incinerator emissions,  The Working Group was comprised of individuals from the
several offices within EPA, including the Office of Research and Development, the
Office of Solid Waste, the Office of Pollution Prevention and Toxic Substances,
and the Office of Water.

      The Working Group selected the document, Methodology for Assessing
Health Risks Associated with Indirect Exposure to Combustor Emissions - Interim
F/nal  (EPA/600/6-90/003;  January, 1990; subsequently abbreviated the Indirect
Exposure Document or IED) as the best currently available guidance on this topic.
Using this document as a starting point, the Working Group identified areas
needing updates and recommended ways to make these updates.  The
recommendations have been compiled into an Addendum  to the Indirect Exposure
Document.

      The Addendum is not meant to be a stand alone document. It is meant to
be used in conjunction with this Indirect Exposure Document.  The Addendum
identifies "Issues" and makes "Conclusions/Recommendations" regarding these
                                 A-l
                                                                printed on flecyefetf Paper

-------
 issues.  The short-term goal of the Working Group was to provide guidance to /;.;.;
 assessors for conducting indirect exposures assessment using state-of-the-sctenco
 tools,  A longer term goal for the Office of Research and  Development is to merge
 the Addendum with the Indirect Exposure Document to provide a single, updated
 indirect methodology document.  A second long term goal is to develop a
 companion document which provides guidance on selecting values for parameters
 used in the methodology.

      The Working Group recommended changes to most chapters of the 1990
 Indirect Exposure Document. The most significant changes were made in the
 following three areas;

      1) Emissions, and air dispersion and deposition modeling:   A
 comprehensive list  of potentially emitted contaminants from combustors was
 compiled and a procedure outlined to narrow the list for a specific facility. An
 updated version of  the COMPDEP model for atmospheric  transport, dispersion, and
 deposition, was described and recommended for use.  Procedures to appropriately
 describe and model vapor phase and particle phase emissions were provided.

      2) Surface water Impacts:  The three-tier option for calculating water
 concentrations in Chapter 9 of the  Indirect Exposure Document was replaced with
 a single model which calculates water and sediment concentrations,

      3] Exposure; This Addendum updates the procedures used to develop
exposure scenarios  on  the basis of the policies described  in the February 26,  1992,
 Deputy Administrator memorandum "Guidance on  Risk Characterization for Risk
 Managers and Risk  Assessors" and the 1992 Exposure Guidelines.  Accordingly,
an emphasis is placed on developing distributions of exposures and identifying high
end and central exposure levels,

      We are seeking comment from the SAB in the following five areas. In areas
where weaknesses  are found, please provide suggestions for improvement.

General Issues

i. The Working Group  recognizes the awkwardness of needing to  reference two
documents to determine the procedures for conducting indirect exposure
assessments, and has recommended that the 1990 iED be updated with this
Addendum  as a single methodology.  In the interim, is the presentation of material
in the Addendum clear? and can the SAB make specific recommendations for
further clarity and completeness?

ii.  The Working Group recognizes that implementation of the full methodology is
very resource intensive and that screening procedures are needed to narrow the

-------
scope and level of detail where appropriate.  The current draft does provide some
genera) guidance on narrowing the list of compounds of concern, However, in the
current form, most decisions about how to establish screening procedures would
need to be made within Agency Programs as part of their implementation
guidance. Does the SAB believe this is appropriate or should further screening
procedures be included here? If so can such  procedures be recommended?

Air Emissions and Modeling Issues

      Chapter 3 of the Addendum presents  a framework for estimating emissions
from stationary combustion sources, and for estimating the atmospheric
concentration and surface deposition flux of the contaminants near the source
using the COMPDEP air dispersion model.  The Agency is not asking for a critical
review of the algorithms inherent to COMPDEP.  Instead the Agency would like the
SAB to focus on application issues such as selection of the appropriate inputs to
the model and proper interpretation and use  of the outputs. Specific  issues are
listed below.

i.  Tables 3-1  to 3-3 present a listing of chemicals that have been measured in the
stack gas emissions of a variety of combustion sources and that may be subject to
the direct/indirect exposure/risk assessment. Is this sufficient in  terms of an initial
listing and a first step  in the process?

