United States
Environmental Protection
Agency
Office of Research and
Development
Washington DC 20460
EPA/630/R-96/001
May 1996
Report on the U.S. EPA
Technical Workshop on
WTI Incinerator Risk
Assessment Issues
RISK ASSESSMENT FORUM
-------
EPA/630/R-96/001
May 1996
REPORT ON THE U.S. EPA TECHNICAL WORKSHOP ON
WTI INCINERATOR RISK ASSESSMENT ISSUES
Risk Assessment Forum
U.S. Environmental Protection Agency
Washington, DC 20460
Printed on Recycled Paper
-------
NOTICE
Mention of trade names or commercial products does not constitute endorsement or
recommendation for use. Statements are the individual views of each workshop participant; the
statements in this report do not represent analyses or positions of the Risk Assessment Forum or the
U.S. Environmental Protection Agency (EPA).
This report was prepared by Eastern Research Group, Inc. (ERG), an EPA contractor (Contract
No. 68-D5-Q028), as a general record of discussions held during the Technical Workshop on WIT
Incinerator Risk Assessment Issues. As requested by EPA, this report captures the main points and
highlights of discussions held during the plenary sessions and includes brief summaries of the work
group sessions. The report is not a complete record of all details discussed, nor does it embellish,
interpret, or enlarge upon matters that were incomplete or unclear. In particular, the five work group
summaries were prepared separately by the work group leaders (with, or without help from group
members) based on their groups' discussions. Thus, the recommendations of the groups might differ
slightly. ERG did not attempt to harmonize all the recommendations.
-------
CONTENTS
Page
Foreword v
SECTION ONE—OVERVIEW ; .1-1
General Summary 1-1
History of EPA Risk Assessment Activities for the WIT Incinerator 1-5
SECTION TWO—CHAIRPERSON'S SUMMARY OF THE WORKSHOP 2-1
Dr. Thomas McKone
General Summary 2-1
Summary of Recommendations on Accident Issues 2-4
References 2-6
SECTION THREE—WORK GROUP SUMMARIES 3-1
Combustion Engineering Work Group 3-1
Dr. Barry Dellinger
Air Dispersion/Deposition Modeling and Accident Analysis Work Group 3-14
Dr. Walter Dabberdt
Exposure Assessment Work Group 3-22
Dr. George Fries
Toxicology Work Group 3-28
Dr. Mary Davis
Ecological Risk Assessment Work Group 3-36
Dr. Glenn Suter II
SECTION FOUR—HIGHLIGHTS FROM COMMENTS 4-1
Peer Reviewers' Preliminary Comments 4-1
Observers' Comments 4-4
APPENDIX A—WORKSHOP AGENDA A-l
APPENDIX B—REVIEWER LIST .. B-l
APPENDIX C—CHARGE TO REVIEWERS C-l
iii
-------
APPENDIX D—PREMEETING COMMENTS D-l
Combustion Engineering D-l
Dr. ELmar Altwicker ».. D-3
Dr. Barry Dellinger D-ll
Dr. Randy Seeker D-13
Air Dispersion/Deposition Modeling and Accident Analysis D-17
Dr. Walter Dabberdt D-19
Dr. Mark Garrison D-23
Dr. Halstead Harrison D-43
Dr. Jerry Havens D-55
Dr. Geoffrey Kaiser D-63
Dr. Robert Meroney D-85
Exposure Assessment D-91
Dr. James Butler D-93
Dr. George Fries D-99
Dr. Thomas McKone D-105
Toxicology D-115
Dr. George Alexeeff D-117
Dr. Mary Davis D-145
Dr. Thomas Gasiewicz D-149
Ecological Risk Assessment D-167
Dr. Peter deFur D-169
Dr. Km Kosalwat D-173
Dr. Steven Peterson '....- D-177
Dr. Glenn Suter D-187
APPENDIXE—REVIEWER WORK GROUP ASSIGNMENTS E-l
APPENDIX F—FINAL OBSERVER LIST F-l
APPENDIX G—PRINTED MATERIALS DISTRIBUTED BY OBSERVERS G-l
APPENDIX H—WRITTEN STATEMENT SENT AFTER THE WORKSHOP H-l
IV
-------
FOREWORD
This report presents information and materials from a peer review workshop organized by
EPA's Risk Assessment Forum for Region 5 and the Office of Solid Waste and Emergency
Response. The meeting was held in Washington, DC, at the Holiday Inn Georgetown on
January 11,1996. Due to severe weather conditions, the meeting was compressed from the
planned two days to a single long day. The subject of the peer review was a draft document
prepared by Region 5 assessing risk at an incinerator operated by Waste Technologies Industries
(WIT) in East Liverpool, Ohio.
The peer review continued a process begun in 1993, when the Risk Assessment Forum
held a workshop to review the project plan for the WTI incinerator risk assessment. In that
workshop, 13 peer reviewers divided into work groups to discuss four major aspects of the
project plan: combustion engineering, meteorology/air dispersion, exposure assessment, and
toxicology. The workshop was attended by more than 100 observers. Workshop participants
recommended that EPA expand the scope of the planned assessment to include more facility
performance data, use additional computer models, include a screening level ecological risk
assessment, and provide a comprehensive analysis of accident scenarios.
In 1994 and 1995, EPA conducted the WTI incinerator risk assessment now under review.
To reflect the larger scope of work recommended by participants of the first workshop, EPA
modified the peer review workshop format as follows: EPA expanded the scope of the air
dispersion work group to cover deposition modeling and accident analysis, added a fifth work
group on ecological risk assessment, and increased the number of peer reviewers from 13 to 19.
Most of the 13 reviewers of the project plan were able to participate in the peer review of the
risk assessment. In conducting the peer review, EPA sought comments on the technical
accuracy, completeness, and scientific soundness of the WTI incinerator risk assessment. EPA
will consider these comments in revising the assessment, which in turn will be used to set final
permit conditions for the WTI facility.
This report summarizes the discussions that took place at the peer review workshop. The
report opens with an overview of the workshop and a history of EPA's WTI incinerator risk
assessment activities (section 1), then presents the chairperson's summary (section 2) and the five
work group chairs' summaries (section 3). The body of the report ends with highlights of the
peer reviewers' preliminary comments and of the comments offered by workshop observers
(section 4). Appendices to the report present the workshop agenda, a list of the peer reviewers,
their charge, their premeeting comments, and their work group assignments (appendices A-E) as
well as a list of observers and printed materials distributed by observers (appendices F-G) and
the written version of a comment that a citizen intended to offer at the workshop but was unable
to do so due to severe, travel-hampering weather conditions (appendix H).
William Wood, Ph.D.
Executive Director
Risk Assessment Forum
-------
SECTION ONE
OVERVIEW
GENERAL SUMMARY
The workshop provided a forum for the expert peer review panel to discuss the technical
accuracy, completeness, and scientific soundness of the draft WTI incinerator risk assessment.
The reviewers were in general agreement on the overall quality of the assessment and
contributed useful suggestions for moving the process ahead to finalize the document.
Overall, comments on the draft WTI incinerator risk assessment were favorable. Indeed,
throughout the workshop, as the expert peer reviewers discussed the assessment as a whole and
specific parts of it, workshop participants repeatedly prefaced suggestions for improvement with
praise for the overall thoroughness, quality, and integrity of the assessment. Noting that they
had been quite critical of the draft project plan for the assessment, the peer reviewers stated that
by contrast they were very impressed with the thoroughness, organization, and clarity of the draft
assessment—and with the seriousness and faithfulness with which EPA had followed the
comments and recommendations of the project plan peer reviewers. Their most substantive
comments pertained to three topics (accident scenarios, cumulative risk, ecological risk) that
were not covered in the initial project plan for the assessment and thus had not benefitted from
previous review. The peer reviewers described most of their other comments as questions of
clarification or as other minor issues not likely to affect the overall results of the assessment.
Members of the Combustion Engineering Work Group, for example, praised EPA's work
on WTI stack emissions, noting that the Agency's efforts to determine the composition of the
waste feed could be labeled heroic. Due to the basic soundness of the analysis, they focused on
attempting to trace the progression of information on chemicals of concern from the waste feed
through emissions estimation (see Combustion Engineering Work Group summary and diagram
in Section 3). The main question generated by this exercise sought information on why EPA had
not prorated measured products of incomplete combustion (PICs) to compensate for unmeasured
PICs, as had been recommended by the peer reviewers of the project plan. The peer reviewers
1-1
-------
recommended that EPA clarify the explanation in the assessment that doing so would cause
chronic effects to be overestimated and acute effects to be underestimated or prorate the
measured PIC values as suggested.
Members of the Air Dispersion/Deposition Modeling and Accident Analysis Work Group
also praised EPA's work, asserting that the assessment does a good job estimating the dispersion
of emissions from the WTI facility, at least during routine operations. They suggested that EPA
work with the CALPUFF model or other models to characterize dispersion of ordinary emissions
under stagnant air conditions, which could magnify air quality problems, and to better
characterize accidental emissions. They also suggested that EPA consider whether additional
release scenarios (e.g., release following pressurization of wastes in containers) might be
important and that EPA explain mitigation measures more clearly. Noting that they had been
unable to trace the process themselves, they also recommended that EPA clarify how it had
obtained the entries presented in the following summary tables in Volume VII:
• Table VIH-l: Probability/Severity Matrix—Typical Meteorological Conditions.
• Table VQI-2: Probability/Severity Matrix—Conservative Meteorological
Conditions.
• Table VHI-3: Probability/Severity Matrix—Calm/Inversion Meteorological
Conditions.
EPA's analysis of accident scenarios engendered lively discussions not just in the Air
Dispersion/Deposition Modeling and Accident Analysis Work Group, but throughout the
workshop. Although some peer reviewers said that the analysis seems reasonable, other peer
reviewers and two workshop observers (see section 4) contended that it is incomplete and
excessively qualitative. In addition to suggesting that EPA use the CALPUFF model to improve
its estimation of accidental emissions, peer reviewers suggested that EPA more clearly address
the location of the nearby elementary school and consider using a compound other than acetone
for future modeling purposes because acetone has been delisted. Several peer reviewers and
workshop observers also expressed reservations about the use of immediately detrimental to life
and health (IDLH) values; because these values pertain to healthy adult workers exposed to
contaminants for up to 30 minutes (during which time they will have donned protective gear or
otherwise protected themselves from further exposure), the use of IDLH values might
1-2
-------
underestimate the risks that accidents pose to a residential population. Members of the Toxicity
Work Group suggested using level of concern (LOG) values instead; they referred EPA to
George Alexeeff s premeeting comments on this topic for further information.
Members of the Exposure Assessment Work Group commented that the exposure
assessment in the WT1 incinerator risk assessment is among the most comprehensive they haye
seen and that EPA did a good job addressing the recommendations of the project plan peer
reviewers. They noted that EPA had not addressed house dust as had been recommended, but
they contended that a rigorous quantitative analysis is not needed because house dust is unlikely
to be an important exposure pathway. They suggested discussing house dust in conjunction with
soil exposure pathways among children. They also recommended that EPA conduct further work
to determine cumulative exposures (i.e., to the combination of WTI emissions and background
contaminant levels) and, on a more long-term basis, to develop more refined methods and
models.
Like the issue of possible accidents, the issue of cumulative risk arose several times
during the workshop. The general consensus was that EPA should address this issue further to
determine whether WTI-related exposures have the potential to increase total exposures to
unacceptable levels. Members of the Toxicology Work Group suggested that simply examining
how WTI-related exposures compare to background exposures might be sufficient to accomplish
this goal.
Members of the Toxicology Work Group also offered a number of specific comments and
suggestions related to the human health risk assessment. For example, they suggested that EPA
use California EPA's slope factor to model lead toxicity, add a table of noncancer endpoints
(while noting in the text that cancer endpoints are more sensitive), discuss the contribution of
exposure to metals in breast milk to total metal exposures, assess exposures in the subpopulation
of individuals who both work in and live near the WTI facility, provide a more quantitative
analysis of uncertainty, and include a discussion of uncertainties related to data gaps.
Finally, peer reviewers discussed EPA's screening level ecological risk assessment
(SERA), the third of the three topics in the assessment that peer reviewers felt needed
substantive work. At the most basic level, peer reviewers were unclear about the goals and
1-3
-------
purpose of the SERA. Members of the Ecological Risk Assessment Work Group, for example,
commented that EPA apparently included a permit limit scenario because it would be needed to
design permit limits for metal emissions. This led the peer reviewers to wonder whether EPA
conducted the SERA to support the setting of sufficiently conservative permit limits rather than
to generate information that regulators and the public could use to understand the ecological
risks associated with the WIT facility. The peer reviewers recommended that EPA clarify the
goals of the SERA and conduct further work if needed to accomplish these goals. They also
suggested that EPA address the issue of cumulative ecological risk and include provisions for
monitoring in any facility permits (although distinguishing between background and WTI
contributions might be difficult).
Members of the Ecological Risk Assessment Work Group also noted that the problem of
data gaps is even greater in ecotoxicology than in human health toxicology. They recommended
that EPA discuss the implications of data gaps more systematically in future versions of the
assessment. They also wondered whether data from recent WTI plant operations might permit
validation of the test burn data used for the SERA. Acknowledging that all ecological risk
assessments suffer from a lack of established data, tools, and procedures, the peer reviewers
noted that research is needed to provide the infrastructure necessary for improved ecological risk
assessments in the future.
At the conclusion of the workshop, the peer reviewers attempted to sum up by asking two
questions:
• If fully implemented, would any of the recommendations of the work groups
change the results of the assessment? The peer reviewers concluded that
recommendations in three areas—accident scenarios, cumulative risk, and
ecological risk—have the potential to change some of the results of the
assessment. Except for those three areas, the. assessment is adequate In its
current form.
• Should the risk assessment present information on conditions that increase the
risks associated with operating the facility and recommend mitigation measures?
This question related to the line between risk assessment and risk management.
The peer reviewers concluded that the WH incinerator risk assessment should
provide information about the contribution of various conditions to the predicted
risks (information that will help risk managers make informed decisions), but it
should not recommend mitigation measures (because those are risk management
decisions based on social and policy factors as well as scientific/technical factors).
1-4
-------
For example, the assessment could state that certain weather conditions
substantially increase the risks associated with operating the WIT facility, but it
should not make recommendations as to operations of the facility during such
weather conditions.
HISTORY OF EPA RISK ASSESSMENT ACTIVITIES FOR THE WTI INCINERATOR
The WTI incinerator, the subject of the risk assessment reviewed at the workshop, is
located in East Liverpool, Ohio, across the Ohio River from West Virginia and about a mile and
a half west of the Pennsylvania border. A permit to store and treat hazardous waste regulated
under Subtitle C of the Resource Conservation and Recovery Act (RCRA) was issued to the
WIT facility on June 24,1983. Because the original permit was appealed, however, it did not
become effective until January 25,1985. On November 30,1990, WTT began constructing the
incinerator.
Due to intense interest in the WIT facility in East Liverpool, EPA's Region 5 initiated a
risk assessment in 1991 before authorizing interim operations. Because site-specific information
was unavailable, the risk assessment was conducted using regional meteorological data and stack
emissions data from other incineration facilities. In accordance with EPA's Office of Solid
Waste guidelines, only the direct inhalation exposure pathway was assessed. This initial risk
assessment, conducted by a contractor and referred to as Phase I risk assessment activities, was
completed and made available to the public in July 1992. The risk assessment indicated that
predicted inhalation exposure levels were below the level of concern. Subsequently, EPA's
Office of Health and Environmental Assessment conducted an additional screening level analysis
of potential cancer risks from dioxin stack emissions, generating preliminary risk estimates for
four exposure scenarios, each of which included indirect exposures through the food chain.
Since these initial, Phase I risk assessments were completed, a full year of onsite
meteorological data as well as WIT incinerator-specific waste composition and emissions data
from trial burns and performance tests have been collected. With input from scientists in a
number of EPA offices, Region 5 prepared a project plan for a Phase II risk assessment, which
was to include a multipathway assessment using the new data. The project plan was peer
reviewed in December 1993 under the auspices of EPA's Risk Assessment Forum.. In conducting
1-5
-------
the Phase n risk assessment in 1994 and 1995, EPA conscientiously tried to follow the
recommendations of the peer review panel to include facility performance data, use a variety of
computer models, include a SERA, and provide a comprehensive analysis of accident scenarios.
Accordingly, the assessment consists of three separate analyses: a human health risk assessment,
a SERA, and an accident analysis.
Since the completion of Phase II risk assessment activities in 1995, EPA's Risk
Assessment Forum has been organizing a peer review of the draft WTI incinerator risk
assessment to obtain comments on its technical accuracy, completeness, and scientific
soundness—and to obtain comments on whether and how well the Agency succeeded in
implementing the recommendations of the project plan peer review panel. EPA will consider
these comments in revising the assessment, which will then be used to set final permit conditions
for the WH facility.
1-6
-------
SECTION TWO
CHAIRPERSON'S SUMMARY OF THE WORKSHOP
Thomas McKone
School of Public Health
University of California at Berkeley
Berkeley, CA
GENERAL SUMMARY
On January 11,1996, EPA held the second of two external peer reviews of documents
related to an assessment of risks associated with the WIT incinerator in East Liverpool, Ohio.
The purpose of this workshop was to bring together a team of scientific experts to comment on
the draft WTI incinerator risk assessment. The assessment was based on a project plan
developed in 1993 and subjected to peer review at a similar workshop held in December 1993.
Most members of this 1996 peer review (Elmar Altwicker, James Butler, Walter Dabberdt, Mary
Davis, Barry Dellinger, George Fries, Thomas Gasiewicz, Halstead Harrison, Pirn Kosalwat,
Thomas McKone, Randy Seeker) also participated in the 1993 review.
The peer review of the draft WIT incinerator risk assessment was carried out in stages. A
few months before the workshop, reviewers received the full assessment report and were assigned to
one of five work groups, each focusing on a specific technical aspect of the assessment. In
December 1995, each reviewer submitted written premeeting comments to his/her work group chair
and to the workshop chair. In January 1996, these comments were circulated to all members of the
peer review panel. At the January 11 workshop, the peer reviewers met in plenary and work group
sessions to discuss the draft assessment and prepare this summary report. In addition to the peer
reviewers, some 30 observers also participated in this public meeting. They included EPA staff and
consultants, citizens from East Liverpool and other communities, representatives of the press,
employees of WTI, and representatives of other industries.
The U.S. National Research Council (1982,1994) divides the practice of risk analysis into
two substantially different processes: risk assessment and risk management. The goal of risk
assessment is to estimate a risk; that is, to produce a specific risk value and explain the precision
2-1
-------
of this estimate. The goal of risk management is-to establish the significance of the estimated
risk, compare the costs of reducing this risk to the benefits gained, compare the estimated risk to
;
the societal benefits derived from incurring the risk, and implement any political and institutional
processes needed to reduce the risk. As a risk assessment document, the main goal of the draft
WII incinerator risk assessment is to give the public and decision-makers adequate information
about the nature and likelihood of any health detriment associated with the WIT facility.
Prescriptions for technological, social, legal, or political control actions are risk management
decisions and are not explicitly discussed in the draft assessment. Neither the draft risk
assessment nor this workshop considered risk management issues.
The draft WII incinerator risk assessment is a large, comprehensive document consisting
of several volumes:.
Volume I: Executive Summary
Volume II: Introduction
Volume III: Characterization of the Nature and Magnitude of Emissions
Volume IV: Atmospheric Dispersion and Deposition Modeling
Volume V: Human Health Risk Assessment Evaluation of Potential Risks from
Multipatfaway Exposure to Emissions
Volume VI: Screening Ecological Risk Assessment
Volume VII: Accident Analysis: Selection and .Assessment of Potential Release
Scenarios
To cover the extensive volume of material in the draft assessment,, the peer review panel
divided into five work groups focusing on the following a,reasi
• Combustion engineering (focusing on emissions)
• Air dispersion/deposition modeling* and accident analysis
• Exposure assessment
• Toxicology
• Ecological risk assessment
2-2
-------
The peer reviewers also met in plenary sessions to discuss cross-cutting issues, such as
accident analysis results. During the first plenary session, the peer reviewers addressed the
following general issues identified in the premeeting comments:
• The scientific quality of the draft WIT incinerator risk assessment is .considerably
better than that of the 1993 risk assessment plan.
« In preparing the assessment, EPA addressed the major recommendations of the
1993 peer review panel.
• Some important uhcertairities (the confidence with which accident scenarios and
impacts can be specified; the influence of data gaps on emission and health
impact estimates, the quantification of noncancer impacts of dioxin-like
compounds) need to be addressed prior to closure.
• Emissions of WIT chemicals of concern from other proximate industrial sources
and even from local residential combustion sources should be factored into the
assessment to facilitate the calculation of better cumulative dose and impact
estimates and the development of validation studies.
» The goals and conclusions of the SERA are vague.
• In the accident analysis, the accident scenarios are incomplete and their
contribution to information on possible health detriment is inadequately
addressed; also, accident-related risks are not fully quantified.
After the first plenary session, the peer review panel broke into work group sessions to
discuss specific areas of the assessment and prepare the work group summary reports included in
section 3 of this workshop report. The reviewers agreed that the work group reports should:
• Focus on scientific issues, not issues of policy.
• Collect and summarize the opinions of the experts (i.e., consensus was not
necessary).
• Make recommendations that EPA can use to finalize the assessment.
• Identify the most important uncertainties and information gaps in the
assessment—those likely to alter the assessment's conclusions on the likelihood of
health detriment.
• Consolidate the reviewers' comments, present these comments as concise problem
statements, and identify the likely impact of these problems on the results of the
assessment.
2-3
-------
After the work groups completed their tasks, the peer review panel met again in a
plenary session for presentation of reports from the work group leaders and discussion of general
recommendations. For the most part, the work group meetings supported and better defined the
issues identified in the reviewers' premeeting comments, as summarized above. The issue of
accidents, in particular, sparked extensive discussion (see below). Some peer reviewers
recommended that a WTI site visit be part of any additional peer review of WTI-related EPA
documents.
As noted by several peer reviewers, the draft WTI incinerator risk assessment is one of
the most extensive and comprehensive risk assessments ever compiled for a stationary
combustion source. The assessment goes to great lengths to address regulatory requirements and
EPA guidance. Although addressing such issues is necessary, the often neglected goal of the risk
assessment process is to address the questions of how precisely we can estimate a source's "true"
impact on public health and how well we can address and answer the concerns of affected
communities. Certainly, these are challenges that remain for future risk assessments.
SUMMARY OF RECOMMENDATIONS ON ACCIDENT ISSUES
The peer reviewers had several general and specific concerns related to characterizing the
occurrence and impacts of accidents. Because accident issues cut across all aspects of the risk
characterization, a separate work group was not assigned to this topic. As a result, comments
regarding accidents appear throughout the work group summaries (see section 3). These
recommendations are briefly summarized below:
• The accident analysis does not address all potentially important accident
scenarios. For example, pressurized jet releases from the incinerator containment
might occur and result in aerosol formation due to mixing of chemicals or heating
by fire. Although the accident analysis in the draft assessment provides a useful
beginning for assessing nonroutine emissions and accidents, it is not well
developed and it lacks the precision and depth needed for reliable estimates of
impact.
• The predicted effectiveness (or failure) of mitigation measures needs to be more
clearly addressed.
2-4
-------
The accident analysis does not adequately communicate the expected value of
accident impacts, nor does it adequately explain the reliability of the estimates
given. Whereas the assessment conveys cancer risk estimates in terms of the
likelihood of detriment (i.e., less than one chance in a million per lifetime), it
conveys accident severity and consequence information in vague terms (e.g.,
"likely" or "unlikely" events, "moderate" to "catastrophic" consequences). These
estimates should be quantified more rigorously.
The accident scenarios do not characterize in any quantitative fashion the
sequence of events that might result in an accident or the likelihood of these
events. The absence of this information hampers use of the accident analysis as a
guide for planning to reduce the incidence and consequences of accidents in an
efficient and cost-effective manner.
The IDLH values used in the accident analysis are designed to provide short-term
protection to healthy workers and do not account for the greater variation in
sensitivity likely to exist in a non-occupational population that includes children.
The peer reviewers recommend that some other measure of accident health
impacts be considered. It was noted that the American Industrial Hygiene
Association's Emergency Response Planning Guidelines (ERPG) levels would
probably have been more appropriate than IDLH values for characterizing the
severity of accident consequences. :(^
The accident analysis would be strengthened by an examination of the safety
record of other hazardous waste facilities such as the Biebesheim facility in
Germany, which is similar to WIT and has apparently reported two release
incidents.
The accident analysis focuses on the acute impacts of the accident scenarios, but
does not address how chemical exposures during an accident could impact an
individual's lifetime exposures to chemicals from the WIT facility.
The atmospheric dispersion analysis used in the accident analysis should explicitly
report the chemical concentration ranges expected to occur at the East
Elementary School under the various accident scenarios.
The chemical release model for accidental fires should be changed to include the
same chemicals and relative emission rate estimation procedures used for stack
emissions. In addition, an improved method for calculating the total emissions
rate from the fire should be developed.
The dispersion modeling performed for the accident scenarios should be re-
examined in light of the peer reviewers' recommendation that calm/stagnant
conditions be reanalyzed with a more appropriate data set in the CALPUFF
model.
The model used to estimate the rate of chemical evaporation from spills is not
appropriate for calm conditions. More appropriate models are discussed in the
report of the work group on atmospheric dispersion.
2-5
-------
Because this facility is located on a flood plain, the risk assessment should include
the likelihood that a flood of sufficient magnitude to inundate the facility, will
occur and that hazardous materials would be released during such a flood.
Because acetone has now been deleted from the list of toxic chemicals used for
emergency planning, the peer reviewers recommend that it not be used as a
sentinel chemical for the accident analysis.
The peer reviewers also offered one important long-term recommendation: that more
sophisticated accidental fire and chemical release models be developed. It is unlikely that such
models can be incorporated in the WIT assessment, but they would be useful for future
incinerator risk assessments.
REFERENCES
National Research Council. 1982. Risk and decision-making: Perspectives and research.
Washington, DC: National Academy Press.
National Research Council. 1994. Science and judgment in risk assessment.
Washington, DC: National Academy Press.
2-6
-------
SECTION THREE
WORK GROUP SUMMARIES
COMBUSTION ENGINEERING
Barry Bellinger, Chair
Department of Environmental Sciences and Engineering
University of Dayton
Dayton, OH
Elmar Altwicker William Randall Seeker"
Department of Chemical Engineering Energy and Environmental Research Corporation
Rensselaer Polytechnic Institute Irvine, CA
Troy, NY
The WIT risk assessment document represents a highly professional and dedicated effort
by EPA and its contractors. In 1993, the combustion engineering panel offered detailed
recommendations for improving the draft risk assessment. EPA made an exceptional effort to
follow the spirit of the recommendations and, in some instances, the Agency's efforts can be
termed heroic. The Combustion Engineering Work Group is confident that the WIT risk
assessment document (at least the part we reviewed in detail) is fair and scientifically unbiased.
EPA used a very detailed procedure to estimate possible stack emissions and did a good
job identifying and discussing uncertainties in the procedure. To our knowledge, this is the first
time a risk assessment, document has included this much detail. As a result, it substantially
improves our understanding of how to conduct these estimates. In general, the procedure
described in the draft risk assessment is a good model for developing emissions estimates for
future risk assessments. The level of detail provided in the draft risk assessment does point to
some weaknesses, however.
*Dr. Seeker reviewed the WIT incinerator risk assessment and provided premeeting comments,
but was unable to attend the workshop.
3-1
-------
The Combustion Engineering Work Group discussed six general types of emissions or
emissions issues:
• Emissions from accidental fires or other hot releases
• Gas-solid partitioning of emissions
• PIC emissions
• Dioxins/furans (PCDD/F) emissions
• Metals emissions
• Fugitive emissions
The first issue, the accidental fire release scenario, was not assigned to the Combustion
Engineering Work Group. Nevertheless, we reviewed this topic in some detail. As discussed
below, the work group believes that the procedure used to estimate emissions from an accidental
fire is largely inadequate. This is not a criticism of EPA's regulatory staff, since appropriate
models are not currently available. In fact, EPA's work clearly illuminates the deficiencies of
existing models for the first time.
Regarding gas-solid partitioning, the work group believes that the procedure used is
probably very inaccurate, but sufficiently conservative to ensure that the risk results present a
reasonable worst case. Regarding the last four emissions issues, the work group believes that a
few concerns remain, but that nothing short of a complete paradigm shift (vide infra) will
significantly affect the results of the overall risk assessment.
NEAR-TERM RECOMMENDATIONS
Some of the Combustion Engineering Work Group's recommendations should be readily
implementable and should improve the results, of this risk assessment. For the near term, we
recommend that EPA:
3-2
-------
Change the chemical release model for the accidental fire scenario to include the
same chemicals and relative emission rate estimation procedures used for the
stack emissions. An improved method for calculating the total emission rate from
the fire should be developed. Perhaps a range of overall destruction efficiencies
(e.g., 90 to 99.99 percent) could be evaluated.
Obtain actual particle size distribution data for stack emissions to improve the risk
assessment's estimates of gas-particle partitioning of PCDD/F and metals (and
other PICs). If it can be demonstrated that the currently used assumptions are
the most conservative, the requirement for additional stack sampling can be
omitted.
Adjust the estimate of normal PIC emission rates to reflect emissions during
abnormal operations (i.e., based on the percentage of operation time during which
emission violations or automatic waste feed cutoffs occur).
f
Obtain actual facility-specific sulfur dioxide (SO2) removal efficiency data over a
wide concentration range so that the SO2 surrogate will better model the behavior
of selenium (Se).
Abandon the assumption that fugitive emissions particles are the same as coal
dust in favor of using actual fugitive emissions particle characterization for the
WTI facility.
LONG-TERM RECOMMENDATIONS
Some of the Combustion Engineering Work Group's recommendations should be
implemented in future risk assessments. Over the long term, we recommend that EPA:
Develop far more sophisticated accidental fire and chemical release models.
Improved dispersion models that better reflect low-level releases and complex
terrain might also be necessary. Existing models are inappropriate for portraying
releases from a hazardous waste incineration facility.
Generate experimental data on the gas-solid partitioning of PCDD/F and other
toxic air pollutants on various types of particles. Existing modeling approaches
are very inaccurate.
Consider using carcinogenicity assay screening of incinerator effluents to assess
overall risk. This approach might be necessary to address the unaccounted-for
fraction of incinerator emissions.
3-3
-------
DETAILED COMMENTS
Accidental Fire Model
Because the WIT incinerator is located in a valley near an elementary school and
residential properties, one .of the most sensitive release scenarios may be an accidental fire.
Whereas a simple spill could result in a moderate release (i.e., due to evaporation), an accidental
fire could produce a catastrophic release. In fact, such a fire involving hazardous waste could be
the most insidious of all combustion releases. In contrast to combustion in the incinerator,
where it is controlled and results in extraordinarily efficient destruction of waste, combustion in
an accidental fire would be very uncontrolled. A raging fire with relatively high temperatures
and good waste-air mixing might produce reasonably good combustion, significant plume rise,
and thorough dispersion of toxic gases. Because chlorine and other halogens are flame
inhibitors, however, a fire consuming material containing chlorinated hydrocarbons would likely
bum much more slowly—potentially resulting in less complete combustion, vaporization of
solvents at the periphery of the fire, and less plume rise and dispersion. This is not to say that
models and experiments have shown this scenario to be tme or untrue. The issue is that we have
not adequately addressed these important safety questions.
The model used to determine the physical characteristics of a pool fire resulting from a
chemical spill is very simplistic. More detailed fire models appear to be available from the
chemical industry, but these models have not been adapted for simulating a hazardous waste fire.
The chemical release model, too, is totally inadequate. Thus, additional model development is
needed. Adapting the general procedures used to estimate stack emissions might yield a
dramatic improvement in fire emissions estimates.
The existing fire model uses empirical equations to estimate burn rate, flame
temperature, flame height, and vertical velocity. It does not account for variations in fuel
mixture or the burn properties of different fuels. The complexity of the fire model is consistent
with the chemical decomposition model for the formation of phosgene and hydrochloric acid, but
it is inadequate for use as a comprehensive release model. A fire model is needed that describes
the structure of the flame and plume in terms of a profile of flame temperature, vertical velocity,
and oxygen and fuel concentrations within the flame. Several research groups, have developed
3-4
-------
models of this type for simulating liquid and solid fuel fires; their models could be adapted for
use in characterizing a hazardous waste fire.
The chemical release model includes emissions of phosgene and hydrochloric acid based
on 30- to 40-year-old experimental studies on the burning of various chlorinated hydrocarbons.
The model does not address the individual components of the waste, their potential for
vaporization without burning, or the likelihood that they will form toxic byproducts other than
phosgene and hydrochloric acid. Accident models often consider only acutely toxic chemical
exposures. Because only limited data exist on the acute toxicity of hazardous waste combustion
byproducts, very few chemicals can be explicitly included in a fire emissions model. A prorating
approach, such as that used in the stack emissions model, might be a viable alternative.
Also, the notion that fire contributes insignificantly to exposure to chronically toxic
chemicals has not been proven. In contrast, given the magnitude of emissions in a catastrophic
event, the contribution of fire might indeed be significant. Adapting the procedure used to
estimate annual stack emissions to estimate toxic chemical emissions might yield improved
estimates of these emissions. To develop quantitative estimates, however, an improved fire
model will be needed to better estimate combustion conditions in the fire.
Better dispersion models are also needed to address the effects of local terrain and
variable plume rise. The Combustion Engineering Work Group is concerned that a slow-
burning, cool fire will produce large low-level releases whose plume rise remains close to the
ground. Under stable atmospheric conditions, the dispersion might be controlled by drainage
flow that follows the contours of the terrain. Although ISC-COMPDEP and CALPUFF are
excellent state-of-the-art dispersion models, they do not model the influence of terrain on the
local air flow. Local terrain might significantly affect short-range dispersion of the release.
Gas-Solid Partitioning
Although phase distribution of PCDD/F is important from the standpoint of risk
assessment, little seems to have been added since the initial, Phase I risk assessment. The
discussion on partitioning appears to be unchanged. The statement (Volume III, page III-ll)
3-5
-------
that "substances in the stack gas will generally be present in either the vapor phase or in the
particle phase" seems to ignore the possibility that many compounds, among them PCDD/F,
could be partitioned in both phases in the stack gas. Although the material on partitioning
appears in Volume m, it is applied only hi Volume V, Table IV-5. The numbers in that table
appear to derive from an assumption of T = T^,^ and initial concentrations in the gas phase.
As noted in the work group's December 1993 review of the risk assessment work plan,
Biddleman (1988) cites DSyR = 6.79 as an average and gives three references; he does not say
that it can be "satisfactorily estimated." In fact, McKay et al. (1982) state that it is an average
empirical value and "may be substantially in error for certain compounds." Thus, this analysis
has two problems:
• It does not clearly show how the results In Table IV-5 (Volume V) were obtained.
What was the source of the vapor pressures of the different PCDD/F congeners in
the subcooled liquid state?
• It seems to ignore chemisorption and potential differences in stack particle
properties that determine partitioning prior to emission from the stack. Although
we are not aware of any description of hazardous waste incinerator ash particles
in terms of their PCDD/F-partitioning properties, laboratory results with
municipal solid waste incinerator fly ash suggest that an idealized physical
adsorption/desorption based on a Langmuir adsorption isotherm is not tenable.
Given the particle emission rates cited (0.07 g/s) and volumetric flow rates, and
assuming fly ash surface areas of less than 10 m2/g, the surface area available in
the stack gas particulate matter would appear to be substantially greater than 10"4
cm2/cm3 (Volume III, page 11-12). WI1 stack particles should be characterized by
surface area and size distribution to permit meaningful in-stack partition
calculations. A more critical review of partitioning must be conducted.
Our understanding is that the risk assessment considers uptake of gases by plants and
animals (not terrain uptake of pollutants adsorbed on particulates) to be the primary source of
risk from WTI operations. EPA appears to have used a conservative gas-solid partitioning
factor. Considering the importance of partitioning, however, the Combustion Engineering Work
Group believes it advisable to generate experimental gas-solid partitioning data for at least
PCDD/F on various types of particles.
3-6
-------
Stack Emissions of PICs
*
The procedure used to estimate normal stack emissions was quite complex (see figure 1).
In essence, it involved:
• Compiling a list of chemicals to be fed into the incinerator.
• Applying an average destruction and removal efficiency (DRE) value to each
chemical to obtain an emission rate.
• Identifying a target list of PICs for stack testing.
, • Including emission rates for detected and nondetected chemicals.
• Combining these results to produce a single list of estimated emissions.
• Prorating the estimated emission rates of known chemicals to account for the
roughly 60 percent of the emissions that are unknown.
Although improving this procedure might be possible, the Combustion Engineering Work
Group considers the procedure satisfactory. Chemicals on the target analyte list that were not
detected in stack tests were included at one-half their detection limit or at their detection limit
for the central tendency and worst case emission scenarios, respectively." The prorating of known
emissions to account for unknown emissions is tantamount to assuming that all emissions are
equally toxic. In the absence of better data, this method is as good as any; modifying the
procedure is unlikely to have a significant effect on the overall assessment.
The work group is still concerned, however, about the nature of the 60 percent of organic
emissions tihat remain uncharacterized (see figure 2). This means that 60 percent of the total
mass of organics are uncharacterized. Although as much as 90 percent of this 60 percent might
be light hydrocarbons such as methane and ethane, more than 99 percent of the number of
organics are probably uncharacterized. The large number of uncharacterized emissions increases
the likelihood that one of the organics is a "supercarcinogen." The lack of full characterization is
exacerbated by the present practice of not testing for chemicals for which no approved EPA
method exists, even though commonly used laboratory methods can reliably detect many of the
chemicals. Even so, using the most comprehensive battery of analytical techniques would likely
characterize only a few percent more of the organic emissions. At present, the best way to assess
3-7
-------
Obtained
Common PIC
List from
Office of Solid Waste
(OSW)
±
Added
PCBs
Deleted Methods
Not Certified by EPA
f
Target List of
94 PICs
Compiled 78
Waste Profiles
+
Deleted 5% of
Minor Compounds
|
Normalized. •
to 100%
*
Normalized to
Measured Chlorine
19 PICs Detected
75 PICs Not detected
Applied Average Trial
Burn DRE to 71
Feed Components
J
Final Emissions .
List of 143 Chemicals
(Non-dioxin PICs)
High-End Estimates
^
Central Tendency Estimate
Feed Components: Average DRE
Detected PICs: 95% UCL *
Not Detected PICs: Mean
1
Feed Components: Average DRE
. Detected PICs: D.L.
Not Detected PICs: 1/2 D.L.
Gas-Solid Partitioned
Gas-Solid Partitioned
Dispersion/Exposure/
Ecological Risk Models
Known Prorated to
Account for 60%
Unknowns
Dispersion/Exposure/
Ecological Risk Models
Figure 1. Stack emissions estimation procedure.
3-8
-------
Unified Effluent Chromatogram
Fixed
Gases
Light
HCs
Volatile*
Semi-Volatiles Non-Volatiles
e.g.,
C02.CO
-99
- 0.0009
e.g.,
methane,
ethane
-0.9
e.g., carbon
tetrachloride,
benzene
% of Total Mass
-0.09
e.g.,
PCDD/F,
PNAs
- 0.009
- 0.009
of Total Number of Chemicals
- 0.09 - 0.9-9
e.g., PNAs.
polymeric soots
- 0.0009
-9-90
Figure 2. Chemical positioning of organic wastes in an incinerator.
3-9
-------
the toxicity of incinerator emissions is to conduct carcinogenicity testing. Simple Ames test
screening of mutagenicity is inappropriate because chlorinated hydrocarbons do not respond to
the test. A somewhat more complex mouse papilloma screening test could be reasonably
performed instead. This represents a shift in the risk assessment paradigm, but it is difficult to
envision a better method for including the uncharacterizcd fraction of emissions in the risk
assessment.
In estimating actual kiln emissions, a factor for process upsets should be included as a
final step in the estimation process. The percentage of operating time under upset conditions
can be estimated from continuous monitoring data, records of automatic waste feed cutoffs, and
records of emissions violations. EPA does not seem to have included this factor in the present
risk assessment, except when discussing worst case assumptions. The Combustion Engineering
Work Group believes that data from a joint National Institute of Occupational Safety and Health
(NIOSH)/EPA study on process upset emissions, although not comprehensive, can be used to
adjust normal emissions rates to reflect process upsets.
PCDD/F Emissions
The PCDD/F data (Volume I, Table m-1) are interesting. The 1994 results, including the
December 1994 performance test results made available at this January 1996 workshop, are
presented slightly differently in Tables A through D. Of particular interest is a comparison
between the February 1994 performance test (PT) and trial burn (TB). Compared to the TB,
the PT generated a higher total PCDD/F value, but a lower toxic equivalent (TEQ) value; in
addition, this PT generated the smallest value for the ratio TEQx7(PCDD/F)x, 0.0058 (see Table
B). Unfortunately, carbon injection locations, carbon injection rates, and adjustments in
electrostatic precipitator (ESP) operating parameters were withheld. This information would
have aided in the interpretation of these results.
Table C lists the particle concentrations (gr/dsef) measured during the PTs and TB.
Calculating the ratio of the average toxic equivalent, (TEQ)x, to the average particle
concentration, (PC)x, produces an interesting trend. Although these are all ECIS-runs, this ratio
declines by a factor of five between 2/94 and 12/94 (see Table D). Because all performance
3-10
-------
Table A
1994 Performance Test (FT) and Trial Burn (TB) Results
for Chlorine, PCDD/F, and TEQ
Mean W , standard deviation (s) and variability (v)1 for (PCDD/F)T and (TEQ)
Total, (PCDD/F)T (TEQ)
Type and Date
PT,2/94
TB,2/94
PT,4/94
PT,8/94
PT, 12/94
5.34
4.60
3.74
1.34
1.58
1.85
1.06
0.83
0.38
0.71
0.35
0.23
0.22
0.28
0.45
-
x
TableB
Ratio of TEQ * to (PCDD/F) *
Type and Date
PT,2/94
TB,2/94
PT, 4/94
PT,8/94
PT, 12/94
PT, 8/931
IRun 3, apparent outlier, deleted
0.0058
0.0146
0.0094
0.0127
0.0140
0.01801
0.031
0.067
0.035
0.017
0.022
0.009
0.012
0.002
0.003
0.014
0.277
0.123
0.053
0.202
0.623
Table C Stack Particle Concentrations (PC), gr/dscf
PT,2/94
TB,2/94
PT,4/94
PT, 8/94
PT, 12/94
0.0013
0.0016
0.0035
0.0018
0.0046
0.0007
0.0001
0.0013
0.0012
0.0022
0.5376
0.0000
0.3731
0.6466
0.4876
Table D Ratio of TEQ * /(PC) * , ng/m3/gr/dscf
PT, 2/94
TB, 2/94
PT,4/94
PT, 8/94
PT, 12/94
23.8
41.9
10.0
9.4
4.8
3-11
-------
parameters have not been made available, this change cannot necessarily be attributed solely to
changes in ECIS-injection quantity and location. Nevertheless, this change suggests that lumping
together PCDD/F data from all 26 runs will not help us understand the results, nor is calculating
a 95-percent upper confidence limit very meaningful. The data raise additional questions as well:
Do the average and high values (Volume m, Table m-2) represent results for repeat analyses (if
they were done)? Some of the values differ by less than 30 percent. Which of the 17 congeners
were not detected? The conclusion that TEQ is at best a weak function of the chlorine content
of the fuel seems reasonable. In addition, using the August 1993 test data represents a
conservative approach.
Metals Emissions
Although the modeling results seem impressive, some of the assumptions are unclear.
The Combustion Engineering Work Group infers that particle size has not been measured since
the March 1993 TB and that EPA's analysis assumes:
'• A large particle mode for ash particles.
• Condensation of vaporized metals forms a second (0.5-mm) mode.
• No submicron mode (a mode smaller than 1 /zm) is present initially (on a mass
basis).
Most ash size distributions appear to be monomodal (on a mass basis), but that does not
preclude the presence of a large number of submicron particles that could play a role in
nucleation/condensation of metal vapors. Thus, the apparent assumption that no submicron
particles initially exist should be justified. If all metals that vaporize subsequently condense to
form 0.5-mm particles, their density should be very different from the (typical) ash particles.
SO2 was used to model removal of Se by air pollution control devices (APCDs) because
of its similar chemistry. Although use of SO2 might be justifiable, the analysis was based on data
involving large SO2 concentrations; this would make mass transport (which is normally rate-
limiting in APCDs) less important for SO2 than for Se. As a result, additional data may be
needed to estimate Se emissions at the WTI facility. One possibility would be to operate the
3-12
-------
incinerator in a low-SO2 mode, monitor removal efficiency as a function of SO2 concentration,
and extrapolate to expected Se concentrations. Alternatively, actual Se data could be collected;
Selsun Blue could be a source of Se for such a test burn.
"Despite some modeling similarities between chromium and aluminum, the former would
seem to be a poor surrogate for the latter, since their chemistries differ.
In the past, some have argued that historical data were unavailable or insufficient to
assist in the estimation of emissions. Although the Biebesheim facility in Germany is not exactly
like the WTI facility, data from this facility have been cited hi previous examinations of PCDD/F
emissions and carbon injection and could be used for a metals emissions comparison as well.
Tillman (1994) and references cited therein should be a good starting point.
Fugitive Emissions
Under the Fugitive Ash Emissions heading, the draft WTI risk assessment mentions that
a monthly fly ash sample (1994) was analyzed for 80 organics and that none was found. The
Combustion Engineering Work Group assumes that the fly ash sample must be some sort of
composite sample. We wonder how the sample was obtained, how it was stored, and what the
detection limits of the analytical method(s) were.
The work group also questions using a coal ash emission factor of 0.107 Ib/ton and
multiplying it by 10. This does not seem justifiable. Could the estimate not be refined based on
the composition of WTI facility fly ash (some fraction of which consists of large carbon particles,
which presumably are much less dense than ash particles)? We believe that sufficient
information might be available to generate a WTI-specific emission factor, especially since the
physical and chemical composition of the fugitive emissions is known. (The work group is not
aware of specific facility data or other methods that could be used to refine the coal ash emission
factor other than through consideration of the physical and chemical composition of the fugitive
emissions.)
3-B
-------
AIR DISPERSION/DEPOSITION MODELING AND ACCIDENT ANALYSIS
Walter Dabberdt, Chair
National Center for Atmospheric Research
Boulder, CO
Mark Garrison
Environmental Services
Raytheon Engineers and Constructors
Philadelphia, PA
Halstead Harrison
Department of Atmospheric Sciences
University of Washington
Seattle, WA
Jerry Havens
Chemical Hazards Research Center
University of Arkansas
Fayetteville, AR
Geoffrey Kaiser
Environmental and Energy Group
SAIC
Reston, VA
Robert Meroney*
Civil Engineering Department
Colorado State University
Fort Collins, CO
CHARGE TO THE WORK GROUP
The following is a list of issues in the charge to the full review panel that specifically
pertain to the review conducted by the Air Dispersion/Deposition Modeling and Accident
Analysis Work Group.
General issues:
Organization of the document
" Dr. Meroney reviewed the WTI incinerator risk assessment and provided premeeting comments,
but was unable to attend the workshop.
3-14
-------
• Scope of the Executive Summary
• Consideration of the 1993 review panel's recommendations
• Major data or methodological gaps
• Recommendations for long-term study
Dispersion and deposition modeling:
• Does the draft risk assessment adequately summarize work performed in response
to the recommendations of the 1993 review panel?
• Comment on the adequacy of the sensitivity tests relating dispersion and
deposition to geophysical variables.
• Is the ISC-COMPDEP model sufficiently conservative?
• Comment on the adequacy of the CALPUFF (and INPUFF) analyses for
assessment of calms and strong inversions.
• Comment on the use of the SLAB model.
• Comment on the overall adequacy of the model sensitivity tests.
Accident analysis:
Comment on the appropriateness of the SLAB and ISC-COMPDEP models for
estimating atmospheric concentrations resulting from fires and the mixing of
incompatible wastes.
Assess the adequacy of conclusions pertaining to severity of consequences and
probability of occurrence.
Have the magnitudes and "directions" of the effects been properly and adequately
characterized?
Comment on the appropriateness of using IDLH values for characterizing the
severity of consequences.
Comment on the adequacy of assumptions of uncertainty pertaining to accident
severity and emission rates, atmospheric-concentration averaging times, and
meteorological conditions.
3-15
-------
PREMEEUNG COMMENTS
Each of the six work group members submitted comments on the voluminous materials in
the draft WTI risk assessment (see Appendix D). Although in general the comments do not
follow the outline suggested in the charge to the reviewers, they do address the essential
elements of the questions and issues raised by the EPA Risk Assessment Forum. Together with
the major recommendations offered at the January 1996 workshop (as summarized below), they
constitute the full body of this work group's .comments and recommendations.
WORKSHOP COMMENTS AND RECOMMENDATIONS
The work group focused its review on two areas: atmospheric dispersion and deposition,
and accident analysis. Note that WTI plans to operate a second incineration unit at the East
Liverpool site; impacts associated with the second unit were, not considered in the risk
assessment document nor discussed by the work group.
Atmospheric Dispersion and Deposition
The work group concluded that overall EPA developed a thorough prediction and
assessment of routine releases for the WTI facility. Recommendations from the 1993 work
group on atmospheric dispersion and deposition were taken seriously and a genuine effort was
made to address the work group's concerns.
One area for which the work group at the January 1996 meeting suggested additional
work should be performed concerned the treatment of calm/stagnation conditions. .Specifically,
the CALPUFF model analysis was limited to "simple terrain" receptors and a greatly simplified
meteorological data set due to data limitations. The work group recommends [STJ* performing
an analysis that utilizes CALPUFF with a "synthesized" calm/stagnation event. The event would
have to be synthesized based on reasonable assumptions regarding the duration and spatial
* ST indicates the short-term nature of the recommendation; L'T indicates th.j recommendation
is of a long-term nature.
3-16
-------
distribution of winds. Possibly such an event could be portrayed using a subset of the multiyear
meteorological data base available from the nearby power plant tower in conjunction with the
data periods concurrently available from that tower an&the-onsite WTI tower. The CALMET
meteorological processor or other wind flow models could be used for this analysis. The
meteorological data base comprising surface and multilevel tower observations that was compiled
for the risk assessment (together with high-resolution terrain data) would provide suitable inputs
to various diagnostic wind flow models. The gridded output fields from the diagnostic model
should then be used to provide the high-resolution meteorological data required as input to
CALPUEF (and other time- and space-variant dispersion models). Maximum chemical
concentrations predicted for the event could then be compared to concentrations predicted by
the ISC-COMPDEP modeling to better understand the impact of calm/stagnation conditions on
predicted concentrations.
In summary, the work group believes that the CALPUFF model was not used in a
manner consistent with its potential for providing enhanced realism of dispersion simulation.
The use of a zero-dimensional wind field (i.e., wind constant in time, height, and horizontal
dimension) precluded any measure of enhanced realism with CAIPUFF and does not provide a
basis for a meaningful assessment (qualitative or quantitative) of the ISC-COMPDEP modeling
results. Our recommendation [STJ is to use a realistic four-dimensional wind field over a
reasonable period of time to assess concentrations under adverse dispersion conditions, and then
to compare these results with ISC-COMPDEP. As presently constituted, the CALPUFF analysis
adds little to the overall risk assessment.
Also, the work group recommends [ST] that extended dispersion modeling performed for
the accident scenario should be re-examined in light of the CALPUFF calm/stagnation analysis.
Accident scenario concentrations should be recomputed based on the occurrence of an accident
during the meteorological event to assess whether ambient concentrations during such an event
are significantly exacerbated by accident concentrations and vice versa. Accident impacts
generally occur over a much shorter timeframe than impacts from typical air quality events
associated with routine plant operations. Also, they occur over much shorter distances than
impacts from stack releases. Thus, calculating accident concentrations based solely on the
stagnation event may not be appropriate.
3-17
-------
The following comments address the adequacy of the ISC-COMPDEP and SLAB models
for application in accident analyses. The risk assessment authors clearly state the limitations of
ISC-COMPDEP in Appendix VII-4, Volume VII: "....ISC-COMPDEP does not simulate
instantaneous or transient releases" (page 5), and only "fire scenarios are modeled using the ISC-
COMPDEP model" (page 7). Thus, the remaining question on the use of ISC-COMPDEP is
whether it is appropriate for fire scenarios. The answer is "perhaps," but a more appropriate
model for fire situations would have been CALPUEF in conjunction with a meteorological/wind
field model such as CALMET or an equivalent model. Regarding the use of SLAB, the work
group could not reach a firm recommendation because the risk assessment document does not
provide a meaningful description of the model's physics. There are hints to SLAB'S features in
the discussion on "Modeling Parameters," and based on these it would appear that SLAB is
acceptable in terms of its ability to treat transient emission conditions. Whether it can treat
transient wind conditions or regions of horizontal or vertical shear in the wind field is unclear. It
would have been helpful if the work group had been provided copies of the two cited references
to SLAB: Ermak (1990) and U.S. EPA (1993); see page 7. Also unclear is the appropriateness
of the very short duration (less than 30 minutes) of some of the accident scenarios in the context
of worst case scenarios; longer scenarios should have been considered.
The use of spatially (x, y, and z) invariant wind fields in the dispersion modeling
performed both with ISC-COMPDEP and CALPUFF/INPTJFF is a cause for concern. There
are two potentially significant consequences of this simplistic approach: (1) worst case conditions
for routine operations may not be adequately described due to the effects of recirculation
conditions occurring during multiple-day events, and (2) the direction of worst case impacts
relative to the stack (for routine operations) or a ground-level source (for accidents) will likely
not be described properly. This could lead to under- or overestimates of human health or
ecological consequences. The calm/stagnation event analysis should help address this concern.
Existing air quality data collected in the vicinity of the WTI facility should be examined
PUT] to provide some basis for comparing the relative increase in ambient chemical
concentrations due to incinerator operation. • Data from existing local, state, or federal ambient
stations should be examined to perform this assessment. The document would be strengthened
by adding consideration of the existing air quality conditions in the valley. Does the valley at
present meet EPA standards for "criteria pollutants"? How often, if at all, have exceedances
. 3-18
-------
occurred for SO2, PM10, O3, etc? What are the trends? WIT does not exist in isolation, and the
permitting process has to start at the margin, not from zero.
The WIT risk assessment also would be strengthened [ST] by providing an accounting of
the safety record at a similar hazardous waste incinerator such as the Biebesheim facility in
Germany. Some concern was expressed that the risk assessment's assumption of "one emergency
incident involving hazardous waste release for every 25 or 30 years of operation" may be
inconsistent with the two hazardous release incidents already reported at the WIT site (on
December 1993 and October 1994) and the "frequent occurrence of kiln overpressures."
Finally, the discussion of the potential health effects associated with inhalation
and paniculate matter does not provide adequate consideration of the impacts of the
additional load on respiratory function, particularly in asthmatic and elderly individuals. The
discussion should be expanded [ST] to include consideration of how the increased emissions will
affect respiratory function. This could be done by comparing the estimated increments to
published reports on the effects of pollution episodes on respiratory function (Dockery et al.,
1984; this is the Steubenville TEAM study). The treatment of noncancer effects through the use
of the Hazard Index does not provide adequate discussion of noncancer health effects. The
atmospheric work group recommends [ST] that the present discussion in the risk assessment be
expanded to include an analysis of the likely range of risks associated with noncancer health
effects.
Accident Analysis
The work group found that potentially important release scenarios have not been
considered. Specifically, investigations should be performed to determine whether pressurized jet
releases from containment can occur (with aerosol formation due either to the mixing/reaction of
chemicals or as a result of heating by fire). The risk assessment/accident analysis should seek to
identify scenarios in which liquid can be driven through an orifice in a vessel or pipework at high
pressure. Such scenarios might occur when there is high pressure storage (if there is any at the
site) or they might occur if vessels are pressurized by some external agent, such as fire. The
issue to be decided is whether there are any circumstances at the site where aerosolization could
i
3-19
-------
occur, allowing a large fraction of the release to remain airborne as fine liquid droplets, thus
increasing the effective magnitude of the source term several-fold over what it would be if only
the vapor component is taken into account. This issue was originally raised by the panel because
there was some concern that the range of such scenarios had not been fully considered in the
WTI risk assessment.
The model used to estimate the rate of evaporation is not appropriate for calm
conditions. Evaporation models are available for extremely low wind speeds. One of these
should be used (e.g., Rife, 1981), and the results used to support the present estimates or to
replace them in an updated accident analysis [ST].
Also, the predicted effectiveness (or lack thereof) of mitigation measures needs to be
more clearly addressed, including the influence of time-varying rates of release [LT] and explicit
or implicit assumptions about the toxicity [ST] of the hazardous chemicals involved (such as
Haber's law).
The work group also discussed the justification of the 20,000-gallon cap on accident
emissions as cited in EPA's Proposed Rule on Accidental Release Prevention Requirements: Risk
Management Programs Under. Clean Air Section 112(r)(7). The proposed definition of a worst
case release is "the largest quantity of a regulated substance resulting from a vessel or process
piping failure." While the largest vessels have capacities of 20,000 gallons each, it is not apparent
that a risk assessment should have the same limitation as a risk management plan. Considering
the nature of this facility and its history, a risk assessment based on a
truly worst case accident scenario should be considered [SI]; this might involve more than a
single storage vessel.
Finally, it is not clear how the final summary tables of risk (Tables VIII-1, -2, and -3)
were developed from the information presented elsewhere in Volume VII. The authors should
provide a reproducible trail of analysis and clearly justify any conservative or nonconservative
aspects of their assumptions [ST].
3-20
-------
REFERENCES
Dockery et al., 1984. Change in pulmonary function in children associated with air
pollution episodes. J. Air Pollut. Control Assoc. 32:937-942.
Rife, R.R. 1981. Calculation of evaporation rates for chemical agent spills. Report
DRXTH-ES-TM-81101. U.S. Army Toxic and Hazardous Materials Agency, Aberdeen Proving
Ground, MD.
3721
-------
EXPOSURE ASSESSMENT
George Fries, Chair
Agricultural Research Service
U.S. Department of Agriculture
Beltsville, MD
James Butler Thomas McKone
Argonne National Laboratory School of Public Health
University of .Chicago University of California at Berkeley
Argonne, IL Berkeley, CA
These comments pertain primarily to sections of Volume V of the draft WTI risk
assessment document concerned with the characterization of human exposures attributable to
\
contaminants in the gas and particle phases of the atmosphere. The Exposure Assessment Work
Group offered three types of comments on this material:
• General issues.
• Comments on the specific issues raised in the charge to the Exposure Assessment
Work Group.
• Comments on the accident analysis section of the risk assessment.
All these should be addressed in the near term. The work group also offered suggestions
for improving future assessments; these are long-term recommendations.
NEAR-TERM RECOMMENDATIONS—GENERAL ISSUES
The exposure assessment is a large and comprehensive document. EPA expended a great
deal of effort to assemble data, construct models, run simulations, and evaluate data. The
resulting draft risk assessment addresses most of the recommendations of the 1993 project plan
peer reviewers. The document contains information required for an informed debate on health
issues, but the information is frequently buried and difficult to track.
3-22
-------
The Exposure Assessment Work Group addressed four general issues related to this <
assessment and future activities:
The surrogate selection process, which utilized the quantity-carcinogenic potency-
bioaccumulation (QCB) scores of compounds potentially present in emissions.
The work group recommends that the Executive Summary include an expression
of the likelihood that an important compound was omitted in the surrogate
compound selection process. This would involve a few simple reality checks that
can be described qualitatively in a single paragraph.
The small amount of information on existing exposures to incinerator-type
contaminants in the area. This omission is consistent with EPA risk assessment
guidance, but it raises questions about cumulative or total exposures to a given
contaminant in the area. Although methodologies for evaluating cumulative
exposures have not been developed fully, the document should address this issue
qualitatively. Such a discussion would facilitate evaluation of environmental
equity issues as well as more effective communication of relative risks.
Total risk from the facility. Total risk encompasses exposure to continuing
emissions from routine operations, episodic exposures to fugitive emissions, and
exposures resulting from accidents. The derivation of risk numbers for these
three sources should remain separate, but the combined risk should be discussed
together in the risk characterization.
NEAR-TERM RECOMMENDATIONS—SPECIFIC ISSUES
Exposure Descriptors
EPA's Exposure Assessment Guidelines document identifies descriptors that should be
used to characterize exposure, including central tendency estimates (representing the center of
the exposure distribution) and high-end estimates (representing individuals above the 90th
percentile in the exposure distribution). For the draft WTI risk assessment, average and
maximum environmental concentrations were modeled for each medium of concern. Similarly,
typical and 90th-percentile values were obtained for most of the exposure factors. If the central
tendency exposure estimates were calculated using average values for both media concentrations
and exposure factors, and high-end exposure estimates were calculated using 90th-percentile
values for both media concentrations and exposure factors, the exposure descriptors were
properly used to characterize exposures. Median values rather than mean values generally were
3-23
-------
used to estimate the central tendency values. This is appropriate because environmental
concentrations are often skewed to the high end. The document does not make clear whether
90th-percentile values were used for all inputs in the high-end exposure estimations.
Estimation Approach :
The draft risk assessment explains the general approach used to estimate central tendency
and high-end exposure values. This approach took into account three factors: concentrations in
environmental media, intake rates, and durations and/or frequencies of exposure. These three
factors appear to have been combined properly to characterize central tendency and high-end
exposures. Many procedures utilize models described in EPA guidance documents. As noted
above, however, in some cases the draft risk assessment does not clearly specify when mean and
90th-percentile values were used. Adding a summary table that identifies the specific factors
used for each exposure descriptor would be useful.
Exposure Sources and Pathways
Exposure assessments should identify all important exposure sources and pathways. The
draft WTI assessment examines a fairly wide range of potential exposure scenarios; the selection
of these scenarios and the conclusions'concerning the importance of the various pathways are
consistent with the current state of knowledge. Adequate justification now exists for omitting
ground-water and surface water pathways, as recommended during the 1993 peer review of the
project plan for this risk assessment. The approach and algorithms used to calculate exposure
doses listed in Volume V, Section VII, of the assessment are in harmony with approaches
suggested in the draft EPA Dioxin Reassessment—as well as with approaches used by other
agencies, such as the California Environmental Protection Agency.
