-------
Train A
Train B
38.0
26
38.9
26.2
30.2
25.8
23.1
27
27
23.9
27.3
27.4
26.5
25.1
25.9
25.1
27
25
31.8
29.2
39.9
26.8
29.2
27
-
25
20.9
33.9
23.1
31
29.7
29
27.6
26.1
30.8
23.3
28.7
26.1
31.5
32.1
35.5
29.7
38.1
28.1
21.0
25.6
38.1
24.8
37.2
26.9
28.6
30
25.7
28.4
33.9
27.7
32.8
25.3
33.8
24.8
27.5
26.9
33.0
29.5
23.8
27.9
42.0
26.9
15.8
24.8
30.8
25.3
36.5
26.9
Average 30.2
Average 26.5
Figure 9. SCR Inlet Velocity and NOx Concentration Maps
500
5 10 50 100
SO3 Concentration, ppm
500
Figure 10. Ammonia Salt Formation
as a Function of Temperature and NH3 and S03 Concentration (2)
5B-36
-------
NOx REDUCTION AT THE ARGUS PLANT
USING THE NOxOUT* PROCESS
Joseph R. Comparato
Nalco Fuel Tech
Roland A. Buchs
North American Chemical Corporation
Dr . D . S . Arnold.
L. Keith Bailey
Kerr-McGee Corporation
-------
NOx Reduction At The Argus Plant
Using The NOxOUTR Process
Joseph R. Comparato
Nalco Fuel Tech
Roland A. Buchs
North American Chemical Corporation
Dr. D. S. Arnold
L. Keith Bailey
Kerr-McGee Corporation
ABSTRACT
Urea injection using the NOxOUT Process was demonstrated at the Kerr-McGee Argus
No. 26 unit. The earlier installation of burner modifications had reduced NOx
emissions from 330 ppm to about 225 ppm. The NOxOUT Process further reduced NOx
emissions to below a target level of 165 ppm.
Testing of the hybrid NOx control system included furnace characterization,
injection optimization, and a 48-hour demonstration test. Process performance
was analyzed from extensive data logged with a computer data acquisition system.
A computer model of the furnace flow dynamics provided information for selecting
injector locations and performance settings. Optimization reduced the ammonia
slip to 2 ppm. CO slip was limited to 6 ppm.
Subsequent long-term evaluation examined the impact on plant operation. The air
heater was inspected for possible accumulation of ammonium bisulfate and was
found free of such deposit build-up. The storage, pumping, and injection
equipment operated reliably. Chemical consumption has been consistently within
expected projections. The successful NOxOUT demonstration is being upgraded to
a permanent installation.
5B-39
-------
The NOxOUT Process for controlling oxides of nitrogen (NOx) emissions was
installed on the Kerr-McGee Argus No. 26 coal-fired boiler in June 1989.
Parametric testing was conducted in August 1989 to characterize and optimize the
process application. The matrix testing concluded with a 48-hour continuous
demonstration run. The achievement of 30% reduction in NOx emissions below the
level of reduction previously accomplished with low NOx combustion system
modifications was demonstrated. The combined result of NOxOUT and combustion
system modifications was an overall NOx reduction of more than 50%.
The process optimization during start-up of the NOxOUT system concentrated on
achieving the required NOx reduction while controlling ammonia slip to below 5
ppm. The purpose of this objective was to prevent potential fouling of the
regenerative air preheater surfaces. The limit was chosen to avoid any
significant formation of ammonium bisulfate from the combination of ammonia with
fuel sulfur products. The demonstration test showed that ammonia slip was held
to 2 ppm. It was also important to prevent any significant increase in carbon
monoxide emissions. A target of less than 10 ppm CO increase was achieved with
a CO slip of 6 ppm.
Following the formal testing, the program continued with Phase II, a four-month
period, that was extended to seven months, to observe the long-term effects of
operating the NOxOUT system. The process equipment performed reliably.
Inspections of the unit conducted during and after the Phase II operation
verified successful control of potential air preheater deposits.
NOxOUT Process Technology
In the NOxOUT process, the products of combustion are treated with an aqueous
solution of chemicals. NOxOUT A, sometimes enhanced with other chemicals,
combines with NOx in reduction reactions to yield molecular nitrogen, water, and
carbon dioxide. The technology initially emerged from research on the use of
urea1 to reduce nitrogen oxides conducted in 1976 by the Electric Power Research
Institute (EPRI). EPRI obtained the first patent on the fundamental urea
process in 1980. The overall chemical reaction for reducing NOx with urea is:
NH2CONH2 + 2NO +1/2O2 —> 2N2 + C02 + 2H2O
Nalco Fuel Tech is the exclusive licensing agent for the EPRI technology. Nalco
Fuel Tech has developed the technology with added know-how and patented
advancements. NOxOUT is the tradename for this post-combustion technology for
NOx reduction.
The NOxOUT technology comprises methods and experience for effectively treating
a wide range of applications. Combustion laboratory testing provides data for
proprietary chemical formulations that extend effectiveness beyond the
conditions limiting the performance of the basic urea process. The NOxOUT A
5B-40
-------
formulation insures consistent product quality control and includes additives
which prevent problems such as injector fouling.
Performance design tools increase confidence in applying NOxOUT to new
applications. Process performance is analyzed using Nalco Fuel Tech's chemical
kinetics computer model (CKM). Process conditions are evaluated using
computational fluid dynamics (CFD) modeling techniques.4 The CFD modeling also
enables the simulation of injector design configurations to evaluate chemical
dispersion effectiveness. Used together, the CKM and CFD models provide a sound
basis for predicting expected performance.
Research in injector development, including laboratory analysis using laser
equipment for measuring droplet size and velocity, provides a database for
selecting injection equipment for a specific application. Process equipment
designs incorporate experience from both demonstration and commercial projects.
The NOxOUT technology was fully applied in treating the Kerr-McGee Argus No. 26
unit. Successful experience with a similar unit in Germany, a 75-MW brown coal
fired power plant operated by Rheinisch-Westfalisches Elektrizitatswerk A. G.
(RWE), provided a basis for confidence.5 However, there are often significant
differences between similar coal fired units. Thus, extensive modeling and data
analysis were conducted in support of the testing.
Argus No. 26 Boiler Description
The Kerr-McGee Argus No. 26 unit (figure 1) is a tangentially fired, pulverized
coal, VU 40 type, ABB Combustion Engineering boiler. Western bituminous coal is
burned in the furnace with three coal elevations, each supplied by a bowl mill
pulverizer. Table I is a typical fuel analysis. The unit has a normal
operating steam output of 710,000 Ib/hr (322,580 kg/hr) at 950°F (510°C).
TABLE I
COAL ANALYSIS
Type Utah Bituminous
Ultimate Analysis As Rec'd Dry Basis
%Carbon 70.52 73.27
%Hydrogen 4.91 5.10
%Nitrogen 1.37 1.42
%Chlorine <0.1 <0.1
%Sulfur 0.47 0.49
%Oxygen 10.32 10.72
%Ash 8.66 9.00
%Moisture 3.75 N/A
HHV, Btu/lb 12,592 13,083
58-41
-------
Flue gas heat recovery is accomplished with an economizer followed by a
horizontal shaft regenerative air preheater. After the air preheater, the
combustion products pass through an electrostatic precipitator (ESP) for dust
control, and then through a sodium-based wet SO2 scrubber. The flue gas is
exhausted without reheat at 120°F from the stack.
In May 1989, the firing system was modified to reduce NOx emissions. As
originally built, the unit had close coupled over-fire air (COFA) for NOx
control. In this configuration, baseline NOx levels were about 360 ppm (dry,
corrected to 3% O2) when firing 60% coal and 40% petroleum coke (330 ppm when
firing 100% coal). The modifications included LNCFS (Low NOx Concentric Firing
System) nozzles, flame attachment nozzles, and the addition of SOFA (Separated
Overfire Air) ports.6 NOx emissions were reduced to a typical value of under
225 ppm under normal operating conditions.
Operation with varied overfire air configurations had a strong effect on the
baseline conditions for NOxOUT treatment. Figure 2 shows the NOx emissions with
different SOFA damper positions. The numbers identifying the SOFA conditions
correspond to the upper/middle/lower damper percent opening.
As overfire air dampers were opened, the combustion air was redirected from the
burner zone to higher elevations. While the total oxygen available for
combustion in the furnace was relatively constant, less oxygen was available in
the burner zone as overfire dampers were opened. Fuel burning was effectively
staged. Fuel-rich conditions were created in the burner zone to promote
reduction reactions that destroy some of the NOx formed from fuel nitrogen.7
Combustion was distributed over a longer portion of the furnace. Peak
temperatures were lowered to avoid the thermal formation of NOx from nitrogen in
the combustion air.
Temperatures in the regions suitable for NOxOUT injection were affected by the
degree of staging. A reduction in peak furnace temperatures to control NOx
also reduced the radiant heat transfer to the furnace walls. Consequently, the
flue gas temperature in the upper portion of the furnace increased as NOx is
reduced with deeper degrees of staging. Some data indicated an increase in
temperatures in the upper furnace (elevation 106') from about 1800°F (982°C)
before modifications, to a maximum of 2200°F (1204°C) with the SOFA dampers
fully open.
The 40/100/100 SOFA configuration was considered the typical operating mode for
the Argus #26 unit. As evident in figure 2, the benefits of additional NOx
reduction began to diminish with deeper staging. Figure 3 is a plot of CO
emissions as a function of NOx level. CO emissions tended to increase as NOx
level decreased. This resulted in part from increasing difficulty in tuning the
burner air flows as more air was redirected to the SOFA ports.
5B-42
-------
The 40/100/100 SOFA staging was chosen as the base condition for applying the
NOxOUT process. In July 1989, the temperature profile in the upper furnace with
this SOFA configuration was measured. An average temperature of 2020°F (1104°C)
and a peak of 2110°F (1154°C) in the center of the plane were observed. The
temperature was of concern since the critical level of NOx increases with
increasing temperature.
Chemical kinetics modeling and data from laboratory and field tests have shown
that a "critical NOx" level exists as a function of temperature (figure 4).3
Critical NOx is also strongly affected by the oxygen concentration and the
presence of reducing species such as carbon monoxide. CO concentrations were
also sampled during the temperature traverse and found to be less than 200 ppm
at the furnace outlet plane. A benefit of the high temperatures is that the
reactions are rapid, requiring less residence time than at lower temperatures.
The tendency for residual formation of ammonia and CO byproducts is also
decreased.
A CFD model (figure 5) of the Argus #26 unit was prepared to provide guidance
for the testing. The upward spiraling flow typical of a T-fired furnace was
predicted. The model provided simulations of the injection trajectories and
chemical dispersion. In applying the results, care was taken to identify
guidelines for preventing droplet impingement on tube surfaces.
The NOxOUT Installation
Injection ports were installed at two levels. The upper level, at elevation
106', provided a region where fine droplets could be promptly evaporated in the
lowest available gas temperature conditions. The lower level, at elevation 90',
allowed the injection of larger droplets to enable greater penetration into the
gas stream, but into higher temperatures. The injectors were designed with
interchangeable atomizing tips to facilitate testing different spray pattern
options.
Skid-mounted pumping equipment was installed on site. Chemical injection pumps
metered the reagents into a mixing header. Dilution water also entered the
mixing header. A rotary positive displacement pump mixed the reagents and water
by recirculation through the header and pressurized the mixture for supply to
the injectors. Air was used as the atomizing medium for the injectors. A
pressure-settable air regulator controlled the atomizing medium conditions.
Figure 6 is a simplified schematic of the process system. An analog controller
provided output to the electronic stroke controlled chemical injection pumps.
It also provided PID loop control of the pressure control valve to maintain a
settable constant mixture discharge pressure.
5B-43
-------
Testing Results
Test series were identified in terms of eighteen test days. The test objectives
were:
Test days Test Series Type
# 1-4 Boiler SOFA Characterization
# 5-9 Upper Level Injection
#10-13 Lower Level Injection
#14-16 48-Hour Demonstration Test
#17-18 Miscellaneous Testing
The demonstration utilized an on-line data logging system to provide continuous
monitoring of the boiler and NOxOUT system operation. Display screens of the
current operating conditions facilitated assessing test progress and decision
making for proceeding with steps in the test program. Analog signals from the
boiler control room and instrumentation from the chemical injection equipment
were transmitted to an analog-to-digital converter. The digital values were
read by an 80286 based micro-computer using THE FIX software by Intellution,
Inc.
The plant's continuous stack emissions monitor provided NOx and CO data,
corrected to a dry basis at 3% O2- Signals from the control room provided data
on the boiler operating conditions. Calculations were performed with THE FIX
software to compute NOx on a mass flow basis. Values for NOxOUT chemical flow
rates were taken from analog outputs from the pumping skid controller.
The main parameter for determining the NOxOUT treatment rate is normalized
stoichiometric ratio (NSR). As can be seen from the basic chemical reaction,
one mole of urea combines with two moles of NOx under perfect conditions. NSR
is the ratio of the actual molar flow of urea to the molar flow required for
stoichiometry, or perfect reaction. NSR values were computed from the chemical
flow rates and NOx massflows identified as baseline conditions for the various
test runs.
Ammonia analysis utilized a manual batch extractive method. The very low levels
of ammonia measurements required a technique with high sensitivity. Filtered
flue gas samples were drawn through heated probes from ports in the economizer
outlet. During the 48-hour demonstration run, 12 point samples, on a 4 port by
3 point insertion grid, were collected. Ammonia was captured in an impinger
train containing dilute sulfuric acid. The impinger samples were cooled to a
controlled temperature, then made alkaline to release the ammonia for
measurement with an ion specific electrochemical cell.
Figure 7 is a plot of the NOx emissions as a function of NSR for various SOFA
settings observed during the boiler characterization tests, series 1-4.
External mix injectors producing 100 micron volume mean diameter droplets were
used in the seven ports available at the 106' level. Over 50% NOx reduction was
5B-44
-------
achieved with an NSR of 2.2 in the 0/0/100 SOFA condition and a high NOx
baseline. However, lower NOx emissions were obtained using less chemical with
deeper staging.
The data at 0/0/100 SOFA suggested, as was expected, that the chemical was not
fully dispersed in the flue gas. It should be noted that the chemical flow for
an NSR of 2.2 at a baseline of 288 ppm is 3.8 times the flow for a NSR of 1.0 at
a baseline of 166 ppm. The curve for the 0/0/100 condition suggests that the
performance was limited by the ability to treat all of the gas. The CFD model
indicated that with injection at the 106' level, a large portion of the gas
would pass below the injection plane.
It was noticed that the stack opacity visibly increased during injection and
persisted for more than an hour after injection was discontinued. A "plume"
appeared that was attached to the stack outlet as opposed to the detached water
vapor plume normal during the cooler times of day. Opacity readings at the ESP
outlet did not increase. It was assumed that the plume was caused by ammonia
slip combining with trace amounts of chloride and/or sulfate in the stack gas.
Traces of chloride and sulfate were present in the stack gas from entrainment of
brackish water from the wet scrubber. Many of the decisions in subsequent tests
were aimed at minimizing ammonia emissions.
The plume was minimized as ammonia slip was reduced in the later injection
optimization series but at the expense of some NOx reduction. An SOj injection
system was installed after the demonstration test series was completed. This
was previously planned to reduce particulate emissions. After installation of
the ESP injection system, the plume was eliminated.
Series 5-9 and 10-13 tested injection at the upper (106') and lower (90')
levels. It was found that roughly equal NOx reduction performance could be
achieved at either level. Large droplet sprays (1000 micron) with high total
liquid flows were effective at the lower, hotter level. The large droplets had
longer lifetimes and evaporated in the cooler upper furnace.
The NOx reduction results are shown in figure 8. Somewhat better performance
was achieved with injection at the lower level. This is in part the result of
improved dispersion of the chemical in the flue gases and a slight quenching
effect from the increased liquid flow. High liquid flows were not desirable at
the upper level since complete evaporation could not be assured prior to
reaching tube surfaces. A trend of increased NOx reduction with increased
liquid flow can be seen in figure 9. Injection was optimized by adjusting
atomizing and liquid pressures and using angled internal mix tips with varied
orientation. Figure 10 shows the progress of NOx reduction as different
injection arrangements were tested.
Ammonia slip control was the principal guide in selecting injector arrangements.
5B-45
-------
The results are seen in figure 11. In general, injectors were operated to avoid
the release of chemical in regions too close to the inlet to the convective
pass. Chemical released where gas temperatures are rapidly quenched would form
ammonia. Thus, the optimization achieved a balance between excessively high and
low temperature zones. Ammonia slip values of 2 ppm were measured in the two
12-point traverses conducted during the 48-hour demonstration run.
CO slip was controlled to 6 ppm during the demonstration run. Figure 12 is a
plot of CO emissions versus NOx emissions for all tests. CO emissions increased
from the 11 ppm baseline shown in figure 3 to 17 ppm. As with the baseline
data, CO emissions tended to increase as NOx emissions were decreased.
The scatter in the NOx reduction data reflect the influence of a number of
factors in operating a coal-fired furnace. Routine adjustments in the burner
dampers would result in changes in baseline NOx. Furnace cleanliness influenced
flue gas temperatures. Figure 13 shows a trend of slightly decreasing NOx
reduction with time after cleaning with furnace wall blowers during the 48 hour
demonstration run.
Phase II operation showed that consistent performance can be achieved. The air
preheater was inspected in January, 1990, and May, 1990, and found to be free of
deposits that could be caused by the NOxOUT system. In June, 1990, changes were
made to the boiler aimed at reducing carbon loss. However, the NOxOUT
application was not adjusted for the new conditions. Ammonium bisulfate
deposits accumulated apparently as the result of an undetected increase in
ammonia slip resulting from changes in the furnace conditions. In October,
1990, the injector operating conditions were adjusted to reduce droplet size and
in November, 1990, changes were made in the operation of the air heater
sootblowers. Subsequent operations have been too short to determine whether the
problem has been fully resolved.
Demonstration Results
NOx emissions during the 48-hour demonstration, using an NSR of 1.1, were
reduced 31% below the test baseline. Ammonia and CO slip were controlled to 2
and 6 ppm, respectively. The equipment operated reliably with minimal need for
operator attention. Phase II extended operation confirmed that the system is an
effective means for reducing NOx emissions from the large coal-fired boiler.
As an outcome of the demonstration, the NOxOUT system for Argus unit #26 is
being upgraded to a permanent installation and integrated with the plant control
system. The process will also be installed on the identical unit #25.
5B-46
-------
REFERENCES
1. Muzio, L. J., and Arand, J. K. "Homogeneous Gas Phase Decomposition of
Oxides of Nitrogen", EPRI Report No. FP-253, 1976.
2. Arand, J. K., Muzio, L. J., Setter, J. G., U. S. Patent 4,208,386, June
17, 1980.
3. Sun, W. H., and Hofmann, J. E., "Post Combustion NOx Reduction with Urea:
Theory and Practice", presented at the Seventh Annual International
Pittsburgh Coal Conference, Pittsburgh, PA, September 10-14, 1990.
4. Michels, W. F., Gnaedig, G., and Comparato, J. R. , "The Application of
Computational Fluid Dynamics in the NOxOUT Process for reducing NOx
Emissions from Stationary Combustion Sources", presented at the AFRC
Committee Conference, San Francisco, CA, October 10-12, 1990.
5. Hofmann, J. E., von Bergmann, J., Bokenbrink, D., Hein, K., "NOx Control
in a Brown Coal-Fired Utility Boiler", presented at the EPRI/EPA Symposium
on Stationary Combustion NOx Control, March, 1989.
6. Buchs, R. A., Bailey, L. K., Dallen, J. V., Hellewell, T. D., Smith, C.
W., "Results from a Commercial Installation of Low NOx Concentric Firing
System (LNCFS)", presented at the 1990 International Joint Power
Generation Conference and Exhibition, October 21-25, 1990, Boston, MA.
7. Morgan, M. E., "Effect of Coal Quality on the Performance of Low-NOx
Burners", presented at the British Flame Days Conference, London,
September 1988.
5B-47
-------
CONVECTIVE SUPERHEATER
PLATEN SUPERHEATER —I
•EL 106'
NOxOUT ™
INJECTOR PORT LEVELS
ECONOMIZER
EMISSIONS SAMPLING
COAL PULVERIZER
400
300
Q_
Q.
8
200
100
COFA
ARGUS #26 COAL FIRED BOILER
NOx Baseline
TO
AIR PREHEATER
FROM
AIR PREHEATER
FIGURE 1
0/0/100
0/50/100
0/100/100 40/100/100 100/100/100
Staging Condition (SOFA Damper positions)
FIGURE 2
5B-48
-------
CO EMISSIONS AT BASELINE NOx LEVELS
60
50
Q.
a. 40
CD
A
A
A
O
c3
O
30
20
10
...A
A
A
A
160
180 200 220
NOx Emissions (ppm)
240
FIGURE 3
D_
D_
500
400
300
200
100
Critical NOx Concentration
700 800
3% Excess Oxygen
NOxOUT Kinetic Model
NOxi=500 PPM
,' NOxi=200PPM
900 1,000 1,100 1,200 1,300 1,400
Temperature (degrees C)
FIGURE 4
5B-49
-------
CFD MODEL OF NOxOUT INJECTION
PLAN VIEW AT ELEVATION 100'
cn
CD
cn
o
lonuentrat ion
O.OOE+Lin
1 .34E-04
J1.69E-04
4. 03 E- 04
5.38E-IVI
9.41E-D4
1 .08E-03
1 .21E-03
1 .3AE-OJ
1 .A8E-OJ
1 .blE-OJ
1 .75E-03
Y
FIGURE 5
-------
NOxOUT INJECTION SYSTEM
NOXOUT-A METERING PUMP
WATER
NOXOUT-34 METERING PUMI'
PRESSURE REGULATOR
-N
MOYNO PUMP
MIXING/METERING SKID
INJECTORS
FIGURE 6
300
250 -
Q.
6
200
150 -
100
NOx Emissions
AT STAGING CONDITIONS
0.5 1 1.5 2
NORMALIZED STOICHIOMETRIC RATIO (NSR)
0/0/100 0/100/100 40/100/100 loo/loo/loo
D A O *
2.5
FIGURE 7
5B-51
-------
240
220 '-
Q. 20°
Q.
180
en
"F 160
LJJ
X
O 14°
120
100
NOx Reduction vs NSR
0.5 1 1.5
Normalized Stoichiometric Ratio (NSR)
level 106 level 90 demo
D A O
FIGURE 8
EFFECT OF TOTAL FLOW
NSR RANGE 098 -1.19
ou —
40 -
O 30 -
|~"
O
Q
UJ 20.
X
O
10 -
0 -
3£
D
D
o Q
DQ Q n
~*~ "*" _n ^ E3
+ + + D
+ + +
a
+
I I I I I I I I I I I | | | I 1
0 420 460 500 540 580 620 660 700
D LOWER LEVEL INJ
TOTAL FLOW (GPH)
+ UPPER LEVEL INJ
FIGURE 9
5B-52
-------
NOx REDUCTION vs INJECTOR ARRANGEMENT
NSR RANGE 098-1 19
au —
40 -
£
z
O 30 -
b
D
Uj »-
1
10 -
o -
B
0 D
B D
D O D Q
+ + + n o o B Q
+ + + Q
D
+
1 6 8 10 12 14 16
TEST DAY
D LOWER LEVEL INJ
4- UPPER LEVEL INJ
FIGURE 10
AMMONIA EMISSIONS
SAMPLED AT ECONOMIZED OUTLET
JU
f s 28 -
^^
^ K
D_ x ~
5>
uJ «-
3
-1 18 -
7 16 -
0
^ 12 -
^ 10 -
LU
O B ~
c 6 ~
LU 4 _
^ 2 -
0 -
C
+
D
+
+
D
O
° D
I I I I I I I I I I I I I I 1
2 4 6 8 10 12 14 1
TEST NUMBER
+ UPPER LEVEL INJ. Q LOWER LEVEL INJ.
FIGURE 11
5B-53
-------
LLJ
Q
g
o
CO
DC
O 10 -
CO EMISSIONS WITH NOx REDUCTION
NSH RANGE 098-1.19
+
D Q
:~o~
LOWER LEVEL INJ
NOx EMISSIONS (PPM)
+ UPPER LEVEL INJ
180
O 48 HR DEMO
FIGURE 12
z
Q
O
^)
Q
LLJ
DC
X
O
EFFECT OF FURNACE CLEANING ON REDUCTION
OPTIMIZED INJECTION DURING 48 HH DEMO
n r
HOURS SINCE LAST SOOTBLOW
~\ T
FIGURE 13
5B-54
-------
REBURNING APPLIED TO COGENERATION NOx CONTROL
C. Castaldini
C. B. Moyer
Acurex Corporation
Mountain View, California
R. A. Brown
Electric Power Research Institute
Palo Alto, California
J. A. Nicholson
ABB Combustion Engineering
Windsor, Connecticut
-------
REBURNING APPLIED TO COGENERATION NO, CONTROL
C. Castaldini
C.B. Moyer
Acurex Corporation
Mountain View, California
R. A. Brown
Electric Power Research Institute
Palo Alto, California
J. A. Nicholson
ABB Combustion Engineering
Windsor, Connecticut
ABSTRACT
New cogeneration systems are increasingly regulated to stringent NOX levels based on control
technology precedents established in California. NOX compliance costs can be a disincentive
to cogeneration markets. This project evaluated reburning to achieve low NOX levels at lower
costs than postcombustion catalytic reduction. Subscale tests were run at the 100,000 Btu/hr
scale to simulate combustion conditions with both rich-burn and lean-burn reciprocating-engine-
based cogeneration and lean-burn turbine-based cogeneration. Results showed NOX reductions
in the range of 50 to 70 percent for rich-burn conditions with a reburn-to-engine fuel ratio of 0.2
to 0.3. Reductions with lean-burn engine conditions were nominal unless the reburn zone was
operated at a locally substoichiometric condition. For rich-burn conditions, introduction of a
metal catalyst into the reburn zone increased the NO, reductions to greater than 90 percent
by presumably accelerating the NO, reduction reactions under fuel-rich conditions.
Full-scale rich-burn reburn tests were run with a 150-kW Caterpillar engine feeding flue gas to
a new design reburn section. Over the range tested, the full-scale NOX reduction results
corroborated the subscale results. Reburn burner stability problems prevented going to
stoichiometric ratios below 0.98, however, so maximum NOX reductions were 50 percent
without the catalyst and 75 percent with the catalyst.
Pilot-scale lean-burn repower tests were run with the boiler fired at a high fuel fraction to
produce a locally substoichiometric condition. Air staging in the boiler was also used to further
improve NOX reductions. NOX reductions of 50 percent were achieved with no air staging at
boiler-to-engine fuel ratios of 1.5 and above. With air staging in the boiler, NOX reductions of
70 percent were experienced. In all configurations, reburning was very effective in destroying
90 percent or more of the CO emitted by the prime mover.
5B-57
-------
INTRODUCTION
The cogeneration of electricity and process steam has grown at a steady rate, stimulated by
favorable economics of on-site generation and by the Public Utilities Regulatory Policy Act
(PURPA). New cogeneration is expected to increase annual gas consumption by 815 billion
cubic feet per year over the next 10 years (1,2). Turbine-powered cogeneration or repower
configurations will contribute about 585 BCF of this growth, or 70 percent. Rich-burn or lean-
burn reciprocating engine-powered systems will contribute about 230 BCF, or 30 percent.
The cost of NO, controls for new cogeneration systems is increasingly taking on a larger
fraction of the total system cost. With increasingly stringent control technologies required
during permitting, the incremental costs of NO, compliance may be decisive in making
cogeneration noncompetitive. This trend is accelerating as a result of two recent regulatory
developments: the top down BACT policy, and Title I of the 1990 Clean Air Act Amendments.
The top down BACT procedure causes permit applicants to consider implementing the most
stringent NO, control technology adopted elsewhere for similar equipment. This is causing
considerable downward pressure nationwide on BACT levels set during permitting because of
the California cogeneration precedent. In several districts in California, selective catalytic
reduction is required as BACT for turbines and nonselective catalytic reduction is required for
rich-burn reciprocating engines. Title I of the 1990 Clean Air Act Amendments promotes NO,
controls for attainment of ozone air quality in areas designated as in extreme, severe, or
serious nonaftainment. This is increasing both the number of sources under control as well
as the severity of new or retrofit control levels.
In many cases in California and elsewhere, consideration of catalytic postcombustion controls
has diminished the return on investment for the cogeneration project to the point where other
energy options are preferred. The present project was initiated by the Gas Research Institute
to evaluate reburning as a means to achieve improved NO, reductions at lower costs than
postcombustion controls.
A market applications study at the outset of the project indicated that two types of
engine/boiler configurations, shown in Figure 1, could gain a significant market share with
reburning. The conventional cogeneration system, shown at the bottom normally feeds the
prime mover exhaust directly to an unfired heat recovery steam generator. For reburn NO,
control, the fuel staging is most easily done with installation of a reburner section in the engine
exhaust gas ducting to the HRSG. This configuration, shown at the top is most readily
packaged for new units. Repowering is a cogeneration alternative for existing boilers that can
be retrofitted with a reciprocating engine or turbine.
For both reburn configurations, developmental testing is needed to identify the preferred reburn
stoichiometry, temperatures, engine-to-reburn fuel ratio, and primary/reburn mixing geometry.
In the present program, testing was done in three stages to address these issues:
• Subscale 100,000 Btu/hr parametric configurational tests for rich-burn,
moderate O2, and lean-burn cogeneration conditions.
• Full-scale 150-MW rich-burn reciprocating engine cogeneration configuration
tests
5B-58
-------
• Pilot-scale one million Btu/hr repowered boiler burner configuration testing
A detailed discussion of these tests, as well as associated market applications studies and
economic comparisons, is contained in References 1 and 2.
SUBSCALE TESTS
The subscale facility used for parametric reburn configurational testing is shown in Figure 2.
The test combustor was assembled in two main sections: a 100,000 Btu/hr down-fired engine
exhaust simulator; and a reburner and burnout section. Doping with nitric oxide and CO was
done between the two sections to achieve NO levels representative of engines or turbines.
Independent regulation of natural gas and combustion air to the reburner and burnout air
downstream of the reburner allowed parametric variation of the reburner stoichiometry, SR2,
and the postreburn stoichiometry, SR3. Combustion air preheat capability was added to study
temperature effects on the reduction reactions.
Initially, a hardware screening series of tests was done to identify the sensitivity of NOX
reduction to burner geometry, and to iterate to the preferred burner design. These tests
showed that NO, reduction was sensitive to the method of reburn mixing with the engine
exhaust. For cases where the mixing was enhanced to promote NO, reduction, the percent
reduction was sensitive to the inlet level of NOX.
Based on the initial screening tests, the reburner design shown in Figure 3 was selected. Early
tests showed the benefit of the bluff body over the flame with a tight spacing to promote
mixing. The forced mixing of the reburn flame with the primary flue gas stream promoted NO,
reduction by exposing the carryover NO, from the engine simulator to the fuel-rich reactants.
With this burner, optimum performance was experienced at a reburner stoichiometric ratio of
SR2 of about 0.8. Figure 4 shows the improvement in NO, reduction with increasing fuel
fraction as the quantity of flue gas generated in the burner becomes a larger fraction of the
engine exhaust volume.
The rich-burn tests showed a significant effect of inlet NO, concentration on NO, reduction
efficiency. Figure 5 shows that for the rich-burn engine with a reburn stoichiometric ratio of 0.8
and a fuel fraction of 20 percent, the reburn efficiency decreases as carryover NO, increases.
This may indicate an increasing depletion of radical species in the fuel-rich region.
Increasing temperatures in the reheat zone is apparently effective in accelerating the reburn
reactions within the available residence time. Figure 6 shows that addition of preheated air to
the reburner improves the reduction efficiency significantly for fuel fractions of 20 and
37.5 percent. There is also a beneficial reburn effect in the downstream zone where burnout
air is injected when the reburn region is operated at an overall substoichiometric condition.
Figure 7 shows an improvement in NO, reduction of over 10 percent with a rich-burn exhaust
when reburn air is added to complete combustion.
As would be expected, the reburner acts as an afterburner for CO destruction. Figure 8 shows
that with sufficient heat addition to the reburn section, the carryover CO can be effectively
destroyed.
5B-59
-------
With lean-burn engine simulation, the NO, reductions were less effective because a
substoichiometric condition was not achieved for the fuel fractions tested. The lean-burn tests
showed that a much richer reburn stoichiometry was most effective compared to the SR2 = 0.8
optimum observed with rich-burn conditions. Figure 9 shows the improvement with richer
reburn conditions. The best reductions achieved were around 35 percent. These moderate
reductions would probably not justify use of the reburn hardware. The effect of fuel ratio was
not significant over the range of 30 to 37.5 percent tested. For the lean-burn conditions of
Figure 9, a fuel ratio of about 100 percent would be required to achieve an overall fuel-rich
reburn zone.
Exploratory tests made during the initial parametric study showed a dramatic increase in NOX
reduction when metal oxide catalysts were introduced into the reburn chamber. The potential
benefits of the concept of catalytic enhancement of NOX reduction was sufficiently strong that
the burner was modified for catalyst inserts, as shown in Figure 10. Figure 11 shows the
reduction resulting from use of a nickel oxide ceramic catalyst added at the end of the reburn
mixing zone. For an overall reburn zone stoichiometry of 0.95 or lower, the reduction of the
carryover NOX from the rich-burn engine was essentially complete. Figure 12 shows the effects
of several catalyst configurations that give variations in effective surface area. Although there
is considerable scatter, the data show that higher effective surface area strongly improves
reduction.
FULL-SCALE RICH-BURN TESTS
Based on the favorable subscale test results, a full-scale rich-burn cogeneration configuration
was tested at the Air and Energy Engineering Research Laboratory of the Environmental
Protection Agency in Research Triangle Park, North Carolina. Figure 13 shows the reburn
reaction chamber fabricated for the testing and the overall laboratory configuration. The
noncatalytic baseline and the catalytic testing agreed fairly well with the subscale tests.
Figures 14 and 15 show the NOX reduction without and with the catalyst section. Due to flame
stability problems experienced with the reburner under fuel-rich conditions, it was not possible
to test below stoichiometric ratios of about 0.99. Since NOX reduction is very sensitive to
stoichiometric ratio at these conditions, this was a constraining factor. The trends indicate that
if the stability problem was resolved, considerably higher reductions would be experienced.
Apart from the burner issue, the reburn reactor section performed well and showed promise
for sustained commercial usage.
LEAN-BURN REPOWER TESTS
The cogeneration tests discussed above centered on reburn-to-primary-fuel ratios of around
0.2 to 0.375, which would be characteristic of a duct reburn section upstream of a HRSG. For
repowering of existing boilers, the fuel ratios are much higher since the prime mover exhaust
is used as combustion air for the boiler and sufficient fuel is added to nearly deplete oxygen.
To simulate these repower conditions, the test facility shown in Figure 16 was tested. The
prime mover simulator had a firing capacity of one million Btu/hr. The exhaust from tne
simulator was directed to the primary boiler test burner. The firing rate of the prime mover
simulator together with heat exchangers and NO or CO doping were adjusted to obtain a
reasonable simulation of lean-burn turbine repowering temperatures and flue gas composition.
The boiler had additional provision for stage air above the test burner.
5B-60
-------
Three different boiler repower burners were tested to study effects of mixing NO,-bearing
combustion air with the primary boiler flame. Despite significant differences in mixing patterns,
the three burners produced comparable NO, emissions reductions. Figure 17 shows NO,
reduction results with and without stage air. The NO, reduction improved with boiler-to-engine
fuel ratio, and reductions in excess of 70 percent were experienced at representative fuel ratios
with boiler staging. Stability tests showed that turbine exhaust oxygen levels of 14 percent or
greater were needed to maintain a stable boiler flame. Repowering is effective in destroying
any carryover CO, as shown in Figure 18. The lower efficiency at low CO levels is due to
residual boiler CO concentrations.
CONCLUSIONS
The following conclusions were reached in this study:
• Reburning, without catalyst assist, reduced NO, by 50 percent at a fuel fraction
of about 30 percent. With this performance the process presents little
economic attractiveness.
• Catalyst, assist reburn was shown to achieve 70 to 99 percent NO, destruction.
This performance is required for reburn to become a viable and competitive
technology for gas-fired engine NO, control.
• Continued research is needed to evaluate catalyst and improved mixing on
NO, reduction potential and applications.
ACKNOWLEDGEMENTS
This project was sponsored by the Gas Research Institute. Dr. F. R. Kurzynske was the Gas
Research Institute Project Manager. The Coen Company assisted in selecting model burner
designs for testing. The Todd Burner Division of Fuel Tech, Inc., contributed the reburner
reactor used in the full-scale testing. The U.S. Environmental Protection Agency made available
the host site for the full-scale testing.
REFERENCES
1. Brown, R. A., Lips, N., and Kuby, W. C., "Application of Reburn Techniques for
NO, Reduction to Cogeneration Prime Movers: Volume I, Rich-Burn
Applications," GRI 88/0341, Gas Research Institute, Chicago, IL, March 1989.
2. Brown, R. W., Moyer, C., Nicholson, J., and Torbov, S., "Application of Reburn
Techniques for NOX Reduction and Cogeneration Prime Movers: Volume II,
Lean-Burn Engine Applications," GRI 90/125, Gas Research Institute, Chicago,
IL, March 1991.
5B-61
-------
Air
Natural gas
H?0 In
1
T
¥
Reburner
Rich A
i
3
9
5
Waste heat
recovery boiler
< » Flue out
Lean
Steam
out
H,0 in
A1r ^.
Air ».
Fuel ^
_ turbine >
| I
IfJ
C
t
IAAAA/I
onventlor
>o 1 1 e r
i
[_
al
Steam
out
Figure 1. Reburning Applied to Cogeneration or Repowering with Gas-Fired Prime Movers
Figure 2. Reburn Subscale Facility Schematic
5B-62
-------
6o
50 I
« 40
30 I
10
SR, = 0.8
I
_L
_L
10
15 20 25 30
Fuel fraction (percent)
Figure 3. Subscale Reburner
35
View port
2-1/2 in plunger
Gas
Figure 4. Effect of Fuel Fraction
5B-63
-------
- 2,000
DIM gu eoodttooni
- 07 pvrtwnl
T, • 1.000'F
• 100.000 Blu/ht
400
800
1,200
1,600 2,000 2,100
Input NO (ppm)
2,800 3,200
Figure 5. Effect of Input NO Concentration
60
55
'I 40
o-35
I 30
U
!2B
- 20
QJ
U
I 15
10
5
& f « 0.375 with preheat
A I - 0.375 no preheat
© f - 0.20 with preheat
• I - 0.20 no preheat
Flue gas conditions
NO. - 1.500 ppm
0, - 0.2 percent
T, - 1,100'F
- 100.000 Btu/hr
0.6
0.65
0.7
0.75
O.B
0.85
0.9
SR, rehurner stoichiometry
0.95
Figure 6. Effect of Reburner Air Preheat
1.0
5B-64
-------
60 _
~ 50 -
•IT "5 _
10
5
SR, - OB
Flue gas conditions
NO, = 1.500 ppm
Oa = 0 2 percent
T, « 1,000'F
FR^, - 100,000 Btu/hr
_L
_L
With burnout air
Without burnout
air
10 IB 20 25 30
Fuel fraction (percent)
Figure 7. Effect of Burnout Air
35
?-]/? 1n.
G«p • I/I In.
Hut gti centflttoni
MO, • 1,500
• 0.?
T3 • I.100T
-l. - 100.000 Btu/
15 20 25 30
Fuel fraction (percent)
Figure 8. CO Level Versus Fuel Fraction
5B-65
-------
!»*» |M»lfl
1-1/7 In. • Dlw«ff
ClD - J't 1".
fit* pi ttf*d< t(MI
wo, • too OP*
tj • I.IWT
n^,B . jf»,«» ltu/h'
03 04 0.5 06 07
09 10
Figure 9. Effect of Reburner Stoichiometric Ratio for Lean-Burn Conditions
SHIELD
AIR
QAS
Figure 10. Burner Configuration with Catalyst Insert
5B-66
-------
2
g
^
<_>
n
D
LJ
c:
X
O
vp
c1;
100-
80-
60-
40-
20-
Q
e u %HF
O 30X FUEL FRAC
• 20X FUEL FRAC
A 1 5X FUEL FRAC
A 1 0X FUEL FRAC
'
' 1
0.80 0.85 0.90
SR3
fcl* •.
^k
*. *°o
•&. .
w
. A
* A •
A A *
^^ 4B
. AA
A A ^
A
1 1
0.95 1.00
Figure 11. NOX Reduction with Catalyst Enhancement for a Space Velocity of 7,500 per hour
o
o
o
Ui
fy
IL.
X
O
K
90-
80-
70-
60-
50-
40-
30
*&& 6
W W
o
° .0 0
o c£
0 ^ *>
o
o
0
Of\
0
o
o
5.0 15.0 25.0 35.0 45
SPACE VELOCITY (1/HR X 10~3)
Figure 12. Effect of Space Velocity on NOX Reduction with Catalyst Enhancement
5B-67
-------
BYPASS ENGINE EXHAUST
LAYOUT FOR REBURN SYSTEM
Figure 13. Full-Scale System with 150-kW Caterpillar Engine
BASELINE - NO CATALYST
£
z
=>
0
1, |
cc
0
2
"'O
60-
50-
40-
30-
20-
10-
0-
V V
V V •
V * J^ ••»
^7
V^ V
V
A 1 25 kw LOAD • V
• 1 00 to. LOAD
• BO kw LOAD
V PILOT SCALE TESTS
1 1 1 i i i _ _ _ i
O.B9 0.91 0.93 0.95 0.97 0.99 1.01 1.03 1.C
SR3
Figure 14. Baseline Reburn NO, Reductions for Full-Scale System
5B-68
-------
100
75
50
25
• 90 kw LOAD
A 100kwLOAD
• 90 kw LOAD
V PILOT SCALE TESTS
V V
vv v
V
I
v v
0.85
0.90
0.95
SR3
1.00
1.05
Figure 15. Full-Scale NOX Reduction with Catalyst
loll
Figure 16. Laboratory Repower Test Facility
5B-69
-------
90 -
BO -
70 -
60 -
50 -
40 -
30 -
20 -
10 -
0 -
1
o EJ
J ° * °
a
° ° °.-?.°€*? ;.:< : •
+ t** *4
i i i i i i i i i ' i i i i
14 te 22 26 3 14 38
FUEL FRACTION,!
D STAGING + NONSTAGING
Figure 17. Effect of Fuel Fraction on NOK Reduction for Staging and Non-Stagin
CO REDUCTION ALL BURNERS
100 -
90 -
80 -
70 -
K
60-
g
^ 50 -
D
" 40-
o
u
30 -
20 -
10 -
n -
COLD & HOT WALL
B " V V V7 7
7 V y VV U
^ V
J
w
7
7
3
9
77
^
0.2 04 06 08 1
(Thousonds)
INITIAL CO CONCENTRATION, ppm (0 7. 02)
V WITH Si W/0 STAGING
1.2
1.4
Figure 18. Effect of Initial CO Concentration on CO Reduction
5B-70
-------
SELECTIVE NON-CATALYTIC REDUCTION (SNCR)
PERFORMANCE ON THREE CALIFORNIA
WASTE-TO-ENERGY FACILITIES
Barry L. McDonald, P.E.
Gary R. Fields
Mark D. McDannel, P.E.
CARNOT
15991 Red Hill Ave., Suite 110
Tustin, CA 92680-7388
-------
SELECTIVE NON-CATALYTIC REDUCTION (SNCR)
PERFORMANCE ON THREE CALIFORNIA
WASTE-TO-ENERGY FACILITIES
ABSTRACT
Concern over NOX emissions from municipal waste combustors (MWC) has increased to the
point where recently the EPA determined DeNOx to be BACT on several MWC facilities.
In addition, in February of this year, the EPA issued new source performance standards
(NSPS) which establish NOX limits for facilities larger than 250 tons/day, at 180 ppm,
corrected to 7% oxygen.*
Three MWC located in California were the first incinerators to install post-combustion
NOX control in the form of Exxon's Thermal DeNOx, a selective non-catalytic reduction
(SNCR) technology. Other examples of SNCR technologies which have been applied or
proposed for NOX control on MWC units include: (1) urea injection (NOXOUT), (2)
cyanuric acid (RAPENOJ, and (3) ammonium sulfate. This paper discusses the practical
(rather than the theoretical) aspects of the DeNOx technology such as: 1) installa-
tion, 2) control strategies, 3) regulatory limits, 4) system performance, 5)
startup/shutdown considerations and 6) secondary effects (i.e., plumes and increased
particulate emissions).
All NOX data presented in this paper is given on a dry basis corrected to 7%
oxygen.
5B-73
-------
SELECTIVE NON-CATALYTIC REDUCTION (SNCR)
PERFORMANCE ON THREE CALIFORNIA
WASTE-TO-ENERGY FACILITIES
INTRODUCTION
In nearly three decades, waste generation in this country has doubled, from 88 million
tons in 1960 to nearly 180 million tons in 1988. This is the equivalent of each
person in the U.S. generating four pounds of waste every day. The EPA now projects
that by 2000, we will produce 216 million tons per year, or close to 4-1/2 pounds per
person per day.
Of the 180 million tons being produced annually in 1988 roughly 76 percent was
landfilled; 11 percent was recycled; and 13 percent was incinerated. With more
stringent regulations involving the siting and operation of landfills the cost of
landfill ing has increased and the available capacity decreased. By 1992 the EPA
projects that the fraction of the nation's waste that is incinerated will have
increased to roughly 19 percent.
Recognizing the growth of incineration, currently there are approximately 130 MWC
facilities operating in the U.S., the EPA has moved to establish controls on the
emissions from these facilities. On February 11 of this year the EPA promulgated
final standards for new and existing MWC. Relative to air emissions, the New Source
Performance Standards (NSPS) established limits for new facilities for: particulate
matter, dioxins/furans, sulfur dioxide, hydrogen chloride, nitrogen oxides and carbon
monoxide. The EPA also promulgated guidelines with the intended effect to initiate
state action to develop state regulations controlling emissions from existing MWC.
The guidelines covered the same air contaminants as those covered under NSPS, with the
exception that there was no guideline given for nitrogen oxides.
The NSPS set for nitrogen oxide (NOX) emissions for new large MWC (those constructed
or modified after December 20, 1989 with a greater throughput than 250 TPD) is 180
ppm, averaged over a 24-hour period.
Currently, the Exxon Thermal DeNOx process had been operational from two to three and
one-half years on three state-of-the-art facilities built in California. It is
understandable that DeNOx was first demonstrated in California since the state and the
5B-74
-------
area regulated by the South Coast Air Quality Management District (SCAQMD), in
particular, are recognized as regions in which emission controls are especially
strict, due to regional air quality.
The first MWC in California to install Thermal DeNOx was the Commerce Refuse-to-Energy
Facility which is operated by the Los Angeles County Sanitation District (LACSD). The
Stanislaus County Resource Recovery Facility which is owned and operated by Ogden-
Martin also employs DeNOx. Finally, the third MWC to have installed Thermal DeNOx was
the Southeast Resource Recovery Facility (SERRF), which is owned by the City of Long
Beach and operated by Montenay Pacific Power Corporation.
THERMAL DeNOx INSTALLATION AND CONTROL
Mass-burn waterwall MSW incinerators are ideally suited, with respect to Thermal DeNOx
performance, as compared to utility boilers. Incinerators generally have an ideal
temperature region (1600-1800 F) in which to inject the ammonia and obtain good NOX
destruction. Furthermore, flue gas velocities are lower giving longer residence times
and there is good mixing due to overfire air ports. These factors all enhance the
performance of DeNOx on MWC furnaces.
Figure 1 provides general information on the current Thermal DeNOx installations at
the three incineration plants. The plants are remarkably similar relative to design
steam flow (each unit is large by EPA NSPS standards, throughput >250 TPD), but it is
easy to observe that the DeNOx designs differ markedly. Some of the unique designs
and operational features are:
Commerce
Stanislaus
Four injection zones are provided. The lower two injection zones
were added to assist in meeting permit conditions at reduced load
and during startup and shutdown.
Although originally equipped with an air compressor to provide 30
psi carrier air, overfire air at 1 psig is presently utilized.
This provides substantial power savings with no loss in perfor-
mance. The system configuration (Figure 2) includes purge air for
unused nozzles and remote zone selection.
Ammonia feed rate is controlled automatically based on stack NOX as
shown in Figure 3. The control logic minimizes ammonia flow and
hence ammonia slip when the emissions are within permit limits.
Reagent flow increases substantially during off nominal periods.
Two injection zones are provided, however, only the upper level is
utilized during normal operation. The lower level is utilized
during startup and shutdown transients.
5B-75
-------
• Ammonia feed rate is controlled automatically by a proprietary
control system.
SERRF
• Two injection zones are provided, however, only the upper level is
utilized during normal operation. Ammonia flow is proportioned
between the upper and lower zones using an algorithm which uses
upper furnace temperature as the only input.
• Ammonia feed rate is controlled automatically based on stack NOX
concentration.
Having worked closely on the SERRF plant it would be helpful to other facilities
considering the Thermal DeNOx technology to report some of the early work conducted
shortly after startup. Initially, NOX control was inadequate and several measurements
were taken to assess why NOX could not be maintained continually below permit limits.
Temperature profiling was performed using suction pyrometry. Sample locations are
shown on Figure 4. Temperature profiling identified three problems which prevented
the DeNOx process from adequately controlling NOX: (1) rapid flue gas temperature
swings, (2) an increasing temperature gradient from the front towards the rear wall
of the furnace, and (3) excess temperatures. Working with Dravo and Steinmuller the
combustion logic and overfire air operation were significantly modified. While these
modifications stabilized temperatures in the furnace the injection location was
determined to be too low in the furnace. Ammonia was being injected into a region
where the flue gas temperature was above the optimum for DeNOx performance and some
of the ammonia was being oxidized. The optimum temperature was located near the next
higher level of boiler nozzle penetrations. Since the upper front wall nozzle
penetrations were already in place, it was relatively simple to connect an ammonia/air
header and insert the proper nozzles. The combinations of these modifications allow
the SERRF boilers to operate in compliance with their NOX limits.
Recent operational data for Commerce has demonstrated that some flexibility in
injection location is possible for operation under steady controlled firing
conditions. Four months of operational data provided the NOX vs. load relationship
presented in Figure 5, for four separate zone combinations. Of particular interest
is the ability of one zone (or combination of zones) to provide low NOX over a wide
operating range. Although DeNOx system performance is regarded to be highly dependent
on the temperature at the point of injection, the actual window can be rather wide
when a removal efficiency of 50% is acceptable.
5B-76
-------
REGULATORY EMISSION LIMITS
Before reviewing the performance of the Thermal DeNOx systems at these three
facilities it is important to understand the regulatory limits or targets that each
facility was designed to achieve. It is interesting to note that although all three
facilities are located in California (two are even located in the SCAQMD) the
regulatory limits for each facility is uniquely different. The difference is not
solely the magnitude of allowable NOX emissions but also of particular significance
is the averaging time designated for each limit. Table 1 presents the NOX regulatory
limits for Commerce, Stanislaus and SERRF.
Each individual unit, (units are similarly sized from a steam throughput standpoint),
at the three facilities have a broad range of NOX limits to comply with. Considering
mass versus concentration limits and the five different averaging periods it is
interesting to note that there is only one common emission limit for all three
facilities. The allowable NOX emissions on a daily basis range from a low of 720
Ib/day at SERRF to a high of 1130 Ib/day at Stanislaus; Commerce has a daily, NOX
1imit of 825 pounds.
It is obvious that lower NOX emission limits are more difficult to achieve. However,
the averaging period and concentration versus mass limits have an important effect.
For example, even though Commerce, in order to avoid an emission exceedence, cannot
exceed 175 ppm for a fifteen minute period, the plant must operate below roughly 120
ppm so as not to exceed the 40 Ib/hour limit. (Note: The 175 ppm limit for Commerce
and SERRF is not in either plant's authority to construct permit but is a prohibitory
limit in SCAQMD Rule 476. Rule 476 limits the NOX concentration from liquid or solid
fuel fired units in the Basin to 225 ppm corrected to 3% 02. This value is equivalent
to 175 ppm corrected to 7% 02.)
COMMERCE NOX LIMITS
The daily NOX mass emission limit at Commerce (825 Ib/day) is equivalent to roughly
34 Ib/hr which translates to about 100 ppm. Consequently, the plant needs to operate
consistently below 100 ppm in order to comply with the daily mass limit. A safety
margin below 100 ppm would be required if frequent upsets resulting in large spikes
of NOX were to occur.
STANISLAUS NOX LIMITS
Stanislaus is unique in that NOX emissions are regulated by both the Stanislaus County
Air Pollution Control District (SCAPCD) and the EPA, due to EPA's PSD permit. The
most stringent limit from a continuous basis is the SCAPCD daily mass limit of 1130
5B-77
-------
Ib/day which is roughly equivalent to 150 ppm. Stanislaus is also unique in that the
plant has a stack ammonia limit of 50 ppm (raw).
NO^ COMPLIANCE TEST RESULTS
Emissions data taken from initial compliance tests and some more recent results are
presented in Table 2. Uncontrolled NOX data is not as plentiful as an analyst might
desire since all three plants are required to operate the DeNOx system when the plants
are on-line and/or burning refuse. To obtain uncontrolled emissions data, therefore,
a variance is required. Uncontrolled emissions are in-line with levels reported in
an EPA study, which reviewed NOX data from twenty-six mass-burn/waterwall facilities.
The study stated that the average uncontrolled NOX concentration was 242 ppm. This
is in the range of the data from Commerce, SERRF and Stanislaus. It should be noted
that the 68 ppm listed for SERRF in the EPA study was incorrect. The study stated
that the low NOX value was due to flue gas recirculation, which as previously stated,
is incorrect.
A limited amount of work was initially performed to evaluate FGR injected in the first
three undergrate zones on the SERRF units. Preliminary indications were that some NOX
reduction was achievable at a recirculation rate of roughly ten percent. Since those
early tests there have been numerous modifications to the SERRF units. In order to
establish a more definitive answer as to the effectiveness of FGR a research plan was
submitted to the SCAQMD. The goals of the research plan are:
1. to quantify the effect of FGRs contribution to NOX reduction during
simultaneous FGR/Thermal DeNOx use.
2. to quantify FGR's contribution to reduced ammonia usage and slip
during simultaneous FGR/Thermal DeNOx use, and
3. to assess the impact of FGR on primary combustion zone location and
on boiler/grate operation.
Work, under a SCAQMD research permit, is currently on-going. Along with the FGR
study, an extensive DENOX optimization program is being conducted.
Carnot conducted a DeNOx optimization program at Commerce. At Commerce the study
evaluated injection level (there were only two injection levels at the time), carrier
air injection pressure and ammonia injection rate. The study concluded that optimum
performance was achieved by injection of an NH3-to-NOx mole ratio of about 1.5 through
the upper elevation of nozzles. Carrier air pressure had no effect on DeNO
performance. Further, it was observed that even when there was substantial ammonia
slip levels at the economizer exit the level at the stack due to the spray dryer
baghouse was held to less than 5 ppm.
5B-78
-------
The controlled NOX data given in Table 2 was taken at nominal full load. The lower
levels achieved by SERRF are due to a higher rate of ammonia being injected as
compared to Commerce and Stanislaus. The higher ammonia injection rate also explains
the higher ammonia slip numbers experienced at SERRF.
STARTUP AND SHUTDOWN TRANSIENTS
With the advent of continuous emissions monitors (CEMS) plant operators are able to
observe emission levels during all operational phases. CEMS have proven to be
invaluable tools, however, some problems, which were not originally anticipated have
developed with the data they provide. Before CEM data were available, emissions were
measured using integrated sampling techniques. Normally emissions tests were
conducted at full load.
CEM data now permits plant operators to monitor emission levels during transient
conditions such as startup and shutdown. Because these periods are transients, the
emission rates are not characteristic of normal steady-state operation. Regulations
in establishing permit limits have only had to deal with what emissions are expected
to be at steady load. Once it was determined that steady state emission levels could
be exceeded during startup/shutdown transients, regulators were forced to modify
emission requirements. As an example, the SCAQMD adopted Rule 429 which recognizing
this problem provided startup/shutdown NOX relief for refinery boilers, refinery
process heaters, gas turbines, utility boilers, industrial boilers, industrial process
heaters and nitric acid plants.
Emission transients can occur for both NOX and CO during startup and shutdown. Since
Thermal DeNOx is a temperature dependent process it is critical that special
procedures be developed to control emissions during these transients. In addition,
regulators need to develop acceptable permit language which provides plant operators
sufficient margin to transition these periods safely.
IMPACT OF AMMONIA SLIP ON PARTICULATE EMISSIONS
As a result of the way particulates are defined by California regulators ammonia use
for NOX control has resulted in higher particulate values being reported. This has
caused concern among plant operators as well as particulate control suppliers who are
being asked to guarantee particulate emission levels but have no way of collecting the
gaseous components that make-up this excess particulate, which we refer to as pseudo-
particulate.
Pseudo-particulate is an artifact of the standard EPA Method 5 sampling procedure.
In the back-half of the sampling train are two impingers containing water. Normally
5B-79
-------
gaseous species pass through the water and when the impinger solution is evaporated,
there is little material found. On plants equipped with NOX control equipment which
results in some ammonia slip, the ammonia is absorbed by the water creating an
alkaline solution. The solution acts as an acid gas scrubber removing S02, HC1 and
N02, forming the associated ammonium salts. When the impinger solution is evaporated
these salts remain leaving the particulate residue referred to as pseudo-particulate.
When test protocols were being developed for Commerce, the SCAQMD accepted a procedure
which excluded the neutral salts caught in the back-half fraction. All of the
particulate tests conducted at Commerce were adjusted to exclude these neutral salts.
Similarly, the Stanislaus County APCD accepted the premise behind the particulate
adjustment and the initial particulate compliance tests at Stanislaus were corrected
for neutral salts.
Recently, however, the SCAQMD in evaluating the test protocol for SERRF concluded that
the neutral salt adjustment was unwarranted. Their logic was that since the gaseous
species combined in the atmosphere forming particulate that it was incorrect to back
them out from the particulate determination simply because the components were gaseous
when they passed through the sampling train. Consequently, particulate tests at SERRF
include this pseudo-particulate fraction. It is interesting to note that the SCAQMD
draws a distinction between plants using ammonia for NOX control and those using
ammonia for ESP performance improvement. When measuring particulates from facilities
using ammonia as an ESP performance enhancement SCAQMD allows the neutral salts
collected in the impinger solution to be backed-out of the particulate determination.
The impact of including pseudo-particulate in the particulate emission determination
is shown in Table 3.
As might be expected, the higher the ammonia slip, the more prevalent this problem
becomes. Individuals considering projects that employ ammonia or other SNCR
technologies, as well as regulators need to understand the impact ammonia can have on
particulates when setting particulate emissions levels.
IMPACT OF AMMONIA SLIP ON PLUME FORMATION
With the wide application of ammonia injection and other SNCR technologies for NO
control, there have been frequent occurrences of plumes from sources which have
chlorine in the fuel. Typically these plumes are detached but once formed continue
for long distances. SERRF has a detached plume and frequently a plume can be observed
at Commerce. Stanislaus was reported as having a plume in the past but due to the new
NO control logic has stated that a plume no longer is visible.
X
5B-80
-------
Analysis of the situation at SERRF in terms of chemical equilibrium calculations
indicates that the plume problem is explainable in terms of ammonium chloride (NH4C1)
condensation in the atmosphere above the stack. These calculations also show that
ammonium sulfate or bisulfate should not be contributing factors.
Principles of chemical thermodynamics show that NH4C1 condensation is governed by the
product of NH3 and HC1 concentrations in the stack ([NH3] x [HC1], the "concentration
product") and the stack and ambient temperatures. The thermodynamic relationship
showing the critical value of [NH3] x [HC1] above which condensation will occur versus
temperature is shown in Figure 6. For any combination of stack temperature, ambient
temperature and concentration product in the stack, there is a dilution vector on
Figure 6 along which the stack conditions will decay as ambient air mixes with the
flue gas leaving the stack.
Once NH4C1 forms, its visibility is dependent upon plume diameter. This is known to
be a logarithmic dependence for simple opacity but becomes more complicated when back
scattering is included, which must be the case for a white plume. The plume diameter
is, of course, related to stack diameter and air infiltration.
Based on a study conducted at SERRF, to avoid NH4C1 formation requires extremely low
values of NH3 and/or HC1 concentrations, such that NH3 x HC1 does not exceed approxi-
mately 10"4 ppm2. This criterion is impractical for SERRF to achieve and total
avoidance of NH4C1 formation therefore does not appear to be an option. Further, the
plume visibility is essentially proportional to the lesser concentration of NH3 and/or
HC1.
SUMMARY
Thermal DeNOx is successfully providing adequate NOX control such that Commerce, SERRF
and Stanislaus can meet their individual NOX emission permit limits. Furthermore, all
three plants operate below the NSPS NOX limits recently promulgated by the EPA.
Critical to the success of this technology is stable combustion and the ability to
inject and properly mix the ammonia at the proper optimum flue gas temperature. When
done correctly, continuous NOX compliance is possible.
By reducing the time intervals by which compliance is monitored, plants are forced to
operate at lower NOX levels to avoid emission upsets associated with variations in
feed quality or equipment upsets. Furthermore, the use of ammonia injection is not
without secondary problems, specifically potentially higher particulate emissions,
depending on what regulatory agencies define particulate to be, and visible plume
formation.
5B-81
-------
Four Side Wall
(8) NH3 Injection
Nozzle Locations
COMMERCE REFUSE-TO-ENERGY FACILITY:
:Unlts:X X X (1)330-400 TPD XX:
Foster-Wheeler (115,000 Ib/hr)
Detroit. ..'xx X. . .X /XXXXXX
.4 Levels oh Both . .XX
Side Walls X ':• XX. .X..
Boiler Cross-Section: iSis'fw) x 18'(d)
Stoker:
•;NHi Injection;
Two Front Wall
(10) NH3 Injection
Nozzle Locations
STANISLAUS COUNTY RESOURCE RECOVERY
FACILITY:
Units:
. Boiler;
Stoker:
NH3 Injection:
Boiler Cross-Section:
(2) 400 TPD XX
Zurn {Not Available) .. ..
Martin v.. ...Y
2 Levels on the
Front Wall
(Kot Available) : .
Front Wall (15)
and Side Wall
(23) NH3 Injection
Locations
SOUTHEAST RESOURCE RECOVERY
FACILITY (SERRH:
Units:
Boiler.
Stoker:
NH3 Injection
(3) 460 TPD . .. . ..:
L&C Stelnmueller (117,170 Ib/hr)
L&C Stelnmueller
2 Levels, Front Wall
and B.oth Side Walls
Boiler Cross-Section: 19'(w) x 18'(d)
Figure 1. Various Ammonia Injection Configurations at Three
California MSW Incinerators Equipped with Thermal DeNOx
5B-82
-------
CARRIER/PURGE AIR
OVERFIRE AIR
FAN - 30" H20
AMMONIA STORAGE
F
*GE J
t^~A
VAPORIZE
"I
INJECTION
ZONE
Figure 2. Commerce Ammonia Receiving, Storage and Delivery System
3
u.
o
80 n
60-
40-
20-
Limit Needed to Meet
Daily NOx Limit
0 50 100 150
NOx - PPMc at 7% Oz
Figure 3. Commerce Refuse-To-Energy Facility
Ammonia Feed Rate vs. NOx
200
5B-83
-------
F1.F2
D
G1.G2
D
E1.E2
D
A1.A2 Bl.BI C1.C2
ODD
Furnace Penetrations for
Ammonia Injection Nozzles
Upp»f«mmofit«ifi)«ctlon penetration
ptarw ffflONT WALLHAS 1S NOZZLES)
Micfcte ammprita ipiection pkne
(23 NOZZLES 1CCATED ON BO1>|
SIDEWAU5J
- (El, -S
Lowera
plan* (21 NOZZLES LOCATED ON BOTH
SIDE WALLS;
Figure 4. North side schematic of a typical SERRF Steinmuller-designed furnace.
Observation ports through which temperature profiling was performed are
shown.
M
o
f«-
4-1
CO
Q.
I
X
O
200 i
150-
100-
50-
• ZONE 3
• ZONE 3,4
^ ZONE 2,3,4
n ZONE 2,3
20
40
1
60
1
80
100
%MCR
Figure 5. Commerce Refuse-To-Energy Facility NO, vs. Load
Utilizing Various Injection Zones.
5B-84
-------
15
10
Q.
Q.
IT
O
T.
X ,
n O
T.
Z
X
D)
O
(5)
(s)
+ HCI (g)
Explanation:
At any given temperature,
condensation will occur if
the log of the product of
mole-fractions XNH3-XHCI,
expressed as ppm2, lies
above the curve.
100
200
300
400
500
600
700
Temperature F
Figure 6. NH4C1 Equilibrium Curve
5B-85
-------
TABLE 1
REGULATORY LIMITS FOR COMMERCE, STANISLAUS, AND SERRF
Plant
Air Quality District
Commerce
Stanislaus
SERRF
South Coast Stanislaus County South Coast
AQMD APCD AQMD
Pollutant
NOX ppm G> 7% 02
NOX ppm G> 7% 02
NOX ppm (? 7% 02
NO" Ib
NOX Ib
NH3 ppm (raw)
EPA-PSD
More stringent of
NOX ppm 0 7% 02
or
NO -Ib
and
More stringent of
NOX ppm @ 7% 02
or
NOX Ib
Averaging
Period
15 min. 175
1 hour
8 hour
1 hour 40
24 hours 825
--
3 hour
3 hour
24 hour
24 hour
--
200
--
1130
50
175
160.5
165
1200
175
116
--
34
720
--
--
--
--
~ —
NOTE: The EPA NSPS NOX limit for MWC which are larger than 250 TPD is 180 ppm NOX
averaged over 24 hours.
5B-86
-------
TABLE 2
COMPARISON OF NOX EMISSIONS FROM THREE CALIFORNIA MSW
INCINERATORS EQUIPPED WITH THERMAL DENOX
Uncontrolled NOX
ppm @ 7% 0,
Ib/hr
Controlled NOX
ppm G> 7% 0,
Ib/hr
Ammonia Slip
ppm (raw)
Commerce
128-217
44-75
104
35.8
-2
Stanisl
Unit 1
298
90.4
93
28.1
3.7
aus
Unit 2
305
96.0
112
36.0
5.0
SERRF
Unit 1 Unit 2
210
74.8
49 72
16.5 22.7
--
Unit 3
259
93.1
54
17.9
35
TABLE 3
PARTICULATE EMISSIONS AND THE IMPACT OF ADDING BACK
THE PSEUDO-PARTICULATE FRACTION
Permit Limit
Test Results
% of particulate
Commerce
5.5 Ib/hr
2.5
88%
Stani
0.0275
Unit 1
0.011
51%
si aus
gr/sdcf
Unit 2
0.011
79%
SERRF
5.0 Ib/hr
Unit 3
1.7
70%
caught in the back-half
of the sample train
Impact on particulate
level if neutral salts
were added back
60%
+ 34%
+38%
N/A
5B-87
-------
USE OF NATURAL GAS FOR NOX CONTROL
IN MUNICIPAL WASTE COMBUSTION
H. Abbasi and R. Biljetina
Institute of Gas Technology
3424 South State Street
Chicago, Illinois 60616
F. Zone and R. Lisauskas
Riley Stoker Corporation
Riley Research Center
45 McKeon Road
Worcester, Massachusetts 01610
R. Dunnette
Olmsted Waste-to-Energy
2128 Campus Drive, S.E.
Rochester, Minnesota 55904
K. Nakazato
Itoh Takuma Resource Systems Inc.
335 Madison Avenue
New York, New York 10017
P. Duggan and D. Linz
Gas Research Institute
8600 West Bryn Mawr Avenue
Chicago, Illinois 60631
-------
USE OF NATURAL GAS FOR NOX CONTROL
IN MUNICIPAL WASTE COMBUSTION
ABSTRACT
Natural gas injection (NGI) technology for reducing NOX emissions from
municipal waste combustors (MWCs) is being developed in a joint
program between the Gas Research Institute (GRI), the Institute of Gas
Technology (IGT), Riley Stoker Corporation (Riley), Olmsted Waste-to-
Energy (Olmsted), and Takuma Company, Ltd. (Takuma). The approach
developed by IGT and Riley (termed METHANE de-NOx) is based on
extensive, full-scale, MWC in-furnace characterization followed by
pilot-scale testing using simulated combustion products that would
result from the firing of 1.7 X 106 Btu/h (0.5 MWth) municipal solid
wastes (MSW). The approach involves the injection of natural gas,
together with recirculated flue gases (for mixing), above the grate to
provide reducing combustion conditions that promote the destruction of
NOX precursors, as well as NOX. Extensive development testing was
subsequently carried out in a 2.5 X 106 Btu/h (0.7 MWth) pilot-scale
MWC firing actual MSW. Both tests, using simulated combustion
products and actual MSW, showed that 50% to 70% NOX reduction could be
achieved. These results were used to define the key operating
parameters.
A full-scale system has been designed and retrofitted to a 100-ton/day
Riley/Takuma mass burn system at the Olmsted County Waste-to-Energy
facility. The system was designed to provide variation in the key
parameters to not only optimize the process for the Olmsted unit, but
also to acquire design data for MWCs of other sizes and designs.
Extensive testing was conducted in December 1990 and January 1991 to
evaluate the effectiveness of NGI. This paper concentrates on the
METHANE de-NOx system retrofit and testing. The results show
simultaneous reductions of 60% in NOX, 50% in CO, and 40% in excess
air requirement with natural gas injection.