ii.  A general procedure is presented that, if applied on a site-specific basis, could
reduce the number and kinds of  contaminants subject to review. Is the procedure
adequate for this purpose, or should the Agency continue to develop a more
detailed/refined screening procedure?  If a detailed procedure is  recommended, can
you provide guidance or suggestions as to a  screening method that would be
appropriate?

iii.  A hierarchical approach for estimating emission rates of the contaminants is
presented.  Is this procedure adequate in terms of providing estimates of average
emission estimates? A more difficult issue is the magnitude, frequency and
duration of increased emissions that may occur during temporary upset conditions,
during startup and shutdown procedures and during emergency events. The
Agency presents default assumptions that address these occurrences based on
information developed by the State of California Air Resources Board.  Are these
default assumptions adequate in terms of reflecting short,-term and longer-term
increases in emissions when mal-operations  occur? Can the SAB recommend
additional assumptions, alternative approaches, and/or databases that may be
used?

tv.  A specific procedure (based primarily on  adsorption theory) is presented to
estimate the vapor phase/particle phase partitioning of sernivolatile organic

                                   A-3

-------
compounds in the ambient air. This V/P ratio is then applied to the air modeling •-.
the contaminant emissions.  Specifically the  V/P ratio expected in ambient air is
assumed to apply at the stack.  Is this method appropriate -or can alternatives be
recommended?
v. Fn order to evaluate the potential deposition flux of particles near the source, a
particle-size distribution must be known,  A default assumption of the particle size
distribution of particuiate matter in combustor emissions is presented in Table 3-7,
!s this default distribution appropriate for use in those situations where site-specific
information is not available? The emission rate of the panicle-bound portion of the
contaminant  is apportioned to the particle size distribution based on the
assumption that the contaminant will be adsorbed to the surface of the particle. Is
the procedure for assigning the  portion of emissions to the particuiate array logical
and technically defensible?

vi.  The Addendum recommends running the COMPDEP model twice in order to
isolate the ambient air concentration of the vapor-phase portion  of the contaminant
from the ambient air concentration and wet/dry deposition flux of the particle-
bound portion of the contaminant. Is this approach appropriate and  sufficient for
purposes of providing concentration terms to be used in estimating  inhalation
exposures, for estimating media concentrations, and for estimating  contamination
of the human food chain?

Soil Impact and Food Chain Issues

i. Impacts to soils have been modeled as a function of contaminants depositing
onto soils in the  particle-phase;  depositions have included wet plus  dry deposition
of particle bound contaminants.  A common shortcoming identified  is the lack of
consideration of vapor phase impacts to soils.  This Addendum recommends the
inclusion of a vapor-phase diffusive flux term to soils, based on the gas phase
mass  transfer coefficient described in the IED and  used there to  estimate a
volatilization  loss rate from soils. Is this approach accurate, appropriate, and
sufficiently described?

ii. Mixing soil depths of 20 cm  for tillage operations {farming, e.g.) and 1 cm for
non-tillage situations (pasture, lawns at residences) were recommended in the
1990 IED, and are also recommended here with additional verbiage on  their use
and interpretation. Are these depths appropriate, and Is the SAB aware of other
methodologies to determine more appropriate soil mixing depths?

iii. A  differentiation between below and above ground vegetation is recommended
in this document, and a  procedure supplied for below ground vegetation is
described. Is the differentiation appropriate, and is the procedure for underground

