PB84-184902
Laboratory-Scale Flame-Mode
Hazardous Waste Thermal
Destruction Research
Energy and Environmental Research Corp.
Irvine, CA
Prepared for
Industrial Environmental Research Lab.
Cincinnati, OH
Apr 84
U.S. DEPARTMENT OF COMMERCE
National Technical Information Service
nib
-------
FE8U-16U9Q2
EPA-600/2-84-086
April 1984
LABORATORY-SCALE FLAME-MODE HAZARDOUS WASTE
THERMAL DESTRUCTION RESEARCH
by
J. C. Kramlich
M. P. Heap
J. H. Pohl
E. Poncelet
G. S. Samuel sen
W. R. Seeker
Energy and Environmental Research Corporation
18 Mason
Irvine, CA 92714-4190
EPA Prime Contract No. 68-03-3113
SuDcontract Task 24-1
EPA Project Officer: C. C. Lee
INDUSTRIAL ENVIRONMENTAL RESEARCH LABORATORY
OFFICE OF RESEARCH AND DEVELOPMENT
U.S. ENVIRONMENTAL PROTECTION AGENCY
CINCINNATI, OH 45268
-------
TECHNICAL REPORT DATA
fPtesie read lnsmicnoru on wc reverz: before completing)
1. REPORT NO. ?-
EPA-600/2-84-086
X RECIPIENT'S ACCESSION NO.
PB8 A 1 8 k 9 0 2
4. TITLE AND SUBTITLE
Laboratory-Scale Flame-Mode Hazardous Waste Thermal
Destruction Research
5. REPORT DATE
April 1984
6. PERFORMING ORGANIZATION CODE -
t
7 AUTHORISI
J.C. Kramlich, et.al.
a. performing organization report no.
9. PERFORMING ORGANIZATION name and adoress
J.C. Kramlich, et.al.
EERC
18 M;son St.
Trv-fna C& 0971A
10. program element no.
CBRD1A
11. contract/grant NO.
Contract 68-03-3113
12. SPONSC >NG AGENCY NAME ANO ADDRESS
USEP-1
InCLitrial Waste Combustion Group
Ci>--:nnatf, Ohio 45268
12,TYPE OF.REPORT ANO PERIOD COVERtD
Research report
14. SPONSORING AGENCY CODE
EPA/600/12
15. S '.'5NTARY noteS
IS - ^ACT
¦ ms research is to investigate the flame mode incinerability of hazardous waste
co" vnds. It was also designed to provide a comparison between flame and non-flame
des uction of expounds and act as a guideline for future work on the development of
an .ccsptable incinerability ranking metfodology.
Two flame reactors were used in order to simulate a wide variety of failure con-
ditions for liquid injection incinerators. The first reactor was called a microspray
and consisted of drcplets injected into a hydrocarbon flat-flame. The second reactor
was a turbulent fia.-s reactor which allowed the investigation of failure conditions in
spray flames such as mixing, atomization and quench phenomenon. The compounds investi-
gated include!: chloroform, 1,2-dichloroethane, benzene, chlorobenzene, and acrylo-
nitrile.
The results indicated that flames without long residence time postflame zones were
capable of d;Svroying all the compounds investigated to high efficiencies (>99.99%) whop
operated corrc-ctly. However, under failure conditions such as poor atomization, low
excess air, lew flame temperature, and quenching, the destruction efficiencies were
typically 90-99.9%.
17. KEY WORDS ANO DOCUMENT ANALYSIS
4. DESCRIPTORS
b.IDENTIFIERS/OPEN ENDED TERMS
c. COSATi Field/Group
Hazardous compound incinerability
mode of combustion failures
Flame-mode destruction
non-flame thermal decomposition
Hazardous waste
disposal.
ib. distribution statement
Release to Public
19 SECURITY CLASS (Thu Report)
Unclassified
21. NO. OF PAGES
160
20. SECURITY CLASS (Thu page}
Unclassified
22. PRICE
EPA Form 2220-1 (R«»- 4-77) prcviouj coition is oasoucre
i
-------
NOTICE
This document has been reviewed in accordance with
U.S. Environmental Protection Agency policy and
approved for publication. Mention of trade names
or commercial products does not constitute endorse-
ment or recommendation for use.
ii
-------
FOREWORD
When energy and material resources are extracted, processed, converted,
and used, the related pollution impact on our environment and even our health
often requires that new and increasingly more efficient pollution control
methods be used. The Industrial Environmental Research Laboratory - Cincin-
nati (IERL-Ci) assists in developing and demonstrating new and improved method-
ologies that will meet these needs both efficiently and economically.
This report presents the research results from laboratory-scale flame
mode combustion of organic compounds. The research objective was to compare
compound incinerability ranking methods and to evaluate differences between
non-flame and flame combustion of organic compounds. The results indicated
that flames without long residence time post-flame zones were capable of
destroying all the compounds investigated to high efficiency (greater than
99.999%) when operated correctly. However, under failure conditions such as
poor atomization, low excess air, low flame temperature and quenching, the
destruction efficiency was only 90-99.9 percent. The data also indicated that
exhaust measurements of carbon monoxide and total hydrocarbons could be used
as an indication of operation in failure conditions.
The result of this study will be useful to regulatory programs responsible
for implementation of Resource Conservation and Recovery Act (1976), to owners
and operators of hazardous waste facilities, and to firms anticipating entry
into the manufacture of incinerator equipment.
iii
-------
ABSTRACT
This constitutes the final report of a program to investigate the flame
mode incinerability of hazardous waste compounds. The objective of the pro-
gram was to generate fundamental flame mode data on a series of compounds in
order to make a preliminary comparison of proposed incinerability rankings.
It was also designed to provide a comparison between flame and non-flame
destruction of compounds and act as a guideline for future work on the develop-
ment of an acceptable incinerability ranking methodology.
Two flame reactors were used in order to simulate a wide variety of
failure conditions for liquid injection incinerators. The approach was to
burn a mixture of selected hazardous waste compounds in reactors operated
under conditions where incomplete destruction was allowed to occur. The first
reactor was called a microspray and consisted of droplets injected into a
hydrocarbon flat-flame. The failure conditions explored with this reactor
were only thermal, since the flame was premixed. The second reactor was a
turbulent flame reactor which allowed the investigation of other failure
conditions in spray flames such as mixing, atomization and quench phenomenon.
The comDounds investigated included hazardous compounds that have a wide range
of incinerability rankings: chloroform, 1,2-dichloroethane, benzene, chloro-
benzene, and acrylqnjjitrile.
The results indicated that flames without long residence time post-flame
zones were capable of destroying all the compounds investigated to high effi-
ciencies (>99.999%) when operated correctly. However, under failure conditions
such as poor atomization, low excess air, low flame temperature, and quenching,
the destruction efficiencies were tyDically 90-99.9%. The data indicated that
exhaust measurements of CO and total hydrocarbons could be used as an indica-
tion of operation in a failure condition. The destruction efficiency was
dependent on compound type and failure condition, although the differences
between compounds was small. When compared to the proposed ranking procedure
no one ranking procedure was able to predict the compound ranking for every
failure condition. However, some procedures were able to predict the ranking
for specific failure conditions. For example, the non-flame temperature
requirements correctly identified the compound order for premixed conditions
when the flame temperature was insufficient to burn the compounds. These data
indicate that incinerability should be tied to likely failure conditions
for the incinerator under consideration and that more data is required for
a wide range of failure conditions and compounds.
iv
-------
ACKNOWLEDGEMENTS
This work was performed under subcontract (Task 24-1) from
JRB Associates. The U.S. Environmental Protection Agency prime contract
number was 68-02-3113. C. C. Lee was the EPA Technical Project Officer and
V. S. Engleman was the JRB Associates Task Manager. Other members of the EPA
Technical Advisory Committee who assisted in program guidance were: A. F.
Sarofim, Massachusetts Institute of Technology; F. W. Marble, California
Institute of Technology, R. M. Fristrom, The Johns Hopkins University;
B. Dellinger, University of Dayton Research Institute; W. Tsang, National
Bureau of Standards; R. A. Carnes, Industrial Environmental Research Labora-
tory - Cincinnati EPA; and E. P. Crumpler, Office of Solid Waste, EPA.
We wish to acknowledge the contributions of R. K, Nihart and H. D. Crum,
who played important roles in the development of the analytical approach and
experimental construction and operation.
V
-------
TABLE OF CONTENTS
Section Page
1.0 EXECUTIVE SUMMARY 1-1
1.1 Introduction 1-T
1.2 Experimental Approach 1-5
1.3 Microspray Results 1-7
1.4 Turbulent Flame Reactor Results 1-10
1.5 Discussion 1-18
1.6 Conclusions 1-23
1.7 Executive Summary References 1-25
2.0 INTRODUCTION 2-1
3.0 EXPERIMENTAL EQUIPMENT 3-1
3.1 Microspray Reactor 3-2
3.2 Turbulent Fls^.e ".eactor 3-9
3.3 Laborator 7 ¦ r ty 3-14
3 4 i.r- Systems 3-14
'. . easurement of Destruction Efficiency (DE) . . . 3-15
3. .1 Other Measurements 3-16
4.0 RESULTS AND DISCUSSION 4-1
4.1 Microspray--Drop1et Decomposition in Flames 4-1
4.1.1 Droplets of Mixtures in an Oxygen-Rich Flame . . . 4-2
4.1.2 Pure Compounds in an Oxygen-Rich Flame 4-5
4.1.3 Compound Mixtures with Low Oxygen 4-8
4.1.4 Summary of Microspray Rankings and Data 4-13
4.2 Turbulent Flame Reactor Results 4-16
4.2.1 Influence of Stoichiometry and Load 4-19
4.2.2 Influence of Theoretical Air 4-25
4.2.3 Influence of Air Velocity 4-28
4.2.4 Influence of Compound Concentration 4.30
vii
-------
TABLE OF CONTENTS (Concluded)
Section Page
4.2.5 Effect of Quench Coil 4-33
4.2.6 Effect of Nozzle Performance
4.2.7 Effect of Auxiliary Fuel 4-32
4.2.8 Turbulent Flow Reactor Data Summary 4-41
5.0 SUMMARY OF RANKINGS, CONCLUSIONS, AND RECOMMENDATIONS ... 5-1
5.1 Summary and Discussion of Rankings 5-1
5.2 Conclusions 5-6
5.3 Recommendations 5-7
6.0 REFERENCES 6-1
APPENDIX A—EXPERIMENTAL FLOW CALIBRATIONS A-l
APPENDIX B—ANALYTICAL TECHNIQUE FOR TEST COMPOUNDS B-l
APPENDIX C—THERMAL DECOMPOSITION MODEL C-l
APPENDIX D—RAW DATA D-l
vili
-------
LIST OF FIGURES
Figure Page
1-1 Fraction of test compound remaining in exhaust when 38 ym 1-8
droplets of mixtures of compounds were injected into lean
(10 percent excess oxygen) H2/air flames as function of
flame temperature
1-2 Fraction of test compound remaining in exhaust when 38 ym
droplets of mixtures of compounds were injected into rich
(stoichiometric ratio = 0.83) H^/air flame as function of
flame temperature. Incinerability order at 1050 K is
chlorobenzene, benzene, acrylonitrile, and chloroform .... 1-11
1-3 Exhaust CO and total hydrocarbons and fraction of test
compound remaining in exhaust as a function of theoreti-
cal air (constant air velocity, variable load, equal
molar mixture of chloroform, benzene, chlorobenzene,
and acrylonitrile added 3 percent by weight to heptane ... 1-13
1-4 Impact of theoretical air on CO and DRE from turbulent
flame reactor. Incinerability order at 150 percent
theoretical air is chloroform, acrylonitrile, benzene, and
chlorobenzene (constant load = 48 kW; variable air flow
rate and burner velocity, equal molar mixture of compounds
added 3 percent by weight to heptane) 1-14
1-5 Impact of theoretical air and load on fraction of test
compounds remaining in exhaust of turbulent flame
reactor (constant air velocity, variable load 42-24 kW;
equal molar mixture of compounds added 3 percent by
weight to heptane) 1-16
1-6 Impact of atomization quality on CO and fraction of test
compounds remaining in exhaust of turbulent flame reactor
(constant air velocity, variable load: 42-16 kW; equal
molar mixture of compounds added 3 percent by weight to
heptane 1-7
1-7 Impact of cooling coil placed in flame on CO and fraction
of test compound remaining in exhaust of turbulent flame
reactor. (Constant air velocity; load = 32 kW equal
molar mixture of compounds added 3 percent by weight to
heptane) 1-19
1-8 Comparison of proposed ranking techniques arid concentra-
tion measured in the experiments under flame failure con-
ditions normalized so most predominant compound shows
full scale 1-22
ix
-------
LIST OF FIGURES (Continued)
Figure Page
3-1 Microspray reactor 3-3
3-2 Details of microspray reactor 3_5
3-3 Schematic diagram of the droplet generator 3_5
3-4 Gas temperature measurements in microspray reactor 3.3
3-5 Turbulent flame reactor 3_13
3-6 Turbulent flow reactor burner detail 3_12
4-1 Fraction of test compound remaining in exhaust when 38 urn
droplets of mixtures of compounds were injected into lean
(10 percent postflame oxygen) h^/air flames as function
of flame temperature 4_3
4-2 Fraction of test compound remaining in exhaust when 38 um
droplets of pure compounds were injected into lean (10 per-
cent postflame oxygen) H2/air flames as function of flame
temperature 4.5
4-3 Fraction of test compound*remaining in exhaust when 38 um
droplets of mixtures of compounds were injected into rich
stoichiometric ratio = 0.83) H^/air flames as function of
flame temperature 4_g
4-4 Model results: fraction of test compound remaining in
exhaust when 38 um droplets were exposed to the tempera-
ture profiles shown in Figure 3-4a and allowed to react
by nonflame thermal decomposition kinetics 4_n
4-5 Model results: effect of oxygen concentration on ther-
mal decomposition for benzene at isothermal conditions . . . 4.14
4-6 Comparison of calculated reactor performance assuming
1) droplet evaporation controls DE, 2) droplet ignition
controls DE, and 3) thermal decomposition kinetics con-
trol DE; with microspray oxidation mixture data and
microspray low oxygen data 4_15
4-7 Microspray incinerability rankings 4_17
4-8 Exhaust CO, total hydrocarbons, and relative heptane as a
function of theoretical air (constant air velocity, var-
iable load: 42-24 kW, equal molar mixture of chloroform,
benzene, chlorobenzene, and acrylonitrile added 3 percent
by weight to heptane 4_22
x
-------
LIST OF FIGURES (Continued)
Figure Page
4-9 Impact of theoretical air and load on fraction of test
compounds remaining in exhaust of turbulent flame reactor
(constant air velocity, variable load: 42-24 kM; equal
molar mixture of compounds added 3 percent by weight to
heptane) 4-24
4-10 Impact of theoretical air on CO and DRE from turbulent
flame reactor (constant load: 48 kW; variable air flow
rate and burner velocity; equal molar mixture of compounds
added 3 percent by weight to heptane 4-27
4-11 Impact of burner air velocity on fraction of test com-
pounds remaining in exhaust of turbulent flow reactor
(156 percent theoretical air; 0.72 gm/sec fuel flow;
3-weight-percent equimolar mixture of test compounds
in heptane; 32.3 kW load) 4-29
4-12 Impact of compound concentration on fraction of test com-
pound remaining in exhaust of turbulent flow reactor (con-
stant air velocity, variable load: 39-26 kW; variable
weight concentration of an equimolar mixture of test com-
pounds in heptane) 4-31/32
4-13 Influence of quench-coil on CO in exhaust of turbulent
flame reactor as a function of load and theoretical air
(constant air velocity, variable load: 42-16 kW; equi-
molar mixture of compounds added 3 percent by weight
to heptane) 4-34
4-14 Influence of quench-coil on fraction of test compound
remaining in exhaust of turbulent flame reactor as a
function of load and theoretical air (constant air veloc-
ity, variable load: 42-24 kW; equimolar mixture of com-
pounds added 3 percent by weight to heptane 4-35
4-15 Influence of atomization quality on CO in exhaust of
turbulent flame reactor as a function of load and theo-
retical air (constant air velocity, variable load:
42-16 kW; equimolar mixture of compounds added 3 per-
cent by weight to heptane). Also, atomization pres-
sure as a function of theoretical air for each of the
nozzles (parameter is nozzle capacity in gal/hr) 4-37
4-16 Impact of atomizer performance on fraction of test com-
pound remaining in exhaust as a function of percent
theoretical air (constant air velocity, variable load:
42-16 kW; equimolar mixture of compounds added 3 percent
by weight to heptane) 4-39
XI
-------
LIST OF FIGURES (Concluded)
Figure Page
4-17 Impact of auxiliary fuel type on fraction of test compound
remaining in exhaust as a function of percent theoretical
air (constant air velocity, variable load: 42-24 kW; equi-
molar mixture of compounds added 3 percent by weight to
No. 2 fuel oil) 4-40
5-1 Comparison of proposed ranking techniques and relative
compound decomposition of compounds under flame failure
conditions 5-4
xii
-------
LIST OF TABLES
Table Page
2-1 Test Compounds and Properties 2-6
3-1 Normal Operating Conditions for Turbulent Flame Reactor . . . 3-13
3-2 Experimental Measurements 3-17
3-3 Chlorine Mass Closure 3-18
4-1 Turbulent Flame Experimental Conditions 4-20
4-2 Atomi-zation Parameters 4-21
xiii
-------
1.0 EXECUTIVE SUMMARY
1.1 Introduction
Permitting procedures for hazardous waste incinerators are defined by
the Resource Conservation and Recovery Act (RCRA). A permit to operate is
issued after a trial burn has been executed or other appropriate test data
obtained which demonstrate that the incinerator satisfactorily converts
hazardous waste into non-hazardous compounds when operated under specified
conditions. Satisfactory conversion is defined in terms of destruction and
removal efficiency (DRE). However, since most hazardous waste streams contain
many compounds, a trial burn which involves the measurement of all of them would
be prohibitively expensive. Consequently, the trial burn involves the measure-
ment of a subset of compounds (the principal organic hazardous constituents—
POHCs) which are present in the input stream. If the DRE of these POHCs is
99.99 percent or greater, and certain other conditions met (e.g., chlorine and
particulate matter removal and emissions standards) then a permit to operate
is granted. Thus, the burden of responsibility rests with the permit writer
who must select the subset of compounds (POHCs) based upon concentration and
incinerability. This constitutes the final report of a project which was
carried out to examine methods of ranking incinerability, and to compare flame
vs non-flame waste destruction.
Several procedures have been proposed to rank incinerability (Cudahy et
al., 1981) namely:
• The heat of combustion.
• The autoignition temperature (AIT).
• A computational approach based upon AIT, compound structure,
and other compound-dependent parameters (Lee et al., 1979, 1982).
e The temperature necessary for a given destruction level within
a given time under dilute premixed conditions (Tgg gg) (Dellinger
et al., 1982; Duvall and Rubey, 1976, 1977; Lee et al., 1979,
1982).
• Susceptibility of the compound bond structure to attack by
flame radicals (Tsang and Shaub, 1981).
1-1
-------
These procedures have their merits but fail to take into account all the
conditions which may exist in actual incinerators. The heat of combustion
for example, of a particular compound may be insignificant if it is present
in small quantities and is mixed with an auxiliary fuel. In addition, some
of these procedures do not consider processes and reactions that occur in
flames. The times and temperatures which exist under non-flame experimental
conditions may be inappropriate for large-scale diffusion flames.
The concept of incinerability is used to describe the relative degree of
difficulty of incineration of the various hazardous organic constituents
present in a given waste stream. If during the trial burn^rvit is demonstrated
that these compounds which are most difficult to destroy have a DRE greater
than 99.99 percent, then it is assumed that compounds ranked more incinerable
under the accepted hierarchy will be destroyed at the same or greater DRE than
the difficult compounds. Thus, there is a need for some ranking methodology
that will aid the permit writer in his selection of difficult compounds. If
the ranking methodology is in error, or is not applicable to a particular
system, then a condition could exist wherein a POHC was destroyed satisfactor-
ily, but other hazardous compounds in the waste stream were not destroyed
sufficiently. Under these circumstances, a trial burn designed to measure
only the POHC may have incorrectly demonstrated the satisfactory operation of
the incinerator.
Because of the nature of flames, waste compounds which experience a flame
environment are rapidly and completely destroyed. This can be demonstrated by
considering non-flame thermal decomposition data obtained under dilute pre-
mixed conditions (Dellinger et al., 1982; Ouvall and Rubey, 1976, 1977; Lee
et al., 1979, 1982}. As an example, non-flame data indicates that chloroben-
zene would decompose to 99.99 percent of its original concentration in 1 sec.
at 1038°K (Lee, et al., 1982}. At typical flame temperatures (approximately
2Q00°K), the time required to obtain the same destruction level is much
-13
smaller (<10 sec. using the same thermal decomposition data) than the
typical 0.10 sec. flame residence time (Perry, et al., 1963).
Thus, non-flame thermal decomposition data obtained under dilute pre-
mixed conditions indicate that temperatures much lower than those encountered
1-2
-------
in typical incinerator flames will destroy all the organic hazardous waste
compounds which have been tested to date. Also, because of high reactant
concentrations in flames, free radicals which must be present to propagate
the flame will contribute to destruction of the compounds in the flame
(Seeker et al., 1981 a). These free radicals will increase the rate of decom-
position above those predicted from dilute decomposition kinetics. Under
ideal flame conditions, in which all of the waste is exposed to flame tempera-
tures, the concept of incinerability has little significance since all hazar-
dous compounds would be expected to be completely destroyed.
Incomplete destruction of a hazardous waste compound in an actual incin-
erator must be caused by conditions which allow some of the material to escape
or bypass the flame since organic compounds are destroyed rapidly in a flame
environment. Most incinerators include long residence time hold-up zones or
after-burners to destroy material which is not completely reacted in the flame
zone. Thus, fncinerability would be expected to be influenced not only by the
chemical properties of the compound but also by its physical properties and
their interaction with the incinerator operating conditions because these may
influence the failure mode. The term failure mode is used to describe those
conditions which might occur in a practical incinerator which preclude com-
plete processing of the waste material by a high-temperature turbulent diffu-
sion flame. Thus, the term in the present context does not include conditions
which may affect other parts of an incinerator (e.g., after-burner o1" scrubber).
It is important to evaluate incinerability under conditions which simulate
those failure modes which could occur in practice.
Various phenomena account for the failure of turbulent diffusion flames,
typical of those used in liquid injection incinerators, to completely destroy
a liquid waste. The destruction efficiency in the flame may be less than
quantitative (100 percent) because of any of the following reasons:
1. Atomization Parameters. When the waste material is injected
as a liquid which must be atomized, poor destruction efficiency
can result from inappropriate atomization. (a) Droplets which
are too large may be produced, (b) Their trajectory may be such
1-3
-------
that they penetrate the flame zone and ignition does not
occur, (c) Droplets which are too small may promote con-
centrated evaporation zones which produce fuel-rich pockets.
