FILE
           UNITED STATES ENVIRONMENTAL PROTECTION AGENCY
                      WASHINGTON. D.C. 204SQ

'December 15, 1989                       EPA-SAB-EC-tO-004


Honorable William K. Reilly
Administrator
U.S* Environmental Protection Agency
401 M street, S»W,
Washington, D.C.  20460


                              OSW's Proposed Controls  for
                              Hazardous Waste Incinerators
                              Products of


Dear Mr. Reilly:

     The Science Advisory Board's Products of Incomplete Combustion
Subcommittee has reviewed the feroposed Controls for. Hazardous Waste
Incinerators^Products of Incomplete Co.mb_us_tion  fPICs)  for  the
Office of Solid Waste.   The Office  of Solid Wast® seeks to propose
and promulgate rules which amend existing standards fo? boilers and
industrial  furnaces  burning hazardous wastes fuels  and  for  all
hazardous waste incinerators.  The Science Advisory Board was asked
to address three questions:

     l.   Whether   limiting  carbon  monoxide   (CO)   and  total
          hydrocarbons  {THC)  is  a  reasonable  approach  to control
          emissions  of PICs,  given the current  data base  and
          statutory time constraints?

     2.   What  is the  feasibility  of  monitoring THC  to determine
          to the aggregate  emission rate of organic compounds?

     3.   Whether the proposed approach to assess the health
          risk  from THC emissions  is reasonable given the
          current data base and statutory time constraints?

     To  address these  questions,  the Subcommittee reviewed  the
documentation provided on  PICs  controls and held open meetings
December  14 and IS,  1988,  and  January  26  and 27,  1989,  in
Washington,  D.C.   A  final publicly  announced  meeting  vas held
September 15, 1989 by conference call.

     The  Subcommittee  also addressed;   atmospheric  dispersion
simulation, selection of CO  and THC concentration limits, selection
of averaging methods and periods, alternate  control approaches and
research iieerk .

     The proposal  for controls was made  even though  OSW has not
established that emission of PICs from hazardous  waste incinerators
currently pose a substantial risk.  EPA's risk assessments  indicate
that emission of jPICs at currently measured level* are not  likely

-------
to  produce  significant human health effects.   However,  since the
current  DR1 standard applies only to designated POHCs,  a 4-nines
 (99.i9%)  DUE does not preclude the  possibility that emission of
PICS could  present significant human health risk

     The Subcommittee findings and recommendations are found in the
attached  report.   The Subcommittee would like to emphasize that,
while the concept of using CO and/or THC concentrations to control
incinerator operations is reasonable, their use to reduce the risk
posed by  the emission of PICs is based on policy  rather than on
science,  because  of  limited  data.   In  addition,  there  is  a
technical obstacle to the  implementation  of this  concept if co
alone is controlled.  When CO is high, THC  (a surrogate for PICs)
may be high  or low? this lack of correlation at high concentrations
of CO limits the usefulness of CO alone as an operational control
for emission of PICs.  Thus at high CO 'concentrations,  a better
measure of  PICs is THC.

     Due  to the limitations of the  emissions data  and  the large
degree  of  uncertainty  introduced  by the  various  assumptions
employed  in the  risk  assessment methodology,  the Subcommittee
considers the methodology only sufficient to provide a risk-based
check  on  the  proposed  THC  emissions   limit  used   when  CO
concentrations are high.  Further, the Agency's evaluation of the
emissions limit provides some evidence of adequate safety,
However, the risk assessment is not sufficient for site-specific
applications.

     The  SAB would like to compliment the Office  of Solid Waste
staff and that of the Atmospheric  Research  and Exposure Assessment
Laboratory for their active and helpful participation in. our review
of the PICs issues.

     The Science Advisory Board is pleased to have been invited to
review this  important issue and looks forward to a written response
from EPA on the implementation of the Board's recommendations.

                                      Sincerely,
                                      Raymond CTtbehr, Chairman
                                      Executive Committee and
                                      chairman, Products of
                                      Incomplete Combustion
                                      Subcommittee, Science
                                    /loft ftrfrtung,
                                      Vic* Chairman,  Products  of
                                      Incomplete Combustion
                                      Subcommittee

cc:  D. Barnes, J. Berlow, D. Bussard, J. cannon, J.  Denit,
     S. Garg, R..Hoi 1 away, D. Hlustick, S. Lowrance

-------
                  REPORT OF THE



   PRODUCTS OF INCOMPLETE COMBUSTION SUBCOMMITTEE



               SCIENCE ADVISORY BOARD



      U.S. ENVIRONMENTAL PROTECTION AGENCY



                  REVIEW OF THE




              OFFICE OF SOLID WASTE




PROPOSED CONTROLS FOR HAZARDOUS WASTE INCINERATORS:



         PRODUCTS OP INCOMPLETE COMBUSTION



                 October 24, Ifif

-------
                                ABSTRACT
                                                  *

      The Products of Incomplete Combustion  (Pics) Subcommittee of
 the EPA's  Science Advisory Board  reviewed the  Office  of Solid
 Waste's (OSW) proposal to control emissions of PICS  from haiardous
 waste incinerators by instituting process controls based on Co and
 THC emission  concentrations.   Because  compounds known  to cause
 adverse  human  health  effects have  been  detected at very  low
 concentrations   in FXCs,  it  is prudent  to  take  precautionary
 measures to control PICs.   However, the  linkages between emission
 concentration,  exposure, and  effects {health and  environmental)
 were not documented.

      The proposal for controls was made even though QSW has not
 established that emission of PICs from hazardous waste incinerators
 currently pose  a substantial risk. EPA's risk assessments indicate
 that emission of PICs at currently  measured levels  are not likely
 to produce  significant human health effects.  However, since the
 current ORE standard applies only to designated  PQHCs, a  4-nines
 (99,99%)  ORE does  not preclude the possibility that emission of
 PICS could  present significant human  health risk.

      The concept of using CO and/or THC as guidance for incinerator
 operational control is reasonable.   The concentration limits for
 CO and THC, the averaging methods,  and the averaging periods EPA
.chose were  selected on the basis of  informed  judgments using the
 best available  data.

      Continuous emissions monitoring for THC with  a cold  system
 appears  to  be  practical  for  routine   operations.     Because
 incinerators may emit more PICs when upset from changes in waste
 quantity or composition  that  can result  from abrupt waste  feed
 shutoff,  a  poorly implemented,  automatic  shutdown strategy  has the
 potential to create more pollution  than  it  stops.

      The Subcommittee found the data base  characterizing  PICs in
 emissions would not allow a correlation  to  be established  with CO
 or THC levels for various combustion devices  and/or  conditions.
 The sparseness  of data introduces large uncertainties into EPA's
 risk assessment.  This uncertainty limits  the usefulness   of one
 approach proposed  by OSW to  control THC  emissions—using site-
 specific quantitative risk assessment to establish  acceptable THC
 emission rates*   Despite  the  limitation* of  the risk assessment
 methodology, however, the Subcommittee  considers the methodology
 sufficient  to provide a risk-based check  on  an alternative proposal
 by OSW—limiting  THC  concentrations  to levels  representative  of
 good operating  practice.

-------
                              NOTICE


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

-------
                 U.S. ENVIRONMENTAL PROTECTION AGENCY

                       SCIENCE ADVISORY BOARD

         Products of Incomplete Combustion (PlCs)  Subcommittee

Chairman

     Or. Raymond c, Loehr
          H* M. Alharthy, Centennial Chair in civil Engineering
          8.614 EOJ Hall
          University of Texas
          Austin, TX  7S712

Vice-chairman

     Dr. Rolf Hartung
          Professor of Environmental Toxicology
          School of Public Health
          University of Michigan
          3125 Fernwood Avenue
          Ann Arbor, Michigan  48108

Members

     Dr. Walter F, Dabberdt
          Senior Manager and Scientist
          National Center for Atmospheric Research
          Atmospheric Technology Division
          Post Office Box 3000
          Boulder, CO  80307-3000

     Dr. Richard Denison
          Senior Scientist
          Environmental Defense Fund
          isis p street, M.W.
          Washington, D.c.  20036

     Dr. Judith Harris
          vice President
          Arthur 0* Little, Inc.
          15W-315 Acorn ParJc
          Cambridge, MA  02140

     Dr. Robert J. Huggett
          Professor of Marine Science
          Virginia Institute of Marine Science
          College of William and Mary
          Gloucester Point, VA  23062
                               iil

-------
     Dr. Kun-ehieh Lee
          Engineering Scientist
          union Carbide Corporation
          Post office BOX 8361
          South Charleston, wv  25303

     Dr, Morton Lippmann
          Institute of Environmental Medicine
          Hew York University
          Lanza Laboratory
          Long Meadow Road
          Tuxedo, New vork  10987

     Mr, Charles 0, Velzy
          Velzy/Weston
          355 Main Street
          Armonk, New York  10SQ4

     Dr. Adel Sarofim
          Department of Chemical Engineering
          Massachusetts Institute of Technology
          Cambridge, MA  02139
Executive
     Dr. K. Jack Kooyoomjian
          Science Advisory Board
          U.S. Environmental Protection Agency
          401 M Street, S,W. - A1Q1F
          Washington, D.C.  20460

SJ^aff Secretary

     Mrs. Marie Miller
          Science Advisory Board
          U.S. Environmental Protection Agency
          401 M Street, S.W. - A101P
          Washington, D.C.  20460

Deputy Director^ .Science Advisory Board

     Mrs, Kathleen W, Conway
          O.S, Invironmentml Protection Agency
          401. It Str«*t, S.W. - A101F
          Washington, D.C.  20460

          Science Advisory Board

     Dr. Donald G. Barnes
          U.S. Environmental Protection Agency
          401 If Street, S.W. - A101F
          Washington, D.C.  20460
                                IV

-------
                         TABLE OF CONTENTS

1.  EXECUTIVE SUMMARY ,...........,...,„"..   1

2.  BACKGROUND  .......................   4
     2.1  Process 	  ,,...„,.  	 ...   4
     2,2  The Charge for the Pica Subcommittee   ,	   5
     2.3  Regulations .......	,	   g
     2.4  Technical ....... 	  .......   7
               2.4.1     Chemistry  	  ..,.„,.   7
               2.4.2     Incinerators	   7
               2.4.3  .   Measurements  ............   8
               2.4.4     Risk Reduction,  Upsets,  and
                         Emissions  .*...,«.,....  10
               2.4.5     Pics in Perspective	10

3.  RESPONSE TO THE FIRST QUESTION	  .  14
     3.1  Comments on the General Concept .	.  »  14
     3.2  The Critical Linkage Between EPA's Proposed Two-Tier
          Approach  and  the capability  for Continuous   or
          Frequent THC Monitoring .  .	...,»»  15
     3.3  current Available Data  Correlating CO and organic
          Emissions .....................  is
               3.3.1     Investigations    of     Potential
                         Correlation  of  CO with  Specific
                         PlCs;   .,....,....,»..  16
               3.3.2     Investigations    of     Potential
                         Correlation of CO with DRE at  Low  CO
                         Levels s	17
               3.3.3     Investigation     of     Potential
                         Correlation of CO  and THC  .....  17
     3.4  Summary	  i?

4.  RESPONSE TO THE SECOND QUESTION  .  	  .......  18
     4.1  Availability of "Total Hydrocarbon" Monitor  ....  is
     4.2  Availability of a Sampling System	  19
     4.3  Ruggedness (Operability and Maintainability) of THC-
          CEM Systems ....................  20
     4,4  Documentation of THC-CEM System Feasibility  ...»  21
     4.5  Summary *	22

5.  RESPONSE TO THE THIRD QUESTION	  23
     5.1  A Brief Description of EPA's Risk Assessment and Two
          Regulatory Approaches ..........  	  23
     5.2  Discussion of Assumptions  Used  in Risk Assessment  .  25
     5.3  Evaluation of the Risk Assessment	  28
               543.1     Is the  Use of  the Risk  Assessment
                         Methodology    Reasonable    and
                         Appropriate as A Risk-Based Check  on
                         the Proposed  Technology-THC  Limit
                         (20 ppm)?  .............  23

-------
               5.3.2     Is the Ose  of the Risk Assessment
                         Methodology    in    Site-Specific
                         Assessments    Reasonable    and
                         Appropriate   to  Support  Facility
                         Operation at CO Levels  in Excess of
                         100 ppm?  .  .  .  ,  ,	* . «  29
     5.4  Summary . . 	  ,.........„...„  29

6.  VIEWS OH OTHER TECHNICAL ISSUES .............  30
     6*1  Risk    Assessment    and    Exposure    Modeling
          Considerations  ..................  30
               6.1.1     Atmospheric  Dispersion  Simulation  .  30
               6.1.2     Recommendations  ..........  31
     6.2  Selection of CO and THC  Levels	32
     6.3  choice'of Averaging Method   ............  36
     6.4  Alternative Approaches  .........  	  37
               6*4.1     Alternative    Measures    of
                         Performance   ............  37
               €.4.2     Alternative  Actions  ,....,..  38
     6.5  Research Needs  ..................  38

7.  CONCLUSIONS AND RECOMMENDATIONS .............  41
     7.1  Conclusions ....................  41
     7.2  Recommendations ..................  44
APPENDIX A—Discussion of Averaging
APPENDIX B—Letter from R. Denision
APPENDIX C—Comments from P. Deisler
APPENDIX D—Glossary of Terms
APPENDIX E —References
                                vi

-------
                    List of Ficnirss and Tables



Figure A -  Benzene  (a PIC) Concentration Versus Those of
            CO  (bottom scale) and THC  (top scale).*...*....   12

Figure B -  Vinyl chloride Concentration versus Those of CO
            (bottom  scale) and THC  (top scale)..,,,..	,.   13

Figure C -  Methyl Chloride Concentration Versus Those of CO
            (bottom  scale) and THC  (top scale)......,»,,.,..  12

Figure D    In-Situ  CO Monitoring Data and Rolling Hour
            Average  Data Uneorrected for Moisture and Oxygen
            at A Multipurpose Rotary Kiln Incinerator
            (Plant A) 	.................	,.	  35

Figure A-l  Ditto ,,»»,			„	...» A-3

Figure A-2  Extractive co Monitoring Data Uncorrected for
            oxygen at a Multipurpose Rotary Kiln Incinerator
            (Plant B)	......	 A-4

Figure A-3  Extractive CO Monitoring Data Corrected  for
            Oxygen with Four Rolling Average scenarios,  for
            A Multipurpose Rotary Kiln incinerator (Plant C) A-s

Figure A-4  Extractive CO Monitoring Data Uncorrected for
            Oxygen at a Multipurpose Multiple-Chamber
            Incinerator (Plant D)......................	 A-7

Figure A-5 instantaneous CO Monitoring Data Corrected for
           Oxygen at a Multipurpose Rotary Kiln Incinerator
           (Plant C)	,.,, .A-IO

Figure A-6  Four Rolling Average Scenarios for co Monitoring
            Data Shown in Figure p-a (Plant C)	-A-ll

Table I     The Number of Appendix VIII Compounds Used to
            Calculate the THC Unit Risk Value and the Basis
            for Their Assumed Concentrations..................24

Table II    Incinerator CO/THC/Oata from Research Tests....... 3 3

Table A-I   Relationship Between the Ratio of Geometric
            Average  (GA) and Arithmetic Average  (AA) and the
            Standard Derivation(s) at Three Observed CO
            Hourly AA Levels Shown  in Figure F...............A-9
                               vii

-------
                      1.  EXECUTIVE SUMMARY


     At the request of the Office of Solid Waste (QSW) , the Science
Advisory   Board's  products   of   Incomplete  Combustion   (PIC)
Subcommittee reviewed the approach  OSW proposed  in  June  1988  for
the control  emission of PICs.   The Subcommittee was charged  to
address the following three questions:

     1,   Whether   limiting   carbon  monoxide   (CO)  and  total
          hydrocarbons  (THC)  is a  reasonable  approach to control
          emission  of  PICs,  given  the current data  base  and
          statutory tine constraints,

     2.   What is the feasibility  of monitoring  THC to  determine
          the aggregate emission rate of organic compounds?

     3.   Whether the proposed approach to assess the health risk
          from THC emissions is reasonable given the current data
          base and statutory time constraints?

The  Subcommittee  also  addressed;     atmospheric   dispersion
simulation, selection of CO and THC concentration  limits, selection
of averaging method and period, alternate control approaches,  and
research needs.

     The Subcommittee's task was to review the documents provided,
to provide advice on the technical and scientific adequacy of the
indicated approaches, and to suggest how to  improve the approaches.
The task was not  to provide  on-going continuing  oversight of the
EPA effort as it may have evolved since the Subcommittee meetings.

     Agency  staff  were present  at  the Subcommittee  meetings,
participated in  the discussions  and heard  the  comments  of  the
Subcommittee members. In addition, Agency staff were provided with
drafts of this report as it  was being prepared.   This report has
been compiled  from information obtained and  discussions held at
the  Subcommittee  meetings   (including  the  publicly  announced
conference call meeting of September 15),  from  written comments
submitted by the  subcommittee members and from comments supplied
from the members mm they reviewed earlier drafts of this report.

     The data base  on PICs  is sparse particularly for full-scale
incinerator*.   (Notes   In  this report the terms "PICs11 and "THC11
refer to emissions of residual organic compounds  and include toxic
compounds Stteh as chlorinated dioxins and benzene as well as non-
chronically toxic compounds  such as ethane  and  methane.)  Those
emissions which have been characterized have shown that individual
PICs are present  over a wide range of concentrations.  Likewise,
the potential toxieities of  the  PICs differ over a wide  spectrum
of  concentrations.     Under  good  combustion  conditions,  the
concentrations of  individual  measured PICs have been found to be

-------
relatively low in relation to the known toxic concentrations,   the
correlation  of  CO,  THC,  or  other parameters  with  combustion
efficiency  and  PICs  emission  concentration  is  weak  for  some
conditions  and for some  combustion devices.   The  relationship
between THC and emission of PICs  below levels of public  health
concern for hazardous waste incinerators indicates that at low CO
concentrations   (less than 100 ppm) THC emissions are  low (less
than 20 ppm).  At higher CO concentrations,  THC may or may not be
high.

     Compounds known to  cause adverse health effects are among the
PICs detected in  the  ppb  and ppt range  in  actual  incinerator
emissions.  One risk assessment, based on measured emission levels
and employing conservative assumptions,  suggests that PICs do not
pose  a significant health  risk.(2,7)    This risk  assessment,
however, considered only  organic compounds  in  the emissions  that
were actually identified and quantified  (for example, these ranged
from 1% to €0% of the total organic  emissions at a specific site),
and was limited to the  direct inhalation route.   Environmental
risks  and  human health risks  resulting from  indirect  routes of
exposure have not been assessed.

     Plant upsets  can increase emission of  PICs.   Therefore, any
proposed controls should minimize unnecessary waste feed shutoffs
which may result in upsets; otherwise, the controls may cause more
pollution  than  they  prevent,    osw's  approach can result  in
automatic shutdowns and plants may  be upset by sudden changes in
the  amount or  composition of waste or  auxiliary  fuel  burned,
Parameters used for control must relate  to actual emission of PICs
and  be practical  as well if they are to  be  of  use.  oxygen
monitoring,  frequent periodic  testing  for THC and  other factors
discussed in  the  report may be additional  guides for operational
controls.   The appropriate control  parameter and level may vary by
class of combustion device.