-------
 vegetation concentration appropriate?
 iv.  The 1990 1ED assumed that wet deposition (i.e., deposition of particle bound
 contaminants via rainfall) would not impact above ground vegetations
 (vegetables/fruits, and animal feeds).  The reference to this assumption was an
 internal memorandum which did not include a technical justification.  The Working
 Group recognized that for some compounds evidence exists that wet deposition is
 roughly equal in magnitude to dry deposition, so that zeroing out wet deposition
 roughly halves the impact of depositions to plants. Subsequent to the  completion
 of the Addendum document, the Working Group was made aware of a key
 publication by Hoffman, et ai. (1992,  Quantification of the interception and initial
 retention of radioactive contaminants deposited on pasture grass by  simulated
 rainfall. Atmospheric Environment 26{AL number 18: 3313-3321).  Data from this
 article suggests that the retention of particles on the vegetations tested ranged
 between 24 and 37% of all depositing, depending on size of particle and
 vegetations.  Without this information, the Working Group recommended the
 following assumptions for percent of contaminant in wet deposition that is retained
 on the plant surface; 1)  100% for contaminants with high sorptive tendencies and
 2) 10% for contaminants with low sorptive tendencies. The Hoffman article
 suggests that an assumption of 30%  retention of particle-bound contaminants,
 regardless of sorptive tendencies of the contaminant, is a more supportable
 assumption.  Which of the two approaches, or other approaches, would the SAB
 recommend?  And is the  SAB aware of other information that would assist in
 developing values for the retention of particle bound contaminants on plants during
 rainfalls?

 v. After contaminants from stack emissions deposit on to soils, they can be
 resuspended as particles or volatilize into the air. Model testing has  shown that
 resulting air concentrations associated with these soil emissions are significantly
 smaller in magnitude compared to concentrations resulting from the stack
 emissions.   Accordingly, the Addendum does not recommend increasing modeled
 air concentrations from stack emissions to reflect additional emissions from soil.
 Does the SAB concur with this recommendation or can alternative approaches be
 recommended?

vi.  Recent literature has emphasized the importance of vapor-phase  transfers to
vegetation, particularly for dioxin-like compounds.   This Addendum recommends
substantial changes for estimating vegetative impacts by vapors.  These changes
 include:

      1, Neglect soil volatilization as an additional reservoir for vapor transfers.
The argument was presented that the soil-plant empirical transfer factor already
includes all routes of soil to plant transfers.


                                   A-5

-------
       2,  Do not include the vapor fraction term, since the air modeling will direcUy
 yield a vapor-phase concentration.


       3,  In external model testing, the Bacci azalea leaf model was shown to
 overestimate the transfer of vapors to plants. An empirical correction reducing the
 value of the Bacci transfer factor  by 90% was recommended based on external
 model validation exercises with dioxin.

 Are these changes technically accurate, complete, and sufficiently described?

 Water Impact Modeling Issues

 i.  A three-tier approach to evaluating aquatic impacts, from screening level, Tier !,
 to site-specific and more complex modeling, Tier 3, was promoted in the  Indirect
 Exposure Document,  This has been replaced with a steady state model for long
 term average risks and a storm event mode! for short term acute risks, is the
 replacement of the three-tier system with these  two models appropriate?

 ii.  The previous IED long-term average water model was quite simplistic,
 accounting for loading and dilution only. The steady-state water model proposed in
 the addendum accounts for several water body loss processes, including  burial,
 volatilization, chemical/biochemical degradation in water column and benthic
 sediment, and sediment burial.  It  also allows for the formation of a reaction
 product, thereby maintaining reservoirs of more than one contaminant in the
 aquatic ecosystem. These processes, of course, require more input data.  Does
 the addition of these processes make the simple screening level approach more
 appropriate for assessing health risks?  And can  the SAB provide comments on
 specific algorithms?

 iti.  The proposed watershed loading model calculates average soil concentrations
 in a manner similar to the old  IED.  One additional process modeled is atmospheric
 diffusion loading to soil.  Long term average erosion of chemical from the
 watershed is multiplied by a sediment delivery ratio and a  pollutant enrichment
 factor before loading to the waterbody. Are the watershed calculations appropriate
 for calculating long-term average loadings via erosion to the water body?

iv.  The proposed storm event water model, like  the previous version, accounts for
loading and dilution only. There are two significant differences. The proposed
version includes a watershed sediment delivery ratio, so that predicted loading of
sediment and chemical is reduced  by typically over  70%,  Second, the proposed
version calculates a peak instream storm flow that includes storm runoff volume
added to average annual stream flow.  The old version diluted storm loads into
base flow only. The additional dilution further reduces the calculated peak

