United States
                  Environmental Protection
                  Agency
Risk Reduction Engineering
Laboratory
Cincinnati OH 45268
                  Research and Development
EPA/600/S2-90/039 May 1991
&EPA         Project Summary

                   Minimization  and Control  of
                   Hazardous  Combustion
                   Byproducts
                   Barry Dellinger, Philip H. Taylor, and Debra A. Tirey
                    Control of emissions  of  toxic
                  organometallic compounds is one of
                  the major technical and sociological
                  issues surrounding the  further
                  implementation of incineration as a
                  waste  disposal alternative.  A
                  comprehensive review of  the status
                  of research concerning the emission
                  of organic products of  incomplete
                  combustion (PICs)  indicated  that
                  available full-scale data  were
                  insufficient in volume and quality to
                  make  firm conclusions.  As
                  enumerated below,  several trends
                  were identified  that appeared to
                  concur with  scientific theory, and
                  several deficiencies  in fundamental
                  knowledge and completeness of full-
                  scale emissions data were observed.
                    PIC emissions  are  a  natural
                  consequence of the kinetically-
                  limited  thermal  degradation of
                  hazardous wastes. Comparison of PIC
                  formation/destruction rates based on
                  theory and nominal incineration
                  conditions  indicate   that  PIC
                  emissions are always several orders
                  of magnitude higher than predicted.
                  This suggests that temporal or spatial
                  excursions from  these nominal
                  conditions are  occurring that are
                  responsible for  PIC  emissions. Low
                  temperatures due  to quenching,
                  residence time short circuits due to
                  nonplug flow  and/or  unswept
                  recesses,  and   locally  high
                  waste/oxygen concentration ratios
                  due to poor  microscale  mixing or
                  overloading are  implicated as failure
                  modes. Relative emission rates  are
                  controlled by failure conditions that
arise in  post-flame regimes of full-
scale systems.
  Consideration of waste  feed
composition and  the reported
emissions  suggested  that a
significant fraction of PICs that may
have been emitted  from full-scale
facilities were not analyzed. Analysis
of PIC emissions data from 17 full-
scale facilities further suggested that
the magnitude  of  PIC emissions
correlated with the completeness of
the  sampling  and   analysis
procedures.
  A comprehensive rank-order
statistical  analysis of  full-scale
emissions data indicated that of the
design  and operational  parameters
evaluated, stack  oxygen concentra-
tion and waste heat load exhibited
the strongest correlation with various
measures  of  PIC emissions.
Regulatory  surrogates  for PIC
emissions (CO and TUHC) and stack
benzene and  toluene concentrations
exhibited only occasional correlation
with various measures  of PIC
emissions.
  Analysis of specifically identified
emissions indicated that simple
halogenated methanes, halogenated
ethanes; halogenated ethenes, and
nonhalogenated  aromatic species
were the most frequently observed
and highest  yield  PICs. The high
yields  of  the  relatively  fragile
halogenated methanes and  ethanes
inferred that many of these PICs were
formed  downstream of the  high-
temperature zones of  full-scale
systems via radical-association
reactions.
                                                               Printed on Recycled Paper

-------
  In conclusion, a series of design,
operational,  and regulatory implica-
tions for PIC control were presented.
From the laboratory  perspective, the
urgent  need for detailed studies of
PIC characterization,  yields,  and
emission mechanisms  at  elevated
temperatures was  emphasized. The
benefits  of Interactions  between
laboratory-scale PiC studies, PIC toxi-
cologlcal studies, and sampling and
analysis methods development for
full-scale   systems  were  also
discussed.
  This   Project  Summary  was
developed by EPA's  Risk Reduction
Engineering Laboratory, Cincinnati,
OH,  to  announce key findings of the
research  project  that  is  fully
documented In  a separate  report of
the  same title  (see Project  Report
ordering Information at back).
Introduction
  Of all the issues surrounding  waste
disposal, the  potential  formation  and
emission  of PICs from  the  thermal
disposal  of  wastes  is  the  most
controversial. Much of the controversy is
due to the lack of knowledge of PICs, the
current  regulation  only  addresses
destruction  and  removal efficiency  of
principal organic  hazardous constituents
(POHCs) contained in the waste stream.
This has led to  the  unfortunate  public
misconception of incineration  as a
"landfill in the sky".
  It would  seem  that  this important
subject and  controversy that has raged
all  through  the  80's  would  be  well
researched and significant strides made
towards answering key issues. However,
we  in reality know very little about PIC
emissions. The purpose of this project
was to define what we really know about
PICs, what we think we  know, and what
we  don't know. The reader is referred to
the final report  for  a  comprehensive
analysis of theories of PIC formation and
a thorough statistical analysis  of pilot-
and full-scale  PIC emissions data.  The
project culminates  with numerous
implications for PIC emissions control.
  Although every  effort has been made
to approach this subject scientifically and
arrive at  logical,  factually-based
conclusions,  there will undoubtedly
continue  to  arise nonfactual  or even
clearly incorrect  statements concerning
PICs.  It is hoped  that this report  will at
least serve to  establish a framework for
rational  discussion  of  the issues
surrounding PICs.
Theories of PIC Formation
  There are only a few issues for which
we  have sufficient data that agree  with
scientific theory that will  allow us  to
advance several theories  about PICs.
These theories are discussed below.