2. Mixing Parameters. In a turbulent diffusion flame the
reactants are supplied separately and reactant contacting
takes place via turbulent mixing. Poor mixing can result
in low destruction efficiencies because the waste material
may not be mixed with oxygen before it escapes from the
flame region.
3. Thermal Parameters. The destruction efficiency may be low
because flame temperatures are too low. This can occur if the
calorific value of the waste/auxiliary-fuel mixture is low
or heat removal rates are high.
4. Quenching Parameters. The reactants can be quenched before
destruction is complete by heterogeneous or homogenous
phenomena. Quench rates are high due to mixing with
excessive excess air levels in duel injection systems
in which the flame impinges on an aqueous jet, or the flame
may contact a relatively cool surface.
Consequently, it is essential to investigate the concept of incinerabi1ity
in flames under conditions which could account for a failure to completely
destroy the waste compound and are typical of real systems.
The primary goal of this study was to compare the incinerabi1ity ranking
procedures which have been proposed with those measured under flame conditions
typical of liquid injection incinerators. The approach utilized was to mea-
sure the exhaust compound concentration under different simulated failure
modes and to compare the ordering of the compounds to those given by several
incinerability ranking procedures. Failure conditions were simulated for all
the parameters expected to influence incinerator performance; i.e., atomiza-
tion, mixing, thermal and quenching. To simulate all of them required two
reactors. A microspray reactor consisting of a laminar premixed flat flame
into which test compounds were injected,was used to investigate thermal
parameters. A subscale turbulent diffusion spray-flame was used to investigate
1-4
-------
atomization, mixing and quenching parameters. Secondary goals included the
generation of fundamental flame mode destruction data necessary to compare
flame and non-flame decomposition. The results are primarily a means of
guiding future experimental work, since further work is necessary to select
a reasonable ranking orotocol.
1.2 Experimental Approach
Extensive investigations are being carried out at the University of
Dayton Research Institute under EPA sponsorship to define the kinetics of
waste decomposition in post-flame regions (Dellinger et al., 1982; Duval! and
Rubey, 1976, 1977). The emphasis of the present study was on the flame zone
itself and the impact of failure conditions associated with mixing, thermal,
quenching, and atomization parameters on the relative destruction of five
compounds. These compounds were selected because they represented a broad
range of incinerability as defined by existing ranking procedures, and because
data within each of the procedures were available for the compounds. The
study was restricted to conditions typical of liquid injection incinerators
No attempt was made to include phenomena associated with waste destruction in
beds such as that exist in fluidized beds, rotary kilns or hearth incinerators.
Two flame reactors were used to study destruction efficiency under different
condi tions:
1. Microspray Reactor. In the microspray reactor,monodisperse
waste droplets were injected into a hot, uniform post-flame
gas. These experiments investigated the destruction efficiency
(DE) behavior and ranking that resulted from individual droplet
evaporation and flame decomposition reactions. The experiment
was designed to bridge the gap between the non-flame thermal
decomposition experiments and the turbulent flame data. As
such, it included two processes in addition to the thermal de-
composition experiments: droplet vaporization dynamics and
flame reactions. The data were used for the'following purposes:
• To determine what portion of the turbulent flame
rankings were due to laminar flame and evaporation
processes.
1-5
-------
• To compare flame (microspray) vs. non-flame (thermal
decomposition) destruction on a fundamental level
without the complicating influence of turbulence.
2. Turbulent Flame Reactor. A turbulent flame reactor (TFR) was
used to investigate OE and ranking in a turbulent spray diffu-
sion flame. The TFR was operated under conditions to simulate
many of the processes occurring in the flame zone of a liquid
injection incinerator; these could be exaggerated to simulate
different failure modes.
Five compounds (chloroform, acrylonitrile, benzene, chlorobenzene, and
1, 2-dichloroethane) were selected as representative of liquid organic hazard-
ous wastes. All the compounds are listed in the 1980 RCRA regulations, Part
261, Appendix VIII (Federal Register: May 19, 1980). The compounds were
chosen to represent a broad range of incinerability based on the most commonly
proposed ranking procedures. They cover greater than 90 percent of the range
in heats of combustion for the listed compounds (.13 to 10.14 kcal/gm). Since
a direct comparison between non-flame thermal decomposition rankings and the
flame-mode destruction was an objective of this study, compounds were selected
for testing for which non-flame data were available. In addition, the selec-
tion also took into account the NBS ranking system, a range of auto-ignition
temperatures and a variety of molecular structures. Two of the compounds are
aromatic, one is a highly chlorinated methane, another is a chlorinated ethane
and one contains nitrogen.
Compound DE was measured in the reactor exhaust by adsorption onto
Tenax-GC, followed by thermal desorption and flame ionization gas chromato-
graphic analysis. The use of Tenax for concentrating the sample provided
the necessary rapid turnover of samples with sufficient separation and sensi-
tivity. The breakthrough volumes of all the test compounds were directly
measured and were found to be greater than the utilized sample volumes.
Benezene and 1, 2-dichloroethane were not separable by the column and hence
mixtures containing both compounds were avoided. Appendix B details the
design, operating procedures, and verification/quality assurance work performed
on the DE measurement technique.
1-6
-------
1.3 Microspray Results
'The microspray was used to investigate the impact of thermal parameters
for two conditions:
• Fuel-lean - Excess oxygen available to oxidize test
compounds.
• Fuel-rich - Insufficient oxygen available to oxidize
test compounds.
In addition, the effect of using pure compounds was compared with that for
mixtures of compounds. The other failure mode parameters (atomization, quench-
ing, and mixing) cannot be effectively investigated in the microspray reactor
and were investigated in the turbulent flame reactor.
Figure 1-1 presents data for two mixtures of four compounds shown separ-
ately in Figure 1 -1 a and 1-lb. In these tests, 38 ym droplets of the two
mixtures were injected separately into a lean (10 percent excess oxygen)
f^/Air/^ flame with different flat-flame temperatures. Exhaust concentra-
tions of the individual test compounds were measured and the data are shown
in Figure 1-1 in terms of the fraction of each compound remaining versus the
measured flat-flame temperature. This temperature is determined by extrapo-
lating the axial temperature measurements to the burner face and is the highest
temperature of the flat flame gas (Appendix D). Under these excess oxygen
conditions, flames were observed to surround each individual droplet for both
mixtures for flat-flame temperatures in excess of 850 K. However, the minimum
droplet ignition temperature was observed at slightly lower temperatures for
the 1, 2-dichloroethane mixture, probably due to the substitution of compounds.
When the flat-flame temperature is greater than the ignition temperature of
the specific compound mixture, the exhaust concentration of the test compounds
were below the detection limit of the analytical technique which indicated a
destruction level in excess of 99.995 percent.
Calculations using non-flame kinetics indicate that almost no decompo-
sition should occur below 800 K for the residence times (^ 1 sec.) available
in the microspray reactor. However, as shown in Figure 1-1, significant
1-7
-------
in
3
"3
c
e
s
a
u
«
i.
ui
ai
ex
s
a
u
1.0
"3 0.8
UJ
CI
c
0.6 ,—
2 0.4 —'
0.2
Visual
lanition
0.8 -
Q Chlorooenzsne
•& 8enzene
# 1,2-Qichloroethane
O Chlorofom
A Acrylonitrile
700 300 900
Flat-Flame Temperature (K)
1000
(a) Mixture containing Oicnloroethane, Chlorobenzsne
Chloroform and Acrylonitrile
Visual
Ignition
700 800 900
Flat-Flame Tenoerature (;<)
1000
(b) Mixture containing 3enzene, Chlorobenzsne
Chloroform and Acrylonitrile
Figure 1-1. Fraction of test compound remaining in exhaust when
38 um droplets of mixtures of compounds were injected
into lean (10 percent excess oxygen) H-/air flames as
function of flame temperature.
1-8
-------
destruction was measured at flat-flame temperatures below 800 K. This destruc-
tion atlow flat-flame temperatures is probably due to a local increase in
temperature around droplets and flame radical attack. For gas temperatures at
and above the point at which the individual droplets were observed to support
flames, all the compounds were destroyed, but below the ignition temperature
the fraction destroyed depended upon the compound. At low flat-flame tempera-
tures for the dichloroethane mixture, the ranking from highest to lowest con-
centration was: chlorobenzene, dichloroethane, chloroform, and acrylonitrite.
At flat-flame temperatures just below the droplet ignition point, again chloro-
benzene was found to be the most difficult compound to be eliminated but the
other compounds showed some rearrangement in ranking; however, the effect of
compound type is small. When benzene was substituted for dichloroethane (Fig-
ure 1-lb), chlorobenzene remained the most prominent compound in the exhaust
followed by benzene, chloroform, and acrylonitrile. Again, just below ignition
there was some reordering of compounds with chloroform becoming the easiest to
eliminate.
These data indicate that for single droplet oxidative conditions where the
flame temperature is too low for droplet ignition, a particular order of com-
pounds does exist in terms of the fraction remaining in the exhaust. This
order is chlorobenzene, benzene, 1, 2-dichloroethane, chloroform, and acrylo-
nitrile. However, this order changes as the temperatures reach the ignition
point. The ordering just below ignition is identical to the ordering suggested
by Tgg gg and auto-ignition temperature.
When 38 um droplets of pure compounds were injected into oxygen-rich
fuel-lean flame products, the droplets were observed to ignite at different
temperatures. For example, visual ignition for chloroform droplets was
observed at 860 K, while dichloroethane ignited at 850 K, acrylonitrile at
800 K, and chlorobenzene at 740 K. Benzene had the lowest ignition temperature
and was observed to ignite at temperatures below 600 K. For pure compounds,
the destruction is controlled by droplet ignition. The observed ignition
temperature does not agree with any proposed incinerability ranking procedures
although the heat of combustion criteria is almost the same with the exception
that acrylonitrile and chlorobenzene are reversed. A potential explanation
for this behavior is that the ability to support droplet flames is determined
to first order, by the heat release available upon droplet combustion (AHc).
1-9
-------
Although second order effects may modify the rankings, for pure droplets the
ranking appears to be dominated by heat of combustion.
The absence of oxygen was the third failure mode investigated with the
microspray reactor. Droplets of equal molar mixtures of compounds were
injected into fuel-rich (stoichiometric ratio = 0.83) F^/air/^' flames of
different temperatures. In these tests, the oxygen was rapidly and completely
consumed by the hydrogen in the flat-flame so that no oxygen was available to
oxidize the test compounds. The fraction of each compound remaining in the
exhaust as a function of the flat-flame temperature is shown in Figure 1-2.
Even with mixtures, the temperature required to destroy the compounds, 105") !'*,
was found to be very similar to the Tgg gg temperatures of the individual
compounds (920 to 1037 K; see Table 2-1) and were much higher than those
required if droplet ignition occurred (Figure 1-1}. The fractional destruction
was strongly dependent upon flame temperature. In fact, the data show that a
very small change in flame temperature above 1050 K produced a substantial
change in the compound concentrations, particularly for benzene. A difference
between the compounds was observed only at a temperature just below the flat-
flame temperature required for complete destruction. At that temperature, the
compound that was most predominant was chlorobenzene, followed by benzene,
chloroform, and acrylonitrile. This ranking was identical to that measured
for the low temperature oxidation data (Figure 1-1). The non-flame Tgg gg did
identify the temperature range required for complete destruction and the most
predominant compounds (chlorobenzene and benzene); however, acrylonitrile and
chloroform are reversed from the Tgg gg ranking.
1.4 Turbulent Flame Reactor Results
The turbulent flame reactor was operated and tested under a number of
conditions. However, many of these conditions resulted in high destruction
efficiency of all the test compounds. Only those parameters resulting in
significant deterioration of destruction efficiency are presented. Data on
high destruction levels are presented elsewhere (Section 4). The conditions
investigated in the turbulent flame reactor which had a strong influence on
destruction efficiency were primarily associated ^n'th three failure parameters:
• Atomization parameter - poor atomization quality
1-10
-------
V)
3
«
-C
X
cr>
ta
4)
C£
C
o
u
(0
i-
1.0
0.8
0.6
0.4
0.2
**fF
— ~ Chlorobenzene
£ Benzene
— A Aery 1 on i tri 1 e
O Chloroform
1
A
A
A
8
600
800 1000 1200
Flame Temperature (K)
Figure 1-2. Fraction of test compound remaining in exhaust when
38 um droolets of mixtures of compounds where injected
into rich (stoichiometric ratio = 0.83) H^/air flame
as a function of flame temperature. Incinerability
order at 1050°K (highest to lowest concentration) is
chlorobenzene, benzene, acrylonitrile, and chloroform.
1-11
-------
• Combustion parameters - high excess air
- low excess air
- low heat release
• Mixing (or turbulence) - swirl
- air velocity
Those parameters found to be of less importance included burner velocity,
fuel type (No. 2 fuel oil) and concentration of hazardous waste compounds
(from 3 to 25 percent).
It was generally found that exhaust concentration measurements of carbon
monoxide and total hydrocarbons were good indicators of flame performance and
compound destruction efficiency. The exhaust CO level in particular appeared
to be well correlated with the exnaust concentration of the test compounds.
This result was expected since the high heat removal rates in the TFR empha-
size flame performance over postflame reaction. Since CO is an intermediate
in the oxidation of hydrocarbons to CO^ (Seeker et al., 1981a), itisdirectly
linked with combustion efficiency. Therefore, an examination of the relative
CO levels for each failure condition indicates the overall combustion effi-
ciency which can be compared to the destruction efficiency of the hazardous
waste compounds. The relationship between exhaust CO, total hydrocarbons
measured by the flame ionization detector, and destruction efficiency measured
for a mixture of compounds is shown in Figure 1-3. The maximum DRE (>99.995
percent) was measured at 30-40 percent excess air, which corresponded to the
minimum in both exhaust CO and hydrocarbon.
Figure 1-4 presents data obtained with the TFR at high heat-release
rates (44 kW). Very high destruction levels (>99.995 percent) were
measured for all compounds at 20 percent excess air at this heat-release
rate with the exception of benzene. It is possible that benzene was a pro-
duct of incomplete combustion of either the auxiliary fuel or one of the
test compounds (e.g., chlorobenzene). The actual source of the benzene
whether it is a product of incomplete combustion or an indication of incom-
plete benzene destruction, has not been determined. Benzene is a possible
intermediate in the formation of soot which was observed in the flame in the
form of luminosity especially at low excess air levels. Because of the
relatively large amounts of heptane present (97 percent) only a small con-
version of heptane to benzene is required to account for the exhaust levels
1-12
-------
0.03
2000
^ Hydrocarbons as Crl^ _
Q Test Compound (Average of four)
3 1500
0.02 ~
o
fcr 1000
-o
500
100
125
150
175
200
225
Percent Theoretical Air
Figure 1-3. Exhaust CO and total hydrocarbons and fraction
of test compound remaining in exhaust as a func-
tion of theoretical air (constant air velocity,
variable load, equal molar mixture of chloroform,
benzene, chloroDenzene, and acrylomtri le added
3 percent by weight to heptane).
1-13
-------
_ 4000
3000 -
« 2000
" 1000
(a) Exhaust CO
Concentration
125 150 175 200
Percent'Theoretical Air
e
CD
U
SJ
99.99
Detection Limit
99.999
100
125
150
175 200
225
Percent Theoretical Air
(b) Destruction and
Removal Efficiency
O Chloroform
A Acrylomtrile
£> Benzene
~ Chlorobenzene
Figure 1-4. Impact of theoretical air on CO and ORE from turbulent flame
reactor. Incinerabi1ity order at 150 percent T.A. is chloro-
form, acrylonitrile, benzene, and chlorobenzene (constant
load = 48 kW; variable air flow rate and burner velocity,
equal molar mixture of compounds added 3 percent by weight
to heptane).
1-14-
-------
of benzene measured at this low excess air condition. However, the benzene
could also be the result of a chlorobenzene reaction.
At higher excess air levels (>150 percent theoretical air) the exhaust
concentrations of CO and the test compounds increased. This is probably due
to lower flame temperatures and increased quenching which can occur when large
amounts of unheated air are present. The lowest DRE level obtained for these
high heat-release rates (44 kW) was 99.9 percent. The compound differences
were small but measureable at 150 percent theoretical air. The ranking from
highest to lowest concentration was: chloroform, acrylonitrile, benzene, and
chlorobenzene. This particular order, which was found to exist for a number
of failure conditions tested with the turbulent flame reactor does not agree
"* JLi,, j
with any of the-pr-oposed rankings, although the heat of combustion did
identify the most predominant compound (chloroform).
The data obtained at low heat-release rates (24-42 kW) are shown in Fig-
ure 1-5. This data set was achieved by lowering the fuel flow rate from the
nominal operating conditions while maintaining the air flow constant. This
drop in load and increase in theoretical air resulted in significant increase
in the fraction of the waste compounds in the exhaust. Under this failure
condition, chloroform and benzene had similar high exhaust concentrations
followed by 1,2-dichloroethane and similar low exhaust concentrations for
acrylonitrile and chlorobenzene.
The data presented in Figure 1-6 indicate that atomization parameters
had significant impact upon compound destruction. In these tests, a nozzle
designed for 1.5 gal/min was operated at .75 gal/min dropping the pressure
from 151 psig to 40 psig. This increases the mean droplet size, affects
fuel air mixing, and may cause some of the large droplets to escape the
flame. The highest compound exhaust concentrations were measured under.these
poor atomization conditions. However, the order of compounds was found to
be identical to other failure conditions for the TFR such as high excess air
at high loads, low excess air at low loads, and quench coils. The chloro-
form was found to be the most predominant compound followed by benzene,
acrylonitrile, and chlorobenzene. —
1-15
-------
Firing Rate (kW)
40 35 30 25
Chloroform
Benzene
Acrylonitrile
Chlorobenzene
Mixture containing
Benzene, Chlorobenzene,
Chloroform and
Acrylonitrile
Chloroform
Acrylonitrile
Chlorobenzene
1,2 Oichloroethane
Mixture containing
Oichloroethane,
Chlorobenzene, Chloro-
form and Acrylonitrile.
150 200
Percent Theoretical Air
Tmpact of theoretical air ana load on fraction of test
compounds remaining in exhaust of turbulent flame
reactor (constant air velocity, variable load 42-24- kW;
equal molar mixture of compounds added 3 percent by
weight to heptane).
1-16
-------
Off Design
.x n v'standard"
Nozzle
100 200 300
Percent Theoretical Air
(a) Carbon Monoxide
0.4 -
Chloroform
Acry-
onitrile
"Standard"
Nozzle
Chlorobenzgne
I »
100 200 300
Percent Theoretical Air
(b) Test Compound Oata with
Standard Nozzle and Off
Design Nozzle
Figure 1-6. Impact of atomization quality on CO and fraction of
test compounds remaining in exhaust of turbulent
flame reactor (constant air velocity, variable load:
42-16 kW; equal molar mixture of compounds added
3 percent by weight to heptane.
1-17
-------
A water-cooled copper coil was placed directly within the flame in the
TFR to provide an extrerng_c.as.e_of.._flame-quenchi.og in order to investigate
destruction efficiencies under this mode of failure. In this failure condi-
tion test, the coil acted to cool the flame and supplied a surface area for
reactants to quench. The presence of the quench surface increased both CO
and the test compound concentration (see Figure 1-7). The dashed line for
the uncooled data was derived from Figure 1-5. The order of the compounds
i was similar to other failure conditions with chloroform being the most pre-
\ dominant and chlorobenzene the least predominant compound in the exhaust.
\However, the positions of acrylonitrile and benzene were reversed from the
\order found in other failure modes.
1.5 Discussion
The combustion of hydrocarbon fuels in turbulent diffusion flames results
in relatively high flame-zone temperatures (between 1600-2000 K) and residence
times are on the order of 0.1 seconds. If the waste compounds investigated
in this study experience these conditions, then they would be quantitatively
destroyed. The results of this study agree with this hypothesis. Turbulent
diffusion spray flame and a laminar reactor burning single droplets were
capable of destruction efficiencies greater than 99.995 percent. In the case
of the turbulent flame reactor under optimized conditions (stable flame, low
CO and total hydrocarbon) the compounds were destroyed mainly in the flame
because post-flame decomposition was minimized due to the fact that the flame
was contained by cold walls. Consequently, it can be concluded that a flame
_^^is an extremely efficient mode of destroying waste compounds and the concept
/ of incinerability under these conditions has little value. If everything is
destroyed it is not possible to rank compounds in terms of difficulty or ease
of destruction. Consequently, a series of experiments were designed to
assess incinerability under several limiting conditions which might typify the
failure modes of practical liquid injection incinerators.
The microspray reactor investigated those conditions associated with
single droplet combustion in the absence of complications due to turbulent
1-18
-------
c
aj
CT
>1
x
o
o
"O
OJ
u
03
S_
i~
o
u
CL
a.
a
o
led Da/
(a) Carbon Monoxide
TOO 150 200
Percent Theoretical Air
in
3
-------
mixing. It was selected in order to study thermal effects separated from
turbulent mixing and atomization. The temperature required to ignite drop-
lets of hazardous waste under oxygen-rich conditions in the laminar premixed
flat-flame reactor was found to be low (850°K) in comparison to typical flame
temperatures (1500-2000°K). Above the ignition temperature, the droplets were
visually observed to ignite and the compounds tested were quantitatively
(> 99.995 percent) destroyed. Even in the absence of oxygen, the microspray
data were consistent with the high destruction efficiencies achievable in a
turbulent diffusion spray flame environment.
The TFR was operated at high heat removal rates by operating with water-
cooled walls in order to minimize post-flame reactions and mixtures up to
25 percent by weight of the test compounds were investigated. Even in the
absence of significant post-flame decomposition, destruction efficiencies
which corresDonded to the detection limits of the analytical systems (99.995
percent) were achieved for all the compounds tested. In the turbulent flame
reactor, a direct relationship was observed between overall combustion effi-
ciency as indicated by exhaust CO and hydrocarbon emissions and the destruc-
tion of the test compounds. Conditions which minimized the CO concentration
in the exhaust gases also maximized destruction efficiency. Under all failure
conditions investigated, exhaust CO concentration increased when the test
compound concentration increased. These results suggest the feasibility of
using exhaust CO and potentially total hydrocarbons to monitor the performance
of liquid injection incinerators once the conditions giving the maximum
destruction efficiency have been defined.
The incinerability or ordering of the compounds was found to depend on
the actual failure condition which caused the inefficiency. When both the
microspray and the turbulent flame reactor were operated under conditions
which simulated failure modes of practical incinerators, measureable differ-
ences in the destruction efficiency of the five test compounds were obtained.