The 1993 peer review of the project plan for this assessment recommended that EPA
include a discussion of exposure via household dust. -The assessment does not consider this
pathway, nor does it explain why this is not a significant route of exposure for sensitive subgroups
(e.g., infants and children). Dermal and ingestion pathways for outdoor soil do not necessarily
3-24
-------
represent how these contacts occur inside houses. House dust likely originates from three
sources:
• Airborne particles that move from outside air to indoor air.
• Surface soil and dust tracked into buildings.
• Sources related to occupant activities, material degradation, and household
products.
The assessment should include a brief qualitative discussion of exposure via ingestion of
house dust as part of its discussion of soil ingestion.
Estimation of Concentrations and Exposure
The assessment identifies the key assumptions for estimating chemical concentrations and
exposures. The magnitude and direction of effect generally are correct, except that the
assumption that fate and transport modeling accurately reflects reality is uncertain and does not
necessarily result in a "likely overestimate." Often, model parameters are derived from a small
number of observations of only a few animal or plant species. The direction of effect is really
unknown until these models are better validated with monitoring data collected for that purpose.
EPA could make progress toward verifying its surrogate selection process by comparing
the QCB scores of chemicals listed in Volume V, Table IV-1, with the relative contribution of
each of these chemicals to the total estimated risk in the actual risk assessment. The
)
assumptions that fate and transport models are accurate should be broken down further (i.e., by
including more components in Volume V, Table VI-20). At .a minimum, the biotransfer,
diffusion, and advection (i.e., deposition) of the fate and transport models should be separated
out and listed as separate assumption categories.
3-25
-------
Conservative Assumptions
This assessment is based on conservative assumptions, and the cumulative impact of this
conservativeness probably results in an overstatement of risk. Including uncertainty analyses for
two representative compounds was a useful way of addressing the overall uncertainties—and
identifying input parameters that have the greatest effect on the final risk estimate. The
discussion on pages VI-14 to VT-15 of Volume V is particularly useful. The evaluation of the
fate and transport models includes both model and parameter uncertainties.
One item that remains unclear in the Executive Summary relates to the ratio of high-end
to low-end exposure estimates attributable to uncertainty. The former reflects heterogeneity,
whereas the latter reflects uncertainty. This provides more confidence about the relative values
of high-end and central tendency exposures than it does about their absolute values. The risk
assessment addresses uncertainty by biasing both the high-end and central tendency values
toward the upper end of their likely range.
NEAR-TERM RECOMMENDATIONS—ACCIDENT ANALYSIS
The Exposure Assessment Work Group has the following comments on the accident
analysis:
• The facility is located on a flood plane. EPA should evaluate the likelihood (1) of
a flood large enough to inundate the facility, and (2) that such a flood would
cause hazardous materials to be released to the environment.
• The key assumptions made in the identification of accident scenarios and ranking
of accident events appear to be reasonable with respect to magnitude and
direction of effects. The report does not adequately express or communicate the
expected value of harm associated with accidents, however. Rather, the report
expresses severity and consequence information using ambiguous phrases, such as
"likely" and "unlikely" events, and "moderate" to "catastrophic" consequences. A
rough calculation, assuming proper interpretation of the tables, suggests a 1 in
1,000 chance per year of an accident that involves approximately 10 fatalities.
Does this mean that, over 10 years, there Is a 0.1 or 10 percent (10x10/1,000)
likelihood of one fatality in the community from accidents? This is a very large
risk compared to the 1 in 1,000,000 limit typically used for cancer risk. Some
clarification would be useful.
3-26
-------
The American Industrial Hygiene Association's Emergency Response Planning
Guideline (ERPG) levels would probably have been more appropriate than IDLH
values for characterizing the severity of accident consequences because IDLH
values are designed for healthy workers rather than the general population. LOG
values would be an acceptable alternative to ERPG values because LOG values
are more stringent than IDLH values for assessing the acute effects of short-term
exposures. The Toxicology Work Group also discussed this issue; see that group's
summary (below) for more on the use of ERPG versus LOG values instead of
IDLH values.
LONG-TERM RECOMMENDATIONS
Although the work group does not have specific recommendations for future research, we
note that no provisions are made for validation and periodic checks to determine the reliability
of the assessment. If feasible, these activities should be conducted. The work group also
recommends that models and parameters be updated as new information is developed. This is
particularly important for physical-chemical parameters (e.g., log K^,), which drive many fate and
transport models.
3-27
-------
TOXICOLOGY
Mary Davis, Chair
Department of Pharmacology and Toxicology
West Virginia University
Morgantown, WV
George Alexeeff Thomas Gasiewicz
Air Toxicology and Epidemiology Section Department of Environmental Medicine
California Environmental Protection Agency University of Rochester
Berkeley, CA Rochester, NY
HAZARD IDENTIFICATION, DOSE RESPONSE, AND RISK CHARACTERIZATION
The Toxicology Work Group noted that, overall, the draft WIT incinerator human health
risk assessment is thorough and comprehensive. Highlights of the work group's discussions are
summarized below. These comments, which supplement the reviewers' premeeting comments
(see appendix D), are divided into near-term recommendations (priority issues and minor
clarifications) and long-term recommendations (for future risk assessments/method development).
Near-Term Recommendations — Priority Issues
Cumulative jR«#
Exposures from WTT emissions occur against a background of existing emissions. To
evaluate the potential impact of the WTI facility on human health, it is necessary to consider the
facility's emissions against existing exposures. Therefore, the Toxicology Work Group
recommends that EPA quantitatively or qualitatively evaluate releases from other facilities in the
area or other data on existing emissions. In addition to the emissions data base, the Total
Exposure Assessment Methodology Study data from Steubenville might be useful.
*Dr. Alexeeff was unable to travel to the workshop, but participated in the Toxicology Work
Group via teleconference. j . '
3-28
-------
Uncertainty
The Toxicology Work Group considers discussions of uncertainty to be an integral part of
risk assessments. The treatment of uncertainty in the WIT incinerator risk assessment is more
thorough than is typical. Nonetheless, the work group recommends adding further discussion of
the uncertainties associated with:
Data gaps, since not all of the chemicals in the stack emissions have been
identified, their toxicities characterized, and/or a valid reference dose (RfD),
reference concentration (RfC), or slope factor determined.
Extrapolating the slope factor and/or hazard index across different routes of
exposure.
Lead Cancer Risk
The draft risk assessment does not treat lead as a carcinogen because EPA does not have
a slope factor for lead. On an interim basis, the Toxicology Work Group recommends using the
slope factor that California EPA has developed to provide a sense of the magnitude of the
cancer risk from lead. To address the uncertainty associated with the lack of a U.S. EPA slope
factor, the work group suggests comparing California EPA's slope factor with those developed by
other agencies. The work group recommends that the discussion of lead's noncancer
(neurobehavioral) effects be retained.
Noncancer Endpoints for Dioxin-Like Chemicals
The draft risk assessment does not estimate noncancer risks associated with dioxin-like
chemicals because EPA has not yet determined which is the most sensitive toxic effect, nor has
the Agency developed an RfD for that effect. The draft risk assessment addresses noncancer risk
by comparing the releases estimated to occur from the WIT facility to estimated background
exposures. Some dioxin effects (reproductive/developmental, immune) occur in experimental
animals at exposure levels lower than those producing cancer, creating concern that these effects
will occur more frequently than cancer. Thus, the risk assessment should explicitly discuss
3-29
-------
differences in how EPA estimates noncancer and cancer risks (e.g., use of hazard index versus
slope factor, assumption of no threshold or a threshold, other assumptions) to put the cancer risk
estimate into perspective. In particular, the risk assessment should point out that the cancer risk
estimate might be more conservative than a noncancer estimate would be (if one were
calculated) because it is derived using a slope factor, which assumes no threshold. .
Complex Mixture Toxicology: Additivity Versus Synergy Versus Antagonism
To address the fact that WTT emissions represent a complex mixture of toxicants, the
draft risk assessment assumes that toxic effects are additive. The Toxicology Work Group
recommends that EPA explain the rationale for this assumption more fully, particularly in light
of recent studies. Jonker et al., for example, compared the acute (24-hour) toxicity of a
combination of four nephrotoxicants in rats compared with that of the individual compounds.
Another study (Food Chem Toxicol 31:45-52,1993) suggests that noncancer effects are not
additive when exposures occur at or below the no effect level.
Near-Term Recommendations—Minor Clarifications
In reviewing the draft plan for the WTI incinerator risk assessment, the Toxicology Work
Group suggested that EPA include health effects data from similar facilities. These data
apparently are not available. To clarify this situation, the work group recommends that EPA
include in the final risk assessment a list of the data sources examined and what was found in
each.
The risk associated with polycyclic aromatic hydrocarbons (PAHs) is based on seven
PAHs and expressed as their potency relative to benzo[a]pyrene. The other PAHs are not
included. As indicated in Dr. Alexeeff s premeeting comments, beginning on page 8, this
introduces uncertainty into the risk estimate.. The Toxicology Work Group recommends that
EPA briefly explain the rationale for and impact of this risk estimation procedure. In addition,
members of the Combustion Engineering Work Group expressed concern that only
benzo[a]pyrene was detected.
3-30
-------
The Toxicology Work Group also recommends:
• Including a list of noncancer endpoints for the chemicals addressed in the risk '
assessment, probably in the form of a table.
• In the characterization of releases, clarifying which chemicals were estimated to be
released and which were actually detected.
» Clarifying why the risk assessment does not address metals (especially
methylmercury) in breast milk.
• Using better terms or acronyms for maximum concentration and area average
exposures.
• Expanding the discussion of endocrine disrupters to clearly indicate which emitted
(estimated or measured) chemicals are endocrine disrupters and what effects
might have endocrine disruption as a mechanism of action.
Long-Term Recommendations—Future Risk Assessments/Method Development
Validation of Dispersion and Uptake Models
The usefulness of a model depends on the ability of the model to accurately predict the
fate of the study chemicals. To validate and improve the models used in the draft risk
assessment, the Toxicology Work Group recommends that EPA conduct followup monitoring of
chemicals of concern in the air, soil, vegetation, and locally produced food. By providing
information about the appropriateness of the models under specific conditions, such monitoring
might also facilitate model selection in future risk assessments.
The Subpopulation of Workers Who Reside in the Area
The draft risk assessment does not consider the total (occupational and environmental)
exposures of individuals who both work in and live near the WTI facility. The Toxicology Work
Group recognizes that occupational and environmental exposures fall under the purview of
different agencies. Nevertheless, both agencies are charged with protecting human health. The
work group recommends that the two agencies jointly develop policies and procedures to
3-31
-------
integrate the assessment of occupational and environmental exposures to ensure that future
assessments adequately consider and protect the health of worker residents.
Subpopulations and Exposure Assumptions
The draft risk assessment uses Superfund defaults for exposure parameters such as
residence time and duration of exposure. The Toxicology Work Group recommends evaluating
the appropriateness of these assumptions given conditions at the WIT site. For instance, do
individuals living in the area have the same mobility as the nation as a whole? Does that
mobility reflect true population movement into and out of the affected area, or is the apparent
mobility a result of relocations within the area? This information is crucial for judging the
validity of the assumption of a 9-year exposure period. It would also shed light on the validity of
the assumption that exposure via breast milk can be treated separately (rather than one of
several exposure routes applicable to the same individual over his or her lifetime).
The cancer risks for the different subpopulations. appear to be spread over different
times. Indicating the length of time used for each subpopulation would be helpful. Ideally, the
assessment would present cancer risks associated with different exposures to the younger
population (-0.75 to 1, -0.75 to 6 or 9, -0.75 to 30) and would analyze exposures from all
pathways (including breast milk).
ACCIDENT ANALYSIS
In reviewing the draft plan for the WTT incinerator risk assessment, the Toxicology' Work
Group expressed great concern that nonroutine and fugitive emissions might be an important
exposure source and recommended that they be a major part of the risk assessment. The
accident analysis represents a valiant start, but this topic is not well developed and many tools
needed to perform a thorough risk assessment of potential accidents are lacking. Many of the
work group's concerns with the accident analysis arise from EPA's use of methods not designed
for analyzing chemical risks.
3-32
-------
Near-Term Recommendations—Priority Issues
Chemicals Selected for Evaluation
Acetone has been deleted from the list of toxic chemicals used for emergency planning
because severe toxic levels are not expected to occur (see Dr. Alexeeff s premeeting comments,
page 16). The Toxicology Work Group recommends that EPA select a different chemical.
Use oflDLH Values for Characterizing Severity of Consequences
The Toxicology Work Group discussed the use of IDLH values at length. While these
values have the advantage of being a comprehensive set of values designed for accident analysis,
they are based on assumptions that are not appropriate to the WTI facility. Specifically, they are
designed to provide short-term protection to healthy workers trained in emergency procedures
and who would be exposed to IDLH levels for a limited time. They do not account for the
higher respiratory rates of children, nor do they account for pre-existing conditions (e.g.,
childhood asthma) that render some of the population more sensitive.
The Toxicity Work Group considered several possible control levels for the accident
analysis, including IDLH, ERPG, EEGL, and SPEGL. All of them have problems or limitations,
and none is ideal. The work group noted that, for some chemicals, the LOG is similar to ERPG,
EEGL, and SPEGL values; however, LOG values exist for more chemicals. The overall
consensus was that IDLH values are inappropriate for the risk analysis of the WTI site and that
uncertainty analysis using the LOG would be a better assessment of the risks. For individual
chemicals, ERPG-2, EEGL, or SPEGL might be more appropriate. LOCs are available for more
compounds, which is an advantage, and are often similar to ERPG-2, EEGL, and SPEGL values.
Long-Term Sequelae of an Accident
The accident analysis focuses on the acute effects of the accident scenarios. It does not
address how chemical exposures during an accident might impact an individual's lifetime
3-33
-------
exposure to chemicals from the WIT plant. The Toxicology Work Group is concerned that the
magnitude of chemical exposure from accidents or other upset conditions might be far greater
than exposures from normal operations.
Subpopidation at Risk
The Toxicology Work Group recommends that EPA perform an appropriate modeling
and dispersion analysis to determine the worst case concentrations of chemicals at the adjacent
East Elementary School. A large number of children attend this school; they might be more
sensitive than the adult worker population, and they might be exposed for more than 30 minutes,
the maximum exposure period assumed in the IDLH values. EPA should evaluate both short-
term acute effects and long-term sequelae.
Near-Term Recommendations—Minor Clarifications
For each accident scenario, the Toxicology Work Group recommends explicitly addressing
the East Elementary School in the discussion of exposure area.
Long-Term Recommendations—Future Risk Assessments/Method Development
Severity Categories
Federal Emergency Management Agency (FEMA) categories appear to be structured for
widespread disasters (earthquakes, floods, hurricanes), with a large difference between minor and
major events. In the case of the WTI incinerator, however, even a minor event would pose a
significant challenge to the local community. Thus, the Toxicology Work Group recommends
i
developing a classification system that:
3-34
-------
Includes an intermediate category. :
Characterizes impact according to magnitude of severity and likelihood of
occurrence within a specified period of time.
Uses easily understood terms.
3-35
-------
ECOLOGICAL RISK ASSESSMENT
Glenn Suter II, Chair
Environmental Sciences Division
Oak Ridge National Laboratory
Oak Ridge, TN
Peter deFur Steven Peterson
Environmental Defense Fund Ecology and Environment, Inc.
Washington, DC Lancaster, NY
Pirn Kosalwat
Applied Sciences and Research Division
KBN Engineering and Applied Sciences, Inc.
Gainesville, FL
Members of the Ecological Risk Assessment Work Group agree that EPA conducted the
SERA in a technically competent manner that conforms with the state-of-practice for SERAs.
Work group members provided premeeting comments covering all aspects of the SERA, and the
work group encourages EPA to consider those comments when revising the document. The
following points address areas where work group members can recommend specific actions to
improve the document.
NEAR-TERM RECOMMENDATIONS
Goals and Purpose of the SERA
The principal problem with the SERA is that its goals and purpose are not clear: As a
result, the implications of the results of the assessment and the appropriateness of-possible
recommendations are unclear. This general problem has the following consequences:
• The purpose of screening assessments is to narrow the scope of subsequent
assessments by eliminating chemicals, scenarios, routes of exposure, or receptors
that are clearly not hazardous. When, as in this case, certain chemicals in certain
scenarios are retained by the screen, the assessment should suggest what
additional data collection or analysis will be performed to resolve those issues.
3-36
-------
That was not done, at least in part because the assessors were uncertain about the
need to resolve the issues.
The work group's attempts to recommend additional activities to resolve the
potential risks identified by the SERA elicited a clarification that called into
question the need to resolve these risks: the scenarios related to unresolved
hazards were included to help define RCRA permit limits, not to produce actual
estimates of risk. If this is true, EPA should clearly state this in the SERA.
Otherwise, EPA should consider how to resolve the hazards.
The lack of an accident analysis in the SERA concerned the work group because
such an analysis might contribute to a determination of the acceptability of the
incinerator. We would guess that atmospheric releases during an accident could
exceed routine emissions and could result in greater exposures because the
deposition could occur during a relatively brief period. An accident might also
result in significant land and water pollution. Direct terrestrial and aqueous
contamination was not considered for routine operations, however, so it was not
clear whether such exposures were out of scope.
The work group's concern about the lack of accident scenarios met with the
following clarification: because RCRA permits do not permit accidents, the goal
of helping to define permit limits does not require consideration of the ecological
risks of accidents. If this is EPA's position, the SERA should clearly state this
because a serious accident could release more contaminants than decades of
routine operation. EPA should also clarify why this logic was applied to the
SERA but not to the human health risk assessment.
In summary, EPA should consider whether an accident analysis is needed for the
SERA and present the results of that consideration. If EPA has a good reason
for omitting an accident analysis, this should be presented. Possible reasons might
include:
— EPA performed an analysis that indicated that ecological risks due to
accidents could not exceed those from routine emissions.
— The human health consequences of an accident are so large that they
clearly overwhelm ecological considerations.
— EPA intends to use accident analyses only to devise contingency plans
(e.g., evacuation plans), not to help determine the acceptability of the
incinerator. Because EPA does not devise ecological contingency plans,
accident analyses serve no purpose in a SERA.
— EPA made a policy decision to not consider accidents relative to
ecological endpoints.
If EPA instead decides to conduct an accident analysis for the SERA, the work
group suggests that EPA resolve questions such as what endpoints to include,
what results will be useful for risk management, and whether transportation
accidents should be considered during the problem formulation stage; these are
3-37
-------
not strictly technical questions. The work: group knows of no precedent or
guidance for this accident analysis, but we believe the greatest difficulty lies in the
source terms and dispersion rather than the ecological assessment. That is, if
accident scenarios can be.defined and the associated release and transport
modeled, the analysis should not be inherently more difficult than that for routine
emissions. If persistent chemicals are released in an accident, chronic as well as
acute effects should be considered.
The permit limit scenario, which seems to drive the SERA, does not appear in the
human health risk assessment. The work group wonders if this reflects a
judgment on the part of EPA that only health risks will contribute to the decision
concerning acceptability of the facility—and that ecological risk considerations will
be considered only to refine permit limits,
The permit-setting goal of the SERA also might explain why the assessment fails
to include nonatmospheric emissions (e.g., disposal of ash, scrubber sludge,
wastewater). The work group recommends that EPA consider whether the
SERA's goals are fulfilled in the absence of consideration of these emissions.
The work group recommends that EPA staff and contractors responsible for
performing the SERA meet with the Region 5 and State of Ohio risk managers
for the WTI incinerator. The meeting should follow a Comprehensive
Environmental Response, Compensation, and Liability Act (CERCLA) data
quality objective (DQO) process to establish the purpose of the SERA and the
types of results needed.
The meeting should address whether the SERA is intended to reveal the risks of
the plant to the environment (i.e., analogous to the health risk assessment), reveal
the consequences of alternative permit limits, determine whether the risk of
severe effects is sufficient to shut down the plant, and so on. Depending on the
decision, the SERA might include a best estimate case, focus on critical risks, or
address accidents or other emissions.
Given the relatively large effort devoted to the SERA and the relatively high
profile review, a nonmanagement purpose of the SERA might be to develop
methods for this type of assessment and to serve as a model for future
assessments. The work group's recommendations are based on the assumption
that this SERA is not intended to serve as a model assessment, so only issues
likely to change the SERA's conclusions are important for this review. If EPA
reconsiders the purpose of the SERA and decides to make it a model assessment,
the work group recommends that the Agency carefully consider all premeeting
comments.
3-38
-------
Other Issues
The Ecological Risk Assessment Work Group recommends that EPA:
• Edit the SERA to tighten it up and make it more accessible to stakeholders (if
the document will be used as a communications piece).
• Indicate in the uncertainty analysis which uncertainties are most important to the
conclusions. For example, EPA should state explicitly that the metals found to be
hazardous are the same as those for which removal efficiencies were
unknown—and that the removal efficiencies of these metals were assumed to be
100 percent.
• Clarify the description of the emission scenarios to indicate their intended
interpretation, the reasons for inconsistencies in the degree of conservatism used,
and the reasons for inconsistencies compared to the human health risk
assessment.
• Briefly address whether, how, and why startup, shutdown, and other nonstandard
operating conditions were or were not addressed in the SERA. This is an
important issue in the minds of many people who have followed the WTI
incinerator debate.
LONG-TERM RECOMMENDATIONS
SERA Methods
The Ecological Risk Assessment Work Group feels strongly that the SERA illustrates the
need to develop data sets, models, and other methods for screening assessments. SERAs should
be relatively quick and inexpensive so that time and effort can be devoted to definitive
assessments that provide realistic estimates of exposure and effects and that resolve risk
management issues. In the absence of the types of data sets and default methods available to
human health risk assessors, high-effort SERAs will be required to reinvent the wheel. Thus,
efforts to develop tools and data for SERAs would not only be efficient use of resources, but
would increase the quality of SERAs by providing tools and data that are consistent, reliable, and
peer reviewed.
3-39
-------
Background Contamination
The Ecological Risk Assessment Work Group recommends addressing the issue of
background contamination in the SERA. Background contamination might be important for two
reasons:
The WTI incinerator is located in an industrialized area. The incremental risk
posed by the incinerator might be small or significant, depending on how it
compares to the magnitude of the background risk.
Given the large uncertainties in the SERA, environmental monitoring would be
desirable; however, background contamination might preclude monitoring of
ambient media and biota to detect the influence of the incinerator. Thus, EPA
should evaluate the practicality of environmental monitoring in this context.
Ecotoxicological Data Gaps
The Ecological Risk Assessment Work Group is concerned about the implications of the
many ecotoxicological data gaps. The work group recommends that EPA attempt to analyze the
implications of the data gaps on the reliability of the SERA. EPA could address the lack of data
for specific endpoint taxa by considering the sensitivities of uncharacterized taxa relative to those
of taxa for which data are available. EPA could address the lack of data for specific chemicals
by determining whether the estimated -risks are credible given each chemical's maximum credible
toxicity.
Other Issues
The Ecological Risk Assessment Work Group recommends that EPA:
Consider the work group's premeeting comments on selection of ecological
chemicals of concern (ECOCs) if the Agency reconsiders the purpose of the
SERA or decides to make it a model assessment. Although the work group
believes EPA's method of ECOC selection yielded acceptable results for this
SERA, the method is questionable.
3-40
-------
emissions to model fugitive emissions, the use of sulfur dioxide as a surrogate for selenium, the
particle size assumptions used for the metals analysis, and the gas-solid partitioning assumptions
used in the PCDD/F analysis. They also wondered which PICs were nondetects and why EPA
had not prorated measured PICs to compensate for unmeasured PICs. On the whole, however,
the peer reviewers responded favorably to this part of the risk assessment, praising several
aspects of EPA's work (e.g., the Agency's efforts to determine the composition of the waste feed
and its discussion of uncertainty in PIC estimation).
Similarly, the peer reviewers praised EPA's efforts to follow the recommendations of the
project plan peer reviewers on air dispersion/deposition modeling. Some voiced concern that the
ISC-COMPDEP model might be so conservative as to mask sensitive effects, commented that the
discussion does not clearly explain how and to what extent EPA used the INPUFF and
CALPUFF models, and suggested that EPA run ensemble types of sensitivity tests to obtain a
range of stochastic distributions. Over the long term, they suggested, EPA should develop
improved methodologies for analyzing potential dense gas emissions and dispersion and more
sophisticated tools for modeling deposition (especially wet deposition).
In their preliminary comments on the exposure assessment, the peer reviewers again
observed that EPA had addressed nearly all of the recommendations of the project plan peer
reviewers. Describing the exposure assessment as comprehensive, the peer reviewers stated that
EPA had appropriately addressed all important descriptors and pathways except for house dust.
The peer reviewers suggested that EPA consider using actual operating experience to validate its
exposure predictions and to determine how WTI-related exposures contribute to total exposures
in the area.
On the topic of toxicology, too, the peer reviewers noted that EPA had generally
followed the recommendations of the project plan peer reviewers. They also commented that
EPA's selection of surrogate chemicals seems reasonable and that the special consideration given
to some chemicals is appropriate, although they suggested that EPA provide more on PAHs and
lead and enumerate specific noncancer endpoints. The peer reviewers also suggested that EPA
consider risks to the subpopulation of individuals who both work in and live near the WTI
facility, that EPA clarify how it assessed the noncancer risks associated with chemicals without an
RfC or RfD, and that EPA more clearly identify gaps in the data.
4-3
-------
OBSERVERS' COMMENTS
During the workshop, observers were given two opportunities to offer questions and
comments. First, at the discretion of each work group chair, observers were invited to participate
in work group discussions. Second, they were invited to make statements to the full workshop
during the afternoon plenary session. Observers raised issues concerning the technical and
scientific aspects of the draft risk assessment as well as the political and personal aspects of the
WH incinerator. Highlights of observer comments include:
• If possible, it would be useful for peer reviewers to visit the WTI incinerator to
obtain a clearer understanding of the layout/impacts; in so doing, peer reviewers
should interview local residents to balance industry information with the
knowledge and experiences of the public.
• The emergency plan to be used in case of an accident at the WTI facility will not
prevent exposure of children to toxic gases.
• During previous incidents, WTI has not acted uniformly responsibly (e.g., WTI
failed to notify public health officials when a gas main was ruptured). Several
incidents (e.g., mistaken placement of caustic waste into the WTI facility's
pollution control system) have engendered fear in local residents and, in some
cases, have prompted evacuations.
• Residents impacted by the WIT facility are people, not just numbers. Use of the
term "moderate risk" to describe an accident involving 10 fatalities is insensitive to
impacted residents. WTI should not be allowed to operate if EPA is not
extremely confident that there will be no life-threatening accident or other
accident that will impact the children attending the nearby school.
*• The accident analysis is the weakest part of the risk assessment. It fails to:
— Include information from other relevant facilities (e.g., the Biebesheim
facility in Germany, other American hazardous waste incinerators, and
chemical plants).
— Provide true worst case scenarios (e.g., a situation in which a liquid
vaporizes and the condensate aerosol ignites to cause a gasoline bomb-
type explosion) and other relevant scenarios (e.g., a major fire in the
bunker where solid hazardous waste is dumped and commingled prior to
burning; accidental burning of radioactive waste).
— Consider the unique location of the WTI facility, which might make offsite
damage and injuries more likely than at other facilities.
— Use sufficiently protective values (IDLE values are inappropriate).
,
4-4
-------
APPENDIX A
WORKSHOP AGENDA
-------
-------
SEPA
United States
Environmental Protection Agency
Risk Assessment Forum
Technical Workshop on
WTI Incinerator Risk Issues
Holiday Inn - Georgetown
Washington, DC
January 11-12, 1996
Final Agenda
Workshop Chair: Dr. Thomas McKone, University of California
THURSDAY, JANUARY 11
7:30AM REGISTRATION/CHECK-IN
OPENING PLENARY SESSION
8:30AM Welcome
Dr. William Wood, U.S. Environmental Protection Agency (EPA),
Risk Assessment Forum (RAF)
8:45AM History of EPA Risk Assessment Activities for the WTI Incinerator
Dr. Harriet Croke, EPA Region 5
9:OOAM Charge to the Reviewers
Dr. Thomas McKone
PREMEETING COMMENTS
9:1 SAM Combustion Engineering
Dr. Barry Dellinger, University of Dayton
9:30AM Air Dispersion/Deposition Modeling and Accident Analysis
Dr. Walter Dabberdt, National Center for Atmospheric Research
9:45AM Exposure Assessment
Dr. George Fries, U.S. Department of Agriculture
10:OOAM Toxicology
Dr. Mary Daw's, West Virginia University
10:1 SAM Ecological Risk Assessment
Dr. Glenn Suter, Oak Ridge National Laboratory
IO:30AM BREAK
i Printed on Recycled Paper
A-l
-------
THURSDAY, JANUARY 11 (Continued)
11 :OOAM CONCURRENT WORKGROUP BREAKOUT SESSIONS
I2.-OOPM LUNCH
I:OOPM CONCURRENT WORKGROUP BREAKOUT SESSIONS
3:30PM BREAK
PLENARY SESSION
3:30PM Observer Comments
4:OOPM Workgroup Status Reports and Discussion (Workgroup Leaders)
6:30PM ADJOURN
A-2
-------
APPENDIX B
REVIEWER LIST
-------
-------
S-EPA
United States
Environmental Protection Agency
Risk Assessment Forum
Technical Workshop on
WTI Incinerator Risk Issues
Holiday Inn - Georgetown
Washington, DC
January 11 -12, 1996
Final Reviewer List
George Alexeeff
Chief, Air Toxicology &
Epidemiology Section
California Environmental
Protection Agency
2151 Berkeley Way - Annex 11
2nd Floor
Berkeley, CA 94704
510-540-3324
Fax:510-540-2923
Elmar Altwicker
Professor
Department of Chemical Engineering
School of Engineering
Rensselaer Polytechnic Institute
110 Eighth Street
Troy, NY 12180-3590
518-276-6927
Fax:518-276-4030
E-mail: altwie@rpi.edu
James Butler
Argbnne National Laboratory
University of Chicago
9700 South Cass Avenue (EAD/900)
Argonne, IL 60439-4832
708-252-9158
Fax: 708-252-4336
E-mail: jpbutler@anl.gov
Walter Dabberdt
Associate Director
National Center for
Atmospheric Research
1850 Table Mesa Drive
P.O. Box 3000
Boulder, CO 80303
303-497-1108
Fax:303-497-1194
E-mail: dabberdt@ncar.ucar.edu
Mary Davis
Professor of Pharmacology & Toxicology
Robert C. Byrd Health
Sciences Center
West Virginia University
I Medical Center Drive
P.O. Box 9223
Morgantown, WV 26506-9223
304-293-4449
Fax: 304-293-6854
E-mail: mdavis@wvnvm.wvnej.edu
Peter deFur
Sen/or Scientist
Environmental Defense Fund
1875 Connecticut Avenue, NW
I Oth Floor-Suite 1016
Washington, DC 20009
202-387-3500
Fax: 202-234-6049
E-mail: peterd@edf.org
Barry Dellinger
Group Leader
Environmental Sciences
and Engineering
University of Dayton
300 College Park
Dayton, OH 45469
513-229-2846
Fax:513-229-2503
E-mail: dellinger@udri.udayton.edu
George Fries
Research Animal Scientist
Agricultural Research Service
U.S. Department of Agriculture
BARC East/Building 201
Beltsville, MD 20705
301-504-9198
Fax:301-504-8438
E-mail: fries@ggpl.arsusda.gov
Mark Garrison
ERM, Inc.
855 Springdale Drive
Exxton, PA 19341
610-524-3500
Fax:610-524-7798
i Printed on Recycled Paper
.B-l
-------
Thomas Gasiewicz
Professor of Toxicology
Department of Environmental
Medicine
School of Medicine
P.O. Box EHSC
University of Rochester
Medical Center
575 Elmwood Avenue
Rochester, NY 14642
716-275-7723
Fax:716-256-2591
E-mail: gasiewt@envmed.rochester.edu
Halstead Harrison
Associate Professor
Department of Atmospheric Sciences
University of Washington
P.O. Box 351640
Seattle, WA 98195-1640
206-543-4596
Fax: 206-543-0308
E-mail: harrison@atmos.washington.edu
Jerry Havens
Distinguished Professor of
Chemical Engineering
Chemical Hazards Research Center
University of Arkansas
700 West 20th Street
Fayetteville, AR 72701
501-575-2055
Fax: 501-575-8718
E-mail: jah@engr.uark.edu
Geoffrey Kaiser
Vice President
Environmental and Energy Group
SAIC
11251 Roger Bacon Drive
Reston.VA 22090
703-318-4626
Fax: 703-709-1042
E-mail; geoffrey.d.kaiser@cpmx.saic.cam
Pirn Kosalwat
Sen/or Scientist
Applied Sciences & Research Division
KBN Engineering & Applied
Sciences, Inc.
6241 Northwest 23rd Street
Suite 500
Gainesville, FL 32653-1500
904-336-5600
Fax: 904-336-6556
E-mail: kbn_jvil@digita!.net
Thomas McKone
COEH, School of Public Health
140 Warren Hall
University of California - Berkeley
Berkeley, CA 94720-7360
510-642-8771
Fax:510-642-5815
Robert Meroney
Civil Engineering Department
Colorado State University
Fort Collins, CO 80523
970-491-8574 .
Fax:970-491-8671
E-mail: rmeroney@vines.colostate.edu
Steven Peterson
Ecologist
Ecology and Environment, Inc.
368 Pleasantview Drive
Lancaster, NY 14086
716-684-8060
Fax: 716-684-0844
William Randall Seeker
Sen/or Wee President
Energy and Environmental
Research Corporation
18 Mason Street
Irvine, CA 92718
714-859-8851
Fax:714-859-3194
E-mail: 74723,1170@compuserye.com
Glenn Suter, II
Sen/or Research Staff Member
Environmental Sciences Division
Oak Ridge National Laboratory
P.O. Box 2008 (MS-6038)
Oak Ridge, TN 37831
423-574-7306
Fax: 423-576-8646
E-mail: swg@ornl.gov
B-2
-------
APPENDIX C
CHARGE TO REVIEWERS
-------
-------
CHARGE TO REVIEWERS
FOR THE WTI DRAPT FINAL RISK ASSESSMENT
The draft final WTI risk assessment is divided into several volumes covering the
scientific disciplines of toxicology, environmental fate and transport, combustion
engineering, atmospheric modeling, exposure assessment, ecofogical risk assessment,
and accident analysis. As a reviewer of the WTI draft final risk assessment, you
should use your best technical knowledge and professional judgment to comment on
the technical accuracy, completeness and scientific soundness of the assessment.
Each reviewer is asked to focus on several specific issues in his or her area of
expertise with comments on other areas invited but optional. Your comments will be
considered in finalizing the risk assessment.
For the peer review workshop reviewers will be organized into 5 work groups:
Combustion Engineering, Air Dispersion and Deposition Modeling and Accident
Analysis, Toxicology, Exposure Assessment, and Ecological Risk Assessment. All
reviewers should be familiar with the Executive Summary (Volume I) and the Facility
Background (Volume II) sections of the draft risk assessment In addition, each work
group should focus on specific Volumes as specified below:
Workgroup
Combustion Engineering
Air Dispersion and
Deposition Modeling and
Accident Analysis
Toxicology
Exposure" Assessment
Ecological Risk
Assessment
Risk Assessment Volumes
Volume III - Facility
Emissions
Volume IV - Atmospheric
Dispersion and
Deposition Modeling
Volume V - Human
Health Risk Assessment
Volume VII - Accident
Analysis
Volume VII - Accident
Analysis
Volume V - Human Volume VII - Accident
Health Risk Assessment Analysis
Volume V! - Screening
Ecological Risk
Assessment
While reviewing these sections of the document, please address the following
general issues.
1. Comment on the organization of the risk assessment document. Does the
layout follow a logical format? Is the presentation of information in the
document clear, concise and easy to follow?
C-i
-------
2. Does the executive summary accurately reflect the data and methodologies
used and the conclusions derived in the risk assessment?
3. Were the major recommendations of the 1993 peer review workshop for the risk
assessment plan addressed?
4. As with any risk assessment, there are -always additional data and method
development efforts that could be undertaken to reduce the level of uncertainty.
However, are there any major data or methodological gaps that would preclude
the use of this risk assessment for decision making? If so, How should they be
addressed?
5. What long-term research would you recommend that could improve risk
assessments of this type in the fijture?
In addition, the following workgroup specific issues should be addressed.
Emissions Characterization
Emissions characterization includes identification of substances of concern and
the development of emission rates for these contaminants. Emission rates were
developed through a combination of site specific stack test data and models, Pleasfe
comment on the following issues with respect to this aspect of the draft risk
assessment.
1. To characterize the nature of the emissions, waste stream profiles were
developed and entered into a database. Several refinements and adjustments
(e.g., the Subtraction Correction Factor for chlorinated compounds) were
applied to the profiles before substances of concern were identified. Please
comment on whether or not these adjustments are appropriate. What is the
anticipated effect on the risk assessment?
2. Comment on the selected chemicals of concern. Have important chemicals
been missed due to the selection technique?
3. Comment on the approaches used to estimate stack emission rates (e.g., use
of the 95% UCL of the arithmetic mean or the maximum detected value,
whichever is smaller, for high end emission rates). Are the approaches
appropriate? Are their effects on the risk assessment adequately
characterized? Comment on the adjustment made to PCDD/PCDF emission
rates to account for brominated dioxin-fike compounds. Also, comment on the
approach to characterizing emission rates from fugitive sources (e.g., use of the
TANKS 2 model for the Carbon Adsorption Bed).
4. Comment on the identified sources of fugitive emissions. Was the approach
used to select these sources appropriate? Have important sources been
C-2
-------
missed? Have emissions from process upsets been given appropriate
consideration?
5. There have been a number of controlled bums at the WTI facility. Please
comment on the adequacy of these data in estimating potential exposure.
Please comment on the assumptions made from the tests in regard to
composition of wastes received at WTI and emissions when the plant operates
in the future.
6. Comment on the use of emission factors from coal, burning to estimate the
emission rate of fly ash from WTI. Are the factors used to adjust the coal
emission rate appropriate? Are the uncertainties introduced from this approach
adequately characterized?
7. Overall, is the identification of the key assumptions used in characterizing the
nature and magnitude of emissions thorough? Are the magnitude and direction
of effect of these assumptions on the overall risk assessment accurately
characterized? Is the uncertainty and variability inherent in this analysis
adequately discussed? Does the sensitivity analysis cover the major
parameters expected to have an effect on the risk assessment?
Dispersion and Deposition Modeling
To develop this risk assessment, computer models have been used with site
specific data on emission rates and meteorological conditions to simulate the air
concentrations and deposition rates for-contaminahts potentially emitted from the WTI
facility. The models used include the Industrial Source Complex - Complex Terrain
Deposition (ISC-COMPDEP), the CALPLJFF, and the INPUFF models. In your review,
please address the following issues.
1. Since the 1993 peer review of the risk assessment plan, a number of efforts
have been completed to reduce the uncertainty associated with the air
dispersion and deposition modeling. These efforts Include the collection of site-
specific data for emission rates and meteorological conditions. Also, a wind
tunnel study was conducted to evaluate the effects of the complex terrain
surrounding the WTI facility. Does the risk assessment document adequately
summarize these activities? Is the link between these data collection efforts,
the air dispersion models, and the risk assessment clearly established?
2. • The results of 12 sets of sensitivity tests indicate that geophysical variables
(e.g., terrain) are more likely to affect dispersion and deposition than emission
variables (e.g., stack temperature). Were these sensitivity analyses adequate?
Comment on the conclusions reached. To further examine the effect of
geophysical variables, wind tunnel testing was conducted to model the terrain
induced flow effects expected near WTI. It was concluded that changes in
peak concentrations attributed to these effects are relatively minor and that the
C-3
-------
1SC-COMPDEP model is sufficiently conservative. Comment on this
conclusion. Have these analyses helped to characterize and/or reduce the
uncertainty in the air dispersion modeling associated with the complex terrain
surroundinp WTI,
3. The ISC-COMPDEP model does not allow for non-steady state conditions such
as calm winds and strong temperature inversions. Therefore, CALPUFF was
used to estimate air dispersion and deposition under these conditions.
However, CALPUFF gave similar peak, 24 hour, and annual average
concentrations as ISC-COMPDEP. Comment on the adequacy of this analysis.
Comment on the conclusions reached. Has this analysis helped to characterize
and/or reduce the uncertainty in the air dispersion modeling associated with
non-steady state meteorological conditions?
4. Atmospheric dispersion modeling was used to estimate air concentrations of
hazardous chemicals for the accident analysis. The SLAB model was used for
vapor releases from spills and the mixing of incompatible wastes. ISC-
COMPDEP was used for releases associated with fires. Comment on the
selection of the models and inputs. Are they appropriate selections?
5. Overall, have adequate sensitivity tests been conducted to demonstrate the
magnitude of variation in concentrations and deposition estimates with model
inputs?
Human Health Risks
Human Health Risk Assessment includes hazard identification, dose-response
evaluation, exposure assessment, and risk characterization. To develop the risk
assessment, potentially exposed populations have been identified and the magnitude,
frequency, and duration of their exposure quantified. This information was then
integrated with the hazard identification and dose response evaluation for the risk
characterization. For this risk assessment, both carcinogenic and non-carcinogenic
health effects have been evaluated. In your review, please comment on the following
issues.
Exposure
1. EPA's Exposure Assessment Guidelines identify certain exposure descriptors
that should be used to characterize exposure estimates. The Guidelines define
high end exposure estimates as those representing Individuals above the 90th
percentiie on the exposure distribution but not higher than the individual in the
population who has the highest exposure. Bounding exposure estimates are
those that are higher than the exposure incurred by the person in the
population with the highest exposure. Central tendency exposure estimates are
defined as the best representation of the center of the exposure distribution
(e.g., arithmetic mean for normal distributions). Comment on whether or not
c-4
-------
the WT! exposure assessment property characterizes each of the exposure
estimates in terms of these descriptors.
2. The factors that go Into estimating a central tendency or high end exposure,
once the population has been defined, include the environmental media
concentration, the Intake rate, and the duration and/or frequency of exposure.
Comment on whether or not the WTI exposure assessment does an adequate
job of describing the logical procedure of combining these factors to develop
central tendency, high end, and/or bounding estimates of exposure for each of
the exposed subpopulations.
3. An important factor in an exposure assessment is identifying all of the important
exposure sources. Please comment on the adequacy of the WTI assessment
in identifying the Important sources and pathways of exposure.
4. Have the key assumptions for estimation of chemical concentration and for
estimation of exposure been identified? Are the magnitude and direction of
effect correct for the assumptions that have been identified?
5. Supposedly, conservative assumptions have been applied in this assessment to
account for uncertainty. Are the conservative assumptions appropriately
factored into the ultimate characterization of what descriptor best applies to
each exposure estimate? Please comment on whether the uncertainties were
confronted in an adequate manner. If they were not, please state what should
be done differently.
Hazard Identification/Dose Response and Risk Characterization
1. To select surrogate compounds for quantitative risk assessment, a two step
process was used in which chemicals were ranked on the basis of emission
rate, toxicity (both cancer and non-cancer), and bioaccumulatjon potential.
Please comment on this selection process. Are the ranking factors
appropriate? Could Important compounds have been omitted from the analysis
based on the ranking procedure?
2. For the majority of the chemicals of concern, traditional approaches to dose
response evaluation were employed (e.g., use of a slope factor for cancer and
use of a RfD/RfC for non-cancer). However for certain chemicals or groups of
compounds a different methodology was used. Specifically, dioxins, furans,
PAHs, lead, mercury, nickel, chromium, acid gases, and particulate matter were
given special consideration. Please comment on the methodology used for
these compounds. Was it appropriate? Have the uncertainties associated with
the methodology been adequately characterized? Comment on the
assumptions used due to a lack of chemical specific data.
C-5
-------
3, Please comment on the selection of the overall population and the various
subpopulations at risk. Were site specific data, such as the informal home
gardening survey, properly utilized to identify these subpopulations?
4. It is stated in the risk assessment that average risk estimates are based on
average emission rates, average air dispersion/deposition within a subarea, and
typical exposure factors. Further, maximum risks are based on average
emission rates, typical exposure factors, and the maximum air concentration
within a subarea. Please comment on this use of the terms average and
maximum risks. Are these 'descriptive terms appropriate given the parameters
used to derive each?
5. Comment on whether or not the non-cancer risks of chemicals of concern have
been adequately addressed by the risk assessment? For example, has an
adequate discussion of endocrine disrupters been provided which either
characterizes their risks or clearly explains why their risks cannot be
characterized? Further, have non-cancer chronic toxicities of dioxins and
furans been adequately addressed in the risk assessment?
6. Please comment on whether or not the uncertainties associated with the
additivity and/or synergy of risks from pollutants emitted together from the WTI
facility are adequately discussed in the risk assessment.
7. Have the key assumptions for estimation of dose and risk been Identified? Are
the magnitude and direction of effect correct for the assumptions that have
been identified? Please comment on whetHer the uncertainties were confronted
in an adequate manner. If they were not, please state what should be done
differently.
8. Please comment on the overall adequacy of the risk characterization. Does the
risk characterization include a statement of confidence in the risk assessment
including a discussion of the major uncertainties. Are the hazard identification,
dose-response assessment, and exposure assessment dearly presented?
Have sufficient risk descriptors which include important subgroups been
presented and discussed?
Screening Ecological Risk Assessment
As with the human health risk assessment, the ecological risk assessment pulls
together elements of exposure analysis and dose-response evaluations to develop a
risk characterization. For the Screening level Ecological Risk Assessment (SERA),
Ecological Chemicals of Concern (ECOC) and indicator species have been identified
to provide conservative estimates of risk. Please address the following issues in your
review.
C-6
-------
1. Are there any components of the SERA which you feel undermine the scientific
validity of the assessment? If so, what are they and can you provide
suggestions to strengthen the identified components?
2. Is the organization of the document dear and does it present the material in a
clear and concise manner consistent with the Framework for Ecological Risk
Assessment (EPA, 1992)?
3. Uncertainties are discussed in numerous sections of the SERA and compose
Section VIII of the SERA. In each case, do these discussions eover.all relevant
and important aspects of the uncertainties which you think should be addressed
in the SERA?
4. In your opinion, what is the weakest and what is the strongest aspect of the
SERA? Can you make any suggestions on how the weakest parts can be
strengthened by the Agency?
5. in Section IJ, are the stressors, ecological effects, and both the assessment and
measurement endpoints adequately characterized? Are the five emission
scenarios'adequate to characterize the exposures for the WTI facility? Are
there other emission scenarios which you think should be included in the
SERA?
6. In Section HI, is the site characterization adequate to support the SERA? Why
or why not?
7. In Section IV, is the tiered process used to identify the ecological chemicals of
concern (ECOC) from the initial list of potential chemicals considered
scientifically defensible? Does application of this tiered approach support the
statement made in the SERA "by focusing on the potential risk from the
selected ECOCs, the SERA provides a thorough screening-level evaluation for.
the WTi facility?" .
9 ' .
8, In Sections V and VI, are the exposure and ecological effects adequately
characterized? Are the most appropriate estimation techniques available used?
Are the assumptions dearly stated?
9. In Section VIII, are there any major elements missing from the risk
•characterization which you think need to be included or which would strengthen
the risk characterization? Does the risk characterization support the summary
and conclusions presented in Section IX?
10. In Section IX, given the assumptions made and the processes, used to select
and evaluate chemicals, receptors, and exposure pathways, do you think the
SERA adequately met its objective of not inadvertently underestimating risk?
C-7
-------
Accident Analysis
The Accident Analysis for the WTI Incinerator involves evaluating the probability
of an emergency incident occurring which results in the release of hazardous waste.
The consequences of this release-are also evaluated using exposure and human
health effects information. Unlike the human health risk assessment which has a
primary goal.of quantifying risks, the accident-analysis typically provides information
that can be used to reduce the likelihood, extent and impact of possible accidents.
Please'comment on the following issues In your review of this aspect of the risk
assessment
1. The WTl accident assessment selected five scenarios for quantitative
evaluation that were considered to be of primary concern. The scenarios are
an on-site spill, an on-site fire, an on-site mixing of incompatible waste, an off-
site spill, and an off-site spill and fire. Please comment on the selection of
these scenarios. Were any significant scenarios missed?
2. Specific chemicals were selected to evaluate each scenario. Please comment
on the selections. Would other chemicals have been more .appropriate?
3. Chemical specific release rates are calculated for each scenario. Please
comment on the procedures used to estimate the release rates. Was an
appropriate approach used?
4. Atmospheric dispersion modeling was used to estimate air concentrations of
hazardous chemicals. Specifically, the SLAB model was used for vapor
releases from spills and the mixing of incompatible wastes. ISC-COMPDEP
was used for releases associated with fires. Comment on the selection of the
models and inputs. Are they appropriate selections? Should other models or
inputs been used?
5. Please comment on the assessment's conclusions on the severity of
consequences and probability of occurrence. Has the report correctly
categorized the severity of the consequences of the different accident
scenarios? Hfas the assessment adequately justified the reported probability of
occurrence of each of the accident events?
6. Key assumptions were made in the identification of accident scenarios and the
description of the conservative and typical events. Included were a description
of the magnitude of the effect of the assumptions and direction of the .effect
Please comment on the assumptions: Are' they justified? Are the descriptions
of the magnitude and directions of the effects correct? Has the accident
assessment adequately confronted the uncertainties involved in doing this type
of analysis? If .not, what else should be done?
C-8
-------
7. Comment on the appropriateness of using IDLH values for characterizing the
severity of consequences in the accident analysis. Comment on the
appropriateness of using 10 X LOG for chemicals for which IDLH values have
not been established.
8. in the accident analysis, IDLH (or 10 X LOG) values were used to determine
the downwind distances over which adverse human health effects might occur,
To evaluate the uncertainty introduced by using the IDLH, a sensitivity analysis
was conducted where these distances were recalculated using the LOG (a more
stringent health criteria). Other sources of uncertainty that are identified in the
accident analysis include concentration averaging times, chemical
concentrations, emission rates, and. meteorological conditions. For most of .
these parameters it is stated that conservative assumptions were used to avoid
underestimating risks. Have the uncertainties inherent in the accident analysis
been adequately characterized? For those parameters where sensitivity
analyses were not conducted, is the conclusion that conservative assumptions
have avoided underestimation valid?
C-9
-------
-------
APPENDIX D
PREMEETING COMMENTS
-------
-------
Combustion Engineering
D-l
-------
-------
Elmar R. Altwicker
This assessment appears to be a considerable improvement over the last one
and the authors are to be complimented for addressing most issues with
thoroughness. My comments focus on certain specific aspects of the emission
portion of the document (Volumes I - III). The Executive Summary could be a bit
more results-oriented.
Facility Description
The facility description is not as complete as it could be, though Figures II-5
and II-6 (Vol. II) give a good over view. The latter is a pre-enhanced carbon
injection system (ECIS) -schematic and while the location(s) of carbon (C)-injection
is confidential, quantities injected appear to be equally so. Since (some) Crinjection
occurs upstream of a electrostatic precipitator (ESP) one would expect some effect on
performance due to a change in dust resistivity. For the 3-field .ESP no rapping cycle
information under ECIS-conditions is provided. From Figures II-6 the ESP-efficiency
works out to 99.63%. Is it different under ECIS-conditions? It is stated that the
removal efficiencies for PCDD/F are assumed to be directly dependent on the
concentration of activated carbon, but there is no way to verify this contention (cf.
below).
Chlorinated Dioxin and Furan (PCDD/F) Emissions
The PCDD/F-data (Vol. I, Table III-l) are of considerable interest. The 1994
results are reviewed slightly differently in the enclosed tables (A-D). Of interest is a
comparison between PT (performance test) and TB (trial burn) for 2/94; the former
gives higher total (PCDD/F), but lower (TEQ); in addition, this PT gave the smallest
value for the ratio (TEQ)./(PCDD/F)., 0.0058 (Table B). However, the ratio (TEQ)-/
X X *"~" . X
(Cl). was the lowest for this TB, though the chlorine feed rate there was the highest
/\. ~ '
(Table C). Unfortunately, all 2,3,7,8-TCDD-emissibn rates are reported (presumably
due to analytical limitations) as "less than" (Table D). Though this means that
these emissions rates are very low, this result does not lend itself to interpretation in
terms of chlorine feed variability and other possible (though not reported)
differences between these tests. Thus, I am not quite sure how this data set (Table
D) was used to generate an. average emission rate of 1.08 x 10r11 and a high-end
D-3
-------
; Elmar R. Altwicker
emission rate of 2.16 x 10'11 g/s for this compound (Vol. Ill, 111-16, Table III-2). Once
the 12/94 results are available, a more sophisticated statistical analysis should be
attempted. It is unfortunate that C-injection locations, C-injection rates and
adjustments in ESP-operating parameters were withheld. Such information would
aid in the interpretation of these results. It is not clear, therefore, that lumping all
26 post ECIS-PCDD/F-results to calculate a 95% UCL is all that meaningful. Another
questions is: are the average and high values (Vol. Ill, Table III-2) within the
analytical results for repeat analyses (if taken)? Some of the differences are less than
30%. Which of the 17 congeners were not detected? The conclusion, that (TEQ) is at
best a weak function of. the Cl-content of the fuel seems reasonable. And using the
8/93 test data represents a conservative approach.
Although phase distribution of PCDD/F is important from a risk assessment
point of view, little seems to have been added between this and the original
assessment. The discussion on partitioning appears unchanged from the previous
one. The statement (Vol. HI, III-ll) that "substances in the stack gas will generally
be present in either the vapor phase or in the particle phase" seems to ignore the
possibility that many compounds, among them PCDD/F, could be partitioned in
both phases in the stack gas. Although the material on partitioning appears in Vol.
Ill, it is applied only in Vol. V, Table IV-5. The numbers in that table appear to
derive from an assumption of T" = Tambient and initial concentrations in the gas
phase. As I noted in my earlier comments (December 1993), Biddleman (1988) cites
ASf/R = 6.79 as an average and gives three references; he does not say that it can be
"satisfactorily estimated". In fact, McKay, et al. (1982) state that it is an average
empirical value and "may be substantially in error for certain compounds". Given
the particle emission rates cited (0.07 g/s), volumetric flow rates, and assuming fly
ash surface areas of less than 10 m2/g, the surface area available in the stack gas
particulate matter would appear to .be substantially greater than 10"4 cm2/cm3
(Vol. Ill, 111-12). The WTI-stack particles should be characterized by surface area and
size distribution to enable meaningful in-stack partition calculations. It is likely
that the nonideality of the ash particle surfaces needs to be described in terms of a
D-4
-------
Elmar R. Altwicker
Freundlich isotherm. To make this approach useful for risk assessment purposes, a
more critical review has to be conducted.
Waste Profile Data/PICs
Why were waste profile sheets from only the first year operation used to
generate waste feed data? It would seem to make sense to compare the estimated
emissions to current sampling/analytical data. Is there a error/reliability estimate
for the generator range of chlorine for use in the correction factor (CF) -equation
„., mol/y Cllanal.
*—r — I / f-< | |
moi/y Cl 1 generator range
How are some of the constituent ranges (0-30, 5-25, etc.) justified? Using these in the
data base refinement (Vol. Ill, Appendix III-l, II-4) assumes that the upper value is
the highest value possible. How certain is one that 0-30% means that the actual
content of a particular compound within that range (Vol. Ill, Appendix III-l, II-9)? Is
"caution of the part of the shipper" the best criterion?
With respect to other PICs and organic residues (Table III-3, III-8) was chloranil
considered as a possible PIC; it is an expected oxidation product of pentachloro- and
other chlorophenols?
It is stated that a quick analysis (to obtain a finger print) is normally
performed on incoming wastes. Can normally be related to some frequency with
which this is done?
On page II-5 (Vol. Ill, Appendix III-l) pumpable and nonpumpable wastes are
defined. The waste profile data sheets (ibid., p. 2, attachment 1) ask (ll.D): is this
waste pumpable? Is there another judgment made later when wastes "... are
aggregated ...?" I attach some importance to the actual sequence of events here.
Both types of waste contain substantial quantities of such compounds as toluene
and MEK, but the firing methods differ.
What is the exact interpretation of Table IV-1 (Appendix III-l)? Presumably
monochlorobenzene (POHC) equals chlorobenzene (PIC). Only ranges are given. Can
we associate these compounds more directly with the PCDD/F-measurements made
during the same trial burn (3/93)? Key assumptions for chapter IV (Table IV-3),
D-5
-------
Elmar R. Altwicker
such as "factor accurately reflects sample loss (2)" and "factor accurately reflects
instrument response (3)" appear to be inadequately. justified in' the text. The
statement that "these data show that there is no large removal effect of a carbon
injection system on the bulk of combustion THC-emissions" seems premature.
Continuous .Monitoring
Several species, i.e., CO, SO2, NOX, etc. are monitored continuously. However,
there is no mention of a) % data capture, and b) number of excursions, if any,
associated with kiln mishaps, etc.
Fugitive Emissions
Under fugitive Ash Emissions, it is mentioned that a monthly fly ash sample
(1994) was analyzed for 80 organics and that none were found. This must be some
sort of composite sample. How was it obtained? How was it stored? What were the
detection limits of the analytical method(s) used?
The use of a coal ash emission factor of 0.107 Ib/ton and multiplying it by 10
seems tenuous. Could the estimate not be .refined in terms of the WTI-fly ash
composition (some fraction of large, carbon particles, present, which presumably'are
of much lower density than ash particles)? It would seem that there might be
enough information to generate this emission factor for this facility.
Most of the other approaches described appear reasonable.
Metals
Though the modeling results appear impressive, some of the assumptions are
not clear. Apparently, there have been no particle size specific measurements since
the 3/93 trial burn. So what were initial number distributions used in the
modeling? If all metals that vaporize subsequently condense to form 0.5 u,m
particle, their density should .be very different from the (typical) ash particles. For
Figure III-6 .(p. 111-42) is the ordinate scale correct? And the basis for the (assumed)
value of the saturation ratio? In spite of some modeling similarities between
chromium and .aluminum, the former would seern to be a poor. surrogate for the
latter; their chemistries differ.
D-6
-------
Elmar R. Altwicker
Elsewhere on the discussion of metal emissions it has been argued that
historical data were insufficient or unknown to assist in the estimation of emissions.
The HIM facility in Biebesheim, Germany, which has been cited previously when
looking at PCDD/F-emissions and C-injection, could be used for metals -emissions
comparison too, though it is not exactly like the WTI-facility. Tillman (1994) and
references cited therein should be a good starting point. Why were waste feed data
not available?
Other
I counted eight kiln outages for full or partial rebricking between 5/27/93 -
1/08/95, with no apparent systematic intervals. Is some thought being given to do
this (rebricking) on a scheduled basis? These and other process upsets can lead to
variable emissions. Some consideration has been given to them. This is a good
starting point, but more refinement is advisable, for example, with respect to actual
duration of emissions after waste cutoff.
D-7
-------
Elmar R. Altwicker
1994 Performance Test (PT) and Trial Burn (TB) Results
for Chlorine, PCDD/F, and TEQ
Table A Mean (x), standard deviation (a) and variability (v)1 for (PCDD/F)T and
(TEQ)
Type and Date
PT, 2/94
TB, 2/94
PT, 4/94
PT, 8/94
Total, (PCDD/F)i
x o v
5.34 1.85 0.35
4.60 1.06 0.23
3.74 0.83 0.22
1.34 0.38 0.28
(TEQ)
x
-------
Elmar R. Altwicker
Table C Chlorine feed rate and ratio of (TEQ) to mean chlorine feed rate, (Cl)
x . x
Chlorine feed rate (TEQ)_/(CI) , x 104
x x
a v
0.16
0.10
0.17
0.21
PT. 2/94
TB, 2/94
PT, 4/94
PT. 8/94
1979
3151
2039
798
342
131 •
359
529
0.173
0.042
0.180
0.660
D-9
-------
Table D Reported 2,3,7,8-TCDD ranges of emission .rates from 1994 TB and PT's; g/s
TB, 2/94 < 2.28 - 3.98 x 10'11 (4)
PT, 2/94 < 2.20 - 6.72 x 10'11 '(5)
t
PT, 4/94 < 3.37 - 6.27 x lO'11 (6)
PT, 8/94 < 2.16 - 5.22 x 10'11 (7)
D-10
-------
Review of Risk Assessment for the WTT Hazardous Waste Incinerator
Barry Dellinger, Ph. D.
Leader, Combustion Engineering Peer Review Panel
Public concern and objections to siting and operating a hazardous waste incineration facility
can be placed in three categories: 1) "nuisance" value, 2) concern over stack emissions of toxic
combustion by-products, and 3) concerns over accidental catastrophic releases of toxic materials.
The nuisance value includes concern over "quality of life" limits such as declining property value,
degradation of aesthetic value of the landscape, increased heavy vehicle traffic, etc. There appears
to be little that can be done from a scientific or policy perspective that can ever resolve the nuisance
issues other than many years of operation without incident As a society, we have not reached the
point of acceptance; in fact, concern is on the increase.
However, science can address the other two areas of concern, which is the purpose of this
risk assessment Two years ago the overall review committee was quite critical of the preliminary
risk assessment The combustion engineering panel, which I chaired, was at least as critical of the
emissions assessment and accidental fire assessment portions of the draft document We furnished
a number of recommendations and suggestions that needed to be implemented in order to
significantly improve the risk assessment
On the basis of a general reading of the entire document the US-EPA should be
congratulated on what initially appears to be a very thoughtful, detailed, and extraordinarily well
documented report This is a thoroughly professional effort in which every possible effort has
been made to be devoid of bias and emotion, 1 have reviewed the combustion emissions section of
the report very closely and find it to be well organized, with the critical issues and assumptions
well presented. At worst, the combustion section represents a summary of what we have learned
about incinerator emissions in die last 15 years of research, and at best, provides a highly advanced
method for assessing incinerator emissions.
I was also impressed by the faithfulness of the EPA in following die recommendations of
the review committee. 3h many cases their efforts were even heroic. Some very difficult tasks
were suggested by the combustion engineering panel that were accomplished and in some cases
approved upon by the preparcis of the report Any criticism that I may have concerning the
assessment of stack emissions is essentially nit-picking and I expect them to have very tittle impact
on die risk assessment results.
Assessment of Stack Emissions Estimation Procedure
As expected, there is little adverse risk from the emissions of toxic metals even when
conservative assumptions are made. The existing data base on metals is well defined, and
modeling is effectively used to fin data gaps when appropriate. While kinetics may play a
significant role in determining the speciation of metals in some cases, the thermodynamic modeling
approach appears to be sufficient to minimize most concerns.
