xvEPA
United States
Environmental Protection
Agency
Industrial Environmental Research
Laboratory
Research Triangle Park NC 27711
EPA-600/7-79-050C
February 1979
Proceedings of the Third
Stationary Source
Combustion Symposium;
Volume III
Stationary Engine and
Industrial Process
Combustion Systems
Interagency
Energy/Environment
R&D Program Report
-------
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 m?:.;imum interface in related fields.
The nine series are:
1. Environmental Health Effotts Research
2. Environmental Projection Technology
3. Ecological Rese ch
4. Environm*1 ;,.«ti Monitoring
5. SOCKX jonomic Environmental Studies
35. 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 INTERAGENCY ENERGY-ENVIRONMENT
RESEARCH AND DEVELOPMENT series. Reports in this series result from the
effort funded under the 17-agency Federal Energy/Environment Research and
Development Program. These studies relate to EPA's mission to protect the public
health and welfare from adverse effects of pollutants associated with energy sys-
tems. The goal of the Program is to assure the rapid development of domestic
energy supplies in an environmentally-compatible manner by providing the nec-
essary environmental data and control technology. Investigations include analy-
ses of the transport of energy-related pollutants and their health and ecological
effects; assessments of, and development of, control technologies for energy
systems; and integrated assessments of a wide-range of energy-related environ-
mental issues.
EPA REVIEW NOTICE
This report has been reviewed by the participating Federal Agencies, and approved
for publication. Approval does not signify that the contents necessarily reflect
the views and policies of the Government, nor does mention of trade names or
commercial products constitute endorsement or recommendation for use.
This document is available to the public through the National Technical Informa-
tion Service, Springfield, Virginia 22161.
-------
EPA-600/7-79-050C
February 1979
Proceedings of the Third
Stationary Source Combustion
Symposium;
Volume 111. Stationary Engine and
Industrial Process Combustion
Systems
Joshua S. Bowen, Symposium Chairman,
and
Robert E. Hall, Symposium Vice-chairman
Environmental Protection Agency
Office of Research and Development
Industrial Environmental Research Laboratory
Research Triangle Park, North Carolina 27711
Program Element No. EHE624
Prepared for
U.S. ENVIRONMENTAL PROTECTION AGENCY
Office of Research and Development
Washington, DC 20460
-------
PREFACE
These proceedings document more than 50 presentations and discussions
presented at the Third Symposium on Stationary Source Combustion held March
5-8, 1979 at the Sheraton Palace Hotel, San Francisco, California. Sponsored
by the Combustion Research Branch of the EPA's Industrial Environmental
Research Laboratory - Research Triangle Park, the symposium papers emphasized
recent results in the area of combustion modification for NOX control. In
addition, selected papers were also solicited on alternative methods for
NOX control, on environmental assessment, and on the impact of NOX control
on other pollutants.
Dr. Joshua S. Bowen, Chief, Combustion Research Branch, was Symposium
Chairman; Robert E. Hall, Combustion Research Branch, was Symposium Vice-
Chairman and Project Officer. The welcoming address was delivered by Clyde
B. Eller, Director, Enforcement Division, U.S. EPA, Region IX and the opening
Address was delivered by Dr. Norbert A. Jaworski, Deputy Director of IERL-RTP.
The symposium consisted of seven sessions:
Session I: Small Industrial, Commercial and Residential Systems
Robert E. Hall, Session Chairman
Session II: Utilities and Large Industrial Boilers
David G. Lachapelle, Session Chairman
Session III: Advanced Processes
G. Blair Martin, Session Chairman
Session IV:
Session V:
Session VI:
Session VII:
Special Topics
Joshua S. Bowen, Session Chairman
Stationary Engines and Industrial Process Combustion
Systems
John H. Wasser, Session Chairman
Fundamental Combustion Research
W. Steven Lanier, Session Chairman
Environmental Assessment
Wade H. Ponder, Session Chairman
11
-------
VOLUME III
Table of Contents
Session V: Stationary Engines and Industrial Process
Combustion Systems
(*) See Volume V.
Page
"Application of Advanced Combustion Modifications to Industrial
Process Equipment Process Heater Subscale Tests," S. C. Hunter,
R. J. Tidona, W. A. Carter and H. G. Buening 3
"Pollutant Emissions From "Dirty1 Low and Medium - Btu Gases,"
R. T. Waibel, E. S. Fleming and D. H. Larson 37
"Some Aspects of Afterburner Performance For Control of Organic
Emissions," R. E. Barrett and R. H. Barnes 79
"Development of Emission-Control Methods For Large-Bore
Stationary Engines," R. P. Wilson 95
"Low NO Combustor Development For Stationary Gas Turbine
Engines," R. M. Pierce, C. E. Smith and B. S. Hinton 137
"A Research Plan to Study Emissions From Small Internal
Combustion Engines," J. W. Murrell and F. Alexander (Abstract)* - 175
ill
-------
SESSION V
STATIONARY ENGINES AND INDUSTRIAL PROCESS
COMBUSTION SYSTEMS
JOHN H. WASSER
SESSION CHAIRMAN
-------
APPLICATION OF ADVANCED COMBUSTION MODIFICATIONS TO
INDUSTRIAL PROCESS EQUIPMENT
PROCESS HEATER SUBSCALE TESTS
By:
S. C. Hunter, R. J. Tidona,
W. A. Carter and H. J. Buening
KVB, Inc.
A Research-Cottrell Company
Tustin, California 92680
-------
ABSTRACT
This is a report of a research program to develop combustion modifica-
tion technology as means of emissions reduction and thermal efficiency improve-
ment on industrial process equipment. The work is an extension to EPA Contract
68-02-2144 which concentrated on operational adjustments. Presented are
results of subscale tests for petroleum refinery heaters.
Subscale tests with natural draft process heater burners firing natural
gas, No. 2 oil, No. 6 oil and shale oil included standard burners, two commer-
cial low NO designs, staged combustion, flue gas recirculation, steam injec-
X
tion, and altered fuel injection. The most effective of these for reducing
NO was staged combustion; reduction obtained with natural gas was 67% from
X
a baseline of 61 ng/J (120 ppm, 3% O , dry) and with No. 6 oil (0.3% N^) was
51% from a baseline of 172 ng/J (307 ppm, 3% 02/ dry). The costs of applying
all modifications were evaluated; stage combustion appears to be the most cost
effective for large heaters while commercial low-NO burners are more cost
X
effective for small heaters burning'No. 6 oil.
These subscale test results are preparatory to full scale testing and
should not be interpreted as achievable technology until full scale test
demonstration is completed.
This report was submitted in fulfillment on Contract 68-02-2645 by
KVB, Inc. under the sponsorship of the U.S. Environmental Protection Agency.
This report covers the period August 5, 1977 to October 30, 1978.
-------
SECTION 1
INTRODUCTION
At the Second EPA Stationary Source Combustion Symposium, KVB reported
on work to investigate the feasibility of applying combustion modifications,
developed for boilers, to other industrial process equipment (Ref. 4). Results
of that program were mixed. Some units responded well with up to 69% "reduction
in NO . For other types, equipment design and process constraints limited NO
X X
reduction to insignificant or negligible amounts. This paper presents additional
work directed to understand and overcome these limitations.
OBJECTIVE AND SCOPE
The objective of this program is to develop advanced combustion modi-
fication concepts requiring minor hardware modifications that could be used by
operators and/or manufacturers of selected industrial process equipment to
control emissions. The development is to be performed for equipment in which
the modifications will be most widely applicable and of the roost significance
in mitigating the impact of stationary source emissions on the environment.
The program involves investigation not only of emissions but also multimedia
impacts and control cost effectiveness.
The program involves both subscale and full scale testing. Subscale
testing is a necessary part of development of new hardware to ensure acceptable
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 the equipment manufacturers and process
industry can employ an emission control technology.
At the conclusion of the study, a final engineering report will be pre-
pared summarizing the accomplishments of the subscale and full scale demonstra-
tion tests. A series of guideline manuals will be prepared to acquaint the
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.
-------
PRELIMINARY SURVEY
The initial task was to review existing source inventories and update
them where possible to more clearly define those processes where controls will
be of maximum benefit. Activity in the preliminary survey task concentrated
on the review of processes to be emphasized in the test program. The review
of source emission NO data and total annual heat input provided the following
X
relative ranking of industrial processes as most important in potential
environmental impact:
NOX Emissions Heat Input
Process Gg/y103 tons/y 1015 J/y10IZ Btu/y
1. Cement kilns 704 776 513 486
2. Wood-bark boilers 141 157 915 873
3. Refinery process heaters 122 134 1580 1500
4. Glass container furnaces 39 43 105 99
5. Steel soaking pits and
reheat furnaces 29 32 538 510
Table I presents additional key information used to develop comparative
data on each process.
Of the five processes considered, four were identified as candidates
for combustion modification. Glass furnaces were excluded because of a lack
of flexibility in the combustion systems.
TEST PROGRAM STATUS
Subscale test work has now been completed on a vertically-fired
rectangular process heater research furnace and a small research dry process
rotary cement kiln. The process heater subscale tests were conducted in
conjunction with a major process heater burner manufacturer. The research kiln
tests were conducted with the help of a major cement industry association.
Instrumentation used is discussed in Section 2.
In chronological order, the modifications tested in the subscale proc-
ess heater were the following:
1. Lowered excess air
2. Commercial low-NOx burners (two designs)
3. Steam injection
4. Staged combustion (two methods)
5. Flue gas recirculation
6. Modified fuel injection
-------
Each of these concepts is defined in Section 3 of this paper. A summary
of test results for each modification is also given in that section. An
analysis of the cost effectiveness of each modification is presented in
Section 4.
A summary of the results of the tests on the process heater is presented
in Table II. This table shows that relatively minor hardware modifications may
be quite effective in reducing NO in process heaters (e.g. reductions in NO
X X
of 50-60% or more were obtained using fairly simple staging techniques as
compared to a maximum 63% reduction for the more complicated flue gas recircula-
tion technique).
Tests on the research kiln are not expected to be representative of
producing cement kilns, however, preliminary indications are that fly ash
injection into the flame zone is effective in reducing NO . Reductions of 28%
in NO emissions were achieved with fly ash injection. A basic objective in
A
these tests was to evaluate the effect of the modifications on cement quality.
Analyses of cement samples were not yet available for discussion in this paper.
Therefore, the results of those tests will be presented in a later paper and
the final report.
-------
SECTION 2
INSTRUMENTATION
The process heater emission measurements were made with instrumentation
carried in a 32 ft x 8 ft mobile laboratory which was described in detail in
the EPA Report, "Application of Combustion Modification to Industrial Combus-
tion Equipment," Contract No. 68-02-2144.
Gaseous species measurements were made with analyzers located in the
trailer. Particulate emission and size measurements were not made during sub-
scale tests to allow larger range of test variables for effect on gaseous emis-
sions. These measurements will be made on full scale units. The emission
measurement instrumentation used is the following:
TABLE III. EMISSION MEASUREMENT INSTRUMENTATION
Model
Species Manufacturer Measurement Method No.
Hydrocarbon Beckman Instruments Flame lonization 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
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.)
8
-------
nylon line. The positive displacement diaphragm sample pumps provide unseated
sample gas to the refrigerated condenser (to reduce the dew point to 35 °F},
a rotameter with flow control valve, and to the O , NO, CO, and CO instru-
mentation. 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 NO , SO and
hydrocarbons, the sample must be kept above its dew point, since heavy hydro-
carbons may be condensible and SO and NO 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 line, electrically traced and thermally insulated to maintain a sample
temperature of up to 400 °F. Metal bellows pumps provide sample to the hydro-
carbon, S0_ and NO continuous analyzers.
£ 3C
-------
SECTION 3
COMBUSTION MODIFICATIONS ON A SUBSCALE PROCESS HEATER
EQUIPMENT CHARACTERISTICS
The process heater subscale testing was conducted in the research
furnace of a major manufacturer of process heater burners. The furnace was
a refractory lined, uncooled rectangular box type furnace 2.4 m (8 ft) wide
by 1.8 m (6 ft) deep by 9.8 m (32 ft) high.
Natural draft burners were installed in the furnace floor firing
vertically upward. The nominal firing rate for the tests was 1.5 MW (5x10
Btu/hr). Furnace draft was controlled manually with a damper in the stack.
View ports for observing flame shape were provided.
The furnace had the capability of firing either oil or natural gas.
Both flows were measured with flow meters. Thermocouples were installed in
the side of the furnace to measure the vertical thermal gradient and the
temperature-time behavior during furnace heat-up.
BURNER COLD FLOW TESTS
A cold flow burner model with the same horizontal dimensions as the
natural draft heater was fabricated at the KVB laboratory. The purpose of the
cold flow tests was to evaluate the fuel and air mixing to provide insight into
fuel injection modifications which could lead to lowered NO emissions when
firing natural gas.
The cold flow model used an induced draft fan to obtain the same air
flow and velocity as that of the actual natural draft burner. The natural
gas was simulated by gaseous CO . Measurements of CO concentration were made
in two axes across the firebox at three axial planes. Contour maps of constant
concentration were prepared to compare the mixing characteristics of the
different gas tips. After the mixing model was prepared, modifications to the
gas tips were made and tested. The gas fuel was simulated by CO and the con-
centration measured with an NDIR CO analyzer. Actual burner fuel/air ratio
10
-------
and the cold flow simulation are realted by the following expression
(F/ft
ner sm. Burner
Tests were conducted with three types of standard gas tips, each pro-
viding a different degree of swirl, supplied by the burner manufacturer. Very
little difference in mixing patterns was found among the three standard types
of gas tips. All produced substantial mixing of the fuel gas and air within a
short distance of the injection plane.
Several modifications to the fuel injection geometry were evaluated.
Modifications which looked promising from the cold flow model were then evaluated
in the hot firing tests in the furnace.
The modifications tested were the following:
1. Turning the gas tips so that the center gas port was aimed radially
outward such that the gas stream 'impinged upon the 41 cm (16 in. )
diameter cylindrical sleeve.
2. Placing a 20 cm (8 in.) diameter 'staging' cylinder with its vertical
centerline coinciding with that of the burner into the flow such
that roughly 25%- 30% of the 'combustion* air flow was introduced
through the cylinder. Two cylinders of different length were used
in separate tests. In one case, the top of the cylinder was 5.4 cm
(2-1/8 in.) above the gas tips. In the second case, the cylinder
top was 30.5 cm (12 in.) above the gas tips.
3. Placing a 7.6 cm (3 in.) wide, 15.2 cm (6 in.) long deflector
upstream of each of the gas nozzles inclined at a 45 degree
angle from vertical and extending from the 'burner' sleeve to
the plane of the gas tip orifices.
All of these modifications were expected to delay mixing of fuel and
air, thereby lengthening the flame in a hot-firing application, lowering peak
temperatures and residence times at high temperatures and, thus, lowering NO
emissions. The concentration of the cold flow test gas (CO ) was again measured
at various radial positions at three heights above the gas injection plane.
The results for the four different test cases at 36 cm (14 in.) above the gas
nozzles are shown in Figure 1 . These curves indicate that the concentric
'staging' cylinder and the radially-outward- facing injection orifices produce
a significant delay in the mixing of the test gr~ and air. The mixing pattern
with the deflectors in place did not vary appreciably from the patterns
obtained for the standard nozzles without modification.
11
-------
HOT FIRING TEST RESULTS
Tests were conducted to evaluate the effect of combustion modifications
on emissions from a natural draft process heater. The reduction in NO emis-
sions and the change in efficiency were evaluated for: (1) lowered excess air,
(2) staged combustion air, (3) low NO burners (tertiary air injection and
X
recirculating tile designs), (4) flue gas recirculation, (5) steam injection
and (6) altered fuel injection geometry. The tests were conducted with natural
gas and No. 6 oil. Only burner baseline measurements were made with No. 2 oil
because of limited furnace availability due to the manufacturer's work schedule.
Baseline Tests
Tests were conducted with each burner prior to implementing any combus-
tion modification. These baseline measurements were made with the burner firing
natural gas, No. 6 oil (0.3% N) and No. 2 oil (0.01% N) . A summary of baseline
gaseous emissions data is presented in Table IV.
Lowered Excess Air (LEA)
The NO emissions from the conventional MA-16 and DBA-16 burners as
x
well as the low-NO (tertiary air) burner for various test conditions including
X.
baseline are graphed as a function of stack excess oxygen for natural gas and
No. 5 oil firing in Figures 2 and 3. The unmodified burners are represented
by heavy curves and the results of the modifications are shown as light curves.
The low NO burner (tertiary air injection) exhibited the lowest level of NO
X X
at the nominal 3% O condition and showed the most dependence on O level
firing natural gas. The NO emissions at 2.7% O was 100 ppm* which dropped
X £
off to 76 ppm at 2.1% O .
The effect of excess O on NO for No. 6 oil firing with a 40 degree
£ X
spray angle is shown in Figure 3 for the MA-16 and the tertiary air natural
draft burners. The spray angle is defined as the total included angle of the
conical jet produced by the oil gun. The minimum value for NO for the MA-16
X
was 282 ppm at 0.75% O . The minimum NO emission for the tertiary air burner
2. x
firing No. 6 oil was 235 ppm at 0.5% 0 .
*A11 concentrations in this report are corrected to 3% excess O_ on a dry basis.
12
-------
Staged Combustion Air (SCA)
Staged combustion is a technique for emissions reduction wherein a
portion of the flame zone is operated fuel-rich and secondary air is injected
subsequently to bring the overall air-fuel ratio to the desired level to assure
complete combustion. Staged combustion has been shown to be an effective
method of NO reduction in other applications. In order to develop staged
X
combustion in a natural draft heater, two techniques were evaluated. In the
first method, four staged air lances were inserted through the furnace floor
positioned 90 degrees apart outside the burner tile on a diameter of 61 cm
(24 in.). This modification is shown schematically in Figure 4. The staged
air lances were made from 3.2 cm (1-1/4 in.) diameter stainless steel pipe with
an orifice plate of 3.0 cm (1-3/16 in.) diameter on the end. The end of the
lance was angled 45 degrees inward to provide better penetration of the flame
by the secondary air. Adjustment of the insertion depth up to 1.52 m (5 ft)
was provided by a locking collar outside the furnace floor.
Nine tests with natural gas fuel and ten with No. 6 fuel oil were
conducted to evaluate the effect of staged combustion on NO emissions and
burner performance. The first tests with natural gas consisted of varying the
injection depth for the staged air. These tests showed no significant reduc-
tion in NO with injection heights greater than 1.22 m (4 ft) approximately.
NO emissions increased significantly as the burner air-to-fuel ratio
(i.e. the A/F of the fuel-rich zone) was increased when firing natural gas.
This trend is shown in Figure 5 as well as the dependence of NO emissions on
overall excess oxygen. The tests at low overall excess O reduced NO emissions
as much as 67% lower than the conventional burner baseline emissions. At
normal O_, NO emissions were 46% below the baseline value.
The staged combustion modification was also evaluated with the burner
firing No. 6 fuel oil. The staged air lances were kept at a fixed insertion
depth of 1.22 m (4 ft) for all tests with oil. A similar effect of burner
on NO emissions was observed for the tests with No. 6 oil as is shown in
x
Figure 6. At the normal O_ condition, a reduction of 35% was achieved. With
the unit operating in the low O mode a reduction of 51% from the baseline
condition was achieved.
13
-------
An alternative method of producing staged combustion was developed
from the cold flow tests described in an earlier section. This technique
employed a central cylinder which introduced the secondary air into the flame
zone after the primary combustion zone. This modification was tested firing
only natural gas.
For this modification the orifice plate was removed from a DBA-16
burner (a conventional burner differing only in tile design from the MA-16
burner) and a 19.1 cm (7-1/2 in.) I.D., 21.6 cm (8-1/2 in.) O.D. cylinder
placed in the burner such that its longitudinal axis coincided with the ver-
tical centerline of the burner. The bottom of the cylinder rested on the base
of the secondary air section of the burner. Thus, all of the primary air flow
(approximately 1/3 of the total air flow) was routed through the cylinder and
the rest through the secondary air registers.
Several different cylinder lengths were tried in this series of tests.
The height of the top of the cylinder above the gas tips was varied from 7.6 cm
(3 in.) to 109 cm (43 in.). The data showed that an optimum height above the
gas tips lies between 23 cm (9 in.) and 94 cm (37 in.). The lowest NO emis-
sion (88 ppm) occurred at 94 cm (37 in.) above the gas tips.
At a cylinder height of 109 cm (43 in.) , excess oxygen was varied from
4.9% to the CO limit of 0.5% (with a CO concentration of 439 ppm). At 1.2%
excess O», the NO concentration was 66 ppm, a reduction of 50% from the
£t A
corresponding O point for the standard DBA-16 burner. At the CO limit, the
NO level dropped to 54 ppm for a reduction of 59% from the CO limit N0x concen-
tration emitted by the standard burner.
Low NO Burner (Tertiary Air Injection)
Tests were made on a low NO burner similar to the conventional MA-16
X
burner but incorporating a tertiary air register above the primary and secondary
air registers. Figure 7 is a schematic of this burner.
Baseline NO measurements firing natural gas were about 100 ppm. Excess
X
O was varied from 4.1% down to 2.1%. CO concentration at the minimum O was
47 ppm. NO at that O setting was 76 ppm, a reduction of 24% from the
X ^
tertiary air burner baseline concentration and 30% below the average MA-rl6
burner baseline emission. The results of these tests are shown in Figure 2.
14
-------
Firing rate changes were also made firing natural gas. At 100% of
capacity (6.5x10 Btu/hr) NO emission was 155 ppm and dropped to 109 ppm at
37% of capacity. A series of air register adjustments were made at approximately
3% O with the tertiary air burner but produced no appreciable reduction in NO
£ X.
levels.
The effect of furnace temperature on NO emissions for natural gas fuel
was also investigated. The NO level tended to rise until a stack temperature
of about 1200 K (1700 °F) was attained. Since many tests were conducted with
stack temperatures less than 1200 K due to the length of time required for
furnace heat-up (about 4 hours) some temperature-related effects were unavoid-
able in the data. However, the effects were fairly small and were also minimized
where possible by conducting a related series of tests (e.g. different excess
O points) over the shortest time possible and making baseline checks periodically
during the day. There was no large temperature effect on NO emissions firing
No. 6 oil.
Tests on No. 6 oil with the tertiary air burner consisted of excess
O variation and air register adjustments. An oil tip with a 40 deg. spray
angle was used for all of the tests with No. 6 oil. The effect of excess 0
on NO emissions for the tertiary air burner using No. 6 oil is shown in
Figure 3. The curve is fairly flat showing baseline NO emissions to be 272
X
ppia dropping to 235 ppm at 0.5% O?f for a reduction of 14%. The CO level at
0.5% O was 57 ppm. These baseline NO values were approximately 15% less
^ X
than the baseline NO emissions from the MA-16 burner with a 40 deg. spray
X
angle tip.
The air register setting which produced the lowest NO had the primary
X
air register 10% open and the secondary and tertiary air registers 100% open.
At 2.9% O , NO emission at this register setting was 200 ppm, a reduction of
£* X
26% from tertiary air burner baseline or 42% from the average MA-16 burner
baseline. The variation of excess 0 at this register setting could not be
completed because of a severe coking problem due to the oil tip being placed
1/4-inch too low in the burner throat.
15
-------
A few tests on the tertiary air burner were also conducted using a
shale oil of high nitrogen content (2.1% by weight). Excess O changes coupled
with relatively minor register adjustments were made. NO emissions varied from
526 ppm at 6.5% 0 to 200 ppm at 0.35% O . The CO concentration at the latter
<- £.
O2 was greater than 2000 ppm. At an excess 0 of 1.2%, NO emission was 295 ppm,
or 33% less than the emission at 3.2% O (439 ppm). Also, at the latter point
all registers were 50% open, increasing the draft and lengthening the flame.
With the primary air register nearly closed, the secondary air register 25%
open, and the tertiary air register 100% open, the NO at 3% O was 329 ppm;
25% less than the 439 ppm measured at 3.2% O with all registers 50% open.
£i
Low NO Burner (Recirculating Tile)
A low NO burner incorporating a self-recirculating tile was evaluated
X
in the research furnace. A special tile was used to achieve some recirculation
of fuel vapors and the products of combustion in the immediate vicinity of the
burner. The recirculation of these gases is intended to lower the flame zone
temperature and, thus, lower thermal NO . The NO level firing natural gas
X X
had a baseline value of 104 ppm (corrected to 3% O , dry) , only slightly lower
than the baseline NO found on the conventional MA-16 burner. Excess O was
X &,
varied from 4.4% down to 0.6%, at which point CO appeared; NO emission decreased
by 20% to 83 ppm. The CO concentration at 0.6% O was 44 ppm.
Baseline NO emissions (at 4% excess O_) firing No. 2 oil were 110 ppm.
X ^
Excess O was varied from 5.1% down to 0.5% (CO was 147 ppm at 0.5% O_) . The
lowest NO emission occurred at an excess O of 1.4% and was 98 ppm, down 11%
X £.
from the baseline value. Further reduction of the excess O_ appeared to have
no significant result on NO emissions.
X
Flue Gas Recirculation (FGR)
Flue gas recirculation has been demonstrated to be an effective method
of NO reduction for industrial boilers. Until these tests, flue gas recircula-
Ji
tion had not been applied to process heaters for NO reduction. It was not
X
possible to duct actual flue gases from the stack to the burner because a high
temperature fan was not available. In order to simulate FGR, an auxiliary
burner was installed which exhausted into a combustion air duct leading to the
burner plenum. A gas-gas heat exchanger was installed to control the tempera-
ture of the combustion air-flue gas mixture. The percentage of recirculated
16
-------
flue gas was varied by adjusting the firing rate for the auxiliary burner.
Flue gas recirculation rates were varied up to a maximum of approximately 40%
when firing natural gas and No. 6 oil.
The effect of FGR rate firing natural gas is shown in Figure 8. The
flue gas recirculation is defined by the following expression:
_ Recirc. mass flow rate x 100
^ r \ji\
Combustion air flow + Recirc. mass flow -f Fuel flow
A reduction in NO of 59% from the baseline condition was achieved with
X
approximately 40% FGR at the normal O level Cv 3%) . The overall 0 level was
reduced until the CO limit was reached. This limiting value of excess 0 was
0.7% 0 . A reduction in NO of 63% was measured with the combination of 40%
£ A.
FGR and low O operation.
Figure 9 presents NO as a function of FGR rate for No. 6 oil firing.
FGR alone resulted in a reduction of 31% at the maximum recirculation rate
C\> 40%) . The combination of FGR and low O operation yielded a reduction of
39% in NO emissions.
x
Steam Injection
The effect of steam injection on NO emissions was evaluated for natural
gas firing using two techniques. In the first method, steam was injected into
the gas manifold of an MA-16 burner and the steam/gas mixture was then injected
radially inward through the standard gas tips. The steam flow rate was varied
up to a maximum of 0.0098 kg/sec (78 Ib/hr). The effect of steam injection
flow rate on NO emissions is shown in Figure 10. The maximum reduction in NO
occurred with the maximum steam flow rate. NO emissions were reduced 33%
from the baseline condition at 0.0098 kg/sec (78 Ib/hr) flow rate.
Since steam for fuel oil atomization is already supplied to the oil
gun, injection through the oil gun is a simpler modification than steam injec-
tion through the gas tips. This second method of steam injection was tried
with a DBA-16 burner. It was hoped that by experimenting with the positioning
of the oil tip relative to the gas tips, WO emissions could be reduced below
2£
the levels of the previous tests.
17
-------
Maximum reduction in NO was achieved at the highest rate of steam
injection114 ppm at 0.0095 kg/s (75 Ib/hr) steam flow. Very little difference
in NO production was observed at the other steam flow rate used (0.0067 kg/s
X
or 53 Ib/hr). Thus, the lowest NO emissions for steam injection through the
X
oil gun was 13% less than the normal, baseline (3% 0 ) NO levels for the
DBA-16 burner. Thus, the influence of steam injection on NO emissions was not
X
nearly as strong for steam injection through the oil gun as it was for steam
injection through the gas tips.
Altered Fuel Injection Geometry (AIG)
Previous work with boilers has shown that NO emissions can be reduced
X
by altering fuel injection geometry to produce locally fuel-rich zones in the
flame. The off-stoichiometric combustion results in lower flame zone tempera-
tures and, thus, lower overall NO production. Based on the results of cold
flow tests in KVB's laboratory the fuel injection geometry was modified for a
DBA-16 natural draft burner.
In the first test series standard gas tips were installed in the DBA-16
with the center firing port facing radially outward rather than inward as is
standard practice. On the basis of the cold flow test results discussed earlier,
this tip orientation was expected to delay mixing of fuel and air, thus producing
a longer, less intense flame and lower NO emissions.
