950R80042
U.S. Environmental Electric Power IERL-RTP-1085
Protection Agency Research Institute October 1980
Proceedings of the Joint
Symposium on Stationary
Combustion NOx Control
Volume III
NOx Control and Environmental Assessment
of Industrial Process Equipment, Engines,
and Small Stationary Sources
(
HUMMUS A ¦ ' • v, ¦
)EPFS
J EPR1
-------
RESEARCH REPORTING SERIES
Research reports of the Office of Research and Development, U S. Environmental
Protection Agency, have been grouped into nine series These nine broad cate-
gories were established to facilitate further development and application of en-
vironmental technology. Elimination of traditional grouping was consciously
planned to foster technology transfer and a maximum interface in related fields.
The nine series are:
1. Environmental Health Effects Research
2. Environmental Protection Technology
3. Ecological Research
4. Environmental Monitoring
5. Socioeconomic Environmental Studies
6. Scientific and Technical Assessment Reports (STAR)
7. Interagency Energy-Environment Research and Development
8. "Special" Reports
9. Miscellaneous Reports
This report has been assigned to the MISCELLANEOUS
REPORTS series. This series is reserved for reports whose
content does not fit into one of the other specific series.
Conference proceedings, annual reports, and bibliographies
are examples of miscellaneous reports.
EPA REVIEW NOTICE
This report has been reviewed by the U.S. Environmental
Protection Agency, and approved for publication. Approval
does not signify that the contents necessarily reflect the
views and policy of the Agency, nor does mention of trade
names or commercial products constitute endorsement or
recommendation for use.
-------
IERL Rip 1085
October 1980
Proceedings of the Joint
Symposium on Stationary
Combustion NOx Control
Volume III
NOx Control and Environmental Assessment
of Industrial Process Equipment, Engines,
and Small Stationary Sources
Symposium Cochairmen
Robert E. Hail, EPA
and
J. Edward Cichanowicz, EPRI
Program Element No. N130
Prepared for
U.S. ENVIRONMENTAL PROTECTION AGENCY
Office of Research and Development
Washington, D.C. 20460
and
ELECTRIC POWER RESEARCH INSTITUTE
3412 Hillview Avenue
Palo Alto, California 94303
-------
PREFACE
These proceedings document more than 50 presentations given at the
Joint Symposium on Stationary Combustion N0X Control held October 6-9,
1980 at the Stouffer's Denver Inn in Denver, Colorado. The symposium was
sponsored by the Combustion Research Branch of the Environmental
Protection Agency's (EPA) Industrial Environmental Research
Laboratory-Research Triangle Park and the Electric Power Research
Institute (EPRI). The presentations emphasized recent developments in
N0X control technology. Cochairmen of the symposium were Robert E.
Hall, EPA, and J. Edward Cichanowicz, EPRI. Introductory remarks were
made by Kurt E. Yeager, Director, Coal Combustion Systems Division, EPRI,
and the welcoming address was given by Roger L. Williams, Regional
Administrator, EPA Region VIII. Stephen J. Gage, Assistant Administrator
for Research and Development, EPA, was the keynote speaker. The symposium
had 11 sessions:
I: N0X Emissions Issues
Michael J. Miller, EPRI, Session Chairman
II: Manufacturers Update of Commercially Available Combustion
Technology
Joshua S. Bowen, EPA, Session Chairman
III: N0X Emissions Characterization of Full Scale Utility
Powerplants
David G. Lachapelle, EPA, Session Chairman
IV: Low N0X Combustion Development
Michael W. McElroy, EPRI, Session Chairman
Va: Postcombustion N0X Control
George P. Green, Public Service Company of Colorado,
Session Chairman
Vb: Fundamental Combustion Research
Tom W. Lester, EPA, Session Chairman
VI: Status of Flue Gas Treatment for Coal-Fired Boilers
Dan V. Giovanni, EPRI, Session Chairman
VII: Small Industrial, Commercial, and Residential Systems
Robert E. Hall, EPA, Session Chairman
VIII: Large Industrial Boilers
J. David Mobley, EPA, Session Chairman
IX: Environmental Assessment
Robert P. Hangebrauck, EPA, Session Chairman
X: Stationary Engines and Industrial Process Combustion Systems
John H. Wasser, EPA, Session Chairman
XI: Advanced Processes
G. Blair Martin, EPA, Session Chairman
ii
-------
VOLUME III
TABLE OF CONTENTS
Session VII: Small Industrial, Commercial, and Residential
Systems
Page
Session VII: Small Industrial, Commercial, and Residential
Systems
"Evaluation of Emissions and Control Technology for
Industrial Stoker Boilers," R. D. Gianmar, R. H. Barnes,
D. R. Hopper, P. R. Webb, and A. E. Weller 1
"Control of Emissions from Residential Wood Combustion by
Combustion Modification," J. M. Allen 39
"Field Tests of Eleven Stoker Coal-Fired Boilers for
Emissions Control and Improved Efficiency,"
P. L. Langsjoen 64
Session VIII: Large Industrial Boilers
"Combustion Modification for Coal-Fired Stoker Boilers,"
K. L. Maloney, K. F. Maloney, and M. J. Pfefferle 83
"Thirty-Day Field Tests of Industrial Boiler Combustion
Modifications," W. A. Carter 99
"Update of NSPS for Industrial Boilers," L. G. Jones .... *
Session IX: Environmental Assessment
"Conventional Combustion Environmental Assessment
Program," W. H. Ponder 118
"Combustion Modification Environmental Assessment,"
C. Castaldini, R. M. Evans, E. B. Higginbotham, K. J. Lim,
H. B. Mason, and L. R. Water land 147
"Utility Boiler Environmental Assessment," R. M. Perhac ... *
*See Volume V, Addendum.
iii
-------
Session X: Stationary Engines and Industrial Process Combustion
Systems
"Characterization and Oxidation of Diesel Particulate,"
D. A. Trayser, L. J. Hillenbrand, M. J. Murphy,
J. R. Longanbach, and A. Levy 188
"Emission Control Methods for Stationary Large-Bore
Engines," R. P. Wilson, Jr *
"Stationary Diesel Emissions — A Comprehensive Analysis,"
J. H. Wasser *
"Emission Reduction by Combustion Modification for
Petroleum Process Heaters," W. A. Carter *
"Kinetics and Mixing in Industrial Afterburners," A. Levy,
A. A. Putnam, H. A. Arbib, and R. H. Barnes 225
"Subscale Tests of Combustion Modification for Steel
Furnaces," R. J. Tidona, W. A. Carter, and S. C. Hunter . . . 274
*See.Volume V, Addendum.
iv
-------
EVALUATION OF EMISSIONS AND CONTROL
TECHNOLOGY FOR INDUSTRIAL STOKER BOILERS
By;
R. D. Grammar, R. H. Barnes, D. R. Hopper,
P. R. Webb, and A. E. Weller
BATTELLE
Columbus Laboratories
505 King Avenue
Columbus, Ohio 43201
1
-------
ABSTRACT
This paper presents the results of a 3-phase program to
evaluate emissions and control technology for industrial stoker boilers.
The paper focuses on the third phase "Limestone/Coal Pellet Development",
while summaries are given of the first two phases, "Alternate Fuels
Evaluation" and "Control Technology Evaluation". Because SO2 appears
to be the most troublesome emission to control for stokers, a limestone/
high sulfur coal pellet was developed and evaluated as a SO2 control
technique. Initially, this pellet with a Ca/S molar ratio of 7 was
successfully fired in a 8 MWth industrial spreader-stoker boiler with
SO2 emissions reduced by 75 percent. However, from both an economical
and operational standpoint, the amount of limestone required had to be
reduced to correspond to Ca/S molar ratio of 3 to 4. Furthermore, the
mechanical properties of this pellet were inadequate to withstand the
severe stresses of an industrial fuel-handling system. Accordingly,
an R&D effort was undertaken to refine the pellet. A refined pellet,
with a Ca/S molar ratio of 3-1/2 with appropriate binders was produced
that had similar or improved physical characteristics of raw coals.
Additionally, economic analysis indicates that this pellet can be pro-
duced for approximately $15/ton above the cost of the high sulfur coal.
This refined pellet was fired in a 200 kW^h laboratory spreader-stoker
boiler achieving sulfur captures as high as 70 percent. However, when
fired in the 8 MWth (25,000 lb steam/hr) stoker boiler, sulfur captures
on the order of 50 percent were achieved.
2
-------
ACKNOWLEDGMENT
The research covered in this report was pursuant to
Contract No. 68-02-2627 with the U.S. Environmental Protection
Agency, Combustion Research Section. The authors wish to express
their appreciation for the assistance and direction given the
program by project monitor John H. Wasser.
We would also like to acknowledge Harold Johnson of
Detroit Stoker, William Engelleitner of Mars-Mineral, Sam Spector
of Banner Industries, and Donald Hansen of Alley-Cassetty Coal
Company for providing advice and assistance to the program.
Finally, we would like to recognize Battelle-Columbus
staff members—John Faught, Tom Lyons, Paul Strup, Don Hupp,
Luis Kahn, and Andrew Skidmore, and acknowledge the cooperation of
John Clayton and his Facilities staff who allowed us to use the
Battelle steam plant boiler during the program.
3
-------
SECTION I
INTRODUCTION
The coal-fired stoker boiler provides an option for industry
to meet its energy needs. This option has not been exercised by a
significant number of industries primarily because oil- and gas-fired
equipment have been, and still are, more environmentally and economically
attractive. However, with the dwindling supplies of oil and gas, the
rising costs of these fuels, and increased attention given to coal
utilization, industry once again is considering the coal-fired stoker
boiler.
In support of our nation's commitments to maintain a clean
environment and to utilize coal, EPA funded a research and development
program to identify and demonstrate improvements in stoker-coal firing
that can provide an incentive for greater industrial use of coal. The
overall objectives of this program were to
• Characterize the spectrum of emissions
from industrial coal-fired stoker boilers
using several types of coal under various
stoker-firing conditions
• Investigate control methods to reduce these
emissions
• Determine the effect of these control methods
and variations in stoker-boiler operation on
the overall performance of the stoker boiler,
and,
• Assess the environmental impact of new
technology on the future acceptability of
stoker boilers.
This program was recently completed and the final report should be
available soon.
4
-------
This program was divided into three phases. In Phase I,
Alternative Fuels Evaluation, emission characteristics were determined
for a variety of coals fired in a 200-kWt stoker boiler. Emphasis was
focused on identifying coals with low pollutant potential, including
both physically and chemically treated coals. The results of this
phase were presented at the Second Stationary Source Combustion
Symposium and contained in its Proceedings. In Phase II, Control
Technology Evaluation, potential concepts for control of emissions for
full-scale industrial stokers were evaluated. Similarly the results
of this phase were presented at the Third Stationary Source Combustion
Symposium and contained in its Proceedings. In Phase III, Limestone/
Coal Pellet Development, a limestone/coal fuel pellet was developed and
evaluated as to its viability as an SO2 control for industrial stoker
boilers. This paper focuses on that effort and summarizes those results.
The executive summaries of Phases I and II of the final report of this
program are also included in this paper.
5
-------
SECTION II
PHASE I. ALTERNATE FUELS EVALUATION
EXECUTIVE SUMMARY
A 200-kW stoker-boiler facility was used to evaluate
characteristics of emissions from combustion of a variety of coals,
including coals that could not be conveniently or economically
evaluated in larger industrial systems. The stoker was initially
operated in an underfeed mode to expand the data base developed in
an earlier EPA program (1). This facility was modified to accommodate
a model spreader stoker more typical of an industrial boiler.
Raw coals with low pollution potential and treated coals were
evaluated. Because there was only one treated coal available during the
time framework of the program, Battelle developed, as part of this pro-
gram, a limestone/high sulfur coal fuel pellet.
Results of the Phase I emission characterization were as
follows.
NO
For the underfeed stoker, less than 10 percent of the fuel
nitrogen was converted to NO, assuming no thermal NO. For the model
spreader-stoker, between 10 and 20 percent of the fuel nitrogen was con-
verted to NO.
Coals naturally high in calcium and sodium and those treated
with these elements retained significant percentages of the sulfur
in the ash. For the eastern bituminous coals, with relatively small
amounts of calcium and sodium but significant amounts of iron, sulfur
retention in the ash was as high as 20 percent. It was observed that
bed temperatures in these laboratory stokers are significantly lower
than those measured in an industrial stoker.
6
-------
CO
CO levels can be controlled by the use of overfire air and
were generally less than 100 ppm.
Particulate Loading
Particulate loadings did not correlate consistently with either
the ash content of the coal nor its size consist prior to feeding. It
appears that the friability and inherent moisture content of the coal may
affect particulate loading since these properties influence the amount of
fines generated.
POM Loadings
POM loadings for continuous operation of the underfeed stoker
were significantly less than those reported earlier (1) for intermittent
operation.
Particle-Size Distribution
For the model spreader, the average stack particle size ranged
between 15 and 30 micrometers.
Treated Coals
No commercially available, chemically treated coals were identified.
Treated coals required pelletization for firing in stokers.
The Battelle Hydrothermally Treated (HTT) coal was available for
laboratory evaluation. The treatment reduced the fuel sulfur from 2.6
percent to 1.1 percent. Because of the relatively high calcium and sodium
residual from the treatment, only 28 percent of the remaining sulfur was
emitted as SO^.
Also, the limestone/coal fuel pellet, with a Ca/S molar ratio of
7, reduced SO2 emissions by over 70 percent. Even at the elevated fuel-bed
temperatures (> 1100 C), the calcium reacts with the coal sulfur and retains
it as a sulfide/sulfate as part of the fuel ash.
7
-------
SECTION III
PHASE II. CONTROL TECHNOLOGY EVALUATION
EXECUTIVE SUMMARY
Potential control concepts were identified and evaluated in
the Battelle 8 MW , (25,000 lb steam/hr) spreader stoker boiler. Con-
tn
trol strategies were limited to:
• Use of compliance coals
• Combustion-system operational modifications
• Minor combustion-system design modification
• Use of treated coal (limestone/coal fuel pellet).
Flue-gas clean-up techniques were not considered. Criteria pollutants
were used as the basis for evaluation.
The Phase II experiments have demonstrated that emission levels
can be reduced by proper control of the stoker operating variables. In
addition, the limestone/coal pellets have been demonstrated to offer
potential for SO2 control. In summary, the major findings are:
• The limestone/high-sulfur coal pellet showed a
sulfur capture of about 75 percent for a Ca/S
molar ratio of 7.
• Sulfur capture efficiencies of around 25 percent
were noticed with Bome eastern bituminous coals.
• High excess air rates at low loads resulted in
increased sulfur retention in the bed ash.
» CO and smoke levels were controlled by providing
adequate excess air. CO levels were low for all
fuels tested, except the limestone/coal fuel pellet.
• Clinker formation may be a limiting factor in
determining the minimum excess air rate.
• NO levels increased slightly with increased
excess air.
8
-------
• An increase in overfire air/total air flow rate
ratio reduced CO and smoke, the latter more
significantly. Particulate loadings were also
reduced with increased overfire air.
• NO was lower for inactive overfire air jets
• Clinker formation occurred readily if bed
depths became excessive, while the danger
of burning the grates existed for operation
with very shallow beds. Bed depths around
6.3 to 7.6 cm appeared to be optimum for
lower ash coals.
3
• POM levels ranged from 13 to 24 yg/Nm .
They were somewhat lower than those of the model
spreader and only slightly higher than those
from a 500 kW^ packaged boiler firing natural
gas and fuel oil (2).
o A higher excess air rate was required for low-
load than for partial- or full-load operation.
Additionally, a greater percentage of overfire
air was required at low load. Low-load smoke
can be reduced by a reduction in underfire air,
coupled with attentive boiler operation.
• At full load fly-ash reinjection increase
boiler efficiency by 1.5 percent. However,
particulate loadings were reduced by 10 to 25
percent by operating without fly-ash reinjection.
• The high-sulfur Ohio coals had to be fired at
higher excess air rates than did the low-
sulfur Ohio and Kentucky coals. The high-ash
unwashed stoker coal and high-moisture Illinois
No. 6 coal could not be fired satisfactorily.
• Improvements in the coal feed system will
increase control of fuel distribution within
the stoker boiler and thus offer an improved
control of the combustion process. Coal particle
segragation by the transport system was signi-
ficant in the 8 MWth stoker boiler.
• Improved grate design to minimize leakage and to
provide more uniform air distribution would also
improve the control of combustion process and
reduce emissions and increase boiler efficiency.
• Improved overfire air systems to increase the
effectiveness of aerodynamic mixing above the
bed is one of the more promising combustion
modification techniques to reduce emissions
and improve performance.
9
-------
SECTION IV
PHASE III. LIMESTONE/COAL PELLET DEVELOPMENT
EXECUTIVE SUMMARY
The Phase III program focused on refinement of the limestone/
coal fuel pellet and evaluation of its suitability as an industrial
stoker-boiler fuel. This program consisted of four major tasks:
1. Pellet Development aimed at developing a fuel pellet
with mechanical strength characteristics that can
withstand weathering and the severe stresses of an
industrial stoker coal-handling and feeding system,
burns at reasonable rates, and captures sufficient
sulfur to be competitive with other control
strategies. Mechanical strength characteristics
were evaluated with standard laboratory tests.
Burning characteristics and sulfur capture were
determined in a fixed-bed reactor simulating the
fuel bed of a spreader stoker.
2. Process Variables Selection combined a mathematical
model analysis with a series of experimental studies
to develop a more comprehensive understanding of
the processes that influence the combustion of the
fuel pellet and control the capture of sulfur.
3. Laboratory Evaluations conducted in both the 200 kWt
model-spreader stoker and the 8 MWt Battelle steam
plant boiler to evaluate the most promising candidate
pellets.
4. Economic Analysis aimed at developing pellet process
costs.
10
-------
The major results and conclusions of the four tasks are:
Pellet Development
• A fuel pellet was produced that, according to
laboratory tests, has mechanical strength and
durability characteristics similar to those of
conventional coals.
• Pellets produced by auger extrusion or pellet mill
processes had better mechanical strength than those
produced by disc pelleting or briquetting.
• Binders that provide some resistance to the weather
were identified. However, no binder was identified
that provided complete weather proofing.
• The fixed-bed reactor experiments indicated a weak
dependency between Ca/S ratio and sulfur capture
for Ca/S ratios above 2.
• Calcium oxide is a superior absorbent to limestone,
but is not economically competitive with limestone.
• Additives do not appear to enhance sulfur capture.
Process Variables Selection
• The mathematical model predicts an optimum coal size
(35-40 mm diameter) for maximum sulfur retention.
• The model indicates a weak dependency on the calcium/
sulfur ratio.
• Scanning electron microscopy and x-ray diffusion are
powerful tools for the study of solid-state reactions
in the pellets. Results indicate that sulfur is
retained predominantly as CaSO^.
• Sulfur may react directly with limestone by solid-
state processes without involving the formation of
S02.
Laboratory Evaluations
Auger-extruded and milled pellets burned better than
briquets and disc-agglomerated pellets.
s Sulfur capture of about 65 percent was achieved at
Ca/S molar ratios of 3.5.
11
-------
• Sulfur capture of about 50 percent was achieved in
the steam-plant stoker. In comparison to the model
spreader, this lower SO2 capture was attributed to
higher temperatures (in excess of 1300 C).
• In the Battelle steam power plant, the fuel pellets
burned as well as low-sulfur coal.
Economic Analysis
• It is estimated that limestone/coal fuel pellets
can be produced for about $15.40/Mg ($14/ton) of
pellets above the costs of the high-sulfur coal.
12
-------
SECTION V
BACKGROUND
During Phase II of this program, Control Technology Evaluation,
SC>2 was considered as the most troublesome emission to control in an
industrial stoker-boiler system. The other criteria emissions could be
controlled with existing technology or were within the current emissions
requirements. As a result, and because the limes tone/coal fuel pellet
offers a means for environmentally acceptable burning of high-sulfur coal
in existing boilers, EPA continued the development of the fuel pellet as
part of this program.
From an industrial point of view, the possibilities of using
limes tone/coal pellets for removing SC^ in situ via a dry process is
more acceptable than the use of scrubbers. The additional costs for
pelletizing the coal/limestone mixture and for the removal of 3 to 4 times
as much ash is far more attractive than the high cost of operating and
maintaining wet scrubbers.
13
-------
SECTION VI
PELLET DEVELOPMENT
The fuel pellets used in earlier studies did not have adequate
strength or durability as up to 50 percent fines were introduced into the
8MWth (25,000 lb steam/hr) boiler. Furthermore, from both an economical
and operational standpoint, the amount of limestone required to capture
a target goal of 70 percent fuel sulfur had to be reduced to correspond
to a Ca/S molar ratio of 3 to 4. As a result, an extensive effort
(166 test samples) was made to investigate:
• pellet production techniques
• binder types
• coal and limestone particle sites
• limestone types
• pellet formulations.
Laboratory test procedures were developed to evaluate the effect of these
variables on the mechanical strength properties of the fuel pellets.
Illinois No. 6 coal was used as the base coal. This coal was
ground to 100 percent through 20-mesh and 50 percent through 100-mesh.
Ground limestone (-50 mesh) was added to the coal with the selected binders
and thoroughly mixed. This mixture was fed to a pellet mill to produce
cylindrical pellets one-half inch in diameter and about three-fourth inch
long.
Table I indicates that pellets were produced with mechanical
strength, durability and weatherability characteristics similar to those
of raw coals based upon the laboratory test procedures. A number of
formulations were identified that could produce satisfactory pellets.
The specific formulation used will depend on economics and availability.
The reader is referred to the final report for detailed
discussion.
14
-------
SECTION VII
PROCESS VARIABLES SELECTION
A mathematical modeling analysis, in combination with a series
of experimental studies, was performed to develop a more comprehensive
understanding of the processes that influence the burning of the coal/
limestone pellets and control sulfur capture. The purpose of the
experimental studies was to provide physical and chemical rate constants
for the model, and to provide a basis for assessing model reliability.
The main objectives of the modeling studies were to predict burning
rates and sulfur capture as a function of pellet properties such as
size, composition, and physical structure under different combustion
conditions. Knowledge of how these parameters affect pellet performance
is important for the development of optimum pellets for specific boiler
applications.
The mathematical model described here is a preliminary model
with a number of simplifying assumptions incorporated into it. An
important purpose of this effort was to demonstrate how modeling can be
practically and usefully employed to understand how coal/limestone
pellets behave and how to improve their performance. The basic structure
of this preliminary model can be expanded to include either additional
processes or refined mechanisms to increase its accuracy.
In the model, the pellet's cylindrical geometry is represented
by an equivalent sphere having the same surface-to-volume ratio. The
burning of the pellet is represented by the shrinking core model developed
bv Wen and co-workers (3-6). It is based on a burning reaction
zone that penetrates into the pellet and is supported by oxygen
that diffuses through the ash layer surrounding the burning core. SO^,
formed at the burning interface between the unburned core and ash, is
assumed to diffuse out through the ash layer where it is captured by the
CaCO^ in the limestone. The calcination process is represented by the
reaction
15
-------
CaCO^ CaO + CO2
and sulfur capture by the reaction
CaO + S02 + 1/2 02 CaSO^ .
S(>2 that does not react with the CaO while diffusing to the outer surface
of the ash layer is considered to be released from the pellet. The
model, in its present form, neglects factors such as heat transfer,
devolatilization of the coal, the reaction between CC>2 and carbon forming
CO, and solid-state reactions. These mechanisms can be incorporated into
extended versions of the present model as appropriate.
The experimental studies associated with the modeling effort
were performed using the fixed-bed reactor described in detail in the final
report. These experiments involved heating pellets in the fixed-bed
reactor in a flowing air stream for various periods of time corresponding
to different levels of fuel consumption in the pellets. The combustion
gases exiting from the reactor were analyzed continuously for SO^, CO,
and 0^ to follow the kinetic rates of the combustion and S02~release
processes. Burned pellets were recovered, weighed, analyzed chemically,
and subjected to examination by metallographic techniques, scanning
electron microscopy, and X-ray diffraction. Some limitations, however,
exist in accurately applying some of these data to the model because of
uncertainties in the fixed-bed reactor experiments. These difficulties
can be eliminated in future work by using the model developed here as
a guide for designing the experiments.
The results of this effort are summarized in the Executive
Summary. Further details on both the fixed-bed reactor experiments and
the model are presented in the final report.
16
-------
SECTION VIII
LABORATORY EVALUATION
The promising pellet formulations identified during the mech-
anical strength and fixed-bed reactor experiments were evaluated in the
model spreader-stoker boiler. This evaluation was based on gaseous
emissions (primarily SO2) and visual observations of the fuel bed. In
addition, 18 Mg of the most promising pellet formulation were fired in
the Battelle steamplant stoker. Criteria pollutants, visual observations,
and ash analyses were used in these evaluations. Both facilities have
been described in earlier symposium proceedings.
MODEL SPREADER EXPERIMENTS
To supplement the fixed-bed reactor experiments, the model
spreader-stoker boiler was used to evaluate the more promising pellet
formulations. Compared to the fixed-bed reactor, the model spreader pro-
vides an improved simulation of the operation of an industrial stoker
boiler and thus evaluates the fuel pellet more realistically.
Table II presents the results of these experiments. In these
experiments, the effect of Ca/S ratio (3.5 and 7), the four pellet pro-
duction techniques, and binder type (cement and methylcellulose) were
investigated. Additionally, for comparison, experiments were conducted
with medium-S Kentucky coal, Illinois No. 6 coal, and the 50/50 pellets
produced during Phase II. Prior to experimentation, the sampling system
and procedures were modified to minimize any reactions that may occur in
the sampling system.
17
-------
Sampling System
The sampling system was a modification of that used during
the Phase II experiments. An in-stack filter was used upstream of the
water trap and the water trap was coupled as close to the stack as
physically possible.
Modifications were made to minimize the presence of water
(especially water with calcium-laden particulates) in the sampling system
that could remove SC^. Provisions were also made to span the instruments
by injecting calibration gases through the entire sampling system before,
during, and after the experiments. This procedure indicated no loss of
SO2 in the sampling train at any time during the experiment.
A comparison of the model-spreader pellet data from the Phase
111 experiments with those from Phase 1 indicate that sulfur capture was
not as great (about 10 to 15 percent lower) for the Phase III experiments.
This small reduction may be attributed to the improved sampling system
where precautions were taken to minimize sulfur capture in the sampling
line. Because of the difficulty in obtaining a representative sample, sulfur
retention based on SO2 emission levels could not be verified from a
sulfur analysis of the bed ash.
Ca/S Ratio
The fixed-bed reactor experiments indicated that the Ca/S ratio
had little or no effect on sulfur capture for Ca/S ratios greater than
3.5. The model spreader data presented in Table II confirm this
observation. Visual observations indicated, as expected, that the pellets
with less limestone (Ca/S - 3.5) burned more uniformly and rapidly
than those with more limestone (Ca/S » 7).
Production Technique
Pellets using the same formulation consisting of Illinois No. 6
coal, limestone (Ca/S * 3.5), and methylcellulose binder, were prepared
by the following production techniques:
18
-------
• Pellet mill (prepared by Battelle staff)
• Auger extrusion (prepared by Banner Industries)
• Disc agglomeration (prepared by Mars Mineral Corporation)
• Briquets (prepared by Evergreen Company).
The pellet-mill and auger-extruded pellets burned satisfactorily,
having sulfur captures of 67 and 63 percent, respectively. The auger
extruded pellets were observed to burn more uniformly than the pellet mill
pellets perhaps because they were more porous (~1.0 g/cc compared to -1.4
g/cc).
The briquetted formulations showed relatively low sulfur reten-
tion (52 percent) — a surprising and unexplained result. These pellets
burned satisfactorily. The disc agglomerated pellets were entirely
unsatisfactory when fired in the model spreader. These pellets disinte-
grated in the combustion zone producing excessive amounts (greater than 50
percent) of fines. Such fines matted the bed causing nonuniform air
distribution. Fuel-bed conditions degraded so rapidly that meaningful
data could not be obtained.
Binder Type
Comparison of sulfur retention data of the auger-extruded and
mill-pellets made with organic (methylcellulose) and inorganic (cement)
binders indicated no significant difference. The binders are used in
very small quantities (less than A percent) and do not have any catalytic
effects. As a result, it appears that the type of binder does not signi-
ficantly effect the combustion behavior of the pellet providing the
physical properties of the pellet are retained. Cement-bound pellets
with satisfactory physical properties could not be made by the disc
agglomeration and briquetting methods.
STEAMPLANT STOKER DEMONSTRATION
Eighteen Mg of the limestone coal fuel pellets with a Ca/S
molal ratio of approximately 3.5 were fired in the steainplant boiler.
Two types of pellets were used — a lower density (0.9 to 1.2 g/cc)
19
-------
pellet produced by Banner Industries using sugar extrusion and a higher
density pellet (C 1.4 g/cc) produced by Alley-Cassetty Coal Company
using a pellet mill. Both types of pellets were fired under a variety
of boiler conditions. Evaluations were based on visual observations,
criteria pollutants, and ash analyses.
PELLET PROPERTIES
Allbond-200 cornstarch and M~167 latex emulsion were used as
binders. The resulting pellet formulation (dry basis) consisted of:
67 percent Illinois No. 6 coal
30 percent Piqua limestone
2 percent Allbond 200 binder
1 percent M-167 latex binder.
These pellets were cylindrical, about 13-mm in diameter and 25- to 75-nun
long. Table III gives the proximate and ultimate analyses and also the
ash-fusion temperature (initial deformation) for these pellets.
Table IV gives the mineral analysis of the ash.
The high CaO in the ash of the treated coal preclude
the usual procedures for evaluating ash characteristics, which are
limited to about 20 percent CaO. The excess CaO above that which
will react with the other ash constituents, principally with Si02, to
form a low-viscosity slag, provides a matrix of solid CaO particles. Thus,
in the ASTM cone fusion determination, this matrix retains the original
shape of the cone, probably even at temperatures above 1650 C, which
explains the anomalous data on the "fusion temperature" of the ash.
The high CaO content physically interferes with flow of the fluid
slag that would result with a smaller addition of CaO. For example, if
the CaO content of typical Illinois No. 6 coal ash were increased to only
20 percent on a normalized basis, the resulting fluxed coal ash would have
a viscosity of only 10 poise at 1430 C, comparable to that of castor oil
at room temperature. With 60.7 percent CaO, the ash will behave as a
solid rather than a liquid because of all the unreacted CaO.
20
-------
The pellets remained sufficiently intact during storage and
handling that an acceptable pellet was fed into the boiler. However, it
was observed that some pellets softened during exposure to rain. Weather-
ability tests on these pellets were rerun showing approximately the same
characteristics. It appears that the weatherability test used during
pellet development has some limitations and that pellets will require some
undercover storage or further formulation refinement for weatherproofing.
EXPERIMENTAL RESULTS
Checkout Runs
Prior to the demonstration test, the fuel pellets were fired for
10 hours to determine the necessary stoker adjustments and to establish
a range of operating conditions.
The limestone/coal pellets were fired without any adjustment to
the stoker mechanism, previously set for a low-sulfur Ohio stoker coal.
The stoker feed mechanism distributed the pellets uniformly over the
grate. This was unexpected since the pellets were all approximately
the same size. It was observed, however, that approximately 50 percent
of the pellets broke randomly into smaller pieces providing a reasonably
good size distribution.
a. Phase II/Phase III Pellet Comparison. Pellets fired in the
Phase III study were significantly superior to those fired in the Phase
11 steamplant runs. They burned more readily at lower excess air rates,
provided improved boiler response (thinner bed), ignited more readily,
and generated lower CO and smoke levels. These improvements are attri-
buted to the fact that the Phase III fuel had a higher heating value,
contained an organic (rather than inorganic binder), contained less ash,
and exhibited superior mechanical strength. However, sulfur retention
was not as high with the Phase III pellets.
21
-------
b. Stoker Coal/Phase III Pellet Comparison. Phase III pellets
appeared to burn equally as well as the low-sulfur Ohio coal that is
normally fired in the Battelle steamplant boiler. The boiler appeared
to be as responsive to the load and could be operated at comparable excess
air levels. Table V compares these two fuels. Emissions are corrected
to 3 percent C^.
c. Effect of Operating Parameters on Sulfur Retention. Because
it was not the intent of the checkout runs to characterize the emissions
for a variety of boiler operating conditions nor was it possible with the
limited supply of fuel pellets, only limited amounts of data were collected
in the checkout runs.
Sulfur retention was observed to decrease for increasing load
as indicated below for relatively constant excess air (about 80 percent).
Boiler Load, Sulfur Retention, Bed Temperature,
percent full load percent C
0.64 50 1315
0.80 48 1405
0.85 47 1425
The bed temperatures were measured with an optical pyrometer sighted on
the combustion zone at the top surface of the bed. Sulfur retention
varies with bed temperature. However, this observation must be tempered
as the combustion conditions were-not closely controlled throughout these
and the observed temperature measurement may not be a good indication of
the actual bed temperature.
At a low-load condition, the excess 0^ was varied from 9.5
percent to 16 percent with no significant change in the S02 retention (46
to 50 percent). Bed depths were also varied from 80 to 160 mm. S02
retention increased somewhat with deeper beds. The increased retention
was attributed to the lower bed temperatures measured for the deeper beds.
22
-------
Demonstration Test
During the limestone/coal fuel pellet demonstration, the pellet
feed rate was maintained at approximately 1360 Kg/hr for a boiler load of
80 percent. Tables VI, VII, VIII, and IX summarize the results of this test.
a. Sulfur Capture. As indicated in Table VI, sulfur
capture was 45 percent during the demonstration test. This sulfur retention
is less than that observed for the model spreader and fixed-bed reactor
experiments firing pellets of similar formulations. Additionally, as
previously discussed, a 75 percent sulfur retention was achieved when
firing a cement-bound pellet with a Ca/S ratio of 7 in the steamplant
during Phase II. The greater sulfur retention of these other experiments
is attributed to the lower bed temperatures, which seldom exceeded 1260 C.
The bed temperatures in the Phase III steamplant demonstration were seldom
less than 1370 C and ran as high as 1455 C. Additionally, with a pulsating
ash discharge stoker, the fuel bed is violently disturbed. Ash can be
recirculated back into the hot zone. Thus, if sulfur is retained in the
ash at a lower bed temperature, it may be released when the ash is exposed
to a higher temperature.
The average SC^ emission level of 1600 ppm during the Method 5
test was verified by the Method 6 wet-chemistry technique. (Wet chemistry
gave an SO^ emission level of 1590 ppm.) In addition, as indicated in
Table IX, the sulfur balance based on the fuel pellet analysis, the
SO2 emission and the sulfur content in the bottom ash (Table VIII)
was complete.
23
-------
b. CO Levels. CO levels from pellet firing were relatively
high compared to those from the firing of conventional stoker coals which
are usually <100 ppm. These higher CO levels may be related to the
nature of the fuel bed and to the fact that the overfire air flow rate
was decreased during the pellet tests. Higher CO levels have been
observed in other pellet firings. Because of the compactness of the
pellet and the limited access of air into it, the capture process first
Involves the formation of calcium sulfide via
2Ca0 + FeS2 S FeO + CO,
which can account for part of this increase in CO.
Another possible explanation for the higher CO levels was that
the overfire air rate was significantly decreased during pellet firing.
In the Battelle boiler the overfire air jets are only 250 mm above the
grate. With the increased bed depth from pellet firing, the overfire
air jets would have impinged upon the fuel bed if the normal flow rate
were maintained. The impingement would increase ash carryover, increasing
particulate loadings.
c. Particulate Loading. The Battelle steamplant boiler
facility has a mechanical collector to control particulates. Depending
on the ash and sulfur content of the coal, the experiments in Phase II
showed that particulate loadings varied between 86 and 258 ng/J (0.2
and 0.6 lb/10® Btu). Generally, for low S, low ash coals, particulate
loadings were less than 129 ng/J (0.3 lb/10® Btu).
The particulate loading from the firing of the fuel pellet
was 258 ng/J (0.6 lb/10® Btu). This loading was not unusually high for
a spreader stoker firing a 33-percent-ash coal. This loading should be
significantly less for a chain-grate stoker. The smoke opacity was
only 20 percent, which would appear low for a particulate loading of
258 ng/J if the fly ash collected was from conventional stoker coal.
However, the fly ash from pellet firing is about 50 percent more
24
-------
dense and considerably more coarse than from conventional coals. For
equivalent mass loadings, optical density varies inversely with particle
size and density. Thus, the apparent discrepancy between smoke opacity
and particulate loading is explained partially by laws of optics. As
indicated in Table VII, about 19 percent of the fly ash was carbon,
a negligible carbon loss.
d. Grate Discharge. Table IX shows that the unburned carbon
content in the grate discharge was less than 2 percent. This indicates that
the fuel pellets were burned essentially to completion. Analysis indicates
that Ca and SO^ were present and could have combined with water to form a
solid mass. Some minor plugging problems were experienced in the ash-
disposal system when steam was used to control dusting during transport of
the ash.
SUMMARY
The steamplant demonstration indicated the limes tone/coal fuel
pellet could be fired in an acceptable manner without modifying the
facility. During the demonstration, sulfur capture levels that would make
the fuel pellet a viable SC^ control were not achieved. The data suggest
that improved SC^ retention could be realized if bed temperature could be
reduced to below 1315 C, perhaps with flue gas recirculation. In addition,
a quiescent fuel bed in a stoker boiler may increase the sulfur retention
in the bed and should reduce particulate emissions.
25
-------
SECTION IX
LIMESTONE/COAL FUEL PELLET PROCESS COST SUMMARY
Table X summarizes an economic analysis of the limestone/
coal pellet process. This analysis considers costs related to raw
materials, utilities, labor, and capital, including profit, Interest,
and income tax. It indicates a process cost of approximately $15.40/Mg
($14/ton) of pellets in addition to the cost of the high sulfur coal.
Increased costs of firing the boiler are not considered. As an example
of such costs, because of the high ash content of the pellet, ash
handling and disposal costs would be higher than for the low-ash con-
ventional coals.
The estimated cost of $15.40/Mg of pellets above the price of
the raw coal 1s based on the best available data. The cost may vary
depending on the type of system used and whether the process may be inte-
grated Into a physical coal cleaning preparation plant. This cost is for
a product with a heating value of 18.6 KJ/g (8000 Btu/lb) and thus adds
Q
about $0.95 per 10 joules ($1 per million Btu) for SO2 control. It
Indicates that the limestone/coal pellet Is cost competitive with other
control strategies.
BASIC ASSUMPTIONS
The following assumptions were used in the analysis.
• Mine-mouth operation
• Limestone and coal ground to 60 to 100 mesh
• Pellet composition:
65 percent high sulfur coal
32 percent limestone
2 percent pregelenized cornstarch
1 percent latex emulsion
• Plant capacity of 54.4 Mg/hr (60 tons/hr).
26
-------
The pellet composition was based on the results of the pellet development
effort.
PROCESS FLOWSHEET
The economic analysis was based on the process flowsheet presented
in Figure 1. In this process
• Coal is taken from a pile instead of directly from
an existing mine operation conveyor
• Limestone is delivered to a pile by truck
• Portland cement is delivered directly to a bin from
a truck by pneumatic feeding system suggested by
Jeffrey Manufacturing Company
• Relatively long inclined conveyors from the coal and
limestone piles are assumed. Costs would be about
35 percent less for horizontal conveyors combined
with bucket elevators.
• A paddle-type mixer, as suggested by California
Pellet Hill, is used
• California Pellet Mill pelletizers and dryers are
costed.
A California Pellet Mill was used in the analysis since cost information was
available. However, pellets can be produced by an extruder at perhaps a
lower cost. Specifications for processing equipment are given in Table 10.
SOURCES OF INFORMATION
Information on equipment included in the flowsheet was obtained
from the following sources:
Front-end loaders — Caterpillar Tractor
Conveyors/elevators — Jeffrey Manufacturing
Storage bins — Butler Manufacturing
Feeders — Jeffrey Manufacturing
Solids mixer — Rapids Machinery
Pelletizers — California Pellet Mill
Coolers — California Pellet Mill
27
-------
COMPARISON TO OTHER CONTROL STRATEGIES
The limestone/coal fuel pellet is an attractive control for two
major reasons:
(1) No major modification of the stoker boiler
facility is required to fire the pellets
(2) The cost of $15.40/Mg is competitive with other
control strategies such as used flue gas scrubbers
or low sulfur coals.
The steamplant experiments indicate that neither the stoker
boiler facility nor its operation will require major modification to
fire fuel pellets. The pellets burn similarly to a lower heating value
coal. In contrast, the addition of a flue gas scrubber is a major facility
modification and increases system maintenance.
Cost comparisons of the various types of control strategies are
difficult to interpret, primarily because of different sets of basic
assumptions and different reference points. However, the pellet process
g
costs of $15.40/Mg or $0.95 per 10 joules ($1 per million Btu) are
competitive with flue gas scrubbers. Foley (7) indicated costs of
between $22 and $33/Mg ($20 and $30/ton) of coal for the gas scrubber
for small to medium-sized industrial boilers based on 1973 figures.
28
-------
REFERENCES
1. Giammar, R. D., R. B. Engdahl, and R. E. Barrett. Emissions from
Residential and Small Commercial Stoker-Coal-Fired Boilers Under
Smokeless Operation. EPA-600/7-76-029, U.S. Environmental
Protection Agency, Washington, D.C. 20460, October, 1976.
2. Giammar, R. D., Weller, A. E., Locklin, D. W., and Krause, H. H.,
Experimental Evaluation of Fuel Oil Additives for Reducing
Emissions and Increasing Efficiency of Boilers, U.S. Environmental
Protection Agency Report No. 600/2-77-008b, Jan, 1977.
3. Wen, C. Y. Noncatalytic Heterogeneous Solid Fluid Reaction
Models. Ind. Eng. Chem., 60 (9): 34, 1968.
4. Wen, C. Y. and S. C. Wang. Thermal and Diffusional Effects in
Noncatalytic Solid Reactions. Ind. Eng. Chem., (8): 31, 1970.
5. Ishida, M. and C. Y. Wen. Comparison of Zone Reaction Model
and Unreacted"Core Shrinking Model In Solid-Gas Reactions - I.
Isothermal Analysis. Chem. Eng. Sci., ^6: 1031, 1971.
6. Ishida, M. and C. Y. Wen. Comparison of Zone Reaction Model and
Unreacted-Core Shrinking Model in Solid-Gas Reactions - II.
Non-Isothermal Analysis. Chem. Eng. Sci., 26^: 1043, 1971.
7. Foley, G. J., et al. Control of S0X Emissions from Industrial
Combustion. Proceedings of First Annual AIChE Southwestern
Ohio Conference on Energy and the Environment, Oxford, Ohio,
October 25-26, 1974.
29
-------
Vent Filter
CPMtMT
(PUEUMATtCMJy —1
ntOM 1VUCK) (V)
HOPPCI&
A\W WrTM OUST
neDUT HUD
6ROUUD UHK5TDUK
FUEL. Oil
FIGURE 1. COAL/LIMESTONE/CEMENT PELLETIZING
PROCESS FLOWSHEET
-------
TABLE X. COMPARISON OF PHYSICAL PROPERTIES OF RAW COAL AND FUEL PELLETS
(a)
Pellet Formulation Compression Post Weathering
Production Coal ~ Limestone' Durability Strength, Weather Durability^ Strength,
Method Type Z Type t Binder Index"1) ib Index^"' Index lb
Raw coal
Illinois *6
100
—
—
—
85 ±
2
74 ± 12
89 ±
1
75
58
Raw coal
E. Kentucky
100
—
—
—
85 ±
2
83 ± 22
94 ±
1
83
94
Saw coal
Lignite
100
—
—
—
77 ±
4
92 ± 22
80 ±
4
34
45
Rav coal
Rosebud
100
—
—
—
84 ±
2
50 ± 15
79 *
2
20
68
CPK lab sill
Illinois #6
70
Pltjua
30
25 Allbond +¦ IX
Polyco 2136
87
112
100
85
>112
Banner extrusion
Illinois #6
70
Plqua
30
1.51 Allbond 200 +
11 M-167.01
94
84
100
62
60
(a) Water added as needed.
0») Percent survival - 100 - percent fines.
-------
TABLE II. MODEL-SPREADER STUDIES
Run Mo.
Fuel
Fuel
Sice,
¦*
Ca/S Ratio
(Approx)
Average
so2,<«>
PP"
Predicted
S02,
PP"
Average
Stack
Te«p, C
Average
Exceaa
Air,
percent
Average
Sulfur
ltetcntlon,
percent
«2.
percent
C02,
percent
CO,
PP"
7-10-78(b)
30/S0 CfM pallets
(cant)
12.5 x 19
7
854
3700
340
77
77
9.2
10.9
20
e-io-7#00
100/50 CFM pallet*
(cownt)
Ditto
4
1116
3700
295
85
70
9.7
11.4
—
79-2
SO/SO CFM pellet•
(caveat)
Ditto
7
1040
3700
350
140
72
11.5 - 15.0
5.9 - 8.0
—
79-3
70/30 CFM pellet*
(ceaenc)
Ditto
3.5
1220
3700
360
110
67
10.0 - 13.8
NA
—
79-4
Medlue-S Kentucky
—
0
1050
900
—
120
-13
9.5 - 19
8.2 - 10
150
79-5
Illinois W coal
—
0
*120
3700
3*0
95
-12
8.0 - 12.7
a.* - 10.8
100
79-6
50/50 CPH pallets
(cnut)
12.5 x 19
7
1240
3700
300
120
67
13.5
8.5
300
79-7
Illinois #6 coal
—
0
3700
3700
325
100
0
10.6
9.2
100
79-8
70/30 CPN pellets
(¦ethocel)
12.5 s 19
3.5
1480
3700
335
90
60
9.5 - 11.5
8.0 - 11.2
150
79-9
70/30 brlqnete
(¦ethocel)
12.5 x 25
3.5
1780
3700
365
80
52
8.2 - 9.8
10.8 - 11.6
85
79-10
70/30 dlac pellets
(¦ethocel)
12.5 die
3.5
—
—
—
—
—
—
—
79-11
70/30 extrmlon
(¦ethocel)
—
3.5
1370
3700
3? 5
60
63
7.4 - 9.2
10.8 - 12.6
90
79-12
70/30 OK pellets
(¦ethocel)
12.5 x 19
3.5
1260 .
(1220)
3700
345
75
67
6.4 - 10.2
10.4 - 12
50
(a)
(b)
(c)
Noraallted to 3 percent Oj.