The organic emissions estimation procedure is generally quite rigorous and closely follows
the recommendations of the panel 11 is somewhat simplified from the more idealistic suggestions
of combustion panel but, nevertheless, is, the most complete and professionally executed
assessment that I have seen. I have a few minor criticisms and questions of concern.
The correction factors applied to THC to calculate total organics is not well documented. I
am aware that EPA gathered some data from a field test a few years ago. Surprisingly it is still not
well recognized that total hydrocarbons (THC), as measured by a conventional analyzer, is not
anywhere near the total organic content of the stack effluent Analyzers really only measure Cl
through C5 chemicals whereas the greatest number of chemicals emitted from an incinerator are
polycydic aromatic hydrocarbons (PAH) with more the five carbon atoms in their structure.
D-ll
-------
Although this point is recognized in the report, the total organic to THC collection factor is critical
to calculating the unaccounted-for fraction of the stack emissions. How complete and reliable is
EPA's data base for calculating this critical factor? How does this basis for calculating the
uncharacterized fraction compart to calculation on a mass balance basis? It would be most useful if
this information could be furnished to the committee at the meeting.
My second criticism is also somewhat of a question. The report states on page IV-8 that
prorating of the emissions rate, of the characterized emissions to account for the uncharaetenzcd
traction was considered but not implemented because it would overestimate the carcinogenic impact
and ignore the toxic (i.e., non-carcinogenic) impact I do not understand why both can't be
included. Furthermore, it is not clear why not including the uncharacterized traction is better than
trying to account for it with some associated error. I would like for this decision to be better
explained to the committee;
My third concern is over the number of chemicals reported as analyzed-for but not detected
in the stack during trial or test bums. Examination of table IU-3 reveals that almost no PAHs were
detected. I have never seen a combustion emissions test when there was not a large number of
PAHs present Was the detection limit for PAHs very high? Could the carbon injection system
have eliminated PAHs entirely? If this is the case, then it is very impressive. I would lite for this
to be explained further with possibly more detail on the stack tests made available to the committee.
As a final point on the subject of stack emissions;, the report makes a good point that it is
the nature, as opposed to quantity, of the uncharacterized traction of emissions that creates the
most uncertainty in risk. How can we ever eliminate this concern? If just one of the
Tmcharacterized chemicals-has the toxiciry of 23,7,8 TCDD, then the calculated risk would
probably increase by orders of mkgnitade. As I see it, this is the only real issue about stack
emissions. This is certainly not to say that the event is likely, but I am not sore that I can say it is
unlikely. Can we arrive at a scientific basis for assessing this uncertainty?
Assessment of Accidental Fire Modeling Procedure
I also reviewed the sections relevant to accidental fires at: the facility. At the last review, the
combustion engineering panel recommended that this issue needed .to be addressed in much greater
detail, although die review of this section is now the responsibility of another panel.
I am disappointed in tins portion of the report The tire model itself is very crude. In
addition, it only includes emissions of HG and phosgene in the risk assessment Considering the
depth of assessment of the stack emissions, how can this data not be included in the fire emissions
assessment? Instead, the approach relies on empirical data generated in 1952. It does not include
the mix of chemical likely to occur at the facility or the type of by-products likely to be emitted.
I suspect that the fault lies less in the authors of this report than in the availability of
appropriate fire models, especially those that include thermal decomposition properties of
hazardous chemicals. Having studied hazardous waste incineration for the past 15 years, my
greatest personal concern about living next to a facility would be exposure from such a fire. In
spite of the peril that could be present from just one accident, I know of no effort to develop
appropriate fire models. This is a serious deficiency in tlic scientific infra-structure that does not
allow us to address a serious issue. We have also had die operating permit of one waste
destruction facility in Ohio revoked based on inadequate fire modeling. I would feel deficient as an
environmental scientist if this were allowed to happen again without the proper caution.
I suspect that the authors of the report have done the best they could with fire modeling using
the available models. This risk assessment is not a research project and they should not be
expected to develop a new model as part of a risk assessment However, the available tools must be
improved before adequate risk assessments can be performed for this facility or any other facility.
D-12
-------
Dr. Randy Seeker
Energy and Environmental Research Corporation
January 2,1996
Preliminary Comments on Risk Assessments fro the WTI Hazardous Waste Incinerator
Facility
Incinerator Stack Emissions
The following are preliminary comments focussed on the emission rates and
chemical speciation used in the risk assessments. In general, I would conclude that the
concerns that I have identified ore (probably) not significant to the overall results of die risk
assessment with the possible exception of selenium behavior as discussed below.
Nonetheless there remains some uncertainties and issues that should be addressed prior to
closure of the study.
Much of the stack emissions rates used in the risk assessments ore based upon
actual emissions. The WTI facility is one of the most extensively studied combustion
facilities in the world. Nonetheless there are still some gaps in the emissions data that
could not be (or have not yet been) obtained from direct emissions but rather were
estimated by analogy to emissions of other species. There are two guiding principles Uiai
must be followed when using emissions data from one species to estimate Uic emissions uf
another: (1) use species with similar physical and chemical behavior and (2) use species
with similar concentrations. While the former principle was generally followed, the latter
was ignored and may lead to some additional uncertainties hi the emissions rates thai may
or may not be significant to the overall risk assessment
Metal Emission Rates
The most important example of this issue is the assumed behavior of selenium. No
direct emissions measurements data are currently available from WTI on selenium.
Selenium is correctly identified as a volatile metal that will volatilize at the kiln
temperatures. The thermodynamics of selenium indicates that it is even more volatile than
mercury under most conditions (mercury is more volatile when chlorine is present).
.Selenium was assumed to be captured in the scrubber with the same efficiency as SC>2 as
measured in the trial bum. There was lilUe support given fur this assumption and I would
questions its validity. While there are data in the literature that indicates, selenium is
D-13
-------
Dr. Randy Seeker
Energy and Environmental Research Corporation
January 2,1996
chemically similar to sulfur and that it can be scrubted with high efficiency in scrubbers,
the use of SC>2 systems removal efficiency data from, the trial burn is challenged due to the
very different concentration levels of interest Scrubljers ore generally moss tranfer limited
devices that work better at high concentrations than (it low concentrations due to gas phase
concentration,driving the mass transfer into the liquid phase. The SC>2 concentration in the
trial burn was over 1700 limes higher than the expected average selenium; hence just based
upon the concentration levels alone the selenium emission estimate is expected to be under
estimated. The authors need to further justify the use of the trial burn sulfur data as an
indicator for selenium and to address the impacts of concentration on the emissions of
selenium. It is not clear without further analysis whether the underestimation is significant
to the overall conclusions of the risk assessment An examination of the selenium issue is
recommended at least with a sensitivity analysis to determine if the risk assessment results
arc sensitive to larger selenium emissions levels. If Uiis sensitivity analysis indicates that
the results are sensitive to estimated selenium emissions then further emissions testing may
be warranted. It is noteworthy that die SERA components of the study have indicated the
importance of selenium at the permit standards.
The same issues arc present but at much less degree witli the approach used to
estimate other metal emissions rates for which emissions data were not available (i.e.,
aluminum, barium, copper, nickel, silver, thallium and zinc). The grouping of metals used
in the study used is somewhat inconsistent with other groupings proposed by the European
Union and the EPA Office of Solid Waste regulatory development office. These other
parties group metals as follows:
^
Volatile -Heand Se
Semi volatile - Pb, Cd and Tl
Low volatile - Be, Ba, As, Ag, Ni, Cr, and Sb
The study as discussed in Appendix 1II-1 of chapter 111, did a good job of examining of the
behavior of other metals relative (q one another and grouping metals with similar behaviors.
Clearly aluminum would be low volatility as assumed in the WT1 study. The volatility of
copper is significantly impacted by chlorine and expected to exhibit both semi volatile and
low volatile behavior depending on the level of chlorine present. The other metals are
D-14
-------
Ai'r Dispersion/Deposition Modeling
and Accident Analysis
D-17
-------
-------
W.F. Dabberdt
Preliminary Comments on Atmospheric Dispersion Aspects of
WTI Risk Assessment
The current risk assessment is improved through the use of improved site-
specific meteorological and emissions data.
What percentile concentration is considered "high-end" and how does this
differ from the central tendency value (is this value the median or the
modal value)?
Does the conclusion that "it is not anticipated that any individual in [the
entire} population would develop cancer as a result of exposure-to routine
... emissions" refer to central tendency or high-end exposure?
Why are the permitted stack emissions limits so much larger (10**4) than
the "expected" emission rates?
What is the impact or effect, if any, of the current risk assessment on the
permitability of the proposed second rotary kiln?
How do the unused December 1994 incinerator performance test data
compare with the post-July 1993 data actually used in the assessment?
.What is the significance, if any, of the OctaCDD estimated emission rate
values being nearly an order of magnitude larger than the largest
PCDD/PCDF values actually measured and "used in the WTI RA" -- as
given in Table 111-1?
The percentage of calm conditions - 22% - actually determined to occur
at the WTI site is a large value, and has significant implications for the
type of dispersion model used in the RA. It is surprising that there is a
large degree of consistency between the CALPUFF and ISC-COMPDEP
values of peak one-hour, 24-hour, and annual average concentrations.
The agreement between CALPUFF and ISC-COMPDEP values applies
to stack-level emissions; were the two models compared for accident
D-19
-------
W.F. Dabberdt
i
i
analyses?
Do the CALPUFF and ISC-COMPDEP models predict concentration-
maxima at the same source-receptor distance and orientation, or are the
peak values the same independent of receptor location?
Did the authors consider wind shear as another source of uncertainty in
the dispersion modeling; the report states the significance of valley
channeling of the wind flow, but this observation is not discussed in the
context of the model(s) performance.
Why are on-site impacts not considered for the accident analyses?
Why is it assumed that "most plausible accident scenarios" would affect
only relatively small areas?
It is not reasonable to assume that the most conservative, yet plausible,
accident scenarios are among those that have occurred in the past 18
years at existing incineration facilities. While this is one approach,
another should be based on a failure analysis of on-site storage and
incineration facilities as well as failure of transport facilities. An historical
analysis would not have projected many notable accidents, such as the
Chernobyl, the metam spill in India, and the oleum spill in the San
Francisco Bay Area.
What is the waste storage volume at the facility, and how does it compare
to the 20,000-gal maximum spill scenario? The larger value would
represent a more conservative case.
The ISC-COMPDEP model does not seem as appropriate to fire
scenarios as does the CALPUFF model.
The calm/inversion meteorological accident-scenario is a good one.
"The probability of occurrence is ranked ... on the probability of the
accident event, the probability of the meteorological conditions, and an
estimated waste composition ..." Does this mean that the overall
D-20
-------
W.F. Dabberdt
probability of occurrence is the joint probability? If not, how is the
probability actually estimated?
The summary of the accident analyses is written in a way that appears
predisposed toward demonstrating negligible risk.
Why would "more extreme events, with potentially greater off-site
consequence..." not "affect the overall conclusions of the Accident
Analysis?"
Why do all accidents where the severity of consequence is "catastrophic"
have accompanying probabilities of occurrence that are "very unlikely?"
D-21
-------
-------
Mark Garrison
BACKGROUND
A risk assessment has been performed for the Waste Technologies Industries (WTI)
Hazardous Waste Incinerator Facility located in East Liverpool, Ohio. The draft risk
assessment is documented in seven volumes that were provided for review by letter
dated November 14,1995:
Volume I: Executive Summary
Volume II: Introduction
Volume III: Facility Emissions
Volume IV: Atmospheric Dispersion and Deposition Modeling of Emissions
Volume V: Human Health Risk Assessment
Volume VI: Screening Ecological Risk Assessment
Volume VII: Accident Analysis
No electronic files were provided for this review.
I have performed a review of the WTI risk assessment based on the charge articulated
to peer reviewers. This charge was expressed as a series of issues and questions, first
of a general nature and then specific to the field of expertise of each reviewer, based on
the reviewer's workgroup assignment. As a member of the Air Dispersion Modeling and
Accident Analysis workgroup, I have focussed on the specific issues and questions for
that group. My review consisted of a careful reading of Volumes I and II, a critical
review of Volumes IV and VII, and a cursory review of the other volumes. My comments
are given below following a re-statement of each of the specific issues.
GENERAL COMMENTS
General Issue # 1: Comment on the organization of the risk assessment
document Does the layout follow a logical format? Is the presentation of
information in the document clear, concise and easy to follow?
Page 1
D-23
-------
Mark Garrison
I was generally favorably impressed by the overall layout of the document. I found that
it was relatively easy to find the answers to questions that came to mind while reading
one part of the document, by looking in the Table of Contents of other volumes. I do
have some comments on re-organizing some of the presentation (particularly Volume
IV, Section IV) and other comments on strengthening the presentation that are
presented below.
i
i
i
General Issue # 2: Does the executive summary accurately reflect the data and
methodologies used and the conclusions derived in the risk assessment?
The executive summary provides a reasonably accurate picture of the results of the
dispersion and deposition modeling that are described in detail in Volume IV (some of
my comments on this Volume, as reflected in more detail below, may be appropriate for
summarizing in the executive summary), and also insofar as the numbers in Tables IV-3
and IV-4 appear to be accurate. Table IV-3 should be modified to indicate that maximum
concentrations are annual averages (as opposed to short-term maxima such as 24-hr
values). Also, Table IV-4 does not need to indicate that emission rates are in g/m2/s
since no area sources were modeled. In the Human Health Risk Assessment (HHRA)
section, Table V-4 provides a concise summary of the results of the analysis but I felt
that, since this table represents the "bottom line" for the whole effort, some additional
t
information should be presented in between Tables V-3 and V-4: for example, a
breakdown of risk components from each of the exposure pathways for the worst-case
pollutant. The interested reader would benefit from some more details.
General Issue # 3: Were the major recommendations of the 1993 peer review
workshop for the risk assessment plan addressed?
The major recommendations of the 1993 peer review workshop are listed in Volume II,
page IV-2, and also in Volume IV, pages I-2 and I-3 (I did not review the workshop
report itself). I believe that the documents that I reviewed addressed the major
recommendations and reflected a genuine, dedicated effort to provide additional
information for the overall project. Some specific comments that I have on aspects of
Page 2
D-24
-------
Mark Garrison
the overall analysis, that should be considered prior to finalizing the risk assessment,
are discussed below.
General Issue # 4: As with any risk assessment, there are always additional data
and method development efforts that could be undertaken to reduce the level of
uncertainty. However, are there any major data or methodological gaps that
would preclude the use of this risk assessment for decision making? If so How
should they be addressed?
I do have a number of comments that refer primarily to organization and presentation
and a few things that may have to be looked at in some depth, however, pending a
satisfactory resolution of these issues I do not believe that there are any major gaps that
would preclude the use of this risk assessment for decision-making.
General Issue # 5: What long-term research would you recommend that could
improve risk assessments of this type in the future?
I believe that research in the following areas would help improve risk assessments of
this type in the future:
Given the critical nature of the deposition pathway, additional research on deposition
approaches would be quite beneficial - particularly in terms of developing and testing
wet deposition algorithms that more accurately reflect the physics of this removal
process.
Non-steady state models have not been subject to rigorous testing, particularly in
complex terrain areas. Additional testing and possibly development of CALPUFF or
other models would help improve risk assessments of this type, particularly if the
research is given the additional direction of improving modeling for accidental releases
of-short duration. .
Page 3
D-25
-------
Mark Garrison
DISPERSION AND DEPOSITION MODELING COMMENTS
Dispersion and Deposition Modeling Issue #1: Since the 1993 peer review of the
risk assessment plan, a number of efforts have been completed to reduce the
uncertainty associated with the air dispersion and deposition modeling. These
efforts include the collection of site-specific data for emission rates and
meteorological conditions. Also, a wind tunnel study was conducted to evaluate
the effects of the complex terrain surrounding the WTI facility. Does the risk
assessment document adequately summarize these activities? Is the link
between these data collection efforts, the air dispersion models, and the risk
assessment clearly established?
The two questions posed here will be dealt with in reverse order. I have reviewed the
meteorological data and wind tunnel issues in some detail, and in general I felt that the
documentation of these efforts was reasonably thorough (although see the comments in
response to issue no. 2 below, related to the presentation of sensitivity results). The link
between these efforts and the overall risk assessment is established through the
prediction of ambient concentration and deposition values. These concentration and
deposition values are absolutely indispensable to the estimation of environmental
concentrations and exposures which are the core of the risk assessment. The link
between the additional data collection and the prediction of ambient concentration and
deposition values, therefore, is the key to estiablishing the link that is referred to in the
question. In that sense the link between additional data collection and the risk
assessment has been clearly established.
Although emissions are also an important ingredient in the exposure assessment, I did
j
not review the collection of the site-specific rales presented in Volume III carefully since
I believe that others are doing so. The point in the risk assessment where
concentration/deposition estimates and emission rates are combined and then utilized
either for direct exposure pathways (inhalation) or indirect pathways (e.g. consumption
of contaminated food) is found in Volume V (the HHRA), especially Chapters VI
(environmental concentrations), VII (exposure doses) and VIII (risk.characterization).
Although I did not review Volume V in detail, the development of the risk
characterization was laid out fairly clearly. One thing that would have been valuable in
Page 4
D-26
-------
Mark Garrison
terms of a review but that I was not able to do because of time constraints would have
been to trace a single chemical through the entire process from emissions estimation to
concentration calculation, dose assessment, and risk characterization.
The question as to whether the data collection efforts were adequately summarized will
be dealt with in detail for the meteorological data and the wind tunnel study (but not the
site-specific emissions characterizations).
Meteorological data
The site-specific meteorological data collection effort included measurements taken at
two on-site towers (10 and 30 meters high), and incorporated measurements taken at a
500-ft (152 meter) tower located approximately eight miles away at the Beaver Valley
Power Station.
I have some reservations about the representativeness of the lower levels of the
BVPSMT data to the area in the immediate vicinity of the WTl. My conclusion, from a
modeler's perspective, after looking at Figures III-5 and III-6 (reproductions of
topographic maps of the area surrounding WTl and BVPSMT) is that lower levels of
both towers are likely to show different types of influences: the BVPSMT is located in a
broader area of the valley, not too distant from a significant bend and widening of the •
river. Furthermore, comparing Figures III-3 (WTl 30-meter wind rose), IH-8 (BVPSMT
10-meter wind rose), and 111-10 (BVPSMT 45.7-meter wind rose), it does not appear to
me as though the WTl 30-meter level "fits in" between the BVPSMT 10-meter and 45.7-
meter levels.
Having said this, however, I believe that the use of the BVPSMT wind data in the WTl
analysis does accomplish an important objective: namely, that since the 152-meter level
allows for more cross-valley flow than the WTl 30-meter data, it probably adds a degree
of realism and a degree of conservatism to the analysis - especially since the base
complex terrain model (COMPLEX-!) is a model that is itself widely regarded as
Page 5
D-27
-------
Mark Garrison
extremely conservative. The closest terrain with which the plume from the WTI stack
would interact (or be predicted to interact based on COMPLEX-I) is of course cross-
valley with respect to WTI, and utilizing meteorological data strictly from the on-site 30
meter tower would understate the frequency with which the plume is transported in this
direction (the plume from the stack is buoyant and therefore rises considerably higher
than the 45.7-meter physical stack height - see the discussion below).
I think that the summary of the meteorological data collection and usage should have a
different focus than on making a case for representativeness of the BVPSMT, especially
the lower levels; the focus should be on the use of the BVPSMT wind data to introduce
a degree of realism and conservatism to the analysis. The degree to which the
BVPSMT data is appropriate is also a function of what parameter is. being used -
temperature profiles are probably more widely applicable to different parts of the valley
than speed and direction profiles.
i
The following are some additional comments on the summary of the meteorological data
and its use in concentration and deposition modeling.
i
• The discussion in Section III.C was a little confusing in terms of what time period
of was covered by the meteorological data used in the modeling. The WTI 30-
meter data were analyzed for the time period April 1992 to March 1993.
BVPSMT data were analyzed for the time periods 1986-1990 and 1992.
Evidently the 4/92 to 3/93 time period was used; were BVPSMT data available
(hopefully) for the same time period? Section III.C.4 should clearly state what
time period was actually used. Given that the data set used in the modeling was
a hybrid consisting of data from two locations, some further analysis of the actual
profiles used (e.g., scatter plots of speed or direction values from one level to the
next, stability-dependent wind roses) would I think be very informative. Since the
WTI stack is 45.7 meters high, the actual use of the meteorological profile in the
modeling means that effectively the 30-meter data collected at the WTI site was
not used at all for the stack modeling.
Page 6
D-28
-------
Mark Garrison
A specific analysis of plume height should be included to better illustrate the
elevations typical of plume transport (particularly stable atmospheric conditions).
A table of plume heights is presented below for different meteorological
conditions (stable conditions assumed, plume heights as calculated by
COMPLEX-I, wind speed shown assumed to be at stack top). Since measured
temperature gradients are used in the analysis, the plume elevation analysis
should reflect observed values (by using average values or possibly ranges).
Potential Temperature
Gradient, °C/100m
1.0
1.0
1.0
.1.0
2.0
2.0
2.0
2.0
3.5
3.5
3.5
Wind Speed @
stack top (m/s)
1.0
2.5
5.0
8.0
1,0
2.5
5.0
8.0
1.0
2.5
5.0,
Plume Rise (meters)
115.5
85.2
. 67.5
57.8
91.7
67.6
53.6
45.9
76.1
56.1
44.5
Plume Elevation
(ftmsl)
1,225
1,126
1,067
1,036
1,147
1,068
1 ,022
997
1,096
1,030
992
Note: default PTG for E stability is 2.0 ° C/100 m; for F stability, 3.5 ° C/100m
The plume elevations as illustrated in this table reveal that under most conditions
the plume from the WTl stack is out of the immediate influence of nearby valley
walls (approx. 1000 ft msl) and in a "transition" zone between in-valley and out-of
valley flow. This provides more support, I believe, for looking to other sources of
Page 7
D-29
-------
Mark Garrison
"out-of-valley" wind data, such as the BVPSMT.
• Using measured temperature gradients for plume height calculations is a feature
of many advanced models (including CTDMPLUS and the new AERMOD
model). The use of measured gradients with simpler models such as
COMPLEX-I (which is incorporated into ISC-COMPDEP) introduces some
degree of realism but diminishes the conifidence with which the claim can be
made that the model is conservative for stable case, complex terrain impacts.
Some comparison should be made between measured gradients and the
defaults built into regulatory models to have some means of judging the effect of
using the measured values. If any "minimum" values for stable conditions were
used in the processing, the values should be identified.
i
Wind tunnel study
The wind tunnel modeling study presented in Volume IV, Appendix IV-6, did not fully
resolve all of the complex technical issues associated with the potential for terrain-
induced downwash at the WTI facility. The authors acknowledge this, but state that
nonetheless the broad picture is understood well enough to utilize the wind tunnel
concentration results in a risk assessment of the WTI site. All three reviewers of the
wind tunnel study (R. Hosker, M. Schatzmann, and R. Britter) generally concur that
adequate experimental methods were employed in the study and that the conclusions
are sound, but also contend that some areas of uncertainty remain. The primary areas
of uncertainty (based on my reading of the comments, and highly paraphrased) are first,
the issue of combined effects of terrain and building downwash; second, the fact that the
study did not address convective or nocturnal, stable cases; and finally, the issue of
marginally separating flow in the upwind terrain. Although I believe that none of these
issues would invalidate the way in which wind tunnel results were analyzed and utilized,
the summary of the wind tunnel work that is contained in Section IV.B.6 would benefit
from a discussion of these issues and why they do not invalidate the stated conclusions.
PageS
D-30
-------
Mark Garrison
The "bottom line" in the analysis of terrain downwash is that the inherently conservative
nature of the ISC-COMPDEP modeling produces peak concentrations for similar
meteorological conditions that are much higher than peak concentrations measured in
the wind tunnel, and therefore the model does not need to be changed to specifically
account for the terrain downwash phenomenon. Although I fully agree with this
conclusion (as a practical although perhaps not scientifically satisfying approach), I think
that the discussion in Section IV.B.6 would leave fewer questions unanswered if the
following issues were addressed (as a minor note, the reference to Figures IV-7 through
IV-9 in this section should be changed to Figures IV-11 through IV-13):
• State that in the ISC-COMPDEP modeling conducted for comparison to wind
tunnel results, neutral atmospheric conditions (i.e. stability D) were utilized for
ISC-COMPDEP, if this is the case (if not, explain why).
• Provide some reference to maximum hourly concentrations over all conditions,
which would further help understand the context of the concentrations being
discussed in the overall picture.
• Since in a risk assessment the spatial distribution of concentration patterns, as
well as long-term averages, are generally more relevant than the value of hourly
maximum concentrations, provide some simple means of demonstrating that
model-predicted spatial patterns are not greatly different than tunnel measured
spatial patterns for relevant meteorological conditions; also, provide a qualitative
discussions of the frequency of the conditions depicted (specific to speed and
direction; reference can be made to wind rose patterns presented elsewhere).
• Provide information related to the maximum concentrations separated by model
algorithm (i.e., the ISC part and the COMPLEX-I part of the model).
• Provide a discussion (only a very brief one is necessary) that addresses the
points made by the reviewers of the wind tunnel study. These can be as simple
Page 9
D-31
-------
Mark Garrison
as: stable and convective conditions were addressed by ISC-COMPDEP;
building downwash was simulated by ISC-COMPDEP and, since the terrain-
distorted wind field produces compensating effects (lower wind speeds,
descending mean streamlines), these effects are not likely to exacerbate building
downwash impacts; and the recirculating region was acknowledged by the
authors of the study but deemed to be insignificant in the broad picture of wind
tunnel results - especially since the wind tunnel results were not used directly,
but compared to model results and found to be not critical.
Most of these points only need to be addressed with respect to the actual stack height at
the facility (i.e. 45.7 meters) and not the other stack heights that were examined in the
tunnel.
Dispersion and Deposition Modeling Issue # 2: The results of 12 sets of sensitivity
tests indicate that geophysical variables (e.g., terrain) are more likely to affect
dispersion and deposition than emission variables (e.g., stack temperature).
Were these sensitivity analyses adequate? Comment on the conclusions
reached. To further examine the effect of geophysical variables, wind tunnel
testing was conducted to model the terrain induced flow effects expected near
WTI. It was concluded that changes in peak concentrations attributed to these
effects are relatively minor and that the ISC-COMPDEP model is sufficiently
conservative. Comment on this conclusion. Have these analyses helped to
characterize and/or reduce the uncertainty in the air dispersion modeling
associated with the complex terrain surrounding WTI.
To answer these questions, I reviewed Section IV which includes a section on sensitivity
simulations (Section IV.B) and a section on uncertainty (IV.D). I am not sure where the
reference to 12 sets of sensitivity tests comes from. 13 sets of results are presented in
Table IV-2, but three of them are base case runs and there are really only four types of
tests listed in the table.
My overall comment on the sensitivity analyses presented in Section IV is that they
provided useful and valuable information for characterizing the uncertainty of modeling
results. I would recommend adding only one additional test, namely, the effect on
concentration and deposition values of using default potential temperature gradients
Page 10
D-32
-------
Mark Garrison
instead of measured values from BVPSMT (see the discussion above related to
temperature profiles). I would also recommend performing the sensitivity tests that were
run with a previous version of ISC-COMPDEP with the most recent version (see the
discussion below). I also believe that Section IV could be strengthened by some re-
organization of the section. I found it difficult to sort out base case results (i.e., what
values were actually used in the exposure assessment?), sensitivity runs with the newer
vs. older versions of ISC-COMPDEP, and sensitivity analyses that involved models or
approaches other than ISC-COMPDEP. My recommendations are as follows:
• Confine the discussion of base case results to one section and one table, and
treat all of the sensitivity analyses separately. Table IV-1 presents base case
results and the results of several sensitivity runs that are first mentioned in
Section IV.A but not discussed in detail until Section IV.B. It should be made
clear that the values presented for the base case are those that were carried
forward for use in the exposure assessment ad risk characterization sections of
the risk assessment, after consideration of sensitivity runs and incorporating any
modifications performed as a result of the sensitivity runs. Since the
concentration and deposition values contained in Table IV-1 for the base case
provide a critical link in the risk assessment, I recommend that the area-specific
concentration and deposition values presented in Tables Vll-14 through VII-17 of
Volume V be presented in Volume IV, in Table IV-1. Since the values in Tables
Vll-14, etc. were actually used in calculating risk, presenting those values in
Volume IV would more clearly establish the link between the modeling and the
risk assessment.
• Present the results of the sensitivity runs in one section and one table. This
would include the values currently in Table IV-1, but it should also include the
results of sensitivity analyses selected from Table IV-2 and re-run with the most
current version of ISC-COMPDEP. I believe that the four sets of analyses
presented in Table IV-2 (i.e., "mass < 0.4 urn at 0.03 urn", "vapor modeled as
0.03 urn particle", "no depletion", and "receptor-specific land use") are worth
Page 11
D-33
-------
Mark Garrison
repeating with the latest 1SC-COMPDEP to eliminate any question that the
sensitivity is affected by different model versions. Table IV-2 could be presented
as-is for some historical perspective, but the "insights" provided would be more
valuable if re-run with the latest version of the model - especially since the
location of the maximum impacts changes significantly between versions of ISC-
COMPDEP. '
• Present the calm/fumigation and terrain downwash discussions in separate sub-
sections.
• The section on uncertainty could remain pretty much as-is.
Other comments.related to the questions posed and to sensitivity:
• I fully support the performance of the wind tunnel study as a means of examining
the possible influence of terrain. The results were somewhat surprising, due
apparently to the shielding effect of the terrain allowing for lower wind speeds at
stack top and increased plume rise, but the conclusion that the model should not
be modified is well supported (my specific comments on the terrain downwash
study and the presentation of results are found in the previous section). Terrain
downwash, it should be noted, is totally unrelated to stable-case plume impacts
on elevated terrain that are associated with the highest concentration
predictions.
• One of the sensitivity tests - for receptor-specific land use types - appears to give
results that would suggest that this approach be used in the "base case"
modeling.
Dispersion and Deposition Modeling Issue # 3: The ISC-COMPDEP model does
not allow for non-steady state conditions such as calm winds and strong
Page 12
D-34
-------
Mark Garrison
temperature inversions. Therefore, CALPUFF was used to estimate air dispersion
and deposition under these conditions. However, CALPUFF gave similar peak,
24-hour, and annual average concentrations as ISC-COMPDEP. Comment on the
adequacy of this analysis. Comment on the conclusions reached. Has this
analysis helped to characterize and/or reduce the uncertainty in the air dispersion
modeling associated with non-steady state meteorological conditions?
The issue of calm winds (associated with temperature inversions) and fumigation events
is a valid one to consider in the WTI setting. The CALPUFF model, with its puff and
"slug" sampling functions that allow for near-source assessments, is quite appropriate
for performing the analysis. The usefulness of the analysis performed here, however, is
severely limited based on the following considerations:
• Limiting the analysis to receptors less than stack top: it would seem to me that
interaction with terrain could also be a concern under calm conditions.
CALPUFF has the capability to handle complex terrain, and 1 am not sure why
complex terrain receptors were not modeled for this sensitivity analysis.
• It is not surprising that CALPUFF produced similar values to the ISC part of ISC-
COMPDEP. Minimum wind speeds were apparently set equal to 1.0
meter/second, and meteorological conditions representative of one hour were
used in the analysis. Since a plume can travel 3600 meters in one hour at 1 m/s,
the features of CALPUFF that make it useful for assessing low wind-speed
conditions are not fully realized. Since the BVPSMT data is available in 15-
minute increments and the wind speed threshold 0.27 m/s, it may be more
valuable to use BVPSMT data in this assessment.
I recognize that the desire was to present a limited analysis to examine these
phenomena, but I think that the analysis was so limited as to not provide significant
additional information. One reason that further modeling was not performed with
CALPUFF was that data limitations prevented the full benefits of CALPUFF to be
realized. An analysis that could provide some of the additional insights possible with
CALPUFF would be to create a "synthesized" stagnation event, possibly based on
Page 13
D-35
-------
Mark Garrison
examining the period of record of the BVPSMT measurements. The event could span
several days and the necessary vertical and horizontal data synthesized to represent
expected wind fields (or rather lack of winds) in such an episode. The maximum hourly
and 24-hr concentrations predicted for this event could be compared to concentrations
/
from ISC-COMPDEP, and provide a more meaningful insight into concentration
predictions for stagnation conditions. One thing that could be evaluated as an
alternative or as a supplement to this analysis is whether the conservativeness of the
%
terrain interaction used in the COMPLEX-I part of ISC-COMPDEP is sufficient to "cover"
these phenomena, similar to the argument made for the issue of terrain downwash.
Dispersion and Deposition Modeling Issue # 4: Atmospheric dispersion modeling
was used to estimate air concentrations of hazardous chemicals for the accident
analysis. The SLAB model was used for vapor releases from spills and the
mixing of incompatible wastes. ISC-COMPDEP was used for releases associated
with fires. Comment on the selection of the models and inputs. Are they
appropriate selections?
(Please see my comments in the accident analysis section)
Dispersion and Deposition Modeling Issue # 5: Overall, have adequate sensitivity
tests been conducted to demonstrate the magnitude of variation in
concentrations and deposition estimates with model inputs? >
Overall, I believe that adequate sensitivity tests have been conducted. My previous
comments contain recommendations for a small number of additional tests and different
ways of discussing and presenting results that can be considered for strengthening the
presentation of the results of the dispersion and deposition modeling,
COMMENTS ON THE ACCIDENT ANALYSIS
Please note that my comments on the accident analysis are based on less actual
experience with these issues than with the issues related to modeling. For the most part,
my review consisted of a careful reading of Volume Vtl and a response to the questions
posed based on whether the information presented was logical and credible.
; Page 14
D-3,6
-------
Mark Garrison
Accident Analysis Issue # 1: The WTI accident assessment selected five
scenarios for quantitative evaluation that were considered to be of primary
concern. The scenarios are an on-site spill, an on-site fire, an on-site mixing of
incompatible waste, an off-site spill, and an off-site spill and fire. Please
comment on the selection of these scenarios. Were any significant scenarios
missed?
The selection process described in Volume VII, Section II appears to be quite thorough
and convincing in the logic of selecting both conservative and typical versions of each
scenario. Considering the type of facility and delivery modes that exist for the WTI
incinerator, I do not believe that any significant scenarios were missed.
Accident Analysis Issue # 2: Specific chemicals were selected to evaluate each
scenario. Please comment on the selections. Would other chemicals have been
more appropriate?
The selection of chemicals for quantitative analysis based on the five identified
scenarios appears to have followed an appropriate screening and selection process, as
described in Section III of Volume VII, supplemented by the rankings documented in
Appendix VII-2.
Accident Analysis Issue # 3: Chemical specific release rates are calculated for
each scenario. Please comment on the procedures used to estimate the release
rates. Was an appropriate approach used?
The procedures used to calculate release rates, summarized in Chapter IV and
explained in detail in Appendix III-3, appear to be reasonable. My only specific
comment in terms of the parameters used to calculate the release rate for spills is that
an average ambient temperature of 68°F will understate emissions on hot summer days
(this is acknowledged in the write-up); a higher temperature may be more appropriate
for estimating worst-case emissions.
Accident Analysis Issue # 4: Atmospheric dispersion modeling was used to
estimate air concentrations of hazardous chemicals. Specifically, the SLAB
model was used for vapor releases from spills and the mixing of incompatible
wastes. ISC-COMPDEP was used for releases associated with fires. Comment on
Page 15
D-37
-------
Mark Garrison
the selection of the models and inputs. Are they appropriate selections? Should
other models or inputs have been used?
The SLAB and ISC-COMPDEP are appropriate models for the scenarios analyzed.
Modeling procedures and results are summarized in Chapter V and discussed in detail
in Appendix VII-4. Although it is not possible to provide a detailed review of the inputs
for all runs for both models, my review of selected inputs revealed that for the most part
the models and inputs were configured appropriately. The following comments are a
result of this abbreviated review (note that the first comment could change the results of
the on-site fire analysis):
• For the on-site fire analysis using "conservative" screening meteorology (i.e. the
54 conditions used widely for screening analyses) the wind direction was set to
270 degrees - i.e. wind from the west that will transport a plume to the east.
Receptors were set up in a line with a y-coordinate of 0.0 and an x-coordinate
ranging from 100 meters to 50 km. For a source located at (0.0, 0.0) this set-up
will identify plume centerline concentrations at the stated downwind distances.
The fire "source", however, was located at (186, 105) which means that the fire
plume completely misses the first few receptors (see Appendix VII-4, Attachment
4.A). I made an independent model run with which I reproduced (approximately)
the results for the pn-site fire in Att. 4.A, and then re-ran the model using
coordinates (0.0, 0.0) for the fire source. The maximum concentrations nearly
doubled, with the maximum occurring at 200 meters from the source- I believe
that locating the source at (0.0, 0.0) is the right approach, and that the modeling
should be modified to reflect this approach. I do not believe that this change will
affect the bottom line, since ISC-COMPDEP with real meteorology predicted
results that are in the range of what would be predicted with the correct source
location. .
• This is a minor point, but one that can cause confusion -.the model run titles in
Attachment 4.A and 4.B do not reflect the on-site and off-site fire scenarios as
Page 16
D-38
-------
Mark Garrison
they should (the 4.A title refers to a storage tank rupture, and 4.B refers to a
truck accident).
• The building dimensions used in 4.A (conservative meteorology) used a
minimum building width, at least according to the dimensions identified in 4.B.
This is probably all right, since the height is still less than this width, but a few
words on why the minimum was selected should appear in the write-up.
• In Table V-5, phosgene concentrations at 100 meters are shown as 3 ppm. This
can't be right and should be corrected.
• Presentation of concentrations is given in different units in different places - g/m3
and ppm. The different units make it difficult to cross-check values from one
table to the next; either both units should be presented, or one set of units
should be used consistently.
• As with the analysis of stack impacts, a table showing plume heights for the fire
.scenarios would be helpful in terms of assessing whether the source
characterizations appear realistic.
The expanded CALPUFF analysis is performed (as discussed above),
Accident Analysis Issue # 5: Please comment on the assessment's conclusions
on the severity of consequences and probability of occurrence. Has the report
correctly categorized the severity of the consequences of the different accident
scenarios? Has the assessment adequately justified the reported probability of
occurrence of each of the accident events?
On balance, the severity of consequences and the probability of occurrence for the
scenarios analyzed that are presented in Chapter VI appear to be well supported
(however, see my comment on Issue 7 below). Severity of consequences may have to
be re-visited based on the comments regarding source placement (see the previous
Page 17
D-39
-------
Mark Garrison
comment).
Accident Analysis Issue # 6: Key assumptions were made in the identification of
accident scenarios and the description of the conservative and typical events.
Included were a description of the magnitude of the effect of the assumptions and
direction of the effect Please comment on the assumptions. Are they justified?
Are the descriptions of the magnitude and directions of the effects correct? Has
the accident assessment adequately confronted the uncertainties involved in
doing this type of analysis? If not, what else should be done?
I believe that the key assumptions discussion and tables presented in each chapter for
the most part identify the magnitudes and direction of the effects correctly. Time
constraints prevent a detailed discussion of each instance where I would have given a
slightly different estimate for the magnitude of effects, although time at the workshop
should be devoted to this issue.
Accident Analysis Issue # 7: Comment on the appropriateness of using IDLH
values for characterizing the severity of consequences in the accident analysis.
Comment on the appropriateness of using 10 X LOC for chemicals for which IDLH
values have not been established.
I believe that it is appropriate to consider some level of impact lower than the IDLH to
further qualify the severity of consequences. An event that produces an impact less
than both the LOC and IDLH is in my mind considerably different than an event that
produces an impact nine times the LOC but less than the IDLH. For the present
analysis, the only chemicals for which there is a large difference between the two are
HCI and phosgene (see Table VII-5 - factor of 5 for HCI, factor of 10 for phosgene). The
additional information presented in Chapter VII related to distances to the LOC should,
in my opinion, be presented directly in the analysis of severity of consequences. Since
the information is available and is presented in the report, its consideration directly in the
determination of severity of consequences (as long as the significance of the IDLH and
LOC thresholds is clearly spelled out) would seem to me to make sense.
Accident Analysis Issue # 8: In the accident analysis, IDLH (or 10 X LOC) values
Page 18
D-40
-------
Mark Garrison
were used to determine the downwind distances over which adverse human
health effects might occur. To evaluate the uncertainty introduced by using the
IDLH, a sensitivity analysis was conducted where these distances were
recalculated using the LOC (a more stringent health criteria). Other sources of
uncertainty that are identified in the accident analysis include concentration
averaging times, chemical concentrations, emission rates, and meteorological
conditions. For most of these parameters it is stated that conservative
assumptions were sued to avoid underestimating risks. Have the uncertainties
inherent in the accident analysis been adequately characterized? For those
parameters where sensitivity analyses were not conducted, is the conclusion that
conservative assumptions have avoided underestimation valid?
I believe that the accident analysis has been conducted with appropriately conservative,
assumptions and inputs (I feel more confident about this conclusion as it relates to the
modeling than to the emissions estimates, with which I have less experience). As stated
in the previous comment, I believe that the LOC values should be considered directly in
the severity of consequences presentation.
ADDITIONAL COMMENTS (NOT SPECIFICALLY SOLICITED)
• In Volume V (HHRA), the presentation of deposition values does not appear to be
consistent with values presented in Volume IV (modeling). To illustrate: Volume IV,
max concentration and deposition values are reported as 0.9111 ug/m3-g/s and
0.3052 g/m2/yr-g/s (surface area distribution); Volume V, Table VII-14, E1 subarea,
identifies a concentration of 0.91 ug/m3-g/s (consistent with Volume IV) and a
depostion value of 0.025 (wet) + 0.0052 (dry) = 0.0302 g/m*/yr-g/s. The deposition
value appears to be about one-tenth of the value reported in Volume IV. Unless I
am missing something, one or the other is right but they both can't be right.
• In Volume III (emissions characterization) emissions from the ash unloading
operation are captured and vented to a baghouse._The only emissions quantified
are those emitted from the baghouse itself. This assumes that 100% of the
emissions are captured. If this is the case, then it should be stated (what is stated is
Page 19
D-41
-------
Mark Garrison
that there is a "fraction escaping capture"). If some emissions are anticipated to
escape capture, then those emissions should be quantified and modeled or, if they
are inconsequential, a statement should be made that they are.
In Volume V (HHRA), Table VIII-3 compares facility impacts of "critera" pollutants to
the National Ambient Air Quality Standards (IMAAQS). The comparisons would be
more meaningful if each pollutant's "significant impact level" was also identified on
this table, since a true comparison to the NA^QS should include total concentrations
(i.e., due to all sources and assumed background).
In Volume IV (modeling) on page IV-9 the reference to Figure IV-1 should be
changed to Figure IV-5, and subsequent references to figures should be
incremented by 1.
Page 20
-------
H.Harrison
page 1 of 4
A Review of:
RISK ASSESSMENT FOR THE WAST2 TECHNOLOGIES INDUSTRIES [WTI]
HAZARDOUS WASTE INCINERATOR FACILITY [EAST LIVERPOOL, OHIO]
SPA Region V
with A.T. Kearney, Inc. Chicago IL
Draft, Dae. 1995
Review by: Halstead Harrison
Atmospheric Sciences
University of Washington
Seattle, WA 98195-1640
Tel: (206)-543-4596
-543-0308 [FAX]
harr i songatmos. Washington. edu
December 20, 1995
Phase I of an EPA review process on potential health -and
environmental effects that nay be associated with a high
temperature toxic waste incinerator presently operating
in eastern Ohio [Dec. 1993] included recommendations that
Phase ±1 should consider potential effects of accidents and
of plume downwasih, that non-steady state pollutant dispersion
models should be* exercised with improved meteorological data,
and that attentipn should be paid to wet and dry removal
processes.
The present Phasie II draft diligently touches all these bases.
The draft's layout is logical and clear. It is not concise,
nor should it be>. With mild reservations, I judge the
executive summary of volumes I-IV reasonably represents the
data, procedurea, and conclusions described in the bulk of
this draft. Theare are methodological gaps and presentational
shortcomings that diminish the usefulness of this risk
assessment for wdsa decision making:
1. WTI does notf live alone, and the permitting process has
to start at the margin, not from zero. Emissions from
local housing and valley industries should be accounted also,
especially from wood smoke, that asphalt plant, and the
refinery. What is the history of air-guality in this valley?
By what increment is WTI expected to degrade this quality?
D-43
-------
H. Harrison
page 2 of 4
2. The draft assessment well describes exercises with the
The COMPDSP dispersion model for annual averages [Vol IV,
IV-28]. Cursory mention is made of exercises to estimate
largest 1-hr averages with the CALPUFF and INPUFF models
[Vol 1, IV-4]. These appear to b« summarized only in a
single two-line table "[Table IV-3, Vol IV, IV-26], without
supporting information on the assumptions that produced those
numbers.
Host of the COMBDEF modeling appears .. sensibly .. to have
b«en conducted with standard 1 gm/s sources. Figures iv-i
through IV-4, and most of the figures lv-4-1 through IV-4-45
appear to be presented as relative concentrations and
depositions, but it is not clear to me whether all of these
figures ara so: their captions should be more explicit.
At any event, converting relative numbers for each tracer
requires absolute emission rates, which are separately
tabulated for many tracers in tables III-l through III-5.
It would be useful if this were made clearer to the reader
by bringing th
-------
H. Harrison
page 3 of 4
4. The summary asserts that aaong metals "the risks are
highest for thallium, selenium, and nickel" [Vol I, 1-8],
. Table. III-4. [VoJj I, 111-14] lists mercury emission* of 0.0014
g/s [44 kg/yr}, higher on a -weight basis than either selenium
or thallium. All *^i«se metals -are 'seriously toxic.
5. The emissions of nitrogen oxides are estimated at 2,4
•g/s £ Table JII<-5, Vol I, III-5]. This converts to a non-
trivial incremenft of about 16 ppb(v) NO2 in a well-mixed plume
1 km wide by lOOi meters high, a^ a wind velocity of i m/s.
Other NOx sources are expected in the valley from cars, a
refinery, and anj asphalt plant. Conditions occur when O3
production at rufral site* is NOX limited at NO2 levels 'below
20 ppb, with O3 at or above Federal primary standard of 120
ppb (Cardelino and Chameides 1990; Chameides et al, 1992] .
It is likely that 'the NOx increment from WT1 will contribute
with other sources towards occasional exceedances of the
Federal O3 standard.
6 . A critique of health effects to be expected from WTI is
beyond my central competence. It is my understanding,
however, that bo£h chronic and acute respiratory effects of
particulate inhalation on huaaij- health have been demonstrated
at low thresholds. [Fierson and Koenig, 1992; Koenig et al,
1994; Larson and Koenig, 1994 J X judge the draft discussion
of non-cancer effects [vol v, vill] to be both inadequate and
excessively compressed into an obscuring "Harard Index" . What
is closer to what we really want to know is the expected
increment, of asthmatic distress ,- especially in children and
the elderly.
This is what hit Oonora.
7. In my judgment, the ambient-air concentration estimates
through CALPtpFl1 and ISC-COHPDEP are optimistic 'under
the rare ..but not implausible case .. of a strong, Donor a-
like inversion, with stagnant winds.
In ay judgment, it is unreasonable to try to control WTI
emissions down to this really worst case: instead, pro- —
visions should be made to shut down all industrial sources,
D-45
-------
H. Harrison
page 4 of 4
and private woodsmoke emissions!, whan the air is severely
stagnant for extended periods. Are adequate, continuous,
and calibrated ^ir-quality monitors, an alerting system,
conservative protocols, and An authoritative regulatory
ae in plac* to do this?-
References:
Card«lino-C-A Chaneid«-W-L £1990]
NATURAL HYDROCARBONS, URBANIZATION, AND URBAN OZONE.
Journal of Geophysical Research, vol.95, no.D9.
pp. 13971-9.
Chameides-W-L Fehsenfeld-F Rodgers-M-O Cardelino-c
Martinez-J Parrlish-D Lonneaan-W Lawson-D-R
Rasmussen-R-A aiauaeraan-P Greenberg-J Middleton-P
W»ng-T [1992]
OZONE FRECURSOK. RELATIONSHIPS IN THE AMBIENT ATMOSPHERE
J.Ge'ophys. Res«a&rch. v97, 05, pp 6037-6055 1992
Koenig-J-Q. Covwrt-D-S . Fierson-W-B . Hanley-Q-S.
Rebolledo-V. Du»»ler-K. McKinney-S-E. £1994]
OX2DANT AND ACID AEROSOL SXFOSORE IK HEALTHY SUBJECTS
AND SUBJECTS WIT£ ASTHMA. PART i: EFFECTS OF OXIDANTS,
COHBINED WITH SOT^FURIC OR NITRIC ACID, ON THE PULMONARY
FUNCTION OP ADOLESCENTS WITH ASTHMA.
Res-Rep-Heal th-Efrf-Inst. 1994 Nov. (70). P 1-36.
Larson-T-V. Koehig-J-Q. £1994]
WOOD SHORE: EMISSIONS AND NONCANCER RESPIRATORY EFFECTS.
An«u-Rev-Public-H«alth. 1994. 15. P 133-56.
Pi«rsen-W-S. Koenig-J-Q. £1992]
RESPIRATORY EFPSpTS OF AIR POLLUTP1ON OH ALLERGIC DISEASE.
J-Aliergy-Clin-Iianunol . 1992 Oct. 90(4 Ft -.1). P 557-66.
D-46
-------
Appendix
H.Harrison
page 1 of 3
APPENDIX
Review of Phase II WTI Risk Assessment
Halstead Harrison
Dec. 22, 1995
In sections 2 and 7 of my review of the draft Phase II WTI
risk assessment report I express doubts that air-quality
simulations are sufficiently conservative for short term
averages and with stagnant winds. It is always unattractive
to state unsupported doubts in the face of what has clearly
been considerable effort by competent people. To support my
judgment I have therefore undertaken a supplementary simula-
tion of pollutant dispersion in a confined river valley.
For this task I have used WPUFF, a Lagrangian-puff dispersion
model that is similar in broad outline to CALPUFF, though the
two models differ in assumptions about wind algorithms and
diffusivities.
Lacking a convenient data base for the topographic relief
in the Ohio river valley, I have adopted a generic valley
[actually a section of the Columbia River, a bit east of
Deshutes],. illustrated in figure 1. The horizontal
dimensions of the modeled domain are 10 x 15 km. The
valley is about 2-4 km wide, which is comparable to the
site near Liverpool, and the highest point of the adjacent
rim is about 500 meters above the valley floor. This is
higher than at Liverpool, but the difference is not relevant
as with stable air at low wind speeds the puffs may not
climb above a few tens of meters. The simulations were
bounded by an inversion lid at 150 meters.
I have simulated three successive days with an assumed wind-
speed distribution illustrated in figure 2. The mean speed
was 0.66 m/s, the mode was 0.50 m/s, and in no period was
the wind speed less than 0.10 m/s. This distribution
resembles stagnation conditions that are observed in
Seattle several days each year.
"Steering wind" directions were assumed as guasi-random,
clustered about the valley axis with successive 12-hr periods
predominantly up- and down-valley. Trajectories are biased
parallel to the valley walls by a mass-conserving algorithm.
Figure 3 illustrates the vertical and horizontal diffusivi-
ties, which were parameterized as proportional to
D-47
-------
Appendix
H. Harrison
page 2 of 3
wind speeds and time-of-day. These curves derive from
angular variances of observed wind-directions and from
vertical temperature gradients measured in the Puyallup
river basin in "class D" stabilities. In severe winter
stagnations [classes E and F] the afternoon maxima of
figure 3 are largely suppressed.
Figures 4 and 5 illustrate surface concentrations averaged
over the final 24 hours of 3-day simulations, with the same
winds. Each was for a 1 gm/s reference source at the base
of the northern wall of figure 1, 7.5 km from the western
edge [just hidden in that figure by the perspective overlap
from the southern valley wall.] The simulations emitted
puffs at one minute intervals. The model advects the puffs
about the valley, with diffusive dilution. At every time
step surface concentrations were averaged into 250 x 250
meter grids. At the last step the resulting concentration
fields were smoothed through a binomial filter with a spatial
coherance length of 1 km.
The two simulations differ only in their source's heights
above the local surface. In figure 4 this height is 80 i
meters [46 m stack + plume rise]; in figure 5 it is 10
meters, to approximate fugitive emissions.
Note in figure 4 that the highest 24-hr surface concentration
associated with the 1 gm/s reference stack source is 3 micro-
grams/mA3 [3047 ng/mA3], which is higher by a factor of 6-15
than the annual averages reported by COMPDEP for the WTI
stack [figure IV-l, vol IV, IV-28], but comparable to the
4.48 micrograms/mA3 listed as a 24-hr average by CALPUFF in
Table IV-3 [vol IV, V-26]. I am-not certain how to interpret
this last number, however, as the draft assessment document
does not identify it as stack or fugitive, or, indeed,
whether it is an absolute number, or relative to a 1 gm/s
emission source. [I take it as likely to be "stack" and
"relative".]
Note in figure 5 that WPUFF estimates the highest 24-hr
surface concentrations associated with a 1 g/s fugitive
emission source to be 70 micrograms/mA3, a factor of 230
higher that with the same emissions from the stack.
The* present simulations do NOT approximate a really severe
stagnation, as might perhaps occur once per decade. The
wind-speed distribution of figure 2 is typical of several
events per year in Seattle weather.- The stack simulations
of figure 4 are likely low owing to the neglect of coherent
D-48
-------
Appendix
H.•Harrison
page 3 of 3
vertical mixing processes from downwash,. from eddies pro-
duced by the complex terrain, and from valley oscillations
or "rolls". The simulations of figure 5 are likely low
owing to overestimates of afternoon ventilations during
very stagnant air, in figure 3. Both simulations are
certainly low owing to spatial averaging necessary to re-
duce the "shot" noise,associated with finite puff numbers,
some of residue of which can be seen to remain in figure 4.
Resolving these reservations is beyond the resources of
this present brief effort. As a guess .. and emphasizing
that it is only a guess .'. an additional factor of 3-5
might be applied to both figures to estimate highest 24-hr
averages from 1 gm/s sources, to be expected once a decade.
D-49
-------
Smoothed Isometric
A generic river valley.
Horizontal dimensions are 10 x 15 km
The vertical scale is exaggerated.
D-50
-------
Mean = 6.61D-81 Skeu = 7.71E-81
S.D. = 3.15E-81 Kurt = 3.22E+08
—
—
H = 72 points
Simulated
Ua 1 ley-Stagnat ion —
Uind-Speeds ~
—
;
8
1.8 — U (ra/s) —> 2.8
Histogram of wind speeds for a 3-day
stagnation, expected once or more per year
D-51
-------
B.I
8.06
8.64
8.02
• 8.68
Ky = Ey U SigY
Kz = Ez U Z
<- Ey
local time —•>
12
24
Horizontal [Ey]. and vertical [Ez] entrainraent
coefficiencts, measured with "Class' D" stability
in the Puyallup river valley.
D-52
-------
Smoothed Isometric
Isometric plot of 24-hour averaged tracer
distributions associated with a 1 gm/s
source 80 meters above the local terrain.
The contours have been smoothed through a
binomial filter with 1 km coherance.
Residual sampling noise is still apparent.
D-53
-------
Isometric plot of 24-hour averaged tracer
distributions associated with a 1 gm/s
source 10 meters above the local terrain.
The contours have .been smoothed through a
binomial filter with 1 km coherance.
D-54
-------
Pre-Meeting Comments (December IS95/
Hazardous Waste Incineraccr
Jerry Havens - Page 1
Pre-Meeting Comments on the Risk. Assessment for the Waste Technologies
Incorporated Hazardous Waste Incinerator
Located in East Liverpool. Ohio
Prepared by: Jerry Havens
Distinguished Professor of Chemical Engineering
University of Arkansas
This, multifaceted study appears to have been considered carefully
and there is evidence that the advice of the scientific community was
sought, received, and acted upon. With an important exception
(discussed below), I found this risk assessment to be realistic and
_,/
comprehensive, and I found it to to be fairly presented. In my opinion,
it is worth the considerable cost and effort expended, and I believe it
deserves to be received by all parties concerned as a balanced attempt
to realistically assess the risks associated with the operation of the
hazardous waste incineration facility.
Following my. summary critique of the sections which I feel most
qualified to address, I offer.suggestions for further, consideration.
The Risk Assessment is composed of seven volumes:
~ Volume 1: Executive Summary"
Volume 2. Introduction
Volume 3. Characterization of the Nature and Magnitude of Emissions
Volume 4. Atmospheric Dispersion and Deposition Modeling of
Emissions
Volume 5. Human Health Risk Assessment.(HHRA): Evaluation of
Potential Risks from Multipathway Exposure to Emissions
Volume 6. Screening Ecological Risk Assessment (SERA)
Volume 7. Accident Analysis: Selection and Assessment of Potential
Release Scenarios
There was not sufficient time to consider carefully all of the
material provided. My review focused on the methods used to identify
the potential for releases of hazardous materials, both routine and
D-55
-------
?re-Meet:ing Comments •, December 19S5)
WTI Hazardous Waste Incinerator
Jerry Havens - Page 2
accidental, and on the methodology for estimating the potential
consequences offsite using atmospheric dispersion models.
After reading the Summary and the Introduction, I read carefully
Volumes 3, 4, and 7. I considered Volumes 5 and 6 only in the light of
those sections' dependence on .the results of Volumes 3 and 4.
My comments are divided into two categories:
1. Estimation of risks associated with routine emissions.
2. Estimation of risks associated with accidental releases.
Risk Assessment for Routine Emissions
Identification and quantitative estimation of the materials present
in routine stack and fugitive emissions is a very difficult task,
requiring realistic, accurate forecasts of the schedules for receiving,
processing, temporary storage, and incineration of several hundred
potentially hazardous chemicals. In addition to the hazards associated
with the potential for release of individual chemicals, requirements for
segregation of the materials to preclude reactive conditions (which
could cause, or increase the severity of a release) considerably
complicate the realization of safe operation. Nevertheless, I believe
the simplification adopted to identify surrogate hazardous chemicals,
based on forecasts of the incinerator facility's receipts, is
reasonable, and I doubt that the information upon which the forecasts
are based would justify a more specific approach.
Following the identification of surrogate chemicals, the potential
rates of release (routine emissions) are specified partly by reference
to the expected operating characteristics of the release"control
technology which is applied (which appears to be state of the art) and
partly by reference to measurements .conducted at the site. This
procedure appears reasonable, and I cannot suggest improved
alternatives.
Given the specification of the materials, amounts, and physical.
states (solid, liquid, and vapor) of the (routine) releases., the
D-56
-------
eeeing Comments (December 1
WTI Hazardous Waste Incinerator
Jerry Havens - Page 3
estimation of the potential -off-site consequences is based on'the use of
atmospheric- dispersion models. I found this part of• the overall risk
assessment to be thorough and comprehensive.
The modeling of routine emissions is based primarily-on the ISC-
COMPDEP model developed by EPA. A technical description of the ISC-
COMPDEP^model is provided in sufficient detail to allow assessment by an
independent reviewer. The ISC-COMPDEP model is supplemented by modeling-
exercises with the CALPUFF and INPUFF models to provide for address of
low wind/calm conditions which are-specified to occur (on average) about
22% of the time. It appears that the ISC-COMPDEP model is the best
modeling procedure available for the modeling of the routine releases
from the WTI facility. The CALPUFF and INPUFF models strengthen the
process by providing for estimates of the dispersion of transient
emissions and emissions in calm conditions. It is important that the
CALPUFF and INPUFF model applications indicate that the maximum distance
indicated for offsite consequences (from routine releases) occurs under
other-than-calm conditions. The modeling methodology provides for
consideration of tfte following aspects which are important for
realistically modeling the atmospheric dispersion of the effluent from
the WTI facility:
1. Terrain features, including the effects of stack height less
than and greater than the surrounding terrain elevations.
2. Use of. on-site and/or near-site meteorological data, including
precipitation data for wet deposition calculations and
turbulence measurements for dispersion estimation.
3. Provision for evaluating the short-term concentration increases
resulting from process upset conditions.
4. Provision for evaluation of the effects of calm wind conditions
and fumigation on short- and long-term concentrations in and
beyond the valley.
5. Provision for evaluation of the impacts of fugitive emissions.
The physical modeling study provided by EPA's Fluid .Modeling
Facility (FMF) effectively addresses the issue of terrain downwash
D-57
-------
?re-Meeting Comments'(December 1955'-
WTI Hazardous Waste Incinerator
Jerry Havens •- Page 4
(which is not explicitly accounted for in the ISC-COMPDEP model). The;
comparisons between the FMF wind tunnel predictions and the ISC-COMPDEP
model predictions greatly strengthen the credibility and reliability of
the process.
Overall, I found the methodology used to identify and assess the
offsite risks due to routine emissions to:be thorough and reasonably
complete.
Risk Assessment for Accidental Releases
In parallel with the methods employed for estimating the
consequences of routine releases, a three-step process was used to
define the scenarios representative, of the (accidental release) risks,
including, specifically, "worst case" scenarios:
1. Identification and selection of accident scenarios.
2. Specification of chemical-specific (accident) emission rates.'
3. Atmospheric dispersion modeling.
First, my comments on the latter steps. I did not have time to
verify the estimates of (accident) emission rates and (fire) heat ;
effects, which were stated to have Been made using EPA methods as well
as the Automated Resource for Chemical Hazard Incident Evaluation'.
(ARCHIE) model developed by FEMA. However, I have no reason to believe
that the correct use of these models would not be adequate to provide
reasonable estimates of the evaporation rates of spilled liquids as well
as the radiative heat effects from pool fires.
Based on the surface meteorological data set developed for use in
the dispersion modeling for routine emissions, three meteorological
conditions were selected for use in the accident analyses: • ' •
1. A "typical" meteorological condition was determined 'to be a
neutral atmosphere with average wind speed (3.2 m/s) . The non-
fire scenarios were modeled for the "typical" meteorology with:
the SLAB model and the fire-related scenarios were modeled with
the ISC-COMPDEP model. '
D-58
-------
?re-Meeting Comments (December 1395-
WTI Hazardous Waste Incineracor
Jerry Havens - Page 5
2. A "conservative" meteorological condition was determined by
evaluating 54 combinations of atmospheric stability and wind
speed to determine which combination resulted in the maximum
downwind ground-level concentrations off-site. The 54
combinations were used to determine the conservative
meteorological conditions and the resulting concentrations using
both the SLAB model (for non-fire scenarios) and the .ISC-COMPDEP
model (for fire scenarios).