     Because  the  available database on CO  and organic emissions
shows  CO   does  not  correlate   well   with  THC   at  high  CO
concentrations, reliance on controlling CO alone would have serious
limitations.

     While  EPA1* risk assessment  approach in  this  case  follows
previously accepted methodologies,  the Subcommittee believes that
it lacks both the precision and accuracy needed to be useful  in a
site-specific  regulatory  context.    Given  the  assumptions and
uncertainties,  it  im the  judgment  of the Subcommittee that it is
not possible to calculate  total THC risks reliably  and that at
present th* method is not  suitable for setting site-specific limits
based upon THC  levels alone*

     Due to the limitations of the emissions  data and the  large
degree  of  uncertainty  introduced  by  th« various  assumptions
employed  in  the  risk  assessment   methodology,  the  Subcommittee

-------
considers the methodology only sufficient to provide a risk-based
check on the proposed  THC  emissions  limit.  Further,  the Agency's
evaluation  of  tile emissions  limit  provides some  evidence  of
adequate safety.   The  concept of using CO and/or THC  to control
PICs is reasonable.

     The Subcommittee addressed research needs. Since only a small
fraction of the total  number of compounds produced  during upset
conditions can be monitored, there is a need to relate the simple
measures of emissions  produced by a CO, THC,  or other detection
surrogates to  risk.   Emission of PICs  from incinerators  are a
potential problem that  forms part of the broader problem of organic
emissions  from  combustors.    Comparative  emissions  and  risk
assessments of different combustion categories would be desirable
in order to assign priorities for risk reduction measures.

     Overall,  the  Subcommittee believes  that the general concept
of using CO and THC for the purpose of ensuring that PIC emissions
are below  levels  of public health  concern  is  reasonable.   The
Subcommittee,  however,  is concerned about the averaging method, the
averaging period, and the concentrations chosen for the CO and THC
standard.  The Subcommittee understands that these parameters and
values were chosen primarily based on informed judgments using the
best available  data.   However,  the  supporting documentation does
not convincingly demonstrate that a CO concentration  of loo ppm is
better than 50 ppm or 150 ppm,  nor that a one-hour rolling average
is better than an eight-hour rolling average  for CO.

-------
                          2.  BACKGRGUHD
2.i  process
     In a memo dated June 28, 1988, Mr. Joseph Cam,  Director of
the Waste Management Division of th* Office of Solid Waste (OSW),
requested a meeting with  Dr.  Donald Barnes,  then  Acting Director
of the Science Advisory Board, to  discuss  how and when SAB could
review  osw's proposed approach  to  control  the potential  for
emissions of products  of  incomplete combustion (PICs)  that occur
during the  incineration of hazardous wastes.(57)   The memo also
described the basic  approach of OSW in the  control of potential
PICs.   The memo  acknowledged that "the approach is   based  on a
number of assumptions  and a  relatively thin  data  base*  However,
we know of no viable  alternative.   SAB  comments on improvements to
this approach or  alternative approaches to control Pic emissions
would be  helpful."   Accompanying  the memo was  a draft  of the
proposed rule and supporting background documents for SAB staff
review.

     Following  initial  consideration  of  the matter,  the  SAB
Executive Committee  approved formation of a "PICs Investigative
Group11 to focus the review and to select reviewers, once the charge
was  clear.    The  PICs Investigative Group  was  composed  of SAB
Executive Committee members, Dr. Raymond Loehr, Dr. Rolf Hartung,
and Dr.  Richard Griesemer.  Background materials were provided in
July 1988.  A teleconference August 30, 1988 sharpened the scope
and focus of the review.

     This activity produced  a charge (Section 2.2) setting forth
the questions to  be  addressed by  the PICs  Subcommittee.  The SAB
and OSW agreed this  review need not address metals,  residues, or
specific organic emissions such as dioxin because  of extensive SAB
reviews of  single  chemical risk assessments and prior reviews of
incineration issues.   Participants identified the  expertise needed
to   address    these   questions    including:       engineering,
instrumentation,  risk assessment,  and human  health.    The Pics
investigative Group also  established a preliminary schedule.

     The SAB PICs Subcommittee held its first meeting December  15-
16, 1988, drafted a report, and held a second meeting  January  26-
17,  1989,  at which  the  report was  extensively revised.   Both
meetings were held in  Washington  and were  open to the public. An
additional publicly announced meeting  was  held by conference call
on September 15,  1989. All revisions made  to the report prior to
its  submittal  to the  Executive  Committee  wer* made  by mail  and
telephone.  The Executive Committee recommended  improvements  and
considered  the  report at a  public meeting  July  17-11, 1989, in
Washington and approved it at a public meeting October  23-24,  1989,
also in Washington.

-------
      The Subcommittee's task was to review the documents         ,
 to  provide  advice  on the technical and scientific adequacy of the
 indicated approaches, and to suggest how to improve the approaches,
 The task was not to provide on-going continuing oversight of the
 EPA effort  as it nay have evolved since the Subcommittee meetings*

      Agency staff  were present at  the Subcommittee  meetings,
 participated in the  discussions and heard  the comments  of the
 Subcommittee members.  In addition, Agency staff were provided with
 drafts  of this  report  as it was being prepared.  This report has
 been compiled from information  obtained  and discussions held at
 the  Subcommittee  meetings   (including  the  publicly  announced
 conference  call meeting of September 15),  from written comments
 submitted by the Subcommittee members and from comments supplied
 from the members as they reviewed earlier drafts o£ this report.

 2.2   The charge for the Pies  Subcommittee

      OSW has developed a regulatory program to control emission of
 PICs based on limiting  CO and THC concentrations in stack emissions
 of incinerators, boilers, and industrial furnaces. In establishing
 concentration limits on  CO and THC that would ensure emission of
 PICs did not pose  a significant  health risk, OSW developed a risk
.assessment  methodology to  conservatively estimate the inhalation
 risk posed  by THC.

      OSW requested that the PICs Subcommittee provide comments on
 the  technical merits of the proposed approach to control emission
 of PICs,  Specifically, OSW requested comments  om

      o    Whether  limiting CO and THC is  a reasonable approach to
          control emission of PICs, given the current data base and
          statutory time constraints?

      o    what  is  the  feasibility of monitoring THC to determine
          the aggregate emission rate of  organic compounds?

      o    Whether  the proposed approach to assess the health risk
          from THC emissions is reasonable, given the current data
          base  and statutory  time constraints?

 2.3  Regulations

     Hazardous  waste  incinerators have been  subject to controls
 under the Resource Conservation and Recovery Act (RCRA) since  1981.
The- existing regulations control emissions of organic compounds by
 requiring a 99.§9% Destruction  and  Removal  Efficiency (ORE) for
 Principal Organic Hazardous Constituents (POHC)  in the waste  feed.
while these standards do not directly control emissions of Products
 of Incomplete Combustion (PIC), the quality of  operation  needed to
 achieve a  99.99%  ORE  also generally results  in low emission of
 PICs.

-------
Trial  burn  data  have  shown,  however,   that  hazardous  waste
incinerators can  operate at CO  levels indicative of  combustion
upset conditions and still achieve 99.99%  ORE.   Under  these high
CO conditions,  the EPA is concerned that Pics could -be  present at
concentrations that pose unacceptable health risks. Therefore, the
Office  of  Solid  Waste  (OSW)   proposes to  amend  the  existing
regulations  to control  emission of  PICs from hazardous  waste
incineration using  the same limits on carbon monoxide  (CO)  and
total hydrocarbons  (THC) that  are being proposed for  control of
PICs from boilers and industrial furnaces (1  - 5)^

     The OSW staff believes that requiring incinerators to operate
at high combustion efficiency will Minimize the potential  health
risk posed  by  emission of  PICs.   Stack gas CO  is  a  conventional
indicator of combustion efficiency and' a  sensitive  indicator of
poor  efficiency under most combustion conditions.    THC  is  a
surrogate for PICs.  When THC is high,  CO is always high. However,
when CO  is  high,  THC may or may not be high.   THC  is   currently
measured as part of routine operations  at a number of incinerators
and is a better indicator  of poor efficiency than CO  when CO is
high.  Therefore,  PICs  and thereby the toxic fraction  of PICs could
be controlled by ensuring that hasardous waste incinerators operate
at  high  combustion  efficiency  through  limits  on  stack  gas
concentrations of CO and/or THC.

     Studies characterizing Pica formed during the incineration of
hazardous waste largely represent  good  combustion conditions where
CO and THC are low.  While  a large fraction (that is,  from 40 to
99%) of the hydrocarbon emissions at any particular facility have
not been identified, and health data do not exist for nany  of the
compounds that have been identified, many identified hydrocarbons
are known to cause adverse human health effects.(4)   EPA's risk
assessments indicate that while emission of PICs at the currently
measured levels are  not likely to cause significant human health
problems,   the  current  4-nines   (99.99%)   DSS  standard  could
theoretically   allow  PICs  emission levels  which could present
significant human health risks,(2,3,4)   These observations are
consistent  with the findings  of  the  SAB  in its "Report  on the
Incineration of Liquid Hazardous Wastes."(7)  Given the  uncertainty
about the health risk emission of PICs posed,  and the  strong public
concerns about risks from incineration, OSW believes  it is prudent
to  institute  additional  regulatory  requirements  controls  to
minimize  the  potential  for  health risks  from  possible elevated
concentrations of PICs.

     The proposal  for controls was made even  though osw has not
established that emission of PICs from hazardous waste incinerators
currently pose a substantial risk. SPA's risk assessments indicate
that emission of PICs  at currently measured levels are not likely
to produce  significant human health effects.  However,  since the
current  ORE standard applies only to designated POHCs,  a  4-nines

-------
 (99.99%)  ORE does not preclude the possibility  that  emission of
 PICS could present significant human health risk.

 2.4  Technical

   2*4.1   Chemistry

     Combustion  is  a chemical process.   In the incineration of
 hazardous wastes, waste and air are combined with heat to 'produce
 major  products,   by-products,  and  unburned wastes.   The  major
 products  for incineration  are simple molecules such as water,
 carbon dioxide, and hydrochloric  acid; these simple molecules make
 up approximately  99.99% of  the emissions.   Organic components in
 the emissions are generally referred to as "Products of Incomplete
 Combustion   (PICs)"  which  include  various  hydrocarbons  (THC)
 including chlorinated hydrocarbons,  Metals present in the waste
 are  not  destroyed  by  incineration and will  be  found in  the
 emissions and in the residue of incineration.

     The  incineration of wastes  proceeds by means of  a series of
 complex parallel  and  sequential  processes,  including  the heating
 and volatilization  of  the waste,  mixing  of the vapor  feed or
 volatile products with  the  oxidant, and  the chemical  reaction of
 the gaseous species.  The oxidation reactions may involve several
 hundred elementary reactions,  but with very  few exceptions, carbon
 monoxide  (CO) is an intermediate  product  between  the carbon in the
 waste being incinerated and the most oxidized form of carbon, which
 is carbon dioxide (CO2),

     On  a weight  basis,  the  majority  of  the  PlCs  are CO  and
 methane. (34)  The others are trace amounts of the  various partially
 oxidized   organics,   polynuclear   aromatic  hydrocarbons,   and
 soot.(34,44)

   2.4.2   Incinerators

     If an incinerator is properly designed and operated to provide
 adequate  time,  temperature and  turbulence, the dominant factor
which impacts  on the  PICs emission level  is the excess air or
 combustion chamber  oxygen level.  When  there  is adequate oxygen
 supply in a properly designed combustion chamber or chambers! and
adequate air/fuel mixing, the  emission of  PICs is extremely ""low,
The concentration*  of PICs increase when   the oxygen content is
 close to the  atoichiometric  requirement.   when less  than  the
 stoichiometric amount of oxygen exists in the combustion  chamber,
 PICs increase even more.(27)

     Very high destruction efficiencies can be attained in the high
temperature oxidizing environment of an incinerator.   However, the
destruction of the organic hazardous compounds in a waste does not
guarantee the absence of by-products formed from the  waste during
combustion.   High concentrations of such  products of incomplete

-------
 combustion  (PICs)  usually are a consequence of a perturbation in
 the  incinerator operation resulting from rapid transients in feed
 rate or composition, failure  to adequately atomize a liquid fuel,
 excursions   in  operating  temperature,   instances  where   the
 combustible mixture fraction is outside the range of good operating
 practice,  or  inadequate  mixing  between   the  combustibles  and
 oxidant.  Modern incinerators are equipped with a large combustion
 chamber or secondary combustion chamber to minimize the impact of
 these perturbations.    Waste  feed  management and- sound  process
 control  systems  further  reduce  or compensate  for any  adverse
 impact.   The  amount  and  composition  of Pies  will  depend  in a
 complex and unpredictable  way on the nature of the perturbation.

      Current data have indicated that the state-of-art incinerators
 can  be  operated  extremely   efficiently.  Under good  combustion
 conditions,  the  combustion  efficiency  (conversion  to  C02)  is
 typically higher  than  99.9%.  The  destruction  efficiency of the
 parent compounds are typically around f§.999%.

      A study of  combustion test data obtained from various types
 of incinerators, boilers,  and process furnaces reveals that under
 normal operating  conditions, about 80% of  the principal organic
 hazardous constituents (POHC) and the major compounds in PICs  (with
 the   exception  of methane)   are  found  in  the  flue  gas  at
 concentrations between 0,1 and 20 ppbv.(io» 11, 12,  13, 14, 15, 22,
 50,  and  51)    This relatively  narrow  concentration range was
 observed  even  though  the  data  were obtained from  tests  on
 facilities  of many different designs,  operating  at  different
 conditions, burning different types of wastes with different POHC
 compounds selected,  and  operated  by different  personnel.   This
 indicates that there is a wide  range of  designs  and  operating
 conditions which achieve good combustion performance.  However, the
 compounds that were analyzed in the gaseous emissions were often
 limited in number  and types,

      The  flue  gas  POHC  and  PiCs levels  in  state-of-the-art
 incinerators, if operated properly, may be  limited by reaction and
 reactor  kinetics,  or by  som« other typ*  of limitation due to
 quenching effects.  Studies  conducted  on  several new/innovative
 incineration technologies indicate that their performance may be
 no better than existing  incinerators.(11)

   2.4.3   Measurements

      Emission of PICs composed of thousands of different compounds,
 some  of which ar*  present  in  very minute quantities and cannot be
detected  and  quantified  without  very  elaborate and  expensive
 sampling and analytical (S&A) techniques.   Such S&A work is not
 feasible in trial  bums for  permitting purposes and can only be
done  in  research  tests.    Very  few  research  tests have  been
 conducted to date  which attempted to identify and quantify all  the
 PICs  in a typical  emission sample.  Such tests were unsuccessful

                                8

-------
because  sampling and  analysis  techniques are  not available  to
identify or quantify many of the potential compounds emitted,  nor
are toxieity data available for all the compounds.(2)

     In view of the large number of possible compounds that can be
produced and their  presence at   concentrations  approaching their
practical detection limits,  it is at present impractical to design
a monitoring scheme to identify and quantify the individual toxic
compounds in  incinerator stack emissions.   What is needed is A
robust,  continuous monitor to measure  a compound  or class  of
compounds, the concentration of which correlates with those of the
toxic PICs.

     Carbon monoxide,  being  an  intermediate  in the  combustion
process and one  for which continuous detectors  are available,  is
a candidate for such monitoring.   The rate of oxidation of carbon
monoxide is slow relative to  that  of most organic  compounds and,
as a  consequence,   perturbations  in combustion  conditions,  will
usually result in an increase in carbon monoxide concentration well
before tftat of other Pies.(13)  CO is  expected  to  persist beyond
the completion  of  combustion of other  combustion  intermediates.
Total hydrocarbons  (THC) provides an alternative measure of PICs,
because the  concentration of THC  may  better correlate with  the
large number of PICs which are hydrocarbons.

     The results of several  studies on the use  of  CO  and THC as
surrogates for PICs are summarized  in a document made available to
the Subcommittee»{3)  Figure A from that document (which can also
be found  on  page 12 of  this  report)  plots  the  concentration of
benzene  as  a  function  of  the  CO concentration  and the total
hydrocarbon  concentration for data obtained  at  several sites.
Benzene was the only compound for  which some general correlation
could be found within the data obtained from different facilities.
Whenever the benzene concentration is high, the CO or THC is also
high; however, there are a  significant number of measurements in
which the CO or THC concentration is high but the benzene  level is
low*   These data indicate that the use of CO and THC as a surrogate
for benzene  will protect against high levels of benzene,  but may
also  give   false   positives,  i.e.,  high   readings   when  the
concentrations  of  benzene  is  low.    The  potential  for  false
positives is  also  seen  in  Figures B and C taken  from the same
reference (and  which  can be  found on  page 12  of  this  report)«
Figures B and C plot vinyl chloride and methyl chloride data from
one   test  facility   against  the  carbon  monoxide  and   THC
concentrations.  Since the data came from a single  test facility,
there wad no impact of design parameters on the results,Benzene,
vinyl chloride, and methyl chloride are individual PICs. while  the
concentrations of  these PICs did correlate  with CO and THC,  the
concentrations of other PICs did  not. The high concentration of CO
or THC may instead  reflect high concentrations of other PICs.

-------
   2*4.4   Risk Reduction, Upsets, and Emissions

     Risks are involved in all human  activities.   Risks  are also
involved  with  any  waste  treatment  and  disposal  operations,
including waste minimization efforts,

     In the case of PICs,  risk may be associated with exposure to
PICs.  Controlling the concentration  of PICs  in emissions  is one
approach to  reducing the risks  PICs  may present.   Incinerators
monitor CO as  an indicator of combustion efficiency.  CO  may be
high for a variety of reasons.  For example, some incinerators may
have high  CO (but low emission  of  PICs)  because of the type of
waste burned,- generally,  this situation is identified in the test
burn and requires no  additional  action to reduce risk*  CO  may be
high for very short periods because of small perturbations  in the
flame zone? typically those perturbations last seconds to minutes
and may actually be over before they are detected.  In such cases,
corrective  action  to  reduce  emission  of  PICs  is  virtually
impossible and very  likely unnecessary.   Longer  lasting high CO
concentrations do call for corrective action and a variety of these
are  possible  (such   as  readjusting  combustion  air,  increasing
turbulence, or decreasing  the rate at which waste is fed) ,  In some
cases,  corrective action requires shutting down  the incinerator to
fix the problem.   These temporary elevations  in CO concentration
are often called upsets,

     High  concentrations  of PICs may be associated with major
upsets in  incinerator operations.   Sudden significant changes in
feed rate or composition can cause such upsets.  While it may seem
counter-intuitive, very strict controls could, by leading to more
frequent shutdowns, actually increase  emission of  PICs rather than
decrease them.   A good control  system will  minimize  both false
positives  (shutting  down  when  the  incinerator  is  operating
correctly) and false negatives (operating the incinerator when it
is running inefficiently).

   2.4.S   PICs in Perspective

     The  emission of PICs  is  a  consequence of  any combustion
process. (44,45,46,47,48,49, and 50)  Emissions from hazardous waste
incinerators contribute a relatively  small fraction of the total
combustion  emissions released  into  the  environment  each year.
However,  with  PICs  the  concerns   are  effects  on  the  local
environment, not  the  aggregate national  emissions, and the local
impacts may vary considerably.