[For example, chlorobenzene was the most difficult to eliminate in the micro-
spray when the temperature was too low to ignite the droplets, but was the
7-20
-------
least difficult to eliminate for a variety of failure conditions in the TFR
'such as poor atomization quality.
Figure 1-8 presents a series of bar graphs which allow a comparison be-
tween incinerability as defined by the various failure modes and the rankings
indicated by procedures based upon Tgg gg, heat of combustion, the NBS method
and AIT. The bar graph shows the concentrations measured in the experiment
normalized so that the most predominant compound shows full-scale and the
lesser concentrations are expressed as a percentage of that maximum concen-
tration. This approach gives an indication of the measured magnitude of the
difference in destruction efficiency between compounds. A comparison of these
relative concentration measurements with proposed incinerability ranking tech-
niques demonstrates that none of the proposed techniques agree with the data
for all failure conditions. However, some of the ranking procedures were
found to be appropriate for specific failure conditions. For example, the
non-flame thermal destruction (Tgg gg) and AIT procedures both agreed with the
compound concentration measurements when the temperature was below droplet ^
ignition temperature and under oxygen-deficient conditions. Heat of combustion
was found to correlate the pure compound data when the microspray was operated
below droplet ignition temperature. In most instance, chloroform was the most
difficult compound to incinerate for the failure conditions investigated with
the TFR, and this was anticipated by only one of the four ranking techniques -
heat of combustion.
Although measureable differences in the destruction efficiency of the
five test compounds were obtained, the differences were not large under any of
the conditions tested. For the most part, the variation in the concentration
(between highest and lowest) of the compounds in the exhaust was typically of
the order of five, although variations larger than ten were measured under some
circumstances. This suggests that the selection of POHC may not be very criti-
cal because the differences between compounds are small. If the permit writer
selects three compounds based upon two or more ranking techniques, and it is
demonstrated that their DRE is greater than 99.99 percent, then it is very
unlikely that any other compounds will be destroyed to a significantly lesser
degree.
1-21
-------
NON-FLAME
TEMPERATURE
INCINERABILITY RANKINGS) T99.99
HEAT OF
C0M8USTI0N
M8S FUME , AUTO IGNITION
RANKING TEMPERATURE
CHLOROFORM ^
1,2 OICHLOROETHANE /
3ENZENE
ACRYLQNITRILE
CHLOROBENZENE 1
PURE COMPOUNO
IGNITION
TEMPERATURE
OXIDATIVE
LOW TEMP
OXIDATIVE JUST
BELOW IGNITION
NO OXYGEN
LOW TEMP
MICROSPRAY
CHLOROFORM /
1,2 OXCHLORQETHAME 3
3EHZENE ^
ACRYLONITILE £
I
CHLOROBENZSIE
HIGH
EXCESS AIR
LOW • , POOR
EXCESS AIR ATOMIZATION
TURBUL£?IT FUME
QUENCH COIL
CHLOROFORM
1,2-0ICXLORO ETHANE
3ENEZENE
ACRYLONITRILE
CHL0R08ENZE2
-------
This study has identified the differences between compound destruction
efficiency caused by failure conditions associated with the flame zone. High
destruction efficiencies have been demonstrated in the flame alone. However,
many incinerators are equipped with post-flame hold-up zones and after-burners
in order to achieve additional thermal decomposition of compounds which escape
the flame zone. In order for an incinerator to fail to destroy a compound,
the material must both escape the flame and the temperature be too low in the
post-flame hold-up zone to destroy the compound (less than Tgg gg). The dif-
ferences in the concentration of compounds in the exhaust of the incinerators
is associated with both the flame and non-flame zones. The thermal decomposi-
tion which occurs in the post-flame zone can alter the ranking in the exhaust.
As an example, consider a flame zone in which the DE of chloroform and chloro-
benzene was 95 percent and 99 percent, respectively (a flame ranking consistant
with the data of Figures 1-6 or 1-7). Utilizing non-flame kinetics and a 1.0
sec isothermal post-flame zone (kinetics shown in Table 2-1), for post-flame
temperatures below about 870 K, the flame zone ranking will persist in the
exhaust. Above 1008 K both compounds are destroyed to 99.99 percent DE.
Hence, there are potential situations, dependent on incinerator conditions,
for either a flame zone or a post-flame ranking to prevail within a given jnit.
It was not the purpose of this study to ascertain why destruction effi-
ciency under flame conditions can be compound and failure mode specific. More
detailed measurements such as fundamental kinetic flame studies, are necessary
to provide a full explanation of the causes of the rankings. It could be
associated with flame inhibition due to the presence of halogens which are-
known to reduce burning rates. Under quenching conditions, these effects
could be enhanced. The formation of products of incomplete combustion (PICs)
as a consequence of the partial destruction of the waste compound, was not
investigated. An alternate method of assessing incinerabi1ity could be based 1
upon the potential to form PICs which are themselves hazardous. \
1.6 Conclusions
1. Under optimum conditions, flames are capable of destroying
hazardous waste compounds with very high efficiencies (greater
than 99.995 percent) without the need for long residence time
high-temperature post-flame zones or after-burners.
1-23
-------
2. Reduced flame destruction efficiencies are the result of
operation under some failure mode such as poor atomization,
poor mixing, or flame quenching.
3. Incinerability, or ordering of compounds in terms of their
relative destruction efficiency, is dependent on the actual
failure condition which caused the inefficiency.
4. Optimum conditions for destruction of hazardous waste
compounds in turbulent diffusion spray flames correspond
to minimal exhaust CO and total hydrocarbons.
5. No one incinerability ranking system appears to predict
correctly the relative destruction efficiency of the five
compounds tested for all failure conditions investigated.
However, several rankings did correctly predict relative
DE for specific failure conditions.
6. More data is required on other compounds and other failure
conditions more appropriate to different types of hazardous
waste incinerators to fully determine the limitations of
incinerability ranking systems and develop an appropriate
incinerability ranking methodology.
7. Future experimental effort should be directed toward
extending the compound data base beyond the current five,
and in particular, the extention of experimental capabilities
to consider additional failure modes (e.g., those associated
with post-flame thermal processes of after-burners).
1-24
-------
1.7 Executive Summary References
Cudahy, 1981. Incinerability, Thermal Oxidation Characteristics and Thermal
Oxidation Stability of RCRA Listed Hazardous Wastes. IT Enviroscience Corp.
Dellinger, B., D. S. Duvall, D. L. Hall, and W. A. Rubey, 1982. Laboratory
Determinations of High Temperature Decomposition Behavior of Industrial
Organic Materials. 75th Annual Meeting of the APCA. Paper No. 82-3.5.
New Orleans, LA.
Dietrich, V. E., 1979. Dropsize Distribution for Various Types of Nozzles.
In Proceedings of the 1st International Conference on Liquid Atomization
and Spray Systems. The Fuel Society of Japan, Tokyo, Japan, p. 69.
Duvall, D. S. and W. A. Rubey, 1976. Laboratory Evaluation of High-Temperature
Destruction of Kepone and Related Pesticides. Technical Report UDRI-TR
-76-wl, University of Dayton Research Institute. EPA 600/2-76
-299.
Duvall, D. S. and W. A. Rubey, 1977. Laboratory Evaluation of High-Temperature
Destruction of Poly-chlorinated Biphenyls and Related Compounds. EPA
600/2-77-228.
Lee, K. C., J. L. Hansen, and D. C. Macauley, 1979. Predictive Model of the
Time/Temperature Requirements for Thermal Destruction of Dilute Organic
VaDors. 72nd Annual Meeting of the APCA. Cincinnati, OH,
Lee, K. C., N. Morgan, J. L. Hansen, and G. M. Whipple, 1982. Revised Model
for the Prediction of the Time-Temperature Requirements for Thermal
Destruction of Dilute Organic Vapors and its Usage for Predicting Com-
pound Destructability. 75th Annual Meeting of the APCA, New Orleans,
June 1982.
Seeker, W. R., M. P. Heap, and T. J. Tyson, 1981a. Gas Phase Chemistry.
Volume I of Final Report for EPA 68-02-2631.
Tsang, W. and W. Shaub, 1981. Chemical Processes in the Incineration of
Hazardous Waste. National Bureau of Standards. Paper presented to
American Chemical Society Symposium on Detoxification of Hazardous
Wastes, New York, NY.
1-25
-------
2.0 INTRODUCTION
According to the Resource Conservation and Recovery Act regulations,
standards for safe operation of hazardous waste incinerators are defined as
achieving at least 99.99 percent destruction and removal efficiency (DRE) of
the hazardous constituents in the waste feed. The interim final rule (Federal
Register: 1/23/81) stipulates that the standard of operation must be demonstra-
ted in a trial burn for only a limited set of compounds in the waste stream or
by submission of acceptable data. The permit writer selects one or more com-
pounds, designated as the principal organic hazardous constituents (POHC),
based upon the component concentration in the waste stream and the relative
ease of incineration. Demonstration of 99.99 percent DRE for the POHC defines
safe operating conditions. This approach assumes that destruction of all the
remaining more easily incinerated compounds is greater than 99.99 percent DRE
at this specified operating condition. The permit operating requirements are
then specified based on these trial burn operating conditions.
The procedure assumes the existance of some measure of ranking the rela-
tive ease of destruction of individual compounds (i.e., degree of incinerabil-
ity). At present, a single ranking list is employed and, hence, the ranking
must be system-dependent; i.e., it must be applicable to all types of incinera-
tors under all operating conditions. Since the proof of compliance is based
solely upon trial burn measurements of the POHC chosen from the ranking, it is
critical that the ranking correctly reflect the relative ease of incinerability
in the system and operating conditions under consideration. If it does not,
the incinerator may be operated under conditions which release hazardous
compounds even though it is quantitatively destroying the selected POHC.
An appropriate incinerability ranking could be obtained from extensive
trial burn data. Such an approach is advantageous because the resulting
ranking would be based on actual incinerator performance. It would also
indicate if ranking was a function of incinerator type or operating condition.
However, such an approach cannot be adopted because of the expense of measur-
ing each waste during a trial burn. In addition, the measurement techniques
required for obtaining complete characterization of incinerator performance
are, at least partially, beyond the state-of-the-art.
2-1
-------
The quantities of compounds escaping from the incinerator will likely
depend on the actual failure condition which allowed the material to escape.
The development of an incinerability ranking without extensive trial burn
data on all compounds requires consideration of the possible mechanisms of
escape. Liquid injection incinerators typically produce the heat necessary
to destroy the compounds in large-scale turbulent-diffusion flames burning
either an auxiliary fuel or the hazardous waste itself. This is followed by
large, well insulated hold-up zones in which wastes that escape the flame are
removed by thermal decomposition, after-burners, and gas cleaning equipment
such as scrubbers. In such an incinerator, the parameters which control waste
destruction efficiency are:
1. Atomization Parameters. When waste material is injected as a
liquid which must be atomized, poor destruction efficiency can
result from inappropriate atomization. Droplets which are too
large may be produced or their trajectory may be such that they
pass through the flame zone without completely evaporating.
2. Mixing Parameters. In a turbulent diffusion flame the reactants
are supplied in separate streams and reactant contacting takes
place via turbulent mixing. Poor mixing can result in low
destruction efficiencies because the waste material may not be
mixed with oxygen before it escapes from the flame region.
3. Thermal Parameters. The destruction efficiency may be low because
flame or post-flame temperatures are low. This can occur if the
calorific value of the waste/auxiliary fuel mixture is low, heat
removal rates are high, or quench rates are high due to mixing
with excessive amounts of air.
4. Quenching Parameters. The reactants can be quenched before
destruction is complete by heterogeneous or homogeneous pheno-
mena. Mixing can cause quenching as explained above, or the
flame may impinge upon a relatively cool surface.
Consequently, it is essential to investigate the concept of incinerability in
flames under conditions which could account for a failure to completely destroy
the waste compound and are typical of real systems.
2-2
-------
A number of incinerability ranking procedures and tests have been proposed
(Cudahy, et al., 1981). These procedures include:
• Heat of combustion per unit mass compound.
• Autoignition temperature (AIT).
• Temperature required to yield a 99.99 percent destruction
efficiency in a specified residence time (normally 1-2 sec)
derived from dilute first order thermal decomposition
kinetics (Tgg>gg).
• Correlation of compound specific properties such as AIT,
structure, etc., with Tgg gg. This approach establishes
an estimated Tgg gg ranking and thus avoids the necessity
of obtaining thermal decomposition data for each compound.
• Susceptibility of bonds to flame radical attack (National
Bureau of Standards: NBS).
• Miscellaneous approaches: molar heat of combustion, heat
of formation per unit weight, Gibbs free energy per unit
weight, ionization potential, flash point, activation energy
derived from thermal decomposition kinetics, and heat of
formation of ion per unit weight.
Each of these ranking procedures approaches the problem of determining
incinerability in a different manner. The advantages and disadvantages, and
the relationship of each to the incineration process will be discussed in Sec.5.
The primary goal of this study was to establish if any of the proposed
incinerability rankings were appropriate to account for compound effects that
might occur for possible failure conditions for the most common (Keitz, et al.,
1983) type of incinerator, liquid injection. The approach utilized in the
study was to measure the relative exhaust compound concentration when dif-
ferent modes of failure were allowed to occur and to compare this ordering
to proposed incinerability ranking procedures. Secondary goals included
generating fundamental flame-mode destruction data necessary to compare flame
and non-flame decomposition and to act as a guide for types of experimentation
needed for establishing an incinerability ranking that could account for
different modes of failure.
2-3
-------
Two flame reactors were employed which allowed the relative destruction
of compounds to be observed under two limiting mixing conditions. A micro-
spray reactor consisting of a laminar premixed flat frame was used to
investigate the effect of thermal and stoichiometric parameters on de-
struction efficiencies of individual droplets where large-scale turbulent
mixing was not important. The second flame reactor utilized was the turbu-
lent flame reactor which is similar to a real system in that it consisted
of a turbulent diffusion spray flame burning a mixture of hazardous waste
and an auxiliary fuel. It was utilized to investigate atomization, mixing,
and quenching parameters.
In the subsequent section the details of the experimental approach uti-
lized in this study are discussed. In section 4 the results from both reac-
tors are presented and discussed in terms of the behavior of different test
compounds under a variety of incineration failure conditions; the flame data
are compared with the proposed incinerability rankings and the conclusions
of the study and recommendations for further effort are discussed in section
5.
Five compounds representative of liquid organic hazardous waste listed
in the RCRA regulations (Appendix 8; 19 May 1980 Federal Register) were
selected for these flame reactor studies. This limited number required that
compounds be selected carefully to provide the maximum amount of comparison
information.
The following criteria were utilized in the selection of test compounds:
1. Selected Compounds Show Variations in Incinerability Rankings. To
provide as thorough a test of each of the incinerability ranking
procedures as possible, the test compounds were selected ta repre-
sent a wide variation within each ranking procedure. Since only
five compounds were selected, it was necessary to give prece-
dence to choosing compounds with wide variations in the heat of
combustion and thermal decomposition rankings.
Z. Nonflame Decomposition Data Exist. A direct comparison between
thermal decompositon rankings and the flame-mode destruction rank-
2-4
-------
ings requires that nonflame thermal decomposition data are avail-
able for all selected compounds.
3. Various Molecular Structures. The structure criteria consisted of
two parts: First, the test compounds were selected to be represen-
tative of the various organic structures classified by the EPA as
hazardous waste (Appendix 8 of the 19 May 1980 Federal Register).
Second, an effort was made to select compounds with similar non-
flame thermal decomposition data but whose structures were entirely
different.
4. Liquid Compounds. Liquids were chosen for this first set of com-
pounds because they represent the largest class of waste generated.
Since only five test compounds were considered, it was also neces-
sary to consider only liquids so that the results for the different
compounds could be compared directly.
These selection criteria were derived directly from the program objectives.
Of first importance was obtaining flame-mode destruction data for comparison
with the several proposed ranking procedures; hence, the first criteria
required that the proposed ranking procedures be well represented by the test
compounds. A second objective required the comparison of flame and nonflame
results. This objective dictated that nonflame thermal decomposition data
be available. Also, compounds with entirely different molecular structure
but similar thermal decomposition data were selected to determine if flame-
mode rankings are structure-dependent. Testing of solid, liquid, and gaseous
compounds in an experiment where only five compounds were used would have
made the comparisons required in the objectives impossible. The rankings
resulting from the tests could not be separated into physical and chemical
causes with such a narrow data base; thus only liquid compounds were
selected.
The list of selected compounds is shown in Table 2-1. All compounds are
liquids at room temperature. The first order thermal decomposition para-
meters (pre-exponential term and activation temperature) are listed, alon^
with the value for Tgg gg for 1.0 sec. derived from the kinetic parameters.
The time required for a droplet of initial diameter of 100 JJm to completely
2-5
-------
TABLE 2-1. TEST COMPOUNDS AND PROPERTIES
Compound
and
Structure
Boiling
Point
A
(sec-^)
Activa-
tion Tem-
perature
Tact (K)
T99.99
(K)
AH _
vap
(cal/gm)
Evap.
Rate
(msec)
NBS
Rank-
ing
AHC
(kcal/gm)
AIT
(K)
(K)
(1)
(1)
(2)
(3)
(4)
(5)
(6)
(7)
Benzene
352
7.4x 1021
48,300
1,007
103.57
14.4
4
10.03
836
Chlorobenzene
405
1.3x 1017
38,200
1,038
77.59
10.8
3
6.6
911
Chloroform CHCI3
355
2.9 x 1012
24,500
925
64.75
9.0
18
0.75
-
1,2-Dichloro-
ethane CH2CI-CH2CI
356
4.8 x 1011
23,000
931
85.3
11.9
28
3.0
686
Acryloni-
trile CH2=CH-CeN
351
2.1 x1012
26,300
1,003
149.84
20.8
-
7.93
754
Highest (8)
NA
NA
48,300
(Benzene)
1,112
(Methane)
241
(Methanol)
33.5
(Methanol)
NA
10.14
(Toluene)
988
(Phenol)
Lowest (8)
NA
NA
11,300
(DDT)
754
(DDT)
46.42
(CCI4
6.4
(CCI4)
NA
0.13
(CHBt3)
363
(CS2)
Rank 1 is most difficult to inciner-
ate. [Adapted from Tsang and Shaub
(1981).]
Heat of combustion.
Autoignition temperature.
Highest and lowest values observed in
each category.
NA - Not appropriate
(1) C/C0 = exp(-kt); k = A exp(-Tact/T). (Values (5)
are from Lee et al., 1979 and 1982 except
chloroform (Dellinger, 1982).
(2) Temperature for 99.99 percent DE at t = 1 sec. (6)
[Values from same sources as (1)].
(3) Latent heat of vaporization. ^
(4) Time for 100 pm droplet to evaporate at 1000 K.
-------
evaporate is also shown. A variety of organic structures are represented.
Benzene and chlorobenzene are aromatic, chloroform is a chlorinated methane,
1,2-dichloroethane is a chlorinated ethane, and acrylonitrile contains
nitrogen. Although AIT data are not available for chloroform, it was in-
cluded because of its very low heat of combustion. Acrylonitrile was
selected because it has almost the same Tgg gQ as benzene, but has a
different structure. For reference, the highest and lowest values for
compounds that have been determined in each ranking system are included to
indicate the range of values covered by the selected test compounds. The
selected compounds represent greater than 90 percent of the range covered
by listed compounds for heat of combustion (.13 to 10.14 kcal/gm), and
encompassed roughly 70 percent of the compounds tested to data for non-
flame decomposition temperature, autoignition temperature, and the NBS
incinerability ranking.
2-7
-------
3.0 EXPERIMENTAL EQUIPMENT
In this section the experimental systems are introduced. Discussion
topics include the reasons for selecting two distinct reactor designs, the
physical description of the two reactors and their associated equipment, the
sampling and analytical systems, and the laboratory facility.
The discussion in the previous section indicated that several potential
escape mechanisms exist for the flame zone of liquid injection incinerators.
This research program addressed the incinerability rankings that resulted
from the laboratory simulation of these failure modes. In order to investi-
gate a variety of realistic failure modes for liquid injection incinerators,
considerable experimental flexibility was necessary. This flexibility was
obtained through two reactor designs: the microspray reactor and the turbu-
lent flame reactor (TFR).
The microspray experiment was designed to study the rankings that result
from individual droplet processes without large-scale turbulent mixing.
These processes include:
• Droplet physical processes such as evaporation, liquid phase reac-
tions, etc.
• Flame zone chemistry in the immediate droplet vicinity.
e Nonflame decomposition chemistry away from the immediate droplet
vicinity.
The experiment was designed to bridge the gap between the non-flame thermal
decomposition experiments and the turbulent flame data. As such, it in-
cluded two processes in addition to those present in the thermal
decomposition experiments : droplet vaporization dynamics and flame
reactions. The data were used for the following purposes:
• To determine what portion of the turbulent flame rankings were
due to droplet flame and evaporation processes.
• To compare flame (microspray) vs. non-flame (thermal decomposition)
destruction on a fundamental level without the complicating in-
fluence of turbulence.
3-1
-------
The TFR included all of the processes that occur in the microspray, and the
following:
• Droplet-droplet interactions in a high number density droplet field.
• Turbulent mixing.
• Thermal quench.
3.1 Microspray Reactor
The microspray reactor was employed in this program in order to in-
vestigate single droplet reactions without limitations associated with
turbulent mixing. Data from this reactor could be easily interpreted in terms
of how thermal parameters influenced compound incinerability in the absence
of atomization, mixing, or quenching limitations. The microspray reactor
(Figure 3-1) is a modification of flame reactors used previously in the
study of the thermal decomposition of pulverized coal particles and r>£jvy
fuel oil droplets (Seeker et al., 1981b; Kramlich et al., 1%1). In these
flame reactors, particles or droplets of the material to be studied (in this
particular case, droplets of hazardous waste) are injected through a laminar,
premixed, hydrocarbon flat flame and thermal decomposition of the material
takes place in a flame environment. The reactor was chosen because ft can
provide a gas-phase environment similar to that experienced by droplets in
the near field of spray flames. Badzioch (1967] first specified the criteria
necessary to study the decomposition of solid fuels appropriate for a flame
environment. For liquid fuels, those criteria can be specified as follows:
• The gas-phase temperature must be in the range of normal flame
temperatures (up to 2000 K) and be well-defined and controllable.
• The fuel must be in the form of droplets in the size range
(20-200 um) representative of commercial fuel injectors.
• The droplets should be well-dispersed in a gas-phase environment
which is similar to the recirculated products in the near-field of
the flame in order to simulate the contacting between liquid and
gas.