3. The "Calm/Inversion" condition assumes that emissions accumulate
in the air immediately'above the source for'one hour during calm
conditions and a stable atmospheric lapse rate and are then
transported downwind with a wind speed of 1 m/s. The limited
mixing in the surface layer imposed by the temperature inversion
is represented by a mixing height of 100 m in the SLAB model,
and the worst-case meteorology is represented by the combination
of 1 m/s wind speed and a stable atmosphere with Monin-Obukov
length of 8.3 m (said to be roughly equivalent to Pasquill-
Gifford atmospheric stability E or F).
Again, I did not have time to verify any of the mathematical model
predictions, either with the models used in the risk assessment or with
alternative models. However, I am familiar with the SLAB model, and I=
•consider it appropriate for the use specified here. The SLAB model has
been validated against several benchmark data sets, and in my opinion
can satisfactorily account for the effects of cloud density on
dispersion. (This requirement is probably important only for the
accidental non-fire releases.) I am not aware of any validation
exercises which test the applicability of the ISC-COMPDEP model to the
fire scenarios, but the provisions for positive buoyancy which the ISC-
COMPDEP model includes are probably as good as any available. In my
opinion, the fire-product dispersion modeling with the ISC-COMPDEP model
is suitable, particularly in view of the strong suggestion that the
dispersion downwind of the fire scenarios considered here do not result
in the maximum distances required for consideration.
D-59
-------
?re-Meeting Comments (December 19S5;
WTI Hazardous Waste Incinerator
Jerry Havens - Page: 6
In my opinion the methods used to estimate the consequences of 'the
(selected) accident scenarios are appropriate, and I have no reason to
doubt that the results provide estimates which are sufficiently accurate
and realistic for the use made in this risk assessment. :
Regarding the identification and selection of accident scenarios,
the assessment identifies (five) scenarios considered to be of primary
concern: :
1. On-site spill of 90% methanol/10% formaldehyde (worst-case
waste), or 90% toluene/10% acetone (typical waste) with
formaldehyde or toluene, respectively, released to the |
atmosphere.
2. On-site pool fire involving 15% tetrachloroethene/85% toluene,
with hydrogen chloride and phosgene released to the atmosphere,.
3. On-site mixing of incompatible wastes, consisting of 15%
tetrachloroethene/85% methanol mixed with waste consisting of
70% nitric acid, with hydrogen chloride released to the
atmosphere.
4. Off-site tanker truck spill of waste consisting of 90%
methanol/10% formaldehyde (worst-case waste) or 90% toluene/10%
acetone (typical waste), with formaldehyde and acetone,
respectively, released to the atmosphere.
5. Off-site pool fire involving 85% toluene/15% tetrachloroethene
with hydrogen chloride and phosgene released to the atmosphere.
I have no basic disagreement with the; methods-used to quantify the
the consequences of these accident scenarios, including the proposed use
of IDLH and LOG for characterizing the severity of consequences.
However, I am concerned that potentially important accident
scenarios have not been given sufficient attention. Specifically, I
noted that the organic waste tank farm consists of several tanks (I
believe six are indicated) with individual volumes of approximately
20000 ga-llons. The tanks are enclosed in. a building that has four vents
to the atmosphere. I found no specification of the (planned or actual)
D-60
-------
Pre-Meeeing Comments (December 1995)
WTI Hazardous Waste incinerator
Je.rry Havens - Page 7
contents of these tanks, nor of the potential for interconnection
(planned or otherwise) between tanks. Although there is an indication
that the tanks are diked and that fire protection is provided, there is
no clear specification of the separation or segregation (by diking) in
the tank farm. I did not find a description of the provisions
(managerial or technical) made to preclude accidental mixing of reactive
chemicals in the tank farm or in the collection system which feeds
effluents from the tank-farm area to the carbon absorption bed.
It is repeatedly assumed in the specification of the five accident
scenarios of greatest concern that explosion and/or BLEVE incidents are
sufficiently unlikely that they are not included. It may be true that
sufficient measures have been taken -to prevent (for example) fire
involvement of multiple tanks in the organic waste'tank, farm in the
event of a pool fire in the enclosure, but -in my opinion the assessment
does not satisfactorily consider this question. There is no information
given to allow specification of the contents of the large tanks in the
organic waste tank farm. Without further information, given the
indicated individual tank size of 20000 gallons, and the siting of as
many as six tanks in a building which is assumed not to be designed to
contain either explosion overpressures or gases which might be released
in the event of emergency pressure relief operations, it is not
justified to dismiss these'potentially catastrophic scenarios as being
"sufficiently", unlikely.
Having opened this can of worms, I would expect that the questions I
have raised were considered in the design and in the plans for operation
of the facility, and that information may be available which would
alleviate the concerns which I have stated. If that is the case, the
Risk Assessment.should address those issues carefully.
.Overall, I am most concerned that in the effort to deal
quantitatively with the technical questions of emission estimation and
atmospheric dispersion, important questions regarding the provision of
good engineering design and operation principles to minimize the
probability (and potential consequences) of catastrophic events have not
been given sufficient attention.
D-61
-------
-------
Geoffrey D. Kaiser
REVIEW OF THE RISK ASSESSMENT
FOR THE
WASTE TECHNOLOGIES INDUSTRIES (WTI) HAZARDOUS WASTE INCINERATOR
FACILITY (EAST LIVERPOOL, OHIO)
by
Geoffrey D. Kaiser
Science Applications International Corporation
November 1995
GENERAL QUESTIONS
1. Comment on the organization of the risk assessment document. Does the layout
follow a logical format? Is the presentation of information in the document clear,
concise and easy to follow?
The organization of the risk assessment is logical. The presentation is generally clear, but the
document can hardly be said to be concise. The presentation is generally easy to follow
except for Volume VII, the Risk Assessment, which often does not clearly explain
assumptions and in which it is often difficult to see how results were obtained. See below for
my extensive.comments on Volume VEL
2. Does the executive summary accurately reflect the data and methodologies used
and the conclusions derived in the risk assessment?
The Executive Summary is clearly written and adequately summarizes those portions of the
report that I have reviewed (chiefly Volumes II, IV and VII). However, on pages VII-6 and
VII-7 of the Executive Summary ( and Table VI-1 of Volume VII) the authors attribute an
accident severity ranking to FEMA which does not seem to actually be in one-to-one
correspondence with the FEMA scheme:
D-63
-------
Geoffrey D. Kaiser
Severity
Ranking
From Page VU-7 of the
Executive Summary
Actual FEMA Definitions
Minor
No -exoeedanee of ][DLH value in
inhabited off-site areas; and
negligible potential for off-site
fatalities or serious injuries
due to heat effects from a fire.
Low potential for serious human
injuries; no potential for
human fatalities; and no need
for a formal evacuation,
although the public may be
cleared from the immediate area
of the spill or discharge.
Moderate
Exceedance of IDLE! values in
inhabited areas over distances of
100 meters or less; injuries due
to heat effects limited to a
distance of 1,000 meters into
inhabited areas.
Up to 100 potential human
injuries requiring medical
treatment or observation; up- to
10 potential human fatalities;
or evacuation of up to 2,000
people. !
Major
Exceedance of IDLE! values in
inhabited areas over distances
between 100 meters and 1,000
meters; injuries due to heat
effects limited to a distance of
1,000 meters into inhabited
areas.
Up to several hundred potential
human injuries requiring medical
treatment; up to 100 potential
human fatalities; or evacuation-
of up to 20,000 people.
Catastrophic
Exceedance of IDLH values in
inhabited off-site areas over
distances greater than 1,000
meters; injuries due to heat
effects extend to distances
greater'than 1,000 meters into
inhabited areas.
More than 300 potential human
injuries requiring.formal
medical treatment; more than 100
potential human fatalities; or
evacuation of more than 20,000
people.
In Volume VII, the authors do not show why they believe that there is a one-to-one
correspondence between the two sets of definitions. In my answer to Question 5 under
Volume VII (see below) I explain why the authors, by adopting the definitions in Column 2 of
the above table, seem to have introduced considerable conservatisms for which I cannot find
an explanation. Furthermore, in the summary Tables; VII-2 and-3 in the Executive Summary
(Tables Vm-1,-2 and -3 in Volume VII), the authors appear also to have introduced
conservatisms into the frequency assignments (see also my answer to Question 5 under
Volume VII). Consequently, the summary of results on the bottom of p. VII-8 continuing on
p. VII-9 of the Executive Summary overstates the risks arising from the accident analysis.
D-64
-------
Geoffrey D. Kaiser
3. Were the major recommendations of the 1993 peer review workshop for the risk
assessment plan addressed? -
The major recommendations of the 1993 peer review workshop in the areas which I have
reviewed are as follows:
o In developing an appropriate meteorological data set for the air dispersion
modeling, it was suggested that site-specific meteorological observations be
combined with Beaver Valley Power Station data collected at multiple
elevations.
o Wet deposition estimates were recommended to be refined using local
precipitation data.
o Fumigation conditions and terrain induced downwash were identified as
having the potential to cause locally elevated concentrations. Further,
evaluation of such conditions by modeling or by conducting wind tunnel
studies was suggested.
o Sensitivity and uncertainty analyses were recommended to estimate the
uncertainty of the model's concentration and deposition outputs.
An impressive amount of work has been done to address these recommendations.
4. As with any risk assessment, there are always additional data and method
development efforts that could be undertaken to reduce the level of uncertainty.
However, are there any major data or methodological gaps that would preclude
the use of the risk assessment for decision making? If so, how should they be
addressed?
D-65
-------
Geoffrey D. Kaiser
I do not think that additional short term (i.e. realistically accomplished within a year or two)
data or method development, efforts would be cost beneficial in the short term. The analyses
presented in the areas that I have reviewed are consistent with the state-of-the-art and further
work would not lead to further insights. In other words, further data or methods development
efforts would not give decision makers additional insights. The one exception is Volume VII,
which could with profit be rewritten to enhance the clarity of explanation and either to remove
or justify currently unexplained conservatisms (see below).
5. What long-term research would you recommend that could improve risk
assessments of this type in the future?
Specifically for risk assessments of the type reported in Volume VII, long term research
devoted to determining the levels of airborne concentration that will cause injury or fatality as
a function of exposure time would be very helpful, especially if it could be expressed in the
probit format. Such data are currently available for very few toxic substances.. :
D-66
-------
Geoffrey D. Kaiser
ATMOSPHERIC DISPERSION AND DEPOSITION MODELING
1. Since the 1993 peer review of the risk assessment plan, a number of efforts have
been completed to reduce the uncertainty associated with the air dispersion and
deposition modeling. These efforts include the collection of site-specific data for
*
emission rates and meteorological conditions. Also, a wind tunnel study was
conducted to evaluate the effects of the complex terrain surrounding the WTI
facility. Does the risk assessment document summarize these activities? Is the
link between these data collection effects, the air dispersion models and the risk
assessment clearly established?
Yes. This part of the work has been well done and is well explained.
2. The results of 12 sets of sensitivity tests indicate that geophysical variables (e.g.
terrain) are more likely to affect dispersion and deposition than emission
variables (e.g. stack temperature). Were these sensitivity analyses adequate?
Comment on the conclusions reached. To further examine the effect of physical
variables, wind tunnel testing was conducted to model the terrain induced flow
effects near WTI. It was concluded that changes in peak concentrations
attributed to these effects are relatively minor and that the ISC-COMPDEP
model is sufficiently conservative. Comment on this conclusion. Have these
analyses helped to characterize and reduce the uncertainty in the air dispersion
modeling associated with the complex terrain surrounding WTI?
a) Sufficient sensitivity studies have been performed.
b) The wind tunnel modeling was conducted thoroughly and with exemplary
professionalism. The conclusions arising from the wind tunnel work are credible.
D-67
-------
Geoffrey D. Kaiser
c) The analyses have helped to reduce the uncertainty in the air dispersion modeling
associated with the complex terrain surrounding WTI.
3. The ISC-COMPDEP model does not allow for non-steady state conditions such as
calm winds and strong temperature inversions. Therefore, CALPUFF was used
to estimate air dispersion and deposition under these conditions. However,
CALPUFF gave similar peak, 24 hour and annual average concentrations as ISC-
COMPDEP. Comment on the adequacy of (this analysis. Comment on the
conclusions reached. Has this analysis helped to characterize and/or reduce the
uncertainty in the air dispersion modeling associated with non-steady state
meteorological conditions?
The CAL-PUFF analysis is credible and helps make conclusions drawn from air dispersion
modeling hi non-steady state conditions more robust. It has helped to characterize the
uncertainties in the modeling.
4. Atmospheric dispersion modeling was used to estimate air concentrations of
hazardous chemicals for the accident analysis. The SLAB model was used for
vapor releases from spills and the mixing of incompatible wastes. ISC-
COMPDEP was used for releases associated with fires. Comment on the selection
of the models and inputs. Are they appropriate selections?
See the answer below to identical question 4 under Volume VII
5. Overall, have adequate sensitivity tests been conducted to demonstrate the
magnitude of variation in concentrations and deposition estimates with model
inputs?
Yes.
D-68
-------
Geoffrey D. Kaiser
ACCIDENT ANALYSIS - VOLUME VH
A. GENERAL COMMENTS
o There appear to be unjustified (or, at least, unexplained) conservatisms introduced into
both the frequency and the magnitude of the consequences. See the answer to
Question 5 below for an explanation of this observation. This means .that the
"probability/severity" matrices Tables VIII-1,2 and-3 consistently overestimate the
risk and that the conclusions in Chapter 8 overstate the risk.
o Throughout volume VII, the word "probability" is used when the word "frequency" is
what is meant. Probability is dimensionless, whereas frequency has dimensions of
events per unit time. The authors should review wherever "probability" is used and
replace it by "frequency" in almost every case.
o The use of "averaging time" is sometimes confusing because it is used in two different
ways. In some places, it is used as the time that is placed in the equation for the
increase in plume width as a function of time. In other places, it is used as the length
of time for which an individual is exposed to the passing plume. Paragraph 11 on
Pages 26 and 27 of Appendix VII-4 is an example where the use of the terminology is
particularly confusing. Perhaps different phrases could be used such as "dispersion
averaging time" and "exposure averaging time." The authors should review the entire
volume to make sure that this distinction is always clearly observed.
D-69
-------
Geoffrey D. Kaiser
B. ANSWERS TO SPECIFIC QUESTIONS
1. The WIT accident assessment selected five scenarios for quantitative evaluation
that were considered to be of primary concern. The-scenarios are an on-site spill,
ah on-site fire, an on-site mixing of incompatible waste, an off-site spill, and an
off-site spill and fire. Please comment on the selection of these scenarios. Were
any significant scenarios missed?
The chosen scenarios provide an adequate foundation for the semi-quantitative risk evaluation
hi Volume VII. I do not think any significant scenarios were missed.
2. Specific chemicals were selected to evaluate.each scenario. Please comment-on
the selections. Would other chemicals have been more appropriate?
I doubt that selecting other chemicals would have led to more insights. The use of
formaldehyde as a worst-case chemical and acetone as a typical chemical is appropriate for the
spill scenarios, HC1 and phosgene for the fire scenarios and HC1 for the inadvertent mixing
scenarios.
3. Chemical specific release rates are calculates for each scenario. Please comment
on the procedures used to estimate the release rates. Was an appropriate
approach used?
a) The procedure used to calculate the evaporation rates of spillages of formaldehyde and
acetone in the conservative weather condition and the calm/inversion weather
condition is not appropriate, since the same evaporation rates are used as for the
typical weather condition - that is, Equation (12) on p. 6 of Appendix VII-3 is used
with the same windspeed for all three conditions, namely 3.2 m/s. This evaporation
rate is too high for the conservative condition (windspeed 1.5 m/s) and is completely
inappropriate for the calm/inversion condition where the windspeed is essentially zero.
D-70
-------
Geoffrey D. Kaiser
There are models for evaporation in calm conditions. One such gives the evaporation
rate Q as follows:
Q = 292(1 + 0.51 Rel/2Sc1/3)ln(l/(l-Pv))(ML/T)dX/2 g/min
where Re is the Reynolds number using a windspeed of 0.03 m/s
Sc is the Schmidt number
Pv is the vapor pressure of the liquid (atmospheres)
ML is the molecular weight of the liquid
T is the ambient temperature (K)
d is the diffusivity of the air (cm2/g) and
X is the diameter of the spill (m).
The windspeed of 0.03 m/s used to calculate the Reynolds number here seems to be
consistent with the spreading speed of 0.03 m/s used to represent the growth of the
cloud in calm conditions on p. 19 of Appendix VIM.
Overall, the authors should reconsider their calculations of evaporation rates in
conservative and calm/inversion conditions. The actual release rate used seems too
high and therefore the estimated downwind distances in these conditions are likely too
high.
b) For the calm/inversion condition, the authors assume that, over a period of one hour,
the vapors evaporating from a pool occupy a volume that is 108m x 108m x 100m,
This will not be true for evaporating vapors that are heavier-than-air. These vapors
will slump and form a cloud that may be only a few centimeters in depth (I have seen
videos of experiments at e.g. Porton Down where this happened). When the wind
picks up, it will entrain vapor through a process of quasi-evaporation. I am not sure
that we know how to calculate this, but downwind concentrations will probably not be
very high.
D-71
-------
Geoffrey D. Kaiser
4. Atmospheric dispersion modeling was used to estimate air concentrations of
hazardous chemicals for the accident analysis. Specifically, the SLAB model was
used for vapor releases from spills and the mixing of incompatible wastes. ISC-
COMPDEP was used for releases associated with fires. Comment on the selection
of the models and inputs. Are they appropriate selections? Should other models
or inputs have been used?
a) The use of SLAB for the spill evaporation scenario in calm/inversion conditions is
inappropriate. If the authors truly believe that, after one hour, the evaporating
material occupies a volume that is 108 x 108 x 100 m (P. 19 of Appendix Vn-4),
which then begins to move downwind once the wind picks up, then a more
appropriate model would be an inversion-lid-limited Gaussian model with a volume
source. The SLAB model as used in the WTI risk assessment almost certainly
overestimates the airborne concentrations.
b) In all of the SLAB runs for this project, the input parameter TAV is set equal to
1,800 seconds = 30 minutes, which represents the exposure averaging time for which
the IDLH is defined. The input parameter TSD (the dispersion averaging time) is set
equal to the duration of release which, near the source, is also the duration of cloud
passage. When TSD is less than TAV, the authors reduce the calculated peak
centerline concentration by the ratio r = (TSD/TAV). Thus, if TSD is 600 seconds,
as it is" in the mitigated runs of SLAB such as Run No. 2, then r = 1/3. This
implicitly assumes that the IDLH is not a constant concentration, but rather a constant
dose, an expression of Haber's law. The authors should explicitly recognize that this
is a major assumption about the toxicity of the vapor and should review the evidence
that it is valid for formaldehyde, acetone, hydrogen chloride and phosgene. It is an
assumption which has a large effect on the predicted IDLH propagation distances for
the mitigated cases and for any other case where TSD is considerably less than TAV.
D-72
-------
Geoffrey D. Kaiser
I haven't had time to check the following out, but it seems to me that, when TSD
exceeds TAV, the authors do not use a similar averaging algorithm- that is, for
exposure times in excess of 30 minutes, toxic vapors are treated as if the IDLH is a
constant concentration and not a constant dose^ Is this correct? If true, this certainly
represents an inconsistent approach to the IDLH, which apparently obeys Haber's law
for exposure times < 30 minutes but not for exposure times > 30 minutes.
c) It is appropriate to apply SLAB to evaporating pools in a steady wind and to dense
vapor jets such as that which may evolve from inadvertent mixing scenarios.
d) The applications of ISC-COMPDEP seem to be appropriate.
5. Please comment on the assessment's conclusions on the severity of consequences
and the probability of occurrence. Has the report correctly categorized the
severity of the consequences of the different accident scenarios? Has the
assessment adequately justified the reported probability of occurrence of each of
the accident events?
a) Regarding frequencies of occurrence, there seem to be a number of inconsistencies
between various parts of the text. Some examples follow:
At the end of Chapter I (page 1-11) it is stated that the probability (it should be
frequency) of occurrence of an event having minor consequences would be classified
as reasonably likely —. However, in Table Vni-1 there is an event of minor
consequence with a likely frequency of occurrence. On Page VIII-1 it is stated that
events with minor consequences are classified as likely to occur. However, in Table
VIII-1, there are seven events with minor consequences, but only one of them is
likely, the rest being only reasonably likely or unlikely. On Tables VIII-2 and VIII-3,
there are several events with minor consequences, none of them with a frequency
exceeding reasonably likely.
D-73
-------
Geoffrey D. Kaiser
On pp VI-10,11, it is stated that spills of approximately 100 gallons are expected to
occur at least once every ten years and spills of 5,000 gallons are expected to occur ;
between once every ten years and once every hundred years. [The sentence in
question reads " — the available information summarized in Table 2 of Appendix VII-
1 would suggest that spills of approximately 100 gallons might be considered at most
reasonably likely (i.e. expected to occur at least once every ten years on average)
while spills of 5,000 gallons might be considered at most reasonably likely (i.e.
expected to occur between once every ten years and once every hundred years on
average)". This sentence needs some attention because it is internally inconsistent, see
in particular the use of "reasonably likely".] However, on page 1-11 the frequency of
occurrence of all accidental spills is estimated to be once in every 25-30 years, so it is
not clear where the once in ten years on pp VI-10,11 comes from. Starting with the
once in 25 -30 years, and using the estimate thait only 5% of spills are of 5,000 '
gallons or more leads to a frequency of once in 500 - 600 years for a 5,000 gallon
spill, much lower than in the sentence quoted above.
There is a clear need for the authors to review their written and tabular summaries of
the risk assessment results and to make sure that mere is consistency throughout
Volume VII and to make it easy for the reader to understand how estimates of
"likely", "reasonably likely" etc. were derived.
b) I tried to derive some of the results in table VI-10. e.g the predicted frequency of a
5,000 gallon spill of formaldehyde in typical weather conditions, which I took to be
the product of the following factors:
Frequency of occurrence of any spill: 1/25 = 0.04/yr
Probability that the spill is 5,000 gallons: 5% = 0.05
probability that the spill is formaldehyde: 0.5% = 0.005 (see p.VI-11)
Probability mat the weather is "typical": 0.57 (see p. V-2)
Product of all of the above: 5.7 x 10~6/yr or less than once every 100,000
D-74
-------
Geoffrey D. Kaiser
years, more than two orders of magnitude below the once in a thousand year
threshold for the unlikely category, yet in Table VI-10, this event is
categorized as unlikely. Even if formaldehyde is taken to be a part of as
many as 10% of all spills (I think this is what the authors actually mean, but it
is not clear on page VI-11; I have assumed 10% in .all the following
discussion) the above product is still about once in 10,000 years, well below
the "likely" threshold. Similarly, if we assume that 90% of all spills contain
acetone, die same reasoning as above would put the frequency of a large
acetone spill in the range of once in ten thousand to once in a thousand years,
i.e. barely approaching the lower threshold for the unlikely category, yet it is
assessed to be reasonably likely (more than once in a hundred years) in Table
VI-10.
-If we proceed to investigate calm/inversion conditions, which occur 1.7% of the time
per page VI-11 and replace 0.57 in the above product by 0.017, the frequency of
occurrence of a 5,000 gallon acetone spill is now 0.017/.57 = 1/33 of what it is in
typical weather conditions - that is, instead of somewhere in the range once in ten
thousand years to once in a thousand years it now becomes once in 330,000 years to
once in 33,000 years. However you look at it, this is very unlikely, whereas the
frequency given in Table VI-10 is unlikely, i.e. more than once hi a thousand years.
c) Turning to the estimate of the frequency of onsite fires, we have the following
calculation:
frequency of a large spill: 1/500/yr = 0.002/yr
probability that spill will ignite: 1/25 = 0.04 (see p. VI-17)
probability of typical weather: 0.57
product of the above: 4.56 xlO'5 = once hi 20,000 years, i.e. very
unlikely, yet in Table VI-10, the frequency of occurrence of a large
fire in typical weather conditions is characterized as unlikely, i.e.
D-75
-------
Geoffrey D. Kaiser
more than once in 1,000 years. Since the ranking of the on-site
mixing of incompatible wastes Has been assumed to be the same as
that for fires (see p. VI-12), the frequency of occurrence of these
mixing events also seems to be overestimated.
d) Turning to offsite events, the frequency of occurrence of a fire involving a 100 gallon
spill is [1/60 (chance of such a spill per year, p.. VI-12)] x [1/20 Oargest of ^
probabilities of ignition given on p. VI-14)] x [0.57 (probability of typical weather)]
= 4.75 x 10"Vyr = ~ once in 2,000 years, i.e. very unlikely. However, this event
is characterized as unlikely in Table VT-11, i.e. occurring more than once in a
thousand years.
e) The overall conclusion is that many of the frequencies in Tables VI-10 and VI-11 are
overestimated; in some cases by more than an order of magnitude. These tables need
to be carefully revisited.
f) Characterization of the severity of the consequences is discussed in Chapter VI of
Volume VII. At the bottom of p. VI-1 and the top of p. VI-2 we are told that
approximately 1,000 - 1,500 (isn't this figure more accurately known?) people live
within a rough semicircle .of radius 1,100 m. Assuming that a contaminant plume has
a radius of 5-10° enclosing an area with concentrations above the IDLH, then
between 1/36 and 1/18 of these people would be affected, or —27 — 83. Within
175 m of the site, there are 25 - 50 people (this is another figure that ought to be
better known), so that between 1 and 3 people might be affected. ;
On Page 1-8, a four tier system for classifying human accidents is presented,
attributed to FEMA. Minor consequences or those for which there is a low potential
for serious human injuries (among other criteria); moderate consequences are those
for which there are up to 100 potential injuries requiring medical treatment; major
consequences are those for which there up to several hundred injuries requiring
D-76
-------
Geoffrey D. Kaiser
medical treatment and observation and, finally, in a catastrophic accident there are
more than 300 human injuries requiring medical treatment. Assuming that the IDLH
indicates, the need for medical treatment, then, based on the first paragraph of f)
above, if the IDLH extends for 1,100 m from WTI, the consequences are moderate
(up to 100 injuries). If the IDLH extends up to 175 m, with the potential for only up
to about three injuries, then the severity would appear to be on the boundary between
moderate and minor.
In contrast, on Table VI-1, an accident with the IDLH extending out to 100 m is
defined as having moderate consequences and an accident with the IDLH extending
out to 1,000 m is defined as major. This ranking appears to overstate the actual
severity of the consequences, at least when compared with the FEMA rankings. The
authors should either return to the FEMA rankings or explain why additional
conservatisms appear to'have been added.
Similarly, consider Scenario 4B, the off-site spill of 100 gallons of formaldehyde
waste. The downwind distances to the IDLH are given in Table VI-8. In
conservative meteorology, the IDLH extends downwind to 630 m. According to our
discussion above, this ought to be characterized as moderate, yet in Table VIII-2 this
scenario is listed in the catastrophic column. [Even according to Table VI-1, it ought
to be no more than major]. Similarly, in calm/inversion conditions, this scenario.
propagates to 1,080 m according to Table VI-8. As we have seen, propagation to
1,000 m leads to less than 100 predicted injuries, so an extra 80 m is not going to
lead to enough extra casualties to warrant a catastrophic ranking (> 300 injuries). I
suspect that this scenario remains in the moderate category, certainly no more than
major.
One can make similar comments about other scenarios. For example, scenario 3A
does not cause the IDLH to propagate more than ~ 500 m downwind in any weather
condition (see Table VI-7), so that it should not have more than moderate
D-77
-------
Geoffrey D. Kaiser
consequences, yet on Tables VTII-2 and VTII-3 it has major consequences. I haven't
investigated other scenarios in detail, but it appe;ars that there is a tendency to
overstate the consequences and each scenario should be reviewed to make sure that its ;
consequences have been correctly characterized.
g) In summary, items a)- f) above seem to indicate that there has been a considerably
tendency to overstate both frequency of occurrence and magnitude of the
consequences in the matrices in Tables VIII-1,2,3. Overall, Figures VHI-1,2,3
overstate the risks that have been predicted by the accident analysis. My own
conclusions (for which I am prepared to provide details - see the attached handwritten
worksheets) are that (1) there are no scenarios for which comprehensive planning and
preparedness are essentially mandatory: (2) almost all of the scenarios lie in the region
where comprehensive planning may be unwarranted and unnecessary and (3) the only
scenarios that lie in the region where comprehensive planning is optional are ID, 3D
and 4D (all large formaldehyde spills) in conservative or calm/inversion conditions/ I
estimate that these all have frequencies of less than lO^/yr - i.e. more than an order
of magnitude less than the frequency threshold that divides unlikely from very
unlikely. This additional order of magnitude should influence the decision as to
whether planning is undertaken.
6. Key assumptions were made in the identification of accident scenarios and the :
description of the conservative and typical events. Included were a description of
the magnitude of the effect of the assumptions and direction of the effect. Please
comment on the assumptions. Are they justified? Are the descriptions of the
magnitude and the directions of the effects correct? Has the accident.assessment
adequately confronted the uncertainties involved in doing this kind of analysis?
If not, what else should be done?
D-78
-------
Geoffrey D. Kaiser
I am not sure what part of Volume VII is being referred to here, at least as far as the
"magnitude of the effect of the assumptions and the direction of the effect" is concerned.
The choice of accident scenarios and the description of conservative events seems reasonable.
7. Comment on the appropriateness of using IDLH values for characterizing the
severity of the consequences in the accident analysis. Comment on the
appropriateness of using 10 X LOC for chemicals for which IDLH values have
not been established.
a) Although there are known problems with the consistency of the definition and the peer
review of IDLH's, I do not know of any source of more consistent data. It is
appropriate to use an endpoint that is meant to be a threshold for injury or
incapacitation that would prevent the individual from taking countermeasures (i.e.
taking significantly lower endpoints would likely overestimate the risk). For a
qualitative/semi-quantitative risk assessment of the type reported in Chapter VII, I do
not think it is possible to find a toxic endpoint that is obviously better. The ERPG
would in principle be better, but it has not yet been derived for acetone or
formaldehyde.
b) IDLH values exist for all of the four chemicals used in the risk assessment -
formaldehyde, acetone, HC1 and phosgene. Therefore, the question- about 10 x LOC
is moot.
8. In the accident analysis, IDLH (or 10 X LOC) values were used to determine the
downwind distances over which adverse health effects might occur. To evaluate
the uncertainty introduced by using the IDLH, a sensitivity analysis was
conducted where these distances were recalculated using the LOC (a more
stringent health criterion). Other sources of uncertainty that are identified in the
accident analysis include concentration averaging times, chemical concentrations,
emission rates and meteorological conditions. For most of these parameters, it is
D-79
-------
Geoffrey D. Kaiser
stated that conservative assumptions were used to avoid underestimating risks.
Have the uncertainties inherent in the accident analysis been adequately
characterized? For those parameters where sensitivity analyses were not
conducted, is the conclusion that conservative assumptions have avoided
underestimation valid?
As can be seen from my answer to Question 5 above, I think that, by the time the authors
come to the final presentation of the results in Tables VIII-1,2 and-3, they have unnecessarily
overstated conservatisms. Therefore, conservatisms have indeed avoided underestimation, but'
by too much! The authors are aware of and discuss the major areas of uncertainty.
C. ADDITIONAL COMMENTS ON VOLUME Vn
1. Effects of Mitigation (Chapter VH.C.l)
The discussion of .sensitivity studies in Chapter VIIC (pp VII-5 - VII-7) is not helpful because
it does hot explain, why the results turn out to be the way they are. For example, on p. VII-
8, it is'stated that "for the mixing of incompatible waste scenario involving 200 gallons of
waste, active mitigation within 10 minutes does not significantly reduce the maximum distance
to the EDLH." The reader has to correlate this observation with the SLAB inputs in Table 4
on p. 32 of Appendix Vn-4, runs no. 55 - 60. Only then does it become clear that the
mitigated and unmitigated cases have the same duration of release, namely 10 minutes, so that
mitigation in 10 minutes is not a meaningful action and of course there is no reduction in the
maximum distance to the IDJLH. Similarly, mitigation in one hour has no meaning for a
scenario that is over hi 10 minutes.
Looking at Attachment 3 of Appendix VII-4 (actual data from SLAB output files), the
distance to the IDLH in typical weather conditions for a 5,000 gallon spill of formaldehyde is :
640 m (unmitigated) and 730 m (mitigated) (RUNG 1.OUT and RUN02.OUT respectively).
However, on Table VII-1 on P. VII-12, both of these numbers are presented as 640 m. No
D-80
-------
Geoffrey D. Kaiser
explanation is given. In fact, this result should set the warning bells ringing because it
doesn't make any physical sense that mitigation could increase the distances within which
people might be injured. Several mathematical factors combine to give this result: a) the
average release rate over 10 minutes (1.99 kg/s) is greater than that averaged over 3120
seconds (1.04 kg/s), see Table 2 on p. 30 of Appendix VII-4: b) from the information on
the RUN1.OUT and the RUN2.OUT printouts in Attachment 3 of Appendix VII-4, the
average vapor phase mole fraction of formaldehyde is 0.742 for the first ten minutes of
evaporation and 0.478 for an evaporation time of 3120 seconds; c) the peak centerline
concentration at 10 minutes is (3120/600)0-2 - 1.39 times higher than that at 3120 seconds.
All these factors together make the 10 minute peak centerline concentration effectively 1.39 x
(1.99/1.04) x (.742/.47S) - 4.13 times higher than the 3120 second peak centerline
concentration. However, SLAB averages the 10 minute release over 30 minutes for
comparison with the IDLH, so introducing a further factor of (1/3), which reduces 4.13 to
1.38, still greater than unity, which is mathematically why the 10 minute mitigated case
propagates further than the 3120 second unmitigated case. The authors need to devote more
thought to the further explanation of this effect.
2. Annual Average Off-Site Air Concentrations (Appendix VII-4 p. 25)
There is a paragraph about annual average off-site air concentrations in the middle of page
25. It is not clear why it has been included. It has no relevance to a risk assessment of
accidental releases.
3. Appendix VII-4, p.27
The paragraph in the middle of the page that begins "If the release is instantaneous or very
short —" is not clearly written. It is not clear to what scenario it applies. It needs to be
rewritten in order to make its point clear (that, because a puff elongates along the wind as it
travels, exposure times may in fact equal or exceed thirty minutes when the cloud has
travelled (say) 10 km and the predicted distance to the LOG may not exceed that predicted
D-81
-------
Geoffrey D. Kaiser
using the continuous, finite duration release model?).
4. Page n-15
The second of the three bullets at the bottom of this page is not always correct - "chemical
mixing with the water, or sinking below the surface of the water, would reduce emission rates
relative to a spill onto the road.". If the chemical reacts with water, evaporation rates driven
by the heat of reaction can be very high.
D-82
-------
Table 1
Ikevked Tables VHI~i»Vnt2 and VHI-3
" - ' " ..-••••••- ^ j.* Frequency/Severity Matrix % - ',, -,"'-, ^
• Frequency of
Occurrence (f)
Common: f ^ 1/yr
Likely: l>f;>0.1/yr
Reasonably likely
0.1>f>0.01/yr
Unlikely
0.01>f;>0.001/yr
Very Unlikely (1)
0.001 >f^ lO^/yr
Very Unlikely (2)
10'4>fk lO'Vyr
Very Unlikely (3)
f.< 10'Vyr
Severity of Consequence
Minor
-
-
(1A),(1C),(4A)
[4A], [4C]
[1A]
(3A),(2A),
[1C]
{1A> {4A} <£
(2B),[2A]
[5A],{1C}
[2B]
Moderate
-
-
-
(IB)
[4C], [5A^
(5B), {4C}
S • ^^
^s.
-{3A}
Catastrophic
-
-
-
-
F4R1
-1HE>J
[ID], [4D]
^{4B}
[3B],{1D}
{3B},{4D}
Key: ( ) Typical weather conditions, Table VIII-1
[ ] Conservative conditions, Table VIII-2
{} Calm/inversion conditions, Table VIII-3
<— Represents change in estimated magnitude of consequences as described in the
accompanying text. Other scenario severities might need to be changed. All of
values of f in the above table have been re-evaluated.
©• Consequence severity could be in either category
D-83
-------
Assumptions in Table 1
Frequency of on-site spill - 0.04/yr
Probability that spill is 100 gallons - 0.35
Probability that spill is 5,000 gallons - 0.05
Probability that spill is acetone - 0.9
Probability that spill is formaldehyde - 0.1
Probability that weather conditions are typical - 0.57
Probability that weather conditions are conservative - 0.1
Probability that weather conditions are calm/inversion - 0.017
Probability of ignition given a spill - 0.04
Frequency of small off-site spill-0.017/yr
Frequency of large off-site spill - 0.0034/yr
Probability of ignition given a spill - 0.05
typical
f, on-site spill small spill / conservative
Examples Si
-------
Robert N. Meroney
PREMEETING COMMENTS ON WTI INCINERATOR RISK ISSUES
General Comments:
1. The organization of the risk assessment document is logical, the text is clear,
concise and easy to follow.
2. The executive summary reflects the data and methodologies used and the
conclusions derived.
3. The 1993 peer review workshop recommendations (as listed on pages IV-2 and IV-
3) related to Ch IV: Atmospheric Dispersion and Deposition Modeling were
addressed.
4. Uncertainty Issues:
• Quantifying Uncertainties: There is far more to uncertainty analysis than just
listing assumptions and admitting models are approximate. Sensitivity tests
are important, but they also do not guarantee consideration of the
cumulative effects of various uncertainties. The use of "conservative"
assumptions throughout would initially appear to accomplish the goal of
protecting public health, but it may also encourage undue expense to protect
the public from non-existent hazard levels! Ranges of confidence and
quantitative statements of error about "realistic" calculations of exposure
would appear to be a more appropriate way to protect health and
environment economically. At least then one knows quantitatively what
factor of safety is being imposed. Finally, the final risks involved should be
phrased in terms of common risks that the community already endures or
accepts, so that the lay person can also appreciate the situation.
• Spurious Correlations: In many cases correlations are used to consolidate
data or identify similarities which can artificially reduce variance through
"virtual" or "spurious" correlations. It is not uncommon for such spurious
correlation to be as high as 50 to 80%! The proportion of reduced variance
D-85
-------
Robert N. Meroney
due to virtual correlation can be estimated for any parameter scheme, and
some indication of the fraction of virtual correlation present is a valuable
measure of the value of any statement of correlation.
5. Long-term research which would improve risk assessments of this type include:
• Statistical analysis and evaluation-of common correlations to evaluate level
of spurious correlation present.
• Wind-tunnel evaluation of fluctuating and instantaneous peak and root-
mean-square (rms) concentrations present during building and terrain
induced fumigation. Instantaneous concentrations will govern probabilities
for flammability and the level of odors. RMS levels imply mixing potentials.
• Examination of the influence of terrain irregularities and slope on the
transport and dispersion of dense-gas emissions using fluid modeling.
These emissions will tend to drain and channel, yet no existing numerical
models properly account for these effects.
• Evaluate the proportion of unsteady gas clouds which are likely to infiltrate
into buildings during an emergency event. Currently most models presume
concentration levels which exist outside buildings will be the same in the
occupied spaces within the building.
• The Huber-Snyder and Schulman-Scire downwash models seem to be
limited to distances downwind exceeding x/Hb > 3, yet many situations exist
where short stacks, fume hoods and flush vents emit gases which re-
impinge or re-entrain on to the original building or the near cavity wake
region. A model or protocol should be developed to deal with these
Situations including average and instantaneous concentration levels.
• The SLAB model used in the WTI calculations does not include the influence
of sloping terrain,- gullies, or gorges on the transport and dispersion of heavy
gas plumes. Lee and Meroney (1988) have demonstrated that depth-
integrated models can predict dispersion in such situations. A series of fluid
model experiments complimented by further depth-integrated calcualtions
should be carried out to determine the worst-case situations for heavy plume
dispersion over irregular terrain.
D-86
-------
Robert N. Meroney
Workshop Specific issues:
ATMOSPHERIC DISPERSION AND DEPOSITION MODELING: CHAPTER IV
1. Section II, C, 8, pp. H-8 to 11-10 and D, 11-10 to 11-13. Downwash Effects:
• The Huber-Snyder and Schulman-Scire models are limited to x/Hb > 3.
What are the possibilities of plumes producing maximum concentrations in
the near cavity region. (If not for the WTI case, then in other
circumstances.)
2. Section IV, B, 6, pp. 1V-16 to IV-18. Sensitivity tests: Terrain Downwash Simulations
• The report of the wind-tunnel tests by the U.S. EPA Fluid Modeling Facility
(Appendix IV-6) together with the associated Peer Review statements
(Appendix IV-7) provide convincing documentation that the effects of
building and plume downwash have been properly considered during the
WTI analysis.
• A major reason for the wind-tunnel study was to document the presence of
terrain fumigation, flow separation, reattachment and associated
recirculation regions. As noted in Appendix IV-6, page 4, a terraced model
was constructed from Vz inch plywood. Normally such steps are considered
appropriate since they help simulate the local roughness effects.
Unfortunately, the use of a terraced model may incorrectly predict the
location and strength of recirculation regions. Meroney (1978,1979) found
that when correct simulation of near surface winds and associated
recirculation regions are required it is appropriate to simulate the terrain with
a smoothly contoured model with simulated vegetation added.
Meroney, R. N., Bowen, A. J., Lindley, D., and Pearse, J., WIND
CHARACTERISTICS OVER COMPLEX TERRAIN: LABORATORY
SIMULATION AND FIELD MEASUREMENT AT RAKAIA GORGE, NEW
ZEALAND, FINAL REPORT, PART II, Department of Energy Contract No.
EY-76-S-06-2438, A001, 219 pp., Report RLO/2438-77/2, 219 pp., May
1978
D-87
-------
Robert N. Meroney
Meroney, R. N..-FIELD VERIFICATION AND LABORATORY SIMULATION
OF AIRFLOW PATTERNS UN COMPLEX TERRAIN, Proceedings of Fourth
Symposium on Turbulence, Diffusion, and Air Pollution, January 15-18,
1979, Reno, Nevada, pp. 592-595
• Only neutral stratification situations were considered during the wind-tunnel model
program under the argument that only "moderate and high-wind conditions" need be
simulated to evaluate terrain effects. It is true, however, that stratification effects can
be .present under moderate and high-wind conditions. In particular if an elevated
inversion exists, then the local terrain may cause strong down-slope winds which can
enhance or degrade recirculation situations, indeed in a valley situation at night local
radiation inversions may further enhance stagnation and decouple the valley flows
from the stronger winds above (see page IV-21, Section D.b, line 7). Meroney and
Grainger(1992, 1993) discuss the influence of stratification on winds over
depressions.
Meroney, R.N., Grainger, Clyde," DISPERSION IN AN OPEN-CUT COAL
MINE IN STABLY STRATIFIED FLOW., Jml. of Boundary Layer
Meteorology, Vol. 63,1993, pp. 117-140.
Meroney, R.N., Grainger, Clyde, NIGHTTIME FLOW AND DISPERSION
OVER LARGE BASINS OR MINING PITS. Symposium on Measurement
and Modeling of Environmental Flows. 1992 ASME Winter Annual Meeting,
Anaheim, CA, November 8-13,1992. FED-Vol. 143/HTD-Vol. 232, pp. 209-
215,
3. Charge Question 1: The risk assessment document does summarize the
additional tasks taken to address the 1993 peer review concerns. The links
between the data collection, the air dispersion models, and the risk assessment
seem adequate.
4. Charge Question 2: The considerations of the effects produced by terrain
downwash have been adequately addressed through the use of fluid modeling and
the subsequent comparison of the numerical model and wind-tunnel results. The
comparisons suggest the ISC-COMPDEP model is conservative. This would
appear to be the case for neutrally stratified conditions; however, the possibility that
stratification or raised inversions might enhance downwash without associated
increases in terrain induced turbulence has not been considered.
D-S
-------
Robert N. Meroney
5. Charge Question 3: Comparisons of the results from the steady state ISC-
COMPDEP model and the non-steady CALPUFF model do suggest that receptor
concentrations will not be significantly increased due to nonstationarity.
6. Charge Question 4: Given that many hazardous chemicals likely to be released
in a spill or accident have either heavy molecular weight or may produce a cold gas
cloud as a result of decompression or evaporation it is commendable that the SLAB
model was used to evaluate hazards. Unfortunately, the version of the depth-
integrated model used presumes absolutely flat terrain with no slope or channeling
due to terrain irregularities. The presence of such terrain perturbations changes the
consequences of such releases significantly; hence, high concentrations may be
carried to significantly greater distances downwind in such situations. A complete
hazard analysis for on- or off-site releases should examine worst case scenarios of
terrain induced transport. Terrain effects have been predicted by depth-integrated
models, but no systematic evaluation of the enhancement of transport has been
completed (Lee, 1988; Lee and Meroney, 1988).
Lee, J.T. and Meroney, R.N., NUMERICAL MODELING OF DENSE GAS CLOUD
DISPERSION OVER IRREGULAR TERRAIN, Proceedings of Eighth Symposium
on Turbulence and Diffusion, American Meteorological Society, San Diego,
California, 25-29 April, 1988, Paper 4.9, pp. 392-395
Lee, J.T. (R. N. Meroney, advisor), A NUMERICAL INVESTIGATION ON DENSE
PLUME DISPERSION OVER COMPLEX TERRAIN, M.Sc. Thesis, August 1988,116
pp.
7. Charge Question 5: The sensitivity tests completed appear to be appropriate and
inclusive. The tests respond to the 1993 Peer Review panel concerns, and the
conclusions drawn from the tests are logical and appropriate.
D-89
-------
-------
Exposure Assessment
D-91
-------
-------
James P. Butler
Comments on the Risk Assessment for the Waste Technologies Industries (WTI)
Hazardous Waste Incinerator Facility (East Liverpool, Ohio)
Based on my review of Volume V: Human Health Risk Assessment: Evaluation of
Potential Risks from Multipathway Exposure to Emissions and Volume VII: Accident
Analysis: Selection and Assessment of Potential Release Scenarios, I can offer the
following comments on technical aspects of the WTI Risk Assessment. My comments
are organized into two sections. The first part addresses issues raised in the Charge to
Reviewers, and the second part contains specific comments and recommendations that
were not considered in the charge.
PARTI
General Issues
1. This is one of the most comprehensive risk assessments that has been conducted for'
a waste disposal facility. Considering the amount of data and complexity of the analysis,
the risk assessment is presented in a clear and logical manner. In particular, there is a
reasonable balance between the extensive exposure-related data in the appendices and
summary tables in the texts (with cross-referencing). It is also helpful having the key
assumptions clearly identified and tabulated in the discussion of uncertainties.
2. The executive summary accurately conveys the general approach, results, and
conclusions of the risk assessment.
3. Most of the major recommendations of the 1993 peer review workshop pertaining to
exposure assessment have been addressed to some extent in the document. However, the
issue of background exposures to mercury has not been evaluated in the human health risk
assessment (discussed below in Part II).
4. The human health risk assessment is generally concerned with the potential for
D-93
-------
James P. Butler
incremental risks due to emissions from the WTI facility. While this approach is
consistent with EPA risk assessment guidance, it does not account for cumulative or total
exposures to a given contaminant.- Because East Liverpool, Ohio has been characterized
as a depressed, industrial city, its low-income population may already be exposed to a
variety of environmental hazards. While the methodology for evaluating multiple and
cumulative exposures and risks has not been fully developed, the document should at least
qualitatively address this issue for contaminants commonly found in industrial areas. To
use the results of this analysis for risk-based decision making, it will be necessary to
consider environmental equity issues related to health effects.
5. One of the most significant long-term research efforts would be to continue, and
expand, the ongoing baseline biomonitoring program.that is being conducted in the
vicinity of the WTI incinerator. There is a need to collect additional environmental
monitoring data for key input parameters in order to validate the exposure modeling
approach. It would also be valuable to conduct follow-up surveys to confirm earlier
findings related to key exposure parameters (garden usage, fish consumption, etc.).
Another useful study would be to collect tap water samples to confirm the assumption that
the water treatment plant is removing any contamination resulting from the incinerator's
emissions. One of the real strengths of this risk assessment is the inclusion of more site-
specific data than is typically available. Building on that database would be beneficial
for future risk assessments of similar facilities,,
Human Health Risks - Exposure
1. Average and maximum environmental concentrations were modeled for each medium
of concern. Similarly, typical and 90th percentile values were obtained for most of the
exposure factors. If the central tendency exposure estimates were calculated using
average values for both media concentrations and exposure factors, and high-end exposure
estimates were calculated using at least 90th percentile values for both media
concentrations and exposure factors, then the exposure descriptors were properly used to
characterize exposures. However, it is not clear in the document which set of values
D-94
-------
James P. Butler
were used as inputs for the exposure dose equations.
2. The general approach to developing estimates of exposure is adequately explained in
the exposure assessment. However, the specific procedure used to develop central
tendency, high end, and/or bounding estimates is not clear (as mentioned above).
Perhaps it would be helpful to have a summary table that identifies the combinations of
factors that make up each exposure descriptor.
3. The most important exposure pathways have been identified and evaluated in the risk
assessment. As recommended in the 1993 peer review workshop, there now is adequate
justification for omitting the groundwater and surface water pathways.
4. The key assumptions needed for estimating environmental concentrations and exposures
have been identified. The magnitude and direction of effect are correct for the
assumptions, with one exception. The assumption that fate and transport modeling
accurately reflects reality is highly uncertain and does not necessarily result in a "likely
overestimate." The direction of effect is really unknown until these models are better
validated with monitoring data collected for that purpose.
5. The uncertainty analysis for two representative compounds was probably the most
useful way to confront the overall uncertainties and to identify input parameters that have
the greatest effect on the final risk estimate.
Accident Analysis
1. Two other scenarios should also be considered. Because there is a history of on-site
worker injuries and fatalities at other commercial hazardous waste incinerators, a
plausible accident scenario involving workers at the WTI facility should be included.
Given the incinerator's location in a floodplain, it would also make sense to evaluate the
impacts of a flooding accident.
D-95
-------
James P. Butler
6. The key assumptions made in the identification of accident scenarios and ranking of
accident events appear to be reasonable with respect to magnitude and direction of effects.
In general, conservative assumptions were made to compensate for the uncertainties
resulting from a lack of information needed to fully characterize the exposure and dose.
7. The American Industrial Hygiene Association's Emergency Response Planning
Guideline (ERPG) levels would probably have been more appropriate than IDLH values
for characterizing the severity of accident consequences. At a minimum, for compounds
with both values, ERPG levels should be compared to IDLH levels to make it clear which
are the more stringent criteria for assessing the acute effects of short-term exposures.
PART II
Background Exposures
The general approach to estimating exposures and doses involves assessing incremental
intakes of chemicals emitted from the facility. For some contaminants, it is essential to
factor in estimates of existing body burdens and intakes from other sources. This has
been done in evaluating potential health effects of lead using the IEUBK model.
However, background intake of methylmercury through the consumption of non-local fish
and seafood should also be evaluated in the exposure modeling. While not the case for
subsistence fishermen, consumption of commercial seafood and fish is the primary source
of methylmercury intake for much of the U.S. population. To decide whether or not an
incremental exposure is acceptable, one needs to know the current intake levels and
existing body burden of mercury for the fish-eating population. It has been calculated
that a significant fraction of women of childbearing age already have an unacceptably
high level of methylmercury in their diets based on estimates of seafood consumption and
Hg levels in the U.S. catch [Stern, A. H., 1994, "Re-evaluation of the Reference Dose
for Methylmercury and Assessment of Current Exposure Levels," Risk Analysis 13, 355-
364]. If this is true, any increase in incremental exposure to methylmercury could
present a health risk, at least for specific sensitive populations.
D-96
-------
James P. Butler
Biomonitoring Results
It was mentioned in the document that site-specific monitoring data indicates that lead and
mercury levels in local vegetable gardens has not been increasing, but are present at
higher levels than observed in other Ohio communities. These results should be included
and discussed in the risk assessment.
Breast Milk Concentrations
Although consumption of mother's milk was found to a significant pathway of exposure
to organics for breast-feeding infants, inorganic compounds were omitted from this
pathway analysis. The point that metals are not expected to accumulate in breast milk,
however, should not be used to rule out mercury which can be converted to the lipophilic
organic form, methylmercury.
Contingency Plans
The accident analysis section of the risk assessment should include a description of
measures taken to prevent on-site accidents, and a summary of emergency response plans
that are in place in the event an accident does occur.
D-97
-------
-------
George F. Fries
General
1. The document is organized in a logical format. I did have a little difficulty
originally in tracking the emission and deposition rates in Section IV with the
exposure analysis in Section V. More cross-references between the sections
would have been helpful for those who wish track residue concentrations from
emissions to humans but these cross-references are not absolutely necessary.
2. The Executive Summary is a good reflection of the full document. There are no
discrepancies between the conclusions in the summary and in the main
document, and I did not find significant omissions. The Executive Summary
would read more smoothly if Section A (Overview) and Section B (Introduction
and Overview of Results) were rewritten to highlight the conclusions. Much of
the material in the Introduction subsections could be omitted because it appears
too detailed and its presence detracts from the readability of these overviews.
This material is covered later in the Volume I.
3. The recommendations of the exposure subgroup were addressed in this
assessment. Volumes VI and VII addressing ecological risks and accidents
were clearly developed in response to the 1 993 recommendations. Concerns
raised by the other subgroups appear to have been addressed, but I am not
familiar enough with the topics to determine of the responses are adequate.
4. This risk assessment is the most comprehensive site-specific assessment that
has come to my attention. I have not identified significant gaps in
methodology or data in the exposure area, where I have the greatest experience
and background. The gaps that may exist are small and the deficiencies that
may exist will be covered be the general conservatism of the assessment. This
assessment is adequate for making decisions.
D-99
-------
George F. Fries ;
5. I have no specific suggestions for long term research. Within the exposure
area, almost all of the parameters can be refined and improved. For example,
the beef bioconcentration factors are based on one animal, and the air-to-plant
factors are based on only a few plant species. Physical parameters, such as
log Kow, also show inconsistencies that reflect difference among laboratories in
techniques. Many models depend on these parameters and more reliable values
would be useful. A recent trend has been to use Monte Carlo analyses to
obtain probability distributions of the potential exposure. Better estimates of
both the means and the variation of the more sensitive parameters would be
useful in carrying out these analyses in future exposure assessments.
Emissions Characterization
No Comments
Dispersion and Disposition Modeling
No Comments
Human Health Risks
Exposure
1. The high-end exposure appears to be estimated only for the subsistence farmer
group in the El area. This group and area were identified as those most likely
to have the highest exposure. The mean exposures of all other groups and
[
areas were lower. It is likely that the relative variations of other groups will be
similar. Thus, the characterization for all groups can be inferred even if the
data is not provided. One might question if a normal distribution is the most
appropriate assumption. Often, residue concentrations in a large sample of
environmentally exposed individuals will follow a log normal distribution.
D-100
-------
George F. Fries
2. The assessment does a good job of describing the central tendencies for the
various groups and the logical basis for combining the factors. The end
exposure has on been estimated for one group, which has the highest predicted
exposure. This, failure has no serious consequences for the reasons noted
above, The exposure assessment follows standard procedures in combining the
environmental media concentrations, intake rates, and duration and frequency
of exposures. Many of these procedures follow models that have been
accepted as policy and the estimates follow from emission rates that follow
from stack tests of this incinerator. A sensitivity analysis is appropriate for
developing the bounding estimates. Overall the approach is appropriate given
to current state of the art.
3. The WTI assessment has examined a fairly wide range of potential exposure
scenarios. The selection of scenarios, and the conclusions .concerning the
importance of the various routes is consistent with the current state of the
knowledge. The only significant future refinement would possibly involve a re-
evaluation of trie relative importance of the various animal food chain
pathways. I cannot now envision any important routes that were not
considered, nor do I feel that the nonfood routes will assume greater
importance.
4. The key assumptions for estimation of chemical concentrations and exposures
have been identified. Generally, the magnitude and direction of effects appear
correct for the assumptions. There are some areas of uncertainty with regard
to the concentration along various points of the chain from emissions to
exposure for specific chemicals such as dioxins, but he appropriate parameters
have been used based on current knowledge. Often these parameters depend
on a small number of observations of only a few animals or plant species that
can lead to uncertainties.
D-101
-------
George F. Fries
5. I agree that conservative assumptions have been applied in this assessment.
The accumulative effect of these conservative assumptions is a very
conservative assessment. This high degree of conservatism has not been
reflected in the risk characterization section of the exposure document. I would
not, however, consider this a serious problem because to the extent that this
document is used for policy decisions, the conservatism of the document will
be health protective.
Hazard Identification/Dose Response and Risk Characterization
I will not comment on this section other than to note the some of the specific
questions to overlap with questions in the Exposure Subsection. My responses
in that subsection may be referred to.
Screening Ecological Risk Assessment
I have no particular comments except to note that in my comments for the
previous workshop, I suggested that it difficult to visualize a scenario in which
the ecological concerns would outweigh the human health. My quick review of
the SERA indicates that the conclusions are consistent with my suggestion.
Accident Analysis
I am not particularly familiar with the subject of accident analysis. My limited
comments reflect intuition rather than specific knowledge.
1. The five scenarios appear to be logical. The classes are broad enough so that
almost every imaginal accident is represented.
2. The specific chemicals selected are appropriate given the waste that the facility
is permitted or likely to incinerate.
D-102
-------
George F. Fries
3. No Comment.
4. No comment, except to note that these are accepted methods to deal with
chemical release and dispersion.
5. Probably logical. The conclusions appear to be based on established
methodology.
6. The key assumptions are reasonably stated.
7. IDLH values are a logical way to deal with one-time accidental exposures. It
appears to be a standard procedure that LOCs are about one tenth the IDLH,
and therefore, ten times LOG would be appropriate when there is no established
IDLH value.
8. I agree that generally conservative assumptions were made.
D-103
-------
-------
Comments on the WTI Draft Final Risk Assessment
for the Technical Workshop on WTI Incinerator Risk Issues
Thomas E. McKone
Exposure Assessment Workgroup
GENERAL ISSUES
This is a large and comprehensive document. The EPA staff and their consultants have
clearly expended an enormous amount of f:ffort in assembling large amounts of data and
constructing models, running simulations, evaluating the quality of the data and the
reliability and uncertainty in the estimates of exposure and risk. However, in reading
the summary for this document, one gets a sense that this effort was expended mainly to
N-
comply with regulatory requirements and guidelines. Even though such compliance is
necessary, if this is the only motivation for the time and energy invested in this
document then it will be hard to argue that the document provides much in the way of
public service. I believe that one important goal of the risk assessment is to address the
impact of. a process or facility on public health and to address the concerns of the
affected community. In reading the risk assessment, it becomes clear to me that there is
sufficient information provided to inform any intelligent debate on community health
issues. However, much of the information relating to these issues is buried in text,
equations, tables and appendices. Thus, it would be useful to include in the executive
summary and in the introduction to the document a summary table of the local
community concerns and a brief description of how these concerns are addressed in the
risk assessment. Failure to express the community concerns in a risk assessment leaves
one with the impression that the concerns of the regulators are all that matter.
The risk assessment is still overly focused on the conservative estimate (with respect to
uncertainty) of exposure and risk at the middle range and high-end of the heterogeneity
scale. This approach still fails to give some sense of the likely or plausible range of
outcomes. For example, when you role a six-side die the expected outcome is 3.5, but
the outcome realization is anywhere from 1 to 6. This issue is particularly relevant in the
section on accidents.
It appears to me that the risk assessment fails to merge the stack and fugitive emissions
and accident exposures into a single measure for expected harm within the affected
community.
D-105
-------
T.E. McKone
In my review of the risk assessment, it appears to me that much of the public health risk
(in terms of likelihood of harm) is associated with accidents and not with routine stack
and fugitive emissions. This does not come across in the executive summary and the
introduction.
Throughout the document the justification for the models and data used are EPA
guidance documents. In all cases where I was able to check the citation, I found that the
model was indeed consistent with the guidance in the EPA documents. This type of i
cross-referencing is a useful approach for consistency and quality control. However, it is
not a substitute for verification and validation. In the longer term there is a need to
apply a criteria of validation against data. Since many of the cited documents are not
final or still under external peer-review .or under EPA-SAB review, citation of these
documents should not be interpreted as consistent with reliability or veracity. Defining
reliability and accuracy will require matching the results against measured data. !
Organization of the Risk Assessment Document
If this document is to continue to inform the debate and planning processes within the
affected community there needs to be some stated plan within the risk assessment
document for how the conditions assumed for the risk assessment will be monitored arid
audited. There should be some discussion of continuing efforts to assess the stack and
fugitive emissions source terms and conditions that give rise to accidents. There should
be guidelines for how the exposure media concentrations and estimated doses in the
high-end groups could be monitored to verify that the predictions of the risk assessment
are within the estimated confidence bounds.
There has been a very visible and consistent effort to include an uncertainty assessment
in ever aspect of the report. Nevertheless, the uncertainty analysis still often seems to ;be
an add-in put at the end of each section and at the end of each chapter instead of being
a more integral part of the assessment. One rather simple change that could improve on
this problem is to move the chapter oh uncertainty; analysis in the Health Risk
Assessment, Chapter DC in Volume V, to be presented before the chapter on risk
characterization. In this way the material on risk characterization could more easily
address the results of the uncertainty analysis.
D-106
-------
T.E. McKone
Adequacy of the Executive Summary
The Executive Summary should include some consideration of the combined effect of
both routine (stack and fugitive) emissions and the releases associated with accidents on
public health. It would be useful to summarize the likely risk of cancer associated with
both stack and fugitive emissions and explicitly define the relative contributions to risk.
It would also be useful to compare the probability of health detriment from stack and
fugitive emissions to those from accidents.
The Executive Summary makes note that the 1983 National Research Council (NRC)
report on risk assessment served as a guide for the framework of the WTI risk
assessment. There are two other NRC reports that provide guidance for the exposure
assessment component of the risk assessment and should be noted here:
National Research Council 1991, Frontiers in Assessing Human Exposure to
Environmental'Toxicants, (National Academy Press, Washington, D.C.)
National Research Council 1991, Human Exposure Assessment for Airborne Pollutants:
Advances and Opportunities, (National Academy Press, Washington, D.C.)
These reports should also be cited on page II-2 of Volume V.
Outside of Volume VH, I could not find a definition of the acronym IDLH used on page
1-11 of Volume 1.