1978 data.
By Method 6.
-------
TABLE III. ULTIMATE, PROXIMATE, AND ASH-FUSION TEMPERATURE ANALYSES FOR
LIMESTONE/COAL FUEL PELLET Ca/S =3.5
u>
Proximate Analysis
(As received), %
Volatiles
Fixed
Carbon
Moisture
Ash
N
(difference)
Ash-Fusion
Heating Temperature, C
Value, Initial Deformation
KJ/g Reducing Oxidizing
42.9
21.8
2.15
33.2
47.0
3.2
.9
2.9
10.5
18.6
1500+
1500+
-------
TABLE IV. MINERAL ANALYSIS OF ASH
Compound
Percent Weight
Silica, Si02
14.20
Alumina* Al^O^
5.42
Titania, TiOg
0.24
Ferric oxide, ^e2°3
7.10
Lime, CaO
51.88
Magnesia, MgO
6.94
Potassium oxide, K^O
0.55
Sodium oxide, Na^O
0.42
Sulfur trioxide, S03
10.77
Phos, pentoxide, PjO^
0.09
Strontium oxide, SrO
0.00
Barium oxide, BaO
0.01
Manganese oxide,
0.06
Undetermined
2.32
100.00
34
-------
TABLE V. COMPARISON OF EMISSIONS FROM COMBUSTION OF A
LOW SULFUR COAL AND LIMESTONE/COAL PELLET
Smoke
Fuel N
Fuel S
Opacity,
Converted,
Emitted,
Coal Type
percent
CO
NO
percent
S02
percent
Low-S coal
10
70
480
18
540
90
Fuel pellet
20
A 00
310
20
1800
45
35
-------
TABLE VI. EMISSION DATA SUMMARY FOR FUEL PELLET DEMONSTRATION
Saoka CO *t ^_ ___ Tu*l * „ „ n PVjal S
Load, O , COj, CO, WO, SO,, Opacity, 31 O2, ** Converted, 2 Ealttcd, Particulate*,
X t I lit X ppk Conputad Meaaured X Computed Measured X og/J
80 «.4 10.5 300 310 1600 20 420 2250 **0 20 4100 2250 55 258
-------
TABLE VII. ANALYSIS OF METHOD 5 FILTER CATCH (Weight Percent)
Ash
C
Ca
COj Fe Total S
81
19
11
4 54
TABLE
VIII.
ANALYSIS
OF
GRATE DISCHARGE (Weight Percent)
Ash
c
Ca
C03 Fe Total S
97.7
1.8
36.5
0.7 5.8 3.9
TABLE
IX.
SULFUR BALANCE
Sulfur Retained in
Computed Fuel S In, Emitted,as S02t Bed Ash as SO2,
lb/106 Btu lb/10 Btu * lb/106 Btu
7.4 (3182 ng/J) 4.1 (1763 ng/J) 3.3 (1419 ng/J)
37
-------
TABLE X. SUMMARY OF LIMESTONE/COAL PELLETIZING PROCESS COSTS
w
00
kasisi 60 tana per hour product with 65 parcant coal, 30 psrceat li«eatoaa, J parent Portland
23 hour* par day, 330 days per y*tr
1380 tooa par day, 45$,400 per year of product
Plaed plant lnvestaent $2,790,000
Working capital 80,000
Interne during construction 230.000
$3,120,000
Itea
la) Materlala
Llaeatona IB toaa/hr, 136,620 tons/year at $8/ton delivered
Pregelatln cornetarch, 9100 tooa/year at $20/ton delivered
Latex aula lor, 1.2 too/hr, 9100 ton year at $150/ton delivered
Utilities
Process wter 12 toea/hr (48 gpa) 21.9 M gallon/year at $0.2/M gal
fuel oil 32 HfBtu/hr, 243 trillion Btu/yr of $3/MMBtu
Power 75 percent of 1917 rv or 1440 W at $0.035/KW-hr
Dleael fuel 5 gph, 37,950 gallon/year at $0.80/gal
Labor ttelated
Direct labor — 7 operator* ? $B/hr plua 25 percent payroll burden
($10/hr total); stsffed 365 daya/yr
Supervision — 15 percent of direct lsbor
Overhead — SO percent of direct labor end aupervlaion
Capital Related
Maintenance — f percent of fixed plant Inveataent
Special pelletlier Balntenance at $0.30/ton plus
$0.55/toa die and rollera
Front-end loader Balntenance at $0.22/hr per aachlna
Taxes and lnenrance — 1.5 percent of fixed plant Investment
Depreciation — 11 year, straight line on fixed plant lnveetaent
Profit, Interest, lacoae tax — 30 percent of total employed capital
TOTAL
Annual Costa,
Dollar
$1,092,960
1,138,500
4,400
728,600
382,500
21,800
613,200
91,980
306,500
167,400
387,100
3,300
41,850
250,000
936.000
Per Ton Product,
Dollars
2.4
3.0
0.01
1.60
0.84
0.06
1.30
.20
.70
0.37
0.85
0.01
0.09
0.56
2.05
$6,167,100
-$14.00
-------
CONTROL OF EMISSIONS FROM RESIDENTIAL WOOD COMBUSTION
BY COMBUSTION MODIFICATION
By:
J. M. Allen
BATTELLE
Columbus Laboratories
505 King Avenue
Columbus, Ohio 43201
39
-------
ABSTRACT
This program was conducted to identify promising methods of
reducing emissions of air pollutants from residential wood burning stoves.
The overall study has included a review of the few ongoing and recently
reported studies related to emissions measurements, causes, and charac-
terization*
The most significant emissions are the hydrocarbons and carbon
monoxide released by wood pyrolysis and the carbon monoxide formed by the
combustion under locally starved air conditions. The hydrocarbons are
especially important as they have been shown to contain polycyclic species
which are suspected as being carcinogenic.
An experimental phase of the Battelle program has included stove
operations in the laboratory, designed to correlate emissions with design
and operating characteristics of the stoves. The burning properties of
different types of fuel wood have also been investigated. The combustion
tests in radiant stoves have been designed to identify those phenomena
which contribute directly and indirectly to the emissions. Continuous
monitoring of the following emissions has been provided: 0^, CO^, CO,
NO, SC^, and total hydrocarbons. In a few tests, batch sampling of stack
gases has been performed to determine particulate emissions, and the
concentration of polycyclic organic species in both particulate and
gaseous emissions. Continuous weighing of the stoves during operation has
provided a measurement of burning rate.
The average emission factors for CO and total hydrocarbons
varied by more than a factor of 10 between different burning modes and
rates. Both emission factors vary inversely with burning rate. At the
higher burning rates, the CO constitutes a larger fraction of the emissions
40
-------
of combustibles. The emission factors also vary inversely with excess air
ratio as measured at the stove outlet, although a large fraction of the
total air may bypass the active burning zone within most stoves. True
down-draft combustion produces low emission factors compared to other
modes of burning, especially with a preheated air supply. Nitrogen oxide
emissions increase with overall excess air in all the naturally drafted
stoves, ranging between 1 and 10 lb/ton fuel.
Combustion modification techniques were found to affect emissions
and therefore are of interest for emission controls. These include fuel
modifications, thermal and flow modifications in the stove design, and
operator techniques.
41
-------
ACKNOWLEDGMENT
The guidance of the Project Officer, Mr. Robert Hall, is acknow-
leged and appreciated. The willingness of the Solar Energy Program of
the Tennessee Valley Authority to interchange experimental data with this
program is also gratefully acknowledged.
42
-------
INTRODUCTION
As we are all aware, there is a significant increase in the use
of wood as a residential heating fuel. This trend was initiated by the
rapidly rising prices of oil, public skepticism as to reliability of normal
fuel supplies, and the widespread impression that wood burning is environ-
mentally clean. Increased wood use is being encouraged by DOE, equipment
manufacturers, and many environmentalists, all with a very limited technology
pertaining to emission factors, and even less technical bases for reducing
the emission. We have started a program for EPA to identify the effects of
several operating variables on objectionable air emissions, and to identify
principles of combustion modification that will reduce emission levels from
residential wood burning facilities. Unfortunately, the residential stove
operator, with his inherent preferences for minimizing his labor and maximizing
overall thermal efficiency is adopting stove designs and operating practices
conducive to increased emission factors. This accentuates the environmental
significance of the continuing increase in quantity of wood burned residentially.
BURNING CHARACTERISTICS OF WOOD
Residential cord wood is generally characterized as either a hard-
wood or softwood specie, either green or air dried, and normally consists of
both split and round pieces. These characteristics all have small but signi-
ficant effects on emissions. The chemical composition and heating value
vary remarkable little between species when measured on a dry basis, the
soft woods being only slightly more calorific due to their slightly higher
resin content.
The most significant burning characteristic affecting emissions
is the phased burning phenomena. The three phases, drying, pyrolysis, and
43
-------
char burning take place successively for any one particle of wood, but all
three phases occur simultaneously during most of the burning in a stove.
As heat is added to a piece of wood, drying of the surface layer occurs first.
This is followed by thermal pyrolysis of this same layer simultaneous with
drying of the underlying wood. The pyrolyzed gas may or may not burn at
the surface. The effluent steam from the surface drying absorbes both latent
and sensible heat. It also tends to block convective heat flow from the
surroundings into the wood. This heat flow blockage retards the wood pyrolysis,
and thus slows the overall rate of wood burning. As the burning continues
and the surface layer becomes converted to a char, pyrolysis gas from the
wood below the char layer must diffuse through the charred surface before it
can react with atmospheric air to burn in a flame. This subsurface pyrolysis
is exothermic, however, and once started can consume a significant portion
of the wood without requiring oxidation at the surface or oxygen penetration
into the wood. The rate of pyrolytic gas evolution often exceeds the availa-
bility of oxygen at the surface resulting in unreacted or partially reacted
species leaving the wood even when flames surround each piece of wood.
Moisture from the core of the wood can, even in some cases, transpire outward
through a hot char layer resulting in CO formation by the water-gas reaction.
This primary burning within a pile of wood thus results in incomplete burning,
with primary combustion products including CO, and other cellulose and resin
pyrolysis products. Subsequent oxidation with a secondary air supply is
required to complete the burning. The limited pressure drop normally avail-
able in the air supply is usually not sufficient to provide the penetration
and mixing required for effective secondary combustion. This results in ultimate
emissions of CO, hydrocarbons, and other organics including polycyclic organic
materials. Some of these materials condense onto the chimney walls as creo-
sote, some condense into submicron particulates within the stove and chimney
passages, and some leave the chimney as a gas. The fraction that is measured
as particulates by the Method 5 sampling procedures depends on several factors
and is typically only a third or less of the total unburned hydrocarbons
leaving the stove. The remainder, which is not deposited in the stack as
creosote, presumable condenses onto and into additional atmospheric particulates-.
44
-------
Types of Stoves
There are at least four generic types of naturally drafted wood
burning stoves as shown in Figure 1. The designs are catagorized as to the
the direction of air flow through the burning wood and wood inventory within
the stove. Each of these types is shown in the figure as it can be adapted
to provide a secondary air supply and permit secondary burning. Although
considered very desireable, effective secondary burning has seldom been
attained in the laboratory, and apparently is rarely attained in residential
applications. The updraft design appears to be most conducive to organic
emissions because the inventory of wood within the stove is extensively heated
by primary combustion products depleted of oxygen. The true down-draft
burning mode can be very clean burning, when up-flow countercurrents within
the bed are prevented. This type of burning delays the rapid release of
pyrolysis products until active and complete burning can be achieved.
Other Published Studies
There have been a few programs which have published experimentally
measured emissions from wood burning stoves. Professor Butcher at Bowdoin
College has measured particulate concentrations after extensive dillution
and flue gas cooling in a collection system. He demonstrated that the
emission factor for these emissions varied inversely with burning rate and
increased linearly with amount of wood charged into the combustion chamber/"^
Monsanto measurements conducted at Auburn University for EPA showed that at
fairly high rates of burning the emission factors were nearly independent
of specie and moisture content of the wood. The two similar stoves used in
that program were not operated in a severely restricted air supply mode,
as is prevalent in residential practice. Even so, as much as one third of
(2)
the carbon fired in the wood was reported emitted as CO. Thus, the increases
in organic emissions associated with starved-air or low rate burning were not
observed. Extensive measurements of organic species and POM's were made,
however, indicating that even at high burning rates these emissions from
stoves are appreciably higher than for open burning in a fireplace. Both
(3)
the Monsanto study and a California Air Resources Board (CARB) study
found that at high burning rates the particulates retained on the Method 5
filter were appreciably less than the organic material collected by the back
half of the Method 5 system. This indicates that only partial condensation
45
-------
of the organic emissions had occured at the sampling location. The CARB
study showed that most emission factors increase with increasing overall
excess air. No distinction was noted between excess air within the active
combustion zone and air that by-passed the primary combustion zone and
simply diluted and quenched the primary combustion products.
EXPERIMENTAL PROGRAM
Facility
The test facility assembled at Battelle for this program consists
of three test sites, each with independent, prefabricated, insulated chimneys
exhausting within a high bay area in the building. Thus, the chimney effects
on draft are not affected either by outside, wind or by the building's tran-
sient pressure fluctuations. Radiant stoves are mounted on electronic
scales, such that total weight losses associated with fuel burning can be
continuously monitored. Normal combustion products (CO NO and CO) are
i., x
continuously monitored with NDIR instruments, together with 0^ to determine
combustion parameters including excess air and total air flow. Total hydro-
carbons (THC) are monitored with a FID instrument calibrated with methane.
A high-temperature heated sampling line is used between the sampling probe
in the stove outlet and the instrument. This measurement is interpreted
as including all organic emissions, whether they ultimately leave as vapor
or particulates. Particulate and gaseous samples are collected simultaneously
during some tests for POM measurements. This is done at a downstream loca-
tion in the stack, using a Method 5 front-half filter, followed by a water
cooled XAD-2 column." For definitive measurements, filter catches and XAD-2
column catches are separately extracted and analyzed by GC/MS for POM content.
Different fuel woods used have been:
• Commercial oak cordwood, splits and rounds,
cured> 20.7 percent moisture
• Commercial pine cordwood, splits and rounds,
not cured, 42 percent moisture
• Douglas fir brands 3/4 inch x 3/4 inch on
1 inch centers, oven dried, null moisture
• Oak lumber, nominal 4 inches x 4 inches, cut
diagonally, thoroughly air dried, 12.4 percent
moisture.
46
-------
The first two are representative of typical residential fuel,
although many stove operators will cure or age their fuel before burning.
The fir brands are widely used as a standardized fuel for test purposes.
They are very reproducible, easy to ignite, and fast burning. Because of
the very high surface area and lack of moisture, the pyrolysis rates early
during the brand burning are unusually high. Dried oak, in the 4x4 shape,
has been used as a fuel closely simulating commercial cordwood in moisture
content, size, and shape, yet reproducible to a degree considered satis-
factory for test purposes.
Test Procedures
Our usual test procedure is to operate the stove as a batch process,
with a test run immediately following a pre-run, burning the same fuel at the
fame burning rate. The weight loss (fuel burned) and emissions are recorded
each minute until the weight loss equals 95 percent of the weight of test
fuel fired. Time averages of the measured flue gas compositions are machine
calculated for each entire run. Emission factors are then calculated
using the averaged gas composition and the weight of wood burned. Effects
of a decreasing rate of weight loss (i.e., decreasing burning rate), and
the changing composition of the fuel being burned during a run are not
accomodated by this calculation technique. Those factors would slightly
increase the emission factors, over those calculated by the time averaging
technique.
MEASURED EMISSIONS
Emission of Combustibles
Figure 2 shows the time variation of hydrocarbon emissions during
a typical run burning the triangular 4x4 oak fuel. The total hydrocarbon
emissions rose to above 10,000 ppmC (measured as methane) during the early
burning, and reduced to less than 1000 ppm when the wood pyrolysis had conver
most of the remaining fuel into char. Simultaneously, the in the flue
gas dropped below 5 percent during most of the wood' pyrolysis, and increased
to about 13 percent during the char burning. This increase in C>2, late in
many runs, is assumed to come from an increasing fraction of the primary
air flow by-passing the active primary burning area.
47
-------
In Figure 3 the carbon dioxide and carbon monoxide measurements
of the flue gas are shown for the same run. It must be recognized that
these flue gas analyses can not be related exclusively to the direct primary
combustion effluents, even though no secondary combustion was identified
in this run. The secondary air admissions dilute and cool the primary
combustion products by an unknown amount, and some of the primary air can
bypass the active burning that takes place within the pile of wood.
There appears to be a consistent correlation of CO and hydrocarbon
emissions, even though their formation mechanisms are not directly related.
Figure k shows this relationship as observed in the operation of several
stoves operated at different rates of burning. At low burning rates (i.e.,
stove outlet temperatures below 400 F) the hydrocarbons approximately equal
the CO emissions (HC/C0=1), whereas at high burning rates (i.e., stove outlet
temperatures above about 600 F) the hydrocarbons are only about 1/5 the CO
emissions (HC/CO=0.2). When these stove emissions get very high, the
CO emissions increase relative to the hydrocarbons more than proportionally
as these correlations would indicate.
Effects of Burning Rate
Figure 5 shows the effects of burning rate on THC emission factors
when burning oak 4 x 4 s in four different stoves. These emission factors
were determined from time averaging of emissions measured over a complete
burning cycle. It is evident that the emission factors obtained at high
burning rates are not representative of stoves operated at the reduced burning
rates often used in residential heating.
The combustion in the stoves is air-contro3led once the wood is
adequately ignited. At reduced rates of air flow into the combustion chamber,
caused by partially closing the inlet air damper, a reduced burning rate of
fuel is obtained. The reduced wood burning rate is frequently accompanied
by increased oxygen content in the flue gas, indicating a larger fraction
of the total air flow bypasses the active combustion region. Thus, the
restricted air-flow burning is accompanied simultaneously by higher CO and
THC emission factors, and higher overall excess air ratios.
48
-------
Effects of Mode of Combustion
Figure 6 shows a simular plot of the emission factors for a fifth
Stove, which was altered to permit operation under three different modes
of air flow through the primary combustion region. In the updraft mode of
burning the primary combustion air entered the combustion chamber through
the walls, just above the hearth supporting the burning wood. Combustion
products left the top of the chamber, which is the principal characteristic
of this burning mode. The local thermal buoyancy effects vithin the burning
wood pile induced the primary flow of combustion air upward through the wood,
permitting most of the total air to bypass the active burning area, as is
typical in an open fireplace. In the side draft mode the primary combustion
products left the primary combustion chamber at the hearth level, thus
drawing more of the total air flow through the burning wood. Although CO and
THC emission concentrations were appreciably different, the emission factors
for these two modes were found to be similar to each other (Figure 6) and
similar to the emission factors observed in other stoves (Figure 5).
Also shown in Figure 6 are emission factors observed during
several runs with true down draft of air flow through the burning wood. In
this downdraft mode, the pile of burning wood was supported on a horizontal
grate, with a baffle arrangement forcing all primary air to pass downward
through the pile. The burning is intended to occur only adjacent to the
grate, the rest of the wood remaining cool. This operation produced low
THC emission factors as shown by the test points plotted separately on
Figure 6. Visual observations of this downdraft burning were made, sighting
through a pyrex window in the top of the combustion chamber- The gas flow
velocity and direction were observed by flame velocity and direction within
the pile of wood. At lower total air flow rates, the principal down flow
of air tended to channel through only one portion of the bed, as local
thermally-induced counter flow currents moved upwards through other regions
of the bed. This counter flow phenomena tends to defeat a principal advan-
tage of the down flow mechanism: namely isolating the primary combustion
products from the inventory of wood thus delaying extensive pyrolysis early
in the burning cycle.
49
-------
It is evident from these observations that the down flow combustion
mode has inherent advantages of low emissions, but the down flow is hard
to maintain over a wide range of burning rates.
Combustion Air Preheat
While operating a stove in the downdraft mode, a pair of runs
was conducted to identify possible effects of preheating of the primary
air on emission factors for combustibles. The tirre averaged emissions for
the entire runs were converted to emission factors as shown in Table 1.
Oc-k logs were burned in these runs. Even with the low factors inherent
with downdraft burning, reductions were obtained when the primary air was
preheated with an outside source. It is not certain if this same effect
would be observed with other modes of combustion. In the downdraft mode
of burning the heated incoming primary air should have the greatest effect,
by promoting early pyrolysis of wood thus adversely affecting emissions
of combustibles. However, these pyrolysis products can not easily bypass
the active burning area at the downstream edge of the bed as in other modes
of combustion. Reductions in emission factors were observed as shown in
Table 1.
Preheating secondary air is assumed to have no effect on emissions
unless secondary combustion is established or can be initiated by the
preheat. When burning wood which emits large quantities of combustibles
early in a burning cycle (dry fir brands) secondary combustion was initiated
when a 5 scfm stream of 442 F secondary air was admitted. This constituted
about 15 percent of the total air supply, and resulted in a small reduction
in the previously very high THC emissions.
Nitrogen Oxides Emissions
Although NO emissions have not generally been recognized as a
serious emission from wood stoves, they should be considered. The temperature
levels within the stoves are not expected to reach the range where fixation
occurs. Although the nitrogen content in wood is not high, the conversion
cf this nitrogen can be significant. Accordingly, our laboratory instrumen-
tation included a monitor for NO^ (as NO) during the test runs.
50
-------
Figure 7 shows the experimental correlation of average NO emission
factors with the time averaged oxygen content of the "flue gas leaving .the
stove. It must be recognized that the excess air or oxygen content leaving
the combustion chamber is only an upper limit as to what prevails within the
burning pile of wood, as in most cases there is no measure of how much of
the primary supply bypasses the primary burning region, or how much
secondary air is supplied which only dilutes the primary combustion products.
Commercial Low Emission Burner
A residential wood burning appliance has been developed under a DOE
contract by Professor Hill at University of Maine. The burner demonstrates
that increased efficiencies can be obtained with an improved design of the
combustion and heat recovery system. The burner is a residential boiler
(water heater) for a central heating system that incorporates several factors
beneficial in controlling air pollution emissions. The combustion is supported
by both forced draft of combustion air and induced draft of flue gases, such
that a high turbulence level can be maintained in the burning areas, yet not
develop a firebox pressure exceeding ambient. It operates only at a fixed,
relatively high burning rate (^150,000 Btu/hr) comparable to one of the
large radiant heaters. Several characteristics of this design favor low
THC and CO emissions:
• High turbulence level in primary burning area
• Primary combustion products isolated from
inventory of wood
• Inventory of wood cooled by water-walled
chamber
• Burning confined to small portion of the
wood inventory
• Refractory lined primary combustion area
• Refractory lined secondary combustion area
• Large pieces of wood utilized to reduce
excessive pyrolysis early during burning cycle
• Burning restricted to high rates, with heat
storage provisions to accomodate lower heating
requirements.
51
-------
A commercial production version of this design of wood burner
was obtained and operated in the laboratory. Table 2 presents the operating
conditions and the significant emissions. During the batch burning, the
overall excess air remained high, with an average oxygen content exceeding
10 percent. The heat recovery efficiency remained high (low stack tempera-
tures) because a large and effective heat transfer surface is provided for
the flue gas to water heat exchange within the boiler.
Table 2 summarizes the emissions performance of this boiler,
together with corresponding data from a radiant heater operating in the up
draft mode presented for comparison.
TECHNIQUES FOR REDUCING EMISSIONS
In reviewing the observations of stove performance on this and
other experimental programs, several combustion modification techniques
have been identified which shew promise of reducing objectionable emissions.
In summary, they are listed below with a few less promising noncombustion
techniques. Most of these techniques still need further development before
they can be reliably reduced to practice, especially in the simultaneous
application of several techniques. Both the prevention of the initial
pollutant formation and destruction of the formed pollutants before their
emission from the stove are considered. Some of these approaches are in
conflict with operator convenience or overal thermal efficiency of the
stove, however, any reduction in hydrocarbon emissions from the stove will
reduce creosote problems in the chimney. The forced draft boiler described
previously is an example of the simultaneous application of several of
these techniques.
Modifications of Fuel
• Utilize a processed wood (pellets or briquettes)
to permit continuous fuel flow into and/or within
the stove. Active burning can then be contained
within a small volume permitting better controls,
even to the extent of developing truly fuel-
controlled burning.
52
-------
• Reduce volatility of the wood, especially at low
temperatures, either by thermal processing
(charring) before burning, or possibly with
additives or fuel admixtures.
• Reduce moisture content, if other provisions
are made to limit excessive pyrolysis when
initially heated.
• Incorporate additives to lower ignition tempera-
ture of hydrocarbons leaving primary combustion
space.
Modifications of Stove Design,
Thermal Control
• Insulate or cool the inventory of wood in the
stove prior to active burning to minimize
premature pyrolysis
• Insulate active combustion zone(s) to aid in
ignition and burning of pyrolysis products by
increasing the temperature level
• Preheat the air supplied to active burning zone(s)
without pyrolyzing the wood inventory within the
stove.
• Provide recoverable heat storage for stove
output, to permit a high rate of burning to
satisfy low rate heat demands. This can be
either integrated within a stove or as an add-
on component, utilizing sensible and/or latent
heat storage.
• Add supplemental fuel, supplemental air, or
both to the secondary combustion system to just
assure complete burning of combustible emissions
from the primary burning zone. The supply
requirements for these supplements change
appreciably during a bum cycle, requiring a
sophisticated sensing and control system.
53
-------
• Provide an afterburner system, independently fueled
with balanced fuel and air supplies, which will
consume all combustible emissions as their emission
rate changes during a burn cycle.
• Provide a heat path from primary burning zone to the
secondary burning zone, to assure ignition when
combustible mixtures are obtained after introduction
of secondary air.
Modification of Stove Design,
Air Flow Control
• Supply energy to the air supplies to provide
control, turbulence, and mixing in the primary
and secondary combustion zones, as by the use
of a forced and/or induced draft fan.
• Control flow of primary air within the stove
to minimize the extensive by-passing of the active
burning zone, especially at low burning rates.
• Mix secondary air with the primary combustion
products immediately as they leave the burning
wood.
• Prevent combustion products from convectively
heating the entire inventory of wood within
the stove.
• Change combustion chamber geometry when operating
at reduced burning rates to maintain a high rate
of burning per unit volume. Movable baffles or
dummy removable side walls would be one approach.
• Terminate secondary air supply thus increasing
primary air supply when flue gas dilution is the
only effect realized. This may also require a
complex sensing system.
54
-------
Modification of Operator Techniques
• Avoid excessive quantities of wood within the
stove at one time.
• Use large pieces of wood after burning is
established, consistent with desired rate of
burning.
• Operate only at high burning rates, with
shutdown times between burns to accommodate
lower heating requirements (i.e., intermittent
stove operation).
Add-On Devices
• Heat storage device as noted above
• Direct combustion afterburner as noted above
• Catalytic afterburner
• Electrostatic precipitator
• Scrubber.
55
-------
REFERENCES
Butcher, Samuel S., and Edmund M. Sorenson. A Study of Wood Stove
Particulate Emissions. Journal of the Air Pollution Control
Association, _29_ (7): 724-728, 1979.
DeAngelis, D. G., D. S. Ruffin, and R. B. Reznik. Source Assessment:
Wood-Fired Residential Combustion Equipment Field Tests. EPA 600/2-79-019
U.S. Environmental Protection Agency, Research Triangle Park, North
Carolina, 1979.
Kosel, Peter. Emissions from Residential Fireplaces. State of
California Air Resources Board, Report No. C-80-027, April, 1980.
56
-------
Underfire Air
OR UP DRAFT
Down draft
Cross draft
S
P-fc
j
SC
B/
B 6
\
r
3
S-FLOW
Types of Stoves
ACCORDING TO
AIR FLOW PATHS
P - Primary air supply
S - Secondary air supply
E - Exhaust to stack
B - Primary burning
SC - Secondary combustion
FIGURE 1, GENERIC DESIGNS OF WOOD STOVES BASED ON FLOW PATHS
57
-------
25.000
20.000
U1
00
20.000 -
° 15.000
<
Od
^ 10.000
z
o
o
5.000
0.000
~ OXTCEN
0 TOT.HC
16.000
12.000 o
8.000 i
4.000
0-000
1 -250
I .500
.750
1 -000
0.000 .250
.500
TIME (HOURS)
FIGURE 2, HYDROCARBON EMISSIONS DURING A RUN BURNING CURED OAK
-------
25.000
C02
CO
20.000
z
o 15.000
h-
ZZL
10.000
:z:
o
o
5.000
0.000
0.000
1 .500
1 .250
1 .000
.750
.500
.250
TIME (HOURS)
FIGURE 3. CARBON MONOXIDE EMISSIONS DURING A RUN BURNING CURED OAK
-------
THC/C0=0.2
THC/CO=I
LEGEND
~ = Low Burn Rate Oak 4x4
o = Med. Burn Rate Oak 4x4
a = High Bum Rate Oak 4x4
o = High Burn Rate Brands
40000
20000
30000
Avg. CO, PPM
10000
60000
90000
Fig. 4 Correlation of THC and
CO Emissions for
Several Stoves
-------
1 "I "S | I I 1 I I ] III'
) 15 20 2
Burning Rate, lb/hr
1 1 ' i
10
20
—i r
5
Fig. 5 THC Emission Factors for
Several Stoves
61
-------
LEGEND
~ = Up Draft Mode
o = Side Draft Mode
a = Down Draft Mode
-r
25
~
r
5
10
i
15
i
20
30
Burning Hate, Ib/hr
Fig. 6 THC Emission Factors for
One Stove, Three
Modes of Burning
-------
Hi
~
O
A
O
LEGEND
Up Draft Mode
Side Draft Mode
Down Draft Mode
Forced Draft Mode
8
pQ
CD H
A
A
NH
~ ~
O
~
©-!
10 15
Avg. 0 in Flue Gas, Percent
Fig. 7 NO Emission Factors
for Different Modes
of Burning
20
-------
FIELD TESTS OF ELEVEN STOKER COAL-FIRED BOILERS
FOR EMISSIONS CONTROL AND IMPROVED EFFICIENCY
By:
P. L. Langsjoeti
KVB Incorporated — A Research Cottrell Company
6176 Olson Memorial Highway
Minneapolis, Minnesota 55422
64
-------
ABSTRACT
This stoker test program was awarded to the American Boiler Manufacturers
Association (ABMA) in late 1977 as a result of the national interest in coal
utilization. The objective of the program is to improve specification data
relating to emissions and efficiency of coal fired stoker boilers. Such data
are required by both industry and government in order to increase coal usage.
Eleven stoker boilers were tested including six spreader stokers, one
vibragrate stoker and four overfeed traveling and chain grate units. Each
units emissions and efficiency were measured under a variety of operating con-
ditions. This paper deals with particulate loading, nitric oxide concentration
and combustibles in the bottom ash and flyash. The effect of stoker design,
boiler loading, excess air, overfire air and coal properties on the three types
of emissions cited above is also discussed.
Test results show that overfeed stokers have lower particulate and nitric
oxide emissions, and lower combustible heat losses than do spreader stokers.
Flyash reinjection is shown to substantially increase particulate loading in
some cases. Overfire air is shown to have little or no effect on nitric oxide
emissions, and flyash combustible content is a function of particle size.
These and other relationships are discussed. More importantly, an attempt is
made to quantify these relationships and provide a broad data base from which
government and industry may draw to implement sound decisions for our coal
future.
Field testing was completed in late 1979. Individual site reports are
available through the EPA or NTIS. A final project report is scheduled for
completion in late 1980.
65
-------
SECTION 1
INTRODUCTION
This paper outlines the objectives and presents some of the preliminary
findings of a coal-fired stoker test program. The program was initiated be-
cause in recent years there has been a preponderance of industrial boiler in-
stallations which have been shop-assembled gas- and oil-fired units purchased
and installed at substantially lower costs than conventional coal burning,
boiler-stoker equipment. Because of this decline, in coal-firing, little or no
work was done to improve specification data and information for consulting
engineers and purchasers of coal-burning equipment. The current implementation
of more rigid air pollution regulations has made it difficult for many coal-
burning installations to comply with required stack emission limits, thus
creating a further negative influence on coal-firing.
The market for coal suitable to be fired in industrial boilers, as re-
flected by sales data, is being held back by critical uncertainties in the en-
vironmental and energy areas, causing potential customers of coal-fired in-
dustrial boilers to shelve plans for capital expansion and conversion. This
has caused a serious reduction in the number of installations of new industrial
coal-fired units. It is highly desirable to remove these uncertainties and
thereby establish confidence among industrial users to order and install stoker
coal-fired boilers. This will lead to significantly increased coal usage.
OBJECTIVES
The objectives of this Program are:
1. To prepare a comprehensive project report: Design and Application
Guidelines for Industrial Stoker Firing.
66
-------
2. To advance boiler and stoker technology by testing various boiler
furnaces and stoker designs for response to changes in coal analysis
and sizing, degree of flyash reinjection, overfire air admission, ash
handling and pollutants emitted.
3. To refine applications of existing pollution control equipment and to
more closely control stack emissions under varied operating condi-
tions through more accurate boiler outlet dust loading data.
4. To contribute to the design of new and improved air pollution control
equipment.
5. To facilitate preparation of reasonable and workable national
emissions standards for small coal-fired boilers by the Environmental
Protection Agency.
6. To facilitate planning for coal supply contracts by users of the
boiler/stoker equipment by developing reasonable emission regulations.
7. To promote the increased utilization of stoker coal-fired boilers by
U.S. industry by insuring the compatibility of these units emissions
with appropriate environmental requirements.
AWARD OF CONTRACT
This stoker test program was awarded to the American Boiler Manufacturers
Association (ABMA) in late 1977 as a result of the national interest in coal
utilization. The program is sponsored by the Department of Energy (DOE) under
Contract Number EF-77-C—01-2609, and co—sponsored by the Research Branch of the
United States Environmental Protection Agency (EPA), under inter-agency
Agreement Number IAG-D7-E681. The program is directed by an ABMA Stoker Techni-
cal Committee which, in turn, has subcontracted the field testing to KVB, Inc.,
of Minneapolis, Minnesota.
SITE TESTING
The test program involved testing eleven coal-fired stoker boilers. The
boilers ranged in size from 13 to 87 megawatts thermal output (45-300 K lb
steam/hr). Six were spreaders, one was a vibragrate, and four were overfeed
traveling grate and chain grate units. Most of the boilers were new "state-
of-the-art" designs.
67
-------
Extensive testing has been completed on all eleven boilers listed in the
table below. The ownership and location of these boilers is being treated
confidentially.
LISTING OP UNITS TESTED
Site
Code
Boiler*
Stoker
No.
Coals
Arrangement of
Heat Trap
A
87
MW
PW
Detroit Spreader
3
Econ. after ESP
B
61
MW
Riley
Riley Spreader
4
Econ. after DC
C
59
MW
BSW
Detroit Spreader
3
Econ. after DC
D
25
MW
B&W
Detroit Vibragrate
3
Econ. before DC
E
55
MW
Riley
Riley Spreader
3
Econ. before DC
F
24
MW
Keeler
Detroit Spreader
2
Econ. before DC
G
22
MW
Zurn
Zurn Spreader
3
No economizer
H
13
MW
Bros
Riley Traveling Grate
1
No economizer
I
21
MW
Wickes
Riley Traveling Grate
2
No economizer
J
21
MW
Keeler
Laclede Chain Grate
2
Econ. before DC
K
15
MW
Riley
Riley Traveling Grate
2
Econ. before DC
*MW is megawatts thermal output;
MWtx3.4 = lO^tu/hr = 103lb/hr Steam
The boilers were tested for particulate and gaseous emissions and for
boiler efficiency over a wide range of firing conditions. Particulate mass
loading was determined at the boiler outlet and after the mechanical dust
collector (if any). Other tests included, but were not limited to, 02, CO2*
CO, NO, S02, SO3 and particle size distribution. Percent coal fines was deter-
mined for most tests, and coal and ash samples were collected for chemical
analysis.
68
-------
SECTION 2
TEST RESULTS
Comprehensive results of this test program are being published by the
Environmental Protection Agency (EPA) and are available through the EPA at
Research Triangle Park, or through NTIS (1-8). This paper deals with nitric
oxide and particulate emissions, and with the combustible content of bottom ash
and flyash.
PARTICULATE LOADING
A total of 291 particulate mass loading determinations were made in field
tests on the eleven stoker boilers. Most of these tests were simultaneous
measurements made before and after the mechanical dust collector. From this
large data base it is possible to make several generalizations about the vari-
ables affecting particulate loading.
Stoker Design
Stoker design is a variable which had a major impact on particulate load-
ing. For example, spreader stokers emitted five times more particulate matter
at the boiler outlet than did overfeed stokers. Also, spreader stokers with
flyash reinjection emitted as much as four times more particulate matter than
spreader stokers without reinjection, when measured at the boiler outlet.
These results were true regardless of the size of the unit.
The range of particulate loadings measured at the boiler outlet under
full load conditions were as follows:
Boiler Outlet Particulate Loading
Spreader's with Reinjection
Spreader's w/o Reinjection
Overfeed's w/o Reinjection
4300-15500 ng/J (10-36 li>/106Btu)
1200- 3100 ng/J ( 3-7 lb/106Btu)
260- 950 ng/J <0.6-2 lb/106Btu)
69
-------
These differences were not as distinct downstream of the mechanical dust
collector. Collector efficiency, which is sensitive to particle size distri-
bution and inlet loading, was highest for spreader stokers with flyash rein-
jection and lowest for overfeed stokers. The range of particulate loadings
measured downstream of the mechanical dust collector under full load conditions
were as follows:
Boiler Loading and Excess Air
The particulate loading at the boiler outlet always increased as load in-
creased, sometimes doubling between loads of 50% and 100% of capacity.
The boiler outlet particulate loading also increased when excess air was
increased. The data base for this observation are few and the magnitude of
the increased particulate loading has not yet been quantified.
Overfire Air
The data indicate that high overfire air flow rates, representing 10% to
20% of the total combustion air, are best for minimizing particulate loading.
When overfire air flows were increased from their lowest non-smoking flow rates
to their maximum flow rates, particulate loadings were observed to either de-
crease or remain unchanged.
D.C. Outlet Particulate Loading
Spreaders with Reinjection
Spreaders w/o Reinjection
Overfeeds w/o Reinjection
200-470 ng/J (.47-1.1 lb/106Btu)
70-600 ng/J (.17-1.4 lb/106Btu)
50-320 ng/J (.11-.75 lb/106Btu)
DROP IN PARTICULATE LOADING DUE TO INCREASED OVERFIRE AIR
Spreader Stokers
Overfeed Stokers
Site A: 25-50%
Site B: 25%
Site D: No Change
Site H: 50%
Site C: No Change
Site E: No Change
Site F: 35%
Site I: 40%
Site J: No Change
Site K: No Change
Site G: No Change
70
-------
Coal Properties
Although several coals were test fired at each site, with only one ex-
ception, their varying chemical properties had little or no effect on particu-
late loading. Even at Test Site C where an Eastern coal and a Western coal
were fired in the same boiler, no change in particulate loadings was observed.
It is recognized that in most cases the coals chemical properties were very
similar and that these results apply only to the specific coals tested.
Physical coal properties did make a difference. An increase in the per-
centage of fines in the coal, at Site K, resulting from passing a washed
Oxl-1/4" coal through a crusher set at 3/4", increased the particulate loading
by 60% at full load. Hie same coal unwashed and reported to have a high clay
content gave a 180% higher particulate loading than when washed, ftiese data
are graphically presented in Figure 1.
NITRIC OXIDE
Nitric oxide levels were measured for 389 test conditions during the test
program. From this large data base, several correlations can be made between
nitric oxide concentration and the test variables.
Stoker Design
Spreader stokers emitted significantly higher levels of nitric oxide than
did overfeed stokers at the same load and excess oxygen. This point is clearly
illustrated in Figure 2 where nitric oxide levels are 1-1/2 to 2 times higher
for spreaders. Even when correction is made for the fact that overfeed stokers
are operated at higher excess oxygen levels than spreader stokers, the spreaders
emit 50% more nitric oxide per MMBtu.
Average Full Load o?
Average NOx
206 ng/J (0.48 lb/106Btu)
133 ng/J (0.31 lb/106Btu)
Spreader Stokers
Overfeed Stokers
6.4%
7.9%
71
-------
Boiler Loading and Excess Oxygen
At constant load, a one-percent increase in excess oxygen resulted in an
average .048 lb NC>2/106Btu increase in NQx. This slope was true for both
spreader stokers and overfeed stokers as illustrated in Figure 2.
At constant excess oxygen, nitric oxide increased with increasing load.
This relationship has not yet been quantified, but it is observed that the
effect is greater on spreader stokers than on overfeed stokers.
Since excess oxygen decreases as load increases under normal stoker
operation, these two influences tend to cancel each other out. Therefore,
under normal operating conditions, nitric oxide concentrations are invarient
with load.
Overfire Air
In general, changes in overfire air operation had no effect on nitric
oxide concentrations. Specific units tested showed small increases or de-
creases in nitric oxide as the overfire air flow was altered, but no consis-
tent trend was established.
Flyash Reinjection
Flyash reinjection did not alter the nitric oxide concentrations. lhis
was observed on all three of the spreader stokers which reinjected flyash from
their mechanical dust collectors.
Coal Properties
Nitric oxide concentration was invarient with coal properties in these
tests. The one exception to this rule, a 36% reduction in nitric oxide when
switching coals at Site I, is suspect.
Although many of the coals tested were very similar, a few do stand out
as true tests. For exanple, at Site C a Western coal and an Eastern coal with
significantly different chemical composition were test fired in the same boiler.
At Site K, coal fines and clay content were major variables. In both these
case studies, nitric oxide concentration remained the same.
72
-------
COMBUSTIBLES IN THE ASH
Contoustibles in the ash are important since they are related to boiler
efficiency- They are of special concern in stoker boilers where they result
in heat losses often exceeding 5% of the unit's heat input.
A total of 511 ash samples were collected and examined for their com-
bustible content in the course of this test program. These samples included
bottom ash, flyash from before and after the mechanical dust collector, and
flyash from the mechanical dust collector hopper.
Stoker Design
The magnitude of the combustibles in the ash was found to correlate with
stoker type. For example, spreader stokers had less combustible material in
their bottom ash than did overfeed stokers. However, spreader stokers had more
combustible material in their flyash than did overfeed stokers. Figures 3 and
4 illustrate this correlation.
The net result was that spreader stokers without flyash reinjection had
combustible heat losses which were several percent greater than those for over-
feed stokers. With flyash reinjection, some of this loss is recovered, but not
You will notice in Figure 4 that combustibles in the flyash of Site C
were considerably lower than they were in the flyash from the other spreader
stokers. The reason for this low combustible level has not been established,
but it is believed to be related to boiler design parameters.
The boiler at Site C was designed with a very low heat release rate of
less than 500,000 Btu/hr-ft2 effective grate area. This is considerably lower
than the 600,000 to 850,000 Btu/hr-ft2 of the other spreader stokers tested.
The Site C heat release per front foot of grate width was also lowest in the
all.
General Combustible
Heat Loss Range
Spreader with Reinjection
Spreader w/o Reinjection
Overfeed w/o Reinjection
2-6%
4-7%
2-4%
73
-------
group. These two design parameters may be at least partially responsible for
the low combustible levels at Site C.
Boiler Loading and Excess Air
Combustible levels were relatively invarient with load. In general they
showed a slight tendency to increase as load increased, but the opposite trend
was also observed in a few isolated cases.
The effect of excess air on combustibles has not been fully examined, but
preliminary indications are that this variable had little effect on combustible
levels as long as excess air was maintained within a normal, non-smoking range.
Overfire Air
Increasing the overfire air pressure has a tendency to reduce the com-
bustible content of the flyash. This reduction is presumably due to improved
fuel-air mixing in the flame zone.
Our examination of the relationship between overfire air and combustibles
in the bottom ash is not complete. However, it appears that conbustibles in
the bottom ash increased at a few sites when overfire air flow was increased.
Flyash Reinjection
Combustible levels in the bottom ash and the flyash were unchanged by fly-
ash reinjection status.
Coal Properties
Conbustibles in the ash showed a correlation with coal properties at only
one site in this program. At Test Site C, where an Eastern and a Western coal
were fired in the same boiler, the Western coal had the highest percentage of
combustibles in its bottom ash and the lowest percentage of conbustibles in its
flyash.
Combustibles vs Size
The combustible content of flyash is a function of the size of the flyash
particles. The largest particles contain the highest percentage of conbustible
matter.
74
-------
Although not a part of the original scope of this program, a limited in-
vestigation was undertaken to quantify this relationship. Selected samples of
flyash were sieved into several size fractions and analyzed for their com-
bustible content.
The results are plotted in Figure 5. The two plots shown relate com-
bustible content to partice size for two unrelated boilers and two different
ash sample locations. Each plot presents the average of data from four samples.
They have one thing in common, the larger particles contain the highest com-
bustible content.
Since mechanical dust collectors remove the largest particles and allow
the smallest particles to pass on through, one would eiqpect the exiting parti-
cles to contain a lower combustible content. Test data from the six spreader
stokers verifies this assumption and illustrates the magnitude of the com-
bustible drop.
Percent Confoustibles in Flyash
D.C. Inlet
D.C. Outlet
Site A
Site B
Site C
Site E
Site F
Site G
58%
60
25
66
67
53
30%
29
16
52
47
32
75
-------
REFERENCES
Gabrielson, J. E., P. L. Langsjoen, and T. C. Kosvic. Field Tests of
Industrial Stoker Coal-Fired Boilers for Emissions Control and
Efficiency Improvement - Site A. EPA-600/7-78-136a, U.S. Environmental
Protection Agency, Research Triangle Park, NC, July, 1978, 106 pp.
Gabrielson, J. E., P. L. Langsjoen, and T. C. Kosvic. Field Tests of
Industrial Stoker Coal-Fired Boilers for Emissions Control and
Efficiency Improvement - Site B. EPA-600/7-79-041a, U.S. Environmental
Protection Agency, Research Triangle Park, NC, February, 1979. 113 pp.
Gabrielson, J. E., P. L. Langsjoen, and T. C. Kosvic. Field Tests of
Industrial Stoker Coal-Fired Boilers for Emissions Control and
Efficiency Improvement - Site C. EPA-600/7-79-130a, U.S. Environmental
Protection Agency, Research Triangle Park, NC, May, 1979. 138 pp.