Theory 0—Plps are a Natural
Consequence of Thermal
Degradation of Toxic Wastes
  This theory  jis so fundamental  and
obvious that  it is  designated  as the
Zeroth Theory of PIC Formation. From a
scientific point of view, any product of the
oxidative or pyfolytic degradation of a
compound, other  than  the  thermo-
dynamically most stable end-products,
should be included in the definition of a
PIC. If we want to  designate PICs  of
environmental concern where toxicity and
concentration are factors, then  we can
define a new  term: toxic,  principal,
product  of incomplete combustion (ToP-
PIC). The question is not whether PICs
are emitted  from   hazardous waste
incinerators,  because  they will  un-
doubtedly be  formed and emitted from
any combustion (source. The true issue is
whether they  |are  of  environmental
consequence,  which  is a very complex
issue.

Theory 1—PIC Emission Rates
are Kinetically, Not Thermo-
dynamically, Controlled
  If an  incinerator  achieves thermo-
dynamic equilibrium, then  one only
needs to know tfie elemental composition
of the  waste/fuel/oxidizer (air)  feed
system and the  temperature to calculate
the emissions  from  the system. For
example, therrnodynamically complete
combustion or pyrolysis of a chlorinated
hydrocarbon  at| 1800°F  results in the
formation of a few stable products,  e.g.,
C02, H2O, HCI,  H2,  CI2,  etc., and trace
quantities of CO and organic byproducts
(Eq. 1). Numerous  calculations of PIC
emission rates! have been performed
assuming an incinerator  at equilibrium.
These calculations typically demonstrate
that observed PIC concentrations are  at
least 10 orders of magnitude greater than
predicted based on equilibrium (Figure 1,
Table 1).
  Some have argued for rather vague
concepts with
practical utility,
ittle scientific  basis or
such as  PIC  emissions
being at "pseudo-equilibrium".  Sufficient
time is not available for equilibrium to be
achieved,  and I PIC emissions  are a
function of  tjme  (i.e.,  kinetically
controlled).  Chemical reaction kinetics
                         undoubtedly  play  a large  role in PIC
                         formation.  However, physical kinetics,
                         such as  vaporization  and  waste/air
                         mixing,  can also contribute  to kinetically
                         limited routes of PIC formation.
                         Theory 2—Deviation from
                         Normal, Average Operating
                         Conditions are Responsible for
                         Most PIC Emissions
                           Modern, well-designed  and managed
                         incinerators operate  at temperatures  in
                         excess  of 18QO°F,  residence times
                         greater than  2.0  s,  and  at least 50%
                         excess air. Detailed elementary  and
                         global  oxidation  kinetic  models
                         consistently  show that  destruction
                         efficiencies (DE's)  for  POHCs under
                         these conditions  are  greater  than
                         99.9999% for  even the  most  stable
                         compounds. Similar  calculations for PICs
                         result  in  yields  <  0.000001  (Table  2,
                         columns 3, 4, and 6).
                           Since  the nominal operating range  of
                         incinerators precludes measurable POHC
                         or PIC emissions, we are  left with the
                         conclusion that temporal  or  spatial
                         excursions from the  measured conditions
                         are responsible  for the observed
                         emissions. These "failure  modes"  may
                         be due  to excursions  in  temperature
                         (thermal), residence time (temporal),  or
                         oxygen  concentration  (mixing).  As
                         examples, a spatial,  thermal-failure mode
                         could be a cold gas boundary layer in  a
                         boiler whereas a temporal, thermal-failure
                         mode  would be  a  simple  flameout.   A
                         spatial, residence-time failure could be
                         deviation  from ideal  plug flow such that
                         residence-time short circuits are possible.
                         A temporal, residence-time  failure  could
                         be loss  of atomization pressure in  a
                         burner such theat very large droplets are
                         produced that do not vaporize until  they
                         have traversed  much of the combustion
                         chamber. A temporal, mixing failure could
                         be poor  waste/air micromixing on the
                         molecular level due  to flash volatilization,
                         whereas a spatial failure could simply be
                         poor burner  placement or  location  of
                         secondary air ports.
Hypotheses of PIC Formation
  In addition to these theories  that the
preponderance of the data support,  a
body of qualitative and semi-quantitative
laboratory  and field  data  have led to
certain  observations  concerning  PIC
emissions. We refer to these as partially
verified hypotheses.