     Present analytical methods do not allow scientists to identify
and measure  all compounds  in incinerator  emissions—or in  many
other  materials.   The  best  studies  characterizing  PICs  have
accounted  for about  60% of the mass,  some studies have  accounted
for as little as 1% of the mass.(4)


                                10

-------
     The available  data show that  a well designed  and operated
incinerator does not emit Pies in substantially greater quantities
than fossil fuel combustion processes.
                                11

-------
to
                                            A foliil llyclrocutiion ConceiitrolHiii

                                                             i,ft


*
c
c .— ,
o „
11
i 3
* D
it O
u
2
0
J!





2A-
M-
t.2-
1.8-

l.A-
1.4-
I.I-
1.0-
0,0-
O.A-
0.4-
1 1 1 1 1 |
A A
A





A *

,— (40 IMC Polnti )
r/— (44 CO Points)
/ 4 *
*«.
i— . 	 > 	 , 	 _* 	 , 	 -*-, 	 . 	 — — -, 	 . 	 T 	 • 	 • 	 1 	
                                         2fll)
-400
                                                                  f|i|»n)
                  FIGURE A  - Benzene  (a  PIC)  Concentration Versus Those  of CO
                                 (Bottom Scale) and THC  (Top Scale)

-------

•"T*
CL
*s
e
4 "
"d
"c
ii
y
e
u
-a
J

U
">,
e
>



100
10

80


70
60

£0
40

30

20

10
rt
0

A

m


-
-

-*
-

-

-

-
•
°0

FIGURE B -


d.
e
^
«*•
a
"S
u
c
0
U
41

__
U
J*
1



5CCO

4CCO



3CCO




2000


1000


0
3


-



•




-


-


I
SO 100 1*0 2CO 210 SCO 250 4CI
I ' t \ • i ' k ' \ f i • '





A •

• A



* A
• JL
^

*AA * * *4
— a — L 	 «— -' — • 	 ' --•-.. ! -.. » -' , !_. .
4CO 800 1200 16CO 2000 2400 2SOQ
•Cafien Mcnoxidt CsnctnffttfScn (pptn)
Vinyl Chloride Concentration Versus Those of CO
(Bottom Scale) and THC (Top Scale)
£0 ICO. IfO 200 250 ICO 350 *e
• i • ' i ' i i • i < i * i i

A •








•A » A

• A

• * A
* 4
iAA, t t J i t * i i i I i t i
          0     400    800   J2QO   I6CO   2000    2400   2100

                    •Carbon Monoxide Conc«nrroflon (ppm)

FIGURE C  -  Methyl Chloride Concentration Versus Those of CO
              (Bottom Scale)  and THC (Top Scale)

                                 13

-------
                3,   RESPONSE TO THE  FIRST QUESTION


     The  first  question the  PICs   Subcommittee  considered  was
"Whether  limiting  carbon monoxide  (CO)   and  total  hydrocarbon
emissions  (THC)  is a reasonable approach  to control  emission of
PICs, given the current data base and statutory time constraints?"

3.1  Comments on the General Concept

     Both  the  laboratory  and field  test  data,  and  both  the
non-flame  and  flame test data  have indicated  that CO is  a good
conservative    indicator    of    combustion    performance,
(3,4,6,10,13,24,25)  Combustion of CO reguires very high ignition
energy.  CO is the dominant combustion by-product.   Combustion of
other  organics  requires  much lower   ignition  energy  and  the
reactions proceed more quickly.  While poor combustion conditions
are always  indicated by high CO levels, a high CO  level  may not
indicate poor combustion conditions.

     THC level, as  measured by a  flame ionization  detector (FID)
with a heated sampling line, is a measure of carbon-hydrogen based
volatile and semi-volatile organics.  While there are indications
that  a FID detector  may  be  inappropriate  for  chlorine  based
compounds such as  carbon tetrachloride, test data  indicated that
this  is  not important  for  regulatory  control purposes.   Carbon
tetrachloride is very difficult to oxidize  by itself and has to be
burned with  large  amounts of  carbon-hydrogen  based high  heating
value  fuel  or  wastes.    Most EPA sponsored  combustion  tests and
industry sponsored  trial burn  tests used  carbon tetrachloride as
a POHC.

     Abundant test data  indicate that CO and  associated carbon
based PICs are the dominant compounds in the  flue gas.  If both CO
and PICs  are low,  the carbon  tetrachloride level also  is low.
Hence,  the available  test  data  indicates  that  CO and/or  THC
concentration may be a good  indicator of the emission level of PICs
in stack flue gas even when carbon tetrachloride is burned.

     Some test data indicate that under certain conditions the flue
gas  THC level  may be  higher than th*  CO  level.    At  other
conditions, the flue gas soot level  is higher than the co and the
THC level.  Further evaluation of those teat conditions (31, 32)
indicated that when the THC level was higher  than  the  CO level,
there was no excess oxygen in the  combustion  region and the system
was in a reducing environment.

     Under extreme  reducing or pyrolyzing conditions where less
than the  stoichiometrlc  amount of  oxygen is available  and the
temperature is high, most organics will be thermally cracked into
soot.   Under this condition, the THC reading may be close to zero
because the carbon is in the soot.   However,  the observed CO level

                                14

-------
will be in the percent range and much higher than the 100 ppm range
which   is   of  interest  for   regulatory  compliance   purposes*
Therefore,  reducing conditions will  be  identified  by  the  CO
compliance  concentration or  low excess  oxygen concentrations and
not by a high THC concentration.

3.2  The Critical Linkage Between EPA's Proposed Two-Tier Approach
     and the Capability for Continuous or Frequent THC Monitoring

     In its September  1988 guidelines,  EPA originally  proposed a
two-tiered approach to applying limits on CO  in the stack exhaust
gas as  a  surrogate for emission of  pics.  (3)  Under Tier  1,  CO
emissions  would be limited  to 100  ppm  (corrected to  7  percent
oxygen, based on an hourly rolling average)i  compliance with this
limit would be demonstrated  initially during the  trial burn and
thereafter through continuous emissions monitoring for CO,  EPA's
rationale for this approach is based  on data demonstrating that at
co levels below 100 ppm, PICs appear to generally pose acceptably
low risk.

     EPA's  Tier II  approach  was developed for  facilities which,
despite operating at higher  CO  levels,  may  nevertheless produce
PICs at acceptably low  levels. (3)  If CO levels are  found to exceed
100 ppm during the trial burn,  the highest hourly CO average would
serve  as  the  co limit  in  the permit, and  operation would  be
permitted up  to this  level  if  total hydrocarbon  (THC)  emissions
measured  during the  trial  burn are  sufficiently  low.    EPA's
rationale for this approach  is that, when CO levels are above 100
ppm, there is virtually no correlation between CO and emission of
PICs? that  is,  when CO is high, emission of  FlCs  may  be high or
low.   Under  Tier  II,  therefore,  THC levels (which serve  as a
surrogate  for emission  of PICs) would need to  be measured  to
determine whether they are acceptably low.  EPA has proposed that
measured THC  emissions levels  would be required  to  either »eet
specified screening limits or be demonstrated on a site-specific
basis not to pose an unacceptable risk.   The agency is considering
modifying this proposal by placing an upper limit of 20 ppmv on THC
emissions,

     While this general  approach appears to  be a reasonable one,
given the available data, it contains one serious deficiency that
must be addressed.  As proposed, THC emissions would be measured
only during the trial burn.   If found  acceptable,  operation at
permitted CO  levels (above 100 ppm)  would be presumed to produce
THC emissions  on a  routine basis that are not greater than those
measured during the trial burn; no verification would be required.
Yet the entire basis of Tier  II is the lack of  any correlation
between CO  and THC when CO  exceeds 100 ppm.  Thus,  THC levels
measured  during  the   trial  burn  cannot   be   assumed  to  be
representative of routine THC emissions.
                                15

-------
     EPA  indicates that  it  is considering  requiring  continuous
emissions monitoring  (CEM) of  TEC,  because  of the limitations of
CO as a surrogate for THC.  Significant questions haves been raised,
however, regarding the current feasibility and reliability of CEM
capability  for THC,   CEM  is  necessary for the viability  of the
entire Tier II  approach.   At  the  very least,  frequent  (weekly or
on some other  time frame determined by appropriate testing) routine
stack testing for THC must be conducted as an alternative to CEM,
to provide a  basis for  assessing  both compliance  with  THC limits
and the correlation (if any)  between CO levels and THC emissions,
waste feed characteristics, and operating conditions,


3.3  Current Available Data Correlating CO and Organic Emissions

     In the past  few years,  EPA  has spent  extensive  effort in
studying the  emission of products  of incomplete  combustion from
hazardous wastes.  In most of those studies,  EPA has only attempted
to analyze and quantify Appendix VIII toxic compounds.  In only one
study, has EPA tried to identify and quantify all compounds found
in incinerator emissions.  (19)

     As discussed  in  Section  2.4.3, the Subcommittee agrees that
due to the  large number of possible  compounds and their typical
presence in the stack flue gas at concentrations approaching their
practical  detection  limits,   it  is  difficult to identify and
quantify  all  the  Pies.   However,  due to  public  concern over
emission of PICs, the Subcommittee considers the available data on
Pics  is  still sparse and more study should be  conducted.   The
ultra-low level  of PICs  and  the  similarity to those emitted from
other  combustion   devices  (22,44,45,46,47,48,49,50),   however,
suggest that this is not an important  or unique problem associated
with hazardous waste incinerators.

     Below is a brief  summary of the data presented by SPA to date,
trying to find correlations among CO, specific PICs, ORE and THC,
Specific relationships normally can be  found  for data obtained at
the same facility  where  most  of  the parameters which have impact
on  combustion  performance are  fixed.    However,  only general
correlation can be found for  data  obtained at  different facilities
where many parameters were changed in the data collection process.

   3.3.1   Investigations of Potential  Correlation of CO with
           Specific PICs:

     a.   One  study  of  four  full-scale incinerators  found that
          three   of   four   PICs   (benzene,  toluene,   carbon
          tetrachloride) were  low  (on the order of 0.1 ng/L) when
          co  was  below  100   ppm;  a  fourth  (trichloroethylene),
          however, did not show this  correlation with CO*(22)
                                16

-------
     b.   Another study of a pilot-scale facility examined il PICSJ
          while  the  two  most  abundant  Pies  showed   a  good
          correlation with CO, the other nine did not.(40)

   3.3,2   Investigations of Potential Correlation of CO with DUE
           at Low CO Levels:

      a.  One  study of  a  bench-scale  facility  revealed  little
          effect on DRE at  CO levels ranging between  15  and 522
          ppm.(42)

      b.  Another study of a  pilot-scale  facility found that DEE
          was fairly constant at Co levels up to 220 ppm. (41)

   3.3.3   Investigation of Potential correlation of CO and THC:'

      a.   Data from 9  full-scale incinerators of various designs
           at many operating conditions and burning various types
           of wastes demonstrated that  when  CO levels were below
           100 ppm, THC levels were  almost always below 20 ppmv.
           Higher CO levels were usually associated with higher THC
           emissions,(33)

      b.   Data from 11 industrial boilers of various designs and
           at different operating conditions, co-firing hazardous
           waste indicate a similar correlation, with the exception
           of one firetube boiler burning natural gas,(43,51)

      c.   Data  from  10 cement  kilns  co-firing  hazardous waste
           indicate the same  correlationi however,  at CO levels
           higher than  100  ppm,  no clear  trend was seen because
           both high and low THC emissions were observed.(43)

3.4   summary

      In the  subcommittee's view,  the available data correlating
CO with THC emissions  at low  CO levels are sufficient to support
the Agency's  concent;  of limiting CO as a  means of ensuring high
combustion efficiency  and reducing total  organic emissions.  The
data do  nojt,  however,  provide  a sufficient  basis for assessing
emission  levels  for   specific   PICs.   The   data  also  do  not
convincingly support EPA's choice of a limit as low as 100 ppmv for
CO.
                                17

-------
               4.  RESPONSE TO THE SECOND QUESTION


      The second question asked of the FICs Subcommittee was, "What
is  the  feasibility of monitoring THC to  determine  the  aggregate
emission rate of organic compounds?"

      The feasibility of continuously monitoring THC to determine
the aggregate emission rate of organic compounds can be subdivided
into several parts:

      a.   Is  there   an  available  detection   system   that  is
           responsive to "total hydrocarbons"?

      b.   Is  there   an  available  sampling  system  that  can
           reproducibly  deliver to  the  detection system a sample
           of stack gas  that  is representative of the actual stack
           gases?

      c.   Are  the detection and sampling  systems sufficiently
           rugged  for  use  in  a  continuous monitoring mode during
           routine  or  trial  burn  hazardous  waste  combustion
           operations?

      d.   Has  the feasibility of  the  overall THC  system been
           sufficiently  well  documented to serve as  a  basis for
           regulation?

      Each of these issues is discussed separately below.

4.1   Availability of a  "Total Hydrocarbon" Monitor

      Accurate determination  of the "aggregate  emission rate of
organic  compounds" requires  a detection system whose response
depends on, and only  onf  the  mass of organic material present in
the stack gas.  The Flame lonization Detector (FID) represents a
commercially available detector that comes close to meeting this
criterion, in  that it is  responsive to  most classes of organic
chemicals, including those likely to be most abundant as PICs in
hazardous   waste   combustor   emissions,   at   sub-ppm    (v/v)
concentrations in air,

      However, the  magnitude  of the FID response depends both on
the  concentration and the composition  of the  organic material
present.   A report presented to the Committee  notes  that "The
response  (relative to methane  taken as  1.0)  varies from  0 for
formaldehyde and  formic  acid to 0.4  for  methylanine  (CH3NH3) and
0.76 for dichloromethane  
-------
 (methane, etc.), some of the methane could have been unburned fuel
rather than a  product  from  the waste  since  auxiliary fuel  is,  in
some  cases,  burned  at  the time  of  sample  collection.    Since
compounds such as methane have high FID response factors, they will
tend to  dominate the magnitude of the total hydrocarbon analyzer
signal.  A 10- or 100-fold increase in the concentration of a low
response  factor  but   highly  toxic,   organic compound  might  go
undetected against the background signal  of high response  factor
nontoxic hydrocarbons.  Many different mixtures of chlorinated and
nc-nchlorinated organics could give rise to exactly the same total
FID response.

      The Office of Solid Waste addressed the response factor issue
by  developing  a  "weighted  average response  factor" based  on a
hypothetical  worst  case  composition  of  PJCs  in  incinerator
emissions.(38)    However,  in  actual  hazardous  waste  combustion
processes,  the  composition  of   the  organic  species  will  not
necessarily  approximate this hypothetical  distribution and the
actual  average  response  factor  will  differ.     Further,  the
composition of organics in the  stack gas,  and  thus the average
response factor, will  vary.   Thus, even for a single incinerator
burning a single waste, it will not be possible to deduce whether
a change in the FID signal represents a change in  the  emission rate
or a change in composition,

      Despite  these caveats,  the Subcommittee thinks it probable
that the average response factor  of the  organic species present in
relatively high concentration  in the stack gas will be sufficiently
constant that the FID response could provide a useful  approximation
or estimate of the level of PICs.   Such  an estimate could serve as
an indicator of good combustion control (but not as  the basis for
a health risk estimate),

4.2   Availability of  a Sampling System

      The FID monitor responds only to organic compounds that reach
the detector in the vapor phase.   In order to achieve an estimate
of  the  aggregate emission rate  of  organic compounds,   it  is
necessary to use a sampling system that delivers, at a minimum, a
constant fraction of  the total organics present in the original
flue gas to the detector.

      The Office of Solid Waste asked the Subcommittee to address
the  question  of using  a   "hot"  (1S06F)  versus  "cold"  (ambient
temperature)  transfer  line to deliver the stack  gas  sample to the
FID.  A  150°F  transfer line may  not  eliminate sorption problems,
although the hot line should allow a broader range of compounds to
reach  the detector.    The proposed  regulations  assume  that  a
constant fraction (25%) of the THC (the "missing carbon") will be
lost due to absorption effects.(38)  No evidence was presented to
the Committee to document that the "missing carbon11 percentage will
be constant.   in fact,  data  were presented which  show "missing

                               19

-------
carbon"  percentages ranging  from 2%  to  71.1%.(4)    It was  not
possible to determine whether the variability wag due to differing
degrees  of  absorption/condensation in the sampling system  or to
differences in the composition of orgmnics present or both.

      With  a  cold transfer  line,  most lower  volatility,  higher
molecular weight  organics will condense out of the  vapor phase.
Cold  sampling systems  typically  include  a  conctensate trap  to
collect  these non-volatile   organics,  along  with  water.    The
relatively  volatile compounds  will  reach the detector and  the
variability in their concentration should be mainly a function of
stack gas composition, rather than sampling train problems.   This
may be advantageous relative to using hydrocarbons as a measure of
good  combustion,  but  not for  risk  estimates because the risk
assessment  relies on the accuracy of mass emission rate of  toe
pollutant.  Using a method which measures only a portion of the THC
is acceptable in view of EPA's proposed technology  based 20 ppm THC
limit, although  it is desirable  to  measure  the total THC  or at
least a relatively constant  fraction of the  total.   However,  the
reliability and operability of the system in all modes of operation
is also important so that a  comparative picture of  the emissions
is available on a continuous basis.  Heated THC systems, although
potentially detecting  a greater  fraction  of the THC,  have been
observed to experience problems attributable to plugging of sample
extraction lines due to heavy particulate loading and/or condensed
organics.  Unheated fie systems  have a longer history of use and
have a much higher availability.  A recent  survey  of continuous
THC  monitoring systems  reported  that  6 facilities  have been
continuously monitoring THCs using "cold11 or "conditioned" systems
for  periods ranging  from 1  to  7 years  and did  not have  any
significant loss of availability due to breakdowns  or malfunctions.
Moreover, the  Subcommittee has learned that continuous THC monitors
are  required  by   regulations  in  Germany and Switzerland  and  the
facilities  reported  no  major  problems  with  routine  (daily)
maintenance.(58)     The  Subcommittee  believes  that  "cold"  or
"conditioned" monitoring systems for hazardous waste incinerators
are available at  this time.

      Under upset conditions, the presence  of high-surface-area
particulate matter  (soot)  in  the stack could act as condensation
nuclei or absorption sites, thus reducing the quantity of organic
material reaching the FID.  However, excessive levels of soot would
be  likely  to be  accompanied  by  other  indications  of  upset
conditions, such  as high CO readings and/or low excess oxygen
readings (see Section 3,1),

4,3   Kuggedness  (Operability  and  Maintainability)  of  THC-CEM
      Systems

      The  feasibility  of monitoring  hydrocarbons using  heated
systems  in hazardous  waste  incinerator  stack  emissions  on   a
continuous  basis  during  routine operations was  not documented.

                                20

-------
While such  systems  (like the leckman 402 and Ratfisch  455)  have
been available  for  a long  time and used in  trial burn test  or
research tests, their continuous usage for any extended period  of
time  (such  as  a  month)  could  not be documented  either in the
U.S.and Europe.  In fact, the Subcommittee heard anecdotal evidence
that the "hot" THC system(s) can present reliability difficulties
even under trial burn conditions.