Gabrielson, J. E., P. L. Langsjoen, and T. C. Kosvic. Field Tests of
Industrial Stoker Coal-Fired Boilers for Emissions Control and
Efficiency Improvement - Site D. EPA-600/7-79-237a, U.S. Environmental
Protection Agency, Research Triangle Park, NC, November, 1979. 115 pp.
Langsjoen, P. L., J. 0. Burlingame, and J. E. Gabrielson. Field Tests of
Industrial Stoker Coal-Fired Boilers for Emissions Control and Efficiency
Improvement - Site E. EPA-600/7-80-064a, U. S. Environmental Protection
Agency, Research Triangle Park, NC, March, 1980. 102 pp.
Langsjoen, P. L., J. 0. Burlingame, and J. E. Gabrielson. Field Tests of
Industrial Stoker Coal-Fired Boilers for Emissions Control and Efficiency
Improvement - Site F. EPA-600/7-80-065a, U. S. Environmental Protection
Agency, NC, March, 1980. 113 pp.
Langsjoen, P. L., J. 0. Burlingame, and J. E. Gabrielson. Field Tests of
Industrial Stoker Coal-Fired Boilers for Emissions Control and Efficiency
Improvement - Site G. EPA-600/7-80-082a, U. S. Environmental Protection
Agency, Research Triangle Park, NC, April, 1980. 114 pp.
Langsjoen, P. L., R. J. Tidona, and J. E. Gabrielson. Field Tests of
Industrial Stoker Coal-Fired Boilers for Emissions Control and Efficiency
Inprovement - Site H. EPA-600/7-80-112a, U. S. Environmental Protection
Agency, NC, May, 1980. 90 pp.
76
-------
9. Langsjoen, P. L., J. E. Gabrielson, W. H. Axtman. Test Results of
Modern Coal-Fired Stoker Boilers for Emissions and Efficiency. In:
Proceedings of the American Power Conference, Volume 41, Chicago, Illinois,
1979. pp. 406-410.
10. Langsjoen, P. L. Characterization of Emissions from Industrial Stoker
Boilers. In: Proceedings of the Air Pollution Control Association Con-
ference on Industrial Boilers, Research Triangle Park, NC, 1979.
11. Mosher, R. N. A Testing Program on Modern Stoker-Fired Boilers for
Emissions and Efficiency. In: Proceedings of the Symposium on Industrial
Coal Utilization, Charleston, South Carolina, 1980.
77
-------
1000
800
cn
C
o
z
»—I
o
<
o
LU
(—
*c
cc
-------
300
SPREADER STOKERS
OVERFEED STOKERS
250
® 200
LU
Q
X
o
o
t—i
cc
150
100
50
6 8
EXCESS OXYGEN, % (DRY)
10
12
Figure 2. Relationship Between Nitric Oxide Concentration and
Excess Oxygen at Maximum Load for Eleven Stoker
Boilers.
79
-------
100
80 _
SPREADER
STOKERS
OVERFEED
STOKERS
o
60
40
20
HiSs
" I III
I I I I
ABCEF6 DHIJK
TEST SITE CODE
Figure 3. Hie Average and Standard Deviation of the Mass
Percentage of Combustible Material in the
Bottom Ash of Eleven Stoker Boilers.
SO
-------
100
80
t/)
<
>-
60
UJ
_l
CD
i—i
H-
oo
3
CO
| 40
UJ
o
a:
LU
a.
20
0
SPREADER
STOKERS
A
A B C E
OVERFEED
STOKERS
}H5
I I I I I I I I I I I
F G D H I J K
TEST SITE CODE
Figure 4. The Average and Standard Deviation of the Mass
Percentage of Combustible Material in the Un-
controlled Flyash of Eleven Stoker Boilers.
81
-------
100
FLYASH COLLECTED AT THE
BOILER OUTLET OF SITE B
CO
O 40
FLYASH COLLECTED FROM
THE BOILER HOPPER
OF SITE D.
o
cc
a.
I I I
I
40 100 300 1000 3000
PARTICLE DIAMETER, MICROMETERS
Figure 5. Relationship Between Particle Size and Combustible
Content of Flyash from Two Boilers.
82
-------
COMBUSTION MODIFICATION FOR COAL-FIRED
STOKER BOILERS
By:
K. L. Maloney, K. F. Maloney, and M. J. Pfefferle
KVB, Inc.
18006 Skypark Boulevard
Irvine, California 92714
83
-------
ABSTRACT
Preliminary results from a program to develop and assess advanced
combustion modification concepts for coal-fired stoker boilers are
presented. Tests on a 100,000-lb/hr steam spreader stoker boiler showed that
overfire air reduces smoke emissions only when injected in a zone extending a
few feet above the fuel bed. Improved overfire air design can permit lower
excess 02 firing for N0X control, while maintaining acceptable smoke and CO
emissions. Staged combustion was applied to a laboratory underfeed
stoker (~240,000 Btu/hr heat input) to reduce N0>x emissions.
84
-------
ACKNOWLEDGMENTS
The authors would like to thank program monitor John H. Wasser of EPA's
Combustion Research Branch for his assistance and guidance* They would also
like to extend their appreciation to Jim Maloney of the State of Wisconsin and
Stan Novotny of the University of Wisconsin for their help in arranging and
performing the field tests.
85
-------
SECTION 1
INTRODUCTION
The restricted availability of natural gas and oil fuel has resulted in
the increased use of coal-fired stoker boilers for industrial applications.
However, the widespread use of stokers could be limited by environmental
constraints if methods are not developed to reduce their emissions. This
situation has created the need for a stoker emissions technology program that
will develop combustion modification methods for minimizing stoker gaseous
emissions, characterize the influence of these modifications on particulate
properties which are important in collector design and optimization, and
assess the ability of new stokers to comply with present and anticipated
pollution regulations. There also is a need for close coordination of this
technology development with the stoker industry to maximize technology trans-
fer.
EPA has funded such a program, under EPA Contract No. 68-02-3166,
"Application and Assessment of Combustion Modification Concepts for Full-Scale
Stoker Coal-Fired Boilers." The major objectives of this recently-begun
program are listed below:
1. Establish combustion modification techniques that are
practical and effective on a stoker boiler.
2. Characterize the dependence of particulate and other emissions
on coal type, coal size, ash reinjection rate, degree of over-
fire air, load, excess O2, etc.
3. Determine to what degree the "optimized" stoker emissions are
dependent upon coal type.
86
-------
4. Examine the relationship between stoker design and emissions
and determine what latitude exists in burning other than the
"design" coal in existing field units.
5. Establish desirable coal characteristics and the requirements
for practical use by stokers.
6. Examine the feasibility of burning cleaned or processed coals.
7. Perform Level I emission assessment of stoker-fired boilers,
with Level II assessment as necessary.
8. Determine the impact of modified combustion techniques on fly
ash collector efficiency and design.
9. Assess the potential for emissions control and overall
environmental acceptability of industrial stoker boilers.
10. Prepare a guideline document for use in the industrial and
commercial application of successful combustion modification
concepts.
Phase I of the program involves tests on two industrial-sized spreader
stokers (100,000 and 300,000 lb/hr steam). Testing on the smaller unit is
underway and is described in Section 2. When testing on the larger unit is
completed, the results will be examined to see how well the various modifica-
tion techniques scale up.
Three institutional-size moving grate and underfeed stokers will be
tested in Phase II. Testing of the first underfeed stoker is currently under-
way in KVB's laboratory and is discussed in Section 3. This unit will also be
used for small-scale tests on advanced combustion concepts that will be used
on the larger boilers in the program if successful. The results from each
phase will be reviewed by EPA and a technology transfer committee, consisting
of representatives from the coal and stoker industries, to decide whether the
study has established significant improvements in environmental status for
each class of stoker. If so, KVB will prepare a guideline document that will
detail the methods and procedures for application of the successful combustion
modification concepts.
87
-------
Some of the combustion modifications to be considered are:
1.
Low excess air (LEA) firing
2.
Steam injection through overfire air (OFA)
jets
3.
Undergrate air humidification
4.
Undergrate air redistribution
5.
Improved OFA addition
6.
Addition of recycled flue gas through OFA
jets
7.
Biased bed loading
8.
Clean coal firing
Throughout the study, the results will be evaluated with respect to these
questions:
1. How dependent are stoker emissions on the coal type, stoker
design, stoker size, and operating parameters?
2. What coal characteristics are desirable for stoker fuel use,
and what alternative reconstituted and processed coals are
suitable for significantly reducing pollutants?
3. What combustion modifications are practical and effective on a
stoker boiler, and what methods are best?
4. How boiler-design-dependent are stoker emissions, and what
latitude exists in burning other than the "design" coal?
5. How can coal size, ash reinjection rate, and overfire air
configuration be optimized to minimize emissions?
6. What are the trade-offs between emissions and boiler
performance for key design parameters?
7. How dependent are particulate size and loading on combustion
parameters or combustion modification method, and how do these
affect collector design?
8. Are POM, PCB, and other organic emissions significant from
fuel-rich stoker beds?
9. What trace elements and other pollutants may be of concern
based on Level I and limited Level II screening tests? To what
degree are they influenced by combustion modifications?
88
-------
A key feature of the program is the creation of an industrial technical
panel, mentioned above, to review program progress and provide practical
advice on technical objectives. Panel members will be drawn from stoker
manufacturers, the coal industry, and plant managers at the respective test
sites. This panel will help the stoker industry make better use of new design
information generated during the study.
This paper presents preliminary results from the field and laboratory
investigations, which have just started. In Section 2, the spreader stoker
field tests will be discussed. In Section 3, the underfeed laboratory tests
will be discussed. Section 4 will summarize the work to date.
89
-------
SECTION 2
FIELD TESTS
FACILITY DESCRIPTION
Field tests have recently begun on a 100,000-lb/hr steam flow spreader
stoker boiler at the University of Wisconsin, Madison, Central Heating
Plant. The boiler (Unit No. 1) was constructed by Babcock & Wilcox in 1952,
and is fired with three Westinghouse Centrafire spreader stokers with a front-
discharge traveling grate, shown in Figure 1. It is a balanced draft unit,
firing western Kentucky bituminous coal from the Fies mine. Other specifica-
tions are as follows:
• Design pressure - 4.9 MPa (700 psig)
• Steam temperature - 656°K (720°F)
• Actual day-to-day operating pressure - 4.2 MPa (600 psig)
• Boiler heating surface - 131.18 m2 (1412 sq ft)
• Economizer heating surface - 518.12 m2 (5577 sq ft)
• Stoker grate size - 4.038 m x 4.94 m (13* 3" x 16' 2-1/2")
2
• Stoker grate area - 19.96 m (214.8 sq ft)
Overfire air is supplied by a separate fan and injected at a pressure of
16 inches HjO through three banks of 13 nozzles, each 1.5 inches in diame-
ter. The lower rear bank is about one foot above the grate; the upper rear
bank is about 4.5 feet above the grate, directed about 15° down from the
horizontal; and the front bank is about two feet above the grate, just below
the stokers.
90
-------
A mobile laboratory has been constructed for the emissions analysis.
This laboratory is housed in a 45-foot semi-trailer and includes the following
instrumentation s
• NO/NO — Teco Model 10A
X
• C>2 — Beckman Model F3M3-1AA
• CO, C02 — Beckman Models 315B
• SC>2 — Dupont Model 400
• HC — Carle Model 211 GC.
In addition, the laboratory contains auxiliary equipment to support EPA
Method 5 and SASS sampling.
TEST RESULTS
During the tests described here, the fuel burned was not the normal Fies
coal but a mixture of the Fies coal and other coal bought 10 years ago during
a mine strike. Due to weathering and a recent fire in the coal pile, the coal
was much higher in fines than normal stoker coal, precluding full-load opera-
tion. These factors limited the testing to the examination, described below,
of the effectiveness of overfire air.
To assess the effect of the individual OFA banks, each bank was turned
off separately and in combination with other banks. Smoke was monitored with
the plant's optical opacity meter. (Each of the five units in this plant has
its own opacity meter.)
The result was a dramatic demonstration of the effectiveness of the lower
rear OFA bank. Any combination of active banks including the lower rear kept
the opacity below five percent, while any combination without air through the
lower rear OFA jets resulted in opacities of 45 percent or greater.
DISCUSSION
These results seem to indicate that the OFA is effective only within a
region a few feet above the bed, as the front and upper rear OFA jets are two
91
-------
and five feet above the grate. Hie rising combustion products keep the air
jet from penetrating across the grate.
It is interesting that one bank of jets on the back wall should have such
a profound effect on smoke emissions since viscous flow theory predicts that
the centerline jet velocity will decrease to less than 20 percent of its
original value within 30 nozzle diameters. For the Madison unit, 30 nozzle
diameters is just 3.75 feet. Thus, the rear OFA jets would appear to cover
less than 25 percent of the grate area before their velocity diminishes to
insignificance. Yet this is enough to make the difference between a clean
stack and opacities of 45 percent or greater.
Two engineers entered the furnace of a twin unit, down at the time, to
get an idea of how quickly the OFA velocity decayed. They found that the jet
velocity had diminished to a barely-discernible breeze four feet from the
nozzles. Turning off two banks increased the velocity of the remaining bank
slightly, but had no measurable effect on the penetration distance.
FUTURE TESTS
A modified OFA system is currently being constructed. It will consist of
three tubes mounted crosswise across and roughly two feet above the grate,
drilled with holes similar to the existing OFA jets. These invasion pipes
will be almost evenly spaced (they will be inserted through observation
hatches) and should cover the grate much better than the existing OFA
system. The significance of improved OFA with respect to NOx reduction is the
possibility of reducing excess air while retaining acceptable CO and smoke
levels. Steam injection will also be tried with the invasion pipes. The
steam should improve carbon burn-out as well as provide improved mixing.
Another test currently underway is examining whether combustion is
uniform over the grate. Eight probes have been installed at the boiler exit
plane, and gas samples from these probes will be used to spot regions of
abnormal CO or Oj on the bed, assuming essentially laminar flow through the
boiler. The results can be used to balance the three stokers to achieve more
uniform combustion conditions.
92
-------
SECTION 3
FACILITY DESCRIPTION
FACILITY DESCRIPTION
To provide a small-scale test bed for advanced combustion modification
concepts and new fuel evaluations, a single-retort underfeed stoker burner was
constructed in KVB's Santa Ana laboratory. The unit was designed to be as
versatile as possible and burns about 20 lb coal/hr (about 240,000 Btu/hr heat
input). A schematic of the stoker is shown in Figure 2; it is equipped with a
lower overfire air torus, comparable to normal overfire air on a full-scale
stoker, and a variable-height upper overfire air torus for staged combustion
concept tests. Recycled flue gas can be mixed with the overfire air, and
different mixtures can be injected in each torus. Steam (50 psig) can also be
injected in each torus and in the undergrate air.
The firebox is lined with refractory. (Removable boiler tubes have also
been fabricated to vary the radiant section's heat removal rate, but these
were not used in the experiments described here.) The flue gas is cooled by a
water-jacketed convective section. Flue gas can be recycled from both before
and after the convective section and mixed to control the temperature of the
recycled flue gas.
Emissions of NO, NOx, SOg/ CO, COg/ 0j, and smoke are monitored by the
laboratory's instrumentation system, which has been described elsewhere.* For
the tests described here, smoke was measured with a Bacharach pump and
~Muzio, L. J., J. K. Arand, and K. L. Maloney, "Noncatalytic NOx Removal
with Ammonia," EPRI Report FP-735, April, 1978.
93
-------
reported on the Bacharach oil scale while an optical opacity meter was being
installed.
TESTS
In the first tests, the height of the upper air torus was varied over its
full range (0 to 34 inches above the coal bed) to examine the effect on NO
emissions. Kentucky bituminous coal was the fuel (the coal analysis is shown
in Table I), and no flue gas was added to either overfire air stream. In the
unmodified baseline condition the NOx emissions were in good agreement with
the field test results from the Madison boiler at 350 ppm.
With an upper/lower overfire air flow ratio of 2, the NOx emissions were
minimized at 176 ppm (3% 02, dry) when the torus was 28 inches above the
grate. NOx emissions peaked at 394 ppm when the torus was 12 inches above the
bed. As shown in Figure 3/ the 02 roughly followed the NOx curve, peaking at
10 percent with the torus 8 inches above the bed. The variation in 02 was
attributed to changing bed conditions. It was noted that clinker formation
caused the 02 to rise one or two percent at otherwise steady conditions.
Carbon monoxide and smoke readings are shown with 02 in Figure 4. The CO
rose slightly as the upper torus was moved from directly above the lower torus
to 20 inches above the bed. The CO then rose quickly from about 90 ppm (3%
02, dry) to 350 ppm at the highest torus setting. At the lowest NOx point,
the CO was a still-reasonable 150 ppm. The smoke posed a problem, however.
In the low-NOx range, the smoke number was above 8 on the Bacharach oil
scale. At the lower torus settings, the smoke number was highly variable,
jumping back and forth between 2.5 and 9. This was felt to be a bed effect;
the addition of a continuous opacity monitor should provide insight as to the
relation between bed condition and smoke emissions.
DISCUSSION
The effectiveness of staged combustion in reducing NO emissions can be
seen by comparing Figure 3 with Figure 5. Figure 5 shows NO emissions vs 02
without staging (i.e., only the lower OFA torus was activated). Reducing the
02 from 10 percent to 6 percent reduced the NO by 25 percent. From Figure 3,
94
-------
the same change in C>2 reduces the N0X by 50 percent. (In Figure 3, the NO2
generally ranged from 0 to 15 ppm). The unstaged tests were run burning a low
sulfur western subbituminous coal; however, the NO emissions are similar for
the unstaged point in Figure 3 (i.e., the point at which the upper torus is
directly above the lower torus) and the comparable point in Figure 5.
An interesting, and unresolved, question is: why does the NOx peak as
the upper torus is raised?
For these tests, accurate measurements of the undergrate air or coal flow
were not available, so it was not possible to calculate fuel/air ratios in the
different stages. Future tests will provide this information.
This work will continue with tests of steam and recycled flue gas in the
OFA jets, comparison of eastern and western coals, and specie and temperature
mapping to analyze the staging process.
95
-------
SECTION 4
SUMMARY AND CONCLUSIONS
Preliminary studies have been performed to date on a 100,000-lb/hr steam
flow spreader stoker boiler and a 240,000-Btu(t)/hr underfeed stoker
furnace. The effectiveness of the various sets of overfire air jets on the
spreader stoker has been investigated, and staged combustion has been imple-
mented on the underfeed stoker.
These tentative conclusions have been drawn:
1. The effectiveness of overfire air jets seems confined to a zone
extending only two or three feet above the fuel bed. This
indicates that currently used overfire air flow rates may be
excessive and that lower excess air firing for NOx control can
be achieved by optimized OFA design.
2. Staged combustion can be an effective technique for lowering
NOx emissions. The major contribution to NOx emission is fuel
nitrogen; running the coal bed more fuel-rich inhibits conver-
sion of fuel N to NOx.
96
-------
C0UAL1IER
SPILL PLATE
CONSTANT
VELOCITY
PAIRED RAMS
1
IR COOLED DISTRIBUTING ROTOR
SILL NOZZLES
FEED RAM
HYDRAULIC
ENGINE
2
Figure 1. Westinghouse Centrafire spreader stoker with traveling grate.
-------
TABLE I. COAL PROPERTIES - KENTUCKY BITUMINOUS
Proximate Analysis
% Moisture 3.85
% Ash 8.97
% Volatile 36.31
% Fixed Carbon 50.87
HHV (Btu/lb) 12,698
Ultimate Analysis (% wt)
Moisture 3.85
Carbon 71.31
Hydrogen 4.79
Nitrogen 1.29
Sulfur 1.01
Ash 8.97
Oxygen 8.74
98
-------
THIRTY-DAY FIELD TESTS OP INDUSTRIAL BOILER
COMBUSTION MODIFICATIONS
By:
W. A. Carter
KVB, Inc.
18066 Skypark Boulevard
Irvine, California 92714
99
-------
ABSTRACT
This paper is based on a field test program sponsored by EPA to evaluate
the long-term effectiveness of combustion modifications for reducing NO^
emissions from industrial boilers. Five 30-day field tests have been con-
ducted so far. The combustion modifications evaluated include low excess air
on a coal-fired spreader stoker, staged combustion air on a residual-oil-fired
boiler, staged combustion air and staged combustion on a pulverized-coal-fired
boiler, low excess air and staged combustion air on a spreader stoker, and a
gas-fired low N0X burner. Reductions in N0X varied from 15 percent with the
pulverized-coal-fired boiler to approximately 70 percent with the gas-fired
low NOx burners. No serious operational or reliability problems were
encountered, and most units demonstrated an increase in boiler efficiency.
100
-------
THIRTY-DAY FIELD TESTS OF INDUSTRIAL
BOILER COMBUSTION MODIFICATIONS
INTRODUCTION
A field test program is being sponsored by EPA's Combustion Research
Branch and their Office of Air Quality Planning and Standards to determine
whether combustion modification techniques which demonstrated reductions on
air pollutant emissions (particularly oxides of nitrogen) during
short-duration tests are feasible over longer periods. In addition, boiler
performance and reliability are being monitored. KVB, Inc., was selected to
conduct this field test program with the purpose of providing data to be used
by EPA in the preparation of a new source performance standard for N0X
emissions from industrial boilers.
The program scope requires testing of seven industrial boilers ranging in
capacity from 50,000 to 250,000 lb steam/hr. Fuels include natural gas,
distillate and residual oil, and coal. NOx control technologies to be evalu-
ated are low excess air, staged combustion air, low NOx burners, and flue gas
recirculation.
Continuous measurements of gaseous emissions (NO, CO, CO2 and Og) were
made with a certified monitor fabricated by KVB. Particulate emissions were
measured in accordance with EPA Reference Method 5.
Five 30-day field tests have been completed. Equipment tested includes
two spreader stokers, a pulverized-coal-fired boiler, a residual-oil-fired
boiler, and a natural-gas-fired boiler.
101
-------
TEST PROCEDURES
The NSPS for industrial boilers may include emissions standards averaged
over a time period as long as 24 hours. A source testing program will be more
supportive of the standard if it utilizes techniques which acquire data over a
long time period or continuously.
The accepted EPA Reference Methods for determining the emission rates of
particulate and N0X are Methods -5 and 7, respectively. Each of these methods
extracts pollutant samples over a short time period. Method 5 will extract a
sample for 60 minutes. Method 7 will extract a sample in less than a
minute. These methods result in source emission rates that are averaged over
short time periods. A measurement of emission rates averaged over longer time
periods had to be obtained in the past by conducting multiple Reference Method
tests. For gaseous pollutants, this approach can be improved upon by
utilizing the continuous monitoring system methodology.
Using a continuous monitoring system (CMS) to collect data in support of
a standard will be more defensible if those data can be correlated with an EPA
Reference Method. The mechanism for doing this is provided within Performance
Specification 2 (PS2), 40 CFR 60, Appendix B. In PS2 criteria are set for
installing and testing a NOx or SOj continuous monitoring system. PS2 also
establishes minimum performance specifications that the system must meet in
terms of eight parameters: accuracy, calibration error, 2- and 24-hour zero
drifts, 2- and 24-hour calibration drifts, response time, and operational
period. The accuracy (relative) parameter compares data from the CMS to data
obtained via an EPA Reference Method test. CMS and Reference Method tests are
run simultaneously, with at least 27 tests for NOx being run. The accuracy is
the mean difference of the two measurement techniques plus a 95 percent con-
fidence interval. This value must be within 20 percent of the EPA Reference
Method. Therefore, each performance evaluation of a CMS conducted according
to PS2 serves the dual purpose of insuring that the equipment is operating
properly and providing a means for correlating the data with an EPA Reference
Method.
102
-------
The data collected from a continuous monitoring system over a 30-day test
period should adequately represent the emission rate that would have been
derived from Reference Method testing if the following criteria are met:
1. The CMS is installed in accordance with Performance
Specification 2.
2. The CMS passes a complete performance evaluation at the
beginning of the test period.
3. The CMS passes the relative accuracy portion of the
performance evaluation at the end of the 30-day test period.
This assumes that daily span gas checks are performed and that there are no
replacements of major components.
General
All boiler operating conditions were recorded including fuel feed rate,
steam production rate, and excess oxygen. This was done via hourly operation
logs and periodic recordings by a technician on site.
The nitrogen (N), sulfur (S), and ash contents, as well as heating value,
were determined for oil and coal samples. For coal, the state, county, bed,
seam, and mine from which the coal was obtained were provided.
Because of coal nitrogen and sulfur variability, coal samples were
collected periodically during the day and composited into a single sample for
analysis. Sample collection and size reduction were conducted according to
American Society of Testing and Materials methods. Samples were analyzed
periodically for ultimate analysis, ash analysis, ash fusion temperature, and
any other properties which may be essential to boiler performance evaluation.
For oil, the number of samples analyzed was significantly lower.
Analyses were necessary only to characterize each batch (tankful) of fuel
burned. The American Petroleum Institute gravity of the oil samples was also
determined. All testing included monitoring for C02 and 02 so that the data
collected on particulate and N0X could be converted to any of the possible
units of a standard. Visible emission readings (Reference Method 9, 40 CFR 60
Appendix A) were taken at least every time a Method 5 or 7 was performed.
When particulate source testing was performed, a Method 9 was completed as
103
-------
often as was reasonable considering manpower and plume variability. One set
of measurements for polycyclic organic matter was obtained.
Particulate Testing
There are no EPA performance specifications for continuous particulate
monitoring equipment. Therefore, particulate testing was performed by repeti-
tion of Reference Method 5. Each day's testing consisted of triplicate runs
of the Reference Method. Initially three triplicate runs, a total of 9 Method
5's, were conducted. The triplicate sets were conducted at the start, mid-
point, and conclusion of the 30-day test. After the second test site, it was
decided to drop the triplicate series at the midpoint of the 30 days because
of the cost involved. Particulate testing was not conducted at the gas-fired
boiler test site.
Gaseous Pollutants (NOY and CO)
A continuous monitor system which met PS2 and PS3 specifications was used
as the source test method for NO„ and CO. The source tests consisted of 30
days of continuous monitoring. The NOx control technologies employed con-
sisted of combustion modifications which allow for a single monitoring point.
Complete PS2 and 3 tests were performed at the beginning of each 30-day
test period. The relative accuracy portion of the performance evaluation was
repeated at the end of the 30-day test period. In addition, an abbreviated
relative accuracy test, consisting of PDS flasks, was conducted at the mid-
point of the 30-day test. A chronology of a typical 30-day test is shown in
Figure 1.
Instrumentation
The continuous monitor shown in Figure 2 was equipped with analytical
instruments to measure concentrations of NO, CO, CO2, and O2• Gaseous mea-
surements were made with the analytical instruments listed in Table I. The
sample gas is delivered to the analyzers at the proper condition and flow rate
through a sampling and conditioning system. A stainless steel probe and
sintered stainless steel filter were installed in the stack to sample the flue
gas. The CMS is equipped with three two-pen strip chart recorders for
104
-------
continuous recording of gaseous concentrations. A 20-channel automatic data
logger was later added to reduce the data reduction task.
IMPLEMENTATION OF LOW N0V OPERATING MODE
One area of concern in the implementation of a low N0X operating mode is
how to preserve a low emission level for extended duration through routine
maintenance. A number of visual observations and supporting measurements are
available to confirm the performance after combustion modification. The items
listed in Table I may be useful as a checklist on NO control combustion
performance.
Visual observations are important, even in coal-fired units, to detect
possible flame impingement, stability, or slagging and fouling problems.
Staged combustion tends to result in longer flames that, if not properly
controlled, can lead to flame impingement, slagging, and tube corrosion in the
lower furnace region. Excess O2 control and flame stability are critical.
They are influenced by combustion uniformity across the burner region, careful
maintenance of burner parts, and bulk gas motion due to inter-burner mixing.
Uniformity of burner mixing and completeness of combustion should be confirmed
by gaseous emissions traverses in the exhaust duct. Local gaseous emission
measurements in the lower furnace region through modified viewing or service
ports may be desirable to establish that local reducing atmospheres that could
lead to tube corrosion are not present.
Satisfactory implementation of a low N0X mode for the long term is very
strongly dependent on a number of critical factors in individual boiler design
and maintenance. Almost any design factor or maintenance item that can affect
the performance of the fuel or combustion air system is critical to the deli-
cate balance of combustion conditions necessary for low N0X« A number of
important design limitations may be encountered that compromise or make com-
bustion modifications more difficult. Table II lists some of these factors,
such as the mill or primary air capacity, which may limit the number of BOOS
or the burner pattern.
105
-------
The preservation of low N0x emissions requires careful attention to a
number of critical maintenance items, mostly associated with the fuel and
combustion air systems.
TEST RESULTS
Five of the planned seven 30-day tests have been completed and final
reports written. The five test sites include two spreader stokers, a
pulverized-coal-fired unit, a residual oil-fired boiler, and a natural
gas-fired boiler. The results of the tests are summarized in this section.
Site 1, Coal-Fired Spreader Stoker
Site 1 was a 100,000-lb steam/hr coal-fired spreader stoker. The N0X
control technology employed on this unit was low excess air (LEA). The LEA
condition was maintained for 30 days with a mean NO emission level of 170 ng/J
(278 ppm @ 3% Oj, dry) with the boiler load between 20.5 and 23.2 MW thermal
output (70,000 to 79,000 lb steam/hr). At the same load, baseline NO emis-
sions are 200 ng/J (360 ppm). A log-probability plot of the NO emissions for
the coal-fired spreader is presented in Figure 3.
Site 2, Residual-Oil-Fired Boiler
Site 2 was a 26.4-MW output (90,000 lb steam/hr) residual-oil-fired
boiler. The NOx control technology employed on this unit was staged combus-
tion air (SCA). The as-found concentration of NOx was 130 ng/J (235 ppm at 3%
Oji dry). Firing in the low NOx mode, with staged combustion air, resulted in
NOx emission reduction of approximately 23 percent to 100 ng/J (181 ppm at 3%
Oj# dry). Staged combustion was accomplished by removing from service the top
burner in the triangular arrangement. All air registers were left open.
Capacity of the boiler was reduced to approximately 60,000 lb/hr due to fuel
pressure limitations. Normally, during low NOx testing the oil gun tips are
changed to allow the boiler to operate at full capacity with burners out of
service. At this site, larger tips were not available at the time of the
test, so the test was conducted at lower capacity. A log-probability plot of
NO emissions is presented in Figure 4 for the boiler operating with SCA. The
106
-------
mean value was 100 ng/J (181 ppm), and 99 percent of the data are less than
130 ng/J (235 ppm).
Site 3, Pulverized-Coal-Fired Boiler
Site 3 was a 76.2-MW (260,000 lb steam/hr) output, pulverized-coal-fired
watertube boiler. The NOx control technology employed on this unit was staged
combustion air and low excess air. The results indicate that staged combus-
tion air and low excess air can be effective techniques for NOx control.
However, additional operational problems such as flame stability can be
encountered. The baseline NO measurement was 498 ng/J (815 ppm @ 3% Og, dry)
with the unit operating at approximately 70 percent of capacity. At approxi-
mately the same load, low NOx operation yielded a NO emission level of
422 ng/J (691 ppm @ 3% O2/ dry) for a reduction of approximately 15 percent.
The average NO emission level for 30 days, firing with staged combustion air
and low excess air at loads varying from 15 MW to 63 MW, was 340 ng/J (557 ppm
@ 3% O2» dry). Boiler efficiency showed an increase of approximately
1 percent under low NOx firing condition. A log-probability plot of the NO
emissions is shown in Figure 5 for the boiler. The mean value is 340 ng/J
(557 ppm).
Site 4, Coal-Fired Spreader Stoker
Site 4 was a 38.1-MW output (130,000 lb steam/hr) coal-fired spreader
stoker. The NOx control technology employed on this unit was low excess air
and staged combustion air. The results indicate that low excess air firing is
an effective long-term N0X control technique for spreader stokers, while the
use of staged combustion air by overfire air adjustment is not. The as-found
concentration of NOx was 240 ng/J (409 ppm at 3% 02, dry) with the boiler load
at 80 percent of design capacity. Firing in the low excess air mode resulted
in a reduction of approximately 19 percent from the as-found condition. Low
excess air firing also resulted in an increase in efficiency of approximately
1.2 percent, and a decrease in particulates of about 22 percent. A
log-probablity plot of the NO emissions is shown in Figure 6 for this
boiler. The mean value of NO is 211 ng/J (360 ppm) with 99 percent of the
values less than 245 ng/J.
107
-------
Site 5, Gas-Fired Low NO]c Burner
Site 5 was a 24.9-MW (85,000 lb steam/hr) watertube boiler outfitted with
low NOx burners. The 30-day test was conducted with the unit firing natural
gas. The mean NO emission level was 33 ng/J, with a geometric dispersion of
1.12 at the high load condition. At low loads (<11 MW) the mean NO emission
level was 44 ng/J with geometric dispersion of 1.14. Tests with an adjacent
identical boiler with standard burners produced NO emissions of 113 ng/J at
boiler loads greater than 11 MW and 95 ng/J at less than 11 MW. A
log-probability plot of NO emissions is shown in Figure 7 for the boiler under
high load conditions*
CONCLUSIONS
Combustion modification techniques that have demonstrated reduction of
air pollutant emissions during short duration tests have been shown to be
effective for extended periods.
Little adverse effect on boiler performance or reliability was observed
during the five 30-day tests.
The continuous monitor system employing an extractive sampling system
provided accurate, reliable data.
108
-------
o
TEST
Particulate
Opacity
Gaseous
Pollutants
DAY
(not to scale)
in
J
1
-u
O
"5
S
Test Period 30-37 Days
J/i
o>
c
SE-j™
o
^d:
O
u
©CN
£ c c
goo
Ego
D
Til
v*
to
-o
O
©
2
¦D
O
"5
2
so
IT
CMS Daily Span Checks
~9TV15| 16l ^ 30l 3ll ^ 36! 37
Figure 1. Chronology of typical source test
-------
Two-pen
Chart
Recorders
Figure 2. Continuous monitor.
110
-------
1000
900
800
700 i
600
500
400
300
200 _
LI HI II I 1—I 1 1 I I I I I 1—I TT
100
9
8
7
6
5 -
4
3 -
2 —
MIL
LOAD: 20.5 - 23.2
x ¦ 170 ng/J
g.d. ~ 1.08
1 I I ¦ ¦ ¦ ¦ I I ' ' I I I I 1 I 1 I I
JL-L
II I I
.01.05.1.2 .5 12 5 10 20 30 40 50 60 70 80
PERCENT LESS THAN
90 95 98 99 99.8 99.9 99.S
Figure 3. NO emissions from site 1, coal-fired spreader stoker
-------
1000
900
800
700
600
500
400
300
200
100
90
80
70
60
50
40
30
20
10
jiii i i i—i—i—i i i i i i i—i—i—n—n—~
x « 100 ng/J
g - 1.12
I I I I I I I I I I I I I I I I I I I I >1
01.05.1 .2 . 5 1 2 5 10 20 30 40 50 60 70 80 90 95 98 99 99.8 99.9 99.
PERCENT LESS THAN
Figure 4. Site 2, residual-oil-fired boiler, staged combustion air (BOOS)
-------
800
700
600
500
400 h-
300
200
TT
t—T
t—i i i i i—r
t i i i i re
100
x
g
340 ng/J
1.13
Low NO^ Operation (SCA)
I
JUL
JLJL
J I I 1 I I I I I I I I I I 11
0.01 0.05 0.1 0.2 0.5
5 10 20 30 40 50 60
Percent Less Than
70 80 90 95 98 99
99.99
Figure 5. NO emissions from site 3, pulverized-coal-fired boiler
-------
i i i i ii i—i—i—i i i i i i i—i—r
i »
n—r
300 —
200 —
£
i
100 —
SITE 4 - GOAL FIRED
STOKER
x « 211 ng/J
9 « 1.06
I I II II I I I I » I I » » I
I I I
1_I L
0.01 0.05 0.10.5 1
5 10 20 30 40 50 60 70 80
90
95 98 99 99.8 99.9 99.
PERCENT LESS THAN
Figure 6. NO emissions from site 4, coal-fired spreader stoker
-------
Li I I I ill—I—I—l I i I I I I—I—I—n n—n
Natural Gas Firing
Load Range >11 MW
200
tr>
c
O
z
100
90
80
70
60
50
40
30 L.
20
10
111! II I
t I I I I I I I I
X
g
I
33 ng/J
1.12
I I 1
I I
I
0.01
0.2
10 20 30 50 70 80
PERCENT LESS THAN
90 95 98 99 99.8 99.00
Figure 7. NO emissions from site 5, low N0X burner/watertube boiler
-------
TABLE I
FACTORS TO CONSIDER IN EVALUATING THE
SUITABILITY OF A LOW NOx FIRING MODE
Visual
Observation
Flame stability
Slagging/fouling
Bulk gas motion
Combustion uniformity
Completeness of
combustion
Flame impingement
Gaseous Emissions
Data
Combustion balance
Burner mixing
Operating O2 level
Potential tube
corrosion
Boiler efficiency
Particulate/
Fly Ash Samples
Carbon carry-
over
Particulates
Ash characteris-
tics
Combustion
efficiency
Precipitator
performance
Equipment
Operating Data
Tube metal tenp.
Steam temperature
Firing rate
Efficiency
Pulverizer per-
formance
Feeder operations
Auxiliary load
Burner front
settings
-------
TABLE II
OPERATIONAL CONSIDERATIONS IN IMPLEMENTING
LOW NOx COMBUSTION MODIFICATIONS
Boiler Design Factors
Mill/PA capacity, maximum and minimum
Burner pattern/mill arrangement
Coal pipe design and coal distribution
Burner design
Windbox design
Overfire air port configuration/mixing
Excess oxygen sensors, number and placement
CO sensors/instrumentation
Fuel/air flow instrumentation
Critical Maintenance/Adjustment Items
Burner impellers
Pulverizer adjustments/wear, parts replacement
Classifier setting
Primary air control/instrumentation
Coal pipe dams/balance
Feeder balance
Feeder bar height
Mill balance
Oxygen analyzer/controls maintenance
Air register control
Air/coal temperature control
Sootblower/air compressor maintenance
Air preheater leakage/maintenance
117
-------
CONVENTIONAL COMBUSTION ENVIRONMENTAL
ASSESSMENT PROGRAM
By:
W. H. Ponder
U. S. Environmental Protection Agency
Industrial Environmental Research Laboratory
Research Triangle Park, North Carolina 27711
118
-------
ABSTRACT
The Environmental Protection Agency's Industrial Environmental Research
Laboratory at Research Triangle Park, North Carolina, has developed and
implemented a major program for the assessment of the environmental, economic,
and energy Impacts of multimedia pollutant emissions from stationary residential,
commercial, institutional, industrial, and utility combustion processes. The
Conventional Combustion Environmental Assessment (CCEA) Program has become a major
source of data and information for Agency use in developing and modifying standards
and control technologies.
This paper presents the theme, objectives, pollutants of concern, current
activities, and some representative data from CCEA Program projects. Included
are data from: 1) a comparative assessment of coal and oil firing in an indus-
trial boiler, 2) environmental assessments of an 820 MW, FGD-controlled, coal-
fired utility boiler and a 342 MW oil-fired utility boiler, 3) a 170-site
field study of combustion sources, 4) dry bottom industrial boilers firing
pulverized coal, 5) residential coal combustion, and 6) wood combustion studies.
119
-------
1. INTRODUCTION
EPA's Industrial Environmental Research Laboratory at Research Triangle
Park, NC, initiated the Conventional Combustion Environmental Assessment (CCEA)
Program In February 1977. The primary purpose of the CCEA Program is to assess
the environmental, economic, and energy impacts of stationary conventional processes
firing coal, oil, wood, derived fuels, waste materials, and combinations of these
fuels• The assessment results achieved in the CCEA Program are used in the Agency
to ensure that the environmental impacts of the expanded use of conventional
combustion processes for energy production are kept within acceptable limits.
This paper presents the theme, objectives, pollutants of concern, current activities,
and some representative data from the CCEA Program*
2. CCEA PROGRAM THEME AND OBJECTIVES
The central theme and focus of the CCEA Program is the assessment of hazardous
pollutants in the gaseous emission streams from conventional combustion processes.
The Program is also concerned with those pollutants which degrade water and land
quality after removal from air emission streams (cross media impacts). In keep-
ing with the CCEA Program theme, three specific objectives have been established
for the Program:
a. Assess the controllability of hazardous pollutants,
including costs, removal efficiencies, schedules
(for emerging technologies), and energy penalties.
b. Assess the environmental impacts of hazardous pollu-
tants from conventional combustion processes operating
at baseline, transient, and modified conditions (pollu-
tants emitted, quantities emitted, biological characteri-
zation, and impact projections).
c. Identify need for risk assessments and provide supporting
information for risk assessments which may be conducted
subsequently by other Agency offices.
120
-------
3. POLLUTANTS
In meeting the three objectives specified above, the CCEA Program is consid-
ering some top priority pollutants including hazardous or toxic organic compounds
and hazardous or toxic metals. For example, the CCEA Program studies have determined
that the emission levels of polycyclic organic matter (POM) from wood combustion
are high enough to cause concern for potential environmental impacts and to warrant
the commitment of additional resources to better define the magnitude of the
problem and to investigate techniques for mitigating the impacts. In addition
to the assessment of the impacts of wood combustion during 1981 and 1982, the
Program will also continue to characterize and assess the emissions of the
top priority pollutants which result from the combustion of other fuels, including
coal, oil, derived fuels, waste solvents, and combinations of these fuels.
4. SUMMARY OF CURRENT MAJOR PROGRAM ACTIVITIES
The CCEA Program is a comprehensive, coordinated effort to assess the
environmental impacts of conventional combustion processes and associated
pollution control technologies. The Program is comprised of several projects
each of which makes a contribution to the overall goals and objectives of
the Program. In the interest of brevity, this section will provide an
overview of some of the major projects.
Emissions Assessment of Conventional Combustion Systems (EACCS) - The EACCS
project is a major contributor to the CCEA Program data base. The EACCS
project, a 4-year effort which will end in September 1980, collected field
data on multimedia emissions from 51 classes of conventional stationary combustion
sources. Prior to initiating field testing, existing data from each class of
combustion sources were assembled and evaluated to determine areas in which
data were scarce, of questionable accuracy, or non-existent. Based on the
results of this evaluation of existing data, field testing of 171 combustion
sources was scheduled and conducted.
121
-------
These field tests have filled data gaps, augmented existing data, and
produced some important findings. For example, the study of stationary internal
combustion sources verified the significant contribution that these sources make to
the national emissions of NO and hydrocarbons. In addition, oil-fired residential
furnaces were found to be significant sources of NO , SO.,, and Ni. Initial data
X
indicate that POM emissions from residential and industrial wood combustion may
be alarmingly high, and extensive investigation of POM emissions from wood combustion
is being planned as a result.
Selected data from the EACCS project are presented in Section 5 of this paper.
The current status of the EACCS reports on various combustion source categories
surveyed is presented below:
1) Reports Published:
° Methods and Procedures Manual for Sampling and- Analysis
EPA-600/7-79-029a (NTIS PB 294675)
° Gas- and Oil-Fired Residential Heating Sources
EPA-600/7-79-029b (NTIS PB 298494)
0 Internal Combustion Sources
EPA-600/7-79-029c (NTIS PB 296390)
° Environmental Assessment of a Coal-Fired Controlled
Utility Boiler (LaCygne Station, Kansas City Power
and Light Company)
EPA-600/7-80-086 (NTIS PB 80-187735)
° Environmental Assessment of an Oil-Fired Controlled
Utility Boiler (Haynes Station, Los Angeles Department
of Water and Power)
EPA-600/7-80-087 (NTIS PB 80-190085)
122
-------
2) Reports Currently in Draft Form:
0 Commercial/Institutional Source Category
° Utility (External Combustion) Source Category
3) Reports Scheduled for Completion in 1980:
° Environmental Assessment of NIPSCO's Mitchell Station
(Wellman-Lord FGD System)
° Environmental Assessment of Louisville Gas and Electric's
Cane Run Station (Dual Alkali FGD System)
° Industrial Source Category
° EACCS Final Summary Report
Environmental Assessment of Stationary Source NO^ Control Technologies - Like
EACCS, this project is also a major contributor to the CCEA Program data base.
The overall objectives of this project are to (1) assess the environmental
impacts of stationary combustion sources and NO controls and (2) identify
cost-effective, environmentally sound NO^ control technologies that can be
used to meet NC^ emission and air quality standards. The project - initiated
in June 1976 and concluded in November 1979 - filled data gaps, augmented the
CCEA data base, and produced significant and useful findings. For example, the
study showed that coal firing produces the greatest discharge severity (see
Glossary) and that the flue gas stream dominates environmental impacts. Among
flue gas components, NO and S09 generally account for more than 50 percent
of the discharge severity associated with the entire flue gas stream. In
addition, the preliminary study results suggest that no definitive trend exists
between POM emissions and low NO operation. Selected data from this project
X
123
-------
are presented in Section 5 of this paper. The current status of the N0x EA
reports is presented below:
1) Reports Published:
° Preliminary Environmental Assessment of Combustion
Modification Techniques (EPA-600/7-77-119a and b;
NTIS PB 276680 and 276681)
° SAM/IA: A Rapid Screening Method for Environmental
Assessment of Fossil Energy Process Effluents
(EPA-600/7-78-015; NTIS PB 276088)
° Environmental Assessment of Stationary Source N0x
Control Technologies: First Annual Report
(EPA-600/7-78-046; NTIS PB 279083)
° Emission Characterization of Stationary N0x Sources
(EPA-600/7-78-120a and b; NTIS PB 284520 and 285429)
° Environmental Assessment of Stationary Source N0x
Control Technologies: Second Annual Report
(EPA-600/7-79-147; NTIS PB 300469)
2) Reports Currently in Draft Form:
° Internal Combustion Engine Special Report
° Environmental Assessment of Stationary Source
N0x Control Technologies: Final Report
124
-------
Combustion Modification Environmental Assessment - The main objective of
this project is to continue the environmental assessment of combustion control
technology initiated under the project above* This project will produce
environmental assessment test programs including individual source test
reports and special reports. The test programs will determine the effects
of combustion modifications through chemical analysis, bioassay testing, and
operating data evaluation. Initiated in December 1979, this project will
assess 10 sites per year over the next 3 years.