                                    A-6

-------
sediment and chemical  concentrations. Is the proposed storm event mode!
appropriate for calculating peak concentrations for assessment of acute or
threshold health risks?
Exposure Issues

i. The Addendum recommends a procedure to define the extent of the study area
on the basis of an tsopleth corresponding to a de minimus risk tevei.  Is this
appropriate, if not could alternatives be suggested?

ii.  The Addendum recommends a procedure to develop distributions of exposure
on the basis of either a series of point estimates of exposure or a Monte Carlo
assessment.  Both would  involve identifying isopleths defining areas of equa! media
contamination levels and characterizing behavior patterns in these areas.  Are these
procedures clear and reasonable?

iii. The Addendum provides some general guidance for conducting a Monte Carlo
assessments in the areas  of separating uncertainty and variability, addressing
dependence among variables and selecting distributions. Is this guidance
appropriate and sufficiently detailed?

iv. The Addendum describes two procedures for estimating breast milk
concentrations on the basis of maternal intake.  Are these the best models
available for this purpose? Are the assumptions sufficiently wei! stated?

v.  The Addendum describes a procedure for estimating population risk for cancer
on the basis of local food  production. Is this approach clear and reasonable?

vi. The 1990 iED document used the 1978 USDA food consumption survey to
estimate diet fractions for locally grown foods.  Some of these estimates do not
appear reasonable. The Working Group strongly recommends trying to base these
estimates on local surveys, but could not resolve how to make these estimates if
this was not feasible.  Can the  SAB provide any further guidance on how to
address this issue?

vii.  Indirect pathways are defined as all pathways other than inhalation.  This
document addresses food chain, water, soil, and breast milk exposures. Should
any other pathways be considered?
                                   A-7

-------
APPENDIX B,     Interim Letter to Administrator (EPA-SAB-IAQC-94-009a)
                                  39

-------
  ^£^\         UNITED STATES ENVIRONMENTAL PROTECTION AGENCY
          *j                     WASHINGTON, D.C.''20460
                                                                    CFSCg CP TH
                                                                      3C:ENC5 AOVISCRY 3CASO
 EP A-S AB-lAQC'94-OC9a
 February 15,  1994

 Honorable Carol M. Browner
 Administrator
 U.S. Environmental Protection Agency
 401 M Street, SW
 Washington, DC 20460

 RE:    Draft * Addendum to the Methodology for Asses--:* Health Risks Associated with
        Indirect Exposure to Combustor Emissions*

 Dear Mrs. Browner:

        On December 3, 1993, the Indoor Air Quality/Total Human Exposure Committee (the
 Committee) of fie Science Advisory'Board reviewed the-draft document * Addendum :o the
 Methodology for Assessing Health Risks Associated with Indirect Exposure to Combusior
 Emissions* (the Addendum), Da view of pressing EPA and public concerns about
 incinerators, this interim letter was prepared to provide you with preliminary information oo
 some of the major findings of the Coir ;ttee,  A more detailed report is in preparation and
 wiH Mow soon (EPA-SAB-IAQC-94-009).

       The assessment of risks from combustors entails a complex range of issues, including
 many different Muds of combustioa devices and raw materials, direct and indirect exposure
 routes, and coccerna regarding transportation and disposal of raw materials and combusuoa
     It Is thus important to emphasize, at the outset, that this letSer and the report in
           address only one of these aspects, aameiy the questions surrounding indirect
 exposure assessment which are the subject of the Addendum.  Indirect exposures are- those
 that occur via.  transfer of aisftorne contaminants into water, soil and the food chain.  Direci
 airborne exposures from combustor emissions are being addressed with other metbodoiopes
 by the Agcacy and are act the subject of this letter.