5B-91
-------
USE OF NATURAL GAS FOR NO CONTROL
IN MUNICIPAL WASTE COMBUSTION
UTILIZATION OF NATURAL GAS IN MUNICIPAL WASTE COMBUSTORS (MWCs)
In 1986, following GRI's successful pilot-scale testing of natural gas
reburning for NOX reduction in coal-fired applications, GRI and IGT
began an investigation of the potential for utilizing natural gas in
MWCs for the control of NOX emissions. At that time the control of
NOX was required in the State of California; however, it was not yet
being seriously discussed elsewhere in the United States. By 1989,
the U.S. Environmental Protection Agency had announced its intention
to set limits for NOV emissions from all MWCs. The limits being
X
evaluated were based on the performance of the thermal de-NOx process,
which uses ammonia injection to reduce NOX emissions. The thermal de-
NOX process has been installed on three MWCs operating in California.
Figure 1 illustrates the NOX reduction approach proposed for MWCs.
This approach, termed METHANE de-NOx, involves the injection of
natural gas, together with recirculated flue gases (for mixing), above
the grate to provide reducing combustion conditions that promote the
destruction of NOX precursors, as well as NOX. Secondary overfire air
(OFA) is then injected at a higher elevation in the furnace, after
sufficient residence time at these reducing conditions, to burn out
the combustibles. Applying this approach to MWCs is challenging
because of the low heat content of the waste being fired, the presence
of significant amounts of NOX precursors (for example, NH3, HCN) above
the grate, and the high excess air levels that are typically used in
these types of combustors. These conditions result in relatively low
temperatures and high oxygen and NOX precursor levels in the primary
combustion zone compared with conditions in the same location in a
coal-fired boiler. Further complexities include the distribution of
air, which includes a relatively large amount through the burnout
grate at the discharge end of the combustor, and a large amount of air
infiltration due to the negative operating pressure of the combustor.
Also, because of the variability of the waste being burned, conditions
in the furnace are typically variable. The initial concern,
therefore, was that if NGI could be made to work at all in MWCs, it
5B-92
-------
might require either large amounts of natural gas, or extended furnace
zones to increase the residence time, or both.
The objectives of the development program were to 1) characterize the
in-furnace conditions of a commercial MWC to define the variability of
operation, the gas compositions within the furnace, and the flow
distribution patterns for oxygen, CO, NOX, and other flue gas species,
2) evaluate the gas-phase chemistry in laboratory furnace simulation
experiments (0.5 MWth) and define regions of operation in which NGI
could be effective using simulated MWC flue gases, 3) design and build
a pilot combustor (0.7 MWth) firing actual MSW, in which the NGI
process could be developed and tested, and 4) design and conduct a
full-scale evaluation of the NGI process on a commercial MWC.
The experimental program was conducted from 1987 to 1989. The
installation of the full-scale field evaluation was completed in late
1990, and NGI testing was completed in January 1991. The remainder of
this paper summarizes the research conducted over the last 3 years
that led to the design of a full-scale system and the results of NGI
testing on the full-scale commercially operating MWC.
RESULTS OF COMMERCIAL COMBUSTOR CHARACTERIZATION
The baseline data were acquired on one of the two units at the Olmsted
County Waste-to-Energy Facility (Figure 2) located in Rochester,
Minnesota. The design of the combustor is an integration of the
Takuma MWC stoker and combustion control technology with the Riley
waterwall furnace technology. Each unit was designed to burn MSW at
the rate of 100 tons/day (90 metric tons/day), producing about
24,000 Ib/h (11,000 kg/h) of 615-psig (42-bar) superheated steam.
The unit was tested while varying load, total stoichiometric ratio
(TSR), allocation of undergrate air (UGA) flow, and OFA location. Two
general types of tests were conducted: in-furnace measurements by IGT
and overall system performance data acquisition by Riley. Test
details have been presented earlier (1) and the results are briefly
described below.
5B-93
-------
In normal operation, with 60% to 80% excess air to ensure complete
combustion, this unit produced about 125 to 175 ppm* NOX- Without OFA
and at lower excess air, NOX emissions were reduced significantly, but
CO and total hydrocarbon (THC) emissions increased greatly. The
baseline data show that NO can be reduced by eliminating OFA and
reducing excess air; however, incomplete combustion results — as
indicated by the high CO levels. The goal of NGI is to reduce NOX
emissions without the corresponding increase in CO emissions. The
furnace characterization data that were acquired also show that it
would be possible to create the substoichiometric NOX reducing
conditions within the furnace with NGI.
Furnace Simulator
A pilot furnace at IGT was fired with No. 2 fuel oil using preheated
air and adding appropriate amounts of oxygen, moisture, and ammonia
(to simulate fuel-bound nitrogen). Thus, the pilot furnace closely
simulated the baseline combustion products from the stoker firing
1.7 X 106 Btu/h (0.5 MWth) of MSW. Tests investigated the impacts of
reducing zone residence time, stoichiometry, and gas temperature;
amounts of natural gas and fuel bound nitrogen; overall excess air;
and the amount of flue gas recirculation (FGR) for mixing the natural
gas with the combustion products. These test details have also been
presented earlier (2.3).
In typical excess air operation (without NGI), the furnace simulator
produced relatively steady NOX levels of 200 to 225 ppm — independent
of residence time. As illustrated in Figure 3, however, residence
time plays an important role when natural gas is injected, because
sufficient time must be available for the natural gas to decompose NOV
A.
precursors. The first 3 seconds after NGI reduced NOX from 225 to
75 ppm. Longer times produce very little additional NOX reduction.
The results showed that if NGI is to be effective, it must be injected
into the MWC such that sufficient residence time at high temperatures
is provided before OFA is injected for combustible burnout. An NGI
level of 15% was found to be sufficient for 50% to 70% NOV reduction.
* All of the NOX and CO emission values presented here are on a 12% O2
and dry basis. For a 3% 02 basis, multiply values by 2 and for a 7%
02 basis, multiply by 1.56.
5B-94
-------
Pilot MWC Combustor
Because of the encouraging furnace simulator test results, it was
decided to make follow-up tests in the pilot combustor at Riley's
Research Center. A pulverized coal combustor at Riley was modified to
simulate the commercial unit at Olmsted, and several different batches
of MSW were tested to investigate the impacts of reducing zone
residence time and stoichiometry, natural gas injection location and
amount, and overfire air injection location. The results have been
presented earlier (3,4) and show that without NGI, NOX emissions
ranged from 110 to 165 ppm — a fairly good simulation of the baseline
results obtained in the commercial combustors. With 10% to 15%
(percent of total heat input) NGI, NOX emissions were reduced by as
much as 70%, depending on the natural gas and OFA injection points and
the residence time in the reducing zone. NO emissions decreased from
100 to 130 ppm at 0.6 seconds residence time and 40 to 80 ppm at
1.2 seconds residence time. These results verify the beneficial
effects of residence time as observed in the furnace simulator tests.
A reducing zone stoichiometric ratio of between 0.8 to 1.0 was found
to be sufficient for effective NOX reduction. With NGI, it was also
possible to operate the unit with significantly lower excess air.
FIELD EVALUATION OF NATURAL GAS INJECTION
In light of the favorable test results obtained from both the IGT and
Riley pilot-scale investigations of NGI, a field evaluation was
undertaken. The NGI technology was retrofitted to one of the Olmsted
units. This facility was also used to acquire all the baseline data
reported here.
The pilot-scale work had demonstrated the potential of NGI for
reducing the emissions of NOX, CO, and THC. A number of issues
remained, however, before it could be commercialized as a viable
emissions reduction technology. The major issues were as follows:
• Can NGI be as effective on a commercial unit,
considering the actual conditions of high excess oxygen
and the variability of feed quality and operating
temperature?
• Can the already low CO and THC levels (<50 ppm) be
further lowered and stabilized on the full-scale unit,
as evidenced in the pilot unit?
5B-95
-------
• Can proper furnace aerodynamics be maintained or
improved? In other words, can adequate distribution of
natural gas in the reducing zone and OFA in the burnout
zone be accomplished in full-scale systems?
• What would be the impact on thermal efficiency,
slagging, corrosion, steam superheat, and other boiler
performance parameters?
• What are the costs and advantages over thermal de-NOx
and/or other alternative NOX control measures?
The results of the field evaluation would help resolve many of these
issues. As with the experimental program, this 15-month effort was
conducted jointly by IGT and Riley in consultation with Olmsted and
Takuma. The work effort was divided into three major activities. The
first involved finalization of site selection and engineering and
design of a flexible NGI retrofit system. The second was the
procurement and installation of the retrofit system. The third was
the field evaluation testing of NGI for emissions reduction, as well
as other impacts, which began in early December 1990 and was completed
in late January 1991.
The primary goal was to reduce NOX to below 70 ppm from the current
uncontrolled level of over 140 ppm without adversely affecting other
emissions such as CO and THC. Additional goals were to maintain or
improve the steam capacity while increasing the boiler thermal
efficiency.
The retrofit METHANE de-NOx system was designed by IGT and Riley based
on the pilot-scale testing results. The primary variables (determined
during the pilot testing) for design of the NGI system are -
• 15% natural gas above grates to create substoichiometric
conditions
t 15% FGR above grates for mixing the natural gas with the
furnace gases
• Variability in reducing zone stoichiometry; reducing
zone residence time; and natural gas, FGR and OFA flows,
injection locations, and velocities.
The retrofit included installation of an FGR system and modification
of the furnace walls to accommodate several nozzles and
sampling/observation ports at multiple levels. The design also
5B-96
-------
provides for acquisition of the necessary in-furnace and flue gas
composition and temperature data, as well as other relevant data.
Recirculated flue gas, taken from the economizer outlet, is used to
introduce natural gas above the stoker.
OFA injectors are installed in two locations in the upper half of the
furnace for combustible burnout. The two elevations enabled testing
of different residence times for the reducing zone. Residence time
has a significant effect on NOX reduction and combustible burnout.
Inserts were employed during the testing to evaluate higher injection
velocities for the OFA, natural gas, and FGR.
FIELD EVALUATION TESTS
Extensive testing was carried out on the 100-ton/day commercially
operating MWC during December 1990 and January 1991. These tests
investigated the impacts of the following variables.
• OFA location - to change the residence time in the
reducing zone
• OFA amount, injector size, and number of injectors — to
optimize combustible burnout
• Natural gas and FGR amounts, distribution, injector
sizes, and injector locations — to modify reducing zone
mixing
• UGA amount and distribution — to modify MSW combustion
profiles.
As indicated earlier, the objective of the testing was twofold:
1. To prove the effectiveness of natural gas in reducing
the NOX emissions on a
without any adverse effects
the NO emissions on a full-scale commercial unit
2. To acquire design data for the application of the NGI
technology to MWCs of other sizes and designs.
As a result, the system was instrumented to provide an extensive data
base for the impacts of NGI on both the furnace side, as well as the
steam side parameters. The following is a list of measurements made
during the tests.
5B-97
-------
• Full spectrum of furnace and steam side operating data
including temperatures, flows, pressures, etc. through a
specially installed computer data acquisition system and
manually
• Gas composition (O2, CO, THC, CO2, NO ) and temperature
profiles in the reducing zone below the OFA injectors
and at the furnace exit above the OFA injectors
• Flue gas composition (O2, CO, CO2, NOX) at the
electrostatic precipitator (ESP) inlet
• Flue gas composition (O2, CO, NOX) in the recirculated
flue gases
• Oxygen concentration in the reducing zone (continuously)
• Ash samples
• MSW samples.
The in-furnace gas composition and temperature measurements were made
using water-cooled gas sampling and suction pyrometer probes that were
installed at various elevations to traverse the furnace. Two sets of
continuous emission monitors were employed. One set of O2, CO, CO2,
and NO analyzers was installed near the ESP to measure the gas
ji.
composition at the ESP inlet; and another set of O2, CO, THC, CO2, and
NO analyzers was installed in the control room to measure the gas
compositions inside the furnace and in the recirculated flue gases.
The gas composition at the ESP inlet was measured continuously for the
duration of each test, while the gas composition in the recirculated
flue gases was measured periodically between the in-furnace traverses.
The moisture contents of the flue gases and the flue gas flow rates
were also measured during some of the tests.
The extensive data that were acquired during the field evaluation
tests have not been fully reduced and analyzed at this writing. The
composition of the actual MSW burned during the tests is also not yet
available. Consequently, the data presented here are limited. The
results will focus on NOX and CO emissions measured at the ESP inlet
and their preliminary relationships with some of the significant
operating parameters. In general, these relationships were consistent
with the pilot-scale results. The data presented here are further
limited to the configurations that provided the optimum results with
NGI. Data are presented for three types of tests. First, these data
5B-98
-------
are presented with the baseline configuration as the unit is normally
operated; second, in the NGI configuration with FGR injected into the
lower furnace and OFA moved up to a higher elevation; and third, also
in the NGI configuration with both FGR and natural gas injected into
the lower furnace and OFA injected at the higher elevation.
Table 1 summarizes the average values of selected operating data, as
well as CO and NOX emissions for these three test configurations.
Data are also presented from the 1987 baseline testing and for one
test with NGI that was carried out at a higher steam flow to maintain
the MSW rate at the current normal baseline value of 7000 Ib/h. The
MSW feed rate and the total flue gas flow rate shown have been
estimated assuming typical MSW composition and heat content. The
actual values might be somewhat different, but the trends are expected
to be unaltered. It must be noted that the steam flow during the 1991
baseline test was about 28,250 Ib/h or 6% higher than the current
normal baseline steam flow of 26,700 Ib/h, and 20% higher than the
1987 baseline level of 23,500 Ib/h. During most of the tests with
NGI, the steam flow rate was maintained at 29,000 Ib/h or 9% higher
than the current normal baseline level (as there was no need for the
additional steam) which automatically decreased the MSW feed rate to
the 1987 baseline value. However, as shown, one test was carried out
with the MSW rate maintained very close to the current normal baseline
level by increasing the steam flow by about 14%. This was to prove
that NGI retrofit may not necessarily require a decrease in MSW feed
rate. Table 1 shows that 12.5% to 14% (total heat input) NGI allowed
a reduction in excess air from over 70% to about 40% which may
increase the boiler thermal efficiency.
The data presented in the table also show that, compared to the 1991
baseline test, NGI decreased the NOX emissions by 60% and CO emissions
by 50%. The NOX emissions were decreased by 40% with FGR alone,
however, the CO emissions were more than double compared with the
average CO with NGI. The CO level with FGR was comparable to the 1991
baseline test value, but higher than the average value for the 1987
baseline tests. Figure 4 illustrates the relationship between NOV and
A
CO emissions for the Olmsted combustor that was found in 1987 for the
baseline operation. The relationship represents baseline operation at
different UGA and OFA flow distributions and excess air levels. The
5B-99
-------
current (1990-1991) data at baseline configuration, as well as with
FGR, show scatter but appear to follow the 1987 trend. The average
NOX/CO values with FGR fall close to the average baseline curve. This
suggests that the effectiveness of FGR in reducing NOX may not be
significantly better than some of the other simpler combustion
modifications that were tested in 1987. The figure also illustrates
the effectiveness of NGI in controlling both NOX and CO emissions
simultaneously. Both NOX and CO emissions were significantly lower
with NGI. The average baseline NOX at 32 ppm CO (expected regulatory
limit) was about 137 ppm while the average NOX with natural gas was
about 50 ppm at an average CO level of about 22 ppm.
SUMMARY OF RESULTS
As discussed, the data acquired during the field evaluation tests have
not yet been fully reduced and analyzed. Based on the current
analysis, however, the following can be stated:
• In general, the relationships between the significant
operating parameters and the emissions were consistent
with those found on the pilot-scale units.
• Proper injection of 12% to 15% (heat input basis)
natural gas simultaneously decreased the NO emissions
to below 50 ppm and the CO emissions to below 25 ppm,
which represents a 60% reduction in NOX and a 50%
reduction in CO compared to the 1991 baseline test
values.
• NGI also allowed a reduction in excess air to 40% (from
the baseline levels of 70% to 80%), which may provide an
increase in boiler thermal efficiency.
• An FGR level of 6% to 8% was sufficient to inject and
effectively mix the natural gas with the furnace gases.
• Because of the reduced excess air requirement, it was
possible (as demonstrated in one test) to maintain the
MSW feed rate at the baseline level by increasing the
steam output to accommodate the additional heat input
with natural gas.
In conclusion, the effectiveness of the METHANE de-NOv process for
A
controlling NOX and CO emissions from MWCs has now been demonstrated
on a commercially operating MWC. Further analysis of the data should
provide additional information for application of this process to MWCs
of other sizes and designs, including refuse derived fuel (RDF).
5B-100
-------
ACKNOWLEDGMENT
Many sponsors played important roles in the development of the METHANE
de-NOx process. Considerable funding and guidance were provided by
the Gas Research Institute, Brooklyn Union Gas Co., Minnegasco,
Northern Illinois Gas Co., Northern Natural Gas Co., Peoples Gas Light
and Coke Co., Southern California Gas Co., and IGT's Sustaining
Membership Program member companies
The Olmsted County Waste-to-Energy officials and plant personnel
warrant special thanks for interrupting commercial operations to not
only accommodate but also vigorously assist the researchers in the
birth of a new process that can serve both the waste-to-energy and
natural gas industries.
REFERENCES CITED
1. Fleming, O.K., Khinkis, M.J., Abbasi, H.A., Linz, D.G. and
Penterson, C.A. "Emissions Reduction From MSW Combustion Systems
Using Natural Gas." Paper presented at the Conference on Energy
From Biomass and Wastes, XII, New Orleans, Louisiana, February
15-19, 1988.
2. Abbasi, H. , Khinkis, M.J., Itse, D., Penterson, C. , Wakamura, Y.
and Linz, D. "Development of Natural Gas Reburning Technology
for NO.. Reduction From MSW Combustion Systems." Paper presented
at the 1989 International Gas Research Conference, Tokyo, Japan,
November 6-9, 1989.
3. Emissions Reduction From MSW Combustion Systems Using Natural
Gas. Task 2. Pilot-Scale Assessment of Emissions Reduction
Strategies. GRI-90/0145 Final Report, Institute of Gas
Technology and Riley Stoker Corp., July 1990.
4. Penterson, C.A., Itse, D.C., Abbasi, H.A., Khinkis, M.J.,
Wakamura, Y. and Linz, D.G. "Natural Gas Reburning Technology
for NOX Reduction From MSW Combustion Systems." Paper presented
at the ASME 1990 National Waste Processing Conference, Long
Beach, California, June 3-6, 1990.
5B-101
-------
Undergrate Air
Overflre Air
Natural Gas/
Reclrc. Flue
Gases
Figure 1. The METHANE de-NOx Process
Figure 2. Olmsted Waste-to-Energy Facility
5B-102
-------
250
200
Q_
Q.
- 150
X
O
100
n
D
50
012345
Residence Time, seconds
Figure 3. Residence time plays a
significant role in the effectiveness
of natural gas
180
160
140
p 120
Q_
D.
-100
80
60
40
20
D
Baseline 87
D
Baseline 91
A
FGR Only
O
FGR + Natural Gas
O
D
10 20 30 40 50 60 70 80 90
CO, ppm
Figure 4. Natural gas injection
simultaneously decreases NO and CO
emissions
5B-103
-------
Table 1
AVERAGE OPERATING DATA - 1990/1991 FIELD EVALUATION TESTS
en
CD
MSW,* Ib/h
Natural Gas, %
Total Heat Input,* 106 Btu/h
FGR, %
Excess Air, %
Total Flue Gas,* Ib/h
Steam Flow, Ib/h
Economizer Exit Temperature, °F
Precipitator Inlet
02, %
CO, ppm at 12%
°2
Baseline
1987
Test
6,450
0
33.5
0
73
44,800
23,500
417
9.3
30
135
1991
Test
7,760
0
40.3
0
76
54,100
28,250
425
10.5
46
117
FGR Only
(Average
Data)
—
0
—
9.5
54
47,100
27,670
423
7.6
47
70
FGR +
At Normal
1987
Baseline
MSW Input
(Average
Data)
6,500
14.0
39.9
9.5
37
45,400
29,000
422
6.5
22
48
NGas
At
Normal
1991
Baseline
MSW Input
Test
7,000
12.4
41.9
10.0
41
48,500
30,500
422
5.9
21
48
*Estimated.
-------
Session 6A
POST COMBUSTION DEVELOPMENTS II
Chair: D. Drehmel, EPA
-------
PERFORMANCE OF UREA NOx REDUCTION SYSTEMS
ON UTILITY BOILERS
Andris R. Abele, Yul Kwan, and M.N. Mansour
Applied Utility Systems, Inc.
1140 East Chesnut Avenue
Santa Ana, California 92701
N.J. Kertamus and Les J. Radak
Southern California Edison Company
2244 Walnut Grove Avenue
Rosemead, California 91770
James H. Nylander
San Diego Gas and Electric Company
4600 Calsbad Boulevard
Carlsbad, California 92008
-------
PERFORMANCE OF UREA NOX REDUCTION SYSTEMS
ON UTILITY BOILERS
Andris R. Abele*, Yul Kwan, and M.N. Mansour
Applied Utility Systems, Inc.
1140 East Chestnut Avenue
Santa Ana, California 92701
N.J. Kertamus and Les J. Radak
Southern California Edison Company
2244 Walnut Grove Avenue
Rosemead, California 91770
James H. Nylander
San Diego Gas and Electric Company
4600 Carlsbad Boulevard
Carlsbad, California 92008
ABSTRACT
Test results from the full-scale application of urea injection for NOX reduction on two utility
boilers demonstrate the sensitivity of urea NOX reduction performance to boiler design,
operating conditions, and urea process variables. The two utility boilers are both gas- and
oil-fired boilers, but of different size and design. The demonstration sites include a
Southern California Edison Company 320 MW tangentially-fired boiler and a San Diego Gas
and Electric Company (SDG&E) 110 MW front wall-fired boiler.
The performance of the urea NOX reduction process at the two sites was dominated by
variables affecting the temperature at the injection location and the mixing of urea with the
combustion products. Varying operating conditions, such as load and firing configuration,
changed the temperature distribution in the boilers as well as initial NOX levels. Such
changes affect the relative location of urea injectors within the urea reaction temperature
window and, thus, the level of NOX reduction achieved. Available injection process
variables, including injector design, solution flow and pressure, injector location and spray
orientation, were used to optimize the distribution of urea within the reaction window at
varying loads to achieve maximum NOX reduction.
Minimum NOX emissions were achieved at both sites by coupling urea injection with
modified combustion conditions. Urea NOX reduction performance at these modified
operating conditions was about 30 percent at NSR = 2.0 over the boilers' load ranges.
Resulting stack NOX emissions at both units were 20 to 45 ppm @ 3 % O2 depending on
load, while ammonia slip was less than 20 ppm.
* Currently with the South Coast Air Quality Management District.
6A-1
-------
PERFORMANCE OF UREA NOX REDUCTION SYSTEMS
ON UTILITY BOILERS
IMPLEMENTATION OF THE UREA NO, REDUCTION PROCESS
The urea NOX reduction process is a selective non-catalytic reduction (SNCR) process which
encompasses a sequence of steps. Aqueous urea solution is pumped to injection nozzles
which spray the chemical into a boiler or furnace chamber. The droplets of injected
solution evaporate and the urea thermally decomposes into reactive species. The urea
droplets and released reactive species mix with the NOx-laden combustion products. Urea
species then react with the combustion products at the proper temperatures to reduce nitric
oxide (NO) to elemental nitrogen (N2).
The NO-reducing reactions are temperature sensitive and occur within a narrow temperature
range. If the urea is released at too high a temperature, the chemical species can actually
be oxidized to NOX. If the urea is released at low temperatures, the NO-reducing reaction
rates are limited and result in poor chemical utilization.
An additional complication in SNCR systems is that these temperature sensitive reactions
must occur not in a well controlled reactor, but in a load-following utility boiler. The
design of these systems must address the issues of temperature variations and mixing
limitations to the extent possible. Since a utility boiler presents a far from perfect reaction
chamber environment, efficient utilization of injected urea is not possible for all boiler
operating conditions. Since the process is imperfect, excess urea must be injected to
maximize the availability of NOx-reducing species within the narrow reaction window
provided within utility boilers. Unutilized ammonia (NH3) will be a result if the injection
temperature is too low. At high injection temperatures, excess NH3 is oxidized to NOX,
defeating the purpose of reducing NOX emissions. Thus, tradeoffs will exist between NOX
reduction and overall process performance.
To understand the effectiveness of the urea injection process, the term Normalized
Stoichiometric Ratio (NSR) was defined as the ratio between the actual amount of urea
injected and the theoretical amount required to react with all the NO present. For example,
a urea flowrate of NSR =1.0 provides the exact amount of urea to react with 100 percent
of the NO present. This Stoichiometric ratio of NSR = 1.0 is equivalent to a urea to NO
mole ratio of 0.5, since one mole of urea (NH2CONH2) potentially has two moles of
nitrogen species (e.g., NH;) available to react with NO.
6A-2
-------
COMPARISON OF BOILER DESIGNS
The two boilers used for demonstrating the urea NOX reduction process are different in
design and size. Both boilers are located in Southern California. The primary fuel for each
unit is natural gas, but each unit is also equipped to burn low sulfur fuel oil. Cross-sections
of the two boilers are shown in Figure 1, and their design characteristics are compared in
Table 1.
Encina Unit 2 is a 110 MW Babcock and Wilcox Company boiler. The unit is fired from
the front wall with ten burners arranged in two elevations of five burners. The unit operates
with balanced draft maintained by forced draft and induced draft fans. Flue gas
recirculation (FOR) injected between water tubes on the back wall of the lower furnace is
a primary means of steam temperature control. Final superheat steam temperature is
controlled by spray attemperation. The final reheat steam temperature is controlled by
distribution dampers in the backpass. A total of sixteen existing observation ports are
available for urea injection in two elevations of the upper furnace. One elevation is located
adjacent to the furnace exit and entrance to the convective pass, while the second elevation
is about 12 feet below, near the arch of the furnace.
Etiwanda Unit 3 is a 320 MW Combustion Engineering boiler. This is a tangentially-fired
boiler with twin furnaces separated by a division wall. Etiwanda Unit 3 operates with a
pressurized furnace. This unit is unique in its downward flow arrangement with the burner
assemblies located at the top of the boiler. The burner assemblies consist of three tiers of
gas and oil burners located in the corners of each furnace. Tilt of the burner assemblies is
a primary means of reheat steam temperature control. FOR is injected into the windbox for
NOX control and for steam temperature control at low loads. Spray attemperators maintain
final steam temperatures. Twelve existing observation ports arranged in two elevations near
the furnace exit were initially used for urea injection. Additional ports were installed based
on initial test results and modeling efforts to improve NO, control performance over a wider
load range.
Etiwanda Unit 3 differs from Encina Unit 2 in a number of ways which can affect urea NOX
reduction performance. These differences include:
• Boiler dimensions and geometry Etiwanda Unit 3 is physically larger than
Encina Unit 2 with a larger furnace cross-section. In addition, Etiwanda
Unit 3 has a divided furnace which limits access to the furnace cross-section
by urea injectors to two walls rather than three walls as at Encina Unit 2;
• Firing configuration Etiwanda Unit 3 is tangentially down-fired while
Encina Unit 2 is a conventional front wall-fired boiler. The firing
configuration, and the furnace geometry affect the furnace flow field and
thus can be expected to affect mixing of injected urea with the furnace
gases;
6A-3
-------
Thermal environment At full load, gas temperatures in the region of the
furnace exit are significantly higher at Etiwanda Unit 3 (2400°F) than at
Encina Unit 2 (2250°F). Since the urea NOX reduction reactions are
temperature sensitive, differences in injector configurations and resulting
performance can be expected;
Combustion conditions The combustion conditions at Etiwanda Unit 3
result in significantly lower initial NOX levels than found at Encina Unit 2.
At full load on gas fuel, for example, NOX emissions at Etiwanda Unit 3 are
as low as 90 ppm (@ 3% O2) compared to 225 ppm (@ 3% O2) at Encina
Unit 2 with all-burners-in-service (ABIS). This is the result of NOX controls
that have been in place since the 1970's, consisting of FOR and two-stage
combustion achieved with burners-out-of-service (BOOS).
SENSITIVITY OF UREA NOX REDUCTION PERFORMANCE
Results from urea injection trials of both Encina Unit 2 and Etiwanda Unit 3 are indicative
of the key factors affecting NOX reduction potential. While boiler operating conditions
directly affected NOX reduction achieved with urea injection, the injection conditions and
configurations could be adjusted to ultimately minimize stack NOX emissions over a range
of conditions on each unit.
Effect of Operating Conditions
Previous urea injection testing at Encina Unit 2 was conducted with the boiler operating
with ABIS(1). Initially, urea injection was evaluated as a cost-effective NOX control
alternative to the combustion modification techniques typically used in the SDG&E system
to meet current NOX regulations. The combustion modification techniques reduce overall
boiler efficiency compared to the higher, NOx-producing ABIS operating mode. With urea
injection, however, NOX emissions could meet existing NOX regulations while operating with
the more efficient ABIS. Subsequent testing has been completed to evaluate urea injection
in conjunction with alternate, or modified, combustion conditions.
The firing configurations evaluated included ABIS, air biasing, BOOS, and fuel biasing.
These alternatives were evaluated to determine the overall NOX reductions possible by
coupling urea injection with modified combustion conditions. ABIS represents conventional
operation with balanced fuel and air for all the burners, resulting in high baseline NOX
emissions. Air biasing was achieved with ABIS by closing the registers to the lower burner
elevation and thus diverting air to the upper level. This results in staged combustion, with
the lower burners operating fuel-rich and the upper burners operating fuel-lean. The effect
of staged combustion achieved with air biasing not only reduced baseline NOX emissions,
but also affected the heat release distribution through the boiler by delaying the mixing of
fuel and air. BOOS operation was achieved by shutting the fuel off to three of the ten
6A-4
-------
burners. This redistributes the fuel to the remaining burners and results in those burners
operating fuel-rich. BOOS operation thus also results in staged combustion and reduced
NOX emissions. Since the fuel distribution is changed with BOOS, the heat release
distribution also changes.
In gas fuel biasing, some of the fuel is diverted from the upper elevation of burners to the
lower elevations. This increases the heat release into the lower furnace and achieves staged
combustion. Compared to air biasing and BOOS operation, which delay fuel and air mixing
by varying air distribution or by discrete changes in fuel distribution, fuel biasing provides
more uniform changes in fuel distribution such that slightly fuel-rich and slightly fuel-lean
zones are created. The result with fuel biasing is a more confined heat release zone due to
more balanced fuel and air mixing and, more importantly, the diversion of fuel to the lower
burner elevation.
The urea NOX reduction performance varied for the different combustion modes at Encina
Unit 2, as the data in Figure 2 illustrate. Corresponding NOX emissions are shown in Figure
3. The data presented in Figures 2 and 3 represent urea NOX reduction performance
resulting from the injection configuration optimized for ABIS operation. No attempt was
made in these trials to optimize performance for each operating condition. Thus, injection
nozzle characteristics and injection configuration were constant. The highest percentage
reductions were achieved with ABIS operation and the lowest with BOOS operation.
Differences in measured performance can be attributed directly to changes in boiler
conditions. The data set presented in the two figures indicates that differences in NOX
reduction performance can be attributed both to the different initial NOX levels produced by
the different combustion configurations and to the effect on the temperature distribution
through the boiler.
Analogous variations in urea NOX reduction performance with changing operating conditions
were documented at Etiwanda Unit 3(2). Figure 4 illustrates the effect of various combustion
conditions on NOX reduction while Figure 5 presents the corresponding NOX emissions
levels. Included in the data presented from Etiwanda Unit 3 are urea injection test results
with normal, as found fuel oil-fired conditions; normal, as found gas-fired conditions; and
modified gas-fired combustion conditions. The modified combustion conditions at Etiwanda
Unit 3 comprised adjustment of burner tilt to horizontal for all loads with increased FOR
flowrate. As in the case for the Encina Unit 2 data set, the urea injection configuration was
not optimized for each operating condition.
The highest NOX reductions achieved at Etiwanda Unit 3 were with fuel oil. Fuel oil-firing
improves NOX reductions due to producing more favorable temperatures in the boiler (due
to differences in heat transfer characteristics between oil and gas fuels). Furnace exit gas
temperatures are about 200°F lower for oil-firing than comparable gas-fired conditions.
NOX reductions over 30 percent were achieved with gas-firing over the load range of 80 to
320 MW. Changes in combustion conditions, however, resulted in variations in NOX
reduction performance. Even with the variations in urea system performance, the lowest
6A-5
-------
NOX emission levels, down to 21 to 45 ppm (@ 3% O2) depending on load, were achieved
by coupling low NOX, modified combustion conditions with urea injection.
Effect of Urea Injection Parameters
Tests to optimize urea injection performance at each unit involved parametric evaluation of
urea injection process variables. The variables considered included: atomizer design,
solution flow and pressure, location and injector orientation at each injection location.
Conclusions from these parametric tests for both units include the following0'2':
• Atomizer design and the resulting spray characteristics (spray distribution
and angle, droplet size distribution, and injection momentum) affect NOX
reduction performance. The effect of these atomizer specific characteristics
are related to the penetration of urea spray into the furnace flow, the
resulting mixing of urea with the furnace gases, and the rate of evaporation
and the ultimate location of release of urea into the furnace gases;
• The location of injectors and their orientation can improve NOX reduction
performance by taking advantage of furnace flow dynamics to enhance
mixing of urea with the furnace gases and maximize residence time at
optimum reaction temperatures.
Because of the fundamental differences in the thermal and mixing environments presented
by the two different units, the injector design and performance characteristics (i.e., droplet
size distribution, spray angle, injection momentum, etc.) were significantly different. In
relative terms, the requirements for Encina Unit 2 compared to Etiwanda Unit 3 were
injectors which produced small urea solution droplets; lower injection momentum to cover
the furnace gas flow across the entire cross-section of the boiler; and spray angle, shape,
and location of ports to inject across the cross-flowing stream. These requirements are
consistent with the characteristic differences between the two units, including:
• Favorable furnace gas temperatures in the region of injection at Encina Unit
2 for urea NOX reduction reactions to occur, thus requiring the fast
evaporation and release of urea from small solution droplets;
• Small furnace cross-section dimensions in the region of injection requiring
only relatively low injection momentum for adequate penetration and mixing
of urea droplets with the furnace droplets;
• More uniform furnace gas flow with less cross-mixing due to the front wall
firing configuration compared to the swirling flow field resulting from
tangential firing, requiring use of ports physically spaced across the boiler.
6A-6
-------
The requirements for Etiwanda Unit 3, on the other hand, were satisfied by urea solution
injection characteristics which included large droplets that would delay the evaporation and
release of urea from the high temperatures at the point of injection for reaction in lower
temperature regions. In addition, the injectors and locations were developed to optimize the
distribution and mixing of the urea solution by taking advantage of the furnace flow
dynamics of the tangentially, down-fired configuration. In fact, in a brief series of trials
to establish a direct comparison for urea injection between Encina Unit 2 and Etiwanda Unit
3, the injectors achieving optimum performance at Encina Unit 2 were found to achieve
essentially no NOX reduction at Etiwanda Unit 3 at full load conditions.
OPTIMIZATION FOR VARYING CONDITIONS
The data from these two utility boilers demonstrate that unit design and operating conditions
can affect urea NOX reduction performance. Since urea systems are designed by necessity
for optimum performance at selected, typical operating conditions, NOX reduction
performance will vary. However, the design of urea injection and control systems can
incorporate adjustable parameters to accommodate intermediate or varying conditions. This
potential to control over varying conditions has been demonstrated at both Encina Unit 2 and
Etiwanda Unit 3.
Multiple Level Injection
At Encina Unit 2, for example, simultaneous injection from multiple levels improved NOX
removal at both high and low loads<2). In a multiple injection configuration, a reduced
dosage of urea (lower NSR) is injected at each elevation. This improves urea utilization
and, in turn, the overall NOX removal. This improved utilization also reduces byproduct
NH3 emissions. Figure 6 compares NOX reduction performance at Encina Unit 2 achieved
with bi-level injection for natural gas and fuel oil-firing. The method of bi-level injection
reduced the sensitivity of NOX removal to load. In addition, similar performance was
achieved for the two different fuels even though the resulting furnace temperature profiles
are distinctly different.
Injection Location and Orientation
Another technique used at both units to adjust for varying operating conditions was adjusting
injection location by varying injector orientation. In practical applications of the urea
injection process, boiler penetrations to accommodate urea injectors will be selected to
provide access into favorable temperature regions for a limited number of conditions or
loads. To maintain urea NOX reduction performance for intermediate loads or changes in
operating conditions, the orientation of the injectors can be used to adjust the relative
location of urea injection. Recent tests were conducted at Encina Unit 2 to evaluate the
optimization of urea injection with the combustion modification technique of fuel biasing
6A-7
-------
to achieve minimum stack NOX emissions. The test results illustrate how varying orientation
from available injection locations can improve performance and how orientation can be used
to maintain NOX reduction performance as operating conditions vary.
Urea NOX reduction performance was evaluated with and without fuel biasing by screening
injection location and orientation. Tests were completed for loads of 80 MW and 50 MW.
At 80 MW with ABIS operation, the highest NOX reduction achieved was 44.3 percent using
the lower level injectors only pointed up and urea injected at a rate of NSR = 2.0. This
reduction resulted in NOX emissions of 50 ppm (@ 3% O2) from a baseline of 91 ppm. With
fuel biasing at the same load, however, the highest NOX reduction achieved was 29.1 percent
using simultaneous bi-level injection with both the upper and lower elevations of nozzles
pointed up and urea injected at NSR = 2.0. The optimum urea injection configurations thus
shifted for the two different firing modes.
The reasons for this shift appear to be a shift in furnace temperature. Furnace exit
temperatures increased about 40°F with fuel biasing. As a result, NOX reduction was
improved by injection at a higher, and therefore cooler, elevation for fuel biasing conditions
than for normal ABIS operation. Although relative urea NOX reduction performance was
decreased with fuel biasing compared to ABIS, stack NOX emissions were reduced from 50
ppm (@ 3% O2) for ABIS and urea down to 38 ppm (@ 3% O2) for fuel biasing and urea.
At 50 MW the data indicate that, for ABIS operation, injecting urea through the lower
elevation with nozzles pointed upward achieved the highest NOX reduction. For fuel bias
operation, however, the best configuration was bi-level injection with the upper elevation
injectors pointed down and the lower elevation injectors pointed up. As for the 80 MW
case, the shift in optimum injection configuration for the two operating conditions suggest
contributing affect of a change in furnace gas temperature. The data also indicate that
significant reductions can be achieved for low initial NOX levels, resulting in stack emissions
down to 23 ppm for an NSR = 1.7.
Dilution Water Flow and Injection Momentum
At Etiwanda Unit 3, three elevations of injection ports were determined to provide coverage
over the unit's normal load range, 80 to 320 MW, as shown in Figure 7. However,
Etiwanda Unit 3 is also routinely operated down to 20 MW. Test results demonstrated that
dilution water flow could be used in conjunction with injector elevation and orientation to
adjust the ultimate fate of urea droplets and achieve NOX reductions at loads less than 160
MW. By varying dilution water flow, the solution concentration, injection momentum, and
resulting droplet size distribution is changed. The parameters directly affect the point at
which the urea is released from solution to react with the furnace gases.
The performance of the urea NOX reduction system at Etiwanda Unit 3 is illustrated in
Figure 8. The optimized system is used together with combustion modifications to achieve
NOX levels of 20 to 45 ppm over the entire load range of 20 to 320 MW. This represents
6A-8
-------
significant reductions in NOX compared to normal, as found conditions also shown for
reference. In addition to the NOX reductions achieved, the available data indicate that
byproduct NH3 emissions below 20 ppm could be maintained up to urea flowrates
corresponding to NSR = 2.0. Figure 9 illustrates typical NH3 emissions measured at
Etiwanda Unit 3.
CONCLUSIONS
The effectiveness of the urea NOX reduction process is sensitive to temperature and mixing
phenomena as well as chemical stoichiometry (NSR). Since the urea NOX reduction process
occurs within the boiler furnace, the ultimate performance of the urea process is thus
dependent on boiler design and operating characteristics. Although the design of SNCR
systems must attempt to address these factors, realistic limitations must be imposed on the
range of expected boiler operating conditions (fuel type, load, burner firing pattern, excess
air, FOR flowrate, etc.) over which the system performance can be optimized.
To accommodate differences in boiler design and variations in operating conditions, urea
injection process parameters can be adjusted and optimized. Improvements in urea NOX
reduction performance and, ultimately stack NOX emissions, can be achieved by modifying
combustion conditions, optimizing injection location and orientation, and adjusting injection
nozzle droplet size and injection momentum. NOX reductions of about 30 percent at NSR
= 2.0 could be achieved over the load range of 20 to 320 MW at Etiwanda Unit 3, resulting
in stack NO, emissions in the range of 20 to 45 ppm (@ 3% O2) when combined with
combustion modifications. At Encina Unit 2, similar reductions and stack NOX levels (23
to 38 ppm @ 3% O^ could be achieved when urea injection was coupled with the
combustion modification technique of fuel biasing. In general, the data trends suggest that
for these gas- and oil-fired boilers, more confined heat release zones provide a more
favorable furnace environment than deeply staged, delayed mixing conditions.
REFERENCES
1. J.H. Nylander, M.N. Mansour, and D.R. Brown, "Demonstration of an Automated
Urea Injection System at Encina Unit 2," in proceedings of the Joint Symposium on
Stationary Combustion NO, Control, EPRI Report GS-6423, July 1989.
2. A.R. Abele, D.R. Brown, Y. Kwan, M.N. Mansour, and J.H. Nylander,
"Demonstration of Urea Injection for NOX Control on Utility Boilers," in proceedings:
GEN-UPGRADE 90, EPRI Report GS-6986, September 1990.
6A-9
-------
en
>
Encina Unit 2
Etiwanda Unit 3
Figure 1. Demonstration Sites
-------
70
60
50
40
30
20
10
NOx Removal (%)
o
NSR • 2.0
ABIS
Air Bias
BOOS
0 20 40 60 80 100 120
Load (MW)
Figure 2. Effect of Combustion Conditions on Urea
NOx Removal at Encina Unit 2, Gas Fuel.
6 A-11
-------
90
80
70
60
50
40
30
20
10
NOx (ppm @ 3% O0)
0
NSR = 2.0
o
ABIS
20
Air Bias
BOOS
40 60 80
Load (MW)
100
120
Figure 3. Effect of Combustion Conditions on
Stack NOx Emission Levels with Urea
at Encina Unit 2, Gas Fuel.
6A-12
-------
60
NOx Removal (%)
50
40
30
20
10
NSR = 2.0
Oil-As Found
Gas-Comb. Mod.
Gas-As Found
0 50 100 150 200 250 300 350
Load (MW)
Figure 4. Effect of Operating Conditions on Urea NOx
Reduction Performance at Etiwanda Unit 3.
6 A-13
-------
90
80
70
60
50
40
30
20
10
0
NOx (ppm @ 3% O? )
NSR = 2.0
Oil - As Found
Gas - Comb. Mod.
Gas - As Found
0 50 100 150 200 250 300 350
Load (MW)
Figure 5. Effect of Operating Conditions on Stack NOx
Emission Levels with Urea at Etiwanda Unit 3.
6A-14
-------
en
80
70
60
50
40
30
20
10
NOx Removal, Percent
50
-©- Gas Firing ~V-Oil Firing
60
70 80
Load, MW
90
-o
100
NOx Removal, Percent
40
30
20
10
50
-©-Gas Firing -V- Oil Firing
60
70 80
Load, MW
90
100
Figure 6. Comparison of NOx Reduction with Bi-Level Injection
for Natural Gas and Fuel Oil Firing at Encina Unit 2.
-------
O)
CO
Urea Injection
Ports
O Unused Ports
Loop 3
El. 84'
El. 641
El. 61'
El. 54'
O
O O
0*0
Loop 2
Division Wall
O
o o
0*0
Loop 2
Side View
Front View
Figure 7. Etiwanda Unit 3- Urea Injection Port Locations
-------
>
-vl
NOx (ppm @ 3% O2)
110
100
90
80
70
60
50
40
30
20
10
0
0
As Found N Ox
x Combustion Modification N Ox
Urea * Combustion Modification NOx
50
100
150 200
Load (MW)
250
300
350
Figure 8. Overall NOx Reduction Performance at Etiwanda Unit 3,
Gas Fuel.
-------
en
00
75
60
45
30
15
NH3, ppm
o
0
O 320 MW
80 MW
O
o
NSR
Figure 9. Typical NH 3 Emission from Optimized
Urea System at Etiwanda Unit 3, Gas Fuel.
-------
TABLE 1. BOILER DESIGN CHARACTERISTICS
Design
Parameter
Capacity (MW)
Firing
Configuration
Burners
Dimensions
Height (ft)
Depth (ft)
Width (ft)
Steam Flow (Ib/hr)
SH Temperature (°F)
RH Temperature (°F)
Steam Press, (psig)
Encina
Unit 2
110
Front
Wall
2 Rows x
5 Burner
Peabody
77.0
20.0
34.0
700,000
1000
1000
1450
Etiwanda
Unit 3
320
Tangential
Down-Fired;
Divided Furnace
3 Elev/
Corner
x 8 Corner
CE
88.1
22.1
60.0 (30/30)
2,305,000
1050
1000
2450
6A-19
-------
WIDENING THE UREA TEMPERATURE WINDOW
D. P. Teixeira
Research & Development Department
Pacific Gas and Electric Company
San Ramon, CA 94583
L J. Muzio
T. A. Montgomery
G. C. Quartucy
T. D. Martz
Fossil Energy Research Corporation
Laguna Hills, CA 92653
-------
WIDENING THE UREA TEMPERATURE WINDOW
ABSTRACT
The results of laboratory tests to widen the effective temperature range while, at the same time,
minimizing byproduct emissions for the urea injection SNCR process are described. Data are
presented showing the effect of a number of additives (methane, combination of hydrocarbons, carbon
monoxide, ethylene glycol, HMTA, and furfural) and initial NOX level (125 and 250 ppm) on NOX
removal efficiency and byproduct emissions (NH3, CO, N2O) as a function of temperature. Several new
phenomenon not previously observed are described. Of particular interest is the strong effect of CO
on N2O emissions during urea injection. In addition, many additives were found to improve NO
reduction but not NOX reduction. In these cases, the presence of additives converted the NO initially
present to NO2 and/or N2O.
6A-23
-------
WIDENING THE UREA TEMPERATURE WINDOW
INTRODUCTION
A variety of technologies is available to control NOX emissions from fossil power plants. One attractive
option is selective non-catalytic reduction (SNCR) with urea (1_). However, the SNCR process, which
has many attractive features, does have several disadvantages. One drawback is the relatively narrow
temperature "window" over which the process is effective. Another potential disadvantage is the
emission, at least under some operating conditions, of undesirable byproducts such as NH3 or CO.
These issues become even more important for units which are cycled frequently or use multiple
fuels-which is the case for many fossil plants.
Results of a series of laboratory tests to address the issues noted above through the use of additives
to the basic urea injection process are described in the sections which follow. The effects of additive
type, additive concentration and initial NOX level on NOX removal and byproduct emissions as a function
of temperature are presented.
PROCESS DESCRIPTION
Conceptually, the SNCR process with urea is quite simple. An aqueous solution of urea is injected
into, and mixed with, the flue gas at the correct temperature. After the mixing has been completed,
the urea then reacts selectively to remove the NOX.
In practical applications, however, the process (and the equipment required) can be much more
complicated. Non-uniformities in velocity, temperature, and NOX concentration at the point of injection,
along with the variation in the physical location of the effective process temperature range within the
boiler, depend on various operating factors including load, type of fuel fired, and length of time on a
particular fuel. These factors often lead to multiple levels of injection and/or use of additives to
accommodate the shifts in temperature.
6A-24
-------
PILOT SCALE TEST FACILITY
A schematic of the pilot-scale facility used for these tests is shown in Figure 1. The pilot scale
combustor fires natural gas, doped with NH3 to control the initial NOX level. The combustor and test
section are refractory lined with the test section being 15 cm in diameter and 240 cm long. At the firing
rates used for these tests, the residence time in the test section is nominally 0.5 seconds, while the
temperature drop along the test section is nominally 250°C/sec (450°F/sec). The SNCR solutions were
injected into the combustion products at the combustor throat through a small air assist atomizer,
above the test section. The atomizer was fabricated into a water cooled holder. The atomizer was
located at the center of the throat with the spray directed downward (i.e., co-flowing with the
combustion products). The solutions were pumped with variable speed peristaltic pumps and metered
with rotameters. In order to maintain a constant thermal environment in the test section, the total
amount of liquid injection was held constant at nominally 1 liter/hr. By diluting a concentrated urea (or
other SNCR chemical) solution with distilled water, the amount of chemical reagent was varied while
a total liquid flow rate of 1 liter/hr was maintained.
Gas samples were taken at the exit of the combustor with a water-cooled probe and transported to a
series of gas analyzers (NO/NOX, N2O, CO, CO2, and O2). The continuous measurement of N2O was
made using an NDIR based technique (2). NH3 was measured using a selective ion electrode
technique.
The pilot-scale tests investigated the effect of temperature, additives, chemical injection rate, and initial
NOX concentration on NOX removal efficiency and byproduct emissions (specifically NH3, CO, and
N20).
RESULTS
During this study, experiments were carried out at initial NOX levels of 125 ppm and 250 ppm and
ISI/NO, molar ratios of 1 and 2. For brevity, most of the results shown in this paper will be from the
tests at an initial NOX level of 125 ppm. Results at 250 ppm will be shown for situations where the
effect of the SNCR chemical, or additive, exhibits different behavior from that observed at the 125 ppm
level.
Baseline Performance of Urea - No Additive
To establish a reference for comparison of results from the various additives, a series of baseline tests
were performed using urea alone. The baseline NOX removal and byproduct emission results over
6A-25
-------
BURNER FLOW SYSTEM
7
It
34
lorn COMBUSTION
ANDCOOL1NOSECIION
EIOMT CONCENTRIC
COOLINQ PROBE |_
PORTS ^
r~
0X3 AND SOLID
INJECTION PORT L_
' .- h.
\ r
cm ADDITIVE IHJECTION
SECTION
"•-IT
r i
L I
THERMOCOUPLE (_|
PORT ^
!~
I
LJ
Dem TEST SECTION
LJ
LI
LI
r
i
LI
n
BURNER
I 1
J
I1-
|
Horn
-------
the temperature range investigated for initial NOX levels of 125 and 250 ppm and a urea injection rate
corresponding to molar ratios of nitrogen to NOX (N/NOX) of 1 and 2 are shown in Figures 2 and 3.
Figure 2 shows the results for an initial NOX level of 125 ppm. Figure 3 shows the same data but for
an initial NOX level of 250 ppm. The narrow effective process temperature range for NOX removal can
be clearly seen in both figures, as can the increasing levels of NH3 and CO byproducts as temperature
is decreased. Also shown are byproduct levels of N2O produced by the process at the test conditions.
Other investigators have also noted N2O byproducts associated with urea injection (3).
Carbon Monoxide Additive
A review of the general combustion chemistry literature showed that CO was a potential compound that
could alter the temperature dependence of the urea injection process. This behavior was also
suggested by the data of reference 4 showing the effect of CO at high concentrations (8000 ppm CO)
on NOX removal. While the use of CO to modify the urea temperature window in power plant boilers
presents several difficult practical application issues, it was felt important to address the effect of CO
since all combustion devices emit some level of CO.
For the data discussed below, the CO additive was introduced by injecting it with the atomizing air.
NO. Removal Temperature Dependance. Figure 4 shows the effect of CO on NOX removal as a
function of temperature at an initial NOX level of 125 ppm and N/NOX ratio of 2. This figure shows
several interesting features:
• CO, even in relatively low amounts, has a significant impact on the NOX removal efficiency at a
given temperature. As CO levels are increased, the NOX removal versus temperature
dependence shifts to a lower temperature regime. Figure 4 shows that, increasing the CO levels
from O ppm to 1000 ppm shifts the peak NOX removal temperature about 200°F lower.
• As CO levels increase, the effective process temperature range is broadened. For the conditions
of Figure 4 when CO is in the 500-1000 ppm range, the window appears to be broadened by
about 100°F.
• Increasing CO also lowers the peak level of NOX removal possible. Figure 4 shows that peak NOX
removal decreases from about 55% to 45-50% as CO increases from 0 ppm to 500 ppm; it further
decreases to about 45% as CO is increased to 1000 ppm. Similar behavior is noted for the other
conditions investigated.
CO Byproduct Emissions. The final CO levels resulting from addition of CO to the urea process are
shown in Figure 5. As can be seen, at the lowest temperature evaluated, 1470°F, CO emissions
increase as the initial amount of CO addition is increased. However, for temperatures at or above
1600-1650°F, final CO levels are practically independent of the amount of CO added.
6A-27
-------
E*.
Q- c
LU
O
80
70
60
50
40
30
20
10
0
-10
-20
NH3
ANOx
(a) N/NOX =
1400 1500 1600 1700 1800 1900 2000 2100 2200 2300
Temperature, °F
(b) N/NO, = 2
1400 1500 1600 1700 1800 1900 2000 2100 2200 2300
Temperature, °F
Figure 2. NOX Reduction and Byproduct Emissions with Urea Injection
(Initial NOX = 125 ppm)
6A-28
-------
E °:
o. c
o. g
w "-
'E x
UJO
100
90
80
70
60
50
40
30
20
10 -
0.5XNII3
ANOx
(a) N/NO, = 1
0
1400 1500 1600 1700 1800 1900 2000 2100 2200 2300
Temperature, °F
.
Q- c
CL g
100
90
80
70
60
50
40
30
20
10
0
1400 1500 1600 1700 1800 1900 2000 2100 2200 2300
Temperature, °F
(b) N/NO. = 2
Figure 3. NO, Reduction and Byproduct Emissions with Urea Injection
(Initial NOX = 250 ppm)
6A-29
-------
"O
-------
NH, Byproduct Emissions. Since minimum unreacted NH3 from the SNCR process is desirable both
from an environmental, as well as boiler impact standpoint, measurements of the byproduct NH3 were
made. Figure 6 shows the results of these measurements for an initial NOX level of 125 ppm and
N/NOX ratio of 2. As expected, NH3 emissions decrease as temperature increases. However, NH3
levels at any given temperature, were found to decrease significantly as CO levels increased.
N;O Byproduct Emissions. The most interesting influence of CO on the urea injection process was on
the N2O byproduct characteristics (Figure 7). The effect of CO on N2O is strongly temperature
dependent. At higher temperatures (approximately 1900°F and above), N2O levels tend to merge to
a similar low level for all combinations of CO, initial NOX and N/NOX. At these high temperatures, N2O
tends to decrease rapidly to very low levels as temperature is increased.
However, at the lower temperatures investigated (1500-1600°F), a very different behavior can be seen;
N2O levels increase with increasing CO levels. For example, at an initial NOX of 125 ppm and N/NOX
= 2, N2O increases from about 10 ppm to 35 ppm as CO is increased from 0 ppm to 1000 ppm.
Although not shown, N2O emissions at these lower temperatures also increase as the amount of urea
(i.e. N/NOJ and initial level of NOX increase. At the highest initial NOX (250 ppm),N/NOx (2), and CO
(2000 ppm) levels investigated, N2O concentrations approach 100 ppm.
At the intermediate temperatures (between 1500°F and 1900°F), there is a transition from the low
temperature behavior to the high temperature behavior. At the lower CO levels, increasing
temperatures first produce an increase in N2O then a decrease as temperature is increased, with an
obvious maximum in the N2O as a function of temperature. At higher CO levels, N2O initially remains
relatively constant as temperature increases, then drops off abruptly.
Implications. There are several important practical implications regarding the influence of CO on the
urea injection process, in particular the N20 characteristics. First, to minimize N2O production in the
urea injection process it is important to maintain low CO levels.
Second, when using urea injection, a "coupling" between the combustion process and the urea
injection process can occur, i.e. CO produced in the burner region influences the SNCR performance.
This may be especially true for low NOX burner systems where, as is well known, there are frequently
trade-offs between the NOX reduction and CO levels.
Lastly, the effect of CO on N2O formation may explain some of the differences in N2O levels reported
by various researchers at a recent workshop on N2O (5).
6A-31
-------
E
Q.
Q.
CO
250
200
150
100
50
0
CO Addition
A 0 ppm
A 65 ppm
* 125 ppm
1400 1500 1600 1700 1800 1900 2000 2100 2200 2300
Temperature, °F
e
a
0
CJ
Figure 6. Effect of CO Additive with Urea on Byproduct NH3 Emissions
(Initial NOX = 125 ppm; N/NOX = 2)
so 1—. r
CO addition
O 0 ppm
• 500 ppm
n 1 ooo ppm
1400 1500 1600 1700 1800 1900 2000 2100 2200 2300
Temperature, °F
Figure 7. Effect of CO Additive with Urea on Byproduct N2O Emissions
(Initial NOX = 125 ppm; N/NOX = 2)
6A-32
-------
Methane - Additive
Methane (CH4) was also investigated as a potential additive to alter the urea/NOx removal temperature
dependance. The results shown are for tests conducted at 1600°F, N/NOX = 2, initial NOX levels of 125
ppm and 250 ppm, and CH4/NOX molar ratios of 0, 0.5 and 1. Figures 8 and 9 show the results.
For the initial NOX levels investigated, both NO and NOX (NO+NO2) levels decrease with the addition
of urea alone. However, when methane is added, while the NO levels continue to decrease for both
initial NOX levels, the effect on NOX differs. At an initial NOX level of 250 ppm, NOX levels continue to
decrease with the addition of CH4. However, at the lower initial NOX level of 125 ppm, while NO levels
decrease with CH4 addition, NOX levels remain constant. At this lower initial NOX level, the effect of
the CH4 is to oxidize NO to NO2, rather than to enhance the SNCR process.
The effect of methane additive with urea on N2O emissions is also included in Figures 8 and 9. At both
initial NOX levels, methane promotes the formation of N2O as a byproduct; the N2O levels increase with
increasing amounts of CH4.
Efforts to explain the significantly different behavior between the two initial NOX cases have to date
been unsuccessful. The possibility of hydrocarbon interference with the N2O measurements, which is
known to occur for the instrument used, was considered but could not explain the results observed.
Multiple Additives
NOV Removal Efficiency. A Japanese patent (6) identifies multiple hydrocarbon additives used with
urea to broaden the temperature window. A specific example was presented for the following
conditions: urea at N/NOX = 4; initial NOX = 990 ppm; temperature = HOOT; and additives consisting
of ethylene glycol, propane and carbon. Without the additives (i.e. urea only) the NOX reduction, as
expected, was low, under 10%. With the additives, the NOX reduction was increased to almost 75%.
A series of tests were performed to verify the performance of the multiple additives under the following
conditions: initial NOX = 790 ppm; temperature of 1400°F; N/NOX = 4; ethylene glycol/urea concentration
of 9.5%; propane to urea of 57%; and carbon/urea of 33%. All concentration ratios are on a molar
basis. The tests were conducted sequentially to evaluate the individual, as well as combined, effect.
The results are shown in Figure 10.
As can be seen in Figure 10, the addition of glycol resulted in an increase of NOX removal from about
10% with urea only to about 20%. Addition of propane increased the NOX removal to almost 55%.
6A-33
-------
150
Q-
Q.
O
C\J
CM
O
O
Initial
NO + NO2 + N20
NO + NO2
Urea
Urea +
0.5 ChM/Uiea
Urea +
1.0 CH4/Urea
Figure 8. Effect of Methane Additive with Urea Injection on NO, NO2, and N2O
(Temperature = 1600°F; Initial NOX = 125 ppm; N/NOX = 2)
300
Initial
NO + NO2 + N2O
NO + NO2
ND
Urea
Urea+ Urea +
0.5 CH4/Urea 1.0 CH4/Urea
Figure 9. Effect of Methane Additive with Urea Injection on NO, NO2 and N2O
(Temperature = 1600°F; Initial NOX = 250 ppm; N/NOX = 2)
6A-34
-------
I
I
DC
o
100
90
80
70
60
50
40
30
20
10
0
Urea
+ Glycol
+ Propane
+ Carbon
Urea
+Glycol
+ Propane
Urea
+Glycol
+Propane
+ Carbon -
D %ANO
M %ANOx
(Reference 6)
Present Tests
Figure 10. Effect of Multiple Additives on NOX Reduciton with Urea
o
TJ
03
cc.
x
O
25
20
15
10
Urea
+ Glycol
+ Propane
Urea
Urea
+ Glycol
Urea
f Glycol
+ Methane
%ANOx
Figure 11. Effect of Multiple Additives on NOX Reduction with Urea
(Temperature = 1400°F; Initial NOX = 760 ppm; N/NOX = 1)
6A-35
-------
Further addition of the carbon actually resulted in a small deterioration in NOX removal. While the 55%
removal did not quite match the 75% value cited in the patent, the results were sufficiently encouraging
that additional tests were conducted.
The next series of tests were done under nominally the same conditions as above (initial NOX of 755
ppm; temperature of 1400°F; ethylene glycol/urea of 9.9%; propane/urea of 60%), but at a lower N/NOX
ratio of 1.0. Results of these tests are shown in Figure 11. No improvements in NOX removal were
noted for the case of glycol-only addition. NOX removal increased to about 15% for the glycol plus
propane case. No change was seen in NOX removal when methane was substituted for propane. It
should also be noted that, in the urea plus glycol plus propane, or methane, cases, the NOX removal
was significantly lower than the NO removal.
The final series of tests considered the multiple additive concept at conditions of greatest practical
interest: temperature of 1600°F; initial NOX levels of 125 ppm and 250 ppm; N/NOX of 2; CH4/NOX
values of 0, 0.5, 1; ethylene glycol/urea of 10%. The results of these tests are shown in Figures 12
and 13. At 125 ppm initial NOX (Figure 12), little benefit of the multiple additives was observed. Some
improvement in the NO removal was noted for glycol addition alone. Very little NOX removal
improvement was noted. As discussed previously, no effect of methane on either NOX or NO removal
was observed other than to increase N20 emissions. A case where methanol was substituted for
glycol at a methanol/urea of 10% was investigated and yielded virtually identical results.
Contrary to the general lack of improvement in NOX performance at 125 ppm, meaningful improvement
in NOX (and NO) removal was observed when glycol was added, and/or when CH4 was added at a
higher initial NOX level of 250 ppm (Figure 13). As will be seen in the next section, NO2 and N2O
formation from the initial NO explains at least a portion of the difference between the NOX and NO
removal levels.
NO0/N0O Characteristics. Data summarizing the NO2 and N2O characteristics of the multiple additive
concept are summarized in Table 1.
6A-36
-------
E
o
n_
6
(M
C\J
O
z:
o"
150
125 -
100 -
Additive: Elhylene Glycol
NO + NO2 + N2O
NO i NO2
Utea N/NO (molar)
E. Glycol/Urea (molar)
CH4/Urea (molar)
Initial
0.0
0.0
0.0
Urea
2.0
0.0
0.0
Urea/Add. Urea/Add. Urea/Add.
2.0
0.1
0.0
2.0
0.1
0.5
2.0
0.1
1.0
Figure 12. Effect of Multiple Additives (Ethylene Glycol and Methane) on
NO, Reduction with Urea
(Temperature = 1600°F; Initial NOX = 125 ppm; N/NOX = 2)
300
NO + NO2 + N2O
Additive: Elhylene Glycol |
Urea N/NO (molar)
E.Glycol/Urea (molar)
CH4/Urea (molar)
Initial
0.0
0.0
0.0
Uiea
2.0
0.0
0.0
Urea/Add. Urea/Add. Urea/Add.
2.0
0.1
0.0
2.0
0.1
0.5
2.0
0.1
1.0
Figure 13. Effect of Multiple Additives (Ethylene Glycol and Methane) on
NOX Reduction with Urea
(Temperature = 1600°F; Initial NOX = 250 ppm; N/NOX = 2)
6A-37
-------
Table 1
NO2 AND N2O MULTIPLE ADDITIVES
T = 1600°F N/NO = 2 Ethylene Glycol/Urea = 10%
Initial NO.
ppm
125
125
125
250
250
250
CHAlrea
molar
0
0.5
1
0
0.5
1
Final NO,
egm
19
24
25
33
37
31
N^O
ppm
13
21
28
26
42
63
N,O + NO,
ppm
32
45
53
59
79
94
Although not shown, virtually identical data were collected for methanol under the same test conditions.
As can be seen, a portion of the original NO appears in the products as NO2 and N2O. NO2 levels
were roughly in proportion to the initial NOX levels and tended to increase as the CH4/urea increased.
Likewise, N20 increased approximately in proportion to the initial NOX and as CH4/urea was increased.
HMTA/Furfural Additives
A review of the patent literature also indicated that the addition of hexamethylenetetramine, C6H12N4
(HMTA), and furfural (C5H4O2) to urea results in a broadening of the effective process temperature
range for NOX reduction (7,8,9).
A series of tests were conducted to evaluate the effectiveness of these compounds. The tests
evaluated HMTA addition alone and in combination with furfural. A temperature of 1650°F was used
for these tests. The quantity of additives used in the tests was estimated based on the data contained
in References 7-9. Test conditions were as follows: initial NOX level of 250 ppm; HMTA/urea of 0.2;
furfural/urea of 3.65 (all on a molar basis). Results of these tests are shown in Figure 14.
Examination of Figure 14 shows that the addition of HMTA alone, or the HMTA/furfural mixture, led to
a meaningful improvement in both NO and NOX removal. However, the improvement in NOX is
considerably lower than the improvement for NO removal. Evaluation of the final NO2 levels (Table
6A-38
-------
cc
X
O
50
40
30
20
10
Urea
Alone
HMTA HMTA/Furfural
Addition Addition
Urea
Alone
D %ANO
M %ANOX
N/NO = 1.0 N/NO = 1.0 N/NO = 1.0 N/NO = 2.0 Urea Alone
N/NO = 1.0 N/NO = 1.8 N/NO =1.8 N/NO = 2.0 Urea and Additive
Figure 14. Effect of HMTA and Furfural on NO and NOX Reduction with Urea
(Temperature = 1650°F; Initial NOX = 250 ppm)
6A-39
-------
2) showed that a portion of the initial NO was being oxidized to NO2. Unfortunately, data for N2O was
not collected during this test series, so a more complete assessment of the impact of HMTA/furfural
on byproducts could not be done.
Table 2
NO2 LEVELS WITH HMT A/FURFURAL ADDITIVE
HMTA/Urea
(Molar)
0.2
0.2
Temperature =
Furfural/HMTA
(Molar)
0
3.65
1650°F Urea/NOx = 1
Initial NO,
(ppm)
15
15
Final NO,
(ppm)
60
58
Since the nitrogen in the HMTA increases the effective N/NOX ratio from 1 to 1.8, Figure 14 also shows
the NOX removal expected for the urea only case at N/NOX = 2. This allows an alternative comparison
of the behavior of HMTA since one alternative to the use of HMTA additives would be to increase the
N/NOX by increasing the amount of urea injected in place of adding the HMTA. As can be seen,
increasing the amount of urea injected provided a comparable degree of NOX removal when compared
to HMTA, or HMTA/furfural addition.
Future Research
Continuation of efforts to find additives or alternative reducing agents to improve the SNCR process
will be pursued in the future. In addition, a series of tests to evaluate the effect of CO additive with
NH3 as a reducing agent will be conducted and compared to the urea plus CO additive results.
CONCLUSIONS
A number of unexpected results were observed when testing various additives to the urea injection
process:
CO shifts and broadens the temperature window even at low CO levels; in addition,
significant changes in the byproduct emissions, especially for N2O, occur.
6A-40
-------
CH4 exhibits markedly different NOX and NO removal behavior depending on the initial
NOX level. Reasons for this behavior are not understood. CH4 addition also leads to the
conversion of NO to NO2 (oxidation) and the formation of N2O.
As with CH4, the use of multiple hydrocarbon additives leads to different NOX and NO
removal behavior, depending on the initial NOX level. The use of multiple additives also
leads to the conversion of a portion of the initial NO to NO2 and N2O.
The HMTA and furfural additives lead to the conversion of NO to N02. As a result, NO
removal improves to a greater extent than the NO,, removal. Further, it appears that the
improvement in NOX reduction can be attributed to the increased N/NOX injection ratio
that results from the addition of HMTA.
In addition to the specific conclusions reached above for the individual additives, overall examination
of the results indicates a more general conclusion: The chemistry involved in urea NOX removal is
more complex than previously thought. As a result, when considering employment of the process to
a specific application, careful consideration of the initial NOX level and the levels of trace combustion
product species, including hydrocarbons and CO, is required.
6A-41
-------
REFERENCES
1. Arand, J. K., Muzio, L. J., Setter, J. G., U.S. Patent 4.208.386. June 17, 1980.
2. Montgomery, T. A., et al, "Continuous Infrared Analysis of N2O in Combustion Products",
JAPCA Vol. 39, No. 5, May 1989.
3. Jodal, et al, "Pilot Scale Experiments with Ammonia and Urea as Reductants in Selective
Non-Catalytic Reduction of Nitric Oxide", 23rd International Symposium on Combustion,
Orleans, France, July 1990.
4. Siebers, D. L. and Caton, J. A., "Removal of Nitric Oxide from Exhaust Gas with Cyanuric
Acid", Paper No. WSS/CI88-66, 1988 Fall Meeting of the Western States Section of the
Combustion Institute, Dana Point, California, October 1988.