1) Reports Published: None
2) Reports Currently in Draft Form: None
3) Reports Scheduled for Completion in 1980:
° Individual Site Reports and Data Supplement
Reports for Sites 1-10 (9/80 - 12/80)
0 Special Report 7/80
0 Special Report 11/80
Environmental Assessment of Wood Combustion - Preliminary data from CCEA
projects and from the work of other researchers have indicated that the
quantities of POM emitted from residential, commercial, and industrial wood
combustion are significant in comparison to quantities emitted when other fuels
are burned. Data in this area are scarce, and there is a pressing need to better
delineate the nature and magnitude of the problem. As a result, the CCEA Program
has undertaken the development of a comprehensive wood combustion research and
development program which began in FY 79 and will be greatly expanded in FY 81.
The wood program will Include the characterization and environmental assessment
of emissions from various residential, commercial, and industrial wood combustors.
125
-------
The program will also include ambient sampling to determine POM concentrations
in a community in which wood is the predominant fuel* In addition, the basic
process of wood combustion under various conditions will be evaluated along with
combustor types and operating techniques to determine feasible approaches for
minimizing potentially harmful emissions from wood combustion processes* This
comprehensive program will be initiated in December 1980.
5. REPRESENTATIVE DATA
The CCEA Program has compiled an extensive environmental assessment data
base dealing with a broad range of conventional combustion equipment and
a variety of fuels and fuel combinations* CCEA field data and the assessment
and evaluation of these and other data have produced significant results
and findings that identify and elucidate environmental problems that are
of concern to the Agency and to Industry as well. The data below are presented
as examples of representative data, findings, and conclusions that are
being developed in the CCEA Program.
EACCS Project - Environmental Assessment of Coal Vs. Oil Firing - A com-
prehensive enviromental assessment of a 10 MW industrial boiler has been
completed under the direction and sponsorship of the CCEA Program. The final
report (EPA-600/7-78-164a, b, and c; NTIS PB 289942, 289941, and 291236)
describes the results of sampling and analysis in a dual fuel process steam
boiler operated by Firestone Tire and Rubber Company. The boiler, equipped
with an FMC dual alkali FGD pilot scrubber, was originally designed to burn
coal and later modified to burn either high volatile eastern bituminous coal
or No. 6 fuel oil.
The difference in environmental impacts between coal and oil combustion
emissions from industrial boilers controlled only by FGD is potentially
significant. Even so, the report concludes that other factors may override
fuel choice in. determining the environmental acceptability of controlled
126
-------
industrial boilers. Such factors include location, type, and number of
other emission sources; background pollution levels; and the potential long-
term accumulation of pollutants to unacceptable levels in the environment.
The dual alkali scrubber operated at SO2 and particle removal effi-
ciencies of 96 and 99 percent, respectively. At these conditions, emissions
of NC>x> CO, and organics (as CH^) from coal firing were about three times as
great as those from oil firing. But the ratio of coal to oil emissions for
SC>2, SO^, sulfate, and total particles was in the range of 0.8 to 1.5,
Models applied to this specific location and plant indicate that estimated
ambient N0x concentration produced by coal firing (double those produced by
oil firing) would exceed the National Ambient Air Quality Standards (NAAQS) for
NO without controls.
Coal firing produced higher trace element concentrations in the scrubber
cake than oil firing, but quantities of heavy metals and toxic substances in
the scrubber cake from both fuels would require disposal controls to prevent
leaching and ground water contamination.
Emissions of cadmium from oil firing and molybdenum from coal firing
were of concern due to projected accumulations in vegetation to concentrations
that could be potentially injurious to people and to animals.
Scrubbing removed coal-generated sulfate more efficiently than oil-
generated sulfate, and 99 percent of coal-fired particles were removed by
scrubbing while only 75 percent of oil-fired particles were removed. This
may be attributed to the fact that almost 98 (wt) percent of coal-fired
particles were greater than 10pm in diameter whereas the oil-fired particles
were smaller. The data suggest the possibility of a net increase across
the FGD process in the mass emission rate of particles less than 3 p m in
size, and further investigation of this is planned in the CCEA Program. For
127
-------
purposes of comparison, controlled particle emissions for both fuels were
well below the old utility boiler New Source Performance Standards (NSPS) of
0.04 g/MJ (0.1 lb/10^ Btu) of heat input but slightly above the revised limit
of 0.01 g/MJ (0.03 lb/10^ Btu). The results of this study are contributing
to the technical basis used by EPA's Office of Air Quality Standards and
Planning (OAQPS) in developing NSPS for industrial boilers.
EACCS Project - Environmental Assessment of an 820 MW Coal-Fired Utility Boiler
and a 342 MW 011-Fired Utility Boiler - The CCEA Program has recently completed
two studies to characterize multimedia pollutant emissions from utility boilers:
one study of an 820 MW coal-fired utility boiler and the other of a 342 MW
oil-fired utility boiler. Level 1 and Level 2 sampling and analysis procedures*
were used in both cases to characterize emissions in gaseous, liquid, and
solid process streams. The major conclusions from each study are presented
below.
The coal-fired boiler studied was the No. 1 unit at Kansas City Power and
Lights's La Cygne Power Station in Kansas. It typically burns a local high
sulfur, high ash subbitumlnous coal. Sulfur dioxide (SO2) and particle
emissions are controlled by eight venturi/absorber scrubber modules using
limestone slurry.
During the study, flue gas was sampled before and after scrubbing.
Emissions were determined for the major species, as shown below. Other streams
examined were the combined bottom and fly ash, scrubber solids, settling pond
overflow, and ash pond overflow.
* These sampling and analysis procedures are described in a previous CCEA report,
EPA-600/7-79-029a, "Emissions Assessment of Conventional Stationary Combustion
Systems: Methods and Procedures Manual for Sampling and Analysis," (NTIS
PB 294675).
128
-------
The data for total particulate matter emissions indicate 91 percent removal
by scrubbing. While scrubber inlet particles were larger than 3 jjm in diameter,
after scrubbing most particles were less than 1 m in diameter. As was the
case earlier for industrial coal combustion, preliminary data suggest an increase
in the mass emission rate of particles less than 3 yB in size across the scrubber,
and additional CCEA studies are planned to verify or refute this finding.
Some POM compounds were identified at the scrubber inlet, but the levels
were considered to be insignificant. The POM compounds identified were
naphthalene, substituted naphthalenes, biphenyl, and substituted biphenyls.
No POM compound was detected at the scrubber outlet.
The major study conclusions, based on assumed typical and worst case
meteorological conditions, are:
6 There is a low risk of violating the NAAQS for 24-hour
and annual average levels of criteria pollutants.
However, units firing high sulfur fuels may exceed the
short term NAAQS for SO2.
0 SO2, SO^, arid particulate emissions from coal-fired
units of the type tested may result in limited adverse
health effects, and should be studied further.
0 Increased concentrations of cadmium and lead in plant
and soil tissues as a result of trace element emissions
could cause plant damage and adverse health effects to
animals consuming vegetation in the affected areas.
0 NO and SO emissions will probably cause plant damage,
X X
since concentrations of both pollutants approach or exceed
the damage threshold range.
129
-------
These conclusions are based on assumed typical and worst case meteorological
parameters. Because the environmental acceptability of emissions from coal-
fired boilers depends largely on site specific factors, extrapolation from one
set of conditions to another should be avoided.
Emissions testing was also performed at the oil-fired No. 5 boiler at
Los Angeles Department of Water and Power's Haynes Power Plant in Long Beach,
California. This unit is capable of firing either low sulfur oil or natural gas.
N0x emissions are controlled by off-stoichiometric firing and flue gas recirculation.
As i6 typical of oil-fired utility boilers, no particulate or S0x controls are
applied at this site.
The fuel oil and flue gas were analyzed during oil-fired operation. No
significant liquid or solid waste streams are produced by the boiler.
Measured emissions of the criteria pollutants and SO3 (shown below)
corresponded well with published emission data from oil-fired boilers (AP-42,
NT1S PB 275525), although measured NO and total organic emissions were
A
somewhat lower. (The reduced NO emissions were the likely result of the
N0x control systems.)
The environmental assessment study concluded that there is a low risk
of exceeding the NAAQS as a result of application of this type of boiler. The
projected emissions of SO2, SO^, and particles seem to be within acceptable
limits. Negligible impacts of trace element burdens on drinking water, plant
tissue, soil, and the atmosphere are projected. The risk of damage to vegetation
posed by criteria pollutant emissions is remote*
Summary data from the coal-fired tests and the oil-fired tests are
presented below.
130
-------
FLUE GAS EMISSIONS FROM A COAL-FIRED UTILITY BOILER
Emission Factor, ng/J (g/kg)
Pollutant
NOx (as NO2 near full load)
CO
so2
so3
so I
Total Organics
Total Particulates
CI"
F~
Before Scrubber
2715 (17.2)a
<520 (12.5)b
3380 + 400 (81.2 + 9.6)c
48 + 24 (1.2 + 0,6)C
22 + 9.0 (0.5 + 0.2)c
2.77 - 4.07 (0.07 - 0.10)d
1090 + 270 (26.2 + 6.5)c
<0.1 (0.002)
0.6 + 0.4 (0.014 + 0.010)c
After Scrubber
>385 (9.3)a
<520 (12.5)b
740 + 90 (17.8 + 2.2)c
10 + 11 (0.2 + 0.3)c
2.7 + 1.9 (0.07 + 0.05)°
1.45 - 2.60 (0.04 - 0.06)d
80 (1.9)
<0.1 (0.002)
<0.14 (0.003)
Measured values are considered to be lower limit values due to potential for
NO degradation in bag samples.
X
Determined by GC analysis of bag samples; values represent detection limit
of 1000 ppm.
Indicated uncertainty represents one standard deviation.
Cj - C,c fractions determined by GC; >C^ fraction determined gravimetrically.
Upper limit values include detection limits of fractions which were not found.
131
-------
EMISSIONS FROM AN OIL-FIRED UTILITY BOILER
Pollutant
NO^ (as NO2 near full load)
CO
so2
so3
so;
Total Organlcs
Total Particulates
ci-
F~
Cr
Ni
Emission Factor» ng/J (g/kg)
116 + 2.12 (5.12 + 0.09)a
6.6 + 3.1 (0.29 + 0.14)a
98 + 7.0 (4.3 + 0.31)a
1.14 (0.05)
1.27 (0.06)
0.42 - 0.58 (0.02 - 0.03)b
7.5 + 1.2 (0.33 + 0.05)a
1.34 (0.06)
0.061 (0.003)
0.002 (0.0001)
0.2 (0.01)
£
Indicated uncertainty represents one standard deviation.
Cj - Cjg fractions determined by GC; fraction determined gravimetrically.
132
-------
EACCS Project - Residential and Internal Combustion Sources - As indicated
in Section 4, a major project of the CCEA Program is the EACCS project. This
project, performed by TRW, Inc., is providing a comprehensive assessment of
emissions from several types of conventional stationary combustion processes.
The goal of the project is to develop extensive baseline data by identifying
and characterizing the gaseous, liquid, and solid pollutants generated by these
sources. The final assessment will be based on appropriate existing emissions
data as well as on new data acquired through source sampling and analysis.
When the project is completed, assessments will have been made of five
major groups of combustion processes:
° Gas- and oil-fired residential combustion sources.
° Gas- and distillate-oil-fired gas turbines and
reciprocating engines (internal combustion sources)
for electricity generation and industrial application.
° External combustion sources for electricity generation.
0 Industrial external combustion sources.
° Commercial/institutional external combustion sources.
Assessments of the first two major groups have been completed, and the final
reports are available.
To date, Level 1 sampling has been completed at about 170 sites, and
Level 2 sampling has been performed at 15 sites. Two EACCS reports present
the results of these efforts for two groups of source categories. "Emissions
Assessment of Conventional Stationary Combustion Systems: Volume I: Gas- and
Oil-Fired Residential Heating Sources" (EPA-600/7-79-029b; NTIS PB 298494)
considers combustion units for space heating with gas or oil input capacities
below 0.12 MJ/s (400 Btu/hr). Gas-fired systems and oil-fired systems account
133
-------
for 58 percent and 38 percent, respectively, of residential units. Residential
combustion systems consume about 15 percent of the fuel used by conventional
stationary combustion systems.
Initially, five gas-fired and five oil-fired residential sources were
tested; mass emission rates of criteria pollutants, trace elements, and
organics (including POM) were determined. Emission data for particles, SO^,
S03, and SO2 were also obtained at the oil-fired sites. Later tests were performed
at one gas-fired and two oil-fired sites to determine the effect of the boiler
on/off cycle on emissions.
Severity factors were calculated for the various species emitted. The
severity factors are defined as the ratio of the calculated maximum ground
level concentration of the pollutant species to the level at which a potential
hazard exists. Concentrations for multiple combustion sources were determined
using a dispersion model for an array of 1000 sources.
The study concludes that residential sources are of potential significance
based on multiple source severity factors. Multiple source severity factors
exceeded 0.05 (the level which may be potentially significant) for NO from
gas-fired sources and for SO^, N0x> and Ni from oil-fired sources. Measured
criteria pollutant emission factors were generally comparable to the EPA emission
factors based on earlier data (AP-42, NTIS PB 275525), except for total hydro-
carbon emissions from oil-fire sources, which were three times greater. POM
compounds that were known to be carcinogenic were not found above the detection
limit of 0.3 p g/m^.
The report recommends additional work to augment the emission data base for
oil-fired sources, especially with regard to SO^ and POM emissions and multiple
source severities. The emission data base for gas-fired residential sources
is currently considered adequate and no further study is recommended at this time.
134
-------
The second EACCS report, "Emissions Assessment of Conventional Stationary
Combustion Systems: Volume II: Internal Combustion Sources" (EPA-600/7-79-029cj
NTIS PB 296390), examines stationary internal combustion (IC) sources for
electricity generation and industrial use. The sources are classified as
(1) gas- and distillate-oil-fired gas turbines and (2) reciprocating engines
(diesel). Six gas turbines (five oil- and one gas-fired) and five distillate
oil reciprocating (diesel) engines were tested* The existing data for gas-fired
reciprocating engines were judged to be adequate. The data for gas-fired turbines
were also considered to be adequate, but one site was included to ensure that
previously unidentified pollutants were not being emitted in unacceptable quantities.
Two major conclusions of the study are that:
0 IC sources contribute significantly to the national
emissions burdens. N0x, hydrocarbon, and carbon monoxide
(CO) emissions from IC sources account for approximately
20, 9, and 1 percent, respectively, of the emissions of
these pollutants from all stationary sources.
° Several pollutants emitted by IC sources are of environ-
mental concern. These include NO (from all sources
examined), total hydrocarbons (from gas reciprocating
engines and distillate oil reciprocating engines),
SO. (from distillate oil reciprocating engines),
S03 (from distillate-oil-fueled gas turbines and dis-
tillate oil reciprocating engines), and trace elements
(from most of the oil-fueled IC sources tested).
Source Assessment Project - Dry Bottom Industrial Boilers Firing Pulverized
Coal - The primary method of coal combustion in U.S. industrial boilers is in dry
bottom units firing pulverized bituminous coal. A multimedia environmental assess-
ment of this source type was recently completed by Monsanto Research Corporation
for EPA (EPA-600/2-79-019e; NTIS PB 80-177207). The study, "Source Assessment:
Dry Bottom Industrial Boilers Firing Pulverized Bituminous Coal," was based on an
extensive literature search and sampling and analysis as well. It concludes that
certain air emissions are released at potentially hazardous concentrations even when
existing controls are applied. The potential impacts of controlled liquid and solid
waste discharges from this source, however, are insignificant.
135
-------
Dry bottom boilers operate at temperatures below the ash fusion temperature.
Ash remaining in the bottom of the furnace is removed as a dry powder. Most
of these boilers are in the industrialized Northeast states, in large cities
and along major waterways. This source category represents about 9 percent of
the total steam-generating capacity of U.S. industry and approximately 49 percent
of the industrial steam generated by coal combustion. The average capacity of the
O
industrial boilers considered in this assessment was 222 GJ/hr (2.1 x 10 Btu/hr).
More than 99 percent of the air emissions result from coal combustion in
the furnace and are emitted from the boiler stack* Major emissions are
the criteria pollutants: particles, SO^, N0x, hydrocarbons, and CO. POMs
are among the hydrocarbon species emitted. In addition, trace elements are
released as part of the particulates or in the vapor phase.
The potential environmental impact of each species emitted after
passing through state-of-the-art controls was individually assessed on the
basis of source severity. Source severity, an indicator of potential environmental
impact, is the ratio of the maximum ground level concentration to a potentially
hazardous concentration. Species with source severities greater than
1.0 were N0x (1.7), S0x (2.2), and POMs (6.0).
Dispersion modeling was applied to determine the affected population,
defined as the population exposed to specified potentially harmful emission
levels from an average source. Estimates of the number of persons exposed to
severities greater than 1.0 are shown below.
Emission Species Source Severity* No. of Affected Persons
2.2
1.7
1,200
2,200
7,500
POM
6.0
* Ratio of the maximum ground level concentration to a potentially
hazardous concentration
136
-------
The study predicts that the total design capacity of boilers covered
in this assessment will increase at an annual rate of 3 to 4 percent
through 1990. Total air emissions and wastewater effluents during this
period will likely remain constant or drop slightly due to increased controls.
The volume of solid wastes is expected to grow as air emission controls are
increasingly applied to this expanding source.
Source Assessment Project - Residential Coal Combustion - Some regions of
the U.S. show a trend toward increased home heating with coal. A recent study
conducted by Monsanto Research Corporation characterized the emissions from
residential coal combustion and evaluated their potential environmental effects.
The study results are presented in the final report, "Source Assessment:
Residential Combustion of Coal" (EPA-600/2-79-019a; NTIS PB 295649).
In 1974 approximately 2.6 Tg (2.9 x 10^ tons) of coal were burned as
a primary source of heat in an estimated 493,018 housing units. Although
this represents only 1 percent of the total U.S. housing units with primary
heating devices, interest in this form of heating is growing. From 1972
to 1975, sales of domestic coal-fired heating stoves increased by 130 percent.
Since 1976, sales of other heating devices such as stoker furnaces have also
increased. With the current shortages of natural gas and oil, these trends
may continue for some time, a likelihood that underscores the importance
of assessing coal-fired home heaters.
Coal-fired residential combustion sources consist of all equipment that
burns bituminous, anthracite, or lignite coal to generate household heat.
These devices produce up to 0.12 MJ/s (4.0 x 10^ Btu/hr) of heat in occupied
structures containing one or two housing units. A wide variety of primary
residential coal-fired heating equipment is available; common types include
steam or hot water boilers, warm air furnaces, and domestic heating stoves.
137
-------
Residential coal combustion generates many atmospheric emissions in
addition to a solid residue. Atmospheric emissions include particles, SC>x,
N0x> CO, hydrocarbons (including POMs), and trace elements. Pollutants are
generated during the combustion process and, with the exception of some of
the NO , are formed from the coal as it burns. Some NO is formed by the
X X
combination of atmospheric nitrogen and oxygen at high temperatures in
the furnace.
The solid residue consists of inert material (ash) and unburned or
partially burned fuel. If the solid residue is disposed of by landfill,
elements may be leached out by rainfall into water supplies.
Unlike larger combustion systems such as utility boilers, which have
tall stacks to disperse emissions and reduce ground level concentrations,
residential units emit pollutants close to ground level where dispersion is
minimal. In addition, several residential coal combustion sources may be
concentrated in a small area, such as a housing subdivision. In these
cases, an additive multiple source effect resulting in increased ground
level concentrations could occur.
The study evaluated the potential environmental effects of air emissions
from coal-fired residential combustion systems on the basis of source severity,
affected population, state emission burdens, and national emission burdens.
The source severity measures the potential health effect of a pollutant
at its maximum ground level concentration. Generally, 0.05 is considered a
threshold level, above which a potential environmental problem may exist.
An average combustion unit was determined for each coal type as a basis for
severity calculations. (For instance, an average bituminous coal-fired unit
burned Appalachian coal at the rate of 0.3 g/s (II tons/year) and was located
2
in an area with an average population density of 132 persons/km .) POM emissions
showed a severity of 2.6 for bituminous coal combustion, while severities of
the other emissions were 0.05 or less.
138
-------
An assessment of the environmental impact of composite emissions from
100 houses burning coal indicated the potential for a thirtyfold increase
in the associated severities. Severities were greater than 0.05 for 16 indi-
vidual elements and 4 criteria pollutants (particles, SO^, an<^ hydrocarbons).
Severities were 91 and 1.7 for POM in bituminous and anthracite coal burning,
respectively.
POM was the only pollutant from a single source projected to have a potential
effect on the exposed population. However, the potential effect on the exposed
population of pollutants from multiple residential combustion sources was
much greater than the effects from a single source. Multiple source emissions
of particulates, SO^, N0x, hydrocarbons, 16 individual elements, and POM
showed severities greater than 0.05.
The study determined the contributions of coal-fired residential combus-
tion to state and national levels of criteria pollutant emissions. In 1974
home heating with bituminous coal had the greatest impact on a state-by-state
basis, exceeding 1 percent of the total state S0x emissions in the District
of Columbia, Virginia, and West Virginia. Criteria pollutant emissions from
anthracite and lignite combustion were all less than 1 percent of the total
state emissions in every state.
While the criteria pollutant annual emissions from residential coal
combustion comprise a relatively small fraction of the total annual emissions
of these pollutants, the levels of POM released may be significant. In
1974 the national annual emissions of POM totalled about 101 Mg (111 tons)
from automatic bituminous-fired units and about 0.9 Mg (1 ton) from automatic
anthracite-fired units. This represents approximately 10 percent of the total
annual estimated national emission of POM from all stationary sources (indus-
trial, residential, commercial/institutional, and utility).
139
-------
Emissions from residential combustion systems are not typically controlled
with add-on equipment but can be reduced by improved design and proper operation.
Factors to be considered include fuel properties and type, firing rate, firing
equipment design, cyclic operation of automatic equipment, and excess air ratios.
CCEA Systems Engineering Project - Volatile Organic Compound (VOC)
Emissions - A recent CCEA study has provided updated volatile organic compound
(VOC) emission factors for utility coal-fired power plants in support of the
Monitoring and Data Analysis Division of EPA's Office of Air Quality Planning
and Standards (OAQPS). OAQPS is responsible for developing the reactive VOC
emission factors, which will be used by the Regions and States in meeting
Prevention of Significant Deterioration (PSD) requirements.
The VOCs analyzed in the study were C^ to C^ hydrocarbons. Level 1
procedures were used to test for stack VOC emissions from 43 utility boilers
firing bituminous coal, lignite, residual oil, or gas. Boiler size ranged
from a small lignite-fired boiler of 9 MW to a large bituminous coal-fired
unit of 910 MW.
Major conclusions of the study were:
° Reactive VOC emissions ranged from 1.7 to 5.1 ng/J
(4.0 to 11.9 lb/10^ Btu) heat input.
° Bituminous-coal- and lignite-fired boilers emitted
1.2 to 3.0 times more VOCs than did residual-oil-
and gas-fired boilers.
0 Of total reactive VOCs, 85 to 95 percent were due to
the C^ to Cg hydrocarbon group.
0 There was no appreciable difference in the quantities
of VOC emissions from bituminous-coal- and lignite-fired
.boilers. Similarly, there was little quantitative
difference between residual-oil- and gas-fired boiler
VOC emissions.
140
-------
° The reactive VOC emission factors and annual reactive
VOC emissions are accurate within a range of + 200
percent, which is consistent with IERL-RTP Level 1
requirements*
The study recommends further testing with Level 2 sampling and analysis
procedures to determine quantities of oxygenated hydrocarbons. Also suggested
is the use of a statistically designed experimental matrix to evaluate the
effects of boiler types and boiler operating parameters on VOC emissions.
EACCS Project - POM Emissions from Industrial Wood Combustion - Preliminary
results from CCEA field testing at five industrial wood-fired boilers indicate
that wood combustion produces substantially more emissions of POM than other
source categories. Field tests, performed by TRW, Inc., found average POM
O
emissions from the five industrial sites to be 0.8 mg/m . This value is
80 times greater than the average POM emissions determined for utility boilers
firing bituminous coal, and 133 times greater than those found for lignite-fired
utility boilers.
POM emissions from these industrial wood-fired boilers are not only
high, but also include emissions of such highly carcinogenic compounds as:
0 benzo(a)pyrene
° benzo(e)pyrene
0 dibenzo(a,h)anthracene
° dibenzo(def,mno)chrysene
° benzo(g,h,i)perylene
0 indeno(l,2,3-cd)perylene
In view of the potential magnitude and severity of these emissions, additional
sampling and analysis are underway to substantiate these preliminary findings.
Once the extent of the problem has been clearly defined, the CCEA Program will
recommend an approach for dealing with it.
141
-------
Source Assessment Project - Residential Wood Combustion - A recent EPA
report, "Preliminary Characterization of Emissions from Wood-Fired Residential
Combustion Equipment," presents results from a comprehensive study undertaken
to characterize emissions from residential wood-burning devices. The report,
EPA-600/7-80-040 (NTIS PB 80-182066)» prepared for the CCEA Program by
Monsanto Research Corporation, focuses on the effects of certain test parameters
on pollutant species. Conclusions show that combustion equipment design influences
certain emissions. Levels of CO and POM were highest from wood-burning stoves,
while NO2 emissions were greatest from fireplaces. The effect of wood type was
also studied, and results showed increased levels of organic compound emissions
from the combustion of green pine.
In this preliminary study of residential wood combustion, one fireplace
and two wood-burning stoves were tested during the combustion of four types
of wood (green and seasoned yellow pine and red oak). Gaseous emissions
were sampled and analyzed for particulates, condensable organics, SO2, N^j
CO, organic species (including POM), and individual elements. BioasBay tests
were also conducted on the stack emissions and bottom ash. Samples were
collected at Auburn University (Auburn, Alabama) during March and April 1979.
The three combustion units tested during this study were a residential
fireplace, a baffled wood-burning stove, and a nonbaffled wood-burning stove.
Because the emphasis is now on energy efficiency, airtight metal stoves, which
are claimed to be 50 to 70 percent energy efficient, are becoming very popular.
Baffled stoves are generally more sophisticated in design than nonbaffled stoves
and are designed to improve combustion efficiency by providing longer retention
time, a secondary combustion zone, and secondary combustion air. Fireplace
design, however, favors more complete combustion because the combustion air is
not as restricted as it is in wood-burning stoves. Indeed, examination of the
wood-burning rate revealed that wood was consumed in the fireplace at a rate
40 percent greater than in the stove, evidence of a hotter fire and better
combustion.
142
-------
The two stoves had similar emissions, but differences were noted between
stove and fireplace emissions. CO and POM emission factors were an order of
magnitude higher from the stoves than from the fireplace, as shown below in the
data obtained during combustion of seasoned oak:
CO Emission POM Emission
Factor Factor
Equipment g/kg g/kg
Fireplace 30 0.025
Baffled Stove 110 0.21
Nonbaffled Stove 370 0.19
Since both CO and POM are products of incomplete combustion, they are expected
to be emitted in greater amount under the poorer combustion conditions of the
stove.
The data on particulate matter and hydrocarbon emissions were highly variable,
and no trends were noted in the emissions of these species. It is unclear why
particulates and hydrocarbons did not exhibit higher emission factors from the
stoves, since they too form as a result of incomplete combustion.
NO2 emissions from the fireplace were approximately four times greater
than NO2 emission from the stoves. Since NO2 emissions depend primarily on
combustion temperatures (as long as sufficient excess air is present for
complete combustion), it is not surprising that higher N0£ emission factors
occurred during the fireplace tests where hotter temperatures prevailed.
No significant differences were noted in emissions between the two airtight
stoves tested. The baffled stove, despite its design, did not increase combus-
tion efficiency, as evidenced by the similar emission factors and energy
efficiencies of both stoves.
The data obtained are in general agreement with other studies of this
source type. Caution should be exercised, however, in extrapolating these
results to other test conditions. Further studies are recommended to provide
information on such variables as wood geometry, firing rate, and air/fuel ratio.
143
-------
Source Assessment Project - NO Emissions from Wood-Fired Industrial
j —x
Boilers - Results from a CCEA study performed by TRW, Inc. (EPA-600/7-79-219;
NTIS PB 80-102288) indicate that wood-fired boilers emit considerably less NO^
(on an energy basis) than fossil fuel boilers of comparable size. The study,
"NOx Emission Factors for Wood-Burning Boilers," points out that this conclusion
is not expected, since most wood-fired boilers operate with high levels of excess
air, a practice normally associated with increased NO^ emissions.
These findings are based on source test data obtained from 14 industrial
boilers firing either wood alone or in combination with oil, coal, or natural
gas* The types of fuel woods studied were mostly wood processing residues,
such as sawdust, chips, Bhavings, edgings, bark, and scraps. The boilers
tested ranged in size from 1.5 to 67 MW (4,450 to 200,000 lb steam/hour).
Test data for each source were used to determine NO emission rates in
X
the units of g/kg fuel and mg/MJ. The boilers were then separated into three
size categories, on the basis of the mean emission rates within the categories.
Emission factors are shown below, based on the test data from boilers in
each size category. Most of the large wood-fired boilers are co-fired with a
conventional fossil fuel. Although this study focused on boilers which use wood
as the primary fuel, emissions data from these co-fired systems were Included in
the development of emission factors. A separate emissions factor was developed
for boilers firing wood as a minor supplement to coal.
Boiler Size 8 NO^/kg Fuel me N0^/10^ joule
<10.0 MW 0.03 9.0
>10.0 MW 1.60 70.0
>10.0 MW* 4.00 170.0
* Wood used as supplement to coal
144
-------
The unexpected finding that NOx emissions from wood-fired boilers generally
decrease with higher levels of excess air can be explained by the fact that,
in wood-fired boilers, high levels of excess air reduce the fire box temperature.
This in turn decreases the rate at which thermal NO is formed. (Thermal NO
X X
is the NO formed by oxidation of atmospheric nitrogen in the combustion air.)
The study found a definite correlation between emission rate and boiler
size. Emission rates for boilers smaller than 10 MW (29,670 lb steam/hr) are
quite similar. This is not surprising in view of the uniform operating parameters
of these smaller boilers. Most operate with more than 300 percent excess air.
In addition, fire box temperatures generally range from 1000 to 1100°C
(1832 to 2012°F). Larger boilers (greater than 10 MW) exhibit less consistency
in operating parameters and emission rates. Excess air use ranges from 62 to
155 percent, and the fire box temperatures are unknown.
6. CONCLUSION
With conventional fuel combustion processes, principally coal combustion,
playing an increasing role in our movement toward national energy independence,
there is a simultaneous increase in the potential for adverse environmental impacts.
The preceding examples of field study data and assessment results are representative
of the extensive activities currently underway in the CCEA Program to ensure that
the country can increase its reliance on conventional combustion processes at
reasonable economic, energy, and environmental costs. The results of the CCEA
Program efforts, then, will be recommendations for control technology and support
for standards development to control adverse effects within acceptable limits.
I
145
-------
GLOSSARY
Discharge Severity - An indicator of potential environmental impact; the ratio
of the measured concentration of a pollutant in a discharge
stream to a potentially hazardous concentration
EACCS - Environmental Assessment of Conventional Combustion Systems
FGD - Flue Gas Desulfurization
GC - Gas Chromatograph(ic)
NAAQS - National Ambient Air Quality Standards
NO^ - Generic formula for oxides of nitrogen; includes NO and NO2
OAQPS - (EPA's) Office of Air Quality, Planning, and Standards
POM - Polycycllc Organic Matter
Source Severity - An indicator of potential environmental impact; the ratio of
the maximum ground level concentrations resulting from a
source to a potentially hazardous concentration
146
-------
COMBUSTION MODIFICATION
ENVIRONMENTAL ASSESSMENT
By:
C. Castaldini, R. M. Evans, E. B. Higginbotham,
K. J. Lim, H. B. Mason, and L. R. Waterland
Acurex Corporation
Mountain View, California 94042
147
-------
ABSTRACT
The Combustion Modification Environmental Assessment (CMEA) was
started in 1976 as part of the Environmental Protection Agency's (EPA)
Conventional Combustion Environmental Assessment Program. The primary CMEA
objectives are to:
• Identify potential multimedia environmental hazards from
stationary combustion sources before and after the use of
combustion modifications to control NO and other
x
combustion-related pollutants
• Develop combustion modification application guidelines
documenting the economic, energy, operational and environmental
impacts of meeting prescribed emission levels
• Identify the most cost-effective and environmentally acceptable
combustion modification techniques to achieve and maintain
environmental goals for NO^
To support these objectives, the emphasis in the CMEA is on field tests to
quantify changes in emissions, energy efficiency, and operation due to the
use of combustion modifications. The field testing uses the EPA
environmental assessment "Level 1" protocol which includes sampling and
analysis for N0x, SO^, SO^, CO, CO^, 0^, trace metals, organics,
and trace inorganic species» During the first 3 years of the CMEA, field
tests were done on three utility boilers, two industrial boilers, a gas
turbine, and a residential warm air furnace. Each source was either
modified in the field for low NO operation or was equipped with low NO
X X
designs. Test results showed no major increase in emissions due to
combustion modifications. Changes in emissions other than NO^ were
148
-------
typically within the accuracy of the experimental methods, or within the
range of changes due to day-to-day variations in fuel composition or unit
operation.
Changes in the severity to the environment of total source effluents
was secondary to the improvement due to NO^ reduction. Energy efficiency
was generally unimpared or improved through the use of combustion
modifications. One exception was a water injection equipped gas turbine for
which a 2 percent efficiency decrease was observed.
The CMEA program has recently been renewed to extend the field test
program to additional sources} advanced combustion modification controls,
alternate fuels, and nonsteady operation. The site selection and field test
status for the extended program are described.
149
-------
SECTION 1
INTRODUCTION
In 1975, EPA's Industrial Environmental Research Laboratory (IERL)
started a major program for the environmental assessment of energy systems
and industrial processes. The purpose of these assessments is to detect and
quantify potential environmental problems with the systems or processes and
identify potential control measures to reduce the environmental problems
found. This information is needed by EPA and other agencies to establish
R&D priorities, to support standards setting activities by regulatory
groups, and to develop environmentally acceptable energy systems.
Environmental assessments of stationary conventional combustion
sources are coordinated by the "Conventional Combustion Environmental
Assessment" (CCEA) program managed by IERL-Research Triangle Park (RTP). A
major component of the CCEA is the CMEA. The CMEA was started by IERL-RTP's
Combustion Research Branch in June 1976 to support the overall EA program by
focusing on stationary combustion sources with combustion modification
techniques to control NO^ or other pollutants amenable to control through
combustion process modification.
The three primary objectives of the CMEA are to:
• Identify potential multimedia environmental hazards from
stationary combustion sources
— Under baseline operation without combustion modification
controls
— Under controlled operation to suppress NO^ or other
pollutants amenable to control through combustion process
modification
• Develop control application guidelines on the economic, energy,
and operational impacts of meeting prescribed emission levels
150
-------
• Identify the most cost-effective and environmentally acceptable
N0x control techniques to achieve and maintain air quality
considering:
— Current and anticipated air quality standards
— Alternate equipment use and fuel use scenarios to the
year 2000
The program approach to address these objectives is illustrated in
Figure 1. Here, the rectangles denote major tasks while the ovals denote
outputs. The flowchart at the left of Figure 1 shows the approach for the
first two objectives, while the approach for the third objective is shown by
the flowchart at the right.
Since 1976, the CMEA program objectives have been accomplished for
the major stationary sources firing conventional fuels and equipped with
conventional combustion modifications. The overall results from the initial
3-year effort are summarized in Reference 1. The emission characterization
effort shown in Figure 1 was documented in References 2 and 3 and updates
were supplied in References 1 and 4. The test program results are
summarized in a series of test reports, References 5 through 11, which will
be available through the National Technical Information Service in the fall
of 1980. The process engineering and impact analysis results are documented
in a series of source-specific reports for utility boilers, industrial
boilers, gas turbines, residential heating systems, and internal combustion
engines (12 through 16). The preliminary air quality analyses to identify
combustion modification R&D priorities were initially documented in
Reference 17. These analyses were subsequently updated in References 1 and
4 using current regulatory policy and control technology information.
The CMEA has recently been extended to augment the initial results by
evaluating more advanced control technologies, secondary NO^ sources,
alternate fuels, and nonsteady operation. Although the emphasis in the
extended program is on field testing, the other program elements in Figure 1
will be updated as more current information becomes available.
This paper presents the field test program results from the first
3—year effort and summarizes the plans for the first year of the extended
program. Section 2 describes the sources tested and the sampling and
151
-------
analysis protocols used in the initial field test program, and Section 3
presents the results. The site selection and field testing for the extended
program are reviewed in Section 4.
152
-------
SECTION 2
INITIAL TEST PROGRAM
The purpose of the CMEA test program is to quantify how source
emissions, efficiency, and operation are affected by combustion modification
techniques. To help formulate the test program, a preliminary environmental
assessment (17) was conducted during the first year of the CMEA. This
assessment surveyed available data, identified data gaps, and set priorities
on sources, fuels, and controls for the initial CMEA effort. The assessment
showed a virtual absence of data on the effects of combustion modifications
on noncriteria flue gas pollutants and on solid or liquid effluents. The
assessment concluded that comprehensive field tests were needed to show the
effects of combustion modifications on vapor phase hydrocarbons, particulate
load and size distribution, sulfur species, vapor and condensed phase trace
elements, and vapor or condensed phase organic species. The preliminary
assessment further concluded that priority in the initial field tests should
be on major source categories, conventional fuels, and conventional
combustion modification techniques.
Based on the results of the preliminary source/control priorities
established in the first year of the CMEA, 19 candidate field tests were
identified. From the 19 potential tests, seven were selected and tested. A
summary of these seven tests is given in Table I. Where possible, the CMEA
tests were done as an augmentation to ongoing tests done as part of other
programs.
The test plan developed for each test called for sampling all
influent and effluent streams with the exception of the ambient air.
Continuous monitors were employed to measure flue gas NO , CO, CO-, and
X /
02. EPA Level 1 procedures (18, 19) were used to sample the flue gas and
all discharge ash streams for trace element and organic species. In
153
-------
addition, flue gas particulate and sulfur species concentrations were
measured in each test. Flue gas sampling upstream and downstream of
particulate collection devices were always performed. An example of the
samples collected during a test are shown in Figure 2. Operating data
sufficient to calculate unit efficiency, cycle efficiency, and verify
consistent operation, were also recorded.
For each test the following environmental assessment sampling
protocol was used:
• Continuous monitoring of flue gas NO^, CO, CO2, and 0^
(SO2 was only measured continuously during one test)
• Flue gas Source Assessment Sampling System (SASS), EPA Method 5
particulate load, and EPA Method 8 (or equivalent) sulfur species
sampling; both upstream and downstream of the particulate
collector, if applicable
• Flue gas grab sampling and onsite gas chromatographic analysis
for C,-C, hydrocarbons; both upstream and downstream of the
i o
particulate collector, if applicable
• Bottom ash slurry sampling
• Particulate collector hopper ash sampling
• Fuel and fuel additive if applicable sample collection,
• Operating data collection
As noted in Table I, the test program was conducted, at a minimum, for at
least two conditions of source operation: baseline (uncontrolled), and low
NO^ operation. In several instances, operation at intermediate levels of
NO^ control was tested. In addition, replicate testing was performed in
selected cases.
A key part of the test program involved close monitoring of source
operating data. This was done not only to ensure that test conditions
remained constant and representative of acceptable source operation over the
duration of sample collection, but also to provide the necessary input to
further process analysis efforts.
Laboratory chemical analyses of samples collected generally followed
IERL-RTP defined Level 1 procedures (18, 19) with a few exceptions and
additions. Level 1 is a semiqualitative screening approach used in
154
-------
environmental assessments to identify areas needing further analysis. - A
simplified schematic of the analysis scheme adopted for SASS train Level 1
samples is illustrated in Figure 3. The analysis scheme for solid (ash)
samples and the general organic analysis scheme is described in
Reference 1.
A specific exception to the Level 1 protocol dealt with sample trace
element analysis. Here, instead of assaying for trace elements by spark
source mass spectrometry, atomic absorption spectroscopy was employed to
determine the 25 more commonly occurring elements listed in Table II.
Another exception dealt with organic analyses of flue gas (XAD-2 extract),
particulate, and liquid/solid samples. Here the analyses were extended,
when feasible, to the determination of polycyclic organic compounds (POM)
and later to the determination of the priority pollutant compounds listed in
Table III.
Following the procedures of the Level 1 analysis, data listed below
can be obtained for each test point:
• Continuous flue gas N0X» CO, CO^, and 0^
• Flue gas SC>2> SO^ and speciated hydrocarbons
• Flue gas particulate load and size distribution
• Flue gas vapor phase trace element composition for the
25 elements listed in Table II
• Flue gas >C^ organic composition in terms of seven compound
polarity fractions and flue gas composition for the species
listed in Table III
• Particulate composition for the 25 elements listed in Table II
and the six ionic species listed in Table IV, as a function of
particulate size
• Particulate organic composition in terms of seven polarity
fractions, and for the species listed in Table III, as a function
of particulate size
• Liquid/solid stream (bottom, hopper ash) composition for the
25 elements listed in Table II and the six ionic species listed
in Table IV
• Liquid/solid stream (bottom, hopper ash) composition for seven
polarity fractions and for the species listed in Table III
155
-------
• Particulate and ash combustible material content
• Fuel proximate and ultimate analysis (heating value, and water,
C, H, 0, N, and S content)
• Fuel trace element content for the 25 elements listed in Table II
The above data satisfy the specific CMEA program needs identified.
Specific attention was focused on obtaining data on emitted POM, SO^ and
condensed sulfate, and trace element levels as a function of particulate
size, especially as these are affected by combustion modification
control applications.
Bioassay testing in accordance with IERL-RTP guidelines (20) was
performed on samples collected during the gas turbine, Crist Unit 7, Site B,
and the Moss Landing Unit 6 tests. The bioassays are conducted on samples
collected during the controlled (for NO^) tests only. The general Level 1
bioassay protocol for the CMEA tests is given in Table V. This test
protocol includes both health effects and ecological effects tests.
However, sample size requirements for certain tests are substantial. Thus
performing certain tests was often precluded. The actual bioassays
performed in the test program are listed in Table VI.
156
-------
SECTION 3
TEST RESULTS
Detailed sampling and analysis results from the initial test program
are documented in References 5-11. For this paper, the results will be
represented in terms of the Source Analysis Model (21). The Source Analysis
Model compares discharge stream specie concentrations to threshold
concentrations of these species. For the purposes of screening pollutant
emissions data to identify species requiring further study, a discharge
severity (DS) is defined as follows:
Ds » Concentration of Pollutant i in Effluent Stream
i Discharge Multimedia Environmental Goal
The discharge multimedia environmental goal (DMEG) values describe
maximum concentrations believed to be safe for short-term direct exposure to
a discharge stream .(21). The DMEGs are used in EPA's environmental
assessment programs to indicate when a specie concentration is sufficiently
high to warrant further evaluation. Stream discharge severity is evaluated
as the sum of the individual specie discharge severities.
To compare waste stream potential hazards, a weighted discharge
severity (WDS) is defined as follows:
WDS » (E. DS.) x Mass Flowrate,
i i
where the discharge severity is summed over all species analyzed. The
weighted discharge severity is an order of magnitude indicator of hazardous
pollutant release and can be used to rank the needs for controls for waste
streams. It can also be used as a preliminary measure of how well a
pollutant control, say a combustion modification NO control, reduces the
X
overall environmental hazard of the source.
157
-------
Tables VII and VIII present a summary of the CMEA test results noting
the pollutants of concern. The tables show pollutant components with
discharge severity (DS) values greater than 1 for any source tested, with an
indication of the magnitude of the discharge severity for each component for
each source. Table VII presents results for the flue gas stream; Table VIII
for the ash streams.
Frcm Table VII, it is apparent that SC^ and NO^ emissions present
the greatest potential hazard from all the combustion sources tested. The
discharge severity for these is especially high from coal-fired sources.
Other species with high discharge severity in most tests include CO^, CO,
As, and SO^ (vapor phase). The only organic emissions of potential
concern noted were those of carboxylic acids; though this would be so only
under the conservative assumption that all the organics analyzed in SASS
train catches consisted of the most toxic carboxylic acid on the EMEG list;
maleic acid. Even so, the discharge severity values are only of order 1.
Several other trace element species and condensed sulfate were flagged of
concern in several tests, though these were not universally noted.
Table VIII shows that the potentially most hazardous species in the
ash streams from coal-fired sources were iron and manganese, followed by
chromium, nickel, beryllium, and barium. Interestingly, lead levels were
high only in particle collector ash streams, particularly the ESP hopper
ash, suggesting that lead partitions to potentially hazardous levels on
passage through a boiler.
Table IX compares the stream discharge severities. The stream
discharge severities for a given source are generally within an order of
magnitude of each other. In Table X, the stream discharge severity is
weighted by the stream flowrate. It is clear that the flue gas stream
dominates the effluent streams.
General conclusions that can be derived from the SAM IA discharge
severity evaluations include:
• NO^ and SO^ represent the potentially most hazardous flue gas
species from the sources tested; the sum of the discharge
severity values for these two species in general accounts for
greater than half the stream discharge severity
158
-------
• The flue gas stream weighted discharge severity dominates the
source total weighted discharged severity for sources which
discharge other (ash) streams (coal-fired sources)
• NO^ control application either reduces flue gas discharge
severity and weighted discharge severity, or, at worst, does not
increase these; this translates to a reduction, or, at worst, no
increase in source total weighted discharge severity
• In general, changes in flue gas discharge severity and weighted
discharge severity due to NO^ control are less significant than
those resulting from day-to-day variations in fuel composition
(especially sulfur)
Carbon Monoxide and Vapor Phase Hydrocarbons
The presence of CO and vapor phase hydrocarbons (HC) in the exhaust
gases of combustion systems results from incomplete fuel combustion. The
various combustion modification controls tested can give rise to conditions
resulting in incomplete combustion, so increased emissions of CO and HC can
be a concern in NDx control application. However, since emissions of CO
and HC are associated with decreased efficiency, combustion sources are
generally operated, even with NDx control, to keep these emissions at a
minimum. The preliminary environmental assessment (17) concluded that
increased emissions of these due to combustion modification control should
not be considered a major concern. Results of the CMEA test program
substantiate this.