Are Major Recommendations of the 1993 Peer-Review Workshop Addressed?
In order to address this question, I went back to our 1993 recommendations and
compiled them into tables. I focus here only on issues that were raised by the Exposure
Workgroup in 1993. With regard to exposure assessment, the 1993 Peer-Review Group
developed recommendations on the WTI Risk Assessment Plan and divided these
recommendations into two categories—priority issues and issues of lesser significance
that needed to be fixed or addressed as part of the EPA research plan. The priority
issues are listed in Table 1. In the column to the right I have noted to what extent I
believe these issues are addressed in the actual risk assessment.
Table 2, which is listed in the section below that provides my comments on the health
risk assessment, covers our comments that dealt with issues of lesser significance that
we felt should be addressed or fixed. Since most of these are related to the overall
health risk assessment. They are included in that section.
D-107
-------
I.E. iMcKone
Table 1 Priority Recommendations
Recommendation from 1993 Workshop
Existing and future exposures to other sources in the area are not
addressed. This issue becomes relevant if there are future efforts to monitor
exposure in this region. The risk assessment can riot be used to estimate
total exposure for this population but only the exposures attributable to
WTI. Also, this plan does not address the regional impacts, that is beyond
50 miles, of this facility in combination with other combustion sources.
Uncertainties need to be confronted quantitatively. The treatment of
uncertainty and the issues of uncertainty and variability should be
integrated as much as possible into all aspects of the report and not just
included as an addendum to the section on risk characterization.
Use of site specific data is to be commended and encouraged.
Compounds selected as surrogates for the risk analysis on the basis of
quantity, toxitity, and KOW should not necessarily be used to carry out
validation studies unless it can be demonstrated that these are indeed also
persistent compounds — as is implied by the exclusion of a persistence
factor from the selection criteria
Evaluating exposures to short-term releases — accidents, fires, equipment
malfunctions, fugitive emissions is necessary.
Where possible, information on feed stocks and ash residuals should be
used together with a mass balance to verify emissions estimations
Was it
addressed?
Not very
well.
Mostly
Yes
Not very
well
Yes
Mostly
The atmospheric dispersion modeling section reflects an exemplary effort to address the
concerns raised by the peer review panel.
Major Data and Methodological Gaps
As far as I can determine the document still provides very little information on existing
and future exposures to incinerator-type contaminants from other sources in the area.
With the newly released EPA dioxin report and with existing data bases on levels of
PAHs and metals in the environment, there is ample opportunity to carry out this sort of
assessment. This issue is particularly relevant if mere are future efforts to monitor
exposure in this region. Until the total exposure assessment for the population is
included, the WTI risk assessment can not be used to inform public health assessment in
this community. In Volume V it is stated that total TEQ exposure due to operation of
the WTI incinerator is compared to the expected background TEQ exposure dose for
individuals living in the vicinity of WTI, but I could not find this comparison.
D-108
-------
T.E. McKone
Long-Term Research Recommendations
The one long-term issue that I believe needs to be-addressed is research and
development of better methods for organizing and presenting information in a risk
assessment. I think the risk assessment would have been much easier to interpret if risk
for each contaminant or contaminant category (i.e. CDD/CDF TEQ, PAH, particles,
acids, metals etc.) is presented so that we could assess the estimated level of dose and
compare it to background and to what toxicologists would consider a safe dose. Since
each of these doses is not known with precision we would expect a range comparison.
Estimate of the added dose
range from WTI
Estimated background
dose range
Estimated safe
dose range
Dose Scale
D-109
-------
T.E. McKone
HUMAN HEALTH RISK CHARACTERIZATION
As was done above, I went back to the 1993 Peer Review Report to compile a list of our
recommendations regarding the exposure component of the WTI Risk Assessment Plan.
Table 2 which is listed below covers our comments that dealt with issues of lesser
significance that we felt should be addressed or fixed. In the column to the right I have
noted to what extent I believe these issues are addressed in the draft risk assessment.
Table 2
Recommendation from 1993 Workshop
The models proposed in the plan are not suited to speciation of inorganic
metals. Mercury is a particular problem.
The process of handling collected ash from the kiln should be included in
the estimation of fugitive emissions.
A number of groups that might be considered as high-end or sensitive
exposure groups were identified. These are, (1) children who live near the
facility and attend school near the facility, (2) elderly, people (who
apparently make up a larger than expected fraction of the East Liverpool .
community), (3) those who are already highly exposed to metals (i.e., lead)
and dioxin-like compounds from other sources, (4) hunters of deer and
waterfowl (5) very active people with higher breathing and food
consumption rates,, and (6) individuals who both work at the WTI facility
and live near it.
There needs to be a survey of gardening patterns wilihin the primary plume
area — that is within East Liverpool itself and up and down the river valley
where contaminants might be blown much of the time.
The Biddleman model is used to determine the relative amount of chemical
in gas phase versus particles in air. There are empirical parameters used in
this model that have been fit using a small set of chemical data.
No discussion of exposure via household dust is included in the plan.
The transfer of chemicals from air to soil is by deposition, which does not
explicitly include diffusion.
The COMDEP model does not count snow as a form of precipitation for
purposes of estimating deposition of contaminants from air to soil.
The exposure concentrations included in this risk assessment should be
reconciled with those that have been compiled in several states.
High-end exposure duration for breast feeding is 270 days is too low to be
a high-end value.
Food consumption rates compiled for the risk assessment are.based on
survey data that is at least fifteen years old and should be made site
specific.
Was it
addressed?
Somewhat
Apparently
Mostly ,
Yes
Yes
NO
Somewhat
No
No
Yes
Yes •:
D-110
-------
T.E. McKone
Table 2 continued
Recommendation from 1993 Workshop
Was it
addressed?
How much uncertainty is added to the assessment by using data from the
similar incinerators and/or from the trial burn as the primary basis for
estimating the source of emissions.
Mostly
There are a number of partition and biotransfer factors that are included
in the model and many of these are likely to be quite uncertain. The
impacts of the uncertainty in these inputs should be addressed by
sensitivity and uncertainty analyses.
Yes
The results of this risk assessment will be more credible, if the variance in
the input values is clearly stated and the impact of these variances on the
final estimates of risk is assessed. At a minimum, this can be done by
listing the estimation error or the experimental variance associated with
the parameters when these values or their estimation equations are listed
in tables.
Yes
A sensitivity analysis should be used to assess how model predictions are
impacted by model reliability and data precision. The goal of a
sensitivity analysis is to rank the input parameters on the basis of their
contribution to variance in the output.
Yes
Variance propagation methods (including but not necessarily Monte-Carlo
methods) should be used to carefully map how the overall precision of
risk estimates is tied to the variability and uncertainty associated with
the models, inputs/and scenarios. •
Yes
Are the High-End and Central Tendency Exposures Properly Characterized?
The exposure media concentrations, intake rate and duration/frequency of exposure
appear to be properly combined in order to characterize high-end and central tendency
exposures.
Sources and Pathways of Exposure
The 1993 Peer-Review Report recommended that.discussion of exposure via household
dust be included in the risk assessment. The risk assessment does not consider this
pathway or explain why this is not a significant route of exposure for sensitive sub
groups such as infants and children. In my view, dermal and ingestion pathways for
outdoor soil do not necessarily represent how these contacts occur inside houses. House
dust likely originates from three sources, (1) airborne particles that penetrate from
outside air to indoor air; (2) surface soil and dust tracked into buildings on shoes or
clothes, by pets, or other vectors; and (3) a variety of sources related to occupant
activities, material degradation, and household products.
D-lll
-------
I.E. McKone
Throughout the report, the population-averaged potential dose (for ingestion or
inhalation routes) or absorbed dose (for dermal contact) is expressed as an average
daily dose rate (ADD), in mg/kg-d either during a lifetime (LADD) or in some cases
during the exposure duration. The general form of the expression used is as follows
[CkJ
ADD = ^ --' "" ' EFxED
.BWJ AT
In this expression [Cj/Qj is the intermedia-transfer factor, which expresses the ratio of
contaminant concentration in the exposure medium i (i.e., personal air, tap water, milk,
soil, etc.) to the concentration in an environmental medium k (ambient-air gases or
particles^ surface soil, root-zone soil, surface water, and ground water) and [IUj/BW] is.
the intake or uptake factor per unit body weight associated with the exposure medium f.
For exposure through the inhalation or ingestion route, [IUi/BW] is Ij the intake rate per
unit body weight of the exposure medium such as m.3(air)/kg-d, L(milk)/kg-d, or
kg(soil)/kg-d. For exposure through the dermal route, [IUj/BW] is replaced by UFj, the
uptake factor per unit body weight and per unit initial concentration in the applied
medium (L(water)/kg-d or kg(soil)/kg-d). EF is the exposure frequency for the exposed
population, in days per year; ED is the exposure duration for the exposed population, in
years; AT is the averaging time for the exposed population, in days; and C^ is the
contaminant concentration in environmental medium k.
From my experience working with the California Environmental Protection Agency (Cal-
EPA), I can report that this approach and the algorithms for calculating exposure dose
listed in Section VH of Volume V are very much in harmony with the approach used by
the Cal-EPA.
Have Critical Assumptions Been Identified?
In order to provide some verification for the surrogeite selection process, the quantity-
carcinogenic potency- bioaccumulation QCB scores of chemicals listed in Table IV-1 of '
Volume V should be compared to the relative contributions of each of these chemicals to
total estimated risk in the actual risk assessment.
D-112
-------
T.E. McKone
In Table VI-2 on page VI-20 of Volume V, there is a need to break out the assumption
that fate and transport models are accurate into more components. At a minimum the
biotransfer, diffusion, and advection (i.e. deposition) of the fate and transport models
should be separated out and listed as separate assumption categories.
Were Uncertainties Confronted in an Adequate Manner?
On pages VI-14 to VI-15 of Volume V, the discussion here is particularly useful. The
authors should be commended for their efforts to address both model and parameter
uncertainties in their evaluaoon of the fate and transport models.
One item that is not made dear in the executive summary is that the ratio of high-end to
central-tendency exposure (or risk) has a value much lower than ratio of the high to low
end of the range of exposure estimates attributable to uncertainty. The former ratio
reflects heterogeneity, whereas the latter reflects uncertainty. This means that we have
more confidence about the relative values of high-end versus central tendency exposure
than we do about the absolute value of the exposure. The risk assessment addresses
uncertainty by biasing both the high-end and central tendency values toward the upper
end of their likely range.
ACCIDENT ANALYSIS
Listed in Table 3 are the 1993 recommendations of the Peer-Review Exposure
Workgroup regarding accidents, along with my assessment as to whether the
recommendation has been addressed in the risk assessment.
Table 3
Recommendation from 1993 Workshop
Was it
addressed?
There are three types of events that should be included in this category—
upset conditions, fugitive emissions, and accidents. These events could I
result in larger annual releases than the routine emissions to which the
majority of the report is devoted.
In the case of releases from accidents it will be necessary to make use of a
combination of fault-tree studies, local transportation accidents data,
reviews of operating experience, and reviews of past experience to
determine both the frequency of accidents and the chemical source terms
associated with these accidents.
Yes
D-113
-------
I.E. McKone
Comment on the Conclusions Regarding Severity of Consequences and Probability
of Occurrence
The report does not adequately express or communicate the expected value of harm for
accidents. The accident severity and consequence information is coded into phases that
are hard to interpret—such as "likely", "unlikely", etc. events and "moderate"
(1 fatality?!) to "catastrophic" consequences. Based on my applications of a little
fuzzy logic and what I read from the tables, I calculate roughly one in a thousand chance
per year of an accident that could kill something on the order of 10 people. Does this
mean that ever 10 years of operation, we have a 0.1 or 10% (10x10/1000) likelihood of
at least one fatality in the community as a result of accidents. If so this is a very large
risk relative to the one in a million chance of cancer per individual.
Has the Accident Assessment Adequately Confronted Uncertainties?
Most of the uncertainties are dealt with in a qualitative manner this leads to concerns
about the reliability of the estimates
D-114
-------
Toxicology
D-115
-------
-------
COMMENTS FROM GEORGE V. ALEXEEFF, PH.D., DART
The document review has been organized in response to questions posed to the
"Charge to Reviewers." Only questions for which I have formulated answers are
included in the attachment.
GENERAL ISSUES
1. Comment on the organization of the risk assessment document. Does the layout
follow a logical format? Is the presentation of the information in the document
clear, concise and easy to follow?
The risk assessment document is well-organized, follows a logical format, and is
presented in a clear, relatively concise, easy to follow manner. The document is very
well-written. 'Finding specific information within the document is difficult. While there is
a lot of cross-references, additional cross-references (particularly with regard to
chemical-specific data), would be helpful. The document could be improved if the
noncancer health effects of concern were highlighted for those substances evaluated in
depth.
The following suggestions would improve the reviewers ability to analyze the
document in depth. The units in the document change between ppm, and mg/m3. It
would be helpful if the other units could be presented in parenthesis so comparisons
could be made more easily. The cancer risks of the emitted substances are presented
in the form of per mg-kg-day. Since inhalation risks are often expressed as unit risks in
the form of per microgram per cubic meter, it would be helpful to have that information
available in the tables as well. Many of the tables in the document, such as, Tables III-
1 and III-4 have an "NL" or not listed under the slope factor column. The notation refers
to whether or not information is available in IRIS or HEAST. It is not clear if the absence
of a slope factor is due to it not being carcinogenic, or the unavailability of a number. It
would be helpful to clarify this in the tables. The presentation of noncancer health
values in Table IV-2 was changed from inhalation RAC values in Table 111-1 to RfD
• Page 1
D-117
-------
COMMENTS FROM GEORGE V. ALEXEEFF, PH.D., D.A.B.T.
values. It is not clear why this was changed, what the conversion factors were, or how it
affects the assessment.
2. Does the executive summary accurately reflect the data and methodologies used
and the conclusions derived in the risk assessment?
Yes, in general, the executive summary accurately reflects the data and
methodologies used and the conclusions derived in the risk assessment. Overall, the
methodology described for the human health risk assessment appears to be consistent
with the methodology used in other risk assessments to evaluate stationary facility
emissions.
4. Are there any major data or methodological gaps that would preclude the use of
this risk assessment for decision making? If so, how should they be addressed?
The primary methodological gap identified was in the use of IDLH values in the
accident analysis. As indicated below, alternate values that appear to more accurately
reflect the toxicity of the substances are available. Application of these other values
may increase the severity ranking of the potential consequences. '
Another methodological issue of concern is the absence of evaluating the
potential impact of typical emission upsets or excursions above the. long-term
background emissions.
A third issue would be to more clearly indicate the summation of th6 cancer
risks, including the breast milk pathway.
5. What long-term research would you recommend that could improve risk
assessments of this type in the future?
• Page 2
D-118
-------
COMMENTS FROM GEORGE V. ALEXEEFF, PH.D., D.A.S.T.
The primary research needed is to develop more appropriate health levels
especially to evaluate potential acute health effects for both intermittent upset conditions
and for accident analysis as well as to assess chronic noncancer risks.
For many of the substances without IRIS or HEAST values, California risk
assessment agencies have developed toxicity-based values to allow for evaluation of
emissions. Further collaboration between the Cal/EPA and U.S. EPA could improve
evaluation of the health effects of complex facility emissions, such as the one described
in the report.
Another primary area is to better identify the input parameters and distributions
of the parameters.
HUMAN HEALTH RISKS
EXPOSURE
1. EPA's Exposure Assessment Guidelines identify certain exposure descriptors
that should be used to characterize exposure estimates. The Guidelines define
high end exposure estimates as those representing individuals above the 90th
percentile on the exposure distribution but not higher than the individual in the
population who has fhe highest exposure. Bounding exposure estimates are
those that are higher than the exposure incurred by the person in the population
with the highest exposure. Central tendency exposure estimates are defined as
the best representation of the center of the exposure distribution (e.g., arithmetic
mean for normal distributions). Comment on whether or not the WTI exposure
assessment properly characterizes each of the exposure estimates in terms of
these descriptors.
The EPA's Exposure Assessment Guidelines are somewhat different from the
ones I am familiar with, the California Air Pollution Control Officers' Air Toxics 'Hot
Spots' Program Risk Assessment Guidelines. The WTI exposure assessment appears
to have properly characterized the central tendency exposure estimates based on EPA
• Page 3
D-119
-------
COMMENTS FROM GEORGE V. ALEXEEFF, PH.D., D.A-B.T.
procedures, but the results for the high end exposure estimates would, be
underestimates under the California Air Pollution Control Officers' Risk Assessment
Guidelines. Some of the areas of concern for the high end estimate are the
assumptions regarding exposure duration, fish consumption, and the breast milk
exposure pathway. In particular, it is unclear how the exposure duration, which appears
to define the length of time in a single residence, relates to living in the particular
community of concern. Furthermore, it does not appear that an evaluation was done on
whether there are individuals that might subsist on fishing to a large part residing in the
community, and if their exposure has been considered. The fish consumption rate is
low in comparison to levels used in the California Air Pollution Control Officers' Risk
Assessment Guidelines and it appears to be low due to the 11-county averaging
procedure. :
2. The factors that go into estimating a central tendency .or high end exposure,
once the population has been defined, include the environmental media
concentration, the intake rate, and the duration and/or frequency of exposure.
Comment on whether or not the WTI exposure assessment does an adequate
job of describing the logical procedure of combining these factors to develop
centra/ tendency, high end, and/or bounding estimates of exposure for each of
the exposed subpopulations.
The WTI exposure assessment appears to do an adequate job describing the
components of the exposure estimates. The breast milk and fish consumption pathways
were difficult to evaluate due the extensive material in both the body of the document
and in the appendices. It is not clear if the breast milk pathway is incorporated into the
final risk estimates.
3. An important factor in an exposure assessment is identifying all of the important
exposure sources. Please comment on the adequacy of the WTI assessment in
identifying the important sources and pathways of exposure.
• Page 4
D-120
-------
COMMENTS FROM GEORGE V. ALEXEEFF, PH.D., D.A.B.T.
The risk assessment appears to have adequately identified the important
sources and pathways of exposure. The evaluation appears to follow standard U.S.
EPA procedures.
The document refers to a concept of potential dose. This concept is confusing
and its necessity is unclear. Since essentially all potency values and reference doses
are based on potential doses as well, the concept does not seem to clarify the issue.
4. Have the key assumptions for estimation of chemical concentration and for
estimation of exposure been identified? Are the magnitude and direction of
effect correct for the assumptions that have been identified?
The document identifies what appears to be the two key assumptions in terms of
impact: fate and transport modeling and chemical specific inputs. In both cases the best
available data were used. It is unclear how the best available data result in likely
overestimates of risk. Upon review of selected parameters, the values appear be within
the reported range of values. Clarification on how the parameters were chosen to be
conservative would be helpful.
Due to the voluminous nature of the documentation it would have been helpful to
state the importance of specific input parameters on the results of the risk assessment.
After reading the risk assessment it was difficult to sense the key drivers of the risk.
5. Supposedly, conservative assumptions have been applied in this assessment to
account for uncertainty. Are the conservative assumptions appropriately
factored into the ultimate characterization of what descriptor best applies to each
exposure estimate? Please comment on whether the uncertainties were
confronted in an adequate manner. If they were not, please state what should
be done differently.
The document discusses the uncertainty in many of the assumptions made in
the analysis. However, the document does not extensively discuss the uncertainty with
regards to data gaps and the absence of information. This leaves the impression that
• Page 5
D-121
-------
COMMENTS FROM GEORGE V. ALEXEEFF, PH.D., D.A.B.T.
the uncertainties of over-prediction are emphasized, while those of under-prediction are
not mentioned in places like the summary. One example is that reference
concentrations are available for only 54 of the 96 compounds listed as fugitive
emissions. Consequently, the exposure to these substances cannot be considered
quantitatively. Another example is that the summary does not indicate that for 77 of the
carcinogenic substances emitted, the U.S. EPA does not have potencies calculated for
them, and consequently their contribution to the carcinogenic risk is not considered in
the evaluation.
With regard to uncertainties of exposure parameters, the choice of input
parameters for Kow being conservatively selected appears unclear. The Kow values
used appear to represent measured values in the literature. The range of values for a
chemical may occur for a number of reasons. However, since the values chosen were
from the range, it is unclear how they tend to overestimate risk as indicated in Table VI-
2. Data gaps in the literature are not adequately addressed in the uncertainty
evaluation.
Lifetime chronic doses could result in an underestimate of risks for the following
reasons. In establishing a chronic RfC or RfD. an exposure of 1 year or more may be
used. The experimental dose or study dose from which the RfC or RfD was derived
may not have occurred over the lifetime of the test subject. In the risk estimate of dose,
doses are averaged over a lifetime of the test subject. Thus, a higher exposure could
occur a few times a year and be averaged over a lifetime. If the exposure had not been
averaged, it may have exceeded the RfC for some period of time less than a year.
. • Page 6
D-122
-------
COMMENTS FROM GEORGE V. ALEXEEFF, PH.D., D.A.B.T.
HAZARD IDENTIFICATION/DOSE RESPONSE AND RISK CHARACTERIZATION
1. To select surrogate compounds for quantitative risk assessment, a two step
process was used in which chemicals were ranked on the basis of emission rate,
toxicity (both cancer and non-cancer), and bioaccumulation potential. Please
comment on this selection process. Are the ranking factors appropriate? Could
important compounds have been omitted from the analysis based on the ranking
procedure?
Yes, important chemicals could have been omitted from the analysis and this is a
key uncertainty that should be discussed in the document. The current summary does
not indicate: 1) many of the surrogate substances emitted (72 of 159 stack emissions,
45 of 96 of fugitive emissions) do not have RfDs or RfCs calculated for them; 2) the 300
substances considered emitted were reduced to less than 100 based on total pumpable
feed processed. While this seems reasonable, the substances not considered do
constitute an uncertainty.
The procedure to identify substances that bioaccumulate and have potential
long-term toxicity appears to be sufficient. However, in many of the risk assessments
the inhalation pathway is often the dominant pathway. For this reason, there may be
important air emissions that do not bioaccumuiate but are significant for the risk
assessment. Were any procedures taken to determine if any such chemicals were
missed? Does the procedure for selecting fugitive emissions result in" identifying the
chemicals of concern for stack emissions? Possibly, a calculation of emission rate
divided by potency would identify the highest ranking chemicals under those
circumstances.
The document acknowledges that some uncertainties are introduced into the
process of identifying the surrogate chemicals. The document states that a
"conservative" method was used to select Kow values and consequently the risk may be
overestimated. It does not appear clear what type of decision process was used in
selecting the Kow values from the ranges available in the literature. Furthermore, it is
D-123
-------
COMMENTS FROM GEORGE V. ALEXEEFF, PH.D., D.A.B.T.
unclear how a risk conclusion can be reached in chemical selection process. It seems
possible that the process could have resulted in overestimating the risk in a few
chemicals and thus resulting in the non-selection of other important chemicals.
Clarification of this issue would be helpful.
The assumptions regarding the selection of surrogate chemicals could use
additional clarification. Table IV-13 should be revised to include the uncertainties
referred to in the text of nature and the magnitude of the fugitive emissions. Table IV-13
should be" revised to include the possibility of overestimating the risk of some chemicals
by using high Kow values and consequently missing some important chemicals. The
procedure of using the Kow value in the formula appears interesting but new to the
reviewer. Other procedures often look at emissions and toxicity only. One uncertainty
is whether the introduction of the Kow value into the equation gets one closer to the
actual risk or not. It is not transparent to this reviewer.
2. For the majority of the chemicals of concern, traditional approaches to dose
response evaluation were employed (e.g., use of a slope factor for cancer and
use of a RfD/RfC for non-cancer). However, for certain chemicals or groups of
compounds a different methodology was used. Specifically, dioxins, furaris,
PAHs, lead, mercury, nickel, chromium, acid gases, and particulate matter were
given special consideration. Was it appropriate? Have the uncertainties
associated with the methodology been adequately characterized? Comment on
the assumptions used due to lack of chemical specific data.
To see a complete picture of the carcinogenic risk and the uncertainty associated
with estimating it, it would be helpful to indicate which of the carcinogenic substances do
not have available slope factors.
The document used a relative potency procedure to estimate the risk of PAH
exposure. However, it appears that the assessment assumes that the remaining PAHs
are not carcinogenic. However, IARC has identified an additional 16 PAHs as possibly
• PageS
D-124
-------
COMMENTS FROM GEORGE V. ALEXEEFF, PH.D., D.A.B.T.
or probably carcinogenic to humans. Also, a larger number of PAH compounds exhibit
strong genotoxicity. It would be helpful to know how much of the PAH fraction has been
evaluated based on these seven compounds. This would allow some quantification of
the uncertainty of the toxicity estimation for these compounds. The OEHHA in Cal/EPA
(Office of Environmental Health Hazard Assessment (OEHHA) 1993 Benzo[a]pyrene as
a Toxic Air Contaminant. Part B. Health. Effects of Benzo[a]pyrene. Office of
Environmental Health Hazard Assessment, California Environmental Protection Agency,
Air Toxicology and Epidemiology Section, Berkeley, CA.) developed relative potency
factors for a total of 25 PAHs; such information could be helpful to further refine the risk
assessment. An alternate comparison, that has been used by the California Air
Pollution Control Officers Association, would be to assume that the remaining fraction is
as potent as BAP and use that in the calculations. This could help bound the risk from
PAHs and could be used as part of the uncertainty analysis.
Since the carcinogenic activities of the various chemicals are added together, the
reasoning provided for not developing a cancer SF for lead does not appear convincing.
The document states that since "neurobehaviorai effects have been observed in children
with blood lead levels below those that have caused carcinogenic effects in laboratory
animals, a cancer SF has not been derived by U.S. EPA." This logic would only be
applicable if one was trying to determine the most sensitive effect of lead. However, the
risk assessment is looking at the toxicity of emissions from a facility. Consequently, the
carcinogenic activity of lead is relevant in ascertaining the impact of the facility. Using
U.S. EPA's methodology, the OEHHA (Health Effects of Airborne Inorganic Lead (Draft
1993) Office of Environmental Health Hazard Assessment, Cal/EPA) has developed a
draft upper bound range of inhalation unit risks of 1.2 x 10~5 to 6.5 x 10~5 (jig/m3)"1 for
inorganic lead. Such information could be derived to determine the contribution of lead
to the overall estimated cancer risk.
It is important that the risk assessment treat inorganic nickel as a carcinogen as
it proposes to do so. The IARC classification is based on a study that was co-funded by
U.S. EPA (International Committee on Nickel Carcinogenesis in Man (ICNCM) 1990,
ISSN 0365-3140 Scand J. Work and Environmental Health 16, no.1). The U.S. EPA
apparently has not updated its classification of nickel compounds since the publication
. • Page 9
D-125
-------
COMMENTS FROM GEORGE V. ALEXEEFF, PH.D.-, D.A.B.T.
of this study. Consequently, it appears that the current U.S. EPA nickel evaluation may
be out of date and would underestimate risk based on the best scientific information
currently available. The approach taken in the risk assessment attempts to account for
the more recent information and appears to be appropriate.
3. Please comment on the selection of the overall population and the various
subpopulations at risk. Were site specific data, such as the informal home
gardening survey, property, utilized to identify these subpopulations?
The document refers to the comparison of background exposure levels to those
of the most highly exposed individual. It would be useful to understand how many other
individuals may be close to the highest exposed, similar to the analysis done for off-site
consequence analysis. Additional information on potential subsistence fisherman would
be helpful.
5. Comment on whether or not the non-cancer risks of chemicals of concern have
been adequately addressed by the risk assessment? For example, has an
adequate discussion of endocrine disrupters been provided which either
characterizes their risks or clearly explains why their risks cannot be
characterized? Further, have non-cancer chronic toxicities of dioxins and furans
been adequately discussed in the risk assessment?
Noncarcinogenic .risk assessment is a difficult area to address completely due to
the substantial data gaps. The report does not address the issue of data gaps to a
great extent. Many chemicals of concern have not been thoroughly tested for
noncarcinogenic effects. Many substances have not been adequately tested for acute
effects, neurotoxicity, reproductive toxicity and many long-term health effects.
Consequently, choosing even the more sensitive reported studies may not result in
health levels that are protective for the untested health endpoints. This is an uncertainty
worth mentioning in the report.
• Page 10
D426
-------
COMMENTS. FROM GEORGE V. ALEXEEFF, PH.D., D.A.B.T.
Some of the statements regarding reference concentrations do not completely
reflect the uncertainty involved. Page 111-3 states that "since the RfD is intended to be
adequately protective of sensitive individuals, application of the RfD to the general
population is conservative." Sensitive individuals are members of the general
population; they generally include pregnant women, children, aged, and individuals with
chronic diseases such asthma. Thus, the statement appears to be an overstatement of
health protection. It would be better to state that the RfD is designed to protect sensitive
members of the population.
It is clear from Table 111-1 that RfCs were available for only approximately 41 of
the 215 substances listed plus an addition 82 values bases on route-to-route
extrapolation. However, route to route extrapolation may miss important irritation
effects or may be masked by poor oral absorption.
The non cancer health effects evaluation focuses on chronic exposures.
However, for many of the substance short-term excursions may be more important.
Exposure to the acid gases and other irritants may exceed irritating levels on a short-
term occasional basis, while the long-term averaged exposure is below irritating levels.
For a key group of substances, RfC values are not available for evaluation. The
recent U.S. EPA health effects document on chlorinated dibenzo-p-dioxins and
chlorinated dibenzofurans related compounds indicated the average levels in human
tissue from background exposure was 28 picograms of TEQ/g. The report also
indicated that waste incineration was a key source for dioxin emissions. Since non
canCer effects are thought to act by a threshold mechanism the emissions could be
considered additive to existing background levels. The sum total would be .of interest to
consider to determine the potential for dioxin-related health effects. The California Air
Pollution Control Officers Association uses the reference exposure level of 3.5 10"6
ng/m3 for the noncancer chronic health effects of chlorinated dibenzo-p-dioxins and
chlorinated dibenzofurans.
• Page 11
D-127
-------
COMMENTS FROM GEORGE V. ALEXEEFF, PH.D., D.A.B.T.
The document states that it is valuable to compare the incremental exposures to
dioxin-like compounds to the expected background exposure levels. However, such a
comparison may suggest that background levels are acceptable or at least unavoidable.
However, if the recent USEPA health effects document on dioxins and related
compounds is correct, then the existing background levels are in large part due
anthropogenic practices, then usefulness of the comparison becomes unclear.
For some of the chemicals listed in the document wifhout U.S. EPA reference
concentrations, the California Air Pollution Control Officers Association has developed
chronic exposure values. These include benzene (RfC of 71), chlorinated dibenzo-p-
dioxins (RfC of 3.5 x 10"6), chlorinated dibenzofurans (RfC of 3.5 x 1Q"6), copper (RfC of
180), dimethylarnine (RfC of 2), 1,4-dioxane, (RfC of 4.0), ethyl acrylate (RfC of 48),
ethylene oxide (RfC of 600), hydrazine (RfC of 0.24), lead (RfC of 1.5), toluene
diisocyanate ((RfC of 0.095) arid vinyl chloride (RfC of 26). If .the emission rates for
these substances were given in the report, it would be possible to determine how they
would be ultimately scored using the procedures described.
The formaldehyde RAC value of 180 jug/rn3 in Table III-4 and an RAC value of
175 fig/m3 in Table IV-6 (RfC of approximately 720 ng/m3) should probably be double-
checked. The value is based on a route to route extrapolation in the risk assessment.
An inhalation RfC value used by the California Air Pollution Control Officers Association
is 3.6 ng/m3 or approximately 20 times lower. Logically, it would not appear that
formaldehyde should have an RfC that is twice* that of acetone or 3 times that of
dichlorodifluoromethane.' The 3.6 jig/m3 value is based on prevention of eye and nose
irritation.
In Table III-6, the assumption that RACs are developed from oral RfDs when
RfCs are not available states that it overestimates the risk. This does not appear to be
sufficiently substantiated since oral absorption may be much less than inhalation
absorption and respiratory effects may be significant but not evaluated in oral studies.
• Page 12
IM28
-------
COMMENTS FROM GEORGE V. ALEXEEFF, PH.D., D.A.B. T.
7. Have the key assumptions for estimation of dose and risk been identified? Are
the magnitude and direction of effect correct for the assumptions that have been
identified? Please comment on whether the uncertainties were confronted in an
adequate manner. If they are not, please state what should be done differently.
In Table lil-6 a number of statements regarding uncertainty are made. Their
basis is in some cases unclear and they may not be correct. The nickel statement
regarding and overestimate of carcinogenicity is not necessarily correct. The U.S. EPA
co-funded the human epidemiological reanalysis on which the IARC determination is
based. The results indicated that nickel oxides encountered in the nickel refining
industry were carcinogenic to humans.
It is unclear why the TEF scheme is thought to overestimate the risk of dioxin
exposure. The approach reflects the best use of the data, the compounds are
considered to be equal to or less potent than TCDD. It is unclear why the uncertainty is
labeled an overestimate.
It is unclear why the PAH relative potency scheme is thought to overestimate the
risk of PAH exposure. The approach reflects the best use of the data, the compounds.
are considered to be less potent than BaP. It is unclear why the uncertainty is labeled
an overestimate. It is more likely that the scheme underestimates the risk of PAHs
since not all carcinogenic PAHs are accounted for in the scheme.
It is unclear why the use of NAAQS values are considered overestimates of risk.
The standards are based on extensive human data. The standards incorporate small
margins of safety due to the high quality of the data. However, they could be
underestimates of risk if an acute standard is applied to a chronic exposure. Especially
in light of the data now available on particulate matter, it is unlikely that use of the
current NAAQS overestimates its toxicity.
• Page 13
D-129
-------
COMMENTS FROM GEORGE V. ALEXEEFF, PH.D., D.A.B.T.
The table does not discuss several important areas of uncertainty that indicate
that the overall risks could be underestimated. These areas include 1) the substances
emitted which do not have RfDs or RfCs calculated for them; 2) the focus on chronic
exposure effects and not the potential acute effects of repeated excursions above the
chronic exposure level; and 3) the incomplete nature of the database for noncancer
health effects resulting in the possible omission of critical adverse health effects.
The risk assessment makes the assumption that 100% of the chromium
emissions are in the hexavalent form. This appears to be an overestimate of emissions.
w
However, the potency slope used for hexavalent chromium in the risk assessment
appears to be based on total chromium exposure. If that is the case, then the
overestimate may not be that significant and would represent the relative difference
between the hexavalent chromium content in the dose-response study and in the facility
emissions.
-8. Please comment on the overall adequacy of the risk characterization. Does the
risk characterization include a statement of confidence in the risk assessment
including a discussion of the major uncertainties. Are the hazard identification,
dose-response assessment, and exposure assessment clearly presented? Have
sufficient risk descriptors which include important subgroups been presented and
discussed?
The uncertainties regarding the data gaps of emitted substances are not
substantially discussed in the document as described above. Furthermore, the impact
of many substances cannot be estimated since toxicity values have not been
established. The document relies on a number of extrapolations from oral to inhalation
toxicity. Extrapolation from oral to inhalation may underestimate risk if absorption is
greater by the inhalation route, or if the effect if respiratory tract is a target organ, as it
would be for irritants. This reviewer believes that such extrapolations are more likely to
underestimate risk. For this reason, additional uncertainty factors are often included in
• Page 14
D-130
-------
COMMENTS FROM GEORGE V. ALEXEEFF, PH.D., D.A.B. T.
such extrapolations. However, the statements in the document suggests cross route
extrapolation will overestimate risk, and do not appear to be substantiated.
While the document acknowledges the potential for endocrine disruption, it does
not quantify the potential risk and it does not acknowledge that by not quantifying the
risk, that the uncertainty is one that tends to underestimate risk.
ACCIDENT ANALYSIS
1. The WTI accident assessment selected five scenarios for quantitative evaluation
that were considered to be of primary concern. The scenarios are an on-site
spill, an on-site fire, an on-site mixing of incompatible waste, an off-site spill, and
an off-site spill and fire. Please comment on the selection of these scenarios.
Were any significant scenarios missed?~
The scenarios appear to be reasonable choices. However, it is unclear why a
release associated, with a function of the facility, i.e., an equipment failure, was not
chosen. The document lists equipment failure as one of the conservative choices.
Later in the document it states that the scenarios were chosen based on guidance, WTI
design characteristics and accident reports in the industry, combined with the' potential
for significant off-site consequences and the potential for occurring within 30 years.
Thus, it appears that judgment was used to select accidents with greater frequency.
One question in this regard is how age of facility was accounted for in the scenario
selection analysis. Are there a sufficient number of older facilities to determine the
probability of failure in the 10 to 30 year age bracket?
Use of the descriptors of the releases, typical verses conservative, are unclear.
They appear to be probability related. Using the F£MA guideline terminology, it seems
that-scenarios were chosen that were either common (several accidents a year), likely
(once every 10 years) or reasonably likely (accidents once every 10 to 100 years). If that
is the case it may be clearer to use the descriptors in the FEMA (1993) guidelines of
common, likely, or reasonably likely in the initial description of the scenarios.
• Page 15
D-131
-------
COMMENTS FROM GEORGE V. ALEXEEFF, PH.D., D.A.B.T.
2. Specific chemicals were selected to evaluate each scenario. Please comment
on the selections. Would other chemicals have been more appropriate?
The choice of acetone to evaluate on-site spill and off-site spill scenarios may
have been inappropriate. Page II1-2 states that "a key factor in evaluating the
consequences of accidental releases is the acute toxicity criterion." The document
further states that "the purpose of the WTI Accident Analysis is to determine the areas
where, if any accidental releases were to occur, serious irreversible health effects are
possible."
On June 16, 1995, U.S. EPA granted a petition to delete acetone from the list of
toxic chemicals under Section 313 of the Emergency Planning and Community Right-to-
know Act (Fed Reg. 60(116):31643). The Federal Register states "It was EPA's belief
that there was insufficient evidence to demonstrate that acetone causes or can
reasonably be anticipated to cause significant adverse human health or environment
effects." The Federal Register further states "...acetone (1) cannot reasonably be
anticipated to cause cancer or neurotoxicity and has not been shown to be mutagenic,
and (2) cannot reasonably be anticipated to cause adverse developmental effects or
other chronic effects except at relatively high dose levels." Further it states "...acetone
causes adverse environmental effects only at relatively high dose levels." It is my
understanding that scenarios developed by U.S. EPA indicated that severe toxic levels
would not be expected to occur in any acetone releases. For these reasons it seems
that acetone would not be expected to be found a concern in the WTI risk assessment.
Thus, if the intent is to determine the impact of a release of commonly transported
hazardous substance, another chemical should be chosen.
The choices of formaldehyde, phosgenes and HCI appear to be appropriate for
the accident analysis.
• Page 16
D-132
-------
COMMENTS FROM GEORGE V. ALEXEEFF, PH.D., D.A.B. r.
5. Please comment on the assessment's conclusions on the severity of
consequences and probability of occurrence. Has the report correctly
categorized the severity of the consequences of the different accident-
scenarios? Has the assessment adequately justified the reported probability of
occurrence of each of the accident events?
As indicated below, under question 7, it is suggested that the severity of
the consequences be reevaluated for formaldehyde, hydrogen chloride and
phosgene. For formaldehyde, reevaluating the accidental impact based on the
AIHA ERPG-2 of 10 ppm indicates that the severity ranking may increase for the
conservative and typical scenario. For, hydrogen chloride, reevaluating the
accidental impact based on the 1-hour SPEGL of 1 ppm or the 1-hour EEGL of
20 ppm may increase the severity ranking for the on-site fire scenario, offrsite fire
scenario and on-site mixing of incompatible wastes scenario. For phosgene,
reevaluating the accidental impact based on the 1-hour EEGL (or AIHA ERPG-2)
of 0.2 ppm would likely increase the severity ranking for the on-site fire scenario
and off-site fire scenario.
6. Key assumptions were made in the identification of accident scenarios and the
description of the conservative and typical events. Included were a description
of the magnitude of the effect of the assumptions and direction of-the effect.
Please comment on the assumptions. Are they justified? Are the descriptions of
the magnitude and directions of the effects correct? Has the accident
assessment adequately confronted the uncertainties involved in doing this type
of analysis? If not, what else should be done?
One assumption used in the accident analysis states that "IDLH values can be
used as a benchmark to evaluate extent of possible off-site health effects." This
assumption is ranked in the report as having a "high" magnitude of effect with and the
direction of the effect "may over or underestimate size of area over which effects may
be observed, depending on derivation of IDLH value." As discussed in response to
, Page 17
D-133
-------
COMMENTS FROM GEORGE V. ALEXEEFF, PH.D., D.A.B.T.
questions 5 and 6, it does not appear that the assumption is correctly characterized.
The IDLH values are not consistent acute toxicity criteria, and the available scientific
data for formaldehyde, hydrogen chloride and phosgene suggest that the IDLH would
underestimate their toxicity.
7. Comment on the appropriateness of using IDLH values for characterizing the
severity of consequences in the accident analysis. Comment on the
appropriateness of using 10XLOC for chemical for which IDLH values have not
been established.
Page III-3 of Volume VII of the report states:
Immediately Dangerous to Life or Health (IDLH) values established by the
National Institute for Occupational Safety and Health (NIOSH) are used in the Accident
analysis as the acute toxicity criteria for evaluating potential off-site consequences,
because they represent a consistent, relatively comprehensive set 'of criteria for
assessing the acute effects of short-term exposures. IDLH values are defined by
NIOSH as the maximum airborne contaminant concentrations from which an individual
could escape within 30 minutes without any escape-impairing "symptoms or any .
irreversible health effects (FEMA 1993). This definition-is consistent with the purpose of
the Accident Analysis. Although the IDLH is primarily used for selection of occupational
respiratory protection levels, the IDLH values represent a consistent, relatively
comprehensive set of criteria that can be used to estimate the areas in an accidental
release situation where people may be potentially exposed to harmful concentrations of
hazardous substances. The IDLH values used in Accident Analysis can be found in
U.S. EPA (1995c). The document also states the U.S. EPA chose to use the IDLH
values for the quantitative evaluation to generate consistent, comparable results.
An evaluation of such an application [PROBLEMS ASSOCIATED WITH THE USE OF
IMMEDIATELY DANGEROUS TO LIFE'AND HEALTT-! (IDLH) VALUES FOR ESTIMATING THE
HAZARD OF ACCIDENTAL CHEMICAL RELEASES - George V. Alexeeff, MichaelJ. Lipsett and
'• Page 18
D-134
-------
COMMENTS FROM GEORGE V. ALEXEEFF, PH.D., D.A.B. r.
Kenneth W. Kizer, American Industrial Hygiene Association Journal: 50(11):598-605
(1989)] suggests that it would be. inappropriate to classify chemicals for the basis of
accidental release planning by using the IDLH.
The paper concluded the IDLH values were developed for the purpose of
respirator selection, riot permissible exposure. In this study, 84 of the 336 IDLH values
were reviewed critically. For 79 of the 84 compounds (94%), the IDLH concentrations
did not appear to be adequately protective for a 30-minute exposure. Comparing IDLH
values to LC50s, 18 compounds were in the same range as lethal levels for animals.
Severe toxic effects might result from exposure to the IDLH concentrations for 45
compounds. All the NAS emergency exposure guidance levels developed for the
military were below the respective IDLH values. In comparison to lethal or severe
toxicity endpoints, the IDLH varied up to four orders of magnitude; and the IDLH values
vary by 200-fold when compared with MAS emergency guidance levels. Thus, IDLH
values represent inconsistent estimates of toxicity. Consequently, the use of IDLH
values as planning guidelines for accidental releases would appear inappropriate.
Several of the IDLH values have been subsequently changed; however, it is likely that
the general criticism still holds as indicated below.
Specifically with regard to the substances evaluated in the document, the report
made the following points. Acetone's IDLH was similar to the RD50 (respiratory rate
depression of 50% in mice) reported for .the compound. As indicated in the report,
exposure to 1/10 the RD50 would be expected to be irritating to the eyes, nose, and
throat but would be tolerable, while 1/100 the RD50 would cause slight to negligible
irritation. The IDLH for acetone was also found to be 2.4 times greater than the 1-hour
emergency exposure guidance level developed by NAS committee on toxicology to
protect military personnel.
With regard to formaldehyde, the IDLH was found to be 32 times greater than
the RD50 reported for the compound and 1/8 the LCSO reported in the rat.
• Page 19
D-135
-------
COMMENTS FROM GEORGE V. ALEXEEFF, PH.D., D.A.B.T.
For hydrogen chloride, the IDLH was found to be reasonably below the LC50 and
RD50l however it was 100 times greater than the SPEGL developed by the NAS.
With regards to phosgene, little was reported except that the IDLH'was 10 times
greater than the EEGL developed by the NAS.
The Executive Summary of the risk assessment document cites FEMA (1993) as
the basis for choosing the IDLH: "FEMA (1993) also presents a four-tier system for
classifying the consequences of accident scenarios. This system has been used as the
basis for developing the following severity of consequence categories in the Accident
Analysis: Minor - No exceedance of an IDLH.... Moderate - Exceedance of IDLH values
in inhabited areas over distances of 100 meters or less ...Major - Exceedance of IDLH
values in inhabited areas over distances between 100 and 1,000 meters... Catastrophip
- Exceedance of IDLH values in inhabited areas over distances greater than 1,000
meters..." However, the FEMA (1993) document actually states:
"Minor accidents are specified herein as those with the potential to have
one or more of the following features: low potential for serious human injuries; no
potential for human fatalities; no need for a formal evacuation, although the
public may be cleared from the immediate area of the spill or discharge;
localized, non-severe contamination of the environment which does not require
costly cleanup and recovery efforts; no need for resources beyond those
normally and currently available to local response forces.
"Accidents are specified herein as of moderate severity when they have
the potential to have one or more of the following features: up to 10 potential
human fatalities; up to 100 potential human injuries requiring medical treatment
or observation; evacuation of up to 2000 people; localized contamination of the
environment requiring a formal but quickly accomplished cleanup effort; possible
assistance needed from county and state authorities; only limited need for
specialized equipment, services, or materials for a rapid and effective response.
• Page 20
D-136 ,
-------
COMMENTS FROM GEORGE V. ALEXEEFF, PH.D., D.A.B.T.
"Major accidents are specified herein as those with the potential to have
one or more of the following features: up to 100 potential fatalities; up to several
hundred potential human injuries requiring medical treatment or observation;
evacuation of up to 20,000 people; significant contamination of the environment
requiring a formal and somewhat prolonged cleanup effort; assistance needed
from county, state, and possible federal authorities; significant need for
specialized equipment, services, or materials for a rapid and effective response.
"Catastrophic accidents are defined as those having the potential to have
one or more of the following features: more than 100 potential human fatalities;
more than 300 potential human injuries requiring formal medical treatment;
evacuation of more than 20,000 people; significant contamination of the
environment requiring a formal, prolonged, and expensive cleanup effort to
protect human health and the environment; assistance needed from county,
state and federal authorities; significant need for specialized equipment,
services, or materials for a rapid and effective response."
The FEMA (1993) also suggested that NAS values be chosen as a first priority
by stating: "Some options, in order of decreasing preference, and by no means
mandatory for use, are as follows: use the NAS/NRC SPEGL or the AIHA ERPG-2
value for the material if one has been established; consult a toxicologist or similarly
qualified individual for advice based on a formal review of the toxicity of the material of
concern; use the highest value among the following:
IDLH value divided by 10 (with "10" being a safety factor)
TLV-STEL
TLV-TWA multiplied by 3 (if a TLV-STEL does not exist)
TLV-C"
Consequently, the FEMA document does not appear to recommend the IDLH.
However, even if the IDLH was used in the screening for surrogate chemicals, other
more specific values could have been'used in the actual scenario analysis.
• Page 21
D-137
-------
COMMENTS FROM GEORGE V. ALEXEEFF, PH.D., D.A.B.T.
SPECIFIC COMMENTS ON THE USE OF THE FORMALDEHYDE IDLH IN THE ACCIDENT ANALYSIS
The choice of the IDLH to evaluate the severe toxicity of formaldehyde may'be
inappropriate. The IDLH for formaldehyde is provided in Table 111-1 as 0.024 g/m3
(which is equivalent to 20 ppm); it would have been helpful if the document provided the
level in ppm in various tables and appropriate places since all the comparisons are
made in the document based on ppm. The basis of the IDLH is Patty (1963) reported
that exposure to 10 to 20 ppm produces almost immediate eye irritation and a sharp
burning sensation of the nose and throat which may be associated with sneezing,
difficulty in taking a deep breath, and coughing; recovery is prompt from these transient
effects and that exposure for 5 to 10 minutes to 50 to 100 ppm might cause serious
injury to the lower respiratory passages in man.
At the present, time the.AlHA ERPG-2 (AIHA 1991) appears to. be a better
toxfctty criterion for evaluating the serious irreversible health effects of formaldehyde.
This value is 10 ppm (12 mg/m?). The value is reportedly based on Brabec, 1981; Kulle
et al., 1987; Sim and Rattle, 1957, and AIHA, 1991.
Sim and Pattle (1957) exposed twelve men to 17.3 mg/m3 (13.9 ppm) for 30
minutes. This concentration of formaldehyde caused "considerable nasal and eye
irritation when they first entered the chamber, but, despite the continued mild lacrimation
for some period of time, there was no marked response (pulmonary or cardiovascular)
to the exposure." The eye irritation was not severe and was absent after 10 minutes in
the chamber (Sim and Pattle, 1957).
The intensity of sensory irritation symptoms diminishes during exposure to
formaldehyde at approximately 5 ppm, however, tolerance is lost after 1-2 hours of
exposure (Brabec, 1981). The ERPG document mistakenly cites the Kulle et al. study
as supporting a 10 ppm ERPG-2. No data on exposures to 10 ppm formaldehyde were
available in the Kulle article. Feinman (1988) states that most people cannot tolerate
exposures to more than 5 ppm formaldehyde in air; above 10-20 ppm symptoms
become severe and shortness of breath occurs. The Sim and Pattle data are based on
only 12 healthy men and are poorly presented. In contrast to the Sim and Pattle finding
• Page 22
t
D-138 ;
-------
COMMENTS FROM GEORGE V. ALEXEEFF, PH.D., O.A.B. T.
that irritation was absent after 10 minutes, Brabec (1981) states that tolerance to the
sensory irritation induced by 5 ppm formaldehyde was lost after 1-2 hours of exposure.
Because of the conflicting evidence and interpretations offered by these studies, as well
as the failure to adjust for duration of exposure and sensitive individuals, the level may
underestimate the impact of exposure to the general population. Despite the
shortcomings of the AIHA value and the uncertainties in the conflicting data, the
ERPG-2 is a more scientifically credible basis for the severe effects of formaldehyde
than the IDLH.
REFERENCES FOR FORMALDEHYDE
American Industrial Hygiene Association (AIHA) 1991. Formaldehyde. Emergency
Response Planning Guidelines. AIHA. Akron, OH.
Brabec, J.B. 1981. Aldehydes and Acetals. In Patty's Industrial Hygiene and
Toxicology, Volume II A, Toxicology. 3rd. ed., edited by D.C. Clayton and F.E. Clayton.
New York: John Wiley & Sons,, pp. 2637-2669.
Kulle, J.T., LR. Sauder, J.R. Hebel, D. Green, and M.D. Chatham 1987. Formaldehyde
dose-response in healthy nonsmokers. J. Air Pollution Control Assoc. 37:919-924.
Patty FA, ed. [1963] Industrial Hygiene and Toxicology, 2nd edition, revised. Vol. !!
Toxicology. New York, NY Interscience Publishers, p. 1971
Sim V.M. and R.E. Pattle 1957. Effect of possible smog irritants on human subjects. J.
Am. Med. Assoc. 165:1908-1913.
SPECIFIC COMMENTS ON THE USE OF THE HYDROGEN CHLORIDE IDLH IN THE ACCIDENT
ANALYSIS
The choice of the IDLH to evaluate the severe toxicity of HCI may be
inappropriate. The IDLH for HCI is provided in Table 111-1 as 0.075 g/m3. (which is
equivalent to 50 ppm); it would have been helpful if the document provided the level in
ppm in various tables .and appropriate places since all the comparisons are made in the
document based on ppm. The basis of the IDLH is Patty (1963) stating that according
to Matt (1889, yes, over 100 years ago), as cited in Flury and Zemik (1931), work is
impossible when one inhales air containing hydrogen chloride.in concentrations of 75 to
150 mg/m3 (50 to 100 ppm); work is difficult but possible when the air contains
• Page 23
D-139
-------
COMMENTS FROM GEORGE V. ALEXEEFF, PH.D., D.A.B.T.
concentrations of 15 to 75 mg/m3 (10 to 50 ppm); and work is undisturbed at; the
concentration of 15 mg/m3 (10 pprri). !
According to FEMA (1993) the best value to use to evaluate serious irreversible
health effects of hydrogen chloride is the 1-hour SPEGL (Short-term Public Emergency
Planning Level) of 1 ppm. The rationale states "...in connection with community
exposure during space shuttle launches, the Committee recommends lower
concentrations, to avoid adverse effects that might occur in a more sensitive
population..." (NRC, 1987). While it appears that no supporting data is cited to justify
the value, the SPEGL essentially incorporates an additional 20-fold safety factor to the
Emergency Exposure Guidance Level of 20 ppm, to protect sensitive subpopulations.
The EEGL of 20 ppm (29.8 mg/m3) is based on NIRC, 1987; and Kane et al., 1979.
The RD50 in mice for a 10-minute exposure to HCI is reported as 309 ppm
(460 mg/m3) (Kane et al. 1979). The NRC applied an uncertainty factor of 10 to the
o to account for interspecies differences yielding a 1-hour EEGL of 31 ppm. The
EEGL was further reduced to 20 ppm because "of the paucity of human data." The 1
ppm value may be an overestimate of the concentration of hydrogen chloride that could
produce an serious health impact. Consequently a level closer to 20 ppm or perhaps
the range should be considered in the accident analysis.
REFERENCES FOR HYDROGEN CHLORIDE
Flury F and Zemik F (1931) Schadliche Gase Dampfe, Nebel, Rauch- und Staubarten
(In German). Berlin, Germany: Verlag von Julius Springer, p. 128.
Kane, L.E., Barrow, C.S., Alarie, Y. 1979. A short-term test to predict acceptable levels
of exposure to airborne sensory irritants. Am. Ind. Hyg. Assoc. J. 40:207-229.
Matt L (1889) Doctoral dissertation (in German). Wurzburg, Germany: Julieus
Maximillian University.
National Research Council (NRC) 1987. Committee on Toxicology. Emergency and
Continuous Exposure Limits for Selected Airborne Contaminants. Vol. 7. National
Academy Press, Washington, D.C.
Patty FA, ed. [1963] Industrial Hygiene and Toxicology, 2nd edition, revised. Vol. II
Toxicology. New York, NY Interscience Publishers, p. 851.
• Page '24
D-140
-------
COMMENTS FROM GEORGE V. ALEXEEFF, PH.D., D.A.B.T.
SPECIFIC COMMENTS ON THE USE OF THE PHOSGENE IDLH IN THE ACCIDENT ANALYSIS
The choice of the IDLH to evaluate the severe toxicity of phosgene may be
inappropriate. The IDLH for phosgene is provided in Table III-.1 as. 0.00081 g/m3, which
is equivalent to 2 ppm.
The best value available to use to evaluate serious irreversible health effects of
phosgene would appear to be the 1-hour NRC-EEGL of 0.2 ppm (0.8 mg/m3). The
basis of this level is Cameron and Foss (1941); Cameron et al. (1942). The guideline
exposure of 20 mice, 10 rats, 10 guinea pigs, 10 rabbits, 2 cats, and 2 goats to 0.2 ppm
phosgene for 5 hours per day for 5 days resulted in no deaths, and minimal pulmonary
edema in the majority of .the animals. In a small percentage of animals: 1 rat, 1 mouse,
1 rabbit, and 3 guinea pigs, massive pulmonary edema was noted. It was therefore
decided that a single 1-hour exposure of humans to 0.2 ppm phosgene would not cause
serious health effects. The AIHA-ERPG-2 level is also 0.2 ppm.
However, the NAS document includes a margin of safety of approximately 5,
since the 1-hour EEGL is based on a 5-hour exposure, and since the concentration/time
product for phosgene has been shown to use an exponent (n) of 1 for the equation
Cn * T = K (Rinehart and Hatch, 1964). However, additional safety factors for
extrapolation from animal data, for approximation of a NOAEL, and for consideration of
sensitive individuals, were not included to protect against pulmonary edema in humans.
This level may therefore underestimate the risk and has some uncertainty. It is
noteworthy that the results by Hatch et al. (1986), indicate the onset of pulmonary
edema in several laboratory species after a 4-hour exposure to 0.2 ppm phosgene and
was not considered in the development of the EEGL. It is also of interest to indicate that
the AIHA-ERPG-3, has also proposed that the maximum airborne concentration below
which it is believed that nearly all individuals could be exposed for up to one hour with
out experiencing or developing life-threatening health effects, is 1 ppm (4 mg/m3). That
is below the IDLH of 2 ppm. Based on the above information it would appear to be
_ • Page 25
D-141
-------
COMMENTS FROM GEORGE V. ALEXEEFF, PH.D., D.A.B.T.
appropriate to reevaluate the accidental impact of phosgene based on the NAS EEGL
(and AIHA ERPG-2) criterion.
REFERENCES FOR PHOSGENE
American Industrial Hygiene Association. 1989. Emergency Planning Guidelines for
Phosgene. Akron, OH.
Cameron, G.R., and Foss, G.L 1942. Effects of exposing different animals to a low
concentration of phosgene 1:1,000,000 (4 mg/m3) for 5 hours. Porton report No. 2349,
Washington, D.C.: British Defense Staff, British Embassy.. Cited in the NIOSH criteria for
a recommended standard for occupational exposure to phosgene. DHEW/PHS/CDC
[NIOSH] Pub. No. 76-137). Washington-, D.C.: U.S. Government Printing Office, 1976.
p.'52.
Cameron, G.R., Coutice, F.C., and Foss, G.L 1942. Effects of exposing different
animals to a low concentration of phosgene 1:1,000,000 (4 mg/m3) for 5 hours. Porton
report No. 2349, Washington, D.C.: British Defense Staff, British Embassy. Cited in the
NIOSH criteria for a recommended.standard for occupational exposure to phosgene.
DHEW/PHS/CDC [NIOSH] Pub. No. 76-137): Washington, D.C.: U.S. Government
Printing Office, 1976. p.52. !
Hatch, G.E., Slade, R., Stead, A.G., and Graham, J.A. 1986. Species comparison of
acute inhalation toxicity of ozone and phosgene. J. Toxicol. Environ. Health. 19:43-53.
Rinehart, W.E., and Hatch, T. 1964. Concentration-time product (CT) as an expression
of dose in sublethal exposures to phosgene. Ind. Hyg. J. 545-553.
8. In the accident analysis, IDLH (or 10 X LOC) values were used to determine the
downwind distances over which adverse human health effects might occur. To
evaluate the uncertainty introduced by using the IDLH, a sensitivity analysis was
conducted Where these distances were recalculated using the LOC ( a more
stringent health criteria). Other sources of uncertainty that are identified in the
accident analysis include concentration averaging times, chemical
concentrations, emission rates, and meteorological conditions. For most of
these parameters it is stated that conservative assumptions were used to avoid
underestimating risks. Have the uncertainties inherent in the accident analysis
been adequately characterized? For those parameters where sensitivity
• Page 26
D-142
-------
COMMENTS FROM GEORGE V. ALEXEEFF, PH.D., D.A.B.T.
analyses were not conducted, is the conclusion that conservative assumptions
have avoided underestimation valid?
Use of the LOG values for phosgene, formaldehyde and hydrogen chloride
appear to be suitable for the accident analysis. This is based on information provided in
response to question 7. The LOG values are very close to the levels suggested in the
response to question 7. Consequently, they would be expected to reflect the impact of
the accidents more appropriately than use of the IDLH.
The uncertainty analysis considers the implication of a shorter averaging time.
This may be a helpful method. Another method would be to consider the relative
exposure-dose based on a concentration17 x time metric. The value of n can be based
on empirical evidence where the value is 2 for formaldehyde and hydrogen chloride and
1 for phosgene (American Institute of the Chemical Engineers, Guidelines for Chemical
Process.Quantitative Risk Analysis, Center for Chemical Process Safety, New York,
NY,, p. 156). The exposure-dose comparisons may provide a helpful way of comparing
whether a slight change in the scenario is significant (Alexeeff, G., Lewis, D., and
Lipsett, M. (1992) Use of toxicity information in risk assessment for accidental releases
of toxic gases, J. Hazard. Mater. 29:387-403). This procedure would take into account
the increased importance of concentration relative to the contribution due to time for
formaldehyde and hydrogen chloride. The major limitation of this, and the suggested
method in the risk assessment, is that at some point exceedance of a certain
concentration may result in exceedance of an additional effect threshold.
• Page 27
D-143
-------
-------
Mary E. Davis
Risk Assessment for the Waste Technologies Industries Hazardous Waste Incinerator Facility
at East Liverpool, Ohio
General Comments
This risk assessment is extensive, in that it covers a wide variety of potential exposure
scenarios and develops risk characterizations for them. The uncertainties of the various elements
have been considered and are well documented. The organization is reasonable given the mass of
the material. Recommendations of the 1993 peer review workshop seem to have been followed.
Hazard Identification/Dose Response and Risk Characterization
1. Selection of surrogate compounds.
The procedure gave equal weight to emission rate, toxicity and bioaccumulation potential.
The basis for this is not clear. The three parameters seem to have similar, broad, ranges and so
equal weighting probably is reasonable.
There is still some confusion as to which chemicals have actually been detected in stack
emissions and which are predicted to be there based upon modeling. It would be less confusing if
potential emissions were labeled as such. For example, Table 1-1 would be more appropriately
labeled "Substances of Concern in Potential Stack Emissions."
2. Special consideration for dose-response evaluation of some chemicals.
The methods used to characterize cancer and non-cancer risks are very different and the
expression and interpretation of the results could lead to confusion. The methodologies for PAHs,
lead, mercury, nickel, chromium, acid gases and paniculate matter seem reasonable. The approach
for dioxins and furans is to not estimate non-cancer risks.
The non-cancer risks of dioxins and furans have not been characterized because EPA has
not yet determined the reference dose/concentration, at which no adverse effects are expected. In
the chapter on toxicity assessment (Will), the context suggests the reference dose has not been
determined because there is not agreement as to which effects are toxic and which effects are
ancillary, and not to be considered as toxic and therefore the reference dose cannot be determined.