5. Second European Workshop on N2O Emissions, Lisbon, Portugal, June 1990.
6. Kuze, T., et al, Japanese Patent 53128023, November 8, 1978.
7. Bowers, E. B., U.S. Patent 4,751.065, June 14, 1988.
8. Epperly, R. E. and Sullivan, J. C., U.S. Patent 4,770.863. September 1988.
9. Epperly, W. R., O'Leary, J. H., Sullivan, J. C., U.S. Patent 4.780,289. October 25, 1988.
6A-42
-------
CATALYTIC FABRIC FILTRATION FOR
SIMULTANEOUS NOx AND PARTICULATE CONTROL
Greg F. Weber and Dennis L. Laudal
Energy and Environmental Research Center
University of North Dakota
Box 8213, University Station
Grand Forks, ND 58202
Patrick F. Aubourg and Marie Kalinowski
Owens-Corning Fiberglass
P.O. Box 415
Granville, OH 43023-0415
Prepared for
Electric Power Research Institute
3412 Hillview Avenue
Palo Alto, CA 94303
-------
CATALYTIC FABRIC FILTRATION FOR
SIMULTANEOUS NO. AND PARTICULATE CONTROL
Greg F. Weber and Dennis L. Laudal
Energy and Environmental Research Center
University of North Dakota
Box 8213, University Station
Grand Forks, ND 58202
Patrick F. Aubourg and Marie Kalinowski
Owens-Corning Fiberglas
P.O. Box 415
Granville, OH 43023-0415
Prepared for
Electric Power Research Institute
3412 Hillview Avenue
Palo Alto, CA 94303
ABSTRACT
The Energy and Environmental Research Center (EERC) at the University of North
Dakota (UNO) has been working with Owens-Corning Fiberglas Corporation (OCF) for
several years evaluating Catalytic Fabric Filtration for simultaneous NOX and
particulate control. Early work sponsored by OCF was presented at the 1989
EPRI/EPA NOX Symposium. Since April 1988, the U.S. DOE Pittsburgh Energy
Technology Center (PETC) has funded development activities at the EERC, with OCF
providing catalyst-coated fabric samples for testing.
The work has involved evaluating samples (1 ft2) of catalyst-coated fabric prepared
by OCF using actual flue gas from the combustion of pulverized coal. Dependent
variables included air-to-cloth ratio, ammonia/NO,, molar ratio, and coal type
(bituminous, subbituminous, and lignite). Flue gas temperature was maintained at
650°±25°F. Resulting NOX removal efficiency and ammonia slip varied significantly
with air-to-cloth ratio. As the air-to-cloth-ratio increased from 2 to 6 ft/min,
NOX reduction decreased from 85-95% to less than 70% with corresponding ammonia
slip values ranging from 5 ppm to 360 ppm. For the short-term (8-hour) tests
completed, the four coals tested did not appear to have a significant effect on
catalyst-coated fabric performance. Bench-scale tests have demonstrated that 90%
NOX reduction can be achieved with an ammonia slip of <5 ppm.
6A-45
-------
CATALYTIC FABRIC FILTRATION FOR
SIMULTANEOUS NOX AND PARTICULATE CONTROL
INTRODUCTION i BACKGROUND
In 1990, the first major reauthorization of the Clean Air Act since 1970 was
enacted by Congress and signed into law by the President of the United States.
Although S02 emissions are still the primary focus of acid rain control, studies in
Europe and the United States investigating the role of NOX in acid rain formation
and ozone chemistry have resulted in NOX control being an important component of
the new Clean Air Act (1,2). Specifically, the Clean Air Act Amendments of 1990
require a two million-ton reduction in NOX emissions by January 1, 1995.
Expectations are that NOX emissions will be regulated more strictly at the local
level (state and local regulatory agencies) than as currently addressed under the
reauthorized Clean Air Act. Therefore, technology capable of achieving higher
levels of NOX control than those demonstrated by low NOX burners must be developed.
For the past six years, the Energy and Environmental Research Center (EERC), using
fabrics developed by Owens-Corning Fiberglas (OCF), has pursued the development of
the catalytic fabric filtration concept as an advanced NOX control technology. The
overall objective of the project is to evaluate the potential of a catalytic
fabric filter for simultaneous NO, and particulate control. Specific goals include
the following:
• 90% NOX removal efficiency with <25 ppm ammonia slip.
• A particulate removal efficiency of >99.5%.
• A bag/catalyst life of >1 year.
• A 20% cost savings over conventional baghouse and SCR control
technology.
• Compatibility with S02 removal systems.
• A nonhazardous waste material.
Even though promising results were obtained in the early bench-scale work funded
by OCF, a continued effort was needed to further develop the product that would
give the best combination of high NO, removal capability, low ammonia slip, high
particulate removal efficiency, and long catalyst/bag life.
Specific activities have progressed from bench-scale experiments using simulated
flue gas (Task A) and flue gas from a pc-fired source (Task B) to pilot-scale
experiments with catalyst-coated bags. Specific parametric and fabric-screening
tests using simulated flue gas (Task A) were conducted in which the fabric weave,
coating composition, and coating process were adjusted to develop acceptable
fabrics for further testing. Task B, which is the focus of this paper, involved
the testing of ten catalyst-coated fabric samples developed by OCF using a
6A-46
-------
slipstream of flue gas from EERC's Participate Test Combustor (PTC). Based on the
results of these bench-scale experiments, tests with catalyst-coated filter bags
are scheduled to begin in the summer of 1991.
RESULTS & DISCUSSION
The purpose of Task B was to further evaluate catalyst-coated fabric samples in
the presence of flue gas generated during pulverized coal combustion. This was
considered necessary to begin evaluating the potential effects of fly ash on
catalytic performance: specifically, the effects of submicron particles, volatile
species, and trace elements that could not be addressed using synthetic flue gas.
Ten catalyst-coated fabric samples (Fabrics #2, #3, #4, #5, #7, #13, #14, #15,
#17, and #18) developed by OCF were selected for testing. The criteria for
selecting these fabric samples for further evaluation were high NOX removal
efficiency and/or low ammonia slip, based on Task A results. Detailed
descriptions of eight catalyst-coated fabric samples were presented in a previous
report (3). Fabrics #17 and #18 were catalyst-coated fabric samples recently
developed by OCF. Fabric #17 was similar to previously tested Fabric #2, except
that a different vanadium source was used to prepare the coating, and
modifications were made to increase the surface area.
The catalyst coated on Fabric #18 was a new iron-based catalyst. Iron compounds
have been shown to be effective catalysts for reducing NOX (4). In addition, it
may broaden the temperature window for the NOX reduction reactions.
Four coals were selected for Task B testing, a medium-sulfur washed Illinois #6
bituminous (the baseline coal), a high-sulfur Pyro Kentucky bituminous, a Jacobs
Ranch subbituminous, and a South Hallsville, Texas, lignite. Each of the ten
fabrics was tested with the washed Illinois #6 bituminous coal at air-to-cloth
ratios of 2, 3, 4, and 6 ft/min. Ammonia slip and S03 measurements were made at
each air-to-cloth ratio. The ammonia/NOx molar ratio was to be held constant at
0.9; however, due to an error in calculating an orifice coefficient, several tests
were conducted at an ammonia/NOx molar ratio of 1.1. Cloth weight in all instances
was 14 ounces per square yard.
Based on the results of the first eight fabric-screening tests, two fabric
samples, #2 and #13, were selected to be tested using the remaining three coals.
For the first 6 hours of the test, the air-to-cloth ratio was held constant at 3
ft/min. However, near the end of each test, the air-to-cloth ratio was adjusted
to 2 ft/min for 1 hour and then 4 ft/min for 1 hour. The ammonia/NOx molar ratio
was held constant at 0.9. The slipstream sample system used to perform the tests
is shown in Figure 1.
The results of the Task B fabric-screening tests are presented in Table 1. These
results are consistent with the values reported for Task A. As expected, there
was a marked decrease in NOX removal efficiency with increased air-to-cloth ratio.
An example of this is shown in Figure 2. Although there was some variability in
the operation of the combustion system, NOX removal efficiency was relatively
constant with time. Fabric #2 appeared to have demonstrated the best overall
performance of the first eight fabric samples tested, with respect to high NOX
removal and low ammonia slip.
6A-47
-------
The results for Fabric #17, with the new vanadium source, compared favorably to
Fabric #2, which is similar in all other respects. The two fabrics are compared
directly in Figure 3. As can be seen, with the exception of the ammonia slip at
an air-to-cloth ratio of 2 ft/min, the results are very similar. Figure 4 shows
the actual ammonia/NOx molar ratio as a function of time for Fabric #17. As is
shown in the figure, the ammonia/NOx molar ratio averaged about 0.95 for the test
at an air-to-cloth ratio of 2.2 ft/min. This may have been the reason for the
higher ammonia slip at the lowest air-to-cloth ratio. Figure 4 data are typical
of the variability in ammonia/NOx molar ratio for all the tests.
For Fabric #18, the results did not seem to be very impressive (an NOX removal
efficiency of 64% at an air-to-cloth ratio of 2 ft/min); however, this is
promising, as the coating process for iron has not been optimized. As stated
earlier, iron presents several potential advantages over vanadium; however,
further development by OCF will be necessary to improve its performance.
From the fabric-screening data, the maximum air-to-cloth ratio that can be used
and still obtain >85% NOX removal efficiency is 3 ft/min, which is consistent with
the bench-scale results using simulated flue gas (Task A). For all the catalyst-
coated fabric samples, there was a marked decrease in catalytic performance at
air-to-cloth ratios of 4 and 6 ft/min.
Following completion of the first eight fabric-screening tests, fabric samples #2
and #13 were chosen to test the effects of coal type on fabric performance. Both
fabrics were tested using the three remaining coals: South Hallsville, Texas,
lignite; Jacobs Ranch subbituminous; and a Pyro Kentucky bituminous at an air-to-
cloth ratio of 3 ft/min, ammonia/NOx molar ratio of 0.9, and temperature of 650°F.
Table 2 summarizes the results from these tests as well as data from the previous
screening tests using the washed Illinois #6 bituminous coal. The data are also
represented graphically in Figures 5 and 6.
From the data, it appears that NOX removal efficiency with Fabric #2 was similar
(85% to 90%) for three of the four coals fired in the pilot-scale combustor. The
exception was observed when firing the South Hallsville, Texas, lignite. Although
an obvious explanation of this result (80% NOX removal efficiency and 121 ppm
ammonia slip) is not apparent, EERC believes that the filtration characteristics
of the South Hallsville fly ash may have contributed to the observed result.
Specifically South Hallsville, Texas, lignite is known to produce an ash that is
difficult to collect in a fabric filter (5). A large number of pinholes were
present in the dust cake at the conclusion of the test. Pinholes may result in
localized areas of very high air-to-cloth ratios which, depending on the number
and size of the pinholes, can limit contact between the flue gas and the catalyst,
resulting in decreased NOX removal efficiency and increased ammonia slip.
For Fabric #13, the results using South Hallsville, Texas, lignite were more
successful, as excessive pinholing did not occur. Although the NO^ removal
efficiency was somewhat lower, about 83% compared to 86% and 90% for the Jacobs
Ranch and Illinois #6 coals, respectively, the data is not conclusive. Therefore,
the effect of coal type, if any, on catalyst-coated fabric performance has not yet
been determined. The results using the Pyro Kentucky bituminous coal with
Fabric #13 are suspect due to an upset in the pilot-scale combustion system.
6A-48
-------
Excessive slagging resulted in an unstable flame in the burner, causing an early
shutdown of the test.
Table 3 presents surface area and catalyst data for each of the catalyst-coated
fabric samples tested. Both were measured prior to exposure to the flue gas and
after completion of the reactivity tests. In all cases, there was a substantial
decrease in surface area after exposure to flue gas. But, for most of the fabric
samples tested, the catalyst concentration decreased only slightly or remained
constant with exposure to flue gas. However, this indicates that the decrease in
surface area is not due to sluffing of the catalyst from the fabric surface. The
decrease in surface area may be due to a slight sintering effect, possible
plugging of the surface pores by submicron aerosols or fly ash particles, or due
to residual carbon burnout in the coatings.
The initial BET surface area for both Fabrics #17 and #18 was higher than previous
fabrics. However, the surface area for Fabric #17 after exposure to flue gas
(which gave results very similar to Fabric #2) decreased to a level that was
essentially the same as that observed for Fabric #2. For Fabric #18 (iron
catalyst), there seems to have been almost a complete collapse of surface area.
The reason for this is not known at this time; however, it was speculated by OCF
that there may be some temperature effects. Figure 7 shows the NOX removal
efficiency as a function of the surface area after exposure to flue gas. One
surface area point does not fit the curve. This data point represents Fabric #7,
and a final determination concerning its validity has not been made. Fabric #7
may be tested again during upcoming pilot-scale activities. Although other
factors such as weave texturization may also be important, the figure shows that
NOX removal efficiency is directly proportional to the surface area. Based on this
data, the minimum BET surface area needed to achieve 85% NO, removal efficiency at
an air-to-cloth ratio of 3 ft/min is about 4-5 m2/g.
For Fabrics #17 and #18, N20 was measured at the inlet and outlet of the catalyst-
coated fabric. The measurements are shown in Table 4. Within the limits of the
instrument, the table shows that there is no apparent conversion of NOX to N20
across the catalyst-coated fabric. Downstream N20 values ranged from 4 to 6 ppm.
This is consistent with results presented by other researchers (6,7) for a
pulverized coal-fired boiler. Additional measurements will be made when pilot-
scale bag tests begin.
6A-49
-------
CONCLUSIONS
Based on the results of Task B testing, several conclusions can be made.
1. There was a substantial decrease in NOX removal efficiency with increased
air-to-cloth ratio for all the catalyst-coated fabric samples tested. It
appears that for the 14 ounce per square yard fabric samples tested, in
the bench-scale system, the maximum air-to-cloth ratio at which 85%-90%
NOX removal can be achieved is 3 ft/min.
2. Although there was some variability in the data, the NO, removal
efficiency appeared to be constant with time over the short (eight hours)
duration of these tests.
3. Of the fabric samples tested, Fabrics #2 and #17 appear to provide the
best performance with respect to NOX removal efficiency and ammonia slip.
4. Although three of the coals, the two bituminous coals and the
subbituminous coal, resulted in similar catalyst-coated fabric
performance, there appeared to be a reduction in NOX removal efficiency
for the South Hallsville, Texas, lignite. This may have been a result of
pinhole formation.
5. When the catalyst-coated fabric is exposed to flue gas, there is a
decrease in the total surface area. A minimum BET surface area after
exposure to flue gas of 4 to 5 m2/g is necessary to provide good catalyst-
coated fabric performance. Therefore, in order to improve performance,
it would be beneficial to increase the surface area of the catalyst or
the catalyst-coated fabric.
6. There does not seem to be any decrease in catalyst-coated fabric
performance using the new vanadium source. Although the NOX removal
efficiency using the iron catalyst is relatively low, it does show
promise, as the coating process for the iron catalyst has not been
optimized.
7. For these initial tests, there is no apparent conversion of NOX to N20
across the catalyst-coated fabric.
REFERENCES
1. Hjalmarsson, A.K.; Vernon, J. "Policies for NO, Control in Europe,"
Presented at: 1989 EPRI/EPA Joint Symposium on Stationary Combustion NOX
Control, San Francisco, CA, March 1989.
2. Bruck, R.I. "Boreal Montane Ecosystem Decline in Central Europe and the
Eastern United States: Potential Role of Anthropogenic Pollution with
Emphasis on Nitrogen Compounds," Presented at 1985 EPRI/EPA Joint
Symposium on Stationary Combustion NOX Control, Boston, MA, May 1985.
6A-50
-------
3. Weber, G.F.; Laudal, D.L. "Final Technical Project Report for April 1988
through June 1989 for Flue Gas Cleanup," Work performed under DOE
Contract No. DE-FC21-86MC10637, Grand Forks, ND, November 1989.
4. Kato, A.; Matsuda, S.; Nakajima, M.I.; Watanabe, Y. "Reduction of Nitric
Oxide on Iron Oxide-Titanium Oxide Catalyst," Journal of Physical
Chemistry 1981, 85, (12), 1710-1713.
5. Miller, S.J.; Laudal, D.L. "Flue Gas Conditioning for Improved Fine
Particle Capture in Fabric Filters: Comparative Technical and Economic
Assessment," Vol II. Advanced Research and Technology Development, Low-
Rank Coal Research Final Report, Work performed under DOE Contract No.
DE-FC21-86MC10637, Grand Forks, ND, 1987, Vol. III.
6. Aho, M.J.; Rantanen, J.T.; Linna, V.L. "Formation and Destruction of
Nitrous Oxide in Pulverized Fuel Combustion Environments between 750° and
970°C," Fuel 1990, 29, 957-1005.
7. Kokkinos, A. "Measurement of Nitrous Oxide Emissions," EPRI Journal
1990, April/May, 36-39.
6A-51
-------
Thermocouples
To Baghouse
To Gas Pump and
Dry Gas Meter
To Sample Conditioner
for Flue Gas Analysis
Figure 1. Slipstream Sample System
100
^ 90-
>^80-
c
CD 70-
]
-------
Fabric *2 •Fabric #17
A/C = 2 ft/min A/C = 3 tt/min A/C = 4 ft/min A/C = 6 ft/min
NH3/NOx Molar Ratio = 0.9
Figure 3. Comparison of the NOX Removal Efficiency as
a Function of Air-to-Cloth Ratio for Fabrics #2 and #17
CO
DC
_00
O
o
o
.A/C..= .3.ft/m.in A/C.=_2.,2.ft/m|n
700
800
900
Time (min)
Figure 4. Ammonia/N0x Molar Ratio as a Function
of Time for Fabric #17
6A-53
-------
Fabric #2
Air-to-Cloth Ratio (ft/min)
Illinois #6 Jacobs Ranch Pyro Kentucky South Hallsville
NH3/NOx Molar Ratio = 0.9
Figure 5. Comparison of the Catalytic Performance
Using Four Different Coals for Fabric #2
Fabric #13
Air-to-Cloth Ratio (ft/min) • 2 ^ 3 |gg 4
o
c
CD
'o
it=
LJJ
"ro
O
E
-------
10
2 -
50
60
Air-to-Cloth Ratio = 3 ift/min
NH3/NOx Molar Ratio = 0.9
70
80
90
100
NO Removal Efficiency (%)
Figure 7. NOX Removal Efficiency as a Function
of Catalyst-Coated Fabric Surface Area after Exposure
to Flue Gas
6A-55
-------
Table 1
RESULTS FROM TASK B — BENCH-SCALE FABRIC-SCREENING TESTS "b
Fabric
No.
2
2
2
2
2
2
2
2
3
3
3
4
4
4
4
5
5
5
5
7
7
7
7
13
13
13
13
14
14
14
14
15
15
15
15
17
17
17
17
18
18
18
18
A/C
Ratio
fft/min)
2
3
4
4.5
2
3
4
6
2
4
6
2
3
4
6
2
3
4
6
2
3
4
6
2
3
4
6
2
3
4
6
2
3
4
6
2.2
3
4
5.5
2
3
4
6
NH3/NO,
Molar
Ratio
1.1
1.1
1.1
1.1
0.9
0.9
0.9
0.9
0.9
0.9
0.9
0.9
0.9
0.9
0.9
0.9
0.9
0.9
0.9
0.9
0.9
0.9
0.9
1.1
1.1
1.1
1.1
1.1
1.1
1.1
1.1
1.1
1.1
1.1
1.1
0.9
0.9
0.9
0.9
0.9
0.9
0.9
0.9
NO,
Inlet
(pom)
765
716
740
735
540
550
590
630
760
710
720
715
695
675
645
730
700
760
730
700
675
650
660
673
686
688
671
703
729
772
838
847
789
761
656
306
292
268
287
372
401
413
381
NO,
Outlet
(ppm)
20
38
83
64
58
83
112
175
226
390
490
171
235
310
436
90
125
190
305
75
95
175
200
34
64
126
209
89
151
228
433
40
68
98
193
27
31
66
101
134
166
212
224
NO,
Removal
Efficiency
(%)
97.4
94.7
88.8
91.3
89.3
84.9
81.0
72.2
70.3
45.1
31.9
76.1
66.2
54.1
32.4
87.7
82.1
75.0
58.2
89.3
85.9
73.1
69.7
94.9
90.7
81.7
68.9
87.3
79.3
70.5
48.3
95.3
91.4
87.1
70.6
91.2
89.4
75.4
64.8
64.0
58.6
48.7
41.2
Ammonia
Slip
(ppm)
187
63
129
121
5
7
22
76
ND
NO
357
87
127
179
288
28
54
76
163
4
13
33
50
64
58
88
108
107
153
256
179
57
58
104
122
45
17
28
73
102
88
122
172
Particulate
Removal
Efficiency
(%)
99.8
99.8
90.4
99.5
99.9
99.8
99.4
99.8
99.9
99.8
99.9
Each catalyst-coated fabric sample was evaluated using a slipstream of flue gas from a pc-fired pilot-scale
combustor firing a washed Illinois #6 bituminous coal.
"NO" denotes data that are not available due to problems encountered with the sampling system.
6A-56
-------
Table 2
RESULTS FROM TASK B — EFFECTS OF COAL TYPE
Fabric
No.
2
2
2
13
13
13
A/C
Ratio
(ft/min)
2
3
4
2
3
4
NH3/NOX
Molar
ratio
0.9
0.9
0.9
1.1
1.1
1.1
NO,
Inlet
(ppm)
Washed 11
540
535
590
673
686
688
Jacobs Ranch
2
2
2
13
13
13
2
3
4
2
3
4
0.9
0.9
0.9
0.9
0.9
0.9
785
760
800
645
680
675
South Hall
2
13
13
13
3
2
3
4
0.9
0.9
0.9
0.9
900
820
810
825
NOX
Outlet
(ppm)
1 inois #6
58
81
112
34
64
126
, Wyoming,
59
75
90
80
105
195
NOX
Removal
Efficiency
m
Bituminous
89.3
84.9
81.0
94.9
90.7
81.7
Subbituminous
92.5
90.1
88.8
87.6
84.6
71.1
Ammonia
Slip
(ppm)
7
58
86
99
Particulate
Removal
Efficiency
(%)
99.8
99.4
99.9
99.9
sville, Texas, Lignite
175
110
140
195
80.6
86.6
82.7
76.4
121
75
99.8
Pyro Kentucky Bituminous
2
2
2
2
3
4
0.9
0.9
0.9
970
930
925
93
130
178
90.4
86.0
80.8
10
99.7
13
0.9
810
170
79.0
30
99.6
6A-57
-------
Table 3
CATALYST CONCENTRATION AND BET SURFACE AREA
FOR EACH OF THE CATALYST-COATED FABRICS TESTED"
Catalyst Concentration"
Surface Areac
Unexposed
fmq/q)
0.03
9.1
8.4
4.7
4.7
5.5
7.6
6.8
8.4
3.4
7.7
13.2
7.1
exposed
(mq/q)
9.0
8.3
3.7
4.2
5.4
6.3
6.1
8.0
3.6
5.7
13.4
7.4
Change
(%)
1.1
1.2
21.3
10.6
1.8
17.1
10.3
4.8
(5.9)d
26.0
(1.5)'
(4.2)d
Fabric No.
Blank
2
2
3
4
5
7
13
13
14
15
17
18
Unexposed and exposed refer to exposure to flue gas.
Catalyst concentration is mg catalyst per g of coated fabric.
Fabric surface area is m2 per g of coated fabric (BET surface area).
( ) Indicates there was a measured increase in catalyst concentration.
Unexposed
(m2/q)
0.56
9.50
10.68
3.31
4.28
5.79
6.62
5.76
6.52
3.09
6.24
14.61
16.60
Exposed
(mz/q)
6.19
5.11
1.54
2.02
3.74
2.74
4.04
4.00
1.90
3.79
5.05
2.19
Change
34.8
52.2
53.5
52.8
35.4
58.6
29.9
38.7
38.5
39.3
65.4
86.8
Table 4
N,0 CONCENTRATION IN THE FLUE GAS
Air-to-cloth
Ratio
(ft/min)
2.2
3
4
5.5
2
3
4
6
Inlet N20
Concentration
(ppm)
Fabric #17
4.0
3.5
4.0
4.0
Fabric #18
5.5
4.5
4.0
3.5
Outlet N20
Concentration
ppm)
5.0
4.5
4.5
4.5
6.0
5.0
4.5
4.0
6A-58
-------
Session 6B
COMBUSTION NOX DEVELOPMENTS II
Chair: R. Hall, EPA
-------
HETEROGENEOUS DECOMPOSITION OF NITROUS OXIDE IN THE OPERATING
TEMPERATURE RANGE OF CIRCULATING FLUIDIZED BED COMBUSTORS
T. Khan
Y.Y. Lee
L. Young
Ahlstrom Pyropower Inc.
8970 Crestmar Point
San Diego, California 92121
-------
HETEROGENEOUS DECOMPOSITION OF NITROUS OXIDE IN THE OPERATING
TEMPERATURE RANGE OF CIRCULATING FLUIDIZED BED COMBUSTORS
T. Khan
Y.Y. Lee
L. Young
Ahlstrom Pyropower Inc.
8970 Crestmar Point
San Diego, California 92121
ABSTRACT
There is growing concern over the increasing atmospheric nitrous oxide concentration. This concern
stems from the realization that nitrous oxide contributes to the depletion of the ozone layer and to
the greenhouse effect. A research program has been developed at Ahlstrom Pyropower Inc. to study
the emission of nitrous oxide from circulating fluidized bed combustors (CFBCs). The program
involves, in part, an investigation into the mechanism of nitrous oxide formation and destruction in
the operating temperature range of CFBCs. This paper describes a study directed at understanding
the decomposition of nitrous oxide on solid materials known to be present in the combustor.
An electrically heated tubular quartz reactor (2.3 cm I.D.) was used to study the decomposition of
nitrous oxide on six different solid materials; alumina, silica, ceramic beads, sulfated limestone,
calcined amorphous limestone and calcined crystalline limestone. Approximately 10 cm3 of each
solid material was placed in turn in the reactor and a mixture of nitrous oxide (200 ppm) in helium
was passed through the reactor. The concentration of nitrous oxide at the reactor outlet was
measured to determine the extent of N2O decomposition. As a basis for comparison, the
homogeneous phase decomposition of nitrous oxide in the reactor was also studied.
Results showed that a significant amount of N2O decomposed even in the absence of any solid
material in the reactor. It was observed that the presence of solid materials in the reactor enhanced
the decomposition of nitrous oxide and that the degree of enhancement was dependent on the solid
material being tested; calcined limestone, for example, was seen to be highly effective in
decomposing nitrous oxide while ceramic beads showed little or no effect.
6B-1
-------
INTRODUCTION
There is growing concern over the increasing concentration of atmospheric nitrous oxide. This
concern stems from the realization that nitrous oxide contributes to the greenhouse effect and to
the depletion of the ozone layer. The mean concentration of N2O in the atmosphere is 330 ppbv
and it is estimated that it is increasing at a rate of 0.2% per year JJJ.
It has been suggested that fossil fuel combustion is a major contributor to the atmospheric nitrous
oxide inventory. Measurements [21 show that nitrous oxide emissions from circulating fluidized
bed combustors (CFBCs) range from 20 to 120 ppm. Based on these emission values, it is doubtful
that nitrous oxide emissions from fluidized beds contribute more than a minor fraction to the global
inventory. Nonetheless, in accordance with its dedication to developing an environmentally safe
product, Ahlstrom Pyropower Inc. has instituted a project directed at the reduction of nitrous oxide
emissions from AHLSTROM PYROFLOW* boilers. The project involves, in part, an investigation into
the formation and destruction of nitrous oxide in circulating fluidized bed combustors.
Knowledge of the principal reactions involved in the formation and destruction of nitrous oxide in
fluidized bed environments is limited at best. In order to minimize nitrous oxide emissions it is
necessary that:
1. reactions that play a dominant role in the formation and destruction of nitrous oxide be
identified and that
2. the effect of process parameters on the kinetics of these reactions be studied in detail.
Studies [3.41 indicate that hydrogen cyanide (HCN), released during the devolatilization of coal, is
a major precursor of nitrous oxide. It is believed that HCN undergoes oxidation to NCO which in
turn reacts with nitric oxide (NO) to form nitrous oxide (N2O). There is relatively little debate
about the importance of this reaction path as a means of formation of nitrous oxide. Doubts about
it being the only major nitrous oxide formation path have however been expressed. De Soete [51
and Arnand and Andersen [61 have reported the formation of nitrous oxide by the reduction of NO
on char surfaces. De Soete [51 has also reported that nitrous oxide may be formed by the oxidation
of char nitrogen (1-5%) during combustion.
Nitrous oxide destruction in the fluidized bed environment may occur through both homogeneous
and heterogeneous phase reactions. Kramlich et al. £4J and Emola et al. [31 have suggested that
the principal nitrous oxide destruction reaction is its homogeneous phase reduction to nitrogen by
hydrogen radicals. Relatively very little is known about the heterogeneous phase destruction of
6B-2
-------
nitrous oxide. It is believed [51 that nitrous oxide reduction on char is one of the major
heterogeneous N2O destruction pathways. Little or no information currently exists on the interaction
of nitrous oxide with solids, other than char, present in a fluidized bed environment.
This paper describes a study directed at investigating the heterogeneous decomposition of nitrous
oxide in the operating temperature range of a CFBC. An electrically heated tubular quartz reactor
(2.3 cm I.D.) was used to study the decomposition of nitrous oxide on six different solid materials;
alumina, silica, ceramic beads, sulfated limestone, calcined amorphous limestone and calcined
crystalline limestone. Approximately 10 cc of each solid material was placed in rum in the reactor
and a mixture of nitrous oxide (200 ppm) in helium was passed through the reactor. The
concentration of nitrous oxide at the reactor outlet was measured to determine the extent of N2O
decomposition. As a basis for comparison, the homogeneous phase decomposition of nitrous oxide
in the reactor was also studied.
EXPERIMENTAL SET-UP
The experimental set-up used in the course of this study (Fig. 1) consists of three major components:
1. An electrically heated quartz tube that serves as a reactor.
2. Mass flow controllers used to deliver a measured amount of a nitrous-oxide/helium
mixture to the reactor.
3. A HORIBA Non-Dispersive Infrared nitrous oxide analyzer.
Reactor
The reactor, for the major part, is a 91.5 cm long, 2.3 cm I.D. quartz glass tube. Caps at the end
of the tube house ports for the inlet and the outlet of reactant and product gas mixtures. The end
caps also house inlet ports for thermocouples used in measuring and controlling the reactor
temperature. A sintered quartz glass filter is provided 50.8 cm from one end of the tube and serves
to support a bed of the solid material being tested. The reactor is heated by a three zone, 61 cm
long electric furnace. The two outermost zones of the furnace are 15.25 cm long and the central
zone is 30.5 cm in length. Each furnace zone is independent of the others in its temperature control.
Thermocouples inside the reactor serve as sensors for controllers that control the temperature of each
furnace zone.
Mass Flow Controllers
Two mass flow controller modules, one for a nitrous-oxide/helium mixture (0.4% N2O) and the
6B-3
-------
other for pure helium were used in the course of this study. Using these controllers it was possible
to feed mixtures of nitrous oxide in helium at predetermined concentrations and flow rates to the
reactor. It may be mentioned here that helium was chosen as a 'balance gas' due to its chemical
inertness and its high thermal conductivity. The high thermal conductivity was necessary to minimize
radial temperature gradients and the heat up zone within the reactor.
Nitrous Oxide Analyzer
A HORIBA Non Dispersive Infrared N20 analyzer was used to measure the concentration of nitrous
oxide in the inlet and outlet gas streams of the reactor. The analyzer was equipped with a 7.8 ^m
wavelength filter.
EXPERIMENTAL PROCEDURE
Homogeneous Phase Decomposition Study
The reactor was heated to the desired temperature and a 200 ppm mixture of N2O in helium was
fed to the reactor at three different flow rates (500, 1000 and 1500 cmVmin). At each condition,
the concentration of nitrous oxide at the outlet of the reactor was measured to determine the extent
of nitrous oxide decomposition. This procedure was repeated for six reactor temperatures; 650, 700,
750, 800, 850 and 900°C. The results obtained are presented in the following section.
Heterogeneous Phase Decomposition
Approximately 10 cm3 of the material being tested (250jim>mean particle diameter>125/tm) was
placed in the reactor and the reactor was heated to 850°C. A 200 ppm mixture of N2O in helium
was fed to the reactor at a flow rate of 500, 1000 and 1500 cmVrnin. At each condition, the
concentration of nitrous oxide at the outlet of the reactor was measured to determine the extent
of nitrous oxide decomposition. A comparison between the results obtained for each solid material
tested is presented in the following section.
RESULTS
Results of the homogeneous phase nitrous oxide decomposition study are shown in Table 1. As
may be seen from the data, nitrous oxide does not decompose to any significant extent below 700°C.
It is also evident that the rate of nitrous oxide decomposition increases with reactor temperature and
residence time. It is most likely that the products of the nitrous oxide decomposition were nitrogen
and oxygen; no nitric oxide (NO) was detected in the outlet stream from the reactor.
Reaction rate constants for the homogeneous phase decomposition of nitrous oxide were calculated
6B-4
-------
from the obtained data. It was assumed, in the calculation, that the decomposition of N2O occurs
via a first order reaction. The reaction rate constant, k, is presented as a function of temperature
in Table 2. Fig. 2 is a plot of -ln(k) versus 1/T. As may be seen, the plot is a
straight line. This indicates that the assumption that nitrous oxide undergoes a first order
decomposition reaction was correct. The rate of homogeneous phase nitrous oxide decomposition
may thus be written as:
d[N20]/dt = -k [N20]
where, [N2O] is the nitrous oxide concentration at time t. The reaction rate constant, k, a function
of temperature, may be expressed as:
k = koexp[-E/RT]
The value of the activation energy, E, derived from the slope of the plot (E/R) is 246.6 kJ/mol.
The frequency factor, ko, may be derived from the y-intercept of the plot, -ln(ko), and is equal to
2.813 x 1010 sec".
The results of the heterogeneous phase N2O decomposition studies are shown in Table 3. Also
included in the table, for the purpose of comparison, are the results from the corresponding empty
tube (homogeneous phase) experiments. The results show the fraction of nitrous oxide that
decomposes on passage through the reactor. The residence times entered at the top of the table
are the residence times of the gas mixture in the entire heated length of the reactor. The numbers
(1) and (2) are used to distinguish between the two types of limestones used; respectively, the
calcined crystalline limestone and the calcined amorphous limestone. The variation of nitrous oxide
decomposition with total reactor residence time, is shown, for each solid material and the empty tube
experiment, in Fig. 3.
As may be seen from the results, the presence of ceramic beads or sulfated limestone in the reactor
does not significantly affect the decomposition of nitrous oxide. The presence of silica sand or
alumina enhances the decomposition of nitrous oxide to a small extent. The most dramatic results
are those obtained in the presence of calcined limestone. It may be seen that nitrous oxide
decomposes completely in the presence of the calcined crystalline limestone at 850°C. As may be
seen from the graphical results, the conversion in the presence of calcined limestone is dependent
on the kind of limestone used. There is an almost 50% difference in the conversions for the two
types of limestones used at a reactor residence times of 3.2 sec. As in the case of the homogeneous
6B-5
-------
phase decomposition studies, no NO was detected in the outlet stream from the reactor.
DISCUSSION AND CONCLUSIONS
Based on the observations and results described in this paper, the following conclusions may be
drawn.
1. The homogeneous phase thermal decomposition of nitrous oxide is a very important pathway for
nitrous oxide destruction in a fluidized bed combustor. A simple calculation shows that under
typical operating conditions in a circulating fluidized bed, that is, a gas residence time of 6
seconds at an average furnace temperature of 870°C, over 60 percent of the nitrous oxide present
at the bottom of the combustor would be destroyed before it reached the combustor exit.
Furthermore, if the average operating temperature of the unit were to be increased by 10°C, the
extent of N2O decomposition would be increased to 70%. It has been seen in measurements on
commercial scale CFBCs [21 that the nitrous oxide emission level does in fact decrease
significantly with increasing bed temperature. It must be realized, of course, that the rate of
nitrous oxide formation is also temperature dependent.
2. Of the solid materials tested, calcined limestone was seen to decompose nitrous oxide most
efficiently. Alumina and silica sand were seen to slightly enhance the decomposition of nitrous
oxide and ceramic beads and sulfated Limestone were seen to have virtually no effect on the
extent of nitrous oxide decomposition. One would expect, in the light of these observations, to
see a dramatic decrease in nitrous oxide emissions with increasing feed Ca/S ratio in a CFBC.
This, however, has not been the case. Studies on a 0.8 MW^, Ahlstrom Pyroflow pilot plant [21
show only a slight reduction in nitrous oxide emissions with increasing feed Ca/S ratio; no
definite relationship between nitrous oxide emissions and feed Ca/S ratio could be detected for
a similar study [21 carried out on commercial scale units.
3. The efficacy of calcined Limestone in decomposing nitrous oxide was dependent on the type of
Limestone used. Calcined crystalline Limestone was seen to decompose nitrous oxide more
effectively than was calcined amorphous Limestone. At a reactor residence time of 3.2 seconds,
the calcined crystalline Limestone was seen to completely decompose the nitrous oxide, where, the
calcined amorphous Limestone was seen to decompose only 50% of the inlet nitrous oxide.
ACKNOWLEDGEMENT
The authors gratefully acknowledge partial funding of the described study by the Finnish Ministry
of Trade and Industry through the LIEKKI program.
6B-6
-------
REFERENCES
1. R.F. Weiss, Journal of Geophysical Research, vol. 86, 1981, p. 7185.
2. M. Hiltunen, P. Kilpinen, M. Hupa and Y.Y. Lee, "N2O Emissions from CFB Boilers: Experimental
Results and Chemical Interpretation." To be presented at the 11th Int. Conf. on Fluidized Bed
Combustion. Montreal, 21-24 April, 1991.
3. P. Ernola & M. Hupa, "Kinetic Modelling of Homogeneous N2O Formation and Destruction in
Fluidized Bed Conditions." Proceedings of the Joint Meeting of the British and French Sections
of the Combustion Institute. Rouen, 1989, p. 21.
4. J.C. Kramlich, J.A. Cole, J.M. McCarthy, W.S. Larder & J.A. McSorley, "Mechanisms of Nitrous
Oxide Formation in Coal Flames." Combustion and Flame. 1989, vol. 77, p. 375.
5. G.G. De Soete, "Heterogeneous NO and N2O Formation from Bound Nitrogen during Char
Combustion." Proceedings of the Joint Meeting of the British and French Sections of the
Combustion Institute. The Combustion Institute, 18-21 April, Rouen, 1989, p. 9.
6. L.E. Amand & S. Andersen, "Emissions of Nitrous Oxide (N2O) from Fluidized Bed Boilers."
Proceedings of the 1989 International Conference on Fluidized Bed Combustion, vol. 1, pp. 49-
56.
6B-7
-------
MASS FLOW CONTROLLERS
GAS SUPPLY CYLINDERS
REACTANT GAS INLET _
^n-
r
A
N20/He
eJ&i
i
61
I
i
cm
»
s
QUA
I
[—*
QUARTZ GLASS TUBULAR
REACTOR
THREE ZONE
ELECTRIC FURNACE
QUARTZ GLASS FRIT
PRODUCT GAS TO
ANALYZERS
THERMOCOUPLES
-oki-
REACTOR BYPASS LINE
Figure 1. Experimental Setup for Quartz Tube Reactor Studies
1.0E-3
Figure 2. -in(k) Vs. 1/T
6B-8
-------
o
eg
'in
o
CL
E
o
o
v
Q
o
CM
0.9 -
0.8 —
0.7 -
c
o'
CM
o 0.6 —
O
CM
0.5 -
0.4 -
0.3 -
0.2 -
0.1 -
0.0
v v Empty Tube
O o Alumina
o — o Ceramic Beads
• • Silica Sand
A A Sulfated Limestone
» » Calcined Limestone (1)
• • Calcined Limestone (2)
T
10
Reactor Residence Time (sec)
Figure 3. Fractional N 0 Decomposition vs. Reactor Residence Time
6B-9
-------
Table 1
HOMOGENEOUS PHASE DECOMPOSITION OF NITROUS OXIDE
Reactor Pressure : 3 psig Inlet N2O Concentration : 200 ppm
Temperature
<*C)
650
700
750
800
850
900
Residence Time
(sec)
11.7
5.8
3.9
11.1
5.5
3.7
10.5
5.3
3.5
10.0
5.0
3.4
9.6
4.8
3.2
9.2
4.6
3.1
NHO Outlet Concentration
(ppm)
200
200
200
197
200
200
185
200
200
150
174
182
78
125
148
12
52
81
Table 2
HOMOGENEOUS PHASE N2O DECOMPOSITION REACTION RATE
CONSTANT VS. TEMPERATURE
Temperature
(*C)
700
750
800
850
900
k
(sec'1)
0.001350
0.007399
0.028640
0.097444
0.301750
6B-10
-------
Table 3
FRACTIONAL N2O DECOMPOSITION VS. TOTAL REACTOR RESIDENCE TIME
Reactor Temperature : 850 C Reactor Pressure : 3 psig
Material
Alumina
Ceramic Beads
Silica Sand
Sulfated Limestone
Calcined Limestone (1)
Calcined Limestone (2)
t=9.6s
0.65
0.61
0.63
0.61
1.00
0.92
t=4.8s
0.41
0.38
0.40
0.38
1.00
0.65
t=3.2s
0.29
0.26
0.28
0.26
0.98
0.50
Empty Tube
0.61
0.38
0.26
6B-11
-------
NOx CONTROL IN A SLAGGING COMBUSTOR FOR A
DIRECT COAL-FIRED UTILITY GAS TURBINE
P. J. Loftus and R. C. Diehl
Energy Technology Office/Textron Defense Systems
(Formerly AVCO Research Laboratory)
2385 Revere Beach Parkway
Everett, MA 02149
and
R. L. Bannister and P. W. Pillsbury
Westinghouse Electric Corp.
The Quadrangle, 4400 Alafaya Trail
Orlando, FL 32826-2399
-------
NOX CONTROL IN A SLAGGING COMBUSTOR FOR A
DIRECT COAL-FIRED UTILITY GAS TURBINE
P. J. Loftus and R. C. Diehl
Energy Technology Office/Textron Defense Systems
(Formerly AVCO Research Laboratory)
2385 Revere Beach Parkway
Everett, MA 02149
and
R. L. Bannister and P. W. Pillsbury
Westinghouse Electric Corp.,
The Quadrangle, 4400 Alafaya Trail
Orlando, FL 32826-2399
Joint EPA/EPRI Symposium on Stationary Combustion NOX Control
Washington, D.C., March 25-28, 1991
ABSTRACT
A modular combustion concept, which emphasizes controlled coal
thermochemistry, has been developed for application in direct coal
firing of utility gas turbines. The approach under investigation is
based on a multi-stage, slagging combustor, which incorporates NOX,
SOX and particulate emissions control. This approach allows raw
utility grade coal to be burned, thereby maintaining a low fuel cost.
The cost of electricity from combined cycle plants incorporating a
direct coal-fired gas turbine is expected to be significantly lower
than that from conventional pulverized coal steam plants.
The first stage, the primary combustion zone, is operated fuel-
rich to inhibit NOX formation from fuel-bound nitrogen and has a jet-
driven, toroidal vortex flow field, which provides for efficient,
stable and rapid combustion at high heat release rates. Impact
separation of molten mineral matter is accomplished in the second
stage, which is closely integrated with the primary zone. The second
stage may also include a slagging cyclone separator for additional
slag removal. This is a novel application for a cyclone separator.
Final oxidation of the fuel-rich gases and dilution to achieve the
desired turbine inlet conditions are accomplished in the third stage.
6B-15
-------
Rapid quenching and good mixing with the secondary air are employed to
avoid thermal NOX formation by minimizing peak flame temperatures and
residence times in the third stage.
The combustor concept has been extensively tested at a thermal
input of 3.5 MWt (12 MM Btu/hr) and a pressure of 6 atmospheres. Both
pulverized coal and coal-water mixtures have been successfully fired.
The combustor has demonstrated stable and intense combustion, with
excellent carbon conversion, efficient slag capture, retention of most
of the coal alkali in the slag and low pressure and heat losses. The
staged combustion NOX control strategy has proved very effective: NOX
emissions are approximately one fifth of the New Source Performance
Standards requirements.
BACKGROUND
The authors' companies are working under Department of Energy
sponsorship to develop the technology base for direct coal-firing of
utility gas turbines. The approach under investigation is based on a
multi-stage, slagging combustor, which incorporates NOX, SOX and
particulate emissions control. This approach allows raw utility grade
coal to be burned, thereby maintaining a low fuel cost. The cost of
electricity from combined cycle plants incorporating a direct coal-
fired gas turbine is expected to be significantly lower than that from
conventional pulverized coal steam plants with flue gas
desulfurization (Pillsbury et al., 1989).
The program objective is to develop an efficient combustor
capable of meeting the New Source Performance Standards (NSPS) for
NOX, S02 and particulates upstream of the turbine. The program is
divided into three key tasks. The first of these is the design,
fabrication and testing of a subscale slagging combustor (6 atm, 3.5
MWC). This task is in progress: combustor testing commenced in late
1988 at the Textron Defense Systems/Energy Technology Office (ETO)
Haverhill test facility. The second task involves testing the final
subscale combustor configuration with a stationary cascade to study
the effect of deposition, erosion and corrosion on air-cooled turbine
vanes. Based upon the data generated, the final task is to design,
fabricate and test a full size combustor (14 atm, 20 MWt) . This paper
discusses the design and performance of the subscale slagging
combustor from the point of view of NOX emissions control.
COMBUSTOR CONCEPT
The three stage combustor is illustrated schematically in Figure
1. The design of the first stage, the primary combustion zone, is
based on Avco Research Laboratory's toroidal vortex combustor concept,
and provides for efficient, stable and rapid combustion at high heat
release rates (Mattsson and Stankevics, 1985, Stankevics et al.,
1983). Coal and preheated air are fed coaxially into the primary zone
through four separate injectors which are inclined upward at
approximately 60° to the horizontal. The coaxial injection promotes
intense coal/air mixing, leading to rapid coal particle heating and
devolatilization, which minimizes carbon burn-out time. The four
inlet coal/air jets converge at the combustor centerline and form a
single vertically directed jet. This jet impacts the center uf the
primary zone dome, where it is turned and a toroidal vortex is formed.
This arrangement forces a high degree of controlled combustion product
6B-16
-------
re-circulation which leads to extremely intense and very stable
combustion of a wide variety of fuels. The toroidal vortex design
gives very high volumetric heat release rates for coal combustion (up
to 40 MWt/m3) . These heat release rates are some three to four times
that for cyclone-type combustors, leading to smaller combustor sizes
and lower wall heat losses. Fuel-rich conditions in the primary zone
inhibit NOX formation from fuel-bound nitrogen. Extensive use was
made of 3-D combustion modelling techniques in the preliminary design
of the combustor (Chatwani and Turan, 1988, Loftus et al., 1988).
The toroidal vortex provides the mechanism for flame
stabilization and also for inertial separation of larger ash/slag
particles. Partial separation of mineral matter and char at the top
of the toroidal vortex zone results in initiation of wall slagging
there, with continued deposition and flow over all exposed wall
surfaces. In order for successful slag deposition in the dome region,
enough coal particle residence time and combustion product re-
entrainment must be provided to ensure rapid coal particle burnout,
resulting in molten, free mineral matter. Larger, partially
devolatilized coal particles will continue to burn, either in
suspension or in the wall slag layer. The primary zone was designed
for approximately 90 percent suspension burning and 10 percent wall
burning. The primary zone particle residence time is of order 100
msec for a 75 micron diameter particle. The primary zone slag layer
provides thermal and erosion protection for the combustor walls, in
addition to a mechanism for oxidation of deposited char. The slag
layer formed from this portion of the mineral matter eventually
reaches the impact separator, where it is collected in the slag
bucket.
The major fraction of mineral removal from the gas is obtained in
the second stage impact separator which is at the exit from the
primary zone. The separation of combustion and slag removal duties
between the two stages has two substantial benefits. First, it
results in maximum removal of carbon free slag: at the primary zone
exit there is a very high carbon conversion fraction—essentially all
the coal char has been oxidized, leaving free mineral matter behind.
Second, due to the low density of the combustion products, a simple
impactor allows high efficiency separation of fine mineral matter
particles at low cost in pressure drop. Overall, the air pressure
drop is optimally distributed, first for combustion stabilization and
second for mineral matter removal.
Pulverized limestone sorbent is used for control of sulfur
emissions. The sulfur control technique used is based on related ETO
work on super-equilibrium sulfur capture (Abichandani et al., 1989).
The sorbent is injected into the primary zone combustion products,
which generally contain a mixture of S02/ H2S and COS, at the exit of
the primary zone, just upstream of the exit nozzle. Reacted sorbent
is removed along with the coal ash in the second stage impact
separator.
Final oxidation of the fuel-rich gases and dilution to achieve
the desired turbine inlet conditions are accomplished in the third
stage. Rapid quenching and good mixing with the secondary air are
employed to avoid thermal NOX formation by minimizing peak flame
temperatures and residence times in the third stage.
6B-17
-------
NOX CONTROL APPROACH
Emissions of nitrogen oxides in the products of combustion are
controlled by adopting the following approach:
• Sub-stoichiometric (fuel-rich) combustion conditions in the
first combustor stage.
• Effective control of the gas temperature and stoichiometry
histories during final oxidation/dilution in the third
combustor stage.
The main source for formation of nitric compounds in the
combustion of coal is fuel-bound nitrogen. Part of the fuel nitrogen
is released with the volatiles in the early stages of combustion and
the remainder is retained by the char residue and released during
subsequent char oxidation. Nitric oxide can be produced by the
oxidation of the nitrogen in the volatiles or in the char. NOX
formation from fuel bound nitrogen is very sensitive to the combustion
stoichiometry. It is known that HCN and NH3 are formed in the gas
upon evolution of coal nitrogen. These can subsequently be oxidized
to NO or can react with NO to form harmless molecular nitrogen,
depending upon the availability of oxidants in the gas. Fuel-rich
operation promotes the formation of molecular N2 as the end product of
the fuel nitrogen, whereas fuel lean operation, with the availability
of oxidants, results in NO formation.
Volatile nitrogen is defined as that which is produced from the
volatile coal fractions and reacts in the gas to form N2, NO, HCN or
NH3. Char nitrogen is that which is associated with a solid, either
as a pyrolysis product of tars or as the original coal char. The
distribution of nitrogenous species between volatile nitrogen and char
nitrogen is critically dependent on the coal particle heating rate,
the peak temperature, the residence time at high temperature and the
nitrogen distribution within the coal (Smart and Weber, 1989) . Fuel-
bound nitrogen is the major source of NOX in conventional PC
combustion, typically accounting for more than 80 percent of total NOX
emissions (Pershing and Wendt, 1979).
For staged combustion to be effective, it is important to avoid
the carry over of either volatile or char nitrogen to the final
oxidation zone, where these can be converted to NO. The intense and
rapid mixing produced by the toroidal vortex design leads to rapid de-
volatilization of .the coal, homogeneous combustion conditions and
efficient oxidation of the char to a fuel-rich gas in the first stage.
These conditions favor conversion of fuel bound nitrogen to molecular
nitrogen and minimize the possible carry-over of volatile or char
nitrogen to the third combustor stage.
For the case of PC combustion, the calculated equilibrium
concentrations of nitrogen oxides in the combustion gas are shown in
Figure 2 for various primary zone fuel air equivalence ratios and
temperatures. This plot includes NH3 and HCN, which have been
converted to total NOX and included in the concentrations shown. (The
contributions from these species are typically small.) NOX
concentrations at the adiabatic flame temperature and at 100 K and 200
K below the adiabatic flame temperature are shown. The NOX
concentrations in the gas corresponding to the NSPS limitsX for
6B-18
-------
bituminous coal (0.6 Ibs per MM Btu) and sub-bituminous coal (0.5 Ibs
per MM Btu) are also shown as a function of fuel-air equivalence
ratio. The strong temperature dependence of NOX is clearly seen: a
temperature drop of 200 K typically reduces the equilibrium NOX by a
factor of three or more. The equilibrium NOX concentration in the gas
becomes less than the NSPS standard at fuel-air equivalence ratios
greater than about 1.2 and at the primary zone nominal operating point
(equivalence ratio in the range 1.3 to 1.4) the equilibrium NOX in the
primary zone is more than a factor of ten less than the NSPS
requirement.
The control of stoichiometry and temperature in the third
combustion stage is key to minimizing the formation of thermal NOX.
The formation of thermal NOX is governed by the highly temperature
dependent reactions between nitrogen and oxygen, the Zeldovich chain
reactions. The rate of formation is significant only at temperatures
above approximately 1900 K (3000°F) and increases with increasing
oxygen concentration. Thus temperatures in the final oxidation zone
should be maintained below this value to avoid thermal NOX formation.
The secondary air for final combustion in the last combustor stage is
added in such a manner that the gas is rapidly quenched and maintained
at a temperature below which thermal NOX can form, while final
oxidation of the unburned species in the gas is completed. As soon as
the final oxidation is complete, the dilution air is introduced, again
with rapid and complete mixing, in order to quench all further NOX
generation.
Kinetic calculations were performed to determine the desired
temperature and operating conditions during final oxidation and the
appropriate split between quench/final oxidation air and dilution air.
These calculations showed that an adiabatic flame temperature of about
1800 K is reached for a fuel air equivalence ratio of 0.6 in the
intermediate zone and that the final oxidation of the gas is completed
within a few milliseconds, see Figure 3. At these conditions thermal
NOX formation is insignificant and the predicted final NOX
concentration in the gas will be only a small fraction of the NSPS
limit. It is important to obtain effective mixing of the secondary
combustion air with the hot fuel-rich primary gases.
Three-dimensional aero-thermal calculations and analysis of the
mixing process in both the intermediate and dilution zones of the
third combustor stage were conducted. The number, size and
orientation of the intermediate and dilution zone jets were varied to
arrive at the optimum mixing performance, expressed as a minimum exit
temperature pattern factor. The final design analysis involved
extending the three-dimensional aero-thermal flow modelling of the
third stage to the full reacting flow field. However, no attempt was
made to optimize the lean-burn combustor from the point of view of NOX
control. The principal purpose of the experimental work was to tackle
the major technical issues in this development effort, which are
related to obtaining efficient primary zone and slag separator
performance.
TEST ARRANGEMENT AND COMBUSTOR OPERATION
The combustor concept has been extensively tested at a thermal
input of 12 MM Btu/hr (3.5 MWt) and a pressure of 6 atmospheres.
Tests have been conducted with both pulverized coal (PC) and coal-
6B-19
-------
water mixture (CWM) fuels. A photograph of the subscale slagging
combustor test arrangement as currently installed at ETO's Haverhill
test site is shown in Figure 4. The nominal test conditions for the
subscale combustor are as listed in Table 1. An oil fired air pre-
heater is used to heat the combustion air in order to correctly
simulate the gas turbine compressor discharge conditions. A
downstream sonic orifice is used to control the combustor chamber
pressure. After pressure let-down, the combustor exit gases are water
quenched before being led to an exhaust stack. The subscale combustor
is water cooled, the cooling water being re-circulated via a cooling
tower. All fuel-rich zone components are lined with a high alumina
refractory. This is both to reduce heat losses in this small scale
experimental combustor and to promote slagging during the relatively
short tests.
Start-up and operation of the system proved to be simple and
reliable. After establishing the correct air flow rates through the
system and allowing the air pre-heater to come up to design
temperature, a methane/oxygen torch in the primary zone is ignited.
The torch is used to ignite a fuel oil flame and is then extinguished.
Fuel oil is then burned for approximately 15 minutes, in order to pre-
heat the refractory liner. The oil is injected via two spray nozzles
in the primary zone. After the refractory liner has reached operating
temperature, the coal (PC or CWM) flow is started, and the fuel oil
flow is stopped. In PC testing, a pneumatic conveying system is used
to feed coal to the primary zone. For CWM testing, a Moyno
progressing cavity pump is used to supply CWM to the combustor. The
CWM atomizers are Parker-Hannifin air-assist atomizing nozzles.
Atomizing air for CWM tests is supplied from a high pressure tube
trailer via a heat exchanger. The heat exchanger warms the expanded
high pressure air back up to approximately ambient temperature.
A detailed fuel specification for the proposed application was
prepared by AMAX Extractive Research and Development. Choice of coal
(and consequently of mineral matter composition), coal particle size
and CWM composition affects certain primary design constraints for the
slagging combustor. These include liquid slag formation, combustion
efficiency, downstream corrosion, erosion and deposition and pollutant
generation. The primary zone of the combustor was designed to operate
at highly fuel-rich (i.e. low flame temperature) conditions. The
flame temperature is obviously even lower for CWM fuels. Consequently
a low ash fusion temperature coal was desirable. The ratio of ash to
sulfur content is of interest: the higher the coal sulfur content, the
higher the ratio of limestone sorbent to ash in the slag to be
separated and the greater the effect of sorbent injection on slag
properties. The preferred coal fuels were determined to be high
volatile eastern bituminous coals. These coals have the advantages of
a high heating value, leading to favorable combustion characteristics
with high flame temperatures and rapid combustion. They also tend to
have low to medium sulfur contents and soluble alkali contents below
0.05 percent. From this general specification, several specific seams
were identified for use in the subscale combustor testing. These
included a low and a high sulfur eastern bituminous coal and a low
sulfur western sub-bituminous coal. Detailed coal specifications are
given in Table 2. The CWM fuels tested were prepared from close to
standard grind (95 percent through 200 mesh) coals.
A full test program was conducted with PC feed before switching
6B-20
-------
TABLE 1
SUBSCALE SLAGGING COMBUSTOR NOMINAL TEST CONDITIONS
Coal Thermal Input
Coal Feed
Atomizing Air/CWM Mass Flux Ratio
Oxidizer
Primary Zone Equivalence Ratio
Total Mass Flow Rate
Exit temperature
Pressure
Sorbent
Sorbent Molar Ratio
3.5 MWt (12 MM Btu/hr)
95% < 200 mesh PC
95% < 200 mesh, 60% solids CWM
1.0
620 K (650°F) pre-heated air
1.3 to 1.4
3.2 kg/s (7 Ib/s)
1300 K (1850°F)
6 atm
-325 mesh limestone
Ca/S = 2
over to CWM feed. From the outset of combustor testing, a stable,
flowing slag layer was formed on the primary zone dome and walls.
Some dissolution of the refractory layer was observed in the early
runs, but after a few hours of operation an equilibrium insulation
layer of slag and refractory was formed. Equilibrium slag layer
thicknesses in the primary zone, where heat fluxes are high, are on
the order of 1 mm. The corresponding thickness in the slag separator
is on the order of 3 mm. No obstruction or fouling of any of the
primary zone coal/air injectors or of the relatively small diameter
primary zone exit nozzle with slag was observed. The impact separator
worked as planned, and a flowing slag layer was observed on the top
and sides of the impactor centerbody and on the slag bucket walls.
TEST RESULTS
A full series of tests with PC fuels demonstrated that the
combustor primary zone produces excellent carbon conversion
performance, see Figure 5. At the nominal primary zone operating
point (fuel/air equivalence ratio of 1.3 to 1.4) the carbon conversion
for PC firing is better than 99 percent. For PC firing, better than
98 percent carbon burnout in the primary zone was obtained for
fuel/air equivalence ratios as high as 1.6. In order to obtain good
carbon conversion performance on CWM fuels, it was necessary to
increase the primary zone aspect ratio (length/diameter). For PC
firing the aspect ratio of the primary zone was 1.25 (L/D = 1.25) .
The optimum configuration for CWM firing was a primary zone aspect
ratio of 1.50. In this configuration, better than 98 percent carbon
conversion was obtained for equivalence ratios up to approximately
1.4. The increase in aspect ratio increases the particle residence
time, thus allowing more time for evaporation of the water in the CWM
droplet. The performance on CWM is slightly worse than that obtained
6B-21
-------
TABLE 2
ANALYSES FOR COALS TESTED TO DATE
Coal Analysis As
Received
% Moisture
% Ash
% Volatile Matter
% Fixed Carbon
% Sulfur
% Chlorine
% Carbon
% Hydrogen
% Nitrogen
% Oxygen
MJ/kg (Btu/lb)
Dorchester,
VA
1.00
6.24
33.30
59.46
0.96
0.04
80.43
4.79
1.72
4.82
32.95 (14,234)
Pittsburgh #8
1.49
7.59
38.28
52.64
2.35
0.14
76.73
5.21
1.34
5.15
32.00 (13,822)
Hanna Seam,
WY
9.09
5.37
38.33
47.21
0.57
0.05
67.09
5.06
1.44
11.33
27.28 (11,784)
for PC, but this is to be expected, given the lower heating value and
flame temperatures of CWM fuels. Measured flame temperatures in the
dome region of the primary zone for PC firing are shown in Figure 6.
The primary zone temperatures at the nominal primary zone operating
point are 2100 to 2000 K (3320 to 3140°F) for PC firing and some 150
to 200 K (270 to 360°F) lower than this for CWM firing.
Figure 7 shows measured primary zone NOX concentrations for
pulverized coal firing. These measurements were made at the exit from
the primary zone, just upstream of the main exit nozzle. The
measurements are compared both with calculated equilibrium NOX levels
for PC firing and also the NSPS limits, as described above. The limit
of resolution of the chemiluminescent analyzer used in making these
measurements is of order 10 ppm. The measured NOX concentrations are
well below the NSPS limits and are in general agreement with the
calculated equilibrium concentrations at 100 to 200 K below the
adiabatic flame temperature. The measured flame temperatures, shown
in Figure 6, are typically 150 to 200 K below the adiabatic flame
temperature.
Corresponding primary zone results and equilibrium calculations
for the case of 60 percent solids CWM firing are shown in Figure 8.
The results for CWM firing are substantially different from those for
PC firing. While the calculated equilibrium NOX concentrations for
CWM firing are lower than those for PC firing, because of the lower
flame temperatures, the measured NOX concentrations at the primary
zone exit for CWM firing are considerably higher than those for PC
firing. The CWM measurements are also considerably higher than the
6B-22
-------
calculated equilibrium concentrations for CWM firing.
This increase in NOX concentrations for CWM firing is also
reflected in the lean zone exit, or exhaust emissions, measurements.
Measured NOX exhaust emissions, corrected to 15 percent oxygen, for
three PC fuels and for 60 percent solids Dorchester CWM are plotted as
a function of primary zone fuel-air equivalence ratio in Figure 9.
These measurements were made at the combustor exit, downstream of the
lean-burn zone. The overall combustor fuel-air equivalence ratio at
the lean zone exit was fixed as the primary zone equivalence ratio was
varied. The dramatic reduction of NOX levels with increased staging
of the combustion is clearly illustrated. The NSPS limit (0.6 Ib/MM
Btu for bituminous coals), scaled for the combustor exit conditions,
is shown for reference. At the nominal design operating point, the
combustor NOX emissions for both PC and CWM firing are well below the
NSPS limit. Not enough information is available to partition the
exhaust NOX emissions between contributions from (1) primary zone NO
generation; (2) lean-burn zone oxidation of volatile or char nitrogen
carried over from the fuel-rich zone; and (3) generation of thermal
NOX in the lean-burn zone. It is obvious, however, that NOX is
generated in the lean-burn zone. For example, NOX levels at the rich
zone combustor exit at equivalence ratios in the range 1.3 to 1.4 (the
nominal primary zone operating point) for PC firing have been measured
at 20 to 40 ppm. The primary zone typically has one third of the
total gas mass flow rate. If no NOX was generated in the lean-burn
zone, the primary zone NOX would therefore be diluted by a factor of
approximately three, giving emissions on the order of 10 to 15 ppm.
The actual emissions at this condition are of order 30 to 50 ppm.
Thus some 20 to 40 ppm NOX are generated in the lean-burn zone. These
20 to 40 ppm are either from thermal NOX in the lean-burn zone or from
lean zone oxidation of char of volatile nitrogen carried over from the
primary zone.
The exhaust NOX emissions for CWM firing are slightly higher than
those for PC firing. At the nominal primary zone operating point, the
CWM emissions are in the range 60 to 80 ppm, compared with 30 to 50
ppm for PC firing. While the precise mechanisms leading to the higher
levels of NOX with CWM firing are not clear at present, several
contributing factors may be identified. As discussed above, the
measured NOX levels at the primary zone exit for CWM firing are much
higher than those measured at the same location for PC firing. In
fact, for CWM firing the measured NOX is in excess of the
thermodynamic equilibrium level. Thus NO destruction would be
expected downstream of the primary zone exit. This indeed appears to
be the case: if the assumption of no NOX generation in the lean-burn
zone is again made, and the primary zone NOX concentration is diluted
by a factor of three, the NOX concentration so obtained is of order
160 ppm, considerably in excess of the measured NOX emission for CWM
firing of 60 to 80 ppm. This suggests that NO is destroyed between
the primary zone exit and the lean zone inlet.
The large differences in primary zone NOX between PC and CWM
firing are indicative of significant differences in temperature,
heating rate and stoichiometry histories in the fuel-rich primary zone
for the two fuels. As discussed above, the partition of the fuel-
bound nitrogen between volatile and char nitrogen and the subsequent
conversion of NOX precursors to molecular nitrogen are strongly
dependent on such parameters. Because of its high moisture content
6B-23
-------
and large size, a CWM droplet will experience both a lower heating
rate and a lower final temperature than a pulverized coal particle.
This may lead to both less complete evolution of fuel-bound nitrogen
and also less efficient conversion of released fuel-bound nitrogen to
molecular nitrogen and consequently to higher NOX emissions.
The post-run appearance of the slag layer in the primary zone
would also indicate that more wall burning occurs for CWM firing than
for PC firing, possibly because of the production of relatively large
coal particle agglomerates on evaporation of the moisture in the CWM
droplet. These larger coal particles will be inertially separated
from the toroidal vortex onto the slagged wall before burning out
completely. Thus the gas phase stoichiometry for CWM burning is
leaner than the global stoichiometry based on air and fuel inputs.
NOX levels at the exit of the primary zone may therefore reflect the
equilibrium levels at leaner conditions, and given enough residence
time, would eventually be reduced to levels indicative of the global
stoichiometry.
Figure 10 shows the exhaust NOX emissions plotted as a function
of the combustor outlet temperature. The nominal design outlet
temperature is 1850°F, at which temperature the NOX emission is some
40 ppm. There is only a moderate increase in NOX emissions as the
outlet temperature is increased to 2000°F.
NOX generation and destruction in staged combustion are
controlled by an extremely complex series of phenomena. Given the
limited amount of experimental information available from a practical
staged slagging combustion system such as the one currently being
tested, it is difficult to completely identify the precise mechanisms
responsible for the results obtained. However, the general concept of
staged combustion for NOX reduction has worked extremely well in this
application, leading to NOX emissions on the order of one fifth of the
NSPS requirements.
CONCLUSION
A three-stage combustion concept has been developed for
application to direct coal-firing of utility gas turbines. A key
aspect of combustor performance is the effective control of NOX
emissions. A subscale combustor (3.5 MWt, 6 atm) is currently being
tested. Results for various coal fuels fired as either PC or CWM have
shown extremely good coal particle burnout leading to effective
slagging in the primary zone. The combustor employs staged combustion
(fuel-rich conditions in the primary zone to inhibit NOX production
from fuel-bound nitrogen; rapid quench/good mixing in lean-burn zone
to reduce peak flame temperature and minimize thermal NOX production)
for NOX emissions control. For primary zone fuel-air equivalence
ratios greater than approximately 1.1 for PC firing and 1.15 for CWM
firing, the subscale slagging combustor NOX emissions are well below
the NSPS limit. Given the high levels of fuel-bound nitrogen in the
coals burned (typically 1.3%), the staged combustion has worked
extremely well to control NOX emissions.
ACKNOWLEDGEMENTS
The work described in this paper is sponsored by the U. S.
Department of Energy through the Morgantown Energy Technology Center
6B-24
-------
under Contract No. DE-AC21-86MC23167. Mr. Donald W. Gelling is the
METC Program Manager.
REFERENCES
Abichandani, J. S., Loftus, P. J., Diehl, R. C., Woodroffe, J. A., and
Holcombe, N. T. (1989) "Nonequilibrium Sulfur Removal from High
Temperature Gases," Proceedings: Sixth Annual Pittsburgh Coal
Conference, Pittsburgh, PA, September, 1989.
Chatwani, A. U., Turan, A., and Stickler, D. B. (1988) "Design and
Sizing of the Primary Stage of a Toroidal Vortex Gas Turbine Combustor
Using a 3-D Flow Field Modelling Code," Western States Section Meeting
of the Combustion Institute, Salt Lake City, UT, March, 1988.
Loftus, P. J., Chatwani, A. U., Turan, A., and Stickler, D. B. (1988)
"The Use of 3-D Numerical Modelling in the Design of a Gas Turbine
Coal Combustor," Heat Transfer in Gas Turbine Engines and Three-
Dimensional Flows, ASME HTD-Vol. 103, pp. 95-105, edited by E. Elovic,
J. E. O'Brien, and D. W. Pepper, New York. Also presented at ASME
Winter Annual Meeting, Chicago, IL, December, 1988.
Mattsson, A. C. J., and Stankevics, J. 0. A. (1985) "Development of a
Retrofit External Slagging Coal Combustor Concept," Proceedings:
Second Annual Pittsburgh Coal Conference, Pittsburgh, Pennsylvania.
Pershing, D. W. and Wendt, J. 0. L. (1979) "Relative Contributions of
Volatile Nitrogen and Char Nitrogen to NOX Emissions from Pulverized
Coal Flames," Industrial and Engineering Chemistry: Process Design and
Development, 18 (1); 60-66, 1979.
Pillsbury, P. W., Bannister, R. L., Diehl, R. C., and Loftus, P- J.
(1989) "Direct Coal Firing for Large Combustion Turbines: What Do
Economic Projections and Subscale Combustor Tests Show?" ASME Paper
89-JPGC/GT-4, Joint ASME/IEEE Power Generation Conference, Dallas,
Texas, October, 1989.
Smart, J. P- and Weber, R. (1989) "Reduction of NOX and Optimization
of Burnout with an Aerodynamically Air-Staged Burner and an Air-Staged
Precombustor Burner," Journal of the Institute of Energy, December
1989, pp 237-245.