X
The tests showed that NO emissions were indeed lower for this tip
x
configuration than for the standard configuration. At 3% excess O , the NO
^ X
concentration with outward-facing firing ports was approximately 94 ppm (dry,
corrected to 3% 0 ) , about 31% lower than the NO emissions from the standard
2, X
tip orientation. At an excess 0 of 1.1%, the NO level was 78 ppm. The CO
2. X
limit occurred at 0.5% O_, compared with 0.3% O2 for the standard configuration
with a CO concentration of 615 ppm. NOX at this point was down to 73 ppm, 29%
below the standard-configuration value at the CO limit.
The flame shape with the reverse tip orientation was shorter than the
normal flame and segmented into four fuel rich regions, one above each of the
gas tips. The flame was quite lazy at low firing rates.
18
-------
Summary of Hot-Firing Test Results
Table II shows that the largest percent reductions in NO occurred with
X
staged air or flue gas recirculation techniques. With SCA, these reductions
seem to be a relatively strong function of excess air whereas with FGR they are
a rather weak function of excess air.
The percent reductions in NO emissions observed for modifications to
the DBA-16 burner are expected to occur for the same modifications to the MA-16
burner and vice versa with the possible exception of AIG* (where the difference
in burner tiles may play an important role in the mixing patterns resulting
from the modified injection scheme).
The simplest modifications studied other than LEA were the central
staging cylinder and altered fuel injection geometry. AIG as implemented in
this test program applies only to gas-firing situations and produces a flame
which may be undesirable for practical application. The central cylinder
technique produced large percent reductions in NO , increased furnace efficiency,
Ji
and is one of the simplest modifications to implement. Although tests with the
staging cylinder were done only on natural gas fuel, it may also be possible
to use the cylinder in fuel oil applications provided a coke build-up on the
cylinder walls can be avoided.
*altered fuel injection geometry
19
-------
SECTION 4
COST EFFECTIVENESS OF COMBUSTION MODIFICATIONS
TO NATURAL DRAFT PROCESS HEATER
Summary
The cost effectiveness of the combustion modifications applicable to
natural draft process heaters has been evaluated and the results are summarized
in Table V. All costs are based on 1978 dollars.
The largest and smallest heater sizes chosen for the present study
[147 MW (SOOxlO6 Btu/h) and 2.9 MW (10x10 Btu/h), respectively] represent
the two extremes in firing rate for refinery process heaters. The intermediate
size of 73.3 MW (250x10 Btu/h) was chosen for this cost analysis because it
is the current size limit above which steam boilers are regulated by federal
emission standards (which do not presently include process heaters).
The total annualized cost per 10 kg of NO reduction shown in Table V
was determined by amortizing the initial fixed capital costs at 20% (corre-
sponding to straight-line depreciation of the capital equipment over 12 years,
and assuming a 10% cost of capital, state and federal taxes totalling approxi-
mately 11%, and insurance charges of 0.5%). The annual capital charge was
added to the annual operating and maintenance cost to obtain the total annual-
ized cost. These costs are shown in Table VI. (Annual operating costs did
not include projected fuel savings or costs resulting from modifications for
reasons explained below.) The total annualized cost was then divided by the
annual reduction in NO emissions to obtain the cost effectiveness values in
Table V.
The annual reduction in NO emissions was calculated for each modifica-
tion from the maximum percent NO reduction listed in Table II using the formula:
X
3 % NOV reduction Average baseline emissions
Annual NOx reduction (10 kg) - 10Q x from conventional burner
(ng/J) 15
x heat input rate (W) x 31.536x10 Sec/y x 0.8 (use factor) x -^
10 kg
20
-------
NO emission reductions were determined relative to the conventional MA-16
burner baseline for the following modifications:
1. Lowered Excess Air (LEA)
2. Flue Gas Recirculation (FGR)
3. Staged Combustion Air - Floor Lances (SCA-L)
4. Steam Injection (STM)
5. Tertiary Air Burner (TAB)
NO emission reductions were determined relative to the conventional DBA-16
X
burner baseline for these modifications:
1. Altered Fuel Injection Geometry (AIG)
2. Staged Combustion Air - Central Cylinder (SCA-C)
Although efficiency changes associated with each modification were
calculated, these values are not appropriate for estimating annual fuel costs
or savings. They are useful only inasmuch as they indicate expected trends
in fuel consumption. This is so because the research heater tested by KVB at
Location 1 had no process tubes and, therefore, the data do not reflect any
inefficiencies or variabilities due to changes in heat transfer to a process
stream.
Table V shows that the simplest modifications are the most cost
effective. The least expensive modifications, AIG and SCA-C, were tested only
in gas-firing application. It is possible that both techniques may be adapted
to handle oil-firing applications as well. The more involved modifications,
FGR and SCA-L, are less cost effective although they produced the largest
percent NO reductions.
Most modifications result in lower costs per metric ton of NO removed
x
as heater size increases. Only STM and TAB cost effectiveness ratios appear
to be relatively independent of size. For the other modifications, both on
natural gas and No. 6 oil-firing, the cost effectiveness decreased as heater
size increased from 2.9 MW (10x10 Btu/h) to 73.3 MW (250x10 Btu/h) according
to the relation
CE at 73.3 MW = /73.3 Y*
CE at 2.9 MW \ 2.9y
where - 0.67 < a < - 0.47, a = - 0.56, and S (standard deviation) = 0.07.
21
-------
Since
-0.56
CE tt (rated capacity)
and since we expect that
NO reduction (metric tons) <* (rated capacity)
x
therefore,
0.44
total annualized cost « (NO reduction) x (CE) = (rated capacity)
X
For example, using the total annualized cost for a FGR system for a
2.9 MW (10x10 Btu/h) heater given in Table VI at $4300, one can calculate
approximately the total annualized cost of FGR for a 73.3 MW heater as
follows:
/ o\°-44
(^To I x (4300> = $17,810
\ 2*9 /
Conclusions
The most cost effective combustion modification for NO reduction in
X
natural draft process heaters appears to be staged combustion air. The central
cylinder technique is the least expensive type of staged air modification,
although the largest percent NO reduction was obtained using the floor lance
technique. Optimization of the central cylinder concept may further improve
its NO reduction potential, however.
FGR is an effective but more costly modification. TAB, AIG, and STM
are all moderately effective in reducing NO . STM costs were the highest of
any modification for large heater sizes. AIG in the present form applies only
to gas fired units, although the concept is adaptable to oil firing. TAB is
currently available, represents moderate NO reduction capability at moderate
cost, and appears more cost effective for smaller heaters firing No. 6 oil.
22
-------
SECTION 5
REFERENCES
1. Unpublished results from API NOx study at KVB, API Project 705.
2. Schorr, J. R., et al., "Science Assessment: Glass Container
Manufacturing Plants, " EPA-600/2-76-269, October, 1976.
3. Ketels, P. A., et al., "Survey of Emissions Control and
Combustion Equipment in Industrial Process Heating," EPA 600/
7-76-022, October, 1976.
4. Hunter, S. C., et al., "Application of Combustion Modifications
to Industrial Combustion Equipment," KVB, Inc., presented to the
2nd Symposium on Stationary Source Combustion, August 29-Sept.
1, 1977.
5. Allen, K. C., Directory of Iron and Steel Works of the United
States and Canada, 33rd edition, American Iron and Steel
Institute, July, 1974
6. Private Communication with Max Hoetzl, Surface Combustion, Inc.,
November 17, 1977.
7. Private communication with Chuck Mellus, Surface Combustion, Inc.,
November 18, 1977.
8. Sittig, Marshall, Practical Techniques for Saving Energy in the
Chemical, Petroleum, and Metals Industries, Noyes Data Corp.,
Park Ridge, New Jersy, 1977.
9. National Emissions Data System, Emissions by SCC, Oct. 27, 1977,
provided by EPA, Nov. 1977.
10. Popper, Herbert, Modern Cost-Engineering Techniques, McGraw-Hill
Book Co., New York, 1970.
11. Private communication from Vern Sharpe, Sharpe Heating and Ventilating,
Alhambra, CA to R. J. Tidona (KVB), June 22, 1978.
12. Private communication from Industrial Gas Engineering, Westmont, IL,
to R. J. Tidona (KVB), June 22, 1978.
13. Typical Electric Bills, 1977, Federal Power Commission, FPC R90.
14. Private communication from refinery heater burner manufacturer to
S. S. Cherry (KVB), March 21, 1978.
23
-------
2.0
1.6
1.2
8
0.8
0.4 _
T
I f
Middle Level
Gas tips toward walls
Cylinder 5.4 cm (2-1/8 in.) above tips
Deflectors
Cylinder 30.5 cm (12 in.) above tips
Burner
Centerline
DISTANCE, cm (in.)
Burner Opening
Figure 1. CO concentration vs. centerline distance at one axial position with four different
modifications.
-------
leo
tJ
r
o
<*>
160
140
120 .
100
e so
a
o
z;
60
40
20
I
T
Fuel: Natural Gas
Firing Rate: 1.52 MW (5.2x10
(CO concentration)
- denotes CO limit
Btu/h) nom.
Q
(31 ppm)
MA-16
O Unmodified
-~O SCA (4 tubes)
FGR (40% nom.)
DBA-16
Unmodified
SCA (Central Cyl.)
ALT. INJEC. GEOM.
LOW-NOX(TERTIARY
AIR) BURNER
All reg. 100% open
-»n Extnd. Sec. Tile,
All reg. 100% open
I
I
I
1
012 345
STACK EXCESS OXYGEN, %, dry
Figure 2. Summary of NO emissions as a function of excess oxygen for
subscale natural draft furnace firing natural gas.
25
-------
360
320
280
240 _.
200
a
dp
ro
+J
m
160
* 120
80
40
I I I I I
Fuel: No. 6 Oil &
Firing Rate: 1.52 MW (5.2x10 Btu/h) nom.
(CO concentration)
- denotes data at or below the CO limit
(294
?pm)
MA- 16
Unmodified
Staged Air (4 tubes)
l"~| Flue Gas Recirculation (40%
L_l nom. )
LOW-NO (TERTIARY AIR) BURNER
registers 100% open (unmod.)
\/
\/ Extended Secondary tile,
All registers 100% open
I
234
STACK EXCESS OXYGEN, %, dry
Figure 3. Summary of NO emissions as a function of excess oxygen for
subscale natural draft furnace firing No. 6 oil.
26
-------
ro
Air
Supply
Manifold
Pilot
Air Supply
Tube
Tile in 18
Sections
s Tips
diam. Air
Manifold
Pilot Gas Conn
Air Supply Tube
"Length adjustable
0.3-1.5 m (1-51)
Expansion Joint
\ J
1.3 cm (1/2") Gas Conn.
1.3 cm (1/2") Steam Conn.
Figure 4. Schematic of staged combustion using an MA-16 burner.
-------
100
I
a
Fuel: Natural Gas
firing tote: 1.5 MW (5.1 Btu/hr x 10 )
Gas Tip: Pattern II
I
I
I
0.6 0.7 O.8 0.9 1.0 1.1
*, FRACTION OF STOICHIOHETRIC AIR TO BURNER
1.2
Figure 5. NO emissions as a function of burner d), (A/F) /(A/F)
actual stoich'
400
300 _
o
Z 200
100
Fuel: No. 6 Oil
Firing Rate:
Tip: 764
1.51 MW (5.1 Btu/hr x 10 )
Teat No.
1/6-12, 3.
2J 1/6-13, 3,
1/6-14, 1.
1/6-15, 1,
1/6-16, 3.
1/6-17, 3.
7 J 1/6-18, 4.
1/6-19, 3.
0.6
0.7
0.8 0.9 1.0 1.1
, FRACTION OF STOICHIOMETRIC AIR TO BURNER
1.2
1.3
Figure 6. NO emissions as a function of burner for No. 6 oil firing.
28
-------
Tertiary Air Register (40% of air)
\
Furnace
Floor
Secondary Air
Register
60% of air
Primary Air
Register
< Oil Gun
Not to Scale
Register Controls
Figure 7. Schematic of tertiary air burner for natural draft process heater.
29
-------
i
100
I 5°
I
i
Baseline
Fuel: Natural Gas
Firing Rate! 1.48 HH
(5.1 Btu/hr x 106)
Gas Tip: Pattern II
10
2O 30 40
RECIRCULATED FLUE GAS, %
50
Figure 8. The effect of flue gas recirculation on NO emissions (natural gas)
200
ISO
o
CM
O
100
50 -
Fuel: Ho. 6 oil
Firing Rate: 1.49 UK
(SO Btu/hr x 106)
Burner Tip: 764
Test no.
) 1/7-10, 3.0% O
) 1/7-11, 3.2% 02
} 1/7-12, 3.0% 02
) 1/7-13, 3.0% O2
> 1/7-14, 2.0% 02
) 1/7-15, 2.0% O2
) 1/7-16, 2.0% 02
) 1/7-17, 0.8% 02
) 1/7-18, 1.0% O2
>1/7-19, 1.0% 02
11/7-20, 2.5% O,
I
10 20
RECIRCUIATED FLUE GAS, %
30
40
Figure 9. The effect of flue gas recirculation on NO emissions (No. 6 oil)
30
-------
100
<*>
ro
4)
a
50
(i/5-i)
3.4%
^(1/5-2)
3.3%
Fuel: Natural Gas
Firing Rate: 1.58 MW (5.4xl06 Btu/hr)
Gas Tip: Pattern II
(Test No.)
Excess O
Steam injection through gas tips
I
I
I
I
o
0.0025
(20)
0.0050
(40)
0.0076
(60)
0.0101
(80)
0.0126
(100)
STEAM INJECTION, kg/s (Ib/hr)
Figure 10. The effect of steam injection on NO emissions for the MA-16
burner firing natural gas.
31
-------
TABLE I.
INDUSTRIAL PROCESS CHARACTERISTICS AND NO EMISSIONS
CO
ro
Source Category
Subheading
(Reference)
Number of units
in U.S.
Average design
capacity per
unit, SI
(customary)
Average design
heat input rate
per unit
MW(Btu/h)
Total actual
annual produc-
tion kg (tons)
Average heat
input per unit
of throughput
(J/kg (Btu/t)
Total annual
heat input
J (Btu)
Average emis-
sion factor
ng/J (lb/10 Btu)
Total NOX
emission
Gg/y (t/y)
NOTES: (a) based
Refinery Process Heaters
Natural Draft
(1)
M400
8.06
(27.5xl06)
...
1.37xlo}|!
(1.3X1015)
68.8
(0.16)
93.4
(103,000)
on unpublished dat
Forced Draft
(1)
^600
11.1
<38xl06)
211xl015
(200xl012)
133.3
(0.31)
28.1
(31,000)
.a from Californ
Glass Container
Furnaces
All
(2)
334
1.57 kg/s
(150 t/d)
(44.6xl06)
12.656X109
(13.953xl06)
8.29X106 (a>
(7..14X106)
105xl015
(99.6xl012)
370
(0.861)
38.89
(42,900)
Cement Kilns
All
(KVB Analysis)
412 (in 1975)
6.60 kg/s
(629 t/d)
39.8
(136x10 )
84.8x10
(93.5X106)
6.04x10° (b'
(5.2xl06)
513X101*
(486xlOi<:)
1372 . .
(3.19) (C)
704
(776,000)
id Air Resources Board sf. udy of SOx
, . _
Steel Furnaces
Soaking Pits
(KVB Analysis)
1435
(175X1012)
56
(0.13)
10.32
(11,375)
emission by KVB.
Reheat Furnaces
(KVB Analysis)
1264 (d)
25.2 kg/s
(100 t/h) l '
88.0
(300x10 )
3.48x10° . .
u.oxioV9'
15 (h)
353x10
(335xl012)
52 (i)
(0.12)
18.23
(20,100)
Hood/Bark
Boilers
All '31
(KVB Analysis)
^ noo
18. B MW heat absorbs
(64,000 Ib stean/h)
23.5 MW
(BOxlO6)
328X10900
(364x10° tons steam)
2.8x10° (1)
(2.4X106)
915xl015
(873xl012)
156 (B)
(0.36)
141
(157,000)
(b) from Reference 3
(c) average of emission factors determined from two field tests conducted by KVB as reported in Reference 4.
(d) from Reference 5
(e) from Reference 6 (average for slab reheat furnaces)
(f) from Reference 7
(g) from References 3, 6 and 7
(h) from Reference 3 (total annual heat input for both soaking pits and reheat furnaces combined) and Reference 8 (fraction of total heat input
to soaking pits, fraction to reheat furnaces)
(i) from KVB tests as reported in Reference 4
(j) based on 1977 NEDS point source listing (Ref. 9) with KVB emission factor
(K) assumes operation at 80» rated load for 11 months out of the year
(1) assumes heat requirement of 1200 Btu/lb of steam
(m) from Reference 4i average for wood + coal and wood + NG boilers.
-------
TABLE II. RESULTS OF TESTS ON A SUBSCALE NATURAL DRAFT PROCESS HEATER
Modification
Lowered Excess Air
Lowered Excess Air
Fuel
Natural Gas
No. 6 Oil
Average
ng/J
58
174
Baseline NOX
ppm, dry at
3% 0_
113
311
Maximum Percent
Reduction in NO
27
10
Staged Combustion (Floor Natural Gas 61.2
Lances) at normal excess
oxygen
Staged Combustion (Floor Natural Gas 61.2
Lances) at low excess oxygen
Staged Combustion (Floor No. 6 Oil 172
Lances) at normal excess
oxygen
Staged Combustion (Floor No. 6 Oil 172
Lances) at low excess oxygen
Staged Combustion (Central Natural Gas 66.8
Cylinder) at normal excess
oxygen
Staged Combustion (Central Natural Gas 66.8
Cylinder) at low excess
oxygen
120
120
307
307
131
131
46
67
35
51
31
59
Tertiary Air Burner, Lowest
NOX Configuration
Tertiary Air Burner, Lowest
NOx Configuration
Recirculating Tile Burner,
Lowest NOX Configuration
Recirculating Tile Burner,
Lowest NOX Configuration
Flue Gas Recirculation at
normal excess oxygen
Flue Gas Recirculation at
low excess oxygen
Flue Gas Recirculation at
normal excess oxygen
Flue Gas Recirculation at
low excess oxygen
Steam Injection
Altered Fuel Injection
Geometry at normal excess
oxygen
Altered Fuel Injection
Geometry at low excess oxygen
Natural Gas
No. 6 Oil
Natural Gas
No. 2 Oil
Natural Gas
Natural Gas
No. 6 Oil
No. 6 Oil
Natural Gas
Natural Gas
Natural Gas
54.6
160
54.6
61.8
59.2
59.2
141
135
54.6
66.8
66.8
107
285
107
112
116
116
252
241
107
131
131
30
42
3
13
59
63
31
39
33
31
44
33
-------
TABLE IV. SUMMARY OF AVERAGE BASELINE GASEOUS EMISSIONS FOR UNMODIFIED BURNERS
Heat Input Rate
MW (106 Btu/h)
Natural Gas
MA-16 1.53
DBA-16 1.52
Low-N0x Burner 1.49
(Tertiary Air
Injection)
Low-NOx Burner 1.47
(Recirculating
Tile)
No. 6 Oil (0.3% N)
MA-16 1.47
Low-NOx Burner 1.43
(Tertiary Air
Injection)
No. 2 Oil (0.01% N)
MA-16 1.41
Low-N0x Burner 1.49
(Recirculating
Tils)
(5.2)
(5.2)
(5.1)
(5.0)
(5.0)
(4.9)
(4.8)
(5.1)
°2 C02 NO NO CO SO2
* % ppm* na/J ppm* ng/J ppm* pom*
3.0 10.7 107 54.6 103 53.8 0 0
3.0 10.3 131 67.0 127 64.9 0 0
3.2 10.4 92 47.1 87 44.4 0 0
3.1 9.9 104 53.0 104 53.0 0 0
3.0 13.3 285 159 278 156 0 1015
3.1 13.7 265 149 261 147 0 1334
3.1 12.7 112 63 108 61 0 46
3.9 12.6 110 61.7 105 58.9 0 38
*Corrected to 3%
dry
-------
TABLE V. COST EFFECTIVENESS ($/10 kg of NOX reduction) OF COMBUSTION
MODIFICATIONS TO A NATURAL DRAFT PROCESS HEATER
(NOT INCLUDING ANNUAL FUEL COSTS/SAVINGS)
Modification
Low Excess Air
Altered Injection
Geom.
(Normal O2>
Altered Injection
Geom.
(with LEA)
Staged Air - Central
Cyl. (Normal 02)
Staged Air - Central
Cyl. (with LEA)
Staged Air - Floor
Lances (Normal 0_)
Staged Air - Floor
Lances (with LEA)
Flue Gas Recircula-
tion (Normal O2)
Flue Gas Recircula-
tion (with LEA)
Steam Injection
(no initial cost)
Steam Injection
(incl. initial cost)
Tertiary Air Burner
2.9 MW
UOxlO6 Btu/h)
Natural No. 6
Gas Oil
0 0
$6.60
$4.70
$200
$100
$1000 $460
$710 $320
$1800 $1200
$1700 $940
$990
$1400
$250 |$60
Heater Size
73.3 MW
(250xl06 Btu/h)
Natural No. 6
Gas Oil
0 0
$0.78
$0.55
$43
$23
$160 $70
$110 $48
$320 $200
$300 $160
$970
$1100
$250 $60
147
(SOOxlO6
Natural
Gas
0
$0.78
$0.55
$33
$18
$130
$87
$320
$300
$960
$1000
$250
MW
Btu/h)
No. 6
Oil
0
$57
$39
$200
$160
$60
Indicates lowest cost effectiveness for each size and fuel, excluding
low excess air and altered injection geometry.
35
-------
TABLE VI. TOTAL ANNUALIZED COSTS (IN $) NOT INCLUDING FUEL COSTS (SAVINGS)
OF COMBUSTION MODIFICATIONS TO A NATURAL DRAFT PROCESS HEATER
(AMORTIZING INITIAL CAPITAL COSTS AT 20%)
Modification
Low Excess Air
Altered Injection
Geom.
Staged Air - Central
Cyl.
Staged Air - Floor
Lances
Flue Gas Recircula-
tion
Steam Injection
(if initial instal-
lation necessary)
Steam Injection
(no initial instal-
lation required)
Tertiary Air Burner
Heater Size
2.9 MW
(lOxlO6 Btu/h)
0
10
300
1900
4300
1900
1300
300
73.3 MW
(250xl06 Btu/h)
0
30
1660
7330
18800
35400
32200
7500
147 MW
(BOOxlO6 Btu/h)
0
60
2560
11800
38100
69900
64300
15000
36
-------
POLLUTANT EMISSIONS FROM "DIRTY"
LOW- AND MEDIUM-Btu GASES
By:
Richard T. Waibel
Edward S. Fleming
Dennis H. Larson
Institute of Gas Technology
Chicago, Illinois 60616
37
-------
ABSTRACT
Data were collected to determine the emissions from "dirty" low- and
medium-Btu gases when combusted on industrial process burners. The fuels
utilized were blended to have the composition typically found for Wellman-
Galusha oxygen (WGO) and air (WGA) fuel gases. Base-line data were collected
for natural gas, ambient WGO and WGA, and hot WGO (700K) and WGA (616K) .
Then ammonia, hydrogen sulfide, coal tar and char were added singly at vari-
ous levels and in combinations to the hot fuels in a parametric study to
determine the effects of these contaminants on pollutant emissions. The
burners used in this study were a forward-flow baffle burner and a gas
momentum controlled kiln burner. These burners were mounted in turn on a
pilot-scale test furnace equipped with water tubes as a load and were each
fired at 1.03 + 0.07 MW (3.50 + 0.25 X 106 Btu/hr) with 10% excess air and
477 K (400°F) air preheat.
Based on a detailed analysis of the experimental data, the following con-
clusions were made:
Low-Btu fuels not subjected to post-gasifier cleanup can yield
NOX levels greatly above the thermal levels for the clean fuels
and for natural gas.
In turbulent diffusion flames, fuel-NOx increases with an increase
in a) the amount of fuel-nitrogen, b) the amount of fuel-sulfur,
c) the level of excess air, and d) the degree of initial fuel/air
mixing.
Attempts to close the fuel-sulfur balance were unsuccessful.
Whether this shortfall is due to sampling/instrument effects or
large concentrations of some unmeasured sulfur-containing species
is not clear. Further work should be done in this area.
Compared to natural gas, heat transfer to the load is reduced for
the low-Btu fuels tested. This heat transfer is not greatly af-
fected by the presence of contaminants (tar and char) at levels
characteristic of raw gasifier effluents.
38
-------
INTRODUCTION
The objective of the research program was to provide and evaluate quan-
titative data on the differences in the environmental quality of effluent
combustion products and furnace efficiency when retrofitting a natural gas/
oil industrial burner with "dirty" intermediate- and low-Btu gases. These
data were collected from the IGT pilot-scale industrial test furnace.
This program was intended to complement a recently completed EPA pro-
gram (1) , which evaluated th6 emissions resulting from burning "clean" low-
and medium-Btu gases in boilers. This previous work provided quantitative
information on emissions using gases processed by ambient temperature
sulfur cleanup systems and enabled correlating emissions to gas composition
and operating conditions.
Work is currently being done by the government and industry to develop
high temperature-post gasification sulfur cleanup systems. The high-temper-
ature cleanup processes leave varying amounts of tars, oils, and ammonia in
the product gas stream. Since these contaminants can contribute to emissions
resulting from combustion of fuel gas, it is important that the magnitude of
the potential problem be evaluated and the results used to determine if
high-temperature sulfur removal systems are feasible from an environmental
viewpoint.
"Dirty" low- and intermediate-Btu gases could have higher flame
emissivities than "clean" gases due to tar-oils and char, which could result
in increased furnace efficiency. However, the hot "dirty" fuel often has a
lower adiabatic flame temperature due to a higher water content in the fuel.
This program was designed to quantify such changes in efficiency and provide
data on which a decision can be made on retrofitting an industrial process
furnace with low- or intermediate-Btu gases.
This program was designed to provide data for two gases and two
burners as a means to broadly scope the potential environmental problem.
ENVIRONMENTAL ASSESSMENT OF COMMERCIAL GASIFICATION PROCESSES
Tars and Oils
A wide variety of coal conversion systems can be used to produce a
low- or intermediate-Btu gas. However, the operating conditions of the
39
-------
gasifier and cleanup system will have a significant effect upon the types of
potential pollutants present in the off-gas. In most of the gasification
processes, the initial treatment of the coal determines the major character-
istics of the raw gas. Initial dimensions of contact time, gas-to-coal ratio,
contacter types (entrained bed, fluidized-bed, fixed-bed, stirred liquid),
and contacter mode (cocurrent, countercurrent, back-mixed reactor), and perhaps
other variables should be included for a complete profile of the types of
pollutants that might be present in the raw gasifier product, particularly
tars and oils.
These liquids would be condensed and removed from the gas stream using
ambient temperature sulfur cleanup systems, but would not be removed when
using the high-temperature processes. If not condensed, they would be burned
during the combustion process with the off-gas. The combustion of these tar-
and oil-containing gases in an environmentally acceptable manner was part of
the overall experimental evaluation.
The basic character of the complex coal organic structure is aromatic.
Therefore, the tars that are expelled from coal during devolatilization in
lower temperature reactors may be expected to contain naphthalenes, indenes,
anthracenes, and similar compounds. Oxygenated compounds such as phenols and
cresylic acids may be expected, in addition to nitrogen- and sulfur-containing
ring structures. In moderate temperature reactors, these complex aromatics
are hydrocracked and possibly hydrodealkylated to simpler BTX (benzene-toluene-
xylene) streams.
Of the conversion processes commercially available (e.g. Koppers-Totzek,
Winkler, Wellman-Galusha, and Lurgi), only Wellman-Galusha and Lurgi produce
significant amounts of tars and oils.
Sulfur
The sulfur that is present in coal is one of the primary reasons that
low- and intermediate-Btu gasification processes are being developed. Much
of the coal in this country contains significant quantities of sulfur, and
present methods of sulfur oxides reduction, such as stack-gas scrubbing, are
not viewed as sufficiently reliable, effective, or operable by many potential
coal users. The concept of low- and intermediate-Btu gasification permits
the removal of the sulfur from the gas before combustion, and overall sulfur
40
-------
oxides emissions may be significantly reduced utilizing this approach. The
other major impetus for the development of low- and intermediate-Btu gasifi-
cation is the higher overall efficiency achieved when the gas is utilized to
produce electricity in combined-cycle operations.
The sulfur that occurs naturally in the coal will be largely driven into
the raw product gas although hard data do not exist on the form of the sulfur
in the off-gas. Thermodynamically, the great majority of the sulfur should
exist as hydrogen sulfide.
Purification systems are available for removing a large quantity of the
hydrogen sulfide that is present in the raw product gas. Currently, much
work is being done to develop and demonstrate the use of high-temperature
sulfur cleanup systems. However, even the most optimistic projections show
an H.S concentration of 100 ppm in the off-gas.