There are data on dose-response relationship of specific toxic effects and a better solution would be
D-145
-------
Mary E. Davis
to estimate the risks of these specific adverse effects individually rather than not estimate non-
cancer risks because the "global" risk cannot be estimated. Reproductive/developmental toxicity is
one non-cancer toxicity for which there are good data, and for which there is serious concern.
There are data that suggest that reproductive effects occur at lower exposures or body burdens, so
that estimating the risk of cancer is not estimating the most sensitive outcome. The problem is the
perception that reproductive/developmental and immune toxic effects occur at doses lower than
those that cause cancer. Therefore, while the risk from cancer is in the range of 3.2X10"5 to
5.8X10"8 the other effects will occur at lesser exposures and therefore will be more prevalent.
3. Population and subpopulation selection.
Seems to be appropriate, in that it addressed both a "representative" or likely exposure and
an upper end exposure.
4. Appropriateness of risk estimate terms "average risk" and "maximum risk"
While reading the risk characterization chapter, it understood the meaning of the terms
"area average" and "maximum concentration" however, one could easily misconstrue the meaning
!
of "maximum concentration" in the tables if one did not read the text closely. Tables should have'
sufficient headings or footnotes so that one can discern the results being presented without having
to refer back to the text. My impression of the text is that the terms "average risk" and "maximum
risk" were used with qualifiers so that it was clear that maximum referred to an even smaller region
that, because of differences of dispersion, had higher chemical concentrations than the subarea asia
whole. I would not use the term "maximum risk" to refer to the risk associated with this exposure,
however, as that would be misleading.
5. Non-cancer risks adequately addressed?
The discussion of endocrine disrupters does explain why the risks cannot be characterized.
However, it is not clear why endocrine disruption has been considered, and which chemical(s) are
endocrine disrupters.
D-146
-------
Mary E. Davis
The non-cancer risks of dioxin have not been characterized, as discussed in #2, above.
Dioxin is a major concern with this facility and its' non-cancer risks should be addressed.
6. Additivity/synergy uncertainties
This was discussed only briefly in V/VHI. It would be useful to include discussion of what
is known about interactions among chemicals, particularly at .relatively low concentrations (such as
interactions mediated by changes in metabolism are unlikely because at low exposures, the amount
of enzyme does not limit the amount of metabolism and therefore formation of reactive
intermediates).
7. Identification of key assumptions
Overall, this seems to be well done. The tables are particularly useful.
8. Overall adequacy of risk characterization
The risk characterization as presented is as readable as I suspect such a document could
be. Much of the information is in tables and table titles, column headings and footnotes sometimes
could be more detailed. The question of non-cancer dioxin risks was not addressed as well as it
should have been and that undermines the adequacy of the risk characterization.
Accident Analysis
General comment:
The impact on East End Elementary School is a major concern for the accident analysis.
The use of population densities seems to hide the fact that an elementary school full of children is
within 1000 feet of the site. How many children are in the school? The truck route also goes by
Garfield School. What kind of school is this? How many students? Are deliveries expected during
D-147
-------
Mary E. Davis
school hours or afterwards? How much is the school exposed under the various meteorological
conditions, or under what conditions is the school more likely to be exposed?
The accident scenarios and chemicals selected seem to be reasonable choices for exploring
the effects of a range of accidents and conditions under which the accidents occur. lathe
discussion of the severity of occurrences, I would have iiiked explicit reference to the East End
Elementary School. I suggest that the discussion of the probability of occurrence include a more
detailed presentation of the rationale for the final probability result (combining the probability of
the accident occurring under the" specific meteorological conditions).
The sensitivity analysis of the IDLH vs. LOG was illuminating and disconcerting. It
presents a reasonable case for not using the IDLH for this type of analysis. The IDLH is designed
to protect healthy adult male workers from severe consequences. Children are likely to be more
sensitive. At elementary school age, many childhood asthmatics have not yet outgrown their
asthma. Also, elementary school children have colds and other respiratory infections more often
and so are less able to deal with additional respiratory challenges. Another component of the
IDLH is that the individual will leave the environment so that exposure is of a short duration. The
prospect of evacuating an elementary school is daunting, although I suspect a "shelter in place"
strategy would probably be the more reasonable response.
D-148
-------
Gasiewicz, t.A.
Comments on WTI Draft Final Risk Assessment (December, 1995)
Thomas A. Gasiewicz, Ph.D., Professor of Toxicology, Department of
Environmental Medicine, University of Rochester Medical Center, Rochester, New
York 14642
I have focused my attention on the following: Executive Summary (Volume
I), Facility Background (Volume II), Human Health Risk Assessment (Volume V),
Screening Ecological Risk Assessment (Volume VI), and Accident Analysis
(Volume VII). Most of my comments have been directed at these volumes. I
have also briefly reviewed Facility Emissions (Volume III) and Atmospheric
Dispersion and Deposition Modeling (Volume IV) in order to obtain general view
of how certain data and principles applied in the Human Health Risk and
Accident Analysis volumes were generated.
I. General Comments:
1. Overall the risk assessment appeared to be well organized and presented in
a logical format. I liked the idea of presenting the detailed discussion of
certain items and the bulk of the data as Appendices, while focusing on the
main and important .thrust of the particular volumes within individual
chapters. Nevertheless, due to the (mostly) thorough nature of the
document and the number of volumes and pages necessary to contain all of
this information, I found the document at times difficult to wade through,
especially when searching for particular information. Better cross-
referencing would have been useful. For example, when a point is made
about a particular piece of data, the page number .and or Appendix page
number would have been useful beyond just giving the Volume or Chapter
number. In most cases the text was well written and concise. However,
each chapter, for the most part, was very thorough and there was enough
redundance of explanation and information to allow each to stand on its
D-149
-------
Gasiewicz, T.A.
own while referring back to other sections for specific details .and data.
2. Executive summary: For the most part the Executive Summary appears to
reflect the data, approaches used, and conclusions derived. It is
emphasized often, and appropriately, that in most cases conservative
approaches and assumptions were used so that the data likely represents
an overestimate of the risk.
3. There were several major modifications suggested by the 1993 workshop.
The recommendation was made for additional performance tests to
develop more reliable estimates of emissions have been performed. These
have added to the reliability and accuracy of the assessment.
Within the context of the Human Health Exposure Assessment, the 1993
workshop also recommended an updating of food consumption data and
the inclusion and/or consideration of other exposed population groups.
These have been adequately addressed and/or considered by the present
document.
The workshop suggested more consideration be given to uncertainties
and variability. In the present document, much additional effort was made
in this document for assessing uncertainty and variability, in particular which
parameters and/or measurements were most likely to be highly variable
and the likely degree of variability. There were also many factors which
contributed to the uncertainly of the either the assumptions being made o r
i
the data being used. These were, for the most part, also appropriately
discussed and/or documented. Some exceptions are discussed in my
specific comments below.
A recommendation was made to consider in more detail upset
conditions-, fugitive emissions, and accidents. For the most part this has
been adequately addressed. Some further considerations are detailed
below. -
For particular chemicals, the physical and chemical form of several of the
D-150
-------
Gasiewicz, T.A.
metals was identified as influencing transport. Where appropriate, and
where data was available, this has been considered by the present
document. In many of these cases, only limited data was available, and the
conservative, but not totally unrealistic, approach was applied.
The workshop recommended that an ecological risk assessment be
conducted. A laudable, and what appears to be a reasonably thorough,
attempt was made at this risk assessment. However, I have made a
number of specific recommendations below.
The consideration of several additional compounds was recommended
by the workshop. In some cases, the contributions to the total risk was
minimal. However, for others, especially for certain polycyclic aromatic
hydrocarbons, the contributed risk was considerable. Thus, this turned out
to be a valuable suggestion by the previous workshop.
The workshop also recommended a consideration of the additive and/or
synergistic effects. Although the present document has considered these
interactions, very little and specific data is available that could be
reasonably used for this risk assessment. The method of adding risks for
these compounds is actually one of the difficulties which, in my opinion, has
not been adequately discussed. My specific comments are noted below.
It would have been useful for a statement to be made if in fact all the
recommendations by the workshop were considered in this document. It is
indicated on p. IV-1 that "..a concerted attempt has been made to
incorporate the recommendations provided by the Peer Review Panel." It
would have been useful to indicate, perhaps in a Table, what these
recommendations were, -if or if not these recommendations were
incorporated, why or why not they were incorporated, and if incorporated,
what was the overall effect on the risk assessment process. From what
has been presented, the reviewer does not know if there were some that
the present document did not consider? This should have been more
D-151
-------
Gasiewicz, T.A.
specifically, and likely very simply, addressed.
4. As indicated, "..there are always additional data and method development
efforts that could be undertaken to reduce the level of uncertainty".. The
question has been put forth of whether there are serious data 'or
methodological gaps in this particular assessment that would preclude its
use.in a decision-making process. Much of the risk assessment described h
this document depends on the.accuracy and reliability of the models used
for- predicting concentrations of particular chemicals in environmental
receptors, i.e. soil, foodstuffs, water, and in human tissues exposed to
these chemicals either directly or indirectly. These predicted concentrations
are given in some of the Appendices of Volume V. Realizing that the
concentrations used for this risk assessment are those predicted to be
contributed by the WTI facility, it would have been extremely useful for an
additional section (in the uncertainty analysis ?) to compare these
concentrations to present, i.e. "background" levels, in environmental
receptors and human tissues. Certainly, there is enough literature already
available to indicate what these "background" levels might be. If data were
available from the locality under consideration; so much the better. The
lack of this comparison, in my opinion, decreases the reviewers confidence in
the models being used, despite their apparent theoretical goodness. One
would predict that the numbers generated from the models would be
substantially less, but within at least 1 to 2 orders of magnitude, than the
determined background levels. If the predicted numbers happened to be
substantially lower or any higher, then it would be apparent that something
is wrong with the particular model, the assumptions made, and/or the data
used. Again, if nothing else, this comparison would have increased
confidence in the models being used. No such comparison was apparent h
this document, and I would recommend that such be made for each
chemical of concern prior to the use of this risk assessment for any decision
D-152
-------
Gasiewicz, T.A.
making process.
For the most part very conservative assumptions have been made in
this particular risk assessment process. One of the most conservative
assumptions is that the risk is additive for all chemical exposures. With
some exceptions, i.e. the dioxins and dibenzofurans, for which there is
reasonably good data, this is probably an unreasonable assumption and for
which there is little or no data. I would recommend that any the risk be
based on the most hazardous chemical group, e.g. the dioxins, without the
consideration of additivity from the other chemicals unless reasonably good
data were available.
5. Long-term research: This reviewer would recommend that more effort be
made to gather real world numbers in terms of concentrations in
environmental receptors and the contributions of various sources to these
concentrations. Although, as indicated above, the models, in most cases
seem theoretically appropriate, real world data would be much better not
only to estimate risks more accurately but to test the models. Thus, more
research should be made into testing the models developed and their
parameters under a real world situation.
II. Comments on Volume I, Executive Summary:
1. General Comments: At least for the Human Risk Assessment and Accident
Analysis volumes, the Executive Summary is an accurate reflection of these
both in specific data and conclusions from the data. Importantly, it has
also been emphasized that in general very conservative assumptions have
been used, and thus the predicted risks are more likely to be overestimated
than underestimated.
2. p. II-3 and elsewhere throughout this risk analysis: Since this is a scientific
analysis, the temperatures should be presented as degrees C, rather than
degrees F.
D-153
-------
Gasiewicz, T.A.
3. pp. 111-1 through III-3 discuss Incinerator Stack Emissions. If not already
done in the specific volume where this is discussed more thoroughly,
somewhere in the document it should be indicated how efficiency of
combustion changes over time and usage, how often 'maintenence
procedures are performed and how these procedures (or the lack of such
procedures) are likely to change the combustion efficiencies and thus the
emissions.
4. p. IV-4: It is indicated that the use of the CALPUFFand ISGCOMPDEP
models indicates that "..the inclusion of calm wind dispersion and fumigation
does not have a significant effect on the peak predicted concentrations
from the WTI incinerator stack". This is an important conclusion, but seems
counter-intuitive. This might be explained in more detail here. Are there any
real data to'back up the suggestions of the models? Here again, as noted
above, although the models are undoubtedly useful, it seems necessary t o
consider real data and compare the predicted to actually observed data
whenever possible. For such important conclusions, if no real-world data is
available a statement should be made indicating such.
5. p. V-3, Fugative Emissions: Here is should be mentioned if there should be
any consideration of groundwater contamination and why or why not. In
addition, although this might have been considered elsewhere, what are the
concerns, if any, to the workers at the WTI site? Are workers who may
also be highly exposed by residence or life-style an additional population to
consider?
6. p. V-10: Here for "Fugative Emissions" it should be stated specifically for
which subgroup the risk estimates are given.
7. V-11:- Regardless of the very conservative assumption of additivity of risks
and hazard quotients across all exposure pathways, it seems inappropriate
do this simply since there is no evidence to indicate (with the exception of
i
the dioxin-like compounds) that the risks are additive. This seems to be an
D-154
-------
Gasiewicz, T.A.
overly conservative assumption. If there is any basis for such an
assumption is should be clearly explained.
III. Comments on Volume II, Introduction:
1. p. II-2: Although this reviewer realizes that this is not the central focus of
these risk assessment document, there is some concern likely to be
generated (and there should be some real concern) that regardless of the
risk estimates from the WTI facility alone, the facility may be contributing t o
an existing problem for the population of the area due to the presence of
other industries and other sources of area pollution. This is especially true
since there are a number of coal-fired power plants in the area. For
example, it would be useful to have some data on the relative levels of
dioxin-like compounds and PAHs in cattle, milk, and fish in the area. In
addition, environmental conditions, especially during inversions, may allow
very high air levels of a variety of air pollutants. These points should be
addressed, even if only in a general sense, somewhere in this document.
2. p. II-4, line 8: The phrase "..if appropriate.." should be briefly explained here.
3. p. II-5: It should be specified what happens to the sand and activated
carbon that is used to treat the contaminated water. Where does this go?
Likewise on p. II-7, it should be indicate where the collected fly ash is taken
for "..treatment and disposal..".
IV. Comments on Volume V, Human Health Risk Assessment:
Comments Re "Charge to Reviewers"
1. General Comments: Well organized and documented. Methodology and
approaches are appropriate. The choice of surrogates is well rationalized.
The food consumption data has been updated as requested by the 1993
review. In addition, the document has now considered the PAHs - this is an
important addition to the document considering the relative contribution
D-155
-------
Gasiewicz, T.A.
these compounds have to the total risk assessment. Fugative emmission
are considered in more detail, although nothing is mentioned of any potential
groundwater . contamination. Additive and synergistic effects are
considered, although little can be done since little data is available.
2. Characterization of the exposure .estimates in terms of "central tendency",
•"high end", and "bounding" exposures: The terms of "central tendency" and
"high end" exposures are certainly well defined at the beginning of Chapter
VII on Estimation-of Exposure Dose. However, throughout the text the
terms of'"average exposure" and "maximal exposure" are used in their
places. This seems appropriate and clearly understandable. The term
"bounding exposure" is rarely used elsewhere in the text. In my opinion, the
risk characterization for a "bounding exposure" seems inappropriate and
not needed.
3. Procedure for combining environmental media concentration, intake rate,
and duration •/ frequency of exposure to develop estimates of exposure:
This procedure was well described, and in most cases was based on sound
and documented rationale. Where assumptions were made they appeared
to be clearly explained with much consideration of uncertainty analysis. In
fact, it took a great deal of effort to wade through the documentation and
explanation of models, etc. because it was done very thoroughly.
4. Identification of Important Sources and Pathway Routes: Very thorough
and clear approaches. All of the significant pathways of exposure
appeared to be identified. With one exception all of the significantly exposed
subgroups appeared to have been identified. One subgroup that was not
considered was WTI workers who may live in the area. Should this .group
have been considered?
5. Key assumptions for estimation of chemical concentrations and estimation
of exposure: Most of the assumptions are conservative and would tend to
overestimate these.concentrations. Where there is a possibility of an
D-156
-------
Gasiewicz, T.A.
underestimate due to an assumption the magnitude of the effect tended to
be low. An exception might be the assumption that inorganic compounds
do not accumulate in mother's milk. This should not be assumed as milk
would be a significant route of exposure especially for compounds such as
methylmercury. If this assumption is made it should be backed up with
referenced data most of which likely exists. Further literature searching is
needed here. In fact, this relates to my main criticism of the document -
lack of referencing real world data for which there is likely much available.
This referencing would allow us to have greater confidence in the models
being used.
6. Uncertainties: This risk assessment process is full of uncertainties and I
think that the document does a reasonable job in confronting these and
appropriately pointing out what they are, why they are uncertainties (i.e.
lack of data, or variation in day-to-day environmental conditions, etc.), how
they might affect the process (either an underestimate or overestimate),
and what the general magnitude of the effect might be. In most cases a
Table is presented at the end of each chapter to identify these uncertainties
and indicate their nature and possible affect on. the risk assessment.
7. Selection of surrogate compounds: The chosen ranking is appropriate.
8. Dose-response evaluation for chemicals: The dose-response evaluation for
the dioxins/furans is based primarily on TEFs. This is appropriate given the
mechanistic data, albeit the uncertainties surrounding the values for the
TEFs. The PAHs are given relative potencies from zero to one, with
benzo(a)pyrene given a potency of 1.0. As an interim approach, this seems
appropriate. The lead data is based on a model predicting blood levels ri
children. Given the greater sensitivity of children, this is entirely appropriate.
The considerations for mercury are based on its inorganic and organic
forms. Given the inportance of the organic forms and the documented
. recommendations, these considerations are appropriate. There are also
D-157
-------
Gasiewjcz, T.A. ;
forms' of nickel that may or may ngt be carcinogenic. The assumptions for
nickel that it ail exists as an inhalation carcinogen are very conservative, but
nevertheless, appropriate for this case, Similarly the assumption for
chromium is very conservative. For acid gases and particulate matter the
NAAQS values are used. This is appropriate.
9. Identification of subpopulations: In many cases, the use of non-site data
was appropriate given the lack of good data and the cost and time of
obtaining good data. In other cases, such as for subsistence farmers and
the consumption of local meat some site specific data, although some more
formalized than others, were used to identify subpopulations at risk and
estimate exposure rates. This is actually better than what I would have
expected and was glad to see some consideration of local data, although it
is hot clear how accurate these data are. Nevertheless, it provides sorne
directions for estimation of these parameters.
10. Average vs maximum risks: The term "average risk" is appropriate given
the nature of the paramaters, i.e. average emission rates, average air
dispersion/deposition, and "typical" exposure factors, There might be some
consideration as to whether the "maximum" risk should be based on
maximal emission rates. This would be a conservative approach - and this
reviewer thinks actually too conservative. Given the facility operations, it is
highly unlikely that maximal emission rates will ever be approached. Thus,
the use of average emission rates for maximum risks seems appropriate,
also given the very conservative assumption of maximum air
concentrations. :
11. Non-cancer risks: It is this reviewers opinion that given the available data,
these risks have been adequately addressed. Given the paucity of data in
most cases, the discussion of "endocrine disrupters" would have been
inappropriate here. Nevertheless, it is an area of concern that should be
addressed at later dates once more specific data is available. For the
D-158
-------
Gasiewicz, T.A.
dioxins/furans, the numbers used for the cancer risk are likely overly
conservative and, at least based on the available data, would likely protect
against the non-cancer risks from these compounds.
12. Additivity and synergy: Given the data available, the discussion is
appropriate. As indicated elsewhere in my review, it is my opinion that the
additivity of risks for all compounds is very very conservative. This should
have been given more discussion.
13. Assumptions for estimation of dose and risk: As noted above the
assumption of additivity for individual chemicals (p.VIII-55, Volume V) is
certainly a very, very conservative assumption. This should have been given
• more discussion to point this out very clearly. Given the paucity of available
data the noncancer health risks due to infant ingestion of breast milk are
not considered. This should be given more discussion given the noted
effects of some of the metals, i.e. lead, methylrnercury, and even dioxins
(some recent data) on learning and behavior.
14. Overall adequacy of risk characterization: With the exception of the
above - seemingly adequate.
Specific Comments
1. UI-5, 1st sentence of first full paragraph: This is not quite true since certain
other congeners which do not have chlorines in the 2, 3, 7, and 8 positions
also have dioxin-like activity. However, the potency of most of these is
very, very low on a relative basis. This reviewer would recommend a
qualification of something like "...as displaying dioxin-like activity of
significant potency".
2. p. III-6, lineS from bottom: Eliminate the term "hydroxylase" here. Also the
next sentence add "Based partially upon differences..." since the receptor
binding data was not the exclusive data upon which the TEFs have been
developed.
3. p. IV-2: Here it is indicated that 31 compounds were not evaluated in this
D-159
-------
Gasiewicz, T.A.
assessment because emission rates were not available. Somewhere in the
document a table should show what these compounds are.
4. Chapter VI, Estimation of Environmental Concentrations: As noted in 1.4
above, some comparison of the predicted WTI-contributed concentrations
to real world concentrations would be useful. In my opinion, this is a very
important deficiency of this document.
5. p. VII-18, 1st sentence: For completeness some brief explanation of how it
was determined that "..plant uptake of vapor-phase dioxin/furans is the
primary contributor to total risk" is needed here. It simply could be
indicated that consumption of plants is the major contribution to the body
burden of dioxin/furans, in cattte. (See p. IX-1- in Volume V as well.)
6. p. VII-18, 4th line from bottom of .2nd paragraph: Shouldn't this be
Appendix IV-4 instead of IV-1?
7. p. VIII-1, 1st sentence: In Chapter VII only the dose estimates for the adult
subsistence farmer in subarea E1 are presented. Are the data for the
other population supgroups somewhere else?
8. p. VIII-2, B, Estimation of Risks Due to Stack Emissions: For the
noncarcinogenic effects it would be useful to list for each compound what
the particular noncarcinogenic effect was. This might either strengthen or
weaken an argument for assuming additivity of risks. Likewise for
carcinogenic effects, it should be indicated whether a particular type of
tumor or total tumors are being considered.
9. p. ,VIII-8, c, Potential.Health Effects..,,and Particulate Matter: For
completeness, it should be mentioned whether or not the maximum
predicted concentrations are during a period of inversions or other
environmental conditions. Again, it would also be useful to know the real
world concentrations of these pollutants in the particular area of interest.
10. p. VIII-2, I, Estimated Risks Due to Fugitive Organic Vapor Emissions: Data
is given for the lifetime cancer risks due to average exposure. What about
D-160
-------
Gasiewicz, T.A.
maximum exposure? (Are these numbers a concern for workers at the
plant who might live close to the facility??)
11. pp. IX-1 and IX-2: Here again, some comparison with real data would
assist in the uncertainty analysis.
12. p. X-3, Cancer Risks, 1st paragraph: There should be some clarification
here to distinguish more specifically the average total cancer risk for each
of the subpopulation groups vs the highest cancer risk for the subsistence
farmer. For the latter, shouldn't this be "the cancer risk for maximal (or
highest end) exposure"? The way is is presently worded it sounds like of
the subpopulation groups the subsistence farmer has the greatest risk for
the average total cancer risk.
13. Appendix V-5, p. 2, Individuals Who Work at WTI: It is stated that
"evaluation of worker exposure ....is beyond the scope of the WTI Risk
Assessment". What if those same workers also live in the area that is
predicted to be the most heavily contaminated? It would seem that this is
a population that should be considered. If indeed this is beyond the scope
.of the Risk Assessment, specific reasoning and rationale should be given.
14. Appendix V-8, p. 1, bottom 7 lines: It is not clear why only 6 year
exposure is used for chilhood exposures. Shouldn't this be extended to
cover the age until the end of high school. Is this inconsistent with what is
stated on the top of p. 12?
15. Appendix V-8, p. 40, Table 17: It is not at all clear why 1) (as noted
above) ED is 6 years for both the child resident and school-age children, 2)
the assumed BW for these groups are different, and 3) the LT value is same
here across all subgroups". This is not obvious here and should either be
explained or cross-referenced to somewhere else in text where it is
explained.
D-161
-------
Gasiewicz, T.A.
V. Comments on Volume VI, Screening Ecological Risk Assessment:
1. p. I-2, bottom 8 lines: The differences among SERA, PERA, and DERA
should be explained more thoroughly in terms of the data base used, the
assumptions made, etc. This is especially important since the present SERA
indicate possible risks for certain chemicals that would have to be
considered in more detail by the PERA arid DERA.
2. p. I-6, line 5 from top: It has not been explained why a 30-year
accumulation of the chemicals has been assumed.
3. p. 1^9, Indicator Species: Based on the chemicals of interest, would it not
be good to choose a most sensitive species? The indicator species chosen,
although perhaps representative of certain groups, may not truly cover a
most sensitive species. Data should be available, again based on the
chemicals of interest) for a most sensitive species for each chemical. If
these species are within the area.under consideration than this most
sensitive species should be represented.
4. IV. Identification of Ecological Chemicals of Concern: Is there any
consideration of the possibility that the existing levels of certain chemicals
may already be high in the area of concern due to the amount of
industrialization present? Will the WTI facility contribute to a problem that
already exists? This should be addressed somewhere in the document. As
this reviewer noted for other volumes, some data on already existing levels
of certain chemicals already existing in the area of concern would be useful.
5. p. IV-4, B, "Development..., last 2 sentences of 1st paragraph: Despite the
discussion, it is not clear to this reviweir why the emission rates used in the
SERA differ from that used in the HHRA. The discussion should be more
precise - but obviously brief.
6. p. IV-11, 1st paragraph: For the aquatic organisms, the discussion should
mention if all stages of development are considered vs just adult animals.
For example, it is well known that developing fish are extremely sensitive to
D-162
-------
Gasiewicz, T.A.
dioxins/furans. I believe that the endpoints examined (in Appendix VI-22) do
cover fish embryo, eggs etc, but this should be mentioned here as well.
7. V, Characterization of Exposure: Here it is indicated that "the potential for
adverse effects to ecological receptors....is a function of (2) the
concentrations of chemicals in the media to which the receptor is likely to be
exposed...". Without a consideration of what is already there from other
contributing sources, this analysis might be considered somewhat useless.
It would be worthwhile to have some real world data regarding the
concentrations of some of these chemicals that may already be present in
the particular environment. This data might also help to have some
confidence in the modeling that is being used to estimate exposures!
8. p. V-61 and in other tables in this chapter: For the total dioxin/furans - is
this as TEQs?
VI. Comments on VII, Accident Analysis:
Comments Re "Charge to Reviewers"
1. Selection of Scenarios: A number of scenarios were initially considered as
potential accident scenarios. These included the ones finally selected for
analysis as well as others including, for example, failure of air pollution
control equipment. A number of criteria were used to make the final
selections. These were well justified and had good rationale. The final
selections.were appropriate.
2, Selection of Chemicals: The selection of chemicals was based on the
substances handled at the WTI facility, acute toxicity indices, volatility, and
estimated maximum concentrations in waste.These criteria were
appropriate. It might have been useful to consider persistence as an
additional criteria for off-site spills, especially where ground and/or surface
water might be contaminated.
3. Chemical Release Rates: With two exceptions the methodology used and
D-163
-------
Gasiewicz, T.A.
assumptions made seemed to be appropriate and mainly conservative. The
exceptions are the assumptions surrounding ambient temperatures and
roadway sites. In both cases worst-case, but nevertheless real,
parameters should have been used.
4. Atmospheric Dispersion Modeling: The selections appeared to be well-
justified and appropriate. However, this is somewhat beyond the limits of
this reviewers expertise. .
5. Severity of Consequences and Probability of Occurrence: Based on the use
of the IDLH values and the assumptions used to determine release rates,
the categorization of severity and probability appear to be correct.
Notably, the probabilities are likely an overestimate based on some
historical data presented.
6. Key Assumptions: I have commented on the use of IDLH and the roadway
sites above and below. The other assumptions seem reasonable, but not
without a significant amount of uncertainty. For example, the 30-minute
exposure assumption may not necessarily be the worst-case assumption.
This might be altered since it may have a significant effect on the ranking of
events. The others are fairly conservative assumptions.
t s
7. Use of IDLH Values: f have commented on these above an below. There is
no specific rational presented for using 10 x the LOG values for chemicals
for which IDLH values have not been established. This seems to be even
more reason for using the LOG values for the analyses.
8. Uncertainties: For the most part the uncertainties appeared to adequately
addressed. As noted in my other comments, there are other areas where,
in my opinion, the best conservative assumptions have not been made to
avoid underestimation.
Specific Comments:
1. p. I-5: The use of IDLH values seems to be inappropriate in as much as
these values, as the document states, "..were originally developed to be
D-164
-------
Gasiewicz, T.A.
protective for healthy adult male workers.." and have not been adjusted to
account for heterogeneous populations. I would have recommended strictly
using the LOG values. Nevertheless, a "sensitivity analysis" was performed
using the LOG values.
2. II, Accident Scenarios: Good rationale has been used for the scenarios used
here. These appear to likely account for most, if not all, of the expected
(predicted) accidents. Indeed, some historical data was presented to
indicate that this would be the case. Fairly conservative assumptions have
been made.
3. HI, Chemicals of Potential Concern: The selection of chemicals is well
documented and based on sound and appropriate criteria. This is based on
part on the use of the WTI facility.
4. IV,1 Characterization of Accidental Releases: Here although the assumptions
are mainly conservative, for some parameters only the average and not.
worst-case values are used. For example, 68°F is used as the ambient
temperature. In fact temperatures have been documented in the 80s.
Depending on the scenario and particular chemical in question the higher
temperature should have been used as a more conservative and worst, but
real, case.
D-165
-------
-------
Ecological Risk Assessment
D-167
-------
-------
P.L.deFur
WTI Review — SERA — December 1995
Peter L. deFur
Environmental Defense Fund
Overall issues from reviewing the Screening Ecological Risk
Assessment (SERA)
1. The exposure data are limited by the paucity of real data from
a range and variety of real operating conditions. Granted that the
facility has only been operating for a limited period of time, but
the reliance on a small number of test burns under fairly known and
controlled conditions is problematic. Some correction needs to be
made for upset conditions, emergency shut-downs, start-ups, and
other non-normal operating conditions that would increase or change
the composition of the emissions.
2. The SERA does not seem to take into account the contribution of
the facility to continued degradation or prevention of restoration
and recovery of sites (habitats) within the impact region.
Specifically, the SERA notes that bald eagles do not occur within
the site,, but does not consider the contribution of the facility to
preventing bald eagles from returning. Other species, both
terrestrial and aquatic may be similarly affected.
3. It is not clear that the SERA considers the combined effects of
all of the chemicals released, or even all the chemicals selected
for analysis.
4. The SERA explains that a 20 km was selected for analysis, but
does not explain the deference that might be expected if the
largest region (50 km) had been selected.
5. The analysis uses arithmetic means in a number of places, e.g.
D-169
-------
P.L.deFur
of the test burn data, yet no analyses are mentioned that other
means (e.g. harmonic, geometric) are not as accurate in
representing the data.
6. The emissions estimate cannot take into account chemicals for
which there are no quantitative estimates, understandably. But is
this not an omission that undermines the validity of the outcome?
7. The SERA does not account for the process used here. Were
outside parties used to suggest approaches, data, species, end-
points, etc. Recognizing that this represents but one part of a
multi-year and multi-stage effort, the process for including
interested and affected parties is still a necessary one in the
SERA. Where is it?
Answers to Reviewers Questions:
1. Does anything undermine the scientific validity of the SERA?
The estimates of emissions rates, including the omission of
chemicals is undermining. Are there other facilities similar to
this one the could be used to provide surrogate data for
comparison?
2. Is the organization clear and does it follow the Framework?
Yes, the organization is fine and this SERA does seem to follow the
Framework quite well.
3. Are all the uncertainties included?
To the extent that chemical interactions are uncertain, the SERA
does miss some important ones. The uncertainty of emissions is at
least discussed.
4. The weakest and strongest points in the SERA; The weakest points
in the SERA are the emission estimates and -the unknown animal data.
The former are addressed above; the latter will remain unknown.
5. The major elements in section II seem adequate. The 5 scenarios
for emissions are not convincing in terms of completeness. I am
left wondering about other ash emissions and about total loadings
D-170
-------
P.L.deFur
to tlie environment from this facility via all routes, both
intentional and accidental. The SERA does not account for the
unexpected.
6* The site characterization seems adequate, but not overly so. I
do not get a good sense of the interactions within the area as a
watershed. What type of water flow drains across the land? How
much flooding is there? Are there numerous diverse habitat niches?
7. The screening method in section IV seems to be adequate, but I
have to see where the question "What did we miss?" is answered.
The SERA cannot capture everything, therefore it must include a
section that looks for the missing parts. It is not clear that the
tiered approach does provide a "thorough screening-level
evaluation." The approach may be more accurate termed
"representative" than thorough. The approach is one based on
indicators and such is limited.
8. Are the exposure and effects adequately characterized? The
characterization may be adequate, but key elements are not obvious.
The species selection, section V, indicates only part of the
explanation for choice. Do these species represent most or some of
the total number of individuals that are resident? Why is the 1 km
distance used here, when the site was described as 20 km?
9. Section IX is adequate in describing the uncertainties, but it
is not a complete risk characterization. A more complete risk
characterization needs to address some risk estimation questions,
process questions and the issues raised by interested and affected
parties.
10. Does the SERA not underestimate risk?
There are several omissions that raise the possibility that risk is
actually greater than indicated. First, the chemical by chemical
approach does not include the interactions among chemicals,
especially metals and organics. Second, if one of the chemicals
not included in the SERA has a high impact, then is not the risk
underestimated? Third, the exclusion of amphibians does not offer
D-171
-------
P.L.deFur
evidence that this group will be unaffected. Organics washed from
surrounding areas into numerous small wetland areas would be an
effective exposure pathway should be investigated.
D-172
-------
Pirn Kosalwat
THE SCREENING ECOLOGICAL RISK ASSESSMENT
FOR THE WASTE TECHNOLOGIES INDUSTRIES (WTO
HAZARDOUS WASTE INCINERATOR FACILITY
1. Are there any components of the SERA which you feel undermine the scientific
validity of the assessment? If so, what are they and can you provide suggestions
to strengthen the identified components?
No, I think the. SERA was well conducted and scientifically defensible.
2. Is the organization of the document clear and does it present the material in a
clear and concise manner consistent with the Framework for Ecological Risk
Assessment (EPA, 1992)?
Yes, the organization of the document is clear and follows the EPA's Framework for
Ecological Risk Assessment. However, the information presented is too repetitive
which makes the report long and tedious to read; and at times, creates confusion.
3. Uncertainties are discussed in numerous sections of the SERA and compose
Section VIII of the SERA. In each case, do these discussions cover all relevant
and important aspects of the uncertainties which you think should be addressed
in the SERA?
Yes, uncertainty analyses presented cover all relevant and important aspects that
should be addressed. When uncertainties occur, the most conservative assumptions
are usually chosen.
4. In your opinion, what is the weakest and what is the strongest aspect of the
SERA? Can you make any suggestions on how the weakest parts can be
strengthened by the Agency?
D-173
-------
Pirn Kosalwat
The weakest aspect of the SERA is the data gap, particularly the toxicity values used
to derive the benchmark values. To strengthen this aspect, the Agency should ensure
that all data bases have been exhausted and if feasible, fund toxicity testing for
chemicals that are deemed critical for this risk assessment.
The strongest aspect is the conservative approach to ensure, that the ecological risks
from the facility will not be underestimated.
5. In Section n, are the stressors, ecological effects, and both the assessment and
measurement endpoints adequately characterized? Are the five emission
scenarios adequate to characterize the exposure for the WTI facility? Are there
other emission scenarios which you think slhould.be included in the SERA?
Yes, they are adequately characterized. The scenarios chosen would probably
overestimate the risks and adequately protect the ecosystem hi the area.
6. In Section m, is the site characterization adequate to support the SERA? Why
or why not?
Yes, except for the land use statistics which contain 15-year-old data, the rest of the
site characterization is considered adequate to support the SERA. This section has
enough details on physical and biological descriptions of the area for an ecological :
risk assessment.
7. In Section IV, is the tiered process used to identify the ecological chemicals of
concern (ECOC) from the initial list of potential chemicals considered
scientifically defensible?
Yes, the tiered process used is appropriate for a pool of potential chemicals this size.
D-174
-------
Pirn Kosalwat
Does application of this tiered approach support the statement made in the SERA
"by focusing on the potential risk from the selected ECOCs, the SERA provides a
thorough screening-level evaluation for the WTI facility?"
Yes.
8. In Sections V and VI, are the exposure and ecological effects adequately
characterized? Are the most appropriate estimation techniques available used?
Are the assumptions clearly stated?
All exposure and ecological effects are adequately characterized using appropriate
estimation techniques with clear assumptions. However, I do not feel comfortable
with some of the uncertainty factors (UF) used to extrapolate toxicity values in Table
VI-1. In particular, an interspecies uncertainty factor of one is used if NOAEL values
are available for three or more species within a class. Since species' responses to
chemicals are highly variable, using one as the UF to extrapolate toxicity values
among species seems inappropriate; an uncertainty factor of at least 5 should be used.
9. In Section VIII (sic), are there any major elements missing from the risk
characterization which you think need to be included or which would strengthen
the risk characterization?
In Section VII (Risk Characterization), it would be interesting to determine cumulative
risks from exposure to all selected ecological chemicals of concern (ECOCs) for each
scenario for each indicator (e.g., by adding hazard quotients for all chemicals for
animal).
Does the risk characterization support the summary and conclusions presented in
Section IX?
Yes.
D-175
-------
Pirn Kosalwat
10. In Section IX, given the assumptions made and the processes used to select and
evaluate chemicals, receptors* and exposure pathways, do you think the SERA
adequately met its objective of not inadvertently underestimating risk?
Yes; in my opinion, this SERA is more likely to overestimate the risk.
Other Comments:
As stated in Comment No. 2, the information presented in this SERA report is too
repetitive. For example, in Section IV, the three paragraphs under Toxicity on pages
IV-23 and IV-24 are identical to the three paragraphs under the same heading on
pages IV-10 and IV-11. The report can probably be condensed to about two-third of
the current size without losing any crucial information. Also, there are a few
grammatical errors in the report. For example, the word "cannot" should be written
as one word, not the two words (i.e., can not) which appear throughout the report.
D-176
-------
Steven C. Peterson.
Technical Workshop on WTI Incinerator Risk Issues
Washington, D.C.January 11-12,1996
Draft Review Comments on Volume VI -Screening Ecological Risk Assessment (SERA)
General Issue*
1. Overall Organization
With exceptions as noted, the draft SERA is generally well written and logically organized,
if somewhat lengthy and cumbersome. The document could be streamlined by (a) transfcirmg some
of the detailed tables and discussion of memodotogy to appendices. and (b) attempting (in the
problem formulation phase) to identify a murower list of potential concerns, thereby focusing the risk .
assessment on the issues of greatest significance. Examples of tables that could be placed in
appendices include the listings of locations of state paries and other areas provided in Section II; log
Kp, and persistence values in Section IV; inodeled concentrations in Section V; and summaries of
eSects in Section VL In addition, some of the detailed discussion of methodology, such as the
inant of CPOCfffl g5r?g"'ng Tflg^bodS *" -S^cri"" TV afl(j
-------
Steven C. Peterson
than background risks. This might lead to a faulty interpretation of the significance of potential
risks, and preclude the use of the SERA for decision making. Further comments on background
comparisons'arc provided below in remarks on Specific Issues.
5. Long-term Research Needs
A significant problem faced by ecological risk assessors at WTI and other sites is the lack of
comprehensive guidance.' The available guidance is too general and does not provide detailed
methodology. Consequently, WIT is similar to most ecological risk assessments in the "recycling" of
information from other government reports and risk assessments, many of which have not undergone
serious peer review. In my view, two of the critical area!, of research in the development of guidance
are (1) further development, compilation and evaluation of toxicily benchmarks for wildlife and other
useful values such as BAFs (bioaccutnulation factors); and (2) development of methods for
extrapolating from potential effects on individual organism* to the prediction of population and
community-level effects. <
Specific, |s, snes
1. Scientific Validity of the SERA
There arc many areas of contention over approaches for evaluating ecological risks of
yhemfeyig jji the environment. For example, risks to populations *n<^ communities frcouentiy arc not
directly addressed, yet these are the levels of biological organisation of fundamental importance in
ecosystems. However, the SERA is no more remiss in this regard than hundreds of other ecological
risk assessments conducted for regulatory purposes. In, fact, the level of scientific rigor in the SERA
is above average, in my experience. While I have a number of specific technical concerns with
various aspects of the SERA, as indicated in the following rernmentc none of mem seriously
undermines the scientific credibility of the report.
2. Consistency whh EPA Framework
While the SERA is generally consistent with the organization and process of ecological risk
assessment as presented in the EPA Framework for Ecological Risk Assessment, the framework
eallg for selection of cndpoints and development of a conceptual model based on identification of
D-178
-------
Steven C. Peterson
strcssors and ecosystems of concern. Hence, these should be presented at the conclusion of the
problem formulation and should be drawn more directly from the information presented in Sections
HI and IV (see comments on Sections H through V, below).
Besides hs adherence to national guidance^ the SERA follows Region V ecological risk
assessment guidance, which calls specifically for a. tiered approach. According to the regional
guidance, more detailed risk assessments should be conducted if potential risks are identified at the
SERA stage. However, the philosophy of taking a highly conservative approach at the SERA stage
to avoid underestimating risks can be taken to an extreme. This approach implicitly assumes mat
resources win be available to conduct further, more detailed phases of ecological risk assessment, in
which overestimation biases presumably -wifl be uncovered and corrected. Given the need to make
timely and cost-efiectivc decisions, time and resources arc not always available for additional tiers of
analysis. Therefore, it is imperative to evaluate risks conservatively but realistically, so that
unnecessary additional analysis or unwarranted remedial measures are not deemed to be required.
Further suggestions for incorporating more realism into the SERA are provided below in comments 5
through 10.
3. Uncertainties of the SERA
The identification of uncertainties is comprehensive, with the exception of uncertainties
regarding the chemical form of metals in stack and fugitive emissions (discussed below in comment
9).
4. Weakest and Strongest Aspects of the SERA
The weakest aspects of the SERA are the methodological problems with selection of ECOCs
(ecological contaminants of concern), and the lack of consideration of background exposure to
ubiquitous chemicals. Comments 5 though 10 offer suggestions for strengthening these aspects of
the SERA.
The strongest elements of the SERA are its comprehensive site characterization and
consideration of exposure pathways, and the sophisticated incorporation of. exposure models.
5. Section n
There arc several areas requiring clarification and reorganisation in this section.
D-179
-------
Steven C. Peterson
The preliminary screening process used to select ECOCs should be moved from Section B.2b to
Section A. 1. Stressors. and the ECOC selection process should be more clearly explained. The
methods used to identify potential receptors wrthin the assessment area should be identified in
Second JL2, Ecological Components.
The results of the site characterization and selection of ECOCs should be used to refine the
conceptual model presented in this section. For example, Section HI identifies the virtual absence of
ecologically significant habitats -within a 1 km radius of the facility. Yet according to the dispersion
modelling this is the area where maximum air concentrations and deposition occur. Given the lack
of significant ecological receptors within the 1 km radius of likely exposure, further explanation is
needed to justify the evaluation of a broad.range of ecosystem components within this area.
, Tt appears from Section IV that many chemical; have been selected as ECOCs even though
they have not been detected in emissions tests. As discussed below in comments on Section IV, the
first screening step typically taken is to remove non^letxted chemicals from consideration. In
addition, as stated below in remarks on Section V, chemicals with predicted concentrations less than
background could also be removed from consideration. Further narrowing of the ECOCs by
eliminating non-detected chemicals and chemicals with concentrations below background might lead
to a more focused identification of relevant pathways and receptors.
In Figure H-2, Conceptual Site Model diagram, the exposure pathways and receptors
identified for stack and fugitive ash appear to be identical and could be combined. Moreover, the
text should state why deposition to soil is not shown as a relevant pathway for organic fugitive
ECOCs (I did not notice any mention of the reason for this until later in the report, page VII-2).
6. Section HI
This section is comprehensive and well-written. The information could be iitiliTrd to refine
the assessment cndpoints identified in Section n, as described above.
7. Section IV
There are several problems with the overall organization and methodology of this section. As
mentioned above, there is too much detailed discussion of the methodology used for ECOC selection
- because of this, it is difficult to develop and tny'ntam comprehension of the principal findings, hi
this regard, it wc^d be usefid to provider overall summary at the end of this section. An overview
D-180
-------
Steven C. Peterson
of the EGOC selection process also is needed, perils presented diagrammaticalTy to capture the
overall purpose and flow of the tiered selection process. ID addition, some of the material included
in this scclion is not realty relevant to thcECOC selection process. For example, methods of
estimating emission rates for metals arc provided in Section B3, yet metals are selected as ECOCs on
the basis of the need to evaluate permit limits and thus, none of the omissions rate data is relevant to
ECOC selection. Rather, the emission estimates are needed for the exposure assessment, and should
be placed in Section V or in an appendix.
Non-detected chemicals are included as ECOCs. As stated above, chemicals that are not
found lo occur at a site typically are not selected as ECOCs. The approach taken in the SERA of
evaluating non-detected chemicals using the detection limit as an estimate of the emission rate is
certainly conservative, but is it realistic or necessary? If the detection limits are not felt to be
sufficiently low to evaluate ecological risks, this can be pointed 6ut and perhaps data quality-
objectives could be set to allow then* detection at ecologically-relevant concentrations.
The ranking algorithms are potentially misleading in that no attempt is made to express
input parameters on a uniform scale. This leads to an inadvertent bias because the calculation of
overall scores is particularly sensitive to extreme values of any given parameter. I believe this
accounts for some of the apparent biases in the ranking, such as the effect of the very low ingestion
TV (toxicity value) on the overall score for dioxin/furans. I would emphasize mat the selection of
ECOCs for WTI is probably not unduly affected by mis statistical bias, but I am concerned about the
need for further review of mis methodology, since acceptance by the workshop and inclusion of mis
approach in the final report is likery to set a precedent that will be followed by others.
Of greater significance are the methods used in the derivation of toxicity values. In general,
the tenacity valties used in thg ECOC sdcctiOT algorithm? ar^ p*rrly dcTTOmiurd ™thg report - the
values are provided in tables, but the source and toxicological basis of the TV is not identified for
each value. I was able to identify the basis of several of the TVs by inspection of Appendix VI-24
and VI-25, where oral toxicity values are provided for ECOCs and benchmarks are derived for use in
the risk assessment For discussion purposes, some of the oral toxicity values and their basis are
provided in the following table.
D-181
-------
Steven C. Peterson
Selected Oral Toiichy Values Used in Screening ECOCs
Chemical
£ezxzo(c)pyieue
BbC2-ctfayIhecyi) Phtbilatt
2,4-D
Bffltchlorobenzene
PentasUofophend
Fotychteinaicd Biphcnyls
2^,7,8-TCDD
TV used in
screening
10
200
02
I
3
0.1
0.00001
TV used! hi the risk
characterization
10
5
10
1.6
1.2
0.32
0.000001
Receptor and Effect
Mouse LQAEL • reduction in fertility
and reproductive capacity
Rat LOAEL - maternal effects
Mammal NOAEL - effects not specified
Rat NOAEL - reproduction
Rat-no effect
dccrcftscd utter size
Rat - no reproductive effect
The table illustrates two important issues related to the selection rfTVs for screening
ECOCs. First, there appear to be discrepancies in the TVs selected for screening and the TVsused
in the risk characterization portion of the risk assessment, Some of these discrepancies arc quite
significant (c.g-, Bis^-ethylhexyl) Phthalate) and could influence the results of the screening.
Second, there is no unifemritv in the selection of NOAELs and LOAELs as the standard for
evaluating effects. According to the statement on page IV' 10, NOAELs were selected if available.
However, using LOAELs for *""« chemicals and NOAELs for others introduces a serious bias in
i
the ranking, bccmis? for any fren i^*mea\ the LOAEL and the NOAEL can differ fay an order of
TtTflgpftiMte Since it is customary to usesafely factors to estimate NOAELs from LOAELs if a
measured NOAEL is unavailable, this approach could have been used to provide a uniform basis fbr
comparison among the TVs.
8. Sections V and VI
Comments on Section V
The first few subsections (A through G) of {Section V represent a refinement and completion
of the problem formulation phase (see coramentson Section fl). In addition, references should be
D-1,82
-------
Steven C. Peterson
provided to justify sane of the statements made in these subsections, namely: page V-4, the list of
metals "known" to bioaccumulatc; page V-6, the importance of "dermal" exposure for earthworms;
and page V-l 1, the higher lipid content of earthworms as compared to seeds.
There is a shift in emphasis in this section from the 20-km radius assessment area to the area
of maximum deposition within 1-km of the site. I agree that this is appropriate based on the air
modeling, but the point should be made more explicitly earlier in the document
Although I cannot comment on the details of the modeling presented in other volumes of the
report, the application of the models in the SERA is sound and presented clearly and logically. My
main concern with this section relates to the lack of consideration of background concentrations of
naturally occurring substances such as metals, and general ambient concentrations of other widely
occurring pollutants such as PAHs and PCBs. For example, many of the metals concentrations in
soils and sediments predicted for both exposure scenarios are well below background concentrations.
It is not uncommon to eliminate chemicals from further consideration in risk assessments if their
site-related concentrations arc below background levels. The-selection of ECOCs could be refined by
eliminating metals that do not occur above background. In addition, some of the organics arc
predicted to occur at concentrations substantially below levels typically found in developed areas
such as the Ohio River valley. Further discussion of background comparisons is provided in
comments on Sections Vn & VIII, below.
Additionally, a default value for small mammal whole-body tissue BAFs of I is used,
because these arc "generally unavailable for most chemicals'1 (page V-26). However, soil to mammal
BAFs are available for some metals, and small mammal tissue levels of many contaminants have
been reported from controlled laboratory feeding studies.
Finally, incidental ingestion of sofl and/or sediment can be an irnrxjrtant pathway for many
wiidlife receptors. Dcfauh estimates of soil ingestion arc available in EPA guidance. I recommend
the inclusion of mis pathway in the SERA.
Comments on Section VI
The selection of the lowest reported effects concentration, use of uncertainty factors, and
other conservative approaches can result in calculated toxkaty benchmarks that areunrealisticaUy
low. While the methods used in this section are gcnerauy appropriate, tojddty benchmarks for
wildlife are not well established and professional judgement plays a larger role in then- selection than
D-183
-------
Steven C.Peterson
in the selection of benchmarks for aquatic life. See comments on Section VII,. below, regarding an
approach to determine if-the tadcity benchmarks are overly conservative. In addition, the chemical
form administered in the toscity test used as a basis for the selected toxicity benchmarks should be
identified, for reasons given in comments on Section VD! below.
AWQC (Ambient Water Qualify- Criteria) should be adjusted .as described in recent EPA
guidance for evaluating dissolved metals.
9. Section VH& VIII
While a. conservative approach is warranted, a'rcaifty check" is needed or predicted risks will
be unrealistic. This is particularly true for naturally occurring Tmfr»fc, for the following reasons. One
of the major factors affecting toxicity of metals to aquatic lifc.and wildlife is the form of the chemical
administered. Typically, highly bioavailablc forms such as metal salts are used in toxkaly tests, yet
metals in nature assume a variety of less soluble and available forms. As a result, mere is an inherent
bias toward cverestinTaticn of risks of exposure to metals^ The SERA fafls to identify this as a
source cfuncertaintywim a poteritiaBy large The exposure
•assessment docs not identify the likeiy form of me metals; emitted by the faculty, ncrdc«s the toxicity
assessment identify- the form of the chemical used in toxicity tests to establish benchmarks.
In order to provide a frame of refcreuue for the predicted site-related risks, it is important to
evaluate the potential risks'occurring through exposure to background levels of metals and other
ECOCs. Id the SERA, for example, some of the lexicological benchmarks for plants and wildlife are
below levels of exposure to metals In soils likely to occur at background (e.g., aluminum., arsenic,
chromium, mercury, etc.). In general, the evaluation of rids relative to reference conditions is an
accepted approach for ecological risk assessments and is isdvocated by EPA guidance. I recommend
that metals predicted to be below background levels in soils be removed from the assessment entirely
during the ECOC screening process. Metals predicted to be above background levels of exposure
should be evaluated relative to background risks to provide the needed perspective on the risks and to
allow interpretation of the significance of overestimation biases inherent in the SERA.
10. Section IX - Does SERA Inadvertently Underestimate Risk?
As mentioned in the report and in my previous comments, the SERA appears more likely to
overestimate than to underestimate potential risks. While this may be consistent with a conservative
D-184
-------
Steven C. Peterson
approach, it is equally important not to unduly overestimate risks if this can be avoided. Resources
arc finite and the overesamation of risks could be interpreted as a need to proceed with costly,
additional tiers of more detailed risk assessments •when in fact they arc unnecessary. I believe this
could be the case with the SERA - the predicted risks for metals arc likely overestimated, based not
only on the use of operationally unlikely emissions scenarios, but on the overcstimation bias of the
screening methods and the lack of consideration of risks relative to background.
D-185
-------
-------
Glenn Suter
Scientific Peer Review of the Ecological Risk Assessment Sections of:
Risk Assessment for the Waste Technologies Industries (WTI)
Hazardous Waste incinerator Facility (East Liverpool, Ohio)
Glenn W. Suter II
Environmental Sciences Division
Oak Ridge National Laboratory
Oak Ridge, Tennessee 37831-6038
December 1995
The following comments address each of the general issues and the specific issues related to the
screening ecological risk assessment contained in the charge to reviewers.
General Issues:
1. The ecological risk assessment (ERA) is generally clear but is not concise. It would have
been nice if someone had taken the time to reduce the repetitions and to make the text more
focused. I lost track of the number of times that the reason for not assessing risks to reptiles
and amphibians was repeated.
The "IDENTIFICATION OF ECOLOGICAL CHEMICALS OF CONCERN' was the most
difficult section to follow. It is in effect a preliminary screening assessment that is
performed prior to the main screening assessment. However, its logic is not clearly risk-
based. Rather, it is presented as being based on an "exposure analysis" (although exposure
is not estimated but toxicity is) and "professional judgement."
ERAs that have multiple endpoints are often hard to follow because the reader has to go
through the exposure analysis for each endpoint, then the effects analysis for each endpoint,
and finally the risk characterization for each, so that the continuity is lost. At ORNL we
have found that reviewers and stakeholders prefer an organization by endpoint. For
U-187
-------
Glenn Suter
example, you present the analyses of exposure and effects and the risk characterization for
plants, then for piscivorous birds, etc. I believe that would have helped here, but it is
probably not worth doing at this point.
2. The executive summary adequately reflects the conclusions.
3. Performance of an ecological risk assessment was recommended. The response was to
perform a screening ecological risk assessment. I am not sure that the intent of the reviewers
was satisfied.
4. The obvious impediment to using this assessment for decision making is the fact that it does
not reach a conclusion-about the ecological risk. A definitive assessment is needed.
5. Screening assessments should be routine, quick, and concise. A program is needed to
address this problem by developing default methods, data sets, models, and assumptions for
screening ERAs. We then need methods to perform conclusive ERAs for these types of
actions, the permitting of future complex effluents.
ERA Workgroup-Specific Issues:
The major problem that I have with this assessment is that it has.no clear purpose. A screening
assessment should either conclude that there is no credible hazard or should lead to a definitive
assessment that actually estimates risks of the hazards that are retained by the screen. This screening
assessment retains some hazards, but it does not prompt a definitive assessment.
1. See the specific comments below.
2. See general comment 1. Although the problem formulation is not identified as such (the first
four sections constitute a problem formulation), that is not really a problem. The
presentation is consistent with the framework.
D-188
-------
Glenn Suter
3. The discussions of uncertainty do not address the magnitude of the uncertainties and they do
not consistently address the significance of the uncertainties. There should be an analysis of
the magnitude of the uncertainties relative to the magnitude of the conservatism. However,
in fairness to the authors, I should point out that the narrative treatment of uncertainties
presented in this document is typical of current ERAs.
4. The weakest aspect of the ERA (ignoring the apparent lack of purpose discussed above) is
the selection of ECOCs. This is a difficult task and would benefit from a careful validation
of the method. The strongest aspect is the well researched paramaterization of the ecological
exposure models and ecotoxicological benchmarks.
5. a. Stressors appear to be adequately characterized.
b: Effects are adequately characterized for a screening assessment.
c. Assessment and measurement ehdpoints are not adequately characterized. The
assessment endpoints are rather vague statements and the "indicators" are closer to being
assessment endpoints. The assessment endpoints should be the things for which risks are
actually assessed and the measurement endpoints should be numerical summaries of actual
measurements used as estimators of the effects. Ask yourself the following questions, for
ecological concerns what is equivalent to lifetime cancer risk of a maximally exposed
individual (the assessment endpoint) and what is equivalent to the cancer slope factor (the
measurement endpoint)? By answering those questions, you can come up with useful
endpoints and eliminate the need for "indicators." For example, you really do not assess
risks to all birds and mammals as a gr.oup (first assessment endpoint). You assess risks to
particular species as representatives of taxonomic and trophic groups (e.g., kingfishers as
representative of piscivorous birds) just as the human health assessors estimate risks to a
reasonable maximally exposed adult as a representative of all adult humans. A measurement
endpoint is not an "evaluation" and it is impossibly optimistic to require that it be a
measurement in your indicator species (Table II-1). It should be something like "the lowest
oral NOAEL for mortality, growth or reproduction in an avian species." Specific comments:
i. The assessment endpoints are different is Table II-1 and Figure II-1.
ii. The "intact and productive food chains" endpoint seems pointless. Some
species are related to this endpoint but not others even though all species are part of
D-189
-------
Glenn Suter
food chains. Earthworms/soil fauna are the only "indicator group" that is-not :
protected except as contributors to food chains. Given the ecological importance of
these-organisms, why not protect them for their own sake?
iii. How do the indicator species represent the Ohio River (p. V-11)?
iv. Red-tailed hawks are not "top predators" (p. V-12). They feed largely on
herbivores so they are no higher than, for example, warblers.
v. What is the passerine/woodpecker group? It is not a taxonomic group,
trophic group, or guild.
vi. Contrary to the statement on p. V-13, small mammals are not a taxonomic
or ecological group, particularly when bats are included as here.
vii. Shrews are not surrogates for bats that feed on emergent aquatic insects or
lepidoptera. Shrews are a worst case for exposure via the soil invertebrate pathway,
but that does not help with bats that consume very few soil invertebrates. If you can
not assess bats, just say so, as you did with reptiles and amphibians.
viii. The large size and small metabolic rate of deer would make them more
rather than less.sensitive based on conventional wisdom (p. V-14). This metabolic
correction is what makes humans more sensitive than mice in standard risk models.
d. The logic for selection of the five emission scenarios is unclear. Why have a high
emission and expected scenario for metals in the stack emissions and not the others? Also
why was the metal stack maximum scenario so much more conservative than the high
emission scenarios for the other classes of emissions? Consistency would seem to be
desirable for comparing the emissions and their constituents. Finally, why were the risks
from the accident scenarios not assessed?
6. The site characterization is adequate but unfocused and excessively long. The discussions of
wetlands, parks, fauna, and flora and associated tables are not really used in the analysis and
are more like the kind of thing that pads EISs. Risk assessments should be more focused.
7. I believe that the results of the selection process are acceptable based on my professional
judgement, but I am not sure that I agree with all of the particulars. As stated earlier, the
logic is not clearly presented. The assumption behind the "exposure analysis" seems to be
D-190
-------
Glenn Suter
that relative risk is proportional to emission rates times some physical-chemical property
that controls exposure divided by toxicity. The most questionable part seems to me to be the
selection of the physical-chemical property which, in effect, substitutes for the entire
transport and exposure model!
A. The authors justify the use of Kow rather than log Kow, (which is conventionally used
in estimating bioaccumulation, because the log-scaled relationships are linear), by indicating
that they want to increase the influence of bioaccumulation on the scores (p. V-l l)r Did they
do a sensitivity analysis to determine that to be appropriate? How influential is Kow in the
scores now?
B. Is it reasonable to use solubility in scoring aquatic contaminants (p. IV-15)? Is
solubility really the physical-chemical property limiting risks to aquatic organisms under
these conditions? I doubt that it is. I believe that Kj is more likely to be a controlling factor,
as is degredation rate.
C. . Freon-like chemicals are eliminated because they are highly volatile (p. IV-19).
However, earlier the authors deliberately include volatile chemicals that had not been
included by the procedure. This does not appear to be consistent. Freons can be eliminated
based on their extremely low toxicity.
D. Selection by chemical group as well as by "exposure analysis" is said to increase
confidence that these chemicals represent the greatest risk potential (p. IV-20). Unless I
missed something, the same criteria are used in both cases. Therefore, there is no real
increase.in confidence.
E. Why are volatile organics included in sediment but not soil? Given that the
chemicals are originating in air, are they any less likely to partition to solids that are damp
than those that are saturated?
Validation of the scoring method would increase confidence in its results. One could apply
the scoring method to a set of reasonably well-characterized chemicals, then do a risk
analysis on them, and finally determine whether the scores and risks are correlated (at least
rank order correlated).
8. The analysis of exposure and effects is generally appropriate. However, I have some
disagreements and questions. My comments do not include the transport modeling
component of the exposure analysis, only the exposure models.
D-191
-------
Glenn Suter
A. Equation V-6 (p. V-27) does not make sense to me. I think the authors modeled it
too closely on the dose rate model. The BAF is simply an empirically derived quotient of
TCx/MCx.' It is not clear to me that the accumulation due to diet should be diminished by
FF or that the BAF is applicable to drinking water. Water does not appear to be a
significant source of exposure for wildlife. If that is correct, then in this model the water is
just acting as a diluent of the dietary exposure. Mammalian BAF studies with which I am
familiar did not include water as part of the exposure, so the model is not equivalent to the
situation in which the BAF was derived. If you can leave respiratory uptake out of the model
as negligible (assuming that is the reason), you can leave water out as well.
B. The paragraph at the bottom of p. V-27 is unclear since it implies that incidental
soil ingestion is handled separately from the model in Eq. V-6 (which does not mention soil).
Apparently soil is one of the dietary items, which is acceptable. However, one should not
have to refer to the tables to figure out what was done.
C. The parameters identified as percentages in Eq. V-6 and V-7 are proportions. I
assume that this is only an error in wording.