Table XI shows CO and HC emissions as a function of NO control
x
application for the tests performed. As noted in the table, emissions of
these either remain relatively unchanged or increase only slightly with the
combustion modifications tested.
Particulate and Particle Size Distribution
The preliminary environmental assessment (17) concluded that the
effects of combustion modification controls on particulate emissions, and
especially on emitted particle size distribution, from stationary sources
had been insufficiently studied. Further, since NO^ controls can produce
combustion conditions conducive to increasing particulate emissions, this
was flagged as an area of potential concern.
159
-------
The data obtained in the CMEA test program on particulate emissions
are summarized in Table XII. The table shows that particulate emissions
generally remain unaffected, or are decreased slightly with the NO^
control applications tested. Particle size distribution data taken for the
coal-fired sources also showed that emitted particle size distribution
remained unaffected or increased slightly (5, 7, 8, 10).
Trace Element Emissions
The preliminary environmental assessment (17) concluded that the
effects of NO^ control application on emissions of trace elements, both
segregating (those \rtiich tend to partition to fine particulate) and
nonsegregating (those which tend to remain equally distributed with particle
size in ash), should be marginal. However, the preliminary assessment noted
that few data existed to substantiate that conclusion.
The data obtained in the CMEA test program, however, do indicate that
combustion modification controls have no measurable effect on trace element
emissions. The test program shows that, within analytical uncertainties,
trace element levels in flyash, bottom ash, and particulate collector hopper
ash streams remain generally unchanged as a function of MO^ control
application.
In addition, data presented in References 5, 7, 8 and 10 also
indicate that:
• Trace element partitioning to fine particulate occurs in
accordance with expectations (17), but
• Changes in trace element partitioning tendencies with combustion
modifications are undetectable
SOj and Sulfate Emissions
High ambient sulfate levels are currently a matter of great concern
in regions of the U.S. with large numbers of combustion sources firing
sulfur bearing coal and oil. The primary reason for this concern is that
increasing ambient sulfate levels are contributing to the problem of acid
precipitation, particularly in the northeast U.S., but in other parts of the
country as well. Ambient sulfates are comprised of directly emitted, or
primary, sulfates and those derived from the atmospheric oxidation of SO^i
160
-------
or secondary sulfates. Although the relative contributions to ambient
sulfate levels from each of these is undetermined, it is clear that an
increase in primary sulfate emissions should be viewed with concern.
Combustion modification NO^ control would be expected to either
reduce or leave unchanged the emissions of primary sulfates from combustion
sources. Such was the conclusion of the preliminary environmental
assessment (17). However, the preliminary assessment also noted that few
data to substantiate this conclusion existed, thus obtaining such data was
given priority.
Table XIII shows the flue gas SO^ and particulate sulfate data
taken during the test program. Also shown is the weight fraction of sulfur
emitted as SO^ and S0£, expressed as the ratio (SO^ + S0^)/(S02 +
SO, + SOf). The table shows that the fraction of sulfur emitted as
3 4
SO^ and particulate sulfate varies from just under 1 percent for
coal-fired sources to 5 to 10 percent for oil-fired sources, in agreement
with previous data (22). The table also shows that the fraction of sulfur
emitted as SO^ + SOj~ remains relatively constant, or decreases slightly
with N0x control application.
Polycyclic Organic Matter and Other Organic Emissions
Just as combustion modification controls have the potential for
increasing CO and HC emissions due to decreased combustion efficiency,
emissions of other organic species can potentially increase also. Since few
data existed on the effects of NO^ control on combustion source PCM and
other organic emissions, and since several species in this pollutant class
are quite hazardous, priority was given to obtaining data on these emissions
in the CMEA test program.
Table XIV shows total SASS train organic determination data for the
tests performed. Infrared analysis of sample extracts showed the organic
species in the samples were in the aliphatic hydrocarbon, ether, ester,
aromatic, and carboxylic acid categories. The data in the table indicate,
though, that as was the case for CO and HC emissions, emission of these
higher molecular weight organics remain relatively unchanged with NO
control application. The seemingly high levels of organic compound
emissions in the Blueray residential furnace test are primarily unburned
161
-------
fuel oil. This was a result of the "on-off" cycling of the furnace during
the test.
Table XV shows the results of the Gas Chromatography/Mass
Spectrometry (GC/MS) analysis of SASS train samples for the tests for which
the analysis was performed. In all these tests at least 11 POM species were
screened. For the Crist Unit 7 and Moss Landing Unit 6 tests the POM
species and organic priority pollutants listed in Table III were analyzed.
Table XV shows that there is a marginal increase in POM emissions
with N0x control application. The emissions levels were generally iti the
order of the detection level of the instrument. It is interesting to note
that, in the Crist and Moss Landing analyses for the organic priority
pollutants, none of the organic priority pollutants were found within the
detection limits.
Bioassay Results
Bioassay testing of samples taken during the Crist Unit 7, Moss
Landing Unit 6, Site B and gas turbine lowest NO^ tests was performed in
the test program. Table XVI suranarizes results from all the bioassay tests
performed and notes the stream total discharge severity for the appropriate
sample assayed. However, from the data presented in the table no real
correlation between stream discharge severity and bioassay test results is
apparent. The fact that most bioassays gave nondetectable toxicity
responses frustrates deriving any correlation.
162
-------
SECTION 4
CURRENT TEST PROGRAM
The initial test program was based on filling priority data gaps
identified early in the CMEA. The identification of these data gaps led to
the testing of major source categories, conventional fuels, and conventional
combustion modification techniques. In the current test program, these
source categories will continue to be evaluated. In addition, the following
priorities will be addressed:
• Advanced NO controls
x
— Evaluation of controls with regard to the impending New
Source Performance Standard (NSPS)
— Evaluation of controls designated Best Available Control
Technology (BACT)
• Alternate fuels
• Secondary sources
« CCEA Program data needs
— Residential oil combustion
— Wood firing in residential, commercial, and industrial sources
— High interest emissions determinations (dioxins,
radionuclides, etc.)
• Nonsteady state operation
As in the initial program, operating data and emissions data will be
evaluated to determine the overall effects of combustion modification
controls.
The goal of the current program is to test approximately ten sources
per year. Whenever possible, cooperative test efforts with other
contractors will be encouraged so that the available data base will be
enhanced.
163
-------
At this time four field tests in the current program have been
completed. The sources tested were two stationary reciprocating internal
combustion engines (one spark ignition and one compression ignition) and two
low emission distillate oil-fired residential furnaces. These sources are
described in Table XVII. Other test programs currently scheduled for this
year include a pulverized coal-fired utility boiler designed to meet the
1971 NSPS, a small wood-fired industrial boiler, two industrial combustion
sources equipped with noncatalytic ammonia injection systems, a process
heater equipped with advanced staged combustion for NO^ control, and a
coal-fired industrial stoker with coal limestone pellets for control.
Results of these programs will be available in individual test reports under
the CMEA program. Additionally, an annual report will be available in
spring 1981 sunmarizing the program results.
164
-------
SECTION 5
SUMMARY
The CMEA field tests conducted to date indicate that:
• For the sources tested, the flue gas stream presents the greatest
potential environmental hazard
• N0x and SO^ are the potentially most hazardous flue gas
pollutants
• Flue gas discharge severity is decreased or, at worst, does not
increase with applying the combustion modifications tested;
changes in emissions due to day-to-day fuel composition changes
are often of greater magnitude than those attributable to NO
x
control
• The effluent streams from the sources tested are not mutagenic,
and, in general, elicit nondetectable toxicity in bioassay testing
• The combustion modifications tested:
— Have no effect, or increase only slightly, emissions of CO
and vapor phase hydrocarbon
— Have no effect on particulate mass emissions
— Have no effect, or tend to increase slightly, emitted
particle size distribution
— Have no measurable effect on trace element emissions or on
trace element partitioning tendencies
— Have no effect, or decrease slightly SO^ and particulate
sulfate emissions
— Have little effect on total higher molecular weight organic
emissions
— Marginal increase in POM emissions; but the emission levels
remained on the order of the detection levels of the
instrument
165
-------
• Emissions of the organic priority pollutants were below the
detection limit for the sources tested
It must be emphasized, though, that the sources were tested only
under steady operation, in short duration tests, and that the controls
tested were the relatively straightforward current technology combustion
modifications. Conclusions on the effects of advanced combustion
modification controls, and on the potential environmental impacts of
combustion sources under unsteady or transient operation must await results
from the current test program.
166
-------
REFERENCES
1. Waterland, L. R., et al., "Environmental Assessment of Stationary
Source N0X Control Technologies — Draft Final Report," Final
Report FR-80-57/EE, Acurex Corporation, Mountain View, California,
April 1980.
2. Salvesen, K. G., et al., "Emission Characterization of Stationary
NOx Sources: Volume I. Results," EPA/600/7-78-120a, NTIS PB
284-520, August 1978.
3. Salvesen, K. G., et al., "Emission Characterization of Stationary
NOx Sources: Volume II. Data Supplement," EPA-600/7-78-120b, NTIS
PB 285-429, August 1978.
4. Mason, H. B., et al., "Environmental Assessment of Stationary Source
NOx Control Technologies, In: Proceedings of the Third Stationary
Source Combustion Symposium, Volume IV, Fundamental Combustion
Research and Environmental Assessment," EPA-600/7-79-050d, NTIS PB
292-542 San Francisco, California, February 1979.
5. Higginbotham, E. B. and P. M. Goldberg, "Field Testing of a
Tangential Coal-Fired Utility Boiler — Effects of Combustion
Modification N0X Control on Multimedia Emissions," Acurex
Report 79-337, Acurex Corporation, Mountain View, California, April
1979.
6. Higginbotham, E. B., "Field Testing of a Low-Emission Oil-Fired
Residential Heating Unit," Acurex Report 79-15/EE, Acurex
Corporation, Mountain View, California, August 1979.
7. Goldberg, P. M. and E. B. Higginbotham, "Field Testing of an
Industrial Stoker Coal-Fired Boiler — Effects of Combustion
Modification N0X Control on Emissions-Site A," Acurex report
TR-79-25/EE, Acurex Corporation, Mountain View, California,
August 1979.
8. Lips, H. I. and E. B. Higginbotham, "Field Testing of an Industrial
Stoker Coal-Fired Boiler, Effects of Combustion Modification N0X
Control on Emissions-Site B," Acurex Report TR-79-18/EE, Acurex
Corporation, Mountain View, California, August 1979.
9 Larkin, R. L. and E. B. Higginbotham, "Field Testing of a Simple
Cycle Stationary Utility Gas Turbine — Effects of Water Injection
for NO* Control on Emissions and Unit Operations," Acurex Report
79-358/EE, Acurex Corporation, Mountain View, California, June 1979.
10. Sawyer, J. W. and E. B. Higginbotham, "Field Testing of a Pulverized
Coal-Fired Utility Boiler — Effects of Combustion Modification N0X
Control on Multimedia Emissions," Acurex Report 79-19/EE, Acurex
Corporation, Mountain View, California, September 1979.
167
-------
11. Sawyer, J. W. and E. B. Higginbotham, "Field Testing of an Oil- and
Gas-Fired Utility Boiler — Effects of Combustion Modification N0X
Control on Emissions," Acurex report 79-361/EE, Acurex Corporation,
Mountain View, California, July 1979.
12. Lim, K. J., et al., "Environmental Assessment of Utility Boiler
Combustion Modification N0X Controls," EPA-600/7-80-075a, b,
April 1980.
13. Liin, K. J., et al., "Environmental Assessment of Industrial Boiler
Combustion Modification N0X Controls," Acurex Draft Report
TR-79-10/EE, Acurex Corporation, Mountain View, California, July 1979.
14. Larkin, R. L., et al., "Environmental Assessment of Combustion
Modification Controls for Stationary Gas Turbines," Acurex Report
TR-79-18/EE, Acurex Corporation, Mountain View, California, September
1979.
15. Castaldini, C., et al., "Environmental Assessment of Combustion
Modification N0X Controls for Residential and Commercial Heating
Systems," Acurex report 79-17/EE, Acurex Corporation, Mountain View,
California, September 1979.
16. Lips, H. I., et al., "Environmental Assessment of Combustion
Modification N0X Controls for Stationary Internal Combustion
Engines," Acurex Draft Report TR-79-14/EE, Acurex Corporation,
Mountain Viev, California, July 1979.
17. Mason, H. B., et al., "Preliminary Environmental Assessment of
Combustion Modification Techniques: Volume I. Summary, Volume II.
Technical Results," EPA-600/7-77-119a and b, NTIS PB 276-680 and
276-681, October 1977.
18. Hamersaa, J. W., et al., "IERL-RTP Procedures Manual: Level 1
Environmental Assessment," EPA-600-2-76-160a, NTIS PB 257 850/AS,
June 1976.
19. Lentzen, D. E., et al., "IERL-RTP Procedures Manual: Level 1
Environmental Assessment (Second Edition)," EPA-600-7-78-201, NTIS PB
293-795, October 1978.
20. Duke, K. M., et al., "IERL-RTP Procedures Manual: Level 1
Environmental Assessment Biological Tests for Pilot Studies,"
EPA-600/7-77-043, NTIS PB 268-484, April 1977.
21. Schalit, L. M. and Wolfe, K. J., "SAM/1A: A Rapid Screening Method
for Environmental Assessment of Fossil Energy Process Effluents,"
EPA-600/7-78-015, NTIS PB 277 088/AS, February 1978.
22. Cleland, J. 6. and G. L. Kingsbury, "Multimedia Environmental Goals
for Enviromental Assessment: Volumes I and II, EPA-600/7-77-136a,b,
'November 1977.
168
-------
23. Homolya, J. B., et al, "A Characterization of the Gaseous Sulfur
Emissions from Coal and Oil-Fired Boilers," presented at the 4th
National Conference on Energy and the Environment, Cincinnati, OH,
October 1976.
24. Surprenant, N. F., et al., "Emission Assessment of Conventional
Stationary Combustion Systems: Volume 1. Gas- and Oil-Fired
Residential Heating Sources," EPA-600/7-79-029b, May 1979.
25. "Compilation of Air Pollutant Emission Factors, Third Edition
Including Supplements 1-7," EPA Publication AP-42, U.S. Environmental
Protection Agency, October 1977.
169
-------
Twt
Mr
Mr
Figure 1. CMEA program approach*
-------
COAL
BOILER
ESP
scales
D
E
&
A
Alt?
F
STACK
B.C
PREHEATED
Sampling Location
A — Coal Scales
B — Sluice Water Inlet
C — Bottom Ash Outlet
D — Airheater Inlet
E - ESP Inlet
F ~ ESP Hopper
G — ESP Outlet
Type of Sample
Grab Sample
Grab Sample
Grab Sample
Gas Sample
Gas Sample
Grab Sample
Gas Sample
Coal
Sluice Water Blank
Bottom Ash Slurry
Continuous Monitors
SASS, EPA 5/8, Gas Grab Sample for
onsite GC analysis of C, - Cc HC
ESP Ash 1 b
SASS, EPA 5/8, Gas Grab Sample for
onsite GC analysis of C^ - Cg HC
Figure 2. Sampling locations.
-------
Liquids
Gases
Solids
Level 1 Sample
Organic
Extract Aqueous
Samples with CH2CI2
GC for bp >100°C
IR
LC/IR/LRMS
Inorganic
• Elements
- SSMS
- Atomic Absorption
» Selected Anions
• Aqueous
- Selected Tests
- Ion Chromatography
Organic
• GC for bp < 100°C
t Organic sorbent
extract
- GC for bp > 100°C
- IR
- LR/IR/LRMS
Inorganic
Elements
- SSMS
- Atomic Absorption
Leachable Material
- Ion Chromatography
- Reagent Test Kit
Organic Extracts
• GC for bp > 100°C
• IR
• LC/IR/LRMS
Inorganic
t GC NH3, HCN, (CN2)
• NO*. continuous Moni-
tor or Method 7
• CO, CO2, O2 continu-
ous monitor or GC
• SO2. SO3 Method 8 or
controlled condition
• Impingers SSMS Atomic
Absorption
Figure 3. Level 1 analysis overview.
-------
TABLE I. CMEA FIELD TEST PROGRAM
Source Category
Description
Test Points
(Unit Operation)
Sampling Protocol
Test
Collaborator
Coal-fired
Utility Boiler
Kingston #6; 180 MW
tangential; twin
furnace, 12 burners/
furnace, 3 elevations;
cyclone, 2 ESP's for
particulate control
Baseline
Biased Firing (2)
BOOS (2)
Continuous NOx, SOji
CO, COj, Oj
Inlet to 1st ESP:
— SASS
— Method 5
— Method 8
— Gas grab (Cj-Cg HC)
Outlet of 1st ESP
— SASS
— Method 5
— Method 8
— Gas Grab (C^-Cg HC)
Bottom ash
Hopper ash (1st ESP,
cyclone)
Fuel
Operating data
TVA
Coal-fired
Utility Boiler
Crist #7; 500 MW
opposed wall-fired; 24
burners, 3 elevations;
ESP for particulate
control
Baseline
BOOS (2)
Continuous N0X, CO
CO2, O2
ESP inlet
— SASS
— Method 5
— Method 8
— Gas grab (C^-Cg HC)
ESP Outlet
— SASS
— Method 5
— Method 8
— Gas Grab (q-C6 HC)
Bottom ash
ESP hopper ash
Fuel
Operating data
Bioassay
Exxon
Oil-fired
Utility Boiler
Moss Landing #6;
740 MW opposed wall-
fired; 48 burners,
6 elevations
Baseline
FGR
FGR + OFA
Continuous N0X, CO
CO21 O2
— SASS
— Method 5
— Method 8
— Gas grab (Cj-Cg HC)
Fuel
Operating data
Bioassay
None
BOOS — Burner out of service 0*A -- Overfire air injection
FGR — Flue gas recirculation LEA —• Low excess air
173
-------
TABLE I. Coneluded
Source Category
Description
Test Points
(Unit Operation)
Sampling Protocol
Test
Collaborator
Coal-fired
Industrial
Boiler
Traveling grate
spreader stoker, 38
kg/s <300,000 lb/hr)
Baseline
LEA + high OFA
Continuous N0X, CO
CO21 O2
Boiler exit:
— SASS
— Method 5
— Shell-Emeryville
— Gas grab (C^-Cg HC)
ESP outlet
— SASS
— Method 5
— Shell-Emeryville
— Gas grab (Cj-Cg HC)
Bottom ash
Cyclone hopper ash
Fuel
Operating data
KVB
Coal-fired
Industrial
Boiler
Traveling grate
spreader stoker, 25
kg/s (200,000 lb/hr)
ESP for particulate
control
Base line
LEA
Continuous N0X, CO
CO2, O2
Boiler exit:
— SASS
— Method 5
— Shell-Emeryville
— Gas grab (Cj-Cg HC)
ESP Outlet
— SASS
— Method 5
— Shell-Emeryville
— Gas grab (Cj-Cg HC)
Bottom ash
ESP hopper ash
Fuel
Operating data
Bioassay
KVB
Oil-fired
Gas turbine
T. H. Wharton Station.
60 MW GE MS 7001 C
machine
Baseline maximum
water injection
Continuous CO
C0j» Oj
— SASS
— Method 5
— Method 8
Fuel
Water
Operating data
Bioassay -
General
Electric
Oil-fired
Residential
Beating Unit
Blueray low NOx
furnace, Medford,
New York
Continuous
cycling
Continuous N0X, CO
COj, Oj
— SAS8
— Method 5
— Method 8
Fuel
EPA/IERL-RTP
BOOS — Burner out of service OFA — Overfire air injection
FGR — Flue gas recirculation LEA — Low excess sir
174
-------
TABLE II. ELEMENTAL ANALYSIS: SPECIES DETERMINED
Antimony (Sb)
Mercury (Hg)
Arsenic (As)
Molybdenum (Mo)
Barium (Ba)
Nickel (Ni)
Beryllium (Be)
Selenium (Se)
Bismuth (Bi)
Tellurium (Te)
Boron (B)
Thallium (Tl)
Cadmium (Cd)
Tin (Sn)
Chromium (Cr)
Titanium (Ti)
Cobalt (Co)
Uranium (U)
Copper (Cu)
Vanadium (V)
Iron (Fe)
Zinc (Zn)
Lead (Pb)
Zirconium (Zr)
Manganese (Mn)
TABLE III. COMPOUNDS FOR WHICH GC/MS ANALYSIS WAS PERFORMED
Acenapthene®
3,3*-Dichlorobenzidine®
Acenapthalene®
Diethyl phthalate®
Anthanthrene®
7,12-Dimethyl benz(a)anthracene®
Anthracene®
Dimethyl phthalate®
Benzidine®
2,4-Dinitrotoluene®
Benzo(a)anthracene®
2,6-Dinitrotoluene®
3,4-Benzofluoranthene®
Di-n-octyl phthalate®
Benzo(k)fluoranthene®
1,2-Diphenylhydrazine®
Benzo(g,h,i)perylene®
Benzo(a)pyrene®
(as azobenzene)
Fluoranthene®
Benzo(e)pyrenea
Fluorene®
4-Bromophenyl phenyl ether®
Hexachlorobenzene®
Butyl benzyl phthalate®
Hexachlorobutadiene®
Bis(2-chloroethoxy) methane®
Hexach1oroeye1opentadiene®
Bis(2-chloroethyl)ether®
Hexach1oroethane®
Bis(2-chloroisopropyl) ether®
Indeno(l,2,3-c,d)pyrene®
2-Chloronaphthaiene
Isophorone
4-Chlorophenyl phenyl ether®
3-Methyl cholanthrene®
Chrysene®
Naphthalene®
Dibenzo(a,h)anthracene®
Nitrobenzene®
Dibenzo(c,g)carbazolea
N-nitrosodiphenylamine®
Dibenzo(a,h)pyrene®
N—nitrosodi—n—propylamine®
Dibenzo(a,i)pyrene®
Di-n-butyl phthalate®
Perylene®
Phenanthrene®
1,2-D icholorobenzene®
Pyrene®
1,3-Dichlorobenzene®
2,3,7,8-Tetrachlorodibenzo-
1,4-Dichlorobenzene®
p-dioxin
Bia (2-ethylhexyl) phthalate®
1» 2,3-Trichlorobenzene®
Compound in calibration standard.
175
-------
TABLE IV. IONIC ANALYSIS: SPECIES DETERMINED
Chloride (Cl~)
Fluoride (F")
Nitrate (NO3-)
Cyanide (CN~)
Sulfate (SO.2-)
4
Ammonia (NH.+)
4
TABLE V. BIOASSAY PROTOCOL
Sample Type
Bioassay Test Protocol
Sample Size
Requirements
SASS cyclones,
10+3
Microbial Mutagenesis
Cytotoxicity, RAM
l.Og
0.5g
SASS cyclones,
1 + filter
Microbial Mutagenesis
Cytotoxicity, RAM
l.Og
0.5g
XAD-2 extract
Microbial Mutagenesis
Cytotoxicity, WI-38® or CHO
50 mi
50 ml
Bottom ash
Microbial Mutagenesis
Cytotoxicity, RAM
Rodent Acute Toxicity
Freshwater Algal Bioassay
Freshwater Static Bioassay
l.Og
0.5g
100 g
50 kg
(200 I
if sluiced)
ESP hopper ash
Microbial Mutagenesis
Cytotoxicity, RAM
Rodent Acute Toxicity
Freshwater Algal Bioassay
Freshwater Static Bioassy
l.Og
0.5g
100 g
50 kg
®WI-38 was used in the initial CMEA program
176
-------
TABLE VI. BIOASSAY TESTS PERFORMED
Bioassay
Microbial
Freshwater
Freshwater
Field Test
Sample
Mutagenesis
RAM
WI-38
RAT
Algal
Fish
Crist Unit 7,
SASS cyclones
X
X
BOOS 2
10+3 ji
SASS cyclones
X
X
1 m + filter
Bottom ash
X
X
X
X
X
ESP hopper ash
X
X
X
X
X
Moss Landing
XAD-2 extract
X
X
Unit 6
BOOS+FGR
Site B Lew NOx
Bottom ash
X
X
X
X
X
ESP hopper ash
X
X
X
Gas Turbine
XAD-2 extract
X
X
Water Injection
BOOS2: Burner out of service test No. 2 N0X control
FGR: Flue gas recirculation
-------
TABLE VII. DISCHARGE SEVERITY SUMMARY FOR FLUE GAS
DS Level*
Kingston
Crist
Moss Landing
T.H. Wharton
Blueray
Component
Unit 6
Unit 7
Unit 6
Site A
Site B
Unit 52
Furnace
so2
+++
+++
++
++
++
+
+
N0X
+++
+++
++
++
++
+
-
co2
++
++
++
-
++
+
—
As
++
++
+
+
++
+
-
SO3 (vapor)
++
++
+
++
+
+
-
C°_
+
++
+
+
+
-
+
SO4 (condensed)
++
++
+
-
-
-
-
Fe
++
++
-
+
-
-
—
Carboxylic acids
-
+
+
+
-
-
+
Be
+
+
-
+
-
-
—
Co
+
+
+
-
-
-
-
Ba
+
+
-
-
-
-
-
B
-
-
+
+
-
-
-
Ti
-
+
-
+
-
-
-
Cu
+
-
+
-
-
-
-
Cd
-
-
-
+
-
+
-
Pb
-
-
-
+
-
-
-
CI
+
-
-
• -
-
-
-
U
~~
"¦'
+
aKey: +++ denotes DS >100
++ denotes DS >10
+ denotes DS >1
- denotes DS< 1 or species not measured
-------
Table VIII. DISCHARGE SEVERITY SUMMARY FOR ASH STREAMS
DS Level3
Kingston
Crist
Ash Stream
Component
Unit 6
Unit 7
Site A
Site B
Bottom ash
Fe
+++
+++
++
Mn
+
++
++
++
Cr
+
+
-
+
Ni
+
+
-
+
Be
+
+
-
-
Ba
+
-
-
-
T1
-
+
-
-
Sn
+
—
Mechanical
Fe
—
_b
++
+
collector
Mn
+
-
+
+
hopper ash
Be
+
—
-
+
Cr
+
—
—
+
Ba
+
-
-
-
Pb
-
-
-
+
Ni
+
«•»
—
ESP hopper ash
Fe
—
+++
++
Mn
+
++
++
-
Be
+
+
-
++
Ba
+
+
-
+
Pb
+
+
+
—
As
+
-
-
++
Ni
+
+
+
—
Cr
+
+
-
—
SO4
+
-
-
+
Se
—
—
—
++
T1
+
-
aKey: +++ denotes DS >100
++ denotes DS >10
+ denotes DS >1
denotes DS < 1 or species not measured
bNo mechanical collector at Crist Unit 7
179
-------
TABLE IX. TOTAL STREAM DISCHARGE SEVERITY: SUMMARY
Floe
Gas
Bottom Ash
Hopper Ash
Baseline
Low H0X
Baseline
Low H0X
Baseline
Low N0X
Kingston Unit 6
640
580
18
16
23
23
Crist Unit 7
740
830
160
170
150
120
Noes Landing Unit 6
160
130
a
a
a
a
Site A
150
210
120
120
110
98
Site B
170
110
b
82
45
120
T. H. Wharton Unit 52
87
52
a
a
a
a
Blueray furnace
70c
25
a
a
'
a
*Saople do applicable to this source
^Staple not analysed
cBaseline residential furnace evaluation baaed on conventional furnace data
(24, 25)
TABLE X. WEIGHTED DISCHARGE SEVERITY (kg/a) — KINGSTON UNIT 6
Baseline
Bias
BOOS
Flue gas
4.5 x 104
3.5 x 104
3.9 x 104
Cyclone ash
19
16
16
ESP ash
6.1
6.1
5.1
Bottom ash slurry
57
53
42
Total source
4.5 x 104
3.5 x 104
3.9 x 104
Bias: biased burner firing, intermediate N0X control
BOOS: burners out of service, low N0X operation
180
-------
TABLE XI. EFFECTS OF CONTROLS TESTED ON CO AND HC EMISSIONS
CO Emissions (ppm^)
HC Emissions (ppm^)
Test
Control3
Baseline
Intermediate
N0X
Low N0X
Baseline
Intermediate
N0X
Low N0X
Kingston Unit 6
Bias, BOOS
29
35
22
0
0
0
Crist Unit 7
BOOS
357
392
608
6.3
19.0
6.3
Moss Landing
Unit 6
FGR, BOOS/FGR
69
10
49
0
0
0.05
Site A
OFA
243
-
483
97
-
c
Site B
LEA
65
-
36
23
-
103
T.H. Wharton
Unit 52
WI
5.6d
-
8.1d
2.3e
-
3.5e
Blueray Furnace
New design
-
-
160
-
-
23
aBias: Biased burner firing; BOOS: Burners out'of service; FGR: Flue gas recirculation;
OFA: High overfire air; LEA: Low excess air; WI: Water injection
^3 percent 02» dry
cData not available
d15 percent 02» dry
e15 percent O2, wet
- Source not tested under this condition
-------
TABLE XII. EFFECTS OF CONTROLS TESTED ON PARTICULATE EMISSIONS
Test
Control®
Particulate Emissions (ng/J)
Baseline
Intermediate
NOx
Low NOx
Kingston Unit 6
Bias, BOOS
228
238
160
Crist Unit 7
BOOS
460
340
360
Moss Landing Unit 6
FGR, BOOS/FGR
23
22
16
Site A
OFA
13
-
24
Site B
LEA
8.0
-
10.0
T.H. Wharton Unit 52
WI
19
•-
16
Blueray Furnace
New design
—
—
1.3
aBias: Biased burner firing; BOOS: Burners out of service;
FGR: Flue gas recirculation; OFA: High overfire air;
LEA: Low excess air; WI: Water injection
- Source not tested under this condition
182
-------
TABLE XIII. EFFECTS OF CONTROLS TESTED ON FLOE GAS SO3 AND SO4 EMISSIONS
SO
j Emissions (ng/J)
so3 + S0=
—Z
SO
2 + S03 + S°l
V A /
Teat
Control®
Baseline
Intermediate
®x
Low H0Z
Baseline
Intermediate
nox
Low N0X
Baseline
Intermediate
H0X
Low N0X
Kingston
Unit 6
Bias, BOOS
4.3
3.9
4.5
8.6
4.4
11.0
0.85
1.24
0.98
Crist Onit 7
BOOS
9.5
4.5
3.6
11.6
9.6
8.2
0.75
0.50
0.36
Moss Landing
Unit 6
FGR,BOOS/FGR
0.96
0.96
1.3
1.6
1.6
1.3
4.1
4.0
3.9
Site A
OFA
2.2
—
12.0
b
—
b
0.70c
—
3.0C
Site B
LEA
0.8
—
0.5
b
—
b
0.46c
—
0.46c
T.H. Wharton
Dnit 52
HI
3.0
—
5.0
4.0
—
b
9.4
—
14.3C
Blueray
Furnace
New design
—
—
1.0
—
—
0
—
—
2.7
aBias: Biased burner firing; BOOS: Burners out of service; FGR: Flue gas recirculation; OFA: High overfire air;
LEA: Low excess air, WI: Water injection
^Not analyzed
cSO^ not included
— Source not tested under this condition
-------
TABLE XIV. EFFECTS OF CONTROLS TESTED ON FLUE GAS SASS ORGANIC EMISSIONS
Organic
Emissions (mg/dscm)
Test
Control®
Baseline
Intermediate
NOx
Low N0X
Kingston Unit 6
BOOS
0.124
—
0.834
Crist Unit 7
BOOS
4.23
2.320
0.722
Moss Landing Unit 6
FGR, BOOS/FGR
4.38
1.37
1.43
Site A
OFA
1.00
—
1.79
Site B
LEA
0.924
—
1.37
T.H. Wharton Unit 52
WI
1.30
—
1.10
Blueray Furnace
New design
—
—
26.3
aBOOS: Burners out of service; FGR: Flue gas recirculation;
OFA: High overfire air; LEA: Low excess air;
Wis Water injection
— Source not tested under this condition
184
-------
TABLE XV. EFFECTS OF CONTROLS TESTED ON FLUE GAS POM SPECIES EMISSIONS (yg/dscm)
Crist Unit 7
Moss Landing Unit 6
Site A
T.H. Wharton Unit 52
Blueray
Furnace
POM Species
Baseline
BOOS 2
BOOS 1
Baseline
FGR
BOOS/FGR
Baseline
Low N0X
Baseline
Water Injection
Cyclic
Chrysene/Benz
(a)anthracene
—
—
0.1
—
0.1
—
—
—
—
—
Fluoranthene
0.1
0.1
0.2
—
0.1
0.1
—
—
—
0.5
0.03
Fluorene
__
—
0.3
—
—
0.1
—
—
—
—
—
Naphthalene
—
0.9
0.3
—
—
—
—
—
—
1.0
—
Phenanthrene/
0.6
1.0
0.9
0.1
0.2
0.7
—
—
0.5
1.0
0.77
Anthracene
Pyrene
0.1
0.3
0.3
—
—
0.2
—
—
—
0.5
0.01
BOOS: Burners out of service; FGR ¦ flue gas recirculation
— Compound not present or below detection level
-------
TABLE XVI. BIOASSAY/DISCHARGE SEVERITY COMPARISON
Test/Sample
Bioassay Result8
Stream
DS
Microbial
Mutagenesis
Cytotoxicity
Rodent
Acute
Toxicity
Freshwater
Algae
Freshwater
Fish
RAM
WI-38
Crist Unit 7
>3 ym flyash )
830b
Neg.
ND
—
—
—
—
<3 vim flyash /
Neg.
L
—
—
—
—
Bottom ash
170
Neg.
ND
—
ND
ND
ND
ESP hopper ash
115
Neg.
ND
__
ND
ND
ND
Moas Landing Unit 6
XAD-2 extract
84
Neg.
—
M
—
—
—
Site B
Bottom ash
82
Neg.
ND
—
ND
ND
ND
ESP hopper ash
117
Neg.
L
—
ND
—
—
Gas Turbine
XAD-2 extract
13b
Neg.
—
L
—
—
—
•Neg: Negative, ND: Mot detectable; L: Law toxicity; M: Medium toxicity
'Total flue gas DS cited
— Test not conducted
186
-------
TABLE XVII. COMPLETED TESTS DURING THE FIRST YEAR OF THE CURRENT PROGRAM
Source
Description
Test Points
Unit Operation
Sampling Protocol
Test Collaborator
Spark ignited natural
gas-fired reciprocating
internal combustion
engine
Large bore, 6 cylinder,
opposed piston, 186.5 kW
(250 Bhp)/cyl, 900 rpa
Model 38TDS8-1/8
— Baseline (pre-NSPS)
— Increased air-fuel
ratio aimed at
meeting proposed
NSPS of 700 ppm
corrected to 15Z
Oj and standard
atmospheric con-
ditions
Engine exhaust:
— SASS
— Method 5
— Gas grab (Cj-Cg HC)
— Continuous NO, NOx, CO,
CO2, 0j, CH4, 1'UUC
Fuel
Lube oil
Fairbanks Morse
Division of Colt
Industries
Compression ignition
dieael-fired
reciprocating internal
coabustion engine
Large bore, 6 cyclinder
opposed piston, 261.1 kW
(350 Bhp)/cyl, 900 rpa
Model 38TDD8-1/8
— Baseline (pre-NSPS)
— Fuel injection
retard aimed at
meeting proposed
KSPS of 600 ppa
corrected to 15Z
O2 and standard
atmospheric con-
ditions
Engine exhaust:
— SASS
— Method 8
— Method 5
— Gas grab (Cj-Cj HC)
— Continuous NO, N0X, CO,
CO2, Oj, iubC
Fuel
Lube oil
Fairbanks Morse
Division of Colt
Industries
Lew N0X residential
condensing heating
system furnished by
Karlson's Blueburner
Systems, Ltd. of Canada
Residential hot water
heater equipped with
M.A.I. lew N0X burner
0.55 ml/s (0.5 gph)
firing capacity;
Condensing flue gas
Low BDX burner
design by M.A.N.
Furnace exhaust:
— SASS
— Method 8
— Method 5
— Gas grab (Cj-Cg HC)
— Continuous NO, NOg, CO,
CO2, O2, ximC
Fuel
Waste water
New test
Rocketdyne/EPA
Low N0X residential
forced vara air furnace
Sesidential warm air
furnace with Modified
high pressure burner and
firebox. 0.83 al/s
(0.75 gph) firing
capacity
Low N0X burner
design and integrated
furnace system
Furnace exhaust:
— SASS
— Method 8
— Controlled condensation
— Method 5
— Gas grab (Cj-Cg HC)
— Continuous NO, NOg, CO,
COj, O2, TUHC
Fuel
New test
-------
CHARACTERIZATION AND OXIDATION OF
DIESEL PARTICULATE
By:
D. A. Trayser, L. J. Hillenbrand, M. J. Murphy,
J. R. Longanbach, and A. Levy
BATTELLE
Columbus Laboratories
505 King Avenue
Columbus, Ohio 43201
188
-------
ABSTRACT
This study is being conducted for the Environmental Protection
Agency to evaluate emissions control on light-duty diesel vehicles by
postcylinder oxidation. The primary objective is to determine the feasi-
bility of thermal or catalytic oxidation as a means of diesel particulate
emissions control.
The program plan includes a review of the state of technology,
detailed chemical and physical characterization of the particulate from
a light-duty diesel engine, bench experiments to define the ignition and
oxidation properties of the particulate, experiments with catalytic igni-
tion of particulates, and experimental evaluation of concepts and devices
for particulate emission control by oxidation in the exhaust of an engine.
The particulate characterization is being carried out using an
Oldsmobile 4.3-liter diesel engine coupled to a dynamometer with direct
and diluted exhaust particulate sampling and measurement. Bench experiments
are being conducted with various types of hot-tube reactors and instrumenta-
tion.
The particulate characteristics being measured include: mass
concentration; soluble organic content; carbon, hydrogen, and ash; trace
mineral content; surface area; size distribution*, and volume concentration.
Results to date show that: as engine load is increased mass concentration
increases substantially and soluble organic content decreases, both hydrogen
content and ash content vary between 1/2 and 2 percent, the surface area is
2
approximately 100 m /g, and the mass median particle diameter increases with
load increase and is in the range of 0.1 to 0.3 ym.
Preliminary catalytic ignition results indicate that the ignition
temperature of the particulate can be substantially reduced (at least 150 C)
by application of small concentrations of metal salt solutions. In addition,
it has been found that the catalytic action of the metal salt is enhanced
by admixing salts such as sodium or ammonium chloride and nitrate. Copper
salts have been found to work best of the materials studied to date.
189
-------
ACKNOWLEDGMENTS
Battel!e-Columbus prepared this report as part of a dlesel
afterburner study for EPA, Contract No. 68-02-2629, with Mr. John Wasser
as Project Officer.
190
-------
SECTION 1
INTRODUCTION
Diesel engines are becoming available in increasing numbers in
passenger car service because of their good fuel economy in comparison to
conventional gasoline engines. Though the diesel exhaust is relatively
clean with respect to unburned hydrocarbons and carbon monoxide, it contains
particulate emissions that are 30 to 50 times greater than those produced
by the catalyst-equipped gasoline engine. These diesel particulate
emissions not only will contribute to already high levels of total suspended
particulate (TSP) in urban areas, but certain components of the particulates
have been identified as carcinogenic, thereby creating a potentially greater
health hazard.
The U.S. Environmental Protection Agency (EPA) has established
regulations on the amount of particulate that may be emitted by each
light-duty diesel vehicle. These standards, based on the presently used
Federal Test Procedure (FTP) with a particulate measurement procedure
added, are 0.6 grams per mile (0.37 g/km) for 1982 model year vehicles
and 0.2 grams per mile (0.12 g/km) for 1985 model year vehicles. These
emission levels are based on the need to reduce (or prevent an increase
in) the total suspended particulate (TSP) levels in urban areas as diesels
become more numerous. It is quite possible that even more stringent
particulate emission standards will have to be set in the future to
control the toxicity problem.
For these reasons, the EPA is interested in the state of
technology of diesel particulate emission control, and in particular, in
the feasibility of controlling the particulate emissions by afterburner
techniques. A program was already under way at Battelle's Columbus
Laboratories to assess the performance of Industrial afterburner emission
191
-------
control devices, hence, it was a logical extension of that program to
include diesel particulate emission control by afterburners.
The objective of the diesel afterburner study is to evaluate
emissions control on light-duty diesel vehicles by postcylinder oxidation.
Because of the special nature of the diesel emissions problem, we have
defined the scope of the program as a study of the ignition and oxidation
characteristics and requirements of diesel exhaust particulates.
192
-------
SECTION 2
PROGRAM PLAN
This study comprises a review of the state of technology,
detailed chemical and physical characterization of the particulate from
a light-duty diesel engine, bench experiments to define the ignition and
oxidation properties of diesel particulate, experiments with catalytic
ignition of particulates, and experimental evaluation of concepts and
devices for particulate emission control by oxidation in the exhaust of
an engine.
STATE-OF-TECHNOLOGY REVIEW
The state-of-technology review is directed towards determining
the current understanding of the nature of the diesel particulate primarily
with respect to its ignition and burning characteristics, and assessing
the current status of development of particulate emission control by
oxidation. It was hoped that specific devices or concepts would be found
in the course of this review that could be evaluated in the program.
However, industry sources were not encouraging in having devices or concepts
sufficiently developed for evaluation at this time.
A report on the technology of thermal and catalytic oxidation
of diesel particulate was published in October, 1979.^
CHEMICAL AND PHYSICAL CHARACTERIZATION
This task of the program is aimed at determining the chemical
and physical nature of the diesel particulate and the exhaust environment
as they relate to the requirements for thermal or catalytic oxidation of
the particulate. The Information developed in this task is being used
both in the evaluation of current prototype technology and in the development
193
-------
control devices, hence, it was a logical extension of that program to
include diesel particulate emission control by afterburners.
The objective of the diesel afterburner study is to evaluate
emissions control on light-duty diesel vehicles by postcylinder oxidation.
Because of the special nature of the diesel emissions problem, we have
defined the scope of the program as a study of the ignition and oxidation
characteristics and requirements of diesel exhaust particulates.
192
-------
SECTION 2
PROGRAM PLAN
This study comprises a review of the state of technology,
detailed chemical and physical characterization of the particulate from
a light-duty diesel engine, bench experiments to define the ignition and
oxidation properties of diesel particulate, experiments with catalytic
ignition of particulates, and experimental evaluation of concepts and
devices for particulate emission control by oxidation in the exhaust of
an engine.
STATE-0F-TECHNOLOGY REVIEW
The state-of-technology review is directed towards determining
Che current understanding of the nature of the diesel particulate primarily
with respect to its ignition and burning characteristics, and assessing
the current status of development of particulate emission control by
oxidation. It was hoped that specific devices or concepts would be found
in the course of this review that could be evaluated in the program.
However, industry sources were not encouraging in having devices or concepts
sufficiently developed for evaluation at this time.
A report on the technology of thermal and catalytic oxidation
of diesel particulate was published in October, 1979.^
CHEMICAL AND PHYSICAL CHARACTERIZATION
This task of the program is aimed at determining the chemical
and physical nature of the diesel particulate and the exhaust environment
as they relate to the requirements for thermal or catalytic oxidation of
the particulate. The information developed in this task is being used
both in the evaluation of current prototype technology and in the development
193
-------
of particulate oxidation principles and methodologies. Particulate
characteristics of special interest include surface area, ash chemistry,
size distribution, trace mineral content, total mass concentration, and
soluble organic content.
PARTICULATE IGNITION AND OXIDATION PROPERTIES
In this task, experiments are being conducted to study the
ignition and burning characteristics of diesel particulate. These
experimental results will establish a basis for developing particulate
oxidation concepts and for evaluating available prototypes. The specific
objectives include determining (1) how readily the particulate ignites,
(2) how rapidly it burns, (3) whether the behavior can be modified, and
(4) what the products of combustion are.
The bench experiments on ignition and oxidation properties are
under way but currently available results are too preliminary to report
at this time.
CATALYTIC IGNITION OF PARTICULATE
The primary activity in this task is an investigation of the
catalytic approach to particulate oxidation. The feasibility of using
inorganic salts, metal oxides, or other materials to promote the ignition
and oxidation of the particulate is being studied in bench experiments
involving a hot tube reactor. The experiments are being conducted to
identify potentially catalytic materials, to explore methods of catalyst
application, and to determine the magnitude of the catalytic effect and
the catalytic material amounts required to achieve that effect.
In this task and in the task on particulate ignition and oxidation
properties, we are also seeking to determine if a relationship exists
between the chemical and physical characteristics of the particulate and
its oxidation properties. Knowledge of such a relationship may contribute
to the development of a successful oxidation-type particulate control
system.
194
-------
EVALUATION OF CONCEPTS AND DEVICES
The objective of this task is to evaluate on the engine any
promising concepts or devices for particulate control by oxidation which
result from the state-of-technology review or from the bench experiments.
195
-------
SECTION 3
EXPERIMENTAL RESULTS
PARTICULATE CHARACTERIZATION
Description of Experimental Facility
Figure 1 is a schematic drawing of the experimental facility
being used in this program. The engine is a 4.3-liter General Motors
passenger-car diesel and is coupled to a 175-hp eddy-current dynamometer.
The exhaust system consists of standard vehicle exhaust pipe, muffler,
and tailpipe components. A dilution tunnel-CVS system combination is
used consistent with present practice for collecting and studying diesel
exhaust particulates. The 560-mm diameter by 5.5-meter long stainless
steel dilution tunnel is installed between the CVS system and its filter
assembly.
Instrumentation includes gas analyzers for exhaust emissions
measurement, thermocouples for engine, exhaust, and dilution system
temperature measurement, stack sampling systems for particulate collection
on filters, and an Electrical Aerosol Analyzer and a Laser Light-Scattering
system for particle size distribution and volume concentration measurements.
The particulate characteristics are being studied primarily by
means of a dilution system. As will be brought out later in this section,
there are limitations to the validity of this approach. However, the
dilution system does provide a controllable means for particulate collection
and measurement and has been established as the accepted method for deter-
mining particulate emission rates, therefore, this approach serves a useful
purpose. Particulates are also being collected directly from the exhaust
system for use in the bench experiments and for characterization.
196
-------
Engine Baseline Operation
Before particulate characterization experiments were begun
the engine was operated at various speed/load conditions to establish
its baseline operating characteristics. In one series of test runs the
engine was operated at wide open throttle over a speed range from 800 to
3200 rpm. The resulting engine performance data were compared with
representative performance data obtained from Oldsmobile Division of
(2)
General Motors. The test engine performance data were close to the
representative engine data except for air-fuel ratio. The test engine
operated approximately 1 to 2 air-fuel ratio numbers richer than the
"representative" data. Richer operation will, in most cases, yield
higher particulate emissions, which, for the purposes of this program
is not an undesirable characteristic.