Previous work with this reactor has shown that it meets these criteria and
the data generated on the decomposition of pulverized coal compare favorably
3-2
-------
t
Sample Port
Mixing Baffles
Flat Flame Reactants
Valve
A
Rotameter
Dispersion
Gas ——"
Vi,'.'.
•'-3>
ja'
m
Gauge
-Droplets
Flat Flame ^
-Drop!et
. Generator
£
Pressure
^ Regulator
N„
Test Compound
Storage
*
Liquid
Purge
Filter
Figure 3-1. Microspray reactor.
-------
with in-flame data from larger scale turbulent diffusion flames (Seeker and
Heap, 1982).
In the reactor, monodisperse droplets are injected into hot postflame
gases. The droplets are sufficiently dispersed to prevent droplet-droplet
interaction. Thus, the reactor behavior is characteristic of an ensemble of
individual droplets. This droplet dispersion is accomplished by ballistical-
ly injecting droplets through an opening in the center of a flat-flame
burner. The burner is shown in Figure 3-2 and consists of an 8.9-cm-square
piece of ceramic honeycomb (Corning Celcor Cordierite, 39 cells/cm }. The
premixed main burner flow enters through the rear of the burner and is
distributed to the honeycomb through the porous stainless steel plate. These
premixed gases form a flat flame that is stabilized within 2 mm of the face
of the honeycomb by heat loss to the burner.
The droplet stream enters the reactor through a 1.3-cm-diameter opening
in the burner center. Monodisperse droplets are generated by the Berglund-
liu vibrating orifice technique (TSI Model 3050). In this technique, liquid
is forced through a small orifice to form a liquid jet; in this study the
orifice diameter was 20 um. A schematic of the droplet generator is shown
as Figure 3-3. The orifice plate is vibrated by a piezoelectric ceramic to
induce an extremely regular breakup of the jet into monodisperse droplets
(Serglund and Liu, 1973). In the present study, the droplet diameter "as
38 um, although diameters between 20 and 200 ym are possible. This diameter
and the droplet spacing were verified by high magnification spark shadow
photography.
The droplets emerge from the generator in a single column and agglomer-
ate unless dispersed. Dispersion is accomplished by passing the droplets
through a second orifice with a small amount of dispersion gas. The oxygen
content of the dispersion gas was adjusted such that the stoichiometry,
including the droplets, was constant throughout the burner. The flat flame
is located at the inlet end of a 10-an-square by 100-cm-long chimney
(Figure 3-1) within which the droplet reactions occur.
Two chimneys are available. One is fitted with Vycor windows to permit
visual observation of the droplet flame. Because of the potential for
3-4
-------
/
K \ \ V " \
Purge
Drain
Dispersion Flow
Liquid
Monodisperse
Droplet Flame
10-cm-Square
Stainless Steel
Chimney
Hydrocarbon
Flat Flame
Ceramic Honeycomb
Porous Stainless
Steel
Dispersion
Orifice
771
Main Burner
S IS x S SN
Orifice in
Vibrating Crystal
Figure 3-2. Details of microspray reactor.
3-5
-------
mZZZZZT/,
J
Dispersion flow-
Orain Tube
Drain PI_
Dispersion
Cap
Liquid Chamber
Teflon O-Ring
Liquid Orifice
AC Signal
Liquid
Chamber ^
Base
Piezoelectric
Ceramic
Figure 3-3. Schematic diagram of the droplet generator.
-------
window seal leakage, a second chimney without windows is used for concentra-
tion measurements. The exhaust gases pass through a tube containing baffles
to mix the gas uniformly prior to sampling.
The main burner gas for the microspray consists of an Hg/Ng/air mixture.
The air is obtained from a compressed air supply, two-stagi pressure
regulated, and double filtered prior to the flow control rotameter. The
hydrogen and nitrogen are obtained from vendor-supplied compressed gas
bottles. The flow control panel and its calibration are the subject of
Appendix A.
In this study, the microspray reactor was used to investigate the
impact of two controllable parameters on the relative thermal destruction
of hazardous waste compounds: temperature and stoichiometry. In
addition, the characteristics of pure waste compounds were compared to
those of mixtures of waste compounds. One potential failure mode dealing
with the flame condition is too low a flame temperature to destroy the
waste constituents. The flat-flame temperature was controlled by varying
the amount of diluent nitrogen in the flame reactants. This flat flame
temperature was the "ambient" temperature within which the droplet flames
occurred. After ignition, however, the droplet flame temperature was
expected to exceed the local flat-flame temperature. Typical axial pro-
files of the gas-phase temperatures as measured with a type-R (Pt/Pt--
13 percent Rh) thermocouple on the center line of the chimney are presented
in Figure 3-4a. Radial temperature profiles (Figure 3-4b) indicated that
the gas-phase temperature across the reactor was relatively uniform except
near the edges of the reactor. These measurements were made with droplet
dispersion flow but without droplets. Since the droplets were dispersed
only around the center of the reactor with a maximum diameter of 3 cm, the
individual droplets all were subjected to an identical temperature history
which altogether avoided the influence of the walls. The temperature history
was controlled by the flame temperature which was implied by extrapolating
the axial temperature measurements to the burner face. The destruction
data was determined as a function of this extrapolated measured flame
temperature.
Compound effects were also investigated for two extremes in stoichiom-
3-7
-------
1600
400
.1 .2 .3 .4 .5
Time (sec)
_J I I I
0 10 20 30 40
Axial Distance (cm)
50
(a) Gas phase temperatures on center
line of microspray reactor for
different flame conditions.
1500
1000
1500
1000
:1500
Ol
u
3
!1000
Ol
CL
jpl500
1000
1500
1000
1.27 cm from flame holder
i t I I I I I I I > I I J I I
6.35 cm from flame holder
I I I I I I I I I I I I I I I
11.43 cm from flame holder
I I I I I I «—I I I I I I I I
/y^VWYWVY^HW^
- Ci 16.5 cm from flame holder ~
H I | | I 1 | | | I I I I I I
44.4 cm from flame holder
J I I I I I I I I I I I I I L
4 3 2 1 CL 1 2 3
Radial Distance (cm)
(b) Gas phase temperature as a function
of axial and radial distance for one
flame condition.
Figure 3-4. Gas temperature measurements in microspray reactor.
-------
etries, rich and lean. For rich conditions, the flat flame was operated with
no excess oxygen (SR = .83). Under these conditions the droplets evaporated
and thermally decomposed without sufficient oxygen available to support drop-
let combustion. For lean conditions the burner stoichiometry and nitrogen
diluent flow were adjusted to maintain 10 mole percent oxygen in the post-
flame gas. In this case, oxygen was available throughout the reactor, allow-
ing the droplets to ignite and burn.
The final parameter investigated in the microspray reactor was mixtures
of compounds. The majority of the data were taken on equimolar mixtures of
four compounds. The use of mixtures assured that the compounds were initially
exposed to the same time/temperature and stoichiometry, and hence any measured
differences could be attributed to the compound itself. A limited set of
data were obtained on pure compounds for comparison.
3.2 Turbulent Flame Reactor
The turbulent flame reactor (TFR) was designed especially for the present
program to provide a turbulent liquid spray flame, including swirl, recircula-
tion, broad drop-size distribution, and high variation in droplet number den-
sity. It was particularly important that the reactor be capable of simulat-
ing the compound escape mechanisms that can occur for flame zones of liquid
injection incinerators. Very high heat removal rates were utilized to quench
postflame reactions. Thus, the destruction which occurred in the turbulent
diffusion flame was emphasized over nonflame decomposition which can occur in
the postflame region. The reactor design is based on a configuration for
which aerodynamic field data are available (Baker et al., 1975).
The reactor consists of a swirling air/liquid spray burner firing into
a 30.5-cm-diameter by 91.5-cm-long water-cooled cylindrical enclosure shown
in Figure 3-5. The water-cooled cylinder is made of 304 stainless steel and
is formed into three interchangeable segments which are joined by flanges and
gasketing. The lowest segment has four sight glass ports, one of which is
used for flame ignition. The reactor top plate contains an exhaust fitting
which includes the sampling ports, and a Vycor plate/mirror arrangement for
obtaining an axial view of the flame.
The burner consists of a pressure-atomized nozzle (Delavan WDA series)
3-9
-------
FLOW CONTROL
BAFFLE
AIR BLOWER
VENTURI
GAUGE
Figure 3-5
j E~H,x,Mfi
CHAMBER
WATER INLET
WATER OUTLET
- STAINLESS STEEL
WATER-COOLED CHAMBER
TURBULENT SPRAY FLAME
- QUARTZ WINDOW
QUARL
SWIRL VANE
SPRAY NOZZLE
(
PRESSURIZED
n2
RESERVOIR
r
Turbulent flame reactor.
-------
located level with the bottom plate of the reactor as shown in Figure 3-6.
The main burner air is introduced through the annular space around the nozzle.
Variable area flow constrictors were fixed into this space to vary burner
air velocity independently of air flow rate. Interchangeable swirl vanes
are placed into the gap between the constrictor and the nozzle shaft. The
vane angle was set to provide a swirl number of unity for the present study.
(Swirl number is defined as the ratio of angular momentum to axial momentum
for the swirling air flow.) To provide a smooth entry of air into the burn-
er and to prevent corner recirculation, a castable refractory quarl is
placed in the lower water-cooled segment. As shown in the figure, this has
the form of a 45-degree cone.
The liquid fuel feed system consists of a pressurized storage tank. The
fuel-flow vs. tank-pressure calibration is discussed in Appendix A. The tank
is pressurized with bottled nitrogen and the pressure is regulated with a
two-stage bottle regulator. The burner air flow is supplied by a high-
pressure-head rotary blower and is metered with a venturi flow meter. The
calibration is discussed in Appendix A.
The normal operating conditions for the TFR were selected by considering
commercial practice, visual flame stability, low exhaust CO (*75 ppm) and
total hydrocarbon measurements (<20 ppm), and high destruction and removal
efficiency of hazardous waste compounds. The conditions chosen for this nor-
mal operating mode are specified in Table 3-1. Under this condition, ex-
tremely high DRE levels of the test compounds were achieved (> 99.995 percent).
The actual test conditions investigated for the TFR were selected to allow
specified failure conditions to exist. Several variations from the normal
operating conditions were investigated which had little influence on the high
flame destruction efficiency of the hazardous waste constituents. These
included auxiliary fuel type (No. 2 fuel oil), burner throat velocity and
test compound concentration (up to 25 percent). Those variations which did
result in significant deterioration in flame performance and were thus used
to investigate the relative compound destruction under failure conditions
included:
3-11
-------
Castable Refractory
Quarl
Interchangeable
Swirl Vane /
Delavan Assembly A*
-t /* *
k Nozzle , A, ,..
Interchangeable
X AirV Flow
* X Con\strictoo
Burner Air
Flow
Figure 3-6. Turbulent flow reactor burner detail.
3-12
-------
TABLE 3-1. NORMAL OPERATING CONDITIONS FOR TURBULENT FLAME REACTOR
NOZZLE
Delavan Pressure Jet, Hollow
Cone, 60°Spray Angle Model
WDA-60 -1.5
NOZZLE PRESSURE
103 psig-
AUXILIARY FUEL
Heptane
FUEL FLOW RATE
1.4 gm/sec (11 Ib/hr)
AIR FLOW RATE
14.3 1/sec (1830 ft3/hr)
BURNER THROAT VELOCITY
7.1 m/sec (23.3 ft/sec)
BURNER SWIRL NUMBER
1.0
EXCESS AIR
30%
BURNER HEAT RELEASE RATE
38 kW (131,000 Btu/hr)
TEST COMPOUND CONCEN-
TRATION
2% by Mass
EXHAUST CONCENTRATIONS
CO1 75 ppm
Total Hydrocarbons: 20 ppm
DESTRUCTION EFFICIENCY
OF TEST COMPOUNDS
>99.995%
3-13
-------
• Low excess air
• High excess air
t High excess air and low load
• Poor atomization quality
• Quench coils within flame
The exhaust concentration of test compounds were measured under these failure
conditions for mixture of compounds in the auxiliary fuel along with continu-
ous measurements of CO and hydrocarbon to indicate overall flame performance.
3.3 Laboratory Facility
The laboratory facility is designed to:
• Facilitate ease of operation
• Prevent laboratory air contamination during routine operation.
• Physically separate working personnel and experiments from inci-
dental workers and bystanders.
• Minimize the hazards associated with equipment malfunctions and
accidents such as spills, broken lines, fires, and utility outages.
To meet these design objectives, the laboratory is divided into a reactor room
and a control room. The reactor room contains all of the facilities for handl-
ing the test compounds, including storage, preparation, and waste handling.
The reactor room is isolated from the general building air flow system and is
independently ventilated. Both reactors were located within ventilated
enclosures. The systems were designed so that both reactors could be shut
down from the control room without entering the reactor room.
3.4 Analytical Systems
The test compound destruction measurements are performed by adsorption
onto Tenax resin followed by thermal desorption and gas-chromatograph/flame-
ionization detector analysis. The details of the system design, calibration,
determination of compound breakthrough volumes, and minimum sensitivity tests
are discussed in Appendix B.
In addition to the destruction efficiency tests, measurements of CO, CO,,
02» and unbumed hydrocarbons are performed.
3-T4
-------
3.4.1 Measurement of Destruction Efficiency (DE)
The equipment and techniques of the DE measurements are discussed in
detail in Appendix B and are similar to those of Parsons and Mitzner (1975),
and Dellinger et al. (1982). The technique, in summary consists of the
adsorption of waste compound from a known quantity of sample gas onto Tenax-GC.
This is accomplished by passing sample gas through an externally water-cooled
Pyrex cartridge packed with Tenax. At the conclusion of sampling, the car-
tridge is thermally desorbed and the contents are analyzed by flame ionization
gas chromatography (GC-FID). The procedure is to insert the cartridge into
the helium carrier gas line so that the helium flows sequentially through the
cartridge, the Porapak-Q column and the FID. The cartridge is placed within
an aluminum block heater which heats the cartridge to 120 C for a sufficient
time to desorb the test compounds and deposit them on the inlet of the room
temperature GC column. Following desorption the column oven is temperature
programed, the compounds separated, and the compound concentration mea-
sured.
The analytical technique is calibrated by sampling and analysis of known
gas-phase concentrations of the test compounds. The sampling flow rate was
optimized and the breakthrough volumes for each compound were determined to
be greater than the sampling volume; these tests are detailed in Appendix B.
One limitation with the present analytical system is that two of the
test compounds, benzene and 1,2-dichloroethane, could not be resolved by the
Porapak column. To avoid this problem two sets of compound mixtures are used
to prevent the simultaneous use of these two compounds. The mixtures are:
1) containing chlorobenzene, benzene, acrylonitrile, and chloroform; and 2)
containing chlorobenzene, acrylonitrile, 1,2-dichloroethane, and chloroform.
The Tenax-GC technique was selected because of its high sensitivity (1.5
ppb or 99.995 percent DE in the TFR) and the relatively rapid sample turn-
around time (1.5 hr). The technique is compatible with both the TFR and
microspray in that test compounds, auxiliary fuel, and products of incomplete
combustion encountered in this study were adequately separated. Repetitive
calibration tests have indicated an inherent experimental scatter that
corresponds to +5 percent at the 90 percent confidence level (See Appendix D).
3-15
-------
The regulations require that trial burn POHC Destruction and Removal
Efficiency (DRE) be 99.99 percent. DRE is based on the efficiency of both
the incinerator and any stack gas cleaning system. In the present experiment
there is no stack gas cleaning equipment before the measurement point, so
experimental destruction performance is reported as Destruction Efficiency
(DE) in this report rather than as DRE.
3.4.2 Other Measurements
In addition to the DE tests, the measurements shown in Table 3-2 were
performed. In any fuel-lean hydrocarbon flame, the overall efficiency is
defined in terms of fuel utilization. Fuel is fully utilized when all fuel
and fuel fragments are oxidized into CO2 and water. Any CO, fuel, or fuel
fragments that exist in the exhaust represent lost efficiency. Thus, the CO
and unburned hydrocarbon measurements are used as an indication of flame
efficiency. The CO2 and O2 measurements are used to calculate the burner
stoichiometry. This calculated burner stoichiometry is compared with the
stoichiometry based on the fuel and air flow rates to detect any operational
flow problems.
A means of independently verifying the DE calculation procedure is to
utilize a tracer whose concentration is not changed by the flame. If the
tracer does not appear at the concentration expected from the fuel flow, then
the DE calculation is being improperly done or uncontrolled loss of tracer is
occurring within the reactor or sampling system. One such tracer is the chlor-
ine present in three of the test compounds. In the flame, chlorine is quanti-
tatively converted to hydrochloric acid, HC1. Stack gas HC1 concentrations
are measured and compared with the flow of chlorine entering with the fuel.
A closed chlorine mass balance is a check for uncontrolled compound losses
and, in particular, a cross-check on the DE calculation procedures.
An HC1 measurement technique (CI2 was not expected nor observed at the
experimental conditions) was utilized which involved collection in midget
impingers containing 0.1 normal NaOH. At the conclusion of sampling, the
liquid is titrated against 0.01 normal mercuric nitrate. A normal acid-
base titration cannot be used because of dissolved CO2 interference;
mercuric nitrate is chlorine-specific.
3-16
-------
TABLE 3-2. EXPERIMENTAL MEASUREMENTS
Species or Variable
Measurement Technique
CO
Anarad nondispersive infrared
analyzer
CM
O
C_>
Beckman nondispersive infrared
analyzer
Hydrocarbons
Beckman Model 402 total flame
ionization detector
°2
Taylor paramagnetic analyzer
Gas Temperature
1. Microspray flame: Type S
(platinum-rhodium) uncoated
fine-wire thermocouple
2. Turbulent flame exhaust:
Type S thermocouple
3. Microspray exhaust: Type B
(copper-constantan) uncoated
thermocouple
Compound Mass Closure
Stack gas HC1 measurement
3-17
-------
A limited series of measurements were performed in the TFR to assure clo-
sure of CI mass balance. Table 3-3 shows the expected ppm of HC1 against the
measured stack HC1. The results indicate that within the experimental error
of HC1 measurement the chlorine mass balance is closed. These data were ob-
tained at a high DE ( 99.995 percent) condition in the TFR.
TABLE 3-3. CHLORINE MASS CLOSURE
HC1 Expected
HC1 Measured
(ppm)
(ppm)
340
339.6
350
336.9
3-18
-------
4.0 RESULTS AND DISCUSSION
Through the use of two flame reactors, the impact of a number of pos-
sible incinerator failure modes on relative compound destruction has been
explored. The microspray reactor provided information on the behavior of
droplets in a laminar environment without turbulent mixing limitations.
The impact of thermal parameters were investigated for the following condi-
tions for pure test compounds and test compound mixtures:
• Fuel-rich flat-flame: insufficient oxygen available to support
droplet flames.
• Fuel-lean flat-flame: post-flat-flame gas contains 10 mole per-
cent oxygen which is available for reaction with test compounds.
Other failure conditions (i.e., those caused by poor atomization, solid sur-
face quenching, and turbulent mixing limitations) cannot be effectively
investigated in the microspray reactor and were investigated in the turbulent
flame reactor.
4.1 Microspray—Droplet Decomposition in Flames
In the microspray flame reactor, droplets of test compounds were injec-
ted through a flat-flame. The droplets were widely dispersed in the laminar
plug flow region of the reactor to prevent interactions between individual
droplets or the flamelets around individual droplets. The impact of several
parameters on the relative decomposition of test compounds were investigated.
The parameters were controlled primarily by the conditions established by the
support flat flame and included the flat-flame temperature, the availability
of oxygen for oxidation of test compounds, and the effect of using droplets
of pure compounds compared with droplets of compound mixtures.
The major emphasis of this study was the relative behavior of different
compounds to flame conditions in order to determine if one compound was more
likely to escape destruction than another. The majority of the data were
obtained using mixtures of compounds. In that way, the test compounds were
assured of experiencing the same time/temperature history. Exhaust measure-
ments of the test compound concentration were made for each compound in the
mixture and compared. These measurements were made at each condition
4-1
-------
established by the support flat flame in order to examine if changes in the
relative compound destruction would occur for different conditions. The
relative order of test compounds measured in the microspray flame environ-
ment will later be compared to proposed incinerability ranking techniques
(section 5).
4.1.1 Droplets of Mixtures in an Oxygen-Rich Flame
In these tests, 38 urn droplets of two mixtures were separately injected
into a lean Hg/air/Ng flame. Exhaust plane concentrations of the test com-
pounds were measured for different flat-flame temperatures. The flat-flame
burner flow rates were adjusted to provide 10 mole percent oxygen in the
post-flame gas and to hold the (cold) gas velocity through the flat-flame
burner at 25 cm/sec. Gas temperature was the only variable; it was controlled
through the use of reactant flow rates while maintaining 10 percent excess
oxygen and 25 cm/sec flame velocity. The flat-flame temperature range was
selected to encompass complete compound destruction at high temperatures and
small destruction levels at low temperatures (i.e., 600-1100 K).
The results for the mixture of chlorobenzene, acrylonitrile, chloroform,
and 1,2-dichloroethane are shown in Figure 4-la. The fraction of each of the
test compounds that escaped the reactor unreacted is plotted against the flat-
flame temperature. The plot indicates that at low flat-flame temperatures
(650 K) all of the compounds except acrylonitrile passed through the reactor
with essentially no reaction. As the flame temperature was increased the
fraction of the compound that reacted increased up to a flame temperature of
850 K, above which destruction of all compounds became complete. At 850 K
and above no trace of compound was analytically indicated which demon-
strated that DE was at least 99.995 percent.
These OE measurements were supported by visual observation of the drop-
let behavior in the reactor. For temperatures at and above 825 K the drop-
lets supported individual flamelets; the minimum droplet ignition temper-
ature is indicated on the figure. Below the ignition temperature a very
faint form of chemiluminescence was observed in the area of the droplet
stream. This indicated that some form of radical reaction not associated
with individual droplet flames occurred below the droplet ignition temperature.
4-2
-------
-------
In Figure 4-lb, similar results are presented for the mixture in which
benzene replaced 1,2-dichloroethane. The qualitative features are identical
to ihose of Figure 4-1 a; i.e., well-defined droplet ignition point, high DE
above droplet ignition, slow destruction reaction below droplet ignition.
The data indicated that the ignition point had shifted from 825 K for 1,2-
dichloroethane mixture to 875 K for the benzene mixture. This was probably
due to the effect of the substitution of benzene for 1,2-dichloroethane on the
droplet ignition temperature. These results indicated that:
9 If a droplet flame was observed to be present, then destruction
efficiency was greater than 99.99 percent.
• The composition of the mixture determined the actual ignition tem-
perature of the droplet; however, if the droplet was ignited, the
destruction efficien was sufficiently high that no compound dif-
ferences were measured (i.e., no ranking could be determined).