      Commercially available  "total  hydrocarbon  analyzers"  using
sample conditioning lines and a cold FID  system  are considerably
more  rugged  than typical  laboratory  instruments such  as gas
chromatographs.  However, the FID-based monitors are still subject
to  corrosion and plugging  under  conditions  that  may exist  in
hazardous waste incinerator gaseous emissions.  Use of the FID  as
a continuous monitoring system may require considerable (probably
daily)  maintenance under the direction of a reasonably experienced
chemist or chemical  technician.

      The "hot" sampling line may also retire periodic (frequent)
maintenance or cleaning to prevent build-up of condensable organics
and losses of THC over time.  This could require the installation
of two parallel  sampling  lines - one in service and the other being
washed/baked-out at any given tine.  Another alternative might  be
frequent replacement of a short section of transfer line close  to
the stack.

      The Subcommittee is convinced that successful implementation
of  a continuous  THC  monitoring  system will  require  rigorous
attention to Quality Assurance/Quality control (QA/QC) protocols,
as well as requiring the careful training of skilled operators.

      A single gaseous organic,  such  as propane, is conventionally
used to calibrate continuous hydrocarbon monitors.   This may also
be appropriate for hazardous waste combustion purposes.  However,
two caveats need to be made,  it is critically important that the
calibration standard(s)  be introduced at or immediately behind the
probej this will be  the only  way to  approximate the  kinds  of
absorption losses that have  been postulated.  Also, some checks of
the  monitoring system  should  be made  using calibration  gases
containing less-volatile organics (perhaps naphthalene,  or tri~ or
tetrachlorofoenzene).     Preferably,   these   compounds  would  be
introduced  into the stack  gas  stream by the method  of standard
additions?  this could provide  a check on  the constancy  of the
"response factor" and also  on absorptive losses.

4.4   Documentation of THC-CEM System Feasibility

      While there are limited data available  on the accuracy and
precision of continuous THC  measurements over  time  periods of days
to weeks, QSW did not provide documentation of the  operability and
maintainability of the FID detector and sampling system by facility
personnel under routine operations until late in the subcommittee's

                                21

-------
review process.  As discussed above, the OSW recently submitted to
the Subcommittee a survey of continuous THC monitoring systems that
indicates that^"cold" or "conditioned gas" systems can be operated
continuously without unusual operation or maintenance problems. (58)
Although the Subcommittee has not  reviewed  the report in detail,
it  appears  to document the  feasibility  of continuous  cold THC
monitoring.

      Notwithstanding  the  field experience with continuous THC
monitoring,  there  is  a  need to  investigate  and  document the
magnitude of  effects such as rate of condensation,  build-up of
non-volatile organics,  flue  gas  moisture  effects,  and effects of
partieulate matter on the medium- to long-term performance of the
system.  The selection of appropriate calibration  compounds and
determination  of precision  and  accuracy  data over various time
periods  and   concentrations  is  also a  subject   for  further
investigation.   There  is evidence  indicating that  a  cold THC
analyzer with  a  flue  gas pre-conditioning system can function
reasonably reliably.  The reliability is improved due to the fact
that those components which cause corrosion and plugging problems
have been removed,  but,  at the same time, these chemicals have been
removed from the system and have not been detected.

4.5   Summary

      The  feasibility  of  using a heated  line to  continuously
monitor total hydrocarbons (THC)  has not been documented. However,
recent survey data appear to  show that  unheated THC monitors  using
sample conditioning systems  (refrigerated  condensate traps) are
feasible and  already in operation at several  facilities  in the
United states and Europe.
                                22

-------
                5.   RESPONSE  TO THE THIRD QUESTION


      The  third question  asked  of  the PICs Subcommittee  was,
"Whether the proposed approach to assess the health risk from THC
emissions is reasonable given the current data base and statutory
time constraints?11

5,1   A  Brief  Description  of   EPA's  Risk  Assessment  and  Two
      Regulatory Approaches

      To support its proposal to  limit THC emissions as a means of
reducing the potential risks posed by PICs,  EPA developed a means
of estimating the human  carconogenic risk posed  by inhalation of
THC at a given level.  Specifically, the EPA derived a unit cancer
risk value for a mixture of compounds assumed by 1PA to represent
THC (elsewhere in this section this mixture will be refered to as
"estimated THC11.)  using a set of assumptions to predict the array
of specific PICs actually present in THC.

      EPA's starting point for deriving the unit risk value is its
historical data  base on  emissions  of individual  compounds from
hazardous waste  incinerators, boilers,  and  industrial furnaces.
Several hundred toxic compounds which have been found in wastes are
listed in Appendix  VIII  of the  Code of  Federal  Regulations (CPR
261).  For each Appendix VIII compound identified in the emissions
data base,  EPA  assumed  that it  is  present  in  THC at  its 95th
percentile concentration, as a "reasonable worst-case value" {EPA's
wording),(4)  However, as Appendix B shows, this assumption may not
be conservative  in  this particular instance.  For each Appendix
viii compound that  has  not  been detected  in  emissions  and  is
therefore not in the data base, but  for which  adequate health
effects data are available to establish a risk-specific dose, EPA
assumed it is present in THC at  a nominal  detection limit of o.l
ng/L.

      The list was further expanded by including methane and ethane
emissions   from   fossil   fuel   combustion   and  formaldehyde
concentrations from municipal waste incinerators also at- their 95th
percentile concentration.  In recognition of the fact that even, the
most complete analyses  of  incinerator  emissions  have failed to
account for all of the emissions, the list of compounds  was further
expanded by 'including all compounds that have been quantitatively
assessed by the Cancer Assessment Group (CAG) of the Agency*

      Table I indicates the number of Appendix VIII compounds (out
of a total of more than 350 and not all are organic  compounds) used
to  calculate the THC unit  risk value and  the basis for  their
assumed concentrations.
                                23

-------
Table I:  The Number of Appendix VIII Compounds Used to
  Calculate the THC Unit Risk Value and the Basis for
             Their Assumed Concentrations
 No. of App. VIII Compounds Detected in
   Emissions (including formaldehyde)              25

 No. of App. VIII Compounds Assumed at
   0.1 ng/L                                        45

 Total No. of App. VIII Compounds Used to
   Calculate the THC Unit Risk Value               70
                           24

-------
      Approximately 70% of the  identified emissions  (on  a weight
basis) from  these facilities are associated with known  systemic
toxicants,* 30%  of the  identified  emissions are associated  with
Known carcinogens,(37)  All  of  the  individual  systemic toxicants
occurred at concentrations that were calculated to be  lower  than
the  Reference  Air  Concentration  (RAC).     The  reference   air
concentration (RAC) is  an estimate (accurate within an  order of
magnitude) of the  concentration to  which humans  could  be exposed
for a lifetime without suffering adverse effects.  EPA assumed that
the individual non-carcinogenic compounds in the mixture did riot
act additively  and that  consequently,  because the exposures to
systemic toxicants were  estimated at sub-threshold concentrations,
EPA reasoned  that exposure to the mixture would not generate health
concerns for systemic toxicity,

      EPA then applied compound-specific unit cancer risk factors
to  each  concentration  value to  obtain a  risk  level  for  each
compound.  A unit  risk  factor is the upper  bound estimate of the
excess lifetime cancer risk associated with  a lifetime of exposure
to  one  unit  of  concentration  (usually one milligram per cubic
meter) .     A  unit  risk  value  of  zero  was  assumed  for  all
non-carcinogens  (e.g., methane).  Finally,  these risk levels were
summed to produce  a weighted 95th percentile  unit  risk value for
estimated THC.   This unit risk  factor was  then used to calculate
the risk associated with  inhalation of estimated THC emissions at
particular  levels  using  a  variety  of  additional  assumptions
regarding dispersion and point of exposure.


5.2   Discussion of Assumptions Used in Risk Assessment

      The estimation  of potential  risks associated  with various
levels  of  total  hydrocarbon   (THC)   emissions  involves  many
assumptions.   In addition, the many  assumptions required  to derive
the unit risk factor and to calculate risks arising from exposure
to estimated THC render the  methodology even more unreliable for
the purposes of site-specific regulation.  These assumptions
are  commonly  used   in  EPA risk  assessments  and  while  the
Subcommittee did not consider it within its charge to challenge the
assumptions.   Additional comments relating "to the general issue of
risk assessment and which are relevant in part to this particular
instance, are found in Appendix C.

      Below are listed some of the assumptions with commentary from
the  Subcommittee  and   Executive   Committee.     Some   of   these
assumptions are clearly conservative, while others are not.   Some
of the conservative assumptions are the following;

      a.    Compounds  that have been  determined  by  EPA to be
           carcinogenic  in any  context, but which  have not  been
           identified in the emissions data bas«, are nevertheless
           assumed  to  occur in emissions  at  their  approximate

                                25

-------
           detection  limits.   A  nore  realistic risk  assessment
           should omit from consideration those compounds which are
           highly unlikely to be present in incinerator emissions,,
           for example diethylsilbesterol.

      b.   Individual  risks  are  calculated  at  upper  plausible
           limits to  risk. A different approach is  described in
           Appendix C.

      c.   Although not measured in the studies of hazardous waste
           incinerator emissions contained  in  the data base, it is
           assumed  that  formaldehyde  is   emitted  at  a  level
           corresponding to the 95th percentile  of the available
           data from municipal waste combustors.

      d.   The receptor is a Maximum Exposed Individual (Mil)  who
           is  postulated  to reside  continually at  the  site of
           maximum   annual   average   calculated  ground  level
           concentration for a full 70 years.

      e.   The estimated  THCs  are assumed to reach  a postulated
           receptor (the MEI) after dilution in ambient air, based
           on the use of conservative dispersion coefficients for
           reasonable worst-case facilities,

      ft   Individual  unit  risk(s)   are  added,  assuming  that
           carcinogenic risks are additive.  Appendix c shows how
           the addition of upper 95% bounds of risk for individual
           compounds  leads to  a  higher percentile bound  for the
           combined risk.

      The following assumptions in the risk assessment are either
neutral  or  insufficient  data  exist to  determine whether their
effect would be conservative or non-conservatives

      a.   Although SPA uses 95th-pereentile emission levels in
           calculating risk from emission of PICs, this risk is not
           significantly conservative compared with one calculated
           using "median11 levels.   Factors leading to this result
           are:   the high degree of skewness of  emission level
           distributions,  the assignment of 0.1 nanograns/liter to
           some supposed PICs (especially diethylstilbesterol  with
           its high contribution  to  risk at the "median"  levels)
           at both levels and the  zero risk contribution of ci and
           C2.

      b,   Synergistic  or antagonistic effects among  carcinogens
           or  between  carcinogens  mnel  non-carcinogens  are  not
           considered.

      c.   Only Appendix VIII compounds for which  adequate health
           effects data exist are considered in calculating the

                                26

-------
     unit risk factor for estimated; THC.   Because this unit
     risk value is applied to the  entire  mass  of estimated
     THC, the aggregate of all other compounds present in the
     emissions  is assumed to  pose  the  same  risk.    Data
     indicate that non-Appendix VIII  compounds are present
     in   estimated  THC  in  at  least the same order  of
     magnitude  as Appendix  Vlll   compounds?  semi-volatile
     compounds not in Appendix VIII are a particularly high
     fraction of  the total.(4)    These compounds  may pose
     lesser   or   greater  risk   than  those   included  in
     calculating the unit risk value.

d.   It is assumed that a single THC unit risk value,  based
     on a hypothetical reasonable worst-case composition, can
     be applied to all incinerators.

e.   In  calculating  RACs  for  compounds  lacking  direct
     inhalation  data,  oral   RfDs  are  used,   assuming  a
     conversion  factor  of   1  between the two  routes  of
     exposure.

f.   Exposure to the  same carcinogenic compounds contributed
     by  other  sources   (combustion  or otherwise)  is  not
     considered in assessing overall risk.

Finally, certain assumptions are clearly non-conservative:

a.   The emissions data  base is derived  in large part from
     facilities operating under good combustion conditions
     (e.g.,    research   tests),   which   is   likely   to
     underestimate, to an unknown degree,  the emissions that
     occur  during   routine  operations?   the   nature  and
     magnitude  of emissions  under the range of conditions
     which  may  be  experienced during the  lifetime  of  a
     facility's  operation  is poorly understood.   A more
     realistic  assessment would not be restricted to good
     operating conditions.  Continuous CO monitoring records
     could be inspected to estimate the amount  of time spent
     in excursions.

b.   Direct inhalation of carcinogenic PICs is the only route
     of exposure considered;  indirect exposure  through other
     routes (e.g., the food chain)  are not  included, although
     available  data  indicate that such routes may produce
     exposures  that may  be   much  greater  than   direct
     inhalation, particularly for environmentally persistent
     compounds.(39)

c.   Reliance on RfDs,  which are based  on   risks  to the
     general population,  may  not adequately protect  sensitive
     members of the population.  RfDs include a factor of ten
     for  the extrapolation  from  animal  studies  to human

                          27

-------
           health effects and another factor of ten because of the
           variability of sensitivity in human populations? in many
           cases  this appears  to  be  adequate,  but  not in  all
           cases.(2,52,53)

      d.   No consideration is given to environmental effects due
           to a lack of sufficient  information.  These effects may
           occur at levels of exposure lower than those affecting
           human populations.(2)

      In the THC risk assessment process, conservative assumptions
are used to  compensate  for many sources of  uncertainty arid areas
of insufficient information.  However,  the  toxicity  of the total
estimated THC mixture cannot be assessed, in part because of the
great likelihood of the emission of unknown compounds. In addition,
humans can  also suffer adverse effects  from pollutants  to which
they are indirectly exposed (through the food chain or by breathing
resuspended dusts which contain  contaminants, etc.)-  other living
creatures may also be affected by pollutants.   To the extent that
the risk assessment excludes these effects, it is not conservative.

5.3   Evaluation of the Risk Assessment

      EPA, has  proposed two different  regulatory  applications of
its risk assessment methodology  {including use of the THC unit risk
value).  The first application involves its use  as  a risk-based
check on the Agency's proposed technology-based THC emissions limit
of  20  ppm, (5)     The  second  application  involves   its  use  in
site-specific  risk  assessments conducted  to  support  facility
operation at CO flue gas concentrations in excess of 100 ppmv.(3)
These two applications  are discussed separately below, since the
Subcommittee reached different conclusions regarding the adequacy
of the two approaches.

   5.3.1   Is the Use of the Risk Assessment Methodology Reasonable
           and Appropriate as A Risk-'Based  Check on the Proposed
           Technology-THC Limit  (20 ppm}?

      EPA has  proposed to limit THC emissions  to 20 ppm,  based
primarily on consideration  of the actual THC levels achieved by
units operating under good combustion conditions that  is, based on
good operating practice.  In further evaluating  the choice of this
value, EPA  employed  the THC unit  risk value described above (as
well as other assumptions regarding dispersion of and  exposure to
THC emissions)  to provide a risk-based check on the  technology-
based limit of 20  ppm, in order to determine whether  such a limit
will be generally  protective of human health and the .environment.

      Despite the  limitations of the emissions data and the large
degree of uncertainty introduced by various assumptions employed
in the risk  assessment Methodology, the Subcommittee considers that
the methodology is sufficient to provide a risk-based check on the

                                23

-------
proposed THC emissions limit, and that the Agency's evaluation of
the emissions limit provides evidence of adequate safety.

      Therefore, considering the results of the THC risk assessment
calculations  presented,  measured THC  emissions of  20 ppmv  are
likely to  present  actual carcinogenic risks below  the suggested
limit of 1 in 100,000 in the majority of cases.  However, given the
assumptions  and uncertainties discussed above,  the  Subcommittee
concludes that  (a)  it is not possible to calculate reliable total
THC risks—to  both human health  and the environment and  (b)  at
present, the  method  is not suitable for  setting specific  limits
based upon THC levels alone.

   5,3.2   Is the Use of the Risk Assessment Methodology in Site-
           Specific  Assessments  Reasonable and Appropriate  to
           Support Facility Operation at CO Levels in Excess of 100
           ppm?

      While  EPA's  risk  assessment  approach follows  previously
accepted  methodologies,  because  of  the  data   limitations,  the
Subcommittee  believes  that  the risk  assessment lacks both  the
precision  and accuracy  needed  to be  useful in  a site-specific
regulatory context. For site-specific applications SPA would need
to replace the assumptions currently used with hard data. Such data
would have to include at  least:  wind direction,  the nature of the
waste  mixture,  and  the  specific  concentrations of  individual
compounds  found  in  the  emissions.  The  Subcommittee  is  not
recommending  a  large  data  gathering  effort  to  QSW, but  only
recognizing   that   the  data  requirements   for  site-specific
applications are very high.

      THC is used as a surrogate for PICs and for those compounds
in the PICs which may cause cancer.  Since THC does not correlate
well with CO when the concentration of CO is higher than loo ppmv,
reliance on these data, however  necessary*  introduces considerable
uncertainty  into risk estimations.   The  Subcommittee therefore
considers  the  Agency's risk  assessment  methodology,  which would
rely on THC concentrations obtained during trial burn tests only,
to be inadequate for site-specific applications,  as proposed under
the original Tier II. (3) The revision made by requesting continuous
THC monitoring using a cold system should be satisfactory. (37)

5.4   Summary

      The Subcommittee considers EPA's risk assessment  adequate to
provide a risk-based check on the proposed THC emissions limit of
20 ppm.  However, because high levels of CO do not necessarily mean
THC is high, the  risk assessment methodology  is inadequate for
site-specific applications, as proposed under the  initial Tier II
approach.(3)
                                29

-------
               6,  VIEWS ON OTHER TECHNICAL ISSUES


6.1   Risk Assessment and Exposure Modeling Considerations

   6,1.1   Atmospheric Dispersion Simulation

      6*1.1,1   Introduction

      The  approach  to  dispersion   modeling   employed  in  the
evaluation of  exposure and  the  risk assessment  associated with
hazardous  waste  incinerators  is  reasonable  and  appropriate,
especially in  view  of  the relatively small  uncertainty in these
estimates when compared with uncertainties associated with other
components of  the risk assessment, such as emission estimates and
risk factors.  There are a  few areas where the procedures are in
need of  some  "tightening"1'  or clarification,  and in one  or two
areas,   minor  revision.   These  include:    so-called dispersion
coefficients;  sensitivity analyses? representative meteorological
data; and low  stack-height considerations.

      6.1.1.2   Dispersion Coefficients

      The  dispersion  modeling discussion refers  to the  use of
•dispersion coefficients' whereas,  in fact, a more  appropriate term
would be "dilution factor."  The coefficient or factor referenced
is the  modeled atmospheric  concentration (C)  normalized  by the
stack mass emission rate  (Q)I  i.e.,  C/Q,   This ratio is a factor
that denotes  the effluent  dilution,  normalized  by  the  rate.
Convention considers dispersion  factors to characterize the rate
(temporal or spatial)  at which  the atmospheric  motions  cause a
volume of  some additive to  spread.   The  two  concepts  are  quite
distinct and use of "dispersion coefficients' as originally  drafted
is inappropriate and misleading.