-------
oxidation:
pyrolysis:
                                    v-z                        y~2
                     CxHyClz + (x + J__ )O2 = x CO2 + z HCI +	  H2O + trace organics and CO fory>z
                                     4                          4
CxHyClz + xO2  = x CO2 + y HCI + —-  CI2 + trace organics and CO
                                  2
CxHyClz = x C •* zHCt + -^-  H2  + trace organics

CxHyClz = xC + yHCI +  -flL C/2 + trace organics
                         2
                                                                    foryz

                                                                    for y
-------
                   Table. 1. Experimentally Measured and Equilibrium-Predicted POHC Concentrations from a Full-
                           Scale Trial Bum (All values are in mole fraction)
                                                            Equilibrium Calculation (STANJAN)
                        POHC
Measured
(Trial Burn)
2.65% O2
                                                                       25% Oz
                                                                      Deficient
                                100% O2
                                Deficient
                                       2x10*

                                       1X10-9
                                                                       1X10-'8
                                                 1x10-™

                                                 5x70-23
                   Table 2. Kinetic Calculations of the Destruction Efficiency for PICs1 for Various Flame and
                           Post-Flame Reaction Conditions   •
                                                               Optimal   Thermal   Overall   Overall
                                                       Optimal   Thermal   Failure  (Optimal)  (Failure)
                   	POHC	PIC	Flame DE2   DE3      DE4      D£5      D£6
                    1,1-Dichloroethane  Chloroethene     99.9999+ 99.9999+  99.9997  99.9999+ 99.9999 +
                    Pentachloroethane  Tetrachloroethene 99.9999+ 99.9999+  27.0164  99.9999+ 99.2702
                    Chlorobenzene     Benzene         99.9999+ 99.9999+  92.2267  99.9999+ 99.9223
                    1P1C destruction calculations based on analytical solution of the following reaction scheme :
                                  POHC	•>- PIC	-*f Combustion Endproducts


                    2 DE calculated for optimal flame conditions (Tf = 2780°F, t, = 0.1 s).
                    3 DE calculated for optimal post-flame conditions (Tp( = 2300"F, tp( = 2.0 s).
                    4 DE calculated for suboptimal post-flame conditions (Tpt = 1340°'F, tpf 2.0 s).
                    5 Overall DE based on 99% destruction of the waste experiencing flame conditions and 1%
                     of the waste bypassing the flame but experiencing optimal post-flame conditions.
                    6 Overall DE based on 99% destruction of the waste experiencing flame conditions and 1%
                     of the waste experiencing failure post-flame conditions.
Hypothesis 1—Flame Reactions
Control Bulk or Absolute
Emissions of PICs While Post-
Flame (Thermal) Reactions
Control Relative Emission Rates
  This  hypothesis  is based  on  the
reasonable assumption (Theory 2) that
flame temperatures are high enough that
any  material  experiencing  nominal
oxidizing flame conditions will  be  totally,
destroyed (POHCs and PICs)  (Table 2,
column 3). The small fraction  of  waste
that  somehow escapes  the  flame is
responsible for PIC  emissions.  Since
compounds  fed  as mixtures will
    experience very  similar  destruction
    conditions,  the'relative thermal stability
    (oxidative or  pyrolytic) of the escaping
    POHCs will  determine  their  relative
    emission rates.! In a similar manner, the
    propensity for formation of PICs and their
    stability under post-flame conditions will
    control  their rjelative emission rates
    (Table 2, column 7). Since the bulk of the
    material is destroyed  in  the  flame,
    operating  ancl  design  parameters
    affecting the  flame will control  overall
    waste destruction efficiency.  Post-flame
    conditions  may,  however,  apparently
    control the  relative  DEs of POHCs  and
    relative PIC yields.
                               There are, however, insufficient data on
                             the range  of temperature,  residence
                             times,  and  oxygen concentrations  that
                             can potentially exist  in  an  incinerator
                             flame to absolutely state that no PICs are
                             emitted from  the flame. Although a  two-
                             zone system seems to be conceptually
                             useful,  there  is  also a transition zone
                             between the flame and  post-flame zones
                             where  the  distinction is  blurred.  In
                             addition,  differing vaporization  rates  of
                             compounds can,  in principle, result in not
                             all species  experiencing the exact same
                             combustion  conditions.  Preliminary
                             calculations  suggest,  however,  that