Stankevics, J. 0. A., Mattsson, A. C. J., and Stickler, D. B. (1983)
"Toroidal Flow Pulverized Coal-Fired MHD Combustor," Third Coal
Technology Europe Conference, Amsterdam, The Netherlands.
6B-25
-------
STAGE
PRIMARY
ZONE
STAGE I I I
LEAN BURN/
DILUTION ZONE
CENTERBODY
STAGE I I
IMPACT SEPARATOR
CYCLONE SEPARATOR
ORIFICE PLATE
—> TO STACK
Figure 1 Schematic diagram of three stage slagging combustor
concept including slagging cyclone separator
10000
NOx (ppm)
NSPS Bituminous Limit
NSPS Subbituminous Limit
1000 r
100
Equilibrium NOx at AFT
AFT 100 K
AFT 200 K
10
1.1 1.2 1.3 1.4
Fuel-Air Equivalence Ratio
1.5
1.6
Figure 2 Calculated thermochemical equilibrium NOX
concentrations in the fuel-rich zone as a function of
fuel-air equivalence ratio for three gas temperatures:
adiabatic flame temperature (AFT), 100 K below AFT, and
200 K below AFT
6B-26
-------
Species Mass Fraction
NO Concentration (ppm)
0.5
1.5
Time (msec)
Figure 3 Variation of species concentrations showing final CO
burnout and NO generation in lean burn zone at a fuel-
air equivalence ratio of 0.6
Figure 4 Photograph of subscale slagging combustor test
arrangement
6B-27
-------
Carbon Conversion (%)
PC L/D - 1.25
CWM L/D - 1.50
CWM L/D • 1.25
80
0.9
1.1 1.2 1.3 1.4 1.5 1.6
Primary Zone Fuel-Air Equivalence Ratio
1.7
1.8
Figure 5 Measured primary zone carbon conversion for PC and CWM
firing as a function of fuel-air equivalence ratio
Measured Flame Temperature (K)
zouu
2500
2400
.. ^Tn^
D Pittsburgh #8
^ Wyoming Rosebud
0 Dorchester
1600
0.6 0.7 0.8 0.9 1 1.1 1.2 1.3 1.4 1.5
Primary Zone Fuel-Air Equivalence Ratio
1.6
1.7
Figure 6 Measured primary zone flame temperatures for PC firing
as a function of fuel-air equivalence ratio
6B-28
-------
NOx (ppm)
10000=
NSPS Bituminous Limit
NSPS Subbituminous Limit
100k
10 c
Equilibrium NOx at AFT
AFT 100 K
AFT 200 K
-X- Measured NOx (PC)
1.1 1.2 1.3 1.4
Fuel-Air Equivalence Ratio
1.5
1.6
Figure 7 Measurements of NOX concentrations at exit of primary
zone for PC firing and calculated equilibrium NOX
concentrations for PC combustion as a function of fuel-
air equivalence ratio
10000
NOx (ppm)
1000 =
100
NSPS Bituminous Limit
NSPS Subbituminous Limit
Equilibrium NOx at AFT
AFT 100 K
AFT 200 K
1.1
1.2 1.3 1.4
Fuel-Air Equivalence Ratio
1.5
1.6
Figure 8 Measurements of NOX concentrations at exit of primary
zone for CWM firing and calculated equilibrium NOX
concentrations for CWM combustion as a function of
fuel-air equivalence ratio
6B-29
-------
600
500
400
300
200^
NOx (ppmv, dry, corrected to 15% O2)
100
0 PC L/D - 1.25
*- CWM L/D • 1.50
NSPS Limit
o —
0.8 0.9
1 1.1 1.2 1.3 1.4 1.5 1.6
Primary Zone Fuel-Air Equivalence Ratio
1.7
1.8
Figure 9 Measured lean-zone exit NOX concentrations (dry,
corrected to 15 percent oxygen) for PC and CWM burning
as a function of primary zone fuel-air equivalence
ratio
100
80
60
40
NOx (ppmv, dry, corrected to 15% O2)
A
20
0
1500
1600 1700 1800 1900 2000
Combustor Outlet Temperature (deg F)
2100
Figure 10 Measured lean-zone exit NOX concentrations (dry,
corrected to 15 percent oxygen) as a function of lean-
zone outlet temperature
6B-30
-------
LOW NOX COAL BURNER
DEVELOPMENT AND APPLICATION
J. W. Allen
NEI-International Combustion Ltd
Sinfin Lane,
Derby, England DE2 9GJ
-------
LOW NOX COAL BURNER
DEVELOPMENT AND APPLICATION
ABSTRACT
The paper describes the development and application of a front wall low NO coal
A
burner in the U.K. power industry.
Target NO emission levels set by European Community Directives, for the U.K.
X
industry, were met both in full scale single burner thermal trials and in the
multi burner boiler operation.
The paper highlights the basic differences between test rig and boiler
installations, not only in combustion performance but also in the boiler
operational effects which influence the selection of materials of construction
for the critical burner parts.
In order to optimise the boiler performance, the characteristics of the low NO
A
burner must be recognised in the boiler operating procedures.
6B-33
-------
INTRODUCTION
Current UK NO emission targets for large combustion plant (i.e. plant with heat
X
input greater than 50 MW thermal., are based on a European Economic Community
(EEC) Directive (88/609/EEC) issued in December, 1988 (1). The Directive
stipulates limits for new large plant and also NO reduction targets to be
X
achieved by the various EEC countries over the decade to 1998. NO limits for
X
the various fossil fuels are given in Table 1.
Table 1
EMISSION LIMIT VALUES FOR NO FOR NEW PLANTS
X
Type of Fuel Limit Values (mg/NmJ)
Sol id in General 650
Solid with less than 10% volatiles 1300
Liquid 450
Gaseous 350
Although these NO levels refer to new plant they have become target norms for
X
the retrofitting of power generation boilers in the UK for low NO operation.
Furthermore the UK is required to reduce NO emission levels by 15% prior to
X
1993 and 301 prior to 1998, based on NO emission levels in 1980.
X
European units for NO concentrations are frequently quoted in mg/Nm3, although
X
most concentration measurements are made in terms of parts per million (ppm).
For comparison purposes the ppm concentration is referred to either a 3% or 6%
dry waste gas oxygen concentration. Table 2 gives the interconversion factors
for terms commonly used for the expression of NO concentrations.
X
6B-34
-------
Table 2
INTERCONVERSION OF NOX CONCENTRATION TERMS
To convert To Multiply by
From >
D6
mg/Nm3 - 0.487 8.14 x 10
mg/Nm3 ppm lbs/10D Btu
-4
-3
ppm 2.05 - 1.67 x 10
lbs/106 Btu 1230 598
Table 2 is based on coal combustion with a dry flue gas 02 content of 6%. To
correct
used:-
correct NO concentrations, at differing 0? levels, the following formula can be
A
N0y (ppm at 02 n J . 21 p_2 m NOY (ppm at 02 m )
A \ I f \ L- ) A \ L. I
21 02(1)
Prior to the privatisation of the electricity industry in the UK the CEGB
announced a E170M programme in order to achieve the reductions in NO emission
X
levels as required by the EEC Directive. The two major privatised power
generators, National Power and PowerGen, are continuing with this programme.
Progress in the conversion of corner fired units in the UK has proceeded quickly
following the successful demonstration of the 'Low NO Corner Firing System
A
(LNCFS)1 installed in a single 500 MW boiler in the CEGB, North Western Region,
in 1985(2),(3). The 500 MW+ corner firing capacity of both National Power and
PowerGen is committed to this low NO system.
X
Conversion of the wall fired units has proceeded more slowly, at the time of
writing around 25-30% of the UK wall fired coal capacity has been converted or
committed to low NO burner retrofit. This slower progress has enabled the
X
power generating and manufacturing organisations to proceed via a well defined
programme based on isothermal and mathematical modelling, single burner full
scale rig testing and the testing of individual burners within an actual boiler
environment, before commencing a full boiler commercial retrofit. All the low
NO burner developments, including corner firing, have been based on combustion
X
staging techniques, which have been demonstrated as capable of achieving the NO
X
6B-35
-------
reduction requirements of the EEC Directive. The burner development and
operations described in this paper relate to a front wall low NO coal burner
X
incorporating both fuel and air staging into the basic design. Although these
burners are capable of meeting the NO emission requirements up to 1998 it is
X
anticipated that a tightening of the regulations within the EEC will occur
before that date. Improved internal staging, furnace staging and, perhaps, post
combustion No reduction techniques will have to be introduced to meet these
X
more strict emission limits.
If post combustion reduction techniques are eventually required, an accepted
basic low NO burner system will enable any future emission regulations to be
X
met effectively both in terms of speed of implementation and minimum capital
cost.
PRINCIPLES OF BURNER DESIGN AND DEVELOPMENT
The current NEI-ICL low NO wall burner design is shown in Figure 1. Air
X
staging is achieved by splitting the combustion air into independently swirled
secondary and tertiary streams. Fuel staging is achieved by means of fuel flow
redistributors (FFR) located in the pulverised coal/primary air stream close to
the burner exit. Situated in this location the FFR produce a fuel lean/fuel
rich profile at the burner mouth. Ignition of the main pulverised coal fuel
(PF) is achieved via a centrally located oil burner with its integral combustion
air supply fan. PF is supplied, from the PF supply piping, via a tangential
inlet and scroll distribution system to the annular burner fuel duct. The
design concepts were developed using isothermal modelling techniques, to examine
both the flow of fuel and the air distribution within the burner system. Fuel
flow work addressed the problem of roping within the burner fuel annulus and
produced an evenly distributed flow into the FFR system which then produced the
required fuel staging effect at the burner exit. Various forms of FFR devices
were tested using flow visualisation techniques. Air distribution and air swirl
were studied in relation to the recirculation and general mixing patterns
produced both in the near burner region and further downstream. Figure 2
illustrates a typical recirculation pattern from an early burner design.
Following the isothermal model work a series of potential low NO burner design
X
configurations were selected for thermal testing, at full scale, in the 88 MW
NEI-ICL burner test facility. The initial full scale tests related to a 37 MW.,
burner design which would be required for several 48 burner 500 MW front wall
6B-36
-------
coal fired units in the UK. During this work, operating parameters, such as
those relating combustion air preheat and heat input to NO levels, were
A
established (Figures 3a and 3b). In this work the principle of good flame
retention at the burner mouth, as a pre-requisite of low NO operation, was also
A
established. This and the effectiveness of the FFR, in controlling overall NO
X
emission levels, is illustrated in Figures 4a and 4b.
In order to relate the test rig burner performance to site boiler performance,
particularly with respect to NO emissions, the test rig was refractory lined in
X
a pattern determined by computer calculations, such that the rig centre line
temperature was similar to the boiler centre line flame temperature, as shown in
Figure 5 (4). To demonstrate the effectiveness of this approach a standard
burner from a 500 MW boiler was rig tested under these conditions and did
reproduce site NO levels of around 700 ppm at 3% Qz . Thus a 1:1 rig factor in
X
respect of NO emission levels was established.
X
Further work was carried out on flame retention, which resulted in successful
patent applications for the burner design(5) and also up-rating of the design
from 37 MW., to 58 MW., without an increase in NO emissions. The 58 MW burner
was also required to operate with a primary air to pulverised coal ratio of
1.2:1 compared to the more usual 1.5/2:1 range. Furthermore the primary air was
vitiated by the use of recycled flue gas into the ball mills for coal drying
purposes. This primary air vitiation and low pa:pf could aid low NO
X
performance of the burner but also adversely affect flame stability and burn
out.
Figure 6 demonstrates the NO performance of this larger burner showing not only
X
the usual trend of increasing NO with waste gas Oz content (with a NO level of
X X
375 ppm at 3% 02), but also that the burner can operate at lower excess air
rates than normally used for coal firing without the generation of high CO
levels. Corresponding with CO levels below 100 ppm the carbon in dust levels
measured on the rig tests were a maximum of 2%. During the thermal trials the
opportunity was taken to collect in-flame gas samples and temperature
measurements. Contour plots of gas and temperature variations are shown in
Figures 7a-7d. These emphasise the importance of the near burner region
aerodynamics in establishing a centrally located reducing atmosphere with the
flame envelope which encourages the formation of Ha rather than NO from the
X
nitrogen contained in the fuel. High NO levels were produced in the outer
X
regions of the flame, close to the burner, corresponding with the mixing of
6B-37
-------
secondary air and the outer layers of the fuel stream. This NO mixed later in
A
the flame with the reductants produced in the flame core, thus producing a low
overall NO emission from the flame.
X
Depending on the particular conditions rig NO levels were in the 300-400 ppm
A
range (related to 3% Oz, dry) which represents an approximate 50% reduction in
NO .
PERFORMANCE OF BURNERS IN SITE INSTALLATIONS
Prior to the possible retrofitting of a full boiler set of low NO burners it
A
was considered prudent to replace just one or two standard burners, with the low
NO designs, in an operating boiler. This preliminary installation would enable
the compatibility of the low NO burners, within a hot multi-burner furnace
X
environment, to be assessed from an operational and durability standpoint. Two
37 MW. . low NO burners were installed, on a 48 burner 500 MW boiler, in the
L n x
wing and centre top row locations and a single 58 MW., burner installed in the
centre top row position of a 32 burner 500 MW boiler. The centre top row
location was considered to give the most hostile conditions regarding burner
component temperatures, particularly in the non-firing mode. The wing position
enabled a qualitative assessment of the burner, in operation, to be made. The
centre top row burners were inspected, in-situ, using a water cooled periscope
inserted into the burner via de-ashing ports, critical components of the burner
were instrumented with thermocouples to provide burner metal temperature
variations in both the firing and non-firing operational modes.
Temperatures recorded from the single low NO burner, installed in the standard
X
burnered furnaces, gave some cause for concern, as in the non-firing mode,
temperatures approaching recommended limits for the material used in the
critical burner areas were recorded, with the normal 10-15% MCR cooling air
equivalent passing through the burner.
Computer calculations of heat flux based on test rig data, of low NO burner
operation, showed that with a full boiler set of low NO burners the temperature
X
of the critical burner components would be satisfactory. The main reason for
this was the lower peak flame temperature of the combustion staged low NO
burner system which also occurred further down stream from the burner exit.
There was also a change in the gas recircul ation pattern at the furnace front
wall as a result of the low NO burner design.
x
6B-38
-------
Periscope observations indicated the possibility of some ash deposition in the
low NO burner installed in a conventional boiler burner system. From both test
X
rig experience and computer predictions it was postulated that the change in
front wall flow patterns from a full boiler set of low NO burners would
X
eliminate this possibility.
Although both rig operating experience and computer predictions indicated that
neither high material temperature or ash deposition would be a problem with a
full set of low NO burners, material specifications for the critical burner
X
components were selected and a minor modification made to the secondary air
stream aerodynamics to provide further assurance. In practice, with the full
boiler set of low NO burners, the computer and test rig predictions, regarding
X
critical burner metal temperatures and ash deposition, were verified. By
carrying out these investigations a considerable data bank was compiled on
potential materials for burner construction covering fabricated, cast materials,
coated materials and ceramics. Data on erosion resistance of these materials
exposed to flowing pulverised coal streams were also obtained. Table 3 compares
the temperatures measured in the single low NO burner and the multiple low NO
X X
burners after the boiler modification.
Table 3
BURNER METAL TEMPERATURES COMPARISONS
BEFORE AND AFTER LOW NO BOILER MODIFICATIONS
A
Burner Component Temperature °C
Before Modification After Modification
mean peak mean peak
Tertiary Air Duct 880 980 868 1011
Secondary Air Duct 870 950 838 915
Oil burner core tube 730 810 707 792
Temperatures in Table 3 relate to the non-firing mode with 10-15% of normal
firing air supply passing through the burner.
6B-39
-------
Also, prior to the installation of a full boiler set of low NO burners, the NO
X A
and CO levels were measured on an unmodified boiler (6). The results are shown
in Figure 8. In general the unmodified boiler NO emissions were in the range
X
550-730 ppm (related to 3% Qz , dry), depending upon the excess air level, with a
mean level of 633 ppm at 3% 02. Thus a 30% reduction in NO would require the
X
boiler to operate at a mean figure of 443 ppm well within the capacity of the
burner, from the rig test data (see Figure 6). Carbon in dust from the
unmodified boiler was in the range 0-6% 3.3% (mean 1.93%) depending upon mill
groups in operation and excess air levels, under similar conditions CO levels
were recorded in the 60-200 ppm range.
Figure 9 shows the results from the initial commissioning trials of the full
boiler set of low NO burners, covering the whole range of mill groups and
X
excess air levels, equivalent to the 2-5% waste gas 02 range and compares them
wit te test rig burer performace. Summarising these early results from te
boiler, the low NO burners, in combination, can operate under the conditions
X
outlined in Table 4.
Table 4
INITIAL COMPLIANT OPERATING RANGE
OF LOW NO BURNERS
A
Oz level
3% 4%
NO ppm 330 430
CO ppm 25 10
C in Dust % 52
NO levels in Table 4 refer to ppm at 3% 02 dry.
X
The results confirm the 1:1 rig factor to boiler factor relating to NO
X
emissions, in the 3-4% waste gas 02 range. The CO emission results in Figure 9
indicate that the 100 ppm CO level would not be exceeded until excess air levels
equivalent to 1.8% 02 were obtained, this compares to 2.6% 02 in the unmodified
boiler. Over the 3-4% waste gas 02 range the CO levels in the boiler were
similar to those in the rig tests, however there is a tendency for a more rapid
increase in CO generation, below 3% Qz , in the boiler compared to the test rig.
6B-40
-------
The average of all the boiler NO level results gave 399 ppm NO which
X X
corresponds to a 37% reduction in NO compared to the mean level of NO from the
X X
unmodified boiler. This reduction should be even greater when burner
optimisation is complete to enable the burners to operate at lower Oz levels
without excessive CO generation in the boiler. Carbon in dust levels increased
in the low NO burnered boiler to an average of around 5% (at 3% 02) compared to
X
2% in the unmodified boiler (see Figures 8&9). The general practice with this
boiler is to over-fire on the bottom rows of burners in the unmodified boilers,
as a means of controlling superheater temperatures and this practice has been
continued on the modified boiler. Some burners are therefore operating at lower
overall air to fuel ratios, however, the increased swirl and hence shorter flame
length of the unmodified burners produces sufficient in furnace time and
turbulence to produce a low C in dust loss overall.
As a result of staged combustion effects low NO burners have a low overall
X
swirl producing increased flame lengths and low furnace turbulence levels. We
now know that higher carbon in dust levels are generated from the burners which
are operating at lower overall air levels. The time, temperature and mixing
history (Oz availability), which controls the combustion reactions within the
boiler, including NO emissions is influenced by furnace geometry and air
X
quality. The 10 m depth (with an approximate 3:1 width:depth ratio) of the
boiler coupled with the use of vitiated air for coal conveying have an adverse
effect on the final burn-out characteristics. Optimisation of the boiler and
burner performance, fully recognising the low swirl characteristics of the low
NOx burners, should improve this situation.
CONCLUSION
Single full scale burner test facilities can be used to indicate multi-burnered
boiler NO emission levels. Combustion staged low NO burner designs are
X X
capable of meeting current legislation relating to NO emission levels.
X
Front wall environments are less hostile to burner components in a low NO
X
system compared to a conventional front wall coal burner system.
Low NO burner characteristics must be fully recognised in the optimisation of
X
low NO front wall burner boiler operations.
X
6B-41
-------
ACKNOWLEDGEMENTS
Thanks are due to the Directors of NEI-ICL for permission to publish this paper
and to many colleagues within NEI-ICL responsible for providing both test rig
and boiler commissioning data.
Thanks are also due to PowerGen Technical and Station personnel for the
provision of boiler operating data and continued enthusiastic interest in the
project.
REFERENCES
1. Official Journal of the European Communities L336 "Council Directive
88/609/EEC of 24th November, 1988 on the Limitation of Emissions of
Certain Pollutants into the Air from Large Combustion Plants"
7th December, 1988.
2. J. W. Allen, W. J. D. Brooks, N. A. Burdett, F. Clarke and G. Foley.
"Reductions in NO Emissions from a 500 MW Corner Fired Boiler." Joint
Symposium on Stationary NO Combustion Control. New Orleans, 1987.
X
3. J. W. Allen "NO Reductions in Coal fired Boilers." Modern Power
X
Systems. June, 1987.
4. Private Communications. M. J. Sargeant, S. Cooper - CEGB, Marchwood
Engineering Laboratories, 1984.
5. UK Patent 8805208
USA Patent 317743
European Patent 89302101.4
6. Private Communication. CEGB
6B-42
-------
Secondary air
control damper
Secondary air
swirl vanes
Tertiary
air vanes
Outer back plate
Sight tube
Conical liner
Core air tube
PA/PF inlet
Secondary
air tube
Tertiary
air tube
Entry chamber
Rodding tube
Support tubes
Secondary/tertiary
air shut off damper
Fuel flow
redistributors
Figure 1. Low NOx Front Wall Coal Burner.
Axial distance (m)
1.5-1
Burner centre line
1.0-
Flame
boundary
0.5-
Central
recirculation
zone
Figure 2. Low NOx Coal Burner Model.
Typical Recirculation Pattern.
6B-43
-------
NOx (ppm)
500 -i
400 -
300
200 -
100
0
0 100 200
Air preheat temp. (°C)
3a Effect of Air Preheat on
NOx (Excess Air = 3% O2 )
300
NOx (ppm)
500 -i
400 -
300 -
200
100
0
I
50%
Burner load
100%
3b Effect of Burner Load on
NOx (Excess Air = 3% 02 )
100% Load = 58MW.
Figure 3. Effect of Air Preheat and Burner Load on NOx.
NOx (ppm)
700 -i
600 -
500 -
400 -
300 -
200 -
100 -
0
Fully lifted
flame
Well anchored
flame
01 2345
% 02 in waste gas
4a Effect of Flame Retention on NOx
NOx (ppm)
700 -i
600 -
500 -
400 -
300 -
200 -
100 -
0
Burner
without
FFR ^-
Burner
with FFR
0
Figure 4.
12345
% 02 in waste gas
4b Effect of Fuel Staging on NOx
Effect of Burner Parameters on NOx
6B-44
-------
Centre cell
gas temp. (°K)
2000 -i
1750-
1500 -
1250-
1000
750 -
500
Test rig
Boiler
I
10
I
12
l
14
16 18
I
20
Axial distance (m)
Figure 5. Comparison of Refractory Lined Rig and
500MW Boiler Centre Line Temperatures.
NOx (ppm)
500 -i
400 -
300 -
200 -
100 -
CO (ppm)
NOx
CO
-100
-80
-60
-40
20
0
% O2 in waste gas
Figure 6. Test Rig Performance of 58MW (Thermal)
Front Wall Coal Burner.
6B-45
-------
CD
DO
I
-P-
O)
O-i
1 -
2-
3-
0 2
O-i
Burner centre line
280
250
I
12
4 6 8 10 12 14
Distance along axis (m)
7a NOx Contours (ppm)
Burner centre line
I
16
0 2 4 6 8 1012
Distance along axis (m)
7c CO Contours (%)
\
14
16 1!
O-i
Burner centre line
1
10
2.5
O-i
1 -
4 6 8 10 12 14
Distance along axis (m)
7b 02 Contours (%)
Burner centre line
! 6 1!
900
900
800
I
8
I
1 A.
4 6 8 1012
Distance along axis (m)
7d Temperature Contours (°C)
16 18
Figure 7. In —Flame Gas and Temperature Contours.
-------
NOx (ppm)
700 -i
600 -
500 -
NOx
400
I
2
CO (ppm)
100 -i
80 -
60 -
40 -
20 -
0
CO
o
% c
8 -
6 -
4 -
2 -
I
3
% 02 at economiser
Unburnt Carbon
i
4
I
2
% 02 at economiser
Figure 8. Unmodified Boiler Performance
6B-47
-------
NOx (ppm)
600 -i
500 -
400 -
300 -
200 -
100 -
0
KEY
NOx
o Boiler
x Test rig
0
I
5
CO (oom)
80 -i
60 -
40 -
20 -
CO
o
% c
10-,
6 -
4 -
2 -
Unburnt Carbon
' ' 1 1 i
01 2345
% 02 at economiser
Rgure 9. Modified Burner Performance on Boiler
During Commissioning, Compared to Single Burner
Test Rig Performance
6B-48
-------
Session 7A
NEW DEVELOPMENTS I
Chair: G. Veerkamp, Pacific Gas & Electric
-------
Preliminary Test Results
High Energy Urea Injection DeNOx on a 215 Mw Utility Boiler
Dale G. Jones, Ph.D., P.E., Noell, Inc.
Stefan Negrea, P.E., Noell, Inc.
Ben Dutton, Noell, Inc.
Larry W. Johnson, P.E., Southern Calif. Edison Co.
J. Paul Sutherland, P.E., Southern Calif. Edison Co.
Jeff Tormey, Southern Calif. Edison Co.
Randall A. Smith, Fossil Energy Research Corporation
-------
Preliminary Test Results
High Energy Urea Injection DeNOx on a 215 MW UlUlly Boiler
by
Dale G. Jones, Ph.D., P.E., Noell, Inc.
Stefan Negrea, P.E., Noell, Inc.
Ben Dutton, Noell, Inc.
Larry W. Johnson, P.E., Southern Calif. Edison Co.
J. Paul Sutherland, P.E., Southern Calif. Edison Co.
Jeff Tormey, Southern Calif. Edison Co.
Randall A. Smith, Fossil Energy Research Corporation
ABSTRACT
Initial tests of a high energy urea injection SNCR DeNOx system have been
completed at Southern California Edison's Huntington Beach Unit 2 gas- and
oil-fired boiler. The SNCR DeNOx temperature window in this 215 MW utility
boiler occurs in narrow cavities and between boiler convection sections. The
Huntington Beach SNCR DeNOx project Is a demonstration of high energy
urea injection in narrow cavities to evaluate various DeNOx alternatives and
to bring such installations in compliance with South Coast Air Quality
Management District regulations for the metropolitan area.
Following contract award in June, 1990, Noell proceeded with injection system
design, installation and start up. Initial tests of high energy injection into the
2nd cavity and other boiler zones were conducted between Jan. 15 and March
5, 1991. Pressurized urea-water mixtures were Injected into cross-flowing flue
gas using high velocity air-driven nozzles. Initial 2nd cavity injection tests
showed that 25% to 40% DeNOx Is achieved at full load despite adverse
conditions of short cavity residence times (i.e. 40 milliseconds) and floor-to-
roof adverse temperature gradients (l.e. about 200 F). Such adverse conditions
in the 2nd cavity also caused unacceptably high levels of NH3 slip.
Additional tests were therefore performed to investigate urea injection into the
1st cavity where the full load temperature is about 2050 F. Using only four (4)
sldewall Injection nozzles, 20% to 25% full load DeNOx was obtained at urea
feedrates from NSR = 2 to NSR = 4 (NSR is moles of NHi injected vs. moles of
Initial NOx). Under these conditions, NHs slip measured upstream from the air
preheater averaged less than 3 ppm, or less than about 1.5% of NHi feedrate,
Noell Is proceeding with further development of advanced injection systems to
be considered for installation and additional testing at Huntington Beach.
7A-1
-------
1.0 Introduction and Background
Initial tests of a high energy urea injection SNCR DeNOx system have been
completed at Southern California Edison's Huntington Beach Unit 2 gas- and
oil-fired boiler. The SNCR DeNOx temperature window in this 215 MW utility
boiler occurs in narrow cavities and between boiler convection sections. The
Huntington Beach SNCR DeNOx project is a demonstration of high energy
urea injection in narrow cavities to evaluate various DeNOx alternatives to
comply with South Coast Air Quality Management District regulations.
Urea (NH2.CO.NH2) reacts at high temperatures with NOx in combustion flue
gases, approximately as follows:
2 NO + NH2.CO.NH2 + 0.5 O2 = 2 N2 + 2 H2O + CO2
Amine radical (NH2) resulting from thermal decomposition of the urea reacts
with NO. The chemical feedrate vs. quantity of NOx is called the normalized
stoichlometric ratio (NSR), defined as the molar ratio of NHi being injected
divided by initial NOx. At Isothermal conditions, the SNCR DeNOx process
operates best over a narrow 'temperature window' between 1600 F and 1900 F.
If the flue gas temperature Is too hot, some of the NH2 radicals form additional
NOx and DeNOx performance decreases. If the flue gas temperature Is too
cold, some of the NH2 radicals form byproduct NH3, called 'ammonia slip* and
DeNOx performance goes down. Thus, a 'temperature window* exists.
This narrow temperature window is the primary drawback of boiler Injection
SNCR DeNOx technology. When boiler operations change, temperatures at an
injection location also change. Therefore, multiple levels of Injection are
usually required to provide good DeNOx performance over a range of boiler
conditions. At low load, the temperature may be too cold, and Injection
should occur at a location closer to the furnace. At high load, the temperature
may be too hot, and Injection should be at a location further from the furnace.
Noell's boiler injection DeNOx system uses high velocity Injection Jets to
provide Intense flue gas mixing. These Jets can overcome distribution problems
typically observed, such as non-uniformities In temperature, flowrate, and/or
composition of the flue gas. As In any chemical process, intimate and complete
mixing is Important. By proper design and operation of the injection system,
close approximation to a well-mixed reactor can be achieved. Noell's boiler
Injection Jets are used for flue gas mixing and operate Independently from
chemical feeding, accomplished using feed pumps for higher or lower levels of
DeNOx. Chemical distribution occurs first In the Injection Jet, and then as the
injection jet(s) mix with cross-flowing flue gas. Noell's boiler injection concept
is Illustrated in Figure 1, which provides results of Jan, 1988 Injection system
flow model testing for the KVA/Basel MSW incineration plant. The left picture
shows 'channelling1, where a smoke stream passes through the flow path
without much mixing. The right picture is similar except that scaled-down
injection Jets were installed Into the sidewall(s) of the flow model to determine
effects on mixing. As can be seen, such high energy injection Jets have a major
Impact on flue gas mixing. Similar full-size Injection Jets were subsequently
installed in the 330 TPD Basel MSW Incinerator. At maximum boiler output at
330 TPD incinerator feed rate, NOx removal of 70% was obtained at urea NSR
= 1.3, along with relatively low levels of NH3 slip. (Reference 1).
7A-2
-------
Figure 1: Photographs of Flow Model Test Results
KVA Basel 330 TPD MSW Incineration Furnace
January, 1988
"Channelling" Effect
(left-hand picture)
Injection Jet Effect
(right-hand picture)
7A-3
-------
Noell has also installed its high energy boiler injection SNCR DeNOx process
at the 325 MW coal-fired power plant of BKB/Offleben in Germany, which was
started up for commercial operation in Sept. 1989. In this coal-fired boiler.
Noell's steam-driven nozzles are used for urea injection to achieve 95 ppm NOx
at full load. At full load, the urea NSR is about 0.64, corresponding to about
32% DeNOx with NH3 slip of less than 1.0 ppm. Due to the SO2 content of the
flue gas, the Offleben requirement is less than 5.0 ppm NH3 slip to avoid
forming ammonium bisulfate deposits in the air preheater. (Reference 2)
In more recent developments, Noell has been awarded a contract by the Public
Service Company of Colorado (PSCC) to design and procure boiler injection
SNCR DeNOx equipment for a Clean Coal III project at PSCC's Arapahoe coal-
fired station. This boiler injection SNCR DeNOx project is being co-sponsored
by the U.S. DOE and by EPRI. Noell has also been awarded a contract by the
Tennessee Valley Authority (TVA) to conduct perform field testing of flue gas
temperatures, and conduct boiler flow model testing of injection system
options for a project being considered by TVA to demonstrate boiler injection
SNCR DeNOx at a large coal-fired power plant
2.0 Description of Huntlngton Beach Unit 2 Boiler
This gas- and oil-fired 215 MW boiler incorporates a pressurized furnace with
front wall-fired burners arranged 6 wide by 4 high. The drum-type natural
circulation steam generator includes pendant secondary superheater and
reheat superheater convection sections. It is In the area of these pendant
sections that flue gas temperatures at full load on gas fuel reach levels of
interest for SNCR DeNOx. Full load superheater outlet conditions are
1,560,000 Ib/hr at 2450 psig and 1050 F. Flue gas from the furnace passes
horizontally through the secondary superheater, a water screen formed by the
rear wall tubes of the furnace, the reheater, and the pendant loop portion of
the primary superheater. Following the rear cavity, the flue gas then passes
vertically downward through the balance of the convection sections, air
preheater and stack. Flue gas recirculatlon fans are provided for accurate
control of superheated steam temperatures. At full load on gas fuel, about 8%
of the flue gas is recirculated to the furnace bottom hopper. A side sectional
elevation of the boiler is shown in Figure 2. The furnace cross section In the
vertical upflow direction is 24 ft wide oy 50 ft. deep.
Detailed description of the boiler convection sections goes beyond the scope of
this report. It is sufficient to say that the flue gas velocities at full load on
gas fuel are such that the residence times in the 1st and 2nd cavities between
convection sections are on the order of 40 milliseconds (msec), and that flue
gas temperatures initially decrease at a rate of about 4 F/msec in the first
pendant section, and then at a rate of about 2 F/msec In the second and third
sections. These narrow cavities and very short residence times are typical for
many gas- and oil-fired boilers, and offer perhaps the most difficult type of
challenge for application of boiler injection SNCR DeNOx. An earlier
publication by Mittelbach, et.al. indicates that at 1800 F or above, flue gas
residence times of about 100 msec would be sufficient to complete most of the
SNCR DeNOx reactions (Reference 3). In the case of the Huntlngton Beach
Unit 2 boiler, this expectation was overly optimistic.
7A-4
-------
Figure 2: Side Sectional Elevation, Huntlngton Beach Unit 2 Boiler
Southern California Edison Company
StCONOARYl UREHEAT
SUPERHEATER SUPERHEATER
7A-5
-------
3.0 Flue Gas Temperatures
Prior to design of the injection system, flue gas temperature data was obtained
using HVT probes at the upper furnace front and side-wall observation doors,
and oy acoustic pyrometer to obtain average flue gas temperature at the Inlet
of the first pendant tube section. The various field measurements of flue gas
temperatures were compared with boiler manufacturer design data as follows:
Table 3.1 COMPARISON OF FLUE GAS TEMPERATURES
Huntlngton Beach Unit 2 at Full Load (Gas Fuel)
Source of Data SSH Inlet 1st Cavity 2nd Cavity
HVT Probe @ Observation Doors 2230 F n/a n/a
Acoustic Pyrometer @ Obs. Doors 2280 F n/a n/a
HVT Probe @ Manway Doors n/a 1910 F (?) 1760 F
Manufacturer Design Sheets 2340 F n/a 1775 F
The field data seemed to be in reasonable agreement with boiler manufacturer
data. Computer-generated prediction of 2nd cavity temperature contours (full
load on gas fuel) were also provided by the boiler manufacturer, which
indicated cooler zones averaging 1700-1800 F near the 2nd cavity floor, hotter
zones of about 1850-1950 F in the middle, and then 1800 F or above nearly all
the way to the 2nd cavity roof. Based on the foregoing, there was no reason
to doubt that the 2nd cavity was the preferred Injection zone. The 2nd cavity
measures approximately 16 ft. high by 50 ft. wide In cross-section.
Following Installation of the 2nd cavity Injection nozzles, further data was
obtained. Temperature profiles from HVT measurements In the 2nd cavity are
provided In Figures 3 and 4, where the strong Influence of burner patterns
under otherwise Identical operating conditions Is easily seen. Burner pattern
adjustment caused average flue gas temperatures to Increase (or decrease) up
to 100-150 F. The entire SNCR DeNOx temperature window Is only 300 F, and
changes of 100-150 F are quite significant As seen In Figures 3 and 4, flue
gas temperatures also decreased up to 200 F from the floor to the roof. This
adverse temperature gradient substantially shortened the 2nd cavity Injection
residence times within the 1600-1900 F SNCR DeNOx temperature window.
4.0 Description of 2nd Cavity Injection System
The Initial full load NOx concentration was generally about 120 ppm (corr. 3%
O2, dry). Except as noted, this Initial NOx was used for NSR calculations.
Tube shields were designed and Installed by Noell on the first row of boiler
tube at the downstream edge of the 2nd cavity. Discussion between Southern
California Edison and Noell confirmed that tube shields would provide a way
to evaluate effects of high velocity Injection Jets on metal thicknesses, without
any metal loss on the boiler tubes themselves. In coal-fired applications of
high energy boiler Injection for SNCR DeNOx, Noell generally recommends the
use of tube shields so that the potential for Increased erosion In specific high
velocity zones can be determined without risk to the boiler tubes themselves.
7A-6
-------
Figure 3: 2nd Cavity Flue Gas Temperatures Near Boiler Centerline
Huntlngton Beach Unit 2 Boiler. Full Load, Gas Fuel
Southern California Edison Company
HVT Temperature (F)
r"
\
A-
S**
\
>
\\
\
Ni
\
\
^
fv
\
rma
\
\.
oo
V
BO
V,
r^~
-— —
DS
X
AI
,-—
/
^
J*
S
^\
\
r~s
\
A
\\
\\
\
\
\
\\
i
i
4 t a 10 12 14 16
Height Above Floor of 2nd Cavity, ft
HVT Temperature (Fl
*-
tf
N
^
—
N
>TE
/0(
Q
-
PII
S-E
,,oJ
n
mo
jt-
U
M
BOOS-We
i -v
B0(
. i
S-Taat
N
Tali
^v
'e«t
^
7
,
en a
X
/
/
/
/
,r
^
'S
, fro
~N
\
•*v^
n Si
k
^
\
\
ie«
^
\\
\
Jli
1
0 J 4 6 » 10 IS
Height Above Floor of 2nd Cavity, ft
Figure 4: 2nd Cavity Flue Gas Temperatures Near Boiler Walls
Huntlngton Beach Unit 2 Boiler. Full Load, Gas Fuel
Southern California Edison Company
7A-7
-------
5.0 Results of 2nd Cavity Injection Tests
System tests Involved selection of pump settings for controlling the urea-water
mixture ratio. The liquid mixture was then pumped to the boiler level and
injected Into the cross-flowing flue gas using air-driven nozzles operating at
sonic Jet velocities. A number of higli velocity Injection nozzles were installed
in the floor zone of the 2nd cavity. By means of aspirated ports, these nozzles
could be extended or retracted up to 8 ft. into the pressurized flue gas zone,
without influencing boiler operations. Two (2) air orifice sizes were tested, the
larger orifice(s) requiring an injection air flowrate of about 2.1% of the full
load flue gas flowrate, and the smaller orifice(s) requiring about 1.2%.
Figure 5 shows the effect of boiler load and burner pattern on percentage
DeNOx for 2nd cavity injection at NSR = 2 for the two (2) sizes of Injection
nozzles. As can be seen, the effect of increasing boiler load with ABIS (all
burners In service) was to increase the DeNOx performance. With normal
BOOS (burners out of service), increasing boiler load at a constant urea
feedrate for NSR = 2 at full load caused a decrease in DeNOx performance.
With the smaller nozzles, reduced DeNOx performance especially at full load
was partially caused by reduced flue gas mixing at higher flue gas velocities.
Figure 5 Illustrates the effect of adjusting the burner pattern from normal
BOOS to ABIS, which causes increased flue gas temperatures (Figure 3 & 4).
The increased flue gas temperatures, in turn, caused a full load DeNOx
performance Increase from 27% to 40%. Since the change In burner pattern
caused 2nd cavity flue gas temperature changes of 100-150 F, and since the
resulting DeNOx Increase (at otherwise identical conditions) was relatively
large, it was concluded that SNCR DeNOx in the 2nd cavity at full load was
operating at the colder edge of the 1600-1900 F temperature window. The
injected urea behaved as if the isothermal temperature was about 1600 F,
regardless that full load HVT temperatures in the 2nd cavity itself averaged
1720-1780 F. These Initial full load results up to 40% DeNOx were achieved
despite adverse conditions of short cavity residence time (i.e. 40 milliseconds)
and 2nd cavity floor-to-roof adverse temperature gradient (i.e. about 200 F).
Despite moderate DeNOx levels which were achieved, such adverse conditions
in the 2nd cavity caused unacceptably high levels of NHs slip.
Further analysis of these initial test program results showed that the hotter
1st cavity or upper furnace zones offered better locations at full load for high
energy SNCR DeNOx Injection than the 2nd cavity.
6.0 Tests of 2nd Cavity Injection Nozzle Supply Pressure
Additional tests were conducted using the larger 2nd cavity nozzles. In these
tests, the boiler was held at full load, and urea NSR feedrate was increased to
determine DeNOx vs. NSR. The results are presented In Figure 6, where it is
seen that with a lower nozzle pressure, the DeNOx cannot be Increased beyond
about 20% regardless how much the chemical feedrate Is Increased. This type
of response curve Is Indicative of relatively poor flue gas mixing, where the
SNCR DeNOx process become mixing limited. At the higher nozzle pressure,
there Is a continuing Increase in DeNOx performance up to about 37% as NSR
is increased up to about 5. This second type of response curve is Indicative of
relatively good flue gas mixing.
7A-8
-------
Figure 5: Effect of Boiler Operations on 2nd Cavity Injection DeNOx
Huntington Beach Unit 2 Boiler, Gas Fuel
Southern California Edison Company
Percent NOx Removal
lb-
10 •
15
JO
J5
5
0
NC
TE:
..
— -,
; — -~
1 — ~-
•
Nc
Full
Si
^
~~ —
rmaJ
Load
nail
j
— _
— ~-.
sod
NSR
^ozzl
Fixed Urea Feedrates for Normal
NSR = 2 at Full Load (215 Mtt
>-"-"
— — _
~~ —
)S
= 2
es
BOO
r)
BIS (
Largi
^
~.
~~-—~
5 Tes
SNS
; Noz
^
— —
~~~~.
ts
R = :
iles
— —
i^-
Nc
Full
Li
-i
— •— .
^
rma
-------
7.0 Results of 1st Cavity Injection Tests
Additional tests were performed to investigate 1st cavity injection at higher
full load temperatures, which averaged about 2050 F in the 1st cavity. This
was several hundred degrees Fahrenheit hotter than the average full load
temperature in the 2nd cavity. The existing 1st cavity sootblowers were
removed and air-driven nozzles were installed Into these existing membrane
wall aspirated ports. Using four (4) sidewall nozzles with known limitations
in flue gas cross-sectional coverage, 20% to 25% full load DeNOx was
obtained with urea feedrates from NSR = 2 to NSR = 4 (Figure 7). For these
same urea NSR feedrates and operating conditions. NH3 slip as measured
upstream from the air preheater was well below 1.5% of the NHi injection rate,
and averaged less than 3 ppm. Despite the very high 2050 F temperature, the
SNCR DeNOx process operated beyond expectations, especially considering the
relatively poor flue gas cross-sectional coverage and mixing afforded when
using only four (4) sidewall nozzles.
8.0 Results of Upper Furnace Injection Tests
Further tests were also performed to determine upper furnace injection DeNOx
as a function of boiler load. Again, only four (4) sidewall nozzles were used
where existing observation doors (aspirated) were available. The chemical
feedrate during these tests was maintained at a constant value which
provided NSR = 2 at full load conditions. As shown in Figure 8, the
percentage DeNOx decreased from a maximum of about 40% at a reduced load
of 120 Mw. At full load on gas fuel, the flue gas temperatures are about 2300
F at the inlet of the first boiler tube bank. This is too hot for SNCR DeNOx,
and as shown in Figure 8, the DeNOx decreased down to about 5% at full
load. NH3 slip characteristics are also shown in Figure 8, where it is seen that
at about 145 MW or 150 MW boiler load, upper furnace flue gas temperatures
are most favorable for optimum SNCR DeNOx performance.
9.0 Further Work In Progress
Noell is proceeding with further development of advanced injection systems to
be considered for installation and additional testing at Huntington Beach.
7 A-10
-------
Figure 6: Effect of Nozzle Pressure on 2nd Cavity Injection DeNOx
Huntington Beach Unit 2 Boiler, Full Load. Gas Fuel
Southern California Edison Company
Percent DeNOx from InlUaJ NOx .110 ppm
215 MW, Gas Fuel
40 •
13
P«'g
33
AK)»i«i (Off*.
30
55
20
10
2 3
Normalized Stolchlometrlc Ratio (NSR)
to
Figure 7: 1st Cavity Sidewall Injection DeNOx vs. NSR
Huntington Beach Unit 2 Boiler, Full Load, Gas Fuel
Southern California Edison Company
7A-11
-------
ro
CO
FV£NAc£ Uflf*
4 435. ZxfcKS, 6/fS FUEL
PI
:r3
^r?
r n C o
o 0 r*
o
p (T) f1
Cr P o
K.O p
o crp.
p
•-^ Tl
DO rt
o - "i
Bog
TJ P K
EJ K «
P >T1 _
D
fD
2
O
X
-------
Results and Conclusions
1. Narrow cavities and very short residence times In many gas- and oil-
fired boilers offer perhaps the most difficult challenges for application of
boiler Injection SNCR DeNOx.
2. Flue gas temperature variations caused by normal boiler operations can
and will have significant effects on boiler injection SNCR DeNOx,
even when there are no changes in boiler steam production or load.
Successful load-following SNCR DeNOx systems must have multiple
injection zones and relatively sophisticated controls.
3. Detailed field temperature measurements and flow model optimization
tests of injection Jets are considered prerequisites for the design of high
performance (boiler-specific) SNCR DeNOx injection systems
4. Despite adverse time/temperature conditions in narrow cavities between
adjacent convection sections in the Huntington Beach gas-fired boiler,
full load DeNOx performance was obtained as follows:
Injection Zone Nozzle Posltlon(s). DeNOx NH3 Slip
2nd Cavity Multiple Floor Nozzles 25%-40% high
1st Cavity Sldewall Nozzles (4) 20%-25% low < 3 ppm
Upper Furnace Sldewall Nozzles (4) 0%-5% zero
5. This initial Huntington Beach test program has shown that SNCR
DeNOx is a function of available DeNOx reaction time plus injection
system cross-sectional coverage and mixing. In this application at full
load with short residence times, injection into the 1st cavity at a flue
gas temperature of about 2050 F appears to provide the best SNCR
DeNOx results.
6. Noell is proceeding with further development of advanced injection
systems to be considered for installation and additional testing at
Huntington Beach.
References
1. Jones, D.G., et. al., 'Two-Stage DeNOx Process Test Data from
Switzerland's Largest Incineration Plant', EPA/EPRI Symposium
on Stationary Combustion NOx Control, San Francisco, California,
March 6-9. 1989.
2. Negrea, S., et. al., 'Urea Injection NOx Removal on a 325 MW Brown
Coal-Fired Electric Utility Boiler in West Germany', 52nd Annual
Meeting, American Power Conference, Hyatt Regency Chicago,
April 23-15, 1990.
3. Mlttelbach, G., et. al., 'Application of the SNCR Process to Cyclone
Firing', Special Meeting on NOx Emissions Reduction of the VGB,
German Power Industry Association, June 11-12, 1986.
7A-13
-------
EVALUATION OF THE ADA CONTINUOUS AMMONIA SLIP MONITOR
Michael D. Durham, Richard J. Schlager, Mark R. Burkhardt,
Francis J. Sagan and Gary L Anderson
ADA Technologies, Inc.
304 Inverness Way South, Suite 110
Englewood, CO 80112
-------
EVALUATION OF THE ADA CONTINUOUS AMMONIA SLIP MONITOR
Michael D. Durham, Richard J. Schlager, Mark R. Burkhardt,
Francis J. Sagan and Gary L. Anderson
ADA Technologies, Inc.
304 Inverness Way South, Suite 110
Englewood, CO 80112
ADA Technologies, Inc. has developed a continuous emissions monitor for use with
advanced NOX control technologies that is capable of simultaneously monitoring ppm levels
of NH3 and NO in flue gas. The instrument can also measure SO2 when it is present in the
flue gas. The instrument is based on ultraviolet light absorption using a photodiode array
spectrometer. It has unique advantages over other ammonia instruments as it directly
measures ammonia as opposed to the indirect chemiluminescent techniques which must
infer the NH3 concentration from the difference between two large numbers. The monitor
has undergone extensive laboratory and field evaluation and data are presented which
demonstrate sensitivity, accuracy and drift of the instrument. The analyzer has been field
tested at a gas turbine with SCR, a coal-fired circulating fluidized bed with ammonia injection,
a refinery boiler with SNR, and a utility boiler with urea injection. The accuracy of the
instrument was determined by comparison with extractive wet chemical measurements.
7A-17
-------
I. INTRODUCTION
ADA Technologies, Inc. has developed a continuous, real-time analyzer for measuring part-
per-million levels of ammonia (NH3) and nitric oxide (NO) in flue gas associated with
advanced NOX reduction systems. A two-year long development program sponsored by the
U.S. Department of Energy resulted in an analyzer that is specific to ammonia, reliable, and
accurate. Other common flue gas components do not interfere with the measurement of
NH3.
This instrument fills the need created by advanced NOX control technologies for an ammonia
slip monitor which can be used as part of the process control system. Ammonia is a primary
ingredient in virtually all of the advanced NOX control processes such as selective catalytic
reduction (SCR) and selective non-catalytic reduction (SNR) technologies. However,
because of severe problems related to the penetration of unreacted NH3 through the flue
gas treatment system, it is extremely important to measure and control the downstream
concentrations of NH3.
The instrument is an effective diagnostic tool for optimizing De-NOx systems, and will be a
valuable component of NOX control equipment in many applications including: coal-, oil- and
gas-fired utility boilers, co-generation plants, refineries, municipal solid waste incinerators,
and research programs.
The monitor has been operated as both an in-situ and extractive instrument. The extractive
mode of operation allows a testing team to evaluate the stratification of NH3 gas across the
diameter of a duct. This capability is particularly important in evaluating whether ammonia is
dispersed uniformly within the flue gas of a SCR or SNR De-NOx system.
II. MEASUREMENT PRINCIPLE
A. MEASUREMENT PRINCIPLE
Ammonia and NO absorb light in the ultra violet (UV) range at specific wavelengths, and the
shape of the absorption spectra are characteristic of the identity of the particular gas. Figure
1 shows absorption spectra for NH3 and NO in a selected UV wavelength region. In this
spectral range, NO absorbs at two characteristic wavelengths, and NH3 absorbs at four
characteristic wavelengths. The two large doublet peaks identify the absorption due to NO,
and the four smaller peaks, which include two characteristic doublets, are due to NH3. The
quantity of light absorbed by a gas is proportional to its concentration, as defined by
principles of Beer's Law. Since the NO doublet located near diode 400 overlaps with one of
the ammonia peaks, this region cannot be used for analysis. However, the NO peak at
diode 850 and the ammonia peaks at either diode 200 or diode 600 do not interfere and
therefore can be selected and analyzed to determine the concentrations of these two gases.
The data available from the multichannel spectrometer allow measurement of these two
gases directly and accurately.
7A-18
-------
3.20
AMMONIA AND NITRIC 0X11113 SIM3CTHA
3.10 -
3.00 -\
2.90 -1
CJ^
s- 2.80 -
(si
§ 2.70 H
H
,_ 2.60 -
re
o
2.50 -
2.40 :
2.30 -
2.20 -
10 ppm NH3
• •••• 10 ppm Mi-la and 200 ppm NO
' ' i—|—i—i—i | i i i | i—i—i—| i > i—|—J—T—i | i i i | i i—i—|—i—i—rn—i—i—i—i—r
0 100 200 300 400 500 600 700 BOO 900 1000
PHOTODIODE ELEMENT #
Figure 1. Absorbance spectra for ammonia and ammonia/nitric oxide mixture.
B. DETECTION SYSTEM
Photodiode array detectors provide a technology to improve upon the design of
conventional scanning monochromator-based spectrometers. The improvement involves
the placement of a series of detectors across the focal plane of a polychromator, each with
its associated readout electronics. The most advanced of these systems use a linear
photodiode array (LPDA) detector. The LPDA is a large-scale integrated circuit fabricated on
a single monolithic silicon crystal. It consists of an array of diodes, or pixels, each acting as
a light-to-charge transducer and a storage device. These detectors are ideally suited for use
in UV spectrometers because they have a large quantum efficiency (40-80%) throughout the
range as well as geometric, radiometric, and electronic stability. The array itself can be
mounted and operated so as to be tolerant of high temperature, humidity, vibration, and
electrical and magnetic fields.
An LPDA spectrometer system, shown schematically in Figure 2, operates by passing a
continuous light source through the sample and into the polychromator. The polychromator
disperses the light across the LPDA, which has replaced the exit slit of a conventional
spectrometer. The array is located in the focal plane of the polychromator so that each
diode corresponds to a particular wavelength resolution of the UV-VIS spectrum. The diode
array provides an almost ideal sensor for the digital acquisition of spectra, as the array itself,
by its presence in the focal plane of the spectrometer, digitizes the spectrum into discrete
intervals. Unlike the scanning spectrometers, whose wavelength accuracy is mechanically
limited, the LPDA spectrometer is limited only by geometric constraints of the detector itself,
7A-19
-------
Deuterlun
Lanp
Gas Cell
Gas
Outlet [~ |
Gas Inlet
^Quartz Pr,sn^ p^
w> Entrance Silt
Colllnatlng Mirror
Peltier Cooler
T
^L-^
I
Xlrror
ife^l
Thernocouple
ontrollor
Figure 2. Schematic diagram of Linear Photodiode Array spectrometer system.
by vibration and thermal expansion of the optical components, and by the stability of the
source. Wavelength accuracy is equivalent to the diode spacing multiplied by the linear
dispersion of the spectrograph. Its geometric registration and, therefore, its wavelength
accuracy and precision, are greater than any mechanically scanned spectrometer
With the PDA detector it is possible to develop algorithms which use the unique structure of
the absorbance spectrum to quantify the concentration of the gas. This approach eliminates
the need to maintain the initial intensity (IJ reference and simplifies and speeds the
calculation. Since the analysis procedure searches for characteristic features of the
absorption spectrum rather than a fixed wavelength, it is less sensitive to drift or lamp
intensity fluctuations.
The photodiode array detector has unique advantages over all the other ammonia
instruments. It provides a direct measurement of ammonia and is, therefore, inherently more
sensitive than the indirect chemiluminescent measurement techniques which must infer the
NH3 concentration from the difference between two large numbers. In addition, the
photodiode array spectrometer has the following unique features.
• The instrument can be built with no moving parts which will reduce
maintenance and increase reliability in an industrial environment.
• The software is written to provide built-in checks for alignment of the optics.
• Changes in light intensity to do create a drift problem.
• Finally, the interferences are well known and can be accurately handled by
the PDA detector.
7A-20
-------
III. LABORATORY EVALUATIONS
A. TEST SET-UP
Performance parameters of the analyzer were determined in a series of laboratory tests.
Gases used in the evaluation were supplied in cylinders containing the individual gases in a
background of nitrogen gas. The concentrations of the gases were certified by the
manufacturer through analysis. Gases were mixed in various combinations and
concentrations using mass flow controllers and manifold system. The gas flow was then
metered into the analyzer for evaluating performance. Tests were conducted using a gas
cell with a path length of 90-cm. The cell was heated to maintain an internal gas temperature
of 300 °F. Results of the evaluation follow.
B. LINEARITY OF NH3, NO, AND S02
The linearity of the response of the analyzer was evaluated by initially calibrating the analyzer
using nitrogen and a span gas for each component of interest. Gas concentrations were
then decreased in steps and resulting analyzer measurements noted. Results of the linearity
evaluation for NH3, NO, and SO2 are shown in Figures 3 through 6.
Ammonia results are shown for two ranges of measurement, 0 to 70 ppm and 0 to 10 ppm.
Figure 3 shows that when calibrated at 70 ppm, measured concentrations are within 1 ppm
of the input concentration. For the low range, Figure 4 shows that measured concentrations
are within 0.5 ppm of the input concentration.
Prior to measuring the linearity of the NO, the instrument was calibrated using two
concentrations of NO because the absorbance of NO requires a second order equation to fit
the calibration curve. Using this technique, the linearity of the instrument is within 2% of the
actual concentration over a concentration range of 0 to 200 ppm as shown in Figure 5. If
only a single gas is used for calibration, there is a maximum 10% deviation from linearity in
the middle of the range.
Figure 6 shows the linearity of the analyzer for S02 calibrated at 80 ppm. For all gas
concentrations, the measured values are within 1 ppm of the input concentrations. The
capability to accurately measure sulfur dioxide provides the basis for eliminating its
absorbance as an interference to the measurement of NO and NH3.
C. LONG-TERM NOISE AND DRIFT
Analyzer noise and drift were estimated by observing instrument readings over a 36 hour
period of time as a mixed gas stream of fixed composition was passed through the
measurement cell. Analyzer measurements for NH3, NO, and SO2 are shown in Figures 7.
The composition of the gas stream was 10 ppm NH3, 55 ppm NO, and 80 ppm SO2.
7A-21
-------
100-
p. 90-
o.
80-
70-
g 60H
o
g 50H
u
"40-
OJ
u
3 20-
n
m
10
—i—
40
—I—
50
20 30 40 50 60 70 80
Input NH3 Concentration (ppm)
—i—
90
100
Figures. Linearity of NH3 measurements when analyzer is calibrated using 70 ppm
standard gas.
d
o
d
u
8-
4-
55 2H
cd
QJ
a
Input NHS Concentration (ppm)
Figure 4. Linearity of NH3 measurement when analyzer is spanned using 10 ppm
calibration gas.
7A-22
-------
250
40 60 80 100 120 140 160
Input NO Concentration (ppm)
180 200
Figure 5. Linearity of analyzer to NO input concentrations when calibrated using two span
gas concentrations.
100
a 8CH
o
"d
d 60
0)
a
d
o
U
N 40-
•a
tu
in
id
0)
a
20-
—I—
40
—i—
60
20 40 60 80
Input SOZ Concentration (ppm)
100
Figure 6. Linearity of SO2 measurements when analyzer calibrated using 80 ppm span gas.
7A-23
-------
au-i
'"IT
a
0,70-4
>s
fl6o4
.2 * —
A * ft - -- —ft 6- ft
^ $ - v ^ v w
00000 BO ppm S0t Input
•
_ e (5 a e — -• Q e
a
£40-
a
o
<-> 30-
tn
cd
O 20-
10-.'
OOOOO 55 ppm NO Input
10 ppm NH3 Input
—i—i—i—i—i—i—i—i—i—i—i—i—i—i—i—i—i—i—i—i—i—
4 8 12 16 20 24
Measurement Period (hr)
T"1
28
32
36
Figure 7. Noise and drift characteristics of NH3, NO, and SO2 measurements over 36-
hours.
As can be seen from Figure 7, with unattended operation, the output is extremely stable for
all three oases Analyzer noise is defined as the short-term peak to peak signal variation,
and is equal to'jt 0.3 ppm for NH3, ±0.15 ppm for NO, and _± 0.1 ppm for SO2. Analyzer
drift is defined as the long-term variation in analyzer signal around an average value.
Analysis of the measurements shows that the drift is ± 0.3 ppm for NH3, and jf 0.3 ppm for
NO drift, and is _± 0.4 ppm for SO2. These noise and drift measurements are well within the
accuracy capabilities of the gas flow delivery system using the mass flow controllers.
D. RESPONSE TIME
The response time of the analyzer is a function of how quickly a sample of gas is delivered to
the light path and the time it takes to process the spectral information into gas concentration
units. Since the data processing time is very short, on the order of a few seconds, the rate of
response becomes directly related to the volume of the gas cell and the flow rate of the gas
through that cell. For example, 90% of full scale response is achieved to a known NO
calibration gas input within five equivalent volume changes of the cell. This rate of gas flow
through the sample system is typically done within 1-minute. The response to ammonia gas
is slightly slower than observed for NO, due to the nature of ammonia gas which requires
conditioning of tubing surfaces during its travel to the measuring cell.
7A-24
-------
E. MINIMUM DETECTION LEVELS
The minimum detection level for a particular gas is defined as twice the noise value. Based
on data shown in Figure 7, the minimum detectable level using a 0.9 meter log gas cell is 0.6
ppm for NH3 and 0.3 ppm for NO.
The minimum detectable level and maximum concentration measurable using absorption
spectroscopy are a function of the path length that the light travels through a gas sample.
Higher gas concentrations can be measured using a shorter path length, but minimum
detection levels increase in proportion. In actual practice, gas measuring cells lengths are
specified based on the particular application and accuracy requirements.
F. INTERFERENCES
Several gases that are typically found in flue gas absorb light in the lower UV region and
present a potential for interfering with the measurement of NH3 and NO. However,
experiments were conducted which demonstrated that at typical flue gas concentrations,
NO2, CO, CO2, O2> and H2O did not interfere with the measurement of NO and NH3. The
most predominant interference is SO2 which, depending upon the concentration, can be
accounted for using spectral subtraction which has been described previously (Durham et
al., 1989). The maximum SO2 concentration that can be accurately subtracted from the
absorbance spectrum depends upon the length of the gas cell. For example in a 0.9 meter
cell, the maximum concentration of SO2 is 80 ppm. If the cell is reduced to 4 cm, then the
maximum concentration increases to 1800 ppm SO2. However, with the smaller cell the
minimal detection limit for NH3 is increased to 13 ppm. Therefore, a gas cell needs to be
selected for the specific application.
IV. FIELD EVALUATIONS
A. GAS TURBINE WITH SCR
The ADA Analyzer was used to evaluate the De-NOx system of a gas-fired co-generation
facility. At this site, the Analyzer was evaluated as both an in-situ and an extractive
instrument. The in-situ instrument avoids sample biasing and minimizes the operating and
maintenance requirements. The extractive version is designed for traversing ducts
downstream of the NOX control system to optimize the ammonia injection configuration.
At this site, ammonia is injected upstream of a selective catalytic reduction (SCR) bed to
control the NOX emissions. The plant did not have an ammonia detector but did monitor the
concentration of NOX at the inlet and outlet of the SCR and measured the quantity of
that was injected. The target NOx emission from the facility was 18 ppm.
7A-25
-------
Verification of the Accuracy of the Instrument
The measurement accuracy of the analyzer was determined by comparing instrument
emission measurements against a standard wet chemical technique. This manual technique
involves extracting a sample of the flue gas from the stack and bubbling it through an acidic
solution which collects the ammonia. The solution is then analyzed in a laboratory using a
selective ion electrode to determine the quantity of NH3 collected. Although this technique is
very manpower intensive, accurate measurements can be obtained if the procedures are
followed carefully. An experienced third party testing firm was contracted to perform the wet
chemical measurements.
Several wet chemical tests were conducted while the analyzer continuously measured NH3
concentrations. The analyzer was used in-situ, while wet chemical tests were conducted
from a different, neighboring port on the duct. In spite of the fact that the measurements
were made at different points in the stack, there is excellent agreement between the two
methods. Figure 8 shows a comparison of the ammonia concentrations measured by the
continuous analyzer and the manual method. The straight line represents a 1:1 correlation.
The numbers inside the data points are the ports where the extractive measurements were
made. The ADA instrument was operated at a port midway between the two orthogonal
ports 1 and 4. The different ammonia levels in the stack were achieved when the facility
operator manually adjusted the ammonia injection rate. This data demonstrates that the
instrument is capable of accurately measuring the concentration of ammonia in a flue gas
stream.
25-
in
I
20-
15-
E
a.
a.
10-
5 5H
o
o
o
Numbers Represent Extractive Sampling Ports
©
Sample Port Configuration
3 / ^~\ 2
ADA
5 10 15 20 25 30
NH3 CONCENTRATION (ppm) BY WET CHEMICAL ANALYSIS
Figure 8. Comparison of NH3 measurements using the ADA In-Situ monitor and extractive
wet chemical analysis at a co-generation facility.
7A-26
-------
Continuous Operation
The instrument was operated on a 24-hour per day basis during the test week. Algorithms
were written to eliminate any detrimental effects due to fouling of the lenses or mirror. During
the operation of the instrument some fouling of the mirror did occur due to the deterioration
of the purge blower. This resulted in a reduced magnitude of light detected by the
photodiode array. However, the algorithms operated as designed to account for loss of light
level, and the fouling had no effects on the measurements of NHg and NO concentrations.
Figure 9a shows a plot of the data obtained during a 24-hour period. The trends in the NH3
and NO measurements show a gradual decline in the NO concentration while the ammonia
slip is increasing. Whenever a sharp change in NO level occurs, there is a corresponding
change in the opposite direction for NH3. The ammonia injection rate is plotted in Figure 9b.
As can be seen there is a strong correlation between the ammonia injection rate and the
ammonia slip. This data indicates the variability that occurs in even a stable combustion
system such as the gas turbine combustor.
Evaluation of the SCR System
The data obtained during the continuous in-situ measurements were reduced to determine
the relationship between the NO level and the NH3 slip. These data, which are plotted in
Figure 10, provide very valuable information relative to the performance of an SCR system. It
can be seen that for higher concentrations of NO there is very little slip and the amount of slip
increases as the NO is reduced. However, at some point any further decrease in NO can
only be achieved with a significant increase in ammonia slip.
This data is extremely important relative to the cost-effective operation of an SCR and the
resulting emissions. If the facility is operating under a permit that specifies only a maximum
NO concentration, without considering the ammonia slip, the minimum level of emissions will
not be obtained. In this example, in order to obtain a 2 ppm reduction in NO from 19 to 17
ppm, the NH3 slip will increase by 20 ppm. It would be more desirable to operate at the knee
of this curve to minimize the total release of pollutants.
Operating at this point would also make economic sense. At an ammonia slip level of 25
ppm, half the injected ammonia is going up the stack unreacted. This means that the cost of
the ammonia is double what it would be if the system were controlled with the slip as a
parameter. This data also demonstrates the importance of a continuous ammonia slip
monitor. Since the performance of the catalyst in the SCR is going to change over time, the
continuous monitoring of the flue gas can be used to identify the optimum operating
conditions at all times.
Evaluation of the Extractive Analyzer
The analyzer was also used in an extractive mode in order to measure gas concentration
gradients in the system. A probe was used to draw samples of flue gas from discreet points
across the diameter of the stack and into the analyzer. Since there was no access
immediately downstream of the catalyst, a traverse was made at the stack. The traverse was
7A-27
-------
25-
-20-
Concentrations of NHj and NO During Continuous Operation
r NO Concentration
-NHj Concentration
-50
-40;
T—i—i—i—i—i—i—i—i—i—i—i—i—i—i—i—i—i—i—i r
0 1 2 3 4 5 6 7 8 9 10111213141516171819 20 21
•60
a.
a.
QL
O
6
1-20°
O
-10
OPERATING TIME (hrs)
Figure 9a. Continuous NH3 and NO measurements from a co-generation facility.
MH3 Injection Rote During Continuous Operation
II I I i—I—I—i—1—1—I—|—|—I—r
01 23456789 101112131415161718192021
OPERATING TIME (hrs)
Figure 9b. Ammonia injection rates during emissions measurements.
7A-28
-------
25-
E
Q.
CL
v—,
Q_
,20-
15-
5 10-
5-
0 1 i r • i i i i—i—r—|—i r~r -T~;—i—i—rn—|—i—i—1-1 ] ~I~T~T i—| "T i1 i i | i i i -i—|- T -i i i | i i i
0 5 10 15 20 25 30 35 40 45 50
OUTLET NO CONCENTRATION (ppm)
Figure 10. Nitric oxide emissions as a function of ammonia slip.
made parallel to the ammonia injection grid. The results presented in Figure 11 show the
presence of strong gradients in both NO and NH,, concentrations across the stack. The
higher levels of NO correspond with lower levels of NH3. Both the gradients and the inverse
relationship between NO and NH3 are due to an improper balancing of the ammonia
injection valves. This shows the usefulness of the extractive instrument in providing a means
to optimize the ammonia injection system.
B. COAL-FIRED FLUIDIZED BED WITH SNR
The ADA Continuous Ammonia Analyzer was field tested at a 49.5-MW coal-fired circulating
fluidized bed co-generation facility. The plant injects ammonia into the primary cyclone for
control of NO . The on-site CEM system incorporates a chemiluminescent instrument to
beasure both NH3 and NOX levels using a thermal converter for ammonia. Flue gas samples
are withdrawn from the center of the stack (approximately 100 feet above ground level) via a
heated in-situ probe. The flue gas is pulled down approximately 100 feet of heated sample
line to an instrument enclosure. Moisture is removed from the flue gas sample before it
entered the NO^ analyzer. In the NH3 measurement mode, a solenoid valve is activated
periodically, forcing the flue gas through a thermal converter which converts the NHL to NO.
The signal generated from the flue gas that by-passes the thermal converter is subtracted
from the signal generated when the flue gas passes through the thermal converter to obtain
the NH3 concentration present in the sample.
7A-29
-------
N
14 4
136
1L4
31.5
36.5
37.3
385
s
c
R
o
©
©
©
©
©
®
©
From
Turbine
NH3
Injection
Valves
Figure 11. Measured concentration gradients for NH3 and NO.
The field test program was performed to determine the accuracy of the ADA Continuous
Ammonia Analyzer for measuring NH3, SO2 and NO in a flue gas environment containing low
levels (5-40 ppm) of SO2. As was done in the previous field study, the NH3 concentrations
measured by the ADA monitor were compared with those obtained using the standard
ammonia wet chemical technique performed by a third party. In addition, a comparison
between the ADA ammonia monitor and the chemiluminescent ammonia monitor determined
how well the two techniques agreed with each other and with the standard wet chemical
method.
Simultaneous NH3 measurements were taken using the wet chemical method, the ADA
ammonia monitor, and the chemiluminescent ammonia monitor. The chemiluminescent
ammonia monitor took samples from the center of the stack through a heated sample probe.
The ADA ammonia monitor measured NH3 directly in the stack through a port positioned at a
90° angle from the chemiluminescent monitor sample probe. The wet ammonia
measurements were performed by positioning the wet ammonia sample probe adjacent to
the ADA in-situ probe. This was done by placing the sample probe through the sample port
90° from the ADA monitor (180° from the chemiluminescent ammonia monitor) and then
bending the sample line to physically contact the ADA in-situ probe.