      6.1.1.3   Sensitivity Analyses

      Applicants  and  regulators would  be helped by sensitivity
analyses  of the  relative  impact of  stack  height  and   ambient
meteorology  on dilution factors.   Selecting worst-case dilution
factors  from  multiple-year  applications  of a dispersion  model
eliminates most of the site-specific meteorological variability and
emphasizes the importance of  effective stack height.  This emphasis
nay be appropriate because stack  height  is an applicant-controlled
variable, whereas meteorology can only be controlled by  relocating
the plant.   Given these considerations,  the type of sensitivity
analysis suggested would be beneficial*
                                30

-------
      6.1.1.4   Representative Meteorological Data

      EPA's  "Information  Requirements"  section  also  addresses
representative meteorological site data, and indicates virtually
any data should be  used if representative  site-specific data are
not  available,(4)    This  is  a  significant  weakness  of  the
requirements section.  In the absence of representative data, other
data should only be allowed on an interim basis if such data meet
identified acceptance criteria? such criteria need to be included
in the document.

      6.1.1.5   Low Stack-Height Considerations

      The preamble  to  the  hazardous  waste  incinerator regulation
presents many tables that illustrate feed rate screening  limits foe-
various compounds in both complex and non-complex terrain*  These
tables illustrate effective staclc heights that vary from 4 m to 120
m.   Inclusion of the  smaller stack-height values  in  the tables
suggests that these are acceptable values.   However, as a general
practice,  small  stacJc  heights  are  to be  discouraged! specific
guidance can  only  be  given  in  combination with  information on
aerodynamic roughness  (z°)  of the local environment and the
height and fetch of  nearby tall buildings.  Such information should
also be required  specifically under  the  data needs enumerated in
Part Four;   Section VI "Information Requirements,"(4)

      Sites where potentially adverse ambient concentrations may
be found are apt to be associated with plants whose effective staclc
heights are small relative to the height  of nearby buildings or
where the fetch to  nearby  buildings  is small (or both).   In such
casest mathematical  dispersion modeling  may not be the preferred
simulation methodology, and fluid modeling in a boundary-layer wind
tunnel may be  desirable.  The draft regulations do not address this
issue and should  be amended to  reflect a selected preference for
fluid modeling in certain  situations.   The Agency already has in
place guidelines for good engineering practices in the application
of fluid modeling, and these  should be referenced and cited as an
acceptable    (or    preferred)   alternative   to   mathematical
modeling.(1,2,8.9)

   6.1,2   Recommendations

      a.  Replace  so-called  'dispersion   coefficients'   in  the
          context presently used with the term  »dilution  factor1
          and clarify related ambiguities.

      b.' Present  a sensitivity  analysis that  illustrates the
          relative impact on dilution factors of:  stack height and
          ambient meteorology.

      c.  Develop   guidelines   for   acceptable   representative
          meteorological site data.

                                31

-------
      d.  Identify  and  support the  role of  fluid modeling  for
          site-specific assessments  involving low stacks  and/or
          tall buildings.

6.2   Selection of CO and TEC Levels                       !

      EPA's proposed strategy for reducing the risk from emission
of Pics is to  stop  all  hazardous waste  feeds  when the allowed co
limit is exceeded.  The incinerator operator is encouraged to set
an alarm level at a lower level at which corrective  action would
be initiated with the aim of avoiding waste feed cutoff.  Such an
alarm level will reduce frequent transient startups and shutdowns
which could become the source of increased emission of PICs.   The
proposed  CO   and   THC   regulatory  limits   should   be  set  at
concentrations that provide enough margins above  the background
levels to provide  the incinerator operator sufficient  leeway to
take corrective action.   The CO and THC regulatory limits are for
measurements corrected  to 7 percent oxygen.   In  cases in which
oxygen  is  added  after  the  combustion  chamber,  incinerators
operating at higher flue gas oxygen levels will have lower actual
or  "uncorrected"  CO  and  THC  concentration  readings because of
dilution*  Those incinerators which operate at high oxygen levels
normally are burning high  BTU waste and require extra  air as a heat
sink.  Flame temperatures will be high.  Some carbon monoxide may
come from dissociation of  carbon dioxide as discussed  below and the
carbon monoxide level may be  relatively high.  THC will be lower
due to better air/fuel mixing.  In this case,  excess air is not a
dilution air, but combustion air.  OSW has taken this into account
by proposing a CO waiver  if THC is less than  20 pp«u

      The Subcommittee was provided with data  (summarized in Table
II)  from  research  tests  on nine  industrial  hazardous  waste
incinerators* (4)    These  show that  six  of   the  units meet the
proposed standards  for CO and THC'by a wide margin.

      The results of tests on such incinerators are not. necessarily
representative of those on other combustion devices.  Two examples
where the  CO and THC concentrations may  not be  related  to the
efficiency of combustion were brought up during the Subcommittee' s
deliberations.   Cement  kilns involve the  countercuirrent flow of
limestone and  the combustion products.   In  calcining  limestones
with a high organic content, CO and THC may be produced  during the
heat up of the limestone  and will  yield high exit concentrations
unrelated to the efficiency of the combustion process.
                                32

-------
u
u
SHE 10
Hltche)!



DuPont



1W1




DM




RUN HO.
I
t
3
Average
1
1
1
Average
1
2
3
4
Average
1
2
3
4
Average
62
{PERCENT)
9.4
10.5
9.9
9.t
9.2
9.6
10.3
9.1
12.4
13.0
13.2
15,6
13.6
10.1
11. 1
11.5
11.2
11.0

AVERAGE CO
MS MEASURED)
1.4
1.0
< 1
1.4
666
422
624
§n
4.3
o.t
1.2
0.6
1.8
1
NA
I
10
4.0
mm OR*)*
(PP. n 021
i.?
2.4
1.2
1.8
?9Q
518
816
708
7.0
1.6
2.2
1.6
3.1
1.3
M
1.5
14
S.6
AVERAGE T1C7J
(AS MEASURED}
< 1
< I
0.6
0.9
75.9
47.6
sea
60.5
2.5
1.9
1.?
o.a
1,7
2.5
2.3
2.1
2.9
2.5
mjRfp
IriifiSf
1.2
1.3
O.J
1.1
90.1
58. 5
76.0
74. fl
4.1
3.3
3,1
2.1
3.1
3.3
3.2
3.0
4.2
3.4
iilfiHfir J*KM61I~iuirs~
(ppn DRV §71021
CO 111C
6.1 1.3
4.1 1.4
16,3 2.3

1,364.4 Ififi.l
1.BS4.4 105.4
1.W5.J 112.9

107.9 4.?
24.2 3.?
4.1 3.9
5.2 5.4
-
110,5 11.6
11.2
3.7 14.7
1, 02fi.fi 230.0
_
       Sources: HRI "Perforaance Eviluatliui of  Full-Scale Hazardous HaUe Inclneralors. Voliwe 2.   loci nerd tors
                Perfaraince Resulti.»  fPA-600/2-84-iail). PB85-129S18, Hov.  1984.
                Nil "Tot*I Mass EBlssloni  Iron  a llaiirdous Uaste Incinerator,"  HRI Project No. B671-L(1>. Hdy

       *   All THC dii« are Measured propane wlttt liie exception of DOM Site  where T11C was measured  as nethnne,
           Healed CMtraction syste« and healed  TtlC nonllor was used.  Ihe THC data for Hits Site MAS converted  to
           propane using the following equal ton;

                              1HC (propanej * iHC («elhaii*|/3 |to account for the f fO response factor)
                        TABLE 11- Incinerator  CO/THC  Data from  Research Tests

-------
u
SITE 10
Flint B





Boss



Up John



Z*p«ta




A. Cyanaald




RIM Ml.
1
2
3
4
5
Average
1
I
3
Average
1
2
3
Average
1 i
2
3
4
Avenge
t
2
3
4
5
Average
02
(KMXMT)
11. B
10.3
10.7
14,3
10.1
U.4
10.4
lO.fl
10.7
10.6
8.1
a. 3
0.4
0.3
e.2
12.0
ti.a
11.9
11.0
10.3
12.4
—
12.7
13.0
12.1

AVERAGE CO
TATttfASUREfif
14.8
< 1.0
6.9
7.2
4,100
866
4.8
9.1
4.7
6.?
10. S
11.2
9.9
10.5
1.275
22.2
7.S
a. a
328.4
6.7
19.3
.
i3.a
14.3
13.5
P|M urn*
(pp. It 021
22,5
1,3
9.4
15,0
5.523
1.114
6.3
12.5
6.4
8.4
11.4
12.3
mo
11.6
1.394
34.5
11.4
13.5
363. i
6.8
31.4
-
23.3
2i,0
22.1
AVERAGE HIC I
7AS MEASURED1
< 1
< I
< 1
< 1
341
69
< 1
0.9
1.0
1.0
8.9
6.0
3.9
6.3
71.0
1.9
< 1
< I
18,7
< 1
< 1
-
< 1
< I
1.0
jMjJWJ*
Wlffiff
1.5
1.3
1.4
2.1
43a
89
1.3
1.2
1.4
1.3
9.6
£.6
4.3
6.9
77.7
3.0
1.5
1.5
20.9
1.1
1.6
-
1.7
i.a
1.6
tflClI$T RECORDED VALUES
« HftV tl*Q2)
co me
34.2 KS
1.3 1.3
14.5 2.6
17.6 2.3
6.935.a 671.7
-
9.8 1.3
21.3 3.2
u.a 3.1
-
7.3 7.9
7.6 6.Q
6.7 4.1
-
1,717.2 235.2
612.9 63.6
13.4 1.5
28.2 4.5
-
40.2 , 2.1
60.7 4.2
-
43.0 1. 7
45.0 1.9
-
            TABLE  II-  Incinerator CO/THC Data from Research Tests (Continued)

-------
          a.
        1
        U.I
        1
  >
  s
  Q.
        5 <
        ~ H
          Q
          O

          S

          i

          O
          S
a
o
ei
a
                                            920 ppmv a i?i-SO =^55=3
         O    710 ppfflv -^=t==i;_yrd —glf
                                     ^HsferJ ^"-"L-VA:




                                                                  I
                                                                  I
8
                     6PM
                                        2PM
                                     TIME OF TOE DAY
FIGURE D  -  in-situ CO Monitoring Data and Rolling  Hour Average Data
   Uncorrected for Moisture and Oxygen at a Multipurpose Rotary
                      Kiln incinerator (Plant A)
                                     35

-------
      Another  situation  that may lead to  high  CO concentrations
that are not representative of poor combustion is that in which the
products  of combustion  are rapidly cooled*   CO is  produced  in
combustion   chambers  operated  at  high   temperatures   by  the
dissociation of  C02.  When the  combustion  products  are rapidly
quenched, the high CO concentrations inevitably  present in a well-
operated high temperature  combustor will persist.  This source of
CO will not correlate with emission of PlCs, but THC will be less
than 20 ppm if combustion  is  efficient.

      The Subcommittee suggests that the facilities be encouraged
to set an alarm  level  which helps an individual facility improve
its operations. Alarm levels  for CO and THC should be designed to
provide enough  margin above  background concentrations  but below
shutoff values  to  provide  the incinerator  operator sufficient
leeway to take corrective  measures.   The; individual operator can
best determine  how much lead time is needed to take corrective
measures.  For liquid injection incinerators, the lead time could
be very short,  while for  rotary kiln incinerators, it could be very
long.  Further the  Subcommittee does not think it is appropriate
to, in effect,  raise the CO  limit by  providing for a phase down
period.  The intention is  to  reduce the number of automatic waste
feed cutoffs which could potentially increase emission of PICs due
to frequent shutdowns and  restarts.

6,3   Choice of Averaging Method

      The Subcommittee believes that there is a need to justify the
particular selection of the proposed averaging technique.  Although
OSW proposes  use of a  one-hour rolling arithmetic average,  OSW
presented no statistical analysis of CO monitoring data from well-
operated  incinerators to   evaluate  the  impact  of  the  proposed
regulation on incinerators  operating under realistic conditions.

      The Subcommittee has  obtained access to data from four plants
which are  discussed  in  some detail  in Appendix A  in  order  to
elaborate on the  issues  that need to be addressed when selecting
an averaging method.

      The Subcommittee believes that the  selection of levels of CO
and THC and averaging procedures in the short terra will involve a
certain element of judgment, it could be argued  that an arithmetic
mean is preferable for the  following reasons.  The geometric mean
is substantially lower (by  a  factor of three) than the arithmetic
mean with CO and THC levels are relatively steady.  The SAS does
not believe that  these lower  values should be  considered for
conformance with the limits because:   (l)  CO and THC are indicators
of  combustion  efficiency,  a  parameter  which  is  based  on  an
instantaneous measure  of CO? (2)  EPA is  allowing for inevitable
spikes in the  CO/THC levels  that  occur  even when facilities are
well designed  and operated by  allowing  the  CO/THC levels to be
averaged; (3) the  flexibility provided by the arithmetic average

                                36

-------
is  sufficient  to  enable the  vast  majority  of  incinerators  to
routinely  meet  the  recommended Tier  I CO  limit;  and  (4)  for
facilities that cannot  easily  meet  the lOOppmv CO limit, EPA is
providing a waiver that  would allow higher CO levies  provided that
the THC levels do  not exceed an hourly rolling average of 20  ppmv,
a THC  level that  the vast majority  (perhaps all)   of the  well
designed and well operated incinerators meet. It should, therefore,
be recognized that the values may need to be revised in the future
as new information becomes available.

6.4   Alternative Approaches

      EPA's proposed approach  to controlling  emission of  PICs is
to stop all hazardous waste feeds when the allowed CO concentration
is exceeded, without considering whether the shutdown will actually
increase emission of PICs.  Three major conditions  may result in
high emission of  PICs at properly designed incinerators.   These
conditions  are  inadequate  oxygen  or  too  much  fuel,  too  high
instantaneous  thermal  load  for  batch  feed,  and  too  low  a
temperature in the incinerator.  Parameters other than CO and THC
concentrations can be used for making control decisions and actions
other than automatic shutdown  can be used to  control  emission of
PICs.

      6.4.1   Alternative Measures of Performance

      while CO monitoring data provide a very good  indicator for
combustion  performance,  CO  alone  may not   be  appropriate  for
combustion control.   Oxygen monitoring should be considered for
all  hazardous  waste  incinerator  operations   because  oxygen
monitoring provides much better guidance to the operator when the
fuel feed  rate  is  getting too high  or the  air supply  rate is
getting too low.

      For incinerators with batch feed, there is a special CO spike
problem.  As long  as there is some oxygen available  (1-3%), little
emission  of PICS occurs.    However,   if  oxygen  is completely
depleted, high emission of PICs is possible.

      Besides   oxygen  control,   possible   alternates  include
combustion chamber temperature or  the rate of  change of combustion
chamber temperature.  A high temperature or high rate of increase
usually  indicates  that  the waste feed  rate is getting too high.
A lov temperature  or  fast decrease in temperature usually indicates
that the waste feed is getting too low or the waste feed does not
have enough heating value.

      while these  alternative  measures of  performance are useful
to operators and such monitoring could be retired,  because  of the
complexity of incinerator operations,  particular concentrations,
temperatures,  or  rates  of  change  should  not  be  made permit
conditions.

                                37

-------
    6.4.2   Alternative Actions

      Once  an  "upset"  condition  has  been  identified,  it  is
desirable  to  correct the  situation without causing  increased
emission of Pics.  With an automatic shutdown  strategy/ the change
in  fuel property or quality  (from hazardous waste to supplementary
fuel), the change  in fuel quantity, and the restart of hazardous
waste feeds all have the potential to upset operations and increase
emission of PICs.

      Therefore, the Subcommittee  proposes  use of  a  corrective
action approach as the first response to high CO and THC followed
by  a facility before considering an automatic waste feed shutoff.
For example, a staged reduction in waste feed  or other operational
controls  (such  as  readjusting  combustion  air  or  increasing
turbulence)  seem more appropriate in reducing emission of PICs than
a shutdown.   It is  probably not possible  to specify corrective
action (or "alarm")  concentrations except on a facility specific
basis,  while the practice of corrective action could be required
based on generally  applicable criteria, the concentrations are best
developed at the facility.

      For example,  high  CO  spikes  due  to momentary depletion of
oxygen can occur for certain types  of waste material.  From trial
burn tests and past operating experience,  the  operator should have
a pretty good idea on the limitation of  batch  feed size.  However,
flame combustion of solid waste is an extremely complex process and
there is always  the possibility that a  certain material performs
differently in  the combustion chamber.   Those CO  spikes  can be
identified from the oxygen monitoring data.   Since there is a time
lag  between  the waste  feed  and the time  that the  CO  analyzer
recorded a super-high CO  spike, or  the oxygen analyzer recorded a
zero oxygen period,  the  automatic cutoff  of all waste feeds will
not  solve the problem but may  create more problems.  If  the CO
spike  causes  the  hourly rolling  CO  level   to  approach   the
permitted level, a  20%  (or  other appropriate number), cutback of
waste feeds as discussed  previously  may avoid an exeeedence and
the  problems  that  an automatic waste shutoff  can  cause without
further upsetting the operation.

6.5   Research Needs

      One   long-term  research   need,   which   hazardous   waste
incineration has in common  with other combustion systems,  is for
the development of continuous fast-response monitors.  Additional
research questions  arei   (a) how to develop reliable continuous
monitors, (b)  how to convert measurements made by the monitors to
indicators of risk,  and   (c)  how  to correct operations of a unit
when emissions approach unacceptable levels.
                                38

-------
      Short-term continuous monitoring research needs  to  include
better definition of the  limits  of  existing  monitors for  CO,  o?
and "total hydrocarbons."  In  addition,  considerable research is
needed on:   (a) developing flue  gas conditioning  systems  for  THC
monitors that would preclude the plugging up  of sample extraction
lines  and also provide a  reasonably accurate measure of mass
emission  rates,  and  (b)  on  developing  procedures  for  quality
control and quality assurance (QC/QA) of monitors  during extended
operations.

      Longer-term research  should  be  undertaken  to  evaluate or
develop alternate monitoring methods for initial  use as research
tools,  and  eventually  for  more   routine  monitoring  or  audit
purposes.  A number of promising techniques already exist, but have
had little or no practical application to combustion sources.  A
few examples of devices with high  potential  for  application  are
Fourier Transform  infrared spectroscopy  (FT1R),  photoionization
detectors, molecular beam mass  spectroscopy, laser spectroscopy of
several kinds including absorption,  fluorescence,  laser activated
infrared spectroscopy and Raman spectroscopy.

      FTIR has the potential to  monitor  approximately twenty-two
flue  gas   components  simultaneously  including carbon  monoxide,
water,  hydrochloric acid,  and  sulfur  dioxide many  individual
organic compounds,  and a measure of "total  organics."   A laser
fluorescence monitor for  PAHs  has been demonstrated in research.
with some additional modification and further testing, development,
and simplification of these systems  might then be ready for use  in
industrial application.

      The current  cost  of some of these devices is relatively high,
but this could be  reduced dramatically with further research.  The
potential  returns on such research are  very  high.   A continuing
program to evaluate the most promising monitoring techniques and
to adapt  them for application  to a  variety  of combustion sources
would  strengthen  the  ability to   control  pollution  from these
sources.

      since only a small fraction of the  total number of compounds
produced during upset conditions can be monitored, there is a  need
to relate the simple measures of emissions produced by a CO,  THC,
or other  detection surrogates  to  risk.   Emission  of Pies  from
incinerators are a potential problem that forms part of the broader
problem  of  organic  emissions  from  combustors.    comparative
emissions and risk, assessments of different combustion categories
would be desirable in order to assign priorities for risk reduction
measures.