0 At temperatures below ignition a form of incomplete oxidative,
radical-assisted decomposition occurred.
The rankings generated by this experiment were dependent upon the flat-
flame temperature:
• Well below ignition (650-750 K), the decomposition ranking in terms
of most difficult to easiest to remove was: chlorobenzene, benzene,
1,2-dichloroethane, chloroform, and acrylonitrile.
a Immediately below ignition temperature the ordering changed to:
benzene, chlorobenzene, acryloritrile, 1,2-dichloroethane, and
chloroform.
•< Temperatures above the ignition point resulted in very high DE,
so no ranking was obtained.
The thermal decomposition (Tgg gg) and AIT ranking (Table 2-1) is chloroben-
zene, benzene, acrylonitrile, 1,2-dichloroethane, and chloroform. Except for
the placement of acrylonitrile the Tgg ranking predicts the microspray
data ranking for temperatures well below ignition and is identical for tem-
peratures just below ignition. However, the flame-mode ignition temperatures
measured in the microspray (825-875 K) were considerably below the tempera-
4-4
-------
ture typical of nonflame thermal destruction (925-1038 K).
A plausible explanation for the different temperature requirements for
destruction is involved with the differences in waste concentration between
the microspray experiments and the nonflame thermal destruction experiments.
In the latter, the compound concentration is sufficiently low that radicals
resulting from the initial attack are recombined in the inert carrier gas.
In the microspray experiment, the locally high compound concentrations in the
vicinity of individual droplets can lead to the appearance of a droplet flame.
The droplet flame temperature can be considerably elevated above the flat-
flame temperature (825-875 K). This high droplet flame temperature and the
high concentrations of flame radicals act to accelerate the test compound
destruction beyond the rate predicted by nonflame thermal destruction data.
4.1.2 Pure Compounds in an Oxygen-Rich Flame
Data were also obtained in which droplets of pure test compound rather
than compound mixtures were injected into the microspray reactor. The experi-
ments were designed to ascertain the difference between decomposition of pure
droplets and droplets composed of a mixture of waste compounds. The
experiments were conducted in exactly the same manner as the mixture exper-
iments just described. Except for chlorobenzene, where a wide range of flame
temperatures were examined, the data were obtained as follows:
1. The reactor temperature was adjusted until the point of visual
ignition was determined.
2. Destruction efficiency measurements were obtained at the ignition
point, and at flame temperatures approximately 100 K above and
below the ignition point.
The results are shown in Figure 4-2 as a plot of unreacted compound
leaving the reactor versus the flat-flame temperature. The visually observed
ignition temperatures for each of the compounds were:
-------
i.o i 1 r
0.8
Aerylonitrile
Chlorobenzene
1,2-Dichloroethane
Chloroform
700 800 900
Flat-Flame Temperature (K)
1000
Figure 4-2. Fraction of test compound remaining in
exhaust when 38 um droplets of pure com-
pounds were injected into lean (10 percent
post-flame oxygen) h^/air flames as function
of flame temperature.
4-6
-------
Compound
Ignition Temperature (K)
Chloroform
1,2-Dichloroethane
Acrylonitrile
Chlorobenzene
Benzene
903
888
833
813
<700
Benzene droplets remained ignited until the flat-flame itself became exting-
uished. Thus, no benzene data are shown on the figure and the ignition
temperature was similar to the heat of combustion (chloroform, 1,2-dichloro-
ethane, chlorobenzene, acrylonitrile, and benzene) ranking except acryloni-
trile and chlorobenzene were reversed.
The following summarizes the similarities and differences between the
pure compound and mixture microspray data:
• In both experiments, the visual ignition temperature was found to
correspond to a shift between low DE at lower temperatures and high
DE at higher gas-flame temperatures.
• For pure compound droplets the ignition temperature was compound
specific.
• The droplets composed of a mixture experienced a single ignition
temperature; this characteristic mixture ignition temperature
(825-875 K) was in the same range as that of the pure compounds
( 700-903 K) which made up the mixture.
For pure compound droplets, a plausible explanation for the ranking
behavior involves the relative ability of a droplet to maintain a droplet
flame. The thermal theory of ignition (Semenov, 1935) holds that ignition
will occur when the heat generated by reaction exceeds the heat loss from the
droplet region generated by reaction. Heat production is strongly coupled
to compound heat of combustion and, thus, heat of combustion may be the best
indicator of relative DE among independently burning droplets of pure com-
pounds. However, other processes such as evaporation rate, specific oxida-
tion kinetics, and halogen flame inhibition may act as second order effects
to modify rankings. For droplets composed of a mixture of compounds all com-
4-7
-------
ponents within the mixture ignite at the same temperature which is character-
istic of the mixture and not the individual compounds. Thus, the actual igni-
tion temperature does not influence the relative compound destruction. For
temperature below the characteristic ignition temperature of the mixture, the
relative rates of oxidative removal kinetics of the individual components
determines the relative decomposition of the compounds.
4.1.3 Compound Mixtures With Low Oxygen
Experiments were performed to measure ranking behavior under conditions
where oxygen was not available for droplet ignition. The data were obtained
by operating the flat-flame at a constant fuel-rich equivalence ratio ($ =1.2).
To obtain lower temperatures,the fuel/air flow was proportionally reduced and
the flow of nitrogen increased to maintain a constant cold burner velocity
and equivalence ratio. The droplet stream was dispersed with the same mix-
ture which was supplied to the burner to insure that oxygen from the disper-
sion gas was at the same concentration as that of the burner gas.
The results are shown in Figure 4-3 as unreacted test compound leaving
the reactor as a function of flat-flame temperature. The data indicated an
abrupt change from zero to complete reaction at a flat-flame temperature of
1050 K. The individual compound behavior was so similar that it was diffi-
cult to derive a ranking from the data. However, the data near 1050 K dis-
played sufficient compound variation to establish the following ranking:
chlorobenzene, benzene, chloroform, and acrylonitrile. This ranking was
most closely matched by the Tgg gg and AIT ranking: chlorobenzene, benzene,
acrylonitrile, chloroform. The decomposition behavior showed an abrupt
change with temperature at about 1050 K. Only small changes in temperature
were necessary to produce a considerable change in DE, as illustrated on the
Figure.
These nonflame decomposition data differed from the flame-mode data in
two important ways. First, the nonflame reaction above 1050 K, where com-
pound consumption was complete, was not accompanied by droplet flame!ets, as
were the corresponding oxygen-rich data. Secondly, the decomposition reac-
tion occurred at a much higher flat-flame temperature (1050 K) than the cor-
4_tj
-------
1.0
0.8
T
— ~ Chlorobenzene
A Benzene
0.6 f— A Acrylonitrile
O Chloroform
0.4 '
0.2
1
A
A
A
8
£
600
1
800 1000 1200
Flat-Flame Temperature (K)
Figure 4-3. Fraction of test compound remaining in
exhaust when 38 ym droplets of mixtures
of compounds were injected into rich
(stoichiometric ratio =0.83) H2/air flames
as function of flame temperature.
4-9
-------
responding 10 percent oxygen data (825-875 K).
Since nonflame thermal destruction appeared to control compound removal
under low oxygen conditions, the data analysis was aided by a simple analyti-
cal model based on combined droplet evaporation and nonflame decomposition
behavior. The simple predictive model assumed the following:
• Time/temperature history of the reactor gas is identical to the
profiles shown in Figure 3-4a.
• Droplet boiling rate was controlled by the heat transfer rate
between the free stream and the dropl'et surface. Calculations
have shown that mass transfer rate of vapor away from the droplet
surface is relatively rapid compared to heat transfer; thus,
heat transfer is expected to be the controlling step.
• Vapor was chemically removed by first-order nonflame thermal
decomposition kinetics.
The model consists of two coupled first-order differential equations. The
details of the derivation and the solution technique are reported in Appen-
dix C. The calculation yields (1) a profile of the fraction of liquid evap-
orated as a function of time, and (2) a profile of the fraction of original
feed reacted as a function time. Results for the reactor exit (i.e., t = 1
sec which corresponds to the experimental sampling point) are shown in Fig-
ure 4-4 for the five test compounds and, for reference, ODT and methane
which have the lowest and highest thermal decomposition temperatures, respec-
tively, of compounds for which kinetic data are available. The principal
conclusions drawn from the model were:
1. For all conditions, droplets were completely evaporated early in
the burner (evaporation time of 38 ym droplets = 5 msec).
2. Predicted exhaust concentrations were unaffected by drop size for
D<400 ym, For 0<200 ym, results were insensitive to changes in
aH » Nu, and boiling temperature for values within the range of
vap
values represented by the test compounds, and were sensitive only
to time/temperature history and kinetic parameters. However, a
large-scale nozzle may have an upper limit of drop size were evap-
4-10
-------
1,2-Dichloroethane
Benzene
Acrylonitri'le
Chlorobenzene
m 0.6 —
\ DDT
Chloroform
1000
1200
600
800
Flat-Flame Temperature (K)
Figure 4-4. Model results: fraction of test compound
remaining in exhaust when 38 um droplets
were exposed to the temperature profiles
shown in Figure 3-4a and allowed to react
by nonflame thermal decomposition kinetics
4-11
-------
oration time becomes a significant fraction of destruction time
(Seeker and Samuel sen, 1981).
The main shortcoming of the model was that it did not include flame-zone
chemistry (flames about individual droplets or about vapor clouds). However,
this did not present a difficulty for the present fuel-rich conditions as
droplet flames could not persist due to the lack of sufficient free oxygen in
the post-flame gas.
Comparison of the model results (Figure 4-4) with the nonflame droplet
decomposition data (Figure 4-3) shows:
• The approximate flat-flame temperature at which compound reaction
becomes active in the experiment (1050 K) was well-predicted by the
model (970-1070 K).
• The data indicated that there was little difference in the tempera-
ture/destruction characteristic for the compounds tested. The
model indicated that the data should be spread over a range of
100 K for the compounds tested. A maximum variation of 400 K is
expected for all the kinetic data developed to date.
The observed similarity in rates between thermal destruction data and the
microspray pyrolysis data was surprising because the thermal destruction
results were obtained in 21 percent oxygen while the microspray results were
obtained with less than 1 percent oxygen. Thermal decomposition (Oellinger
et al.t 1982) results for PC3 indicated that reaction rata was first order
in 0^ between 2.5 and 35 percent at atmospheric pressure. Assuming the
first-order dependence applied to other organic compounds and taking benzene
for an example, the rate expression can be written in terms of 0^-
where C 3 benzene concentration (ppm)
p0 = oxygen partial pressure (atm)
k' = rate constant (sec"^ atm~^) defined by
4-12
-------
where = partial pressure of Oj in atmospheric pressure air (atm).
k * pub7ished rate constant based on air.
Thus, the above rate equation expresses both the correct rate for air and the
first-order oxygen concentration dependence. This equation was integrated
for-a constant temperature. Figure 4-5 shows the calculated unreacted frac-
tion as a function of temperature for a one-second residence time. The three
curves correspond to 21, 2.5, and 0.25 percent 0^ These calculations indi-
cated that a two-order of magnitude change in oxygen concentration resulted
in less than a 100 K shift in the destruction curve. Thus, the microspray
results were relatively insensitive to oxygen concentration and the low
concentration data would be nearly indistinguishable from the nonflame cal-
culations based on 21 percent 02-
The results suggest that nonflame thermal decomposition kinetics con-
trol both the absolute reaction rate and the ranking. Only a snail variation
in thermal behavior was observed between the compounds. The nonflame thermal
decomposition kinetic data would indicate approximately a 100 K variation
between the most and least incinerable test comoounds. A plausible explanation
is that the actual decomposition kinetics are modified by the high concentra-
tion of compounds present and the interactions between the compounds in the mi xture.
4.1.4 Summary of Microspray Rankings and Data
Microsoray Destruction Process. The overall DE of the microspray reac-
tor was governed primarily by the flat-flame temperature (droplet thermal
environment) and the availability of free oxygen in the region surrounding
the droplet. When sufficient oxygen was available the absolute destruction
was governed by the ability of the droplets, whether they were composed of a
pure compound or a mixture, to support a droplet flame. At low flat-.flame
temperatures the thermal environment of the droplets was too cold to establish
a droplet flame and low DE resulted. At sufficiently high flat-flame tempera-
tures the droplets establish flamelets and DE was essentially complete. Fig-
ure 4-6 shows a plot of flat-flame temperature versus fraction unreacted compound
4-13
-------
x 0.8
0.6
2.5% 0
800 1000 1200
Isothermal Reaction Temperature (K)
Figure 4-5. Model results: effect of oxygen concen-
tration on thermal decomposition for
benzene at isothermal conditions.
4-14
-------
I
on
300 400 500 600 700 800 900 1000 1100 1200
Flat-Flame Temperature (K)
Figure 4-6. Comparison of calculated reactor performance assuming 1) droplet
evaporation controls DE, 2) droplet ignition controls DF, and
3) thermal decomposition kinetics control DE; with microspray
oxidation mixture data and microspray low oxygen data.
to
3
IT)
cn
c
ro
E
CD
cc
c
o
u
n)
5-
1.0
0.8
0.6
0.4
0.2
Low-Oxygen
Microspray
Data
Ignition
Evaporation
Microspray
Oxidation
Mixture Data
Thermal
Decompo
sition
-------
leaving the reactor. Three theoretical regions are shown; each indicates
where the microspray data would be expected to fall if evaporation (based on
droplet evaporation time), ignition (based on AIT), or thermal decomposition
kinetics were the rate-controlling step governing compound release. The micro-
spray oxidation mixture data shown on the figure indicate that the control-
ling process is an ignition phenomena, which agrees with the interpretation of
the visual data discussed above.
When insufficient oxygen was available to support droplet flames the
reactor DE appeared to be governed by nonflame thermal destruction kinetics.
Figure 4-6 shows that the temperature requirements for complete destruction
in the microspray operated fuel-rich was compatible with nonflame thermal deconv
position kinetics. In summary, when free oxygen is present, ignition is the
rate-controlling process; and when insufficient free oxygen is available
thermal decomposition kinetics appear to be rate controlling.
Microspray Rankings. The relative decomposition of the test compounds
is shown in Figure 4-7 with values for several of the ranking procedures.
The figure was prepared with the concentration of the most prominent com-
pound shown as full scale and the remaining compounds expressed as a
percentage of the maximum concentration. None of the ranking procedures
exactly predicted any of the microspray rankings. However, the nonflame
thermal destruction (Tgg ggj and AIT procedures both closely agreed with
tha compound concentration measurements when the temperature was below
droplet ignition temperature and under oxygen-deficient conditions. Heat
of combustion was found to closely correlate the pure compound data when
the microspray was operated below droplet ignition temperature.
4.2 Turbulent Flame Reactor Results
A description of the turbulent flame reactor (TFR) and its range of com-
bustion variables was supplied in section 3. The purpose of the turbulent
flame experiment was to simulate the modes of escape of waste from the flame
zone of liquid injection incinerators. In the flame zone of such a system
finite rate mixing between fuel and air is expected to be the controlling
mechanism of combustion efficiency. In the TFR, exactly the same processes
govern the combustion efficiency because both are recirculation-stabilized
4-16
-------
NON-FLAME
TEMPERATURE
HEAT OF
NBS FLAME
AUTOIGNITION
INCINERABILITY RANKINGS
T99.99
COMBUSTION
RANKING
TEMPERATURE
CHLOROFORM
u
1,2 DICHLOROETHANE
i
BENZENE
ACRYLONITRILE
CHLOROBENZENE
Ivivivivivl
MICROSPRAY
OXIDATIVE
LOW TEMP <
OXIDATIVE JUST
3ELOW IGNITION
NO OXYGEN
.LOW TEMP
PURE COMPOUND
IGNITION
TEMPERATURE
CHLOROFORM
3
1,2 DICHLOROETHANE
BENZENE
ACRYLONITILE
CHLOROBENZENE
J
iij
Figure 4-7. Mlcrospray 1nc1nerab1l1ty rankings.
-------
turbulent diffusion liquid spray flames. The use of a high heat removal rate
for the TFR insured that the measurements would emphasize flame zone reactions.
Flame zone rankings were established for a variety of modes of inefficiency
in the TFR. As a result of the similarities between the flame zone of a liquid
injection incinerator and the TFR,the rankings established in this series of
tests should relate closely to the rankings from the flame zone of real incin-
erators undergoing the same forms of inefficiencies.
The approach used to specify the turbulent flow experimental conditions
was to locate highly efficient combustion conditions characterized by visual
stability, high DE, low exhaust carbon monoxide, and low unburned hydrocarbons.
There are termed "optimal" conditions. Poor operation was induced by perturb-
ing the highly efficient condition by a variety of means (e.g., changing air
velocity, introducing cold surfaces, etc.). These low efficiency conditions
are called "nonoptimal." The ranking that resulted from each form of ineffi-
ciency was then determined by comparing the relative DE measurements from the
five test compounds.
The OE data are, for most of the cases, presented twice in companion plots:
§ A linear plot of fraction unreacted compound is provided with a scale
of 0.0 to 0.03 to facilitate comparisons between different plots.
• A logarithmic plot is provided to quantify the data points
whose DE are too high to be resolved on the linear plots.
For each test condition simultaneous CO, CO^, and 0^ measurements were
obtained. The CO measurement was taken as a measure of the overall reactor
efficiency. The results indicated that the CO measurements correlate with
the measured destruction efficiency. For this reason, the CO data are repor-
ted along with the DE results for most of the test conditions- The C02 and
02 measurements were used primarily to check stoichiometry and reactor feed
rates and are not included in this report.
The conditions investigated in the turbulent flame reactor which had
a strong influence on destruction efficiency were primarily associated
with three failure conditions:
t Atomization parameters—poor atomization quality.
4-18
-------
• Quenching parameters—quenching on cold surface.
t Mixing parameters'—low excess air
—high excess air
—low heat release rate.
Those parameters found to be of less importance included burner velocity,
fuel type (No. 2 fuel oil) and concentration of hazardous waste compounds
(from 3 to 25 percent). For all data,the swirl number was unity. The spe-
cific experimental conditions are outlined in Table 4-1.
All tests where fuel-flow was varied required the use of several nozzles;
this was necessary so that atomization pressure, and therefore atomization
quality, was never significantly changed. Table 4-2 shows the flow rate
ranges for which particular nozzles were used. Only during the nozzle per-
formance tests were the constraints of this table not followed.
4.2.1 Influence of Stoichiometry and Load
These measurements constituted the baseline condition and were obtained
by varying the fuel flow rate at constant air flow and velocity. The atomi-
zation quality of the pressure atomized Delavan WDA-senes nozzles changed
with fuel flow. To maintain consistent atomization quality over the range of
fuel flow rates, it was necessary to substitute nozzles of different capacity
in accordance with Table 4-2.
Unburned hydrocarbon, CO, and relative heptane measurements were obtained
to characterize the relative efficiency of the flame. Heptane demonstrated a
significant volumetric breakthrough on the Tenax under the normal analytical
procecure (see Appendix B). Despite the fact that the Tenax did not quantita-
tively capture heptane, the calibration indicated that the analytical response
was linear in heptane concentration. Thus, the heptane measurements were
relative rather than absolute and were included only to differentiate be-
tween high and low heptane destruction conditions. These measurements, shown
in Figure 4-8, indicated that a region of high combustion efficiency existed
between approximately 120 and 150 percent theoretical air. At higher fuel
flows, mixing was not sufficient to eliminate all of the fuel-rich pockets,
even though the burner was operated at an overall fuel-lean stoichiometry.
4-19
-------
TABLE 4-1. TURBULENT FLAME EXPERIMENTAL CONDITIONS
Experimental
Cases
Theoretical
Air
(percent)
Fuel
Flow
(gm/sec)
Load
(kW)
Veloci ty
(111/sec)
Nozzle
Oelavan
WDA 60° Series
(nominal capacity
in gallons/hr)
Fuel
Compound
Concentration
in Fuel
(weight
percent)
Text
Figures
Baseline
110-210
.93-.54
42-24
7.1
(1)
Hep-
tane
3
4-8,4-9
Theoretical Air
Constant Load
120-200
1.08
48
25-15
1.5
B
B
4-10
Air Velocity
156
.72
32
7.1-17
1.0
B
B
4-11
Compound
Concentration
B
B
B
B
B
B
10,25
4-12
Cold Surface^)
Impingement
B
B
B
B
B
B
B
4-13,
4-14
Nozzle Perfor-
mance
110-320
.93-.34
42-16
B
(3)
B
B
4-15,
4-16
Auxi1iary
Fuel
B
B
B
B
B
Oil
B
4-17
B Variable identical to baseline experiment.
(1) Nozzles selected to maintain operation within capacity limits: see Table 4-2.
(2) Repeat of the baseline experiment with quench coll within the flame.
(3) 1.5-gallon/hr nozzle was operated at flows below design.
-------
TABLE 4-2. ATOMIZATION PARAMETERS
Heptane Flow
Percent
Theoretical
Air1
Nominal Nozzle
Flow
(gal/hr)2
Atomization
Pressure
(psig)
Load
(kW)
gm/sec
(gal/hr)
.93
1.3
120
1.5
121
42
.86
1.2
130
1.5
103
39
.79
1.1
142
1.5
87
36
.72
1.0
156
1.0
161
32
.65
0.9
173
1.0
130
29
.56
0.8
195
1.0
103
25
.50
0.7
223
0.75
140
23
.43
0.6
260
0.75
103
19
.36
0.5
312
0.5
161
16
^Based on the nominal air flow of 16.9 gm/sec (30.5 scfh).
2
Sizing parameter based on design flow rate.
4-21
-------
Firing Load (kW)
2000
<~)
4J •>-
CJ 4-1
01
C
O
.a
s_
(O
u
o
i.
-a
Nonloptima1
Nonoptimal
Optimal
Hydrocarbons "~
Heptane
1000
Percent Theoretical Air
Figure 4-8. Exhaust CO, total hydrocarbons, and rela-
tive heptane as a function of theoretical
air (constant air velocity, variable load:
42-24 kW, equal molar mixture of chloro-
form, benzene, chlorobenzene, and acrylon-
itrile added 3 percent by weight to heptane).
4-22
-------
At very fuel-lean conditions the excess air lowered the flame temperature and
likely quenched portions of the flame prior to the complete consumption of
reactants. These measurements indicated that the burner operated in a manner
consistent with practical systems. Under "optimal" conditions, consumption of
fuel and fuel fragments was high and the flame efficiency, as defined by fuel
consumption, was near unity. Under "nonoptimal" 'conditions the inefficiency
of the flame was characterized by release of unburned fuel and fuel inter-
mediates.