Figure 12 shows the comparison of the NH3 concentrations measured by the ADA ammonia
monitor, the chemiluminescent ammonia monitor, and the wet chemical ammonia method.
All data were corrected for 7.8% moisture and 5% oxygen. These conditions were measured
in the stack at the time of sampling. The sample points are averages taken over the wet
ammonia method sampling time. Measurements of different NH3 levels were attempted
7A-30
-------
when the facility operators manually adjusted the ammonia injection rate. However, the
vaporizers were not functioning properly at the time of the test, and the ammonia control
valves were opened fully.
As shown in Figure 12, the wet chemical and the ADA methods agree well. This test also
shows the effectiveness of the ADA processing package in eliminating the interfering effects
of SO2 on the NH3 measurements. The chemiluminescent ammonia monitor response was
approximately 3-5 times higher than the standard wet chemical method. This inaccurate
measurement of the ammonia slip could result in the injection of an insufficient quantity of
ammonia to react with NOX.
a
'5-
d
o
S4-
-U
fl
QJ
O
fl ,
o 3"
O
Chemiluminescent Indirect
2 3
Time (Hours)
4
Figure 12. Ammonia slip measurements on a coal-fired fluidized bed boiler using three
methods.
C. REFINERY BOILER WITH SNR
The ADA analyzer was used to measure NH3 and NO emissions from a thermal De-NOx
system used on a refinery boiler gas stream. Ammonia gas was injected into the hot exhaust
gas from a furnace in order to effect the NOX reduction reaction. The gas stream contained
several hundred parts per million SO,. Therefore, a gas measuring path length was chosen
to most effectively accommodate the 1lue gas SO2 content, while still providing the necessary
degree of accuracy for NH3 and NO measurements.
7A-31
-------
Accuracy Determination
The analyzer was again used in both an in-situ and extractive mode to gather data. The
facility performed several wet chemical NH3 evaluations while the analyzer operated in-situ.
These results compared as follows:
Wet Chemistry ADA Analyzer
51 ppm 60 ppm
171 ppm 225 ppm
These results indicate good agreement between the methods, especially given the rapid
short-term changes in NH3 emission levels observed in the flue gas stream using the real-
time analyzer.
De-NOx System Evaluation
Ammonia slip and NO emissions data were collected as De-NOx system variables were
adjusted. Figure 13 shows the relationship between NO emissions and NH3 slip measured
over a range of operating conditions. Because of the proprietary nature of the information,
the data are plotted in relative concentration terms. This figure has a very similar shape as
the plot obtained from the SCR tests in that there is a point of diminishing returns relative to
the amount of ammonia injected. This is represented by the point where only minimal
reduction in the concentration of NO is obtained at the expense of significant increases in
ammonia slip . Figure 14 shows the relationship between NH3 slip and NH3 injection rates.
Data such as these, when collected in combination with otner process information, can
produce a significant data base for use in characterizing a De-NOx system, and for
troubleshooting purposes.
The data presented on the De-NOx system evaluation were collected in only a few days of
testing. These results demonstrate the ability of a real-time analyzer for effectively
characterizing emissions from a full-size control system.
D. UTILITY BOILER WITH UREA SNR
The final field test program was conducted during a demonstration of urea injection into a
utility boiler. This program was conducted during October to December, 1990 and is
described in the paper by Abele (1991) which is presented at the 1991 NOX Control
Symposium. During this program, the instrument was successfully operated during the test
program. The calibration of the instrument was checked at the beginning and end of the
program. After nearly two months of operation, the calibration constants had drifted less
than 2%.
7A-32
-------
co 6
c
u
o
O
U
O
1234567
Ammonia Slip, Relative
10 11
Figure 13. Nitric oxide emissions as a function of ammonia slip at a refinery boiler.
12
10
55 6
.2
o
E ,.
2 4 6 8 10
Ammonia Injection Rate, Relative
12
Figure 14. Relationship between ammonia slip and ammonia injection rate for refinery SNR
system.
7A-33
-------
V. STATUS
ADA continues to provide testing services and analyzers for evaluations of De-NO^ systems.
ADA has been working toward commercialization of the analyzer technology with instrument
manufacturers. ADA will be participating in a round-robin performance evaluation of
commercially available analyzers with regulatory agency involvement beginning in March.
ADA highly endorses such programs and will report results at upcoming meetings.
VI. REFERENCES
Durham, M.D., T.G. Ebner, M.R. Burkhardt, and F.J. Sagan (1989). "Development of an
Ammonia Slip Monitor for Process Control of NH~ Based NOX Control Technologies",
presented at the AWMA International Specialty Conference on Continuous Emission
Monitoring-Present and Future Applications, Chicago, IL November 12-15.
Abele, A. (1991). "Performance of Urea NOx Reduction System on Utility Boilers", EPRI-EPA
1991 Joint Symposium on Stationary Combustion NOX Control, Washington D.C.,
March 25-28
7A-34
-------
ONTARIO HYDRO'S SONOX PROCESS FOR
CONTROLLING ACID GAS EMISSIONS
R. Mangal and M.S. Mozes
Ontario Hydro Research Division
800 Kipling Avenue
Toronto, Ontario
M8Z 5S4 Canada
and
P.L. Feldman and K.S. Kumar
R-C Environmental Services and Technologies
US Highway 22 West
Branchburg, New Jersey
USA 08876
-------
ONTARIO HYDRO'S SONOX PROCESS FOR
CONTROLLING ACID GAS EMISSIONS
R. Manga! and M.S. Mozes
Ontario Hydro Research Division
800 Kipling Avenue
Toronto, Ontario
M8Z 5S4 Canada
and
P.L. Feldman and K.S. Kumar
R-C Environmental Services and Technologies
US Highway 22 West
Branchburg, New Jersey
USA 08876
ABSTRACT
An in-furnace slurry injection process for the simultaneous control of SO, and NO, from power plant flue gases has been
developed at Ontario Hydro's 640 MJ/h (0.6 x 10* BTU/h) Combustion Research Facility. The process known as SONOX
involves the injection of an aqueous slurry of a calcium-based sorbent such as limestone, dolomite, hydrated lime, etc and
a nitrogen-based additive into the furnace at temperatures ranging between 900 to 1350°C. Coals varying in sulphur
content from 0.54 to 2.8% with NO, emission levels of 450-620 ppm were studied. Operating parameters have been
optimized for maximum SO, and NO, capture. Under optimized operating conditions the technique removes up to 85%
of the SO2 and effective NO, removal is 63-80%. The specific removal levels obtained depend upon the type of sorbent
and nitrogen-based additive, temperature, stoichiometry and coal. The effluent gas stream has been characterized for NH,,
HCN and N2O. The solid waste produced is composed of fly ash, CaSO4 and CaO which can be collected by the ESP.
Due to the high dust loading that results from the process, the ESP performance deteriorates somewhat. A levelized cost
estimate indicates that a SONOX system is about half the cost of a wet FGD system to own and operate. Negotiations
are in progress to demonstrate this process on full scale boilers.
INTRODUCTION
In December 1985, the Ontario government announced a more stringent acid gas emission policy limiting Ontario
industries in atmospheric emission of SO, and NO.. Ontario Hydro's limits were reduced to 430,000 tonnes/year starting
in 1986 and to 215,000 tonnes/year starting in 1994. This regulation is challenging in that Ontario Hydro must stay below
the regulated tonnage limit regardless of changes in the demand for energy or in other forms of generation. Although the
regulation limits the amount of SO2 emissions, the level of NO, emissions is not specifically regulated and Ontario Hydro
is free to trade between SO2 and NO, as long as the aggregate emissions of the two (SO2 and NOJ does not exceed
215,000 tonnes/year and no more than 175,000 tonnes/year may be SO,(1,2). Specific NO, legislation is now being
negotiated between the Federal and Provincial Ministers of the Environment
Consequently, Ontario Hydro embarked on a program to curtail acid gas emissions from its coal burning plants. This
program was initiated to meet the above mentioned regulations.
Several options are being considered to reduce both SO, and NO.. In the case of SO,, some options include: sorbent
injection processes, burning low sulphur coals with flue gas conditioning, wet flue gas desulphurization and the limestone
dual alkali process. Ontario Hydro is committed to two scrubbers being in operation at the beginning of 1994. For NO,
control, the options can be classified as non-retrofit and retrofit technologies. Non-retrofit options would be to reduce
NO, emissions by installing fossil replacement generation that has lower NO, emission rates than are currently generated
by existing stations and to reduce coal generation. Burning natural gas is an example. Retrofit options include: low NO,
burners, selective catalytic reduction and selective non-catalytic NO, reduction processes-(additive injection).
7A-37
-------
Of the options considered to meet the above regulations in-fumace sorbent injection and selective non-catalytic NO,
reduction processes were investigated extensively at Ontario Hydro's 640 MJ/h Combustion Research Facility. As a result
Ontario Hydro's SONOX process which injects a calcium-based sorbent slurry and an additive to simultaneously abate
S02 and NO, was developed and is the subject of this paper.
The SONOX process is an in-fumace injection technique of an aqueous slurry of a calcium-based sorbent and a soluble
additive injected at temperatures ranging between 900 and 1350°C. The calcium-based sorbent reacts with SO2 and the
additive reacts with NO,. The furnace which serves as the chemical reactor provides sufficient residence time and
favourable temperature for the reactions. The following reactions represent globally, the SOj/NO, (SONOX) removal
paths:
CaCO3 -> CaO + CO2
CaO + SO2 + 1/2 O2 - > CaSO4
NO + Reagent (Additive) - > N2 + H2O
The technique provides excellent distribution and mixing with the flue gas for the above reactions to be efficiently
completed(3). A schematic of the process is shown in Figure la. The process steps can be visualized as follows:
• Atomizauon of Ca sorbent and additive;
• Water droplet evaporation;
Particle disintegration for the Ca sorbent and thermal cracking of the additive;
Calcination of the Ca sorbent;
Development of reactive sorbent and additive (CaO and
SO2 and NO, capture.
The above steps are Illustrated in Figure Ib for limestone.
EXPERIMENTAL
Combustion Research Facility
The study was conducted at Ontario Hydro's Combustion Research Facility (CRF) designed for a maximum coal feed rate
of about 20 kg/h (44 Ib/h) U.S. bituminous coal at a firing rate of 640 MJ/h (0.6 x 10* BTU/h) (Figure 2). The furnace
is a refractory-lined cylindrical chamber, fully equipped for monitoring gas and wall temperatures. There are multiple
ports for flame observation and for insertion of solid sampling probes. There are also probes to determine slagging and
fouling rates. The pulverized coal is delivered down-draft to the burner with the combustion air which can be electrically
preheated to temperatures up to 350°C (662°F). Gas burners on each side of the coal burner are used to heat the furnace
to operating temperatures before beginning to feed the coal.
The coal burner, designed and constructed by Research Division staff, is equipped with a vortex generator and four air
vanes to assure good mixing and adequate residence time of the fuel-air mixture in the combustion zone. The combustion
gases in the furnace are cooled by water and/or air circulating in the cylindrical Inconel jacket around the furnace. This
cooling system is equipped with temperature sensors and flow meters to control furnace quenching rates.
The combustion gases leaving the furnace are further cooled by a series of air-cooled heat exchangers prior to entering
the resistivity probe housing and ESP. The ESP consists of a cubic stainless steel chamber, and is equipped with two sets
of interchangeable cells. One set has an 11-plate electrode with 2.5 cm (1 in) spacing, the other a 5-plate electrode wiih
5 cm (2 in) spacing. The design specific collection areas (SCA, m2/m3/s) for the two sets of cells are 39 (0.2 ftVcfm) and
17 (0.09 ftVcfm) respectively for baseline firing conditions using a high volatile U.S. bituminous coal.
7A-38
-------
The CRF instrumentation permits systems temperatures, and flue gas composition (O,, COj, CO, SO, and NO.) to be
monitored continuously. Gas temperatures in the furnace are measured with a suction pyrometer and flame temperatures
with an optical pyrometer. Flow rates and pressures are measured by flow meters and manometers. All measuring and
monitoring systems are linked to a computerized data acquisition system. Paniculate mass loading in the flue gas before
and after the ESP is measured with an isokinetic sampling system. In-situ resistivity is measured with a point-plane
resistivity probe situated in the resistivity probe housing and particle size distribution of the ash is measured with a Pilot
Mark 3 Cascade Impactor.
A more complete description of the facility is given elsewhere/4/.
SONOX Hardware
A general overview of the hardware used is shown in Figure 3. A positive displacement pump pumps the slurry/additive
mixture from a continuously stirred mixing tank under a pressure of 650 to 720 kPa. Recirculation and a static mixer
upstream of the furnace kept the panicles in suspension and prevented settling. A small metering pump delivered the
slurry/additive mixture to the atomizer through which fine droplets were injected into the flue gas stream.
Injection was in the middle of the furnace through a twin-fluid high pressure nozzle (5 or 3 mm) with an internal mixing
chamber, shown in Figure 4. Operating pressures range between 40 to 60 psig. The stainless steel nozzle was purchased
from Turbotak Inc. The MMD of the droplets was about 12 |im for the 5 mm nozzle and approximately 6 ^im for the
3 mm nozzle. The nozzle was equipped with a cooling jacket which was necessary to avoid evaporation of the water and
hence drying of the slurry causing deposition of particles.
Fuels and Sorbents
Several coals ranging in sulphur content from 0.54% to 2.8% were used with the SONOX technology. These coals
include a 0.54% beneficiated western Canadian coal, supplied by Unocal Canada, a 1.1% S coal resulting from a blend
of western Canadian and eastern U.S., a 1.7% S eastern U.S. bituminous and a 2.8% S coal from Nova Scotia, Canada.
The proximate and ultimate analyses of the coals are shown in Table 1.
Sorbenis used include two local calciuc limestones from Ontario, namely Beachville and PL Anne. A Beachville hydrated
lime was also studied. Also from Ontario, a dolomitic stone was used supplied by E.C. King. A Mosher limestone from
Nova Scotia was used with the Nova Scotia coal. The chemical and physical properties of the raw sorbents are shown
in Table 2. These analyses were performed by ORTECH International - a research foundation in the province of Ontario.
Of the additives used to remove NO., the three best are described in this paper and are labelled A, B and C.
Procedures
After steady state was achieved with the baseline coal, injection of the sorbent slurry/additive into the middle of the
furnace was initiated. Temperature-lime and radial profiles simulating Lakeview and Lambton TGS were studied.
Changing the quenching rate allowed the effect of residence time to be studied. Data collected during each test include
system temperatures, and pressures, slurry/additive-feed rates and stoichiometry, flue gas constituents concentrations (CO2,
O2, CO, SO2 and NOJ, in-situ ash resistivities and particle size distribution. Coal, sorbents feed and fly ash samples were
collected during the tests. Analysis of samples include chemical composition and panicle size distribution. In selected
runs, NH3, N2O and HCN were monitored. Calcines and sulphated calcines were analyzed for CaO, CaCO3 and CaSO.
content. Porosity, mass median diameter and BET surface area of some samples were also determined. The analytical
methods used are described in reference(4).
7A-39
-------
RESULTS AND DISCUSSION
The most important parameters that were found to affect process performance (SO2 and NO capture) are classified under
the following categories:
Sorbent/Additive
• Chemical and physical characteristics;
Concentration; and
Addition rate (stoichiometric ratio).
Injection Parameters
Mode of injection;
• Droplet size, distribution and mixing with the flue gas;
• Temperature; and
• Residence time.
Coal
SO, and NO, concentration.
These parameters were optimized for maximum SOj/NO, capture on the pilot furnace. It is important, however, to address
some of the advantages of the SONOX process and to mention that negotiations are in progress to demonstrate SONOX
on the full scale. Some of the advantages are:
SONOX provides a low cost solution to the removal of acid gas from flue gases;.
SONOX is suitable for retrofit application;
SONOX is applicable to coals with various SO2 and NO, levels; and
SONOX requires short lead time for installation.
Sorbents Comparison
For SOj control using alkaline-based sorbents, sorbent composition and physical properties are important factors in
determining overall capture performance(5,6,7,8,9). Significant variability in the reactivity of the various sorbents has been
observed and it was recognized that surface area and porosity play a vital role in sorbent reactivity. Figure 5 illustrates
the effect of porosity on sulphur capture for various sorbents. Pt Anne limestone with an initial porosity of 55% gave
significantly higher removal than Beachville limestone with an initial porosity of 17% (70% removal for Pt. Anne
compared to 55% for Beachville) at a Ca/S ratio of 3.0. The Nova Scotia limestone slurry was used with the Nova Scotia
coal. Thus a direct comparison of process performance between this sorbent and the local calcitic stones was not possible.
Data indicate, however, that similar sulphur capture can be obtained with Nova Scotia limestone (porosity 57%) and the
Pi. Anne limestone (porosity 55%) even if they are used for two different coals (2.8% and 1.7% sulphur content).
Since the additives for NO.-removal are water soluble, only ihe effect of concentration and chemical composition were
evaluated.
Effect of Injection Parameters
Injection parameters that influence SO^NO, capture include: atomizer type, injector location, atomizing air pressure, and
particle size distribution or mass median diameter (MMD) of the atomized droplets. High atomization air pressure
improves the quality of atomization and promotes an early release of the sorbent/additi ve to engage in the sulphation/NO,
reduction reactions. It also increases the discharge momentum of the droplets leading to enhanced penetration and mixing
with the flue gas stream. These experiments were conducted with the Turbotak nozzle.
The effect of atomizing air pressure on droplet size is illustrated for limestone slurry in Figure 6. SO2 capture was found
to be a function of droplet size distribution, and quality of atomization and mixing with the flue gas. The best mixing
was observed while spraying a 40% aqueous Pt. Anne limestone slurry into the furnace cocurrently at an injection location
which was close to the flame zone where increased turbulence exists. Increasing the atomizing pressure from 40 psig 10
70 psig reduced droplet MMD from 12 nm to 6 |im and improved SOj capture from about 62% to 70% at Ca/S ratio of
3.0.
7A-40
-------
Effect or Temperature and Injection Mode
(a) Slurry Injection for SO, Control
The effect of temperature on SO2 capture was evaluated for the different sorbents (Pt. Anne limestone, Beachville
limestone, Beachville hydrated lime, Nova Scotia limestone and E.G. King dolomite) while burning the 1.7% S
eastern U.S. coal, the 1.1% S eastern U.S./westem Canadian coal blend and the 2.8% S Nova Scotia coal. The
results are shown in Figure 7a. Cocurrent injection gave higher SO2 capture than the countercurrent mode and the
opumum injection temperature for the recurrent mode was found to be 1200°C. The comparative performance for
the different coal/sorbent pairs was done with the Turbotak 3 mm nozzle as is illustrated in Figure 7a.
The highest capture, 85% was observed with hydrated lime to be followed by 83% with the E.G. King dolomite, 65-
70% with the porous Pt. Anne limestone and 55% with the Beachville limestone at a Ca/S ratio of 3.0 while burning
the 1.7% S U.S. coal. Under the same operating conditions, using the same limestone, SO2 capture from the western
Canadian/U.S. coal blend was slightly less than from the U.S. coal as is shown in Figure 7a. Injecting the Pt. Anne
limestone with the high sulphur Nova Scotia coal (2.8%) resulted in 76% SO2 removal at a Ca/S ratio of 3.0.
Sulphur removal efficiency was 58 to 63% using a 2.8% S Nova Scotia coal with the porous Nova Scotia limestone,
at a Ca/S ratio of 2.2. (Because of the presence of grits with this limestone and limited pump capacity, this was the
highest rate at which this sorbent could be fed to the furnace.) However, this compares favourably well with the
60% capture obtained at a Ca to S ratio of 2.5, using the porous PL Anne limestone with the 1.7% S U.S. coal.
Replacing 5% of the calcium from the PL Anne limestone by an equivalent amount of dolomite (dolomite doping)
resulted in 80% SO2 capture, up by 10% from what was achieved with pure Pt. Anne limestone.
(b) Additive Injection for NO. Control
The effect of temperature on NO, removal is shown for the three additives. A, B and C, in Figure 7b while they were
being injected cocurrently only. The data indicate that additives A and B show a common optimum at around
1100°C, while additive C shows a "flat" profile between 975 to 1100°C. At 1100°C, additives A and B removed
90 and 84% NO. respectively, while between 975 to 1100°C additive C removed 77 to 80% NO.. This can be quite
a desirable feature for full scale boilers where load is constantly changing resulting in changing temperatures. The
reason for additive C behaving differently from the others is not fully understood and further studies may be able
to provide an explanation.
Slip Gases
The concentration of nitrogen containing species such as ammonia (NHj), hydrogen cyanide (HCN) and nitrous oxide
(N2O) in the slip gases during additive injection has been investigated.
Results indicate that NH3 slippage for additive A ranged between 7 - 26 ppm and for additive C up to 49 ppm. HCN
was found to be between 3 - 9 ppm. With no NO. removal additive present the N2O produced ranged from 10-25
ppm at an initial NO. concentration of - 550 ppm. Decomposition of additive A has a side reaction which could
lead to the formation of N2O. The amount of N2O produced when additive A was injected ranged from 59 - 150
ppm at 1100°C and an additive/NO stoichiometric ratio of 2.0. These data demonstrate that 11 to 27% of the NO.
is converted to N2O thus the effective NO, removal for additive A is 63 to 80% instead of 90%. It was found that
NjO formation is affected by injection temperature, additive stoichiometry and NO. level in the flue gas. More
studies are required to optimize operating conditions for minimum conversion of NO. to N2O. Additives B and C
showed an increase in N2O levels of 5 - 15 ppm from the baseline.
(c) SONOX Process for SO/NO. Control
Simultaneous capture of S02 and NO, was undertaken by adding additive A to an aqueous slurry of PL Anne
limestone and dolomite doped PL Anne limestone while burning the 1.7% S eastern U.S. bituminous coal with an
initial SO2 concentration of 1350 - 1400 ppm and NO, concentration of 550 ppm. The results are illustrated in
Figure 7c for the following optimized conditions:
7A-41
-------
40% aqueous calcium-based slurry (Pt. Anne and dolomite doped)
Ca/S ratio = 3.0
Additive A concentration of 13.5% (w/w) in slurry
Addinve/NO mole ratio = 2.0
Injection mode: cocurrent
Nozzle: Turbotak 3 mm, droplet size = 6 ^m MMD
The graph of Figure 7c shows the effect of temperature on SOj/NO, capture for additive A combined with PL Anne
and dolomite doped PL Anne slurries. SO2 capture for the PL Anne slurry and additive A at the optimun
temperature of 1200°C is 70% and nominal NO, capture is 90%. With the 5% dolomite doped PL Anne slurry and
additive A, SO2 capture is 80% and nominal NO, capture is still 90%.
Effect of Stoichiometry
(a) Ca/S Ratio for SO. Control
The effect of Ca/S ratios on sulphur capture and sorbent utilization was studied while using the Pt. Anne (porous)
limestone with the U.S. coal, ihe U.S.-western Canadian blend and the Nova Scotia coal. The Beachvillc (non-
porous) limestone, dolomite and hydrated ume were studied only with the U.S. coal. Injecting the Pi. Anne
limestone with the U.S. coal was done at 1200°C and 1300°C while all other coal-sorbent combinations were done
at 1200°C. In all cases injection took place cocurrently using a 40% aqueous slurry. Ca/S ratios varied from 1.5
to 3.0 and the furnace quenching rate was held constant at 500°C/s. The results are shown in Figure 8a. Sulphur
capture and sorbent utilization are plotted vs Ca/S ratios for the various sorbem-coal pairs.
Sulphur capture decreases, but sorbent utilization increases with decreasing Ca/S ratios for all coal-sorbent pairs
tested. At the optimum temperature of 1200°C, dolomite and hydrated lime showed the highest capture. Dolomite
removed 78% of the SC^ at a Ca/S ratio of 1.5 for a calcium utilization of 52%, while hydrated lime removed 75%
and 83% SOj at Ca/S ratios of 1.5 and 2.5 respectively. Sorbent utilization was 50 and 33%.
At all Ca/S ratios, the more porous PL Anne limestone outperformed the less porous BeachviUe limestone both in
terms of sulphur capture and sorbent utilization. At 1200°C using the U.S. coal with the PL Anne limestone at a
Ca/S ratio of 3.0, sulphur capture and sorbent utilization were 65 to 70% and 22 to 23% respectively as compared
to 55% and 18% with the Beachville limestone. Using the PL Anne limestone with the high sulphur Nova Scotia
coal, sulphur capture at a ratio of 2.0 is 72% and at a ratio of 3.0 is 76%. Under the same operating conditions at
a Ca/S ratio of 1.5 sulphur capture for the Pt. Anne and Beachville limestones dropped to 50 and 31 respectively,
but utilization increased to 33 and 21%. With the Nova Scotia coal and Pt. Anne limestone at a Ca/S = 1.5, sulphur
capture is 64% with a sorbent utilization of 43%.
(b) Additive/NO Ratio for NO. Control
The effect of additive jjormalized stoichiometric ratio, NSR (NSR = moles of additive injected to the theoretical
moles required to remove 100% NOJ for the three additives. A, B and C, was studied while burning the eastern U.S.
bituminous coal. In all cases injection of each additive took place cocurrently at 1100°C while NSR was varied from
1.2 to 3.0. The concentrations of the additive solutions were as follows: A -13.5% by weight, B - 5.6% by weight,
and C - 16.1% by weighL The baseline NO, from the U.S. coal was 500-550 ppm.
NO, capture is illustrated in Figure 8b. NO, capture by A and C increases with increasing NSR up to 1.7 to a
maximum of 90% (nominal) and 80% respectively, and by B up to NSR = 2.0 to a maximum of 84%. Reagent
utilization drops with increased Stoichiometry for all three additives. The best utilization with A was 55-56% at an
NSR of 1.2 to 1.5, with B, 56% at a NSR of 1.0 and with C, 41 to 42% at a NSR of 1.5 to 1.7.
7A-42
-------
(c) Ca/S - Add/NO for SONOX
The effect of Ca/S mole ratio and additive/NO normalized stoichiometric ratio was studied by injecting the 5%
dolomite doped Pt. Anne limestone combined with additive A. The coal burned was the 1.7% S eastern U.S.
bituminous and injection was carried out cocurrently at the optimum temperature of 1200°C. The results in Figure 8c
show thai at a Ca/S ratio of 3.0, 80% SO2 capture is achieved and at an additive to NO stoichiometric ratio of 1.7
to 2.0, a nominal NO, capture of 90% is achieved.
Low Sulphur Coal Application
The development of the SONOX technology has been carried out mainly on a medium S (1.7%) eastern U.S. bituminous
coal and a high S (2.8%) coal from Nova Scotia with SO: emissions of 1350-1400 and 1700-1725 ppm and NO, emissions
of 550 and 450-520 ppm respectively.
The effectiveness of the SONOX process was also demonstrated on a western Canadian Obed coal sample, prepared by
UnocaJ Canada. The sulphur content of the coal is 0.54% with initial SO2 concentration of 349 ppm. NO, level initially
measured 620 ppm. A 40% aqueous dolomite doped Pt. Anne limestone slurry (10% dolomite) with additive A was
injected cocurrently in the pilot furnace and the effects of injection temperature and stoichiometry observed. The results
are illustrated in Figure 9.
In Figure 9a, SO^NO, capture as a function of injection temperature is plotted for constant stoichiometries, Ca/S = 3.0
and additive/NO normalized stoichiometric ratio of 3.0. The results indicate that the optimum temperature was around
1100°C for both pollutants with SO2 removal being 81% and nominal NO. removal being 89%.
The effects of Ca/S ratio and additive/NO stoichiometric ratio is shown in Figure 9b. Removal of both acid gas
components increases with increasing Ca/S and add/NO ratios. Optimum Ca/S ratio for SO2 is 2.0 to 2.5 and for NO,,
optimum add/NO stoichiometry is 2.0. Utilization of both sorbents improves with decreasing addition ratios as is shown
in Figure 9b. Under optimized operating conditions (injection temperature = 1100°C, Ca/S = 2.0-2.5 and add/NO = 2.0)
80% SO2 and 85% NO, was removed from the flue gas stream. Sorbents utilization and 32-40% and 43% respectively.
These results indicate that the SONOX technology is applicable to coals with various levels of sulphur content and NO,
levels.
Impact on Ash Characteristics, Collectibility and Deposition
The SONOX process produces increased amounts of waste composed mainly of CaSO<, unreacted CaO and fly ash. Any
impact on ESP performance and deposition on the radiant section and convective passes will depend on the type and
chemical composition, the particle size distribution and amount of Ca-based sorbent injected and waste produced.
Waste Characteristics
Particle size distribution of isolrinetically collected waste samples from the injection of various limestone sorbent slurries
while burning a 1.7% S U.S. bituminous coal are compared to that of an ash sample from the same coal in Figure 10.
The mass median diameter of the baseline ash is about 8 (am compared to 6 \im for the Pt. Anne and 9 \in\ for the
Beachville limestone slurry. The slightly finer waste resulting from the injection of the very fine Pt. Anne limestone is
not expected to affect panicle migration velocity and ESP collection efficiency.
In Table 3 a typical waste from slurry injection is compared to the baseline ash and to a waste from dry sorbent injection.
High levels of calcium compounds and the quantity produced must be considered for handling and disposal. CaO content
of a typical slurry waste is 302 g/kg and CaSO4 content is about 220 g/kg.
Dust Electrical Resistivity and ESP Performance
The resistivity of the baseline fly ash measured in-situ with about 10 ppm SO, naturally occurring in the flue gas from
the medium sulphur eastern U.S. bituminous coal is about 10* ohm.cm. During injection of all slurries, resistivities
consistently increased by one to two orders of magnitude to 109 to 1010 ohm.cm yet the electrical operating conditions of
7A-43
-------
the ESP were not seriously affected and collection efficiencies were not seriously degraded (see Table 4) from a baseline
level of 89% during slurry injection. Dry injection on the other hand results in a resistivity of 10" ohm.cm and an 8%
drop in collection efficiency. It is possible that due to the increased moisture level in the flue gas (up to 23% relative
humidity) a thin acidic film forms around the panicles and acts as a conditioning agent aiding the ESP in its performance.
Inlet mass loading to the ESP has increased 2 fold from a baseline level of 1.4 g/m3 with a resulting increase in paniculate
emissions by a factor of about 2 times during slurry injection. Thus the main problem with the SONOX process is the
high dust loading to the ESP which depends on the Ca/S ratio.
Slagging and Fouling Properties of the Waste
Soft deposits, which form at low temperatures and are generally characteristic of deposits found on air heaters and
economizers were observed on the furnace walls and heat exchanger surfaces. These deposits could be easily blown away
by compressed air suggesting that, conventional soot blowing equipment may suffice for full scale application of the
SONOX process.
SONOX COMMERCIALIZATION ISSUES
Electrostatic Precipitator Performance Following SONOX Application
The application of the SONOX technology in the upper furnace region affects the nature of paniculate mauer entering
the existing electrostatic precipitator. While the additives for NO, control do not add to the paniculate content entering
the ESP, the calcium sorbents for SO2 control in the furnace result in higher paniculate loading depending on coal sulphur
content and Ca/S ratios. The precipitator inlet loading can double for most applications. In addition to the increase in
inlet paniculate loading, an increase in paniculate resistivity is to be expected because of the uptake of SO, from the flue
gas by free lime in entrained solids. While dry sorbent injection technologies increase paniculate resistivity from about
109 ohm.cm to the 10" levels, paniculates from the slurry injection process show resistivity levels of about 10* 10'°
ohm.cm due to the higher moisture content in the flue gas. Hence, the electrical operation of the ESP is expected to
remain unaffected and only the solids loading will have to be dealt with.
Precipitator upgrades will be needed in most cases following sorbent injection in order to handle both high loadings and
increased resisuvity. Research-Cottrell has conducted a detailed study on behalf of the Electric Power Research Institute
and proposed solutions for the precipitator degradation problems following furnace sorbent uijection(lO). The most
economical solution is humidification and subsequent evaporative cooling of flue gas to restore resisuvity to pre-injection
levels. At the lower temperature, due to increased gas density, the precipitator can be operated at increased power
compared to the pre-injection level operation at 150°C. The precipitator can thus be operated at higher collection
efficiency to overcome the increased loading effect. The humidification concept for restoring precipitator operation has
been successfully carried out at two full-scale plants by EPRI and DOE(ll). The humidification concept has also been
demonstrated earlier by Research-Cottrell at the pilot scale in a CONOCO supported program.
The requirements for cooling to restore ESP performance are significantly reduced for the SONOX process because of
reduced paniculate resistivity. We expect the stack paniculate emissions to be restored to pre-injection levels by operating
at ESP inlet gas temperature between 110 to 120°C.
Economics
Economic analysis of the SONOX technology indicates that capital costs can vary between 30 to 60 $/KWe, including
moderate precipitator upgrade costs, for combined SO2 and NO, removal rates at SO to 70% each. This can be compared
to the wet FGD capital costs of ISO to 400 S/KWe, the higher cost numbers being applicable to smaller plants in the 150
MW size range. The operating costs of SONOX will be higher because of higher sorbent consumption when compared
to wet FGD. A levelized cost estimate, however, indicates that a SONOX system is about half the cost of a wet FGD
system to own and operate.
SONOX technology has been demonstrated at the pilot plant level. Application of the SONOX concept on a full-scale
coal-fired boiler does impact the overall system and the following questions need to be addressed to assure a successful
commercialization path:
7A-44
-------
• what is the optimum nozzle array configuration and slurry size distribution to assure proper gas-slurry
contact?
what is the optimum sulphation and NO, removal temperature window in the upper furnace region?
•vhat is the effect of increased solids loading on boiler tube erosion?
what is the effect of increased loading and resistivity on ESP performance, and what is the best precipitator
upgrade approach?
what is the best approach to increased solids handling of the calcium-rich ash?
Many of the answers to the above questions can be obtained from the experience with full-scale dry furnace sorbeni
injection systems already operating in Germany and other parts of Europe. Ontario Hydro/Research Cottrell are currently
seeking to demonstrate the SONOX technology on a full-scale coal-fired utility boiler.
SUMMARY AND CONCLUSIONS
The SONOX process, an in-furnace injection of a calcium-based sorbent and a nitrogen-based additive is a very efficient
way of removing SO2 and NO, from flue gases. This technique facilitates unproved distribution and mixing of the
sorbent/additive with the gas flow, reduces deactivation of the sorbent/additive and allows sufficient residence time at
favourable temperatures for the reaction between CaO and SO2, and NH2 and NO to be efficiently completed. The process
was developed at Ontario Hydro's 640 MJ/h (0.6 x Iff BTU/h) Combustion Research Facility. Coals studied ranged in
sulphur content from 0.54 to 2.8% and calcium sorbents used include two local calcitic limestones and a hydrated lime
from Ontario, a local dolomitic stone and a limestone from Nova Scotia. NO, levels in the flue gas ranged between 450-
620 ppm and several nitrogen-based additives were investigated. The following is a summary of the findings:
Sorbents chemical and physical properties are very important in determining the degree of SO^NO, removals.
Dolomite with a high magnesium content was very effective in removing SO2 as was the case for hydrated
lime. PL Anne limestone with an initial porosity of 55% was superior to Beachville limestone with an initial
porosity of 17%. Five percent dolomite doped Pt. Anne limestone increased SO2 capture from 70% to 80%.
The nitrogen-based additives did not vary substantially in their ability to remove NO,.
• Injection parameters were found to be also very important in removing SO2 and NO,. High atomizing air
pressure which improves the quality of atomization, promotes and early release of the sorbeni/additive mixture
and increases the discharge momentum of the droplets, increased SOj/NO, capture significantly. In the case
of SO2 removal, increasing the atomizing air pressure from 40 to 70 psig increased SO, capture from 62 to
70% for the PL Anne limestone.
The optimum injection temperature for SO2 control was 1200°C while NO. was 1100°C. However, with the
SONOX technology (simultaneous control of both SO2 and NOJ the optimum temperature was found to be
1200°C. Injecting 5% dolomite doped PL Anne limestone slurry and additive A at the optimum temperature
of 1200°C resulted in 80% SO, capture and nominal NO. capture is 90%. However, the effective NO.
removal is 63 to 80% because 11 to 27% of the NO, is converted to N2O. Hydrated lime removed up to 85%
SO2 from the flue gas.
Both SO2 and NO, improves with increasing Ca/S and Add/NO stoichiometric ratios. Optimum Ca/S and
Add/NO stoichiometric ratios were found to be 2.5 to 3.0 and 1.5 to 1.7 respectively. Burning the 1.7% S
eastern U.S. bituminous coal and injecting 5% dolomite doped PL Anne limestone at a Ca/S ratio of 3.0 and
additive A at a normalized stoichiometric ratio of 1.7 removed 80% SO2 and nominally 90% NO, at the
optimum temperature of 1200°C.
7A-45
-------
SONOX was also found lo be very effective for low sulphur coal application. Firing a low sulphur western
Canadian Obed coal supplied by Unocal Canada with a sulphur content of 0.54% and injecting 10% dolomite
doped Pt. Anne limestone slurry and additive A (Ca/S = 2.0-2.5 and add A/NO = '..7-2.0), removed 80% SO2
and nominally 85% NO, from the flue gas.
Particle size distribution of the waste from the Pt. Anne slurry was slightly finer than the baseline ash. The
waste contains fly ash and calcium compounds (CaO, CaSO4, etc) and the quantity produced must be
considered for handling and disposal systems.
Ash resistivities increased by one to two orders of magnitude from 10* ohm.cm to 10' to 10'° ohrn.cm but
ESP collection efficiencies were not seriously affected. The increased level of the flue gas moisture is
believed to act as a conditioning agent.
Slagging does not appear to be a problem and the soft deposit formed on the furnace walls and heat
exchanger surfaces was easily removable.
A levelized cost estimate indicates that a SONOX system is about half the cost of a wet FGD system to own
and operate and negotiations are in progress to demonstrate this process on the full scale.
FUTURE WORK
Studies are planned whereby other nozzles will be investigated. Other additives that have the potential for high NO,
removal while at the same time ensuring cost effectiveness of the SONOX technology will be studied Fundamental
studies to better understand the SOj/NO. removal paths will be undertaken. Activating and recycling waste from the
process is being investigated and utilization studies arc being conducted at the University of Calgary.
More importantly, negotiations are in progress to demonstrate this process on full scale boilers.
The work described in this paper was not funded by the U.S. Environmental Protection Agency and therefore the contents
do not necessarily reflect the views of the agency and no official endorsement should be inferred.
ACKNOWLEDGEMENTS
The authors wish to express a special thanks to Ontario Hydro's New Business Ventures Division for their dedicated
efforts in conducting negotiations to commercialize the SONOX technology. In particular, we recognize the efforts of
Mr. F. Schneider and Mr. R. Kozopas.
REFERENCES
1. Taborek, R., Dawson, C.W., and Stuart-Sheppard, IJL, "Acid Gas Emission Control - The Requirements, Technology
and Hardware" Ontario Hydro, Design and Development Division, Special Report, March 1986, 3799H.
2. Bumham, C.. "Ontario Hydro's Acid Gas Control Programs". Paper presented to the Standing Committee on General
Government, June 15, 1989.
3. Mangal, R., Mozes, M.S., Thampi, R., and MacDonald, D., "In-Fumace Sorbent Slurry Injection for SO2 Control"
Presented at the Sixth Annual International Pittsburgh Coal Conference, September 25-29, 1989, Pittsburgh, Penn.
4. Mozes, M.S., Mangal, R., Thampi, R., and Michasiw, D.L., "Pilot Studies of Limestone Injection Process Phase I:
Simulating Lakeview TGS Quenching Rate". Ontario Hydro Research Division Report No 86-62-K, May 30, 1986.
7A-46
-------
5. Kirchgessner, D.A., Gullett, B.K., and Lorrain, J.M., "Physical Parameters Governing the Reactivity of Ca(OH), with
SO2". Presented at the 1986 Joint Symposium on Dry SO2 and Simultaneous SOyNO, Control Technologies, June
2-6, 1986, Raleigh, North Carolina.
6. Dismuk;-.-;, E.G., Berttel, R., Gooch, JP., and Rakes, S.L., "Sorbent Development and Production Studies". Presented
at the 1986 Joint Symposium on Dry SO2 and Simultaneous SOj/NO, Control Technologies, June 2-6,1986, Raleigh,
North Carolina.
7. Szekely, J., Evans, J.W., and John, H.Y., "Gas Solid Reactions". New York, Academic Press, 1976.
8. Simmons, G.A., "Rate Controlling Mechanism of Sulphation". Proceedings 1986 Joint Symposium on Dry SO2 and
Simultaneous SO^NO, Control Technologies, Vol 2, EPRI CS^966, December 1986.
9. Mozes, M.S., Mangal, R., and Thampi, R., "Sorbent Injection for SO2 Control: (A) Sulphur Capture by Various
Sorbents and (B) Humidification. Ontario Hydro Research Report No 88-63-K, July 1988.
10. Helfritch. D.J., et al., "Electrostatic Precipitator Upgrades for Furnace Sorbent Injection", EPRI Final Report
GS 6282, April 1989.
11. Altman, R.F., "Precipitation of Particles Produced by Furnace Sorbent Injection at Edgewater", 8th Symposium on
the Transfer and Utilization of Paniculate Control Technology, March 1990, San Diego, California.
7A-47
-------
Stack
Sorbent.
Slurry
+
Additives
Ln
Esp
Disposal
a) Schematic of SONOX Process
Heat
*•
Water Drop
Evaporation
Heat>
0 Calcination
Limestone Slurry
Atomization
Dry Limestone
Particles
Particle
Disintegration
-Calcination
-High Pore Structure
Development
-Sintering Process Avoided
Sulphation
b) Chemical and Physical Steps
FIGURE 1
SONOX PROCESS
7A-48
-------
CD
Furnace
2) Burner Assembty
3) Air Supply
7) Heat Exchangers
?) Fttter Unit & Coal Bin
6) Door To Control Room
7) Resisttvity Housing
(a) Electrostatic Predpltator
(9) To Exhaust
(to) Propane Gas Control
(ft) Sortwnt Injection System
(fg) Isoklnedc Sampling System
(g) Water Injection System
(u) Furnace Quenching Pipe*
(1%) HumidifcaBon Chamoer
FIGURE 2
COMBUSTION RESEARCH FACILITY
-------
Air In
ft
Cooling
Water In
| J Positive Displacement ^ |
Recirculating Pump
Stanc Mixer
FIGURE 3
SONOX HARDWARE
Slurry In
Air In
Water In
Water Out
Internal
Mixing Chamber
FIGURE 4
TURBOTAK "EXTENDED" NOZZLE
7A-50
-------
70
Ca/S
3.0
•
•
•
2.2-2.5
o
o
A
a
Coal
US
US-W.Can.
N.S.
US
Sorbent
PA
PA
N.S.
B.
I
Q.
re
O
C/3
60
50
40
10 20 30 40 50
Porosity
FIGURE 5
CAPTURE VS LIMESTONE POROSITY
60
7A-51
-------
en
IV)
12
Turbotak 3 mm Nozzle,
40 % Apueou* Slurry ol Pt. Anne Limestone
10
E
o>
y 6
Q.
O
30 40 50 60
Atomizing Air Pressure, psig
a) Droplet Size vs Atomizing Air Pressure
70
Ca/S - 3.0
Slurry Rowrata 70 ml/min
70
O
65
60
2468
Slurry Droplet Size, \im
b) SO2 Capture vs Droplet Size
10
12
FICURE 6
SO2CAPTURE VS SLURRY DROPLET SIZE (ATOMIZING MEDIA - AIR )
-------
Ul
80
70
« 60
o
8"
5?
40
30
— »—
Coal
U.S.
U.S.
U.S.
U.S-WC
Nova Scotia
Nova S
-------
•-J
>
I
01
75
100
90
80
70
£ 60
O* 50
88 40
30
20
10
US Coal (1.7% S)
Initial NO, Cone. - 500 - 550 ppm
A, SR . 2.0
B, SR . 2.0
C, SR - 2.0
900 1000
1100 1200 1300
Temperalure,°C
1400
o
x
100
90
80
70
60
50
40
30
20
US Coal (1.7% S)
Initial NOX Cone. . 500 - 550 ppm
Ca/S Ratio - 3.0
AddvNO Stoichiometry . 1.7
Dolomite Doped P.A. Limestone
P.A Limestone
• SO 2 Removal
° NO x Removal "
100
90
80
0)
70 3
CL
co
60 o
ox
50 Z
5?
40
30
20
" Effective NOX Removal 63-80 % due to N2O formation
900 1000 1100 1200 1300 1400
Injection Temperature, °C
Effective NOX Removal 63-80 % due to N2O formation
FIGURE 7b
NO- CAPTURE - EFFECT OF INJECTION
TEMPERATURE
FIGURE7c
SONOX PROCESS
SO>/NO. CAPTURE - EFFECT OF INJECTION
TEMPERATURE
-------
Ul
en
40 % Aqueous Slurry
Co Current Injection
Droplet MMO - 6fim
-o- 1200°C
—•- 1300°Q
2 3
Ca/S Ratio
BHL -Beachville Hydrated Lime
B • Beachville Limestone
PA • Pt. Anne Limestone
US -U.S. CoaJ
65
60
55
c 50
g
To 45
N
^ 40
en -3C
O J0
5s 30
25
20
15
U.S. - WC - U.S. Western Canadian Coal Blend
D * Dolomite
NSC - Nova Scotia Coal
.-US
D-US
PA-NSC
PA-US
0 PA - US - WC
OB-US
1
Ca/S Ratio
FIGURE 8a
SO2 CAPTURE - EFFECT OF Ca/S RATIO
-------
Ul
05
U.S. Coal
Initial NOX Cone. « 500 - 550 ppm
Injection Temperature - 1100 °C
23
Additive Stoichiometry
-------
en
100
80
(0
§
DC
ox
-------
100
40% Slurry
_ 1.7% S Eastern US Coal
Ca/S-3:1
2
a.
a
Ft. Anna Slurry
I
2.5 5.0 7.5 10.0 12.5 15.0 17.5 20.0
Sieve Opening, (am
FIGURE 10
PARTICLE SIZE DISTRIBUTION OF BASELINE
FLYASH AND SLURRY WASTES
7A-58
-------
en
CD
TABLE I
CHARACTERISTICS OF COALS
Proximate Analysis, g/kg
Ultimate Analysis, g/kg
Moisture
Ash
Volatile Mattel
Fixed Carbon
Healing Value
MJ/kg
US
Coal
14
80
357
548
32
Nova Sea 13
Coal
12
96
314
577
31
W Can US
Coal Blend
32
99
321
548
29
UNOCAL
Coal
33
135
367
465
28
Carbon
Hydrogen
Nitrogen
Sulphur
Ash
Oxygen
US
Coal
756
57
16
17
80
74
NovaScoiia
Coal
756
50
12
28
96
58
W Can US
Coal Blend
751
47
14
11
71
101
UNOCAL
Coal
673
50
15
5
135
122
TABLE 2
CHEMICAL AND PHYSICAL PROPERTIES OF SORBENTS
LJ2O. g/kg
NajO
K2O
MgO
CaO
F«2°3
AI203
Si02
Ca(OH)2
LOI
BET area, nf/g
MMD. |im
p. g/cm
POROSITY. %
Beach villa
Limestone
<003
003
06
80
5240
01
220
<12.0
4340
1.3
86
26
170
Pi. Anne
Limestone
.
0.1
05
48
5354
2.1
43
21.0
29
39
2.3
550
Mosher
Limestone
(Nova Scotia)
0.1
l.t
538.0
64
65
25.0
1 86
110
25
570
E.G. King
Dolomite
.
04
<\0
212.1
300.9
2.3
10
7.9
4610
06
33.0
25
420
Beachville
Hydrated
LJme
0002
03
<005
7.6
1410
1.5
22
5.1
7880
126
82
2.1
264
TABLE 3
WASTE COMPOSITION
1.7% S US Coal with Limestone Sluuy
Ca/S = 2.5
Temperature = 1200 °C
CaO
CaSO<
CaCOj
MgO
LCH
kg Waste/I 00 kg Coal
Baseline
9*9
34
39
30
9
47
88
Sorbent Slurry Injection Waste
g/kg
302
257
44
50
17
Dry Injection Waste
g/kg
316
220
36
44
17
TABLE 4
WASTE RESISTIVITIES AND ESP PERFORMANCE
Injection Temperature 1200 °C (Dry = 1100 °C)
40% Limestone Slurry
Coal
U.S.
US.
US.
us
Sorbent
Beachville Limestone
(Slurry)
Pt. Anne Limestone
(Slurry)
BeachvHIo (Dry)
SO2
Removal
%
55
62
43
Flue Gas
Relative
Humidity
%
8-10
23
-19
8-10
Ash
Resistivity
ohm-cm
83 x 107
47 x 109
l.lxtO10
1 1 x 10n
ESP
Performance
Efficiency
%
89
88
87
at
-------
PILOT PLANT TEST FOR THE NOXSO FLUE GAS TREATMENT SYSTEM
L.G. Neal
Warren T. Ma
NOXSO Corporation
P.O. Box 469
Library, Pennsylvania 15129
Rita E. Bolli
Ohio Edison
76 South Main Street
Akron, Ohio 44308
-------
PILOT PLANT TEST FOR THE NOXSO FLUE GAS TREATMENT SYSTEM
L.G. Neal
Warren T. Ma
NOXSO Corporation
P.O. Box 469
Library, Pennsylvania 15129
Rita E. Bolli
Ohio Edison
76 South Main Street
Akron, Ohio 44308
ABSTRACT
The NOXSO process is a FGT system that employs a reusable sorbent. A fluidized bed of sorbent
simultaneously removes SO2 and NOX from flue gas. The spent sorbent is regenerated by treatment
at high temperature with a reducing gas. Adsorbed NOX is evolved on heating the sorbent to
regeneration temperature. The concentrated stream of NOX produced is returned to the boiler with the
combustion air.
NOXSO Corporation, MK-Ferguson, W.R. Grace & Co., and Ohio Edison will conduct a pilot test
of the NOXSO system at Ohio Edison's Toronto station. The plant treats 12,000 SCFM of flue gas
containing 2300 ppm SO2 and 350 ppm NOX, which is roughly 1/20 the size of a commercial module.
The paper summarizes the system design.
An additional test of the NOX recycle concept will be conducted at the Babcock & Wilcox Research
Center in Alliance, Ohio. The test apparatus is a 6 million Btu/hr small boiler simulator. It is a
scaled-down version of B&W's single cyclone front wall fired boiler design. The proposed test plan
and the data from previously reported NOX reduction tests using a pc-fired system at the Pittsburgh
Energy Technology Center are included.
7A-63
-------
INTRODUCTION
The NOXSO Process simultaneously removes SO2 and NOX from the flue gas of coal-fired boilers
using a dry, regenerable sorbent. Three previous tests of the NOXSO Process have been conducted.
The first was a bench-scale program conducted at TVA's Shawnee Steam Plant for the purposes of
establishing process chemistry and kinetics, quantifying sorbent attrition rates, and establishing the
corrosion properties of different metals for use in specific applications within the NOXSO Process.
The kinetic tests were all performed in a fixed bed reactor (1.2). Funding was provided by NOXSO
and by the U.S. Department of Energy's (DOE) Pittsburgh Energy Technology Center (PETC). The
second and third test programs were funded and conducted by DOE at PETC with technical guidance
provided by NOXSO Corporation. The second test program was designed to test laboratory data in
a scaled-up process, 3/4 MW in size (3). The third test program was a life-cycle test to determine
sorbent physical and chemical performance over repeated cycles of adsorption and regeneration (4).
The current test program is a 5 MW pilot plant that will provide the data necessary to scale up to a
full size (100 MW) module (5). The pilot plant is currently under construction at Ohio Edison's
Toronto Station and is scheduled to begin operation in May 1991. NOXSO Corporation is responsible
for operation of the pilot plant while funding comes from DOE, the Ohio Coal Development Office,
NOXSO Corporation, W.R. Grace & Co., and MK-Ferguson Co. A brief comparison of these four
test programs is given in Table 1. Detailed information on test facility design, test results, and data
analysis can be obtained from the previously referenced reports.
PROCESS DESCRIPTION
The NOXSO Process is a post-combustion flue gas treatment technology that simultaneously removes
both SO2 and NOX from the flue gas generated by coal-fired utility boilers. The process utilizes a high
surface area 7-alumina substrate impregnated with sodium to achieve removal efficiencies of 90% for
SO2 and 70%-90% for NOX. A process flow diagram is shown in Figure 1, and a description of the
process is given below.
Flue gas leaving the boiler passes through the combustion air preheater, the electrostatic precipitator,
and into the NOXSO flue gas treatment system. The flue gas is first cooled to 120°C by vaporizing
a water stream sprayed directly in the ductwork. The cooled flue gas is then passed through a
fluidized bed of sorbent where the SO2 and NOX are simultaneously adsorbed. The clean flue gas
7A-64
-------
flows through a cyclone where attrited sorbent is separated and returned to the adsorber bed. Finally,
the flue gas is returned to the power plant duct and exhausted through the stack.
After the sorbent is loaded with SO2 and NOX, it is removed from the adsorbers and pneumatically
conveyed to a sorbent heater. The sorbent heater is a three-stage fluidized bed where a hot air stream
is used to heat the sorbent to 660°C. During the heating process, NOX and loosely bound SO2 are
desorbed and transported away in the heating gas stream. The hot air stream exiting the sorbent heater
is recycled back to the boiler replacing a portion of the combustion air while providing an energy
credit to the NOXSO Process. At normal boiler operating conditions, the recycled NOX will either be
reduced by hydrocarbon fuel or suppressed by the formation of additional NOX so that a steady-state
equilibrium concentration of NOX is attained.
Once the sorbent reaches a regeneration temperature of 660°C, it is transported from the sorbent
heater to a moving bed regenerator. In the regenerator, sorbent is contacted with natural gas in a
countercurrent fashion. The natural gas reduces sulfur compounds on the sorbent (mainly sodium
sulfate) to primarily SO2 and H2S with some COS also formed (less than 2% of total inlet sulfur).
Approximately 48% of the sodium sulfate is reduced to sodium sulfide which must subsequently be
hydrolyzed in the steam treatment vessel. The moving bed steam treatment is obtained from the
reaction of steam with Na2S. The regenerator off-gasses are sent to a Glaus plant where SO2 and H2S
are reacted to form elemental sulfur. The sulfur is sold as a by-product of the NOXSO Process.
From the steam treatment vessel, the sorbent is fed to a sorbent cooler. The cooler is a three-stage
fluidized bed where the sorbent is cooled to 120°C using an ambient air stream. The warm air exiting
the cooler is further heated in a natural gas fired heater before being used to heat the sorbent in the
fluidized bed heater. The cooled sorbent is returned to the adsorber completing one full cycle.
PROCESS CHEMISTRY
The NOXSO sorbent is prepared by spraying Na2CO3 solution on the surface of 7-alumina sphere (1.6
nominal diameter). Both sodium and alumina contribute to the NOXSO sorbent's capacity to adsorb
SO2 and NOX from flue gas. Our laboratory tests show that the presence of steam in the flue gas helps
the SO2 sorption. The reaction of the sodium can be described as follows:
7A-65
-------
Na2C03 + A1203 2NaAl02 + CO2
2NaAlO2 + H2O <—> 2NaOH + Al2O3 (2)
2NaOH + S02 + -O2 — > Na2SOt + H2O (3)
2NaOH + 2NO + — O2 <—> 2NaNO3 + H2O (4)
2NaOH +• 2NO2 + — O2 — > 2NaNO3 + H2O (5)
Zj
Adsorbed nitrogen oxides are decomposed and evolved on heating the spent sorbent to regeneration
temperature. The concentrated stream of NOX produced on heat-up is returned to the boiler with the
combustion air. This results in no significant increase of NOX concentration in the boiler flue gas
because of the reversibility of NOX formation in the boiler (1.2).
The spent sorbent can be regenerated at high temperature with many kinds of reducing gases, such as
H2S, CO, H2, natural gas, etc. The regeneration reaction, for example, using natural gas at 610°C
is described below:
O2 + CO2 + 2H2O (6)
4Na2SO3 + 3CHi — > 4Na2S + 3CO2 + 6H2O (7)
A1203 + Na2S03 <—> 2NaAlO2 + SO2 (8)
A12O3 + Na2S + H2O <—> 2NaAlO2 + H2S (9)
Although sulfite has not been identified in our studies, it is a likely intermediate in sulfate reduction.
A detailed discussion on the existence of sulfide during regeneration had been given by Gavalas it.al.
(6) who used CO to study the regeneration of alkali-alumina. The SO2 and H2S produced from
regeneration are then converted to elemental sulfur in a Claus-type reactor.
7A-66
-------
S02 + 2H2S <— > XS3/X + 2H20 (10)
The sulfur produced is a marketable by-product of the process.
PROOF-OF-CONCEPT PILOT TEST
Background
On May 10, 1989, a consortium assembled by NOXSO Corporation signed a cost-shared contract with
the DOE/PETC to conduct a POC test of the NOXSO process. The consortium consists of NOXSO,
MK-Ferguson, W.R. Grace & Co., Ohio Edison and the Ohio Coal Development Office. The POC
project will take approximately three years to complete, and the test will be conducted at a coal-fired
Ohio Edison plant in Toronto, Ohio.
POC Test Site
The POC unit will treat flue gas from either Boiler #10 or Boiler #11 at Ohio Edison's Toronto
Station. Two sources of flue gas will be tapped so that the POC test can continue as long as one of
the boilers is operating. A slipstream of flue gas will be taken downstream of the Toronto plant's
electrostatic precipitators. The Toronto boilers are pc-fired and burn Ohio coal containing 3.7%
sulfur. The flue gas typically contains 2300 ppm SO2 and 350 ppm NOX.
POC Test Schedule
Detailed design engineering has been completed and the major pieces of equipment have been ordered.
Construction began in April 1990 and will be completed in May 1991. The test will run through
February 1992.
POC Process
The process flow diagram for the POC has shown in Figure 1. The system differs from a commercial
application of the NOXSO technology in two important areas. First, the POC facility does not include
a Claus plant, which in the commercial design would be used to produce a sulfur by-product from the
concentrated stream of SO2 and H2S produced in the regenerator. This is because Claus technology
is commercially available and therefore does not require testing at pilot scale. Second, the POC does
not include NOX recycle to the coal combustor. In the commercial design, NOX in the air leaving the
sorbent heater is recycled to the combustor as part of the combustion air. Since NOX formation in the
7A-67
-------
coal combustor is a reversible reaction, addition of NOX to the combustion air suppresses the formation
of NOX in the combustor. However, NOX recycle is impractical in the POC test since the POC treats
less than 10% of the flue gas produced by Toronto Unit 10 or 11.
POC Test Unit Design
Data from three previous tests of the NOXSO process were used to design the POC test facility. A
comparison of the three previous test programs was given in Table 1. The design specifications for
the major equipment in the POC test facility are listed in Table 2.
Materials of Construction
During development of the NOXSO process, some corrosion problems were encountered, particularly
in the regenerator. Different materials of construction were tested to withstand the high temperature
environment of SO2, H2S, elemental sulfur, and sulfated sorbent. Corrosion results were documented
in an earlier report (2), the practical results of the test program are discussed here.
In tests performed at the Shawnee Steam Plant, sorbent was heated with electrical resistance heaters
made of Inconel 600, Monel 400, type 316 and type 316L stainless steel (SS). All these materials
exhibited severe corrosion in areas of sorbent contact attributed to hot sulfation of nickel. It should
be noted that the temperature of the heating elements themselves were substantially higher than the bed
temperature of 600°C. The reactor, made out of either type 316 or type 316L SS, showed scale on
the inside surfaces after use. When the reactor was made of type 446 SS or alonized type 316L SS,
there was no scale and only a slight discoloration of the metal surfaces observed.
In the LCTU, the regenerator was made of alonized type 304 SS and showed no visible evidence of
corrosion at the end of 330 regeneration cycles. Based on these results, it was felt that either 446 SS
or alonized 304 or 316L SS would be satisfactory for the POC regenerator.
The sorbent heater also encounters hot sulfated sorbent and will therefore be made of type 304 SS.
The bottom bed of the sorbent heater where the temperature is 660°C will be alonized. All other
vessels will be made of standard A-285 or A-283 grade C carbon steel, as no corrosion problems are
anticipated.
7A-68
-------
The other area in the NOXSO process that requires special consideration for materials is between the
flue gas cooler and the adsorber. In this area, sub-acid dewpoint corrosion can occur. All previous
NOXSO tests have cooled the flue gas indirectly while at the POC the flue gas will be cooled by a
direct water spray in the ductwork. The flue gas temperature in this portion of the system will be
112°F so that an acid-resistant epoxy coating will be used to line the ductwork from the cooler to and
including the bottom of the adsorber. This epoxy has not been tested previously by NOXSO, but there
exists ample literature that supports its use as an acid resistant material in other similar applications.
NOX RECYCLE
NOX recycle will be implemented at the full-scale commercial demonstration plant. The concept of
NOX recycle has been tested previously using the 500 Ib/hr coal combustor used for the 3/4 MW tests
and also using a tunnel furnace capable of being fired with a variety of fuels including gas, fuel oil,
coal, and coal-water mixtures.
Previous NOX Recycle Results
NOX recycle was tested by spiking the combustion air with varying concentrations of bottled NOX and
measuring the outlet NOX concentration from the combustor. The net NOX production rate was
determined by a material balance on the combustor as shown schematically in Figure 2. The NOX flow
rate at the exit of the combustor minus the NOX feedrate into the combustor equals the rate that NOX
is produced in the combustor, which is defined as the net NOX production rate (R). For data reduction
purposes, the NOX production rate (R) and the NOX feedrate (F) were normalized with respect to
conditions at zero NOX feed according to R*=R/R0 and F*=F/R0 where R0 is the NOX production rate
at F = O. Results from the 500 Ib/hr combustor are compiled in Table 3. The measured data are
NOX concentration at the exit of the combustion system and the flow rate of NOX fed into the
combustor with the combustion air. Data provided in the other columns were calculated.
A plot of R* versus F* is shown in Figures 3 and 4 for both the 500 Ib/hr combustor and the tunnel
furnace, respectively. In each case, the data fall in a straight line, but with different slopes. The two
lines are described by the equation R* = 1 - aF*. The parameter "a" is the slope of the line and also
represents the fraction of NOX fed to the combustor that is destroyed, The value of "a" is 0.65 for
the 500 Ib/hr combustor and 0.75 for the tunnel furnace. The data for the tunnel furnace include both
natural gas combustion and coal-water slurry combustion.
7A-69
-------
These results demonstrate that the nature of the fuel has no affect on the effectiveness of the
combustion system to reduce NOX fed through the combustion air. Also, the NOX reduction capability
of a combustion system is independent of the amount of NOX fed with the combustion air. Finally,
the most important variables are those associated with the combustor design. NOX recycle will be
extensively studied at the Babcock & Wilcox Research Center in Alliance, Ohio.
Pilot-Scale NO. Recycle Test
The power plant selected for the NOXSO full-scale demonstration (Ohio Edison's Niles Station, Niles,
Ohio) uses cyclone burners. Since the destruction efficiency of NOX recycle has not previously been
tested with cyclone type burners, a demonstration of NOX recycle with this type of coal combustor is
necessary for the proper design of the NOXSO full-scale plant.
Pilot-scale NOX recycle tests will be done using Babcock & Wilcox's 6 million Btu/hr Small Boiler
Simulator (SBS) shown in Figure 5. The water-cooled furnace is a scaled-down version of B&W's
single-cyclone, front-wall fired boiler design. The cyclone has been in operation since February 1985.
The SBS cyclone furnace simulates a large cyclone unit very well. A comparison between the SBS
cyclone furnace and commercial units is given in Table 4.
The NOX recycle tests will begin with three loads and three excess air levels to establish the baseline
of the NOX emission from the SBS furnace. NO will then be injected in multiples of the baseline NOX
production levels. The NO concentration at the air inlet duct to the cyclone will be measured to
document the inlet level. Stack NOX will be measured to determine NOX destruction occurring in the
flame. The series of tests with different NO injection rates will also be performed at three furnace
loads and three excess air levels. This test result will assist the determination of a second injection
location for the next series of tests.
In the second series of tests, NO and NO2 will be injected separately for two furnace loads and two
excess air levels. Volumetric flowrate of the injected NO and NO2 will be based on the proportion
of these gasses that are present in the NOXSO sorbent heater off-gas. The addition of methane to the
air stream to assist the NOX destruction (7) will also be tested. The NOX recycle test will be finalized
by burning the coal from the Niles plant in the SBS furnace. Since the coal-ash slagging
characteristics are important to the power plant operation, the use of Niles plant coal will assess the
change of the coal ash's "flowability" in the Niles plant when the NOX recycle stream is installed.
7A-70
-------
FUTURE WORK
On December 21, 1989, NOXSO Corporation, in association with MK-Ferguson Company, W.R.
Grace & Co., and Ohio Edison, received an award from DOE's Clean Coal Technology Program to
conduct a $66 million, full-scale commercial demonstration of the NOXSO technology. The U.S.
DOE will provide $33 million and the remaining funds will be provided by the Ohio Coal
Development Office, the Electric Power Research Institute, the Gas Research Institute, the East Ohio
Gas Company, and the aforementioned NOXSO development team. The 115 MW demonstration plant
will be installed at Ohio Edison's Niles Power Plant in northeastern Ohio. Construction is scheduled
to begin in early 1993 with plant startup scheduled in May 1994. This project is the final step in the
program to commercialize the NOXSO technology.
REFERENCES
1. J.L. Haslbeck, CJ. Wang, L.G. Neal, H.P. Tseng, and J.D. Tucker. Evaluation of the NOXSO
Combined NOX/SO2 Flue Gas Treatment Process. NOXSO Corporation Contract Report
submitted to U.S. DOE Report No. DOE/FE/60148-T5. November 1984.
2. J.L. Haslbeck, L.G. Neal, CJ. Wang, and C.P. Perng. Evaluation of the NOXSO Combined
NOX/SO2 Flue Gas Treatment Process. NOXSO Corporation Contract Report submitted to U.S.
DOE Report No. DOE/PC/73225-T2. April 1985.
3. J.L. Haslbeck, W.T. Ma, and L.G. Neal. A Pilot-Scale Test of the NOXSO Flue Gas Treatment
Process. NOXSO Corporation Contract Report submitted to U.S. DOE Contract No. DE-FC22-
85PC81503. June 1988.
4. J.L. Haslbeck, J.T. Yeh, W.T. Ma, J.P. Solar, and H.W. Pennline. Life-Cycle Test of the
NOXSO Process: Simultaneous Removal of NOX and SO2 from Flue Gas. Presented at the 1989
AWMA Annual Meeting, Anaheim, California. June 1989.
5. J.L. Haslbeck, M.C. Woods, R.E. Bolli, R.L. Gilbert, and C.P. Brundrett. Proof-of-Concept
Test of the NOXSO Flue Gas Treatment System. Presented at the EPA/EPRI 1990 SO2 Control
Symposium. New Orleans, Louisiana. May 8-11, 1990.
6. G.R. Gavalas, S. Edelstan, M. Flytzani-Stephanopoulous, and T.A. Weston. Alkali-Alumina
Sorbents for High-Temperature Removal of SO2. AIChE Journal Vol. 33, No. 2, p. 258. 1987.
7. J.T. Yeh, J.M. Ekmann, H.W. Pennline, and CJ. Drummond. New Strategy to Decompose
Nitrogen Oxides from Regenerable Flue Gas Cleanup Processes. Presented at the 194th ACS
National Meeting. New Orleans, Louisiana. Aug. 30 Sept. 4, 1987.
7A-71
-------
NOx RECYCLE
TO CLAUS PLANT
REGENERATOR
AIR
NOXSO PROCESS FLOW DIAGRAM
FIGURE 1
7A-72
-------
-t-—
I E,
Adsorber
R+F
Combustor
FIGURE 2. SCHEMATIC DIAGRAM OF NITROGEN
OXIDE RECYCLE.
7A-73
-------
CC
bJ
*
CC
o
f-
o
o
o
cr
QL
o
LJ
cc
o
-2.0
1.0 2.0 3.0
NORMALIZED NOx FEED RATE, F*
4.0
FIGURE 3. NORMALIZED NOx REDUCTION
DATA-PC COMBUSTOR.
*
OC
UJ
CC.
o
h-
o
o
o
oc.
o.
x
O
z
o
N
QL
O
+ 1
0
-I
-2
-3
-4
-5
-6
-7
-8
0 5 10 15
NORMALIZED NOX FEED RATE, F*
FIGURE 4. NORMALIZED NOx
REDUCTION DATA-
TUNNEL FURNACE.
7A-74
-------
STACK
STEAM
REHEATER
DEPOSITION —
PROBE
SUPERHEATER
FOULING TUBE
DEPOSITION PROBE
FLUE GAS
RECIRCULATION
FURNACE ARCH
PRIMARY AIR
AND COAL
TERTIARY AIR
SECONDARY
AIR
SLAG TAP
MOLTEN SLAG
SLAG COLLECTOR
AND FURNACE
WATER SEAL
FIGURE 5. SMALL BOILER SIMULATOR (SBS) SCHEMATIC
7A-75
-------
Table 1. Comparison of NOXSO Test Programs
Operating Parameter
Coal Burned, Ibs/hr
Flue Gas Volume, SCFM
Adsorber Type
SO2 Inlet Concentration, ppm
NOX Inlet Concentration, ppm
SO2 Removal Efficiency, %
NOX Removal Efficiency, %
Reducing Gas for Regeneration
Operating Mode
Test
TVA
NA
0.35
Fixed Bed
2300
600
90
90
H2S, H2, CO
Batch
Program
3/4 MW
500
1200
Fluid Bed
1465-5000
470-720
90-99*
80-92*
H2, H2+CO, CH4
Batch
Test Program
Operating Parameter
Coal Burned, Ibs/hr
Flue Gas Volume, SCFM
Adsorber Type
SO2 Inlet Concentration, ppm
NOX Inlet Concentration, ppm
SO2 Removal Efficiency, %
NOX Removal Efficiency, %
Reducing Gas for Regeneration
Operating Mode
LCTU
40
120
Fluid Bed
1450-3000
240-800
60-90*
60-90*
H2, CH4
Continuous
POC
NA
12000
Fluid Bed
2300
350
**
**
Natural Gas
Continuous
NA = Not applicable, i.e., small slipstream was drawn from the power plant ductwork.