-------
droplet vaporization does not significantly
affect relative DEs.

Hypothesis 2—Most PIC
Emissions are the Result of
Pyrolysis Pathways
  This  is essentially a  narrowing of the
focus  of  the failure  mode  concept
presented  as the  second theory of PIC
formation.  Theoretical  calculations
concerning the oxidation  of the  most
stable  organic  PICs  indicate that
temperatures below 1450°F for  post-
flame residence times of 2.0 s  (Table 2,
columns 5 and 7) or residence times less
than 0.1 s, for post-flame temperatures of
2300°F are necessary for measurable
PIC  breakthrough.  Since these
temperatures are  so  low, it seems
reasonable that  pyrolysis pathways
(which are much  slower than  oxidation
reactions)  may be responsible for a large
fraction of POHC  and  PIC emissions.
This  hypothesis  has  been  partially
supported  by  comparison of  actual
versus  predicted  CO  emissions. Since
CO  is only destroyed under  pxidative
conditions  (primarily by OH),  its  stack
concentration may be considered an
indicator of the fraction of the incinerator
that is acting as an oxidizer. Theoretical
calculations  indicate  that  CO levels
should  be on the order of  a  few ppb
instead of the observed levels of 1 to 100
ppm, suggesting a significant fraction of
the CO does not experience an oxidative
environment.  Full-scale emissions  of
undestroyed POHCs also agree best with
a theoretical prediction of POHC stability
based  on pyrolysis kinetics.
  Recent  bench-scale  experiments
simulating  the burning of plastics  in  a
rotary  kiln suggest that rapid thermal
degradation results  in a "transient puff"
of hydrocarbons and chlorocarbons that
can overload  the system such  that
intimate   waste/air  mixing  is  not
completed.  Consequently,  high
concentrations of  various PICs and soot
are  observed, apparently due to  a
transient pyrolysis condition.
  Some researchers have argued that
upset  pyrplysis conditions in  full-scale
incineration facilities  actually  result in
lower  PIC remissions. Other researchers,
however, have shown that the high levels
of soot that- are  formed may  act as  a
reservoir  for  the PICs  by  surface
absorption. Although air emissions may
be  mitigated  by  a  high  efficiency
particulate control device, the  total  PIC
yield  and  environmental  burden  is
increased. The impact is  simply shifted
from the air to the land.
  A significant number  of  cases  exist,
however,  where emissions  do  not
correlate with pyrolysis kinetics. In these
instances,  relative  POHC and PIC
emission  rates may  be facility  specific.
For example, very  low  temperature
excursions may exist for which slower
oxidation  reactions can still contribute to
emissions.  Poorly  atomized  aqueous
wastes could result in slow vaporization
of water  soluble  species, resulting in  a
residence time failure.