Nitrogen
Nitrogen that is present in coal may be considered as fixed nitrogen.
Fixed nitrogen may be defined as nitrogen that is chemically bound to other
species in contrast to molecular nitrogen (N_) that is present in the air.
The nitrogen in coal tends to gasify simultaneously with the carbon (2).
Generally, the nitrogen is expected to react with the hydrogen during gasifi-
cation to form ammonia. The existence of ammonia in raw gasifier effluents
has been confirmed by many investigators.
Hydrogen cyanide (HCN) is also present in the raw gas effluents. Pub-
lished data report concentrations of less than 10% of the ammonia concentration.
Thus, the major contribution in NO formation will be ammonia.
' J x
41
-------
TEST FACILITIES
PILOT-SCALE FURNACE
The experimental work was carried out in the pilot-scale furnace which
is 14 feet* long and has a cross-sectional area of 21.3 sq ft. There are
33 panels or "sampling doors" along one sidewall that allow insertion of
probes at any axial position from the burner wall to the rear wall. The
facility can be used for firing burners rated up to 6 million Btu/hr (6 MBtu/hr)
Combustion air temperatures up to 1000°F can be generated with a separately
fired air preheater.
The furnace is also equipped with 58 water cooling tubes, each of which
can be independently inserted through the roof, along the sidewalls. Varying
the number of tubes, their location, and the depth of insertion allows control
over the magnitude and character of the load that can be placed on the furnace.
The amount of heat absorbed by each tube can be determined by measuring the
water flow through each tube and the temperature difference between the inlet
and outlet.
In addition to the combustion air preheater, a separately fired fuel
preheater is available that can heat 12,000 SCFH of low-Btu gas to any desired
temperature up to 800°F. Temperatures up to 1200°F are attainable with lower
flow rates.
LOW-Btu GAS GENERATING SYSTEM
The low- and medium-Btu gases are generated using a special gas-generating
and fuel-preparation facility that can produce varying ratios of hydrogen and
carbon monoxide. Natural Gas, carbon dioxide, and steam are passed through re-
action retorts contained in a vertical cylindrical furnace. After compression,
the product gas is blended with nitrogen, methane, carbon dioxide, and/or
steam, as required, to obtain the specified composition of the fuel gas to be
tested.
*
It is EPA policy to use metric units. However, in this report,
English units are occasionally used for convenience.
42
-------
Table I gives the composition of the Wellman-Galusha oxygen (WGO) and
Wellman-Galusha air (WGA) fuel gases, which were chosen to be simulated as
test gases for the program.
DOPANT SYSTEM
Although a number of potential pollutants might be present in the raw
gasifier products, the data now available on their occurrence and/or concen-
tration are poor. Consequently, the trials concentrated on the fate of species
for which some data exists (tars, oils, ammonia, particulate, and hydrogen
sulfide). These contaminants were "doped" into the hot experimental fuel gas.
Although coal-tar does not have an identical chemical analysis to that of tars
and oils found in raw off-gas, it does contain all the aromatic and oxygenated
compounds that are found. Thus, realistic characterization of the combustion
process and pollutant emissions can be anticipated. Analyses of the tar and
char are presented in Tables II and III.
Figure 1 gives a schematic diagram of the doping system. Raw coal-tar
from a coke oven was used as the tar introduced into the hot gas stream.
Ammonia and hydrogen sulfide were blended into the fuel gas stream from
cylinders. The flow of these dopants was adjusted using rotameters. The tar,
which was a liquid at room temperature, was forced from a container under
nitrogen pressure through a nozzle in the hot fuel feed line where it was
steam atomized. Char was screw-fed into the hot fuel. Doping rates were
controlled by varying the screw speed.
INSTRUMENTATION
The instrumentation used during this study is fully described in EPA
report 600/7-78-191. A listing is provided here:
Suction pyrometer with Pt/Pt-13% Rh thermocouple for gas temperature
Beckman 742 Polarographic Oxygen (0?)
Beckman Paramagnetic Oxygen (0^)
Beckman NDIR Methane (CH )
Beckman NDIR Carbon Monoxide (CO)
Beckman NDIR Carbon Dioxide (C09)
43
-------
Varian 1200 Flame lonization Chromatograph (Total CH and C~ to C )
Beckman NDIR Nitric Oxide (NO)
Beckman UV Nitrogen Dioxide (NO )
Thermo Electron Pulsed Fluorescent Sulfur Dioxide (S0?)
Hewlett-Packard Thermoconductivity Chromatography, Hydrogen (H ),
Nitrogen (N2>, Argon (Ar), CO, C02> C^ to C5> Oxygen (02)
Beckman Chemiluminescent NO-NO
x
Molectron PR-200 Radiometer for radiation intensity
Research Appliances Corp* Model 2414 "Staksamplr" for EPA Method 5
particulate measurements
Anderson Mark III multistage, cascade impactor for particulate
size distribution.
44
-------
RESULTS
Summaries of all the test data for the ported baffle "burner and the fuel
momentum controlled burner are presented in Tables IV and V respectively. The
following sections present the detailed findings by fuel type.
NATURAL GAS: BASE-LINE TESTS
Baffle Burner Tests
A baffle burner, representative of the forward-flow type, was the
first burner to be utilized. The burner tested, illustrated in Figure 2, is
full scale and is available as an off-the-shelf item from the manufacturer
(Bloom Engineering). The burner consists of a centrally located gas nozzle,
surrounded by a high-temperature refractory baffle that has ports for the
injection of combustion air into the furnace. The flame patterns produced
by this burner can be altered by changing the angles of these air ports or
their diameters with the insertion of different baffles.
This type of burner is found on many large-scale industrial process
heating furnaces such as steel reheating, batch glass melting, aluminum
holding, and tunnel kilns. The baffle design selected for testing produces
a flame-to-furnace length ratio equal to the flame-to-preheat section length
ratio found in a five-zone steel slab reheat furnace.
The rate of heat absorption and efficiency (approx 35%) of the preheat
section of a steel reheat furnace was simulated on the pilot-scale furnace
by inserting water cooling tubes along the furnace sidewall.
Fuel Momentum Controlled Burner Tests
A fuel momentum controlled kiln burner (FMCB) was the second burner
tested and the pilot-scale test furnace was set up to specifically simulate
a cement kiln. The critical operating parameters were a) a length sufficient
to simulate the calcining and reaction zones, b) the firing density, and c)
the heat absorption profile.
A typical kiln consists of preheat, calcining, reaction, and cooling
zones. The cross-sectional area (20 sq ft) and length (14 ft) of the IGT
pilot furnace allow for simulation of the two zones of primary importance:
the calcining and reaction zones. It is the heat transfer in these zones
that is sensitive to fuel type. These two zones occupy about one-third of
45
-------
the overall kiln length, with the calcining zone being twice the length of
the sintering zone. The flame usually extends three-quarters of the length
of these zones.
Firing densities in rotary kilns range from 10,000 to 20,000 Btu/hr-ft .
3
A firing density value of 12,500 Btu/hr-ft was used for these tests and is
typical of many kilns. This requires a firing rate of 3.5 million Btu/hr for
our test furnace volume. Cement kilns require about 5 to 7 million Btu/ton
of clinker. Assuming 6 million Btu/hr, our furnace would simulate a production
rate of 1150 Ib/hr.
This rate of heat absorption and this efficiency were simulated on the
pilot-scale furnace by inserting water cooling tubes along the furnace side-
walls, while firing natural gas at 3.5 million Btu/hr. Figure 3 is a schematic
of the kiln burner fuel injector.
WELLMAN-GALUSHA OXYGEN: EMISSION STUDIES
Prior to the doping studies, base-line emissions were obtained for
natural gas and Wellman-Galusha oxygen (WGO) fuel gas on the baffle burner and
kiln burner and are shown in Table VI. Both fuels were fired at 1.03 + 0.07
MW (3.50 + 0.25 X 10 Btu/hr) with 10% excess air. The above variation in
fuel heat input represents the range of firing rates from run to run and not
the firing rate fluctuation during a given run, which was minimal. For all
tests, the furnace was at positive pressure. All concentrations presented are
dry analyses at the flue entrance. For NO levels under 100 ppm the repro-
A.
ducibility of the reported values was * 5 ppm while at the higher levels it
was + 10 ppm. Recent studies (3) have shown that quenching may cause inter-
ference in NO measurements. Our results, however, have not been compensated
Jv
for such effects.
With natural gas and WGO on the baffle burner a 0.063-m inside diameter
(ID) (nominal 2-1/2 inch Schedule 40) fuel nozzle was used. Fuel velocities
at the nozzle exit were 40 m/s (130 ft/s, ambient WGO) and 110 m/s (360 ft/s,
hot WGO). Natural gas velocity was 9.5 m/s (28 ft/s). Combustion air entered
the furnace at velocities of 19 m/s (63 ft/s) for the low-Btu fuels and 23 m/s
(76 ft/s) for natural gas.
46
-------
On the kiln burner a 0.043-m ID (nominal 1-1/2 inch Schedule 10) axial
fuel nozzle with a 0.019-m ID (7/8 inch, 16 gauge) tube as the radial injector
was employed for natural gas and ambient (322 K) WGO. For hot (700 K) WGO, a
0.108-m ID (nominal 4 inch Schedule 10) axial fuel nozzle with a 0.064-m
(nominal 2-1/2 inch Schedule 10) radial injector was required to achieve
comparable fuel velocities as well as to maintain flame stability. The amount
of radial flow for natural gas on the kiln burner was 95% of the total. This
value was selected to give a stable flame of a size compatible with the fur-
nace dimensions. For both ambient and hot WGO, the flow was 22% radial, chosen
to give a flame length comparable to that of natural gas. With the fuels
studied, gas injected radially was at sonic velocity while the axial component
entered at 85 m/s (280 ft/s) for ambient WGO. The axial velocity was 52 m/s
(170 ft/s) for hot WGO on the larger nozzle. With natural gas, the axial flow
was 1.2 m/s (4 ft/s). Air velocity with the kiln burner was 3.1 m/s (10 ft/s).
In order to determine the effects of the various potential sources of
fuel-NO on the measured NO levels, char, tar, and ammonia were individually
X X
doped into hot WGO. The doping system and the char (0.66 weight percent
nitrogen) and tar (0.55 weight percent nitrogen) analyses were presented in
the preceeding section. The results of the char and tar doping tests are
presented in TablesVII and VIII. For char, the doping rate varied from 0.13 to
0.86 g/s or 0.4 to 2.7 grains/SCF, while the tar levels of 0.36 to 0.73 g/s
correspond to 1.1 to 2.3 grains/SCF. The total NO levels measured for char
X
or tar are comparable and show a NO increase from 0 to 30 ppm. The fact
X
that bound sulfur in the tar (0.47 weight percent sulfur) and char (1.64
weight percent sulfur) is converted to sulfur oxides is evidenced by the
SO levels. Increases in NO levels above undoped thermal NO levels cannot
2, X X
simply be ascribed to fuel-nitrogen conversion when fuel-sulfur is also pre-
sent. Work of IGT (unpublished) and others (4) has shown that in turbulent
diffusion flames thermal NO , as well as fuel NO , is enhanced by fuel-sulfur.
X X
In the tar doping studies the problem of char-like residue collecting inside
the fuel nozzle was encountered. On the baffle burner, this residue amounted
to about 3% by weight of the doped tar, whereas on the kiln burner the residue
was around 7% of the tar input.
47
-------
The effects of various levels of fuel-nitrogen, in the form of
ammonia, on the conversion of fuel-nitrogen to NO are shown in Tables IX
and X and Figures 4 and 5 for the baffle and kiln burners with 10% excess
air. Ammonia was metered into the hot WGO at levels of 0.02 to 1.00 volume
percent of fuel input. The oxygen level at the flue was maintained at
1.8 + 0.1% (dry analysis), corresponding to 10% excess air. The fraction of
ammonia converted to NO decreases as the fraction of ammonia in the fuel
x
increases. The results are comparable for both burners with the kiln burner
giving a somewhat higher conversion.
The effects of changing the amount of excess combustion air are also
shown in Tables IX and X and Figures 4 and 5 for the baffle and kiln burners
with 20% excess air. Here, ammonia constituted from 0.20 to 1.09 volume
percent of the fuel input. The oxygen level at the flue was held at
3.3 + 0.1% (dry analysis) to keep the excess air level at 20%. A comparison
of Figures 5 and 6 shows that the fraction of ammonia converted to NO in-
creases with an increase in the availability of oxygen. Again the kiln
burner gives a slightly higher ammonia-to-NO conversion than the baffle
X.
burner.
The effects of ammonia doping on flue oxygen levels were determined
by setting the flue oxygen to the level required for 10% excess air with 1.0%
ammonia addition and then reducing the ammonia level in five steps to 0.2%.
For a given reduction of ammonia the oxygen level rose by an amount consis-
tent with the reaction -
NH3 + (0.75 -I- |)02 = f NO + (-1 ~2 f)N2 + 1.5 R^
where f is the fraction of ammonia converted to NO .
x
Because raw hot gasifier effluents contain both fuel-nitrogen and fuel-
sulfur, the effects of the latter on fuel nitrogen conversion to NO were
X
determined by adding various levels of hydrogen sulfide to 1.0% ammonia-
doped hot WGO. The results for both burners are shown in Figure 6. Hydro-
gen sulfide was metered in from 0.02 to 2.89 volume percent of fuel input.
Oxygen in the flue was kept at 1.8%, corresponding to the 10% excess air
level. As can be seen, small amounts of fuel-sulfur significantly enhanced
48
-------
fuel-nitrogen conversion to NO . Above about 1.5%, and up to 2.9%, hydrogen
3C
sulfide did not add greatly to further total NO enhancement. These results
x
were essentially the same for both burners.
A review of the literature (5) indicated that the degree of initial fuel/
air mixedness had an appreciable effect on fuel-nitrogen conversion to NO .
x
With 1.0% ammonia-doped hot WGO at an excess air level held at 10% on the kiln
burner, the amount of radial flow was varied from 0% to 36% of the total. The
results, shown in Figure 7, confirmed the effect of initial mixing on fuel-
NO emissions. The ammonia conversion to NO rose sharply from 10% radial
x x
flow to 36%. The thermal NO was not appreciably affected over the same
X
range.
Following the single-dopant tests, so-called parametric studies were
performed, wherein the various contaminants were added in combinations of
two. The results of the ammonia-plus-tar and ammonia-plus-char parametric
trials are presented in Table XI. The doping rates were representative
of those used in the single-dopant tests. The total NO levels were found
to be about 5% to 20% higher than expected on a simple additive basis (de-
rived from the single-dopant studies), indicating some kind of synergistic
effect.
The char-plus-tar parametric and the ammonia-plus-char-plus-tar
("dirty") doping results are shown in Table XII. Doping levels were again
consistent with previous tests, and excess air was held at 10%. Total NO
A
levels in the case of char-plus-tar appear to be purely additive while the
results of the "dirty" doping trials show the same kind of NO enhancement
X
as seen in the amonia-plus-char and ammonia-plus-tar tests. Total fuel-
nitrogen conversion to NO is slightly higher on the kiln burner than on the
X
baffle burner (~10% versus ~9%) if one assumes that thermal NO levels are
X
not greatly affected by the sulfur in the dopants.
WELLMAN-GALUSHA AIR: EMISSION STUDIES
As with Wellman-Galusha oxygen (WGO) fuel gas, base-line data were ob-
tained for clean Wellman-Galusha air (WGA) fuel gas at ambient and elevated
temperatures on both the baffle and kiln burners. These data are presented
in Table XIII along with the base-line natural gas data previously shown in
Table VI. The fuels were fired at 1.03 + 0.07 MWfc (3.50 + 0.25 X 10 Btu/hr)
49
-------
with 10% excess air. For the baffle burner, natural gas and ambient (322 K)
WGA were burned on the same 0.063-m ID (nominal 2-1/2 inch) fuel nozzle as
WGO. Hot (616 K) WGA required a 0.078-m ID (nominal 3 inch Schedule 40)
fuel nozzle to overcome the stability problems encountered on the 0.063-m ID
nozzle. WGA fuel velocities were 63 m/s (206 ft/s) for ambient WGA and 83
m/s (273 ft/s) for hot WGA with combustion air at 19 and 30 m/s (62 and 99
ft/s), respectively.
On the kiln burner, the 0.108-m ID (nominal 4 inch) axial fuel nozzle
was used for both ambient and hot WGA. The amount of radial flow was selected
to give flame lengths comparable to that of natural gas; ambient WGA was fired
with 10% radial flow, while hot WGA required 16% radial flow. Radial gas
velocities were sonic while axial fuel velocities varied from 33 m/s (107
ft/s) for ambient WGA to 74 m/s (244 ft/s) for hot WGA. Air velocity was
3.1 m/s (10 ft/s).
Only "dirty" doped, hot WGA was studied. The results are shown in
Table XIV. If the additives did not greatly affect thermal NO levels then
X
the baffle burner is somewhat more efficient in converting fuel-nitrogen
to NO than the kiln burner, namely, ~11% versus ~9%, respectively. This
is opposite to the WGO results, where the kiln burner appeared to be
slightly more efficient, in converting fuel nitrogen.
PARTICULATE STUDIES
Besides measuring the effects of the various dopants on gas-phase pol-
lutants, total particulates in the stack were measured as well. The instru-
mentation and sampling technique used were described in a preceding section.
Results of these total stack-particulate measurements are presented in Figure
8 for both WGO and WGA on the baffle and kiln burners. The fraction of
char-derived particulates surviving in the stack increased from ^5% at a
char input of 0.02 g/s to ~11% at a char input rate of 1.1 g/s. At all
char doping rates, it was visually noted that a qualitatively large number
of glowing particles were deposited on the furnace hearth. The presence of
ammonia and/or tar did not affect the measured total stack-particulates from
char.
50
-------
In addition to total particulates, particle size distributions were
measured for char-doped WGO on the baffle and kiln burners using a cascade
impactor and methods described previously. The results are shown in Tables
XV and XVI. On the baffle burner, with a char doping rate of 1.14 g/s, 63%
of the particles are smaller than roughly 0.7^1, while on the kiln burner,
for a char rate of 0.76 g/s, only about 5% of the particles are below 0.7^.
This shift in particle size depends more on char input level than on burner
type for the cases here studied. This conclusion is supported by the data
obtained from the total particulate measurements y wherein an increase in
"coarse" particles trapped in the cyclone preseparator relative to "fine"
particles staining the filter was observed with decreasing char doping rate
for both burners.
FURNACE EFFICIENCY
Aside from the pollution aspects of burning low-Btu fuel gases as
substitutes for natural gas, potential'users of low-Btu gas are also very
interested in retrofitting problems that might be encountered. One phase
of any retrofit evaluation is the effect of firing low-Btu gas on furnace
efficiency and heat transfer to the load. As noted in an earlier section,
water-cooling tubes were positioned along the furnace wall to simulate the
load of a steel reheat furnace for use with the baffle burner and then to
simulate the calcining and reaction zones of a cement kiln for use with the
kiln burner.
For the baffle burner, fired at 1.03 MW (3.5 X 106 Btu/hr) , furnace
thermal efficiencies, defined as the heat absorbed by the load divided by
the fuel heat input, for clean ambient and clean WGO were found to be 30%
and 31% as compared with the natural gas base-line value of 35%. Average
flue temperatures were measured to be 1541 K (2314°F) for ambient WGO, 1564 K
(2356°F) for hot WGO, and 1436 K (2126°F) for natural gas.
For clean ambient and clean hot WGA on the baffle burner, furnace ther-
mal efficiencies were 24% and 29% with corresponding average flue temperatures
of 1453 K (2156 °F) and 1504 K (2246°F).
With the kiln burner fired at 1.03 MW (3.5 X 10 Btu/hr), the base-
line natural gas efficiency was 31% with an average flue temperature of
1532 K (2298°F). Clean ambient and clean hot WGO yielded efficiencies of
29% and 30% with average flue temperatures of 1524 K (2284°F) and 1545 K
(2322°F). ...
-------
On the kiln burner, clean ambient and clean hot WGA gave furnace ther-
mal efficiencies of 24% and 28% with average flue-gas temperatures of
1479 K (2203°F) and 1553 K (2336°F).
In simulating the hot raw gasifier off-gas, the addition of dopants
to the clean fuel might be expected to affect overall furnace thermal eff-
iciency in two ways. First, the ammonia, char, and tar are fuels themselves
and will therefore affect the fuel heat input. Second, char and tar could
affect flame emissivity.
In the studies conducted, the level of contaminant doping was such that
the maximum contribution to the total fuel heat input was less than 6%. Flue
temperature measurements were essentially constant for a given fuel/burner
with and without doping, indicating that the doping levels employed did not
significantly affect fuel heat input. In any case, calculations of furnace
efficiencies included the contributions of doped material to the total
fuel enthalpy.
For ammonia additions of 1.0 volume percent, no effects on furnace
thermal efficiency were observed for WGO or WGA on either burner. Char
addition of 0.13 g/s (0.4 grains/SCF of WGO, 0.3 grains/SCF of WGA) also
had no effect on efficiency for both fuels on the burners studied.
At a tar feed rate of 0.58 g/s (1.8 grains/SCF of WGO, 1.4 grains/SCF
of WGA), furnace thermal efficiencies for WGO and WGA were increased by about
1.0% to 2.0% on both burners. This enhancement is probably due to the in-
crease in flame luminosity that was visually observed.
52
-------
DISCUSSION
FUEL-NITROGEN EFFECTS ON NO
x
In premixed flames, NO levels associated with thermal fixation of atmo-
spheric nitrogen are dependent primarily on flame temperature and, secondarily,
on the amount of combustion air (6). In turbulent diffusion flames, thermal
NO has also been found to vary with the initial degree of fuel/air mixing (5) .
For the fuels and burners studied, base-line thermal NO levels, presented in
X
Tables VI and XIII, are mainly ordered by adiabatic flame temperatures. Mixing
effects on thermal NO associated with the different aerodynamic character-
X
istics of the two burner types appear to be similar except for natural gas
where a higher thermal NO level is found with the baffle burner than with
X
the kiln burner, suggesting (after Reference 5) that natural gas/air mixing
is somewhat better on the baffle burner.
Another source of NO in combustion is chemically bound nitrogen in the
x
fuel. Since fuel-nitrogen bonds are much weaker than the bond in molecular
nitrogen, fuel-nitrogen can give rise to higher amounts of N0x than from
thermal fixation (5,6). In a flame environment, fuel-nitrogen is generally
believed to react through the competitive paths (6,4):
NH. + Ox = NO (1)
i
NH. + NO = N2 (2)
where NH. is some fuel-nitrogen intermediate, usually considered to be atomic
nitrogen (7) or a cyano or amine derivative (8), and Ox is an oxygen-containing
species such as 0, OH, or 02- In fuel-lean combustion, fuel-nitrogen appears
in the exhaust gases mostly as NO and N2; under fuel-rich conditions, signif-
icant HCN and NH- can also be found (8).
Factors that affect fuel-nitrogen conversion to NO are availability of
A.
oxygen, initial fuel-nitrogen concentration, temperature^ and general fuel
type (6,8). As the amount of oxygen available for combustion increases, the
conversion of fuel-nitrogen to N0x increases (9,6). Where fuel/air mixing is
incomplete, this conversion is strongly affected by local oxygen concentra-
tions as well as local flame temperatures (10). Combustion of fuel-nitrogen
under locally fuel-rich conditions can lower the amount of N0x formed (10) with
the fuel-nitrogen preferentially forming N2 (6). As initial fuel/air mixing is
53
-------
improved, the conversion efficiency of fuel-nitrogen to NO is enhanced (5).
X
For example, one group (11) reported a doubling in fuel-NO when going from
X
diffusional to premixed combustion.
With increasing levels of fuel-nitrogen, the fraction converted to NO
x
decreases (11,12) even though the absolute amount increases. Unlike thermal
NO , temperature does not greatly affect fuel NO in premixed flames (6),
A X
probably because overall fuel-nitrogen reactions are exothermic (5) and are
therefore less temperature dependent.
Recent results (8) imply that fuel-nitrogen conversion to NO depends on
X
fuel type; that is, ammonia conversion to NO in hydrocarbon combustion was
X
found to be much greater than with a hydrogen/carbon monoxide fuel. The
authors attributed this to a difference in the intermediate fuel-nitrogen
species (NH. in Equation 1), depending on whether the main fuel was a hydro-
carbon or not (8).
In order to gauge the magnitude of several of the above parameters on
fuel-nitrogen conversions to NO for raw gasifier effluents, char, tar, and
X
ammonia were added to hot clean Wellman-Galusha fuel gases. To ascertain the
contribution of each dopant to total NO emissions, single dopant and combin-
Ji
ation tests were also performed. The tests were performed using two industrial
burners: a forward flow and a kiln burner.
With char-doped hot WGO on both burners, we have seen (Table VII) that
total N0x levels were not greatly increased above thermal at the char feed
rates employed (maximum increase 30 ppm). The contribution of char-nitrogen
(0.66 weight percent) to these increments is difficult to interpret, owing to
the small change (if any) over thermal NO relative to measurement reproduc-
X
ibility (+ 5 ppm), and considering the known, but in this case unmeasurable,
enhancement of thermal NO by fuel-sulfur (1.64 weight percent of char) in
X
turbulent diffusion flames (9). Measured SO at the flue entrance accounted
for about 50% of the char-sulfur in all cases. Attempts to close this sulfur
balance were unsuccessful.
As with char, hot WGO tar doping results, shown in Table VIII, cannot be
unambiguously analyzed because of the presence of sulfur in the tar (0.47
weight percent). For the tar feed rates employed, measured NO exceeded the
X
clean thermal values by about 30 ppm on both burners. This increase in NO
x
54
-------
cannot be solely accounted for by tar-nitrogen (0.55 weight percent) even
assuming a 100% conversion to NO . The implication is that tar-sulfur is
jt
enhancing thermal NO . The contributions to total NO of sulfur-enhanced
x x
thermal and fuel NO cannot be apportioned. If all tar-nitrogen were con-
X
verted to NO , then sulfur enhancement of thermal NO could be up to 10 ppm
X X
over the clean, undoped value. Conversely, enhanced thermal NO could be
X
much higher, with fuel NO contributing relatively little to the observed
X
increase. Measured S0? accounted for only 20% of tar-sulfur in all cases,
while the tar residue trapped in the fuel nozzles accounted for another 10%
of the fuel-sulfur.
When fuel-sulfur is absent, as in the case of ammonia doped hot WGO, the
effects of fuel-nitrogen on NO are more clearly evident. Varying the amount
2v
of ammonia in the fuel shows that the fraction of fuel-nitrcgen converted to
NO decreases with increasing fuel-nitrogen content, as can be seen from
Figures 4 and 5, even though absolute NO' levels increased. This observation
A
is in agreement with the references cited at the beginning of this section.
For example, on the baffle burner at 10% excess air, ammonia doped at 1.0
volume percent of fuel input yielded a 7% conversion to NO , while 0.4%
X
ammonia gave a 14% conversion. With the kiln burner, the corresponding con-
versions were 8% and 16%.
Examination of Figures 4 and 5 shows that an increase in excess air,
from 10% to 20%, enhanced fuel-nitrogen conversion to NO as expected. On
X
the baffle burner with 1.0% ammonia in hot WGO, the conversion efficiency to
NO increased from 7% to 8% with the increase in excess air. At the 0.4%
x
ammonia level, the conversion went from 14% to 16%. Similar results were
obtained with the kiln burner. For 1.0% and 0.4% ammonia, the respective
increases in conversion with increased excess air were 8% to 10% and 16%
to 17%.
At the radial flow chosen to give the proper flame length for the kiln
burner (22% of the total hot WGO flow), fuel NO is only slightly higher than
X
for the baffle burner at comparable ammonia doping rates and excess air. In
other words, the anticipated mixing/aerodynamic effects of the different
burner types were not evident at the operating conditions employed. However,
changing the amount of radial flow drastically affected ammonia conversion to
55
-------
NO , as can be seen from Figure 7. Increasing radial flow from 22% to 36%
X
resulted in about a 50% increase in ammonia conversion, indicating that im-
proved fuel/air mixing raises fuel-nitrogen conversion, as expected from the
brief literature survey presented earlier (5). (This effect was also confirmed
when fuel-nitrogen conversion was found to be 50% higher on a highly mixed
high-forward-momentum burner than on the kiln and baffle burners in other
tests done at IGT.) Raising the radial flow from 0% to 15% lowered fuel NO .
x
In this region, the apparent loss and then recovery of fuel/air mixedness is
probably due to a trade-off between increasing radial mixing and decreasing
axial fuel momentum, the net effect of which is to decrease overall mixing up
to about 10% radial, where radial flow becomes the dominant mixing parameter
due to the radial flow penetration of the axial flow.
Since the ammonia-to-NO conversion is nearly the same for the baffle
A
burner and the kiln burner (22% radial) with the same doping rate, it may be
inferred that the baffle burner gave about the same degree of initial hot WGO/
air mixing as the kiln burner at 22% radial flow.