       To grapple with these complex exposure pathway issues, the Agency needs, to be able
 to estimate the environmenai fate of oomb«stor emissions, and their consequent potential for
 human exposures.  Hi* dbk requires the development of models to predict accumulations of
chemical contaminants in the environment and idenojocaiioii of the chemicals,, environmental
compartments,  and exposure pathways most iiMy to be of concern so thai, appropriace actions

-------
 can be taken before there is widespread -and/or irreversible damage.  The Addendum we
 reviewed is i critical part of die Agency's effort :c a
-------
       Many of the above issues have been noted by previous SAB committees in their
review of hazardous waste and domestic waste incineration, la addition, various incineration
studies and emission data sets exist in other countries.  It is highly recommended that the
Agency compile and review these previous efforts as a way of focusing its future directions.

       In summary, the Committee is very aware of the difficulties inherent in the * state of
the science" nature of the work which the Addendum effort entails, especially when the work
must be done under the combined pressures of severely limited resources and pabilc demands
for 'something" to be done quickly.  Hie Committee, however, does not recommend the
release of dbe Addendum as an 'EPA Methodology* due to the substantial scientific
uncertainties in the models and the absence of information  in the Addendum concerning those
uncertainties and limitations.
                                      Sincerely,
                 Maonos
Executive Committee
Science Advisory Board
                               .
                            Chair
 r
 5r. Joan M. Daisey, Chair
• Irtdooc Air Qya&y/TotaiHumari Exposure
 Cocimirtee
 Science Advisorv Board

-------
APPENDIX C     Miscellaneous Comments and Specific Corrections and
                   Suggestions for Clarification of Text

Page  1-2: last sentence should read "...to determine air concentrations..." instead
       of "to determine  an air concentrations,.,'1

Page  1-3: third line should read "If one were,,." rather than "It one were,.,."

Page  2-1: rephrasing the statement "...potential number of health effects  cases..."
       can be  more clearly stated  as  "...estimated number of excess cancer cases..."
       since it refers to carcinogens.

Page  2-1: Emphasis is placed on defining the boundaries for the exposed
       population.  Are  food distribution patterns such that regional or even
       national level exposures should be anticipated? As the number of
       incinerators increases^ how would the general population R's level of
       exposure be affected?

Page  2-5: the term "non-negligible" should be defined  (e.g., exposure less than
       the RfD),

Page  2-7: typo on first line of text (population density).

Page  2-8: It is unclear whether individual  or group risk is of interest. Although
       addressed subsequently, some clarification would be  appropriate at this
       point.

Page  2-8: A statistically-based survey is recommended. This recommendation,
       while well intended, may not be practicable.  How much data would be
       needed  to identify deviations from larger survey databases.  Would it be
       feasible to characterize all segments of the population  including high-risk
      groups?

Page 2-9: The "other residents" should be identified. Listing as group I = 1 to N
       would be preferable.

Page 2-10: 6th line from the top "individual  special  interest groups"  probably
      mean "subgroups of special interest"  in  this context. Also, last sentence on
                                       40

-------
      first paragraph should  probably read "Judgment will be needed  to
      decide. „

Page 2-11: 6th line from the top "characterized"; al  14th line  from the top
      ",..may  change in the future.,,"  and "...may  be performed..." instead of
      "...may  be desired..."

Page 3-2: Add a statement to first paragraph under 3.2 "focus on chemicals that
      are potentially toxic to humans	are relatively long-lived in the
      atmosphere, and therefore have a definite propensity for bioaccumulmting,"

Page 3-10, top of page;  log Know can indicate partitioning soils and sediments.
      Doesn't this also indicate lower bioavailability?

      Pages 3-10  & 3-11 - Wouldn't it be better to use as average emission
      factors, factors weighted according to operating conditions with appropriate
      weighing for higher emissions during upset and start-up conditions rather
      than simply some  arbitrary arithmetic average?

      Also, the next to last sentence on the page "For purposes of exposure
      assessment  ,..., ranges and average values  should be developed" also applies
      to section A,

      Operational Facilities, This deserves greater emphasis and should be stated
      earlier in  the  discussion rather  than at the end.

Page 3-10;   on last paragraph, third line "...stack   sampling,  analytical, and
      quality  control, quality assurance protocols...."    should  be   "...stack
      sampling, analytical, and quality control/quality assurance protocols and
      procedures..."