For another series of baseline tests, a number of engine speed/
load conditions were selected to represent typical vehicle road-load
operation. Nine modes were selected representing 25 mph, 35 mph, and 55
mph at light, medium, and heavy loads. These modes were run and the data
are summarized in Table 1.
Particulate Characterization Methods
Particulate characteristics which have been measured in this
program are:
• Mass concentration
• Soluble organic content
• Carbon, hydrogen, and ash
• Trace mineral content
• Surface area
• Size distribution
• Volume concentration.
Mass concentration of the particulate is determined by weighing
the particulate collected on a 100-mm diameter Teflon-impregnated glass
fiber filter sampling from the dilution tunnel using conventional EPA
Method 5 stack sampling systems. These sampling systems consist of a
197
-------
vacuum pump, a dry-gas meter, a flow-indicating orifice, a condenser train,
a flow control valve, and appropriate thermocouples and manometers.
Sampling time, sampling rate, and sample temperature can be varied over
reasonable ranges as desired. Concentration of the particulate in the
exhaust system is calculated from the dilution tunnel data using the
dilution ratio which is determined from CO^ measurements made in the exhaust
gas, dilution air, and dilution tunnel.
The soluble organic content of the collected particulate is
determined by solvent extraction. In this extraction process, soluble
organic material is removed from the filter (or from loose particulate)
by Soxhlet extraction using toluene as the solvent. The extraction is
carried out for 32 to 48 hours at a 10-minute cycle rate. After extraction
the solvent is concentrated by rotary film evaporation and a known fraction
removed for gravimetric analysis.
The carbon and hydrogen measurements are made using a Perkin-
Elmer Elemental Analyzer. The ash was determined by weighing the sample
before and after oxidizing it in a muffle furnace.
The trace mineral content is determined by spark source mass
spectrograph.
Surface area is measured by the BET procedure.
Size distribution and number and volume concentration measurements
are made by Electrical Aerosol Analyzer and Laser Light-Scattering instru-
ments .
Characterization Results
A major thrust of the particulate characterization experiments
has been to determine if the engine operating conditions alter the
chemical or physical nature of the particulate in a manner that might be
taken advantage of in devising an oxidation technique. Thus, a number of
engine tests were run using the operating modes shown in Table I. It was
noted early in the experimental work that the total mass concentration and
the soluble organic content of the particulate were significantly affected
by engine load and speed. To define more completely these influences,
additional load conditions were run at each engine speed.
198
-------
Figure 2 shows the effect of engine speed and load on particulate
mass concentration. The particulate concentration is plotted against
exhaust temperature rather than horsepower because it is believed that
temperature bears a more direct relationship to particulate formation.
The trend of increasing particulate concentration with increasing load
(3)
has been observed by others. It is interesting to note that as load
(temperature) is Increased, specific fuel consumption decreases (see Table
I), meaning the fuel is burned more completely, but the particulate concen-
tration increases sharply.
Figure 3 shows the effect of engine speed and load on the
particulate soluble organic content. As with the particulate mass, the
soluble organic content is plotted against exhaust temperature rather
than engine horsepower. Engine speed does not seem to have as significant
an influence on the soluble organic content as it does on the mass concen-
tration. As one might expect if condensation plays a substantial role in
the association of the organic material with the particulate, the soluble
organic fraction decreases significantly as exhaust temperature is
increased by increasing load.
In the course of collecting data to determine the trends shown
in Figures 2 and 3, we noted that the soluble organic content measurements
were sensitive to certain parameters of the dilution and sampling proce-
dure. An investigation was carried out to determine the specific effects
of sampling rate, sampling time, sample temperature, and dilution ratio on
the soluble organic content measurement.
Figures 4, 5, and 6 show some of the results of this investigation.
Figure 4 shows the effect of sampling rate on soluble organic fraction;
Figure 5 shows the effect of sample temperature; and Figure 6 shows the
effect of sampling time. The sample temperature effect is similar to that
(4)
which has been noted by General Motors researchers. The dilution ratio
test results were not consistent enough to indicate a trend.
The conclusion to be drawn from these data is that artifacts are
being created in the sampling process. It would require considerably more
investigation than is warranted in this program to establish the nature
and magnitude of the artifact or artifacts. Fortunately, the sampling
199
-------
process effects that are seen in Figures 4, 5, and 6 do not appear to alter
significantly the qualitative relationships between soluble organic content
and engine speed and load. In this program it is more important to determine
if differences exist due to engine operating mode differences than it is to
determine precise values for key particulate characteristics. Furthermore,
we must also establish whether these differences which do appear to exist
result in differences in the ignition and combustion characteristics which
are being studied in the bench experiments.
Loose particulate samples are also being collected from the engine
exhaust system, both to provide larger amounts of particulate for the bench
experiments and to characterize the particulate as it exists in the exhaust
pipe before passing through the muffler and tailpipe and into the dilution
tunnel. The present method being used to collect this loose particulate is
to brush it out of the exhaust pipe with a nylon brush assembly made to fit
the pipe inside diameter. The procedure has been to clean the 7-foot exhaust
pipe section before a test run, and then to allow the particulate to collect
on the surfaces during a period of 6 to 7 hours or for the duration of a
day's testing. The deposited particulate is then brushed into a glass
container which is sealed from air and light.
This procedure was followed for a number of test runs including
both single speed/load runs and runs where either the speed or the load or
both were varied during the test period. Table II summarizes characteri-
zation data obtained on these loose particulate samples. Several of the
samples are being used in the bench experiments. One noteworthy charac-
teristic of these samples is that in most cases the soluble organic content
is substantially lower than in filter samples collected from the dilution
tunnel under the same or similar engine operating conditions as shown in
Column 8 of Table II. We theorize that this is due to the organic material
continuing to associate with the particulate throughout the exhaust system
and possibly even into the dilution system. However, it is also possible
that the particulate deposited on the exhaust pipe walls will lose some
organic material as It continues to be exposed to the hot exhaust gases.
Table III presents the results of the trace element analysis
performed on two loose particulate samples by spark source spectrograph.
200
-------
The major elements appear to be carbon, iron, sulfur, phosphorus, calcium,
silicon, and zinc. The two samples represent widely different engine
operating conditions and substantial differences in some of the consti-
tuent elements are noted. Further trace element analyses may be performed
if the bench experiment results indicate a need for such data.
Only one surface area measurement has been made at this time.
2
The resulting value, by the BET procedure, was approximately 100 m /g,
which appears to agree with published data.^
The size distribution measurements by Electrical Aerosol Analyzer
showed a shift in the mass median diameter as engine load was changed.
Figure 7 shows this shift. As engine load (represented by exhaust tempera-
ture) is increased at constant speed, the mass median diameter Increased
also. This phenomenon indicates agglomeration is occurring with increasing
load which could help to explain the mass concentration increase with load
increase. Particles too small to be captured on a filter at lower loads
agglomerate to larger particles as load increases thereby increasing the
filter-collected mass. Others have also observed this effect.^ Figure
7 also shows that higher speed operation results in smaller mass median
diameter, a phenomenon probably related to residence time, i.e., less time
available for agglomeration at higher engine speeds.
The Laser Light-Scattering Instrument is intended to be used in
this program to provide an instantaneous measure of particulate mass
concentration in the dilution tunnel. When concepts or devices for parti-
culate control by oxidation are being evaluated on the engine, there may
be transient effects on the particulate which could not be detected by
the filter-collection method.
In the preliminary evaluation of the Laser-Light Scattering
instrument, data were obtained by both filter collection and light scattering
with the engine operating at constant speed and different loads. Figure 8
shows the results of this experiment. The light-scattering data are seen
to follow the filter data fairly closely.
201
-------
CATALYTIC IGNITION
Background and Experimental Approach
A basic problem with the particulate when entrained in the diesel
exhaust is that it is too dilute to bum. This can be remedied by
collecting the material on a filter or trap. In this concentrated form
the average heat loss is reduced and conditions for combustion are more
favorable.
There are a number of design concepts for trapping and oxidizing
the particulate emissions from diesel engine operation. Although the
requirements for successful application of each of these concepts can be
described qualitatively, neither the character of the particulate, the
service requirements, nor the limitations imposed by automobile application
are well enough described to permit firm choice among these concepts or the
complete rejection of any. In this study it is assumed that the particulate
has been trapped in a fashion suitable to the catalysis demonstrated here.
In the present work, the opportunities for catalytic promotion of
ignition are being surveyed by using catalytic salts that are deposited
directly on previously trapped particulate. In principle this would be
done directly in the exhaust system of the automobile, in the bench experi-
ments we have removed the collected particulate from the exhaust system,
catalyzed it, and examined its ignition properties in a separate system.
Lowered ignition temperatures achieved in this way facilitate the onset of
combustion which, because of its highly exothermic nature, produces a large
local increase in temperature at which the sample is rapidly burned.
With suitable sample geometry, only the ignition need be catalytic
and this is the effect being monitored here. The situation for achieving
ignition in some volume element of a reactor can be described with the help
of the theory employed by Wagner^ and Frank-Kamenetskii^ . Usually the
ignition temperature lies well above the temperature at which the first
significant heat effect due to oxidation is detected* and at ignition the
excess heat generated over that lost by the volume element causes that
element to suddenly rise in temperature well above the environment tempera-
ture. In so doing it passes from activity limited oxidation to transport
202
-------
limited conditions and usually exceeds the kinetics that can be expected for
the catalytic surface reaction processes.
The only form of catalysis that has been described in any detail
for oxidation of carbon particulate is that developed years ago for removal
of soot accumulation from flues and furnaces by the use of inorganic salts
(9)
deposited on the carbon particle. The U.S. Bureau of Mines Bulletin 360
provides an understanding of the level of development of this form of
carbon particulate oxidation catalysis. More recentlysalt additives
were studied as catalysts for the oxidation of high purity graphites.
It is of interest to note that English patents for Inorganic
soot removers began to appear in 1856 covering the use of ordinary alkalis
or salts such as sodium chloride* quicklime plus soda ash, magnesia, copper
salts, and others. American patents started In 1892 covering similar
compounds and especially chlorides of a number of elements such as zinc.
From the earliest patents the use of chlorides has been favored for such
(9)
purposes. Bulletin 360 of the U.S. Bureau of Mines lists 59 compositions
tried by them and exhibiting decreases in ignition temperatures of as much
as 287 C. By this means the ignition temperature was lowered from 613 C
to a minimum of 326 C in normal air.
In all of these tests the volatility of the salt used was an
important characteristic. Initially, the salt was applied by vaporizing
it and allowing it to deposit onto soot that had been accumulated previously
on a test screen. The action by which the salt catalyzed burning was not
specified, but some relatively contemporary publications suggest mechanisms
that do not seem to have been further investigated.
For instance, Cassel^^ noted that soot deposited on the etched
or ground surfaces of Jena glass ignited much more easily and burned more
rapidly than soot on adjacent smooth surfaces of the glass. From evidence
for oriented crystal growth of soot particles on surfaces he suggested
that the ground surfaces Interrupted crystal growth during soot deposition
because of surface Irregularity and also helped to prevent secondary crystal
growth during oxidation (burning). He reasoned that the salt deposits
condensed on soot similarly promote ignition and burning by maintaining
the highly dispersed soot structure prior to and after ignition.
203
-------
(12) (13)
Alternatively, Taylor and Neville and later Day, et al.,
noted large increases in soot burning rate following deposition of soot
onto surfaces previously coated with salts by evaporation. Both of these
investigations concluded that the salts probably functioned by hastening
the decomposition of carbon-oxygen surface complexes. With this barrier
removed the carbon surface was believed to be more rapidly attacked by
oxidizing gases. A brief review of the succeeding years' publications
has failed to disclose any further discussion of these or other hypotheses
for the action of the salts. It is of interest to note that the promotion
effect is claimed either for salts vaporized onto the soot or for soot
condensed (deposited) onto the previously deposited salt.
In the more recent work of Duval and coworkersinorganic
cations were impregnated into graphite carbon from water solution and then
the rates of oxidation were noted. The results showed that the intro-
duction of cations at 120 ppm concentration produced rates of oxidation
up to 470,000 times that of the untreated graphite. In this study the
effective cations were identified as representing those elements that have
variable oxidation states, and can exist in defect states of oxidation.
By this means they serve as oxygen carriers to the graphite interface.
The differences in relative effectiveness of the various metal
salts in these two studies^'^^ seems to underline differences in mechanism
of action that have occurred, either because of the difference in method of
application or because of the differences in reactivity and structural
stability of soot and graphite. The variable valence postulate for the
catalytic action on graphite is the same as that widely used to explain
the catalytic action of high-surface-area oxide catalysts In catalytic
hydrocarbon oxidizers. The large increases in oxidation rate illustrated
by these catalytic applications indicate that considerable catalytic
assistance is available for the burning of dlesel-exhaust particulate.
For the present application, the catalytic effect should start
at temperatures less than that at which the unassisted thermal oxidation
has been detectable and the heat release rate due to this catalytic
oxidation must be sufficient to achieve the ignition temperature. In the
(9)
U.S. Bureau of Mines study trials of uncatalyzed ignition of furnace
204
-------
soot established the ignition temperature at about 615 C, and it was
observed that insignificant oxidation rates were obtained below 480 C.
The reduction of inlet oxygen concentration from normal air to 8 percent
had little effect on the ignition temperature for untreated samples.
Both the design of the catalytic chamber and the activity of the
catalyst influence the ability of any device or concept to achieve ignition.
This study will attempt to define the most favorable form of catalysis and
a chamber design which will maximize the repeatability of the catalytic
ignition at acceptable performance levels.
Experimental Method
The reactor used in these experiments is shown in Figure 9. A
sample of diesel soot, 0.15 g, is placed in a stainless steel pan that is
about 10-mm wide, 47-mm long, and about 3-mm deep. One thermocouple
junction, chromel-alumel 26 gauge, is buried in the sample, the other is
in the gas stream just upstream of the sample. The entire quartz reactor
up to the standard taper joint is placed in a tube furnace that is heated
gradually from room temperature to 600 C in about 1.5 hr. During most of
the heating the sample lags behind the indicated gas temperature by about
15 to 25 degrees centigrade. On ignition the sample temperature rises
abruptly at a rate of 100 to 200 degrees per minute and the occurrence of
ignition on some portion of the sample can be confirmed by observation of
the orange-yellow glow produced as the combustion site moves through the
sample. Since the gas temperature typically is rising about 5 to 10
degrees per minute, the ignition temperature can be identified with an
uncertainty of only a few degrees. Typically, the indicated temperature
rise, Tg-Tg, reaches a maximum of about 120 to 200 C, and this lags behind
the true combustion-produced temperature rise by an amount depending on
the actual location of the hot zones in the sample with respect to the
thermocouple. The arrangement is sensitive to a temperature change of
less than one degree so that the occurrence of ignition somewhere in the
sample is detected readily. Sample traces of the temperature plots
obtained during experiments are shown in Figure 10.
205
-------
The air flow was held at 1 liter per minute for the first 12
runs; from Run 13 onward this flow was increased to 2 liters per minute
to accommodate a few trial measurements of CC>2 production.
Catalytic salts were added to the diesel soot as methanol
solutions and the solvent was removed before trial by heating the sample
16 to 24 hours at 115 to 130 C. In each case the salt solution was made
up 1 x 10 molar so that a few milliliters of these solutions sufficed
to impregnate 0.20 g of soot; after drying, 0.15 g of the sample was used
for the experiment.
The soot samples used in this study were brushed from the walls
of the exhaust pipe after various experiments. They were identified by the
date of collection and by the engine operation data for that date.
These loose samples were extremely fluffy and it was found
difficult to get good thermocouple contact with them. It was found that
samples dispersed in methanol had a more suitable density after removal
of that solvent at 110 to 130 C and so the uncatalyzed samples were examined
in this form. Since all catalyst applications were made from methanol
solution in the same way, we believe the uncatalyzed performance represents
a correct blank for comparison with the catalyzed samples.
In this phase of the bench experiments the primary Intent was
to survey the parameters that control catalytic ignition; most of the
data listed in this report were obtained using a single collection iden-
tified by the date 12/12/79. Table IV provides a comparison of that sample
with a few other samples that were tried in order to compare the responses
of different particulate to catalytic ignition.
Results
The activities of a few metal chlorides for promoting ignition
of the diesel particulate are compared in Table V. The superiority of
copper chloride 1b surprising in view of the relatively high oxidation
activity that has been demonstrated by other investigators for catalysts
based on Co, Mn, and Ft at lower temperatures than used here. In each
case the question arises concerning the temperature required before a
206
-------
catalytic form of the metal can be obtained under the conditions used
here where no preactivation treatment is employed. Persistence of some
part of the water of crystallization in a few cases may be retarding the
decomposition of the catalytic salt during the heating period prior to
ignition.
A further investigation of the parameters controlling catalytic
ignition was made by varying the anion of the salt used. Table VI shows
that good performance is not limited to the chloride as indicated by the
(9)
earlier U.S. Bureau of Mines study , presumably because in the present
case volatilization of the salt is not relied upon to deposit the salt on
the accumulated soot. The superiority of copper salts over the cobalt
equivalents is further demonstrated in this table. Figure 11 shows that
the closely competitive performance of chloride and nitrate salts of copper
continues over a range of catalyst concentration.
(9)
The earlier U.S. Bureau of Mines study had shown some catalytic
effect for salts such as NaCl and so a series of experiments was tried with
different amounts of sodium and ammonium salts added to the copper chloride
solution in order to determine how such additions might affect the activity.
The various trials of Table VII represent experiments in which the anion
(e.g., chloride) concentration is varied over a wide range while holding
the copper concentration constant, and also permits comparison of the use
•f +
of Na versus NH. cations. The results favor the use of Na rather than
+ 4
NH^ salts and demonstrate that substantial increase in catalytic effect
results from addition of NaCl or NaNO^ to the catalyst solution.
Conclusions
1. Diesel particulate is readily brought to ignition
at 380 to 400 C by application of minor concentrations
of metal salt solutions to the previously collected
particulate. The catalytic action of the metal salt
is enhanced by admixing salts such as sodium or
ammonium chloride and nitrate. This represents a
lowering of ignition temperature of about 150 C
compared to the untreated particulate. Detectable
promotion of oxidation rates is found at even lower
temperatures.
207
-------
Among the metals studied to date, copper salts work
best as catalysts for particulate oxidation. The
other metals, which include some known to have high
catalytic activity for oxidation, exhibit low
activity compared to those expected for well-activated
preparations.
The catalytic effect of copper salts in the present
application does not require the use of chlorides as
noted in previous studies of furnace soot oxidation.
The previous preference for chloride apparently arises
because of the salt volatilization required for the
furnace application.
The catalytic effectiveness of a metal salt in the
present application may be limited by the requirements
for its activation as a catalyst. Since preactivation
before use is not practiced here, the facility with
which the metal can achieve active form during heat-up
(prior to about 350 C) is very important. The superiority
of copper salts is suggested to arise for this reason.
The mechanism of catalytic action exhibited in this
study is not known and an investigation of that mechanism
is required for further clarification of the requirements
for catalytic ignition of diesel particulate oxidation.
208
-------
SECTION 4
FUTURE PLANS
Current and future activities on this program will include addi-
tional characterization of particulate in the exhaust system; continuation
of the bench experiments on particulate oxidation characteristics and
catalysis; ignition and catalytic experiments in the engine exhaust; and
evaluation of devices and concepts on the engine.
For the additional particulate characterization we plan to con-
centrate on particulate collected by various means directly from the
exhaust pipe. Mass concentration and soluble organic content will be
measured under different engine operating conditions, and analytical
techniques such as infrared spectroscopy, scanning electron microscopy,
and thermogravimetric analysis, will be explored for their suitability to
yield useful results. We also plan to conduct more surface area and carbon,
hydrogen, and ash measurements.
The catalysis bench experiments will include an investigation of
the mechanism of catalytic action as well as trials of a greater variety
of particulate samples and simulation of exhaust conditions.
The oxidation bench experiments will be aimed at establishing
a measurement methodology and using it to characterize a variety of
particulate samples collected from the engine under various engine opera-
ting conditions and sampling methodologies.
In a final phase of the program, catalytic materials, techniques,
and concepts studied in the bench experiments will be evaluated in the engine
exhaust system. In addition, devices or concepts which emerge from the
state-of-technology review will be evaluated to the extent that sufficient
detailed information or actual prototype devices can be obtained. The
objective of these on-engine evaluations will be to establish potential
feasibility of concepts not to develop prototype systems.
209
-------
REFERENCES
(1) Murphy, M. J., L. J. Hillenbrand, and D. A. Trayser. Assessment
of Diesel Particulate Control: Direct and Catalytic Oxidation.
EPA 600/7-79-232b, U.S. Environmental Protection Agency, Research
Triangle Park, North Carolina, 1979.
(2) Private communication from S. Gupta, Oldsmobile Division of General
Motors.
(3) Khatri, N. J., J. H. Johnson, and D. G. Leddy. The Characterization
of the Hydrocarbon and Sulfate Fractions of Diesel Particulate
Matter. SAE Publication PT-79/17, 1979.
(4) McDonald, J. S., S. L. Plee, J. B. D'Arcy, and R. M. Schreck.
Experimental Measurements of the Independent Effects of Dilution
Ratio and Filter Temperature on Diesel Exhaust Particulate Samples.
SAE Publication P-86, February, 1980.
(5) Lee, K. B., N. W. Thring, and J. M. Beer. Combustion and Flame,
(6): 137, 1962.
(6) Verrant, J. A. and D. B. Kittelson. Sampling and Physical Charac-
terization of Diesel Exhaust Aerosols. SAE Publication PT-79/17,
1979.
(7) Wagner, Carl. Chem. Tech. (Leipzig), ]L8, 28-34 (1945).
(8) Frank-Kamenetskii, D. A. Diffusion and Heat Transfer in Chemical
Kinetics. Plenum Press, New York (1969).
(9) Nicholls, P. and C. W. Staples. U.S. Dept. of Mines Bulletin 360
(1932).
(10) Heuchamps, C. and X. Duval. Carbon, 4., 243-253 (1966) and Amoriglio,
H. and X. Duval. Carbon, j4, 323-332 (1966).
(11) Cassel, H. M. J. Am. Chem. Soc., .43, 2055 (1921).
(12) Taylor, H. S. and H. A. Neville. J. Am. Chem. Soc., 43, 2055 (1921).
(13) Day, J. E., R. F. Robey, and H. J. Dauben. J. Am. Chem. Soc,, 57,
2725-6 (1935).
210
-------
Muffler
Cold Trap &
Filter
dilution Air Filters
Pump
Sample Filter
and Pump
Tailpipe
CO Y
Filters
Exhaust Pipe
Dynamometer
Pump
Dilution Tunnel^
Filter-ilH
Electrical
Aerosol
Analyzer
CVS System
Laser Light-
Scattering System
Engine Heat
Exchanger
Dynamometer
Controls
Particulate
Sampling
Systems
©
Engine Fuel
System
FIGURE 1. ENGINE EXPERIMENTAL FACILITY
211
-------
830 RPM
500
1170 / u
RPM /1 830 RPM
m 400
o 300
£ 200
¦5 100
UJ
500
200 300
Exhaust Temperature, C
400
100
FIGURE 2. INFLUENCE OF EXHAUST TEMPERATURE
AND ENGINE SPEED ON PARTICULATE
EMISSIONS CONCENTRATION
a?
~-
c
0)
c
o
o
o
c
<19
O)
k.
O
0)
2
3
O
to
30
20
10
1 A
830 rpm-pN
I I
-1830 rpm
A
—
° V
1170 rpm-/
I
Y^° , Zl*
100 200 300 400
Exhaust Temperature, C
500
FIGURE 3. INFLUENCE OF EXHAUST TEMPERATURE
AND ENGINE SPEED ON SOLUBLE
ORGANIC CONTENT
212
-------
40
Engine Speed - 1170 rpm
10 hp
c 30
c 20
to
O)
20 hp
Sample Rate, cfm
FIGURE 4. EFFECT OF SAMPLING RATE ON SOLUBLE
ORGANIC CONTENT MEASUREMENT
30
c
0)
o 20
(J
o
c
to
O)
o 10
0)
JO
_3
o
cn 0
1 CL 1
1 1
1
1170 rpm
10 hp
o
—
O
—
1170 rpm
20 hp
0
1830 rpm
40 hp
i i
i i
l
0
20
40
60
80
Sample Temperature, C
100
120
FIGURE 5. EFFECT OF SAMPLE TEMPERATURE ON
SOLUBLE ORGANIC CONTENT MEASUREMENT
213
-------
50
40-
c
®
•>->
c
o
o
a
'E
u.
o
©
.o
j3
o
w
30-
20-
10-
0
J 830 rpm
20 hp
1170 rpm
10 hp
830 rpm O
10 hp D
~
10 20 30
Sampling Time, Minutes
FIGURE 6. EFFECT OF SAMPLING TIME ON SOLUBLE
ORGANIC CONTENT MEASUREMENT
214
-------
£ 0.30
51
£ 0.25
0)
E
.2
% 0.20
o
C
(0
S 0.15
03
2
00
t/J
CO
0-1 °0 100 200 300 400 500
Exhaust Temperature, C
FIGURE 7. INFLUENCE OF EXHAUST TEMPERATURE
AND ENGINE SPEED ON PARTICULATE MASS
MEDIAN DIAMETER
w
E 100
03
E
I 75
(0
L.
+*
c
m
o
§ 50
0
v
JO
1 25
t
CO
Q.
I 0 100 200 300 400 500
K Exhaust Temperature, C
FIGURE 8. COMPARISON OF FILTER AND LASER
PARTICULATE MASS CONCENTRATION
DATA
830 rpm
1830 rpm
• Filter Data
O Laser Data
1830 RPM
215
-------
Air In
Thermo-
couple |£
Entries
Sample at S, In View of Window C
Measure Temperatures
At G =Tg
Ts-Tg =AT
FIGURE 9. ARRANGEMENT OF QUARTZ REACTOR FOR
MEASUREMENT OF IGNITION TEMPERATURES
-------
o
600
500-
Run 16
Uncatalyzed
03
l.
3
(0
u.
QJ
a
£
a>
h-
co
(0
e?
^ 400-
300-
200-
100-
Igmtion—
Ts =545C
10 20 30
40 50 60
Minutes
100
Run 32
Catalyzed
With Mix
1CuCI2'2H20+8NaCI
« 300
Ignition
Ts =39OC
I I
10 20 30 40 50 60 70
Minutes
80 90 100
FIGURE 10. SAMPLE TEMPERATURE TRACES DURING
IGNITION TRIALS
217
-------
460
o
0)
k_
3
¦*->
(0
i—
0)
a
E
0)
l-
c
o
5
"c
O)
440
420-
400
380
T
T
T
O Cupric Chloride
+ Cupric Nitrate
Uncatalyzed Ignition
Temperature = 504-545 C
Observed Ignition Temperature
Range With NaCI Addition,
0 50 100 150 200 250 300
Applied Catalyst Concentration, Micromoles Cu/g Particulate
FIGURE 11. EFFECT OF SALT CONCENTRATION
ON IGNITION TEMPERATURES
21ft
-------
TABLE I. ROAD LOAD ENGINE OPERATION
Teat Number
1
2
3
4
5
6
7
8
9
Road Speed, mph 25
25
25
35
35
35
55
55
55
Engine Speed, rpm 830
830
830
1170
1170
1170
1830
1830
1830
Engine Bhp 2.08
4.15 10.4
4.97
9.94
19.9
10.1
20.1
40.3
Bafc, lb/bhp-hr 1.44
0.89 0.56
0.88
0.58
0.47
0.63
0.55
0.43
Exh Man Temp, C 112
129
179
108
164
249
164
218
345
Tailpipe Temp, C 86
95
127
104
126
183
136
174
267
HC,
ppm C 198
155
95
75
97
85
73
75
63
N0X,
ppm 117
138
165
189
238
152
210
255
335
CO,
ppm 267
298
315
305
237
406
154
155
365
TABLE II.
CHARACTERISTICS OF LOOSE PARTICULATE
SAMPLES
Sample
Date
Speed/Load
Condition,
rpm/hp
Soluble
Organic
Content,
wt X
H
wt 2
C
wt X
Ash
wt X
Exh<«>
Temp,
C
Filter
Soluble
Organic
Content,
wt Z
10/25/79
1170/10
_ _
(162)
11/15/79
1170/5, 10, 20
6.7
—
—
—
213
15
11/16/79
830, 1170, 1830/10
9.0
—
—
—
170
18
12/12/79
1170/20
10.5
—
—
—
240
16
12/19/79
1170/10
7.7
--
—
--
162
16
1/23/80
1170/5
7.2
2.0
77.1
2.0
120
20
1/24/80
1830/40
6.5
0.4
77.8
1.0
350
3
2/6/80
830/varlable
2.7
0.4
70.0
0.8
182
12
2/7/80
1830/variable
5.1
1.0
89.6
2.0
240
14
2/8/80
1170, 1830/varlable
1.9
0.5
82.0
0.9
220
12
2/14/80
1170/10, 20
1.8
0.5
71.0
0.6
200
12
2/15/80
1170, 1830/20
4.8
0.7
71.9
1.5
225
7
4/24/80
830/vaciable
—
—
—
(180)
4/25/80
1830/varlable
—
—
(240)
(a) At same or (laliar conditions.
219
-------
TABLE III. MASS SPECTROGRAPHS ANALYSIS OF PARTICULATE SAMPLES
Sample Daaltnatioi/*^ Sample Deilimetlon^
Element
11/8/79W
11/9/798^)
Element
n/e/79t«>)
ll/9/79BCc
U
10
10
<2
<2
Ba
<0.03
<0.05
Cd
<3
<3
B
10
3
la
<1
<1
C
Major
Major
So
<3
<3
r
3
<3
8b
<2
<2
Ma
500
500
Ta
<2
<2
MS
ISO
ISO
I
0.5
<0.5
A1
1500
1500
C*
<1
<1
81
3000
1500
Ba
30
30
P
-x*
-2X
La
<1
<1
S
-4J
-1*
Ca
<0.5
<0.5
CI
SO
50
Pt
*1
<1
K
100
50
Nd
<3
<3
Ca
3000
3000
<3
<3
8c
<3
<3
Eu
<2
<2
Ti
300
3
Gd
<3
<3
V
0.3
0.3
Tb
<1
<1
Cr
500
300
Dy
<3
<3
Mn
300
300
Bo
<2
<2
Fa
-30*
"102
Er
<3
<3
Co
5
3
Tn
<3
<3
HI
150
150
Tb
<3
<3
Cu
1500
1000
Lu
<1
<1
Zn
2000
1500
Hf
<3
<3
6a
<1
<1
Ta
<3
<3
Ga
<2
<2
V
3
<3
Aa
<0.5
<0.5
it
<2
<2
8a
<2
<2
O*
<3
<3
Br
<1
<1
Ir
<3
<3
Bfc
<5
<5
ft
<3
<3
Sr
3
0.5
Au
<1
<1
T
<0.5
<0.5
B<
<5
<5
Zr
<2
<2
11
<2
<2
Kb
<0.5
<0.5
Pb
30
30
Mo
<3
<3
Bi
<1
<1
*u
<2
<2
Th
<1
<1
Mb
<0.5
<0.5
U
<1
<1
M
<3
<3
(t) Valuae given Id ppm unlaea otherwise noted,
(b) 830 tpm/2 hp.
(e) 1B30 tfrn/id hp.
220
-------
TABLE IV. COMPARISON OF PROPERTIES OF DIESEL
PARTICULATES AND OTHER CARBONS
Sample
Surface
Area,
Soluble
Organic
Content,
wt %
Hydrogen,
wt %
Ignition
Temperature, C^®)
Identification
m2/g
as rec'd
as rec'd
As Rec'd
Catalyzed
Diesel particulate
10/25/79
100
—
—
468
370
Diesel particulate
21/12/79
—
10.2
—
504-545
413-421
Diesel particulate
1/23/80
—
7.2
2.0
485
400-423
Diesel particulate
1/24/80
—
6.5
0.4
466
395
Ultra-F, high
purity graphite
10
Nil
Nil
660
635
Cabot CSX-150
-1100
495
428
(a) In each case the sample has been slurried in methanol, then dried. For
the catalyzed trials, cupric chloride was dissolved in the methanol and
added to the extent of 1.18 x 10"^ mole/g soot.
— Not measured.
221
-------
TABLE V. COMPARISON OF IGNITION TEMPERATURES FOR
VARIOUS METAL CHLORIDES
Run
Catalytic Salt
Used(fl)
Ignition
Temperature,
C
Maximum
Temperature
Rise, (<*)
AT C
16 and 38
None(b*
504-545
50-121
44
MnCl2'4H20
456
125
45
CoCl2»6H20
461
122
36
H2PtCl6(c)
455
237
13
CuC12*2H20
421
139
-4
(a) The salt used in each case was equivalent to 1.18 x 10
moles of the metal per gram of particulate. Since each was
added in methanol and then dried at 110 to 130 C, the water
of crystallization probably was lost on drying.
(b) Diesel particulate collected 12/12/79, slurried in methanol,
and dried 110 to 130 C.
(c) In methanol-ethanol, not previously activated to decompose
the salt other than the 110 to 130 C drying conditions.
(d) This is not the combustion temperature, but the overall
maximum rise sensed by thermocouple reflecting the relative
rate and completeness of oxidation.
222
-------
TABLE VI. COMPARISON OF THE EFFECT OF THE ANION ON
THE ACTIVITY OF THE CATALYST
Run
Catalytic Salt
Used
-------
TABLE VII. EFFECT OF THE ADDITION OF SODIUM AND AMMONIUM
SALTS TO THE COPPER CHLORIDE CATALYST SOLUTION
Maximum
Ignition Temperature
Catalytic Salt Temperature, Rise,(c)
Run Used(a) C AT C
16 and 38
None(b)
504-545
50-121
13
CuCl2*2H20
421
140
25
2NH4C1*CuC12*2H20
428
165
26
2NH4N03'CuC12'2H20
435
178
28
2NaN03*CuCl2*2H20
397
175
27
2NaCl»CuCl2*2H20
381
184
29
INaCI* CuCl2"2H20
407
169
31
4NaCl'CuCl2*2H20
391
143
32
8NaCl2CuCl2'2H20
390
162
(a) The formula represents the nominal molar ratios of ingredients
dissolved in the methanol solvent before addition to the
diesel particulate. In each case, the copper concentration
was 1.18 x 10"^ mole/g particulate.
(b) Slurried with methanol, then dried.
(c) This is not the combustion temperature, but the overall maximum
rise sensed by thermocouple reflecting the relative rates and
completeness of oxidation.
224
-------
KINETICS AND MIXING IN INDUSTRIAL AFTERBURNERS
By:
A. Levy, A. A. Putnam, H. A. Arbib, and R. H. Barnes
BATTELLE
Columbus Laboratories
Columbus, Ohio 43201
225
-------
ABSTRACT
Industrial afterburners can be effective control devices for
limiting the emissions of organic species to the atmosphere. For
practical and effective operation one must balance the demands of
fuel and engineering economics with afterburner size and complexity,
operating and maintenance costs, and meeting other specific requirements,
such as turndown capabilities. In this paper combustion phenomena in
afterburners are considered from the points of view of the turbulent mixing
necessary for a compact system and that of the kinetics involved in the
reaction process. The mixing aspects are considered first on the basis
of the implications of the constructional details of industrially available
afterburners, and second from the potential for applying mathematical
modeling techniques in the design of more effective afterburners. Examination
of the constructional details in terms of the components of generic burner
type, approach section, and fume incineration section shows that the mixing
phenomenon can be considered from a relatively simple point of view that
can be quite amenable to mathematical modeling. The chemical aspects of
afterburner systems are analyzed with respect to hydrocarbon oxidation
processes. Special attention is devoted to the quasi-global and global
kinetics of these oxidation processes. Appropriate equations for calculating
chemical performance based on theoretical and laboratory data are examined.
Some generalized kinetic predictive procedures are also discussed.
226
-------
ACKNOWLEDGMENT
This study was carried out under Contract No. 68-02-2629, Mr. John H.
Wasser, Project Officer.
227
-------
SECTION 1
INTRODUCTION
The afterburner, sometimes referred to as fume incinerator, thermal
oxidizer, catalytic oxidizer, was one of the first control devices used by
industry. Many industrial processes emit organic fumes of one type of
another, and incineration has always been the "obvious" way to eliminate these
fumes. How well or how efficiently these afterburners operate was never con-
sidered very seriously until the role of organics in the photochemical smog
cycle became apparent. At that point it became very obvious that serious
steps had to be taken to monitor and control the emissions of these hydro-
carbons (organics) (n.b. these two words are often used interchangeably in the
context of this paper; they are not synonymous of course, except as their
principal oxidation products are carbon dioxide and water vapor).
At the last Stationary Sources Combustion Symposium Barrett and Barnes
discussed the nature of afterburner operations. (1) In essence their pre-
liminary environmental assessment showed that in-service afterburners were
less efficient than desired, i.e., median efficiencies of 50-75 percent. On
the other hand however the lower afterburner efficiency probably had minor
impact on national organic nonmethane emissions, but could have a marked im-
pact in local areas.
As a consequence of that work, and as part of this program on afterburner
technology it was deemed appropriate to examine the operation of the after-
burner in terms of its two key areas—mixing and kinetics. In principle the
afterburner is s simple combustion device. Organic vapors, maybe only at con-
centrations of a few thousand parts per million, coming out of industrial
processes, i.e., varnish cookers, paint-bake ovens, degreasing operations,
wire-coating, etc., are mixed with air, led into a reaction zone where the
temperature is sufficient to oxidize the organic, and the products, CO2 and
water vapor, are emitted to the atmosphere.
228
-------
This obviously is an over simplistic picture of the afterburner process.
Mixing dilute steams of organic vapor with air requires special mixing de-
vices. Bringing the mixture to temperature and carrying out the oxidation
can be accomplished several ways, i.e., by passing the mixture (1) through a
flame, (2) through a thermal oxidation zone, or (3) through a catalytic re-
actor. Hence—our interests in this paper in considering the mixing and
kinetic processes which influence afterburner design, operation and
efficiency.
Conceptually, afterburners can be considered as being divided into 3 sec-
tions: an auxiliary fuel combustion section, a fume and combustion product
mixing section, and an oxidation (or reaction) section. This afterburner con-
cept is illustrated in Figure 1. Physically, the afterburners sections
shown in Figure 1 may be merged and all the processes may occur in one
chamber.
In the combustion section, an auxiliary fuel is fired to supply the heat
to warm the fume to a temperature that will promote oxidation of the organic
vapors. Usually, a portion of the fume stream supplies the oxygen. (Part of
the fume stream must be bypassed or the fuel/air mixture will be too lean to
sustain combustion.) Both gaseous and liquid fuels are used to fire after-
burners. Gaseous fuels have the advantage of permitting firing in multiple-
jet (or distributed), burners. Oil combustion has the disadvantage of pro-
ducing sulfur oxides (from sulfur in the oil) and normally produces higher
nitrogen oxides emissions.
The mixing section is designed to provide intimate mixing between the
combustion products (from combustion of the auxiliary fuel) and the remaining
fume gases. To insure good mixing it is necessary to provide high velocity
gas flow to produce turbulence. Gas velocities in afterburners range from
25 to 50 feet per second. Ideally, the temperature profile at the outlet of
the mixing section would be flat. In thermal-type afterburners, the following
temperatures are often used as guidelines:
Odor control: 900-1350°F
To oxidized hydrocarbons: 900-1200°F
To oxidized carbon monoxide: 1200-1450°F.
The oxidation section provides time for the organic vapors in the by-
passed fume to be oxidized. Oxidation sections typically have length-to-
229
-------
diameter ratios of 2 to 3. Depending on the type of pollutant, residence
times ranging from 0.2 to 1.0 seconds are required for thermal units. The
residence time in most practical afterburner systems is dictated primarily
by chemical kinetic considerations.
Catalytic afterburners provide a catalytic surface to promote oxidation
for organic vapors. Consequently, catalytic afterburners operate at lower
temperatures than the thermal types and require less fuel. The preheat tem-
perature for catalytic devices varies with gas composition and type of con-
taminant to be oxidized, but is generally in the range from 650 to 1100°F,
lower than the 1200 to 1500°F of most thermal afterburners.
A comparison of temperatures required in thermal and catalytic after-
burners to convert various compounds to CO2 and H2O vapor is given in Table 1.
Catalysts in afterburners typically consist of either a metal mesh,
ceramic honeycomb, or a ceramic matrix with a surface deposit of finely
divided platinum or platinum family metals. In industrial .processes, 10 to
3
100 ft of catalyst bed are used per 1000 scfm of gas flow. In afterburners,
3
however, the requirement is in the range of 1 to 2 ft per 1000 scfm.
Catalytic afterburners have the disadvantage that performance efficiency
deteriorates as the unit is used, and require periodic replacement of the
catalytic material.
230
-------
SECTION 2
MIXING PHENOMENON IN INDUSTRIAL AFTERBURNERS
The aim in an industrial afterburner is to consume a pollutant down to a
safe lower level while balancing the economics of auxiliary fuel requirements,
afterburner size and complexity, operating and maintenance costs, and meeting
any other specific requirement, such as turndown capabilities. While the
kinetics of the specific reactions that are desired establish minimum time-
temperature-concentration requirements, the mixing patterns in the after-
burners are critical in providing these conditions. Furthermore, the com-
plexities of providing mixing patterns of desired types will control the size
of the afterburner, the amount of auxiliary fuel required, and the pressure
drop through the afterburner.
The mixing problem in afterburners is considered in this analysis from
two points of view. First, the design features of commercially used after-
burners are evaluated, from the point of view of their effect on mixing.
Second, to provide a basis for evaluating the potential of using mathe-
matical modeling in afterburner design, such use of mathematical modeling is
considered. In a future report, the information in the literature on fluid
dynamic performance of components of afterburners pertinent to mixing will be
reviewed.
DIRECT FLAME AFTERBURNER SYSTEMS
An effective direct flame afterburner provides (a) contact between the
contaminants in the air and the burner flame, (b) time for the combustion
process to be completed, (c) sufficiently high temperature for the complete
oxidation of the combustibles, and (d) flow patterns that ensure adequate
mixing while preventing excessive quenching. To accomplish this, the con-
taminated gases are delivered to the afterburner, where they are mixed with
the burner flame or flames in the upstream part of the unit, normally a
231
-------
refractory-lined chamber. They then pass through the remainder of the com-
bustion system, where the combustion process is completed prior to discharge
to the atmosphere. Figure 2 is a sketch of a typical fume afterburner using
a mixing plate or grid burner. An array of line burners, multijet burner, a
nozzle-mix burner, or a premix burner could be used in the same general loca-
tion. Variations on the recuperator from no recuperator to more extensive
ones can also be made. A catalytic afterburner would have the catalyst sec-
tion inserted at the downstream end of the combustion chamber, and the combus-
tion chamber would not run as hot.
In the following discussion, emphasis is placed on three areas important
to the turbulence process; namely, the burner, the approach section to the
burner, and the combustion chamber. Since the burner can be an off-the-shelf
item, with many other uses, and since it may dictate many features of the
approach section and combustion chamber, which are intimately related in con-
struction, the burners are discussed first. The aim at this point is not to
consider the details of mixing, but to consider the generic types of com-
ponents that enter into the mixing portion of an afterburner system.
It will be observed that while a wide variety of each of the components
exists, the burners considered as hot gas generators may be reduced to arrays
of jets in their influence on mixing. Thus, the fume afterburner systems may
be considered to be composed of hot jets and baffle areas, in an environment
in which the temperature range is less than in the typical furnace or boiler.
This should lead to a simplification of the overall analytical problem for the
afterburner designer.
Burners
The purpose of the burner is to provide a hot jet or jets of gas that
will raise the temperature of the fume polluted air to the desired level and
will also promote mixing of the product gases with the fume polluted air. As
much of the fume polluted air as possible is mixed with the fuel gas (or fuel
oil) in the burner, to provide "through-the-flame" destruction of the pollut-
ant, and minimize overall system size. On the other hand, to minimize fuel
costs, only sufficient heat is desired to provide adequate incineration tem-
perature for the entire polluted air flow. This requirement minimizes the
burner(ft) size and tends to reduce the amount of primary polluted air. Other
232
-------
conflicting requirements such as minimizing cost by using a small number of
burners (maybe one) and minimizing mixing problems by using many small
burners enter the problem of burner choice. Finally, an afterburner manu-
facturer may have the choice of using his own burner design, or a variety of
off-the-shelf items. Because the optimum choices are not clear-cut, and vary
with type of fume to be incinerated, amount, incinerator location, effective-
ness requirements, and fixed and operating costs, there is a wide choice of
burner possibilities available in the marketplace.
Six general types of gas burner arrays are distinguished herein: line
burners, multijet burners, ring burners, grid or plate burners, distributed
source burners, and discrete source burners. Three different types of gas
burner mixing systems are also distinguished; namely, premix, delay mix, and
nozzle mix. The burner types are discussed below. The sketches are largely
adapted from Reference (2) or trade literature.
Line Burners—
Line burners, such as in Figure 3, along with multijet burners and grid
burners, are distinguished by being an array of components, and being confined
to a rather narrow cross section normal to the flow in the afterburner system.
Because these are composed of a multiplicity of elements, their performance
may be analyzed on the basis of a single component, with the properly chosen
boundary conditions! The components or line elements are made up into lines
or grid arrays. Then profile plates are added to block part of the remaining
open area, and thus to balance off the air flow between burner elements and
the interspaces. As a result, intimate mixing of the fume polluted primary
air with the fuel takes place. The fume polluted secondary air is then
aspirated into the flame from slots defined by the profile plates along the
burner elements. Large recirculation patterns result behind the profile
plates; these produce additional mixing.
Multijet Burners—
In the typical multijet burner, such as in Figure 4, small individual raw
gas burners aspirate the required amount of fume polluted air to burn as in-
dividual flames. An array of these burners is then placed in the combustion
duct, with the remaining fume polluted air bypassing the burner array and
mixing downstream with the products of combustion. A problem with this system
233
-------
is ensuring adequate mixing downstream, because of the large blockage of the
burner array and consequent large distance required for turbulent diffusion
and mixing of primary products and secondary polluted air.