FUEL-SULFUR EFFECTS ON NO
x
As noted in the discussion of the char and tar results, fuel-sulfur is
known to affect thermal NO . In turbulent diffusion flames, characterized by
relatively poor initial fuel/air mixing, fuel-sulfur enhances thermal NO
X
while in premixed flames an inhibition occurs (9) . Besides thermal NO , fuel-
x
sulfur also affects fuel NO . In premixed flames it may enhance, inhibit, or
X
have no effect on fuel NO depending on the point of sampling and/or the
X
burning mixture's residence time in the combustion apparatus, while in turbu-
lent diffusion flames, fuel-sulfur has been found to enhance fuel NO (9).
X
In order to determine the effects of various levels of fuel-sulfur
(hydrogen sulfide) on fuel-nitrogen conversion to NO , hot WGO, doped with
X
1.0 volume percent ammonia, was fired on both burners with the results shown
in Figure 6. Neglecting fuel-sulfur/thermal NO interactions, the anticipated
enhancement of fuel-nitrogen conversion during turbulent diffusion combustion
is evident at hydrogen sulfide levels of 0.5 to 2.9 volume percent fuel input.
Fuel-sulfur effects on fuel NO are essentially the same for both burners.
X
This is not surprising, because the kiln burner, operated at 22% radial flow,
appears to give the same degree of initial hot WGO/air mixing as the baffle
56
-------
burner. This was also implied by the ammonia doping tests as previously
noted. Further, as fuel-sulfur levels are increased, the enhancement of NO
X
appears to reach a maximum, as suggested by Figure 6-
On both burners, measured SO,, corresponds to about 80% of the sulfur
input (as hydrogen sulfide) at all doping rates. The fate of the remaining
sulfur is uncertain. If it were present as some other species, a possible
candidate is 803. Although equilibrium considerations predict negligible
amounts of S0~ (13) in hot (T> 1300 K) combustion gases, relatively high con-
j
centrations of SO are possible under combustion conditions, where rapid
cooling of combustion gases can "freeze" SO (13,14) at superequilibrium values.
Even so, reported SO, levels are usually only a few percent (15), though levels
as high as 10% have been recorded (16). Such high levels are possible where
quenching of S0_ takes place by rapid cooling or by short residence time in
the combustion chamber.
With hot WGO fired on the baffle and kiln burners, combustion takes place
by turbulent diffusion, resulting in wide variations in local species concen-
trations and temperatures. The S0?, readily formed from the added H^S (17),
might form SO,, in two ways: a) by reaction with 0 atoms in high temperature,
fuel-lean regions; and b) in lower temperature regions (T <^ 1000 K) where the
right-hand-side of the equilibrium process, S02 + 1/2 0^ = S03> is not
negligibly small (13). Although the SO /SO approach to equilibrium is slow at
J £
lower temperatures (15), the presence of N0x can catalyze the formation of S03
by (15,18) -
NO + 1/2 02 * N02 (3)
N02 + S02 * S03 + NO (4)
For the tests performed with ammonia and hydrogen sulfide-doped hot WGO,
high concentrations of NO were present, making more plausible the possibility
35.
of high SO,. That SO levels measured were not severely depressed by some
artifact of the sampling system/instrumentation is supported by subsequent IGT
tests performed on a high-forward-momentum burner where measured S02 accounted
for 95% of the hydrogen sulfide added to a 1.0 volume percent ammonia-doped
low-Btu fuel (ambient WGO plus 25% NZ). Because the mixing characteristics of
this kind of burner are superior to the baffle or kiln (22% radial) burners,
one would expect less low-temperature formation of S03 if Reaction 4 is of any
57
-------
importance. The results are in good qualitative agreement with this tentative
mechanism, though more research on this possibility is required before any
definite conclusions can be made.
58
-------
REFERENCES
1. Shoffstall, D. R. and R. T. Waibel. Burner Design Criteria for NO
Control From Low-Btu Gas Combustion. EPA Final Report, EPA-600-7-^7-094b,
Dec. 1977.
2. IGT Process Research Division. HYGAS : 1964 to 1972, Pipeline Gas From
Coal Hydrogenation (IGT Hydrogasification Process). Final Report,
FE-381-T9-P3, Washington, D.C., July 1975.
3. Matthews, R. D., R. F. Sawyer and R. W. Schefer. ES&T, 11 (12): 1092,
1977.
4. Seery, D. J. and M. F. Zabielski. Combustion and Flame, 28: 93, 1977.
5. Appleton, J. P. and J. B. Heywood. Fourteenth Symposium (International)
on Combustion, The Combustion Institute, 1973. 777 pp.
6. DeSoete, G.G. La Rivista Dei Combustibili, 29: 35, 1975.
7. Baynes, B. S. Combustion and Flame, 28: 81, 1977.
8. Takagi, T., M. Ogasawara, M. Daizo, and T. Tatsumi. Sixteenth Symposium
(International) on Combustion, The Combustion Institute, 1977. 181 pp.
9. Wendt, J. 0. L., T. L. Corley, and J. T. Morcomb. Interaction Between
Sulfur Oxides and Nitrogen Oxides in Combustion Processes. Second
Symposium on Stationary Sources Combustion, New Orleans, Aug. 29 Sept. 1,
1977.
10. Sarofim, A. G., J. H. Pohl, and B. R. Taylor. Mechanisms and Kinetics
of NOX Formation: Recent Developments. 69th Annual Meeting, AIChE,
Nov. 30, 1976.
11. Lisauskas, R. A. and S. A. Johnson. NO Formation During Gas Combustion.
CEP, Aug. 1976. 76 pp.
12. Merryman, E. L. and A. Levy. Fifteenth Symposium (International) on
Combustion, The Combustion Institute, 1975. 1073 pp.
13. Sternling, C. V. and J. 0. L. Wendt. Kinetic Mechanisms Governing the
Fate of Chemically Bound Sulfur and Nitrogen in Combustion. EPA Final
Report, PB-230895, Aug. 1972.
14. Chigier, N. A. Prog. Energy Combust. Sci: 1, 3, 1975.
15. Cullis, C. F. and M. F. R. Mulcahy. Combustion and Flame, 18: 225, 1972.
16. Hedley, A. B. J. Institute of Fuel, 40: 142, 1967.
17. Wendt, J. 0. L. and J. M. Ekmann. Effect of Sulfur Dioxide and Fuel
Sulfur on Nitrogen Oxide Emissions. EPA Progress Report, Grant R-802204,
Sept. 1974.
18. Levy, A., E. L. Merryman, and W. T. Reid. ES&T, 4: 653, 1970.
-------
en
o
CLEAN SYNTHETIC
LOW-BtuGAS
ROTAMETERS
HYDROGEN
SULFIDE
FUEL
PREHEATER
SCREWFEEDER
PRESSURE
AMMONIA
W DIRTY" LOW-stu GAS
TO BURNER
TAR
Figure 1. Doping System to Synthesize "Dirty" Low-Btu Gases
-------
OBSERVATION PORT
FUELSN »- -V H-
N02ZLE ASSEMBLY BAFFLE
Figure 2. Assembly Drawing of Baffle Burner
61
-------
°J I-1/2 in.
7/8 i
X''
\ \ \\X\\\\\\
56- 11/16 in. J
/
3 in.
* r If y f 7\7\
GAS INLET GAS 1NLET
5/16 in.
Figure 3. Schematic Diagram of the Kiln Burner Fuel Injector
-------
15.0
o
t
U_
U.
o
o
2
£L
CJ
UJ
oc
5.0
1.0
O IO% EXCESS AIR
A 20% EXCESS AIR
1
i
QO
0.2 0.4 0.6 0.8
PERCENT NH3 IN FUEL
1.0
Figure 4. Ammonia Conversion for Wellman-Galusha Oxygen Fuel
Gas on the Baffle Burner with 10% and 20% Excess Air
63
-------
12.0
O 10% EXCESS AIR
20% EXCESS AIR
I.I
PERCENT NH3 IN FUEL
Figure 5. Ammonia Conversion for Wellman-Galusha Oxygen Fuel
Gas on the Kiln Burner with 10% and 20% Excess Air
64
-------
cn
O BAFFLE BURNER
A KILN BURNER
I
0.4
0.8
1.2 1.6 2.0
H2S IN FUEL,%
2.4
2.8
3.2
Figure 6. Effect of t^S on Ammonia (1.0% by Volume) Conversion
to NO with Wellman-Galusha Oxygen (10% Excess Air)
-------
800
3OO
20 30
RADIAL FLOW, %
Figure 7. Effects of Radial vs. Axial Flow on Ammonia (1.0%) Conversion
to NO with Wellman-Galusha Oxygen Fuel Gas on the Kiln Burner
X
66
-------
o»
(O
0.14
0.12
0.10
3 0.08
D
O
0.06
0.04
0.02
0.0
WGOWGA
O O BAFFLE BURNER
D A KILN BURNER
0.0 0.2 0.4 06 0.8
CHAR INPUT, g/s
1.0
1.2
Figure 8. Particulate Emissions for Char Doping of Wellman-Galusha
Fuel Gases on the Baffle and Kiln Burners (10% Excess Air)
67
-------
TABLE I. FUEL COMPOSITION FOR LOW- AND MEDIUM- Btu GASES TESTED
CT1
oo
Adiabatic
Heating Flame
Temp, H rn rv -a u n
Fuel
Wellman-Galusha
Oxygen
Wellman-Galusha
Air
* 10% excess air at 477 K (400°F). The adiabatic flame temperature for
ambient (298K, 77°F) natural gas is 2231 K (3356°F).
Temp,
°F
90
800
90
650
CO
39.2
29.7
26.9
25.2
H2
40.4
30.6
14.3
13.4
co2
.16.2
12.4
7.4
6.9
CH4
0.9
0.7
2.6
2.4
N2
1.4
1.1
46.9
44.1
n Value, Temp,*
2U Btu/SCF K (°F)
1.9
25.5
1.9
8.0
267
202
159
149
2248
2190
2045
2044
(3587)
(3483)
(3222)
(3220)
-------
TABLE II. TAR ANALYSIS
Ultimate Analysis
Ash
Carbon
Hydrogen
Sulfur
Nitrogen
Oxygen (By Difference)
wt % (Dry Basis)
0.0
84.39
5.65
0.47
0.55
8.94
General Analysis
Solids
Heavy Fraction
Light Fraction (Toluene,
Benzene, Xylene)
wt
JlejcejLyed_)_
27.1
55.3
17.6
TABLE IIL CHAR ANALYSIS
Ultimate Analysis
Ash
Carbon
Hydrogen
Sulfur
Nitrogen
Oxygen (By Difference)
wt % (Dry Basis)
22.88
66.30
1.75
1.78
0.72
6.57
Proximate Analysis
Moisture
Volatile Matter
Ash
Fixed Carbon
wt % (As Received)
7.8
13.6
21.1
57.5
Sieve Analysis
Screen
200
230
270
325
Pan
wt % Retained
67.7
3.3
4.9
3.6
20.5
69
-------
TABLE IV. SUMMARY OF TEST DATA FOR BAFFLE BURNER
Inputs
Fuel Type
Natural Gas
WGO* (322K, 10%)
Excess Air
WGO* (700K, 10%)
Excess Air
WGO* (700K, 20%)
Excess Air
WGO* (700K, 10Z)
Excess Air
Tar Doping
WGO* (700K, 10%)
Char Doping
WGO* (700K, 10Z)
Ammonia/Tar
WGO* (700K, 10Z)
Ammonia/Char
WGO* (700K, 10Z)
Char/Tar
WGO* (700K, 10Z)
Amnonia/Char/Tar
Rate
m3/s
0.027
0.111
0.140
0.134
0.134
0.134
0.134
0.134
0.134
0.134
0.134
0.134
0.134
0.1J4
0.141
0.141
0.141
0.138
0.138
0.138
0.139
0.138
0.138
0.138
0.136
NH3
%
0
0
0
0.19
0.38
0.60
0.81
1.03
0
0.19
0.38
0.60
0.81
1.03
0
0
0
0
0
0
0.99
1.00
0
1.00
1.02
H2S
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
Tar
g/i
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0.36
0.50
0.71
0
0
0
0.39
0
0.46
0.47
0.38
Char
;
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0.13
0.24
0.86
0
0.13
0.13
0.13
0.07
Flue-Gas Composition (as measured)
NO
112
38
30
310
360
400
420
430
38
290
355
410
435
465
52
55
58
30
33
44
520
470
40
500
510
H02
ppm
10
10
5
10
10
15
20
20
5
10
10
20
20
20
7
5
4
3
2
4
10
43
8
42
15
CO
10
200
115
55
55
53
55
60
45
45
45
40
40
40
108
93
93
59
100
90
108
65
63
75
24
C02
Z
10.9
24.0
24.0
24.0
24.0
24.0
24.0
24.0
22.0
22.0
22.0
22.0
22.0
22.0
24.0
24.0
24.0
24.0
24.0
24.0
24.0
24.0
24.0
24.0
24.0
02
1.9
1.8
1.8
1.9
1.8
1.9
1.9
1.9
3.3
3.3
3.2
3.2
3.2
3.3
1.8
1.9
1.9
1.8
1.7
1.9
1.7
1.8
1.9
1.8
1.9
SO,
ppm
1
1
2
4
6
29
2
4
7
8
7
HC
< 1
< 1
< 1
< 1
< 1
< 1
-c 1
< I
< 1
< 1
< 1
Particulate,
g/ro3 g/s
0.037 0.009
0.070 0.017
0.263 0.064
0.026 0.006
WGO* + 15Z FGR
(700K, 10Z)
Amoonia/Char/Tar
WGA* (616K, 10Z)
Anraonia/Char/Tar
0.138 1.00 0
1.22 0.13
510 20
45
24.0 1.9
4 < 1
WGO* (700K, 101)
Hydrogen Sulfide
WGA+ (322K, 10Z
Excess Air)
WGAf (616K, 10Z
Excess Air)
0.
0.
0.
0.
0.
134
134
136
175
175
1
1
1
0
0
.04
.04
.02
0
0,
2.
0
0
.02
.52
.53
0
0
0
0
0
0
0
0
0
0
480
655
950
16
22
29
33
50
1
2
40
50
37
260
35
24.0 1.8 83
24.0 1.8 2546
24.0 1.9 11146
18.4 1.3
18.4 1.4
0.186 1.04 0
0.58 0.35
610 20
40
18.4 1.4
8 < 1
0.044 0.016
* Wcllman-Galusha oxygen fuel gas.
* Wellman-Galusha air fuel gas.
+ Wellman-Galusha air fuel gas. Uses 3-inch nozzle. (All others used 2-1/2 inch nozzle.)
70
-------
TABLE V, SUMMARY OF TEST DATA FOR KILN BURNER
Fuel Type
Natural Gas
WG03'b (316 K, 10%
Excess Air)
WGO (705 K, 102
Excess Air)
WGO (705 K, 20%
Excess Air)
WGO (705 K, 10%
Excess Air)
Hydrogen Sulfide
WGO (705 K, ICtt)
Char Doping
WGO (705 K, 10Z)
Tar Doping
WGO (705 K, 10Z)
Ammonia/Tar
WGO (705 K, 10Z)
Ammonia/Char
WGO (705 K, 10Z)
Char/Tar
WGO (705 K, 10%)
Anmonia/Char/Tar
WGC + 15Z FGR
(705 K, 10Z)
Amonia/Char/Tar
WGAC (322 K. 10Z)
Clean
WGA (620 K, 10%)
Hot/Clean
WGA (620 K, 10%)
Ammonia/Char /Tar
Rate
m^/s
0.027
0.104
0.127
0.127
0.127
0.127
0.127
0.127
0.127
0.127
0.127
0.127
0.127
0.127
0.127
0.127
0.127
0.124
0.127
0.134
0.136
0.136
0.136
0.129
0.144
0.136
0.136
0.136
0.164
0.177
0.172
Inputs
NH
0
0
0
0.20
0.40
0.63
0.85
1.09
0
0.20
0.40
0.63
0.85
1.09
1.09
1.09
1.09
0
0
0
0
0
0
1.07
0.98
0
1.02
1.02
H,S
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0.02
0.54
2.89
0
0
0
0
0
0
0
0
0
0
0,
Tar
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0.
0.
0.
0.
0
0.
0.
0.
Char
ft 1 c
g/s
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0.13
0.29
0.67
38 0
60 0
73 0
63 0
0.13
48 0.35
25 0.35
57 0.13
NO
70
35
24
270
360
420
460
500
40
270
370
420
480
550
470
700
920
30
35
55
50
52
58
540
480
65
540
525
Flue-Gas
NOo
ppm -
5
5
4
20
30
30
40
40
5
20
30
30
40
50
30
40
80
5
2
5
5
3
4
40
30
5
40
30
CO
150
405
34
30
34
32
25
34
25
25
25
25
25
20
34
34
34
30
40
75
50
57
70
75
33
45
40
50
Composition
C02
10.
24.
24.
24.
24.
24.
24.
24.
22.
22.
22.
22.
22.
22.
24.
24.
24.
24.
24.
24.
24.
24.
24.
24.
24.
24.
7.
2
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
,0
24.0
24
.0
°2
2.1
1.9
1.8
1.8
1.8
1.8
1.8
1.8
3.3
3.3
3.3
3.3
3.3
3.3
1.7
1.7
1.7
1.8
1.9
1.7
1.8
1.8
1.7
1.9
1.8
1.9
1.8
1.8
(as measurei
S02
pplr.
--
--
--
_-
--
_..
--
--
90
2579
12800
3
6
10
1
1
2
1
3
6
7
4
HC
_.
<1
<1
<1
<1
<1
<1
-------
TABLE VI. BASE-LINE DATA FOR CLEAN FUELS:
NATURAL GAS AND WELLMAN-GALUSHA OXYGEN (WGO),
3.5 MILLION Btu/hr WITH 10% EXCESS AIR
Fuel NO * CO
Fuel Type Temperature. K ppm
Baffle Natural Gas 298 135 10
Burner WGO 322 53 200
WGO 700 38 115
Kiln Natural Gas 298 83 150
Burner WGO 322 44 405
WGO 700 31 34
NO plus N0« (Dry, corrected to 0% excess air).
-------
TABLE VII. CHAR DOPING:
WELLMAN-GALUSHA OXYGEN (700 K),
3.5 MILLION Btu/hr WITH 10% EXCESS AIR
Baffle
Burner
Kiln
Burner
Char Input,
g/s
0
0.13
0.24
0.86
0
0.13
0.29
0.67
NO
X
38
36
38
53
31
38
41
66
CO
ppm
115
59
100
90
34
30
40
75
so2
0
4
6
29
0
3
6
10
NO plus NO (Dry, corrected to 0% excess air).
TABLE VIII. TAR DOPING:
WELLMAN-GALUSHA OXYGEN (700 K) ,
3.5 MILLION Btu/hr WITH 10% EXCESS AIR
Baffle
Burner
Kiln
Burner
Tar Input,
8/s
0
0.36
0.50
0.71
0
0.38
0.60
0.73
*
NO
X
38
65
66
68
31
60
60
68
CO
ppm -
115
108
93
93
34
50
57
70
so2
0
1
1
2
0
1
1
2
NO plus NO (Dry, corrected to 0% excess air).
73
-------
TABLE IX. AMMONIA CONVERSION TO NO
ON THE BAFFLE BURNER 5
A
% in Fuel
0.02
0.11
0.19
0.38
0.60
0.81
1.03
1.42
0.19
0.38
0.60
0.81
1.03
*
Dry,
Excess Air
10
10
10
10
10
10
10
10
20
20
20
20
20
corrected to 0%
, Fuel NO , NH,
' x 3
ppm
77
213
312
367
416
443
454
509
305
382
459
489
524
excess air.
Conversion,
52
29
26
15
11
9
7
6
26
16
13
10
8
TABLE X. AMMONIA CONVERSION TO NO
ON THE KILN BURNER X
% in Fuel
0.03
0.10
0.20
0.40
0.63
0.85
1.09
0.20
0.40
0.63
0.85
1.09
*
Excess Air
10
10
10
10
10
10
10
20
20
20
20
20
Fuel NO , NH3
ppm X
123
243
287
396
462
516
560
291
421
480
563
658
Conversion
60
38
23
16
12
10
8
23
17
12
11
10
Dry, corrected to 0% excess air.
74
-------
Table XI. PARAMETRIC DOPING:
WELLMAN-GALUSHA OXYGEN (700 K),
3.5 MILLION Btu/hr WITH 10% EXCESS AIR
Baffle
Burner
Kiln
Burner
Inputs
NH-,
vol %
0
1.03
0.99
1.00
0
1.09
1.07
0.98
Tar
rr 1 c
g/s
0
0
0.39
0
0
0
0.63
0
Char
0
0
0
0.13
0
0
0
0.13
NO *
x
38
492
580
561
31
591
635
558
CO
ppm
115
60
108
65
34
34
75
33
SO,
0
0
2
7
0
0
1
3
NO plus NO- (Dry, corrected to 0% excess air).
TABLE XII. PARAMETRIC AND "DIRTY" DOPING:
WELLMAN-GALUSHA OXYGEN (700 K),
3.5 MILLION Btu/hr WITH 10% EXCESS AIR
Baffle
Burner
Kiln
Burner
NH3,
vol %
0
1.03
0
1.00
0
1.09
0
1.02
Inputs
Tar
g/s-
0
0
0.46
0.47
0
0
0.48
0.25
Char
0
0
0.13
0.13
0
0
0.35
0.35
NO,
38
492
53
593
31
591
77
635
CO
ppm
115
60
63
75
34
34
45
40
S02
0
7
8
0
0
6
7
NO plus NO (Dry, corrected to 0% excess air).
75
-------
TABLE XIII. BASE-LINE DATA FOR CLEAN FUELS:
NATURAL GAS AND WELLMAN-GALUSHA AIR (WGA),
3 5 MILLION Btu/hr WITH 10% EXCESS AIR
NO plus N02 (Dry, corrected to 0% excess air).
NO
x
TABLE XIV. "DIRTY" DOPING:
WELLMAN-GALUSHA AIR (616 K) ,
3.5 MILLION Btu/hr WITH 10% EXCESS AIR
CO
Baffle
Burner
Kiln
Burner
Fuel iype
Natural Gas
WGA
WGA
Natural Gas
WGA
WGA
298
322
616
298
322
616
ppm
135
18
26
83
28
22
10
260
35
150
200
30
Inputs
Baffle Burner 0
Kiln Burner 0
NH
%3'
0
1.04
0
0.99
Tar
0
0.58
0
0.45
Char
rr le-
' g/S "
0
0.35
0
0.13
NO "
X
26
672
22
522
CO
ppm
35
40
30
35
SO
2
0
8
0
7
NO
plus N0? (Dry, corrected to 0% excess air),
76
-------
TABLE XV. BAFFLE BURNER CASCADE IMPACTOR
RESULTS FOR CHAR-DOPED WELLAMN-GALUSHA OXYGEN (700 K) -
DOPING RATE: 1.14 g/s
0
1
2
3
4
5
6
7
Filter
10.7
7.3
5.0
3.2
1.8
1.1
0.7
0
> 17.1
-17.1
- 10.1
- 7.3
- 5.0
- 3.2
- 1.8
- 1.1
- 0.7
TT jr . ,__
11.1
1.4
0.4
2.9
2.5
8.9
4.7
5.6
62.5
*
Cumulative
88.9
87.5
87.1
84.2
81.7
72.8
68.1
62.5
0
Wt % of particulates passing through stage.
TABLE XVI. KILN BURNER CASCADE IMPACTOR
RESULTS FOR CHAR-DOPED WELLMAN-GALUSHA OXYGEN (700 K)~
DOPING RATE: 0.76 g/s
*
Stage Size (pi) wt % % Cumulative
51.5
46.6
37.4
30.2
24.4
12.3
4.5
4.3
0
~~
0
1
2
3
4
5
6
7
Filter
7.9
5.3
3.6
2.4
1.2
0.7
0.5
0
> 12.6
- 12.6
- 7.9
- 5.3
- 3.6
- 2.4
- 1.2
- 0.7
- 0.5
48.5
4.9
9.2
7.2
5.8
12.1
7.8
0.2
4.3
Wt % of particulates passing through stage.
77
-------
SOME ASPECTS OF AFTERBURNER PERFORMANCE
FOR CONTROL OF ORGANIC EMISSIONS
By:
Richard E. Barrett
and
Russell H. Barnes
Battelle-Columbus Laboratories
Columbus, Ohio 43201
79
-------
SOME ASPECTS OF AFTERBURNER
PERFORMANCE FOR EMISSIONS CONTROL
ABSTRACT
The initial phase of this program is intended to conduct an emission
assessment of afterburner control systems based on available data. This
phase would be a prelude to intended laboratory and field experimental ef-
forts.
This paper reports on a portion of the Phase I environmental assess-
ment. Firstly it reports on the use of existing data to estimate the poten-
tial national usage of afterburners, based on emissions. Secondly, it re-
ports on an evaluation of field test data from the files of one local air
pollution control agency. Results of the analyses show that in-service
afterburners appear to be less efficient than are units reported on in much
of the literature. The lower afterburners efficiency has little impact on
national organic nonmethane emissions, but may have marked impact in local
areas.
80
-------
SOME ASPECTS OF AFTERBURNER
PERFORMANCE FOR EMISSIONS CONTROL
INTRODUCTION
Organic emissions are of concern because of their participation in re-
actions leading to the production of oxidants and, because in certain locales,
further reactions produce irritating smogs. One type of device that can be
utilized for the control of hydrocarbon or organic emissions from some sta-
tionary sources is the afterburner or fume incinerator.
Afterburners can be applied to some, but not all organic emission
sources. Generally they can be applied to organic emission sources having a
well defined and contained emission stream, such as chemical, metallurgical,
surface coating, and agricultural processes; they cannot readily be applied
to sources having scattered and uncontained organic emissions, such as
burning landfills and coal refuse piles, pipe leaks, and uncontained venting.
Among the sources to which they are applied are:
Resin kettels
Varnish cookers
Sulfuric acid manu-
facturing
Phosphoric acid manu-
facturing
Paint-bake ovens
Wire-coating process
Soap and synthetic
detergent industries
Glass manufacture
Frit Smelters
Food Processing Equip-
ment
Fish canneries
Animal-matter rendering
Electroplating
Insecticide Manufacture
Oil and solvent refining
Chemical milling
Coffee roasting
Meat smokehouses
Fertilizer plants
Rotogravuring
Degreasing operations
Dry Cleaning
Fiberboard drying and curing
81
-------
With such widespread use of afterburners, it is important to national
and local air quality efforts to understand just how successful afterburners
are at controlling emissions. Unfortunately, the overall performance of
afterburners is frequently based on limited research studies, or on tests of
new units. These test measure afterburner performance with well-tuned units
operating at peak efficiency. Consequently, efficiency values above 90 per-
cent or 95 percent, and sometimes as high as even 99 percent, are usually
reported. However, there remains the strong suspicion that performance of
typical in-service afterburners is not as good as that reported in the ide-
alized tests. Because afterburners are not income-producing devices, it is
suspected that their maintainence, etc., receives less than adequate atten-
tion. This paper considers the extent of application of afterburners, and
reports on afterburner performance based on results of a limited number of
afterburner emission tests conducted by a local air pollution control agency.
APPLICATION OF AFTERBURNERS AS RELATED
TO NATIONAL HYDROCARBON EMISSIONS
Based on the work of others^ ' as summarized in Table 1, it is estima-
ted that 106,548,900 metric tons of volatile organics are emitted in the U.S.
each year from stationary (and natural) sources. Of this total, 25,212,400
MT/yr (metric tons per year) are volatile nonmethane organics. Further ex-
amining the total emissions of organics, these emissions may be divided be-
tween natural and man-related sources as follows:
Volatile organics
Natural sources 85,300,000 MT/yr (80.1 percent)
Man-related sources 21,248,900 MT/yr (19.9 percent)
Volatile nonmethane organics
Natural sources 9,100,000 MT/yr (36.1 percent)
Man-related sources 16,112,400 MT/yr (63.9 percent).
Limiting the analysis to volatile nonmethane organics, 62.2 percent
of these emissions are from sources not considered amenable to the applica-
tion of air pollution control devices (e.g. natural sources, fossil fuel
82
-------
extraction, open burning). The remainder, about 37.8 percent of the volatile
nonmethane organic emissions, originate from processes amenable to the appli-
cation of emission control devices. Figure 1 graphically illustrates the
above distribution of volatile nonmethane organics between natural sources
and between man-related sources for which emission control devices are con-
sidered applicable or not applicable.
Further considering the volatile nonmethane organic emissions from
sources for which control devices are applicable, Reference 1 estimates that
about 75.7 percent of these emissions (28.6 percent of total volatile non-
methane organic emissions) could be controlled with the application of con-
trol devices. The remaining 24.3 percent of such emissions are estimated to
be emitted due to control devices being less than 100 percent efficient.