Page 3-10, 11: The  wording at the bottom of p. 3-10 is somewhat confusing.
      When read  carefully, it says that if you use concentrations corrected to 12%
      CO2 or 7%  O2 in  the calculation of mass emission rate, then you should use
      the corresponding volumetric flow rates corrected in the same way.  The
      words "this  calculation" on p. 3-11 could refer to any number of possible
      combinations  of concentration and flow rate.

Page 3-11: 6th line  of first paragraph "In cases where..."

                                      41

-------
 Page  3-13, paragraph 2: Combustion emissions from hazardous waste incinerators
       are likely to be only part of the concern for local communities.  Risk
       assessments for each facilities should examine probabilities of spills and
       leakage of hazardous waste as well as accidents during transport.

 Page  3-13:  Increased emissions during start-ups and shut-downs, malfunctions
       and perturbations.  Some real-world data are critically needed here.  How
       often do these things occur and how much do they contribute to total
       emissions?  The recommended default values suggest that they re  very
       important (p. 3-15).  For nearby populations, increased emissions of irritant
       chemicals are likely to  be of much more concern than  dioxin if levels
       frequently reach thresholds.  Also, levels should be kept  well below
       thresholds because such threshold  are generally based  on data from  healthy
       worker populations and the general population can be  expected to include
       significant percentages  of people who are much more sensitive, e.g., by one
       to two orders of magnitude.

 Page  3-13: Option "D" is not clear.  What emission test reports are these? If they
       are not done via EPA protocol, then how should they be interpreted?  Are
       you suggesting that a mass balance of the system is an acceptable option, for
       estimating emissions.  If so, the Committee does not agree but at a
       minimum, it should be  made  clear,

 Page  3-14: - Similarly, some  information on the removal efficiency of control
       devices overtime, as they age, is needed, as well as information on how
       frequently they  fail and what occurs when they fail

 Page  3-16, Section 3.5;  Short-Term Impacts Due to Meteorologic Conditions. It
       is stated that short-term increases  in emissions and atmospheric inversion
       conditions are unlikely  to significantly affect indirect exposure, but may be
       important to consider in evaluation of acute effects from  inhalation.  Two
       points here; (1)  Is this  treated in the guidance for direct  exposures?  (2) Are
       chemicals with irritant  and other short-term acute  effects the focus for these
       situations?

Page  3-17, first paragraph of 3.6.1:   The text  refers  to lateral and vertical
       "concentrations"; I believe "concentration distributions"  would fit better,

Page 3-18:  Suggest C rather than X be used  for concentration in the  equations,

                                      42

-------
Page 3-21, first paragraph of 3.6.7:  The text says that building wake effects are
      important if the plume height and the stack height are "greater than"
      certain values.  It should say "less than".

Page 3-21: In principle, the restrictions on algorithms from building wake effects
      under complex terrain conditions apply only to those complex terrain
      receptors located within the building cavity, not to the downwind receptors.
      In this regard, the Shulman and Scire algorithm is more compatible with
      complex terrain modeling.

Page 3-25: The algorithms described in Equations 3-5 to 3-10 do not depend upon
      the surrounding terrain.  The algorithms describe phenomena on  a  relatively
      small scale near the surface.  Perhaps  the turbulence parameters  differ in
      flat vs. complex terrain, but the implication of the statement is than the
      algorithms would depend upon the terrain.

Page 3-25, Bottom of page; Definition of Dyw units should he g/m -s not g/m  /s
      which is g-s/m .

Page 3-27, third paragraph of 3.7.2:  The text refers to "Julienne" day;  it should
      be "Julian".  Also in Table 3-5, on p. 3-28.

Page 3-30: Need to state clearly both here and at the beginning of this major
      section that vapor phase deposition is not being modeled here.  To the
      extent that it is possible, the error produced by this assumption needs to be
      discussed and put in context

Page 3-30, continuation paragraph at the top of the page:  The text recommends
      model receptor spacing out to 5 1cm, and from 10 km to 50 kin, but doesn't
      offer guidance between 5 and 10 km.  Consider 500m resolution between  1
      km and 5  km, and 1000m resolution from 5 km to 10 km.  The jump from
      200m to 1000m is too severe.