* = In the 3/4 MW and LCTU tests, removal efficiencies cover a wide range since
operating conditions were intentionally varied to study their effect on process
performance.
** = Pilot plant is under construction.
7A-76
-------
Table 2. POC Major Equipment Specifications*
Fluidized Bed Adsorber
Diameter 10.5 ft
Temperature 120°C
Settled Bed Height 2 ft
Sorbent Residence Time 45 min
Superficial Gas Velocity 3 ft/s
Transport Disengaging Height 7.7 ft
Material of Construction Carbon Steel
Fluidized Bed Sorbent Heater
Number of Stages 3
Diameter 7.7 ft
Settled Bed Height 0.9 ft
Sorbent Residence Time 30 min
Superficial Gas Velocity 3 ft/s
Transport Disengaging Height 2.8 ft
Material of Construction Type 304 SS
Fluidized Bed Sorbent Cooler
Number of Stages 3
Diameter 5.7 ft
Settled Bed Height 1.2ft
Sorbent Residence Time 20 min
Superficial Gas Velocity 3 ft/s
Transport Disengaging Height 4.3 ft
Material of Construction Carbon Steel
Moving Bed Regenerator/Steam Treater
Diameter 4 ft
Bed Height 10.3 ft/6.8 ft
Sorbent Residence Time 30 min/20 min
Material of Construction Alonized Type 304H SS
Air Heater
Design Flow (Air)
Temperature Rise
Type
Pneumatic Conveyor
Sorbent Circulation Rate
Lift Distance
Adsorber Cyclone
D-50
D-100
Gas Flowrate
* At base case operating conditions.
14,300 Ibs/hr
330°C
Natural gas fired in duct burners
9,994 Ibs/hr
80ft
20/xm
100 MHI
16,257 ACFM @ 120°C
7A-77
-------
Table 3. NOX Reduction Data; 500 Ib/hr Combustor (3)
Test >
No.#
1
2
3
4
5
6
7
8
9
10
Tests 1
Tests 4
Tests 7
F
JOxExit NOxExit NOX Fed R
ppm Ih/hr Ib/hr Ib/hr R* F*
550 3.59
1370 8.94
875 5.71
650 4.24
850 5.55
930 6.07
700 4.56
1100 7.17
1200 7.82
820 5.34
0 +3.59 1.0
14.09 -5.15 -1.43
8.29 -2.58 -0.72
0 +4.24 1.0
4.66 +0.89 0.21
5.49 +0.58 0.14
0 +4.56 1.0
6.64 +0.53 0.12
7.98 -0.16 -0.04
1.60 +3.74 0.82
3. Coal feedrate = 223 Ibs/hr, Flue gas flowrate =
moles/hr (dry), and Temperature = 2500°F.
6. Coal feedrate = 352 Ibs/hr, Flue gas flowrate =
moles/hr (dry), and Temperature = 2500°F.
10. Coal feedrate = 431 Ibs/hr, Flue gas flowrate =
moles/hr (dry), and Temperature = 2500°F.
0
3.92
2.31
0
1.10
1.29
0
1.46
1.75
0.35
122.1
160.0
180.4
Table 4. Comparison of Baseline Conditions Between
the SBS Facility and Commercial Units
Cyclone Temperature
Residence Time at full load
Furnace Exit Gas Temperature
NOx Level
Ash Retention
Unburned Carbon
Ash Particle Size (MMD; Bahco)
SBS
>3000°F
1.4 sec
2265 °F
900-1200 ppm
80% -85%
< 1 % in Ash
6-8 microns
Typical
Cyclone-Fired Boilers
>3000°F
0.7-2.0 sec
2200°-2350°F
600-1400 ppm
60% -80%
1%-20%
6-11 microns
7A-78
-------
THE PRACTICAL APPLICATION OF TUNABLE DIODE LASER INFRARED
SPECTROSCOPY TO THE MONITORING OF NITROUS OXIDE AND OTHER
COMBUSTION PROCESS STREAM GASES
Frank E. Briden
Air and Energy Engineering Research Laboratory
U.S. Enviornmental Protection Agency
Research Triangle Park, North Carolina 27711
David F. Natschke
Richard B. Snoddy
Acurex Corporation
4915 Prospectus Drive
Durham, North Carolina 27713
-------
THE PRACTICAL APPLICATION OF TUNABLE DIODE LASER INFRARED
SPECTROSCOPY TO THE MONITORING OF NITROUS OXIDE AND OTHER
COMBUSTION PROCESS STREAM GASES
Frank E. Briden
Air and Energy Engineering Research Laboratory
U.S. Environmental Protection Agency
Research Triangle Park, North Carolina 27711
David F. Natschke
Richard B. Snoddy
Acurex Corporation
4915 Prospectus Drive
Durham, North Carolina 27713
ABSTRACT
There are a number of gases associated with combustion process streams which are difficult
to monitor because of their physical properties and interferences from other gases. Tunable
diode laser infrared (TDIR) spectroscopy offers a reliable, specific means for the continuous
monitoring of many of these gases. Some of the gases that can be efficiently monitored by
this technique are N2O, NO, NO2, H2O, H2O2, O3, NH3, HCN, SO2, SO3, OCS, CO2, CO,
HCI, HBr, HF, CH3CI, CH4, CH3OH, and C2H5OH, to name a few.
This technique requires the use of sophisticated electronic components, but provides an
extremely rugged, simple to operate, stable, sensitive, and reliable instrument. This paper
describes how the Air and Energy Engineering Research Laboratory of the Environmental
Protection Agency at Research Triangle Park, NC, designed, built, and tested, with a coal
burning furnace, a TDIR monitor for N2O. The present diode mount is limited to the
simultaneous use of only two 2 diodes and therefore only two analyte gases per optical cell.
Newer mounts allow the simultaneous use of four diodes. The conversion of the system for
other gases will be described. TDIR in-stack monitoring and long-range atmospheric
monitoring will also be discussed.
7A-81
-------
INTRODUCTION
The measurement of atmospheric N2O and its sources is of great interest since it is a
potential contributor to global warming and its atmospheric concentration is increasing. The
principal sampling method uses an evacuated container to collect a grab sample of the gas
stream of interest, so the containers could be taken back to a laboratory and analyzed later.
The original data indicated a linear relationship between the concentrations of N2O and NOx
in the stack gases. The validity of this data began to be questioned in the mid-1980s when
studies showed the detection of N2O when none was expected. Muzio et al. reported on the
formation of N2O as a sampling artifact while studying natural gas flames injected with SO2
and NH3. (1) Another report showed that the artifact could be reduced by drying the gas
before sampling, and the artifact could be effectively eliminated by removing the SO2. (2) It
was evident that a gas-phase aqueous reaction between SO2, NOX, and H2O was generating
N2O in the sample container. These reactions have been known since the 18th century and
reported as early as 1924. ^ Discovery of this sampling artifact led to research on the
development of sampling and analysis techniques which would provide accurate results.
One project in this area, by the Air and Energy Engineering Research Laboratory, used a
heated sample line and then filtered and desiccated the gas before it was analyzed by an on-
line GC/ECD (for N2O) and continuous emission monitors (for O2, CO2, CO, and NO). This
research indicated that the N2O concentrations were less than 5 ppm and were not a
function of the NOX concentration. ^
This project was undertaken to demonstrate the ability of a laser diode system to accurately
and correctly measure the concentration of N20 in stack gas in real time, and to verify the
lower N2O concentrations reported with modified sampling methods.
7A-82
-------
EXPERIMENT
The detection of trace gases using second-derivative spectroscopy was first used in 1978 by
Reid et al. at McMaster University. ^4^ Second-derivative or modulation spectroscopy
consists of using a modulated source to scan the absorption line of interest. The detector
output is amplified using a phase-sensitive amplifier referenced to twice the modulation rate.
In addition to significantly reducing the background noise by rejecting all signals which are
not in phase with the reference signal, operating the amplifier at twice the modulation rate
produces a pseudo-second-derivative signal as the output. This signal is proportional to the
absorption of the line being scanned but the signal must be calibrated for each line of
interest. A beamsplitter, lock cell, and a second detector are used to provide a feedback
signal to correct for any drift in the source. For the feedback circuit, a phase-sensitive
amplifier referenced to the modulation frequency reduces the noise level and provides the
stabilization signal. ^4' In this system, an infrared diode laser modulated at 2000 Hz was
used as the source. The output frequency of a diode laser can be broadly tuned by adjusting
its operating temperature and finely tuned by varying the applied current. This particular
diode is tunable over the range 2200 to 2215 cm"1. Figure 1 diagrams the optical system.
The cold head contains part of the cooling system for the diode and also provides an
insulating vacuum for the diode since it is operated at 28 K. A monochrometer is used to
isolate the laser line of interest. The beamsplitter deflects a portion of the laser light through
a lock-cell containing a high concentration of N2O, and into a detector to generate the
stabilization signal. The rest of the laser energy passes through the beamsplitter, into the
analytical cell, and then into the analytical detector. The analytical cell is a two-pass, 0.5 m
cell with an external retroreflector. Both detectors are single element mercury cadmium
telluride photoconductive detectors with low noise preamplifiers. The first and second
derivative signals are generated by setting the reference channel of the phase-sensitive
amplifier to either the "f" (first derivative) or "2f" (second derivative) mode. In the "2f" mode,
the reference channel of the phase-sensitive amplifier is driven internally at twice the input
frequency, eliminating the need for an external, stabilized 4000 Hz reference signal. The
output signal from the amplifier (for either mode) is a pure DC signal reflecting the magnitude
7A-83
-------
of the input signal which is in phase with the reference signal. Any AC component is the
result of noise in the system and is reduced by the AC filter at the output. This AC filter has
a variable time constant which can be adjusted from 1 ms to 100 sec. A higher setting of the
time constant will reduce the noise level, but will also eliminate the corresponding time
variations in the analytical detector signal. The output signal is then displayed on the chart
recorder. The change in the magnitude of the signal, as measured from the baseline
(determined using dry nitrogen gas), is directly proportional to the concentration of N2O in the
analytical cell.
Before beginning the tests, the N2O line with the least interference from the other gases
commonly found in stack gas (H2O, CO2, CO) at various pressures, temperatures, and
concentrations was determined. Theoretical spectra were calculated using the FASCODE
algorithm which was developed by the Air Force Geophysics Laboratory. (5) Examples of
these spectra are shown in Figures 2, 3, and 4. During this work, the gas pressure in the
analytical cell was maintained at 5000 Pa by continuously pumping on the outlet side of the
cell with a vacuum pump and limiting the flow at the cell inlet port. This kept the pressure-
broadening of the lines to a minimum and, during sampling of furnace gases, cooled the
furnace gases to reduce thermal-line-broadening. The line at 2208.75 cm"1 was chosen for
this work. Initial tests using mixed gases from cylinders verified the detection of N2O and no
response to the CO2, CO, SO2, and H2O vapor.
The equipment was moved from the laboratory and connected to the Innovative Furnace
Reactor (IFR), a furnace designed to evaluate various methods of scrubbing stack gases. It
is a down-fired, tunnel-fired furnace burning powdered coal. Figure 5 diagrams the system .
During these tests, the IFR was being used to evaluate the efficiency of powdered lime to
reduce SO2 emissions. The stack gases were sampled at two different positions (see Tables
1 and 2), one at the end of the furnace before the gas is filtered in the bag house, and the
other near the roof just before the gases were vented to the atmosphere. These are
indicated in Figure 5 as #1 and #2, respectively. The gases at the two sampling positions
are significantly different. At position # 1, the gases reflect the actual combustion products of
the furnace. After leaving the furnace, the gases are diluted and cooled to protect the bag
house filter elements and the roof-mounted blower from damage due to excessive heat.
Although the gases sampled at position # 2 reflect what is discharged to the atmosphere, the
gases have been diluted, cooled, and filtered.
7A-84
-------
The operating parameters of the TDIR system are listed in Table 1.
These operating parameters are typical for each sampling position, but the actual values
were adjusted slightly to optimize the system each day. The system was calibrated each day
using N2O in dry nitrogen at concentrations of 0.108, 0.514, 0.970, 1.99, and 4.82 ppm. A
sample of the data collected from sample position #2 is in Figure 6. This section of the chart
paper shows the time variations in the N2O concentration which is attributed to fluctuations in
the coal feed rate. Also visible are the areas where dry nitrogen is used to verify the
baseline. The addition of powdered lime to the stack had no effect on the measured N2O
concentrations. The average concentration is 0.9 ppm with a maximum excursion of 1.0 ppm
and a minimum of 0.8 ppm. Figure 7 shows data collected at sample position #1. There are
several differences evident in this chart. First, the level of N2O is much lower, about 0.3
ppm. Second, as the system is switched from sampling dry nitrogen to stack gas, there is a
spike in the N2O concentration which is not seen in the data from position # 2. Third, the
two spikes at the end of the trace are observed each time the coal feed is stopped and only
air is blown into the burner section of the furnace.
INTERPRETATION
The calibration data were fitted using a linear function to correlate a given deviation from the
baseline to concentration. The results are summarized in Table 2. These concentrations are
much lower than those in the original N2O database and are also lower than the more recent
data indicated. The higher concentrations in the stack at position #2 are caused by the
formation of N2O in the baghouse. The concentration is reduced, by dilution of the gas
stream in the baghouse and after the baghouse, to cool the gas before it is vented. The
data from position #1 is a more accurate measure of N2O produced by the furnace since it is
sampled before there is any chance of dilution and the gas temperature (300 °C) is high
enough to keep the water in vapor form. It is assumed that both the higher temperature and
the reduced time between sampling and analysis work to reduce the amount of N2O
generated as an artifact.
The spikes in the data from position #1 are the result of N2O generation in the filter unit and
the short section pipe connecting the heated sample line to the furnace. The filter and
connector pipe were not heated and would cool off when the furnace gases were not flowing
7A-85
-------
through them. This permits water condensation and the formation of N2O in these unheated
parts. When gas was subsequently drawn from the furnace, the small volume of gas in the
pipe and filter would precede the hot furnace gases into the analytical cell and cause a spike
in the output. The fact that this effect was not observed in the data from position #2
indicates that the gas components had already interacted, producing N2O, and could not
generate more N2O in the filter set. It is assumed that this reaction most likely took place in
the baghouse where the ash and lime reaction products were collected and the temperature
fell below 100 °C causing the water vapor to condense and initiate the reaction.
The fluctuations in the N2O concentration both during furnace operation and at the end,
when the coal feed unit was turned off, were well correlated to similar fluctuations in the
concentration of CO which was continuously measured as part of the SO2 scrubbing tests.
This may indicate that the N2O is a result of a lower concentration of oxygen in the furnace
which also generates more CO.
CONCLUSIONS
In this study, it was found that the N2O concentration, immediately after the combustor
(position #1, Figure 5) varied above and below ambient which was measured at 280 ppb.
However, conditions in the baghouse caused an increase of N2O up to about 3 times
ambient (position #2, Figure 5). The major source of N2O in the stack gas appears to be its
formation when the water vapor condenses and reacts with other components of the stack
gases.
This study also shows great promise for the use of laser diode modulation spectroscopy for
other applications where continuous monitoring of one or more trace gases is required. The
system is easily modified to monitor other gases by replacing the diode with one that will
operate in the spectral region of interest. By operating both diodes simultaneously and
adding more optical components, the current system can be configured to simultaneously
monitor two gases in the sample stream. There are also cold head systems available which
will allow the use of four diodes simultaneously and therefore the monitoring of four distinct
trace gases.
This method may also be used to directly measure species concentrations in the stack by
using optical windows mounted in the stack access ports (e.g., the sulfun'c acid
measurements of Pearson and Mantz. (6)). Measurements of atmospheric contaminants over
7A-86
-------
long path lengths are feasible and could provide significant information on the generation,
distribution, and dissipation of pollutants which are not generated from single sources. It is
proposed to use this technique to monitor methane emissions from landfills or pasture land.
7A-87
-------
REFERENCES
(1) L.J. Muzio, et al. "Errors in Grab Sample Measurements of N2O from Combustion
Sources." JAPCA, Vol. 39 No. 3, 1989, pages 287-293.
(2) L.J. Muzio and J.C. Kramlich. "An Artifact in the Measurement of N2O from
Combustion Sources." Geophysical Research Letters, Vol. 15 No. 12, 1988, pages
1369-1372
(3) W.P. Linak, et al. "N2O Emissions from Fossil Fuel Combustion." In Proceedings:
1989 Symposium on Stationary Combustion NOX Control, San Francisco, CA, March
6-9, 1989, Volume 1, EPA-600/9-89-062a (NTIS) PB89-220529), June 1989.
(4) R.S. Eng, et al. "Tunable Diode Laser Spectroscopy: An Invited Review." Optical
Engineering, Vol. 19 No. 6, pages 952-953
(5) FASCODE Fast Atmospheric Signature Code (Spectral Transmittance and
Radiance), H.J.P. Smith et al., AFGL-TR-78-0081 Air Force Geophysics Laboratory,
Air Force Systems Command, United States Air Force, Hanscom AFB, MA 01731.
(6) E.F. Pearson, A.W. Mantz. "A Tunable Diode Laser Stack Monitor for Sulfuric -Acid
Vapor." EPA-600/Z-80-174 (NTIS PB80-202 690), 1979.
7A-88
-------
Q.
if
O
>
Chart Recorder
Lock-in Amplifier
for Signal Analysis
Lock-in Amplifier
for Reference Analysis
Laser
Control
Module
Oscilloscope
Figure 1. Laser diode setup
7A-89
-------
-vl
CD
O
2200
2201
N20 1 PPM, 50 MB, 25 C
'9 8.0
7.0
6.0
I 5.0-
Q
^ 4.0-
£
O 3.0-
2.0-
1.0-
0.0-
n
11
I .11 .
I , 1
i • i • i • i < i •
2202 2203 2204
WAVENUMBER (CM'1)
Figure 2. N2O Spectrum at 25 °C
2205
2206
-------
N20 PPM, 50MB, 100 C
CO
? ™-j
6.0-
5.0-
~T~
& *o-
Q
S •
E 3-°-
0 :
2.0-
1.0 -i
0.0-
22
A ^^^^
111 . ll ll
I
L . ill Jv. i A Jv i
DO 2201 2202 2203 2204 22
D5
WAVFNUMBER (CM'1)
Figure 3. N2O Spectrum at 100 °C
-------
CD
o
10.0
N2Q. OQZ M20, CO AT 25 C, 50 MB
2208 2208.2
22084 2208.6 22088
WAVENUMBER (CM'1)
Legend
10X002
10 « too
1000 PPM CO
2209 2209.2
Figure 4. Combined Spectrum
-------
Coal Feeder
CD
CO
Sorbent Feeder
Sorbent/Slurry
Injection Probe
Sampling Ports
Rool
N2O Sampling Port #2 —
SO? Sampling Port
o
Q.
I
Q.
o
OJ
O
CEM Sampling Port I
Fan
Stack
1
Baghouse
Figure 5. Innovative Furnace Reactor
-------
CD
o
At
Baseline'
Figure 6. Position #2 N2O Data
-------
CD
cn
I I
N2 Baseline
Figure 7. Position #1 N2O Data
-------
TABLE 1. OPERATING PARAMETERS
Parameter
Current, mA
Temperature, K
Frequency, Hz
Scan Width", mA
Sensitivity, mV
Time Constant, sec
Sample Position #1
217
28
2000
5
0.01
3
Sample Position #2
217
28
2000
5
0.025
3
* A current scan width of 5 mA equates to a frequency shift of 0.75 cm"1
TABLE 2. OBSERVED N2O CONCENTRATION
Data Sample Position #1 Sample Position #2
ppm ppm
Average 0.30 0.74
Maximum 0.46 1.27
Minimum 0.14 0.75
(± 0.053) (± 0.025)
7A-96
-------
Session 7B
NEW DEVELOPMENTS
Chair: C. Miller, EPA
-------
IN-FURNACE LOW NOX SOLUTIONS
FOR WALL FIRED BOILERS
By
R.C. LaFlesh, D. Hart, and P. Jennings
ABB Combustion Engineering
Michael Darroch
City of Jamestown, New York
-------
ABSTRACT
Since the early 1940's, several thousand Type R pulverized coal burners have
been installed and are operating in wall fired boilers ranging up to 160 MWe
in size. In response to the low NOX Emission requirements, ABB Combustion
Engineering Services, Inc. has undertaken development of the RO-II coal
burner based on proven Type R wall firing technology with additional NOX
control capability.
Extensive laboratory tests were conducted at a large pilot scale (50 x 10
Btu/hr) in order to optimize the RO-II coal burner configuration.
Specifically, a number of coal nozzle/air register configurations were
evaluated in terms of their combined ability to meet specific emissions and
operational performance criteria. The RO-II burner reduced NOX from a
baseline uncontrolled level of 0.9 #/106 Btu to 0.5 #/106 Btu during the
laboratory trials.
This paper will review laboratory development activities and report on RO-II
field demonstrations currently in progress.
Background
As a result of the recent Clean Air Act and specific local regulations,
boiler operators are addressing the need to reduce stack gas emissions.
Current attention is focused upon controlling acid rain precursors, oxides of
nitrogen (NOX) and sulfur dioxide (SO,,). Under Phase I of the Clean Air Act,
a number of pre NSPS coal burning wall fired boilers will be required to
reduce their NOX emissions by the mid 1990's. The proposed Federal upper
limit for NOX emissions from wall fired units is 0.50 Ib/MBtu fired.
ABB Combustion Engineering Services, Inc. (ABB-CE) has been actively
developing and commercially demonstrating low NOX technologies for coal fired
tangential and cyclone boiler arrangements. In order to meet the NOX
reduction needs of coal wall fired boilers, ABB-CE has embarked on an
extensive low NOX coal burner development and commercial demonstration
program building on its substantial wall fired experience base with the
ABB-CE Type R burner.
The Type R horizontal burner was developed by Combustion Engineering Inc. in
the early 1940's. This burner was designed to burn pulverized coal, oil, or
7B-1
-------
gas, is applicable to single wall or opposed wall firing in either single or
multiple burner arrangements. In terms of experience, several thousand Type
R burners have been installed and operated in a wide variety of boiler
configurations ranging up to 160 MWe in capacity. Individual burner
capacities have ranged from 20 MBtu/hr to 120 MBtu/hr. As a result of this
extensive field experience, ABB-CE has established Type R design standards
which delineate proven materials of construction and fabrication techniques,
Type R operating procedures are also firmly established.
The Type R coal burner, illustrated in Figure 1, has several key hardware
features. The centrally located coal nozzle has spiral rifling along the
inner walls to promote swirl of the pulverized coal/primary air stream which
is initially established by a tangential inlet nozzle. A convergent nozzle
tip is located at the end of the coal nozzle. Five (5) deflector vanes,
located near the tangential inlet nozzle, can be adjusted in terms of
incident angle to vary coal/primary air stream swirl which in turn,
influences final luminous flame shape. On the combustion air side, the total
combustion air flow passes through an adjustable angle flat blade swirler
assembly. Combustion air angular momentum can be varied to optimize the
burner's flame stabilizing aerodynamic recirculation zone, directly
influencing both flame stability and flame shape.
Laboratory Development Program
In order to respond to low NOX requirements for wall fired-coal boiler
retrofit market, ABB-CE embarked on a laboratory development program with the
objective of developing a new low NOX wall fired burner product.
The new burner, named the RO-II burner, would be capable of meeting the
following performance targets:
• NOX less than 0.5#/106 Btu Fired
• Zero or nominal increase in carbon loss and/or CO emissions under
low NOX conditions.
• Acceptable flame envelope (length).
t Zero or nominal increase in fuel system or combustion air windbox
static pressure(s).
At the onset of the development program, ABB-CE assessed the NOX reduction
potential of the Type R burner design; upon review it was decided to
7B-2
-------
incorporate certain key design features of the Type R design into the new
RO-II burner design. These features specifically included the tangential
inlet, spirally rifled coal nozzle and an adjustable coal stream deflector
vane assembly. The Type R combustion air register assembly was determined to
not offer any advantages in terms of reducing total NOX so alternative air
register assemblies were reviewed for incorporation into the new low NOX RO-
II burner design.
ABB-CE selected a patented, commercially available, air register for
incorporation into the RO-II burner. Key features of the register are
highlighted in Figure 2. These features include:
1. Two separate plenums which permit staged introduction of combustion
air.
pilot air which is introduced concentrically adjacent to the
centrally located coal nozzle
main air which surrounds the pilot air stream
2. Involute (spirally shaped) air inlets for each plenum which swirl
total combustion air flow.
3. Separate flow control dampers for both the pilot and main air
streams.
4. Integral instrumentation which permits burner operators to balance
combustion air flow to multiple burner arrays located within a
common windbox.
5. Unique helical flow vane assembly which enhances combustion air
swirl and improves air distribution within the register.
6. A shadow vane assembly which enhances combustion air swirl but more
importantly protects the flow vane assembly and fuel nozzle from
damage due to flame radiation in multiple burner installations.
Photo 1, an end-on view of the RO-II register assembly, highlights the
involute (spirally shaped) air plenum, for both the pilot and main combustion
air streams, and the shadow vane assembly. Photo 2 highlights the flow vane
assembly utilized in the RO-II register. The helical vane arrangement is
shown separate from the air register. Note that the pilot combustion air
stream passes through six (6) vanes at the rear of the burner (i.e. the
widest part of the vane assembly), the main combustion air stream passes
through eight (8) vanes near the burner front (i.e. the narrowest part of the
vane assembly). It should also be noted that the register design requires
minimal maintenance since the only moving parts are the pilot and main air
7B-3
-------
dampers. These same dampers also provide the register with the ability to
compensate for burner to burner combustion air flow imbalances in multiple
burner/common windbox arrangements.
The RO-II development program was largely comprised of extensive combustion
trials of potential RO-II firing system hardware. These trials were
conducted in one of ABB-CE's front wall fired large scale laboratory test
furnaces. ABB-CE's development philosophy was to conduct tests with hardware
designed to operate at a heat input rate of 50 x 106 BTU/HR. This rate is
identical to the design heat input rate of the burners to be installed in two
units in Jamestown, NY. By adopting this development philosophy, ABB-CE
could confidently accelerate the process of transitioning laboratory hardware
developments into commercial application.
Prior to conducting the laboratory combustion trials, ABB-CE evaluated the
air register's near-field aerodynamics. The objective of these tests was to
define key aerodynamic characteristics of the register in order to support
the design of compatible coal nozzle configurations. Recirculation zone size
and strength as well as the air register's potential to control stoichiometry
in the burner near field (through internal air staging) were assessed. These
aerodynamic properties were consistent with the low NOX objectives of the RO-
II development program.
Laboratory combustion trials began following the register aerodynamic study.
The focus of these trials was to evaluate the combustion performance of a
variety of air register/coal nozzle configurations. The performance of each
configuration was evaluated in comparison with the overall performance
targets for the RO-II burner. It should be noted that the air register
configuration remained fixed throughout the trials. Development activities
concentrated on combining advanced low NOX Type R coal nozzle arrangements
with the existing air register design.
The combustion trials generated the data necessary to assess RO-II burner
performance. Flue gas 02, NOX, and CO concentrations were measured at each
test condition, along with coal/primary air static pressure at the coal
nozzle inlet, windbox and furnace static pressures, and total combustion air
and primary air mass flows. Qualitative assessments of flame shape, length,
and stability were also made throughout the development program. In
addition, flyash samples were taken for subsequent carbon in ash analysis.
7B-4
-------
Furnace horizontal exit gas temperatures were also quantified using suction
pyrometry.
The combustion test program parametrically evaluated a number of key RO-II
design and operating variables. Some of the variables investigated included
coal nozzle/tip configurations, firing rate (MCR and reduced load), excess
air level, coal/primary air velocity at the coal nozzle tip exit, pilot and
main air damper position (pilot/main air flow split) and coal stream swirl.
All laboratory trials were conducted with a Pennsylvania bituminous coal
having 10% ash, with a fixed carbon to volatile ratio of 1.65 and a fuel
nitrogen content of 1.5% by weight. Coal preparation for the laboratory
tests was consistent with typical utility practice; the pulverized coal grind
averaged 70.3% through 200 mesh (75 microns), with 0.6% remaining on a 50
mesh (300 microns) screen.
The laboratory test furnace utilizes a dilute phase (1.5 2.0 # primary
air/# coal) indirect coal feed system. A schematic of the feed system is
shown in Figure 3. Figure 3 highlights the fact that a gravimetric feeder is
employed to accurately quantify coal feed rate. The figure also illustrates
the location of static pressure taps in the coal feed system. These
pressures were documented throughout the test program for comparison to
performance targets.
Photo 3 shows the installed RO-II Burner register as viewed from outside the
furnace. Note the use of the tangential entry fuel nozzle inlet,
characteristic of both the Type R and RO-II burner designs. Photo 4 shows
the installed RO-II from the furnace side and highlights the shadow vanes and
divergent refractory quarl similar to typical field installations.
Note also in Photo 4 that there is refractory material on the furnace walls.
The laboratory test furnace has atmospheric pressure water cooled walls. The
furnace gas temperatures and heat release profile are adjusted by altering
the refractory configuration depending upon test objectives. The refractory
configuration selected for these trials was chosen to create a furnace
thermal environment where relatively high levels of thermal NOX would be
generated. In addition to refractory modifications, the test furnace was
intentionally operated at a volumetric cubic heat release rate of 39,800
Btu/hr/ft3. This volumetric heat release rate in effect far exceeded a
7B-5
-------
typical coal-designed boiler's volumetric heat release range of 9,000-16,000
Btu/hr/ft3. As a result of this (and combined with the refractory insulation
thickness and pattern in the furnace), measured furnace gas outlet
temperatures (horizontal furnace outlet plane) were in the 2500 2700°F
range, far exceeding typical boiler horizontal furnace gas outlet
temperatures of 1900 2000°F. The implication of high temperature furnace
operation during the RO-II laboratory trials is that NOX generated thermally
via the Zeldovich mechanism (1) was projected to be conservatively higher
than would be expected in subsequent field RO-II installations.
Eleven different coal nozzle configurations were evaluated during the
combustion trials. Baseline tests were conducted with a conventional Type R
nozzle; ten advanced Type R nozzle configurations were also evaluated. The
baseline nozzle (Figure 4) was comprised of the tangential fuel inlet, coal
stream deflector vanes, and a spirally rifled nozzle with a convergent tip.
A furnace side view of the baseline Type R coal nozzle is shown in Photo 5.
Combustion test data from the "Baseline" RO-II configuration is shown in
Figure 5 which depicts NOX (ppm corrected to 3% 02) as a function of flue gas
02 concentration. As is characteristic of a diffusion flame burner, NOX
increases with increasing excess air level. The primary point of the figure
is that at a nominal excess air level of 20% (approx. 3.5% 02), measured NOX
was approximately 650ppm (approx. 0.9 #/MBtu). Under all excess air
conditions, NOX exceeded the target value of 0.5 #/MBtu.
The most optimum coal nozzle arrangement of the ten tested is shown in
Figure 6. As shown in the schematic, the optimum RO-II coal nozzle retains
the tangential fuel/primary air inlet, deflector vane assembly, and spirally
rifled nozzle of the Type R design. The optimum RO-II arrangement includes
the addition of a venturi diffuser assembly, which is a channeled flow
control device, and a convergent nozzle tip with axial rifling vs. spiral
rifling as in the baseline case.
Photo 6 is a "furnace side" view of the optimum nozzle arrangement.
Figure 7 graphically depicts the NOX emission performance of a number of the
tested RO-II coal nozzle concepts. Data in this figure highlights the fact
that the coal nozzle design employed had a dominant influence on NOX levels
observed. One can summarize the data contained in Figure 7 by directing
7B-6
-------
attention to the solid line plotted in the center of the graph. All data
below the solid line represents the NOX performance of the venturi diffuser
concept, all data above the line represents alternative tested concepts.
Clearly, the venturi diffuser concept generated lower total NOX at any given
operating excess air level, as compared with all other tested coal nozzle
concepts. Most importantly, at a nominal flue gas 02 concentration of 3.5%
(20% excess air), total measured NOX was 360 ppm (corrected to 3% 02),
meeting the overall project goal of 0.5 #/MBtu NOX. The venturi diffuser
coal nozzle assembly, as a result of its success in meeting the NOX reduction
target established for the project, has been chosen as the coal nozzle design
to be utilized in the RO-II burner.
Beyond its NOX reduction capability, the RO-II burner met all other
established performance targets. These targets were set to ensure that the
firing system hardware developed in the laboratory would be retrofitable to
most existing wall fired boiler arrangements. Most units, for example, have
fan limitations in terms of achievable windbox to furnace delta static
pressure. The RO-II coal burner is capable of operation at less than 3.0"
W.C. static windbox to furnace delta pressure at MCR. Most existing boiler
F.D. fan systems are capable of achieving at least that pressure differential
at MCR. In a similar vein, primary (coal transport) air static pressure at
the coal nozzle inlet is a critical factor from a retrofit standpoint. Any
low NOX burner installation should operate within existing coal feed system
pressure limitations. The RO-II burner operated at MCR with a primary air
static pressure at the nozzle inlet of less than 4.5 inches W.C., an
acceptable operating primary air static pressure for most existing wall fired
installations.
Many low NOX coal firing system laboratory tests and field demonstrations to
date have reported that, under low NOX conditions, carbon in fly ash levels
tend to increase (2, 3, 4). In some cases, CO emissions also increase under
low NOX conditions. These results are, of course, very dependent on coal
type, coal particle size distribution, and furnace configuration. In
practical terms, most low NOX coal firing systems must strike an acceptable
balance between NOX reductions and carbon in fly ash/CO increases. In the
case of the RO-II coal burner, operated at 0.5 #/106 Btu, both carbon in fly
ash and CO emissions did increase, however, the final emission levels
documented were within acceptable operating ranges. For example, under
baseline, high NOX conditions, carbon in fly ash and CO were 1-2% and 30-
7B-7
-------
BOppm, respectively. Under low (0.5 # MBtu) NOX conditions, carbon in fly
ash and CO increased to 3% and 40-70 ppm, respectively. These laboratory
results indicate that nominal increases in carbon in flyash may be expected
in RO-II field applications, again dependent on coal type and furnace
configuration.
Several low NOX coal firing systems evaluated to date for wall fired boiler
applications have experienced increased flame lengths as compared to pre-
retrofit cases (5,6). As in the case of the relationship between NOX, carbon
loss, and CO, one must in most situations strike a balance between NOX
reductions and increasing flame length. Operating experience with the RO-II
coal burner to date is good in this regard. Baseline (high NOX) conditions
produced a luminous, stable flame about 12' long. Under low NOX (0.5 #/MBtu)
conditions, flame length increased to approximately 16'-18' long. The
increase in flame length was deemed acceptable because since the field units
targeted for the first RO-II coal demonstrations can accommodate a similar
increase in flame length without direct flame impingement on rear wall tube
surfaces. Future boiler retrofits will be assessed on an individual basis
not only to ensure compatibility between furnace depth and the luminous flame
volume of the RO-II low NOX coal burner, but also to ascertain potential for
changes in post-retrofit boiler thermal performance.
Field Experience
Following successful laboratory development trials, the RO-II coal burner has
presently been retrofitted to three (3) field installations. Figure 8 is a
schematic of the as-installed RO-II coal burner. The tangential inlet,
spirally rifled coal nozzle with venturi diffuser assembly and convergent
nozzle tip can be seen in the figure. The pilot and main air plenums,
helical flow vanes, and shadow vanes are also depicted.
The current RO-II field installations are listed in Figure 9 with other
pertinent information. City of Jamestown Unit 10 and BPU Kansas City are
currently undergoing start-up and demonstration testing.
Conclusions
ABB-CE's RO-II coal burner, specifically designed for retrofit wall fired
7B-8
-------
boiler applications, has undergone extensive laboratory testing and is now
commercially available. In laboratory trials, the burner was shown to meet
the NOX target of 0.5 #/MBtu firing Eastern U.S. bituminous coal while
limiting increases in carbon loss, CO, and flame length to commercially
acceptable levels. The burner also demonstrated the ability to operate
within the capacity of most existing boiler combustion air fan and coal
delivery systems in terms of static pressure requirements. The RO-II burner
offers advantages in terms of its simplified construction and operation. In
addition, the RO-II burner has the ability (via adjusting the main/pilot air
damper system) to equalize burner to burner combustion air flow imbalances in
multiple burner/common windbox plenum arrangements.
References
1. Zeldovich, Y. et al. (1947), Oxidization of Nitrogen in Combustion,
Academy of Sciences of the USSR, Institute of Chemical Physics,
Moscow-Leningrad, Translated by M. Shelf, Scientific Research
Staff, Ford Motor Co.
2. Beard, P. et al "Reduction of NOX Emissions form a 500 MW Front
Wall Fired Boiler" 1989 Joint EPA/EPRI Symposium on Stationary
Combustion NOX Control.
3. Grusha, J. and McCartney M., "Development and Evolution of the ABB
Combustion Engineering Low NOX Concentric Firing System 1991
Joint EPA/EPRI Symposium on Stationary Combustion NOX Control.
4. Kinoshita, et al "New Approach to NOX Control Optimization and
Unburnt Carbon Losses" 1989 Joint EPA/EPRI Symposium on
Stationary Combustion NOX Control.
5. Clark, M.J. et al "Large Scale Testing and Development of the B&W
Low NOX Cell Burner" 1987 EPA/EPRI Symposium on Stationary
Combustion Nitrogen Oxide Control.
6. LaRue, A. et al "Development Status of B&W's Second Generation Low
NOX Burners The XCL Burner" 1987 EPA/EPRI Symposium on
Stationary Combustion Nitrogen Oxide Control.
7B-9
-------
Figure 1: Type R Coal Burner
Photo 1: End-On View of the RO-II Register Assembly
Figure 2: Exploded View of RO-II Burner Assembly
Photo 2: Helical Flow Vane Assembly
Figure 3: Coal Feed System Schematic
7B-10
-------
Photo 3: Installed RO-II Burner Register as Viewed
from Outside the Furnace
TANGENTWL
FUELPFB MARY AIR
INLET
SPIRALLY-BIFLED TIP
MI
v
DEFLECTOR
VANES
Figure 4: "Baseline" Type R Coal Nozzle Schematic
2D%EA
02, %
Figure 5 "Baseline" Nozzle Assembly, NOx vs. O
Photo 4: Installed RO-II Burner from the Furnace Side
I
Photo 5: Furnace Side View of the "Baseline" Type R
Coal Nozzle
INLET
'I
^- L-
VEWTURI DtfFUSER
ADJUSTMCWT BOOS
STUFFWG BOX Fi
AXIAL NOZ2LE ADJUSTMENT
X \
SPTRlALLY-fOOZD
NOZZLE
DEFLECTOR VANES
NOZZLE TP WITH
AXIAL RJFL**S
Figure 6: Venturi Diffuser Nozzle Assembly, Test
Equipment Schematic
7B-11
-------
02. %
At and Below the Line - Venturl Dlfluser Concept!
Above the Line - Other Tested Concepts
Figure 7: RO-II "Advanced Coal Nozzle Concepts"
NOx vs. O,
Photo 6: "Furnace Side" View of the Optimum Nozzle
Arrangement for the RO-II Burner
Customer Unit
Bd ol Public Utll 9
City of Jamestown
Bd of Public UtIL 10
City of Jameetown
Bd ot Public Utll. Oulndaro
Kansas City Unit 2
Unit
Type
CE-VU40
CE-VU40
Rlley
Steam Flow
Ib/hr
165.000
165.000
1.126.000
No. ol
Burners
4
4
9
Fuels
E. Bit
E. Bit
Sub-Bit.
Natural Gas
Propane
Figure 9: RO-II Experience List
Figure 8: RO-II Burner
7B-12
-------
NOx REDUCTION ON NATURAL GAS-FIRED BOILERS
USING FUEL INJECTION RECIRCULATION (FIR) -
LABORATORY DEMONSTRATION
Kevin C. Hopkins, David 0. Czerniak
Carnot
15991 Red Hill Ave., Suite 110
Tustin, CA 92680-7388
Les Radak
Southern California Edison Company
2244 Walnut Grove Avenue
P.O. Box 800
Rosemead, CA 91770
Cherif Youssef
Southern California Gas Company
3216 North Rosemead Blvd.
El Monte, CA 91731
James Nylander
San Diego Gas & Electric Company
P.O. Box 1831
San Diego, CA 92112
-------
NOx REDUCTION ON NATURAL GAS-FIRED BOILERS
USING FUEL INJECTION RECIRCULATION (FIR) -
LABORATORY DEMONSTRATION
ABSTRACT
Increasingly stringent NOx regulations on industrial and utility boilers may require
implementation of expensive post-combustion NOx control techniques. Fuel Injection
Recirculation (FIR) is a relatively low cost NOx reduction strategy for natural-gas
fired boilers in which the fuel is diluted prior to combustion with air, steam, or
flue gas. This technique is different from conventional flue gas recirculation (FGR)
because it is conceptually believed to impact prompt as well as thermal NO formation
mechanisms and is therefore capable of greater NOx reductions. Furthermore, the two
technologies when applied in conjunction are additive is terms of NOx reduction.
As a preliminary step towards full scale implementation of FIR, a laboratory
demonstration was performed to determine the feasibility of the technology. FIR was
demonstrated on a 2.0 MMBtu/hr test facility designed to simulate burners used on full
scale utility boilers. The test facility employed combustion air preheat, FGR,
staged-air firing, and was modified to inject flue gas, air, or saturated steam into
the fuel stream prior to combustion. The effectiveness of FIR was determined at
varying injection rates, firing rates, air preheat levels, FGR rates, and excess 02
conditions.
Results show that FIR is more effective that FGR in reducing NOx, and that a
additional 50% NOx reduction was achieved when FIR is used in conjunction with FGR.
The test program demonstrated that in a full-scale application, FIR may be capable of
reducing NOx to low levels, at an attractive cost relative to post-combustion control
retrofits.
7B-15
-------
INTRODUCTION
Carnot was contracted by the Southern California Gas Company, the Southern California
Edison Company (SCE), and the San Diego Gas and Electric Company (SDG&E) to perform
a laboratory demonstration of a potential new NOX reduction technology for gas-fired
boilers which has been designated Fuel Injection Recirculation (FIR). As a
preliminary step towards full-scale implementation, this demonstration program was
performed to determine the feasibility of the technology.
Fuel Injection Recirculation involves recirculation of a portion of the boiler flue
gas and mixing it with the gas fuel at some point upstream of the burner.
Additionally, the FIR concept can be expanded to include the fuel injection of any
inert diluent such as steam or air. This method conceptually is believed to be
capable of greater NOX reductions than can be achieved by conventional Flue Gas
Recirculation (FGR), which is mixed with the combustion air. Furthermore, it is
anticipated that when implemented on a utility boiler, the two technologies would be
to some extent, additive in terms of NOX reductions, ultimately resulting in very low
NOX emissions. The principal motivation for pursuing this concept is the potential
cost benefit in comparison post-combustion NOX control technologies such as SCR and
urea injection, which are presently being considered to meet the stringent new NOX
limits specified in the South Coast Air Quality Management District Rules 1135 and
1146. The FIR concept is also attractive because full-scale application of FIR would
require relatively few modifications to existing equipment.
The approach taken for this laboratory demonstration program was to apply the FIR
technology on a test facility which incorporates many key design and operational
attributes of burners in use on utility boilers, and which employs NOX control
techniques commonly used in these large scale boilers. The primary emphasis of the
this feasibility study was a practical evaluation of FIR over ranges of important
operating conditions such as firing rate, air preheat, overfire air, and FGR.
7B-16
-------
TECHNICAL OBJECTIVES
Throughout this study, FIR was evaluated primarily in terms of flue gas concentrations
of NOX, 02, C02, and CO, and in terms of burner stability and flame characteristics.
The specific technical objectives of the investigation were as follows:
1. Evaluate the NOX reduction effectiveness of FIR using a
laboratory-scale burner similar in design and thermal
environment to burners used on electric utility boilers.
2. Evaluate the NO reduction efficiency of FIR alone, and in
combination with FGR.
3. Evaluate the effect of FIR on minimum operable 02 level, and
on burner stability.
4. Evaluate the effect of reduced firing rate on the
effectiveness of FIR.
5. Evaluate the effect of air staging on the effectiveness of
FIR.
6. Compare the effect of air relative to flue gas as the FIR
diluent.
7. Compare the effect of steam relative to flue gas as the FIR
diluent.
BACKGROUND
Fuel Injection Recirculation (FIR) is a potential new NOX control strategy for natural
gas-fired boilers which is defined as the injection of any inert diluent into the fuel
gas at some point upstream of the burner. The concept originally involved the
extraction of flue gas from the exit of the boiler, cooling it if necessary, and
finally compressing it for injection at gas header pressures into the fuel line.
Operating expenses and equipment costs may be reduced by injecting other diluents such
as air or steam, or by lowering gas header pressures through burner modifications.
FIR and Prompt NO Formation: NOX formation in natural gas-fired boilers is associated
with two mechanisms known as thermal NO and prompt NO. Thermal NO refers to the high
temperature reaction of nitrogen and oxygen from the combustion air. This mechanism,
which is commonly termed the "Zeldovich" mechanism after its discoverer, is thought
to occur in the post-flame or burned gas zone. Low excess air firing, flue gas
recirculation, burners-out-of-service (BOOS), and air staging are commonly used on
utility boilers to control thermal NO formation.
The existence of another NO formation mechanism was first suggested by Fenimore whose
measurements showed that reactions other than the Zeldovich mechanism were taking
place, and that some NO was being formed in the flame region. Because of the early
7B-17
-------
formation of NO, Fenimore coined the name "prompt" NO. Fenimore proposed that C2 and
CH radicals present in hydrocarbon flames contribute to the formation of prompt NO.
The formation of prompt NO is greater in fuel-rich flames, and decreases with the
increase in local 02 concentrations. Similar experiments have shown that prompt NO
formation is a function of flame temperature as well as stoichiometry. Other
measurements made in flat flame burners demonstrate that prompt NO can account for 10-
40 ppm of the total NO formed. In utility boiler systems, prompt NO is assumed to be
less than 50 ppm while the thermal NO contribution can be as high as 125-200 ppm.
Thermal NO control techniques such as FGR and BOOS can decrease NO to concentrations
approaching prompt NO concentrations. The South Coast Air Quality Management District
Rule 1135 for utility boilers will require NOX emission limits translating to about
25 ppm, and therefore the control of prompt NO formation is important if new emissions
limits are to be met without installation of expensive post-combustion control
techniques.
FIR appears to be a effective and relatively inexpensive technique for the control of
prompt NO formation. It is believed that FIR reduces prompt NO formation by diluting
the fuel prior to combustion thereby reducing the concentration of hydrocarbon
radicals which produce prompt NO. In addition, FIR also acts like FGR in reducing
thermal NO production. It is anticipated that FIR in combination with FGR, could
reduce NOX emissions to levels below 25 ppm by controlling both NO formation
mechanisms.
TEST DESCRIPTION
Test Facility: The laboratory facility selected for this evaluation of FIR was an 80
hp Scotch fire-tube boiler. This boiler was modified to incorporate many significant
components of a full-scale utility boiler furnace. The test facility comprised the
fire-tube boiler, which is capable of firing up to 3.0 x 106 Btu/hr on natural gas,
a forced draft fan, a separately fired air preheater (APH), a 5 1/2" diameter gas fuel
ring, a ceramic quarl, and a windbox with a sixteen blade variable air register. Off-
stoichiometric firing was achieved by diverting a portion of the pre-heated combustion
air to the overfire air (OFA) ring placed downstream of the burner face. A separate
fan was used to recirculate a portion of the flue gas to the combustion air (FGR).
The FGR flowrate was determined by measuring the windbox 02 concentration along with
the flue gas 02 concentration. The mass flowrate of the flue gas recirculated was
subsequently determined from stoichiometric calculations.
Natural gas was supplied to the boiler via a 10 psig supply, and metered using a
rotameter. The maximum firing used in this study was 2.0 x 106 Btu/hr. The burner
consisted of 3/8 inch ring with 11 equally spaced holes drilled radially, each of
7B-18
-------
0.189 inches diameter. The ceramic burner quarl, six inches long with a nine-inch
exit diameter, was geometrically similar to those used on small Peabody ring burners
in utility boilers. The air register vanes were set initially to target a baseline NOX
level characteristic of full-scale units. The air register vanes were set at 20° off
radial and were not varied throughout the remainder of the tests.
The FIR concept was tested using three fuel diluents: flue gas, air, and saturated
steam. Most of the testing was performed using flue gas as the diluent. The flue gas
injection system consisted of a 5 hp rotary lobe type compressor capable of a delivery
pressure of up to 8 psig at a flow rate of 30 scfm of flue gas. Flue gas, extracted
at the stack plenum, was compressed and injected into a 2 inch fuel line through a
sparger. FIR tests with air injection were performed using the same configuration as
above with the inlet to the blower disconnected from the stack plenum.
Steam injection was accomplished using a separately fired 2-1/2 hp Parker Boiler
providing saturated steam at approximately 180 psig. The flow rate was controlled
using a gate-valve and was metered using an Annubar flow sensor. Steam was injected
through the sparger into a heat-traced fuel line.
Test Conditions: The principal objective of this laboratory demonstration program was
to determine the effectiveness of FIR in reducing NOX at conditions characteristic of
large industrial or utility boilers. Conditions and parameters which significantly
impact NOX on full-scale units include combustion air temperatures, off-stoichiometric
firing, excess air levels, load variations, flue gas recirculation to the combustion
air, burner configuration, and air register orientation. It was not practical to
systematically investigate the influence of each of these characteristics in the
laboratory facility. Once baseline configurations were established, the burner
hardware and the air register orientation were not changed throughout the testing.
Excess Air Levels: Tests were performed at a "minimum" or "nominal" excess 02
condition. The minimum 02 condition was defined by the following criteria:
1. the excess air level producing 200 400 ppm CO, or
2. an excess 02 concentration of « 0.3 %
The second criteria was necessary because at some test conditions, CO did not exceed
100 ppm even at extremely low 02 concentrations. The 0.3 % 02 concentration was
necessary as a lower safety limit for those tests where CO remained below 100 ppm.
The nominal 02 condition was defined as the amount of excess air necessary to increase
the minimum 02 concentration by 0.5 %.
Flame Characteristics: Since an important objective of this test program is to
determine the limits of applicability of FIR with respect to flame characteristics,
7B-19
-------
the test series involving fuel dilution with flue gas, steam, or air, the diluent was
added to the point of flame instability. Flame stability and general flame
characteristics were determined primarily form observations. The flame was considered
to be unstable if any of the following was observed:
1. Any tendency for the flame to lift-off from the burner face
and re-attach downstream on the OFA ring.
2. Excessive fluctuations in furnace draft
3. Excessive fluctuations of NOX, CO, or 02 concentrations.
'X'
RESULTS AND DISCUSSION
The results of the Fuel Injection Recirculation (FIR) test program are presented in
this section. The NOX results presented below are expressed in ppm corrected to 3%
02 on a dry basis. The NOX reductions achievable, and the limitations in terms of
flame stability are considered for FIR used in conjunction with varying firings rates,
flue gas recirculation rates, air preheat levels, and both with, and without overfire
air. For each test series, the injection rate of flue gas was increased until the
limit off flame stability was reached. The flame stability limit is defined as the
maximum injection rate at which the flame remains attached to the burner face.
(Higher injection rates would cause the flame to detach from the burner face and re-
attach to the overfire air ring).
For the purposes of later comparison, the "baseline" condition is defined by the
following parameters:
firing rate: 2,000,000 Btu/hr ± 2 %
Op condition: minimum (defined by CO ~ 200-400 ppm)
OFA condition: nominal (defined by « 10% of total air)
APH temperature: 480 495 °F
Windbox FGR: 0 %
The baseline NOX concentration for this test facility was 87.6 ppm @ 3% 02. Without
OFA, the NOX concentration was 167.6 ppm @ 3% 02. The use of OFA reduced NOX by 48%.
This is consistent with full-scale NOX reductions attainable using NOX ports and/or
burners-out-of-service (BOOS). The effects of other parametric variations are
presented below.
Summary of Baseline Characteristics
• The baseline NO concentration is 87.6 ppm 0 3% 02 with
approximately 10% overfire air with a combustion air
temperature of approximately 490 °F.
• NOX is very sensitive both to excess air level and to
combustion air temperatures, especially at lower FGR rates.
7B-20
-------
• The measured NOX vs FGR relationship is typical of full-
scale units.
• The NO vs firing rate relationship is characteristic only
of smaller industrial boilers.
Flue Gas as FIR Diluent
The effect of Fuel Injection Recirculation using flue gas as the diluent is presented
in this section. The amount of FIR injection is expressed in two ways. First as a
percent fuel dilution defined as the percentage of the volume of flue gas injected to
the total volume flow through the burners. Alternatively, for the purposes of
comparison to conventional flue gas recirculation, it is expressed as the percent of
the weight of the flue gas injected to the total weight of the air and fuel.
FIR vs Windbox FGR: The effect of FIR without windbox flue gas recirculation (FGR),
and at an optimum and maximum FGR rates are presented in this section. The firing rate
is 2.0 x 106 Btu/hr with nominal OFA at the minimum 02 condition. The results are
shown in Figure 1-A and 1-B.
Figure 1-A shows NOX concentration vs FIR injection rate expressed as percent fuel
dilution. NOX decreases uniformly with increasing FIR injection. With no windbox
FGR, the rate of decrease is approximately 1.7 ppm per % fuel dilution. At higher
windbox FGR rates, the rate of decrease is approximately 0.6 ppm per % fuel dilution.
The decreasing effectiveness at higher windbox FGR rates indicates that FIR reductions
are partially thermal NOX reductions and that the two techniques are to some extent
redundant. However, since further decreases are measured even at the maximum windbox
FGR rate, the two techniques also appear to be additive.
This additive effect can be more clearly seen in Figure 1-B where the effect of FIR
on NOX is plotted as a function of the total flue gas recirculated (to windbox and to
fuel). For each of the three data sets shown on the graph, the windbox FGR is held
constant while the FIR flowrate is increased. The dotted line on the graph defines
the relationship between NOX and the windbox FGR alone. At both the 15% and 23%
windbox FGR rates, FIR injection is capable of additional reductions of approximately
50%. Table 1 summarizes the maximum reductions achievable with FIR when used in
conjunction with FGR. Furthermore, it is evident that FIR alone is more effective
than FGR: 5% of the flue gas injected into the fuel results in lower NOX than 23% flue
gas injected into the combustion air. This is shown graphically in Figure 2 where NOX
reduction is plotted vs the total flue gas recirculated. The NOX reduction curve
rises more steeply with FIR than without. It should be re-stated here that the flue
gas recirculation to the fuel requires significantly higher compression that
recirculation to the combustion air.
7B-21
-------
As postulated earlier, FIR is believed control prompt, as well as thermal NO, both by
reducing peak flame temperatures and by lowering the concentration of certain
hydrocarbon radicals which are thought to produce prompt NO. The concentration of
prompt NO formed in utility combustion systems is thought to be 25 ppm or less. The
tests performed in the present study are not intended to distinguish between prompt
NO reductions and thermal NO reductions, or even to confirm the existence of prompt
NO. It is not possible to conclude whether the additive NOX reductions are due to
more efficient mixing of flue gas with the fuel and air, or whether FIR actually
suppresses prompt NO formation. What can be concluded however is that FIR is more
effective than windbox FGR, and that together there is a measurable additive benefit.
The use of FIR does not significantly affect flame stability up to a fuel dilution
ratio of approximately 35% Higher injection rates create a tendency to lift off the
burner face creating a "boiler rumble" and large fluctuations in NOX and 02 and furnace
draft. At lower injection rates, the appearance of the flame is not significantly
different from the flame appearance with no FIR injection, other than decreased
brightness which is indicative of lower peak flame temperatures.
The Effect of Overfire Air on FIR: The effect FIR when used without overfire air is
shown in Figure 3-A and Figure 3-B. FIR is equally effective with, or without
overfire air. Without OFA, FIR reduces NOX concentrations by 60% at 0% FGR and 15%
FGR.
It was also expected that overfire air would affect flame stability by decreasing the
burner throat velocities. The tests demonstrated that overfire air does not affect
flame stability. Figure 3-A shows that the limit of flame stability is approximately
at 35% fuel dilution regardless of the OFA rate. Figure 3-B shows that the effect of
overfire air has a decreasing effect at higher FGR rates. For example, at 15% windbox
FGR with the maximum FIR injection rate, 10% air staging results in less than a 5 ppm
NOX reduction.
The Effect of Firing Rate On FIR: The effect of FIR at three firing rates is
presented in Figure 4. FIR injection using flue gas results in approximately the same
NOX reductions at 1.0, 1.5, and 2.0 x 106 Btu/hr. The slopes of the curves on Figure
4 are not a function of the firing rate. With no windbox FGR, FIR reduces NOX at
approximately 5 ppm/%fuel dilution up to 35% fuel dilution. At an optimum windbox FGR
rate, the slope decreases to .6 ppm/%fuel dilution up to 35% fuel dilution.
It is important to note that reduced firing does not significantly affect flame
stability. The limit of flame stability occurs at approximately 35% fuel dilution at
each firing rate tested. It is difficult to extrapolate this characteristic to the
full-scale application primarily due to the non-characteristic NOX vs firing rate
7B-22
-------
relationship, i.e. the relative increase in NOX at the mid-firing rate. It is also
important to remember that the minimum 02 condition at the lower firing rates results
in significantly higher 02 concentrations. The air register vane setting is likely
to affect flame stability and the minimum 02 condition, however the effect of air
register adjustments was not examined during this test program.
Air As FIR Diluent
The original concept of Fuel Injection Recirculation involved injecting flue gas into
the fuel. In principle any diluent could have the same affect on prompt NO formation.
The advantage of using air as a fuel diluent is that compressing dry air up to fuel
pressures is less expensive than compressing hot flue gas. In addition, problems with
moisture condensation in the fuel delivery system are eliminated if air is used
instead of flue gas. The effectiveness of air injection was explored in a limited
test matrix intended to compare air to flue gas as FIR diluents.
Air was injected as an FIR diluent at the following conditions: high combustion air
temperatures, at a nominal overfire air rate, and at two FGR rates. The results are
shown in Figure 5, where the results for flue gas injection are re-plotted for
comparison. These results demonstrate that air injection is not as effective as flue
gas injection in overall NOX reductions. For the 0% FGR case, NOX actually increases
at low air injection rate. The characteristic is not measured at the 15% FGR
condition. Table 2 shows that the overall NOX reductions achieved using air injection
are less than half of the reduction measured using flue gas injection.
Steam as FIR Diluent
Steam is another fuel diluent which in principle should reduce NOX much the same way
as flue gas. The use of steam as an FIR diluent for full-scale application may be
attractive on a cost basis since it would require no additional compressors. Provided
that steam could be extracted at relatively low pressures, the impact on boiler heat
rate should not be prohibitive. The use of steam injection as a means of NOX control
on large boilers is not a new technique. However, it is usually injected into the
combustion air upstream of the burner rather than into the fuel.
Particular experimental difficulties precluded a more expanded test matrix with steam
injection. The primary difficulty was the high fluctuation in steam flow: the
flowrate fluctuated by approximately 25 %. This made measurement of steam flow rate
difficult and caused high fluctuations of NOX and especially CO. Figure 6 shows a
example time trace taken from data logger records. Note that the NOX has been
corrected to 3% 02. NOX, CO, and 02 fluctuated regularly at the same frequency of the
steam generator fluctuation. The period of the fluctuation was approximately 4
minutes. As the steam flow cycled to a maximum, about 62 Ib/hr, the NOX reached a
7B-23
-------
minimum, and CO was in excess of 1000 ppm. At the minimum steam flow, about 48 Ib/hr,
the NO reached a relative maximum, and CO reached a minimum. Since the fuel flow
could not be adjusted for changes in back pressure, the fuel flow also cycled causing
small fluctuations in 02. Despite the fact that the steam generator flow rate could
not be held constant, the results generated are still valuable. The steam flow cycled
in a very regular, repeatable manner, and accurate data were obtained by averaging the
continuous emissions data over many cycles.
The results of the steam injection test are presented in Table 4-11 and in Figure 7.
The steam injection tests were performed without overfire air. When steam injection
was used in conjunction with overfire air, excessively high CO emissions resulted as
well as poor flame stability. Overall NOX reductions are 54% without FGR, and 36%
with 15% FGR. Figure 7 presents a comparison of steam injection and flue gas
injection. Also shown on this figure are the minimum and maximum NOX concentrations
corresponding to the maximum and minimum steam flow rate. The results show that with
no overfire air, steam injection is nearly as effective as flue gas injection.
CONCLUSIONS
Fuel Injection Recirculation (FIR) was demonstrated on a laboratory scale test
facility designed to simulate the significant combustion characteristics of full-scale
utility natural gas burners. FIR was evaluated in terms of NOX reductions and burner
stability. While, the absolute values of N0x emissions results presented in this
report should not extrapolated directly to full-scale boilers, relative NOX reductions
and general trends measured on the sub-scale facility, should be representative of
results expected on full-scale units. The major conclusions drawn from the laboratory
evaluation are presented below:
Baseline Characteristics
• At test conditions typical of utility boilers, the baseline
NOX concentrations on the sub-scale facility are
representative of full scale units.
• The measured N0x dependencies on FGR, air staging, air
preheat temperatures, and excess air levels are
representative of trends seen in full scale units.
• The measured relationship between NOX and firing rate is
typical of smaller package boilers.
Flue Gas as FIR Diluent
• FIR is an effective NOX reduction technique to be applied to
natural gas-fired boilers, and NOX reductions achieved using
this technique are additive to those achieved by windbox FGR
and air staging.
7B-24
-------
• FIR is more effective than windbox FGR, per pound of flue
gas recirculated, in reducing NOX emissions.
• FIR in combination with air staging and windbox FGR results
in additional NO reduction of approximately 50%. NO
concentrations below 25 ppm were achieved at full load with
nominal air staging, 15% FGR and 35% fuel dilution.
• FIR has no adverse effects on maintaining minimum 02 levels.
• FIR is equally effective at reduced firing rates and when
used without overfire air.
• FIR operates with good flame stability at high combustion
air temperatures and nominal air staging at FIR levels up to
35% fuel dilution. However, the maximum level of FIR
consistent with acceptable burner stability decreases with
decreasing combustion air temperature.
• With no air staging, FIR operates with good flame stability
at low combustion air temperature up to a 35% fuel dilution.
Air as FIR Diluent
• Air as an FIR diluent is less effective than flue gas and
leads to flame instabilities at lower injection rates.
Steam as FIR Diluent
Steam as an FIR diluent when applied in combination with air
staging results in poor flame stability and high CO
concentrations.
Steam when applied with no air staging is nearly as
effective as flue gas as an FIR diluent.
CO concentrations are generally higher with steam than with
air, or flue gas as the FIR diluent.
7B-25
-------
STEAM INJECTION, O OFA, O% FGR
HI
CO
LL
o
©
E
Q.
a
x"
O
1.1
1.0
0.9
0.8
0.7
0.6
0.5
0.4
0.3
0.2
0.1
0.0
TEST 186, 187
CO 6-1000 ppm
19 November 1990
< n
\ !' i | :
I i (I
NOx@3%O2
0-250 ppm
O2 0-10%
10:40
10:55
TIME
Figure 6. Example Emissions Time Traces with Steam Injection
180
160
140
O 120
CO
FIRING RATE: 2.0 x 1C6 Btu/hr
APH: 490"F
NOOVERFIREAIR
—•— STEAM INJECTION
FLUE GAS INJECTION -
100
80
60
40
20
0
15% FGR
0 5 10 15 20 25 30 35 40 45 50
PERCENT FUEL DILUTION
Figure 7. Effect of Steam Injection
7B-26
-------
cf1
CO
©
Q.
Q.
x"
O
110
100
90
80
70
60
50
40
30
20
10
0
INJECTION
FIRING RATE: 2.0 x 106 Btu/hr
APH: 490° F
NOMINAL OFA
MINIMUM O2
—A- AIR INJECTION
FLUE GAS INJECTION
10 15 20 25 30 35 40 45
PERCENT FUEL DILUTION
Figure 5. Effect of Air as FIR Diluent
50
7B-27
-------
O
<£
CO
Q.
0.
x"
O
110
100
90
80
70
60
50
40
30
20
10
0
NOMINAL OFA
MINIMUM O2
APH - 490 °F
—A— 2.0 x106 Btu/hr
—•- 1.5 x106 Btu/hr
- O 1.0 x106 Btu/hr
10 15 20 25 30 35 40 45 50
PERCENT FUEL DILUTION
Rgure 4. Effect of FIR at Three Firing Rates; NOx vs. Dilution
7B-28
-------
180
160
140
O 120
CO
®
100
0. 80
0.
O w
z
40
20
0
I | I r I 7 | T T I 1 [ I I I
O
O- — -, ^ 0% WB FGR
\
\
\
\
FIRING RATE: 2.0 MMBtu/hr
APH: 488 "F
MINIMUM O2
-A—NOMINAL OFA
-O- NO OFA
NO
N
0% WB FGR
\
15% WBFGR
~~O
15% WB FGR
10 15 20 25 30 35 40 45 50
PERCENT FUEL DILUTION
Figure 3A. FIR With and Without Overfire Air
CM
O
a?
n
©
E
Q.
a.
x~
O
z
IOU
160
140
120
100
80
60
40
20
n
! 0 FIRING RATE: 2.0 MMBtj/hr
~ APH: 488DF
"\ MINIMUM Og
\\ —A— NOMINAL OVERFIRE AIR
\ '\ — O- NOOVERFIREAIR
\O : ^^
\
\
\ .
\
- \. <$ "...
\ ' •. \ 0% FIR >'""-...
V ""-^ I o'vQ.
\ ' ... | \
'. \A \O 15% WBFGR
\ \ NO
0% WB FGR A V. \
^^-""-i. VN
'..!,..,! . . . I . , . , I . . , I . , , , I . , , , '
10
15
20
25
30
PERCENT FLUE GAS RECIRCULATION
(FGR + FIR)
Rgure 3B. Effect of FIR with and Without Overfire Air
7B-29
-------
o
D
O
UJ
DC
X
O
LU
O
DC
UJ
O.
100
90
80
70
60
50
40
30
20
10
0
FIRING RATE: 2.0 x 106Btu/hr
APH: 4S8°F
NOMINAL OFA
MINIMUM O2
10
15
20
25
PERCENT FLUE GAS RECIRCULATION
Figure 2. Maximum NOx Reduction with FIR
30
7B-30
-------
cf
vP
5-*
«
©
a.
a.
x"
O
110
100
90
80
70
60
50
40
30
20
10
0
0% WB FGR
23% WB FGR
FIRING RATE: 2.0 MMBtu/hr
APH: 488±6°F
NOMINAL OFA
MINIMUM O2
0 5 10 15 20 25 30 35 40 45 50
PERCENT FUEL DILUTION
Rgure 1A. Effect of FIR at Three FGR Rates; NOx vs. Percent Fuel Dilution
O
5?
CO
©
a.
a.
x"
O
z
110
100
90
80
70
60
50
40
30
20
10
0
0% WB FGR
0% FIR
FIRING RATE: 2.0 MMBtu/hr
APH: 488±6°F
NOMINAL OFA
MINIMUM 02
7
15%WBFGR
23% WB FGR
10
15
20
25
30
PERCENT FLUE GAS RECIRCULATION
(FGR + FIR)
Rgure 1B. Effect of FIR at Three FGR Rates; NOx vs. Total FGR
7B-31
-------
TABLE 1
MAXIMUM NO. REDUCTIONS WITH FIR
Windbox
FGR,%
0
15
23
NO
0% FIR
89.2
40.6
35.3
. & 3% O,
Max FIR
37.8
21.9
17.0
% Reduction
57.6
46.1
51.8
TABLE 2
COMPARATIVE NOS REDUCTIONS;
AIR INJECTION VS FLUE GAS INJECTION
Air Injection
0 MAX 0
FIR FIR %Reduction FIR
Flue Gas Injection
MAX
FIR %Reduction
0% FOR
94.1
73.£
21.6
89.2
33.1
62.9
15% FOR 41.0
31.2
23.9
40.6
21.9
46.1
NOTES 1. Firing Rate = 2.0 x 10" Btu/hr
2. Nominal OFA
3. Minimum O2
4. APH = 490 °F
7B-32
-------
ADVANCED REBURNING FOR NOX CONTROL
IN COAL FIRED BOILERS
S. L. Chen
W. R. Seeker
R. Payne
Energy and Environmental Research Corporation
18 Mason
Irvine, California 92718
(714)859-8851
-------
ADVANCED REBURNING FOR NOX CONTROL
IN COAL FIRED BOILERS
ABSTRACT
This paper summarizes an experimental study which was conducted to investigate the
chemical constraints of the reburning process and identify advanced reburning
configurations for optimal NOX reduction in coal-fired boilers. Tests were performed
initially on a bench scale tunnel furnace to characterize and optimize the fuel-rich
reburning zone and fuel-lean burnout zone independently. Based on the results, an
advanced reburning process was designed which integrated reburning with selective
reducing agent injection to enhance the burnout zone efficiency. The concept was
subsequently tested in a pilot scale facility and yielded over 80 percent reduction in
NOX emissions.