Figure 1 shows these data graphically.
Based on data in Reference 1, afterburners are potentially applicable
to sources representing 18.3 percent of all volatile nonmethane organic
emissions. These sources emit 4,618,500 MT/yr of such emissions. By as-
suming an afterburner efficiency of 90 percent for nearly all afterburner ap
plications, the authors of Reference 1 calculated a control efficiency of
87.9 percent for all sources amenable to the use of afterburners. Thus, for
sources for which afterburners are applicable, the controlled volatile non-
methane organic emissions are 4,057,800 MT/yr and the uncontrolled emissions
are 560,700 MT/yr. Figure 2 repeats the data shown on Figure 1 but includes
a further breakdown of emissions from controllable sources into sources that
are controllable with afterburners and sources that are not controllable with
afterburners.
It should be recognized, however, that afterburners would not be used
on all sources to which they might be applied; other control devices might
be selected due to lower cost or better compatibility with the process. Thus,
if afterburners were applied to one-half of those sources amenable to the
application of afterburners, the national figures for afterburner controlled
sources would be:
83
-------
inlet emissions 2,309,250 MT/yr
outlet emissions 280,350 MT/yr
controlled emissions 2,028,900 MT/yr
EVALUATION OF AFTERBURNER TEST RESULTS
An evaluation was made of the results of 73 field tests on afterburners
conducted by a local air pollution control agency. These tests were con-
ducted over a period of 14 years and do not appear to be limited to tests of
newly installed units. In fact, a few of the tests were made as a result of
citizen complaints regarding the emission source. Hence, these tests would
appear to be more representative of in-service afterburner performance than
are most other sources.
The results of these tests were revealing in the poor afterburner per-
formance recorded for many of the tests. Figure 3 shows the distribution of
afterburner efficiencies based on emissions of total nonmethane organic
species. The median efficiency was about 76 percent. About 38 percent of
the tests gave afterburner efficiencies of 90 percent or higher. Another 18
percent of the tests gave efficiencies of 70 to 90 percent, and 25 percent
of the tests gave efficiencies from 0 to 70 percent. Finally, 19 percent of
the tests recorded afterburner efficiencies below zero percent, that is,
outlet emissions exceeded inlet emissions.
Figure 4 shows the distribution of afterburner efficiencies for the
same tests but based on reactive organics. (Reactive organics include aro-
matics, phenols, carbonyls, and organic acids.) Afterburner efficiencies
for reactive organics were lower than for total nonmethane organics; the
median efficiency was only 50 percent. Twenty-nine percent of the tests
gave efficiencies, based on reactive organics, or 90 percent or above; an-
other 14 percent were in the 70 to 90 percent range; and 25 percent were
in the 0 to 70 percent range. Finally, 33 percent of the tests gave negative
efficiencies.
84
-------
The negative efficiencies based on reactive organics is understandable
in that, if reactive organics are only a small fraction of total organics at
the inlet, an ineffective afterburner could convert a fraction of the non-
reactive organics into reactive organic species and produce a negative ef-
ficiency. A negative efficiency based on total nonmethane organics is not
so easily explained. Efficiencies that are only slightly negative (as a
number were) might be explained by limits in the accuracy of organic sampling
and analysis and/or of flow rate measurements. However, efficiencies that
are negative by more than about 30 percent cannot readily be explained in
this way. A few afterburners were operating at off-design conditions (1000 F
to 1100 F conbustion zone temperatures versus 1300 F to 1500 F normally used);
such units may have been producing nonmethane organics from the fuel gas due
to poor combustion conditions.
Figure 5 is a cross tabulation of afterburner efficiencies for both
total nonmethane organics and reactive organics. From Figure 5, it can be
seen that 30 tests (56 percent) reported about the same efficiencies for both
total and reactive organics. Also, 16 tests (29 percent) showed a higher
efficiency for controlling total nonmethane organics and 8 tests (15 per-
cent) showed a higher efficiency for controlling reactive organics. The
large degree of scatter of data shown in Figure 5 shows that if, both total
and reactive organics are considered important, it is necessary to test for
each as the efficiency for controlling one type of emission may not permit
estimation of control efficiency for the other type of emission.
IMPACT OF LESS EFFICIENT AFTERBURNERS
The impact of afterburners being less efficient than is generally
assumed for enviornmental assessment studies has been evaluated, both on a
national and a local basis.
Considering the national viewpoint, Table 1 and Figure 2 illustrates
national emissions of volatile nonmethane organics from stationary sources
with an assumed afterburner efficiency of 90 percent. National emissions of
85
-------
volatile nonmethane organics from sources utilizing afterburners was
estimated to be 280,350 MT/yr. However, if afterburners are only 76 percent
efficient (as shown earlier), instead of 90 percent efficiency, the outlet
emission could be 596,000 MT/yr. This value is 315,650 MT/yr greater than
the former value. Hence, based on the assumptions that:
1. the test data available are representative of the
performance of all afterburners
2. afterburners are applied to 50 percent of the sources
for which they are considered amenable,
volatile nonmethane organic emissions from afterburner controlled sources are
presently underestimated by 315,650 MT/yr on a nationwide basis. This value
is equal to about 1. 3 percent of total national volatile nonmethane organic
emissions, and equal to about 2.0 percent of national man-related volatile
nonmethane organic emissions. Such an error in estimating national emissions
is probably not significant, as many of the values used in compiling Table 1
were not known to such accuracies.
Now, considering the impact on specific locales, it can reasonably be
assumed that, for well populated areas, nearly all volatile nonmethane or-
ganic emissions are from sources amenable to controls. That is in populated
areas uncontrolled sources such as open burning and solid waste disposal to
other than incinerators would be absent and natural sources would be present
in much smaller proportions then in the nation as a whole. Considering that
afterburners are applied to one-half of the stationary sources in a populated
area, the impact of poorer afterburners performance (76 percent versus 90
percent) would be an underestimation of volatile nonmethane organic emis-
sions from stationary sources by up to 27 percent, depending on the distri-
bution of types of industry. Such an error could have a significant impact
on local air pollution control strategies, and in the evaluation of air pol-
lution control efforts.
86
-------
CONCLUSION
It has been shown that afterburners are candidate organic emission con-
trol devices for application to emission source emitting about 29 percent of
the volatile nonmethane organic emission generated by man-related stationary
sources. Further, the analysis of afterburner efficiency conducted as a
part of this program shows that typical in-service afterburners are probably
considerably less efficient then are the well tuned units used in the re-
search studies normally reported in literature.
It has been shown that the error in estimating volatile nonmethane or-
ganic emissions due to the demonstrated poorer performance of afterburners
is probably not significant of a national basis (causing an error of about
2 percent), but may be very significant on a local basis (causing an error
that may exceed 27 percent). Hence, at least as it affects local areas,
it is important that the effort to understand and interpret the actual per-
formance of in-service afterburners be continued.
REFERENCES
1. Cavanaugh, E.G., Owen, M.L., Nelson, T.P., Carroll, J.R. 3nd Colley,
J.D. . Hydrocarbon Pollutants from Stationary Sources. EPA-600-7-77-
110, U.S. Environmental Protection Agency, Washington, D.C., 1977.
318 pp.
87
-------
TOTAL
VOLATILE
NONMETHANE
ORGANIC
EMISSIONS
25,212,400 MT/yr
100%
MAN-RELATED
SOURCES
16,112,400 MT/yr
63.9%
SOURCES
AMENABLE
TO CONTROL
9,542,200
37.8%
NATURAL
SOURCES
9,100,000 ilT/yr
36.1%
SOURCES NOT
AMENABLE
TO CONTROL
6,570,200
26.1%
CONTROLLED
EMISSIONS
7,220,300
28.6%
UNCONTROLLED EMISSIONS
2,321,900 9.2%
TOTAL
UNCONTROLLED
EMISSIONS
17,992,100
71.4%
FIGURE 1. NATIONAL VOLATILE NONMETHANE ORGANIC
EMISSIONS FROM STATIONARY SOURCES
-------
00
MAN-RELATED
SOURCES
16,112,400 MT/yr
63.9%
25
TOTAL
VOLATILE
NON;1ETHANE
ORGANIC
EMISSIONS
,212,400 !1T/yr
100%
NATURAL
SOURCES
9,100,000 MT/yr
36.1%
TOTAL
UNCONTROLLED
EMISSIONS
17,992,100
71.4%
a. SOURCES AMENABLE TO CONTROL WITH AFTERBURNERS; 4,618,500 MT/yr; 18.3%
b. SOURCES AMENABLE TO CONTROL BUT NOT WITH AFTERBURNERS; 4,923,700 MT/yr; 19.5%
c. CONTROLLED EMISSIONS FROM SOURCES AMENABLE TO CONTROL WITH AFTERBURNERS; 4,057,800 MT/yr; 16.1%
d. CONTROLLED EMISSIONS FROM SOURCES AMENABLE TO CONTROL BUT NOT WITH AFTERBURNERS; 3,162,500 MT/yr;
12.5%
e. UNCONTROLLED EMISSIONS FROM SOURCES AMENABLE TO CONTROL WITH AFTERBURNERS; 560,700 MT/yr; 2.2%
f. UNCONTROLLED EMISSIONS FROM SOURCES AMENABLE TO CONTROL BUT NOT WITH AFTERBURNERS, 1,761,200 MT/yr;
7.0%
FIGURE 2. NATIONAL VOLATILE NONMETHANE ORGANIC EMISSIONS FROM STATIONARY
SOURCES SHOWING AFTERBURNER CONTROLLABLE EMISSIONS
-------
TOO
90
80
70
60
50
40
30
20
10
0
PI
n n n n
>0 0/10 20/30 40/50 60/70 80/90
10/20 30/40 50/60 70/80 90/100
UNIT EFFICIENCY, PERCENT
FIGURE 3. AFTERBURNER EFFICIENCY BASED ON TOTAL ORGANICS
90
-------
a:
UJ
D-
100
90
80
70
60
50
40
30
20
10
0
n n
n n
>0 0/10 20/30 40/50 60/70 80/90
10/20 30/40 50/60 70/80 90/100
UNIT EFFICIENCY, PERCENT
FIGURE 4. AFTERBURNER EFFICIENCY BASED ON REACTIVE ORGANIC5
91
-------
EFFICIENCY FOR DESTRUCTION OF REACTIVE ORGANICS. PERCENT
1
UJ
on
UJ
Q.
t/}
z
QC
0
UJ
z
jc
UJ
z:
0
J
I
°
U_
o
0
I-H
h-
CJ
OS
LU
Q
a:
O
n
o
z
UJ
ti
o
U-
u_
UJ
90-
100
80-
90
70-
80
60-
70
50-
60
40-
50
30-
40
20-
30
10-
20
fl-
it)
<0
on- 80- 70- 60- 50- 40- 30- 20- 10- 0- Q
100 90 80 70 60 50 40 30 20 10
^ 1 1 1 1 1 i i i ' i
X 14\ 2 -
\ \
\ \
. 2 \ 2 \ 1 1
\ \
\ 1 1 12-
\I \
\ \ 16 TESTS; 29*
\ \'
\V
1 ^ > 1
\"
N
\ 1\
\ \
\ \ 2 .
\ \
1 1 1 \ 1 \ 1
\ \
\ *v
8 TESTS; 15% \ ^\
^V y v
\ ^ \
\ \
2 \10 \-
\
i i ii ii i ii i\
FIGURE 5. COMPARISON OF AFTERBURNER EFFICIENCIES
FOR TOTAL AND REACTIVE ORGANICS
92
-------
TABLE 1. SUMURY or NATIONAL ORGANIC EMISSIONS
FKOK STATIONARY SOURCES*
ID
Total Atmotpherlc Emlailoni
(MT/yr)
IV
b
b
Storage, & Dlatrtbution
Fossil Fuel Refining
Volatile
2 510 000
2 071 000
2,173,500
Volatile
Nonne thane
2 071 000
2,173,500
Organic
7,300
77,300
269,000
* c
Fossil Fuel Combustion
VI Fossil Fuel Feedstock
Chemical Processing
VII Noncombustlon Organic
Chemical Utilization
VITI Agricultural 6, Forest
Products
IX Open burning Sources
(agricultural 6, prescribed
forest burning)
724,000 383,900
1,400,000 1,077,000 45,800
3,529,000 3,529,000
508,000 508,000 3,324,000
3,010,000 3,010,000 973,000
X Natural Sources 85,300,000 9.100,000 1,500,000
XI Solid Waste Disposal £ 2,690.000 2,443,000 640,000
XII Municipal Seuage Disposal8 -
XIII Other Sources1'
9]7,000
9J7.000 234,000
106,548,900 ?5,212,400 7,040,400
Emissions from Processes Amenable to Controls
Emissions from All Processes
percentage
percentage
Emissions from Processes Amenable to Control with Afterburners
Emissions from All Processes
Controllable Emissions from Processes Amenable to Controls
Total Emissions from Processes Amenable to Controls ' percentage
Controllable Emissions from Processes Amenable to Afterburner Controls
Total Emissions from Processes Amenable to Afterburner Controls
Uncontrolled Eaititoni fro* Ptocvisat
Am*n«bl« to Air Pollution Control!
Controllable Enlnloni from Sources
Amenable to Controls
Percwlta8e
a) Data compiled from "Hydrocarbon Pollutants from Stationary Sources", EPA Report No, EPA 600/7-77-110,
September, 1977
b) Emissions are mostly methane and considered a low pollution hazard
c) Controllable fraction is 1C engine sources
d) Only control is elimination of source
e) Not controllable
Ao. ruble
Volatile
Orcanlca
2,510,000
1,714,000
2,071,000
2,173,500
(117,000)
317,000
1,400,000
(1,117,000)
3,529,000
(3,162,000)
508,000
(371,000)
115,500
(115,500)
14,338,000
(4,882,500)
13.5
(4,6)
to Control with
(HT/yr)
Volatile
Nona-thane
Organic
-
-
2,071,000
2,173,500
(117,000)
68,200
1,077,000
(853,000)
3,529,000
(3,162,000)
508,000
(371,000)
115,500
(115,500)
9,542,200
(4,618,500)
37.8
U8.3)
: No. EPA 600/7-77-110,
Afterburners)
Organic
Partftculetea
-
7,300
77,300
269,000
(0)
-
45,800
(45,800)
-
3,324,000
(582,000)
108,800
(108,800)
3,832.200
(736,600)
54.4
(10.5)
(Controllable Emission! from Sources
Amenable to Control with Afterburners)
(MT/yr)
Volatile
Volatile Nonmethane
Organic! Organic!
1,830,000
1,030,000
1,363,000 1,363,000
1,400,000 1,400,000
(116,000) (116,000)
314,000 67,500
1,270,000 953,000
(1,106,000) (844,000)
2,868,000 2,868,000
(2,666,000) (2,666,000)
504,000 504,000
(167,000) (367,000)
64,800 64,800
(64,800) (64,800)
10,643,800 7,220,300
(4,319,800) (4,057,800)
74.2 75.7
(88,5) (87. 9)
Organic
^articulate!
-
3,650
69,600
243,600
(0)
-
40.900
(40,900)
-
3 , 300 , 000
(576,000)
75,000
( 75.00U)
3.654,100
(691,900)
97.4
(93.9)
f) Except for Incinerators, only control is elimination of
sources (uncontrolled open burning)
g) No appreciable organic air emissions
h) Major sources are destructive fires and are not considered
controllable
-------
DEVELOPMENT OF EMISSION-CONTROL METHODS
FOR LARGE-BORE STATIONARY ENGINES
By:
Robert P. Wilson, Jr.
Arthur D. Little, Inc.
Acorn Park
Cambridge, Massachusetts 02140
95
-------
ABSTRACT
The research work presented herein was undertaken in order to develop
combustion modifications which substantially reduce N0x emissions of large-
bore engines, and which result in equivalent or lower fuel consumption and
carbonaceous emissions. The scope of the project covers NO control techno-
X
logy for diesel and spark ignition engines, bore sizes ranging from 8 to 20",
and both 2 and 4 cycle charging methods.
In Phase I, a compendium of 40 emission control concepts was prepared,
including methods which have shown promise for automotive engines. We also
developed new methods using fundamental assumptions about the pollutant
formation processes in spark ignition and diesel combustion environments.
In Phase II, a ranking procedure was used to screen down the list to those
concepts that are most promising and, therefore, suitable for testing in
Phase III. The primary tool used in Phase II was a mathematical simulation
of the combustion and pollutant formation process in spark gas engines.
Also important in the selection was the practical feasibility, side effects,
retrofit feasibility, and relative cost of each concept.
In Phase III, under major subcontracts, Cooper Energy Services and
Fairbanks-Morse will use single and dual cylinder laboratory engines to
test selected emission control methods. In Phase IV, the methods which prove
effective in the laboratory will be applied to several engines operating in
the field. Systematic measurements of BSFC and emissions will be made over
an extended period in order to demonstrate the level of NO reduction and the
reliability of the control technology. The program will conclude with an
assessment of the costs and benefits of widespread adoption of the control
methods.
This paper has been prepared under Contract No. 68-02-2664 by Arthur
D. Little, Inc. under the sponsorship of the U.S. Environmental Protection
Agency, covering work completed during the first 12 months of the program.
96
-------
ACKNOWLEDGEMENTS
The project team at Arthur D. Little, Inc. includes Elia Demetri,
W. David Lee, Philip G. Gott, Donald Hurter, William Raymond, and
Arthur Fowle. Professor William J. McLean of Cornell University
directed the development of the combustion model for spark ignition
engines. Also we wish to acknowledge the assistance of Dr. Douglas Taylor
of Ricardo, Professor Adel Sarofim of Massachusetts Institute of Technology,
Professors Philip Myers and Gary Borman of the University of Wisconsin,
Professor Lou Conta (University of Rhode Island), Charles Netwon and
Eugene Kasel of Fairbanks-Morse, and Fred Schaub and Mel Helmich of
Cooper Energy Services.
The Diesel Engine Manufacturers Association (DEMA) has been cognizant
of major aspects of this project starting with the initial proposal in 1976.
DEMA not only has an obvious interest in engine emissions R&D, but also is
uniquely able to assist the program based on practical experience.
97
-------
SECTION 1
INTRODUCTION AND ENGINE DESCRIPTION
INTRODUCTION
The effort described herein deals with a category of stationary combustion
engines which is generally out of the public eye but which contributes a
significant fraction of the total air pollution burden, particularly to the
NO total. There are currently over 10,000 stationary reciprocating engines
X
operating in the U.S. in the 8-20" bore class, ranging from 80 to 700 HP/cylin-
der. The estimated annual fuel use by these engines is 1.1 quads, or 1.5% of
the U.S. energy budget; however, the cumulative NO emissions are dispropor-
tionately high at about 5% of the U.S. total. Among the many applications of
these large reciprocating engines, the most significant single category is the
gas pipeline compressor function, typically a two-cycle turbocharged engine of
14-20" bore. Other important categories in terms of fuel use are the gas
gathering and electric generator engines of somewhat smaller bore size. Table I
summarizes the population of stationary large-bore engines.
All of the combustion conditions which make large-bore engines relatively
highly efficient prime movers (some operate as low as 6000 Btu/BHP-hr) also
make them produce relatively high levels of NO (about 4 Ib/MMBtu) : high flame
X
temperature due to compression preheating, low wall heat losses, air- fuel ratio
in the range most conducive to NO formation, and relatively extended duration at
peak temperature due to low rpm. Interest in developing emission controls for
large bore engines has recently intensified, as evidenced by the proposed EPA
standards (equivalent to about 9 gNO /BHP-hr, or a 30% reduction from current \-
levels) and by recent hearings ef the California Air Resources Board related
to even lower target NO levels. The present study is part of a long-range
X
research program to develop combustion modifications for large bore engines
for the 1985-1990 time frame.
98
-------
Large-Bore Engine Characteristics
The combustion process of the large-bore engines being considered here is
distinguished from that in the more familiar spark ignited automotive engine
by several characteristics. First, both the speed (300-1200 RPM) and the
cylinder dimensions (8-20" bore) of the large-bore engines are a factor of 3
to 7 different than automotive engines. Second, most low-speed spark-ignited
engines employ direct cylinder injection of gaseous fuel. Since gas injection
terminates only 60° CA before TDC, the possibility of imperfect fuel/air mix-
ing must be considered. The limited available information on the details of
the combustion process indicates that the pre-combustion mixture of air,
residual burnt gases, and gaseous fuel is not uniform, but rather exhibits a
spatial non-uniformity in fuel-air ratio which we refer to as "unmixedness."
Although the charge is non-uniform, all portions of the charge are reached by
the flame, leaving little uriburned fuel. Hydrocarbon emissions are typically
only 0.5-2.0 g/BHP-hr for spark gas engines. In the two-stroke engines, the
swirling flow of low turbulence level (induced by the loop scavenging process)
apparently assures that the mixing process is adequate if not complete.
Most two-cycle gas engines are adjusted to operate quite fuel lean by
automotive engine standards; a mean fuel-air equivalence ratio between 0.7 and
0.8 is not uncommon. Reasons for using lean mixtures are improved cycle
efficiency (lower SFC), moderation of pressure rise rates, and prevention of
knock due to end gas autoignition. Such lean mixtures, of course, require a
strong ignition source, particularly for methane fuel. In this regard it is
interesting to speculate that the presence of richer than average regions of
charge near the spark source due to unmixedness may promote more reliable
ignition.
While the low turbulence intensity and large cylinder size in the gas
engines lead to combustion durations in the range 5-15 msec, the relative
duration in crankangle degrees is only 10-30° CA because of the low engine
speed. Thus, at 330 RPM, a 20° CA combustion duration and a spark advance of
about 10° BTDC is typical. The advance in spark timing is limited by the
onset of knock, which is attributed to small quantities of less knock-
resistant higher paraffins (such as butane) in natural gas, which is largely
99
-------
composed of highly knock resistant methane.
With respect to formation of nitric oxide, the conditions described above
could hardly be better selected for maximum NO production. Experience with
automotive engines indicates that maximum NO is to be expected for the lean
high-temperature mixture conditions produced in the large-bore engines. Futher-
more, the low surface-to-volume ratio results in near-adiabatic conditions.
The low engine speed allows sufficient time at high temperatures for the rate
controlling 0 + N * NO + N reaction of the Zeldovich mechanism to produce
nitric oxide concentrations which approach equilibrium levels in the hottest
portions of the charge. As in the automotive engine, these high NO cone en-
X
trations are preserved during the expansion stroke because the NO decomposi-
tion reactions are relatively slow in the temperature range which is character-
istic of expansion.
The current NO emissions of large-bore engines are clustered in the
X
12-15 g/BHP-hr range. This is equivalent to approximately 4 Ib/MM Btu, which
is unusually high for combustion devices, as illustrated in the following
table:
NO EMISSION RATE OF LARGE BORE ENGINES
x
COMPARED TO OTHER COMBUSTION DEVICES
Device
Large Bore Engines
Automotive SI Engine
Coal Fired Utility Boilers
Industrial & Commercial Boilers
Relative Fuel
Use, Nationwide
1
14
15
17
Typical NO Emission
(Ib/MM^tu)
4
2
0.7
0.4
(Oil-Fired)
Industrial Furnaces 12 0.3
(Gas-Fired)
Residential Furnace and
Water Heater 10 0.1
The factors which couple high NO to high efficiency comprise the basic
X
constraint on emission control techniques: efforts to reduce NO are likely
X
to reduce efficiency as well, unless efforts are made to keep heat release
100
-------
near TDC. Simple methods to reduce NO , such as EGR and spark retard, are
X
known to have BSFC penalties; the effort of the proposed program will be on
developing NO control techniques which potentially do not have a BSFC penalty,
X
such as modified fuel/air preparation, chamber shape modifications, and NO
X
decomposition.
Program Methodology^
The work described herein was undertaken in order to develop and
screen potential emission control concepts prior to engine tests. The aim of
Phase I has been to generate an investory of existing and new emission control
concepts for large-bore stationary engines. The scope included both spark
gas engines as well as diesel engines in the medium speed (300--1200 rated
RPM) range. This inventory of concepts was then subjected to a critical
screening process in Phase II, where the concepts were compared based on
their potential merit, considering emissions reduction, effect on brake speci-
fic fuel consumption (BSFC), practical feasibility, and ultimate cost to
users. In order to estimate the effect of various NO control measures on
X
both emissions and fuel consumption, it was necessary to construct a model
which would adequately account for the conditions peculiar to large-bore spark
gas engines. It was considered particularly important to include the effect
of unmixedness or dispersion in the local fuel-air ratio, as this feature of
large-bore engines had not been generally included in models of automotive
engine combustion. A description of the model is presented in Wilson et al(1979).
Since the focus of the program is on exploring the feasibility of
achieving substantial NO reductions for future engine designs, emphasis was
X
placed on combustion modification concepts, such as torch ignition, strati-
fied charge, water-fuel emulsions and novel injection systems rather than on
"external" adjustments alone, such as EGR and timing. Exhaust gas treatment
and combinations of concepts were also studied. In Wilson (1978), each concept
for emission control is described with schematics showing the manufacturers'
preferred and alternative configurations.
101
-------
SECTION 2
EMISSION CONTROL MECHANISMS FOR
- SPARK IGNITION ENGINES
COMBUSTION COW)ITIOWS AND MO PRODUCTION IN SPARK-IGNITION (SI) ENGINES
X
In a large-bore SI engine, the main portion of the charge can be assumed
to be premixed, and therefore combustion is believed to occur by propagation
of a turbulent flame front outward from the spark. The burned gas at the
flame front is thought to engulf eddies of fresh mixture by a turbulent mixing
process. Ignition sites appear on the boundaries of the fresh mixture eddies
and then a flame can be considered as traveling across each eddy at the
laminar flame speed, as suggested in Figure 2-1. The heat released by this
process serves to propagate the flame in two ways: (a) the turbulent mixing
of reactants into products is enhanced by sudden expansion of eddies follow-
ing ignition and (b) the heated combustion products ignite entrained elements
of reactants.
The NO formed in an SI engine is essentially related to the conditions
in the hot gas left in the wake of the flame front. Each successive element
of combustion gas which is produced by the passing flame front generates NO
X
depending on its initial flame temperature, fuel-air ratio, rate of com-
pression heating, and rate of quenching. Figure 2-2 shows the initial NO
production rate in ppm/msec as a function of temperature and fuel-air ratio.
NO formation rates of 100 ppm/msec or greater are practically unavoidable
X
(for at least very short times) because, as shown in Figure 2-2, the adiabatic
flame temperatures predicted for typical engine conditions (800°K compression
temperature, 40 atm) are on the order of 2500°K depending on fuel-air ratio.
The nitric oxide accumulated by a given element of combustion products will
be the time-integral of the formation rate along a temperature history, as
depicted by the path shown in Figure 2-2. Most critical to the cumulative
NO emission are two factors:
102
-------
The conditions (flame temperature, fuel-air ratio) at which each
mixture element ignites.
The subsequent temperature history, particularly the compression
heating (which is the greatest for earliest burned elements), and
the subsequent rate of cooling.
The first few elements to burn are thought to produce relatively high NO ,
J*.
because they subsequently undergo greatest compression heating and experience
a relatively long residence time at high temperature.
Under certain conditions, some decomposition of NO can occur during
the expansion stroke. This decomposition is driven by the difference between
actual NO level and equilibrium NO level.
GENERAL APPROACHES TO NITRIC OXIDE CONTROL
The various techniques for reducing the level of NO emitted by an engine
2v
can be visualized in terms of the ($, T) histories of burned products. In
Figure 2-3, three basic categories of NO -suppression techniques are illustrated:
X
stratified combustion, lean combustion, and temperature suppression.
Stratified Combustion
In the stratified charge engine, the fuel is intentionally maldistrib-
uted or layered so that one segment of the mixture is lean (<£ ~ 0.6) and the
other segment is rich (cj) = 1.2-1.3) during and just after the combustion pro-
cess. Both segments are outside of the fuel-air equivalence ratio range
<(> = 0.8 to 1.1 where NO production rates are 100 ppm NO /msec or greater.
x x
After combustion the burned gas segments intermix at a point well into the
expansion stroke, producing the overall average value of $. It is also impor-
tant to burn the > 1.2 segment first. The fuel-rich segment is set-up in
the vicinity of the spark plug, so that the elements burned as the flame
moves out from the spark are subject to low-NO formation rates (due to lower
X
temperature and the lack of oxygen). Igniting this fuel-rich segment has a
second advantage in that righ gas is more suitable for ignition than the over-
all charge if it were homogenous.