Hypothesis 3—A Significant
Fraction of the Observed PICs
are Formed Outside the High
Temperature Zones
  A detailed study of the types of PICs
emitted from full-scale hazardous waste
incinerators,  boilers, and kilns suggests
that as high as 50% of observed PICs are
likely  formed  by  radical-radical  and
radical-O2 association reactions (Table 3).
These types of reactions are important as
the  system  temperature  is  lowered
because high activation  energy
unimolecular  and  radical-molecular
reactions are not favorable. This may
explain why thermally fragile PICs such
as chloroform, 1,1,1-trichloroethane, and
carbon tetrachloride  are observed in the
effluent of full-scale systems. In addition,
it has been proposed  and limited full-
scale  data suggests that  formation of
poly chlorinated  dibenzo-p-dioxins
(PCDDs) is due to  surface  catalyzed
reactions in the  cool zones of the
incinerator  (i.e., transfer ducts,  air
pollution control devices).
  Some studies, however, show evidence
of  analytical artifact whereas  others
suggest that volatile, fragile chlorocarbon
PICs  may be stripped from contaminated
scrubber water. Both laboratory and full-
scale  research on  the mechanism  of
PCDD formation is  incomplete, and no
quantitative study of gas-phase formation
of PCDD from  possible precursors ha's
been performed.

Hypothesis 4—A Significant
Fraction of Observed  PICs are
Due to Fuel/Waste Interactions
  Analysis of full-scale  data reveals that
many of the observed PIC  emissions are
aromatic and  polynuclear  aromatic
(PNAs) species, i.e., benzene, toluene,
naphthalene, etc. These compounds are
commonly observed in  the effluent from
most combustion  devices.  Further
analysis,  however,  indicates  that the
presence  of chlorine-containing wastes
increases the emission of these species
as well  as  that of partially chlorinated
hydrocarbons. Although chlorination  of a
stable  hydrocarbon  is  kinetically
unfavorable, chlorinated  radicals may
participate  in condensation-type
molecular growth  reactions  resulting  in
formation of chlorinated aromatic species
and  PNAs.  Chlorine can  also  increase
overall  PIC  emissions  through  the
traditionally  proposed  effect of flame
inhibition.
  Not  enough  parametric full-scale
studies have been performed to confirm
the  hypothesis,  however.  Although
preliminary  laboratory  studies  support
this concept, the chemistry of chlorinated
hydrocarbons  and other  hazardous
materials is only  beginning  to  be
addressed.
Hypothesis 5—Carbon Monoxide
and Total Unburned
Hydrocarbons are Surrogates
for PIC Emissions
  From  a  scientific  viewpoint, the
emission  rate  of  a surrogate should
correlate (viz., have a  statistically
significant relationship  through   a
continuous function)  with the  emission
rate of a specific PIC or total PICs. From
a regulatory viewpoint,  it appears that a
reasonable surrogate must  only have a
concentration greater than that of the PIC
of  interest such  that,  if  the CO
concentration is below a given level, the
PIC will be below a given environmentally
acceptable level.
  The  concept  of  CO as PIC  surrogate
arises  from the fact that CO is  more
stable  than any  organic  PIC.  Total
unburned  hydrocarbons  (TUHC) has
been  proposed  as a  PIC  surrogate
because TUHC, in principle, is the total of
undestroyed  POHC  and  organic  PICs.
Results of some full-scale tests indicate
that  CO and TUHC meet the regulatory
definition of a surrogate for some PICs at
some  facilities.  A  statistically significant
rank/order  correlation between  CO,
TUHC, and  various measures of PIC
emissions was  observed at a few sites.
This  is encouraging  considering  the
limited range of conditions tested and  the
number of data  points available.
  The  concept  that the extreme stability
(slow destruction kinetics) of CO makes it
a suitable  PIC  surrogate  is complicated
by  the fact that the concept  does  not
recognize the formation kinetics of  CO
versus organic  PICs. Under some  failure
conditions,  PIC yields may be  high
whereas CO formation  has yet to reach
its'maximum.

-------
Barry Dellinger, Philip H. Taylor, and Debra A. Tirey are with the University of
   Dayton Research Institute. Dayton. OH 45269.
C. C. Lee is the EPA Project Officer (see below).
The complete report, entitled "Minimization and Control of Hazardous Combustion
   Byproducts," (Order No. PB 90-259 854/AS; Cost: $39.00, 'subject to change)
   will be available only from:
       National Technical Information Service
       5285 Port Royal Road
       Spring field. VA 22161
       Telephone: 703-487-4650
The EPA  Project Officer can be contacted at:
       Risk Reduction Engineering Laboratory
       U.S. Environmental Protection Agency
       Cincinnati, OH 45268
United States
Environmental Protection
Agency
Center for Environmental Research
Information               :
Cincinnati OH 45268
      BULK RATE
POSTAGE & FEES PAID
         EPA
   PERMIT No. G-35
Official Business
Penalty for Private Use $300
EPA/600/S2-90/039

-------