Destruction efficiency data for the corresponding conditions were
obtained for both the benzene and the 1,2-dichloroethane test compound mix-
tures. Figure 4-9a shows the data for the benzene mixture. The results are
plotted as fraction of compound remaining in the reactor exhaust as a function
of reactor stoichiometry and firing load. The general trends show that
DE and overall combustor efficiency are strongly related. The "optimal"
and "nonoptimal" conditions defined from Figure 4-8 correspond respectively
with the high (120-150 percent theoretical air) and low DE measurements in
Figure 4-9. The rankings obtained from these results, and in general all of
the following results were divided into optimal and nonoptimal conditions. For
optimal conditions the results indicated the following:
9 Absolute DE was always greater than 99.99 percent and was generally
greater than 99.995 percent.
• No repeatable ranking was observed. Rather, the compounds appeared
to be ranked in a random manner.
For nonoptimal conditions the results indicated:
• The DE for individual compounds was normally no worse than 98
percent. When one compound was released in significant quantities,
the other compounds tended to be within a factor of ten of the same
concentration.
At high DE, neither the absolute DE nor the ranking were repeatable. Instead,
the DE for each compound varied in a random manner between 99.98 and 100 per-
cent. Characterization of the measurement system (Appendix D) has shown that
this variation was not due to an analytical problem. Rather, it was attri-
buted to the random fluctuations that are inherent in a turbulent flame.
4-23
-------
Firing Load (kW)
-------
Over the 10-minute sampling period, 99.99 percent DE corresponded to 60 msec
of zero efficiency operation for an otherwise completely efficient flame.
Thus, the absolute DE and the ranking at high-efficiency conditions were sub-
ject to individual flame fluctuations and were expected, even over a 10-
minute sampling period, to show apparently random fluctuations. At lower DE,
conditions release was governed by large-scale escape mechanisms against
which the random fluctuations observed for the high-DE conditions were un-
measurable. This result may have serious implications to trial burn data.
Practical systems would be expected to behave in a similar manner to the
TFR under high efficiency conditions because the turbulent flames in each have
similar properties. Extension of these TFR results to a practical incinerator
would imply that high-efficiency operation would result in nonrepeatable
rankings. Thus, the POHC would not be guaranteed to be the lowest DRE com-
pound during the trial burn.
The corresponding experiment for the 1,2-dichloroethane mixture produced
the results illustrated in Figure 4-9b. The trends demonstrated by the ben-
zene mixture data in Figure 4-9a are also apparent here:
• High DE was measured between 120 and 150 percent theoretical air.
• Chloroform tended to be the most predominant compound at non-
optimum conditions. Chlorobenzene was normally the most easily
destroyed compound, and the remaining compounds showed various
rankings between chloroform and chlorobenzene.
• No repeatable ranking was observed for the high-efficiency condi-
tions.
4.2.2 Influence of Theoretical Air
In the data just discussed, the stoichiometry variation was obtained by
changing the fuel flow at a constant air flow rate. This procedure caused
the burner heat release rate to change (from 42 to 24 IcW) simultaneously
with stoichiometry. In this section, the effect of stoichiometry on DE and
ranking were investigated at a constant burner heat release rate (48 kW) by
varying air flow rate at a constant fuel flow rate.
4-25
-------
The data in Figure 4-10 were obtained by holding the fuel flow constant
and varying the air flow (fuel flow = 1.08 gm/sec; 3-weight-percent test com-
pound in hepatane; load = 48 kW). This resulted in a variation of both air
velocity and stoichiometry. The results demonstrated that a high-OE condition
was obtained at 120 percent theoretical air. The CO data indicated that
flame efficiency decreased as theoretical air was both increased and decreased
from this value. The DE data decreased from an optimum of greater than 99.995
percent, except for benzene, when air flow was increased, but the measurements
were not carried to a sufficiently fuel-rich condition to observe a cor-
responding degradation in 0E.
At nonoptimum conditions the ranking (difficult to easiest) was chloro-
form, acrylonitrile, benzene, and chlorobenzene. This ranking agrees with the
variable fuel flow rate data (Figure 4-9) in choosing chloroform and chloro-
benzene as the most difficult and most easily destroyed compounds. At the
optimum condition, all of the compounds were at unmeasurably low concentra-
tions (>99.995 percent DE) except benzene. This behavior can be interpreted
as indicating that benzene was a product of incomplete combustion (PIC)
resulting from the parent fuel or from one of the other test compounds (e.g.,
chlorobenzene). The actual source of the benzene, whether it is a product of
incomplete combustion or an indication of incomplete benzene destruction, has
not been determined. Benzene is a possible intermediate in the formation of
the soot which was observed in the flame in the form of luminosity, especially
at low excess air levels. Because of the relatively large levels of heptane
present (97 percent) only a small conversion of heptane to benzene is required
to account for the exhaust levels of benzene measured at this low excess air
condition. However, the benzene could also be the result of a simple trans-
formation of one of the test compounds, e.g., chlorobenzene.
The addition of excess air to turbulent spray flames decreases fuel and
fuel-fragment consumption efficiency through a reduction in flame temperature
and through thermal quenching. Thermal quenching can occur in a diffusion
environment when cold air is rapidly mixed with a burning packet of gas so
that the combustion reaction is extinguished prior to completion. The same
mechanism is likely to adversely affect DE.
The results of Figure 4-10 for optimum conditions indicated that the
4-26
-------
c
0)
CD
>)
X
O
o
"C
OJ
+J
u
QJ
s.
1.
o
<_>
E
o.
o.
o
o
4000
3000
2000
1000 -
(a) Exhaust CO
Concentration
125 150 175 200
Percent Theoretical Air
9
99
.9
99.99
Detection Limit
999
100
25 150 175 200
Percent Theoretical Air
225
(b) Destruction and
Removal Efficiency
O Chloroform
A- Acrylonitri le
^ Benzene
D Chlorobenzene
Figure 4-10. Impact of theoretical air on CO and DRE from turbulent flame
reactor (constant load: 48 kW; variable air flow rate and
burner velocity; equal molar mixture of compounds added 3 per-
cent by weight to heptane).
4-27
-------
variation in DE between the most- and least-prevalent compound was no greater
than a factor of ten. This was a small variation compared to nonflame ther-
mal decomposition kinetics where many orders of magnitude can separate com-
pound concentrations at specific temperatures. For example, the kinetic data
in Table 3-4 indicates that at 925 K and one second residence time the OE for
benzene is 20 percent and for 1,2-dichloroethane is 99.95 percent, a differ-
ence of over 3 orders of magnitude in emission rate. One plausible mechanism
of incomplete destruction under high excess air conditions is flame quench-
ing. Flame quenching occurs when burning packets of gas are rapidly mixed
with cold air so that the combustion reaction is abruptly stopped. As an
example, consider two test compounds A and B. These are present in equal
concentration in the center of a fuel-rich packet; they are of equal concen-
tration in the center of a fuel-rich packet; they are of equal concentration
because 1) A and B entered the reactor in equal concentrations, and 2) the
center of the packet has not experienced a combustion reaction. A diffusion
flame exists at the packet boundary where free oxygen diffuses into the
packet gas. Turbulent quenching occurs when the packet is rapidly diluted
with cold air so that the boundary diffusion flame is extinguished and the
test compounds are diluted beyond their capacity to support further com-
bustion. Thus, given an equal prequench concentration, an instantaneous
quench, and no postflame reaction, the two compounds are released by the
flame in equal concentrations. In reality, second-order effects such as
relative susceptibility to quench or halogen flame inhibition may be super-
imposed on the first order quenching phenomena. This would result in a
relatively small variation in DE, similar to the factor of ten variation
that was experimentally observed.
4.2.3 Influence of Air Velocity
The effect of burner air velocity on DE and ranking was investigated by
varying the area of the burner throat while holding fuel flow, total air flow,
and atomization quality constant. This was accomplished by using three sepa-
rate inserts of different internal diameter as described in section 3.2.
The results shown in Figure 4-11 were obtained at the 156 percent theo-
retical air optimal condition (fuel flow = 0.72 gm/sec; 3-weight-percent test
compound in heptane; load = 32 kW) with three velocities: 7.1, 14.1, and
4-28
-------
io n ~ is
Air Velocity (m/sec)
99.
99.9 ~
C
QJ
O
s.
-------
17.2 m/sec. The results are shown as fraction remaining in the exhaust as a
function of burner air velocity. The data indicated that the optimal opera-
tion of the reactor was relatively insensitive to inlet air velocity over the
range investigated (7.1-17.2 m/sec). All of the compounds had DE equal to
or greater than 99.9 percent. The rankings from this high-efficiency condi-
tion were random, as was expected due to the nature of the escape mechanism
for optimal conditions as previously discussed.
4.2.4 Influence of Compound Concentration
In the baseline test experiment described in section 4.2.1, the test
compounds constituted only 3 percent by weight of the total feed, the balance
being the heptane auxiliary fuel. Hence, the auxiliary fuel determined the
thermal environment and as a result, the test compounds each experienced
identical thermal histories. In practical liquid injection incinerators,
enough auxiliary fuel is added to bring the total heating value up to that
necessary for stable combustion and to provide the desired post-flame tem-
perature. As a consequence, the waste can influence and, at elevated levels,
control the flame temperature. Also, the presence of significant quantities
of waste compound in the fuel could alter the combustion chemistry. Examples
of such behavior include the inhibition of flames by halogens (Biordi et al.,
1975) which are present in many waste streams, and increased sooting caused
by the high aromatic content of certain wastes (Glassman and Yaccorino, 1981).
In the present experiment, the effect of elevated test compound concen-
tration was determined by performing experiments with 10- and 25-weight-
percent waste concentration. These results are plotted, along with the base-
line 3-percent results, for chloroform, acrylonitri1e, benzene, and chloroben-
zene in Figure 4-12. The experiments were performed at constant air velocity
and variable fuel flow rate and load. Despite almost an order of magnitude
change in compound concentration, no consistent variation in fractional
destruction was observed. The increased test compound concentration did not
result in any significant deterioration of the flame's ability to destroy the
test compounds. Rather, the majority of the results showed behavior charac-
teristic of optimal performance (i.e., high OE and random ranking).
4-30
-------
A. CHLOROFORM
01
o
s_
(1)
Q.
LU
o
90
99
99.9
99.99
03
02
01
~ 3% Waste
O 10% Waste
_ ^ 25% Waste
100 125 150 175 200 225
Percent Theoretical Air
rr—
I r
I "
~ 9
_ o A
t>
o o
A
~
A
T
i
0
t l
i
100 125 150 175 200 225
Percent Theoretical Air
B. BENZENE
~ 3% Waste
£ 25% Waste
0
u
I
100 125 150 175 200 225
Percent Theoretical Air
c
O)
(J
s_
O)
CL
99.99 -
125 150 175 200
Percent Theoretical Air
Figure a-12. Impact of comDound concentration on fraction test
compound remaining exhaust of turbulent flow reactor
(constant air velocity, variable load: 39-25 kW;
variable weight concentration of an equimolar mixture
of test compounds in hon!:ane).
4-31
-------
C. ACRYLONITRILE
,03
.02
.01
0
~ 3% Waste
O10% Waste
£25% Waste
i
i
100 125 150 175 200 225
-------
4.2.5 Effect of Quench Coil
The composition of fuel feed in an incinerator can vary considerably
from time to time which can lead to considerable changes in flame shape and
size. One problem associated with large changes in flame shape is the poten-
tial of flame impingement on a solid surface. The extreme limit of this
behavior would be impingement on a cold surface. In the present experiment,
heterogeneous quenching by cold-surface impingement was investigated by plac-
ing a water-cooled coil directly into the flame.
The data were obtained by repeating the baseline test described in sec-
tion a.2.1 in which air velocity was held constant and fuel flow was varied
to obtain various firing loads and stoichiometrics. A water-cooled quench
coil was placed directly into the flame zone. The presence of the coil
decreased the overall efficiency of the flame. In Figure 4-13 is shown CO
for both the coil-quencti conditions and the corresponding quench-free data.
The discontinuities in the quench-coil CO data at 156 and 210 percent theo-
retical air corresponded to points where nozzle size was shifted to prevent
changes in atomization quality. At these points slight changes in atomiza-
tion quality and the positioning 'of the flame relative to the coils was
expected which could influence flame performance. The deviation between the
quench and no-quench data was greatest at the previous optimal 1ow-theoretical-
air/hign-load condition. At this condition the quench coil was fully emersed
in the flame. As fuel flow was decreased, the size of the flame became smal-
ler, and eventually the flame separated from the quench coil. At this condi-
tion the coil no longer affected the flame zone as confirmed by the virtually
identical data for CO above 200 Dercent theoretical air.
The cooling coil changed what otherwise was an optimum condition into
a nonoptimum condition, as indicated by CO measurements. Figure 4-14
shows that this quench-induced nonoptimum condition also resulted in sub-
stantially degraded 0E performance. The plot demonstrates that measured test
compound emissions are considerably elevated over the corresponding non-
quenched data (dashed line obtained from data of Figure 4-9). Resolved
compound DE data were obtained at two previously optimal conditions, 130
and 156 percent theoretical air. The samples obtained at stoichiometrics
4-33
-------
5000
4000
<
LU
+J
C
(U
CJ
i.
0)
a.
3000
o
= 2000
Q.
Quench Coil
A*
o
1000
No Quench Coil
100
Percent Theoretical Air
Figure 4-13. Influence of quench-coil on CO in exhaust of turbulent
flame reactor as a function of load and theoretical air
(constant air velocity, variable load: 42-16 kW; equi-
molar mixture of compounds added 3 percent by weight
to heptane).
4-34
-------
.06
.05
Chloroform
.04
X
LU
C
o>
c
re
B
0)
cz
Acrylonitrile
c
o
•»- . 02
4->
LL.
.01
Benzene
£r
Chlorobenzene
Hi
Maximum
Uncooled
.Data
¦r>—r
100
125
150
175
200
225
Percent Theoretical Air
Figure 4-14. Influence of quench-coil on fraction-of test compound
remaining in exhaust of turbulent flame reactor as a
function of load and theoretical air (constant air
velocity, variable load: ^2-24 kW; equimolar mixture
of compounds added 3 percent by weight to heptane).
-------
outside this range had high levels of test compound and other interfering
species and the chromatogram peaks could not be resolved. These results,
while not quantitative, indicate that DE performance was poorer outside of
the 130 to 156 percent theoretical air range than the reported data within
the range.
The compound ranking, from most difficult to easiest to destroy, was found
to be: chloroform, acrylonitrile, benzene, and chlorobenzene. This ordering
was similar to the measured high-excess-air ordering (Figure- 4-9; chloroform
most difficult, chlorobenzene easiest).
4.2.6 Effect of Nozzle Performance
The pressure-atomized nozzles used in this study function by the breakup
of liquid as it is forced through a small orifice under pressure. The atomi-
zation quality (droplet size) of the spray is primarily determined by the
liquid pressure. These nozzles are designed to operate at liquid pressures
in excess of 50 psig. At reduced pressures the mean droplet diameter for the
spray increases (Dietrich, 1979) and the ballistic velocity of the individual
droplets decreases. For the baseline experiment data presented in Figure 4-9,
atomization pressure was held constant by interchanging nozzles of various
capacities as the fuel flow was changed. Specific nozzles and corresponding
flow rates are identified in Table 4-2.
In the present series of experiments, the CO and DE results for this on-
design atomization condition were compared with the results where a single
nozzle of relatively large capacity (1.5 gallon/hr) was used. The low atomi-
zation pressures used with this large nozzler resulted in off-design atomiza-
tion performance.
The effect of nozzle performance on CO is shown in Figure 4-15 where CO
corrected to 0% 0^ is plotted against percent theoretical air. The "on-design"
curve corresponds to the data of Figure 4-9 and was obtained -by interchang-
ing nozzles to prevent the atomizing pressure from declining to the point
where poor atomization quality is expected. The "off-design" curve was
obtained by using the same capacity nozzle through the entire fuel-flow vari-
ation. Note that the large-capacity nozzle became "on-design" below 156 per-
cent theoretical air; hence, the two sets of data become identical in this
4.-36
-------
150
Solid=0n-Design
0.75
100
Dashed=0ff-Design
50oq
4000
Off-Design
3000
2000-
On-Design
100C-
O
250
300
150
200
100
Percent Theoretical Air
Figure 4-15. Influence of atomization quality on CO in
exhaust of turbulent flame reactor as a func-
tion of load and theoretical air (constant air
velocity, variable load: 42-16 kW; equimolar
mixture of compounds added 3 percent by weight
to heptane). Also, atomization pressure as a
function of theoretical air for each of the
nozzles (parameter is nozzle capacity in gal/hr).
4-37
-------
region. At the top of the figure the atomizing pressure for the on- and off-
design conditions are plotted. As theoretical air was increased (fuel flow
decreased), the divergence between the two pressures was reflected in the
increased divergence between the on-design and off-design CO plots.
The corresponding DE measurements are shown in Figure 4-16. The OE per-
formance was substantially degraded for 156 percent theoretical airand higher
excess air conditions. Chloroform at 310 percent theoretical had a OE of 50
percent, by far the lowest measured in the course of this study. The ranking
obtained from these data was (most difficult to easiest to destroy) chloroform,
benzene, acrylonitrile, and chlorobenzene. This ranking was similar to the
rankings from the baseline experiment (Figure 4-9) and the quench-coil condition
(Figure 4-14) in that chlorobenzene was the most easily destroyed compound and
chloroform the most difficult.
The poor DE performance probably was caused by either the increase in
droplet diameter or the decrease in ballistic velocity associated with low
nozzle pressure. Large droplets may not have sufficient time to completely
evaporate in the flame, or they may strike the cold wall and escape the flame.
The lower droplet velocity could change the flame shape which can affect over-
all flame efficiency and OE.
4.2.7 Effect of Auxiliary Fuel
In addition to mixing and quenching, the turbulent-flow experiment was
designed to consider the effect of auxiliary fuel test compound interactions
on the experimental ranking. The turbulent flow tests previously described
were performed for a simple aliphatic auxiliary fuel, heptane. The present
series was performed with a chemically complex petroleum distillate: No. 2
fuel oil. In these experiments the baseline experimental test shown in Fig-
ure 4-9 was repeated with No. Z fuel oil replacing heptane as the auxiliary
fuel. The results are shown in Figure 4-17 as fraction test compound remain-
ing in the exhaust versus burner firing load and theoretical_air. The simi-
larities with the corresponding heptane data included a high OE optimal con-
dition between 120 and 150 percent theoretical air, and a ranking in which
chloroform was the most and chlorobenzene the least prominent compound. At
low-theoretical air/high-load conditions, chlorobenzene became the most promi-
4-38
-------
0.6
0.5 -
CO
3
c
ro
E
OJ
as.
0.3
c
o
Z 0.2
u
0.1
"Standard"
Atomization
Chloroform,
Chlorobenzene
Benzene
Acrylonitrile
100 150 200 250 300
Percent Theoretical Air
350
Figure 4-16. Impact of atomizer performance on fraction of
test compound remaining in exhaust as a func-
tion of percent theoretical air (constant air
velocity, variable load: 42-16 kW; equimolar
mixture of compounds added 3 percent by weight
to heptane).
4-39
-------
.03
~ Chlorobenzene
£ Benzene
Acrylonitrile
O Chloroform
UJ
c
c
o
u
<0
LU
100
125
150
175
200
225
Percent Theoretical Air
Figure 4-17. Impact of auxiliary fuel type on fraction of
test compound remaining in exhaust as a func-
tion of percent theoretical air (constant air
velocity, variable load: 42-24 kW; equimolar
mixture of compounds added 3 percent by
weight to No. 2 fuel oil).
4-40
-------
nent compound. This differs from the heptane results in which chloroform
was also the highest concentration compound at low-theoretical-air conditions.
4.2.8 Turbulent Flow Reactor Data Summary
The following observations were supported by the TFR data:
• Certain experimental conditions can be classified as having optimal
flame performance. These conditions can be characterized by low CO
and unburned hydrocarbon concentrations and high destruction effi-
ciencies (>99.995 percent). This observation supports the conten-
tion that flame efficiency is correlated with DE. Although this
observation could be qualified by the presence of an afterburner or
a postflame region, this notion suggests that a potential indirect,
continuous, real-time monitoring of incinerator performance could be
developed by monitoring species such as CO and total hydrocar-
bons which reflect flame efficiency. However, much careful charac-
terization work is necessary to insure that these potential techniques
are not overly conservative and that the response time of the instru-
ment is compatible with the response time of the incinerator.
• Certain flame parameters did not, within the values used for this
experiment, significantly deteriorate the performance of the opti-
mum flame. These include burner air velocity, auxiliary fuel type,
and high-compound concentration.
• Flame parameters that resulted in degraded DE performance included
high theoretical air, low theoretical air, low heat release rates,
cold surface impingement, and degraded nozzle performance.
• Any compound differences measured near optimum conditions appeared
to be random.
t The rankings observed for non-optimum operation were dependent on
experimental conditions, on the cause of the inefficiency (see
section 4.3). In particular, the ranking (most difficult to easiest)
of chloroform, benzene, acrylonitrile, and chlorobenzene appeared in
many cases where quenching phenomena were expected.
4-41
-------
5.0 SUMMARY OF RANKINGS, CONCLUSIONS, AND RECOMMENDATIONS
5.1 Summary and Discussion of Rankings
One of the program objectives was to intercompare the various ranking
procedures. The development of these rankings proceeds from the assumption
that incinerability is related to the relative thermal stability'of the com-
pounds (Cudahy et al., 1981). The thermal stability is either measured
directly or inferred from another compound property. The concept of thermal
stability is presumed to be universal; that is, it applies equally well to
flame and post -flame processes, and to decomposition reactions in dilute and
concentrated waste streams. The ranking procedures were developed as a means
of estimating the relative thermal stability of waste compounds. The proposed
procedures include.:
Thermal Decomposition Kinetics. Under this approach the thermal stabil-
ity is assumed to be related to the dilute phase nonflame thermal decomposi-
tion kinetic rate. This is measured in an experimental plug-flow reactor and
reported as the temperature required to produce a 99.99 percent DE over a spe-
cified residence time (Tgg gQ) (Lee etal., 1979, 1982; Duval 1 and Rubey, 1976,
1977; Dellinger et al., 1982).
Autoignition Temperature (AIT). The AIT is used to determine the ten-
dency of a compound to initiate combustion without the presence of an external
ignition source. AIT has been found to correlate with thermal destruction
data (Cudahy et al., 1981). The principal advantage to this approach is that
AIT data are much more easily and rapidly obtained than thermal decomposition
data and a much larger data base is presently available. If thermal stability
is assumed to be measured by thermal decomposition kinetics, then AIT is pro-
posed as an approach to estimating thermal stability.