For the large bore engine, stratified charge is not expected to produce
dramatic NO reductions because in current turbocharged engines the mixture
X
is typically already outside of the critical 4> = 0.8-1.1 range. For example,
103
-------
starting from overall * - 0.75, division into segments of 1.0 and 0.5 actually
is expected to increase the NO emission. This is borne out by the predictions
X
shown in Figure 2-4, which show a 27% ?K»x increase for a 1.0-0.5 stratified
mixture. The explanation is that the = 1,0 segment adds more N0x than is
subtracted by the <|> = 0.5 segment. Bluaberg (1972) corroborates these predic-
tions with his model for small-bore engines, showing an increase of NO from
A
1500 ppm to as high as 2600 ppm for the 0% EGR case. Blumberg emphasizes that
for lean-burn engines (such as the large bore spark gas type), stratification
in fact becomes quite risky since M) can increase from improper stratifica-
A
tions which place near-stoichiometric segments in the first elements to burn.
Lean Combustion
Here the mixture is made to burn more fuel lean than normal ( = 0.6-
0.7), which reduces the exposure of products to high NO-production rates
accordingly. Available data suggests that a 40% NO reduction can be obtained
X
by shifting the equivalence ratio from 0.75 to 0.65. The computer model
developed to simulate combustion and NO formation in a large-bore engine was
X
applied to a representative 20" bore engine in order to project the NO reduc-
X
tions achievable by increasing air-fuel ratio. The results are shown in
Figure 2-5. As the air is increased from a normal condition at $ = .78 to the
"lean" condition at $ = .70, the NO was predicted to drop from 3600 to 2600
X
ppm (28%). Experimental data from a single cylinder engine generally agrees
with the predictions on the effectof excess air, but show a slightly greater
reduction (35%).
The basic problem does not lie with the technology for making the mixture
lean. An increase in the trapped air mass in the cylinder can be accomplished
by increasing the degree of turbocharging. The air mass is increased rather
than the fuel decreased in order to maintain BMEP or engine power. The basic
problems with burning mixtures at 4> =0.6 to 0.7 are stimulating reliable
ignition (preventing misfire) and limiting the combustion duration. Therefore,
each lean combustion concept involves a remedial measure for misfire and com-
bustion duration. One of the most promising concepts for minimizing combustion
duration under lean conditions is torch ignition, as illustrated in the follow-
ing figure:
104
-------
Cylinder
Auxiliary Fuel
Spark Plug
Source: Fairbanks-Morse.
FIGURE 3-4 TORCH IGNITION DEVICE FOR 2 CYCLE GAS ENGINE
Temperature Suppression
By cooling the intake air or adding inerts, such as water vapor or EGR,
the peak flame temperature is reduced. This shifts the NO-production rates to
lower values. These concepts are also applicable to diesel engines.
NO Decomposition
X . . _
In addition to these three categories of techniques, N0x decomposition
can be considered. In the late stages of expansion and in the exhaust manifold,
the equilibrium NO-levels are less than a few hundred ppm. The actual NO level
(which is a few thousand ppm) is relaxing toward this level at a negligible
rate. Methods for enhancing or catalyzing the NO decomposition process were
considered. There are two distinct approaches to carrying out N0x decompo-
sition, each of which has been demonstrated for industrial boilers in Japan
(where NO regulations are severe):
Gas Phase Reaction: Conducted at high-temperature (1000-
(1400°K or 1300-2000°"F). This approach if applied to
engines would require no catalyst, but would require costly
reheating of the exhaust gas to at least 1300°F, partic-
105
-------
ularly for 2-stroke engines. Expected NO reductions
X
are in the 50-60% range for a 2:1 NH_/NO mole ratio,
according to Lyon and Longwell (1976) and Muzio et al
(1977).
Catalyst-Induced Reaction; Conducted at moderate
temperature (500-650°K or 440-700°F). This approach
if applied to engines would require costly catalyst replace-
ment every 1-3 years (about $10/HP for platinum) and would
entail some pressure drop. The expected NO reductions,
X
however, are higher (80-90%) and 25% less ammonia is needed
(NH : NO ratio of 1.5), according to Bartz (1977).
A 3C
Both approaches require a costly ammonia storage and injection system.
At this writing, these approaches are receiving attention in connection with
possible California NO standards.
X
The following table summarizes the emission control methods which have
been compiled for large-bore spark ignition engines in this study.
Category
Concepts Considered
Potentially Practical
Concepts Considered Impractical
Lean
Combustion
Torch ignition
Multiple spark plugs
Increased turbulence
High-energy spark
Diesel fuel ignition
Feedback control w/0,,
Shock wave ignition
Hydrogen enrichment
Catalytic piston
Optical ignition
Pyrophoric jet ignition
sensor
Stratified Divided chamber stratification
Combustion Open chamber stratification
(Auxilliary injection port
or carburetor)
Degraded mixing
Temperature
Suppression
Charge refrigeration
Retarded timing
EGR
LNG injection
Intake water injection
Increased engine speed
Decomposition
Ammonia reduction agent
Nitrogen plasma injection
Ozone injection
Chemical absorption
Metal exhaust catalyst
Catalytic piston
Modified cooling rate
106
-------
IMPACT OF COMBUSTION MODIFICATION ON FUEL CONSUMPTION
Stratified combustion, lean combustion, and temperature suppression
(the basic techniques of NO-suppression illustrated in Figure 2-3) all have
the effect of (a) delaying the establishment of a vigorous propagating flame
and (b) lowering the rate of combustion which, in turn, can adversely affect
BSFC and carbonaceous emissions. The timing of an SI engine is normally
adjusted so that the peak pressure occurs just after top center, so that the
heat released by the burned fuel appears as close as possible to top center
(before piston movement). That portion of the heat which is released to the
gases during the expansion stroke (denoted "late burning") does not con-
tribute fully to the work. The heated gases from "late burning" expand through
only a portion of the volume change. The efficiency loss due to extended com-
bustion duration has been studied by Lyn (1960). Figure 2-6, taken from
Lyn's findings, suggests that the efficiency penalty is about two percentage
points for each 10°CA extension of the combustion duration. Partial compen-
sation for this penalty can be obtained by advancing the ignition point.
The significance of BSFC degradation cannot be overestimated, since the
fuel costs for large-bore engines represent a substantial portion of the over-
all costs. In this project we have arbitrarily adopted a goal of achieving
maximum NO reductions within a 3-4% BSFC penalty. The basis for this goal
3t
is the engine adjustments such as retarded timing alone can substantially
reduce NO but only with a 6-8% fuel penalty [see Youngblood et al. (1978)].
107
-------
SECTION 3
EMISSION CONTROL MECHANISMS FOR DIESEL ENGINES
COMBUSTION CONDITIONS AND NO PRODUCTION IN DIESEL ENGINES
x
The formation of NO in a diesel engine must be described in terms of
X
the sequence of combustion processes which occur as the spray mixes and burns
in the chamber. Following the excellent descriptions put forth by Lyn (1962)
and Austen and Lyn (1960), the four stages of combustion are as follows:
Compression and preignition mixing
Ignition and flame propagation
Spray combustion
Residual combustion and mixing
These processes are more complex than in the premixed charge of an SI engine,
because of spray formation, fuel-air mixing and radiation. In the following
sections, we will describe each stage and will comment on the factors affect-
ing NO -formation.
0 x
Compression and Preignition Mixing
At the start of fuel injection, the air has been heated to 800°K* by
compression and endowed with turbulence due to piston motion, "squish," and
swirl. Liquid fuel is injected into the large-bore diesel starting at about
10°BTDC*, and ignition occurs about 7°CA (3 msec)* later, after a portion of
the fuel has evaporated, mixed with air, and undergone a chemical ignition
delay which is more or less characteristic of the compression temperature.
According to Andree and Pachernegg (1969), ignition occurs whenever the
evaporated fuel has been exposed to a temperature surplus (over the reference
ignition temperature, for a certain integral time (/ AT dT > 335°K-msec),
*Temperatures and crank angles quoted are typical; actual values for specific
engines vary widely.
108
-------
There is no NO formation up to the ignition point; however, the subsequent
X
NO production is highly influenced hy such factors as compression ratio,
charge precooling, timing of injection, and turbulence.
Ignition and Flame Propagation
The first phase of combustion is relatively brief, having a high heat
release rate which is often termed the "spike" on a plot of heat release rate
versus time (see Figure 3.1). The reason that this first phase proceeds so
swiftly is that multiple ignition sites occur and flame kernels propagate
through all the portions of the chamber which contain a premixed flammable
air fuel charge. The high-speed photographs of Rife and Heywood (1974)
suggest that first combustion occurs in the zone where the vapor plume impinges
with the wall. For the large-bore diesel we estimate that about 5-10% of the
heat release is "prepared to burn" (evaporated and mixed) and attributable to
the "spike." A disproportionate amount of NO , however, is thought to be
X
formed in regions ignited by the spike for the same reasons that the zones
near the spark plug of SI engines exhibit high NO :
X
Earliest-produced combustion products have longest residence
time at high temperature.
Subsequent compression heating of earliest zones leads to
elevated temperature and NO rates.
X
The nature of premixed combustion makes the spike-affected
zones more adiabatic; i.e., the quenching rate is limited
because neighboring elements are burnt products rather than
cooler air.
If a large number of fuel elements are all suddenly ignited and burned,
the resulting high rate of pressure rise increases NO by compression heating
of burned gases.
The first guideline for avoiding high NO combustion conditions, there-
2C
fore, is to minimize the heat release by the spike; that is, to minimize the
accumulated premixed fuel vapor which is available for burning at the time
of first ignition. Retarded fuel injection accomplishes this, because the
higher compression temperatures at the retarded crank angle shorten the
109
-------
ignition delay. Fumigation, water emulsion, pilot injection, and cetane
number modifications to the fuel are optional methods which can be considered
to shorten ignition delay. Changes in fuel injectors will also affect the
amount of fuel vapor available at the fuel-air ratio most conducive to igni-
tion (4> = 1.1).
Spray Combustion
Overlapping with the flame-propagation stage is the second stage of
combustion which is characterized by mixing-controlled combustion of the
fuel droplet array and unmixed fuel vapor pockets. These fuel elements are
dispersed into the product/air mixture by swirl, by wall impingement, and
by spray penetration; and they burn as fast as the droplet array can evaporate
and find air. For the large-bore medium speed diesel engine, the photographic
evidence of Taylor and Walsham (1970) suggests that the spray is vaporized
completely before it reaches the wall, but that wall impingement of the
vapor plume occurs. Rife and Heywood (1974) concluded that mixing (and there-
fore the combustion rate) is controlled by the rate of air entrainmant by the
fuel spray or vapor plume, rather than the ballistic dynamics of single drop-
lets. The burning history of each fuel element resembles a downward ramp as
shown in Figure 3-1. The cumulative heat released from all elements follows
the familiar "Wiebe" rate, again shown in Figure 3-1. This stage of com-
bustion lasts 10-15 msec, depending on the duration of fuel injection and the
rpm.
Nitric oxide production during this mode of combustion depends on the
distribution of fuel-air ratio within the various mixing-limited subzones, as
illustrated in Figure 3-2. The mixing-limited subzones can be described by a
fuel-air ratio distribution which presumably centers around stoichiometric
fuel-air ratio, based on fundamental considerations about diffusion flames.
The guideline for avoiding high-NO producing combustion conditions for this
second stage of burning is to reduce the residence time of diffusion burned
zones at temperatures above 2200°K. This can be approached through (a) reduc-
ing the post-combustion compression by retarded injection timing (coupled with
increased mixing to insure that burning is complete prior to expansion),
,(b) forcing off-stoichiometric conditions such as occur in a prechamber engine,
and (c) increasing swirl or turbulence to lean out the hot burnt gases with air.
110
-------
Product-Air Mixing and Residual Combustion
In this final stage of combustion, the elements of combustion products
continue to intermix with each other and with the remaining air in the chamber.
Any residual fuel vapor, CO, or hydrogen found in elements with insufficient
air are burned to CO,, and H_0 in the process of mixing. The entire cylinder
contents gradually become more uniform in excess air and temperature. A given
burned gas element is quenched from 2600°K to below 2200°K in about 0.3 msec
for the high-rpm engines simulated by Murayama et al (1977). During this 0.3
msec the NO is being produced at a rate of about 1000-3000 ppm/msec according
X
to Figure 3-2.
NO production in this stage can be minimized by increasing the mixing
3t
rate so that relatively cool air can quench NO-production elements. It is
important that this increased mixing rate be timed to occur after combustion
rather than before combustion. Otherwise the heat release rate (and NO ) in
X
stage I will increase.
Summary
In summary, the unique combustion factors of large-bore diesel engines
are as follows:
Larger fuel orifices (.028" to .008") lead to larger drop sizes
(SMD 22\i versus 15u).
More fuel jets (10 versus 4).
Low Swirl.
Less heat release in the premixed stage (spike)approximately 10%.
Longer duration at near-constant volume (600 rpm versus 2400 rpm) .
Lower compression ratio (12 to 14).
Higher degree of turbocharging (up to 280 BMEP).
Leaner full load operation (4> = 0.5 versus = 0.7).
Peak pressure more critical (structural-welded, not cast).
Lower surface area per unit volumeless specific heat transfer
and larger quench fraction.
Ill
-------
APPROACHES TO NO CONTROL FOR DIESEL ENGINES
x
To summarize the discussion up to this point, NO reduction is best
X
achieved for both premixed (stage 1) and spray combustion (stage 2) by retard-
ing the injection and compensating for this by reducing the ignition delay and
increasing the mixing rate. Particular attention must be given to minimizing
the time spent by the initial early-mixed charge at 2200-2600°K (near-stoichio-
metric flame temperature).
Approaches to NO control for diesel engines are in the table below.
X
Category
Concepts Considered
Potentially Practical
Concepts Considered Impractical
Modified
Fuel
Injection
Water-fuel emulsion
High injection rate
Pilot injection
Bi-fuel injection
Air-assist atomization
Altered
Mixing
Pattern
Prechamber-variable area
Modified air motion/chamber
shape
Circumferential injection
Initial charge modification
Temperature
Suppression
Charge refrigeration
EGR/retarded timing
Intake water injection
Increased engine speed
NO
x
Decomposition
Ammonia reduction agent
Nitrogen plasma injection
Ozone injection
Chemical Absorption
Metal exhaust catalyst
Catalytic piston
Modified cooling rate
112
-------
Water-Fuel Emulsions
Data for small-bore diesel suggests that a 50% water/fuel ratio by
volume can reduce NO in the exhaust by about 50% when injected as an emulsion
x
(see Figure 3.3). Large-bore engines have not been operated at water-fuel
ratios above 25%, however, the results are promising (see Figure 3-4). The
introduction of water as an emulsion in diesel fuel is thought to lower N0x
production by the following mechanisms: First, latent heat and enthalpy is
absorbed by the water, which lowers the local temperature of the products of
combustion, lowering the rate of NO production. In addition, it has been
X
hypothesized that the atomization process is altered by so-called "micro-
explosions" of emulsified droplets. It is not clear to what extent this
affects NO significantly.
X
Practical problems of emulsions which must be addressed include the
following:
(a) The injection nozzle and cam must be changed if necessary to
accommodate the increased volume of fluid to be injected, while
maintaining injection duration and atomization quality. An
increase in the number of holes is one option.
(b) Preventing "slugs" of water from separating out of the emulsion
Non-emulsified water in fuel systems has been responsible for severe
damage to piston crown and cylinder heads.
(c) The effect of the fuel oil-water mixture on the durability of the
engine and the availability of water is of concern. Lestz et al
(1975) indicate little adverse effect in fuel components even at
existing high water injection rates (300-500% water/fuel ratio).
Demineralized water is required to minimize deposit build-up,
apparently.
(d) Prevention of excessive cylinder pressure by retarded timing.
113
-------
SECTION 4
COMPARISON OF EMISSION CONTROL METHODS
COMPARISON OF METHODS BASED ON POTENTIAL NO REDUCTION
A
Since the primary objective of the project is emissions reduction, the
most critical consideration in comparing the candidate concepts is what frac-
tion of the normal NO emission can probably be eliminated by each concept.
X
Fuel consumption (BSFC), however, is also an important consideration. It is
well established that 40% NO reduction can be achieved on most engines by
A
applying methods which entail 6-8% BSFC penalty (e.g., timing and EGR).
Therefore, in order to take BSFC into account while placing primary emphasis
on NO , we have projected the maximum feasible NO reduction without exceed-
X X
ing 4% BSFC penalty. Based on underlying mechanisms of hydrocarbon and CO
emissions, it is felt that 4% BSFC limit would also constrain these emissions
to acceptable levels.
Three sources of information were used to project NO reductions:
A.
Emission test data for previous experimental attempts to apply
the concepts to large and small-bore engines (50% weighting
factor).
Engineering judgment by those experienced in combustion and
emissions control (25% weighting factor).
Mathematical model predictions using a computer simulation of
a large-bore spark gas engine (25% weighting factor).
These sources of information were weighted as noted and a consensus projection
was reached in this manner. Table II lists the projected NO reductions
using this procedure. Note that when engine data was not available, the model
projections and "engineering judgment" received equal weight.
114
-------
Excluding the four exhaust treatment methods, relatively little variation
apparently exists in the NO -reduction potential, based on Table II (the
X
range is 16 to 37%). There were no concepts in the range of 40-60% reduction.
Reaching this target, however, is still a rational goal for combustion
modification, because:
These projections are reasonably conservative. For example,
the lean operation could have been taken at (J> = 0.60-0.65
where greater NO reductions would be expected. Small-bore
X
engine applications and isolated large-bore experiments have
shown 50-70% reductions are possible.
Combinations of techniques may prove feasible.
COMPARISON OF EMISSION CONTROL CONCEPTS BASED ON FEASIBILITY AND COST
In order to identify those concepts with enough promise to be advanced
to engine experimentation, in addition to NO reduction projections, a critical
review was made of each concept which involved quantitative cost factors, such
as:
added costs of the modified engine;
cost of developing the concept;
added maintenance costs;
other added costs of operating the modified engine.
In addition, there are qualitative factors having to do with practical
feasibility which were also considered, as follows:
applicability of the concept to important engine classes and
major manufacturers' engines (weighting factor, WF = 7);
practicality of retrofitting engine in the field ,(WF = 5);
adverse side effects anticipated, such as noise, corrosion,
odor, or poor starting (WF = 3);
the need for special materials (WF = 2);
operating complexity, reliability, and the feasibility of
unattended operation without unscheduled breakdown (WF = 7);
115
-------
effect of reciprocating engines' salability in competing against
gas turbines (WF = 3);
whether manufacturers are already implementing the concept on
their own on a proprietary basis, so that EPA efforts would be
redundant (WF = 5).
A ranking of concepts according to feasibility was carried out by a formal
procedure. Seven engine manufacturers and Ricardo were contacted for practical
engineering judgments regarding the feasibility of 22 emission control concepts.
Weighting factors were applied to each criterion and the scores were normal-
ized so as to permit an overall figure of merit or "feasibility index."
Strengths and weaknesses of each concept were identified relative to other
concepts. The principle findings were, as follows:
Thet« was no concept which could absolutely not be implemented
on at least one engine type.
There was relatively little variation in overall feasibility
among the 22 concepts. Numerically, on a acale of 0 to 1.00,
the feasibility index ranged from .62 to ,81 for spark
ignition engines and from .56 to .89 for diesel engines. Each
concept appears to have offsetting strengths and weaknesses.
Engine manufacturers' greatest concerns centered on unscheduled
downtime (reliability), requirements for maintenance"of sophis-
ticated electronic components, and avoiding storage and handling
of additional fluids (e.g., liquid nitrogen, demineralized water
for emulsion, or chemicals for scrubbing).
In summary, apart from cost-effectiveness considerations, provided that con-
cerns about reliability and "second fluid" can be met to manufacturers'
satisfaction, it would be difficult to rule out any of the concepts based on
feasibility or lack thereof.
OVERALL RANKING CONSIDERING NO REDUCTION POTENTIAL, ADDED COSTS, AND
FEASIBILITY X
The pertinent figures of merit for each of the concepts are presented in
Table III by engine type. Note that there are certain concepts with noticeable
discrepancies when ranked by different criteria, for example:
116
-------
NO Cost Feasibility
Torch Ignition 1 65
Turbulence 71 1
Degraded Premix 52 7
Charge Refrigeration 2 10 1
Emulsion 16 5
This points out the need to balance or tradeoff the three considerations in
deciding which concepts are to be tested in Phase III. It must be emphasized
that the feasibility indices are quite closely grouped, and both the NO and
3C
the cost projections are uncertain to about a factor of a third the value
quoted in either direction (i.e., 30% means 20-40%). An additional consider-
ation is the relative cost to experiment with the concept in Phase III.
Figures 4-1, 4-2 and Table IV combines the figures of merit in three ways,
each of which may be useful to the decision:
Cost-effectiveness (ANO * Acost)How much NO is reduced for
X X
the added operating cost? Figure IV displays the results.
Feasibility-effectiveness (ANO x feasibility)How much is NO
X X
reduced for a given degree of practical feasibility? Figure 4-2
displays this index.
All factors (ANO x feasibility * Acost)This is one of many
conceivable ways to combine the three indices and can be
recommended only for its simplicity. Table IV displays this
index.
The bias of the third (combined) figure of merit can be assessed, by examining
the rankings of the five concepts mentioned above, as follows:
Torch ignition
Turbulence
Degraded premix
Charge refrigeration
Emulsion
Nfr
Alone
1
7
5
2
1
Cost
Alone
6
1
2
10
6
Feasibility
Alone
5
1
7
1
5
All Factors
Combined
2
1
4
8
5
117
-------
REFERENCES
Andree, A., and Pachernegg, S.J., "Ignition Conditions in
Diesel Engines," SAE Paper 690253, Detroit, Michigan,
January 1969.
Austen, A.E.W. and Lyn, W.T., ''Relation Between Fuel Injection
and Heat Release in a Direct-Injection Engine and the Nature
of the Combustion Processes," Proceedings of the Institute of
Mechanical Engineers, p. 47, January 1960.
Bartz, D.R., "Catalyzed Ammonia Reduction of NOx Emissions from
Oil-Fired Utility Boilers," Paper P-195, NOX Control Technology
Workshop, Asilomar, California, October 1977,
Blumberg, P.N. and Kummer, J,T., "Prediction of NO Formation in
Spark-Ignition EnginesAn analysis of Methods of Control,"
Comb. Sci. and Tech., Vol. 4, p. 73, 1971.
Blumberg, P.N., "Nitric Oxide Emissions from Stratified Charge
Engines: Prediction and Control," Comb, Sci and Tech., Vol 8,
p. 5, 1973.
Dale, J.D., Smy, P.R., and Clements, R.M., "The Effects of a Coaxial
Spark Igniter on the Performance of and the Emissions from an
Internal Combustion Engine," Combustion and Flame ^1 173, 1978.
de Soete, G., "Etude Parametrique des Effects de la Stratification i
de la Flamme Sur les Emissions d'Oxydes d'Azote," Insitut Francais
due Petrole, 32, p. 427, May-June 1977.
Dietzman, H.E., and Springer, K,J,, *:Exhaust Emissions f,rom Piston and
Gas Turbine Engines Used in Natural Gas Transmission," Southwest
Research Institute, AR-923, January 1974.
Herrmann, R., and Magnet, J.L., "Reduction de la Pollution Atmospherique
par les Gas d'Echappement des Moteurs Diesel," CIMAC Conference, 1977.
Kasel, E.A. and Newton, D.L., "U.S.C.G, Pollution Abatement Program:
Icebreaker Smoke Reduction," DOT/USCG Report CG-D-179-75, 1975,
Khan, I.M. and Greeves, G., "A Method of Calculating Emissions of Soot
and Nitric Oxide from Diesel Engines," SAE Paper 730169, 1973.
Kuroda, H., Nakajima, Y., Sugihara, K., Takagi, Y,, and Muranaka, S,,
"The Fast Burn with Heavy EGR, New Approach for Low NOX and Improved
Fuel Economy," SAE Paper 780006, February 1978.
Lyn, W.T., "Calculations of the Effect of Rate of Heat Release on the
Shape of Cylinder-Pressure Diagrams and Cycle Efficiency,"
Proc. Auto. Div., Inst, Mech. Engr.., No, 1, p. 34, 1960.
Lyn, W.T., "Study of Burning Rate and Nature of Combustion in Diesel
Engines," 9th Symposium (Int'l) on Combustion, 1963.
118
-------
Lyon, R.K. and Longwell, J.P., "Selective Non-Catalytic Reduction
of NOX by NH3," NOX Control Technology Seminar, SR-39, Electric
Power Research Institute, Palo Alto, California, February 1976.
Also, Lyon, R.K., U.S. Patent 3,900,554, August 19, 1975.
McGowin, C.R. , "Stationary Internal Combustion Engines in the United
States," EPA Report R2-73-210, PBr-221-457, April 1973,
Middlemiss, I.D., "Characteristics of the Perkins 'Squish Lip'
Direct Injection Combustion System," SAE Paper 780113, February 1978.
Murayama, T., Miyamoto, N., Sasaki, S,, and Kojima, N,, "The Relation
Between Nitric Oxide Formation and Combustion Process in Diesel
Engines," 12th International Congress on Combustion Engines
(CIMAC), Tokyo, 1977.
Muzio, L.J., Arand, J.K,, and Teixeria, D,P,, Sixteenth Symposium
(Int'l) on Combustion, The Combustion Institute, p, 199, 1977,
Pozniak, D.J., "Exhaust-Port Fuel Injection for Chemical Reduction
of Nitric Oxide," SAE Paper 750173, 1975,
Rife, J. and Heywood, J.B., "Photographic and Performance Studies
of Diesel Combustion with a Rapid Compression Machine," SAE
Paper 740948, 1974.
Sakai, Y., Miyazaki, H,, and Mukai, K., "The Effect of Conbustion
Chamber Shape on Nitrogen Oxides," SAE Paper 730154, 1973.
Schaub, F.S. and K.V. Beightol, "Effect of Operating Conditions
on Exhaust Emissions of Diesel, Gasr-Diesel and Sparkr-Ignited
Stationary Engines," ASME 71-WA/DGPr-2, 1971,
Taylor, D.H.C. and Walsham, B,E,, "Combustion Processes in a Medium^
Speed Diesel Engine," Proceedings of the Institute of Mechanical
Engineers, 184, p, 67, October 1969,
Wall, J,C., Heywood^ J,B,, and Woods, W.A,, "Parametric Studies of
Performance and NOX Emissions of the Three^Valve Stratified Charge
Engine Using a Cycle Simulation," SAE Paper 780320, Detroit,
February 1978.
Wilson, R.P., Jr., Muir, E.B,, and Pellicciotti, F,A,, "Emissions
Study of a Single-Cylinder Diesel Engine," SAE Paper 740123, 1974.
Wilson, R.P., Jr., "Potential Emission-Control Concepts for Large-Bore
Stationary Engines," Phases-I and II, EPA Contract 68^02^2664, November
Wilson, R.P.,Jr., Fowle, A.A., Raymond, W.J., McLean, W.J., "Model for
Nitric-Oxide Formation in a Large-Bore Spark Gas Engine," Prepared
for SAE Congress, Engine Modeling Session, 1979.
119
-------
Wyczalek, F.A., Harned, J.L., Maksymiuk, S., and Blevins, J.R.,
"EFI Prechamber Torch Ignition of Lean Mixtures," SAE
Paper 750351, February 1975.
Youngblood, S.B., Offen, G.R., and Cooper, L., "Standards Support
and Environmental Impact Statement for Reciprocating Internal
Combustion Engines," ACIJREX report TR-78-99, EPA Contract
68-02-2807, March 1978.
120
-------
Laminar Flame Propagation
Ignition Site at Interface-
Hot Combustion Products
Spark
Flame Propagation
Contact of Product and Reactant
Eddies
Burned Gas
Fresh Charge
Factors Which Affect NO Production
(See Fig 2-2)
initial Flame Temperature Determined by
Compression Temperature at Which Flame
Ignites, Fuel/Air Ratio, Fuel Type, EGR.
Temperature History Determined by
Compression Heating Due to Combustion,
Wass Heat Loss, Mixing With Cooler Gas, and
Cooling Due to Piston Motion.
Ignition and
Combustion Zone
Fresh Mixture
Flame Propagation
Flame Front
Heat Loss
Figure 21. Combustion and No formation in SI engines.
121
-------
Adiabatic Flame Temperature for
Precompression Temperature = 820°K
Domain of Significant NO Formation
Ignition Point f
ath Representing
Cooling Due to
Expansion and
Heat Loss
2,000
2,100 2,200
2,300 2,400
Temperature, °K
2,500
2,600
2,700 2,800
Figure 22. Initial nitric oxide formation rate . (40 atm pressure)
122
-------
1.4
1.3
1.2
1.1
___ Adiabatic Flame Temperature for
Precompression Temperature = 820° K
Domain of Significant NO Formation Rates
0.6
0.5
0.4
Stratified Combustion (rich portion)
\\YT
Temperature
Suppression (2 ^\\\\\\\v
4 J Combustion (lean port!
Conventional
Temperature Suppression
Lean Combustion
Stratified Combustion
2,000 2,100 2,200
2,300 2,400 2,500
Temperature, °K
2,600 2,700 2,800
Figure 23. Basic categories of nitric-oxide supression.