Ring Burners—
Ring burners are of the shape implied by the name. Raw gas is impinged
on a rod (Figure 5a) or on a ledge (Figure 5b), to stabilize the flames, and
increase the rate of mixing. There is no sharp demarcation of primary air
and secondary air such as there is in the multijet burner. However, the con-
cept of multiple individual flames is the same.
Grid Burners—
In a grid burner such as shown in Figure 6, a grid perforated with large
holes is used to pass the fume contaminated air. On the downstream side of
the blockage region, fuel is injected through drilled ports at a low Reynolds
number. Some polluted air is recirculated into the blockage region to burn
with the fuel. Further fuel combustion may take place immediately down-
stream, followed by spreading of the flame through the remainder of the fume-
contaminated air.
The grid is intended to cover the entire cross section of the combustion
duct.
Distributed Source Burners—
The distributed source burner (Figure 7) is quite similar to the multijet
burner (Figure 4). However, the burners act independently and the flames are
separated by the fume polluted gases moving between the burners. The burners
may be either nozzle mix, delayed mix, or premix in type (see below).
Discrete Source Burners—
Discrete source burners, such as shown in Figure 8, may be more economi-
cal to build and control. However, the mixing problem is increased as com-
pared to a multiple source system with interspace for the secondary polluted
air to come through. The single burners may be placed in an end-wall posi-
tion, or directed in from a side wall, toward the axis or with some tangential
component.
234
-------
Premix Burners—
A premix burner is defined as a burner in which the gas and air have been
previously mixed, or a burner in which the gas and air are mixed before they
reach the nozzle.
In a premix burner, the premixed fuel and air are usually supplied to
the region from an inspirator or Venturi mixer (Figure 9), an aspirator or
suction-type mixer, or a fan mixer. The burner may be a small port or ported
manifold type, a large port (or pressure type), a tunnel burner, or a flame-
retention type pressure burner. For high firing rates with turbulent flow,
the flame will not hold at the end of the duct. Therefore, a variety of
flame-holding systems are used. A flame retention type premix burner might
be used singly in a large size or as a multiple set in a pattern. Another
type of premix burner in which one mixing chamber supplies a multiple set of
ports, is similar to the ones used in most residential gas-fired heating units.
Delayed-Mixing Burners—
Delayed mixing burners as those in which the fuel and air leave the
burner nozzle unmixed and thereafter mix relatively slowly, largely through
diffusion. Figure 10 shows such a burner. This results in a long luminous
flame called a diffusion flame, luminous flame, or a long flame. Because of
their slow mixing, they are not used as discrete sources in fume afterburners.
In smaller sizes as a distributed source, they may be used successfully.
Nozzle-Mix Burners—
A nozzle-mix burner is one in which fuel and air are not mixed until just
as they leave the burner port, after which mixing is usually very rapid.
Figure 11 shows such a blower. The flame cannot flash back to this type of
burner. Nozzle-mix burners combine the advantage of the relatively short
flame of the premix burner and the lack of flash-back problems of the diffu-
sion flame. The short flames are obtained by three different methods:
(1) multiple high velocity air jets parallel to the fuel jet (the air jets
aspirate the fuel in around them and form short flames because of the small
jet diameters and potential core lengths), (2) nonparallel impinging or
interlacing air and fuel jets (a heavy recirculation zone may be formed, and
(3) a burner similar in spirit to (2), but one in which part of the air is
235
-------
mixed rapidly with the fuel, following which the highly turbulent product jet
aspirates the remaining air into itself.
Approach Section—
As used in this discussion, the "approach section" and the "combustion
chamber" comprise the fume afterburner shell between the heat exchanger
components; a generic type of burner would be used to inject the hot products
of combustion between the approach section and the combustion chamber. The
burner could draw all its fume laden air from the approach section, with the
remainder of the fume laden air mixing downstream with the burner products in
the combustion chamber, or from a separate supply of either fume laden air or
uncontamlnated (less economical) air. In any case, the configuration of the
approach section has a strong effect on the uniformity of flow the burner(s)
and to the secondary air passages, and thus affects both the performance of
the burner(s) and the combustion chamber. In other words, the approach sec-
tion configuration affects the completeness of fume combustion in a given
space. Symmetric approaches such as in Figure 8a result in uniformly
distributed downstream flows. On the other hand, the turn such as shown
in Figure 2 can lead to severe maldistributions. Several of the approach
sections that might result in problems are discussed below.
Slotted Duct System-^
Figure 12 shows an afterburner approach section with the fume-air mixture
being discharged through slots into the products of combustion of a nozzle mix
burner. The multiplicity of slots with a reasonable pressure drop should re-
move the effects of the upstream turns, especially if the duct turns are
fitted with properly designed turning vanes. However, considering the cross-
sectional view, it is seen that the passage to supply the fume-air mixture to
the bottom three slots is rather restrictive. With some designs, the flow
through the top slots may be twice that through the bottom slots because of
such a restriction. This leads to a maldistribution in the end of the burner
tunnel and in the combustion chamber.
We note that Figure 8a shows a variation on slot principle In which ori-
fices supply the fume-air mixture to the burner, and then the remaining fume-
air mixture is supplied to the fume combustion chamber by an annulus.
236
-------
Because of the symmetrical approach section in this case, however, there
would be no maldistribution.
Approach With Abrupt Turn—
Figures 2, 5b, 7, and 8b show approach sections with abrupt turns. This
may not be a problem with the distributed sources and high pressure drops of
Figures 2 and 7. Considering the flow pattern shown in Figure 13b, however,
one can see that there could be a severe flow maldistribution in Figure 5b,
for example, unless proper precautions were taken by inserting vanes,
perforated plates, or other corrective devices, or by providing sufficient
combustor chamber volume to compensate for the maldistribution.
High Velocity Approach Section—
Figure 9 shows an extreme case of a high velocity approach section; less
extreme cases could also result from too rapid expansion of the supply duct
into the burner-combustion chamber system. In the case of Figure 9, a massive
recirculation zone would exist around the entrance and extend into the com-
bustion region of the outer burner elements. The ultimate result would be
the necessity for a larger combustion chamber section and probably less ef-
ficient use of the thermal input.
On the other hand, in some designs, a high velocity product-jet may be
desired to aspirate in the fumes, which enter along the length of the jet and
intimately mix with the product gases.
Combustion Chamber
The purpose of the combustion chamber in a fume afterburner is to provide
the necessary time and temperature to completely consume the pollutants. With
perfect mixing of the secondary polluted air with the products of combustion
of the fuel and the primary polluted air [well stirred reactor concept in (1)]
a simple plug flow system would minimize the rest of the combustion chamber
size. Further analysis shows that a minimum total size can be reached with
the proper balance of well stirred sections and plug flow sections. Because
of various economic and constructional considerations, type of pollutants and
variability of supply rates, necessity of turn-down capacity, and other prob-
lems, this ideal situation does not exist. Thus there are departures from the
ideal solution. In this section, examples of the various types of combustion
chambers encountered in practice are considered.
237
-------
Simple Combustion Chambers'—
The simplest combustion system is the rectangular or the cylindrical
duct. Figures 2 and 7 show such systems using mixing plate burner and a
distributed source burner; Figure 13b shows the use of discrete source
burners. With the distributed source, the ideal plug flow situation is
closely approximated on a gross scale. With the discrete source, considera-
tion has to be given to the effect of the discrete jet. Figure 14 presents
three examples of jet-induced flow patterns. In Figure 14a, the blockage
around the entrance results in a recirculation zone with a length of the same
order of magnitude as the width. When intense swirl is added, a central re-
circulation zone is added to the flow pattern (Figure 14b). The third
example (Figure 14c) shows that even if there is no blockage, but just a slow
velocity uniform stream surrounding a high velocity jet (as from a burner), a
recirculation zone can be set up. These recirculation zones may be helpful in
moving hot products of combustion upstream where they can serve to start the
reaction in the fume polluted air. On the other hand, they take up space that
contributes nothing to the needed reaction time.
Combustion Chamber With Baffling—
Combustion chambers such as in Figures 2 and 8b may be equipped with
symmetric downstream baffles. These chambers may be circular, or square with
baffles on all sides, or rectangular with baffles on two opposite sides. It
is obvious that the baffles lead to increased pressure drop. It is not
obvious that they necessarily lead to the desired degree of mixing; an annulus
of unreacted, partially mixed fume-air mixture could pass through the entire
combustion chamber.
Rectangular combustion chambers can also be asymmetric, having one or
more bridge walls to mix the products of combustion and the fume-laden air.
The gross flow patterns are simple to deduce when zero swirl burners are used.
When a high swirl burner is used, the presence of the swirl induces further
mixing and a complex flow pattern that may be difficult to deduce.
Side-Fired Combustors—
There are several possible arrangements for side firing a combustor
chamber. The gas (or products of combustion) can be fired tangentially into
the combustion chamber, resulting in flame impingement on the wall and rapid
238
-------
mixing with the axial flow of fume polluted air through the chamber. There
may be single or multiple recirculation annuli, depending on the end condi-
tions of the chamber. The combustion gases may also move in radially and
impinge on the opposite wall, and spread in both directions. No net swirl is
present. Variations of this design can have multiple entrances.
Compact Combustion Chamber—
The burner flames of Figure 2 may be moved into close proximity with the
entrance from the heat exchanger, resulting in a combustion chamber section
similar to Figure 13a, in which the flow is turned 180°. This results in a
compact recuperator—combustor system. With the distributed combustion, the
arrangement does not cause any segregation problems. However, as shown in
Figure 13, this 180° turn can result in a reduction in effective residence
time by using up volume for the recirculation zone. The same results follow
for a configuration with a 90° turn. If a multiple burner or a discrete
source burner were used, the abrupt turn can cause an additional problem in
obtaining uniform combustion of the fumes.
Regenerative Fume Incinerator—
Figure 15 shows a fume incinerator with a regenerative system for heat
recovery rather than a recuperator system (as in Figure 1). Burners fire into
the combustion chamber from each end. In each of the three sections shown,
fume laden air will flow in one direction (say, downward) through the packed
bed, mix with the burner products, burn out the fumes, and then pass into the
opposite bed (say, the bottom bed), where heat will be given up to heat the
bed. When this bed is sufficiently hot, the entire flow direction is reversed.
The fume laden air will be heated as it moves through the hot bed (say, upward
through the bottom bed), react with the burner products, and pass through the
second bed (top) where the products give up heat. This cycle is periodically
repeated. With the direct cross flow of the fume containing gas it is clear
that the problem is to ensure good mixing with the combustion products of the
burner in the short path available.
239
-------
SECTION 3
MATHEMATICAL MODELING OF AFTERBURNER MIXING
The purpose of this section is to consider the potential of mathematical
modeling techniques in designing more cost-effective furnace afterburners.
While many currently available afterburners are based on good design practices
stemming from both field experience and research, no examples of the use of
mathematical modeling (in the present context) in afterburner design are
known. Yet, because of certain characteristics of the afterburners, there
appears more potential in using this approach for afterburner design than for
many other types of boilers and furnaces.
Considerable progress has been made during the last 20 years in the
numerical modeling of turbulent flows, both reacting and nonreacting. (3)
The results are generally obtained from the solution of time-averaged con-
servation equations, in finite difference form. However, if one is not care-
ful, a prediction procedure can rapidly become too expensive for engineering
use. Thus, when devising these mathematical models, or choosing a specific
model to use, an acceptable balance between economy and accuracy as achieved
through physically acceptable simplifications is sought. The scope of this
section on mathematical modeling in afterburners is (a) to describe some
relevant physical features of turbulence in mixing and reacting flows, (b) to
outline the principles involved in the modeling of turbulent combustion, and
(c) to suggest a balanced method that could reasonably be employed in after-
burner systems studies, stressing particularly the simplifying assumptions
involved and their justification. It is not intended to present any detailed
or specific programs; these can be developed from the literature references.
This review indicates that, compared to the current level of mathematical
modeling problem for combustors, the configuration factors for afterburners
may add to the complexity of the problem. On the other hand, kinetic con-
siderations which are only touched on briefly indicate a simple approach to
this aspect of mathematical modeling is possible, compared with the usual
240
-------
combustion problem. Furthermore, some physical simplifications in handling
the flow problems appear possible.
SOME PHYSICAL FEATURES OF TURBULENCE
The physics of turbulence in any flow system, including that in after-
burners, is very complex; yet, the effective use of turbulence is the key to
the adequate mixing required in a compact afterburner system. To get the
essence of the phenomena across clearly, the present description is quite
simplified. For exhaustive treatment, one can refer to Tennekes and
Lumley (A) and Hinze (5). Turbulent fluid motion is a time-dependent and
inherently three-dimensional condition of flow in which the various quantities
show a random variation with time and space coordinates. The momentary value
of any property 4> is conventionally decomposed into an average value $ and a
fluctuation component so
$ ¦ 4> +
-------
stretching, is called the "energy cascade". At high Reynolds numbers, the
fine scale motions responsible for dissipation are isotropic.
Correlations, Scales and Spectra
The time average of a fluctuating variable (1) is defined by
/.t0+T
i/
i / ~ dt . (2)
t
o
Time averages (mean values) of fluctuations 4* an^ of their combinations will
be denoted by an overbar; $ is zero by definition.
Consider two fluctuating variables, A + a and B + b. If ab f 0, a and b
are said to be correlated; if ab ¦ 0, the two are uncorrelated. Figure 17
illustrates this concept. (4) a has the same sign as b for most of the time;
this makes ab > 0. The variable c, on the other hand, is uncorrelated with
a and b (ac m 0, be ¦ 0) because it has statistically the same chance of hav-
ing values of the same or the opposite sign as a and b. The correlation
coefficient Rab, defined by:
R , H —, "L.. , (3)
ab (7 • b*)1/2
2 2
gives a measure for the degree of correlation between a and b (a and b are
the variances of a and b, respectively). If Rab « ±1, the correlation is
perfect.
To get an idea of the length scales of the fluctuating motion we consider
the correlation between the same fluctuating quantity (say, velocity com-
ponent, v, is the y-direction) at two different points in space separated by
a specific distance (say, r in the x-direction). The (transverse) correlation
coefficient is then given by substituting v at x for a and v at x + r for b.
Physically, the correlation with separation r is a measure of the strength of
eddies whose length in the direction of the vector "r is greater than the
magnitude of r (since eddies smaller than this will not contribute to the
correlation). A typical transverse correlation coefficient is sketched in
Figure 18.
242
-------
A length scale of the eddies could be defined as the distance r" beyond
which the correlation is practically zero. However, a more convenient'
definition is:
actually called the "integral scale". When we speak of the length scale of
the energy containing eddies we mean a length of order L. This value of L
will be comparable to that of F if the R curve does not deviate too much from
the rectangular. The difference becomes substantial if there is a certain
constant periodicity in the flow pattern, so that the R curve oscillates about
the r-axis at high values of r.
The Fourier transform of the space correlation coefficient is called the
wave number spectrum. Much basic theoretical work on turbulence is concerned
with the behavior of one- and three-dimensional wave number spectra, repre-
senting the average amount of energy of the turbulent motion between wave
numbers k and k + dk.* A typical isotropic turbulent energy spectrum is
plotted in Figure 19. These spectra are useful when dealing mathematically
with the transfer of turbulent energy from the low wave numbers (large wave-
lengths—eddy sizes) to the high wave numbers (small wavelengths—eddy sizes)
where it is dissipated.
TURBULENCE AND COMBUSTION
Because they have the greatest specific surface area, and because the
stretching process augments the concentration gradients, almost all the mole-
cular mixing occurs across the interface of the smallest vortices (4). Since
the smallest eddies are eliminated by diffusion as rapidly as they descend
from larger ones, the vortex decay time is characterized by the time scale of
large vortices. Thus the decay time is of the order L/U, where L and U are a
characteristic dimension and velocity of the flow.
For a turbulent diffusion flame, molecular mixing is a prerequisite to
combustion. Since fuel and air are initially contained in separate eddies,
*The wave number, k, is 2tt/X, where \ is the wavelength.
(4)
o
243
-------
this can only occur after a period equal to the eddy lifetime. Now the
latter are usually long: for a typical engineering flow, L/U might be
0.5 (m)/10 (m/s) - 50 milliseconds. The time scale for overall fuel/oxidant
reaction is the order, at most, of a millisecond. Therefore, turbulence, and
not chemical kinetics, is rate controlling. Similar arguments apply for pre-
mixed combustion, except that in this case the mixing is between separate
eddies of burnt and unburnt gas.
Consider mixing of two species a and b, with concentration fluctuations
ca ¦ c^ - c (where C ¦ C + c). c decreases by molecular diffusion. To an
observer moving with the fluid (6):
(5)
dc D c
dt .2
x2
where D is the diffusion coefficient, and X the eddy diameter (— is the time
scale of molecular diffusion). X decreases by stretching (Figure 9). Its
decay is assumed proportional to its dimension and inversely proportional to
the time scale of the large scale motion (4),
dX n XU
dt * " r ¦ w
Combining (5) and (6) and integrating, with X - L and c ¦ cQ for t ¦ 0 (birth
of eddy), gives:
„ L , /. . 2LU , co \
mix 2U V "d~ c"~) • <7)
It can be concluded that the turbulence time scale is of primary importance in
the mixing process. The influence of molecular diffusion is secondary, al-
though essential.
FUNDAMENTALS OF TURBULENT COMBUSTION MODELING
The Flow
As already mentioned, turbulent flows in afterburners are three-
dimensional and time-dependent. Calculation over their entire length and
time scales is both impossible (with available computers), and unnecessary
from a practical viewpoint. As a consequence, mean predictions are sought
by solving time averaged equations. Making substitutions of the type shown
in Equation (1) for the variables in the continuity and Navier-Stokes
244
-------
equations and averaging, unknown correlations of the form-pU^Uj appear in
the derived flow equations because convective terms are nonlinear. They are
apparent stresses (for instance, -puv represents an extra mean shear stress
on the face dxdz of an infinitesimal control volume with sides dx, dy and dz
in the coordinate directions). These extra turbulent stresses are called
Reynolds stresses; the time-mean Navier-Stokes equations, in which they
appear, are called the Reynolds equations. Turbulence models relate these
unknown stresses to known or determinable properties. Having more unknown
than equations in the customary description of turbulence is called the
"closure" problem.
Much of the work on turbulence modeling employs an isotropic turbulent
or "eddy" viscosity concept. This assumes that the turbulent stresses act
like the viscous stresses; that is, they are proportional to the velocity
gradient. In general form:
where 6^ is the Kronecker delta (6=0 for i f j, 6=1 for i » j).
The turbulent viscosity ut is not a property of the fluid but depends on
the dynamic condition of the flow, and its value varies from point to point.
If could be expressed in terms of known or calculable quantities, then the
derived flow equations could be solved.
Prandtl proposed in 1925 (for a simple shear flow):
where I is a characteristic length scale of the turbulent motion, called mix-
tn
ing length. £m must be prescribed algebraically; but in boundary-layer flows,
a few simple rules usually serve for its prescription.
Other models require the solution of a differential equation to determine
the turbulent viscosity. The local variation of vt is assumed to be given by
the Prandtl-Kolmogorov formula:
(8)
(9)
yt" Vkl/2*
(10)
245
-------
where k is the turbulent kinetic energy and is a constant. In this model,
the length scale I is still prescribed algebraically. On the other hand, k is
determined from the solution of a differential equation expressing the pro-
cesses by which it is transported.
Two-equation models of turbulence require the solution of two transport
equations, one for k and another one for a quantity from which £ can be
derived. These models employ for yt the same formula prescribed above. They
do not require I be prescribed algebraically, and this is an advantage in
modeling flows with recirculation, in which it is rather difficult to deter-
mine the length scale profile by measurement.
More advanced models 4pr n°t introduce the turbulent viscosity concept,
but involve the solution of transport equations for the Reynolds stresses.
This is probably the optimum level of closure in terms of physical interpreta-
tion, and it has the potential to be generally applicable to a wide range of
flows. However, Reynolds stress closures are still in a development stage,
and are computationally very expensive.
Equation (10) contains the constant C^, and other constants appear in the
transport equations for k and other variables of the turbulence models. These
"constants" are semiempirical quantities, derived by matching experimental and
numerical results by computer optimization. They are functions of the specific
model employed, and for the same model they depend on the type of flow solved,
and sometimes vary with the various investigators. Therefore, when applying
one model of turbulence, care should be taken to employ the set of constants
which is appropriate to the model and the flow type.
The specification of a length scale I is a dimensional necessity which
stems from the derived flow equations and the assumption of Equation (10).
This scale can be visualized as being the size of the largest eddies of the
flow, and thus having the order of magnitude of an integral scale L [Equa-
tion (4)]. This of course does not imply that a single value of I can be
assigned at each point for a specified flow condition. The actual value of J,
will be a function of the specific turbulence model employed. However, it is
plausible that length scales predicted with different models will be locally
of the same order of magnitude.
246
-------
Thermal and Chemical Properties
In the afterburner design considerations, the boundary can be fixed to
exclude the burner reaction region up to the point where the secondary fume
laden air mixes into the products. As a result, the thermal and kinetic con-
siderations can be simplified. Nevertheless, when dealing with a noniso-
thermal reacting system, the conservation of energy and a conservation equa-
tion for each significant chemical species have to be considered. Again using
scalar quantities of the form of Equation (1) for the composition terms,
conservation equations are produced in which terms similar to the turbulence
stress terms appear. A turbulence model is now required to cover these addi-
tional terms.
Any detailed chemical model describing the combustion of higher hydro-
carbons should take into consideration a large number of species and reaction
steps. Each species involved in the modeling of the chemical kinetics neces-
sitates the solution of an additional conservation equation, and hence in-
creases the computer time required. Moreover, uncertainties exist in both the
mechanisms and the reaction rates for the various species. For these reasons
it is common practice to assume a global, one-step reaction between fuel and
oxidant. This simple assumption allows the heat release rate to be well
evaluated, but cannot be used to predict pollutant formation and emissions.
Improvement is possible by supposing a two-step reaction in which the hydro-
carbon reacts to form CO and H2O followed by the oxidation of CO to CO2 (7).
A three-step mechanism was used by Arbib, et al. (8), for hydrocarbon attack
(to CO and H2), CO oxidation and water formation. All the above-mentioned
investigations employed global reaction rate expressions.
Instantaneous global reaction rates are usually expressed as
f\, *\j
R - K C. C , (11)
j fu ox
where K is the rate constant, usually written the Arrhenius form. The time-
averaging of Rj represents one of the central difficulties of combustion model-
ing. Calculations have often been performed by the simple expedient of
expressing the time-averaged reaction rate in terms of the individual average
values. But, since the concentration and temperature fluctuations can be
large, this practice is not, in general, justifiable. In addition to the
highly nonlinear dependence of reaction rate on temperature, a situation can
247
-------
occur in which the instantaneous values of fuel or oxidizer concentrations
vanish for finite periods of time, leading to a time-average rate of reaction
which is less than the one computed using time-averaged values of the concen-
tration (the "unmixedness" problem).
The Computation
Once the differential equations have been set up, they have to be solved
numerically. To this end, adequate algorithms exist today. They usually in-
volve writing the conservation equations in a common form, and solving them by
the finite difference procedure.
Computational codes to solve these equations are described by Gosman,
et al. (9), and Gosman and Pun (10), for two-dimensional recirculating flows;
by Patankar and Spalding (11), for boundary-layer type flows; and in the
TEACH-3E program developed at the Imperial College (and so far unpublished for
three-dimensional recirculating flows. A considerable degree of sophistication
is being achieved by using these techniques to predict combustor performance,
with a fair amount of success (12).
RECOMMENDATIONS
Because of the great variety of afterburner systems and the wide range of
fuel and waste gas combinations in them, it is rather difficult to choose a
specific model which would suit all or most of them. However, a number of
recommendations can be made in the light of previous experience.
Flow Geometry
Most industrial afterburners have a complicated geometrical form, and the
flow in the combustion region presents zones of recirculation. Therefore, in
principle, the full three-dimensional conservation equations would need to be
solved. Although methods are available for doing that, they would be presently
uneconomical for applications of this kind. Moreover, in many combustion
chambers the burners generate in their immediate vicinity very nearly
axisymmetric flames, which are often relatively insensitive to the details of
the flow elsewhere. This would suggest that a two-dimensional procedure for
axisymmetric flows with recirculation would be suitable for most afterburner
systems.. This would lead to elliptic conservation equations in cylindrical-
polar coordinates.
248
-------
Turbulence Model
The most established turbulence model to date is the two-equation model
in which equations are solved for the turbulence kinetic energy, k, and its
dissipation rate, e. The turbulent viscosity may then be related to k and e
by dimensional arguments as
vt - cyp isi
e
where is a constant of the model. The effective exchange coefficient
r ^ for a variable 6 may be expressed as
^ef f
T. pff - —, (13)
~.eff
where 0, ,, is a turbulent Prandtl or Schmidt number.
~ »eff
Most two-equation models of turbulence provide predictions of roughly
the same level of accuracy. The reason why the k-e model has been favored by
more workers than any other lies partly in the relative ease with which the
exact equation for e can be derived, and partly in the mainly esthetic fact
that e appears directly as an unknown in the equation for k. In addition,
the equation for £ has the advantage to contain fewer terms. The k-e model
also appears to give better results for flows near walls (13).
A usual approximation is to assume a. ,c « 1 in Equation (13) for any
9 > er 1
variable This makes the effective diffusive transport coefficients,
all equal (» Veff) at every location for all variables. This approximation
is very good, since usually the *re all of order unity.
Kinetic Model
The chemical aspects of afterburner systems are reviewed in the next sec-
tion. All that needs to be Btated here is that for some applications, in par-
ticular those in which the waste gas contains mostly inerts and only the tem-
perature distribution in the combustor is important, a global, one-step
reaction between the fuel and the oxidant is sufficient. CO and H2 inter-
mediates have a significant role in highly loaded combustors, but is is
sufficient to ignore them when one seeks only heat release information. The
overall reaction is not influenced by trace species (e.g., N0X) chemistry.
When this approach is not possible, it is of course necessary to keep the
249
-------
number of chemical species involved to a minimum, since every one requires
the solution of an additional differential equation.
The time averaged reaction rate could be acceptably approximated by using
individual term averaged terms In Equation (11), in applications of this kind.
This Is because, due to the minute fuel concentrations, the chemical kinetics
is relatively slow and the fluctuations in concentration are smoothed out by
mixing (14). Moreover, use of a term average version of Equation (11) would
plausibly be justified whenever global reaction-rate expressions are employed,
if these were determined experimentally through average concentration and
temperature measurements inside a combustor having a configuration similar to
the one being modeled.
250
-------
SECTION 4
KINETIC PROCESSES IN AFTERBURNERS
OXIDATION CHEMISTRY OF AFTERBURNER SYSTEMS*
Hydrocarbons in the gas phase react very slowly with oxygen at tempera-
tures below 200°C; however, as the temperature is increased a variety of
oxygen-containing compounds begin to form. As the temperature is increased
further, CO and ^0 are formed as major products and compounds such as CO2,
®2°2» an<* begin to appear. In the range from 300 to 400°C a faint light
often appears. This may be followed by one or more blue flames that suc-
cessively traverse the reaction vessel. At yet higher temperatures, 500°C or
above, explosive reactions can occur.
Hydrocarbon combustion at lower temperatures is usually initiated by the
reaction
RH + 02 R* + H02
Methane, the simplest of the paraffin hydrocarbons, requires some 40 or
more kinetic reaction steps to account for its rate of oxidation. Methane, as
essentially all hydrocarbons, is oxidized via an overall two-step set of
reactions—first the conversion of methane to CO, and second the oxidation of
CO to C02.
As the complexity of the hydrocarbon changes, i.e., unsaturated hydro-
carbons, aromatic hydrocarbons, the type and number of intermediate organic
species increases. Thus aldehydes are formed from unsaturates via
R CH - CHR' + 0, -*¦ R CH-CHR' ~ RCH0 + R'CHO
L I I
0-0
*This paper only addresses the homogeneous thermal oxidation processes and does
not include catalytic processes. The reader is referred to Reference (15) for
a detailed report on Chemical Aspects of Afterburner Systems.
251
-------
The oxidation of aromatic hydrocarbons in general has not really been in-
vestigated extensively. The major steps in the mechanism for benzene oxida-
tion are thought to be
C6H6 + °2 " C6H5 + H02
C6H5 + °2 * C6H5°2
C6H5°2 + C6H6 * C6H5° + C6H5 + 0H
C,HcO + C,H, -»• C,HcOH + C,Hc
65 66 65 65
The phenol, C^H^OH, which forms in high yields because of its great stability
is further oxidized to C^H^COH^, which is in turn oxidized according to the
scheme shown in Figure 21. The products on the right in the figure can then
be oxidized further. Oxidation of the acetylene generally occurs through a
series of chain reactions with formaldehyde and formic acid as intermediates.
The high-temperature oxidation of acetylene is also complicated by the tendency
of acetylene to polymerize.
QUASI-GLOBAL KINETICS
Kinetically the three controlling parameters for afterburner operation
are time-temperature-concentration. Hydrocarbon reactions can be represented
globally by the equation
Ca*b + (a + 4) °2 " a C02 + 2 V
where CaH^ is any hydrocarbon. The process of converting CaH^ to CO2 and H2O
is kinetically very complex. Kinetic models for hydrogen (17), methane (18)
and a number of straight chain hydrocarbons and partially oxygenated
organics (19) have been described by a kinetic mechanism consisting of
69 reactions involving 31 species. Edelman and Fortune (20) have taken these
complex models and developed a quasi-global finite rate combustion model which
reduces the number of steps and species involved, and hence the complexity,
in adapting the oxidation models to a wide variety of hydrocarbons.
In general, both the homogeneous gas-phase and catalytic reactions involv-
ing the destruction of organics or hydrocarbons in afterburner systems can be
represented by the chemical equation
a F + 6 O2 -*• Products
252
-------
where F denotes the hydrocarbon fuel and a and $ are stoichiometric factors.
The chemical reaction rate for the disappearance of the hydrocarbon can be
written as
I ne (14)
a dt F o2
Since oxygen is in large excess in afterburners this rate expression is
readily reduced to a first order rate
1 dnF a
a dt k "F * (15)
HOMOGENEOUS OXIDATION REACTIONS
Homogeneous Oxidation Rate Data
CH^—A number of investigators have reported global rate constants for
CH4. Expressions developed by the different investigators are presented in
Table II. The equation of Dryer and Glassman is considered to be the best
rate expression and is the one recommended here for afterburner applications.
C2H5—For ethane oxidation in the presence of excess oxygen the rate
expression,
_M7.18 e-32,900/RT
was obtained by Glassman, Dryer, and Cohen. (25) The temperature range over
which this relationship applies is about 900 to 1050°K.
Higher CnH2n+2 Hydrocarbons—Flame speed and kinetic measurements indicate
that oxidation rates for paraffinic hydrocarbons in the series from propane
(C3Hg) beyond decane (C^o^22^ are weH within an order of magnitude of those
of propane.
The burning rate of propane has been measured in terms of CO2 formation
to be
' dt8' * "2"9 * 10l° a"15'000/*1 f°'35 fcj i°jo (|f) moles/cm^-sec
253
-------
where f represents the mole fraction. (26) The temperature range over which
the data were obtained was from about 1400 to 1800°K.
Using the detailed chemical kinetic mechanism of Chinitz and Bauer (27)
which involved 31 chemical species participating in 69 elementary reaction
steps, Edelman and Fortune (20) give the overall rate expression for propane
oxidation as
d t C3Ho] o *
—ir---ktC3H8]°-5 [02]
where
.8 * 1°9 [fgf - 0.5]
T0.5 p0.2 e-13,700/RT
with F in atmospheres. The applicable temperature range is estimated to be
from 800 to 3000°K, and is based on the conversion to CO and H2»
Complete combustion to CO2 and H2O must take into account the kinetics of
the H2 and CO reactions. A more representative expression'for k,
k - 5.52 x 108 p~0'815 T e"12'200/'T
has been suggested by Glassman (25) and Engelman (28). The above rate expres-
sion has also been suggested as being applicable to the overall oxidation of
hydrocarbons represented by the generalized chemical equation
a„ b.
+ 2°2 2*2 + aC0
with the rate given as
[ ^tb] - -5.52 x 10® p-0*815 T e~12»20°/T [CaHb]0,5 [02] moles/cm3-sec .
A general overall mechanism based on the above rate expression is given in
Table III.
Nettleton (29) developed a global rate expression for hexadecane. His
relationship is based on second order kinetics where
. % ,n14 -13,200/T 3, ,
k ¦ 10 e cm /mole-sec
Figure 22 compares Nettleton's rate expression for the removal of hexa-
decane with Hemsath and Susey's data. (30) The two expressions differ by a
factor of 10 in time required for complete removal of hexadecane. For
254
-------
afterburner design purposes however either set of equations allows one to
design for effective removal of the hexadecane.
Global rate constants for a number of organic oxidation reactions com-
piled by Seshadri and Williams (31) are listed in Table IV. In general there
is a strong similarity in the rate constants for the different species.
Hemsath and Susey (30) have made oxidation rate-constant measurements in
actual thermal-type afterburner systems. Their rate constants, based on first
order kinetics, are presented in Table V.
A number of values for activation energies have been reported for various
hydrocarbon oxidation reactions. Typically these activation energies range
from about 30 to 50 kcal/mole for most organics, and run about 78 kcal/mole
for CO. [The reader is referred to Table 17 of Reference (15) for specific
values.] Falconer and Van Tiggelen (32) have observed a correlation between
activation energies for hydrocarbon oxidation and the weakest C-H bond in the
hydrocarbon. These results are summarized in Figure 23. For unsaturated
compounds, a reaction at the multiple bond would normally control the activa-
tion energy for subsequent branching. In the case of an unsaturated hydro-
carbon containing an easily abstracted hydrogen atom, as, for example,
propylene, the reaction path with the lower activation energy would be ex-
pected to predominate in the chain branching. By extrapolating Figure 23 an
activation energy of about 20 kcal/mole might be expected for propylene, based
on a carbon-hydrogen bond strength of 77 kcal/mole in the parafflnic portion
of the molecule.
Aromatics—Global rate data for the oxidation of toluene based on first-
order kinetics are presented in Table VI. The values for benzene given in the
table are based on second-order kinetics.
Benzene--
Lee, Jahnes and Macauley (33) have examined the oxidation of several
organics, including benzene, in a plug flow laboratory reactor. In Figure 24
we compare their measurements with the expression repeated for benzene in
Reference (31). The rate expression of Reference (31) shows fairly good
agreement with the measurements of Lee, et al.
255
-------
Toluene—
Figure 25 presents concentration-time plots for the oxidation of toluene
comparing Hemsath and Susey's laboratory data with their derived rate ex-
pression and with another expression [Reference (34)]. The plots show a
significant difference in the reaction time plots suggesting that the ex-
pression in Reference (34) may have too high an Arrhenius factor.
CO—A number of investigators have determined global rate constants for
CO oxidation. The important CO oxidation reaction is
CO + OH -»• C02 + H
with the direct oxidation reaction
CO + 02 C02 + 0
being very slow.
Rolke, et al. (2) and Williams, et al. (35) have reviewed the early
global rate data for CO oxidation. Selected results and more recent data are
summarized in Table VII. A marked variation is observed in the different
rate constants. The only actual afterburner rate data for CO are those by
Hemsath and Susey. (30) These latter results are recommended for afterburner
design applications.
PREDICTIVE METHODS
As one considers the potential number of organics that can be destroyed
in fume incineration and, the diversity of chemical structure and reactivity,
one wonders whether rate data have to be run on every compound. The
answer to this question is yes and no.
In general increasing the number of carbon or oxygen atoms in a molecule
or increasing the residence time decreases the temperature requirements. In
the other direction increasing the aromaticity or the unsaturation in com-
pounds, results in more stable species and requires higher temperatures for
burnout.
Predictive methods for afterburner kinetics have limited usefulness, in
part because in more cases than not, mixing may control the overall efficiency
of burnout. On the other hand some predictive, correlation techniques are
available. Lee, Hansen and Macauley (37) have developed a correlation model
256
-------
for estimating reactivity based on structure and autoignition temperatures of
the compound.
Essentially their method calculates destruction temperatures for 99,
99.9, and 99.99 percent destruction from a set of correlation equations, then
uses these temperatures to calculate the A and E values in an Arrhenius equa-
tion, The correlations work out surprisingly well for the 15 compounds they
were able to run in the laboratory.
257
-------
SECTION 5
REFERENCES
(1) Barrett, R. E., and Sticksel, P. R., A Preliminary Environmental
Assessment of Afterburner Combustion Systems. In: Proceedings of the
Third Stationary Source Combustion Symposium, San Francisco, California,
1979, Vol. Ill, pp 79-93.
(2) Rolke, R. W., Hawthorne, R. D., Garbett, C. R., Slater, E. R., Phillips,
T. T., and Tovell, G. D., Afterburner Systems Study, Final Report
under U.S. EPA Contract EHS-D-71-3. U.S. Government Printing Office,
PB 212560 (1972).
(3) McDonald, H. (1979). Combustion Modeling in Two and Three Dimensions—
Some Numerical Considerations. Prog. Energy Combust. Sci., 5, pp 97-122.
(4) Tennekes, H., and Lumley, J. L. (1972). A First Course in Turbulence.
The MIT Press.
(5) Hinze, J. 0. (1975). Turbulence, 2nd Ed., McGraw-Hill.
(6) Spalding, D. B. (1976). Mathematical Models of Turbulent Flames; a
Review. Combust. Sci. Technol., 13, pp 3-25.
(7) Schefer, R. W., and Sawyer, R. F. (1976). Pollutant Formation in Fuel
Lean Recirculating Flows. NASA CR-2785.
(8) Arbib, H. A., Goldman, Y., Greenberg, J. B., and Timnat, Y. M. (1980).
A Numerical Model of High Intensity Confined Hydrocarbon Combustion.
Combustion and Flame (in press).
(9) Gosman, A. D., Pun, W. M., Runchal, A. K., Spalding, D. B., and
Wolfshtein, M. (1969). Heat and Mass Transfer in Recirculating Flows.
Academic Press.
(10) Gosman, A. D., and Pun, W. M. (1973). Calculation of Recirculating Flow.
Heat Transfer Report No. HTS/74/2, Imperial College, London.
(11) Patankar, S. V., and Spalding, D. B. (1970). Heat and Mass Transfer in
Boundary Layers, 2nd Ed., Intertext, London.
258
-------
(12) Abou Ellail, M.M.M., Gosman, A. D., Lockwood, F. C., and Megahed, I.E.A.
(1978). Description and Validation of a Three-Dimensional Procedure
for Combustion Chamber Flows. Progress in Astronautics and Aeronautics,
Vol. 58, pp 163-190.
(13) Launder, B. E., and Spalding, D. B. (1974). The Numerical Computation
of Turbulent Flows. Computer Methods in Applied Meclianics and Engi-
neering, 3, pp 269-289,
(14) Lockwood, F. C. (1977). The Modelling of Turbulent Premixed and Diffu-
sion Combustion in the Computation of Engineering Flows. Combust.
Flame, 29, pp 111-122.
(15) Barnes, R. H., Saxton, M. J., Barrett, R. E., and Levy, A., Chemical
Aspects of Afterburner Systems. EPA-60G/7-79-096, U.S. Environmental
Protection Agency, Research Triangle Park, N.C., 1979, 117 pp.
(16) Bradley, J. N., Flame and Combustion Phenomena, Mathuen & Co. Ltd.,
London (1969).
(17) Libby, P. A., Pergament, H. S., and Bloom, M. H., A Theoretical Investi-
gation of Hydrogen-Air Reactions, Part I, GASL TR-250, August 1961.
(18) Chinitz, W., Pyrodynamics, 3, 1966, p 196.
(19) Chinitz, W., and Baurer, T., An Analysis of Non-Equilibrium Hydrocarbon-
Air Combustion, Paper 65-19, presented at 1965 Fall Meeting, Western
States Section, The Combustion Institute.
(20) Edelman, R. B., and Fortune, 0. P., A Quasi-Global Chemical Kinetic
Model for the Finite Rate Combustion of Hydrocarbon Fuels With Applica-
tion to Turbulent Burning and Mixing in Hypersonic Engines and Nozzles,
AIAA Paper No. 69-86, New York (1969).
(21) Nemeth, A., and Sawyer, R. F,, The Overall Kinetics of High-Temperature
Methane Oxidation in a Flow Reactor, J. Phys. Chem., 73, 2421 (1969).
(22) Kozlov, G. I., On High-Temperature Oxidation of Methane, Seventh
Symposium (International) on Combustion, p 142, The Combustion Institute,
Pittsburgh, Pa. (1959).
(23) Williams, G. C., Hottel, H. C., and Morgan, A. C., The Combustion of
Methane in a Jet-Mixed Reactor, Twelfth Symposium (International) on
Combustion, p 913, The Combustion Institute, Pittsburgh, Pa. (1969).
(24) Dryer, F. L., and Glassman, I., High-Temperature Oxidation of CO and
CH^, Fourteenth Symposium (International) on Combustion, p 987, The
Combustion Institute, Pittsburgh, Pa. (1973).
259
-------
(25) Glassman, I., Dryer, F. L. , and Cohen, R., Combustion of Hydrocarbons
in an Adiabatic Flow Reactor: Some Considerations and Overall Cor-
relations of Reaction Rate, paper presented at Joint Meeting of the
Central and Western States Sections of the Combustion Institute,
San Antonio, Texas (April 21-22, 1975).
(26) Hottel, H. C., Williams, G. C., Nerheim, N. M., and Schneider, G. R.,
Kinetic Studies in Stirred Reactors: Combustion of Carbon Monoxide
and Propane, Tenth Symposium (International) on Combustion, p 111,
The Combustion Institute, Pittsburgh, Pa. (1965).
(27) Chinitz, W., and Baurer, T., An Analysis of Nonequilibrium Hydrocarbon/
Air Combustion, Pyrodynamics, 4, 119 (1966).
(28) Engleman, V. S., Bartok, W., Longwell, J. P., and Edelman, R. B.,
Experimental and Theoretical Studies of N0X Formation in a Jet-
Stirred Combustor, Fourteenth Symposium (International) on Combustion,
p 775, The Combustion Institute, Pittsburgh, Pa. (1973).
(29) Nettleton, M. A., Ignition and Combustion of a Fuel of Low Volatility
(Hexadecane) in Shock-Heated Air, Fuel, 53, 88 (1974).
(30) Hemsath, K. H., and Susey, P. E., Fume Incineration Kinetics and Its
Applications, AIChE Symposium Series No. 137, Vol. 70, 439 (1974).
(31) Seshadri, K., and Williams, F. A., Effect of CF^Br on Counterflow
Combustion of Liquid Fuel With Diluted Oxygen, in Halogenated Fire
Suppressants, R. G. Gann (Ed.), ACS Symposium Series 16 (1975).
(32) Falconer, W. E., and Van Tiggelen, A., A Kinetic Study of Hydrocarbon-
Oxygen-Nitrogen Flame Systems and Molecular Weights of Chain Carries,
Ninth Symposium (International) on Combustion, p 689.
(33) Lee, K. C., Jahnes, H. J., and Macauley, D. C., Thermal Oxidation
Kinetics of Selected Organic Compounds, Jour. Air Pollut. Control
Assoc., 29 (7), 749-751 (1979).
(34) Perry, R. H., and Chilton, C. H., Chemical Engineers' Handbook, 5th Edi-
tion, McGraw-Hill (1973).
(35) Williams, G. C., Sarofim, A. F., and Lambert, N., Nitric Oxide Formation
and Carbon Monoxide Burnout in a Compact Steam Generator, in Emissions
from Continuous Combustion Systems, W. Cornelius and W. G. Agnew (Eds.),
Plenum Press, New York (1972).
260
-------
(36) Howard, J. B., Williams, G. C., and Fine, D. H., Kinetics of Carbon
Monoxide Oxidation in Postflame Gases, Fourteenth Symposium (Inter-
national) on Combustion, p 975, The Combustion Institute, Pittsburgh,
Pa. (1973).
(37) Lee, K. C., Hansen, J. L., and Macauley, D. C., Predictive Model of the
Time-Temperature Requirements for Thermal Decomposition of Dilute
Organic Vapors, Paper 79-10.1, 72nd Annual Meeting of the Air Pollution
Control Assoc., Cincinnati, Ohio, June 24-29, 1979.