      The presence of spikes of CO during operation of rotary kilns
raises the question on the best averaging  method for  emissions.
There is insufficient data on whether any PICs are emitted with the
episodic  co  emissions  and  if these emissions are  at levels  of

                                39

-------
concern,  A lack of correlation of PICs with CO during the spiKes
would favor  a geometric averaging, but  more data are needed  to
provide an evaluation of the problem*

      The current incinerator permitting process is costly and time
consuming. A better understanding of the relationship of operating
parameters to emissions is needed  in  order to judiciously select
the parameters to control and  to establish their limits.   A goal
should be to  use continuous monitors such as oxygen and temperature
to provide an early warning of equipment malfunction to provide the
operator with time to take corrective actions.
                                40

-------
               7.  CONCLUSIONS AND RECOMMENDATIONS


7.1   Conclusions

      The Products of Incomplete Combustion (FICs) subcommittee of
the  EPA'S  Science Advisory  Board reviewed the  office of  solid
Waste's (OSW) proposal to control  emissions of PICs from hazardous
waste incinerators by instituting  process controls based on CO and
THC  emission concentrations.  Because compounds  known to  cause
adverse  human health effects have  been  detected  at very  low
concentrations  in Pies,  it  is  prudent  to  take  precautionary
measures to control PlCs.  However, the linkages between emission
concentration, exposure,  and effects  (health  and environmental)
were not documented.

      The proposal for controls was  made even  though OSW has not
established that emission of PICs from hazardous waste incinerators
currently pose a  substantial risk.  EPA's risk assessments indicate
that emission of PICs at currently measured levels are not likely
to produce significant human  health  effects.   However, since the
current DRl standard  applies  only to designated  POHCs, a 4-nines
(99,99%) DRl  does  not preclude the  possibility  that emission of
PICS could present significant human health risk,

      overall, the concept of using CO and THC for the purpose of
regulating PICs is reasonable.  However, EPA has not convincingly
documented the superiority of the selected averaging period, the
concentrations chosen for the CO  and  THC  standard,  and  has not
evaluated emissions problems  associated with unnecessary automatic
shutdowns,  other values or approaches may be better.  Individual
conclusions,  referenced  to the relevant sections' of this report
appear below.

1.     PICs, including compounds known  to have  adverse effects on
      human health,  have been detected at concentrations in the
      ppbv and  pptv  range  in the  emissions  of  hazardous waste
      incinerators,   boilers,   industrial   furnaces   and  other
      combustion sources.  (Section  5,1)

2.     Carbon monoxide  (CO) is a  good,  but  conservative indicator
      of combustion performance.   Poor combustion conditions are
      always indicated by high Co levels.  A high CO concentration
      may not indicate poor combustion conditions*   (Section 3.1)

3,     CO does not correlate with THC when CO exceeds  100 ppm.  In
      addition,  THC concentrations measured during the trial burn
      cannot  be  assumed to  be  representative  of  routine THC
      emissions  even  if  CO  concentration  remains unchanged.
      (Section 3.2)
                                41

-------
4.    Even  frequent routine  stack testing  for  THC  may not  be
      adequate to provide a basis for assessing compliance with THC
      limits.  (Section 3.2)

5,    continuous  emissions  Monitoring  (CEM)  of THC  is  desirable
      because of  the  limitations  of CO as a surrogate  for  THC,
      (Section 3.2 and 4.3)

6.    While there are data relating CO  and THC emissions at low co
      concentrations sufficient to support the concept of limiting
      CO to  ensure  high combustion efficiency and  reduced  total
      organic emissions, the existing  data base is not sufficient
      for assessing emission concentrations for  potential  PICs,
      (Section 3.4)

7.    Although the commercially available flame ionization detector
      (FID)   responds  to those classes of organic chemicals most
      abundant as PICs in hazardous waste combustion emissions, the
      magnitude of the FID response varies with the composition of
      the organic material  present. While the FID response could
      provide  an approximation  of the  PICs concentration good
      enough to serve as an indicator  of good combustion control,
      it  is not appropriate  as  the  basis  for a health risk
      estimate.   (Section 4.1)

8.    In principle,  a hot transfer line is better than a cold one.
      However,  the iso°c transfer  line proposed by EPA has not been
      validated   for  reliability  and  maintenance  problems.
      Anecdotal evidence was presented that THC systems can present
      reliability difficulties even during trial burns.
      (Section 4.2)

9.    A recent survey was  presented to  indicate  hydrocarbons in
      hazardous waste emissions are being monitored on a continuous
      basis  during  routine  operations in several  facilities for
      periods ranging from l to 7 years using unheated FID systems.
      A cold THC  system  may  be  more practical to serve  as a
      combustion  performance  indicator.   A  hot THC system would
      detect a  larger  fraction  of the THC if  operability and
      maintenance problems could be overcome. (Section 4*3 and 4.4)

10.   Use of a "cold" or "conditioned gas" FID as a continuous THC
      monitor  will,  however,  require  routine  maintenance  and
      rigorous  attention  to QA/QC protocols,  and the  careful
      training of skilled operators.   (Section 4.3}

11.   Despite the limitations of  the  data  and the  uncertainties
      introduced  by  the   assumptions,  the   risk  assessment
      methodology is  sufficient  to provide a risk-based check on
      the proposed THC emissions  limit.  (Section 5.3.1}
                                42

-------
12,   Although the  toxicity of the  total  THC mixture cannot  be
      assessed, the measured THC emissions of 20 ppmv  are likely
      to present carcinogenic risles  below the suggested  limit  of
      1 in 100,000 in most  cases.  Thus, EPA'a  evaluation of the
      emissions limit provides evidence of  adequate safety.

13.   The approach to dispersion modeling employed by  the Office
      of Solid Waste  in the evaluation of exposure and  the  risk
      assessment associated with hazardous waste incinerators  is
      reasonable and appropriate.   (Section 6*1*1)

14.   Sensitivity analyses of the relative  impact of stack height
      and ambient meteorology on dilution factors would be helpful
      to applicants and regulators.  (Section 6.1.1*3)

15.   In cases where stack heights are low  relative to the height
      of nearby buildings or where the fetch  to nearby buildings
      is small, fluid modeling in a boundary-layer wind tunnel may
      be preferable to mathematical dispersion modelling.  (Section
      6.1.1.5)

16.   While  CO  monitoring   data   provide  a  good  indicator  of
      combustion performance, CO alone may not  be  sufficient for
      combustion control purposes.  Oxygen monitoring data provides
      better guidance to the operator*  (Section 6.2)

17.   Other controls,  related to change in temperature,  can be used
      as alternatives to oxygen control.  (Section 6.2)

18.   Sudden changes in fuel feeds  can  cause upsets to incinerator
      operation.  Such  upsets may  produce  increased emissions of
      PICs,   (Section 6.2)

19.   The most likely cause for continuous high CO concentrations
      in a large-scale incinerator with multiple feeds  is that the
      total waste feed is too high; the same is generally true for
      small liquid injection  type  incinerators,  low incinerator
      operating temperature may also  cause high  co.  (Section 6.2)

20.   Unnecessary shutdown  of the  waste feed to incinerators may
      be counterproductive  to control of PICs  in  some cases and
      should  be discouraged.   Alternate  approaches,  including
      taking corrective  measures to  avoid  an automatic wastefeed
      cutoff or phased shutdown,  may be more effective*  Facility
      operators should be encouraged  to set an alarm level to alert
      of impending waste feed cutoffs and  take remedial measures
      to avoid them.

21.   A major research  need  which  hazardous waste incineration
      shares with other combustion systems, is for  the development
      of continuous fast-response monitors that could be  used  for
      feedback control.   (Section  6.3)

                               43

-------
22.   combustion  devices  differ  and  the  differences  between
      devices, such  as  those between cement kilns  and  hazardous
      waste incinerators, need to be  accounted  for  in developing
      a strategy for controlling PICs.  (Section 3,3.3)

23.   The risk  assessment methodology  should  not be applied  to
      specific sites as proposed under  Tier  II  by using THC data
      obtained during trial burn  tests only because of the lack of
      assurance that THC will remain low during routine operations
      if carbon Monoxide alone is continuously monitored. Routine
      monitoring of THC is necessary.  {Section 5.2)


7.2   Recommendations

l.    EPA should conduct more studies to better define whether or
      not a  problem exists with the emission of PICs, the source(s)
      of the problem if  it exists, and how to  minimize the problem.

2.    To assure that THC remains low even when CO is high, the Tier
      II approach  should  require GEM of THC*   This is necessary
      because  CO  concentration  does  not  correlate  with  THC
      concentrations when CO exceeds  100 pp».   Therefore,even if
      a CO  limit is  set and monitored for,the THC concentrations
      measured  during  the  trial burn  cannot  be assumed  to  be
      representative of routine THC emissions.   (Section 3.2)

3.    EPA should develop and validate  a heated sampling system.
      Until  such a line is validated,  the continued use of a cold
      transfer line may be  appropriate  because the  cold transfer
      line  has been shown  to  operate successfully under  the
      temperature and operating conditions expected.   {Section 4*2)

4.    EPA should investigate and  document the magnitude of effects
      such  as  rate  of  condensation,  build-up  of  non-volatile
      organics,  flue  gas   moisture   effects,   and  effects  of
      particulate matter on  the  medium- to long-term performance
      of the  monitoring  system.   The  selection of appropriate
      calibration  compounds and  determination of  precision and
      accuracy data over various time periods and concentrations
      is also a subject for  further investigation*   (Section 4.4)

5,    EPA should revise  its  discussion of  dispersion modeling to
      employ standard terminology.  (Section 6,1.1.2)

6.    EPA should  include sensitivity analyses  in the supporting
      documentation for the  regulation.  (Section 6,1.1.3)

7.    EPA should expand on the data needs enumerated  in  Part Four
      of its document to include more information on terrain.(4)
      (section 6.1.1.3)

                                44

-------
8.    Fluid modeling  in a boundary-layer  wind tunnel should  be
      considered where  staclc heights are  small relative to  the
      height of  nearby buildings  or where  the fetch to  nearby
      buildings is small.   In such cases, mathematical dispersion
      modeling may  not  be the preferred  simulation  methodology,
      (Section 6.1.1.5)

9.    An oxygen concentration limit should  not be used as a permit
      condition although it provides useful guidance to operators
      for decisions on  corrective actions.  Controls  related to
      temperature may  be considered  as alternatives  to  provide
      information for corrective action,  (Section 6,2.1)


10,   A corrective action system  should be considered as  a first
      response to CQ/THC concentrations that approach permit limits
      in order to avoid unnecessary automatic waste cutoffs because
      sudden changes  in fuel  feeds may  create major upsets  in
      incinerator operations which may lead to the release of more
      PICs to the environment than are associated with a CO spilce.
      Nonetheless, an automatic waste shutoff should be triggered
      when CO and THC levels reach the permit limit.(Section 6.2)

11,   EPA should conduct research on how to develop reliable THC
      continuous monitors,  how to  convert measurements made by the
      monitors  to  indications  of  risk,  and  how  to  correct
      operations of a unit when  emissions approach unacceptable
      levels.  (Section 6,3)

-------
                                                  APPENDIX A

A Discussion of Averaging as it Applies to Emissions Data Controls

     One problem results from occasional momentary high CO spikes
which may give one-hour rolling average CO values above the current
proposed CO standard.  If the CO remains higher than normal for a
prolonged period of time, it is likely that there is a combustion
problem  and the  waste feeds should  be shutoff  to inspect  the
problem.  The  current proposed  standards adequately address this
concern.    However,  since  CO is  a  conservative  indicator  for
combustion  performance,  CO  spikes  will be  observed  during  the
normal incinerator operation.  Any  disturbance in the flame zone
may produce a CO spike which can be large or small.  The CO spike
data will form a skewed (lognormal)  distribution with most of the
data being  very low  (around  20-30  ppmv),  but some  of the spike
values will be very large (higher than 1,000 ppmv).  Depending on
the cause, those spikes may last  from  only a few seconds to one or
two minutes in the combustion chamber, but will show up as a wider
peak at an  extractive  CO  monitor due  to the baffling and damping
effect of the air pollution control equipment, the sampling line,
and the flue gas sample conditioning system.

      There are many reasons for the generation  of such spikes.
The major ones aret

      (a) the combined effect of a  particular batch feed has too
          high a volatilization rate and too high heat content;

      (to) purging clean a plugged feed line;

      (c) loss of feed in one of the burners due to  line plugging
          or  failure  in  the  flame  management  system and  the
          associated safety shutdowni

      (d) switching feeds and the associated shutoff and startup,•
          and

      (e) waste  feed  rate  change  due  to control response  or
          mechanical response.

     The CO spikes generated are lognormally distributed.  Based
on some  preliminary discussion  with  rotary  kiln operators,  the
Subcommittee offers the following as a typical scenario.  A 5,000
ppav CO spike  may occur once every few months.   A 2,000 pprav CO
spike may occur a few times  per week.   A 500 ppmv CO spike may
occur a  few times per day.   There  will be times when CO spikes
occur more or less frequently.

      The impact of those CO spikes  on the rolling hour average is
shown in Figure A-l for an  in-situ mounted CO  analyzer {Plant A).

                              A-l

-------
Figure  A-2 shows  similar monitoring data  for an  extractive  CO
analyzer at a different facility (Plant  1)»  As can be seen, those
rare  high  CO spikes will cause incinerator operation  to  be shut
down  if the CO standard is not set properly.  Such CO spikes cannot
be  entirely  avoided,  due to random  statistical variation as the
result of complexities of a  multiple-purpose rotary kiln operation
and high sensitivity of CO as a combustion performance indicator.
During Plant A's RCRA trial burn compliance test, it was shown that
as  long  as the flue gas  oxygen  level is higher than  1%,  the CO
spikes did not cause deterioration in destruction performance.

      In Figure A-2, each  temperature peak represents a batch feed.
As  can be  seen,  the majority of the batch  feeds did  not  cause a
high  CO  spike problem.   Since the solids  volatilization  and the
associated combustion  phenomena are  very complex,  and due to the
fact  that  CO is a very sensitive  indicator,  certain  batch feeds
will generate high CO spikes.  The same situation applies to burner
switchover,* it may or may not generate high CO spikes.

      During Plant B's RCRA trial burn compliance test, the batch
feed  rates were  accelerated  to  produce  as many high CO spikes as
possible (over ten spikes higher than 2,000 ppmv uncorrected were
produced).  During the  trial burn test, flue gas organics level was
continuously  monitored with a MS/MS  (two  Mass Spectroscopies in
series for  better compound speciation)   mounted in  a  mobile van.
There was no deterioration in destruction performance during those
high  CO spikes observed.  However, there is insufficient data on
other compounds to be confident that no  other PICs were present.

      Figure A-3 shows a two-hour period, extractive CO monitoring
data  (adjusted from 13% to 7% oxygen) at a third plant (Plant C).
Four methods  to  calculate the  rolling average data were used and
will be discussed.  Figures A-4 and A-5 extended the data to a ten-
hour period to show the long-terra trend.

      Figure A-4 shows extractive CO  monitoring data (uncorrected)
at a  multiple-chamber  incinerator  with  multiple feeds (Plant 0).
During one  of Plant D's  RCRA trial  burn  compliance  tests, pint
bottles  of  four different  materials   (solid, slurry,   and  two
liquids)  were burned.   Forty pint  bottles (ten of each material)
were burned in controlled order (one bottle followed with another
of different content) during each vosf sampling period.  Figure A-
6  shows  data for  three  VOST  sampling periods  under the same
operating conditions.   The  resulting measurements were different
even  though  the  content  and operating  conditions  were the same.
The highest CO spike observed was the result of emergency shutdown,
since the  facility  was operated under a permit which allows a CO
spike no higher  than 150  ppm.   The test results differed because
each bottle had been subjected to a different  flame condition and
broke  in  a different  manner.   However, test data indicated no
deterioration in destruction performance.

-------

                 ==30 J^^^LJ^ J—\=^f=^.-^^ 30 -^



                                            ii ,|   j-"—;

O
t£
m
o 1
ol
z <:
O
of
8
  i

  I

  I
  O
  u

               O





... ./"-,.::

                            TIME OF THE DAY
Figure  A-l     In<-Situ CO Monitoring Data  and Rolling Hour
Average Data Uncotfrectecl for Moisture  and  Oxygen in a
Multipurpose Rotary Kiln Incinerator  (Plant A)
                          A-3

-------
  •*]
                                        EXTRACTIVE ACTUAL FLUE GAS CO MONITORING DATA (ppmv)
to
I
*fc
  W
s-
91 W

sew
H- CW
ts n
C rt
^ H-
•o <
O At
(fl
£0
H- H-
o rt
H- W
1

»1 3
a» n
ft O
O H
N H
  O
— •> O
a* a
3
ft Hi
W0

•*•*»

  O
  ifl
     O
          M
          8

-------
      The Subcommittee recommends that EPA evaluate the impact of
those  occasional  CO  spikes  which  would  occur  in  real-world
incinerator operation.  The Subcommittee also recommends that EPA.
evaluate the use of geometric  average  (GA) as an  alternative to the
arithmetic average (AA) .   It the intent of  the regulation is to
regulate the  total amount of  CO discharged at an incinerator, an
arithmetic mean is more appropriate and will account for the total
amount of GO  discharged.   Since the Subcommittee understands the
intent of the regulation is to use CO as  a surrogate combustion
performance   indicator,   then  a   geometric  mean  may  be  more
appropriate,  because the  CO data are  lognorraally distributed.

      The relationship between  the  GA and  AA can be expressed by
the following equation:            z
                               AA*1 +• S*

where  "S"  is the standard  deviation of the data  points used to
calculate AA and  GA.   Table K-l shows the ratio  of GA to AA and
its  relationship  with standard deviation, S,  at  three CO hourly
rolling average levels  observed  in Figure A-3.  The three levels
were:   AA at  32  ppmv  between  0 to  20  minutes,  AA at  112 ppmv
between 20  to 50 minutes,  and AA at 185 pptav between 60 and 80
minutes .

      During normal CO  variation between 10  to 60 ppmv  (as shown
between 0  to  20  minutes in  Figure F)   the  AA was  32  ppmv,  the
standard deviation(s) was about  10 ppmv, and the GA was 31 ppmv,
which was almost  the  same as  AA  (GA/AA =0,96),   If the standard
deviation increased to  20 ppmv, the CO spikes would have to vary,
repeatedly,  between approximately 10 to 100 ppmv.   Even so,  the
ratio of GA/AA would be 0.87.