123
-------
+8
+6
+4
+2
to
g
o
u.
3
1.2
1.0
0.8
0.6
SI?
0.4
0.2
Conditions
Rich Zone Burned First
Overall 4> = 0.75
Start3°
Fixed Duration 15°
Step Stratification
a = 0 (Uniform)
(100%)
0.75
0.75
50% 40% 30%
1.00 0,09 1.18
0.50 0.53 0.57
Extent of Stratification
20% Rich Zone Size
1.30 Rich Zone
0.61 Lean Zone tf>
FIGURE 2-4: PREDICTED EFFECT OF STRATIFIED
CHARGE
124
-------
2 -
I 0
6
O
a
-2 1-
Fixed Combustion.
Duration 15°CA
5000
4000
3000
2000
T
Data for Single Cylinder
Experimental Engine
CR = 7.29
8% Residual
7° BTDC Timing
Fixed Fuel Rate
in-Cylinder NOX
V .1
210% 200%
190%
I
Fixed
Combustion
Duration 15°CA
180% Air
.66 .68 .70 .72 .74 .76 .78 .80 .82
.84
FIGURE 2-5:
EFFECT OF EXCESS AIR
(TWO STROKE SPARK GAS ENGINE)
125
-------
Heat Release
Profile
TDC
* Assuming ubrake-7find-10%
Source: Lyn (1960).
Combustion
Duration r?jnd
30°CA. 54.9% 44.9%
40°C.A. 52.5 42.5
50°C.A. 50.8 40.8
60°C.A. 48.6 38.6
70°C.A. 46.4 36.4
Range of
Typical Engines
Figure 26. Effect of combustion duration on efficiency.
126
-------
Stage 1
Burning Rate of Premixed Charge
(10% of Fuel)
Burning Rate of Spray
Mixing Controlled
90% of Fuel
Rate of Injection
Evaporation Rate
-10
Ignition
10 20
Crank Angle. Degrees
Figure 31. Illustration of two stages of combustion.
127
-------
1.4
1.3
1.2
1.1
Adiabatic Flame Temperature
for Pre-Compression Temperature = 820°K
Domain of Significant NO Formation
Ignition
Peak NOX Rate
After
Compression
Heating
Quenching and
Product/Air
Intermixing
0.7
0.6
_ Conditions at which Combustion Initiates
0.5
0.4
2,000
2,100 2,200 2,300 2,400 2,500
Temperature, °K
2,600 2,700
2,800
Figure 32. Nitric oxide formation rate for a non-uniform fuel distribution.
(40 atm pressure)
128
-------
1.0
0.8
3
U_
QC
0.6
0.4
0.2
D
O
0.2
0.4 0.6 G.8
Water/Fuel Ratio By Volume
1.0
First Author
Type
RPM
O
O
A
V
O
A
+
n
0
A
V
D
Greeves
Murayama
Vichnievsky
Wasser
Marshall
Abthoff
Owens
Vichnievsky
Thompson
Valdmanis
Vichnievsky
Storment
Storment
Last
Last
Dl
Dl
Dl ("AGROM")
IDI
Dl
Dl
Dl
IDI
Dl
Dl
Dl ("MONO")
Dl
Dl
Dl
IDI
2000
1200
2000
1800
13-Mode
2000
2300
2500
1500
2600
2000
2500
2000
1750
2300
Emulsifier
T-Valve
Twisted Blade
Gear Pump
Chemical
Chemical
"DA" Type
"DY"Type
Vortex
Vortex
FIGURE 3-3: EFFECT OF EMULSIONS ON HIGH-SPEED DIESELS (1200-2600 RPM)
129
-------
1.00
1 0.98
0.96
CD
-S 1.0
cc
0.8
a>
DC
0.6
0.4
0.2
A
O
V
O
0.2
Projected
JL
_L
0.4 0.6 0.8
Water/Fuel Ratio By Volume
1.0
O
A
V
D
First Author
Taylor
Taylor
Bastenoff
Semt
Type
Ol (Single)
Dl (Multi)
Dl (PC-2.5)
Dl (PC-2.5)
RPM
1000
1000
500
500
Emuteifier
Gear Pump
Gear Pump
Westhalea
Homogenizer
Bore
216 MM. (8.5")
-270 MM. (10.5")
400 MM. (15.75")
400 MM. (15.75")
FIGURE 3-4: EFFECT OF EMULSIONS OS MEDIUM-SPEED DIESELS (400-1000 RPM)
130
-------
40
Emulsion
Torch
30
Dual Fuel
c
o
u
3
T3
0)
tr
Turbulence
High Rate
_ O
* Multiple Spark
* High Energy
Spark
Degraded Premix/
Modifed
Shape
-o
o>
.92
10
EGR
Oft
Refrigeration
Divided Chamber
Prechamber O
Circumferential Q
Injection
O Pilot
Open-Chamber
Stratified *
C Slope =0.7
10 out of 16 concepts are above this line)
Legend:
Spark Gas
O Diesel
I
10
20 30 40 50
Projected 10-Year Cost (S/HP)
60
Figure 41 . Cost effectiveness of emission control concepts.
131
-------
40
30
o
T)
CD
cc
0*20
R>
CD
T3
-------
TABLE I. STATIONARY LARGE BORE ENGINE POPULATION
Application
Predominant
Engine
Size
Estimated Estimated
Installed Annual
Capacity Fuel Use
(106 HP) (1014 BTU)
Representative
Engine Models
Spark Gas Pipeline
Transmission
2000-4000 HP
(14-20" bore)
Gas Gathering, 800-1600 HP
recomprssion, (8-10" bore)
and storage
Diesel Deep oil well
drilling rigs
and oil trans-
port
1500-2500 HP
Baseload elec- 2000-3000 HP
tricity gener- (8-10" bore)
ators for muni-
cipal utilities
Standby gener- 4000-8000 HP
ating sets for
nuclear and
hospitals
Industrial
power and
water/sewage
pumping
1000-2000 HP
TOTALS
8.0
4.0
6.5
4.0
2.3
1.4
26.2
3.4
1.7
3.0
1.7
0.7
11.1
Cooper GMV.GMW
Ingersoll Rand KVS
Dresser Clark TVC,
HBA
Worthington UTC
Superior 510,825
Waukesha VHP
Waukesha VHP
Electromotive 645
Superior 510,825
Fairbanks-Morse 38D
Same as above
Pielstick PC-2
0.6 Same as above
Source: Arthur D. Little, Inc. estimates based on interviews with engine
manufacturers, Youngblood et al(1978), McGowin(1973), Dietzman and
Springer(1974)tand FPC News, 22 Oct 1976.
133
-------
TABLE II
Projected TXO^ Reductions (Maximum with Under 4% BSFC penalty)
Concept
SI:
Torch Ignition
Multiple Spark
Increased Turbulence
Diesel Fuel Ignition
Feedback Control
Degraded Premix
Divided Chamber
Open Chamber Stratified
High Energy Spark
Charge Refrigeration
SI or Diesel:
Timing
EGR
NH w/Catalyst
Ozone w/ Scrubber
Nitrogen Plasma
Absorption
Diesel i
Water/Fuel Emulsion
Pilot Injection
Prechamber
High Injection Rate
Modified Chamber
Circumferential
Injection
Engine
Large
Bore
35%"
30
_ «
__
35
15
33
__
45
15
20
29
16
on
Data
Small
Bore
~50%~
60
-
42
26
50
40
__
__
(80%)
__
65
20
40
60
50
Based on Model
28%
28
20
__
__
27
32
19
28
30
15
46
__
"Engineering
Judgment"
30%
22
18
20
,_
15
23
13
15
25
15
20
50
50
50
28
17
28
22
23
22
Weighted
Consensus 21
34%
25
19
27
21
27
16
20
31
15
27
65
50
50
37
16
24
25
20
22
References
1,5
17
18
19
20
16
2,3
4
6
7
__
8,9
10,11
11,12
13,14
11,15
1McGowln et al(1973)
2Blumberg and Rummer(1971)
3Wilson et al(1978)
''Youngblood et al(1978)
%yczalek et al(1978) and.
Sakai et al(1976)
6Youngblood et al
(1978) engine #70
7Pozniak(l975)
8See Figure 3-3
9See Figure 3-4
10Mlson et al(1978)
nHermann(1977)
12Wilson et al(1973)
13Kasel & Newton(1977)
ll*Khan, Greeves, Wang(1973)
15Middlemiss(1978)
16Dale et al (1978)
17Kuroda(1978)
18Schaub & Beightol(1973)
19Desoete(1977)
20Wall et al(1978)
21Weighting Factors from left
to right are 50%, 0, 25%,
25%
-------
TABLE III
OVERALL RANKING OF EMISSION CONTROL METHODS
SPARK
IGNITION
DIESEL
EXHAUST
TREATMENT
Method
Torch Ignition
Multiple Spark
Increased Turbulence
Dual Fuel
Feedback Control
Open Chamber Stratified
Degraded Premixing
Divided Chamber
High Energy Spark
Charge Refrigeration
EGR
Timing
Emulsion
Pilot
Prechamber
High Rate
Modified Shape
Circumferential
EGR
Timing
NH Catalyst
Ozone
Nitrogen Plasma
Absorption
% NO Rank
Reduction (NO )
34% 1
25 4
19 7
27 3*
-
16 8
21 5
27 3*
20 6
31 2
27 3*
15 9
37% 1
16 7
24 4
25 3
20 6
22 5
27 2
15 8
65% 1
50 2*
50 2*
«
10-Year Rank
Added (cost)
Cost
!$14/hp ~3
21 .4
-3 1
5 2
28 6
65 11
29 7
64 10
24 5
61 9
47 8
$i7/hp -i j
42 | 4 ;
59 6
31 3
29 2
68 7 '
47 5
'
247 ~77
934 3
714 2
"**
Feasibility Rank
Index (Feasibility
F
.73 5
.74 4
.81 1*
.65 9
.72 6
.62 10
.68 7
.67 8
.79 2
.81 1*
.75 3
-
.68 5
.89 1
.56 7
.80 2
.71 4
.60 6
.75 3
-
.75 3
.70 4
.80 1
.78 2
A BSFC
(±2%)
1
0
- 1
-2
+ 4
+ 1
+ 4
0
0
f 1
+ 4
_ 2
+ 2
+ 3
+ 2
+ 2 ;
+ 2
+ 1
+ 4
Vl
'( + 4
+ 4
+ 2
CO
en
-------
TABLE IV
Ranking of Emission Control Methods based
on NO Reduction, Feasibility and Cost
x
SPARK
IGNITION
DIESEL
EXHAUST
TREATMENT
Method
Torch Ignition
Multiple Spark
Increased Turbulence
Dual Fuel
Feedback Control
Open Chamber Stratified
Degraded Premixlng
Divided Chamber
High Energy Spark
Charge Refrigeration
EGR
Timing
Emulsion
Pilot
Prechamber
High Rate
Kbdified Shape
Circumferential
EGR
Tin ing
NH Catalyst
Ozone
Nitrogen Plasma
Absorption
All Factor*
F -ANOX
ACost
1.7
0.9
00
3.8
-
0.2
0.5
0.3
0.6
0.4
0.5
1.5
0.3
0.2
0.6
0.5
0.2
0.4
-
0.20
0.03
0.06
*~
Rank
3
4
1
2
-
9
6'
8
5
7
6*
1
5
6*
2
3
6*
4
-
1
3
2
"
Tie-score
136
-------
LOW NOx COMBUSTOR DEVELOPMENT
FOR STATIONARY GAS TURBINE ENGINES
By:
R.M. Pierce, C.E. Smith, and B.S. Hinton
Pratt & Whitney Aircraft Group
Division of United Technologies Corporation
West Palm Beach, Florida 33402
137
-------
ABSTRACT
This exploratory development program is being accomplished to identify,
evaluate, and demonstrate dry techniques for significantly reducing produc-
tion of NOX from thermal and fuel-bound sources in burners of stationary gas
turbine engines.
Utilizing the low-NOx combustor design concept, "Rich Burn/Quick Quench",
that was identified previously in the first two phases of this four-phase
program, the design of a full-scale prototype combustor has been completed.
The essential features of the NOX reduction concept, which were deter-
mined in early bench-scale experiments, are reviewed and the design of the
full-scale combustor is described.
The experimental program now underway to evaluate performance and NOX-
reduction characteristics of the full-scale prototype combustor has pro-
gressed through preliminary checkout testing.
138
-------
INTRODUCTION
The overall objective of the work described in this report is to identi-
fy, evaluate, and demonstrate combustion control techniques for significantly
reducing the production of oxides of nitrogen in stationary gas turbine en-
gines. While the precise role of the gas turbine engine in future utility
and industrial powerplant applications is still emerging, it is clear that
the number of units in operation, which has increased dramatically in recent
years, will continue to grow. The higher cost of petroleum and the concern
over air quality are forcing many sectors of energy industry to re-examine
traditional business patterns. Because of the inherent flexibility of in-
stallation and operation, short manufacturing lead times, and lower capital
costs, the gas turbine engine is projected for use in utility combined cycles
and industrial cogeneration applications, as well as the presently used modes.
Operation on nitrogen-bound liquid hydrocarbon and low-BTU gaseous fuels, and
the capability to meet increasingly stringent exhaust emission requirements
are prerequisites to such expanded use.
An exploratory development program was undertaken to achieve significant
reductions in both thermal and fuel-bound NOx in combustors representative of
the designs employed in both current and future stationary gas turbine en-
gines. The investigations have addressed dry combustion control techniques
and have been directed toward combustor designs that are suitable for use in
a 25 megawatt (nominal) stationary gas turbine engine. The program goal for
NOx *-n combustors fired with gas or oil containing no appreciable bound ni-
trogen is 50 ppmv (at 15% 02); for combustors fired with an oil containing up
to 0.5% nitrogen (by weight) the goal for NC^ is 100 ppmv (at 15% 02). The
NOX goals are to be attained while maintaining CO emission levels less than
100 ppmv (at 15% 02).
The program is being accomplished in four phases and is currently near-
Ing completion. The first phase, which consisted of an analytical investiga-
tion of combustion concepts considered to have potential for reducing the
production of NOx, has been completed. In the second phase of work, a number
of promising low-NOx-product ion concepts were bench-tested to select the best
candidate for implementation into the design of a full-scale, 25 megawatt-
size, utility gas turbine engine combustor. The results of the program,
through Phases I and II, are summarized in reference 1. In Phase III, which
has been completed, a full-scale low NOx combustor was designed and fabri-
cated. In the last phase of work, now in progress, the NOx-reduction capa-
bility of the prototype full-scale combustor is being examined experimentally
at conditions simulating the operating range of a 10:1 pressure-ratio sta-
tionary engine.
139
-------
PHASE I AND PHASE II RESULTS
In the first two phases of work, a review and analytical study were con-
ducted to identify concepts that might have potential for reducing the pro-
duction of NOX from thermal and fuel-bound sources of nitrogen in stationary
gas turbine engine combustors. The most promising of these were selected for
experimental evaluation in bench-scale hardware. Of the two successful de-
sign concepts that emerged from this experimental study, a concept based on
rich burning was selected as the basis for the full-scale combustor design
executed in Phase III and evaluated experimentally in Phase IV.
Concept Selection
By general classification, about half the concepts evaluated in the
Phase II bench scale program were based on fuel-lean burning, and half were
based on fuel-rich burning. Because the conversion of fuel-bound nitrogen
to NOx occurs very readily in fuel-lean combustion, those concepts based on
fuel-lean burning were viewed as generally limited to use with fuels con-
taining no appreciable fraction of chemically bound nitrogen. Concepts
based on fuel-rich burning, on the other hand were shown to produce low con-
centrations of the NOx formed from both fuel nitrogen and atmospheric nitro-
gen sources. Two of the design concepts examined in Phase II were found to
provide substantial reductions in NOx concentration levels, and were judged
to be generally acceptable with regard to the fulfillment of conventional
engine-combustor design and operating requirements. One of the two concepts
was based on lean burning, the other was based on rich burning.
Following the bench-scale experimental program, the various NOx control
concepts that had been evaluated were reviewed relative to their potential
and ultimate suitability for incorporation into a stationary gas turbine
engine. Also, an assessment was made of the anticipated differential between
the originally stated and attainable program goals. This effort was con-
sidered essential to determine if the stated goals were realistic or if they
should be revised as dictated by the experimental results obtained in Phase
II or by other information obtained in the conduct of the program.
It was concluded that the program goals as originally stated were indeed
realistic and that the two successful concepts identified in the bench-scale
experimental program were capable of meeting the required NOX and CO concen-
tration levels. Of the two concepts, the lean-burning approach had been
found to meet the emission goals only for fuels containing no more than a
trace concentration of chemically bound nitrogen. The rich burning concept
showed significant potential and had been demonstrated to meet program goals
for NOx reduction with fuels containing nitrogen in concentrations up to 0.570
by weight.
140
-------
Based on this assessment, the rich-burning concept was selected for im-
plementation into the design of the full-scale combustor in the final two
phases of the program1
For review, the key elements of the rich burning concept are identified
in figure 1. A premising chamber is provided, in which the fuel is prevapor-
ized and premixed with air to form a homogeneous rich mixture. The prepared
mixture is introduced into a primary zoue section of the combustor and burned
without the further addition of airflow. The rich burning process is ter-
minated in a final step involving very rapid dilution, which provides the
airflow needed to achieve an overall lean exit plane equivalence ratio. The
success of this concept, which does not differ in its essential features from
many previous proposals for a rich burn, quick quench approach to NOx reduc-
tion, has been largely a matter of execution and of the selection and refine-
ment of techniques for achieving the idealized conditions called for in the
basic concept.
The arrangement of the rich burner bench scale hardware is shown in
figure 2. A single, high velocity premixing passage is provided, terminating
in a swirler that serves to stabilize the flame in the primary zone of the
combustor. All the air entering the primary zone comes through the premixing
passage. At design point, the primary zone operates fuel rich. It is fol-
lowed by a dilution section has has been designed for very rapid quenching of
the fuel rich gases leaving the primary zone.
Tests of the rich burner were conducted at elevated pressures and temp-
eratures, simulating actual engine operating conditions. In figure 3 data
are shown from tests conducted at 150 psia, at inlet air temperatures of
650°F and 750°F. By staging the amount of air that entered the premixing
tube, it was found that low NOg concentration levels could be achieved over a
range of overall (exit-plane) equivalence ratios. At the primary air settings
shown (7% and 14%), NOx concentrations of 60 ppmv and lower were demonstrated
using No. 2 fuel with 0.5% nitrogen (as Pyridine). Even lower concentrations
were demonstrated using non-nitrogenous fuel. In figure 4, representative
bench-scale data points are presented for the rich burning concept, demon-
strating low NOx concentration levels over a wide range of operating condi-
tions, using No. 2 fuel.
Tests of the rich-burning concept were also conducted using a low-BTU
gaseous fuel. The exhaust emission data and the composition of the gaseous
fuel mixture (synthetically prepared) are shown in figure 5 and figure 6; the
arrangement of the bench scale hardware is shown in figure 7. The emission
characteristics of the combustor, measured for low BTU gaseous fuel, were
very similar to those obtained previously for No. 2 fuel. A minimum NOjj con-
centration of about 80 ppmv (uncorrected) was measured at the bottom of the
"bucket" in the NOx curve. By varying the primary air setting (which was not
attempted in the tests conducted), it should be possible to achieve this same
concentration level at any desired operating point over a wide range of over-
all (exit-plane) equivalence ratios, in keeping with the results of staging
tests conducted for the same concept using No. 2 fuel.
141
-------
PHASE III. FULL-SCALE COMBUSTOR DESIGN
In Phase III the design of a. full-scale combustor incorporating the
successful NO^-reduction concept demonstrated in the bench-scale screening
experiments of Phase II was carried out. Execution of the design of the full-
scale combustor was based in large part upon the data generated in the bench-
scale combustor program. It is important to point out, however, that while
these results may provide a full-characterization of the bench-scale combustor
itself, they cannot be used to specify the complete design of the full-scale
combustor. Scaling criteria dictate that there can be no exact and complete
correspondence between a prototype combustor and its subscale model, with re-
gard to physical dimensions, operating conditions, and combustion performance.
In lieu of direct scaling, a partial modeling approach has been taken, as
described in this section. In the basic features of the full-scale combustor,
and in the areas of primary air staging (to control stoichiometry), combustor
aerodynamics, liner cooling, and residence time requirements, an attempt has
been made to reproduce the essential processes of the rich burning concept, as
identified and defined parametrically in the bench-scale test results. The
design of the full-scale combustor has been executed separately, drawing upon
analytical modeling techniques and upon the bench testing of key components
(particularly the full-scale premlx tube) to verify that the essential pro-
cesses of the concept have been successfully reproduced.
In the discussions that follow, the rich burning concept will be referred
to throughout by the descriptive name "Rich Burn/Quick Quench",
Design Requirements
The objectives adopted for the design of the full-scale prototype com-
bustor reflect many of the requirements of conventional gas turbine combus-
tion systems (temperature rise, pressure drop, and others), as well as the
stated emission goals of the current experimental development program. It is
intended that the N0x-reduction technology generated in this program be com-
patible with current state-of-the-art design practice for stationary gas
turbine engines in the 25-megawatt-size range. The design requirements for
the full-scale combustor are presented in Table I.
Emission-Reduction Concept (Rich Burn/Quick Quench)
The basic features and demonstrated results (from Phase II bench-scale
testing) of the Rich Burn/Quick Quench Concept, in summary form, are as
follows:
Arrangement - Two combustion zones are arranged in series: a fuel-rich
primary zone and a fuel-lean secondary zone, separated by a necked-down
142
-------
"quick quench" section. A diagram of the bench-scale configuration as tested
is shown in figure 2, with the major zones identified.
Critical Features - Three key requirements for low exhaust emissions
have been identified, using distillate and low Btu gaseous fuels:
o all air entering the fuel -rich primary zone must be premised with
fuel to prevent diffusion burning (in particular, liner cooling
airflow cannot be discharged into the primary combustion region) ;
o minimum NOx concentration levels are obtained at primary zone
equivalence ratios near 1.3;
o quick-quench air is added at a single site, and must be introduced
in a manner that produces vigorous admixing, approximating a step
change in composition and temperature.
Emission Characteristics - The emission characteristics, or "signature"
of the basic concept are shown in figure 8, as generated at a constant air-
flow setting by varying the burner fuel flowrate. This "signature" has two
notable features:
o a peak in the CO curve, to the right of which (at 0.2 exit plane
equivalence ratio and higher) measured concentration levels are low;
o a minimum point or "bucket" in the NOX curve, which corresponds
approximately to a primary zone equivalence ratio of 1.3. The NOx
curve "bucket" represents the unique low-emission design point of
the basic emission signature.
Variable Primary Zone Airflow - Variable geometry can be employed to
shift the low-emission design point over a broad range of exit plane equiva-
lence ratios, as shown in figure 3. As described in reference 1, the NOx
"bucket" can be shifted in this manner while maintaining an essentially fixed
CO characteristic.
Residence Time Requirements - The minimum NO^ concentration levels
attained (at the bottom of the NOX curve "bucket") have been shown to decline
with increasing primary zone residence time, and with an increasing level of
primary zone turbulence. This characteristic results in basic design trade-
offs among primary zone length (residence time), combustor pressure drop, and
resultant NOx concentration levels.
Basic Features of the Full -Scale Combustor
To initiate the design of the full-scale combustor, studies were con-
ducted to determine what methods might be employed to successfully reproduce
the critical features of the bench-scale combustor and accomplish the essen-
tial processes of the Rich Burn/Quick Quench concept. As stated, the bench-
scale combustor hardware cannot be "scaled -up" directly to produce a full-
scale design. However, the parametric data generated for the bench-scale
combustor does serve to identify the critical features of the bench-scale
143
-------
design, and to characterize the essential processes of the basic concept.
To achieve emission characteristics in the full-scale design comparable to
those demonstrated in bench-scale, it is necessary to execute a second design
(in larger scale) that successfully sets up the same basic physical processes
and preserves the critical features of the smaller combustor.
In the following sections, the basic features of the full-scale config-
uration are described, and discussions of the various procedures followed in
the execution of the detailed design of the combustor are presented.
With reference to figure 9, the basic features of the configuration are
as follows:
a. A single centrally mounted premixing tube is provided having a
velocity versus length schedule similar to that of the smaller
tubes employed successfully throughout the bench scale test program.
Variable vanes (not shown) are provided at the premixing tube in-
let to regulate the primary zone airflow. The premixing tube is
offset slightly with respect to the centerline of the combustor in
order to be in-line with an engine diffuser passage.
b. An extended length primary zone is provided for increased residence
time.
c. A primary liner cooling scheme is provided that does not call for
discharging spent cooling air into the combustion region of the
primary zone. Airflow from the primary-liner convective cooling
passage is discharged into the combustor through the quick-quench
slots.
d. The quick-quench section is designed to provide strong mixing, so
that an abrupt termination of the primary-zone rich burning process
can be achieved. An area ratio of 2.8 to 1 was adopted in the
"necked-down" section of the combustor, matching the optimum value
determined in the bench-scale tests.
e. The aft dilution zone of the combustor is combined with the engine
transition duct to provide a maximum allocation of the available
combustor length for the oxidation of CO while still maintaining an
extended-length primary zone in the interest of achieving low NOX.
In the remaining discussion of the design study that has been carried
out to determine the detailed configuration of the full-scale combustor,
activities are described in four major areas. In figure 10 these areas are
identified, and the logic of the overall design study is depicted.
Primary Air Staging
The bench-scale test results from Phase II have consistently shown that
minimum NOX concentration levels are achieved when the primary zone equiva-
lence ratio is maintained near a value of 1.3. In order to achieve this
value over a broad range of combustor exit plane equivalence ratios (engine
144
-------
power settings), a method of varying the amount of airflow admitted to the
primary zone is required. At the baseload setting, slightly more than 20%
of the total combustor airflow is required in the primary zone; at idle,
approximately 107» is required.
The method of primary air staging selected for the full-scale combustor
is depicted in figure 11. A variable damper, consisting of two sets of
vanes (one movable, one fixed) is mounted at the inlet plane of the premix
tube. The variable damper can be adjusted to achieve a 2:1 variation in pre-
mix tube airflow. At the full-open setting, only a nominal pressure drop
(less than 0.1%) is incurred by airflow passing through the vanes. A large
number of narrow vanes is employed, to minimize wake formation in the in-
coming airflow. In going from the full-open to the full-restricted setting,
the total damper travel required is only about 10 degrees (or 0.25 inches
at the maximum diameter).
Combustor Aerodynamics
The combustor internal airflow distribution is determined by several
factors, which include the relative areas of openings in the combustor liner,
the pressure/velocity distribution of the approach airflow, and the combustor
internal geometry cross-sectional area as a function of length. The full-
scale prototype combustor must meet a prescribed schedule of internal equiva-
lence ratios and therefore must be designed for a specific internal airflow
distribution.
The Rich Burn/Quick Quench concept calls for a unique "necked-down"
shape that produces locally high velocities in a quick-quench section for the
purpose of vigorous mixing. An analysis of the effect of these high velo-
cities on the combustor pressure drop and airflow distribution shows that
significant "mixing losses" are incurred in the quick quench section, and
that these losses must be considered in tailoring the liner hole pattern to
achieve the required airflow splits (these mixing losses are believed to be
desirable and, in general, to be indicative of the high rate of mixing
achieved in that section of the combustor).
To ensure an accurate determination of the liner hole areas required in
the full-scale prototype combustor, a computer model was formulated to simu-
late the aerodynamic processes described above. The model accepts as input
a prescribed fractional airflow distribution, the inlet air temperature and
pressure, the fuel flowrate, and the required liner pressure drop. The
cross-sectional area profile of the combustor is also input, and an external
pressure distribution may be specified. The calculation is performed in a
downstream-marching fashion, beginning with an initial guess for the premix
tube airflow in pounds per second. At each of several stations along the
length of the burner, the pressure drops associated with various components
and processes are computed. These pressure drops include the following:
1) premix tube entrance and blockage losses (both at the variable damper and
at the ff.l injector); 2) swirler pressure loss; 3) momentum pressure loss;
4) mixing loss in the quick quench section; 5) mixing loss in the dilu-
tion zone. At the exit plane a check is made on the overall pres-
sure drop. If it agrees with the specified input value, the solution is
145
-------
complete. Otherwise a new value for the premix tube airflow rate is assumed,
and the computation is repeated. The final solution includes the total air-
flow that can be passed through the combustor for a given overall pressure
drop and specified distribution, and the schedule of hole areas required to
achieve that distribution.