D. The selection criteria for toxicity data to be used as wildlife screening benchmarks is
unclear. On p. VI-11 the most sensitive species available is indicated. On p. VI-9&10
wildlife are preferred over laboratory species in the first paragraph of Sec. F, but in the next
paragraph "the lowest available and most applicable toxicological value" are indicated.
How are these potentially conflicting criteria prioritized? Other considerations are
mentioned, but their relative importance is not indicated. For example, is an acute or
subchronic test with an "applicable" species preferred over a chronic test with a less
applicable species? If you simply used professional judgement to select the best benchmark
value, taking certain factors into consideration, just say so.
E. The frequent lack of toxicity data is dismissed rather lightly on p. VI-11. See the
response to question 9, below.
9. A. The uncertainty due to lack of toxicity data for some chemicals is not treated in a
consistent rigorous manner. The authors should either prorate the uncharacterized chemicals
among the characterized ones based on similarity or greatest toxicity (which is what we did
in the synfuels technology ERAs, performed for the EPA ten years ago) or conduct a
consistent analysis similar to the one presented verbally for risks of organics in air to plants
D-192
-------
Glenn Suter
(p. VIH-3). That analysis would address the following questions: (a) what is the
concentration of the uncharacterized chemicals relative to the characterized ones, (b) what is
the level of relative toxicity that would be necessary for the uncharacterized chemicals to
pose a significant risk, and (c) what is the likelihood that the chemicals would have the
necessary relative toxicity?
B. The risk characterization supports the conclusion that most but not all hazards have
been screened out.
10. I believe that the assessment does not underestimate risks.
This is probably the most difficult type of ecological risk assessment to perform, a predictive
assessment with a large number of potential contaminants in an atmospheric emission. There is no
opportunity to focus on the properties of a single chemical as in registration of pesticides or
industrial chemicals and no opportunity to collect, analyze and test contaminated media as at a
contaminated site. In addition, the atmospheric route of exposure is the most poorly characterized
route for ecological receptors. Although the comments presented are largely negative, that is because
like most reviewers I did not take the time to point out the many good and correct analyses and
statements. The assessment is, as I expected, competently performed but not ground-breaking or
flawless.
D-193
-------
-------
APPENDIX E
REVIEWER WORK GROUP ASSIGNMENTS
-------
-------
v>EPA
United States
Environmental Protection Agency
Risk Assessment Forum
Technical Workshop on
WTI Incinerator Risk Issues
Holiday Inn - Georgetown
Washington, DC
January 11-12, 1996
Final Workgroup Breakout Sessions
Combustion Engineering
Barry Dellinger, Workgroup Leader
Elmar Altwicker
Toxicology
Mary Davis, Workgroup Leader
George Alexeeff
Thomas Gasiewicz
Air Dispersion/Deposition Modeling
and Accident Analysis
Walter Dabberdt, Workgroup Leader
Mark Garrison
Halstead Harrison
Jerry Havens
Geoffrey Kaiser
Ecological Risk Assessment
Glenn Suter, Workgroup Leader
Peter deFur
Pirn Kosalwat
Steven Peterson
Exposure Assessment
George Fries, Workgroup Leader
Thomas McKone
James Butler
t Printed on Recycled Paper
E-l
-------
-------
APPENDIX F
FINAL OBSERVER LIST
-------
-------
vvEPA
United States
Environmental Protection Agency
Risk Assessment Forum
Technical Workshop on
WTI Incinerator Risk Issues
Holiday Inn - Georgetown
Washington, DC
January 11-12, 1996
Final Observer List
Ann Anderson
Manager
AT. Kearney
222 West Adams
Chicago, IL 60606
312-223-6230
William Bailey
Environmental Manager
'Waste Technologies, Inc.
P.O. Box 919
East Liverpool, OH 43920
216-385-7336
Fax:216-385-7813
Robert Barton
Senior Engineer
Midwest Research Institute
425 Volker Boulevard
Kansas City, MO 64110
816-753-7600 Ext. 1425
Fax:816-753-8430
Pamela Blakely
Meteoro/og/st
U.S. Environmental Protection Agency
77 West Jackson Boulevard (DRP-HA)
Chicago, IL 60604
312-886-4447
Fax:312-353-4788
Dorothy Canter
Science Advisor, OSWER
U.S. Environmental Protection Agency
401 M Street, SW
Washington, DC 20460
202-260-3100
Fax: 202-260-2989
Paul Connett
Professor
Chemistry Department
St. Lawrence University
Canton, NY 13617
315-379-9200
Harriet Croke
Chief
Ohio/Minnesota/Wisconsin
Permitting Branch
U.S. Environmental Protection Agency
77 West Jackson Boulevard (DRP-HA)
Chicago, IL 60604
312-353-4789
Fax: 312-353-4788
Mohamad Elnabarawy
Sen/or Environmental Specialist
3M Company
P.O. Box 33331 - Building 41
St. Paul, MN 55133
612-778-5151
Fax: 612-778-7203
Craig Evans
Air Pollution Control Engineer II
Pennsylvania Department of
Environmental Protection
400 Market Street - 12th Floor
(RCSOB)
Harrisburg, PA 17105-8468
717-787-9256
Fax: 717-772-2303
Michael Firth
Project Env/ronmentaf Scientist
DuPont
P.O. Box 80027
Barley Mill Plaza
Wilmington, DE 19880-0027
302-992-6766
Fax: 302-892-7637
Rick Gillam
Environmental Engineer
Waste Management Division
U.S. Environmental Protection Agency
345 Courtland Street, NE
Atlanta, GA 30365
404-347-3555 Ext. 6331
Fax:404-347-1918
E-mail: gillam.rick@epamail.epa.gov
Printed on Recycled Paper
F-l
-------
Dwight Hlustick
Waste Combust/on Specialist
Office of Solid Waste
U.S. Environmental Protection Agency
401 M Street, SW (5303W)
Washington, DC 20460
703-308-8647
Fax: 703-308-8617
Adam Johnston
Environ
4350 North Fairfax Drive
Suite 300
Arlington, VA 22203
703-516-2300
Fax:703-516-2345
William Kappleman
Environ
4350 North Fairfax Drive
Suite 300
Arlington, VA 22203
703-516-2300
Fax: 703-516-2345
Melvin Keener
Executive Director
Coalition for Responsible
Waste Incineration
1133 Connecticut Avenue, NW
Suite 1023
Washington, DC 20036
202-775-9869
Fax: 202-833-8491
E-mail: mekeener@aol.com
Steven Knott
Chemist
Risk Assessment Forum
U.S. Environmental Protection Agency
401 M Street, SW
Washington, DC 20460
202-260-2231
Fax: 202-260-3955
Gary Liberson
Principal
Environmental Risk Sciences, Inc.
1000 Thomas Jefferson Street, NW
Suite 506
Washington, DC 20007
202-965-5188
Fax:202-965-5187
Ranjit Machado
Manager
Environ
4350 North Fairfax Drive
Suite 300
Arlington, VA 22203
703-516-2358
Fax:703-516-2344
Mario Mangino
Toxicologist
Waste, Pesticides and Toxics Division
U.S. Environmental Protection Agency
77 West Jackson Boulevard (DRP-HA)
Chicago, IL 60604
312-886-2589
Fax:312-353-4788
Craig Matthiesson
U.S. Environmental Protection Agency
401 M Street, SW
Washington, DC 20460
202-260-9781
Fax: 202-260-0927
Charles Maurice
Environmental Scientist
Office of RCRA
Waste Management Division
U.S. Environmental Protection Agency
77 West Jackson Boulevard (HRP-8J)
Chicago, IL 60604-3590
312-886-6635
Fax:312-353-4788
George Meaney
Technical Information Specialist
National Clearinghouse for Workers'
Safety and Health Training
10000 New Hampshire Avenue
Silver Spring, MD 20903
301-431-5427
Fax: 301-434-0371
Debbie Nolan
Technical Information Specialist
National Clearinghouse for Workers'
Safety and Health Training
10000 New Hampshire Avenue
Silver Spring, MD 20903
301-431-5425
Fax:301-434-0371
Michael Parkes
Public Relations Manager
Waste Technologies, Inc.
P.O. Box 919
East Liverpool, OH 43920
216-385-7336
Fax:216-385-7813
Amy Porter
Reporter
Bureau of National Affairs ,
1231 25th Street, NW
Washington, DC 20037
202-452-4106
Fax:202-452-4150
Krish Ramamurthy
Chief
Technical Support Section
Pennsylvania Department of
Environmental Protection
400 Market Street - 12th Floor
(RCSOB)
Harrisburg, PA 17105-8468
717-787-9256
Fax:717-772-2303
Alan Rubin
Sen/or Scientist
Water Environment Federation
601 Nythe Street
Alexandria, VA 22314
703-684-2438
Fax: 703-684-2492
Ruth-Ellen Schelhus
Environmental Specialist
NISC Program
1280 Maryland Avenue - Suite 580
Washington, DC 20024-2142
202-651-3109
Fax:202-651-3140
F-2
-------
Joseph Scire
Vice President
Earth Technology
196 Baker Avenue
Concord, MA 01742
508-371-4270
Fax: 508-371-2468
E-mail: jss@src.com
Fred Sigg
Vice President/General Manager
Waste Technologies, Inc.
P.O. Box 919
East Liverpool, OH 43920
216-385-7336
Fax:216-385-7813
Alonzo Spencer
President
Save Our County, Inc.
P.O. Box 1242
East Liverpool, OH 43920
216-385-4584
Fax:216-385-4584
Terri Swearingen
Coordinator
Tri-State Environmental Council
RD#I - P.O. Box 365
Chester, WV 26034
304-387-0574
Fax: 304-387-0574
Gary Victorine
Thermal Destruction Expert
U.S. Environmental Protection Agency
77 West Jackson Boulevard (DRP-HA)
Chicago, IL 60604
312-886-1479
Fax: 312-353-4788
S.T. Washburn
Principal
Environ
214 Carnegie Center
Princeton, NJ 08540
609-243-9817
Fax: 609-452-0848
Raymond Wayne
Public Relations Specialist
Waste Technologies, Inc.
P.O. Box 919
East Liverpool, OH 43920
216-385-7337
Fax: .216-386-2160
Michael Wigmore
Attorney-at-Law
Swidler & Berlin
3000 K Street, NW - Suite 300
Washington, DC 20007-5116
202-424-7792
Fax: 202-424-7643
Daniel Woltering
Principal
Environ
4350 North Fairfax Drive,
Suite 300
Arlington, VA 22203
703-516-2320
Fax:703-516-2345
William Wood
Director, Risk Assessment Forum
Office of Research and Development
U.S. Environmental Protection Agency
401 M Street, SW (8101)
Washington, DC 20460
202-260-1095
. Fax: 202-260-3955
F-3
-------
-------
APPENDIX G
PRINTED MATERIALS DISTRIBUTED BY OBSERVERS
The following materials were distributed by observers Terri Swearingen and Paul Connett.
These observers also distributed nonprint materials (e.g., a video tape and a ruler); these could
not be reproduced in this written report.
-------
-------
TRI STATE ENVIRONMENTAL COUNCIL
For immediate release: January, 101996
Contact: Tom Webster (617)-638-4641 (Office), (617J-542-1676 (Home)
Ellen Connett (315)-379-9200
Richard Wolf (216J-385-2133
WTI: EPA RISK ASSESSMENT OBSCURES OBVIOUS DANGERS
While 21 expert peer reviewers are flying into Washington, D.C. to attend the Technical Workshop on
WTI Incinerator Risk Issues, January 11 & 12,1996, citizens from East Liverpool, Ohio will have driven 8
hours through appalling weather conditions to attend this same workshop. They will be presenting their
criticism of the USEPA's Health Risk Assessment for the WTI hazardous waste incinerator.
Spokesperson, Terri Swearingen, said "We weren't going to allow this weather to keep us from pointing
out that the EPA's Risk Assessment is an enormously expensive exercise to obscure the obvious! Anyone who
has visited East Liverpool and is not "beholden" to the hazardous waste industry can see with their own eyes
that the incinerator is located ridiculously and dangerously close to where people live and where children go to
school."
According to Dr. Paul Connett, Professor of Chemistry, St. Lawrence University, who has been helping
citizens unravel the complexities of several thousand pages of the Risk Assessment, "The strength of a chain is
the strength of its weakest link. In this case the weakest link is clearly the accident analysis. This analysis is
full of holes. The authors did not even consider the track record of WTFs sister plant in Biebesheim, Germany,
which has the same Von Roll technology. This is a serious omission because the German plant has had a long
history of accidents and fires."
School Nurse, June Connolly, said that if one of the major fires that had taken place in Biebesheim
occurred at WTI, the children for whom she is responsible could've been killed. "It astounds me," she said,
"that the authorities here have an emergency plan for these children which consists of herding them into the
gymnasium and duck taping the windows."
According to Dr. Michael McCawley, Professor of Air Pollution Engineering, Dept of Civil
Engineering, West Virginia University, Morgantown, WV, "The accident analysis was deficient on three major
fronts: l)It did not include a "fault tree analysis". 2) It did not include an accident analysis of chemical plants
which, in theory, should be safer than hazardous waste facilities. 3) While they modeled a fire of spilled
liquids, they did not consider the vaporization of these same liquids prior to ignition. Such a scenario could
produce a giant explosion, much like a gasoline bomb."
Tom Webster, School of Public Health, Boston University, was able to demonstrate that the food chain
dioxin exposure analysis was seriously underestimated.
Alonzo Spencer, Save Our County, stated," The question comes down to two things: 1)Whether or not
these peer reviewers, with their lofty credentials, can empathize with the ordinary person, the person who lives
with the dangers every day. 2)Whether the EPA, even at this late hour, can admit they have made an
horrendous mistake and shut this plant down immediately."
In addition to their oral testimony, the citizens- gave each of the peer reviewers a copy of a 17 minute
videotape they had prepared, documenting an explosion in an American hazardous waste incinerator; a list of
accidents and fires occurring in the Biebesheim facility(WTFs sister plant in Germany); a series of video clips
showing the proximity of the WTI incinerator to people's homes and the elementary school. Citizens also gave
each member of the panel a list of statements made by scientists, regulatory officials and politicians who have
visited the location., a copy of the Ohio State Law passed in 1984 which would have prevented the siting of the
WTI facility if a permit was applied for today, and lastly, a ruler for each member to remind them of Terri
Swearingen's comment on CBS 60 minutes, "You don't have to be a rocket scientist to work out that the WTI
location is unacceptable. You only need a ruler."
G-l
-------
-------
Ladies and Gentlemen,
Tarn a life-long resident of East Liverpool, Ohio. I am a high school graduate and
. had one year of business college. I don't have letters following my name to signify a
degree, like all of you. But I do have the important title of MOM. My husband, Bob and
I have five children. Matthew is nineteen and a freshman in college. Ryan is seventeen
and a junior in high school. Elizabeth is fourteen and a freshman in high school. Sierra
is nine and Alex is six.. They are in third grade and Kindergarden at East Elementary-
School, only 1100 ft. away from WTI. Our home sits on the bluff over looking WTI,-
800 ft. away, which is the subject of your latest science project.
From this little bit of information, you all can see that I have a vested interest in the
outcome of your work. If you inadvertently leave one T not crossed or one I not dotted,it
' could mean a life or death sentence for my family. I am not some hysterical housewife
with nothing better to do. Over the last few years, I have had to learn as much as I could
about this most unwelcomed neighbor. I have had to do a lot of reading about lead,
mercury and dioxins ( something I knew nothing about before WTI). This has taken
cherished time away from my family. I have experienced the terror of being evacuated
from my home and pulling my children out of their school to run out of harm's way.
During construction of this plant, workers struck a gas main and it ruptured. Even our
policemen didn't know what the problem was.
I have also been chased from my neighborhood because of the very nauseous fumes
from a chemical spill. Image your entire town smelling like cat urine. Not a very pretty
picture, is it?
Or imagine driving across a bridge and looking across the river towards your home,
and seeing the red flashing lights of fire trucks. I didn't think we would ever get home.
Our hearts were racing faster than the engine to our van. We had left our kids at home
for a little while to visit friends. Upon our arrival home, we saw firetrucks everywhere
at WTI. Remember, we have a bird's-eye view of the plant. Panic tries to take over at
this point, because we have no where to call for answers. My little ones seem excited
over the sight of all the firetrucks and commotion, but my older kids are giving me that
all to" familiar look. "Are we going to have to leave home again?" Our house is filling
with concerned friends and neighbors now and the phone is ringing off the wall. We
can't take it off the hook, because that might scare someone even more when they can't
reach us.
Last January, we were taking down our Christmas tree, when we hear this very loud
roar. We look outside and see what looks like Old Faithful to us. I"m going crazy
running for the video camera and the phone together. My oldest son says "Mom, WTI's
blowing again." The phone starts ringing "No , I don't know what's wrong".
Vallentine's Day, and the day after too, more firetrucks..What's wrong now? This is
getting ridiculous!
July 18th, a truck of "caustic WASTE" was delivered to WTI by the same trucking
company that delivers their "caustic MATERIAL". The truck driver knew the company
was expecting the 'CAUSTIC MATERIAL' AND WAVED THE TRUCK THROUGH
WITHOUT AN INSPECTION. The material was then put into the pollution control
system. As the truck was leaving the plant, an employee looked at the truck manifest
and realized the mistake. The plant was shut down and the pollution control system had
G-3
-------
to be drained and cleaned I brought all of this up to Gary Victorine of USEPA and he
said it was just a miscommunication. Would an air traffic controller accept that excuse?
I could go on and on and on with stories like this. But frankly, I"m tired of this game.
I don't want to play any more. The stakes are too high.
I would like you to look at this map of the US and tell me where each of you live.
I bet you are all a safe distance away from the world's largest hazardous waste
incinerator. I want my family to be a safe distance away, too. Moving is not an option.
This is my home. I am an American citizen, not- some wealthy, foreign-owned
corporation.
PLEASE, when you are doing your job on this risk assessment, remember each of
those numbers on your paper has a name and a face and a life behind it
PLEASE, FOR OUR CHILDREN'S SAKE !!!
Sincerely,
Sandy Estell, MOM
G-4
-------
CITIZENS CRITIQUE OF THE ACCIDENT ANALYSIS IN THE U.S. EPA's RISK
ASSESSMENT FOR WT1
1) Introduction
While we have a number of concerns and criticisms about many parts of the
U.S.EPA's Risk Assessment, our major concern is for the chapter on Accident Analysis
(Chapter 7). The strength of a chain is the strength of its weakest link, in our view, the
accident risk analysis is by far the weakest link in the entire assessment. It is grossly
inadequate! At the end of this analysis we attach a few comments on other parts of the risk
assessment including, comments by Tom Webster on the food chain exposure analysis.
2) Weaknesses of the accident risk analysis
2.1 The accident history of the Von Roll hazardous waste incinerator in Biebesheim,
Germany was not reviewed even though it has almost identical technology to the WTI
plant. Moreover, it has had a long history of accidents and fires which could have thrown
important light on the potential for fires and accidents at the WTI plant (see video).
2.2. A number of American hazardous waste incinerators were not reviewed even
though their operation history would be significant for accident analysis. For example,
a) PCI, Shakopee, MN, was not included even though it was destroyed in an explosion
which led to its closure by local authorities (see video). '
b) Caldwell Systems, Lenoir, N.C. was not included even though it was also destroyed in
an explosion and closed down by local authorities.
c) LWD, Calvert City, Kentucky. LWD was listed as Clay County but this small
incinerator was closed many years ago. On the other hand, LWD has a major and
troublesome incinerator in Calvert City which did not appear on the EPA list.
2.3 Where American facilities were reviewed, the list of accidents was incomplete.
For example, the ENSCO incinerator in El Dorado, Ark. had a major explosion in
December, 1994. This was not included (see attached newspaper account).
2.4 Since the analysis was prepared there has been a major explosion in the Ross
incinerator in Grafton, Ohio.
2.5 The authors relied heavily on self reporting by the facilities and the records of
regulatory authorities. These two sources frequently combine to give a very cosmetic
account of accidents with respect to off site damage and impacts on citizens. To get a
more balanced picture, the reviewers need to examine local newspapers and to interview
residents who have often kept extensive records of plant accidents, including photographs,
videos and diaries.
2.6 The accident history of chemical plants should also have been reviewed. This is a
critical omission in our view because a hazardous waste incinerator is really a chemical
plant handling dirty chemicals. One would anticipate fewer accidents to occur in chemical
plants because the operators are dealing with known chemicals with precisely known
dangers. Knowing why these accidents occur could throw important light oh potential
accidents in a plant handling hazardous waste. Moreover, studying the impact of fires and
explosions involving known chemicals would be very valuable in attempting to estimate
G-5
-------
the impact of similar chemicals handled by WIT. We have attached a list of accidents in
the chemical industry by R. Andurand.
2.7 The scenarios chosen for analysis were not worst case even though the authors
claim they were. For example, the fires examined involved liquids burning. The authors
did not consider a situation were the liquid first vaporizes and the condensate aerosol
becomes ignited causing a gasoline bomb type explosion. Such an explosion would have
devastating consequences in the East Liverpool location.
2.8 The scenarios did not include a major fire hi the bunker where solid hazardous
waste is dumped and co-mingled prior to burning. Again, this was an important omission
because numerous fires have occurred in the bunker in WTT's sister plant in Biebesheim,
Germany including two large fires (see video). This becomes even more important when
one discovers that WIT has already had several smaller fires in the bunker in its short time
of operation.
2.9 In considering the experience with American incinerators the reviewers have
contented themselves with the notion that when major accidents have occurred, even those
involving deaths of plant workers, little-or no off site damage has occurred, however, such
analysis ignores the unique location of the WTI incinerator. The reviewers before
concluding no off site damage, should have compared the location of each plant vis a vis
housing, schools and topography with the location of the East Liverpool plant
2.10 The authors of this report appear to be insensitive to the very special
problem of the location of the East Elementary School. In particular, they did not appear
to be aware that the emergency evacuation plan for these children is to herd them into the
gymnasium and duck tape the windows. Had they been more sensitive to this issue, they
would've realized how inappropriate it was to project casualties based upon IDLH values.
Since these IDLH values are derived from the levels from which a healthy adult worker
can escape within thirty minutes. Even a factor often applied to these values would
probably not be adequate to estimate a level a small panicking child could withstand.
Replacing the IDLH values divided with a more protective value of a 100 would rriake a
huge difference to their analysis.
2.11 We are told by a professional in the field that an accident analysis is not
really an analysis unless a fault tree analysis is constructed. Such an analysis would
require a much greater in depth understanding of each step in the WTI process and each
aspect of operator training. Someone once said, "Those who do not learn from history, are
doomed to repeat it." Prior to the Three Mile Island accident, risk experts had predicted
that the chance of such an accident occurring would have been once in a billion years. A
post-mortem fault tree analysis of the plant's operations greatly reduced this prediction!
2.12 No scenario was considered in which radioactive waste was accidentally
burned at the plant, even though it has been well established that radioactive waste has
been burned at numerous incinerators, including Rollins, LA, and ENSCO, El Dorado, AR.
2.13 Despite two requests by citizens, the dispersal of chemicals liberated in an
accident was not modeled using a wind tunnel experiment, even though equipment had
G-6
-------
been set up to model stack and fugitive emissions. Citizens were told that it would have
been too expensive. Instead, the authors have relied on a computer model.
2.14(a) When the authors modeled fires involving organochlorine wastes, they
failed to acknowledge that such fires have the potential to create considerable quantities of
dioxins and fiirans and other toxic by-products. While these products may not pose an
acute effect at the time of the accident, they could considerably contribute to
contamination of residents' back yards and the playground of the elementary school. Even
if these risks were not calculated in the accident analysis, they should have been passed on
to other sections of the risk assessment.
2.14(b) We are puzzled why the authors estimated that the burning of 5,000
gallons of liquid waste (tetrachloroethene and toluene) should take 117 minutes on site, but
only 7.2 minutes off-site, especially since the burning of 100~gallons of this waste took the
same length of time (2.8 minutes) in both locations.
2.15 Unfortunately, the accident analysis reads more like a PR exercise than a
genuine attempt to gauge the likelihood, and the impact, of a major accident occurring at
WTI. The use of words like "moderate" to describe an accident involving up to 10
fatalities may be appropriate when considering federal emergencies, but it is highly
insensitive and inappropriate when discussing the chances of local residents or children
being killed by an operation they have not chosen.
2.16 Even if we take the accident analysis at its face value with all its many
shortcomings, it is highly unlikely in our view that any permitting body in possession of
this analysis would have allowed WTI to have been built at this location. The state of
Ohio, the Tri-state region, the United States of America are simply not that small that
residents should have to live near or children go to school with this kind of risk hanging
over their heads. A more appropriate and thorough accident analysis, in pur view, would
make the location of this plant unthinkable. We wonder how long our regulatory agencies
can support the unthinkable.
2.18 We invite all the peer reviewers to consider for a moment what it would be
like if they had to live in Sandy EstelPs house, or if they had to send their children to the
East Liverpool elementary school (see video). At this point, we are asking more than
expertise, we are asking for empathy. It's time to call this agony to an end. The plant
must be shut down.
2.19 Whatever the financial remuneration to WTI by the U.S. government to
compensate for their costs would be a small price to pay to remove the shadow from the
lives of the residents who live near this facility. Simply put, no child in the United States
should have to go to school 1100 feet from a hazardous waste incinerator.
2.20 No amount of number crunching on other parts of this risk assessment can
compensate for the totally unacceptable risks posed by accidents at this location. To dwell
on other risks, until the accident analysis is satisfactorily completed is to obfuscate the '
issue. These other risks are somewhat like choosing what kind of deodorant you'll wear to
your own funeral.
G-7
-------
3. Attachments
We are submitting the following additional material:
a) A list of statements made by scientists, regulatory officials, politicians and others who
have commented upon the location of this facility.
b) A 17 minute video tape which illustrates the kind of accidents that have occurred at
American incinerators (Shakopee, MN ), a list of accidents which have occurred at the
Biebesheim incinerator (sister plant to WTI), as well as shots depicting the location of
local residences and the school.
c) A copy of the Ohio state law which was passed in 1984, seven years before WTI was
built, which forbade the building of such a facility within 2000 feet .of schools, homes,
hospitals or prisons.
d) A ruler!!
e) Pat Cosmer's (Greenpeace scientist) comments on the dioxin emissions.
f) Tom Webster's (Department of Public Health, Boston University) comments on risk
assessment methodology and food chain exposure analysis.
g) A list of accidents in the chemical industry published in 1979.
h) An article from an El Dorado newspaper on the December 1994 explosion.
i) An article from an Ohio newspaper on the explosion at the Ross incinerator in Grafton,
Ohio.
j) Correspondence from the Tri-State Environmental Council to the EPA which raised
many important questions about the risk assessment prior to its completion, including
requests to model accidental releases of pollutants using wind tunnel experiments. Sadly,
most of these questions and issues were ignored. So much for citizen participation.
k) An important statement made by Deputy Surgeon General, Dr. Barry Johnson, to the
U.S. Congress, January 24,1994, where he indicates that of 72,000 papers published on
incineration, only one discussed the conduct of a population based study conducted in a
community living in the vicinity of an incinerator. In other words, the claims that an
incinerator poses no health threat to a community is a theoretical statement which has not
been validated, as is the case with the current risk assessment. To quote Dr. Johnson, "The
scientific information on human health impacts of incineration isn't often available
because the relevant studies haven't been conducted."
Note 1. The deposition velocities illustrated in figure H-3, page 11-54, Volume IV, are
based upon wind tunnel experiments. Field data give much higher values for particles in
the 0.01 to 1 micron sized particles. See figure 6 and reference 10 in Chapter 3 in Health
Effects of Municipal Waste Incineration, ed. Hattemer-Frey and Travis.
Note 2. The extrapolation in figure III-l, page 111-22, Volume IV, is mathematically
invalid.
G-8
-------
WTI, EAST LIVERPOOL, OHIO
EPA RISK ASSESSMENT OBSCURES OBVIOUS DANGERS
We believe that the EPA's Risk Assessment for the WTI hazardous waste incinerator has been
an extremely costly exercise designed to obfuscate the obvious.
The obvious (common sense) analysis of the WTI incinerator is that no matter how good the
technology, it should not have been built where it is: in a flood plain, 300 feet from the nearest homes
and 1100 feet from an elementary school, where the authorities have a totally inadequate evacuation
plan.
This obvious (common sense) analysis has essentially been confirmed by numerous scientists,
regulators, politicians and others who have visited the location. Below, after giving the key section in
Ohio State Law, which was passed in 1984, and would prevent the siting of such a facility at this
location today, we offer a series of statements made by those who have either visited the site or have
responded to a description of the location.
WHAT THE GOVERNING LAW SAYS ABOUT THE LOCATION:
Ohio Revised Code Section 3734.05(d)(6) states "The Board shall not approve an application for a
hazardous facility installation and operation permit unless it finds and determines:....(g) that the active
areas within a new hazardous waste facility are not located or operated within:
(i) two thousand feet of any residence, school, hospital, jail, or prison;
(ii) any naturally occurring wetland; or
(iii) any flood hazard area if the applicant cannot show that the facility will be designed,
constructed, operated, and maintained to prevent washout by a one hundred year flood or that
procedures will be in effect to remove the waste before flood waters can reach it."
WHAT SCIENTISTS HAVE SAID ABOUT THE LOCATION:
The United States is not so small that it should have to site one of its twenty commercial hazardous
waste incinerators so close to where children go to school. The ultimate fail safe of a facility like this
is the location where you put it. Here, we have no fail safe. WTI would have to be a perfect machine,
run by perfect people. Such a machine does not exist. It is only a matter of tune before a serious
accident will engulf these children. Only those who have not visited the location, or those who are
totally devoid of empathy, could tolerate allowing WTI to run for even a single day.
Dr. Paul Connett, Professor of Chemistry, St. Lawrence University, New York.
The WTI facility is the worst siting decision I have seen in my 25 years of practice in public health.
Locating a major hazardous waste incinerator 300 feet from the nearest residence and 1100 feet from
an elementary school with 400 children amounts to administrative incompetence if not malfeasance in
office and does violence to commonsense It is well known that the area where the WTI facility is
located is prone to frequent episodes of atmospheric stagnations, conditions which prevent pollutants
G-9
-------
released into the atmosphere from dispersing, instead resulting in sudden high pollutant buildups It
is truly hard to imagine a more inappropriate region for such a facility, to say nothing of the additional
matter of placing it next to an elementary school.
What if an accident were to happen at WTI? How would the nearby residents and the children in
the school escape? It is likely that there would be no time for an evacuation, and this problem has been
recognized for the school (I don't know what the residents, who are even closer, are to do.) The
emergency "plan" calls for a strategy of "sheltering in place." An examination of the details of this
plan reveals in the starkest fashion the underlying futility of truly protecting these small children.. The
plan assumes that all 400 children can be herded safely and efficiently into the school cafeteria within 3
minutes, the room sealed by stuffing wax paper and tin foil in the cracks and taping with duct tape, and
the air conditioning, heating and ventilation systems turned off so that outside air would not be
entering the building. If an explosion were to shatter even one of the windows, sealing the room would
be impossible. Even without a broken window, however, it is unlikely that toxic gases could be kept
out of the room, now making it more a tomb that a safe haven...
Dr. David Ozonoff, M.D., MFH, Professor of Public Health and Chair, Department of
Environmental Health, Boston University School of Public Health, June 21,1993
Considering all of the information available as to effects of exposure to the various chemicals that WTI
proposed to handle at the facility, I fail to see how morally or ethically, they can go ahead Should a
for-profit corporation be allowed to force this degree of hazard upon a community?..
Of greatest concern are the adverse health effects to children living and going to school in the area. It
is inconceivable that such arrogance, to site a toxic waste handling facility in an inhabited area and
close to a school, would persist
Janette Sherman, M.D., Internal Medicine, specializing in Occupational Medicine and
Toxicology and author of the book, "Chemical Exposure and Disease :Diagnostic and
Investigative Techniques.
While it is true that one can criticize the specific relevance of all of these studies [toxicological studies
of substances like dioxin] hi one respect or another, the bottom line is still the potential risks which
they imply, risks which are not discountable, partly because they are not quantifiable. It is obvious that
the WIT incinerator could create a high risk situation for the valley, and it is also obvious that there
remains a tremendous number of unknowns.
Daniel Vallera, Professor and Director, Experimental Cancer Immunology, Department of
Therapeutic Radiology, University of Minnesota, Minneapolis, MN 15 April 1992.
jTjhe physical proximity of the facility to an elementary school (1100 feet) and a residential
neighborhood (300 feet) violates reasonable norms for siting an operation that processes massive
amounts of hazardous materials and emits some of them at biologically significant levels.
Barry Commoner, Director, Center For The Biology Of Natural Systems, Queens College,
CUNY, Flushing, N.Y. in a letter to President Bill Clinton, June 23,1993
G-10
-------
[Tjhe location of the incinerator in a river valley 300 feet from houses and 1100 feet from an
elementary school could turn an accident into a catastrophe.
Tom Webster, Research Associate, CBNS, Queens College, CUNY and Member, U.S.EPA Dioxin
Peer Review and Risk Characterization Committee, in a letter to Pres. Bill Clinton, 23 June 1993
The location of the incinerator in a flood plain makes a mockery of siting standards adopted in this
country. In addition, the top of the stack is located 300 feet from a residential community and 1100
feet from an elementary school. Theoretically, the emissions may be low enough so that levels of
metals and bioaccumulative substances will not reach critical levels hi 30 years. However, the
experience with virtually every other commercial incinerator in the United States is that there will be
accidents and periods when emissions far exceed permitted levels. It is likely that in the next 10 years
the children hi that elementary school will be subject to acutely dangerous emissions from the
incinerator. If the explosion that rocked the Waste Management incinerator in Chicago last year
would occur at the WTI facility, those children would have been enveloped by a cloud of hydrochloric
acid. Such an accident should not be permitted to be a possibility.
Robert Ginsburg, Ph.D., Environmental Health Consultant, Chicago, Illinois, 17 June 1993
letter to President Bill Clinton
I have read the materials that you sent me on the WTI proposed incinerator. My particular interest is hi
the area of neurotoxicology of lead in children. ...The company's statement is that the annual
emissions of lead will be 4.7 tons. ...I can say that the risk to the brains of the children attending
school near the stack, during their most critical period of development, will be put at substantial risk.
These children have no say in this matter and it is our responsibility as adults and government to speak
for them.
Herbert Needleman, M.D., Professor of Psychiatry and Pediatrics, University of Pittsburgh,
December 4,1991.
[The number of days with valley stagnations here is very bad..Donora, Pennsylvania is not that far and
the weather conditions that occurred hi Donora in 1948 that resulted in 20 deaths due to air pollution
can occur here. If you couple that with any accidents.... then you have an incident similar to the one
that happened in Bhopal, India.
Michael McCawley, Ph.D., Professor of Air Pollution Engineering, Department of Civil
Engineering, West Virginia University, Morgantown, WV
WHAT REGULATORY OFFICIALS SAY ABOUT THE LOCATION:
Arthur Davis, Chief of Pennsylvania's Department of Environmental Resources since 1987, in a
March 29,1993 Pittsburgh Post Gazette interview concerning the WTI incinerator, told how he
would handle a situation like WTI in the state of Pennsylvania:
Post Gazette: Would you have approved the WTI incinerator as presently sited, if it were located in
Pennsylvania?
G-ll
-------
Davis: As I understand the thing, at that location, WTI -would not qualify for a permit under our law.
It "would be disqualified because of its location in a flood plain and its closeness to a school. So on
those two grounds, the exclusionary criteria -would deny WTI the opportunity to file a full application.
As a human being, I would never agree to the school being that close.
Valdas Adamkus, Regional Administrator, U.S. Region V, Chicago Tribune, responding to caUs
for his resignation, May 7,1993.
From June, 1988 through January, 1991,1 chaired the Ohio Hazardous Waste Facility Board which
permits all hazardous waste facilities in my home state. It was one of my predecessors who made what
I consider to be the most irresponsible environmental decision hi Ohio history with the permitting of
the WTI hazardous waste incinerator in East Liverpool, in 1984. Had it come before me, I am
confident in saying that my four fellow Board members and I would have laughed the proposal right
out of the room.
In my opinion, WTFs East Liverpool site is one of the worst possible sites in all of Ohio for storing
and incinerating toxic chemical compounds. It is on the banks of the Ohio River that is a drinking
water source for millions. Any mishap at the site, either accidental or through operational violations,
will immediately impact the River. Because of the River and the residences and school immediately
adjoining the site, there is absolutely no margin for error at WTT. For this reason alone, it was both
technically and legally indefensible for the State of Ohio to permit WH to operate there. The fact that
the Ohio River Valley surrounding East Liverpool is prone to atmospheric inversions that trap air
emissions is a separate ground which, standing alone, should have led to the site's prompt
disapproval...
Attorney Richard Sahli, Former Chairman, Ohio Hazardous Waste Facility Board from June
1988 through January 1991. December 6,1993
In summary, regardless of site preparation, flood hazard areas are inappropriate for a hazardous waste
management facility. Based on this and other considerations, it appears that siting considerations and
alternatives were not fully and properly evaluated in this case.
Ohio Department of Natural Resources, in a letter concerning WTI, December 29,1982.
A top of the line, quality waste analysis plan is really needed, especially when one considers that a
school is located only 1100 feet from the site. There can be no significant margin for analytical or data
interpretive errors, given that human error will always manifest itself in the form of routine facility
accidents What may be "okay" for storage facilities is insufficient for a major commercial
incinerator parked next door to a school. As a Waste Management Division Director employed in
Region V is reported to have stated, "Let's put the "E" back in the EPA".
Allen Debus, U.S.EPA Region V, in a memo concerning WTI, March 19,1992
G-12
-------
The following is a statement by John Higgins, Regional Environmental Engineer for the State of
Massachusetts, in a November 20,1986 letter explaining the rejection of a permit to build a
solid waste incinerator in Hofyoke, Massachusetts, because of the location:
The proposed facility would be located in a river valley surrounded by elevated terrain. This
topography would augment any air pollution problems from the facility. The residents of the elevated
areas surrounding the plant would be exposed to higher levels of pollution from the plant than if the
land were level. The upper valley meteorology of the area could lead to persistent trapping of
pollutants, which would raise short term exposures. The pollutants would not disperse as quickly as
they would if the plant were located on level terrain. The health effects of the pollutants emitted from
this plant would be heightened by its location because residents of impacted areas would be exposed to
higher levels of pollutants for longer periods of time..
WHAT POLITICIANS SAY ABOUT THE LOCATION:
The potential impact on the people of this community...on their health, on their children's health, on
the investment they have made in their homes and businesses...is too great to proceed without study
and caution.
And the very idea of putting WTI in a flood plain you know its just unbelievable to me!
Vice Presidential candidate AI Gore, July 19,1992
I mean the federal government should not permit incinerators where you are going to have on-site
storage of garbage in a flood plain—that should not be done. You ought to have some jurisdiction at
the state level about how close they get to schools and other things which are really troubling.
Presidential candidate Bill Clinton
The fact is the USEPA has been doing an end run around the regulatory process for 10 years, thinking
the Ohio Valley is already so polluted no one would notice a little more.
An overwhelming majority of East Liverpool's citizens do not support WTI. In fact, many elected City
Councils in Ohio, West Virginia and Pennsylvania on on record as opposing WTI, as are health groups,
labor unions(UAW, Steel Workers, Communications Workers) and others. -
Encouraging environmental tech is one thing, building a hazardous waste incinerator on the banks that
supplies drinking water for 10 million people, next to an elelmentary school and neighborhood, is just
absurd.
Senator Howard Metzenbaum, Ohio, December 15,1992
Be advised that I have never agreed with the location of the facility and have stated that the students at
East End Elementary School should be relocated. If a problem at WTI should happen to occur during
school hours, what plan would be sufficient enough to'ensure the safety of the children? Just the
G-13
-------
mention of a "safe room" with tape around the doors is enough to fuel the thoughts of a prudent person
to consider the relocation of the students.
James Traiicanr, Member of Congress, in a letter to a member of the East Liverpool Board of
Education, Re. the WTI Evacuation Plan, April 13,1993.
We don't want WTI! The dream of the residents of the state is to leave it better than it was found. Our
dream and WTI are absolutely inconsistent. ...We are all entitled to our health. [TJhis affects all
generations. This [threat] is the result of big business and greed.
The Honorable Gaston Caperton, Governor of West Virginia, June 2,1991.
U.S.EPA wants this facility as much as WTI does...this permit is too important to the agency to deny
no matter what the consequences are to East Liverpool.
Doug Applegate, US House of Representatives, 7/25/83
WHAT OTHERS SAY ABOUT THE LOCATION:
I believe the widespread effects of a potential disaster on the site are too great and too many questions
remain unanswered. Moreover, the board herein, seems blinded by the technological sophistication of
the plant when time and time again human error has managed to overcome "fail-safe" technology from
Bhopal to Chernobyl to the Rhine. One cannot and should not presume the worst will occur.
However, proper planning requires fallback measures to ensure that even minor accidents by plant
management standards do not turn the southern portion of this state into a wasteland. There cannot be
any reason good enough in my view, for the plant's current proposed location.
Justice J. Locher, Ohio Supreme Court, dissenting opiniion
The WTI case reveals something of a noble lie in our country. Citizens are assured that the federal
government will require careful permitting and operation of hazardous waste facilities. They are
encouraged to participate in the review of such facilities and promised that they will have a voice.in
decision of siting and permitting. Most communities do little to test these promises. The WTI
families did and discovered that there is little that a community can do to oppose powerful interests
supporting the hazardous waste market. What we do in this case and dozens like it will determine
whether these promises will be finally honored or discarded by the government.
WIT is an embarrassing example of lax siting regulations for waste facilities. Located on the
banks of the Ohio River, WTI was placed in a residential area only 1100 feet from an elementary
school WTI is about money! There is no better case for new environmental equity laws than
WTI. Ultimately, WTI sends a chilling message that individuals are insignificant players in the global
marketplace.
Jonathan Turley, Professor of Law and Director, Environmental Crimes Project, George
Washington University, Congressional Symposium on the WTI Incinerator, December 6,1993.
G-14
-------
W JJWQERMLE » 60TH6K5EBG • HAMBURG • LEWES- ILK. • LONDON • LUXEMBOUR6 • MADRID • MONTREAL • OSLO • PALMA DE MAI LORCA
.PARIS • ROME • SAN FRANCISCO • SAN JOSE - COSTA RICA • SEATTLE* STOCKHOLM • SYDNEY • TORONTO • VANCOUVER • VIENNA
. WA8HINGTON.»WOBLO.MRKBASS-ANTARCTICA»ZORK!H
Route 7, Box 113, Eureka Springs, Arkansas 72632
Dlf (501) 253-8440 Fax (501) Z53-S540
patcostnefdgreen2.dat.de
5 Jan 95
Dear Paul, Ellen and Terri,
Attached are .two brief discussion papers that I regard as
pertinent to the WTI risk assessment:
* WTI Dioxin Emissions Before and A£t@r Carbon Injection
This paper documents that at least the first series of tests
conducted after installation of the carbon injection system
were conducted at much lower feedrates than the feedrates
of the pre-CI5 trial burn. I do not know whether similarly
low feedrates were also used during the subsequent test
burns.
* Limitations of Historic and Current Methods fx>r Measuring
Polychlorinated Dioxins and Furans in Incinerator Emissions,
Once again, I suggest that the quantification of PCDD/Fs in
stack gases .suffers from great uncertainty.
In the absence of data such as waste composition, waste
feedrates and stack gas flowrates,' it is not possible to reach
any useful .conclusions about the validity of the identities or
quantities of PICs used in the risk assessment. Obviously they
have presented an extensive list of PICs ... about-137, I
believe. Since they report PIC emissions, grams per second,
rather than PIC concentrations, grams per cubic meter, it is not
possible to compare their data with those of other trial burns or
published studies.
Given the conclusion by Huang and, Beukens (1995) that de ncwo
synthesis from carbon particles is the dominant mechanism for
PCDD/F formation, I would be very curious about the
characteristics and fate of the injected carbon. Where does at
end up? [Huang, H., and Beukens, A. Oh the mechanisms of dioxin
formation in combustion processes. Chemosphere (1995) 31(9):
4099-4117]
Based on their highest TEQ emission rate, they are releasing
P.CDD/Fs from the stack at a rate of about 0.12 gram per year.
This exceeds the acceptable annual dioxin intake for wore than
400 million people, based on the currently proposed risk-specific
dose of OiOl pg/kg/day and assuming that one cancer death in a
million ie acceptable.
RECYCLED PAPEfi
G-15
-------
In their estimate of fugitive emissions, they have failed to
consider contributions from many potential sources, including the
largest likely source — the kiln. . Oh page V-13 of Vol, IJ of
the risk assessment, they admit as follows:
Kiln overpressure events trigger automatic waste feed cutoffs
(AWFCOs); ... due to the frequent occurrence of kiln
overpressures {WTI 1994), a detailed evaluation of these
events has been conducted . . . However, since these emissions
occur from the kiln seals, PCDD/PCJDF are unlikely to be
associated .with these releases ..."
Of course, such AWFCOs can be expected to be highly
significant sources of fugitive emissions, including products of
incomplete combustion, such as the POKD/Fs as well as many
others. Contrary to their conclusion that PCDD/Fs are unlikely
to be formed, the conditions,, which include high soot formation/
are extremely conducive to PCDD/P formation, as described by
Huang and Beukens
In general, fugitive emissions are considered to be much
greater sources of wastef eed components than are the stack gases .
For example, at Chem-Security's PCB disposal facility in Swana
Hills, Alberta, Canada, fugitive emissions accounted for over 98
percent of the PCBs released at the site in 1994. Of some 95,136
kilograms of PCBs processed, the facility released 34 kilograms
into the environment at that site, a loss rate of approximately
0.036 percent. [Clearstone Engineering Ltd., An . Assessment of
Fugitive PCB Emissions, Calgary, Canada, June 9, 1995.]
Obviously, such losses of more volatile waste components can be
expected to be much higher.
I hope these comments are of some use. Proper detailing of
the emissions component of the risk assessment would have
required prior detailing of all of the trial and test burns.
Best regards,
C
t~
Pat Costner
G-16
-------
Preliminary Comments on the "Risk Assessment for the Waste Technologies
Industries (WTI) Hazardous Waste Incinerator Facility (East Liverpool, Ohio):
Draft"
10 January 1996
Tom Webster
Department of Environmental Health
Boston University School of Public Health
80 East Concord St., Boston MA 02130
associated with indirect exposure to combustor emissions."
Member, Peer Review-Risk Characterization Committee, USEPA Dioxin Reassessment
General Comments
The WTI risk assessment initially gives the appearance of extreme thoroughness, A great deal of
time and money has clearly been put into this effort. But my general impression is that this
attention to detail obscures the bigger picture. The apparent lack of attention to these larger issues
contrasts sharply with the growing dala requirements and sophistication-or over-sophistication-of
risk Assessment techniques. I will address some of these larger issues first.
It should be recalled that indirect routes of exposure were omitted from the preliminary risk
assessment for WTI released by EPA a few years ago. Partly in response to public concern and
legal controversy^ indirect exposure assessment is a major thrust of the draft health risk
assessment. While it is certainly true that the public has grown increasingly concerned over
exposure via the food chain, these concerns need to be seen from a larger perspective. Health risk
assessments for incinerators have evolved from simply examining inhalation-as was common
practice a decade ago-to a more thorough consideration of exposure pathways. However, at the
same time the public has grown more skeptical of the value of health risk assessments and their use
by_ regulatory agencies. This skepticism has at least two components. Fkst, the public is
mistrustful of the policy uses of risk assessment. Indeed, many suspect that health risk
assessments are used to justify policy decisions that have already been made* Second, the public is
becoming more aware of the scientific limitations of health risk assessment, limitations which are
often not clearly acknowledged. Unless these problems are addressed, I believe that such
assessments will be largely a wasted effort.
Health risk assessments that contain the correct caveats may provide some useful, albeit limited,
information, but only if put in the right context. Unfortunately, I believe that the current uses of
risk assessment has tended to usurp other modes of judgment. In my opinion, risk assessment is
illegitimate unless conducted in conjunction with other considerations. First, alternatives must be
considered, a notion which often seems forgotten these days despite the language of the National
Environmental Protection Act. Why should a facility be considered "acceptable" if a viable, less
damaging alternative is available? Critics have long maintained that alternative policies for
managing hazardous waste were never seriously considered in the case of WTI; they are not
addressed in the current document either. There are also important political and ethical issues
surrounding the notions of "acceptable.^ In a country where individual rights are supposed to be
protected, risk assessment/management appears willing tp sacrifice the health and well-being of
some for the profit of others with little in thejway of due" process. How are decisions about
"acceptability" made? Who makes them? Shouldn't the oublic, i.e., those who will bear the
G-17
-------
consequences, make such decisions? The opportunity to speak at public hearings and provide
written comments is not sufficient Finally, although many try to distinguish between risk
assessment technique and "risk management", this is largely a false dichotomy. Risk management
decisions are embedded throughout risk assessment, including the document under review.
An example of the latter is the primary reliance of NIOSH's IDLH values for assessing the
potential effects of accidents, These values are designed for workplace hazards and are based on
the ability of healthy young men to flee the immediate dangers posed by the accident. Application
of them to the general public is inappropriate. They should certainly not be used in relation to
school age children, elderly or the infirm. It's true that the risk assessment later uses the more
conservative levels of concern but relegates them to the uncertainty analysis. Indeed, the risk
assessment has a general tendency tends to bury the bad news or split it up into smaller, perhaps
more "acceptable" pieces. Another example is the decision to treat breast-feeding as a risk to a
limited number of people associated with a specific activity, placed in the same category as
subsistence fishing. However, most of the medical community believe that breast-feeding is or
ought to be a normal part of every human's life. Such decisions do little to inspire public
confidence. •
Non-Cancer Effects of Dioxin-like Compounds
In my view, one of the serious flaws of the WTI risk assessment is its treatment of the potential
non-cancer effects of dioxin-like compounds. The document states that "Because the threshold
levels for exposure to 2,3,7,8-TCDD and dioxin-like compounds below which toxic effects are not
observed has not been established, the USEPA does not currently list RfD or RfC values for
dioxin-like compounds"(V:IlI-6). This is at best a half-truth and has the appearance of hiding
unpleasant information. The dioxin reassessment actually concluded that any RfD would likely be
below "background" exposure. This conclusion was not changed materially by the last year's
SAB review. The risk assessment cannot hide behind the fact that the agency has/promulgated an
official RfD as these results are widely known.
Having side-stepped this issue, the risk assessment instead estimates an incremental dose. After
comparing this with the estimated average background, they conclude that any incremental effect is
very smaU. This is not a completely useless comparison, although it has its flaws. First, as noted
below, I think there is good reason to believe that the risk assessment underestimates exposure to
dioxiri-like compounds. Second, a proper comparison needs information regarding both
variability in the background exposure and the shape of the dose-response curve, about which we
know little.
But there is a larger issue. With these sorts of comparisons,, individual sources look smaller as the
background gets bigger, hardly the macro picture that EPA should be encouraging. Indeed, I
believe that the "background" exposure is due to the combined emissions from a large number of
sources, mixed together due to the relative environmental persistence and long-distance transport of
these compounds. Examination of the short-range impact of a single source is an inherently
flawed. This problem is not unknown to the agency: "Evaluation of indirect exposures from single
sites is too narrow a basis for decisions regarding stationary combustor risks" (USEPA 1994).
General limitations of incinerator risk assessment which need to be acknowledged
up front.
Public skepticism of risk assessment is also fueled by growing awareness of its scientific
limitations. These limitations are not usually acknowledged in a forthright manner. They do not
prevent many risk assessors from making fairly strong conclusions about the public health and
environmental impacts of a facility. If usurpation is the first flaw of risk assessment, hubris is the
G-18
-------
second. These limitations must be stated openly and up front, not buried in the back chapters on
uncertainty. Here are a few
1) Limited knowledge of what is emitted. The document acknowledges that only some of the
emissions have been characterized. The proposed accounting for other dioxin-like compounds-
brpminated, but not mixed halogcnated, etc.~is not very convincing.
2) Limited knowledge of fate and transport: Many of these compounds are chemically reactive in
the environment, producing other compounds which may be more or less toxic.
3) Limited knowledge of the toxicity of individual compounds. With no or incomplete data, the
toxic effects of a compound are treated as zero. This is apparent in the discussion of compounds
for study, The discussion of the possible risks of endocrine disrupters, a very hot topic ipR-?A.
environmental health (and of great concern to the public) is a good illustration.
4) Limited knowledge of dose-response and human effects.
5) Limited knowledge of the toxicity of mixtures of compounds.
6) Lack of attention to distant impacts.
7) Lack of attention to cumulative effects.
Technical Comments on Indirect Exposure Methodology
The limited time I had to review the risk assessment prevents me from giving a full technical
review here. Discussed here are some problems. I would refer the committee to another USEPA
document (USEPA 1994) for more technical comments on methodology for indirect exposure.
Emission Factors
The analysis docs not for account for upsets, which may lead to substantial increases of
emissions of dioxin-like compounds. As a result, the range of emissions appears quite small
Vapor/Particulate Partitioning
Bidleman's review provides a reasonable starting point for thinking about vapor/pardeulate
partitioning, remembering that partitioning depends on the ambient temperature, Jf a pollutant is
paniculate-bound at stack temperatures, then use of the stack PSD seems appropriate. However, if
a substantial fraction of a pollutant is vapor at ambient temperatures, it should equilibrate between
vapor and ambient particulate. The stack PSD may not be very relevant in this case. The dry
deposition rates implied by the air modeling appear unreasonably low.
Biotransfer of dioxin-like compounds from vapor to plants
The risk assessment uses the Bacci model divided by an "empirical" correction factor of 40
applied to all dioxin-like compounds. I do not think this is justified, leading to a possibly
substantial underestimation of exposure, especially for congeners other than 2,3,7,8-TCDD. EPA
examines this issue in Estimating Exposure to Dioxin-Like Compounds (EPA/600/6-88/005Cc),
Volume HI. In particular, EPA contrasts the results of McCrady et al with Bacci. Comparison of
the grass and azalea results in Table III of McCrady et al. suggests 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 etal. do 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
photodegradatioh of other compounds and the differences between plants. McCrady et al assert
G-19
-------
that the photodcgradation of 2,3,7,8-TCDD accounts for much of the discrepancy from Bacci's
results. If this is the reason, it is likely to be less of a problem for many other congeners which are
thought to be less susceptible to photodegradation. (Also see USEPA1994).
EPA attempts to validate their overall air-to-bccf model in the same document by comparing
rural air samples and beef samples (from different locations), I consider the data used to be Kir too
sparse to place much confidence in these comparisons.
wuvwAuuiiigr fmujrjjj
Although the uncertainty analysis is billed as presenting probable bounds on risk, it should
be clearly noted that it actually looks at pj
Uncertainty analysis
uncertain,
t actually looks at parameter uncertainty AND variability in exposure.
Uncertainty in biological models, typically the overwhelming source of uncertainty, is not
addressed It mixes in variability of population based factors; (e.g., consumption) which is OK for
population risk estimates but not necessarily justified for examining susceptible groups. It docs
not address model specification error, a potentially huge problem. The variance in crucial
parameters (e.g., biotransfer) is based on the variance in Kow, etc., rather than also taking into
account the uncertainty of the model itself (often large for such correlation models). Correction of
these problems would lead to much wider overall variance.
REFERENCES
USEPA (1994). Review of Draft "Addendum to the Methodology for assessing health risks
associated with indirect exposure to combustor emissions." EPA-SAB-IAQC-94-009b.
G-20
-------
Location
Annex
LIST OF ACCIDENTS IN THE CHEMICAL INDUSTRY
by
R. ANDURAND
in "The safety report and its application in industry", Annales dies Mines, 7-8, July/August 1979, 115-38
Country Dale Product Damage*
involved
Cause*
Hull
Cleveland
Manhattan
District
Project
Ludwigshafen
Newark < Warren
Oil Port)
Wilsura '
Whiting.
Indiana
New York
GB
USA
USA
FRG
USA
Germany
USA
A
1921
1944
1946
1948
I9S1
19S2
I9S5
1956
Hydrogen
LNG
Uranium
hexafluoride
Dimethyl-
ether
Not specified
Chlorine
Not
specified
Ethylene
Windows shattered within 3 km radmv Pressure felt within a 7 km radius and tremors 'up
to 70 km.
136 deaths. Nearby roads M»cpt r>> burning gas. Windows shattered, pavements ripped up.
drain coven blown across rixtn>>ps One fire engine blown into the air.
Two deaths, three people senouslv injured, thirteen slightly injured. Explosion of UFg and
very hot water in a laboratory HF aerosol carried up to 100 m by the wind.
245 deaths. 2.500 injured. W agon ruptured near a dimethyl-ether factory followed by
explosion and fire (Cost: 80 million FF).
No record
Seven deaths in an escape of 1 5 tonnes, coming from a storage tank.
Two deaths, thirty injured following a detonation in a pressurised container. Storage tanks
pierced by the burst bumed tor eighi days (Cost: 80 million FF).
• 1. 100 m^ of ethylene escaped inio the atmosphere causing an explosion in ihe air.
Explosion of
confined ips
Explosion ol'
confined teas.
fireball
Explosion
Explosion o!
non-conlined wpor
cloud
Explosion 01
non-confined ea»
cloud
Toxic product
Detonation
Explosion ol'
non-confined
vapour cloud
G-21
-------
xion
N.Y. State
Siful. California
LaBurc. Los
Kentucky
Beriin. NY State
Louisiana
Texas
Texas
Texas
Massachusetts
PJerreUue
Louisiana
Texas
Feyrin
LaSalle
Not specified
Fernhald. Ohio
Santo*
Hawthorne. N.J.
Buenos Aires
Antwerp
Lake Charles
Louisiana
Bankstown. New
South Wales
Perm
Not specified
Pins
Hull
RJukan
Country
USA
USA
USA
USA
USA
USA
USA
USA
USA
USA
France
USA
USA
France
Canada
FRO
USA
Brazil
USA
Argentina
Belgium
USA
Australia
Holland
GDR
France
GB
Norway
Date
1958
1958
1961
1962
1962
1963
1963
1964
1964
1964
1965
1965
1965
1966
1966
1966
1966
1967
1967
1967
1967
1967
1967
1968
1968
1968
1968
196S
Product
involved
Nhro-
mcthane
Not specified
Chlorine
Ethylene
oxide
Propane
Ethylene
Propylene
Ethylene
Ethyl ene
Vinyl
monomere
chloride
Uranium
Hexafluoride
+ fluorising
agent
Ethylene
Propylene
Propane
Styrene
Methane
Uranium
Hexafluoride
Coal gas
Not specified
Propane
Vinyl
roonomcre
chloride
Isobutane
Chlorine
Light
hydrocarbons
Vinyl
chloride
monomere
Petro-
chemical
products
Acetic acid
Gas
Damages
Two hundred injured when a tank wagon detonated and caused a big crater (Cost: 5
million FF).
Two dead when vapor coming from an overflowing tank ignited and ravaged 70 per cent of
the installation.
One dead in a cloud of 27.5 tonnes, released by a tank wagon.
One dead, nine injured. Explosion 'equivalent* to 18 t of TNT.
No records.
Long lasting fire
Fire and explosion in a tow-pressure polymerisation unit for polypropylene (Cost: 30
million FF).
Fire explosion subsequent to an escape of gaseuous ethylene (Cost: 15 million FF).
Two dead in a fire following rupture of a high-pressure ethylene pipeline (Cost: 20 million FF).
Seven flfad. 40 injured. Rupture of an observation window under tension and pressure.
The escaping gas caught fire and exploded (Cost: 25 million FF).
No deaths of injuries or evacuations. Fire subsequent to escape of 300 kg of a mixture
caused by chemical corrosion of a sleeve of a distilling column at theCEA pilot plant. Pilot
run stopped for eight days. Aerosol of hydrofluoric acid carried over 200 metres by the
wind. No external contamination: uranium remained confined in the building natural
uranium).
12 injured by fire and explosion of ethylene from ruptured pipes (Cost: 15 million FF).
Explosion and fire subsequent to pipe break in a propylene polymerisation plant
(polypropylene) (Con: 30 million FF).
16 dead and 63 injured. A valve blocked by freezing during sample-taking from a storage
sphere permitted formation of a gas cloud which exploded near a motorway.
11 dead after explosion following the breaking of an observation window (Cost: 20 million FF).
Three dead, eighty three, injured. Circumstances not specified.
One injured (Six days in hospital). Eight people under observation. Human error: the
maintenance man thought he was unscrewing the top of a valve but unscrewed the valve
itself. Escape of 1 .7 1 of UFfi in vapour. Duration forty minutes (Lead Company of Ohio).
300 injured, eighty buildings of various sizes within a radius of 2 km either destroyed or
damaged.
Two dead, sixteen injured. Building explosion.
100 injured. Fire destroyed four hundred houses in the neighbourhood.
Four dead. 33 injured. Fire lasted three days.
Seven dead. A leaking 10 inch shutter valve released a cloud which exploded. Secondary
fires and explosions continued for two weeks.
Five people intoxicated by evaporations. Evacuation of a large pan of the town.
Two dead. Seventy five injured. Pressure wave broken windows 2 km away.
24 four dead.
400 people evacuated. Explosion shook houses in the neighbourhood.
Two dead, thirteen injured.
Windows of cars and shops shattered. No statement on type of gas or circumstances.
Causes
Detonation
Fireball
Toxic product
Explosion of non-
confined vapour
cloud
Explosion of non-
confined gas cloud
Explosion of
confined vapour
Explosion of
confined vapour
Explosion of
confined gas.
fireball
Explosion of
non-confined
vapour
Explosion of
confined
vapour
Product of of
only chemical
toxicity
Explosion of
confined gas
Explosion of
confined gas and
fireball
Explosion of non-
confined gas cloud
Explosion of
confined vapour
Explosion of
confined gas
Toxic gas +
dangerous
derivatives (HF)
Explosion of
confined gas .
Not specified
Fire
Fire
Explosion of a
non confined gas
cloud
Toxic product
Explosion of non-
confined vapour
cloud
Toxic product
Explosion of
confined vapours
Explosion of con-
fined vapours
Exploaon of
confined vapoun
. \
Annex
•»».
xeuu\
u*
G-22
-------
Location
Soldatna. Alaska
Tamytown
Lievin
Grandes
Armoises
Teeside
Not specified
Puerto la Cruz
Long Beach.
California
Escombreas
Repesa
Crete, Nebraska
Basle
Philadelphia
Osaka
Mitcham,
Surrey
St. Thomas
New Jersey
Port Hudson
Blair.