Page 3-31, continuation sentence at the top of the page: The text talks about
      airborne vapor absorption into plants, and particle deposition on to both
      plants and soils.  It is not clear why vapor uptake by soils is omitted, or for
      that matter, why uptake of both gases  and particles by water surfaces is not
      mentioned.
                                      43

-------
 Page  3-35, last paragraph:  The text recommends using the "background plus local
       source" value for the term ST for urban incinerators, instead of the "urban"
       value,  Why?

 Page  3-36, first complete paragraph, line 6:  The text includes the word
       "emission" after "partition"; it should be deleted,

 Page  3-36:  The conclusion that volatile components  are solely in the vapor phase
       should not be based solely on Henry's constant. An additional factor is the
       adsorptive properties of the particles.

 Page  3-37, 3-38, particle size distributions and table 3-7:  How are these
       measured?  If there  are not measured with a dilution stack sampler, the
       mass in particles will be underestimated and the size distributions will not
       be correct since condensation and coagulation will occur as the plume  cools
       and  becomes detected,

 Page  3-38, table 3-7, column 5: Shouldn't this be "Available surface areas," with
       units given
Page 3-39, first complete paragraph, lines 7 and 8:  The test referring to Column
      4 says "... the total mass of particles is 15 jim",  I suggest a rephrasing: "...
      the total mass of particles is of this size", which meshes with the earlier
      sentences.

Page 3-39, first complete paragraph, line 20, first word:  "Contaminate" should be
      "contaminant",

Page 3-40: What is considered an "adequate general vicinity" when using off-site
      meteorological data? Should some guidelines be provided to the assessor?

Page 4-1:   For homes located near incinerators, some indoor dust will be
      incinerator particles that penetrate building shells via infiltration and open
      windows and settle out as dust.  The concentrations  of pollutants may be
      greater  in these dusts than those tracked in  from outdoors because outdoors
      dusts will be diluted by soils.  It  is not appropriate to assume that deposited
      dusts in nearby homes have the same  concentrations of contaminants as the
      top 1 cm of soil,
                                      44

-------
Page  4-2, Conclusions/Recommendations: The first sentence is not clear,
      specifically the term "Volatilization." Isn't "Loss of volatiles due to
      deposition" meant here?

Page  7-1: Dermal exposure from contaminated soil.  Is this ever significant? In
      the IED document,  p, 7-11, even if AP-1, DDI, is only 1 pg/kg for BaP in
      soil at a concentration of ~ lxl(T9 mg/g  (lxlO_6p.g/g) = 1 pg/g soil.  At
      what  kinds of concentration would this be important, compared to ingestion
      or normal background? The  calculation suggests that this pathway would
      only he important if soil levels reach the order of micrograms of BaP per
      gram  of soil. Are soil levels likely to reach this?

      Also,  is there any evidence that trace metals can be absorbed via the skin?

Page  9-1:  The input requirements for the calculations are very substantial and
      the calculations extensive:  Some guidance is needed regarding situations in
      which these calculations are needed. Is there  good experimental evidence
      that supports the accuracy of that model?

Page  11-2:  What kinds of emission rates are needed to get sufficiently high
      concentrations of something like dioxin into water and then into fish and
      finally into people so that levels are sufficiently  high to matter?

Page  15-2:  On the inhalation dose, this assumes that people are outdoors 24
      hours a day. In actuality, they spend most of their time indoors.  Buildings
      provide varying degrees of protection. For homes near incinerators, this
      might make a difference of an order of magnitude in estimated exposures
      for some chemicals.  Why isn't this taken into consideration?

Page  15-8:  The discussion on mass balance is welcome.  It should be included in
      the discussions elsewhere in the Addendum,

The citation to Junge is not in the reference list.