      Between  20  to 50  minutes,  the observed AA was 112 ppmv and
GA was  39  ppmv,  which  was  35% of AA.   This  means the standard
deviation of the data  points  had to be 300 ppmv.  As  shown in
Figure F,  the  increase in  AA was caused by  only  two  large CO
spiJces,   if the  operation  was seriously faulty,  there  would be
repeated CO spikes and the  standard deviation would be narrower.
For example, if the CO spikes had varied  repeatedly between 50 and
250  ppmv,  the standard  deviation would be  about 50  ppmv.   The
observed GA would then be approximately  102 ppmv, which  is 91% of
the observed AA,
                              A-5

-------
200
                    ARITHMETIC (M) GEOMETRIC (GA) AVERAGES
                                                       \
                                                  INSTANTANEOUS CO
                                               ,	,	,	,,
                                                                             140
                                    TIME  (MINUTES)
     FIGURE A-3  Extractive CO Monitoring Data Corrected  for Gxyfjen with
               Four  Roll-inq Averaqe Scenarios, for a Multipurpose
                       Rotary Kiln Incinerator (Plant C)

-------
  VOST no.  3  tube start —
 VOST no. 2 tube end
   _.,.t^.T.^r._. rr. ; ; , i , i i i i .' t r-^-t-! * T ?! {t   	
   I ;- ... I - . . „! r • • • i • • • • r- »-"•? ' •—•-f	' { * • '  •
VOST no,  2 tube start •	•• 4rj •'"^'"^Ik:

                           "-r:r:;#ii-.f!.'
                           ; «i : ; * • t "f • » r •
                                        :. ^ o
                                           i-
                 * T	I' •'"'••• ' "" * J~ ~ "* * '  	" »*•)*.«*»-' * ••« T»«*

 1OO""'"*ri
 VOST? no.  i tube end

   Pi'' .I'!'!
Emergency'Shutdown,  High CO : l:\
   f r-i=fzi :(:;:•

  VOST no.  1 tube start -^_^
                                         '— o
  a
  o
 o
 in ,
 1-1

_L
o
o
rH


I
                              O
                              U)
O

I
           (Aimld) VIVO
         svo am* TVOLDV
                                                 z
                                                       c
                                                       0)
                                                       o
                                                       05
                                             8g
                                             S-3

                                             > ,

                                             «a
                                             «ij
                                             Ti
                                             o

                                             6.
                     A-7

-------
      Between 60 to 80 minutes, the  observed AA  was  185  ppmv and
<3A was 44 ppmv, which was 24% of the AA.  This means  the  standard
deviation of  the data points  had  to be 750  ppmv.   As  shown  in
Figure A-3, the  increase in AA was  caused by  the  two  additional
large CO  spikes.   If the operation was seriously faulty,  there
would also be repeated, CO spikes.   For  example,  if the CO spikes
had  varied repeatedly   between  30  to   500  ppmv,  the  standard
deviation would be about 200 ppmv.   The observed GA would then be
approximately  125  ppmv  which  is  68% of the  observed AA.   The
difference between GA  and AA will be less if the standard deviation
is smaller.

      As  shown  in the  above  discussion,  the geometric average
removes the effect of a  few very  large values  which  will occur
occasionally  in  incinerator operation.   However,  the  difference
between the geometric  average and arithmetic average will  be small,
if the combustion  conditions is seriously  faulty as  indicated by
repeated CO spikes or continuous high CO levels.

      EPA should also  evaluate  the averaging time, which will have
major impacts as well.  Figures A-5  and  A-6 illustrated the impact
of averaging  time (one  hour  and three hours  were used)  to the
arithmetic average and to the  geometric average for  operation at
Plant C.

      In Figure A-5, continuous CO values are plotted for a period
of 600 minutes  (10 hours).  For the entire period,  97.3% of all
values (adjusted to 7% oxygen) were less than 200 ppmv and 96.5%
were less than 100 ppmv.

      The one-hour and three-hour rolling geometric averages  (l hr
GA, 3 hr GA) and arithmetic averages  (1 hr AA,  3 hr  AA)  are plotted
for the same  600 minute  period in  Figure A-6.   During periods of
relatively constant operation,  the CO is more normally distributed
and the AA and GA approach each other. Examples are shown at times
0-130, 200-240, and 340-400 minutes.  Although the instantaneous
CO value exceeded 100  ppmv less than  3.5% of the time, the one-hour
AA  appears to  exceed  100 ppmv  about 25%  of  the  time.   The
arithmetic average, therefore, exaggerates relatively short events.

      In summary, a good  control strategy should not only address
the  limit, and averaging  times, but also the  frequency of both
false positives  (shutting down when the incinerator is  operating
correctly) and false negatives (operating the incinerator when it
is running  inefficiently) ,   A problem is  that too  many  false
positives nay result In higher overall emissions.  As was  discussed
earlier, the  CO  level is  a conservative indicator of combustion
performance,   so modifying the  standard  to reduce false  positives
might not increase the false negatives.
                               A-8

-------
          AA =• 32 ppmv       AA - 112 SOtnv         AA *= 135 somv


           _S_     GA/AA     	S_  GA/AA             S
5 .99
10 .95
15 .91
20 .35
30 .73
50 .34 50 .91
100 .74
ISO .59
200 .48
250 ,40
300 .34




« 'i-ii i 1 e-s_ei




50 .96
100 .38
ISO .73
200 .68
250 ,60
300 .S3
500 .35
700 .26
750 .24
800 .23
TABLE  A-l  Relationship Between the  Ratio of Geometric Average  (GA)
   and Arithmetic Average (AA) and the  standard Deviation  (S) at
       Three  Observed CO Hourly AA Levels Shown in Figure F.

                                                    /
                                  A-9

-------
I
I-"
O
w r-

gg

B
ri

g
         ^

         P
             260
             240 -4
      §?   200
      s IL
      oi S
      e^   iso
      s «

      1°
      2 x
      Oo
                                                                              400
                                                                                                      600
                                                         TIME (MINUTES)
                        FIGURE A- 5  Instatiianeous CO monitoring data correeied for oxygen at a myliipurpose rotary kiln

                                     incinerator {Plaru C),

-------
                                A«THMCTie(AA) ic GEOMETWC(GA) AVERAGES
 D V
 Z fi
 *™* fV
 o> S

 B?
 S id


 1^
 00
 <-> D»
 OT f*
 < o
 OH
 w p

 ^S
 B?b

 gS
 c &


 <<

*w< «rf*
a Q
             0
                                                   TIME (MINUTES)
                     FIGURE  A 6 Four rolling average scenarios for CO monitoring daia shown in Figure 3A (Plant C).

-------
 ENVIRONMENTAL DEFENSE  FUND
 1616 P Street, NW
 Washington, DC 20036
 <202) 387.3500
                                             July U, 1989
National Headquarterr
257 Park Avenue South
New York, NY JOOlO
(212) 505-2100
1405 Arapahoe Avenue
Boulder. CO 80302
(303)440-4901
5655 College Avenue
Oakland, CA 94618
(415)658-8001
1108 East Main Street
Richmond, VA 23219
(804) 780-1297
12S East Hargftt Street
Raleigh, NC 27601
(919)821-7793
Kathleen White Conway
Deputy Director, SAB
1418 South Monroe Street
Arlington, VA  22204

Dear Kathleen:

      As I mentioned co you on the photic earlier  today, I
have looked  in some detail at Che derivation of the THC UTIXC
risk value used by EFA in its risk assessment for assessing
the risk of  PIC emissions froa hazardous waste incinerators.

      One of the assumptions used in the risk assessment
that has been considered by both EPA and che PIC
Subcommittee to be very conservative is ch« use of
95eh-percentile emission levels for  individual FIG
compounds,   I have examined this assumption more  closely
after noticing that EPA also used 5Sth*p*rcentile values for
non-carcinogenic Cl and C2 hydrocarbons.  Se* Table  18 of
the "Background Information Cod-menC for Che Development of
Regulations  for PIC Emissions from Hazardous Waste
Incinerators."

      The raw Cl and C2 hydrocarbon  emission data used by
EPA were actually measured in fossil fuel burning devicts,
and are listed in Attachment 1.  As  you can see,  the 95th
pereeneile values used by EPA (17575 ng/L for Cl  and 34200
ng/L for C2) are heavily influenced  by the single high
values measured in the wood boiler,-test #1,

      To test the actual conservatism of EPA's use  of the
95th peteentile values, I calculated the THC unit risk value
that would be predicted using median (rather than f5ch
petcentile)  emission values for both individual carcinogenic
PICs and Cl  and C2 hydrocarbons. While my calculation did
not include  all of the carcinogenic  compounds EPA included
(see Table 14 of che Background Information Document), it
included all of the compounds chat  contributed significantly
to the calculated risk, the Bass of  PIC emissions,  or both.
The compounds I Included account far 96% of the total THC
risk and 98% of the total THC moss  in EtA's calculation,

      My analysis Is shown in Attachment 2.  It shows thac
us* of asdlan values predicts a THC  unit risk value thac  is
almost 90% of that predicted using  the 95th percentile
emission values.

-------
July 14, 19S9
Page 2
     In short, and perhaps contrary to intuition,  IPA's assunption is  not at
a.11 conservative.  Thin is because ElA's use «f 9Sth pereentila values for Cl
and C2 hydrocarbons  (which do not contribute at all to risk)  downwardly skevs
the weight fractions, and therefore the weighted unit risks,  of the individual
carcinogenic PIC compounds.

     Given that this factor was assumed by tha PIC' Sttbcouaiitte« to very
significantly contribute to the overall conservatism of EfA'a risk assessment,
we may need to re* evaluate the language pertaining to this issue in our
report,

     Flease feel free to call ne  if you have questions on this natter.

                                             Sincerely,
                                             Richard A. Denison, Fh.D»
                                             Senior Scientist
Attachaents

cc;  SAB FIC  Subcommittee
     Jack Kooyoomjian
     Bob Hollowmy

-------
         All    I        i    \
        t~TH^Ji*^<2M^l    L
           \ \. ^frW   \*  ~"     %„   i^»
             d AND Ct GQHCENTRAT1QNS BASED OH OATA  IH TABLE I
Case no.
from Taoli 2 Ft«d type Test
2 Wood 1
2
3 iasiHne
Low NOX
4
5 Caal/watir 1
2
• 6 Caal/wtter/on 1
2
8 Caal Dry bottom
Wet bottom
Cyclont
Stoker.
Lignltt Dry bottom
Cyclone
Stoker
Rtsldual oil Tanfintlally find
Wall fired
Natural gas Wall find
10*
("9/U
18,500
2,500
0
0
200
3,500
3,000
0
0
3f33Q
3,330
610
970
2,570
6,440
1,780
1,450
5,910
330
929
(ng/U)
36,OCO
5,000
3,600
2,500
0
10,500
5,800
900
0
0
630
270
5,770
310
SCO
330
260
520
330
0
1  Data froo "Total Mass Emissions from a Hazardous waste Incinerator,"
   Final Report, HR1 Project Ho* 8671-1(1).
                                     B-2

-------
Recalculating  risk using median instead ot 95th pereentile emission values

£The eoapovnds listed below account far 96% of the  risk, 9i% of the total
mass, and 90%  of  the non-C^/C-, mass  chac were calculated using the 95th
pereentile emission values.]
                                       Weight
                            Median
Benzene
Carbon Tetrachloride
Chloroform
Chlo ronethane
1,2-Dichloroethane
1,1-Dichloroethylene
DiethylsCilbestrol
Formaldehyde
Hexachlorobenzane
2,3,7,8-HxCDD
Kethylene Chlatide
2,3,7,8-PeCDD
2,3,7,8-TCDF
2,3,7,8-TCDD
Other TCDDa
1,1,2,2 -Te trachloroe than*
Tetraehloroethylen*
1,1,2-Trichloro*ehane
Tr ichla ro 
-------
Benzene
Carbon Tetrachlori.de
Chloroform
Chloromeehsne
t.2-Diehloraethana
1,1-Dichloroethylene
Formaldehyde
Hexachlotobenzene
2,3,7,8-HxCDD
Hethylene Chloride
2,3,7,8-FeCDD
2,3,7,8-TCDF
2»3f7.8-TCDD
Ochet TCDDs
1,1,2,2*Tecraehloroetham
Tecrachloroecfaylene
1,1,2-Triehloroathane
Tzrichloroathylwne
2,4,6-Trichlorophenol
Vinyl Chloride
Cl Hydrocarbons
C2 Hydrocarbons
Hightail










0

0
0
0

s








Median
48
5.1
31
474.6
4.8
2.6
398.5
2.3
.0000425
21
.0000053
.0000190
, 0000425
0.00029
5.7
7,3
15
6.2
8.3
1.7
1615
575


1,
1.
9.
1,
I,
8.
Weight
Fxa
49e-02
SSe-03
62*-03
47e-0l
49e-Q3
Q7t»G4


7.
1.
2.
3.
2,
3,
tJhit
Risk
10«-06
50e-05
30e-05
30* -06
60e-05
30e-04
1.24e-01 l,30e*Q5
7.
1.
6.
.14* -04
.32e-OS
,52*-03
1.64«-09
5.
1,
9
1
,90*-09
.32e-08
.QQ«-08
,77e-03
2,27e-03
4
*
2
5
5
1
.66«.03
,92*-Q3
,58e-03
.28e*04
,01e-01
.7S«-01
5.
2,
4,
2.
5.
OOe-06
OOe-«-00
20e-06
, 5Qa+01
,00»*00
S.QOe+01
S,
,00c-01
S.fO«-05
4.
1
1
5
5


.80*-07
.60«-05
.30*-Q6
. 60«-06
.00* -06
.«
__


1.
2.
2.
4.
3.
2.
1.
3.
2.
2.
4
2
6
4
1
i
7
2
1
2


Unit
Risk
06*-07
37*-08
21*-07
86e-07
87*-08
66«-07
6U-06
,S7«-09
,S4«-08
.74* -08
.1U-08
,95«-08
.SOe-07
. 50«.Q3
.04«-07
.09* -09
,45* -08
.50a-09
,44«-08
.64e-09
--
--
%
of
Total
Risk
2.
0.
5.
12.
1.
7.
42.
0,
0
0
80
63
85
86
02
04
.53
.09
.70
.72
1-09
0
17
1
2
0
1
0
0
0


.78
,45
.19
.76
.01
.97
.07
.38
.07

--
TOTAL
3222.1
1.000
3,7S*-Q6
100

-------
Benzene
Carbon TecrachXoride
Chloroform
Chloromethane
1,2-DichloroeChane
1,1-Dichloroethylene
Hexachlorobenzene
2,3,7,8-HxCDD
Methylene Chloride
2,3,7,8-PeCDD
2,3,7,8-TCDF
2,3,7,8-fCDD
Other TGDDs
1,1,2,2-Tetraehloroethane
Tetrschloroethylen*
1-, 1,2-Triehloroe than*
Triehloreethylenc
2,4,6-TrlchloEophenol
Vinyl Chlorid*
Cl Hydrocarbons
C2 Hydrocarbons








0.

0.
0.
0.

1







Median
48
5.1
31
474,6
4.8
2.6
2,3
0000425
21
0000053
0000190
0000425
0,00029
5.7
7,3
15
6.2
8,3
1.7
1615
575

1
1
1
1
1
9
8
1
7
1
6
1
1
2
2
5
2
2
6
5
2
Weight
Fxn
.70**Q2
.aie-03
.10* -02
.68*-Ql
.70* -03
.2U-Q4
.15* -04
.Sla-QS
.44«-03
.384-09
.73*-09
,51* -08
,03*-07
.02e-03
.59*-03
.31*-Q3
.20*-03
.94* -03
.02* -04
-72«-01
.04* -01

7
Unit
Risk
.10e-06
l.SOe-OS
2
3
2
3
5
2
4
2
5
5
5
5
4
1
I
5
.30e-05
,30*-0i
,60e*05
.30* -04
.OOe-06
.OOe-t-00
.20*-06
, 5Q*+Q1
, 00*+OQ
, OOe^-01
.OOc-01
.fOe-05
.80* -07
,iQ*-05
Weighted
Unit
Risk
1.
2.
2,
5.
4.
3.
4,
3,
3,
4.
3.
7,
5.
1.
1.
a,
21*-07
716-08
53«-07
55e-Q?
42e-08
Q4e-Q?
07e-09
Ole-08
12e-08
69e-08
3£«-Q8
53e-07
14e-08
Wa-07
24e-09
,50* -08
.30e-06 2.85e-Q9
. 60«'06 l,65e-08
5.00*-0fi


--
--.
3,


.Qle-Q§
--
.,
i of
Total
Risk
4,
1.
10,
22.
1.
12,
0.
1.
1.
1.
1,
30,
2.
4.
0,
3.
0.
0,
0


87
09
18
37
78
25
16
21
26
85
36
35
,07
80
,05
,43
,12
.66
.12

--
TOTAL
2823.6
l.OQQ
100

-------
                                                  APPENDIX  C


 Conunents  from Paul  Diesler of the Executive Committee

 1.     If you have the upper  95%  bounds  of risk  PA  and  P8 at
 concentrations  (or doses, exposures)  CA  and  Cg,  this  means that
 there  is only  one  chance  in  twenty  that  the  risks   will,
 individually, equal or exceed the upper  95%  bounds.   If A  and B
 occur together at  CA and CB,  the sura of the risks, PAB^PA+PB     is
 not  at the 95%  upper bound, but, rather  at a higher percentile.
 In this case, if  A and B act independently, the  joint percentile
 is 1-(1/20) (V2Q)=1~Q.0025=0.9975;  or 99.75%.   (If  A  and 1,  and
 therefore PA and P8, are correlated in some way, constrained, etc,..
 the  calculation  is  less simple  but  the  result  is  the same
 qualitatively,  if instead of A and B one has A,B,e,D,  etc..  the
 percentile  becomes still greater,    fhis is  the situation when
 adding up the  risks of many PICS  in hazardous waste  incinerator
 emissions,

 2.     To obtain an upper bound risk  for  a mixture of &,B,C, etc.
 at the 95% percentile  (and so  to be comparable to other risks at
 the  same  level of  confidence)  just calculate the expectations of
 mean risks, __  __    _    _
            p= pA + p8 + pc
 at CA,  Cf, Cc, etc...; the mean for the mixture is then


            P - PA + P^  +  P! + etc.

 (at the low levels of risk involved,  and assuming  independence—no
 synergism, antagonism).  Similarly the variance of the risk for the
mixture can be obtained from  the  sum of  the variances  of  the
 individual risks,  from which the standard  deviation  (the  square
 root of the variance)  can  be obtained.  Given the mean and standard
deviation of the mixture's risk, the upper 95% bond of risk for the
mixture can be estimated—putting the estimate of the  upper bound
of risk   for  a  mixture on  the same basis  as   other  such  risk
estimates.   (Notei   The regression  or curve fitting method with
which the upper  95% percentile risks are calculated should also
estimate,  if asked, the mean risks expected.)

-------
3*   Partial ReconstructJOJOL. of the 95th Percentile^Calculations
    ca.se_I.L  If all PICs were, in fact, combustion, products
       Sm  =  sum of median concentrations of carcinogenic PICs
       S9S =•  sum of 95th percentile concentrations of carcinogenic
       CB  =   sum of median concentrations of Cl and C2
       c?s =  sum of  95th percentile concentrations of ci and C2
       rm =  unit risk of the carcinogenic PICs in the median case
       r9S =  unit risk  of  the carcinogenic PICs in the  95th
              percentile case
       Now:
                                      s.13  x  10*
       B.    rw =   	  -      9.23   X  10"
       c*    rH  =   0.88 r,5

       where rm and r9S  are  the weighted unit risks  in  the median
       and 95th percentile cases, respectively.
       From A, e, and Cs

                    f\ QQ
                    Vr * OO
       Now:
            C,,  *  1,615 + 575  -  2,190
            SB 4- CB  - 3,222.2
            Therefore,  S,,, = 1,032.2
            Also        CM  =  17,575 + 34,200 = 51,775

                              C-2

-------
       The Cl and C2 emissions are highly skewed to the right
 (C95 » cm) .  Suppose the other Pics are skewed and that
S95 = 103^ (where m>l) .

                 could they be highly skewed?
       suppose the other (other than Cl and C2) PICs distributions
are similarly shaped,* then  _    _
                            rn * r95 as a limit*

       Then :    _      _

                *m  •   *m    '    «*• + 51'775
       0.88*   ----   ----    ---- • --------- • --- •
                     a*.       3222.2

       and, with Sm - 1032.2,  m * 29

       or, therefore,  the other PICs  are  highly skewed.*... too .  in
Case I enough so that relative to the skewness of Cl and C2 ,  Sm  is
only 88% of S95.