Nebraska
Crescent City
Emmerich
Not specified
Not specified
Sao Paolo
St. Louis
Virainia
Not specified
Not specified
Country
USA
USA
France
France
GB
Libya
Venez-
uela
USA
USA
Spain
USA
Switzer-
land
USA
Japan
GB
Virgin
Islands
USA
USA
USA
USA
FRG
Holland
Holland
Brazil
USA
USA
Mexico
Date
1968
1968
1968
1969
1969
1969
1969
1969
1969
1969
1969
1969
1970
1970
1970
1970
1970
1970
1970
1970
1971
1971
1972
1972
1972
1972
1972
1973
Product
involved
Pressurised
liquified gas
Propane
Ammonia
Ammonia
Cyclohexane
LNG
Light hydro-
carbon
Mineral Oil
Petroleum
Pressurised
liquified gas
Ammonia
Liquified
nitric product
Petrol
products
Gas
Propane,
Natural
gas
Petroleum
products
Propane
Ammonia
Propane
Not specified
Butadiene
Hydrogen
Gas
Propylene
Gas
Butane
Vinyl
monomer
»I»IA_^.I-
Damages
Two people seriously injured.
3.500 people evacuated.
Explosion of road tanker in the process of unloading. Escape of 19 tonnes. Six dead,
twenty people living in the neighbourhood hospitalised for poisoning.
During transfer of NH j from a fixed to a mobile tank a hose ruptured. Escape of 4 tonnes.
Vegetation burned over a surface of 2.000 x 450 metres. Sixteen cows, one dog and
various chicken killed near living quarters.
Two dead, twenty three injured.
12 injured.
Five dead. Considerable damage to windows and ceiings in town.
One dead, eighty three injured. The cover of a 2.600 1 tank was blown off in a suburban
area.
Four dead, three injured. 5.000 people evacuated. The shock wave broke windows within a
radius of several km.
An escape of liquified propylene gas caused a refinery fire that burned for six days.
Six dead. Escape of 64 t of ammonia from a wagon.
Three dead, twenty eight injured. The pressure shook windows up to 1 km away.
Five dead, twenty seven injured. Explosion in an oil refinery.
•92 dead. Gas explosion on a subway construction site in Osaka.
Substantial destruction of private property in the neighbourhood: roofs cracked, windows
broken, fences overturned, dwellings destroyed by fire, two cars destroyed.
25 injured. The explosion shook practically the whole island.
40 injured. The shock waves shook windows in an area of 150 knr.
No human casualties. Windows were broken up to 18 km away. The derivative cloud was
ignited by an electric motor at a cold storag eunit in its trajectory.
Overflow of a dryogenic tank of 32.000 1 for two and a half hours. Escape of 145 tonnes.
Animals and fish killed. Three foliage burned over 40 hectares of woodland. Low cloud of
2.50 to 9 metres thickness stretching over 365 hectares at 2.500 meters from the tank.
Affected area: one house, one farm; two dogs killed at 1,770 metres distance.
Derailment of a wagon; the commercial centre of the town was destroyed.
Four dead, four injured, many buildings in the area damaged.
38 dead, seventy five injured, five hundred people evacuated; window frames dislocated
within a radius of 15 km. Accident in a 20-story tower.
230 injured: windows damaged up to 3 km from the site of the accident which occurred
during shunting of a wagon.
21 dead, twenty injured, an island pill out of action completely.
350 injured. Thirteen tank wagons of butane exploded.
One dead, sixteen injured.
Seven dead. Vapours released from a reactor exploded. Evacuation of hundreds of people
within a radius of several hundred meters (Cost: 10 million FF).
1.000 people evacuated.
Causes
Not specified
Not specified
Formation of a
toxic aerosol
Formation of a
aerosol
Fireball
Not specified
Not specified
Explosion of
confined vapours
Explosion of non-
confined vapours
Fireball
Toxic product
Detonation
Detonation
Detonation of
confined gas
Explosion of
confined gas
Explosion of
Explosion of non-
confined vapours
Explosion of non-
confined gas
cloud
Toxic product
Explosion of
vapours emitted by
flash fire and
boiling liquified
gas
Explosion of
confined gas
Not specified
Explosion of
confined gas
Explosion of
confined vapours
Explosion of non-
confined vapours
Explosion of
confined gas and
fire
Explosion
Explosion of non-
confined vapours
3
3
a
X
5
a
X
Lodi
chloride
USA 1973 Methanol
Four dead, twenty four injured. The explosion damaged buildings over a large area and Explosion of
w«w in hundreds of windows. Ois were covered with a ihower of debris and crashed by non-confined
enonnoui pteca of concrete. vapours
G-23
-------
Locanoo
GZad&cck/Ruhr
Sheffield
St. Amaadlet
Tokuyama
California
f*/Or**n»
vOlognc
New York
Western
Bohemia
Potchefstroora
Falkirk
Texas
Los Angeles
Beaumont.
Texas
Not specified
Flixborough
Rotterdam
Not specified
Nebraska
Floride
Wenatehce
Not specified
Marseille
Not specified
Antwerp
Philadelphia
Not specified
Oak Ridge
Scvoo
Couaty Date Prodna
*4to«M^^H4
•woncD
FRC 1973 Camcae
CB 1973 Gas from
France 1973 Propane
Japan 1973 Ethytoe
frc« tort v*«ui
Lutn lytJ vmyi
moooor
chloride
FRG 1973 Vinyl
monomer
chloride
USA 1973 Pressurised
liquified gas
Czecho- 1973 Gas
Slovakia
South 1973 Ammonia
Africa
CB 1973 Inflammable
liquid
USA 1974 Isoprese
USA 1974 Organic
peroxides
USA 1974 Isoprene
Czecho- 1974 Ethylene
Slovakia
CB 1974 Cydohexane
Holland 1974 Petro-
chemical
products
Romania 1974 Ethylene
USA 1974 Chlorine
USA 1974 Propane
USA 1974 Monomethyl
amtnotiimte
Holland 1975 Ethylene
France 1975 Petro-
chemical
products
South 1975 Methane
Africa
Belgium 1975 Ethylene
USA 1975 Crude oil
Holland 1975 Propyiene
USA 1976 Uranium '
bexafluoride
Italy 1976 TCDD
Damagn Cams
Fotr dead, two missing, thirty seven injured following the overturning of a truck carrying Not specified
liquified presurised gas-
One dead, four injured. Explosion of non-
confined gas
Explosion of non-
confined gas
Explosion of
non-confined gas
2flO Etre
Four dead. 35 injured. Explosion of
non-confined
gas cloud
One dead, three injured: the explosion broke windows in a large area around the complex. Explosion in §
confined area a
X
Seven dead, seven injured. Whole town gas supply was cut -for two days. Unknown
Six dead. 13 injured. Ethylene escapinj from a compressor exploded, causing extensive Explosion of
damage to buildings and at the plant. ' non-confined gas
cloud
Eight dead, two injured. Vapours from a storage tank exploded in a boiler house when a Explosion of
marine lighter was refuelled. Cost: 50 million FF. confined vapour
14 dead. lot injured. Explosion of
• non-confined gas
cloud
Reaction of UF, and oil from a vacuum pump in a type 30 (2 tonnes) container during Explosion of
transport: ilqutfUFg under pressure. Two injured. Eight days' stoppage. ' UIv* htdro-
cafouu product
Complete evacuation of the area until now (1979). Abortions authorised exceptionally. Aerosol of solid
Deconammatioo made very difficult because of the non-soluble nature of the product. toxic product .'jg
G-24
-------
Location
Country D«te Product
involved
Dtmafes
Closes
Be«k
Baton Rouge,
Louisiana
Sandelfjord
Pierre Benite
Brachead
Mexico City
U mm Said
Mexico City
Not specified
Pierrelaite
Jacksonville
Rockingham
North
Carolina
Gela
Not specified
Not specified
Pierrelatte
Pasacabalo near
Cartagena
Cadarache
Seoul
Los Alfaqun.
South Tarragona
Not specified
Portsmouth
Waverley.
Tennessee
Holland
USA
Norway
France
GB
Mexico
Quaiar
Mexico
Taiwan
France
USA
USA
Italy
India
Italy
France
Columbia
France
South
Korea
Spain
USA
USA
USA
1976
1976
1976
1976
1977
1977
1977
1977
1977
1977
' 1977
1977
1977
1977
1977
1977
1977
1977
1977
1978
1978
1978
1978
Naphta
Chlorine
Inflammable
liquid
Acrolin
Sodium
chloride
Ammonia
Pressurised
liquified gas
Vinyl
• monomor
chloride
Vinyl
monomor
chloride
Uranium
hexafluoride
+ hydro-
fluoric acid
Pressurised
liquified gas
Uranium
hexafluoride
Ethyiene
oxyde
Hydrogen
••
Ethyiene
Uranium
hexafluoride
+• hydro-
fluoric acid
Ammonia
UFfi +
hydrofluoric
acid
Explosives
Propylene
Grain dust
UF, »
hydrofluoric
acid
Propane
14 dead, 30 injured when an escape caught fire. The explosion damaged windows of shops
and houses- Cost: 100 million FF.
10.000 people evacuated. Mississipi banned for navigation over 80 km to the north.
Rupture of underground piping: fire and explosion killing six people and causing 100
million FF damage.
Escape from container of a wagon in the Rhone following human error (21 tonnes*. River
fauna destroyed from Pierre Benite to Vtenne (320 tonnes).
Fire and explosion. Circumstances not specified.
Two dead, 102 people treated for poisoning. Gas entered the drainage system.
Seven dead, many injured. The explosion superficially burned the villages up to 2 km
around. The international airport of Doha was closed for two days.
90 injured. No details.
Six dead. 10 injured.
One dead, nine injured.
Neither deaths nor injuries nor poisoning. Comhurex factory. Subsequent to human error
rapture of a valve in "6 o'clock" position on a type 48 container. Expulsion of 7.1 t of
UF6 liquified under pressure.
2.000 people evacuated.
Derailment of a 29 train. Four type 48 (12 tonnes) containers of UF6 involved in the
accident. Fire of ammonium nitrate, fertiliser and ground nuts. The containers held: no
escape of UF6.
One dead, two injured
20 injured. The explosion shook a fertiliser factory, an oil refinery and a village.
Three dead. 22 injured. Shop windows and doors smashed. Car blown several mores up
into the air.
Neigher deaths nor injuries nor poisoning, f Break of a shutter damp on a tank that was
overfilled wiih UF< under hydrostatic pressure, in the course of warming up. The passage
from the solid to the liquid state causes a volume increase in the order of 25-30 per cent.
Release of 1.200 kg of "natural" L'F6. confined in the building.
30 dead. 22 injured. The villagers in the neighbourhood suffered the effect of the gas. The
installations of the state factory Acobal » ere destroyed at the time of shift change (some
hundred workers were then present) ii has not been proved that NH3 was the cause of the ,
accident but rather a fire.
Neither deaths nor injuries nor poisoning- During warm-up of a thermic trap which was
overfilled with UFg following human error a crack developed in the panioning wall
through hydrostatic pressure which put LF, in contact with cooling fluid and produced an
aerosol of hydrofluoric acid and UO,F, ».nich moved the filters of the extractor fan: two
neighbouring workshops, becoming aepressunesed. were invaded by HF aerosol. Restart
of the workshops after one week. UF6 released: 20 kg. Natural uranium.
58 dead. 1.300 injured. Explosion of tram. Circumstances not specified.
216 dead, many disappeared, several hundred injured following an escape of liquified
propylene under pressure, following road accident of a tanker near a camping site in
summer.
Explosion followed by large fire. Number of dead and injured not specified. Criminal
attack not excluded.
Number of people poisoned not specified No deaths, no injuries. Rupture in the piping of
a hydraulic jack of a lorry which carried a 48G type container (thin panioning wall, not
used in Fiance) which held 9.6 1 of liquified L'F6 under pressure. In the fall of the container •
a cylinder clamp cuased a 19 cm long fissure in the panioning- wall when it struck the
ground violently and expelled the whole LTft content.
12 dead and at least 50 injured when a lank wagon which had derailed exploded.
Explosion of non-
confined vapour
cloud
Toxic product
Fireball.
explosion of
confined vapour
Toxic product
Detonation
Toxic product
Fireball :
i
Unknown
Unknown
Chemically toxic
product:
fluorhydric acid
Not specified
Derailment
Explosion of
confined vapour
Explosion of
confined gas
Explosion of
confined gas cloud
Toxic product:
gasuous hydro-
fluoric acid
Toxic product
Toxic product:
gaseuous acid
hydrofluoric
>
3
3
a
Explosions
Explosion of non-
confined gas
Dust explosion
Toxic product:
aerosol of
hydrofluoric acid
Explosion of
vapour emitted by
flash of pressurised
G-25
-------
Location
lOttOgstQWD*
Florida
Baltimore
ParitfPaisy)
New York
Pierre Bcnite
XUuopcc
Ban try Bay
Warsaw
Islamabad
Couozry
USA
USA
France
USA
France
Mexico
Ireland
Poland
Pakistan
Due
1978
1978
197S
1978
1978
1978
1979
1979
1979
Product
involved
Chlorine
Sulphur
trioxide
Gas
Not specified
Acroltin
Presurised
liquified gas
Hydro-
carbon
vapours
Not specified
Not specified
Damages
Eifht *frt. 10 injured, evacuation of 3.500 people in an area of 10 km2 following spread
of chlorine escaping from a derailed tank wagon. The enquiry concluded that a criminal
attempt was likely.
Toxic fumes drifting up to IS km. Mor ethan 100 people treated fro nausea.
13 dead. 13 injured. GO flats destroyed. Cars damaged by flying debris. Series of explosions
in a building and in underground piping after rupture of a gas pipe.
130 injured: explosion of a deep-freeze lorry near Wall Street.
Escape of some 100 kg of acroleine into the atmosphere. Inconvenience for several
thousand people at Pierre Benite and Oullins (tear gas and nauseous gas).
100 dead. 150 injured. Explosion of a lorry carrying 10.000 litres of LPG in a collision of
12 vehicles on a motorway 85 km north of Mexico City. 85 people died within minutes
from the explosion.
48 disappeared (41 sailors and seven workers) in the explosion of the oil tanker Betelgeuse
at the quayside. The oil tanker was not equipped with an inering system to inject inert gas
as the tanks are emptied. An efficient inverting system only became obligatory when the
"convention of safeguarding human life at sea" of 1974 came into force (only 15
countries, France among them, out of 25 ratified the convention).
41 dead. 77 injured, several hundred people evacuated. Under the effect of the explosion
most of the windows of the neighbouring buildings were sent flying in splinters within a
radius of 200-300 metres. The accident occurred in the basement of the savings bank
building. There were no gas pipes in the building. Welding work was going on in the
basement.
26 dead. 50 injured, several buildings affected by the pressure. Explosions of an anisan
shop in the Raja Bazar at Rawalpindi.
Causes
Toxic product
Toxic product
Explosion of
confined gas
Not specified
Toxic product
Explosion of non-
confined gas
Explosion of gas
in confined
volume, perhaps
preceded by fire
Explosion of
confined gas
probably
Explosion of
instabie solid
chemical
products
Cresrview,
Florida
USA 1979
Ammonia.
Chlorine
4.500 people evacuated within a radius of 2,500 metres. Derailment of a convoy of 28 tank Toxic gases
wagons (NHjC^) on leaving the bridge over the Yellow River.
G-26
-------
A publication of Work On Waste USA, Inc., 82 Judson. Canton, NY 13617 315-379-9200 APRIL 1994
The Health Impacts of Incineration
Excerpts of Testimony
by Barry L. Johnson, Ph.D.
Assistant Surgeon General, Assistant Administrator
Agency for Toxic Substances and Disease Registry, Public Health Service,
U.S. Department of Health and Human Services
Before the
Subcommittee on Human Resources and Intergovernmental Relations
Committee on Government Operations, United States House of Representatives
January 24,1994
[The Subcommittee's transcript of the Hearing on The Health Impacts of Incineration" is not yet available.
To request a copy, call 202^225-2548, and ask to be put on mailing list.]
Part 1 of 2
Good morning. I am Barry Johnson, Ph.D., Assistant Administrator, Agency for Toxic Substances and Disease Registry
(ATSDR). I am accompanied today by Maureen Lichtveld, M.D., Senior Biomedical Officer for Public Health Practice, ATSDR.
We welcome this opportunity to present testimony on the health impacts of incineration: what we know and what we don't.
Our testimony is derived from ATSDR's responsibilities and findings under the Comprehensive Environmental Response
Compensation, and Liability Act, as amended (CERCLA, or Superfund) and the Resource Conservation and Recovery Act
(RCRA, Section 3019).
I will endeavor to respond to the eight issues listed hi your letter of invitation to ATSDR. Because ATSDR is a federal
public health agency, our responses to your issues will be given in a health context. As the Subcommittee knows, there are
many scientific, technology, and policy issues that attend incineration of wastes. Our focus will be only on the public health
issues. But as preface, I can share with you that many communities have expressed concern to ATSDR about the potential
implications of incineration of wastes. Their concerns are usually expressed as health effects questions about then- health. As
this testimony will describe, ATSDR often finds itself unable to answer citizens' questions about associations between
incineration of wastes and public health impacts. A llC SCientlllC IniOTmatlOn
on human health impacts of incineration
isn't often available because the relevant
studies haven't been conducted.
Incineration of wastes should be viewed from a public health perspective in the larger context of generation and
management of wastes. Wastes become a public health concern when they are improperly managed and disposed of. Therefore,
in a public health context, the most protective action is not to produce waste. Waste elimination or minimization comports with
prevention or reduction of health consequences of wastes...
3 What data currently exist on health impacts from incinerator emissions of dioxin, furans,
lead, mercury, and other chemicals you think most relevant? What is the range of health
effects and their intensity at likely emission levels?
There are very few data on the actual human health impacts of incinerator emissions on the health of
communities near incinerators. Epidemiologic investigations have rarely been conducted, nor have studies of
disease and illness patterns been undertaken. For example, ATSDR staff conducted a recent literature
search of the 10 most frequently used computerized environmental data bases. As part of the
G-27
-------
I
search over 1,000,000 entries were identified. Approximately 72*000 of those entries dealt with I
incineration. Of these only one single entry
discussed the conduct of a population-based
study conducted in a community living in
the vicinity of an incinerator. That study of residents living near
Caldwcll Systems Incinerator in North Carolina was conducted by ATSDR. ISeeNotes in WasteNot #277.]
In the absence of human health data, reliance is placed on using toxicity data for individual substances released into the
environment. The effect of any toxic substance depends on factors such as duration of exposure, concentration of the substance in
the environment, biological uptake, and persons' susceptibility factors (e.g., age). AH these factors have to be considered in any
estimate of impact of incinerator emissions...
Adequate information does not exist to support speculation on what, if any, human health effects might be associated with
incinerator emissions. However, our experience with public health associations related to hazardous waste sites would suggest
the need to conduct two kinds of human health investigations. One kind of investigation would look at cancer, birth defect, and
respiratory disease rates in areas thought to be impacted by releases from incinerators. These studies would combine health data
from many geographic areas. A second kind of study would be site specific. Community health surveys would help clarify
whether any unusual exposure or morbidity is occurring 'that might be associated with a given incinerator.
4 What data do you have or gather on additive, multiple, and synergistic impacts when there is
exposure to more than one chemical, as would be the case with incinerator emissions? Do
you expect those impacts would be greater than from single chemical exposure alone?
There are few data available in the scientific literature on
specific interactions of the contaminants that may be released
from waste incinerators (dioxin, furans, lead, mercury), in the
absence of specific studies using combined contaminants, and limited understanding of the mechanisms of actions for some
substances, it is prudent to assume that the effects of exposure to these contaminants is additive.
5 What data exists on the sensitivity of various populations, by age, gender or ethnic
background, to these chemicals?
Infants and children are arguabty the most sensitive segment of the human population to toxic exposures; Infants
and children are at special risk because they play outdoors, they ingest or mouth foreign objects, they are smaller (greater
chemical doses per pound) than adults, they breathe more air (greater volume and breathing rate per pound) than adults, they are,
nutritionally challenged (because of protein-calorie requirements to support rapid physical growth) and they are undergoing
developmental changes that make them especially vulnerable to chemical exposures. Moreover, they have the longest life
expectancies, during which long-term adverse health effects may become manifest. Certain disorders may not become evident
until a child reaches a particular developmental stage, which, may occur long after the damage was done. Some of the largest
cnv ironmcntai health programs (e.g., lead, asbestos) arc directed at children.
People of reproductive age. AH women of reproductive age must be included in this population because the
most severe effects usual ly occur during the very early stages of pregnancy,, often before a woman knows she is pregnant. In
addition, pregnant women, especially those with multiple pregnancies, as well as the developing fetus have increased protein-
calorie requirements to support rapid physical growth.
The developing fetus is particularly sensitive to chemical exposures. Exposure to chemicals has the greatest impact on
those functions undergoing the most active development at the time of exposure. Animal studies and some human studies show
that there are critical fetal developmental stages during which chemical exposure can cause permanent and devastating effects.
There is also a small, but growing, scientific literature that implicates some toxicants as causing effects on male
reproductive processes. For example, laboratory animal studies have shown that exposure to lead causes adverse reproductive
outcomes in male rats, leading to effects on neurologic function in the offspring of the males. Similarly, PCBs in fish and
waterfowl have been reported to cause feminine features in males of these species.
CONTINUED TO WASTE NOT# 277
.WASTE NOT # 2.76. A publication of Work on Waste USA, published 4S times a year. Annual rates are: Croups
•X: Non-ProfitsS5O, Students & Seniors S35; Individual S4O: Consultants & For-ProHts SI25; Canadian SUS50; Overseas S65.
Editors: Elkn .C- Paul Connett. S2Judson Street. Canton. NY 13617. Tel: 315-379-9200. Fax: 315-379-044S.
G-28
-------
A publication of Work On Waste USA. Inc., 82 Judson, Canton, NY 13617 315-379-9200 APRIL 1994
The Health Impacts of Incineration
Excerpts of Testimony
by Barry L. Johnson, Ph.D.
Part 2 of 2
5. (Continued): Elderly persons and persons with chronic illnesses. Elderly persons and the chronically ill tend
to be more susceptible to respiratory irritants. Long-standing public health policies such as immunization guidelines for
influenza support this notion.
The elderly are also nutritionally challenged often due to reduced protein-calorie intake and combined with the metabolic
changes that occur during this life stage. Underlying illnesses such as is the case in the chronically ill may increase their
susceptibility to particular toxicants. For example, persons with chronic diseases of the kidney system may experience more
harmful effects from exposure to renal toxicants such as lead and cadmium compared to a healthy individual.
Moreover, elderly persons and those with chronic illnesses are often socially isolated and potentially less aware of
environmental emergencies. Because of special physical challenges, they may also require special services during time of
evacuation in the event of such an emergency.
Minorities. Preventing adverse health effects in minorities exposed to hazardous substances is a priority for the
Agency for Toxic Substances and Disease Registry (ATSDR). Minority populations, particularly African Americans, Hispanics,
and Native Americans suffer disproportionately from preventable morbidity and mortality. Regardless of income, education, or
geographical locale these populations are often in poorer health than then* white counterparts. This disparity is often associated
with inadequate access to health care for preventive services as well as early diagnosis and treatment of disorders including those
that may be associated with exposure to hazardous substances. Their disadvantaged economic status also frequently affects
priorities on nutritional status. Occupational chemical exposures may increase this population's susceptibility to adverse health
effects resulting from other exposures to hazardous substances. In addition, certain pre-existing genetic disorders (G6-PD
deficiency, sickle cell anemia) may compound the impact of such exposures...
7. What are the most serious data gaps that prevent us from determining the exact health
impacts from incineration?
The data that impede an accurate assessment of the public health impact of incineration can be divided into two categories.
* those associated with the technology and the facility itself
* and those related to environmental health.
Following are examples of some key data gaps in both'categories. Also listed are actions that should be considered in order to
ensure the protection of the public's health. These data gaps and recommended actions are based on ATSDR's experience hi
providing consultations concerning hazardous waste incinerators. Key data gaps associated with the incineration
technology/facility include:
I. The often inadequate identification and quantification of waste feed as well as fugitive emissions associated with
specific incineration facilities.
2. The deposition rates to soil and water for all the potential incinerator stack emissions are not well known.
3. The identification and quantification of emissions during incinerator process upsets are frequently not measured.
4. When stack emissions are analyzed for metals the specific metal compounds or species present are not usually
identified.
5. Concentrations of contaminants in environmental samples around incinerator facilities, e.g., soil, water, and
ambient air are typically not measured.
6. There are limitations in the current stack testing, air monitoring, and air modeling methods. Some of these
methodologies needed further validation.
7. Often there is a lack of data on the concentration of contaminants present in foods that are grown near a facility,
'such as vegetables from gardens, cattle, fish or shellfish, etc.
...The second category of data gaps concerns the area of environmental health. Key data needs in this area include:
* limited demographic and health data on the surrounding community.
* lack of environmental daia such as the types and concentrations of contaminants present and the
G-29
-------
environmental media contaminated.
* limited number of exposure, health monitoring and surveillance activities in communities living near
operating incinerator facilities.
* data gaps in our knowledge about adverse health effects from specific hazardous substances.
* toxicologic data on the mixtures of substances released from incinerators.
Efforts by federal and state environmental and health agencies are underway to address a few number of these data gaps. In addition
to these efforts, attempts should be made to coordinate and collaborate in order to maximize the results in each individual area of
datanmL.
Comments' from Waste Not:
We thought that the testimony of Dr. Barry Johnson, Assistant Surgeon General, on the Health Impacts of Incineration at a
Hearing held by members of the U.S. Congress was significant enough for us to reprint From Dr. Johnson's testimony it is
clear that when the industry says that incineration is a proven technology, they are clearly not referring to the health impacts.
Also, it is important to remember that die U.S. EPA does not have a shred of scientific evidence to support their often repeated
refrain that "the proper operation of a well-designed, well-maintained and properly operated incinerator would not endanger human
health and the environment"
Dr. Johnson refers to the Caldwell Systems hazardous waste incinerator that operated in Lenoir, Caldwell County, N.C.,
from 1978 to 1988. During the time this incinerator operated, citizens bitterly complained to every agency responsible for their
health and environment that the incinerator was making them sick. The reaction to their complaints was that state officials
repeatedly told the community that the Caldwell incinerator operated in accordance with ail state and federal regulations. By the
time the incinerator was shut down, at least five incinerator workers suffered permanent and irreversible brain damage.
Community residents say they have a high rate of cancer and suffered severe respiratory problems while the incinerator operated.
Many of them have now moved away. The group which has the worst health problems are the employees who worked at the
facility. ATSDR conducted a health study of the Lenoir community three years after the incinerator stopped operating.
According to L.C, Coonse, a major watch-dog of the Caldwell incinerator
the ATSDR's health study of residents who
lived near the Caldwell Systems incinerator
in N.C. compared the health impacts by
using, as a control community, another
incinerator community only 7 miles away.
As Dr. Barry Johnson stated, this was the only study available in the ATSDR literature search for health impacts on an
incinerator community. In the "control community1' that ATSDR used, the Broyhill furniture company operates a hazardous
waste incinerator. According to L.C. Coonse, "We would have preferred comparisons to a pristine population" instead of
comparing health impacts to another community where thousands of pounds of volatile organic compounds are burned in the
Broyhill incinerator each year. L.C. told us that he informed ATSDR that a hazardous waste incinerator was operating in the
control community, but ATSDR knowingly dismissed this fact, thus compromising the integrity of the ONE study now
available to current and future researchers. L.C. noted that in ATSDR's first report, the Caldweil Sysiems community, compared
to the control community, "still showed increased respiratory problems."
For more information:
1. Request a copy of ATSDR's report, Studv of Symptom and Disease Prevalence. Caldwell Systems Inc.. Caldwell
Countv. N.C.. January 1993. ATSDR's tel #: 403-639-0700.
2. For the transcript on the hearing, The Health Impacts of Incineration, held on January 24, 1994. call the U.S. Congress,
Subcommittee on Human Resources and Intergovernmental Relations at tel # 202-225-2548. We were told it would
take six months to a year before the transcript is available, but you can call and asked to be placed on the mailing list.
3. See Waste Not # 163. ' - -
4. See Scandal in North Carolina, a 31-minute video produced in July 1990 (before ATSDR considered doing a health
studv in the community). Available forS25 from Video-Active Productions . Rt. 2, Box 322. Canton, New York 13617,
Tel:~3!5-3S6-S797.
5. Contact L.C. Coonse, 71 Pinewood Road., Granite Fails, N.C. 2S630. Tel: 704-396-3288.
33333 JjJ33a33333333333333333a333333333333333333a»3»J33J3333aaa3333a33333333J33JJJ;aa3maj»J»m3J3Ja3J3J3a333333a33333J3333133333
WASTE NOT # 277. A publication of Work on Waste USA, published 48 times a year. Annual rates are: Groups
& Non-Profits S50-. Students & Seniors 535; Individual 540: Consultants & For-Profits 5125; Canadian SUS50; Overseas 565.
Editors: Ellen & Paul Connert, S2Judson Stnet, Canton, NY 13617. Tel: 315-379-9200. Fax: 315-379-0448.
G-30
-------
^Plrnburgh
PENNSYLVANIA
Tri-State Environmental Council
Oaober 10,1994
Rd#lBox365
Chester, WV 26034
Dorothy Patton, PhD
Executive Director and Chair
Risk Assessment Forum
OfBce of Research and Development
U.S. EPA
Washington, DC 20460
RE: U.S. EPA Phase II Risk Assessment for the Waste Technologies Industries (WTI) Toxic Waste
Incinerator
Dear Dr. Patton,
We hope that you will help us obtain information about EPA's Phase II risk assessment for the WTI
incinerator. It is well known that there are insurmountable shortcomings in all risk assessments because
of the enormous uncertainties in the risk assessment process. One such uncertainty is the absence of
information about chemical effects on human health. In his book Calculated Risks, even successful risk
assessor Joseph Rodricks acknowledges how little data is available by stating: "Toxicologists know a great
deal about a few chemicals, a little about many, and next to nothing about most." These uncertainties lead
to the development of assumptions, which are nothing more than formalized guesses to nil in gaps in the
science. Since sets of assumptions can be chosen so as to reach virtually any desired conclusion, the
uncertainties become political opportunities. The risk assessment may become, in borrowed words, "a
pseudo-scientific rationale for a. political decision thathas already been made."
Given our knowledge and understanding of current risk assessment practices, we have many concerns
about the Phase II risk assessment for WTI. We hope that you will be able to secure answers to our
questions and provide us with information regarding the risk assessment. Following are a few of our
questions and concerns:
1. What is the current status of the Phase II risk assessment?
• When can we expect a draft?
• When will it be completed?
2. An August 10,1993 letter from Dorothy Patron, Executive Director of U.S. EPA's Risk Assessment Forum
states that in die second part of me peer review process, EPA will convene a peer review workshop of
up to 15 experts to review the draft.
• Will the citizens have an opportunity to nominate peer reviewers or have any input?
• When will the workshop be held?
G-31
-------
3. Exacfly what will be considered in the Phase n risk assessment?
•• Non-cancer health effects such as effects on* the endc>ainerreproduarver immune and
neurological systems?
How will me risk assessment address emissions of chemicals such as cadminum, dioxin,
lead, mercury, PBBs, PCBs, pentachtorophenol, styrenes, hexachlorobenzene, phenol,
furans etc., mat are reported to have reproductive and endocrine-disrupting effects??
(Because effects are being seen at or near existing background levels, current thinking is that
we must limit emissions of these substances to zero. Any additional release of these types of
substances into the environment is unacceptable.)
• Actual exposures in various distinct situations? Other factors during which time individuals
may receive higher exposures such as spills, leaky valves, equipment maintenance
procedures, planned or unplanned start ups and shut downs, automatic waste feed cut ofis,
etc? Effects of fires, accidents and explosions occurring at the plant'
• Hazards to humans and animals varying greatly in genetic characteristics, age, sensitivity,
and pre-existing health conditions? Exposure and effect on infants, small children, the
elderly, and people with chronic illness such as asthma, emphysema, heart conditions, etc?
Effects on the fetus and consideration that infants and fetuses are generally more susceptible
to toxic effects of chemicals than adults.
• Additive, cumulative and synergistic hazards of daily exposure to WIT emissions?
Simultaneous exposure to other pollutants in the Ohio Valley environment? The ability of
many chemicals to enhance or amplify (promote) the effect of past exposure to carcinogens?
• Relationship between dose and response at low level exposure?
• Effects from substances notyet studied, ie., products of incomplete combustion (PICs) that
have never even been identified? (Between 20-70% of the PICs have never been identified,
let alone what effects they have on human health and the environment)
4.. In order to keep the assumptions conservative, will the emission levels in the risk assessment
equal the permitted levels inthe permit for each substance?:..-.-
In light of the fact that we will be exposed to WITs toxic emissions on a continuous basis, without our
consent, we believe that these are reasonable questions to which we deserve an answer. We sincerely
appreciate your help and efforts to secure the answers for us. We look forward to receiving the
information.
Respectfully,
G-32
-------
OHIO
Cleveland
E. LIVERPOOL
5flE
PENNSYLVANIA
Tri-State Environmental Council
October 11,1994
Rd#lBox365
Chester, WV 26034
/ ' -
William Snyder
Chief Fluid Modeling Branch
U.S. EPA
Research Triangle Park, North Carolina 27711
Dear Mr. Snyder,
We are in receipt of a copy of the video tape of the wind-tunnel model of the WTI incinerator. Have you
conducted an experiment using the same WTI wind-tunnel model for a ground level release as a result of
a mixing accident or spill involving a volatile substance? We request that you conduct an analysis for a
ground level release during an accident or explosion involving a substance such as hydrogen cyanide or
hydrogen sulfide (or a combination of substances) to determine if the concentration would be such that it
would be a killing concentration for the children at the school or in their homes nearby.
Given the extremely sensitive location of the WTI incinerator, and the nature and volume of toxic waste to
be stored and treated on site, we cannot afford to ignore the probability (a statistical determination based
on an actuarial study or actual experience) of fires, explosions and accidents at the site. In feet, the history
of the incineration industry proves that these are. common at these facilities. What is unique about the WTI
toxic waste incinerator is that it is located in the middle of a residential neighborhood and parked next
door to a school. To protect the children, we must consider the worst case scenario for an accident at WTI.
We want to know what would happen as a result of an accident involving the release of poisonous gases
such as phosgene, methylisocyanate or hydrogen cyanide. We also need to know how likely it is that such
an accident will occur over the next 20 years?? (What is the probability of a serious accident at WTI over
the next 20 years?)
In light of the fact that you already have the WTI wind-tunnel model built, we hope that you will conduct
the analysis we have suggested. You owe it to the children and their parents to do so.
We look forward to the results of your study, and express our sincere appreciation in advance.
Sincerely,
<•'
Perri Swearingen
cc: Senators
Representatives
G-33
-------
-------
OHIO
E. LIVERPOOL
SHE
PENNSYLVANIA
Qevetand ^^
Tri-State Environmental Council
^Pittsburgh
December 13,1994
Rd #1 Box 365
Chester, WV 26034
IBYFAX 3^op.M. I
CERTIFIED MAIL TO:
Dorothy Cantor, PhD
Chair, WTI Technical Workgroup
Office of Solid Waste and Emergency Response
U.S. EPA
Washington, DC 20460
RE: WTI Wind Tunnel Study
Dear Dr. Cantor,
Thank you for your November 21,1994 response to our letters concerning the WTI Risk Assessment and the wind
tunnel study. This follow-up letter concerns the WTI wind tunnel study.
Today's newspaper headlines in Arkansas concern the explosion and subsequent fire at the Ensco toxic waste
incinerator in El Dorado that resulted in at least three immediate injuries. According to the El Dorado, Arkansas
News-Times, "An explosion at the Ensco hazardous waste incinerator rocked houses in El Dorado and southeast
Union County and was heard as far away as Farnerville, LA." One witness to the explosion said, "It mushroomed just
like a hydrogen bomb. If then wasn't anyone hurt, it was a miracle." The explosion, which reportedly occurred in the rotary
kiln unit in the waste feed system, was heard as far as 35 miles away. Residents who live nearby said the explosion
violently shook houses, knocking pictures off of walls and items off of shelves. The cause of the accident and the
extent of the damage have yet to be determined.
Today's newspaper headlines in Iowa concern an explosion at a farm chemical plant that produces nitrogen-based
fertilizers near Sioux City, Iowa. According to Associated Press (AP) reports, "An explosion rocked a farm
chemical plant south of Sioux City today, killing at least five people and rupturing huge ammonia tanks. Hundreds
of people were evacuated." Although it has yet to be confirmed, it is believed that there are at least five dead and
17 injured. The explosion knocked out four nearby electricity generating stations, ruptured two ammonia tanks
capable of holding one million gallons, and sent up a'cloud of ammonia gas. Hundreds of people were evacuated
from an Indian casino about 10 miles away, and the 500 people in Homer, Nebraska, to the south, also were
evacuated. The cause of the accident and the extent of the damage have yet to be determined.
In our letter of October 11, 1994 to which you responded, we requested that the EPA conduct an analysis for just
such an accident or explosion causing a ground level release of poisonous gases such as phosgene,
methylisocyanate, hydrogen cyanide or hydrogen sulfide at the WTI facility. As we stated in our first letter, we also
need to know the probability of such an accident at WTI over the next 20 years.
G-35
-------
We have several comments and questions relative to your response. We would very much appreciate answers to
each specific question.
1. What are the chances during the next 20 years for the occurrence of a serious accident involving the release of
poisonous gas at WTI?
2. Does the EPA believe that it would be a beneficial and worthwhile effort to conduct a wind tunnel analysis to
determine the impact of a ground level release accident as described above?
3. We appreciate the fact that you have said that a scale ratio of 1 to 100 would be more appropriate for the
analysis of an accidental ground level release than the scale of the existing wind tunnel model which is 1 to 480.
But would it not be possible to do a first level approximation or a preliminary screening analysis using the
existing model to get some indication of the impact of such an accident? Wouldn't it be more scientific to
conduct such a study first before making a sweeping judgment that it could not be done using the existing '
model? The preliminary analysis may raise a red flag signaling the need for further study using a more accurate
scale ratio.
In rationalizing why EPA is not doing the ground level release study, you also indicated that the model is at the
wrong wind direction. Why not just rotate the model? There is no unidirectional aspect unless the original was
a very crude model that did not account for all the hills. Why can't the model simply be rotated to correct for
wind direction? It would seem that with the use of computers, the entire model could easily be transferred
from one scale to another, and wind direction and speed could be adjusted. The difficult part of the task has
been completed. Now just repeat it on a scale of 1 to 100.
4. Obviously it suits WTI's purposes for EPA to focus entirely on stack emission numbers, which can easily be
manipulated. But look at todays headlines! Our research of the track record and history of incinerators,
including those similar to WTI, indicate that fires, explosions and accidents at these facilities are common, and
that accidental releases pose a greater threat to public health than routine emissions. Has the possibility of using
a model for accidental releases been discussed or rejected? What would the cost of such a study be? How much
would it cost to do it right? Considering what is at stake, can we afford NOT to conduct such a study?
We are interested in very specific answers to each of the questions we have raised. We look forward to your
response.
Sincerely,
cq Senators, Congressmen
G-36
-------
OHIO
Cleveland
LUVEWCOL
Sat
^RtnbwQh
Tri-State Environmental Council
January 3,1995
Rd#lBox365
Chester, WV 26034
BY FAX AND MAIL
Dorothy Canter, PhD
Chair, WT1 Technical Workgroup
Office of Solid Waste and Emergency Response
U.S.EPA
Washington, DC 20460
RE: Waste Technologies Industries (WTI) Risk Assessment
Dear Dr. Canter,
This is a follow-up to our October 10 letter and your November 21, 1994 reply. We are disturbed by your
response, including the fact that you chose not to answer all of our questions. Given what is at stake here, we
have a. right to be concerned. We are not trying to be difficult or cause trouble — we just want to be sure our
children and our families are not at risk from WTL Following are additional questions, comments and
concerns raised by your response.
1. Why is WTI allowed to operate prior to die completion of the Phase II risk assessment?
2. Will the risk assessment consider the psychological impact on the children who attend school and live
beside WTI, or the psychological impact on their parents?
3. How much money has the U.S. EPA spent to date on WTI risk assessment activities, including .but not
limited to the construction, and analysis of a WTI wind tunnel model at RTF; activities related to the Risk
Assessment Forum and peer review process; all meetings and workgroups conducted for the WTI risk
assessment; all telephone communications; payments for services of A.T. Keamay, Environ or any other
consultants; etc.?
4. When it is available, will you please send a copy pf the draft Phase II risk assessment report that is currently
expected to be distributed to interested parties in Spring 1995?
5. Will you please notify us of the date when the external peer review panel will hold the open workshop to
review the report?
G-37
-------
6. In your November 21 response, you state that "both cancer and non-cancer health effects of chemicals of concern an
being evaluated in the Phase H Risk Assessment." Please identify which chemicals are considered to be "chemicals
of concern" and specifically, which chemicals will be considered in the WIT risk assessment?
7. You failed to answer our previous question concerning endocrine disrupters. How will the risk assessment
address emissions of chemicals such as cadmium, dioxin, lead, mercury, PBBs, PCBs, pentachlorophenol,
styrenes, hexachlorobenzene, phenol, furans, etc., that are reported to have reproductive and endocrine-
disrupting effects??
8. How will the results of EPA's dioxin reassessment affect the WTI risk assessment, especially since it has
been reported that the adverse effects of dioxin are being seen at or near current background levels?
9. We are aware that certain chemicals have toxicity profiles but do not have a R£D. You state that lead is
one such substance for which there is no threshold, but a computerized model has been developed for
establishing potential risks from multiple source environmental lead exposure. This model will be Used to
estimate what impact exposure to lead emissions from WTI will have on the children around the facility. Is
it possible to develop a similar computer model for some of the other chemicals of concern, including the
endocrine disrupters?
10. You state that "routine non-stack releases" wfll be included in the risk assessment Does this include fugitive
emissions of chemicals from Storage and hatifTKng?
11. In your November 21 response you state, "Chemicalsfor which vetifieddata exists only on cancer effects wilt be
evaluated for cancer effects but not for non-cancer toxidties. likewise, chemicals for which verified data exist only on non-cancer
toxic endpoints will be evaluated only for those endpoints. Also chemicals for which the agency does not have any verified toxicity
data will not be evaluated in the risk assessment." You further state, "EPA will not be evaluatingpotential synergstic •
hazards because the current scientific data base on sjnergstic effects (or on antagonistic effects) of exposure to multiple
contaminants is not robust enough to determine if such effects are occurring" Are we to understand that if the EPA does
.not have verified data on a chemical then you assume the risk from that chemical is zero? If you do not
evaluate the chemical in the risk assessment then you are assuming the risk is zero. In addition to the
chemicals for which EPA does not have verified toxicity data, what about the 20 to 70% of the PICs that
have not even been identified yet? We must assume that at least some of the unidentified substances are at
least as toxic as dioxin. Doesn't this risk assessment method leave out many risk factors? How can you say
that EPA is using conservative assumptions when you are ignoring so many risk factors?
12. What potential products of incomplete combustion (PICs) were present in the stack samples during the
February 1994 emissions testing at WTI, and at what levels were they present?
13. You stated that in August 1994 the facility tested stack emissions for the presence of more than 80 possible
organic PICs during their performance test Please identify which PICs were present and in what quantity.
14. Please provide the PIC emission estimates that have been developed for contaminants of concern based
on the expected combustion efficiency of the facility and the known waste feeds from the first year of
operation. When completed, please also send the comparison between the measured PIC emission rates
and the predicted value.
G-38
-------
15. You failed to answer our previous question concerning permit limits for rlymirals evaluated in the WTI
risk assessment. To keep the assumptions conservative, will the emission levels used in the risk assessment
be incorporated into WTI's permit as the emission limit for each substance?
We really hope that you will take the time to answer these important questions. You just don't know how much
we appreciate honest, complete and straightforward answers. We hope we r?n count on you to provide them.
Sincerely,
Terri Swearingen
cc: U.S. EPA Administrator Carol Browner
U.S. EPA Deputy Administrator Fred Hansen
U.S. EPA Assistant Administrator Elliott P. Laws
Senators
Congressmen
G-39
-------
-------
Qevetandl
Tri-State Environmental Council
January 3,1995
Rd #1 Box 365
Chester, WV 26034
BY FAX AND MAIL
Carol Browner
Administrator
U.S. EPA
401 M Street, SW
Washington, DC 20469
RE: U.S. EPA Risk Assessment Activities for Von Rofl/WTI Toxic Waste Incinerator,
East Liverpool, Ohio
Dear Ms. Browner,
We have some serious concerns to which we would like you to respond. On October 10,1994, we wrote to
the U.S. EPA regarding the Von Roll/WTI Phase II Risk Assessment. Nearly six weeks later on November 21,
1994, the Chair of the WTI Technical Workgroup, Dr. Dorothy Canter, responded. We are distraught with the
quality of EPA's reply. After reading Dr. Canter's response, it is evident that in conducting a risk assessment
EPA has thrown us a 20 foot rope over the edge of cliff when they are fully aware that we are 25 feet away! In
our October 10 letter, we asked very direct questions. Dr. Canter's reply did not 'answer some of our most
important questions, and she failed entirely to address our concerns about endocrine disrupters. The document
she sent was completely oblivious to all of the new information on environmental hormones. It is dreadful that
EPA has provided such a shoddy answer. Dr. Canter's response was an insult to our intelligence. The same
response with an explanation of RFDs and cancer potencies alone might have been acceptable three or four
years ago. It is not acceptable today with our current knowledge and understanding of the effects of
environmental hormones. Dr. Canter totally ignored our specific questions related to endocrine disrupters,
which are some of our greatest concerns.
After reading Dr. Canter's response, we question whether EPA is making a good faith effort to determine the
real risk from WTI. In her reply, Dr. Canter reveals that:
• EPA will not evaluate effects of chemicals for which the agency does not have verified data.
• EPA will not evaluate non-cancer toxicity of chemicals for which verified data exists only for cancer
effects.
« EPA will not evaluate cancer effects of chemicals for which verified data exists only for non-cancer effects.
» EPA will not evaluate synergistic hazards of exposure to multiple contaminants.
G-41
-------
• EPA will not evaluate antagonistic effects of multiple contaminants.
As Dr. Canter states, "the current scientific data base... is not robust enough to determine if such effects are occurring". La
essence, if EPA does not know about the chemicals in question, they just assume the risk is zero! EPA's
solution to their lack of knowledge on many chemicals is to ignore them!
It appears that the EPA is spending taxpayer money to complete a risk assessment only to prove the facility .is
safe rather than to prove it is not safe. The EPA's complicated risk assessment attempts to provide a veneer
of science while creating a huge bog of impenetrable material in order to obscure the obvious. The EPA is
going to enormous effort and expense to protect this permit. With the kind of taxpayer dollars EPA is
spending on risk assessment activities to justify WTI, we at least expected specific answers to our questions.
WHY is the EPA spending millions of American taxpayer dollars to promote die interests of a foreign owned
hazardous waste company that is being investigated by the SEC and FBI, and whose executives are under
indictment by their own government?
The public needs to know whose side the EPA is on. Is the EPA on the side of the American taxpayer, or
is it on the side of this perverted foreign corporation that is currently under investigation for corrupt and
illegal practices involving organized crime?
Frankly, we are baffled! We hope that you will address these issues, and answer each specific question::
Attached also is our most recent correspondence to Dr. Canter. We would be grateful if you would see that we
get a satisfactory response. Given what is at stake herej we are gravely concerned. We are not trying to be
difficult or cause trouble — we just want to protect our children and our families. We truly look forward to
your response.
Sincerely,
Terri Swearingen
cc: Senators, Congressmen
Enclosures:
Wall Street Journal, December 2, 1994: Von Roll Finds Fresh Headaches as SEC Investigates Unit's
Alleged Mafia T.fnks
New York Newsday, December 30,1994: Did Swiss Firm Pay Off Mob?
The Daily Times, December 16,1994, Editorial: Doubts Cast on Fitness of WTI Ownership
G-42
-------
LLJ
DQ
>.
LUQC
O
Q.
cr
a.
z
O
LU
LU
Z
52
o
LJJ
oc
O
DC
U LLJ
O O
r Z
O LU
O
O,
CC <
CO
O
LU N
I <
G-43
-------
THE GOVERNING LAW:
Ohio Revised Code Section
3734.05(D)(6): "The Board shall
not approve an application for a
hazardous facility installation and
operation permit unless it finds and
determines: . . . (g) that the active
areas within a new hazardous waste
facility ... are not located or
operated within:
(i) two thousand feet of any
residence, school, hospital, jail, or
prison;
(ii) any naturally occurring,
wetland; or
(iii) any flood hazard area if
the applicant cannot show that the
facility will be designed, con-
structed, operated, and maintained
to prevent washout by a one
hundred-year flood or that pro-
cedures will be in effect to remove
the waste before flood waters can
reach it.*'
— Effective August 1, 1984
WHY DOES THE STATE
OF OHIO IGNORE ITS
OWN LAWS?
"[The investigative report of the
Ohio Attorney General released in
June, 1993] concludes that WTI's
changes of ownership have resulted
in unlawful installation and opera-
tion of the facility by the current
owner, in violation of the three
Ohio provisions of law which pro-
hibit ownership and operation with
out a permit." — Jack Van Kley,
Environmental Enforcement Sec-
tion, Ohio Attorney General's Of-
fice, Columbus, Ohio, on Septem-
ber 22, 1993.
"[T]he WTI facility is the worse
siting decision I have seen in my
twenty-five years of practice in pub-
lic health. Locating a major hazar-
dous waste incinerator 300 feet
from the nearest residence and 1100
feet from an elementary school with
400 children . . . amounts to ad-
ministrative incompetence if not
malfeasance in office . .' . and does
violence to common sense."
— Dr. David Ozonoff
G-44
-------
This page intentionally left blank.
G-45
-------
Former Dragon faring well at Panola Junior
E1D(
NEWS-
Volume 106, Number 160
El Dorado,
One hurt in Ensco
ByKULFERGUSOH
An explosion at the Easco haz-
ardous waste plant Monday
night rocked houses in El
Dorado and southeast Union
County and was heard as far
away as Faraerville, La., but
only injured one plant employ-
ee.
The plant, located on Ameri-
can Road Just outside the El
Dorado city limits, was
swarmed by emergency person-
nel from around the county
shortly after the 7:53 p.m. explo-
sion. El Dorado police, the
Union County Sheriffs Office,
the El Dorado Fire Department
and the Arkansas State Police '
responded to the incident
"It mushroomed just like a hy-
drogen bomb. If there wasn't no-
body hurt, it was a miracle.'1 •
one witness to the explosion
said.
Christy Ibert, the plant em-
ployee injured in the blast, was
being treated in the emergency
area of the Medical Center of
South Arkansas for burns at
10:15 p.m., a hospital spokes-
man said. The spokesman said
Ibert was in stable condition.
but did, not know if she would
be'admitted to the hospital
Ibert, according to sheriff's
deputies, was in the plant con-
trol room at the time of the
blast. The control room is near
where the explosion occurred,
according to deputies.
A state police spokesman said
about two hours after the blast
that there appeared to be no
.danger to anyone in surround-
ing areas and no evacuation
was carried out or planned, ac-
cording to the Associated Press.
"State police (at the site} said
that the threat of problems has
been eliminated," state police
spokesman Wayne Jordan said
at 10'pjn. "The lire is probably
out to the extent it's not going to
cause any danger,"
Union County Sheriff Huey
Hatard echoed that judgment
"At this time, there's no dan-
ger, there's no toxics in the air,
there's nothing been released,"
Havard said.
Operations at the incinerator
had been shut down, and "we.
have been informed there's no
danger whatsoever, at this
time." '
The sheriff said the explosion
was heard as far as 30 or 35
miles from El Dorado.
Jordan said a trooper at the
scene reported that acetylene
gas tanks "that are 'used to Ore
the burning process exploded."
"There are still oxidizers
burning." he said about 1H
hours after the blast
State -Trooper Roland Pon-
thieux, a hazardous-materials
expert, responded to the scene,
Jordan said.
He said Ponthieux reported
that the oxidizers that were
burning were used in combina-
tion with the acetylene to cre-
ate a hot fire in the incinerator,
but he didn't know what the
oxidizers were.
Sgt. David Smith of the El
Dorado Police Department said
authorities planned to "try and .
let it burn oaf before moving
in closer to the explosion
scene. He said the department
was deluged with calls right af-
ter the explosion.
Plant general manager Steve
Darnell said the explosion oc-
curred in the rotary kiln unit of
the waste feed system. The sys-
tem was described -by another
cmployec-as the part that
grinds up waste before it is in-
cinerated.
. • Darnell Was interviewed at
the plant's front gate about 9:45
The general manager said "or-
ganic-type wastes" were being
burned at the .time of the blast.
Several small fires burned into
the night He said an explosion
"fail-safe" system prevented
any chemical release and chan-
neled the explosive power up-
ward and away from areas
where employees were located.
"There was no release of any
Blast site
Shortly after the explosion, siw
chemical fumes from the plant
and there was no danger to the
community," Darnell said, "Our
first concern was to make sure
all of our employees were out of
the area and the area was se-
cured."
Darnell added that an inspec-
tor (com the state Department
of Pollution Control and Ecol-
tOOBl
(TIH S'.SIilKS
8r6C C98 T09 XVH SZ-SI 96/80/10
G-46
-------
College, 8A
December 13,1994
51 ado
TIMES
Arkansas
28 Pages In 2 Sections - 35 cents
Cloudy
Details, 2A
plant explosion
ake rises (center) from the blast site, to the right of ttfe taller stick, at the Ensco plant
ogy was on the scene. • was mixed into the .system. we'll be on the scene for a
.He said that as soon as the es^- "I thought so'meone had while."
plosion occurred, plant person- picked the Building up and Havatd, who was at the Union
nd immediately, put ,£be com- moved it," tic employee-said of County Criminal Justice Facil-
pany's emergency plan into bis station several hundred feet ity just one block, away at the
operation. ' ' from the rotary kiln. • time of the blast, said "I
Another plant employee, on El Dorado Fire Chief Ben thought they had blown the side
site at. the plant, said he had Blankenship said "at this time of the jail oft"
been told the incident hap- it appears to be stabilized.
pened when too much oxygen We're working with Ensco, bat See ENSCO, Page 3A
G-47
-------
.5
00
^ 3 2i i SIN
"ISJiiS*!
-------
SrSooi." g^3 .2*3§-ae
IMss^lsj-sils
-------
APPENDIX H
WRITTEN STATEMENT SENT AFTER THE WORKSHOP
The following statement was sent by a member of the public who was unable to attend
the workshop due to severe weather conditions along the East Coast.
-------
Testimony of Alonzo Spencer, President
Save Our County, Inc.
P.O. Box 1242
East Liverpool, Ohio
Before the WTI Risk Assessment Peer Review
January 11 and 12,1996
Georgetown, Washington D.C.
H-l
-------
-------
Mr. Chairman, Members of the Committee;
My name is Alonzo Spencer. I am President of Save Our County, Inc., a local
grassroot environmental organization, located in Bast Liverpool, Ohio.
For almost fifteen years, we have been struggling to protect our community and our
people from the deadly threat presented to us by the construction and operation of the
Waste Technologies Industries (WTJ) Hazardous Waste Incinerator located in our
community.
All across the country low income populations and communities of color have been
targeted for the location of hazardous waste facilities. While such facilities frequently
arrive promising jobs and prosperity to a community, what they really bring is a threat to
the health, environment and safety of families that live in these areas.
The main reason for addressing you today is to inform you that the EPA maybe using
you to try to justify that which is unjustifiable.
The EPA, thus far, has been unable to produce evidence that can be used to justify
allowing WTI/Von Roll to be built, let alone go into limited commercial operation. They
have; failed from a legal, moral or ethical position, and now they are turning to the
scientific community seeking help to make that which is unholy to somehow become
holy.
The WTI case suggests something of a noble lie in our society. We have even a
scheme to transfer R.C.RA permits without authorization from the EPA, to transfer
operational control of a facility without authorization from EPA, to handle hazardous waste
without a R.C.R.A. permit, to fail to disclose one of their corporate family member's
environmental compliance record and connections to organized crime, to approve a
R.C.R.A. permit without the landowner's signature, and to add the landowner's signature
despite the landowner's objection.
EPA officials have played a significant role in the irregularities in the WTI permitting
process. The culpability of particular individuals must be addressed for allowing this plant
to be built on the banks of the Ohio River, a source of drinking water for literally
thousands of families. For knowingly building it in a flood plain, over two high yielding
aquifers, one being contaminated by a previous spill of toluene, ethylbenzene and xylene.
To this day no remediation has taken place to clear up that contamination. It should also
be taken into account that this plant is within a few hundred feet of the nearest resident.
The plant is located in an area that experiences air inversions approximately once every
six days.
H-3
-------
The most astonishing collateral effect of this environmental misconduct is the fact
that this plant sits less than eleven hundred feet from an elementary school, kindergarten
through fifth grade. Those children are exposed to the dangers of this plant from stack
emissions and accidents threatening their lives each and everyday they attend school.
We now have the EPA .turning to risk assessments for some justifications for its
actions. Let me give you a brief history of the origin of risk assessments.
HISTORY - a reminder; Risk assessment began in secret in the late 1940's. When
atmospheric testing of nuclear bombs became more and more intensive in the 1950's
there was a build-up of radioactive strontium (and other radioactive elements) in the
atmosphere of the northern hemisphere. Strontiuin-90 and company were being washed
down in rain and drifting down in dew. They entered the food chain. In the early 1950's
(reference Merrill Eisenbud) some scientists in New York City began to worry. They
went to the market and bought some lamb chops and some Muenster cheese, ran them
through' then* still somewhat rudimentary analytical equipment and found that they
contained radioactivity. The U.S. Atomic Energy Commission embarked on the most
expensive and extensive monitoring study ever launched to that time, to find out the
extent of strontium contamination of the bones of people. WHY? Because strontium was
chemically similar to calcium and of course it would go into the skeleton of animals and
people along with the calcium in food. There was also the very disquieting knowledge that
radioactivity in human and animal bones caused cancer.
This was knowledge then 25 years old. Many of the women who had painted watch
dials with radioactive radium paint and tipped their paintbrushes in their mouths had been
slowly dying from the early 1920's and still developing cancers in the 1950's. The animal
radium experiments of the previous 25 years had shown the same thing. The relationship
between the amount of radium and the dose of radiation to the skeleton and the
appearance of bone cancer and other forms of cancer was well known.
The strontium levels in children's skeletons particularly, were obviously increasing
as the Cold War grew worse. Was there a way to get a risk assessment (a term not yet
invented) on the number of cancers that might result from the radioactive strontium in
human bones? How high could the level go? How many cancers could result? This was
a period of high secrecy - the public afraid but unaware of the research efforts about
radioactive fallout for years.
The answer to the questions was to use the data from the radium dial painters
(information still in use to this day) and from old and new animal radium experiments
added to new experiments on radioactive strontium in animals. The comparison could
then be made between the radiation dose, relationship between human and animal data for
radium with animal data for radiation dose from strontium - then there was only one
H-4
-------
unknown in the equation - the effect of the radioactive strontium dose on human
populations. Enough money was spent, enough scientific observation went into that
equation to provide a reliable cancer risk assessment for radioactive strontium in fallout
The studies also contributed to the evidence for the halting of Russian, British and
American atmospheric testing. Strontium-90 did not reach dangerous levels in human
populations, remaining below the level of natural radioactivity that has always been part of
the human skeleton. That experience showed mat with a reliable scientific foundation,
risk assessment for human cancer was appropriately done and appropriately used.
How does this experience relate to the effluent from the WIT Hazardous Waste
operation.
FOR CANCER RISK ASSESSMENT PURPOSES DO WE HAVE? A human
population under surveillance for a period long enough to know the cancer rate in relation
to their exposure to a single component of the effluent? Dioxin? No. Any other
component of the effluent? Only Radioactive elements. Conclusion - Except for
radioactive effluent, assumptions based on animal studies of varying and unpredictable
chemical mixtures from the WTI Hazardous Waste plant, do not provide a useful scientific
basis for human cancer risk assessment.
IS RISK ASSESSMENT RELIABLE FOR DISEASES OTHER THAN CANCER?
Respiratory disease - The long history of acute, high level exposure to factory and power
plant exposures associated with immediate death and long-term chronic respiratory
disease - e.g. Donora, Pa., (1948), Meuse Valley, Belgium (1934), London, England (1950)
etc where air inversions were the precipitating events have all had exposures to
unidentifiable mixtures of hazardous chemicals.
There is some reliable prevalence data for chronic respiratory disease hi human
populations, good animal data for exposure to specific respirable chemical compounds.
But there is no reliable way to compare a dose response relationship as in the
radium/strontium relationship with variable, non-reproducible mixtures of chemicals
coming from the WTI stack. The assumptions to be added to any model of risk
assessment for the latter situation are imaginary and unscientific.
There is no available comparison data for assessing the effects of acute exposures on
the chronically 511 population particularly those with chronic respiratory disease and the
newborn (Harrisburg, 1968)
Reproductive Effects -.The reliable scientific foundation for developing a risk
assessment model for the multitude of reproductive effects does not exist in terms of
human/animal comparison for most specific chemical compounds and exists not at all for
variable mixtures of compounds in effluents under varying climatic conditions and
H-5
-------
operating conditions.
Immune diseases - No reliable epidemiologic data that can be applied to a
scientifically based risk assessment process.
CONCLUSION: The further we move from the firm scientific foundation established
for the early development of risk assessment methods for radioactive fallout, the weaker
the conclusions become for providing a reliable and credible understanding for the public,
for regulators, for operators of hazardous waste plants. .
The most serious problem of all and the most serious and unpredictable danger we
have is NEVER knowing what is coming out of the WI1 stack. What is the use of a risk
assessment using as many imaginary assumptions as any hysterical neighbor can dream
about when the next Donora type inversion comes our way. When that happens as it
inevitably will there will be no analytical data available to describe what is being added to
the atmospheric soup from the WTI stack. There will be no monitoring on the synergistic
effect of the mix of hundreds of chemicals coming out of the stack right next to the homes
and the school
To use risk assessment judgments for policy decisions concerning the WTI plant flies
in the face of rational scientific conclusions.
H-6 * U.S. GOVERNMENT PRINTING OFFICE: 1996 - 750-001/41006
------- |