Page 5-13: Why assume that free ranging chickens (or chickens raised in an open
      coop)  have  a lower soil intake in their diets than pigs?  One might assume
      instead that chickens consume even more soil  since they peck directly at the
      ground and need  to ingest soil particles and very small pebbles to aid in the
      breakdown of kernels in their stomach.  Chickens are also raised for their

                                      45

-------
      eggs. Is there transfer to and accumulation of contaminants in eggs?
      Ignoring a pathway component because there is no data available cannot be
      easily justified.  If the data are not available it should be obtained,

 Page 15-15; Characterization of uncertainty. Some of the uncertainties are
      qualitative in nature, i.e., based on assumption or limited information  about
      the population.  These are often more important than those that are
      evaluated with modeling.  This should be discussed,

 Chapter 2, Human Exposure Scenarios; Regarding Study Area,  Based upon
      Exposure Potential, not concentration Potential,  Therefore, impacts >50
      km could be important for some IE situations*

      Certain sections of the Addendum are not re viewable as they  stand. They
 should either be deleted or repaired.  For instance, the sections on comparing
 model results for PCDD/PCDF  with rural "background" levels in various
 environmental media (pages 5-10, 15-11,12) contain neither sufficient references
 nor information to make any reasonable judgement as to their validity.  These
 comparisons begin with a "profile of air concentrations  crafted to be typical of
 rural environments" (p, 5-10),  Concentrations of PCDD and PCDF were estimated
 from  these data using the Addendum's models.  These results were compared with
 "background" levels measured in beef. Given the persistence of PCDD and PCDF
 in the environment, this  is  not  a completely unreasonable idea.  However, such
 comparisons are not reasonable unless we have some notion of variability.  Where
 are the  error bars on these averages? Furthermore, a major problem with this
 approach is ensuring that the data are comparable (see Webster and Connett,
 1990). This question cannot be evaluated with the supplied data and lack of
 references (For various reasons, most ambient air samples  are taken in urban
 areas.  Comparisons of such values with concentrations in beef grown  on feed
 from rural  areas may bias the results.  This is why study of a specific setting is
probably better,  even though field data can be difficult to use under  the best of
circumstances).
                                     46

-------
 APPENDIX D     Earlier SAB Reports Concerning Hazardous and Domestic
                  Waste Incineration,
 U.S. EPA/SAB, Report on the jncmeration of Liquid Hazardous Wastes, by the
 Environmental Effects, Transport and Fate Committee of the Science Advisory
 Board, April 5, 1985,

 U.S. EPA/SAB, Review of Technical Documents Supporting Proposed Revisions to
 EPA Regulations for the Disposal/Reuse of Sewage Sludge  Under Section 405(d) of
 the Clean Water Act, Report of the Environmental Engineering Committee, SAB-
 EEC-87^015, January 1987.

 U.S. EPA^SAB, EPA'S Risk  Assessment Methodology for Municipal Incinerator
 EmisdonsjJfeyL Findings and Conclusions. Report of the Municipal Waste
 Combustion Subcommittee of the Environmental Effects, Transport and Fate
 Committee of the Science Advisory Board, SAB-EETFC-87-027, April 9,  1987.

 U.S. EPA/SAB, Reviewofj^_^unicipial_Wast^_CQmbus.tiQn Research Plan. Report
 of the Municipal Waste Combustion Subcommittee of the Science Advisory Board's
 Environmental Effects, Transport and Fate Committee, SAB-EET&FC-88-023,
 Final Report, April 11, 1988.

 U.S. EPA/SAB, Eya1ua|ion^f.S^ie,|itifl^Igjue^R^lat6d to Municipal Waste
 Combustion. Report of the Environmental Effects, Transport and Fate Committee
 of the Science Advisory Board, SAB-EETFC-88-25, Final Report, April 26, 1988.

 U.S. EPA/SAB, Review of Proposed Sewage Sludge Incineration Rules (40 CFR
 Parts 257 and 503). Report of the Municipal Sludge  Incineration Subcommittee of
 the Environmental  Engineering Committee, EPA-SAB-EEO89-Q35, September 20,
 1989.

 U.S. EPA/SAB, Review of OSW Proposed  Controls for Hazardous Waste
 Incineration Jro^uctsjo^                    Report of the Products of
 Incomplete Combustion Subcommittee of the Science Advisory Board  , EPA-SAB-
EC-90-OG4, January 1990.
                                    47

-------

-------