       If all  the  above is  true  (check the distributions) , this
result has nothing to do with the zero  risk of Cl and C2  but only
with the relative skewness of the distributions of Cl, C2, and the
carcinogenic PICs,

_   _   (Note:  for various  r^/rm >  1,  m is still large unless
r95/rm   is  unbelieveably  large)  —   Must   check  skewness   of
carcinogenic PICs  (that is,  actual ratio of
         Case II;  Allow for effect on r95/r_of diethvlstilbesterol
                                           10=
       At the median case, DBS contributes more than 1/2 of total
risk;  same  is  true of  rm  since Cl  and C2  contribute no  risk.
Therefore at the 95th percentile case DBS contributes very little
to the average  iw  because its concentration  (O.l  nanogran/liter)
does not change but that of the non-Gl/C2 PICs rises.   Thus,  DES
has  the  effect of  decreasing average r^/r,,  compared to  what  it
would be if OSS concentrations behaved naturally,  skewness of the
non-Cl/C2 PICs  still needs to be great to compensate here.
                              C-3

-------
Glossary of Tenas
                                                  APPENDIX D
AA     - ARITHMETIC AVERAGE
ADI    - ALLOWABLE DAILY INTAKE
C      - MODELED ATMOSPHERIC CONCENTRATION
CE     - COMBUSTION EFFICIENCY
CEM    - CONTINUOUS  EMISSIONS   MONITORING/CONTINUOUS   EMISSIONS
         MONITOR
CEMS   - CONTINUOUS EMISSION MONITORING SYSTEM
CO     - CARSON MONOXIDE
C0_    - CARBON DIOXIDE
  v£
ORE    - DESTRUCTION REMOVAL EFFICIENCY
EPA    * ENVIRONMENTAL PROTECTION AGENCY
FID    - FLAME IONIZATION DETECTOR
FTIR   - FOURIER TRANSFORM INFRARED SPECTROSCOPY
GA     - GEOMETRIC AVERAGE
GOP    - GOOD OPERATING PRACTICE
MEI    - MAXIMUM EXPOSED INDIVIDUAL
MWI    - MUNICIPAL WASTE INCINERATION
02     - OXYGEN

ORD    - OFFICE OF RESEARCH AND DEVELOPMENT AT SPA
OSW    - OFFICE OF SOLID WASTE AT EPA
PAH    - FOLYCYCLIC AND/OR POLYNUCLEAR AROMATIC HYDROCARBONS
PCDD   * POLYCHLORINATED DIBEN2Q-P-DIOXIN
PCDF   - POLYCHLORINATED DIBENZOFURAN
PCP    - PSNTACHLOROPHENOL
PIC    - PRODUCTS OF INCOMPLETE COMBUSTION
PICS   - PRODUCTS  OF INCOMPLETE  COMBUSTION  SUBCOMMITTEE  OF THE
         EXECUTIVE COMMITTEE OF THE SAB
PNA    - POLYCYCLIC AND/OR POLYNUCLEAR AROMATICS
POHC   - PRINCIPAL ORGANIC HAZARDOUS CONSTITUENTS
POM    - POLYCYCLIC AND/OR POLYNUCLEAR ORGANIC MATTER
PPM    - PARTS-PER-MILLION (BY VOLUME OFTEN IMPLIED)
PPBV   - PARTS-PER-BILLION BY VOLUME
PPTV   - PARTS-PER-TRILLION BY VOLUME
Q      - STACK MASS EMISSION RATE
QC/QA  - QUALITY CONTROL AND QUALITY ASSURANCE
RAC    - REFERENCE AIR CONCENTRATION
RCRA   - RESOURCE CONSERVATION AND RECOVERY ACT
RfD    - REFERENCE DOSE
S      - STANDARD DEVIATION
SAB    - SCIENCE ADVISORY BOARD
TCDD   - 2,3,7,8, - TSTRACHLORODIBlNEO-P-DIOXIN
THC    - TOTAL HYDROCARBONS
TSDF   - HAZARDOUS WASTE TREATMENT, STORAGE AND DISPOSAL FACILITY
VOC    - VOLATILE ORGANIC COMPOUND
VOST   - VOLATILE ORGANICS SAMPLING TRAIN
Z.     - AERODYNAMIC ROUGHNESS

-------
                                                  APPENDIX E
References

1.    EPA Memorandum dated November 16, 1988 from Bob Holloway to
      Jack Kooyoomjian  entitled "SAB Review of  PIC  Controls for
      Hazardous Waste Incinerators."

2.    Draft Preamble  for Hazardous Waste  Incinerator Regulation
      (40 CFR, Parts 260, 261, 264 and 270), dated June 14, 1988.

3.    Guidance  on  Carbon Monoxide  Controls for  Hazardous waste
      Incinerators, Midwest Research institute Draft Final Report
      to EPA,  dated September 9, 1988,(Supereeeded  by reference
      #37)

4.    Background  Information  Document  for the  Development  of
      Regulations   for   PIC   Emissions   from   Hazardous   Waste
      Incinerators, Midwest Research Institute Draft Final Report
      to SPA,  dated July 1989.

5.    Briefing for SAB Entitled, "Proposed Controls for Hazardous
      Waste  Incinerators:    Products  of   Incomplete  Combustion
      (PICs)," dated December 15-16, 19S8.

6.    LaFord,  R.K,, J.L. Kramlich, and  W.R,  Seeker, "Evaluation of
      Continuous Performance  Monitoring  Techniques for Hazardous
      Waste Incinerators," APCA Journal. 33. (6); 658? June 1985.

7.    Report to  the Administrator on  the  Incineration of Liquid
      Hazardous Wastes — Environmental Effects, Transport and Fate
      Committee — USEPA Science Advisory Board, April 1985.

8.    U.S.E.P.A.  (1981a):    Guidelines   for   fluid  modeling  of
      atmospheric  diffusion.    EPA-600/8-81-009,  Office  of Air
      Quality Planning and Standards, Research Triangle Park, North
      Carolina  27711.

9.    U.S.E.P.A. (1981b);  Guideline for use of fluid modeling to
      determine   good   engineering   practice   stack   height.
      EPA-45Q/4-81-003,   Office  of  Air  Quality  Planning  and
      Standards, Research Triangle Park, North Carolina  27711,

10.   Kun-chieh  Lee,  "Research  Areas  for  Improved  Incineration
      Systen Performance/1  Journal^of the  Air Pollution Control
      Association   (JAPCA),  Vol.  38,  No.  12,  December   1988,
      pp. 1542-1SSQ.

11.   "Hazardous Waste  Incineration,  A  Resource  Document," The
      American  Society  of Mechanical  Engineers,  Co-sponsored by
      Air Pollution Control  Association,  American  institute of
      Chemical Engineers,  and U.S. Environmental Protection Agency,
      January 1988,

12.   E.T.  Oppelt,  "Incineration of hazardous waste,  A critical
      review," JAPCA 371558(1987).

-------
13.   J.D. Kramlich,  et al*f Laboratory Scale Flame-Mode Hazardous
      Thermal  Destruction Research,"  Energy  and  Environmental
      Research Corporation, EPA-600-/2-84-086,  NTIS PB-84-184902,
      April 1984.

14.   A. Trenholm,  "Performance Evaluation of Full-scale Hazardous
      Incineration,!" EPA, P1-85-1295QQO,  1984  VolS. 1-5.

15.   "Handbook, Permit Writer's Guide to Test Burn Data, Hazardous
      Waste Incineration/1 EPA 625/6-86/012, September 1986.

16.   W. Tsang, W,  Shauto, "Chemical Processes in the Incineration
      of Hazardous Materials,11 National Bureau of  Standards,  in
      J.H. Exner, ed.,  Detoxification of Hazardous Waste. Ann Arbor
      Science, pp.  41-60(1982).

17.   K.C. Lee,  et al.r  "Revised Model  for  the Prediction of the
      Time-Temperature  Requirements for  Thermal  Destruction  of
      Dilute Organic Vapors and Its Usage for Predicting Compound
      Destructibility," Paper presented at the 75th Annual meeting
      of the Air Pollution Control Association, New Orleans, LA,
      June 1982.

18.   W.M, Shaub,  W.  Tsang,  "Dioxin  formation  in incinerators,"
      Environmental Science Technology I7j721 (1983).

19,   "Total Mass Emissions From a Hazardous Waste Incinerator,"
      Midwest  Research  Institute   report   to  EPA-QRD,  NTIS  PB
      87-228508, August 1987.

20.   D.S.  Duvall,  W.S.  Rubey,"Laboratory Evaluation  of  High
      Temperature  Destruction of  Polychlorinated  liphenyls and
      Related Compounds," university of Dayton Research Institute,
      EPA-600/2-77-228, NTIS PB-269139, December 1977.

21.   B.  Dellinger,   et  al., "Determination  of  the  Thermal
      Decomposition  Properties of  20  Selected  Hazardous Organic
      Compounds,  M  University  of  Dayton  Research  Institute,
      EPA-600/2-84-138, NTIS PB 84-232487, August 1984.

22.   A. Trenholm  and  C.C. Lee, "Analysis  of  PIC and Total Mass
      Emissions  From  an Incinerator,11 Proceedings of  the 12th
      Annual  EPA Research Symposium on Land  Disposal,  Remedial
      Action,  Incineration,   and  Treatment  of Hazardous Waste,
      Cincinnati, Ohio, April 21-23, 1986,  EPA 600/9-86-022, August
      1986.

23,   "Non-steady  State  Testing  of  Industrial  Boilers Burning
      Hazardous  waste,I1*   EPA/600-9-86-022, Accurex, 12th Annual
      EPA Research Symposium,  April  1986.

24.   D.  Bose,   s,M. Senkan,   "On  the combustion  of chlorinated
      hydrocarbons   I.—Trichlorethylent,"  combustion  Science
      Technology 35sl  (1984).

-------
2S,   w* D, Chang, S.B* Karra, 3, K. Senkan,  "Molecular beam mass
      spectroscopic   study   of   trichlorethylejie   flames,11
      gnvironmentaj. Science and Technology 20S1243 (1986).

26.   W.D. Chang, S.B. Karra, s.M.  Senkan,  "A detailed mechanism
      for  the  high temperature  oxidation of G^HCLj,"  Combustion
      gcienge. Technology 49:107 (1986),
27.   B. Bellinger,  g£ al. ,  "pie  Formation Under  Pyrolytic  and
      starved Air Conditions ," EPA/6QO/S2-86-QG6, July 1986,

28.   K.C.  Lee, "Incineration of hazardous waste, critical review
      discussion paper," JAPCA 37:1011 (1987).

29.   B.N.  Ames, at al • f "Ranking Possible Carcinogenic Hazards,"
      Science 236:271 (1987).

29-a. B.N.  Ames, "Dietary Carcinogens and Anticarcinogens , " science
      221:1256(1983).

30.   0. Zimmerman et a\. t  "Anthropogenic Emissions Data  for the
      1985 NAPAP Inventory,  "Final Report  by Alliance Technologies
      Corp to EPA's Air and Energy Engineering Research Laboratory,
      RTF,  North Carolina, USEPA, ORD (This study was conducted in
      cooperation with the National Acid Precipitation Assessment
      Program, IPA Contract No. 6S-02-4274, Work Assignments  26 and
      28),  September 1988.

31.   W.P.  Linak,  et; al. f  "On the occurrence of transient puffs in
      a  rotary kiln  incinerator simulator,  I.   prototype  solid
      plastic wastes f" JAPCA 37: 54(1987).

32.   W.P.  Linak et al. r "On the occurrence of transient puffs in
      a  rotary  kiln incinerator simulator, II.  contained  liquid
      wastes on sorbent," JAPCA 37:934 (1987).

33,   Trenholm, A.f  P, Gorman and G* Jungclaus, MRI, "Performance
      Evaluation  of  Full-Scale  Hazardous  Waste  Incinerators.
      Volume    2.        Incinerators    Performance    Results,"
      EPA-600/2-84-1811S, PS85-129518, November 1984.

34.   MRI,    "Total   Mass  Emissions  from   a   Hazardous  Waste
      Incinerator," MRI Project No.  8671-L(1),  May 1987.  (Same as
      119)

35.   H.M.  Freeman, editor,  Standard Handbook of Hazardous Waste
      Treatment and Disposal,  McGraw-Hill  Book Company, 1989.  (See
      especially Section 8.5 Cement  Kilns, by John  P. Chadboume,
      Ph.D).

36.   Sylvia  1C.  Lowrence, Director,  EPA Office of Solid  Waste,
      Memorandum to Hazardous Waste Division Director, Regions l-x,
      entitled  "Use of  Omnibus Authority to Control Emissions of
      Metals,  HCL and  PICs  from  Hazardous  Waste Incinerator,11
      February 27, 1989.

-------
37.   u.s.s.p.A.  (April  1989).   Guidance  on P£C  Controls  for
      Hazardous Waste Incinerator, Volume 5 of the Hazardous Waste
      Incineration  Guidance  Series,  Office  of  Solid   Waste,
      Washington, D.C. (Draft Final  Report),

38.   Castaltini et al.,   "Guidance  on the Control of Products of
      Incomplete   Combustion   (PlCs)    from  Hazardous    Waste
      incinerator," December 1988.(Superceeded by reference #37)

39*   Minnesota Pollution Control Agency, Supplemental Health Risk
      Assessment Technical  Work Paper,  Volume I:   Folychlorinated
      Dioxins/Polychlorinated Furons, Winina County Rescue Recovery
      Facility, May 1988.

40.   Chang,  D,,   and M.  Sorbo,  "Evaluation of  a  Pilot-Scale
      circulating  Bed Combustor as  a  Potential  Hazardous Waste
      Incinerator," presented at the  Incineration and Treatment of
      Hazardous Waste 12th Annual  Symposium, Cincinnati,  Ohio,
      March 1987.

41.   Waterland,  L.R.,  "Pilot-Scale  Investigation  of  Surrogate
      Means of Determining  POHC Destruction, ** Final Report for the
      Chemical  Manufacturers  Association,  Acurex  Corporation,
      Mountain view, California, July 1983.

42.   EPA,  Engineering  Assessment   Report  on  Hazardous  Waste
      Cofiring in  Industrial  Boiler, Volume  I, EPA  600/2-4-1772,
      PB85-187838/AS.

43,   Hlustick,  D.   Memo  dated  October  10,  1988,  to s.  Garg
      entitled "Summary of Total Hydrocarbon Measurements in Cement
      Kilns,"  found  in  EPA  Background  Document on Boiler and
      Industrial Furnaces.

44,   Edwards, J.B.,  Combustion:  The Formation  and Emission of
      Trace Species, Ann Arbor Science, Ann Arbor, MX, 1977.

45.   Harris, J., J.  Perwak,  and S.  Coons,  "An Exposure and Risk
      Assessment for Benzo(a)Pyrene  and Other Polycyclic Aromatic
      Hydrocarbons,"  A.D.  Little, Inc.,  EPA  contract Ho,  68-01*
      6160, 1982.

46.   Kites, R.JU,  "Sources  and Fates of  Atmospheric Polycyclic
      Aromatic  Hydrocarbons,11  Indiana  University,  Atmospheric
      Aerosol, pp.  187-195,   ACS  Symposium Series  167, American
      Chemical Society, Washington,  D.C., 1981*

47.   Jackson, 8., L. Scinto, and C. Shih, "Emissions of Reactive
      Volatile Organic Compounds from Utility Boilers,"  TRW,  Inc.,
      EPA-600/7-80-11, NTIS PB80-200343, May 1980.

-------
48.   Junk,  G.A., and C.S.  Ford,  "A Review of  organic Emissions
      From Selected Combustion Processes," Antes Laboratory,  Iowa
      State University,  Chemosphere, vol. 9, No.  4,  pp.  137-230,
      1980*

49.   Lao,  R.C.,   R.S.   Thomas,   M-   Lanoy,   and  S.tt.   Lee,
      "Investigation of  PAH and Polychlorinated Organic Pollutant
      Emissions from Wood Combustion Sources," Environment Canada,
      Po-lynuclear Aromatic Hydrocarbons: Formation, Metabolism and
      Measurement,   Proceedings  of  the  seventh  International
      Symposium,, pp. 745-755, Battelle Press,  Columbus, 1983.

50.   National Research Council, Polycyelie Aromatic Hydrocarbons:
      Evaluation of Sources and Effects,  "Chapter 1:   Polycyclic
      aromatic  hydrocarbons   from  mobile  sources   and  their
      atmospheric concentrations," Chapter 2;  "Polycyclic aromatic
      hydrocarbons  from natural and stationary  sources and their
      atmospheric concentration," EPA Contract No.  68-01-4655, NTIS
      PB34-15523 3, National Academy Press, Washington, D.C., 1983.

51.   Background information  Document for  the  Development  of
      Regulations to Control  the  Burning of Hazardous Wastes in
      Boilers and Industrial Furnaces,  3 vols., for USEPA, Office
      of Solid Waste,  by Engineering Science,  Inc., January 1987*

52.   Dour son,  M.L,   and  J.F.  Stava,  "Regulatory  History  and
      Experimental   Support  of  Uncertainty  (Safety)  Factors,"
      Regulatory Toxicology and Pharmacology 2, 224-283 (1983).

53.   Calabrese, E.J.,  "Uncertainty Factors and Interindividual
      Variation," Regulatory  Toxicology and  Pharmacology £, 190-
      196 (1985).

54.   Lee, Eun-Chieh,  "Incineration of Hazardous Waste:  Critical
      Review  Discussion  Papers,"  Journal of  the Air Pollution
      control  Association,   Vol.   37,  No.   §f  pp.  1011-1017,
      1987.(Same as 28)

55*   Lee, Kun-Chieh, "Research Areas for  Improved Incineration
      System Performance,11  Journal of the Air Pollution control
      Federation,  Vol.   38,  No.  12,  pp.  1541-1550,  December
      1988.(Same a« 110)

56.   staley, L.J*,  M.K. Richards,  6.L. Huffman, R.A. Olexsey, and
      B.  Dellinger,  "Relationship  Between CO,   POHC,   and PIC
      Emissions  from  a simulated  Hazardous Waste Incinerator,1*
      Journal of the Air Pollution Control Association, Vol. 39,
      NO. 39, pp. 321-327,  1989.

57.   Carra,  Joseph  of  USEPA*s  Waste  Management  Division,
      Memorandum requesting review by  the SAB,  June 28, 1988.

58.   THC Monitor Survey, MRI. Draft Final Report June 20,  1989.

-------