Several cases were run with the aerodynamic model for the purpose of
sizing the holes in the quick quench section of the combustor and in the
dilution zone. The results verified that a major source of combustor pressure
drop is the "mixing loss" in the quick quench section. The model assumes
one-dimensional flow and computes as "mixing loss" the total pressure drop
due to mass addition (from the momentum equation). In the quick-quench
section, the mass added through the penetration holes is assumed to have zero
axial velocity. This flow must be accelerated, along with the approach flow
from the primary zone, to a uniform axial velocity consistent with the cross-
sectional area of the "necked-down" (quick quench) section of the burner.
The smaller the diameter of the "necked-down" section, the greater the re-
quired acceleration, and the greater the resultant total pressure drop.
To illustrate the results described, a representative case run with the
aerodynamic model is presented in Table II. Predictions for the prototype
combustor operating at 5.5 percent pressure drop and at a baseload power
setting are shown. The data include computed flow properties at selected
stations along the length of the combustor. The stations are identified in
figure 12. It may be seen from the tables that there is a progressive de-
cline in total pressure, caused by the losses incurred at the various sta-
tions. A major source of pressure drop and a controlling factor in the pre-
dicted aerodynamic characteristics of the combustor is the loss incurred in
the quick quench section. The quick quench section has a throttling effect
on the combustor flowrate. The higher the axial velocity in the necked down
passage, (i.e., the smaller the diameter, for a given primary air setting)
the lower the quantity of airflow (the total of primary air and quick quench
air) that can pass through that section without an increase in burner pres-
sure drop.
These predicted results have been verified experimentally in tests of
the bench-scale combustor, as shown in figure 13. Good agreement with the
experimental data was demonstrated.
Liner Cooling Scheme
A critical feature of the Rich Burn/Quick Quench concept is the elimina-
tion of non-premixed air from the primary zone of the combustor. The pri-
mary liner cooling scheme depicted in figure 9 calls for convective cooling
of the outside surface of the liner, and for the discharge of spent cooling
air into the quick quench section of the combustor. Cast fins are provided
on the cooled side of the liner to increase the effective surface area. An
outer shroud is placed around the cast liner to maintain a high air velocity
along the outside surface.
To ensure that the intended design can be properly implemented, and that
the required cooling effectiveness can be achieved, an analysis of the aero-
146
-------
dynamic and heat transfer characteristics of the convective cooling channel
was conducted. Model predictions for the maximum heat-load condition (which
occurs at an equivalence ratio of about 1.2) indicated that a peak primary
liner temperature of 1536°F can be expected when 43.2% of the total burner
airflow is used for cooling. This relatively high percentage of the total
airflow is readily available for cooling because it also serves as quick
quench air.
To verify the results of the analytical studies and to assess the effect-
iveness of the convective cooling technique, a short series of bench scale
experiments was also carried out. The data generated in these experiments
(using reduced scale hardware) were used as a standard of validation for
the analytical model predictions with regard to the influence of burner
airflow rate, inlet pressure, and inlet air temperature on the primary
liner wall temperature level.
The predicted effect of inlet temperature on wall temperature is shown
in figure 14. The agreement between the model and data is very close. An
increase in inlet temperature from 400°F to 600°F roughly increases the
maximum wall temperature (at an overall FA of 0.070) from 1300°F to 1600°F.
Premix Tube
Good fuel preparation (effective prevaporization and premixing) is of
paramount importance in the design of the full-scale combustor. If the air-
flow entering the primary zone has not been admixed with fuel to form a homo-
geneous mixture, diffusion burning will take place between the incoming air
and the fuel -rich gases already present. Because diffusion burning proceeds
at near -peak flame temperatures, it is inevitable that significant concentra-
tion levels of NOx will be formed in the primary zone under these circumstances.
In order to provide uniform premixing (and prevaporization) of the fuel
and air that are introduced into the primary zone, a number of candidate de-
signs for the full-scale premix tube have been proposed and evaluated (both
analytically and experimentally) during the Phase III design effort. In the
course of these evaluations a considerable body of design data has been
gathered. These data have been assembled to form a premix tube design system,
In this section a brief description of the design system is presented.
At the present time several of the premix tube designs described in the
following discussions are principal candidates for evaluation in tests of the
full-scale combustor.
a. Atomization
Atomization of the liquid fuel and optimization of droplet sizes is of
paramount importance to the designer for two reasons. First, fuel vaporiza-
tion is dependent on fuel drop size: the smaller the fuel droplet, the faster
it vaporizes. Because vaporization is usually one of the attainable goals of
a premix system, atomization determines the premixing length requirements for
vaporization. Second, even if complete vaporization is r.3t accomplished,
147
-------
small droplets ( < 20jUrn ) can be expected to behave like vapor in the combus-
tion process. Thus small premised fuel droplets in air can approach the per-
formance of a perfectly premised, prevaporized system.
Of the various atomization processes, air atomization has the most po-
tential to produce fine droplets in premix tubes. In order to optimize air
atomization, three types of fuel injection or combinations thereof can be
used:
1) downstream axial injection (low fuel velocities)
2) upstream axial injection
3) cross-stream (radial or tangential) fuel injection
All these types of injection provide a high relative velocity between the
fuel and air, thus promoting good atomization.
Empirical correlations for drop size (resulting from air atomization)
must be a function pf the following parameters:
Vf - viscosity of fuel
crf - surface tension of fuel
Pj - density of fuel
p - density of air
Va - velocity of air (relative to fuel)
df - characteristic initial dimension of fuel
(diameter> thickness, etc.)
^1L - airflow-to-fuel flow ratio.
Wf
All other parameters have been shown to have a negligible influence on the
Sauter Mean Diameter (SMD).
The last parameter, Wa/Wf, is a droplet interference and interaction
term that can be eliminated from the list by the following reasoning. If all
of the airflow passing through the premix tube is used in the atomization
process, the air-to-fuel ratio can range from about 10 in fuel-rich premix
tubes (0 = 1.3) to about 20 in fuel-lean premix tubes. It has been shown
(reference 2) that for values greater than five the air-to-fuel ratio does
not play a significant role in the atomization process. Thus we have elimin-
ated the term Wa/Wf from further consideration.
The empirical correlation that follows was derived from references 2-11
which include theoretical analyses and experimental data for liquid jets,
sheets and droplets. The correlation has the form;
SMD = K(df)a(i;f)b(CTf)c(Pf)d(Pa)e(Va)f
148
-------
where K, a, b, c, d, e, and £ are constants. Table III gives a list of the
exponents a thru f from the various references. In reviewing the references,
it was apparent that some of the constants were remarkably consistent (par-
ticularly b, c, and f) while others varied. By the use of dimensional analy-
sis, three exponents can be calculated from three selected exponents. The
following equation was derived:
(1) SMD =K(df)-375(^f)'25(orf)-375(Pf)"-125(Pa)"-5(Va)-1'°
The proportionality constant K was determined to be 48 in reference 7.
Equation (1) allows the designer to predict actual SMD values provided
the value of df is known. Also, equation (1) allows the designer the capa-
bility of providing the effect of changing pertinent parameters. It should
be noted that air velocity is the single most important parameter in the atom-
ization of a liquid fuel. As a typical reference, an air velocity of 400 f/s
at ambient conditions will shatter a thin kerosene jet (.062 in) into drop-
lets with a SMD of 16 H m .
b. Distribution
In addition to atomization, the proper distribution of fuel in a premix
tube mast be achieved. Poor fuel distribution results in incomplete atomi-
zation due to droplet interaction effects, slower vaporization, and mixture
nominiformity. If a premix system is properly optimized, the fuel must be
uniformly distributed throughout the airstream by the time the mixture enters
the main combustor.
In bench-scale premix tubes (1 inch diameter), experience has shown that
centrally mounted pressure atomizing fuel nozzles are capable of properly
distributing the fuel. In larger (full-scale) premix tubes, two techniques
appear to offer greater potential for a uniform fuel distribution. First, a
centrally mounted injector can be used in combination with an inlet-plane
mixing device such as a swirler. An example of this type of fuel distribution
system is shown in figure 15. The swirling airstream either centrifuges
larger droplets outboard or transports smaller droplets by turbulence. Ex-
treme care must be exercised in the design of this type of distribution system
both in the avoidance of reverse flow zones and the avoidance of excessive
wall wetting by the fuel. Second, multiple injection sources can be used with
or without mixing devices. Figure 16 shows two types of premix tubes using
multiple injectors, one with and one without a mixing device.
Multiple radial fuel injectors mounted on the wall of the premix tube
can also be employed. This approach offers the advantage of providing a
uniform fuel distribution without the complexity of a mixing device. Radial
injection also eliminates all internal blockage and provides a "clean" premix
tube design. However, the provisions for fuel penetration must be carefully
determined to properly distribute the fuel without excessive wall wetting.
Designs of this type can be undertaken using the three penetration design
curves for radial fuel injection from references 12, 13 and 14. These are
shown to be in fairly good agreement in figure 17. Data from reference 4 are
also plotted in figure 17.
149
-------
Another promising candidate for optimum fuel distribution is the radial
"spoke" design shown in figure 18. Each spoke has multiple orifice injectors
which tangentially feed the fuel into the airstream. The injection system
shown has 12 spokes and 36 individual orifices spaced on an equal area basis.
Reference 15 employed a similar fuel injection system and obtained excellent
premixing results.
c. Pressure Loss
In order to design a premix tube that passes the desired airflow and
meets the requirement for overall combustor pressure drop, an assessment must
be made of the pressure losses of the various parts of the premix tube itself.
As an example, three types of pressure loss occur in the premix tube shown in
figure 18: internal blockage loss, diffuser boundary layer loss, and swirler
dump loss.
Internal blockage loss is calculated from the one dimensional momentum
equation. Diffuser boundary layer loss can be calculated from numerous
diffuser pressure recovery maps found in the literature. Swirler dump loss
is ordinarily calculated from the one dimensional equation of motion assuming
a one dynamic head loss based on the discharge area of the swirler. By
summing the losses of the various components and iterating to a specific over-
all loss, the required "size" of the premix tube can be determined.
150
-------
Full Residence Time (FRT) and Engine-Compatible Designs
The very low NOX concentration levels achieved under the Rich Burn/Quick
Quench concept were shown in the bench-scale experimental program to depend
upon an extended residence time in the fuel-rich primary zone section of the
combustor. By varying the diameter and the length of the primary zone in a
series of bench scale tests, data were obtained that indicate a trade-off
between attainable NQjj concentration levels and the primary zone residence
time. These results were utilized in the design of the full-scale combustor.
Bench-scale combustor configurations similar to the one in figure 13 were
used to generate residence-time data. Primary zone diameters of three,
five, and six inches are illustrated. Two lengths were tested, 9 and 18 in-
ches (measured from the premix tube swirler to the centerline of the quick-
quench slots), and in one configuration an enlarged premixing tube (designed
to pass 70 percent more airflow) was evaluated. The results obtained are
presented in figure 19 in terms of the trade-off between the minimum attain-
able NOX concentration* levels and the primary zone residence time. Normal-
ized values of these two parameters are shown, because a direct application
of raw bench-scale data in the design of the full-scale combustor is not
considered appropriate. Dissimilarities in the two combustors (particularly
a difference in surface-to-volume ratio of about two-to-one) preclude the
possibility of an exact correspondence between the NOX concentration levels
measured in the bench-scale combustor (for a given value of the bulk resi-
dence time) and those that can be expected in the full-scale configuration.
In the design of the full-scale combustor, the general relationship be-
tween residence time and NOX concentration levels shown in figure 19 was
assumed. It was also assumed that the absolute levels demonstrated in the
bench-scale combustor (50 to 60 ppmv over a broad range of operating condi-
tions, as illustrated in figure 4) could be achieved in the full-scale com-
bustor as well. To select a design-point value of the primary-zone residence
time, several factors were considered:
a. To provide an absolute value of residence time equal to that which
had been utilized in the bench-scale combustor, a primary zone
length about 2.5 times greater than the nominal length available in
a representative 25 megawatt engine combustor would be required.
b. Primarily because of the lower surface to volume ratio, it was
reasoned that a lower value of residence time might be required in
the full-scale combustor. For the initial configuration, a value
equal to half the residence time utilized in the bench-scale com-
bustor was selected as generally acceptable.
* The "minimum attainable NOg concentration" is measured at the bottom of
the "bucket" in the characteristi
concept, illustrated in figure 8.
the "bucket" in the characteristic NOX curve of the Rich Burn/Quick Quench
151
-------
c. Because more than one value of primary residence time is required
to establish whether the data being obtained fall on the negative-
slope portion or the flat portion of the curve in figure 19, it was
decided that two configurations of the full-scale combustor, differ-
ing in primary zone length, should be tested.
Based on the above considerations, two configurations of the full-scale
combustor were designed. The first, which is depicted in basic form in
figure 9, provides a primary zone residence time about half as great as that
utilized in the bench-scale combustor. The second, shown in figure 20 repre-
sents a lower value of primary zone residence time (intended to provide a
second data point). Because of the greater residence time it provides, the
first configuration has been designated as the Full-Residence-Time (FRT)
version of the full-scale combustor. The second configuration, which meets
the basic length requirements of a representative 25 megawatt engine, has
been designated the Engine-Compatible version of the full-scale combustor.
The construction of the full-scale combustor hardware was completed under
Phase III. A photograph of the FRT combustor during construction is shown in
figure 21. The premixing tube and primary liner shroud were not attached in
this figure. The fully-assembled configuration is shown in figure 22, except
for the premixing tube damper mechanism. A view of the damper is shown in
figure 11. In the course of the test program, it is planned that the FRT
configuration will be modified to produce the short-length Engine-Compatible
design.
152
-------
PHASE IV - VERIFICATION TESTING
In Phase IV, which is now underway, the experimental evaluation of the
full-scale combustor will be accomplished. Both the FRT and Engine Compatible
configurations will be tested over a range of conditions spanning the oper-
ating requirements of a commercially available 25 megawatt stationary gas
turbine entine. Three fuels will be used in the test program: No. 2 distil-
late; No. 2 with 0.5% N (as pyridine); and a distillate cut shale oil. The
first tests are being conducted to provide preliminary verification that the
design concept has been successfully transferred from bench-scale to full-scale
hardware. Initial tests are being conducted at intermediate pressure using
all three fuels.
Initial test results have indicated that the basic emission signature of
full-scale combustor is the same as that associa ted with the bench-scale con-
figuration (figure 8). Although NOX concentration levels measured in the
early tests have not been as low as those achieved with the bench-scale com-
bustor, values lower than the program goal have been demonstrated. At the
same time, however, the p remix ing performance has been inadequate (poor dis-
tribution of the fuel and minor damage to the p remix tube swirl vanes as a
result of preignition). Currently, revisions are being made to the design of
the full-scale premixing tube. The initial configuration (the basic arrange-
ment is shown in figure 18), has been replaced by the design shown in figure
20. The better fuel preparation characteristics (improved atomization and
distribution), and higher internal velocities provided by this design are
viewed as critical elements in the attainment of the full NOX reduction poten-
tial of the design concept.
153
-------
REFERENCES
1. Hosier, S. A.: "Advanced Combustion Systems for Stationary Gas Turbines,"
EPA-600/7-77-073e, July 1977, presented at Second Stationary Source Com-
bustion Symposium, August 1977.
2. Rizkalla, A. A. and Lefebrve, A. H.: "The Influence of Air and Liquid
Properties on Airblast Atomization," Joint Fluids Engineering and ASME
Conference, Montreal, Quebec, May 13-15, 1974.
3. Adelberg, M.: "Mean Drop Size Resulting from the Injection of a Liquid
Jet Into a High-Speed Gas Stream (Including Corrections to August 1967
Paper)," AIAA Journal, Vol. 6, No. 6, June 1968.
4. Ingebo, Robert D. and Foster, Hampton H.: "Drop-Size Distribution for
Crosscurrent Breakup of Liquid Jets III Air streams," NACA Technical Note
4087, October 1957.
5. Weiss, Maledm A. and Worsham, Charles H.: "Atomization in High Velocity
Airstreams," ARS Journal, Vol. 29, No. 4, April 1959.
6. Nukiyama, S. and Tanasawa, Y.: "Experiments on the Atomization of Liquids
in an Air Stream," Droplet-Size Distribution in an Atomized Jet, transl.
by E. Hope, Rept. 3, March 18, 1960, Defense Research Board, Department
of National Defense, Ottawa, Canada; transl. from Transactions of the
Society of Mechanical Engineers (Japan), Vol. 5, No. 18, February 1939.
7. Kurzius, S. C. and Raab, F. H.: "Measurement of Droplet Sizes in Liquid
Jets Atomized in Low-Density Supersonic Streams," Rept. TP 152, March
1967, Aerochem Research Labs., Princeton, N. J.
8. Lorenzetto, G. E. and Lefebrve, A. H.: "Measurements of Drop Size on a
Plain-Jet Airblast Atomizer," AIAA 1976.
9. Ingebo, Robert D.: "Effect of Airstream Velocity on Mean Drop Diameters
of Water Sprays Produced by Pressure and Air Atomizing Nozzles," Gas
turbine combustion and Fuels technology, ASME, November 27 - December 2,
1977. Edited by E. Karl Bastress.
10. Dombrowski, N. and Johns, W. R.: "The Aerodynamic Instability and Dis-
integration of Viscous Liquid Sheets," Chem. Eng. Sci., Vol. 18, 1963.
154
-------
11. Wolfe, H. E. and Andersen, W. H.: "Kinetics, Mechanism, and Resultant
Droplet Sizes of the Aerodynamic Breakup of Liquid Drops," Aerojet -
General Corporation, Downey, California Report No. 0395-04 (18) SP/April
1964/copy 23.
12. Donaldson; Coleman; Snedeker; and Richard: "Experimental Investigation
of the Structure of Vortices in Simple Cylindrical Vortec Chamber,"
ARAP Report #47, December 1962,
13. Chelko, Louis: "Penetration of Liquid Jets into a High Velocity Air-
stream," NACA E50F21, August 14, 1950.
14. Koplin, M. A.; Horn, K. P.; and Reichenbach, R. E.: "Study of a Liquid
Injectant Into a Supersonic Flow," AIAA Journal, Vol. 6, No. 5, May 1968
pp. 853-858.
15. Tacina, Robert: "Experimental Evaluation of Premixing/Prevaporizing Fuel
Injection Concepts for a Gas Turbine Catalytic Combustor," Gas Turbine
Combustion and Fuels Technology, ASME, November 27 - December 2, 1977,
Edited by E. Karl Bastress.
155
-------
Liquid Fuel
With Bound
Nitrogen
Premlxlng
Chamber
(Fuel Rich) (Fuel Lean)
Primary
Zone
Dilution
Zone
Air
Figure 1. Rich Burning Concept Burner Components
-Primary
Zone
Dilution Zone
Quick Quench
Slots
Figure 2. Rich Burner Arrangement
156
-------
320
0.16
0.20
0.24 0.28
FA - (Wet Fuel)
0.32
0.36
Figure 5. Variation in Emission Concentrations with
Fuel-Air Ratio for Tests Conducted with
Low BTU Gaseous Fuel
Figure 6. Variation in Low BTU Gaseous Fuel Composition
with Test Point Number
158
-------
Figure 7. Rich Burner Arrangement for Low BTU Gaseous Fuel Test
1000
600
400
Corrected 200
[NOX, CO]
ppmv 100
60
40
20
10
i
CO
0.1 0.2 0.3
Overall Equivalence Ratio
0.4
Figure 8. Rich Burner Characteristics (50 psia, 600°F,
0.5% Nitrogen)
159
-------
Pre mixing
Tube
Primary Liner Quick Aft Transition
J Quench Dilution Duct
Section Piece
Figure 9. Basic Layout of the Full-Scale Combustor
Aerothermal
Model
Heat-Transfer
Model
Full-Scale
Combustor
Design
Bench
Scale-Up
Testing
Full-Scale
Component
Evaluation
Flow
Visualization
Droplet
Vaporization
Trajectory
Model
Figure 10. Major Elements of the Full-Scale Combustor
Design Process
160
-------
Figure 11. Premix Tube Variable Damper Mechanism
161
-------
01
no
Figure 12. Identification of Stations Referred to in
the Aerodynamic Model Calculations
-------
Predicted
Experimental
= 1.7 psi
AP = 4 psi
4 5
Station
Figure 13. Comparison of Predicted and Experimental Pressure Drop
Characteristics of the Bench-Scale Combustor
163
-------
Liner
Waif
Temp -
°Fx10?
16
15
14 -
13
12
11
0
Predicted
Tinlet =
AP/PT = 4.6%
A
0.05 0.06 0.07 0.08 0.09 0.10
Primary-Zone Fuel-Air Ratio
Figure 14. Comparison of Predicted and Experimental
Results for the Heat-Transfer Model
Preswirl Vanes
Air Boost
Fuel Nozzle
3.2 in. dia
Figure 15. A Typical Centrally Mounted Fuel
Injector with a Mixing Device
164
-------
-Sprayring
Injector
Swirl
Vanes
5 In. dia
(Ret)
Center body
(a) Without Mixing Device
Swirl
Vanes
Radial Fuel
Injector
(b) With Mixing Device
Figure 16. Typical Multiple Fuel Injector Premix Tubes
165
-------
120
1 I
Schetz and Padhye
I
Kolpin, Horn and
Reichenbach
Cheiko
Ingebo and
Foster
Penetration Distance
Diameter of Jet
Liquid Density
Air Density
Liquid Velocity
Air Velocity
40
60
Figure 17. Liquid Jet Penetration in Airstream
166
-------
Spoke Fuel
Injector
3.2 in. dia
Swirl
Vanes
Figure 18. Radial "Spoke" Premix Tube Design
167
-------
NOX/NOX
Goal
1.0
0.8
0.6
0.4
0.2
i-No. 2 Fuel Oil
Containing 0.5% N
No. 2 Fuel Oil
I I
> 24 6 8 10 12 14 18
Residence Time/Residence Time to Meet NOX Goal
Figure 19. Variation in Minimum NOX Concentration with Primary
Zone Residence Time (Normalized Bench-Scale Data)
Figure 20. Engine-Compatible Version o£ the
Full-Scale Combustor
168
-------
\
Figure 21. Full-Scale Combustor During
Assembly (FRT Version)
169
-------
Figure 22. Full-Scale Combustor Fully
Assembled (FRT Version)
170
-------
TABLE I. DESIGN REQUIREMENTS FOR THE FULL-SCALE
PROTOTYPE COMBUSTOR
Type Combustor: can (1 of 8, internally mounted)
Basic Dimensions: 10 inch diameter, 20 inch length
Design Point Requirements:
(Baseload) (Idle)
Airflow - 31 Ibm/sec 7.8 Ibm/sec
Pressure - 188 psia 40 psia
Inlet Temperature - 722°F 285°F
Temp. Rise - 1160°F 625°F
Pressure Drop: 3% combustor, 2.57=, diffuser
Lean Blowout: 0.006 fuel-air ratio (burner exit)
Exhaust Emissions (max. at any setting):
(0% Fuel N) (0.57o Fuel N)
NOX 50 ppmv 100 ppmv
at 157» 02 at 157= 02
CO 100 ppmv 100 ppmv
at 157o 02 at 157o 02
171
-------
TABLE II. AERODYNAMIC MODEL CALCULATIONS
FOR FULL-SCALE PROTOTYPE COMBUSTOR
o Configuration for 5.57, Pressure Drop
o Baseload Power Setting (Damper Open)
PO
Station
1
Wa(cum) - pps 6.035
Equiv. Ratio (local) 0.0
Tt -
PS -
PT -
OF
psia
psia
Velocity - fps
Mach
No.
722
187
188
123
.30
,00
.4
0.074
2
6.035
1.300
722
185.72
186.94
163.4
0.098
3
6.035
1.300
722
185.94
186.12
61.7
0.037
4
6.035
1.300
3686
185.78
185.93
108.7
0.036
5
6.035
1.300
3686
184.97
185.93
271.6
0.090
6
18.105
0.433
2505
175.15
180.98
578.4
0.224
7
18.105
0.433
2505
177.74
178.38
191.1
0.074
8
29.658
0.265
1875
171
177
524
.71
.66
.4
0.228
-------
TABLE III. THE EFFECT OF IMPORTANT PARAMETERS
ON DROPLET SIZE
Drop
2r
2r
2r
HMD
2r
2r
SMD
SMD
SMD
2r
2r
SMD
a b
.5 .33
.66
.5 .25
.16 .34
.375 .25
.375 .25
.166 .333
.375 .25
c
.16
.33
.25
.41
.50
.375
.33
.375
.33
.50
.375
d
-.16
-.33
-.25
-.84
-.5
-.375
-.37
.25
-.16
-.125
-.125
e
-.33
-.66
-.25
-.25
-.3
-.875
-.16
-.66
-.5
f
-.66
-1.33
-.75
-1.33
-1.0
-.75
-1.0
-1.0
-.75
-.66
-1.33
-1.0
Reference
3
3
4
5
6
7
8
2
9
10
11
SMD - K(df)a(Vf)b(<7f)c(pf)d(pa)e(Va)f
173
-------
A RESEARCH PLAN TO
STUDY EMISSIONS FROM
SMALL INTERNAL COMBUSTION ENGINES
By:
James W. Murrell, Dr.P.M.
and
Frankie Alexander
Systems Research And Development Corporation
P. 0. Box 12221
Research Triangle Park, N. C. 27709
-------
ABSTRACT
This paper examines some of the requirements for Investigating
the environmental status of small internal combustion engines. These
engines range in size from 1^ horsepower to 15 horsepower and power a
variety of equipment used by homeowners and industrial .members.
With the general growing concern in EPA of identifying sources
of potentially carcinogenic emissions, there exists a possibility that
these small internal combustion engines are a problem source. Research
to characterize emissions from this source has largely been limited to
criteria pollutants, even though the small internal combustion engine
is an incomplete combustion; therefore some carcinogens and other
hazardous compounds are probable.
The basic requirements addressed for an integrated research
design include:
a) Analytical Equipment;
b) Experimental System Design; and
c) Statistical Experimental Design.
176
-------
TECHNICAL REPORT DATA
(Please read Instructions on the reverse before completing)
1. REPORT NO.
EPA-600/7-79-050C
2.
3. RECIPIENT'S ACCESSION- NO.
4. TITLE AND SUBTITLE proceedings of the Third Stationary
Source Combustion Symposium; Volume m. Stationary
Engine and Industrial Process Combustion Systems
6. REPORT DATE
February 1979
6. PERFORMING ORGANIZATION CODE
7. AUTHOR(S)
Joshua S. Bowen, Symposium Chairman, and
Robert E. Hall, Symposium Vice-chairman
8. PERFORMING ORGANIZATION REPORT NO.
9. PERFORMING ORGANIZATION NAME AND ADDRESS
10. PROGRAM ELEMENT NO.
See Block 12.
EHE624
11. 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 Pi
Proceedings; 3/79
PERIOD COVERED
14. SPONSORING AGENCY CODE
EPA/600/13
is. SUPPLEMENTARY NOTES IERL-RTP protect officer is Robert E. Hall. MD-65, 919/541-
2477. EPA-600/7-77-073a thru-(F73e and EPA-600/2-76-152a thru -152c are pro-
ceedings of earlier symposiums on the same theme.
ie. ABSTRACTTne proceedings document the approximately 50 presentations made during
the symposium, March 5-8, 1979, in San Francisco. Sponsored by the Combustion
Research Branch of EPA's Industrial Environmental Research Laboratory-RTP,
the symposium dealt with subjects relating both to developing improved combustion
technology for the reduction of air pollutant emissions from stationary sources,
and to improving equipment efficiency. The symposium was in seven parts, and
the proceedings are in five volumes: I. Utility, Industrial, Commercial, and Resi-
dential Systems; U. Advanced Processes and Special Topics; IE. Stationary Engine
and Industrial Process Combustion Systems; IV, Fundamental Combustion Research
and Environmental Assessment; and V. Addendum. The symposium provided contra-
ctor, industrial, and government representatives with the latest information on EPA
inhouse and contract combustion research projects relating to pollution control,
with emphasis on reducing NOx while controlling other emissions and improving
efficiency.
17.
KEY WORDS AND DOCUMENT ANALYSIS
DESCRIPTORS
b.lDENTIFIERS/OPEN ENDED TERMS
c. COSATI Field/Group
Air Pollution
Combustion
Field Tests
Assessments
Combustion Control
Fossil Fuels
Boilers
Gas Turbines
Nitrogen Oxides
Efficiency
Utilities
Industrial Pro-
cesses
Hydrocarbons
Air Pollution Control
Stationary Sources
Environmental Assess-
ment
Combustion Modification
Trace Species
Fuel Nitrogen
13B
21B
14B
21D
13A
13 G
07B
13H
07C
18. DISTRIBUTION STATEMENT
Unlimited
19. SECURITY CLASS (This Report)
Unclassified
21. NO. OF PAGES
180
20. SECURITY CLASS (This page)
Unclassified
22. PRICE
EPA Form 2220-1 (»-73)
177
------- |