7B-35
-------
INTRODUCTION
Reburning is an NOX control technology which uses fuel to reduce N0.1"A The main heat
release zone can be operated normally to achieve optimum combustion conditions without
regard for NOX control. With reburning, a fraction of the fuel is injected above the
main heat release zone. Hydrocarbon radicals from combustion of reburning fuel react
with nitric oxide to form molecular nitrogen. This process occurs best in the absence
of oxygen. Thus sufficient reburning fuel, between 15 and 20 percent of the total heat
input, must be added to produce an oxygen deficient reburning zone. Subsequently, air
is provided to combust fuel fragments which remain at the exit of this zone. Since
reduced nitrogen species NH3 and HCN are also present, air addition may allow a further
NO,, reduction.
X
Previous studies showed that 60 percent reduction in NOX emissions could be achieved
with natural gas reburning.5 Recently research has been conducted to examine and
enhance the NOX reduction chemistry in the burnout zone.6 The burnout zone can be
considered as an excess-air "flame" burning the remaining fuel fragments from the
reburning zone. Oxidation of the fuel fragments, particularly CO, could generate a
significant amount of radicals via chain branching:
CO + OH = C02 + H
H + 02 = OH + 0
0 + H20 = OH + OH
These radicals play an important role in the conversion of XN species to N2 or NO
during burnout.
Figure 1 is an experimental examination of the burnout zone chemistry, in particular,
the conversion efficiency of NH3 to N2. The rich zone was assumed to supply 600 ppm
each of NO and NH3, or an N to NO ratio of 1.0. Under excess air conditions, ammonia
gas was mixed with various amounts of CO and injected at temperatures between 1300 and
2200°F. The solid symbols represent the injection of NH3 alone, which is basically a
7B-36
-------
simulation of Thermal De-N0x. For the open symbols, 0.2 percent CO was included with
NH3, thereby yielding a burnout like environment. The presence of CO lowered the
optimum temperature for NOX reduction from 1800°F to 1500°F. It is readily apparent
that a reduction in the burnout temperature from the 2200-2400°F normally employed in
the reburning process would increase the conversion efficiency of NH3 + NO to N2
because of the presence of CO.
This paper summarizes the results of a pilot scale study which was undertaken to
investigate the possibility of positive synergism between the injection of selective
reducing agents, such as ammonium sulfate, to provide the reducing specie NH3,and
combustion modifications, such as reburning,to serve as the source of CO.
EXPERIMENTAL
The 3.0 MWt, down-fired tower furnace5 used in the pilot-scale investigations was
refractory-lined and water jacketed with inside dimensions of 1.2 x 1.2 x 8.0 m. The
four main diffusion burners each consisted of an inner pipe for axial primary fuel
injection and an outer pipe, equipped with swirl vanes, for the main combustion air.
This four burner array produced relatively uniform velocity and composition profiles
at the primary zone exit. The furnace contained seven rows of ports for reburning fuel
and burnout air injection. The temperature profile was manipulated by insertion of
cooling panels, positioned against the furnace walls. The reburning fuel and burnout
air injectors were designed to maintain jet mixing similarity between the pilot-scale
furnace and a full scale boiler based on empirical correlations for entrainment rate
and jet penetration.
Exhaust gas samples were withdrawn through a stainless steel, water-jacketed probe and
analyzed for NOX (chemiluminescence), 02 (paramagnetic), C0/C02 (NDIR), and S02 (NDUV).
A water jacketed probe with an internal water quench spray near the front end was used
for extracting in-flame samples. Gas phase HCN and NH3 species were collected in a gas
washing unit and subsequently analyzed for CN" and dissolved ammonia using specific ion
electrodes. Gas temperatures were characterized with a suction pyrometer.
RESULTS
Recent studies6 have suggested that the key parameters for the enhancements of burnout
zone chemistry in staged combustion or reburning are:
f Reaction temperature (850°C)
0 CO levels (0.5% or less), and
• NH3 species.
7B-37
-------
Advanced Reburninq
Apparently the conventional reburning process does not provide the required
environment. An advanced reburning process, which combines reburning with selective
NOX reduction (SNR) via ammonium sulfate injection, was designed. Figure 2 shows two
hybrid schemes with 20 percent and 10 percent gas reburning, respectively. With 20
percent reburning (SR2 = 0.9), the burnout air was divided into two streams to yield
an SR3 of 1.03 and an SRt of 1.15. With 10 percent reburning, the reburning zone
stoichiometry (SR2) was 1.03 and the burnout air stoichiometry (SRt) was 1.15. In both
cases, an aqueous solution of ammonium sulfate was atomized with the final burnout air
an injected at 850°C at an N to NO molar ratio of 1.5.
Verification Tests
Figure 3 shows the advanced reburning results obtained with natural gas as the primary
fuel. The natural gas fired at 4.5 x 106 Btu/hr was doped with NH3 to yield primary
NOX levels of 600 and 400 ppm (dry, 0 percent 02). Twenty and ten percent advanced gas
reburning were applied, respectively. Similar final emissions, approximately 125 ppm
NOX, were achieved with both concepts. Experiments were subsequently carried out with
an Indiana coal as the primary fuel. The Indiana coal produced an uncontrolled NOX
emission of 800 ppm (dry, 0 percent 02) at 15 percent excess air. The primary NOX at
SR, = 1.13 was 680 ppm. Figure 4 presents the results and indicate that as seen in
the bench scale studies6, both advanced concepts were equally effective in NOX
reductions. It is apparent that there exists a tradeoff between natural gas premiums
and the cost of ammonium sulfate.
Ammonia Slip and $0x Emissions
The injection of ammonium sulfate into the furnace has a potential of producing
unwanted emissions such as NH3 and S02/S03. A series of exhaust measurements were made
to evaluate the slip of ammonia using selective ion electrode and the emissions of S02
and S03 via controlled condensation during the Indiana coal tests. Exhaust NH3
concentrations were negligible in all cases including those obtained with Utah coal and
natural gas as the primary fuel. Higher S02 emissions were obtained with 10 percent
gas reburning. However, the uncontrolled S02 level was maintained with 20 percent gas
reburning due to dilution. No increase in S03 emissions was observed for both cases,
suggesting favorable conversion of the sulfate to S02.
Thus, there exists a control strategy to prevent the increase in S02 emissions due to
injections of ammonium sulfate. For the application of advanced reburning to high
7B-38
-------
sulfur coals, 10 percent gas reburning is recommended, whereas for low sulfur coal
applications, the 20 percent gas reburning concept is preferred.
CONCLUSIONS
In summary, these results suggest that selective reducing agents can be combined with
combustion modification techniques to provide NOX reductions that are larger than those
that are possible by applying the technologies simultaneously but separately. By using
the stoichiometry control associated with reburning to produce a slightly fuel rich
region for selective reducing agent injection, reductions can be achieved at relatively
low temperatures without the use of stainless steel or other catalysts.
ACKNOWLEDGEMENTS
This work was primarily supported by the U.S. Department of Energy, Pittsburgh Energy
Technology Center (Contract No. DE-AC22-86PC91025) with Dr. Richard Tischer as the
Project Manager. We also would like to acknowledge the contributions of our colleague
Mr. Loc Ho in conducting the experiments.
DISCLAIMER
Reference herein to any specific commercial product, process, or service by trade name,
trademark, manufacturer, or otherwise, does not necessarily constitute or imply its
endorsement, recommendation, or favoring by the United States Government or any agency
thereof. The views and options of authors expressed herein do not necessarily state
or reflect those of the United States Government or any agency thereof.
REFERENCES
1. Myerson, A. L., et al., Sixth Symposium (International) on Combustion, The
Combustion Institute, 1957, p. 154.
2. Reed, R. D., "Process for the Disposal of Nitrogen Oxide." John Zink Company,
U.S. Patent 1274637, 1969.
3. Wendt, J. 0. L., et al., Fourteenth Symposium (International) on Combustion, the
Combustion Institute, 1973, p. 897.
4. Takahashi, Y., et al., "Development of Mitsubishi 'MACT' In-Furnace NOX Removal
Process." Presented at the U.S.-Japan NOX Information Exchange, Tokyo, Japan, May
25-30, 1981. Published in Mitsubishi Heavy Industries, Ltd. Technical Review,
Vol. 18, No. 2.
5. Chen, S. L., et al., 21st Intl. Symp., Combustion Institute, 1986, p. 1159.
6. Chen, S. L., et al, JAPCA. Vol. 39, No. 10 (1989).
7B-39
-------
Z
UJ
O
5 80
ox 60
X
O
2
40
20
SR - 1.1
(NOx)p - 600 PPM (DRY, O% O2)
- 1.0
O NH3 + 0.2% CO
• NH3 ONLY
1400 1600 1800 2000 2200
PEAK INJECTION TEMPERATURE <°F)
Figure 1. NH3 conversion in the "burnout zone."
FUEL + AIR
FUEL + AIR
20%
NAT. GAS
AIR
AIR AND
110
0.90
1.03
1.15
10%
NAT. GAS
AIR AND
1.13
1.03
1.15
\
Figure 2. Advanced reburning concepts.
7B-40
-------
Reburning
V77\ Advanced Reburning
i
600
~ 500
CM
0
o 400
•o
^ 300
Q.
Q.
0* 200
z
4 f\f\
100
n
,
Primary NOX
^H
^
Ml
_
^/,
.g Primary NOX
= 4
1 I
^ t>
O ^
00
Q)
cc
^
2
%,
20% Gas
10% Gas
Figure 3. Results obtained with natural gas as primary fuel
7B-41
-------
800
7, 600
oc
Q 400
Q.
Q.
200
UNCONTROLLED NO
INDIANA COAL
CO
O
0
CM
CO
O
O
UJ
CC
CO
CO
Figure 4. Pilot scale results with Indiana coal
7B-42
-------
LARGE SCALE TRIALS AND DEVELOPMENT OF FUEL STAGING IN A 160 MW COAL
FIRED BOILER
H. Spliethoff
Universitat Stuttgart
Institut fUr Verfahrenstechnik und Dampfkesselwesen
Prof. Dr. techn. R. Dolezal
Pfaffenwaldring 23
7000 Stuttgart 80, Germany
-------
LARGE SCALE TRIALS AND DEVELOPMENT OF FUEL
STAGING IN A 160 MW COAL FIRED BOILER
H. Spliethoff
Universitat Stuttgart
Institut fur Verfahrenstechnik und Dampfkesselwesen
Prof. Dr. techn. R. DoleZal
Pfaffenwaldring 23
7000 Stuttgart 80, Germany
ABSTRACT
In a study under the contract of the Saarbergwerke AG it is planned to achieve NOX
emissions near 200 mg N02/m3, i.e. 98 ppm NO without expensive DENOX technology.
By application of retrofit primary methods (air staging, flue gas recirculation)
the NOX emissions from the coal fired boiler Fenne 3 (slag tap furnace, 160 MW
electric power) could be reduced from 900 to 520 ppm NO at 5% 02- In the year 1988
the boiler was equipped with an arrangement for fuel staging. Reburning fuel is
coal gas with 50 % H2 and 25 % Cffy.
Experiments from September 1988 to July 1990 showed that reburning can reduce NOX
emissions from 520 ppm to 180 ppm NO (5% 02). The influence of different parameters
(primary zone stoichiometry, reducing zone stoichiometry etc.) was investigated.
The reduction zone stoichiometry and the reburn fuel mixing were pointed out to be
the most important parameters for low NOX emissions by reburning /!/.
In order to optimize reburning the following work has been done:
t distribution of flue gas concentrations was measured (primary zone,
reducing zone, burnout zone),
• reburning fuel mixing was optimized by three-dimensional fluid flow
computations,
• fuel staging with synthetic gases was examined in a 0.5 MW test facility
and
• the influence of ammonia addition into the reduction zone was investigated.
By optimizing the reburning gas injection and by addition of ammonia to the
reduction zone the NOX emissions could be reduced to a minimum of 130 ppm NO (5%
02) up to now. Reburning has only a slight impact on the burnout of the coal. The
carbon content in the fly ash is less than five percent.
7B-45
-------
INTRODUCTION
In the last years there had been large efforts to lower NOX emissions from
stationary combustion sources. For combustion systems with a thermal load of more
than 300 MW NOX emissions of 200 mg/m3 NO calculated as N02 (98 ppm NO) at 5% 02
(molten ash furnace) or 6% 02 (try ash furnace) are demanded in Germany.
Applied and commonly used techniques for NOX abatement can be devided in
• combustion modifications,
• selective non catalytic reduction (SNCR) by ammonia or urea and
• selective catalytic reduction (SCR) by ammonia.
Due to the short period for retrofitting existing power plants and equipping new
power plants with NOX abatement techniques, most German hard coal fired power
stations are or will be soon equipped with the SCR DENOX technology.
Measures to influence the NOX emissions of coal furnaces by combustion
modifications are:
• optimized boiler operation (low oxygen operation),
• flue gas recirculation,
• air staging (single burner or in the furnace) and
• fuel staging, reburning (single burner or in the furnace).
In the past years air staging has proved to be an effective method for NOX
reduction. For German lignite it seems possible to achieve the required NOX
emissions without expensive DENOX-technology by improved air staging in the furnace
/2/. A further technique of minimizing NOX emissions is a method called fuel
staging, reburning or In-Furnace NOX Reduction. Results of fuel staging in test
facilities are very promising.
A published application of reburning to coal combustion furnaces is the MACT
process. By fuel staging at a coal dust furnace NOX emissions of less than 150 ppm
could be achieved /3/.
Figure 1 shows the principle of fuel staging. In the first zone, which is the main
heat release zone, the fuel can be burnt under fuel lean conditions to ensure
complete burnout. The addition of reburning fuel creates a fuel rich, NOX reduction
zone. The reduction of nitrogen oxides is initiated by hydrocarbon radicals. In the
final zone the combustion is completed by addition of air.
DESCRIPTION OF THE PROJECT "BRENNSTOFFTRENNSTUFUNG (BTS)H
To lower the NOx emissions in coal dust furnaces the project "Combined minimizing
of NOX production and reduction of formed NOX - Brennstofftrennstufung (translated:
Fuel Splitting and Staging)" has been initialized.
Coal is divided by a devolatilization process in a reduction gas with volatile
nitrogen and the remaining coal (char). Both fractions are burned in a fuel staged
combustion with char as primary fuel and pyrolysis gas as reburning fuel.
7B-46
-------
The project consists of several steps:
• Investigation of reburning at a 0.5 MW gas fired combustion facility with
synthetic fuel,
• Large scale tests of reburning with coal gas as reburning fuel in a slag
tap furnace,
• Investigation of the process "fuel splitting and staging* in a small scale
test facility.
The investigations of reburning in the 0.5 MW combustion facility with synthetic
fuels and the trials at the 160 MWe] slag tap furnace are subjects of this report.
Results of performance and emissions of the process "Fuel Splitting and Staging"
in a small scale test facility are soon expected.
MECHANISMS GOVERNING NOX PRODUCTION AND REDUCTION AT FUEL STAGING
Figure 2 shows the NOX production and NOX reduction mechanism for the three zones
of a fuel staged combustion with coal dust as primary fuel and gas as reburning
fuel.
In the main heat release zone the formation of NOX is mainly due to the fuel
nitrogen. During devolatilization of coal a part of fuel nitrogen is released with
the pyrolysis gases, the other part remains in the coal char. The amount of
nitrogen released with the pyrolysis products depends on coal properties (volatile
matter content) and temperature. The volatile nitrogen and char nitrogen are
converted to NOX in a different way and in different amounts.
The volatile nitrogen quickly forms the intermediate species HCN, which is then
converted in a slow reaction to NH3- Depending on the fuel/air ratio and on
temperature, NH3 is either reduced to molecular nitrogen or it forms NO. The degree
of nitrogen oxide formation from the volatile fuel nitrogen can be affected by
primary combustion modifications, such as air staging or flue gas recirculation.
The production of NOX from Char-N is generally low with conversion rates between
10 and 20 percent. The heterogeneous production of nitrogen oxide is less sensitive
to process parameters as the formation from volatile sources. Therefore it is
assumed, that Char-N is responsible for minimum NOX emissions, which can not be
lowered.
In the reduction zone the nitrogen oxides formed in the main heat release zone are
reduced by homogeneous reactions. If the reburning fuel contains hydrocarbons, the
gas phase reduction of NO is initiated by CH-j in a fast reaction
NO + CHi —> HCN + products. (1)
This fast step is followed by the relatively slow conversion of HCN to NH-j. This
reaction is significant for the overall reduction.
NH-j then either forms NO by reaction with 0 or OH radicals
NHi + 0 / OH —> NO + products (2)
or is reduced by NO to N2
7B-47
-------
NHi + NO —> N2 + products (3).
Because of the fuel rich atmosphere in the reduction zone reaction (3) is
predominant.
Investigations of Bose /4/ confirm, that the gas phase reactions are dominant in
fuel rich combustion zones and that the heterogeneous reduction is of minor
importance for coal dust combustion.
The gas phase nitrogen reactions in the first and second stage are quite the same,
as to be seen in figure 2.
By addition of air the N-containing species NO, HCN and NH^ are converted to NOX
in the burnout zone. NO and HCN are almost completely transformed to NOX, NH-j only
in a very small amount /5/. If the burnout air is added to the flue gas at
temperatures of about 900 °C, a further NOX reduction is possible.
REBURNING WITH SYNTHETIC COAL GASES IN A TEST FACILITY
In order to study the reduction efficiency with a pyrolysis gas as reburning fuel
experimental investigations were carried out under the contract of the
Saarbergwerke in a gas fired combustion facility at the University of Karlsruhe.
The synthetic pyrolysis gas consists of 60% H2 and 30% CH4- The watercooled
combustion chamber is described elsewhere /6/. The residence time in the reducing
atmosphere is about one second, the flue gas temperature at the location of gas
injection is about 1300 °C, at the location of air injection about 900 °C. The
stoichiometric ratio of the first fuel lean zone is \\ = 1.1 with a measured NOX
level after the first stage of 600 ppm. The overall stoichiometric ratio was kept
constant at ^3 = 1.2.
The keypoint of the tests was to evaluate the influence of ammonia addition to the
reburning fuel, as pyrolysis gases contain nitrogen species such as NH3-
Furthermore the pilot scale results are compared to the results of reburning in the
slag tap furnace in order to demonstrate optimization potential for the large scale
application. The experiments at a pilot scale test facility allow the variation of
parameters which cannot be changed at a utility power plant.
Earlier investigations showed, that the addition of a nitrogen species such as NH3
to a reburn fuel makes no difference at the optimum stoichiometry X2> but outside
this optimum the N containing reburn fuel resulted in higher NOX emissions /7/.
Figure 3 shows the final NOX emissions and the corresponding measured nitrogen
species after the reduction zone for using a reburn fuel containing no NH3, 1.5%
and 3 % NH3- For pure pyrolysis gas (0% NH3), NOX is reduced from 600 ppm (5% 02)
after the primary zone to 115 ppm after the burnout zone at X2 = 0.85. The addition
of 3 Vol% ammonia results in a shift of the optimum stoichiometry to A2)0pt = °-89
and a further reduction of the total NOX emissions to 60 ppm NOX (5% 02). The
corresponding N-species of the reduction zone show an increased reduction of NOX,
the concentration of NH3 rises drastically for X2 < A2)0pt, while the HCN emission
7B-48
-------
is not affected by the increased NH3 input. At the optimum stoichiometry without
ammonia, N-species of 130 ppm NO and 20 ppm NH3 are converted to 115 ppm NOX in the
burnout zone. For the maximum NH3 addition (3%) 60 ppm NO and 100 ppm NH3 form 60
ppm final NOX emissions.
Further experiments at the University of Karlsruhe /8/ outside this project with
natural gas as reburn fuel showed similar trends as in the case of ammonia
addition.
In contrast to other investigations the addition of ammonia to the reburning gas
enhances the reduction efficiency of reburning significantly. The discrepancy of
the presented results to those of other authors are believed to be caused by the
high temperature of about 1300 °C in the reduction zone, optimized mixing injection
and a residence time of one second. These conditions favour the formation of NH3
rather than HCN in the reduction zone for all three cases studied. While the NO of
the reduction zone is completely converted to NOX in the burnout zone, the
conversion of NH3 to NOX is small. The high conversion of HCN to NOX can be
avoided. This is in agreement to Tagaki, who reports a low conversion rate of NH3
to NOX and a high rate of HCN to NOX /5/.
INVESTIGATION OF REBURNING IN A 160 MW SLAG TAP FURNACE
In order to show the effectiveness of NOX reduction with pyrolysis gas as reburning
fuel and to find out the main parameters, the fuel staged combustion was applied
to a 160 MWe] power plant.
Furnace design and performance of the trials
Figure 4 shows the furnace of the steam generator and the zones of the fuel staged
combustion. The furnace consists of two molten ash chambers. The two burner rows,
consisting of four air staged burners, are arranged in two stages at each chamber.
To lower the NOX emissions of the molten ash chambers, the old unstaged burners had
been retrofitted by air staged burners. As a second method to reduce NOX by primary
measures, flue gas recirculation to the pulverizer mills had been installed. The
achievable NOX emissions by primary NOX reduction had to be evaluated as the basic
emission level before starting reburning.
After the fuel lean combustion of coal dust in the molten ash chambers reduction
gas can be injected to the flue gas by twelve nozzles for each chamber. The
arrangement of reburning fuel injection is shown in figure 5. The flue gas at the
end of the first zone has a temperature of about 1400 - 1500 °C. The injected fuel
is coke oven gas, which mainly consists of H2 (50%) and CH4 (25%). The addition of
reburning fuel causes the formation of fuel radicals, which start the NOX reduction
process. The residence time of the flue gas under fuel rich conditions in the
reduction zone is about one second at maximum thermal load.
By addition of burnout air at the end of the separated flue gas channels behind the
chambers the combustion is completed.
7B-49
-------
The entire experimental program from September 1988 till September 1990 included
trials without reburning to determine the initial emissions, trials with coal gas
as reburning fuel and experiments with ammonia addition into the reduction zone and
to the burnout zone. During the experiments about 100 process variables were
measured for On-Line monitoring and stored for later data analysis. Besides the
operational flue gas analysis in the furnace and at the stack, flue gas
concentrations and temperatures were measured in cross sections behind the
chambers, in the reduction zone and in the burnout zone for a better understanding
of NOX formation and destruction and to point out possibilities for optimization.
As the results of NOX emissions by reburning are a function of the stoichiometry
of the main heat release zone, the reburning zone and the burnout zone, the
stoichiometries of the zones had to be calculated accurately. While the air flows
and the reburning gas flows were measured, a measurement of the pulverized coal
flow was not available.
The air stream, necessary for the stoichiometric combustion of coal, is
proportional to the ratio of thermal power and the efficiency of steam generation.
Vair,stoich. = A * Pth / ^F
The thermal power P^h can be calculated by the superheater and reheater Jetstream
and the temperatures and pressures necessary for determining the corresponding
enthalpies. The efficiency of steam generation r?p is dominated by the heat loss of
the flue gas. The variable A gives the necessary air for combustion of coal with
a thermal input of 1 MW. A is constant for a large range of coals and not varying
with changing water or ash contents of the coal.
The stoichiometries computed by this method were verified by comparison with the
stoichiometries calculated from flue gas composition.
Results
Primary methods. The results of the primary NOX reduction (air staging at the
burner, flue gas recirculation) are summarized in figure 6. The NO emissions are
plotted as a function of the recirculated flue gas stream. Each point in figure 6
relates to a value, measured every ten seconds.
The application of air staging is for this slag tap furnace the more effective
method for reducing NOX emissions than the application of flue gas recirculation.
By air staging at the burner without flue gas recirculation the NOX emissions could
be lowered from 644 ppm to 500 ppm NO (5% 02). When 10% of the whole flue gas was
recirculated to the mills, air staging caused a reduction from 560 to 490 ppm NO.
By application of different methods for NOX reduction at the same time the
effectiveness of the single measure decreases.
7B-50
-------
The initial emissions for the reburning trials were 500 - 550 ppm, which could be
obtained by air staging at the burner and by flue gas recirculation. The initial
emissions refer to an unstaged operation in the furnace, what means that the
stoichiometry of the chambers and the overall stoichiometry were kept constant at
1.2.
Reburning results. Figure 7 shows the result of reburning with varying gas streams.
Each value corresponds to a trial of at least two hours. At a steam generation
power near full load the NOX emissions without reduction gas are 520 ppm for a
stoichiometry of 1.2. By air staging in the furnace and at a constant thermal load
the NOX emissions could be lowered to 460 ppm (\\ = 1.1, ^3 = 1.2). The reduction
of the thermal power caused in this test no significant change of NO emissions.
Other tests showed a maximum influence of reduced thermal load of 20 ppm NO for the
staged case. The reduction of the thermal power corresponds to the heat input of
the maximum gas stream.
By increasing the gas stream at a constant first zone stoichiometry, the NO
emissions decrease sharply. By supplying twenty percent of the total heat input by
the reburning fuel, NOX emissions of 180 ppm (5% 03) could be achieved. The unburnt
carbon in the fly ash was 4%.
The dominating parameter for reburning is the stoichiometry of the reduction zone.
Figure 8 shows NO emissions for trials in 1989 and 1990 without measures for an
improved reburning as described later. The trials were performed at different
primary zone stoichiometries, burnout zone stoichiometries and different thermal
loads. If sufficient air is provided for the coal combustion in the molten ash
chambers, reburning caused no increase of unburnt carbon in the fly ash. The
operation of the first zone with a stoichiometry greater than 1.09 for the
existing, non optimized coal dust distribution to the burners secured a
satisfactory burnout of the coal below the 5% threshold value.
Figure 9 compares the measured NO concentrations in the reduction zone without
reburning gas and with a reburning fuel of 20% of the total thermal input. Without
reburning gas an uniform distribution of NO concentrations of 550 ppm (at 0% 02)
was measured in the cross section before burnout injection. By addition of
reburning fuel of 20 % the cross section measurements showed NO concentrations
between 100 and 300 ppm NO. The concentrations of NO are corresponding to the
measured concentrations of CO, H2 and CmHn (Figure 10). Near the furnace wall on
the side of the gas injection (left side in figure 9 and 10) and in the middle of
the furnace the concentrations of the combustible species are maximum. The non-
uniform distribution is mainly caused by an incomplete mixing of the reburning gas
with the flue gas from the molten ash chambers. Further cross section measurements
of flue gas concentrations behind the chambers show that the coal dust distribution
to the burners also contributes to an unbalanced distribution in the reduction
zone. In the scope of the investigations the coal dust/air distribution was not
optimized, but it is assumed that a control of coal dust supply to the individual
burners can contribute to obtain lower NOX emissions.
7B-51
-------
Mixing calculations. Experimental investigations of Kolb /6/ with natural gas as
reburning fuel pointed out the influence of mixing on the NOX emissions for a fuel
staged combustion. By an optimized mixing of reburning gas he could achieve a 50%
reduction compared to the case without optimization. The effect of mixing phenomena
on the results at the test facility of the University of Karlsruhe was minimized
by an optimized mixing. The reduction zone in the slag tap furnace "Fenne 3" at an
optimum mean stoichiometry consists of areas with stoichiometries, which differ
from the optimum stoichiometry, so resulting in higher NOX emissions.
In order to improve the mixing of the reburning gas and to optimize NO reduction,
mixing of the reburning fuel was calculated by three-dimensional fluid flow
computations.
The grid used for the computations is shown in figure 11. Because of the symmetry
of the furnace the fluid flow was calculated for a half of one chamber. As the
combustion of coal dust is mainly completed in the chambers, the computation
disregards heat transfer processes by reaction and radiation.
The choice of the computation domain considers the asymmetric distribution of the
velocities (Figure 12) at the location of reduction gas injection. This is caused
by the return of flue gas from the chambers to the upstreaming gas in the first
furnace duct. In the cross section above gas addition an non-uniform distribution
of velocities can be seen with maximum velocities near the side wall and the wall
opposite to the gas nozzles. At the wall near the gas nozzles recirculation takes
place. In the following cross sections the velocities are more balanced, but still
showing basically the same tendencies.
The calculated stoichiometries in figure 13a confirm the measured distribution at
a cross section at the end of the reduction zone. As it was evaluated in the test
facility with a reburning fuel containing ammonia, NOX reduction is optimum at \2
= 0.9 and satisfactory for a reduction 0.82 < \2 < 0.92.
The computations indicate that the area with a stoichiometry for a satisfactory
reduction covers only 15% of the cross section. In 35 % of the cross section the
flue gas atmosphere is fuel lean.
In order to improve gas injection the cooling air duct of the gas nozzles should
be connected to the existing flue gas recirculation. Before installation the
influence of an increased mixing momentum on the stoichiometry distribution was
computated, as shown in figure 13b.
With flue gas as additional mixing momentum the area with a satisfactory reduction
covers 60 % of the cross section at the end of the reduction zone. These results
of calculation were the reason to install a provisional connection of the flue gas
recirculation to the gas nozzles. A comparison of measured and calculated
stoichiometries showed a good agreement /9/.
Trials of improved reburning. The impact of an increased mixing momentum on the
final NO emissions is shown in figure 14. The decrease in NO emissions in this test
was about 25 ppm. The effect of the more uniform distribution of reduction gas on
7B-52
-------
the NO concentrations measured at the end of the reduction zone is depicted in
figure 15. With flue gas as additional mixing momentum the average NO
concentrations are reduced by 40 ppm. An increased reduction of local NO
concentrations seems to be equalized by an increased conversion of the N-species
of the reduction to NO in the burnout zone.
The recirculation of flue gas provided the possibility of ammonia addition into the
reduction zone. In order to quench the flue gas, water or ammonia water can be
injected into the flue gas. For these tests a 15% NH3 concentration was used.
The results confirmed the positive effect of ammonia on NOX reduction. The
experiment shown in figure 16 was carried out at a reduced thermal load in order
to examine a wider range of reducing zone stoichiometries. In the case without
ammonia addition (with flue gas) no NOX minimum could be determined, with ammonia
injection the NOX emissions were minimum at \2 = 0.89. Only for very fuel rich
conditions in the reduction zone \$ < 0.85 (reduction gas fraction > 25%) ammonia
addition leads to higher NO emissions. Figure 16 also demonstrates the effect of
burnout stoichiometry ^3. A decrease of X3 from 1.2 to 1.1 causes a decrease in the
NO emissions for the case with and without ammonia addition. The unburnt carbon in
the fly ash was less than 4%.
Laser measurements /10/ of NH3 concentrations in the flue gas at the end of the
furnace detected in no case a measurable ammonia slip.
The addition of ammonia to the burnout air had only a positive effect for higher
NO emissions or stoichiometries \2 > 0-92 (Figure 17). The temperature of the flue
gas after burnout air injection is between 1000 and 1150 °C, measured at full
thermal load over the complete cross section of the furnace.
The reported results refer to a two chamber operation. In one chamber operation
lower emissions could be determined, as shown in figure 18. Each value in figure
18 corresponds to one test over several hours.
The difference between one chamber and two chamber operation is the possible use
of an air stream to the chamber out of operation as a further burnout air, so that
in one chamber operation the burnout air can be added in two stages. In one chamber
operation minimum emissions of 130 ppm at 5 % 02 could be obtained at
stoichiometries of the burnout zone beetween >3 = 1.05 - 1.1 (without regarding the
air from the chamber out of operation).
In figure 19 the unburnt carbon in the fly ash is plotted as a function of the
reduction zone stoichiometry for the one chamber tests.
CONCLUSIONS
By application of reburning to a slag tap furnace a NO reduction from 520 ppm to
minimum emissions of 130 ppm were obtained. The investigations pointed out the
strong influence of reduction zone stoichiometry on the NO emissions. Mixing of
reburn fuel has to be optimized and burnout zone stoichiometry should be as low as
possible to achieve low NOX emissions.
7B-53
-------
For the slag tap furnace "Ferine 3" there exists a further NOX reduction potential
by
• optimizing the reburn fuel mixing into the reduction zone,
• optimizing of the coal dust distribution to the burners,
• arranging the burnout air injection in at least two stages and by
• addition of ammonia above the reburning gas injection.
Measures to increase the fineness of the coal dust would allow to minimize the
reburning fuel fraction.
ACKNOWLEDGEMENTS
This work was conducted under the contract of the Saarbergwerke AG with financial
support of the federal Ministry of Research and Technology (BMFT), Germany.
REFERENCES
1. H. Spliethoff. "NOx-Minderung durch Brennstoffstufung mit kohlesta'mmigen
Reduktionsgasen." VDI-Bericht 765, 1989, pp. 217-230
2. K.R.G. Hein, D. Kallmeyer. "Stand der NOx-Minderung bei braunkohlebefeuerten
GroBkesselanlagen." VGB Kraftwerkstechnik, June 1989, pp 591-596
3. M. Araoka, A. Iwanaga, M. Sakai. "Application of Mitsubishi "Advanced MACT "
In-Furnace Removal Process." 1987 Joint Symposium on Stationary Combustion NOx-
Control, New Orleans 1987
4. A.C. Bose, J.O.L. Wendt. "Pulverized Coal Combustion: Fuel Nitrogen Mechanics
in the rich Post-Flame." 22ndt Symp. (Int.) on Combustion, The Combustion
Institute, 1988, pp 1127-1134
5. T. Tagaki, T. Tatsumi, M. Ogasawara. "Nitric Oxide Formation from Fuel Nitrogen
in Staged Combustion: Roles of HCN and NHi." Combustion and Flame 35, 1979, pp
17-25
6. T. Kolb, W. Leuckel. "Reduction of NOx Emission in Turbulent Combustion by Fuel
Staging / Effects of Mixing and Stoichiometry in the Reduction Zone."
22nd Symp. (Int.) on Combustion, The Combustion Institute, 1988, pp 1193-1203
7. S.L. Chen, J.M. McCarthy, W.D. Clark, M.P. Heap, W.R. Seeker, D.W. Pershing.
"Bench and Pilot Scale Process Evaluation of Reburning for In-Furnace NOx-
Reduction"
21st Symp.(Int) on Combustion, The Combustion Institute, 1986, pp. 1159-1169
8. J. Ritz, T. Kolb, P. Jahnson, W. Leuckel. "Reduction of NOx Emission by Fuel
Staging Effect of Ammonia Addition to the Reburn Fuel." Joint Meeting of the
British and French Section of the Combustion Institute (1989), Rouen, France
9. H. Spliethoff, B. Epple, D. Renner. "Einmischung von Reduktionsbrennstoff oder
Reduktionsmitteln in technische Feuerungen" 6. TECFLAM Seminar, Stuttgart 1990
10. H. Hemberger, H. Neckel, J. Wolfrum. "LasermeBtechnik und mathematische
Simulation von SekundarmaBnahmen zur NOx-Minderung in Kraftwerken." 3. TECFLAM
Seminar, Karlsruhe 1987
7B-54
-------
Main Fuel / Air
Reduction Fuel
Burnout Air
Primary Zone
X > 1
Reduction Zone
X < 1
Burnout Zone
X > 1
Figure 1. Principle of fuel staging (reburning)
MAIN HEAT RELEASE ZONE REDUCTION ZONE
BURNOUT ZONE
COAL DUST
AIR
REDUCTION GAS
BURNOUT AIR
CharN I
FuelN
\
Volatile N j OH, (
Figure 2. NOX production and reduction for a fuel staged combustion with coal as
primary fuel and gas as reburning fuel
7B-55
-------
CD
cn
CO
300
E
OH
200
d
_o
'to
100
X
O
1.5
NH3 (VolX)
o* NO/NO,
0.80 0,85 0,90 0,95
Stoichiometry Reduction Zone \2
O
N
"8
•-
I
-------
Burnout
Air
Coaldust
burner \
Reduction Gas
Figure 4. Furnace of the 160 MWe] steam generator Fenne 3
Molten Ash Chamber 1
A
o
o
oo
o
o
(M
I f t f f I I M t I t
I t t M I M t M 1
Molten Ash Chamber 2
Coal Gas
Cooling Air /
Flue Gas
9068
Figure 5. Reduction gas nozzles
7B-57
-------
1 1
ex
ex
•—
CM
O
5?
e
_o
°S
'6
I
.? Unstaged Burner Operation
i
v
>'>_ _
**. m-
";;; 1
i Staged Burner Operation
i t 4£v"
*• m «j3S5*
'-'^: !»• •'^^&'
1 ?f
0 10000 20000 30000 40000 50000 60000
Flue Gas Recirculation (to the mills ) f m3/h 1
Figure 6. Results of air staging (burner) and flue gas
recirculation (to the mills)
r-,
0-
OJ
o
.V
in
0
z
DCJU
550
500
450
400
350
300
250
200
150
100
50
n
LOAD
92% ^ Unstaged
927 MJ?^ D S^S6*1 Operation
73% mFumaCe
Reburn Fuel Fraction
•81% ^& 10%
' 86% ^ 14%
C*
'91% o^ 19%
Stoichiometry Primary Zone
Figure 7. Reduction by reburning -
influence of reburn fuel fraction
btau
'E 550
D.
Q.
1-1 500
450
400
350
300
^250
O
.v 200
in
o 150
inn
Two Chamber Operation
Non optimized Reburning o <^
<3£>
o°o o
%^O °
&* 0 0<>
^^^^
&&%
^^r
*•
.8 .9 1 1.1 1.2
Stoichiometry Reduction Zone
Figure 8. Trials in 1988 and 1989
7B-58
-------
Cmm
without
1200
ppm
NO (05$ 02)reburning
'0
I
280 . .
ppm With
NO reburning
4200
[mm]
Figure 9. Cross section measurement of NO in the reduction zone with and
without reburning gas (half cross section behind one chamber, gas injection
is located 12 meters below the depicted cross section on the left side)
Figure 10. Cross section measurement of unburnt gas in the reduction
zone (half cross section behind one chamber, gas injection is
located 12 meters below the depicted cross section on the left side)
7B-59
-------
Figure 11. Grid for fluid
flow computations
Figure 12. Calculated distribution
of velocities
0,93 < X < 1,0
0,87 < X < 0,93
0,82 < X < 0,87 '
0,77 < X < 0,82 I
X > 1,0
t MM !
0,93 < X < 1,0
-N- ^
0,87 < X < 0,93
HIM I
a) Without Flue Gas Momentum b) With Flue Gas Momentum
Figure 13. Calculated distribution of stoichiometries without and
with flue gas as mixing momentum
7B-60
-------
350
|—I
E 325
a
D.
"-1 300
275
250
225
200
~ 175
C\J
O
.V 150
in
^^
~ 125
100
Boiler Load 92%
Reburn Fuel
Fraction 19%
X Without Flue Gas Momentum
V With Flue Gas Momentum
.8 .85 .9 .95 1
Stoichiometry Reduction Zone
Figure 14. Effect of flue gas momentum
on final NOX emissions
280
ppm Without Flue
NO Gas Momentum
0
middle of the furnace /
4300 /•
3800 /•
Momentum
3300 /
2800 /
2300 /
Figure 15. Effect of flue gas momentum on local
NOX emissions in the reduction zone
7B-61
-------
JDU
'e 325
a.
Q_
1-1 300
275
250
225
200
~ 175
OJ
O
.\- 150
in
i 125
i on
Boiler Load 78-86%
V With Flue Gas Momentum
A With NH3 Addition to Flue Gas
V A \3 = 1,2
T A \3 = 1,1
/
/
//
/ /
/* /^ '
^^—£>
'"vT^^
--*-
25% 20% 15% Reburn Fuel
Fraction
JDK)
1 — 1
E 325
Q.
D.
1-1 300
275
250
225
200
~ 175
OJ
o
.V 150
in
o 125
i not
Boiler Load 78 - 86%
X Without Flue Gas Momentum
O With NH3 Addition to Burnout Air
25% 20% 15% Reburn Fuel
Fraction
100.6 .85 .9 .95 1 11JIJ.8 .85 .9 .95 1
Sto i ch i omet ry Reduction Zone St o i ch i ome t r y Reduction Zone
Figure 16. Effect of ammonia Figure 17. Effect of ammonia
addition to the flue gas addition to the burnout air
5001 in
'E 450
CL
CL
^ 400
350
300
250
200
~ 150
OJ
o
.v 100
in
o 50
~z.
Pi
One Chamber Operation
NH3 Addition to the Flue Gas
\3 = 1,05 - 1,2
A
.
A
&
A
A &
A
A 4^A
i V
9
n
.v 8
^
x 7
M
H
x B
L_
5
c J
.^
c 4
0 ^
J3
(0 g
0 J
ID 2
c
-D 1
D
n
One Chamber Operation
NH3 Addition to the Flue Gas
A
A X3 = 1,05 - 1,2
A
A
A
A A A
& A A
A A
"
^ A
A
AA
i .b .9 i 1.1 ^7 .8 -;g i ,•;•,
Stoichiometry Reduction Zone
Figure 18. NOx emissions for one
chamber operation with ammonia
addition to the flue gas
Stoichiometry Reduction Zone
Figure 19. Unburnt carbon in the
fly ash for one chamber operation
(Corresponding to Figure 18)
7B-62
-------
COMPUTER MODELING OF N20 PRODUCTION BY COMBUSTION SYSTEMS
Richard K. Lyon, Jerald A. Cole, John C. Kramlich,
and Wm. Lanier
Energy and Environmental Research Corporation
18 Mason
Irvine, CA 92718-2798
-------
COMPUTER MODELING OF NaO PRODUCTION BY COMBUSTION SYSTEMS
Richard K. Lyon, Jerald A. Cole, John C. Kramlich, and Wm. Stephen Lanier
Energy and Environmental Research Corporation
IB Mason
Irvine CA, 92718-2798
ABSTRACT
The observed rate of increase of NaO (0.181/. to 0.26'/. annually) is a
matter of environmental concern. While it is generally agreed that this
increase is a result of human activity, there is no consensus as to the
relative importance of different sources. Several studies have suggested that
pulverized coal fired combustion systems might be responsible, but the high
levels of NeO found in these studies were later found to be an artifact, the
results of chemical reactions which occur during sample aging. Measurements in
which precautions are taken against this problem show very low NeO levels for
flue gas from pulverized coal firing but do show substantial NeO concentrations
for fluid bed combustion.
In this paper computer modeling calculations are done for two mechanisms
of NeO production, the selective reduction of NO by HCN and sample aging. The
former plausibly accounts for the production of NS0 in fluid bed combustion and
may also be responsible for the small but apparently real amounts of NS0 found
in flue gas from pulverized coal firing. Calculations for sample aging,
however, show that preventing this mechanism from producing small amounts of
NE>O may be substantially more difficult than was initially believed. Thus
sample aging may also account for the small amounts of NS0 presently found in
flue gas from pulverized coal firing.
There have been speculations in the literature that the flue gas from
pulverized coal firing may be an important indirect source of N^O, i.e. it was
speculated that chemical reactions which occur during sample aging may also
occur in the flue gas after it is released to the atmosphere. Our calculations
indicated that this does occur but only to a very minor extent.
7B-65
-------
INTRODUCTION
The observed rate of increase of N^O (0.181/. to 0.267, annually) is a
matter of concern both because NP0 is a greenhouse gas and because it has a
major and unfavorable influence on the ozone layer (1,2,3). While it is
generally agreed that this increase is a result of human activity, there is no
consensus as to the relative importance of different sources. While McElroy's
calculations ( 3 , *t ) suggest that denitrification of chemical fertilizers could
account for the observed increase, others have criticized his calculations as
an order of magnitude too high (5,6). Weiss and Craig (7), Pierotti and
Rasmussen (8), Hae et al (9), and C. Castaldinin et al (10), have all reported
measurements of N^O emissions by large stationary combustion systems, i.e.
pulverized coal fired utility boilers and the like (11). For combustion systems
fired with fuels containing chemically bound nitrogen (i.e. coal and heavy oil)
NF0 levels of approximately 25'/. of the NO emissions were found and there was a
strong suggestion that emissions at this level would be sufficient to explain
the observed increase.
Recent experimental and computer modeling studies (12,13), however, cast
doubt on this conclusion. In all the studies mentioned above, grab samples of
flue gas which contained both NO and SOs were analyzed by GC several hours or
days after being taken. Table 1 shows literature values for the rate constants
and/or equilibrium constants of a number of chemical reactions. These
reactions are all well established processes. Figure 1 from reference 13 shows
the results of modeling calculations done with this set of reactions. The
prediction of these calculations is, that as the sample ages, the NO in the
sample is converted to NO^, which undergoes solution phase reduction by sulfite
ion, first to nitrite ion and then to the N0~ ion, with the N0~ ions then
reacting with each other to form NeO. The amount of NE0 which this completely
a prior model predicts is in reasonable agreement with the amount observed
during the aging of a sample.
Thus it is entirely possible that the NeO concentrations reported in
references 7 -11 are largely or entirely artifacts. As discussed in references
I'*, 15 and 16, recent measurements have been done in which precautions to
prevent this artifact were taken. For conventional utility combustion systems
N.-.?0 levels of only Ippm were typically found, but considerably higher levels
have been found for fluid bed combustion systems. While NeO emissions of Ippm
would not appear to be of environmental concern, the mechanism by which they
are formed is still of scientific interest and the higher levels found for
fluid bed combustors are potentially an environmental concern.
One of the issues to be addressed in this paper is the mechanism by
which this N,?0 formation occurs. The other issue to be addressed herewith
relates to the environmental importance of the NO/NOs/sulfite reaction
mechanism. As is pointed out in reference 16 the absence of NeO in the flue
gases which combustion systems discharge to the atmosphere does not necessarily
mean that these systems are not important sources of NaO. If the
NO/NOe/sulfite mechanism is important in nature, the NO and 502 emissions of
combustion systems may cause substantial NE0 production after the flue gases
enter the environment.
7B-66
-------
COMPUTER MODELING METHODS
Calculations were done with the reaction rate model shown in Table 1
using an Acuchem program (17). Additional calculations were also done with the
model shown in Table 2 using the PC version of ChemKin developed by Albert
Chang of Stanford University (18).
RESULTS AND DISCUSSION
Mechanism of Direct N20 Production during Pulverized Coal Firing
As discussed above in recent measurements of N50 in flue gases of
pulverized coal fired systems precautions were taken against NS0 formation
during sample aging. Since these measurements show greatly reduced but still
apparently real amounts of N^O one might conclude that some small production of
NpO does in fact occur during pulverized coal firing. Since it is well proven
that fluid bed combustion produces large amounts of NeO one might plausible
concluded that whatever mechanism is involved there, is also operative to a
small degree during pulverized coal firing. Alternatively one might conclude
that the precautions taken against the production of NaO during sample aging
were largely but not completely effective.
The production of N^O by sample aging shown in Figure 1 is
oversimplified in one important respect: in Figure 1 it was assumed that all
the NOx in the sample is initially present as NO. Figure 2 shows calculations
for the removal of NOx from the vapor phase by reaction with sulfite ion
solution for two cases, a gas mixture containing 600ppm NO and one containing
5^*0 ppm NO plus 60ppm N0e. While the former shows a slow steady decay of the
NOx, in the latter case there is an initial drop which consumes much of the
N0;=. Figure 3 shows the corresponding calculations for the production of hlNDs.
in the liguid phase. As one might expect, when N0a is not initially present,
HNOa is formed slowly and only after an induction period, while when N0e is
initially present, there is a burst of HNOe formation at the start of the
reaction. As shown in Figure ^ when N0e is initially absent, NeO is produced
only after a significant induction, but when it is present, the formation of
Nfc.0 begins immediately. Indeed when NOK is initially present the sample need
only age for 12 seconds to produce 2ppm NeO.
Thus for samples which initially contain NOS it is considerably more
difficult to avoid the production of NE0 by sample aging. Consequently, if one
tests one's experimental procedures using synthetic gas mixtures which contain
NO but no IMOH, these procedures may appear adequate to prevent NaO production
during the sampling process, but still fail for real flue gases which do
contain NeO.
In this regard, it is interesting to note, that in reference 15,
measured NsO/NOx ratios of 0.01 or less were found for 10 different coal fired
installations, but for a gas turbine a value of 0.21 was found. If the NaO
found in these measurements is a result of inadequate precautions against
sample aging, one would expect the highest N;=0/N0x ratio to be found for the
installation in which the NOx contained the largest fraction NO^.. It is well
known that the NOx emitted by gas turbines can contain a much larger fraction
of N0a than found in other combustion systems.
7B-67
-------
Indirect NgQ Production during Pulverized Coal Firing
As mentioned above there is a question of whether or not the NOx and S02
in flue gas may not represent an indirect source of NaO. When flue gas exits
the top of a stack, it both mixes with the atmosphere and cools to a
temperature that allows some of the water vapor it contains to condense. Thus
two competing processes occur, i.e. formation of an aqueous phase allows the
processes which produced N^O in aging laboratory samples to occur in the flue
gas, but mixing with the ambient atmosphere will rapidly quench those
processes. Thus one can imagine two ways in which flue gas can act as an
indirect source of NeO; some NS0 production can occur immediately after release
to the atmosphere and a much slower N£,0 production might occur after the mixing
with the atmosphere via NOx and S0e reacting in clouds.
The former is a complex process and would be difficult to model
accurately but from the calculations shown in Figure ^ it seems likely that it
is a real but minor source of NeO. In order to do calculations for the
production of NeO by reaction of NOx and S0e once they have been diluted to
ambient atmospheric concentrations a set of typical conditions was assumed.
Thus ambient concentrations of 6ppb and 10 ppb were assumed for NOS and S0e
respectively. L, the ratio of liquid phase to gaseous phase, was taken at ^.8
x lO"7, a typical value for a cloud. It was also assumed that the reaction
of NOp with SOE to form Nf?0 was in competition with other reactions and that
the most important of these was the reaction of N0e with OH to form HNOa. The
ambient concentration of OH free radicals in the cloud was assumed to be 1.7 x
lO'6 molecules/cc and a rate constant of 1.1 x 10-»! was used for the reaction
NOs. + OH = HN03. (IB)
Figure 5 shows the results of these calculations for a case in which the
aqueous phase was assumed to have an initial pH of 7. The NOE + OH = HN03
reaction is found to be faster than NE0 formation by a factor of more than
101*. Assuming an initial pH of less than 7 made NaO formation even less
important. Thus production of N^O from NOx and SOe after they have mixed in
the ambient atmosphere is trivial and combustion systems are indirect sources
of N^O only to the minor extent that NaO forms prior to the mixing of the flue
gas with the atmosphere.
N50 Production during Fluid Bed Combustion
While the very small concentrations of NeO currently being found in the
flue gas of pulverized coal fired systems may or may not be real, the fact that
fluid bed combustion can produce large concentrations of NaO seems to be well
proven. Reference 19 reports an interesting set of experiments which may
provide an explanation for this high NeO production. In reference 19 it is
reported that substantial NO reductions can occur in the free board of a fluid
bed combustion system and that these reductions can occur at temperatures as
low as 1050C.K and reaction times as short as O.S sec. Since these NO reductions
occurred in the presence of V/, Os, some form of selective noncatalytic
reduction is clearly involved, but the observed NO reduction does not appear to
be due to reaction with NH3. Thus the mechanism by which the NO was reduced is
unclear.
7B-68
-------
Figure 6, quoted from reference 20 shows the result of flame modeling
calculations done with a reaction mechanism very similar to that shown in Table
2. The model's prediction is that there exists a narrow range of temperatures
in which HCN selectively reduces NO, the product of this reduction being N20.
Reference 20 also reports experimental results which confirm this prediction.
Based on these results reference 20 suggested that NeO in the flue gases
from pulverized coal firing was produced by the following mechanism. Nitrogen
containing char is produced in the primary combustion. Some of this char
escapes the primary combustion zone and reacts to form HCN down stream at lower
temperature where the reduction of NO by HCN to form NaO is favorable. This
reaction only produces NsO in a narrow range of temperatures because at
temperatures above this range N^O decomposes and at temperatures below the
range the HCN/NO reaction does not occur.
Looking at Figure 6 one might suppose that this mechanism for NeO
production is not applicable to fluid bed combustion systems because they
operate below the temperature window. Figure 7, however, shows that the
temperature window for N^O production is a sensitive function of the reaction
time. Selective reduction of NO to N^O by HCN can occur in the free board of a
fluid bed combustion system and thus may be the explanation of the NO removal
reported by reference 19.
Practical Implications
Fluid bed combustion is generally regarded as a developing technology
and hence the fact that fluid bed combustors may emit N^O might seem to be a
potential rather than an actual problem. There is, however, one application in
which fluid bed combustion is a major industrial process, fluid bed catalytic
cracking. Within the cat cracking process the catalyst used to "crack" higher
molecular weight hydrocarbons to smaller molecules becomes coated with coke
and catalytic activity is restored by fluid bed combustion of the spent
catalyst. The temperature of this combustion is low to protect the catalyst
and consequently any NE0 produced would survive. Further, the amount of
nitrogen in the coke which is available for NaO is substantial, since
chemically bound nitrogen in the hydrocarbon feed goes preferentially into the
coke. Thus, since a major fraction of the world's total oil production goes
through the fluid bed cat cracking process, it is quite possible that this
process contributes significantly to anthroprogenic NeO emissions.
7B-69
-------
CONCLUSIONS
Recent measurements of the N&0 levels in flue gas from pulverized coal
firing typically show'concentrations of a few ppm. These NaO levels may be
real and the result of the reduction of NO by traces of HCN, or they may be an
artifact, a result of the fact that it is more difficult to prevent NE0
production by sample aging than was initially believed.
While there has been speculation that the emissions of S0e and NOx by
pulverized coal firing may indirectly be a substantial source of NS0, our
modeling calculations indicate that indirect NeO production is a minor process.
Fluid bed combustion, however, can produce substantial emissions of NaO
and our modeling calculations suggest that these emissions can plausibly be
explained in terms of the reduction of NO by HCN. It is regrettable that no
data are presently available for the production of NeO by fluid bed catalytic
cracker regenerators, since these installations may be a substantial source of
N,=,Q.
7B-70
-------
REFERENCES
1 Weiss., R.F., J. Beophy. Res., 86,7185-7195 (1981).
2 Khalil, M.A. and Rasmussen, R.A., Tellus, 35B, 161-169 (1983).
3 Mat-land, G., and Rotty, R.M. J.A.P.C.A., 35, 1033-1038 (1985).
4 McElroy, M.B., as reported by J. E. Bishop, The Wall Street Journal, p.9,
Nov. 13, 1975.
5 Crutzen, P.J., Geophys. Res. Lett., 3, 169-172 (1976).
6 Liu, S.C., Cicerone, R. J., Donahue, T.M., and Chameides, W.L., Geophys.
Res. Lett., 3, 157-160 (1976).
7 Weiss, R.F. and Craig, H., Geophys. Res. Lett., 3, 751-753, (1976).
8 Pierotti, D. and Rasmussen, R.A., Beophys. Res. Lett., 3, 265-267 (1976).
9 Hao, W.M., Wofsy, S.C., McElroy, N.B., Beer, J.M., Toqan, M.A., J. Geophy.
Res., 92, 3098-3194 (1987).
10 Castaldini, C., Water land, L.R., and Lips, H.I., EPA-600-7-86~003a, 1986.
11 Ryan, J. V., and R. K. Srivastava, EPA/IFP workshop on the emission of
nitrous oxide from fossil fuel combustion (Ruei1-Malmaison, France, June 1-2,
19B8), Rep. EPA-600/9-89-089, Environ. Prot. Agency, Research Triangle Park,
N.C., 1989. (Available as NTIS PB90-126038 from Natl. Technol. Inf. Serv.,
Springfield, Va.)
12 Muzio, L. J., and Kramlich, J. C., Geophysical Research Letters, 15, 1369-
1372, (1988)
13 Lyon, R. K., and Cole, J. A., Combustion and Flame, 77, 139 (1989)
14 Muzio, L. J., Montgomery, T. A., Samuelsen, G. S., Kramlich, J. C., Lyon,
R. K., and Kokkinos, A., 23rd Symposium (International) on Combustion, in
press.
15 Kokkinos, A, ECS UPDATE, Spring-Summer 1989, No 15 pp 8-10
16 Linak, W. P., et. al., Journal of Geophysical Research, 95, 7533-7541
(1990)
17 Braun, W., Herron, J. T. and Kahaner, D. K., Int. J. Chem. Kin. 20 51-62
(1988)
18 Baulch, D. L., Drysdale, D. D., Home, D. S. and Llyod, A. C., Evaluated
Rate Constants, Butterworth, 1976
19 Walsh, P. M., Chaung, T. Z., Dutta, A., Beer, J. M., and Sarofin, A. F..
19th Symposium (International) on Combustion, 1281-1289 (1982)
20 Kramlich,J. C., Cole, J. A., McCarthy, J. M., Lanier, W. S., and McSorley,
J. A., Combustion and Flame, 77, 375-384, (1989)
7B-71
-------
1000
DO
800 -
600
E
Q.
0.
400
200
EXPERIMENTAL
RESULTS
150 200 250 300 350
TIME, MINUTES
Figure 1. Experimental and Kinetic Calculations of N?0 Formation in Sampling Containers
-------
ppm
DO
I
-vl
CO
[NO2]0 = 60ppm
Figure 2. Modeling of NOX Removal from the Gas Phase by
Reaction with Sulfite ion
-------
-xl
DO
40
30
20
10
ppm
A
A
A
J L
0 10 20 30 40 50 60 70 80 90 100 110
t, sec
+ [NO2]o = 0 A [NO2]o = 60ppm
1000ppm SO2, 600ppm NOx, 0 or 60ppm NO2,
40C, 6.52rnole% liquid water
Concentration of HNO2 expressed
as ppm based on gas phase
Figure 3. Modeling of HN02 formation with N02 initially
present and absent
-------
-si
00
Al
01
40
30
20
10
ppm
0
0
Time to form 2ppm N2O • 12 seconds
100
200
300
t, sec
400
500
[NO2]o = 0 -*- [NO2]o = 60ppm
1000ppm SO2, GOOppm NOx, 0 or 60ppm NO2
40C, 6.52mole% liquid water
600
Figure 4. Modeling of N20 formation with N02 initially
present and absent
-------
-J
CD
-^J
CO
10
20
30 40 50 60 70
TIME, seconds X1000
[NO2]/[NO2]i -3-[N2O]/[NO2}\ X -\0000 —&~
80
90 100
[HNO3]/[N02]i
10ppb SO2, 6ppb NO2, Initial pH = 7
40C, L = 4.8E-7 ccL/ccG
pH at 10E+5 sec - 3.33
Figure 5. Competition between N20 formation and HN03
formation after the flue gas mixes with the
atmosphere
-------
800-
-------
-vl
en
I
~sl
CO
200
150 -•
100 -
ppm
1.05
t = 0.02 sec
11 115. 1.2 1.25 1.3
T, K (Thousands)
1.35
1.4
t = 0.04 sec
t = 0.10 sec
t = 0.20 sec
200ppm HCN, 600ppm NO, 10% O2, 5% H2O,
balance inert
Figure 7. Calculation of the Effect of Reaction Time on N20
Formation
-------
TABLE I
Chemical mechanism, rate constants and equilibrium constants at 25°C
(rate constants are in units of L/mol/s or L /mol /s)
12
Gas Phase Reaction
1 . NO + NO + 0
N02 + N02
Rate Constant
6.73 E + 3
Liquid Phase Reactions
N02 + HS03-
N02- + HS03
HS0
HS0
H0)
H2S03
2N02 + H20 = HN02
HNO,
HN02 + HS03-
NOS03-
H20
H
NOS03-
HNO + HNO
+' H20) = HNO
N2O
NOS03-
I- HS03-
9 . HNO (SO3) 2~ + H*
10. HNO(S03)2~ + H2
H20
f HNO(S03)
(+ H20) =
i = HONHSO,
2-
3.00 E+5
5.00 E+5
7.00 E+7
2.40 E+0
5.00 E+l
3.00 E+4
8.50 E+l
1.90 E-2
1.50 E-6
Equilibrium Processes
11.N02(gas) N02(aq)
12 . S02(gas) S02(aq)
Henry's Law Constants
H = 0.01 M/atm
H = 1.30 M/atm
13. S02(aq)
14.HNO,
H + HS03-
, H f NO -
15. HS04- = H' S04
Equilibrium Constants
K = 1.54 E-2 M
K = 5.10 E-4 M
K = 1.20 E-2 M
7B-79
-------
TABLE 2
ELEMENTARY REACTIONS USED IN MODELLING
REACTION
1 NH3+M=NH2+H+M
2 NH3+H=NH2+H2
3 NH3+0=NH2+OH
4 NH3+OH=NH2+H20
5 NH2+H=NH+H2
6 NH2+0=NH+OH
7 NH2+OH=NH+H20
8 NH2+02=HNO+OH
9 NH2+NONNH+OH
10 NH2+NO=N2+H20
11 NH2+HNONH3 + NO
12 NH2+NNH=N2+NH3
13 NNH+M=N2+H+M
14 NNH+NO=N2+HNO
15 NNH+OH=N2+H20
16 HNO+M=H+NO+M
17 HNO-t-OH=NO+H20
18 NH+02=HNO+0
19 OH+H2=H2O+H
20 H+02=OH+0
21 0+H2=OH+H
22 20H=0+H20
23 H+02+M=H02+M
H20/21./
24 H+H02=20H
25 0+HO2=02+OH
26 OH+H02=H20+02
27 H02+NO=N02+OH
28 N02+H=NO+OH
29 N02 + ONO+02
30 N02+M=NO+0+M
31 0+0+M=02+M
32 N20+H=N2+OH
32 N20+M=N2+O+M
33 N20+0=N2+02
34 N20+0=NO+NO
35 CO+OH=C02+H
36 CO+H02=C02+OH
37 CO+02=C02+0
38 CO+0+M=C02+M
39 NCO+0=NO+CO
40 NCO+NO=N20+CO
41 NCO+H=NH+CO
42 NCO+NH2=NH+HNCO
43 0+H2=HNCO+H
44 NCO+OH=NO+CO+H
45 HNCO+OH=NCO+H20
4t> HNCO+H=NH2+CO
47 HCN+OH=HNCO+H
-------
Session 8
OIL/GAS COMBUSTION APPLICATIONS
Chair: A. Kokkinos, EPRI
-------
LOW NOx LEVELS ACHIEVED BY IMPROVED COMBUSTION
MODIFICATION ON TWO 480 MW GAS-FIRED BOILERS
Mark D. McDannel, P.E.
Sheila M. Haythornthwaite
CARNOT
15991 Red Hill Avenue, Suite 110
Tustin California 92680
Michael D. Escarcega, P.E.
Barry L. Gil man, P. E.
Southern California Edison Company
2244 Walnut Grove Avenue
Rosemead, California 91770
-------
LOW NOX LEVELS ACHIEVED BY IMPROVED COMBUSTION
MODIFICATION ON TWO 480 MW GAS-FIRED BOILERS
Mark 0. McDannel, P.E.
Sheila M. Haythornthwaite
CARNOT
15991 Red Hill Avenue, Suite 110
Tustin, California 92680
Michael D. Escarcega, P.E.
Barry L. Gil man, P.E.
Southern California Edison Company
2244 Walnut Grove Avenue
Rosemead, California 91770
ABSTRACT
While most applications to meet new and emerging NOX regulations have focused on
retrofit technologies (low-NOx burners, urea, SCR), there are still opportunities for
additional NOX reduction via improved combustion optimization.
Southern California Edison, as part of their compliance efforts for a new NOX rule,
which ultimately requires NOX limits of approximately 20 ppmc, retained Carnot to
assist them in designing and conducting a combustion optimization program on two 480
MW gas-fired boilers. As a result of detailed combustion optimization test programs
on the two boilers, NOX was reduced by 24 to 56% over the load range at an average
cost-effectiveness of $.59/lb NOX. Through increased windbox FGR, improved BOOS
patterns and overfire air, NOX levels at full load were reduced from 91 to 62 ppmc.
These reductions will help SCE meet current and near-term NOX limits, and will
substantially reduce construction and operating costs of any future SCR systems.
8-1
-------
INTRODUCTION
While most applications to meet new and emerging NOX regulations have focused on
retrofit technologies (low-NOx burners, urea, SCR), there are still opportunities for
additional NOX reduction via improved combustion optimization.
Southern California Edison, as part of their overall compliance plan for South Coast
Air Quality Management District (SCAQMD) Rule 1135, retained Carnot to assist them in
designing and conducting combustion optimization programs on two 480 MW gas-fired
boilers (Alamitos 5 and Redondo 8). This paper presents the results of the two test
programs, which provided immediately implementable NOX reductions of 24 to 56% over
the unit load range at an average cost-effectiveness of $.59/lb NOX.
Included in the paper is a description of the technical and regulatory background on
NOX emissions from the two boilers, a description of the two boilers, a description
of the approach taken in designing and executing the program, the results of the
program, and a discussion of the results.
BACKGROUND
All of SCE's boilers in the South Coast Air Basin are subject to SCAQMD Rule 1135,
which includes system-wide 24-hour average NOX limits that start at 1.10 Ib NOx/MW-hr
(approximately 100 ppmc*) in 1990 and steps down to 0.25 Ib NOx/MW-hr (approximately
23 ppmc) in 1999. Additionally, Alamitos 5 and Redondo 8 are subject to rule 475,
which was passed in 1970 and limits NOX on gas fuel to 125 ppmc (approximately
1.38 Ib/MW-hr) for a 15-minute averaging period. Figure 1 presents a summary of NO
limits on these two boilers.
When the 125 ppmc limit was imposed, SCE implemented off-stoichiometric combustion
(overfire air ports and/or burners out of service) on 24 boilers in the South Coast
Air Basin, and additionally implemented flue gas recirculation (FGR) to the windbox
on four of these boilers, including Alamitos 5 and Redondo 8. Implementation of these
ppmc = parts per million by volume, corrected to 3% 02, on a dry basis
8-2
-------
techniques reduced NOX levels from approximately 900 ppmc to 100 ppmc on both
Alamitos 5 and Redondo 8.
In SCE's overall Rule 1135 compliance plan, there are a number of NOX reduction
efforts either planned or already evaluated on these two units. On Alamitos 5, urea
injection and installation of one row of low-NOx burners have been tested, and the
installation of a Selective Catalytic Reduction (SCR) system is planned. On
Redondo 8, an SCR system consisting of blocks of (honeycomb) catalyst placed in the
duct between the economizer and air preheater is scheduled for 1991. It is within
this context that combustion optimization was evaluated and implemented.
UNIT DESCRIPTION
Alamitos 5 and Redondo 8 are two of four identical 480 MW Babcock & Wilcox opposed-
fired units operated by SCE in the South Coast Air basin. The units are capable of
firing either natural gas or fuel oil. This program addresses gas firing only, since
Rule 1135 has limited application to oil firing and since oil is rarely burned.
Relevant details on the boilers are listed below:
• Manufacturer: Babcock & Wilcox
• Rated Capacity: 480 MW (net)
• Steam temperature: 1,000°F superheat and reheat
• Steam pressure: 3500 psig (supercritical)
• Burner arrangement (see Figure 2):
-- Opposed fired
-- 32 burners, 16 per wall
-- 4 rows of 4 burners each on each wall
-- furnace split by division wall
• NOX control:
-- third elevation of burners out of service
-- FGR to windbox
-- OFA ports
• Newly installed Rosemount digital control system
• 02 trim system in service
• CO trim system installed but not yet in service
PROGRAM DESCRIPTION
The objective of the program was to determine what level of NOX reductions could be
achieved by modifying and optimizing combustion and boiler operating conditions prior
8-3
-------
to the installation of SCR or other back-end NOX reduction technologies. Specific
benefits expected were:
1. Help meet Rule 1135 limits immediately.
2. By reducing inlet NO levels, reduce the size and cost of
future SCR installations.
A comprehensive program involving five discrete phases was designed. The five phases
are listed below, followed by a brief description of each phase:
• Records search
• Interview operating staff
• Physical inspection and repair
• Optimization test program
• Load following tests
Records Search
The first step of the program was to review available test and operating data on the
units to help plan the test program.
Interview Operating Staff
Interviews were held with station engineers, maintenance and instrumentation
supervisors, shift supervisors, and boiler operators to familiarize test personnel
with unit operation and to familiarize station personnel with the objectives of the
program. Unit operation was observed with at least two different shifts of operators.
Physical Inspection and Repair
Prior to performance of the combustion optimization test programs, thorough boiler
inspections were conducted during maintenance outages. The objectives of the
inspections and outages were to:
1. Evaluate the condition of all fireside operating equipment
including fans, dampers, and burners.
2. Identify any equipment requiring repairs or adjustments, and
verify that repairs were made.
3. Allow the test crew to become familiar with boiler design
and equipment.
4. Wash boiler to provide a known cleanliness lever.
8-4
-------
Performance of the inspections and repairs ensured that equipment problems would not
adversely impact unit operation during the test program.
Optimization Test Program
The optimization test programs consisted of 105 tests on Alamitos 5 and 51 tests on
Redondo 8. The test matrices were designed to evaluate the impact on NOX emissions
from the following variables:
• Unit load
• Excess 02
• Flue gas recirculation (FGR) to the windbox
• Overfire air ports
• Alternate BOOS patterns
t Air register throttling to selected burners
• Superheat/reheat proportioning dampers (Alamitos 5 only)
• Fan balancing
Each test included collection of gaseous emission data at the economizer exit, a full
set of unit operating data from the control room, and external unit data as needed
(damper positions, air register settings, windbox 02, etc.). For most tests,
North/South composite data was collected. This involved collecting average gaseous
data from the North side, average gaseous data from the South side, and a composite
sample. For selected tests, full 32-point gaseous traverses were performed. When
test conditions were established and unit data were collected, the impact of test
variables on unit heat rate was watched carefully. The need to isolate one test
variable at a time to determine its impact on combustion did result in some test
conditions where operation was not optimum; this was considered in evaluation of the
results.
Load Following Tests
The test programs on both units were concluded with two sets of load following tests.
These tests involved establishing recommended low-NOx operating conditions and
monitoring NOX, 02, and CO while ramping boiler load between 160 MW and 480 MW. The
purpose of these tests was to determine if the low-NOx operating modes could be
maintained, and expected NOX values seen, over the entire load range with no
operational problems.