Heat of Combustion. This procedure was initially proposed because 1)
increased chlorine substitution tends to yield higher thermal stability, as
measured by Tgg gg, and 2) heat of combustion tends to decrease with chlorine
substitution. This yielded a potential relationship between heat of combus-
tion and Tgg gg (Cudahy et al., 1981).
5-1
-------
Property Correlation. Attempts have been made to generalize the results
of the thermal decomposition experiments so that a ranking similar to the
thermal decomposition ranking can be established without performing experi-
ments (Lee et al., 1979; Lee et al., 1982). The approach is to correlate
Tgg gg with autoignition temperature and other compound specific physical and
structural properties. The results have indicated that the strongest corre-
lation is with autoignition temperature.
Susceptibility to Radical Attack. Tsang and Shaub (1981) have assumed
that attack on the weakest compound bond by flame radicals is the process
controlling destruction. The rankings are established by the grouping of com-
pounds according to the bond most susceptible to radical attack.
Miscellaneous Approaches. These include molar heat of combustion, heat
of formation per unit weight, Gibbs free energy per unit weight, ionization
potential, flash point, activation energy, and heat of formation of ion per
unit weight. Various arguments have been presented as to why each of these
can be correlated to Tgg gg and, hence thermal stability (Cudahy et al., 1981).
The results of the present study indicate that no single ranking is
appropriate to describe the relative compound measurements from the labora-
tory-scale turbulent spray flame. This implies that the concept of thermal
stability ranking is not universal but is condition-dependent.
Both the microspray and the TFR were operated at high-efficiency, high
DE conditions, and at conditions which simulated practical incinerator fail-
ure modes. The microspray reactor was selected to separate thermal and stoi-
i
chiometric effects from turbulent mixing effects. In the microspray, high-
efficiency OE resulted when the gas temperature was sufficiently high to
sustain individual flames around each droplet for conditions where free, oxygen
was available to the droplet. When insufficient oxygen was available to sup-
port droplet flames, high OE occurred when the gas temperature exceeded the
level required for nonflame thermal decomposition to become important. Under
both the high-efficiency conditions, the DE was too high to be measured
(>99.995 percent) and no ranking was determined.
In the TFR, high efficiency resulted in either complete decomposition of
the test compounds, or a very high OE with a random ranking. This lack of
5-2
-------
ranking appeared to be caused by the random nature of the turbulent flame and
only a brief excursion from high-efficiency conditions was necessary to cause
a measured DE below 99.99 percent. It should be emphasized that this process
is not unique to this experiment and will likely affect the behavior of any
large-scale turbulent diffusion flame, such as the flame zone of a liquid
injection incinerator.
Under various failure conditions for both the microspray and the TFR,
measurable differences between the five test compounds were observed. It was
found that the incinerability or ordering of the compounds was dependent on
the actual failure condition. For example, chlorobenzene was the most diffi-
cult to eliminate in the microspray when the temperature was too low to ignite
the droplets, but was the leas^Jdifficult to eliminate for a variety of fail-
ure conditions in the TFR such as poor atomization quality. Figure 5-1 pre-
sents a series of bar graphs which were prepared in the manner described in
Section 4.1.1, which allows a comparison between incinerability as defined by
the various failure modes and the rankings indicated by procedures based
upon Tgg gg, heat of combustion, the NBS method, and AIT. The bar graph
shows the concentrations measured in the experiment normalized so that the
most predominant compound shows full scale and the lesser concentrations are
expressed as a percentage of that maximum concentration. This approach gives
an indication of the measured magnitude of the difference in destruction
efficiency between compounds. A comparison of these relative concentration
measurements with proposed incinerability ranking techniques demonstrates
that none of the proposed techniques agree with the data for all failure con-
ditions. However, some of the ranking procedures were found to be appropriate
for specific failure conditions. For example, the nonflame thermal destruc-
tion (Tgg gg) and AIT procedures both agreed with the compound concentration
measurements when the temperature was below droplet ignition temperature and
under oxygen-deficient conditions. Heat of combustion was found to correlate
the pure compound data when the microspray was operated below droplet ignition
temperature but in sufficient oxygen. In most instances investigated in the
TFR under specified failure conditions, chloroform was the most difficult
compound to incinerate, and this was anticipated by only one of the four
ranking techniques—heat of combustion.
5-3
-------
NON-FLAME
TEMPERATURE
heat of
NBS FLAME
AUTOIGNITION
INCINERA8ILITY RANKINGS
T99.39
COMBUSTION
RANKING
TEMPERATURE
CHLOROFORM
55T
*
1,2 OICHLOROETHANE
I
3ENZENE
n
ACRYLONITRILE
CHLQROBENZENE
r
MICROSPRAY
OXIDATIVE
LOW TEMP
OXIDATIVE JUST
BELOW IGNITION
MO OXYGEN
.LOW TEMP
PURE CQMPOUNO
IGNITION
TEMPERATURE
CHLOROFORM
1,2 QICHLQRQETHANE
3EKZENE
ACRYLQNITTLi
CHLOROBEN2BE
m
TURBULENT nJ>ME
HIGH | , LOW
LXCE33 AIR | EXCESS AIR
POOR | I
ATOMIZATTON QUENCH COIL
CHLOROFORM
1,2-OICHLORO ETHANE
3ENEZENE
ACRYLONITRILE
CHLQROBENZENE
No.2 FUEL OIL No.2 FUEL OIL
turbulent fume
HIGH
EXCESS AIR
LOW
OCcSS AIR
POOR
ATOMIZATTON
POOR
ATOMIZATTON
CHLOROFORM
1,2 OICHLOROETHANE
. BENZENE
ACRYLONITRILE
CHLOROBENZZNE
l
i
!>>l
s
m
Figure 5-1. Comparison of proposed ranking techniques and
relative compound decomposition of compounds
under flame failure conditions.
5-4
-------
Although measurable differences in the destruction efficiency of the five
test compounds were obtained, the differences were not large relative to the
extreme differences that can occur in thermal destruction experiments (e.g.,
at 925 K and 1 sec benzene DE = 20 percent, acrylonitrile DE = 99.995 per-
cent). For the most part the variation in the concentration (between highest
and lowest) of the compounds in the exhaust was typically of the order of
ten, although larger variations were measured under some circumstances. , This
suggests that the POHC selection may not be very critical because the dif-
ferences between compounds are relatively small. If the permit writer selects
three compounds based upon two or more ranking techniques, and it is demon-
strated that their DRE is greater than 99.99 percent, then it is very
unlikely that any other compounds will be destroyed to a lesser degree.
Nonetheless, to be assured that the incinerator is quantitatively destroying
all compounds requires measurement of the most difficult to destroy compounds
under all potential failure conditions.
This study has identified the differences between compound destruction
efficiency caused by failure conditions associated with the flame zone. High
destruction efficiencies have been demonstrated in the flame alone. However,
many incinerators are equipped with post-flame hold-up zones and afterburners
m order to acmeve additional thermal decomposition of compounds which
escape the flame zone. In order for an incinerator to fail to destroy a
compound, the material must both escaDe the flame and the temperature be too
low in the post-flame hold-up zone to destroy the compound (less than Tgg gg).
The differences rn the concentration of compounds in the exhaust of the incin-
erators is associated with both the flame and nonflame zones. The thermal
decomDosition which occurs in the post-flame zone can alter the ranking in the
exhaust. However, this occurs only if the temperature in the post-flame zone
is between the TgQ gg of the two compounds. For example, the TFR data indi-
cated that chloroform with a Tgg gg of 930 K'is the most likely compound
to escape the flame and chlorobenzene is the least likely with a Tgg gg of
1039 K. If the post-flame zone temperature is less than 930 K, then the flame
zone ordering will prevail in the exhaust. If the temperature is greater
than 1038 K, then both compounds are quantitatively destroyed if the resi-
dence time is greater than 1 second. However, if the temperature is between
5-5
-------
930 and 1038 K, then chloroform is destroyed leaving chlorobenzene intact.
Hence the postflame rank will prevail if, and only if, the temperature in
the postflame is between the two compounds; in this case, a 100 C tempera-
ture range. It should be noted that the temperature in the post-flame zone
is not uniform and the temperatures referred to above are minimum tempera-
tures for a residence time of one second.
5.2 Conclusions
This study represented a first attempt to assess the appropriateness of
the various proposed ranking procedures to flame-mode destruction. In these
tests the destruction of only five waste compounds was investigated under a
variety of conditions which simulated liquid injection incinerator failure
modes. The results support three broad conclusions with regard to flame-
mode destruction and the resulting incinerability rankings.
Conclusion 1: Flame Destruction of Waste Compounds. Flame destruction
of hazardous waste compounds differs from nonflame thermal decomposition due
to the addition of processes such as local heat release, finite rate mixing,
ignition, free radical attack and high flame temperature thermal decomposi-
tion. The experiments have shown that under optimal conditions of mixing,
temperature, and stoichiometry flames are capable of destroying hazardous
waste compounds to very high efficiencies (greater than 99.995 percent) with-
out the need for high-temperature, long residence-time post-flame zones.
Reduced flame destruction efficiency is the result of operation under some
failure mode such as poor atomization, poor mixing, flame quenching, etc.
Conclusion 2: Ranking. No single ranking procedure was found to be
appropriate for all of the flame-mode failure conditions that were studied.
Rankings were dependent on the particular reactor used and on the particular
conditions used for each test. This implies that single-parameter ranking
procedures cannot a priori be assumed to rank incinerability within a given
unit.
Conclusion 3: Turbulent Flame Behavior Correlation. 3ehavior in the
turbulent flow reactor was characterized into two general modes: optimum and
nonoptimum. Optimum behavior was characterized by high overall combustion
efficiency, which is defined as complete consumption of fuel and fuel
5-6
-------
fragments. This condition can be verified easily by low CO and unburned
hydrocarbon measurements. Conversely, nonoptimum conditions are defined by
incomplete consumption of fuel and fuel fragments, and are indicated by
relatively high CO and unburned hydrocarbon exhaust concentrations. The
results of this study indicate that a high compound destruction efficiency
is characteristic of optimum operation in a turbulent spray flame and the
lower destruction efficiency occurs during nonoptimal performance. This cor-
relation between flame efficiency and destruction efficiency offers a poten-
tial means of monitoring incinerator performance in a real-time, continuous
manner relying on measurements of CO and THC.
5.3 Recommendations
This irrrtial study into the nature of the flame-mode incineration pro-
cess indicates a number of areas where supplemental effort would be valuable.
1. Additional Compounds. The five compounds selected for the present
study represent a wide range of values within the proposed ranking
systems. However, further testing of an additional group of com-
pounds would allow an evaluation of whether the original five com-
pounds represent the limits of behavior. Additional testing would
also be directed toward finding if any subclass of compounds were
unexpectedly difficult to incinerate. The TFR results indicated
that chloroform was the least incinerable compound for many condi-
tions. This behavior was not initially expected and a number of
plausible explanations have subsequently been put forth. Additional
compounds should be selected which will test the various possible
explanations for the predominance of chloroform in the exhaust.
2. Incinerability Ranking. No one incinerability ranking system
appears to predict correctly the relative destruction efficiency of
the five compounds tested for all failure conditions investigated.
The fact that neither flame reactor yielded a single ranking for all
experimental conditions indicates that a single universal ranking is
unlikely for the wide variety of incinerator designs in use. Since
no single ranking is appropriate, the results of this study suggest
that POHCs must be selected by consideration of the actual system.
5-7
-------
A system-oriented engineering analysis methodology could be devised
which would predict the likely failure conditions associated with
each incinerator and the most likely POHC for each failure condi-
tion. The results of this study indicate that the POHC for a wide
variety of failure conditions could be selected by considering a
few rankings of incinerability. Specifically, relative compound
destruction data from a number of flame failure conditions corre-
lated with nonflame decomposition data. However, many relative
compound effects from the turbulent flame reactor were not corre-
lated with any incinerability ranking (heat of combustion was
closest). This suggests that another incinerability ranking is
required which considers the specific failure condition associated,
with a turbulent flame environment. If the dominant mechanism of
escape from the turbulent diffusion spray flame is thermal quench-
ing then a susceptibility-to-quench experiment may be required to
correlate the data.
3. Further Types of Experimentation. The experiment in this study
concentrated on the flame zone. *A logical extension would be to
consider additional failure modes associated with other portions of
the incinerator. Such experiments might examine how rankings are
modified by post-flame thermal hold-up zones, afterburners, or
scrubbers. These data would more fully determine the limitations
of incinerability ranking systems and aid the development of an
appropriate incinerability ranking methodology.
5-8
-------
6.0 REFERENCES
Badzioch, S., 1967. Thermal Decomposition in Combustion of Pulverized Coal,
by N. A. Field, D. W. Gill, B. B. Morgan, and P. G. W. Hawksby. The British
Coal Utilization Research Association, Leatherhead, England.
Baker, R. J., P. Hutchinson, E. E. Khalil, and J. H. Whitelaw, 1975. Measure-
ments of Three Velocity Components in a Model Furnace With and Without Com-
bustion. The 15th Symposium (International) on Combustion, The Combustion
Institute, Pittsburgh, PA. p. 553.
Berglund, R. N. and B. Y. H. Liu, 1973. Generation of Monodisperse Aerosol
Standards, Env. Sci. and Tech. 2» 147.
Biordi, J. C., C. P. Lazzara, and J. F. Papp, 1975. Flame Structure Studies,
of CH3Br—Inhibited Methane Flames. II. Kinetics and Mechanisms. The 15th
Symposium (International) on Combustion, The Combustion Institute, Pittsburgh,
PA. p. 917.
Cudahy, 1981. Incinerability, Thermal Oxidation Characteristics and Thermal
Oxidation Stability of RCRA Listed Hazardous Wastes. IT Env.iroscience Corp.
Dellinger, B., 1982. Personal Communication.
Del linger, B., D. S. Duval 1, D. L. Hall, and W. A. Rubey, 1982. Laboratory
Determinations of High Temperature Decomposition Behavior of Industrial Organic
Materials. 75th Annual Meeting of the APCA. New Orleans, LA.
Dietrich, V. E., 1979. Dropsize Distribution for Various Types of Nozzles.
In Proceedings of the 1st International Conference on Liquid Atomization and
Spray Systems. The Fuel Society of Japan, Tokyo, Japan, p. 69.
Duvall, D. S. and W. A. Rubey, 1976. Laboratory Evaluation of High-Temperature
Destruction of Kepone and Related Pesticides. Technical Report UDRI-TR-76-wl,
University of Dayton Research Institute. EPA 600/2-76-299.
Duvall, D. S. and W. A. Rubey, 1977. Laboratory Evaluation of High-Temperature
Destruction of Poly-chlorinated Biphenyls and Related Compounds. EPA 600/2
-77-228.
Glassman, I. and P. Yaccarino, 1981. The Temperature Effect in Sooting Dif-
fusion Flames, The 18th Symposium (International) on Combustion, The Combus-
tion Institute, Pittsburgh, PA. p. 1175.
Kramlich, J. C., G. S. Samuelsen, and W. R. Seeker, 1981. Carbonaceous Par-
ticulate Formation from Synthetic Fuel Droplets. Western States Section of
the Combustion Institute. Fall Meeting, Tempe, Arizona. WSS/CI-81-52.
Lee, K. C., J. L. Hansen, and D. C. Macauley, 1979. Predictive Model of the
Time/Temperature Requirements for Thermal Destruction of Dilute Organic
Vapors. 72nd Annual Meeting of the APCA. Cincinnati, OH.
6-1
-------
Lee, K. C., N. Morgan, J. L. Hansen, and G. M. Whipple, 1982. Revised Model
for the Prediction of the Time-Temperature Requirements for Thermal Destruc-
tion of Dilute Organic Vapors and its Usage for Predicting Compound Destruc-
tability. 75th Annual Meeting of the APCA, New Orleans, LA.
Parson, J. S. and S. Mitzner, 1975. Gas Chromatographic Method for Concen-
tration and Analysis of Industrial Organic Pollutants in Environmental Air
and Stacks. Env. Sci. Tech., 9^> 1053.
Seeker, W. R. and M. P. Heap, 1982. Flame Combustion Processes. Volume II
of Final Report for Contract EPA 68-Q2-2631.
Seeker, W. R., M. P. Heap, and T. J. Tyson, 1981a. Gas Phase Chemistry.
Volume I of Final Report for EPA 68-02-2631.
Seeker, W. R., G. S. Samuelsen, M. P. Heap, and J. D. Trolinger, 1981b. The
Thermal Decomposition of Pulverized Coal Particles. The 18th Symposium
(International) on Combustion. The Combustion Institute, Pittsburqh, PA.
pTTU:
Semenov, N. N., 1935. Chemical Kinetics and Chain Reactions, Clarendon Press,
Oxford, England.
Tsang, W. and W. Shaub, 1981. Chemical Processes in the Incineration of
Hazardous Waste. National Bureau of Standards. Paper presented to American
Chemical Society Symposium on Detoxification of Hazardous Wastes, New York,
NY.
6-2
-------
APPENDIX A
EXPERIMENTAL FLOW- CALIBRATIONS
In this Appendix the techniques used to monitor the reactant flows to
the microspray and turbulent flow experiments are discussed. In addition,
the flow calibration procedures are described.
A.l Microspray Reactor
Gas flows for the support flat-flame were monitored by Fisher-Porter
rotameters. Calibration and use of all rotameters was at a constant
40 psig pressure, as indicated by gauges mounted at each rotameter outlet.
Rotameters were initially calibrated according to the procedures pre-
sented in the vendor literature. The air rotameter was of the viscosity
compensated design, so calibration depends only on gas density. Air mass
flow was obtained by:
*= s-"w/10°
where m = actual mass flow of air (gm/sec)
S = rotameter scale reading (percent of maximum)
m = mass flow at full scale reading (gm/sec)
imgX
The maximum mass flow is calculated from the operating gas density and the
rotameter design parameters by:
%x = ""l v/(Pf " p2} p2/(pf ' Dl)pl (A"2)
where m^ = rotameter maximum mass flow for air at 70 F and 14.7 psia, a
published value (gm/sec).
pf = rotameter float density (for the stainless steel floats pf =
8.02 gm/cc).
c>2 = gas density at the operating condition (gm/cc).
p.| = gas density at the standard handbook condition (gm/cc).
Thus, for the air rotameter the calibration condition requires only a density
correction that can be calculated from the new pressure, temperature, and
A-l
-------
molecular weight. The air calibration was verified by use of a calibrated
dry test meter.
The nitrogen and hydrogen rotameters were of spherical float design so
that gas viscosity became a calibration variable. These calibration proced-
ures require the use of detailed correction charts and are relatively
involved. Details can be found in the vendor literature. The flow charac-
teristics were confirmed by bubble-flow meter measurements.
Liquid flow to the droplet generator was controlled by the pressure
placed on the fuel reservoir. This pressure forced liquid through the orifice
in the droplet generator. Thus, the pressure-flow relation can be expected
to be governed by Bernoulli's equation:
aP/p = V2/2 (A-3)
where aP = fluid gauge pressure
p = fluid density
V = fluid velocity at orifice
This can be arranged to yield
mL = CD2y^P (A-4)
where = liquid flow
C = system constant
0 = orifice diameter
Because of manufacturing variations in the orifices and natural wear, the
orifice diameter is not known well enough to directly use this equation.
Rather, liquid flow was directly measured by timed collection and weighing
and the relation was used to extrapolate between measured points and between
liquids of differing density.
Dispersion gas flow for the droplet generator was measured by a small,
in-line Dwyer rotameter.
A.2 Turbulent Flow Reactor
Burner air flow was measured by a Barco 3-in. ID venturi.
A-2
-------
A magnahelic gauge was used to indicate the pressure drop between the
venturi taps. The calibration constant was obtained from the basic
flow equations and was found to compare exactly with the manufacturers
calibration curve. The calibration was:
Qair = 44^" (A"5)
where Q -r = flow rate of burner air (standard cubic feet per minute).
&P = magnahelic reading (inches water).
A direct calibration checkwas notpossible because of the high air flows. How-
ever, an indirect check was performed through burner stoichiometry measure-
ments.
Burner fuel flow was set by the pressure on the fuel delivered to the
nozzle. The mechanism governing flow was the same as for the droplet generator:
pressurized fuel forced through an orifice. Thus, the controlling equation
is identical to equation A-4. Because of manufacturing variations, each nozzle
was calibrated at several pressures. The results were well described by Equa-
tion A-4, which was subsequently used to interpolate between calibrated points
and to extrapolate the calibration to new fuels.
A-3
-------
APPENDIX B
ANALYTICAL TECHNIQUE FOR
TEST COMPOUNDS
B.l Introduction:
The Tenax-GC sampling/analysis technique was selected for the measure-
ment of the test compound destruction efficiencies for the following reasons:
1. High Sensitivity: The minimum sensitivity of the technique
has been experimentally determined to be about 1.5 ppb, which cor-
responds to ^99.995% DE for the turbulent flow reactor at nominal
operating conditions. Direct grab sample/GC injection techniques
are simpler and are more rapidly performed but cannot measure to
99.99 percent DE.
2. Simple Operation/Rapid Turnaround: The principal competing tech-
nique, XAD-2, requires chemical extraction of the sample, a concen-
tration step for low concentration samples, and GC analysis. With
the Tenax-GC technique the sample cartridge is placed directly in
the GC, desorbed and analyzed in a single operation. Thus, sample
turnaround can be as low as one hour.
There are two principal limitations on the technique as it is presently
employed. First, the compounds can pass through the Tenax with only partial
adsorption (volumetric breakthrough). This is a particular problem for com-
pounds with molecular weights below 150. Since all the test compounds fall
below this molecular weight limit, breakthrough volumes were carefully checked
for each. In general, XAD-2 avoids this problem because of higher compound
specific breakthrough volumes. Second, the use of packed columns for separa-
lon of test compounds on the GC reduces the resolution of separation. The
typical peak width is such that the successful separation of more than five
compounds can be difficult due to peak interferences.
The technique, in summary, consists of the adsorption of test compound
from a known volume of sample gas onto cooled Tenax. At the conclusion of
sampling the Tenax cartridge is placed in the carrier gas line upstream of a
packed column. The cartridge is electrically heated and the trapped compounds
B-l
-------
are desorbed into the carrier gas. The compounds are trapped at the upstream
end of the room temperature packed column. After desorption is complete the
column is temperature programmed, and the compounds are separated and analyzed
by a flame ionization detector (FID). The system is calibrated in two ways.
First, dilute concentrations of test compounds in air are prepared in an 11-
liter dilution tank. This gas is sampled and analyzed by the Tenax technique
as if it were a normal sample. Second, direct liquid injections of test com-
pounds onto the GC column were performed with a microliter syringe; this by-
passed the Tenax. Comparison of the results of the two techniques indicates
the presence or absence of systematic calibration errors and indicates if
volumetric breakthrough has occurred. The occurrance of volumetric break-
through was further tested by placing two cartridges in series and determining
if any compound appeared on the second cartridge during adsorption.