261
-------
Exhaust
Cross-Over Duct
I Combustion
- Chamber
Raw Gas Burner
Waste Gas
X fum*
CembuMion
kaiofl
Mixing
Stcttan
FIGURE 1. SCHEMATIC OF TYPICAL AFTERBURNER
FIGURE 2. FUME AFTERBURNER WITH MIXING-PLATE
BURNER AND RECUPERATOR
Profile Pttt*
FIGURE 4. HIRT MULTIJET GAS BURNER
Fuel Ch'
FIGURE 3. MAXON LINE GAS BURNER WITH
PROFILE PLATES
262
-------
Exhaust
Slack
LphiiiiityJ
fill*
iu
Burnar
Gu
isi
^ V ' iy**£f?
futi-Aieh
J Puma
I lnl«t
is) Midland Rot*
(to) JohnZmk
FIGURE 5. RING BURNERS
FIGURE 6. NORTH AMERICAN FLAME
GRID BURNER
FIGURE 7. AFTERBURNER WITH
DISTRIBUTED BURNER
Fual-
•hhaum
Pima
'bhwnt
FIGURE 8. DISCRETE SOURCE BURNERS
263
-------
FimI
Air RHUHt
FUm»
ArrMtine
Poftt
f
Su««m
Noxzf*
Air
Ges-
>¦-'! T r I
*
_ *
fir. ¦:
[R
kmm
FIGURE 9. PREMIX BURNER
FIGURE 10. DELAYED MIXING BURNERS
Gm ¦
FIGURE 11. NOZZLE-MIXING BURNERS
Noai#
Mix
Burrwr
11 ¦
J
*tja
A-1
7ZZZ2ZZ
Croti Section A-A
FIGURE 12,
AFTERBURNER APPROACH SECTION
WITH SLOTTED DUCT
264
-------
(«> Low Swirl
A. U(and
8hwl*d Araai R*pt«nm
Rtcirculding Fkwn—
DMdZonw
8. LBand
FIGURE 13. RECIRCULATION REGIONS WITH
180* AND 90* TURNS
Limit e( tta R*efeu(M(«n
-------
b
c
FIGURE 17. CORRELATED AND UNCORRELATED
FLUCTUATIONS
R*i
FIGURE 18. TYPICAL TRANSVERSE
CORRELATION COEFFICIENT
Indapandant of Condition of Formation
Dapandant
on
Condition
of
Formation
Wavanumbar k
Enargy
Eddie* of
ComaintiYQ
Character
Universal Equilibrium R*nga
| Diwipatton
Range
Inertial
Range
FIGURE 19. TYPICAL ISOTROPIC TURBULENT
ENERGY SPECTRUM
/ \ *hcho
MM \ W4tfcvd*v
|-OM \
•OH
AlCChoH
i
— *co
1 J
RCOi
I
KCOiH
#«f*Acidt
—i
\
RCOOH
ACitfl
FIGURE 20. SUMMARY OF PRINCIPAL
CHEMICAL REACTIONS INVOLVED IN
HYDROCARBON OXIDATION
266
-------
1.00
0.500
W 0.30
w>
0.10b
Hexadecane
0.05 b
0.03 h
0.001 0.005 0.009 0.013
Time, sec
1.00
o>
c 0.30
c
<5
£
cc
c
o
o
BJ
Hexadecane
0.03 0.07 0.11
Time, sec
0.15
FIGURE 22. HEXADECANE KINETICS
A - Reference 29
B - Reference 30
Hydtoqulnorit
^c«o
* CiHi
S>H
FIGURE 21. REACTION STEPS IN BENZENE
OXIDATION THROUGH
HYDROOUINONE AS AN
INTERMEDIATE
± 40
8
i
so
Benzene,
Ethane 0
Neopentanef
Methane
n-Butane
Isobutane
•o K ioo
Bond Strength, kcal mole'1
FIGURE 23. DEPENDENCE OF ACTIVATION
ENERGY UPON BOND STRENGTH
OF WEAKEST C H BOND<32'
267
-------
O)
c
CO
£
O)
cc
c
o
o
(0
Benzene
a
10,000i
001
0.06 014 0.22
Time, sec
o
E
n
1000
Mi
a
500
c
"o
c
100
0
u
50
-------
TABLE I. COMPARISON OF TEMPERATURES
REQUIRED TO OXIDIZE VARIOUS
COMPOUNDS TO CO2 AND H20(2)
Ignition Temperature, °F
Compound
Thermal
Catalytic
Benzene
1076
575
Toluene
1026
575
Xylene
925
575
Ethanol
738
575
MIBK
858
660
MED
960
660
Methane
1170
932
Carbon Monoxide
1128
500
Hydrogen
1065
250
Propane
898
500
TABLE II. GLOBAL RATE CONSTANTS FOR CH4 OXIDATION TO CO
Temperature
Investigators Rate Expression Range Reference
(21)
(22)
(23)
Nemeth and
Sawyer
Kozlov
Williams et al.
= -6 X 10'° (CH,)"04 [O2]1'4 e"57,000/RT
dt
mole/cmJ-sec
= -7 x io* [CH4]05 [O2}15 r1 e-60,000/RT
dt
d[CH4)
dt
mole/cmJ-sec
= -5.3 X 1015 e-57'000/RT fen, fSj f°H25o
mole/liter-sec
W
(= mole fraction, and P/RT is in moles/liter
d[CH<]
Dryer and = -10'" e4MOO/RT [C^f [O2)0#
Glassman
mole/cm -sec
>1200°K
1200-1400°K
1450-1750°K
1100-1400°K
269
-------
TABLE III. GENERAL OVERALL MECHANISM FOR COMPLETE OXIDATION
OF HYDROCARBONS*20)
k = ATb exp(-E/RT)*
Forward
Reaction A b E/R
C.Hb + | 02 -1 H, + nCO 5^1°' • C&b Co2 1 12.2 X10'
C0 + 0H = H + C02 5.6X10" 0 0.543 X10J
OH + H2 = HjO+H 2.19X10" 0 2.59 X10J
OH + OH = O + HjO 5.75 X1012 0 0.393 X 103
O + Hj = H + OH 1.74X10" 0 4.75 X10J
H + 02 = 0 + 0H 2.24 X10M 0 8.45 X103
M + O + H = OH + M 1X10" 0 0
M + 0 + 0 = 02 + M 9.38 X 1014 0 0
M + H + H = H2+M 5X10" 0 0
M + H + OH = HjO + M 1X1017 0 0
•Units: cm'/mole-sec for bimolccutar reactions; cmVmole-sec for tcrmolecular reactions.
270
-------
TABLE IV. RATE CONSTANTS FOR OXIDATION OF MISCELLANEOUS
ORGANIC COMPOUNDS^1)
Hydrocarbon
Vapor
Chemical
Formula
Activation
Energy,
kcal/mole
Preexponential
Factor,
cmJ/mole-sec
Isopropyl ether
[(CH,)2CH]20
50
1.2 X10'7
Butyl-vinyl ether
CH,(CH2)3OCH:CH2
31
5.0 X 10u
Methyl methacrylate
c3h8o2
37
2.4 X 10'5
Polymethy! methacrylate
(CjHg02)B
42
7.6 X 10,s
Methanol
CHjOH
40
4.6X10"
Heptane
C7H16
38
2.2X10"
Decane
CioH22
37
1.6 X 101J
Hexadecane
C,6H34
35
8.2 X1014
Iso-octane
CiHi«
35
4.5 X 1014
Kerosene
CioH2o (approx.)
35
5.4 X1014
271
-------
TABLE V. RATE CONSTANTS MEASURED IN THERMAL
AFTERBURNER SYSTEM^)
Compound
Preexponential Constant,
k, sec*1
Activation Energy,
Ea, kcal/mole
Hexane
4.5 X 101J
52.5
Cyclohexane
5.13 X 101J
47.6
Natural gas
1.65 X 10,:
49.3
TABLE VI. GLOBAL RATE DATA FOR THE OXIDATION OF AROMATICS
Compound
Preexponeniial Term,
A
Activation Energy,
Ea, kcal/mole
Reference
Benzene
6.0 X 1014 cmVmole-sec
36
(31)
Toluene
6.56 X 1013 sec"'
56.5
(30)
272
-------
TABLE VII. GLOBAL RATE CONSTANTS FOR CO OXIDATION
Investigators Rate Expression, mole/cm3-sec Temperature Range References
Williams, Hottel, and
Morgan
Howard, Williams,
and Fine
d (C°l = -i 8 x 107 e-^,oooiky f fo.5 f0.5 (P/RT)2
i.oxiu e 'CO 02 TH20 1
d d°°' " ~1J X T°14 'C0' f°2J1/2 fH2°J1/2 e*30'000^1
1450-1750°K
840—2360°K
(23)
(36)
Dryer and Glassman
d = -3.9 x 1014 e"40'000^7 [CO]1-0 [H2Oj0-5 [02J0*25
1030-1230°K
(24)
Hottel, Williams, Nerheim
and Schneider
Hemsath and Susey
d [dC01 = -1.2 X 10" e-'MOO/RT fO.3 f co f 0.5q (p/RT)1.8
^ = toj
"dfco =-15 x 10" ,-»««« fco f» s sec"1
1250-1550°K
<1400°F (1033°K)
>1400°F (1033°K)
(26)
(30)
(30)
Alt concentrations ( | in mole/cm3, f is the mole fraction.
-------
SUBSCALE TESTS OF COMBUSTION MODIFICATION
FOR STEEL FURNACES
By:
R. J. Tidona, W. A. Carter, and S. C. Hunter
KVB, Inc.
18006 Skypark Boulevard
Irvine, California 92714
274
-------
ABSTRACT
This is a report of a research program to develop combustion modification
technology as means of emissions reduction and thermal efficiency improvement
on industrial process equipment. The work is an extension of EPA Contract 68-
02-2645, which concentrated on operational adjustments. Presented are results
of subscale tests for steel furnaces.
Subscale tests with a standard steel furnace burner firing natural gas
and No. 2 oil were conducted to determine the effects on N0X emission and
furnace efficiency of water injection into the flame zone, steam injection
into the flame zone, flue gas recirculation, and lowered excess air* With
natural gas fuel the largest N0X emission reduction was obtained using flue
gas recirculation (88% reduction). With No. 2 fuel oil the largest reduction
occurred using steam injection (89%).
The costs of water injection, steam injection, and flue gas recirculation
were evaluated. Steam injection was found to be the most cost effective
combustion modification technique for three heater sizes firing either natural
gas or No. 2 oil.
275
-------
SECTION 1
INTRODUCTION
At the Third EPA Stationary Source Combustion Symposium, KVB reported on
subscale process heater combustion modification tests. The report summarized
the effects on NOx emissions of several types of modifications/ including
staged combustion air, flue gas recirculation, lowered excess air, altered
injection geometry, and low NO^ burner installation. The work showed that
staged combustion air appears to be the most cost effective combustion modifi-
cation for process heaters. Both staged air and flue gas recirculation
produced NOx emission reductions in excess of 60% below baseline emission
levels.
As a continuation of this subscale test work, which is part of a program
to develop advanced combustion modification techniques for industrial process
equipment, KVB evaluated several of the same combustion modifications on a
subscale steel furnace. This paper presents the subscale steel furnace test
findings. It is emphasized that these results have not been demonstrated at
the full-scale level and, therefore, the modifications cannot now be
considered as proven concepts.
OBJECTIVE AND SCOPE
The objective of the program of which these steel furnace tests are only
one part is to develop advanced combustion modification concepts requiring
minor hardware modifications that could be used by operators and/or manufac-
turers of selected industrial process equipment to control emissions. The
development is aimed at equipment on which the modifications will be most
widely•applicable and of the most significance in mitigating the impact of
276
-------
stationary source emissions on the environment. The program involves
investigation not only of emissions but also multimedia impacts and control
cost effectiveness.
The program includes both subscale and full-scale testing. Subscale
testing is a necessary part of development of new hardware to ensure accept-
able performance/ which is a vital aspect of emissions control. Full-scale
testing is also necessary on more than one process design configuration (e.g./
forced draft and natural draft) before equipment manufacturers and the process
industry can employ a given emission control technology.
At the conclusion of the study, a final engineering report will be
prepared summarizing the accomplishments of the subscale and full-scale
demonstration tests. A series of guideline manuals will be prepared to
acquaint equipment manufacturers with the most promising emission control
methods that have been demonstrated and to offer technical guidance that can
be directly applied in their process equipment design.
MODIFICATIONS EVALUATED
KVB evaluated the following modifications to the subscale steel furnace
located at the technical center of a major steel furnace burner manufacturer:
1. Lowered excess air.
2. Steam injection when firing No. 2 oil fuel.
3. Water injection when firing natural gas fuel.
4. Flue gas recirculation.
The details of the experimental arrangement and the results obtained are
given in the following sections of this paper.
The cost effectiveness study indicates that steam or water injection
offers the best N0X reduction capability for the least cost. However/ flue
gas recirculation gives the largest N0X reduction when firing natural gas.
Excess air variations did not affect N0X emissions significantly except at a
high excess oxygen level or a very low excess oxygen level/ neither of which
is a practical operating condition.
277
-------
SECTION 2
EMISSIONS SAMPLING AND TEST APPARATUS
EMISSIONS SAMPLING
The research steel furnace emission measurements were made with
instrumentation, carried in a mobile laboratory, which was described in detail
in the EPA Interim Report entitled "Application of Combustion Modification to
Industrial Process Equipment," Contract No. 68-02-2645.
Gaseous species measurements were made with analyzers located in the
trailer. Particulate emission and size measurements were not made during
subscale tests to allow a larger range of test variables for effects on
gaseous emissions. These measurements will be made on full-scale units. The
emission measurement instrumentation used is listed in Table I.
GAS SAMPLING AND CONDITIONING SYSTEM
The flue gas sampling system uses positive displacement diaphragm pumps
to continuously draw flue gas from the stack into the laboratory. The probes
are connected to the sample pumps with 0.95 cm (3/8 in.) or 0.64 cm (1/4 in.)
nylon line. The positive displacement diaphragm sample pumps provide unheated
sample gas to the refrigerated condenser (to reduce the dew point to 35°F), to
a rotameter with flow control valve, and to the 02, NO, CO, and C02
instrumentation. Flow to the individual analyzers is measured and controlled
with rotameters and flow control valves. Excess sample is vented to the
atmosphere.
To obtain a representative sample for the analysis of N02, S02 and
hydrocarbons, the sample must be kept above its dew point since heavy
278
-------
hydrocarbons may be condensible, and SO2 and NC>2 are quite soluble in water.
For this reason, a separate electrically-heated sample line is used to bring
the sample into the laboratory for analysis. The sample line is 0.64 cm
(1/4 in.) Teflon, electrically traced and thermally insulated to maintain a
sample temperature of up to 400°F. Metal bellows pumps provide sample to the
hydrocarbon, SOj, and N0X continuous analyzers.
TEST APPARATUS
The testing discussed in this report was done in a small research steel
£
furnace with a maximum firing rate of 0.6 MW (2 x 10 Btu/hr) located at the
test facility of a major manufacturer of steel furnace burners. Both natural
gas fuel and No. 2 oil were fired in a standard burner provided by the manu-
facturer. A schematic of the experimental apparatus is presented in Figure 1.
The test apparatus consisted of a burner firing into a test furnace, with
provisions for flue gas, steam, and water to be introduced into the burner
flame. The test furnace served as a combustion chamber with a residence time
of about two seconds when firing at 0.6 MW. The furnace could operate at
1978K (3100#F) and was outfitted with numerous access ports for visual
observation and temperature measurement. The furnace temperature was main-
tained at 1533K (2300°F) throughout the test by exposing more or less of the
water-cooled probes to the furnace interior. This was done to simulate condi-
tions in an actual furnace.
The 0.6 MW (2 x 10® Btu/hr) burner was used to simulate the commonly used
2.4 MW (8 x 10 Btu/hr) version. The burner can be fired on natural gas.
No. 2 fuel oil, or both simultaneously.
The recirculation of flue gases into the burner flame was accomplished by
passing a portion of the furnace exhaust's flue gases by means of a blower
through a stainless steel air-to-air heat exchanger and combining this flow
with the combustion air flow. The temperature of the recycled flue gases was
maintained at HiSSK (500°F) by adjusting the flow of cooling air through the
heat exchanger. Good mixing of the combustion air and flue gases was assured
by employing a diffuser between the combustion air/flue gas plenum and the
burner.
279
-------
A 20-cm (8-inch) diameter orifice was added to the furnace exhaust stack
to stop ambient air entrainment in the exiting flue gases, resulting in
inaccurate flue gas 02 readings, and place the furnace and the flue gas recir-
culation (FGR) ductwork under positive pressure to reduce the infiltration of
ambient air through the heat exchanger and blower.
All gas flows (combustion air, PGR, steam, atomizing air, and natural
gas) were measured with the aid of orifices and manometers and are considered
to be accurate to within 5%. All liquid flows (No. 2 fuel oil and water) were
measured with rotameters which had been calibrated with the fluid to be
measured. The installation of an analyzer in the PGR ductwork just
upstream of the combustion air plenum became necessary to determine the degree
of flue gas dilution with infiltrating ambient air.
Water was injected into the flame zone only when firing natural gas, and
steam was injected only when firing No. 2 oil. During natural gas firing,
water was injected through the unused oil port. During oil firing, steam was
injected through an unused gas annulus. Temperature measurements of all
flows, including the flue gas temperature, were made using type "K" (61K to
1589K) thermocouples. Flame temperature profiles were obtained using a type
"R" (256K to 2033K) aspirated thermocouple. Flame temperature profile mea-
surements were made for each of the modified conditions as well as for
baseline conditions firing both natural gas and No. 2 oil.
280
-------
SECTION 3
COMBUSTION MODIPICATIONS
The overall results of the combustion modification tests are most
encouraging from the standpoint of NO emission reduction potential. The
maximum NO reductions obtained for each modification are summarized in
Table II# The average baseline N0X emission for a steel furnace burner firing
natural gas and No. 2 oil is given in Table III.
Gaseous emissions were measured at baseline conditions and at various
excess air settings at full capacity and half capacity (nominally). In addi-
tion to excess air variation, steam injection and water injection were tried
at full capacity firing No. 2 oil and natural gas, respectively, to reduce N0X
emissions. Flue gas recirculation was also tested firing each of the two
fuels.
Figures 2 and 3 show the effect of excess oxygen on NO missions when
firing natural gas and No. 2 oil. For both fuels, NO emission peaked at about
2% 02« As the 02 was increased beyond 2%, the NO concentration tended to
decrease. The NO concentration also decreased at excess 02 levels below 2%,
but the trend was less pronounced. (The high furnace temperatures which
occurred at low excess 02 conditions on several occasions caused NO emission
at these conditions to be higher than it would have been if the temperatures
had been held constant.)
There is an apparent considerable spread in the data for NO emission
versus stack excess oxygen. Figures 2 and 3 suggest a family of curves for NO
vs. 02» This indicates that another important factor is involved in deter-
mining NO levels. It is believed that this factor is combustion air humidity,
and that each curve in the "family" of curves represents a constant combustion
281
-------
air moisture content. Unfortunately, precise moisture data were unavailable
at the test site. Dry bulb and relative humidity data were obtained from a
weather station approximately ten miles from the test site. These data were
used in the construction of Figures 4 and 5. In these figures the moisture in
the combustion air was added to the H^O injected through the burner.
Figures 4 and 5 reveal the sensitivity of NO to change in H20 injection
rate, particularly when firing No. 2 oil. The maximum percent NO reduction
obtained by injecting water with natural gas was 47% as compared with 89%
reduction of NO obtained by injecting steam with No. 2 oil. It was not prac-
tical to try steam injection with natural gas or water injection with No. 2
oil with this particular burner design.
Flue gas recirculation resulted in large NO reductions for both natural
gas and No. 2 oil fuels (see Figures 6 and 7). The greatest decrease in NO
using the FGR technique was observed when firing natural gas (88% reduction).
282
-------
SECTION 4
COST ANALYSIS OP COMBUSTION MODIFICATIONS
INITIAL CAPITAL COSTS
Capital Costs of Steam and water Injection for Steel Furnaces
For a plant which has steam generating capability but no steam piping' to
the furnace to which the steam injection modification is to be applied, the
capital costs have been determined previously by KVB for process heater appli-
cations. For three heater sizes these costs are shown below in 1980 dollars:
2.9 MW
73.3 MW
147 MW
< 10 x 10® Btu/hr)
(250 x 106 Btu/hr)
{500 x 106 Btu/hr)
$3,500
$19,000
$32,000
Although these costs were developed for a process heater modification it
is not expected that they will differ substantially for a steel furnace modi-
fication. They involve only straightforward piping changes to get the steam
from existing headers to the furnace itself.
Flue Gas Recirculation Capital Costs for Steel Furnaces
The capital costs determined for flue gas recirculation systems for
process heaters are used here to estimate the cost of an FGR system for steel
furnaces. There are only two substantial differences between tun FGR system
for a refinery process heater and a steel furnace system:
1. A heat exchanger may be needed in a steel furnace application
in order to cool the flue gases to a temperature which can be
sent through the recirculating fan. This heat exchanger could
283
-------
act as a regenerator, increasing the efficiency of the unit and
offsetting its cost at least in part.
2. A burner plenum would not be required in a forced-draft steel
furnace, whereas it was required in a natural draft process
heater for which there was not existing forced air injection
capability.
The initial installed costs of a 20% PGR system are shown in Table IV in 1980
dollars.
ANNUAL OPERATING COSTS
Annual Operating Costs for Steam and Water Injection in Steel Furnaces
The total annual costs of the steam and water injection modifications,
including water cost, steam generation cost, additional fuel requirement cost
due to efficiency loss (calculated in the Appendix), and maintenance, are
shown in Table V for three heater sizes using a 0.005 Kg/s/burner
(40 lb/hr/burner) injection rate. One observes that the annual operating
costs of steam and water injection are, for all practical purposes, equal.
Thus, the average of the total annual costs of steam and water injection is
used here for costing both modifications.
The costs of water and steam, shown in Table V,^ include water supply and
treatment costs as well as steam generation costs (in the case of steam injec-
tion) . As shown in Table V these costs are small in comparison to the cost of
additional fuel input required by each modification.
Additional annual costs in the form of increased fuel requirements
brought about by steam and water injection must also be considered. The
additional fuel requirement is calculated in Appendix A for a subscale steel
furnace with a maximum firing rate of 0.59 MW (2 x 106 Btu/hr) and 0.005 Kg/s
(40 lb/hr) steam injection. The additional heat required is directly propor-
tional to the steam or water flow rate. The relationship is given below:
Ah = NC__, ni
S STM STM
Ah = NC m
w WATER WATER
284
-------
Where = 2387 Btu/lb and CSTM = 1275 Btu/lb, N is the number of burners
in the furnace, and Ah is the incremental heat input requirement in units of
Btu/hr. The cost increase on an annual basis may be determined, assuming an
80% use factor, as follows:
Cost h
Increase in Total Annual Fuel Cost = & x .. .. „——— x 8760 — x 0.80
Unit Heat Input y
The cost per unit heat input for typical natural gas fuel is $2.20/10® Btu
(1), the cost for No. 2 oil is $4.55/10® Btu, and the cost of No. 6 oil is
g
$3.90/10 Btu (2). The calculation of Ah is explained in Appendix A. In
Figures 8 and 9 the annual costs of steam and water injection when firing
natural gas or No. 2 oil are given as a function of 1^0 flow rate per burner.
FGR Annual Operating Costs
The electrical cost of fan operation, the incremental fuel costs, and
maintenance costs are the chief components of the annual operating costs of a
flue gas recirculation system. In the special case of steel furnaces, heat
exchanger maintenance costs would be added to those used in other applications
such as process heaters since the flue gases used would be much hotter (1366K
or 2000°F). The annual maintenance costs are estimated to be ~10% of initial
fixed capital costsi
The additional fuel costs resulting from the use of FGR are determined
for 20% FGR and 2% excess O2 in the stack. We emphasize that efficiencies
calculated here assume that the flue gas temperature change from combustion
zone to reinjection point is all due to heat loss to the external environment;
i.e., there is no regenerative capability of the PGR system.
The additional fuel costs are directly proportional to the mass flow rate
of the recirculated flue gas. Ihe relationship used to calculate those costs
is the following;
Cost
A Cost = Ah„„ x 8760 x 0.8 x ' ' „—r— r
FGR Unit Heat Input
285
-------
is calculated in Appendix A for these typical test conditions: flue gas
PGR
temperature of 558K (546°F), flue gas density of 0.673 Kg/m3 (0.042 lbm/ft3),
and specific heat of 1.089 kJ/kg-°C (0.26 Btu/lbm-°P).
The cost of electricity to operate the PGR fan has been estimated from
data obtained by KVB. The total annual costs (not annualized) taking into
account the cost of additional fuel requirements, fan electrical costs, and
maintenance cost are given in Table VI for three furnace sizes for natural gas
or No. 2 oil firing.
Again, it is emphasized that the incremental fuel costs shown here were
determined for the worst case in which nearly all of the sensible heat of the
recirculated flue gas is lost to the furnace surroundings* This situation
would probably not prevail in a practical, full-size steel furnace. However,
it is impossible to predict, with the data available at this time, how much
heat may be retained in the furnace.
TOTAL ANNUALIZED COSTS
The initial fixed capital costs of combustion modifications are
annualized making the following assumptions:
1. Straight-line depreciation of capital assets over a 12-year
life span.
2. After-tax rate of return of 15%.
3. State and federal property taxes totalling 11% of the initial
capital cost.
4. Insurance charges of 0.5% of the initial capital cost.
5. Debt/equity ratio of 0 (100% equity) for financing of initial
fixed capital costs.
6. Annual income tax rate (state and federal) of 50%.
7. Investment tax credit of 10% (applies only to the first year of
operation).
The annualized capital costs must then be added to the annual operating
costs to give the total annualized cost of combustion modifications.
286
-------
Total Annualized Costa of Water or Steam Injection
The calculation of total annual expenses and total annualized cost of the
water or the steam injection modifications to a steel furnace are shown in
Tables VII and VIII for Mo. 2 oil firing and natural gas firing.
The cost effectiveness of a combustion modification is defined as the
total annualized cost of the modification divided by the annual N0X emission
reduction potential of the modification (in thousands of Kg). The annual N0X
emission reduction potentials for steam and water injection and for flue gas
recirculation firing No. 2 oil and natural gas are given in Table IX. The
cost effectiveness of steel furnace combustion modifications for two different
fuels and three furnace sizes are given in Table X.
The annual fuel cost turns out to be the most significant item in the'
cost effectiveness calculation for steel furnaces. These costs were calcu-
lated for steel furnaces based on the annual incremental fuel requirements of
combustion modifications. Certain assumptions were made in the calculation of
those fuel requirements. They are explained along with those calculations in
Appendix A.
287
-------
SECTION 5
CONCLUSIONS
The results of the tests at the subscale steel furnace are summarized
below:
1. Large NO emission reductions were obtained when firing natural
gas and No. 2 oil by the method of injection and by the
flue gas recirculation technique.
2. Excess air variations did not affect NO emissions significantly
except at a high 0£ level, which is a less efficient mode of
operation*
3. From the standpoint of NO reduction capability, without regard
to efficiency considerations, the steam injection technique
appeared to give the best results when firing No. 2 oil, and
PGR gave the best results when firing natural gas.
4. Final calculations of the relative cost of combustion
modifications indicate that steam or water injection offers the
best NO removal capability for the least cost.
286
-------
REFERENCES
American Gas Association Quarterly Report of Gas Industry Operations,
American Gas Association, Second Quarter, 1979.
Energy User News, October 22, 1979, p. 15,
289
-------
GENERAL ARRANGEMENT
To Atmosphere
N>
VO
O
20 cm (8") Dia. Orifice
Flue Gas Recirculation
0, Measurement
~ 533K
(SOOT)
Water Cooled Probes
Combustion Air
Furnace
Temperature
TJ 1533K (2300-F)
Atomizing Air
r
Oil
Injector
l-J u
No. 2
Fuel Oil
Gas
Gas
Or
Steam
Water
Steam
-—' Test Furnace ' '
(Furnace Interior: 0.6 m (2') I.D. x
6.1 m (20*) length)
Cooler
Figure 1. Subscale steel furnace test schematic.
-------
300
(C0=140ppm)
4/4-21
Or—
4/4-23
250
4/4-22
200
4/2-3
4/2-1
CN
O
4/2-2
4->
ID
>,
ki
*0
4/9-2
150
4/10-2
4/10-1
4/10-3
(CO=131ppm)
4/9-1
§
100
Fuel: Natural Gas
T™™ * 1533K ± 89K (2300°F ± 160°F)
FURNACE
Firing Rate = 0.59 MW (2.0 x 10^ Btu/h)
50
5
4
3
2
1
0
Stack Excess Oxygen, %, dry
Figure 2. NO ©mission as a function of stack excess oxygen
for a subscale steel furnace firing natural gas.
291
-------
400
4/5-2
4/5-4
4/14-2
350
4/5-1
4/13-2
300
4/6-2
4/13-3
13-1
250
4/6-4
4/6-5
4/6-3 (CO«227ppm)
u 200
4/6-6
4/6-1
NOTE: For tests 4/6-1 to 4/6-6
tfurnace"1460±35'k
FURNACE(2I68±63.F)
150
100
Firing Rate « 0.55 MW (1.9 x 10 Btu/h)
Firing Rate ¦» 0.30 MW (1.0 x 10^ Btu/h)
0
2
3
1
4
5
Stack Excess Oxygen, », dry
Figure 3. NO emission as a function of stack excess
oxygen for a subscale steel furnace firing
No. 2 oil.
292
-------
300
Fuel: NG
Firing Rate: 2x10 Btu/hr
2* 0„ (0.59 MW)
2300°F t 50°
(1533K ± 28K)
250
FURNACE
200
150
100
50
.200 .300 .400 .500 .600
Water Mass Flow Rate/Fuel Mass Flow Rate
.700
100
Figure 4. NO emission as a function of water injection rate
for a subscale steel furnace firing natural gas.
293
-------
400
No. 2 Oil
1.8x10 Btu/hr (0.55 MW)
2% 0„
-2300BF ± 50°
(1533K ± 28K)
350
FURNACE
300
o 250
¦° 200
150
100
50
0
100
.200 .300 .400 .500
Steam Mass Flow Rate/Fuel Mass Flow Rate
600
700
Figure 5. NO emission as function of steam injection rate
for a subscale steel furnace firing No. 2 oil.
294
-------
250
200 _
a>
CM
*
ro
O
z
150 —
100 —
50
4.0% 0
2.0% 0
0.5% 0
2.0%, Water
Injection
@ 4.8 gph (5.0 g/1)
\
Fuels NG
Firing Rate: 2x10
Btu/hr. (0.59 MW)
\
5 10 15
% Flue Gas Recirculated
Figure 6. NO emission as a function of percent flue
gas recirculated for a subscale steel
furnace firing natural gas.
295
-------
300
250
200
CM
*>
n
>r
•O
150
100
No. 2 Oil; 1.9x10 Btu/hr
4* 0,
4,
2% 0,
(0.55 MW)
50
10 15
% Flue Gas Recirculated
20
25
Figure 7. NO emission as a function of percent flue gas
recirculated for a subscale steel furnace
firing No. 2 oil.
296
-------
60,000
T
T
600,000
50,000 —
Steam Injection
Water Injection
»j
FUEL; No. 2 Oil
N = Number of burners
g 40,000 —
o
oo
c\
£ 30,000
w
8
I
20,000
10,000
N=5:N=50
500,000
400,000
300,000
200,000
100,000
2.5 5.0 7.6 10.1 12.6
(20) (40) (60) (80) (100)
STEAM OR WATER MASS FLOW, g/s (lb/hr)
Figure 8. Annual additional fuel requirement cost with
steam or water in a steel furnace firing
No. 2 oil.
297
-------
30,000
T
T
25,000 —
g 20,000 -
j* 15,000
10,000 —
5,000
Steam Injection
Water Injection
FUEL: Natural Gas
N = Number of burners
N=5:N=50
N=25
10.1
12.6
(100)
STEAM OR WATER MASS FLOW, g/s (lb/hr)
Figure 9. Annual additional fuel requirement cost with
steam or water injection in a steel furnace
firing natural gas.
300,000
250,000
200,000
150,000
100,000
50,000
298
-------
TABLE I. EMISSION MEASUREMENT INSTRUMENTATION
Species
Manufacturer
Measurement Method
Model No*
Hydrocarbon
Beckman Instruments
Flame ionization
402
Carbon Monoxide
Beckman Instruments
IR Spectrometer
865
Oxygen
Teledyne
Polarographic
326A
Carbon Dioxide
Beckman Instruments
IR Spectrometer
864
Nitrogen Oxides
Thermo Electron Co.
Chemiluminescent
10A
Sulfur Dioxide
DuPont Instruments
UV Spectrometer
400
Smoke Spot
Bacharach
ASTM D2156-65
21-7006
299
-------
TABLE II
SUMMARY OF SIGNIFICANT TEST RESULTS,
SUBSCALE STEEL FURNACE BURNER
Test
Number
Fuel
Combustion
Modification
Firing
Rate
(% Cap.)
%2
NO
(ppm) *
% Reduction in NO
From
Nearest Baseline
4/3-11
NG
Water Injection
100
2.2
98
47
4/4-13
NG
FGR
100
o
•
CM
38
88
4/3-12
NG
FGR + Water Inj.
100
1.8
24
87
4/7-2
No. 2
Steam Injection
100
2.1
24
89
4/8-10
No. 2
FGR
100
2.0
57
77
*NO corrected to 3% 02» dry
TABLE III. AVERAGE BASELINE NO EMISSION,
x
SUBSCALE STEEL FURNACE BURNER
(Including All Baseline Tests at Location 4)
NO
Number
Coefficient
Fuel
ppm*
X ng/J
of Tests
of Variation
NG
222
114.6
11
0.19
No. 2
277
153.4
8
0.23
*ppm corrected to 3% O^, dry
x- . Std. deviation
TCoefficient of variation » ..
300
-------
TABLE IV. INITIAL INSTALLED COSTS (IN $) OF FLUE GAS RECIRCULATION
2.9 MW
73.3 MW
147 MW
Item/Heater Size
(10x10® Btu/h)
(250x106 Btu/h)
(500x10® Btu/h)
Fan, Motor & Drive
2,000
10,000
23,000
Damper
500
500
1,000
Ductwork & Heat
Exchanger
2,000
10,000
20,000
Duct Insulation
2,000
8,500
15,000
Instrumentation &
Control Systems
10,000
15,000
15,000
Engineering/Design
2,000
10,000
20,000
Totals
$18,500
$54,000
$94,000
301
-------
TABLE V. TOTAL ANNUAL COSTS OF STEAM AND WATER INJECTION
2.9 MW (lOxlO6 Btu/hr)
73.3 MW (250xl06 Btu/hr)
147 MW (500 x
106 Btu/hr)
Costs
No. 2 Oil
NG
No. 2 Oil
NG
No. 2 Oil
NG
Water
$ 22
$ 22
$ 600
$ 600
$ 1,100
$ 1,100
Additional Fuel
3,000
1,400
75,000
37,000
155,000
73,000
Total
$3,022
$1,422
$81,000
$37,600
$156,100
$84,000
Steam
$1,700
$1,000
$41,000
$21,000
$ 83,000
$43,000
Additional Fuel
1,500
750
40,000
20,000
80,000
39,000
Total
$3,200
$1,750
$81,000
$41,000
$163,000
$82,000
Average
$3,100
$1,600
$81,000
$39,300
$159,500
$83,000
-------
TABLE VI. ANNUAL OPERATING COSTS OF 20 PERCENT FLUE GAS RECIRCULATION
FOR A STEEL FURNACE FIRING NATURAL GAS OR NO. 2 OIL (1980 DOLLARS)
FLUE GAS TEMPERATURE = 533K (500°F)
Costs
2.9 MW (10x106 Btu/hr)
733 MW (250xl06 Btu/hr)
147 MW (500xl06 Btu/hr)
Additional Fuel (NG/No.2)
17,100/35,380
428,000/885,000
856,000/1,770,000
Fan Electricity
400
3,600
7,700
Maintenance
1,850
5,400
9,400
TOTAL (NG/No. 2)
19,350/37,630
437,000/894,000
873,100/1,787,100
-------
TABLE VII. TOTAL ANNUALIZED COSTS OF WATER OR STEAM INJECTION
Annual Operating
Cost (No. 2/Natural Gas)
$3,100/1,600
$81,000/39,300
$159,500/83,000
State and Federal
Taxes (11% of IFC)
385
2,090
3,520
Insurance (0.5% of IFC)
18
95
160
Depreciation (Straight
Line over 12 years)
290
1,585
2,667
Total Annual Expenses
(No. 2)
$3,793
$84,770
$165,847
Total Annual Expenses
(Natural Gas)
$2,293
$43,070
$ 89,347
INITIAL FIXED COSTS (IFC)
3,500
19,000
32,000
(WATER OR STEAM)
ROR=i=l5%,n=l2
Capital Recovery
Factor®.1845=CR
Annual Income
Tax Rate=50%
Investment Tax
Credit=l0%=i
(1st year onSy)
Total Annual
Capital Factor*
(ACF)=.2773
Annual Capital
Charge (=IFCxACF)
971
5,269
8,875
TOTAL ANNUALIZED COSTS (1980 DOLLARS)
No. 2 Oil
4,764
90,039
174,722
Natural Gas
3,264
48,339
98,222
*ACF - CR + T (CR- —) -
n n
where CR = capital recovery factor =
l-(l+i)"n
and T = 1.0 (for debt/equity ratio of O)
304
-------
TABLE VIII. TOTAL ANNUALIZED COSTS OP FGR
1 ¦¦ 1 —1
Annual Operating Costs
(No. 2/NG) -
37,630/19,350
894,000/437,000
1,787,100/873,100
State and Federal Taxes
(11% of IFC)
2,035
5,940
10,340
Insurance
(0.5% of IFC)
100
270
470
Depreciation (Straight
Line over 12 Years)
1,540
4,500
7,830
TOTAL ANNUAL EXPENSES
(No. 2)
41,305
904,710
2,678,840
TOTAL ANNUAL EXPENSES
(NG)
23,025
447,710
891,740
INITIAL FIXED COSTS
(ROR =i=15%,n=12
Capital Recovery
Factor®.1845
Annual Income Tax
Rate=t=50%
Investment Tax
credit«=i =10%
(1st yearconly)
18,500
54,000
94,000
Total Annual
Capital Factor
=.2773
Annual Capital
Charge
5,131
14,976
26,069
TOTAL ANNUALIZED COSTS (1980 DOLLARS)
No. 2
46,436
919,686
2,704,909
NG
28,156
462,686
917,809
305
-------
TABLE IX. BASELINE NC>x EMISSIONS FROM A STEEL FURNACE
Modification
Fuel
Heat Input
MW
NO Concentration
ng/J
Annual Emission
103Kg NO
f - ¦ ¦
Reduction
Percent
Annual
Reduction
103 Kg NO
Steam
No. 2
2.93
153.4
11.3
89
10.1
Steam
No. 2
73.2
153.4
283
89
252
Steam
No. 2
147
153.4
568
89
506
Water
NG
2.93
114.6
8.5
47
4.0
Water
NG
73.2
114.6
211
47
99
Water
NG
147
114.6
425
47
200
FGR
No.2
2.93
153.4
11.3
77
8.7
PGR
No.2
73.2
153.4
283
77
218
FGR
No.2
147
153.4
568
77
437
FGR
NG
2.93
114.6
8.5
88
in
•
FGR
NG
73.2
114.6
211
88
186
FGR
NG
147
114.6
425
88
374
Annual NO Emission = ——^ = 0.0252 x MB x ^
xv J
-------
TABLE X. COST EFFECTIVENESS QF COMBUSTION MODIFICATIONS
ON A STEEL FURNACE ($/10 Kg OF NOx REDUCTION)
INCLUDING ANNUAL INCREMENTAL FUEL COSTS
Modification
2.9 MW (lOxlO6 Btu/hr)
Furnace Heat Input
73.3 MW (250xl06 Btu/hr)
-"it " r ttsaasasBaeaaBag.tgc aaaaeBasBasasagaa
147 MW (500xl06 Btu/hr)
STEAM INJECTION
No. 2 Oil
472
357
345
NG
323
192
194
WATER INJECTION
No. 2 Oil
1,191
909
874
NG
816
488
491
FLUE GAS RECIRCUL
ATION
No. 2 Oil
5,337
4,219
6,190
NG
3,754
2,488
2,454
-------
APPENDIX A
CALCULATION OP INCREMENTAL FUEL REQUIREMENTS OF
COMBUSTION MODIFICATIONS TO A STEEL FURNACE
The calculation of the incremental heat requirement of steam or water
injection or flue gas recirculation when applied to a steel furnace necessi-
tates the assumption that these modifications have no effect on furnace
thermal efficiency other than the additional thermal losses caused by having
to heat the injected materials to combustion temperatures. Thus, the effects
of the combustion modifications on convective or radiative heat transfer
rates, which also affect furnace efficiency, are not considered in this
report.
In an actual application, the convective heat transfer rates from the
combustion gases to the steel itself will probably increase because of the
higher mass flow of gases through the furnace brought about by the injection
of additional material. This may partially offset the efficiency degradation
associated with the added thermal load of the injected material.
In addition, the radiative heat transfer rate may also be increased by
flue gas recirculation or by steam or water injection. The increase in the
partial pressure of COj and t^O resulting from the injection of these mate-
rials would tend to increase the emissivity of the combustion gases and,
therefore, the radiative heat transfer rate to the steel.
The effects of altered convection and radiative heat transfer need to be
studied further in order to develop more meaningful efficiency assessments of
the combustion modifications discussed in this section.
The incremental heat requirements for steel furnace combustion
modifications are calculated in the order of increasing complexity, beginning
308
-------
with steam injection, followed by water injection and, finally, flue gas
recirculation. The percent increase in heat load is assumed equal to the
percent increase in fuel required.
The incremental heat requirement per burner of steam injection at 5.0 g/s
(40 lb/hr) injected steam flow rate is defined as follows:
~ heat required to take steam from the injection conditions
to the furnace bulk gas temperature, TF
= m (h - h )
STM F IN
For PIN = 1 atm, TIN=273K, PF=1 atm, and Tp=1755K (typical measured
operating temperature)
hf = 3384 kJ/kg (1,455 Btu/lb) and
hIN = 419 Wkg (180 Btu/lb)
Thus,
Ah = 0.015 MW (51,000 Btu/hr)
STM
= 2.56% of experimental burner capacity
of 0.586 MW (2 x 106 Btu/hr)
The incremental heat requirement per burner for 5.0 g/s (40 lb/hr)
injected water flow rate is determined below. This requirement includes the
heat necessary to raise the water temperature to the boiling point, complete
the phase change to steam, and heat the steam to the furnace bulk gas
temperature.
^WATER ~ ^STM + ™&20 J^fg + ^H^U) (TB.P. " TIN*]
Where
Ah 88 heat of vaporization of water at p 38 1 atm,
T - 373K (212°F) = 2256 kJ/kg (970 Btu/lb)
CpH ^ - specific heat of water
309
-------
Tb>p = 373K (212°F) = boiling point of water at p=1 atm
Tjjj = injected water temperature
For Tin = 294K (70°F),
Ah, = 0.028 MW (95,480 Btu/hr)
WATER
= 4.75% of experimental burner capacity
The heat losses associated with flue gas recirculation in a steel furnace
arise from the cooling of the flue gas temperature to the reinjection
temperature. In the experimental arrangement most of this cooling occurred in
an air-gas heat exchanger, and no heat was recovered; i.e., all of the heat
was lost to the ambient air. It is emphasized here that in a practical
application of flue gas recirculation much of this heat could be retained
within the furnace proper by combustion air preheat or some other means of
waste heat recovery. The following diagram illustrates the calculation of
furnace efficiency for the steel furnace.
L
FURNACE
Tp-1755KU700°F)
¦OUT' RADIATION
310
-------
One observes from this drawing the increased heat loss from the steel
furnace with FGR as compared to a furnace without FGR. In the calculations
which follow we assume that there are no factors influencing furnace
efficiency other than this heat loss. This assumption may not be strictly
valid, however, for two reasons:
1. The convective heat transfer coefficient of the combustion
gases in the furnace should increase with the increased mass
flow through the furnace due to FGR, thereby increasing the
convective heat transfer to the steel.
2. The recirculation of flue gases containing large amounts of the
radiative species C02 and may increase the emissivity of
the combustion gases within the furnace, thus increasing
radiative heat transfer rates to the steel. (This is also
especially true for the case of 1^0 injection where the volume
fraction of HjO in the combustion gases is significantly
increased.)
The incremental heat (fuel) requirement for 20% FGR is determined below:
^FGR ~ "VgR^^FGR^F^FGR*
where (Cp)FQR = specific heat of the flue gas
tFGR = flue 9as temperature at point of injection
into the furnace
m„__ = recirculated flue gas mass flow rate
FGR
For Test #4/4-13,
Tf = 1755K (2700°F), TpGR = 559K (546°F)
(cp)FGR - 1*09 kJ/Kg-K (0.26 Btu/lbm-°R),
and m = 0.051 kg/s (400.8 lbm/hr)
FGR
and so
Ah - 0.066 MW (224,464 Btu/hr)
FGR
= 11.2% of burner heat input capacity
Thus, with no waste heat recovery, and neglecting the effects of the
combustion modifications on convective and radiative heat transfer, the addi-
tional fuel requirements are summarized in Table A-1.
311
-------
TABLE A-1. INCREMENTAL FUEL REQUIREMENTS OP
COMBUSTION MODIFICATIONS TO A STEEL FURNACE
Percent Increase In
Modification
Fuel Consumption
Steam Injection
2.56
Hater Injection
4.75
Flue Gas Recirculation
11.20
312
-------
TECHNICAL REPORT DATA
(Please read JnUnictions on the reverse before completing)
1 , REPORT NO.
2.
3. RECIPIENT'S ACCESSION NO.
A. TITLE AND SUBTITLE
Proceedings of the Joint Symposium on Stationary
Combustion NOx Control. Vol. 3. NOx Control and
Environmental Assessment of Industrial Process*
5. REPORT DATE
6. PERFORMING ORGANIZATION CODE
7. AUTHOR(S)
Symposium Cochairmen: Robert E. Hall (EPA) and
J.E. Cichanowicz (EPRI)
8. PERFORMING ORGANIZATION REPORT NO.
IERL-RTP-1085
9. PERFORMING ORGANIZATION NAME AND ADDRESS
See Block 12.
10. PROGRAM ELEMENT NO.
EHE624
H i. CONtRACT/GRANT NO.
NA (Inhouse)
12. SPONSORING AGENCY NAME AND ADDRESS
EPA, Office of Research and Development
Industrial Environmental Research Laboratory
Research Triangle Park, NC 27711
13. TYPE OF REPORT AND PERIOD COVERED
Proceedings; 10/6-9/80
14. SPONSORING AGENCY CODE
EPA/600/13
15.supplementary notes ePA-600/7-79-050a through -050e describe the previous sympo-
sium. (*) Equipment, Engines, and Small Stationary Sources.
i6. abstract The proceecftngS document the approximately 50 presentations made during
the symposium, October 6-9, 1980, in Denver, CO. The symposium was sponsored
by the Combustion Research Branch of EPA's Industrial Environmental Research
Laboratory, Research Triangle Park, NC, and the Electric Power Research Institute
(EPRI), Palo Alto, CA. Main topics included utility boiler field tests; NOx flue gas
treatment; advanced combustion processes; environmental assessments; industrial,
commercial, and residential combustion sources; and fundamental combustion re-
search. This volume relates to NOx control and environmental assessment of indus-
trial process equipment, engines, and small stationary sources.
17. KEY WORDS AND DOCUMENT ANALYSIS
a. DESCRIPTORS
b.IDENTIFIERS/OPEN ENDED TERMS
c. COSATI Field/Group
Pollution Flue Gases
Combustion Engines
Nitrogen Oxides
Boilers
Tests
Assessments
Pollution Control
Stationary Sources
Environmental Assess-
ment
13B
2IB 2 IK
07B
13A
14B
18. DISTRIBUTION STATEMENT
Release to Public
IB. SECURITY CLASS (ThisReport)
Unclassified
21. NO. OF PAGES
317
20. SECURITY CLASS (This page)
Unclassified
22. PRICE
CPA Form 2220-1 (»-73) jj
L3
------- |