The following section describes the analytical equipment, the calibration
tests, the operating procedures, and miscellaneous verification work.
B.2 Analytical Equipment:
Sampling and Adsorption: The equipment is illustrated in Figure B-l.
Exhaust gas samples were collected from the stack by an uncooled 6.35 mm OD
(0.25-m.) stainless steel probe. After leaving the stack the probe was heat-
taped and insulated to maintain a 200 C temperature. The sample passed through
a filter (47 mm Geman ) in a 200 C oven. The hot sample stream passed through
an externally water-cooled Tenax cartridge. The cartridge consists of a 120 mm
long by 10 mm ID Pyrex tube packed with 0.65 ± 0.02 grams of Tenax-GC (40-80
mesh). The adsorbant was held in place with small plugs of silianized glass
wool. The sample line was connected to the cartridge by Ultra-Torr fittings.
The entire cartridge was placed within a heat exchanger that flowed cooling
water around the cartridge. The sample subsequently passed through a gas dryer,
a rotameter, and a dry test meter. All connections upstream of the cartridge
were either 316 stainless steel or 6.35 mm 00 (0.25-in.) Teflon.
Desorption and Analysis: This system is based on a Perkin-Elmer Sigma-2
gas chromatograph with a Sigma-10 integrator/data station. The column is a
0.5-long, 3.18-mm OD (1/8-in.) Teflon tube packed with Porapak-Q. The 30 cc/
min. carrier gas flow is maintained through the Tenax cartridge, the column,
8-2
-------
Burner
///////; //////////-,
H
LmmJ
Oven (200 °C)
Probe
DO
I
CO
Electrically
Heated Sample To
Line Vent
r-T
'////////////////
Dry Test
Meter
II
Cooling Water
ill
K
y.w///f/\
I/.',- •:*:/:;
> / / / / / / / / /
'I'
Rotameter
<&
Tenax
Silica
Gas
Drier
Pump
Figure B-l. Sampling and adsorption system.
-------
and the FID, as shown in Figure B-2. The desorber is made from a 58-mm square
aluminum block. The dimensions are detailed in Figure 8-3. Eight 500-watt
electrical cartridge heaters are imbedded in the block. The voltage to these
is controlled by a variable transformer.
8.3 Operating Procedures:
Prior to use the Tenax cartridges are conditioned under a 20 cc/min.
helium flow at 200 C for 45-min. Both before and after sampling the tube ends
are covered with celophane and refrigerated.
After the reactor condition has been set the cartridge is placed in the
adsorption train, the cooling water started, and the sample flow opened. A
sample flow rate of 0.23 liters/min. has been found satisfactory as discussed
below. At the conclusion of sampling (2.3-1 iters or 10-min.) the cartridge is
removed and refrigerated.
For analysis, the cartridge is connected into the GC carrier gas line
and placed in the desorption block. The block is raised to 120 C and held
at this temperature for 5-min. This time and temperature have been found
sufficient to quantitatively desorb all of the test compounds. During
desorption the GC oven is maintained at room temperature. As the compounds
desorb, they are collected at the inlet of the GC column. The compounds do
not start to separate at room temperature. If this was not the case, peaks
would be broadened by the deposition of newly desorbed compound at the column
inlet after separation had started. At the conclusion of the 5-min. desorp-
tion time the column was heated to 120 C, held at this temperature for 25 min.,
and programmed* at 5 C/min. to 150 C when samples containing heptane or No. 2
fuel oil were analyzed; or to 180 C for all other cases. Under the present
analytical conditions, benzene and 1,2-dichloroethane have nearly identical
retention times; because of this, these two compounds were never used in the
same experiment. The FID gas flows were 71 cc/min. hydrogen and 442 cc/min.
air.
The integrator output record consists of a chromatogram detector signal
trace and a table of integrated peak areas. The trace is compared with the
tabulated output to insure that the compound peaks are free of interferances,
that the baseline boundary conditions were correctly constructed by the
B-4
-------
GC Oven
Porapak-W
Column
Desorption Block
FID or TCD
Helium Carrier
I JJHf/J/IIftfU/A
T. C. Readout
A. C. Power Variable Transformer
Figure B-2. Desorption and analysis equipment.
-------
SCALE: 0.75
13mm
70nun
75mm
20mm
Omn
158nm
Tubular Heater
Electrical Terminal
—\
Pixatron
Crew
Set Screw Hole
Thermocouple Hole
10mm
ca
i
cn
FRONT VIEW TOP VIEW
Vitron O-Rings
Reduce for Fitting
Tenax Cartridge
Welded
Ultra Torr Fitting
Figure B-3. Cartridge and cartridge heater detail.
-------
integrator, and that the presence of unanticipated peaks could be noted.
Integrated peak area values were converted into ppm by use of the calibration
curves and destruction efficiencies were calculated using the ppm anticipated
from zero destruction operation.
A potential problem with chromatogram interpretation was found in the
study. Under some conditions unidentified peaks were found in the chroma-
tograms in addition to those of the waste compounds. This raises the po-
tential that peaks could be misidentified. This problem was aggravated by
the fact that waste compound chromatographic retention times were dependent
on compound concentration. In response to this problem, a procedure for
interpreting chromatograms was developed.
During calibration a plot of liquid volume vs. retention time was pre-
pared for each compound except chlorobenzene. Chlorobenzene was omitted
because at its long retention time no potential interfering peaks ever
appeared. This plot is shown as Figure B-4; it illustrates that retention
time decreased with increasing compound concentrations. The following
procedure was used to identify the peaks.
• A tentative peak identity assignment was made on each of the
chromatograms.
• The tentative identity assignment was verified through comparison
of measured peak area and retention time with Figure B-4.
With this procedure, the mi sidentification of a peak was possible only if both
the area and the retention time of the unknown peak match the values shown in
Figure B-4. While misidentification error is still possible, the use of
Figure B-4 greatly reduces the likelihood of an actual occurrence.
In practice, few chromatograms caused any difficulty. Certain con-
ditions caused gross flame inefficiency, as indicated by the presence of
significant amounts of hydrocarbon fragments. These conditions were also
characterized by low DE. Because of the interference between the hydro-
carbons and the waste compounds, no quantitative results could be obtained.
An example of a chromatogram from which no quantitative data were obtained
is presented in Appendix D.
B-7
-------
30
25
20
Benzene
15
1,2 Oichloroethane
Chloroform
10
5
Acrylonitrile
0
0.8
0.4 0.6
Liquid Volume (microliters)
0
Figure B-^. Relationship between retention time and compound volume.
Retention time is based on the time when the column oven
first reaches 120°C.
B-3
-------
B.4 Calibration Techniques:
Test compound calibration was performed by three independent procedures.
This was necessary to determine the inherent scatter and reproducibility of
the measurement technique, and to locate any systematic errors that appeared
in any of the calibration tests. The three techniques were 1) direct syringe
injection of test compound onto the GC column; 2) syringe injection of liquid
test compound into a Tenax cartridge, followed by routine desorption and anal-
ysis; and 3) preparation of known standards in a dilution tank, followed by
routine Tenax sampling and analysis of the tank contents.
Direct Column Injection: Liquid samples were withdrawn with a calibrated
microliter syringe and directly injected onto the column through the injection
septa. Chromatograph conditions were identical to .the nominal operation pro-
cedure except:
1. The Tenax cartridge desorber was not a part of the system.
2. Rather than follow the prescribed oven temperature programing,
an isothermal temperature of 120 C was used (except for chloro-
benzene where 180 C was used). The isothermal temperatures
increased the speed at which calibrations were performed. The
programing is necessary during normal sampling to separate
light hydrocarbons from the earliest test compound peaks.
Because of the linearity of the FID amplifier, the change in
temperature programing does not affect the calibration.
Table B-l lists the FID mass sensitivity for the various compounds tested,
relative to methane.
Syringe Injection Onto Tenax: Liquid test compound samples were directly
injected onto packed Tenax cartridges. These were subsequently desorbed and
analyzed. The results in the present study were used as an initial, qualita-
tive test for breakthrough volume. After injection, helium was drawn through
the cartridge. For some tests a second cartridge was placed behind the first.
Breakthrough was detected by the appearance of test compound on the second
cartridge and, simultaneously, by the loss of response from the analysis of
the first cartridge.
B-9
-------
TABLE B-1. MASS SENSITIVITY RELATIVE TO METHANE FOR THE FID
Compound
Relative
Sensi-
tivity
Compound
Relative
Sensi-
tivity
Acrolein
0.54
Ethyl Acrylate
0.59
Phenol
1.22
Hexachlorobenzene
0.0
Benzene
1.23
Toluene
1.09
Carbon Disulfide
0.0
Vinyl Chloride
0.69
Acrylonitrile
0.59
Methyl Ethyl Ketone
0.68
1,2-Dichloroethane
0.32
Chlorobenzene
0.77
Chloroform
0.09
B-10
-------
Dilution Tank: Dilute samples were prepared by evacuating an 11-liter
glass tank and injecting a known amount of sample into the tank as the tank
was rapidly repressurized. After repressurization the tank was allowed to
equilibrate and the tank pressure and temperature were noted for calculation
of the correct dilution factor. A portion of the tank contents are pumped
through the Tenax sampling system. The remainder of the calibration test is
identical to the normal sampling and analytical procedure.
B.4 Calibrations
Figures B-5 through B-11 show the calibration plots for 1,2-dichloro-
ethane, benzene, chlorobenzene, chloroform, acrylonitrile, acrolein, and
heptane. These are plotted as microliters liquid vs. integrator peak area.
The two plots shown on each graph correspond to the direct injection calibra-
tion and the Tenax calibration. The close agreement between the two curves
for all compounds except acrolein and heptane indicates that breakthrough
volume was not exceeded for these compounds. This was confirmed by the
absence of measurable test compound on the second cartridge. Acrolein and
heptane both demonstrated significant breakthrough.
B.5 Optimization
The Tenax calibration apparatus was used to optimize the response with
respect to total sample flow rate. In addition, the breakthrough volume for
each of the test compounds was determined.
Flow Rate Optimization: Previous work with Tenax has indicated that
maximum breakthrough volume is a function of sample flow rate. Either exces-
sive or too-low flow can result in breakthrough. The experimental sample
flow rate was selected for the present study by varying sample flow for a
gas containing 45.8 ppm benzene and determining the fraction captured. The
results are shown in Figure B-12 and indicate that 0.19 to 0.28 liters/min.
sample flow optimizes capture.
Volumetric Breakthrough: The breakthrough of each compound was measured
for the standard 0.23 liters/mm. flow rate and 10-min. sampling period, and
for 3.7 liters total volume. These results are shown in Table B-2 and indi-
cate that only acrylonitrile and chloroform show significant breakthrough at
B-11
-------
1200]
A Direct Injection
O Tenax
7000
600
400-
200
0.1 0.2 0.3 0A 0.5 0.7 0.8 0.9 1.0 1.1
Volume (microliters)
Figure 8-5. Calibration curves for 1,2-dichloroethane.
B-12
-------
7—i—i—i—i—i—i—i—r
3000 -
S 2000
fO
i-
0>
c 1000 —
Direct Injection
,0 Tenax
J I I L
0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1.0
Volume (micro!itersX
Figure B-6. Calibration curves for benzene.
B-13
-------
A Direct Injection
O Tenax
0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1 .0
Volume (microliters)
Figure B-7. Calibration curves for chlorobenzene.
B-H
-------
500 ~
£ 400
<
s_
o
300 -
(O
S 200
c
T 1 1 r
-A Direct Injection
..O Tenax
100 -
J.
0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1.0
Volume (microliters)
Figure B-8. Calibration curves for chloroform.
B-15
-------
A Direct Injection
O Tenax
I 2000
-------
& Direct Injection
O Tenax
1500
S- 1000
o>
500
~cr
0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1.0
Volume (microliters)
Figure B-10. Calibration curves for acrolein.
B-17
-------
A Direct Injection
O Tenax
2000
<
S 1000
-------
10C-
90-
c
0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1.0
Flow Rate (Liters/min)
Figure B-12. Adsorption efficiency of a 45.8 ppm benzene
stream onto Tenax as a function of flow rate.
B-19
-------
TABLE B-2. VOLUMETRIC BREAKTHROUGH MEASUREMENT
Compound
Percent Breakthrough
V 3 3.7 Liters
V = 2.3 Liters
Acrylonitrile
17.4
0.4
Benzene
0.05
0.0
Chloroform
3.40
0.06
Chlorobenzene
0.0
0.0
1,2-Dichloroethane
0.05
0.0
Flow Rate =0.23 liters/min.
Concentration = 100 ppm
8-20
-------
3.7 liters, and at the normal 2.3-liter volume the maximum breakthrough for
acrylonitrile was an insignificant 0.4 percent.
Sample Storage; Each GC analysis required about 1.5 hours. Thus, a
backlog of unanalyzed samples occasionally accumulated. In no case was a
sample held longer than 24 hours before analysis. However, a series of tests
were performed to determine if 24 hours was a safe period for storage. A
standard gas was prepared and adsorbed onto two cartridges in parallel. The
first was analyzed immediately and the second was stored under nominal stor-
age conditions for 24 hours and analyzed. The results, shown in Table B-3,
indicate that no significant loss of test compound occurred. The scatter
between the two analyses is typical of the measurement technique.
B.6 Uncertainty, Accuracy, and Precision
The problem of sample repeatability and precision really involves two
questions:
• What is the repeatability of the analytical technique given a time-
steady, known concentration to measure?
• What is the time-steadiness of the experiment, assuming a perfect
measurement technique?
The first of these questions was addressed by the repeated analysis of known
calibration standards. An example of such a series is shown in Table B-4
for benzene. The resulting standard deviation of the relative error is 2.1
percent. Thus, for the measurements to be accepted at the 90 percent confi-
dence interval, the relative error is approximately + 4.2 percent. The
error for all calibration data as a group indicated an approximate +5.0 per-
cent at the 90 percent confidence interval for the Tenax procedure.
The data indicate that the inherent time unsteadiness of the experimental
DE measurements was small under nonoptimum conditions and substantial under
optimum conditions. For the optimum conditions the observed time unsteadi-
ness exceeded the +5.0 percent uncertainty associated with the analytical
system and, thus, led to the conclusion that optimum DE measurements and
rankings were random, time unsteady, and were probably related to the sta-
tistical nature of a turbulent flame. However, none of these optimum data
B-21
-------
TABLE B-3. SAMPLE STORAGE STA8TLITY
Analysis
Compound
After 24
Immediate
{ppm)
Hours (ppm)
Acrylonitrile
16.7
18.2
Chloroform
16.2
14.5
1,2-Dichloroethane
16.2
16.9
Chlorobenzene
8.77
9.5
B-22
-------
Table B-4 Calibration Repeatability Data for Benzene
Response (peak area)
Relative
Error
Expected
Obtai ned
1141
1101
.0350
1141
1126
.0131
1470
1399
.0483
1470
1451
.0129
2129
2090
.0183
2129
2138
-.0042
3118
3146
-.00898
3118
3153
-.0112
3118
3136
-.0058
B-23
-------
were used to establish the ranking presented in this study.
For nonoptimum conditions, the OE measurements and rankings were
repeatable. One example of a direct repeat is found in Figure 1-4. The
two data sets shown in the neighborhood of 150 percent theoretical air are a
direct repeat at identical experimental conditions. They were separated by
a small amount during plotting for clarity. This repeat demonstrates that
nonoptimum rankings were repeatable.
A further means of evaluating the repeatability of nonoptimum rankings
is to compare the rankings for different data points within identical failure
condition. For example, Figure 1-6 shows data for off-design nozzle operation.
Both sets of rankings (210 and 315 percent theoretical air) show the same
ranking for this failure condition. Figure 1-7 demonstrates another example
in which the rankings from two individual data sets agree for the failure
condition caused by the quench coil.
It was recognized that an unqualified presentation of a ranking list
was unsatisfactory in that it fails to delineate the quantitative "closeness"
of compounds ranked consecutively. Thus, the rankings were summarized in bar
graph form, of which Figure 1-8 is an example. These figures give a quantita-
tive indication of both the ranking order, and the relative spread separating
the compounds. The analytical uncertainty in the ranking figures has been
found to be +5 percent. The repeatability of the experimental conditions
was not statistically established, although the limited direct repeat data
indicate that approximately a + 10 percent confidence interval can be placed
on each of the bar graphs.
B-24
-------
APPENDIX C
THERMAL DECOMPOSITION MODEL
The model was designed to account for behavior in the Microspray Reactor
that resulted from the simultaneous evaporation of test compound droplets,
and the thermal destruction of the vapor.
Droplet evaporation is governed by the rate of heat transfer to the drop-
let. The rate of heat flow to a sphere is:
is the droplet diameter, cm.
T» = free stream temperature specified from microspray thermo-
couple measurements, K.
Tgp = boiling point of the droplet liquid.
Thus, the evaporation rate for an individual droplet is Equation (A-l) divi
ded by the latent heat of vaporization, and the evaporation rate per unit
volume can be obtained by multiplying by the number of droplets per cubic
centimeter:
H = latent heat of vaporization, cal/gm.
The heat transfer coefficient is related to the droplet Nusselt number by:
"" hAg (Tod ™ ^"gp)
(A-l)
where q = heat flow, cal/sec,
2
h = heat transfer coefficient, cal/cm -sec-K
2
Aq = droplet surface area = -n-Dp /4 where D^
(A-2)
3
where Q = the evaporation rate, gm/cm - sec
3
n = droplets/cm
h = NukH/DQ
(A—3)
C-l
-------
where Nu = Droplet Nusselt number S 2
Kh = gas film heat transfer coefficient, cal/cm-sec-K.
This is substituted into Equation (A-4) to yield:
q = r"CH"°D ' V (A—4)
The reaction equation is essentially a conservation equation for vapor
phase compound that relates the source term from vaporization and the sink
term due to reaction to the change in concentration as the gas moves through
the burner. The equation is
= q - kc {A-5)
where V = gas velocity, cm/sec
3
C = vapor concentration of compound, gm/cm
k = reaction rate coefficient, sec~\
Substituting Equation (A-4) into (A-5) and dividing by V, the equation can be
cast into the functional form:
§ • F(C,x) (A-6)
A companion equation for droplet evaporation can be obtained from (A-4):
dC,
Tx Q a G(x) (A~7>
where C-j = the equivalent concentration of unevaporated material,
gm/cm^,
F,G are the functional forms defined by Equations (A-5) and
(A-4), respectively.
Equations (A-6) and (A-7) are numerically integrated subject to the follow-
ing constraints:
1. Initial conditions: C(0) =0
C-|(0) = equivalent inlet spray concentration.
C-2
-------
2. Constants: n = estimated from experimentally observed
spray shape.
H, Tgp, V are assumed constant.
To> = a known function of X determined from
thermocouple measurements in the absence of
spray.
k = evaluated from thermal decomposition data and
the local value of T®.
4. When C-j reached zero (all liquid evaporated) the problem was
reduced to integration of (A-5) with Q = 0.
The specific model output was a profile of fraction feed evaporated and frac
tion feed reacted as a function of X. The model was used to examine the
effect of droplet diameter, temperature profile, and compound type on stack
gas concentration. An example of output is shown in Figure C-1. The follow'
ing points are noted:
• The droplets are completely evaporated by one-cm into the reactor.
• Because of the declining temperature the reaction essentially
ceases after about 20 cm so that the reaction extent is unchanged
over the final 80 cm of the reactor.
• The exit concentration (at 100 cm = 0.77) was identified by the
entrance plane or gas flame temperature (900 K) when these
results were plotted as a single point on Figure 3-5.
3. Droplet diameter: This is related to C, through:
C-3
-------
900
Temperature
800
700
1.0
T
0.8
Fraction
Unreacted
0.6
0.4
Fraction
Unevaporated
0.2
0
100
Distance (cm)
Figure C-l. Typical model output.
C-4
-------
APPENDIX D
RAW DATA
This section presents the raw test compound data and describes the
procedure for converting the raw data into destruction efficiencies for the
two experiments.
D.l Sample Chromatoqrams
A selection of chromatograms typical of various microspray and turbulent
slow reactor conditions are presented below. Figure D-l shows a relatively
high efficiency microspray condition with few identifiable peaks. The numbers
printed next to each peak are the retention times in minutes. A typical mod-
erate efficiency microspray condition is shown in Figure D-2. A chromatogram
typical of high-efficiency turbulent flow reactor operation is shown in Fig-
ure D-3. Few peaks are present and.only one peak, associated with a fuel
fragment at 45.18 min. is measurable. Figures D-4 and D-5 show chromatograms
for moderate DE heptane fueled turbulent flow reactor data in which the test
compound mixture contained benzene and 1,2-dichloroethane, respectively. A
low-DE turbulent flow chromatogram is shown in Figure D-6. At very low effi-
ciency operating conditions significant quantities of fuel and fuel frag-
ments are released by the flame in addition to the test compounds. These
conditions can result in an uninterpretable chromatogram, as shown in Fig-
ure D-7.
D.2 Calculation Procedures
Chromatogram peak areas were related to moles for each compound by use
of the calibration figures in Appendix B. This number is converted into
mole fraction based on dry gas by:
Mole Fraction (dry) =
-------
0-4
>»
4
3
If
K
a
cj
£
*
K
*
*
*
s
a
*
«
«
*
*
IT
-------
a
O
O
-------
a*
Figure D-3. Chromatogram for a high-DE turbulent flow reactor condition.
-------
c:
in
lf>
m
Figure D-4. Chromatogram for moderate-DE heptane-fueled turbulent
flow reactor condition.
-------
&
Iff
•
av
c
CO
Figure 0-5.
Chromatogram for moderate-DE No. 2 oil-fired
turbulent flow reactor condition.
-------
t*
a
c
u
CO
CM
n ~
CI
CO
Figure D-6. Chromatogram showing a poor-DE turbulent flow
reactor condition.
-------
O H
Figure D-7. Chromatogram showing very low efficiency turbulent flow reactor
operation with significant fuel fragments present.
-------
complete combustion model and the dry mole fractions are corrected to burner
(wet) mole fractions by:
Burner Mole Fraction = (Dry Mole Fraction) - (1 - Water Mole Fraction) (D.2)
The mole fraction of compound that would be present at the burner exit if
efficiency were zero is calculated by:
Zero Efficiency _ (Fuel Flow) (Moles Compound/Mass Fuel)
Mole Fraction ~ Total Reactor Molar Flow w-j)
The (Moles Compound/Mass Fuel) is calculated from the fuel mixture composi-
tion and the (Total Reactor Molar Flow) is calculated from the fuel flow, the
air flow, and the complete combustion model. The fraction unreacted compound
is the ratio of Equation D-2 and D-3.
D-9
-------
|