8-5
-------
RESULTS
The results are presented separately for the two units, as follows. For the sake of
brevity, detailed impacts of individual test variables are presented only for
Redondo 8; similar results were obtained for Alamitos 5.
Redondo 8
The tests identified two modifications to baseline operation (as described under Unit
Operation) that resulted in significant NOX reductions over the full load range, and
two further modifications that resulted in small additional NOX reductions. The
modifications which reduced NOX significantly are:
t increasing flue gas recirculation to the windbox to
the maximum achievable; and
t minimizing excess 02 until CO formation is seen.
Modifications which produced smaller NOX reductions are:
• taking burner pair 6 out of service (while leaving the
air registers open); and
• opening of the OFA ports at 480 MW and during load
following tests.
The results of combining these techniques are summarized in Table 1, detailed in Table
2, and illustrated in Figure 3. Note that Figure 3 does not include the opening of
the OFA ports, which were only evaluated during the load following tests.
Increased Flue Gas Recirculation effects are shown in Figure 4. Test points on
Figure 4 are scattered somewhat due to the inclusion of all test variables. However,
the trend of NOX reduction with increased GR is clear. This is most notable at 480
MW. Higher GR was limited at this load because of a fan amp limit. If fan capacity
could be increased to enable 25% GR, the projected NOX would be approximately 40 ppm
@ 3% 02 (see dotted extension line on graph).
Minimizing excess 02 was performed at 250, 360, and 480 MW. The 02 setpoint for
minimum 02 was determined by gradually lowering excessive air until 100 to 200 ppm of
CO was seen consistently at that condition.
Table 2 shows the percent reduction attributable to minimizing 02 at the various
loads. This reduction increases with lower load, and more reduction may be possible
at 160 MW, where significant CO formation had not begun.
8-6
-------
Taking burner pair 6 out of service reduces the NOX fairly uniformly across the load
range, as shown in Table 2. Figure 5 shows graphically the impact on NOX of taking
6 OOS. The reduction caused by this modification is small, but the improvement in
boiler operation is significant. Figure 6 shows the CO level both with and without
6 OOS. At 480 MW extremely high CO was created with all burners in service; this was
removed by taking 6 pair OOS.
Another impact of this modification was to improve the excess 02 balance between the
north and south sides of the boiler. A series of tests led to the conclusion that
Burner 6 south is starved for air. This results in lower 02 and higher CO levels on
the south side. Taking Burner 6 out of service improved both 02 and CO balance
between the two sides.
Opening the overfire air ports at 480 MW reduced NOX by 7 ppm, or 11%. This condition
was established while at full load. Opening the OFA ports was not evaluated at other
loads due to difficulty in determining the positions of the ports early in the test
program. Once the open position was established by observing NOX reduction at 480 MW,
the ports were kept open for one set of load following tests. Figure 7 shows the
reduction achieved across the load range by opening the NOX ports. While this
reduction is small, the modification does not impact boiler operation, and could
easily be made a permanent operating condition.
Load following tests showed that optimum low-NOx conditions could be maintained over
the full unit load range, without any operating problems. The results of the load
following tests are shown in Figure 7. NOX levels are shown with NOX ports both open
and closed. A slight reduction with NOX ports open is seen over the entire load
range.
Other variables that were investigated during the program were air register throttling
on inboard burners to provide increased air flow to starved outer burners, and
alternate BOOS patterns. These tests provided insight into unit operation, but
implementation caused undesirable effects such as increased NOX, difficult operation,
or a large 02 or CO imbalance between the north and south sides of the boiler.
Alamitos 5
The tests on Alamitos 5 identified three modifications to baseline operation (as
described under Unit Operation) that resulted in significant reductions in NOX
emissions over the full load range: increased flue gas recirculation to the windbox,
opening of the OFA ports, and taking burner pair 6 out of service (while leaving the
8-7
-------
air registers open). The results of combining these three techniques are summarized
in Table 3, and illustrated in Figure 8.
The results show that substantial NOX reductions were achieved across the load range,
with the percentage reductions decreasing as unit load increases (from 56% at minimum
load to 27% at maximum load).
Table 4 shows the incremental reductions achieved by each of the three techniques.
The reductions achieved by each technique were cumulative across the full load range.
The largest reductions (11 to 36%) were achieved by increasing FGR to the windbox.
Reductions of 10 to 18% were achieved by taking Burner Pair 6 OOS, and reductions of
1 to 9% were achieved by opening the NOX ports.
Load following tests showed that these conditions could be maintained over the full
unit load range, without any operating problems. The results of the load following
tests are shown in Figure 9.
Other variables that were investigated during the program were excess 02 level,
superheat/reheat proportioning damper position, air register throttling on lower
burners to provide increased combustion staging, air register throttling on selected
burners in an effort to overcome an air/fuel imbalance, FD and GR fan biasing and
balancing, and alternate BOOS patterns. Those tests provided insight into unit
operation, but did not provide substantial NOX reductions.
Reductions in excess 02 did provide some NOX reductions, but the existing boiler 02
curve is so low (1% 02 over most of the load range) the 02 levels could only be reduced
approximately 0.2% before the onset of CO. Placing the CO trim control system in
service will allow maintenance of minimum 02 levels over the load range, and should
result in additional NOX reductions of 2 to 5% (based on minimum 02 tests conducted
during this program).
The tests also identified a significant north/south 02 imbalance in the furnace. A
series of tests led to the conclusion that the imbalance is mostly due to burner 6
North (an upper, rear, corner burner) being starved for air. The problem was
partially alleviated by taking the burner pair out of service for NOX control.
DISCUSSION
This section presents discussions on the potential impact of the three recommended
combustion modification techniques (increased windbox FGR, Burner 6 out of service,
minimum excess 02) on unit operation, including heat rate. This discussion applied
to both units.
8-8
-------
Heat Rate
Any change in operation should be evaluated in terms of its impact on unit heat rate.
Operating costs for a NOX technique can become significant if they have a significant
impact on boiler efficiency. Emissions data, unit operating data, heat rate factors
and fuel cost factors were combined to determine an operating cost in terms of $/lb
NOX reduced for increased FGR, taking Burner 6 out of service, and opening the NOX
ports. The cost benefit of reduced excess oxygen levels was also considered.
Tables 5 and 6 summarize the heat rate penalties, and present the operating cost of
the techniques combined in dollars per pound of NOX reduced. On Alamitos 5, heat rate
penalties of $0.34 to $0.83/lb NOX were seen. On Redondo 8, the only load at which
a cost is seen is 250 MW. Here NOX costs $0.31/lb reduction. At all other loads, the
heat rate is improved by reducing excess 02.
The results showed two areas in which unit heat rate penalties were incurred, and one
in which heat rate was improved: increasing FGR to the windbox increased auxiliary
power consumption, and taking Burner 6 out of service increased average excess 02
levels as measured by the test van. Minimizing 02 reduced the NOX level and improved
heat rate by lowering the excess air used.
It should be noted that these cost-effectiveness values are so low in part because
these techniques involve an incremental extension of NOX reduction techniques already
implemented on the boilers. Costs for boilers which do not already have windbox FGR
or some form of off-stoichiometric firing would be higher.
Other Impacts on Unit Operation
None of the four low NOX techniques used in this study had any deleterious effects on
unit operation that were detected during the test programs. When the techniques were
implemented unit load was stable, flame appearance and stability were acceptable, and
there were no significant changes in tube metal temperatures.
There are some areas in which the techniques might impact unit operation in the long
run. The most important may be a loss in load capacity safety margin while operating
with Burner 6 out of service. In the baseline condition there are 24 firing burners,
and with Burner 6 out of service there are 22 firing burners. If a burner pair trips
at full load, there would be either two or four fewer firing burners in service
(depending upon whether it is an upper or lower burner pair that trips). With Burner
6 out of service, it would be more likely that available unit load would be curtailed
if a burner pair tripped. Prior to implementing Burner 6 DOS, the magnitude of the
possible curtailments would need to be determined and an evaluation made of the
relative value of reduced NOX emissions vs. the risk of increased load curtailments.
8-9
-------
Another area of concern with taking Burner 6 out of service is that the increased heat
release rate per firing burner (an increase of 9% would occur) might cause overheating
in the burner throat area. This would have to be evaluated prior to implementation.
Implementation of increased flue gas recirculation to the windbox should be
coordinated with appropriate safeguards, since the booster fans have a high enough
capacity that they can blow out the flames at lower loads. New digital controllers
have been installed on the booster fans, the hopper control dampers, the FGR fans, and
superheat/reheat proportioning dampers. With the new booster fan controllers, curves
of damper position vs. unit load can be programmed in. However, to protect against
injecting too much FGR there should be a windbox 02 monitoring system. Such a system
could be either used for operator information or tied into the control system to
provide an alarm and/or feedback signal.
Operating with the OFA ports open should not provide any operation problems. As noted
before, it is currently difficult to access the OFA ports to open or close them. The
ports should be welded open. The chains currently installed do not allow easy
operation.
An important aspect to consider in applying combustion optimization techniques is the
boiler control system. These two boilers have newly installed digital control systems
that allow effective and safe control of the fuel and air systems within close
tolerances. On boilers with older control systems it may not be possible to achieve
such tight control.
CONCLUSIONS
The major conclusions of the program are:
1. Improved combustion optimization can provide
significant NOX reductions (23 to 56%) beyond those
achieved to meet compliance with the first generation
of SCAQMD NO, rules.
X
2. The incremental operating cost of these NOX reductions
is negligible (average of $.59/lb NOJ compared to
retrofit technologies. In some cases operating
savings are achieved due to excess 02 reductions.
3. These techniques can be implemented safely with no
adverse impact on unit operation.
8-10
-------
IS
©
a.
Q.
1SO
165
150
135
120
105
90
75
60
45
30
15
NOTE:
ASSUMES UNIT HEAT RATE --
9,000 Btu/kW-hr
1.5
•5
1970 1980 1990
YEAR
2000
Figure 1. Gas Fuel NOx Limits on Alamltos Unit 5 and Redondo Unit 8
WEST FIRING WALL (VIEW FROM INSIDE)
D
0
75 0
0
3S0
D
©
©8S
0
04S
D
0
5N0
0
1N0
D i
© ^
06N \
© •;
_2N
0 ;
« M< W ¥< M< -.W ..v
PLAN VIEW
FIRING WALL
FIRING WALL FIRING WALL
DIVISION
WALL
33'
FIRING WALL
48'
Figure 2. Burner and NOx Port Locations on Alamitos Unit 6
(Redondo 8 Is a Mirror Image)
8-11
-------
I
Q BASELINE
O BEST. BUHNER B 003
100 200 300 400
UNIT LOAD, MW NET
Figure 3. NOx versus Load for Baseline and Best
Conditions for Redondo Unit 8
100
90
80
70
60
50
40
3D
20
10
0
A 160 MW
O 250 MW
O 360 MW
D 080 UW
INCREASING FGR
WINDBOX 02, %
Figure 4. NOx vs. WindboxO for Redondo Unit 8
8-12
-------
o
*
n
®
I
100
M
SO
70
80
50
40
JO
20
10
O BA8ELME, BURNER S M SERVICE
A BEST. BURNER B M SERVICE
O BEST, BURNER 6 009
NOTE 3RD ELEVATION OF
BURNERS DOS FOB ALL TE3T3
100 200 JOO 400
UNIT LOAD, MW NET
Figure 5. Impact on NOx of Taking Burner 6
Out of Service for Redondo Unit 8
I
o"
u
—D— BASELINE, BURNER i IN SERVICE
- - A - BEST. BURNER • M SERVICE
- O - BEST, E DOS
NOTE: 3RD ELEVATION OF
BUHNER3 OO3 FOR ALL TE3T3
100 150 200 250 300 ISO 400 450 500
UNIT LOAD, MW NET
Figure 6. Impact on CO of Taking Burner 6 Out of
Service for Redondo Unit 8
8-13
-------
-O— BEST WITH OFA PORTS CUOSiD
-A- - BEST WITH OFA PORTS OPEN
100 200 300 400
UNIT LOAD, MW NET
Figure 7. NOx versus Load for Load Following Tests
at Best Conditions for Redondo Unit 8
I
O BASELINE, CLEAN
A BEST, CLEAN
- - -O- - - BASELINE. DIHTY
---O - B€ST, DIRTY
NOTE:
CLEAN AND DIRTY REFER
TO FURNACE CLEANLINESS
100 200 300 400
UNIT LOAD, MW NET
Figure 8. NOx versus Load for Baseline and Best
Conditions for Alamltos Unit 5
8-14
-------
C\J
CO
a
a
x"
O
100
90
80
70
60
50
40
30
20
10
100 200 300 400
UNIT LOAD, MW NET
500
Figure 9. NOx versus Load for Load Following Tests at Best
Conditions for Alamltos Unit 5
8-15
-------
TABLE 1
SUMMARY OF NOX REDUCTIONS
ACHIEVED IN REDONDO 8
COMBUSTION OPTIMIZATION PROGRAM
Unit Load
MW Net
Basel ine,
ppm NO 0 3% 02
Ib/MW-hr1
lb/MW-hr2
Best Case,
ppm NO @ 3% 02
lb/MW-hr1
lb/MW-hr2
% Reduction
(ppm 0 3% 02)
160
26
0.38
0.32
20
0.22
0.22
23%
250
39
0.55
0.47
24
0.28
0.29
38%
360
63
0.84
0.72
30
0.39
0.35
52%
480
88
1.19
1.02
55
0.73
0.64
38%
1 First lb/MW-hr number is calculated from plant CEM data divided by plant MW data
2 Second lb/MW-hr number is calculated from trailer N0x ppm and Rosemount heat rate
by I/O method
TABLE 2
PERCENT REDUCTION ACHIEVED BY THREE
NOX REDUCTION TECHNIQUES AT REDONDO UNIT 8
Unit Load
MW Net
Increased GR
to windbox
Minimize 02
Take Burner
6 OOS
Combining all 3
techniques
160
5%
21%*
5%
23%
250
26%
13%
7%
38%
360
43%
12%
5%
52%
480
37%
7%
6%
34%
* At 160 MW, 02 could be reduced further before significant CO formation
8-16
-------
TABLE 3
SUMMARY OF NOX REDUCTIONS
ACHIEVED IN ALAMITOS 5
COMBUSTION OPTIMIZATION PROGRAM
Unit Load
MW Net
Clean Furnace
Baseline NOX:
ppm @ 3% 02
Ib/MW-hr
Best Case NOX:
ppm @ 3% 02
Ib/MW-hr
% Reduction
Dirty Furnace
Baseline NOX:
ppm @ 3% 02
Ib/MW-hr
Best Case NO :
ppm @ 3% 02
Ib/MW-hr
% Reduction
150 250 360
32 51 59
0.42 0.61 0.67
---- * 29 35
0.35 0.40
4*3
-------
TABLE 4
PERCENT REDUCTIONS ACHIEVED
BY THE THREE NOX REDUCTION TECHNIQUES
ON ALAMITOS 5
Unit Load
MW Net
Increase GR to
windbox
Take Burner 6 DOS
Open NOX ports
Combined techniques
150
36%
18%
9%
56%
250
36%
10%
9%
43%
360
29%
13%
9%
41%
480
11%
13%
1%
27%
8-18
-------
TABLE 5
HEAT RATE PENALTIES ASSOCIATED WITH NO, REDUCTION TECHNIQUES
REDONDO UNIT 8
Load
Increase FGR
(higher aux. power)
Burner 6 DOS
(higher 02)
Minimum 02
Net heat rate
penalty (gain)
Avg. heat rate,
Btu/kW-hr*
Base hourly fuel
cost, $/hr**
Efficiency penalty (gain),
$/hr
Ib/hr NOX Reduced
$/lb NOX Reduced
160 MW
0.06%
0.12%
-0.48%
(0.30%)
10,209
5,717
($17)
12
(1.42)
250 MW
0.16%
0.04%
-0.04%
0.16%
9,645
8,439
$14
45
0.31
360 MW
0.33%
-0.04%
-0.32%
(0.03%)
9,327
11,752
($4)
135
(0.03)
480 MW
0.21%
0 . 08%
-0.32%
(0.03%)
9,415
15,817
($5)
167
(0.03)
* Average of data collected during test program
** Assumes $3.50/MMBtu fuel cost
8-19
-------
TABLE 6
HEAT RATE PENALTIES ASSOCIATED WITH NOX REDUCTION TECHNIQUES
ON ALAMITOS 5
Load
Increase FGR
(higher aux. power)
Burner 6 OOS
(higher 02)
NO Ports Open
(higher 02)
Net heat rate
penalty
Avg. heat rate,
Btu/kW-hr*
Base hourly fuel
cost, $/hr*
Efficiency penalty,
$/hr
Ib/hr N0x Reduced
$/lb NOX Reduced
150 MW
0.33%
0 . 28%
-0.10%
0.51%
10,880
5,710
$29
35
0.83
250 MW
0.20%
0.12%
0.14%
0 . 56%
9,820
8,590
$48
66
0.73
360 MW
0.21%
0.04%
0.12%
0.37%
9,430
11,880
$44
99
0.44
480 MW
0.06%
0.12%
0.12%
0.30%
9,320
15,660
$47
137
0.34
* Average of data collected during test program
** Assumes $3.50/MMBtu fuel cost
8-20
-------
NOx REDUCTION AND OPERATIONAL PERFORMANCE OF TWO FULL-SCALE
UTILITY GAS/OIL BURNER RETROFIT INSTALLATIONS
N. Bayard de Volo
L. Larsen
Energy Technology Consultants, Inc.
Irvine, California
L. Radak
R. Aichner
Southern California Edison Co.
Rosemead, California
A. Kokkinos
Electric Power Research Institute
Palo Alto, California
-------
NOx REDUCTION AND OPERATIONAL PERFORMANCE OF TWO FULL-SCALE
UTILITY GAS/OIL BURNER RETROFIT INSTALLATIONS
N. Bayard de Volo
L. Larsen
Energy Technology Consultants, Inc.
Irvine, California
L. Radak
R. Aichner
Southern California Edison Co.
Rosemead, California
A. Kokkinos
Electric Power Research Institute
Palo Alto, California
ABSTRACT
In 1989-90 Southern California Edison Company replaced the original burners
firing natural gas and residual oil fuels in two large, opposed-fired boilers of
different capacities and design. The replacement burners were manufactured by Todd
Combustion, Inc of Stamford, Connecticut. The principal objectives of the retrofit
were: 1) to improve flame shape and stability, and 2) to achieve NOx emission levels
with all burners in service at full load, in combination with Flue Gas Recirculation
(FGR), equal to or less than the levels previously achieved by Off-Stoichiometric
firing with FGR.
Tests were conducted on both boilers, firing gas and oil fuels separately, to
define the flame shape and stability and the NOx emissions over a wide range of
load, excess air and FGR rate for both pre- and post-retrofit configurations.
Further reduction in NOx emissions achievable with the new burners firing in an Off-
Stoichiometric mode, with FGR, was also determined over the same range of
operational variables.
This paper is an interim status report presenting preliminary results of the
pre- and post-retrofit testing program funded by SCE and EPRI.
8-23
-------
INTRODUCTION
In 1987 Southern California Edison Company (SCE) initiated projects to replace
existing gas/oil burners on two large boilers, Alamitos Generating Station, unit o,
and Ormond Beach Generating Station, Unit 2. The principal motivation in eacn case
was to improve flame quality (stability, attachment, etc.) over the load range, out
especially at low firing rates. Additional motivations included improving boiler
efficiency and reducing NOx emissions.
In order to define the actual improvements achieved by each retrofit, SCE
instituted a program to perform comprehensive testing of both boilers before and
after the burner retrofits. EPRI provided additional funds to expand the parametric
testing and to promote the dissemination of the NOx technology results to the
electric utility industry. Energy Technology Consultants, Inc. (ETEC) was retained
to provide consulting services to plan and conduct the testing program, to analyze
the test results and to report on the program findings. This paper is written to
present some preliminary results comparing pre- and post-retrofit NOx emissions for
natural gas and oil fuels. The program is still in progress and a considerable
portion of the post-retrofit testing remains to be completed for both gas and oil
fuels. Nevertheless, because there is currently so little public information
available on full-scale, Low-NOx gas/oil burner performance, it was thought to be
useful to present these preliminary results at this time.
Considerable success has been achieved by utilities having to comply with
restrictive NOx regulations applying to existing gas/oil fired units by implementing
Off-Stoichiometric (O.S.) firing. In this mode of operation, selected burners are
taken out-of-service (BOOS) while fuel flow is compensatingly increased to the
remaining burners to maintain boiler load requirements. As a result, the active
burner combustion process is made fuel rich and consequently NOx formation is
reduced. Although NOx emissions can be significantly reduced in this manner for
both gas and oil fuels, operational performance can also be degraded somewhat as a
consequence of having to raise excess air levels to maintain acceptable CO
concentrations on gas fuel and plume opacity/particulates on oil fuel. In addition,
a degradation in flame holding and stability can also result. SCE has employed O.S.
firing on all of its units for many years achieving significant reductions in NOx
emissions but has also experienced the deterioration of boiler performance and
combustion on selected units.
The basic concept of low NOx burners is to achieve fuel rich combustion, and
hence reduced NOx formation, by controlling local mixing of fuel and air. This
approach offers the promise of equaling or exceeding the NOx reduction capability of
O.S. firing while avoiding the possible performance and operational deficiencies
associated with the latter approach. The potential gains however must be balanced
against the capital cost of the burner retrofit in comparison to O.S. firing which
is implemented operationally without equipment expenditure.
This paper should be of interest to utilities who anticipate having a future
need to reduce NOx emissions from their gas/oil fired boilers. The subject program
represents one of the few instances in which data are to be developed for a low NOx
burner utility boiler installation and for which a comparison of the relative NOx
reduction capabilities and overall performance of the two NOx control approaches can
be established. It is for this reason that SCE and EPRI have jointly funded the
program reported herein.
8-24
-------
PRE-RETROFIT OPERATION
Unit 6 at the Alamitos Generating Station (AGS-6) is a B&W, opposed-fired,
gas/oil fuel boiler/turbine/generator set rated to produce 480 MWe. The boiler was
designed with 16 two-burner cells, arranged in two rows of four cells on the front
and rear furnace walls. Ring type gas spud burners and constant-differential,
pressure-atomized, swirl-tip oil burners were provided. Dampered overfire air
ports, fed from the windbox, were provided above each top elevation burner cell.
Two gas recirculation fans were originally provided to extract flue gases from the
economizer exit and direct those gases to the furnace hopper area as an aid to
controlling steam temperatures at low firing rates (FGR).
The boiler began operation in 1966 and subsequently became subject to a Los
Angeles County APCD regulation limiting NOx emissions to 225 ppm (dry, 3% 0?) for
natural gas fuel and 325 ppm for fuel oil. The uncontrolled NOx emission with gas
fuel at full load was approximately 700 ppm. NOx emissions were reduced to within
the regulatory limit on both fuels by implementing O.S. firing. The optimum firing
configuration was determined to be with the bottom burners of the upper cells (i.e.
3rd elevation) out of service for both gas and oil fuels and with the OFA ports
closed.
Subsequently, the APCD NOx emission limit for natural gas was reduced to 125
ppm and 225 ppm for oil. Two booster fans were installed to extract flue gas from
the main gas recirculation fan outlets and to inject the flue gas into the
combustion air through orifices in the flow-metering air-foils within the air ducts
between the air preheaters and the windbox as depicted schematically in Figure 1.
The combination of windbox FGR (WFGR) and O.S. operation achieved compliance with
the reduced emissions limits for both fuels and the boiler has been operated in this
mode ever since.
Unit 2 at the Ormond Beach Generating Station (OBGS-2) is a Foster Wheeler,
opposed-fired, gas/oil fuel boiler/turbine/generator set capable of producing 800
gross MWe. The boiler was constructed with two sets of 2-burner cells at each of
four elevations on the front and rear furnace walls. Each two-burner cell is fed by
one gas and one oil supply pipe/valve, however, each individual burner had its own
air register control. Each burner had a constant-differential pressure-atomized,
swirl-tip oil gun and a cane-type gas burner with (8) eight canes fed from an
external ring manifold.
The boiler was originally designed to produce NOx emissions below 500 ppm
(dry, 3% OJ for both gas and oil fuels. This was to be accomplished by including
overfire air (OFA) ports fed by the windbox. In 1969 it appeared that the Ventura
County APCD intended to establish a NOx emission limit of 250 ppm (dry, 3%02) for
both fuels. During construction of the OBGS units (1 & 2) WFGR was added to both
units. For each unit one dual-inlet fan extracted flue gas from the economizer
outlet ducts and injected the gas into the two combustion air ducts leading to the
windbox. The general configuration is depicted schematically in Figure 2. The WFGR
injection is accomplished through an array of perforated pipes located within each
air supply duct a few feet upstream of the rear windbox.
Upon commercial operation of OBGS-2 in 1973, compliance with the 250 ppm NOx
limit was achieved with either fuel at full load by a combination of FGR, OFA and
limited O.S. firing. In 1975 the Ventura County APCD reduced the allowable NOx
emissions with gas fuel to 125 ppm (dry, 3% 0?). Because oil fuel was used
exclusively for several years, compliance with the 125 ppm limit for gas fuel was
not demonstrated until 1977. Compliance was achieved by operation with 8 BOOS,
8-25
-------
maximum FGR (around 18%)) and load restriction to about 720 gross MWe. The use of
the OFA ports was discontinued.
Both units at OBGS have experienced severe boiler vibration under a variety of
"normal" operating conditions, possibly aggravated by the use of low-NOx firing
procedures. The optimum operating modes were determined on the basis of compliance
with NOx emission limits and acceptable vibration control, and consisted of maximum
FGR at full load (throttled back at reduced load) and with 8 out of 32 burners out
of service (3rd elevation-gas fuel, 2nd elevation-oil fuel).
Several substantial efforts were made to alleviate the incidence of boiler
vibrations, including installation of burner air register shrouds and readjustment
of boiler back-pass dampers. These efforts were partially successful in reducing
vibration.
As with the ACS units, operation at OBGS increasingly emphasized reduced load
operation at times of off-peak-demand. SCE determined that the flame conditions at
lower loads (ca 250 MWe) were not as secure as they desired. In addition, the OBGS-
2 steam system was modified in 1985 to permit continuous generation as low as 50
MWe. This increased the concern with flame stability (lift-off, etc.) at the
extremely low firing rates.
LOW NOx BURNER RETROFIT
In 1986, the Steam Generation Division at SCE, in conjunction with the System
Planning and Research Department, contracted with Todd Combustion (formerly a
Division of Fuel Tech, Inc.) to provide 32 gas/oil burners to replace the existing
burners at AGS-6, principally to improve low-firing-rate flame conditions but also
to provide reduced NOx emissions. Shortly thereafter, the Steam Generation Division
solicited competitive bids to provide 32 gas/oil burners for installation on OBGS-2.
The contract was also awarded to Todd Combustion. Again, the emphasis was on stable
combustion at all firing rates, with low-NOx and increased efficiency as additional
objectives.
Prior to installation of the Todd burners at AGS-6, SCE obtained a Permit to
Construct from the South Coast Air Quality Management District (SCAQMD), which
stipulated that the NOx emissions post-retrofit must not exceed 113 ppm on gas fuel
and 203 ppm on oil fuel. An additional requirement was that NOx emissions over the
load range must be at least 10% below comparable emissions pre-retrofit, and that CO
emissions could not increase.
The Todd Dynaswirl® burner relies upon control of the combustion air in
several component streams, as well as the controlled injection of fuel into the air
streams at selected points, for maintaining stable, attached flames with low NOx
generation. Figure 3 schematically illustrates the internal configuration of the
burner.
For gas firing, fuel is introduced through six pipes, or pokers, fed from an
external manifold. The pokers have skewed, flat tips, perforated with numerous
holes and directed inward toward the burner centerline. Gas is also injected
through a central gas pipe with multiple orifices at the furnace end. A single oil
gun is located along the burner centerline, inside the gas pipe.
Primary and secondary air streams flow from the surrounding windbox plenum
through a spun cone inlet to the burner. A shut off damper is provided at the
burner inlet. The primary air stream flows into the burner and down the center of
the venturi around the center fired gas gun where it mixes with the center gas
8-26
-------
forming a stable flame in front of the swirlers. The secondary air flows into the
burner flows near the outer walls of the venturi where it mixes with fuel from the
gas pokers and is ignited by the stable center flame. The testing air stream is
controlled by a separate slide damper and flows between the venturi evase and the
burner throat quarl. A piezometer ring is provided at the venturi vena contracta
for comparison to pressure at the burner inlet; the pressure signal of about 2.5
times the windbox to furnace pressure loss provides an accurate measurement of
combustion air flow rate.
The oil gun is a conventional constant-differential, pressure-atomized burner.
The original single orifice swirl tip was replaced with a multi-orifice proprietary
design to reduce boiler vibration, however the turndown ratio is still of some
concern, and efforts continue to improve the turndown while maintaining good flame
quality and low NOx emissions. A swirl impeller is attached to the oil gun support
pipe just at the end of the primary sleeve section.
In performance of the retrofit contract, Todd Combustion performed flow model
analyses of the windbox air flow distribution. Based upon those analyses, baffles
and turning vanes were installed at selected points in the windbox to improve the
uniformity of air flow to all burners.
Following selection of the Todd Dynaswirl burner for retrofit to OBGS-2, SCE
obtained a "Permit to Construct" from the Ventura County APCD. The permit
conditions specified that the new burners would produce no increase in the emissions
of NOx, CO, total particulate and Volatile Organic Compounds (VOC), over the
operating load range, as compared to pre-retrofit emissions. Windbox modifications
to improve air flow uniformity were also made on this unit.
TEST METHODOLOGY
Comprehensive measurements of gaseous emission species (NOx, CO, 02) were made
for the pre- and post-retrofit testing phases of both boiler retrofits. The scope
and conduct of both boiler test programs were essentially identical.
Gaseous emissions were measured by an extractive sampling/conditioning/
measurement system contained within a mobile van. Gaseous analyses included
chemiluminescent (NOx), non-dispersive infrared (CO, C02) and fuel cell (oxygen)
types. All measurements were made after drying the sample gases.
The sample flue gas was extracted through stainless steel probes located in a
matrix across the economizer exit ducts. Measurements could be made of any single
probe sample or a composite of any combination of probes. Composite samples ensured
an equal portion from each probe by passing each individual sample through a
valve/bubbler prior to mixing within a common manifold.
At AGS-6, a similar matrix of probes was located in the air supply ducts
between the air foils (FGR injection) and the windbox. At OBGS-2 the FGR/Air
mixture was measured by sampling from pressure-tap tubing located adjacent to each
burner air register.
The FGR rate was calculated as the volumetric percentage of the flue gas
extracted from the exit ducts and injected into the combustion air. The calculation
was made based upon the dilution of gas species caused by the mixing process, i.e.
the comparative concentrations of 02, C02 and NOx within the flue gas alone and the
flue gas/air mixture supplied to the burners.
8-27
-------
Since the OBGS-2 permit required a demonstration that particulate and
hydrocarbon emissions did not increase following the burner retrofit, tests were
conducted to measure TSP (oil fuel only) and VOC (both fuels) pre- and post-
retrofit. TSP was measured using a modification of EPA Method 5, in which the back
end catch was analyzed in addition to the front end catch (filter plus probe
washing). VOC was measured by capturing flue gas in Tedlar bags and analyzing for
C2 to C8 by GC/MS. Triplicate measurements of Total Suspended Particulate (TSP) and
Volatile Organic Carbon (VOC) were made for each of four load levels from 250 to 700
MWe. Analyses were made to determine the carbon content of the TSP filter catch and
the organic hydrocarbon content of the back-end catch.
Each test was conducted with operation as close to steady state as possible,
with the load blocked on manual control. The boiler fuel, air and steam controls
were generally on "automatic" except that excess air trim and FGR settings were
manually controlled. In general, each test lasted from 30 minutes to 2 hours,
depending upon the complexity of gas measurement desired. In addition to the
emissions measurements, considerable data were recorded regarding operating
conditions (e.g., fuel and air flows, pressures and temperatures, control/damper
settings, steam conditions, motor amps, boiler excess 02 and stack opacity).
TEST RESULTS
This section of the paper presents a brief discussion of selected test results
acquired to date. As pointed out previously, although pre-retrofit testing has been
completed, only limited test data have been acquired for the post retrofit, low NOx
burner configuration for the two units. Due to the limited extent of this latter
data and some present uncertainty in calculated WFGR rates (discussed below), it is
premature to draw definitive conclusions as to the demonstrated NOx control
capabilities of the two Todd burner installations and comparison with the pre-
retrofit NOx control configurations. This paper should be viewed therefore as an
interim status report which will be superseded by a future publication documenting
the completed program.
The testing of both units was constrained by the necessity to continue to
comply with the regulatory NOx limits of 125 PPM and 225 PPM respectively on gas and
oil fuels. This constraint prevented testing to determine the NOx reduction
capability of the Todd burner by itself in the absence of the utilization of WFGR at
higher loads, since emissions compliance could not have been maintained. This same
constraint applied to the pre-retrofit testing relative to demonstrating the
individual control capabilities of WFGR and O.S firing on the two units. Some
estimate of these individual influences for both NOx control configurations for
Alamitos #6 have been made using historical data and FGR effectiveness trends as
discussed later in the paper.
ALAMITOS UNIT #6
Figure 4 shows representative test results acquired for the Todd burner
installation on AGS-6 over the load range. The calculated WFGR rate and measured
average exhaust gas 0, concentration associated with each test data point is
indicated. In general, the data reflect the maximum NOx reduction capability of the
installation. The indicated 0, levels at the higher loads ( >260 MW) are the
minimum achievable within the SCE constraint of maintaining exhaust gas CO
concentration below 300 PPM. The lower load minimum 02 levels are constrained by
the necessity to maintain a minimum level of air flow as dictated by safe operating
procedures. The indicated WFGR rate at the highest loads is near the maximum
8-28
-------
capability of the WFGR system for the test conditions. At the lower loads, the
indicated maximum WFGR rate is constrained by flame stability concerns although no
flame degradation in this regard was noted for the indicated levels.
The upper data point shown in Figure 4 at 480 MW applies to the all- burner-
in-service (ABIS) mode of operation which was the intended employment by SCE for the
for Todd burner installation. The level of NOx emissions achieved represents a
reduction of 87% for the combination of burner and 16% WFGR from the uncontrolled
level of approximately 700 PPM (best estimate based on historical data, could
possibly be higher). At 19% WFGR, the maximum capability of the FGR system, NOx
emissions would have been in the range of 75 PPM (extrapolated from Figure 5 data)
representing an 89% reduction from uncontrolled baseline.
The curve in Figure 4 is for O.S. operation with 8 BOOS. Although the O.S.
mode of operation was not intended by SCE at the time for normal employment, SCE
wanted to demonstrate the maximum NOx reduction achievable since it now must comply
with a significantly reduced emission limit. As Figure 4 indicates, the O.S. mode
of operation combined with 19% WFGR resulted in a further full load NOx reduction of
35% (from 75 PPM to 49 PPM) which represents a 93% NOx reduction from the
uncontrolled baseline level. This NOx control mode has been implemented by SCE for
normal operation.
A comparison of pre and post retrofit test results for a range of WFGR rates
is shown in Figures 5-7. The measured average exhaust gas 0? concentration
associated with each data point is indicated. The single data point shown in Figure
5 for the Todd burner operating in an ABIS mode indicates that less NOx reduction
was achievable than for the pre-retrofit O.S. mode.
With respect to the O.S. mode of operation, most of the post retrofit data
acquired thus far have been for higher WFGR rates than for the pre-retrofit data and
the minimal overlap for the two sets of data prevent a direct comparison over a
range of WFGR rates. However, the data do seem to demonstrate consistent trends
indicating that the Todd burner is capable of achieving lower NOx levels in an O.S.
mode than was possible pre-retrofit. This result appears to be due primarily to the
burner's capability to operate at lower 0, levels (discussed later) since both sets
of data show a clear trend of decreasing NOx with decreasing excess 02. This may be
only a partial explanation and the Todd burner may in fact produce lower NOx
emissions than pre-retrofit operation at identical excess 02 and WFGR levels. A
regression analysis will be performed on the expanded future data base to more fully
assess this question.
The WFGR rates were determined according to the procedure previously outlined.
There is a degree of uncertainty associated with the indicated values, however,
since a comparison between the calculated rates determined by the different methods
(02 or NOx dilution) showed random differences in the range of 10-15%. Since FGR
exerts a strong influence on NOx level, this degree of uncertainty could result in
appreciable error in the data as plotted and misleading apparent trends. This
potential deficiency will be more fully assessed in the continuing program and it is
believed that the relative level of uncertainty in calculated WFGR rates can be
reduced.
Figure 8 shows a comparison between pre and post retrofit NOx control
performance capability for the various control configurations. The NOx levels for
uncontrolled baseline and BOOS configurations are estimated based on 20 year old
test data. The indicated NOx levels for the other configurations are either current
measurements or extrapolations from these measurements. The comparison is tentative
since it is based on current limited data but is presented to provide the reader
8-29
-------
with a present estimate of the Todd burner NOx control capability for the Alamitos
unit as well as a comparison with the pre-retrofit control capability. The
comparison indicates that for like configurations, there is little difference in pre
and post retrofit NOx control capability in absolute terms, the maximum being either
91% or 93%. However, in relative terms, the difference of 16 PPM is significant to
SCE's NOx emission reduction objectives.
The demonstrated percent reductions are measured from an uncontrolled NOx
baseline level of 700 PPM. Experience with implementing O.S. firing has shown that
the percent reduction achievable on a particular unit is dependent on the magnitude
of the initial, uncontrolled NOx emission rate and decreases as this rate is
reduced. Therefore, it is likely that lower NOx control capability could generally
be expected for Todd burner installations on boilers exhibiting lower uncontrolled
NOx emission rates.
Figure 9 compares pre and post retrofit C0/02 trends. As shown, the Todd
burner demonstrated significantly improved performance over that achievable for the
pre-retrofit NOx control configuration. This gain in minimum achievable excess 02
level is partly responsible for the lower NOx emission rate obtainable with the Todd
burner retrofit and also offers a benefit in terms of boiler thermal efficiency.
The improved C0/02 performance of the Todd burner installation can be
attributed in part to improved air/fuel flow uniformity to the burner arrays on the
two firing walls. This was achieved by a combination of windbox modifications made
in conjunction with the burner installation and balancing of the burner fuel and air
flows during shakedown testing. Therefore, part of the NOx and heat rate gain can
be credited against the windbox modifications independently of the burner
installation and the remaining part to the burner itself. The relative contribution
of these two factors has not yet been assessed but answering this question is useful
in terms of comparing the NOx control capability of O.S. firing (whose
implementation could be accomplished in conjunction with windbox modification) with
the installation of a Todd LNB.
Figure 10 is a plot of recorded CEM data (note scale is in LB/HR) acquired
post retrofit during the month of August, 1990 for unit operation over the normal
load range in both AGC and operator control modes. The significant data scatter can
be attributed to the normal variability of key parameter settings such as excess 02
and FGR rate and instrumentation variability. A similar plot has been prepared for
the pre-retrofit NOx configuration for the same period in 1987. Figure 11 shows the
best curve fits for each of the mentioned data sets and also a replot of the lowest
obtainable post retrofit NOx emission demonstrated as shown previously in Figure 4
(all in LB/HR).
The plots illustrate that single point data acquired in controlled testing of
the maximum NOx control capability configuration can significantly underestimate
achievable operational emissions as monitored by a CEM for demonstration of
regulatory compliance purposes. A comparison of the upper two curves also confirms
that the Todd burner installation was successful in reducing NOx emissions during
normal AGC operation.
Figure 12 shows pre-retrofit NOx emissions at selected loads on oil firing for
the ABIS and BOOS modes of operation. Post retrofit oil firing data have not yet
been acquired and the data are shown for general interest.
In terms of operational performance, the Todd burner installation has
satisfied all of SCE's original objectives with the exception of turndown on oil
firing which has not yet been demonstrated. Flames are stable over the load range
8-30
-------
including minimum load and do not exhibit any tendency to lift off under normal
operating conditions. In addition, operating excess 02 level has been significantly
reduced for gas firing thereby yielding a meaningful improvement in boiler thermal
efficiency.
ORMOND BEACH UNIT #2
Figure 13 shows representative pre and post retrofit test results over the
load range for OBGS-2 firing gas fuel. The data points apply to minimum excess 0,
levels and approximately to the same near maximum WFGR rate at each load level. The
data indicate that the Todd burner installation reduced NOx emissions to below
obtainable pre-retrofit levels for the ABIS mode of operation and a further
increment in NOx reduction was achievable for O.S. operation (third row BOOS).
Uncontrolled full load NOx emissions are believed to have been in the range of
1200-1500 PPM and therefore the controlled full load emissions for any of the
configurations (LNB or original burner with O.S. and with WFGR) represent a
reduction of at least 92%. This magnitude of percent NOx reduction is nearly
identical to that achieved on AGS-6. Unlike that unit however, post retrofit ABIS
NOx emissions at OBGS-2 are lower than the best obtainable pre-retrofit NOx
emissions by approximately 10% at full load. The test results in the O.S. mode
shows an incremental reduction of 20% from the pre-retrofit level at full load as
indicated in Figure 13.
The general range of pre and post retrofit CO concentrations measured verses
excess 02 is shown in Figure 14 for gas fuel at loads of 550 MW and above. The C0/02
trends are approximately the same for the pre-retrofit O.S. and post retrofit ABIS
modes of operation while post retrofit operation in an O.S. mode exhibited higher CO
concentrations at comparable 0, levels. These results are at variance with those
demonstrated for AGS-6 which showed an improvement in the C0/02 post retrofit trend
for the O.S. operating mode in comparison to pre-retrofit results. CO
concentrations for this latter unit operating in an ABIS mode have not yet been
measured. The results are surprising since the windbox modifications made to
improve air flow uniformity were expected to result in an improvement in the C0/02
trend as compared to pre-retrofit conditions.
A comparison of pre- and post-retrofit NOx emissions for oil firing is shown
in Figure 15. The data indicate that the Todd burner achieved lower NOx emissions
at full load operating in an O.S. configuration than was obtainable for pre-
retrofit. Since the data are limited and there is some uncertainty in the indicated
WFGR rates, further analysis is required to confirm this result.
For gas fuel there was no increase in measured VOC emissions for operating
conditions consistent with lowest-NOx emissions, (O.S. operation, low excess 02 and
high FGR rate). Similarly for oil fuel there was no measured increase in either
solid carbon or condensible hydrocarbons, again under lowest-NOx operating
conditions.
The post-retrofit condition of the flames was substantially better than pre-
retrofit under all operating conditions, even at 50 MWe with all air registers open,
high FGR rates (up to 40%) and high excess air (25% of rated flow). Under all
conditions the flames were closely attached to the burner tip/throat area and were
steady and symmetrical. Prior to retrofit the flames were frequently detached from
the burner throat by as much as three to four feet, pulsated irregularly and were
occasionally irregular in shape.
8-31
-------
Prior to the burner retrofit, severe boiler vibration (rumbles and furnace
wall pulsations) were experienced under certain "normal" conditions of load, excess
air, FGR rate and burner firing pattern. Although the severe vibration could
usually be avoided, or corrected by an experienced operator, the condition was of
concern to the operating and engineering staff. Following the burner retrofit, the
unit generally operates more smoothly and the most severe vibrations no longer
occur. It should be noted that simultaneously with the burner retrofit, the FD fans
were modified from constant-speed with inlet vane flow control to variable speed
with no inlet vanes. Although it is uncertain whether the fan modifications
contributed to the reduced vibration, the change has definitely reduced the
operating noise level and has significantly improved the control and steadiness of
the air flow.
CONCLUSIONS
It is premature in view of the limited post-retrofit test data acquired thus
far to draw definitive conclusions relative to the pre and post retrofit NOx
emission control performance comparison. It is possible, however, to make some
observations on the basis of the data that have been acquired which are expected to
be valid at the conclusion of the program.
1) Full load gas fired NOx emissions for both units with the Todd burner
installation combined with approximately 20% WFGR have been reduced by 93%
from the uncontrolled baseline NOx level. This reduction was achieved by
operating in an O.S. mode with 25% BOOS.
2) The pre-retrofit NOx control configuration of O.S.operation (25% BOOS)
combined with 20% WFGR demonstrated nearly the same NOx reduction as post-
retrofit from the uncontrolled baseline level for full load gas fired
operation. The difference in demonstrated relative NOx control capability
amounting to a further reduction of about 20% from the pre- retrofit level
could be meaningful for utilities facing very stringent NOx emission
control regulations such as SCE.
3) Achievable NOx emissions employing either control configuration during
normal AGC operation will be significantly higher than that demonstrated
in the controlled testing conducted in this program.
4) The C0/0? performance demonstrated by the Todd burner installations owed
conflicting trends in comparison to the pre-retrofit test results. VOC
emissions on gas fuel and particulates on oil fuel did not increase with
the installation.
5) The burner retrofit demonstrated significantly improved operational
performance relative to pre-retrofit in terms of flame holding, stability
and boiler vibration.
8-32
-------
00
I
CO
CO
FUEL OIL ATOUIZER
AIR SLIDE ACTUATOR
Figure 1: Todd Dynaswirr Low NOx Burner
Todd
Combustion
GENERAL ARRANGEMENT
DYNASWRL - LN BURNER
C/W CENTEfl FIRED CAS GUN
AND POKERS
-------
CXI
CO
Front
windbox
4
3
Burner
level 2
1
Air from FD fans (2)
Damper
FGR fan (2)
Figure 2: Alamitos - Unit 6 Air/FGR Configuration
-------
00
CO
01
Front
Windbox
FGR
Injection
Array (2)
Flow
Measurement
Venturi (2)
Figure 3: Ormond Beach Unit 2 Air/FGR Configuration
-------
o
CO
Q.
QL
8
100
80 -
60
40
20
ALAMITOS #6
GAS FUEL
• ABIS
A 3rd Row BOOS
Uncontrolled NOx Emission 700 ppm
44
100
16%WFGR
0.8 % Excess O.
500
600
200 300 400
GROSS GENERATION (MW)
Figure 4: Minimum Achievable Post Retrofit NOx Emission Over the Load Range
O
CO
Q.
Q.
x"
O
z
I^U
110
100
90
80
60
50
40
2.0
A A 2.2
—
/
A2.2
—
-
-
ALAMITOS
GAS FUEL,
" Pre Retrofit
^2.3
A 2.8
/ .\ / \ / \ ^k ' 2
1.8 A2.2 *
A1.9
A2.0 Kl'7
Ai.e
#6 A
480 MW 1-3 A145
- BOOS
- • Post Retrofit - ABIS Aa9
- ^ Post Retrofit - BOOS M A
i
1.0
i i i i
8
10
16
18
12 14
WFGR, %
Figure 5: Comparison of Pre and Post Retrofit NOx Emission at 480 MW
20
8-36
-------
I£U
^ 100
0"
^
CO
(8) 80
TJ,
C
Q. 60
o
""^
Z 40
20
ALAMITOS
GAS FUEL
A Pre Retrc
A2.6 A2.7 ApostRett
!— 1 i— i /\fc. 1
A2.5 A2.6
A1.8 A1.8
A2.3
~
A2.6
2.1AA1.9
A1p8
^i 4
A 2.5 % Excess 0 2
A. 95
-
A1.4 1' A
^K ^& M
A1.4
I I I I I
0 5 10 15 20 25
5 #6
., 360 MW
>fit - BOOS
ofit - BOOS
1.3
AL6
J1.0 A1-3
I '85
30 3
WFGR, %
Figure 6: Comparison of Pre and Post Retrofit NOx Emission at 360 MW
au
80
^
Cf 70
^
*? 60
©
S 50
£
a 40
x"
Q 30
Z
20
in
ALAMITOi
A GAS FUEl
2'8 A Pre Retrc
•^2.3 A Post Retr
A.8
A2.3% Excess 02
A2.5
1.7 A A>.2
^1.7 1.8 ^21 A3.0
' 2'3AA?27A,.o A2.1
IQ A . Z_i2.6
* A 'a ^ A c
0.7 A 1.3 A A ^.6
0.9 *'
I I I
5 #6
_, 260 MW
>fit - BOOS
ofit - BOOS
*f
i
10
40
20 30
WFGR, %
Figure 7: Comparison of Pre and Post Retrofit NOx Emission at 260 MW
50
8-37
-------
1,000
800
E 600
0.
0.
x"
O
Z 400
200
0
- 700
ppm
-
-
-
<*•
:';X
v^'
"t
Pre Retrofit * * Post Retrofit
79% Reduction from ^^_
150 ^Ml Baseline 174(^^1
Uncontrolled BOOS BOOS ABIS ABIS BOOS
Baseline 19%FGR 19%FGR 19%FGR
Figure 8: Comparison of Alamitos #6 Pre and Post Retrofit NOx Emission at Full
Load on Gas Fuel
700
600
500
Q.
400
O
300
200
100
ALAMITOS #6
GAS FUEL
250-480 MW
Post Retrofit
(O.S.)
0.5
1.5 2
EXCESS 02, PERCENT
2.5
Figure 9: Comparison of Pre and Post Retrofit CO Emission
8-38
-------
500
450
400
350
300
. 250
X
O
Z 200
150
100
50
0
ALAMITOS #6
GAS FUEL
50 100 150 200 250 300 350 400 450 500
LOAD, MW
Figure 10: CEM Data for the Month of August, 1990
550
500
450
400
350
300
250
200
150
100
50
ALAMITOS #6
GAS FUEL
Beat Fit of August, 1987.
CEM Data, (Pre-Retroflt)
Post Retrofit
Minimum Achievable
Best Fit of August, 1990.
CEM Data, (Post-Retrofit)
50
100
400 450 500 550
150 200 250 300 350
LOAD, MW
Figure 11: Comparison of Pre and Post Retrofit "Best Fit" Curve of CEM Data
and Post Retrofit Minimum Achievable NOx Emission
8-39
-------
170
160
150
CO
©
140
130
Q.
Q.
120
110
100
ALAMITOS #6
OIL FUEL
o ABIS
A BOOS
O12.6%FGR
3.5 %Q,
Ol7.9
3.5
2.8 % O2
Q24.2
3.5
20.7
3.2
\18.4
3.6
250
300 350 400 450
LOAD, MW
Figure 12: Pre Retrofit Oil Fuel NOx Emission
500
ORMOND BEACH #2
GAS FUEL
A Pre Retrofit, O/S Firing
Post Retrofit, ABIS
A Post Retrofit, O/S Firing
Uncontrolled NOx Emission
1200-1500 ppm
200
800
400 500 600
GROSS GENERATION, MWe
Figure 13: Comparison of Minimum Achievable Pre and Post Retrofit NOx Emission
over the Load Range for Gas Fuel
8-40
-------
1,200
1,000
.^ 800
I. 600
Q.
O
° 400
200
Pre Retrofit (O.S.)
Post Retrofit (ABIS)
ORMOND BEACH #2
GAS FUEL
550 MW - 750 MW
0.5 1 1.5
BOILER EXCESS 02, % (dry)
Figure 14: Comparison of Pre and Post Retrofit CO Emission
200
180
160
j
140
ORMOND BEACH #2
OIL FUEL
O Pre-Retrofit (ABIS)
A Pre-Retrofit (O/S Firing)
• Post-Retrofit (ABIS)
± Post-Retrofit (O/S Firing)
(§) 120
Q 100
I. 80
Q.
X" 60
O
Z 40
20
0
38, 2.05
O
A
31,3.78
33, 2.55
26, .77'
27, 2.0 '
24, .73
25,2.950
7, 2.53
438,2.13
41,1.0^
. 20, 2.22
200 400 600
GROSS GENERATION, MWe
Figure 15: Comparison of Minimum Achievable Pre and Post Retrofit NOx Emission
Over the Load Range for Oil Fuel
800
8-41
-------
COMPARATIVE ASSESSMENT OF NOx REDUCTION TECHNIQUES
FOR GAS- AND OIL-FIRED UTILITY BOILERS
Gary L. Bisonett
Steam Generation Department
Pacific Gas and Electric Company
San Francisco, California 94106
Mike McElroy
Electric Power Technologies, Inc.
Berkeley, California 94705
-------
COMPARATIVE ASSESSMENT OF NOx REDUCTION TECHNIQUES
FOR GAS- AND OIL-FIRED UTILITY BOILERS
Gary L. Bisonett
Steam Generation Department
Pacific Gas and Electric Company
San Francisco, California 94106
Mike McElroy
Electric Power Technologies, Inc.
Berkeley, California 94705
ABSTRACT
Pacific Gas and Electric Company conducted a comparative assessment of commercially available
NOx control technologies that might be applicable to our gas- and oil-fired boilers. One phase
of the assessment, cofunded by EPRI, was a comparative cost and feasibility analysis of various
commercially available technologies, including combustion modifications, low NOx burners, and
selective catalytic reduction. The results of this study are being incorporated into efforts
to identify a cost-effective system wide NOx control strategy for our system. The comparative
assessment was conducted based on a typical boiler in our system to allow technology comparisons
to be made on a consistent basis. Once the information for each technology was developed, the
site specific factors that affected each technology were identified so that the results could be
generalized and modified for other boilers in our system. One aspect of the project was to
develop a computer program, also cofunded by EPRI, to help PG&E compare various NOx control
strategies for possible application in our system. The computer program provides a first-cut
analysis of NOx reduction costs given different projected NOx limits and compliance strategies.
8-45
-------
COMPARATIVE ASSESSMENT OF NOx REDUCTION TECHNIQUES
FOR GAS- AND OIL-FIRED UTILITY BOILERS
INTRODUCTION
Pacific Gas and Electric Company (PG&E) performed a multi-faceted engineering program to
identify and evaluate options for reducing NOx emissions from its gas- and oil-fired electric
generating units. The program, involving the 39 boilers in the PG&E system, had two primary
goals: (1) Evaluate and compare the technical and economic merits of commercially available
retrofit NOx control technologies and their applicability to PG&E's boilers; and (2) Develop a
computer model to assist PG&E in developing an optimum system-wide NOx control strategy.
The program was prompted by concerns for lower NOx emission requirements for California
utility boilers.
The program was performed with cofunding and technical participation from the Electric Power
Research Institute (EPRI). The involvement of EPPJ was in recognition that the PG&E program
would be a valuable case study for the utility industry, and the results could assist other
utility companies planning or engaged in similar NOx control assessments.
PG&E is one of the largest investor owned gas and electric utilities in the United States.
PG&E's fossil fuel fired electric generating capacity is centered in seven stations located
throughout the Company's service territory which encompasses much of northern and central
California. PG&E's gas- and/or oil-fired boilers total over 7,600 megawatts of electrical
capacity, and represent a wide cross-section of manufacturers, furnace designs, combustion
systems, equipment sizes, and vintages. PG&E's 345 MW opposed-fired boilers (manufactured
by Babcock and Wilcox) comprise one-third of the capacity, and were the focus of the program.
NOx control measures have been previously implemented on these and several other PG&E boilers,
including overfire air, flue gas recirculation, low excess air operation, and biased firing.
The California Clean Air Act which was passed in 1988 requires local air pollution control
districts to develop plans to attain ambient air quality standards in California. The
California ozone ambient air quality standard is 25 percent more stringent than the Federal
ozone standard. This requires a very aggressive program on the part of regulators to develop
plans to attain the California ozone standard. PG&E's goal is to work closely with regulators
to identify emission reduction plans that are both cost effective and responsive to the air
quality needs of the communities we serve.
Since the completion of this study, PG&E has continued to develop site-specific information to
identify cost effective strategies for reducing NOx emissions. This program is ongoing and will
continue as information from other installations, R&D, and the regulatory process becomes
available.
8-46
-------
PG&E NOx CONTROL ASSESSMENT PROGRAM
Program Scope
The PG&E program consisted of the following work elements:
1. Verify existing boiler NOx emissions as a function of load for each boiler,
using existing field test data, supplemented as necessary with NOx
emission predictions based on furnace heat release rate correlations.
2. Compile detailed listings of boiler-specific operating and physical data
that are related to NOx formation.
3. Evaluate the applicability and NOx reduction potential of operational
modifications (e.g., bumers-out-of-service and biased firing) for the
entire PG&E boiler population. This work was based upon previous
experience with such controls within PG&E and elsewhere in the utility
industry.
4. Assess the technical feasibility of retrofitting state-of-the-art low-NOx
combustion systems for three selected boilers, and develop NOx reduction
and cost factors for the technically feasible options.
5. Perform limited field tests on one unit (Contra Costa Unit 6) to validate
predictions of NOx reduction achievable by combustion modifications.
6. Conduct comprehensive technical and economic assessments for low-NOx
combustion and Selective Catalytic Reduction (SCR) for a selected boiler
(Contra Costa Unit 6).
7. Rank each potential NOx control option evaluated during the study by cost,
NOx reduction potential, and technical risk. Also, identify the site
specific factors that influenced the rankings.
8. Construct a NOx emission forecast model which utilizes the above results to
identify the NOx controls required to meet specified system-wide or
regional emission limits at minimum cost.
9. Develop hypothetical NOx compliance strategies for different levels of
system-wide NOx reduction utilizing the NOx emission forecast model.
Contra Costa Unit 6 was selected for the retrofit feasibility study (Item 4 above), and for
detailed engineering and cost evaluations (Item 6), because it is representative of a boiler
design that constitutes one-third of the PG&E fossil system capacity. Less detailed feasibility
studies where also performed on two other PG&E boiler designs which posed distinctly different
retrofit situations (Moss Landing Units 6 & 7, and Pittsburg Units 5 & 6). Each of the three
selected boilers were already operating in a reduced-NOx mode (with flue gas recirculation to
the windbox and combustion staging) which was the baseline condition for the feasibility and
engineering studies.
8-47
-------
For the three plant sites, operation with natural gas and residual oil was considered. Fuel oil
with a 0.5 percent sulfur content, based on the maximum allowed by regulatory requirements, was
assumed. Since fuel nitrogen content is not constant and variations would affect the NOx
reduction attainable by a given combustion NOx control, values of 0.3 percent and 0.6 percent
(by weight) nitrogen in the oil were considered for purposes of NOx predictions.
Description of Study Boilers
Contra Costa Unit 6 - The unit is a forced draft, opposed-fired, drum-type boiler manufactured
by Babcock &. Wilcox with a rated generating capacity of 345 MW (gross). The unit was built in
1964. The unit fires oil and natural gas through 24 circular register burners arranged in two
rows of six burners on each firing wall. The furnace contains two division walls separated from
the furnace end walls and each other by two columns of burners. An elevation drawing of the
boiler is provided in Figure 1. In 1973-1974, overfire air ports were installed to reduce NOx
emissions in order to meet new NOx emission limits. Overfire air ports were installed in the
windbox, one above each burner column, for a total of twelve ports. In addition, the existing
hopper gas recirculation system was upgraded to mix up to 18 percent flue gas into the secondary
air duct feeding the windbox.
Moss Landing Units 6 and 7 These two identical units, rated at 750 MW (gross), began operation
in 1967-68. These units, manufactured by Babcock & Wilcox, are forced-draft, supercritical
boilers. The units are opposed wall fired and were originally equipped to fire oil or natural
gas with 3-nozzle cell burners arranged in a two-high by four-wide array on each firing wall (a
total of 24 burner throats on each wall). In the early 1970's, the existing hopper gas
recirculation system was modified to permit operation with up to 18 percent flue gas
recirculation with provisions to direct recirculated flue gas to the windbox for NOx control.
Also, the top nozzles of the upper four cell burners on each wall were modified to pass air
only, acting as localized overfire air ports to provide an additional NOx reduction.
Pittsbure Units 5 and 6 - The two identical units, designed by Babcock and Wilcox, began
operation in 1960-61. The units are forced draft, natural circulation boilers, with a rated
generating capacity of 330 MW (gross when fired with either natural gas or oil fuel. The units
were designed for future coal firing with a conversion to balanced draft. The boilers are
opposed fired with 24 burners arranged in two-high by six-wide array on each wall. In the early
1970's, the units were modified to reduce NOx emissions by adding flue gas recirculation to the
windbox and installation of overfire air ports above the top burner row.
Program Participants
A majority of the work was performed by outside contractors selected on a competitive basis.
The major participants and their areas of prime responsibility are as follows:
• EPRI - Cofunding and participation in project technical direction.
• Babcock & Wilcox Company - Retrofit evaluation of low-NOx combustion
equipment options and Selective Catalytic Reduction.
• Fossil Energy Research Corporation - Development of NOx Emission Forecast
Model
• KVB. Inc. - Compilation of current (baseline) boiler NOx emission factors,
and evaluation of NOx reduction via operational modification.
8-48
-------
• Electric Power Technologies. Inc. - Provide technical and administrative
support to PG&E, including assistance in program planning, selection of
subcontractors, and analysis of results.
• PG&E - Overall project management and NOx reduction field tests at Contra
Costa Unit 6.
NOx CONTROL TECHNOLOGIES EVALUATED
The NOx control technologies that were considered in the NOx control evaluation include:
1. Operational Modifications to Existing Equipment
2. Combustion Equipment Modifications
• Two Stage Combustion (TSC)
• Reburning
3. Postcombustion NOx Control
• Selective Catalytic Reduction (SCR)
Operational Modifications. The operational modifications evaluated were: (1) low excess air;
(2) bumers-out-of-service (BOOS), including selected gas spuds out of service for natural gas
firing, (3) fuel biasing, (4) optimization of existing overfire air ports (where installed); and
(5) optimization of existing windbox flue gas recirculation (where installed). Other
modifications considered, but not found to be cost-effective, were reduced combustion air
preheat and water injection.
Combustion Equipment Modifications. The combustion equipment modifications were commercial
combustion systems, offered by B&W. Each involved retrofit of low-NOx PG-DRB burners,
installation of dual register overfire air ports, and installation of a compartmentalized
windbox. Conceptually, the systems differed primarily in the arrangement and number of burners
on the firing walls, location of overfire air ports, requirements for additional furnace height,
and the control and distribution of air and fuel among the overfire air ports and burner
elevations. Each system was evaluated for a range of flue gas recirculation rates, both within
the existing FOR capacity and under conditions of increased FOR capacity. The scope of
modifications and retrofit equipment associated with each combustion technology is summarized in
Table 1.
Four versions of rebuming were evaluated:
(a) In-Fumace NOx Reduction (IFNR)
(b) Pseudo-In-Fumace NOx Reduction (Pseudo-IFNR)
(c) Derate In-Furnace NOx Reduction (Derate-IFNR)
(d) Dual-Mode In-Furnace NOx Reduction (DM-IFNR)
8-49
-------
Versions (b), (c) and (d) were essentially compromise designs which attempt to minimize boiler
modifications [e.g., minimize or eliminate need for additional furnace height] compared to a
non-compromise, full rebuming system [version (a)]. Pseudo-IFNR utilized minimum furnace
residence time criteria for rebuming reactions, Derate-IFNR involved a load reduction on the
unit to satisfy rebuming residence time requirements, and DM-IFNR involved operation in an IFNR
mode below a certain load and TSC operation at higher loads.
A limited evaluation of B&W's XCL burners was also performed, as this technology became
commercial during the course of the study.
Selective Catalytic Reduction. The postcombustion SCR technology was a commercial system
offered by B&W through a licensing agreement with Babcock-Hitachi in Japan. The scope of
modifications and retrofit equipment are summarized in Table 1.
RESULTS
Operational Modifications
Maximum NOx reductions achievable from implementation of operational modifications to existing
combustion equipment were predicted to range from approximately 10 percent to as high as 60
percent from boiler to boiler (at full load).
The range reflects the varying degrees of NOx control already in place, and the site-specific
factors that influence the applicability and performance of these controls. The NOx reductions
typically associated with each control technique are as follows:
Operational Modification NOx Reduction
Low Excess Air 5-10 percent
Bumers-Out-Of-Service 15-60 percent
Fuel Biasing 20-50 percent
Overfire Air Optimization 10-15 percent
FOR Optimization 5-20 percent
In general, due to the low cost of implementing operational changes, these options should be
considered as a first NOx control alternative.
Combustion Equipment Modifications
State-of-the-art low-NOx combustion controls, aimed at achieving minimum NOx emissions via
modifications to combustion equipment — specifically, TSC and rebuming — were not universally
applicable to all boilers in the PG&E system. Moreover, the predicted NOx reductions with these
technologies, where technically feasible, varies considerably from unit to unit. Predicted NOx
reductions range from 20 percent to as high as 70 percent from existing levels, reflecting the
impact of site-specific factors, associated compromises in NOx control system design, and
specific NOx control design and operating conditions. These NOx reductions were calculated from
existing "baseline" boiler operating conditions in which the current use of flue gas
recirculation and various degrees of conventional combustion staging already result in reduced
NOx emissions. Larger percentage NOx reductions would be expected if the study boilers had not
been previously equipped with these NOx control measures.
8-50
-------
The boiler-specific results concerning technical feasibility are summarized in the following
paragraphs. The predicted NOx emissions are summarized in a separate subsection below.
Contra Costa Unit 6: TSC could be applied, with burner rearrangement and significant ductwork
and windbox modifications. The relatively tight furnace, originally designed with minimum
residence times, would not accommodate any version of reburning without major extensions in
furnace height. The change in furnace height required for implementation of IFNR is illustrated
schematically in Figure 2.
Moss Landing Units 6 & 7: Application of low-NOx combustion systems is difficult due to
the 3-nozzle cell burner design, and the physical interferences from steam headers and mixing
equipment located halfway up the furnace walls in the windbox. A TSC system could be installed
but would require major modifications to the firing walls, including complete rearrangements of
the burner array and windbox to accommodate new burners and overfire air ports. Pseudo-IFNR is
the only rebuming option determined to be feasible, but would require a substantial increase in
furnace height as well as firing wall modifications similar to TSC. For both control options,
use of XCL burners instead of PG-DRB burners could reduce the retrofit complexity and cost.
Pittsburp Units 5 & 6: The relatively high residence time in the furnace (originally designed
for future coal conversion) greatly enhances retrofit feasibility. TSC can be retrofitted with
only minor modifications to the overfire air ports (the PG-DRB burner would fit into existing
burner openings). IFNR can also be applied without major furnace modifications~an additional
row of burners and new overfire air ports would be required.
Selective Catalytic Reduction
It is feasible to retrofit Selective Catalytic Reduction (SCR) to the Contra Costa Unit 6 to
achieve postcombustion NOx removals of approximately 80 percent. The design conditions and
operating parameters were concluded to be similar to SCR units operating in Japan.
Two possible SCR arrangement were evaluated for Contra Costa Unit 6: (1) Base Case -single SCR
reactor located in the existing air heater location, requiring relocation of air heaters and FD
fans towards the stack; and (2) Alternate Case - two SCR reactors located above the existing
air heater locations, with air heaters and fans undisturbed. Schematics of both configurations
are shown in Figures 3 and 4.
8-51
-------
NOx Reduction Summary for Control Options
The predicted NOx reductions for the combustion modification options are summarized in Table 2
for the three boilers evaluated.
Figure 5 compares the NOx reductions predicted for combustion modifications and SCR applied to
Contra Costa Unit 6.
Plant Impacts
For SCR, and the advanced combustion systems that were technically feasible, there appear to be
no adverse impacts on power plant performance, operation, or reliability that would preclude
their implementation. However, potential impacts were identified and incorporated into the
overall evaluation of control options. The potential impacts considered include:
Combustion Modifications
Increased auxiliary power for higher FOR rates, where required.
Potential increase in furnace tube wastage due to reducing
conditions.
- Boiler control system complexity.
Changes in furnace excess air and resulting effects on plant heat
rate.
- Boiler startup and shutdown procedures.
- Potential for flame impingement.
- Burner turndown.
- Restrictions on rate of load change.
Potential localized connective pass tube overheating.
Selective Catalytic Reduction
- Potential air heater plugging when burning oil fuel.
- Increased minimum load or economizer bypass to maintain minimum SCR
temperature.
- FD fan upgrading to overcome increased system pressure drop.
Boiler startup and shutdown procedures.
- Increased maintenance for SCR catalyst replacement and air heater
cleaning.
- Air heater wash water treatment.
- Ammonia emissions.
8-52
-------
Cost of NOx Control
The cost of retrofitting combustion modifications and SCR (1989 dollars) were evaluated
according the standard EPRI Economic Premises. Capital costs ($/kW) included all materials,
engineering, installation, contingencies, and home office fees for a turn-key retrofit project.
Levelized costs (mills/kWh) included all operating and maintenance labor and materials,
administrative costs, and carrying charges. Levelized costs reported herein are for a base case
30-year levelization period and 30 percent capacity factor (other assumptions were evaluated in
the study to examine cost sensitivity to these parameters).
Low-NOx combustion system costs estimated for Contra Cost Unit 6 ranged from approximately
$40/kW to $50/Kw, with total levelized costs ranging from approximately 3 to 4 mills/kWh. These
cost estimates are higher than generic cost estimates in the open literature.
The capital cost of SCR ranged from approximately $72/kW to $82/kW, and total levelized costs
range from approximately 3 to 8 mills/kWh, depending on specific design and operating
assumptions.
A comparison of the costs of technically feasible NOx control options (TSC and SCR) for Contra
Costa Unit 6 are compared in Table 3. Approximately 30 percent of the Engineering & Material
costs for TSC-are for low-NOx burners, burner accessories, and overfire air ports. For SCR,
approximately 40 percent of the Materials & Engineering cost is for the SCR reactor vessel,
including the casing, framework, and initial catalyst charge.
General Observations. The results of the study reinforce the following considerations
regarding the evaluation of utility boiler retrofit NOx controls:
1. The selection of an optimum NOx control approach for a specific boiler is
rarely obvious, without first performing detailed engineering and cost
analysis of the available technology options.
2. To provide a meaningful comparison of NOx control options, it is imperative
that a systematic approach be used which analyzes each potential control
technology under the same technical and economic premises.
3. Relying on generic technical and cost data is not advisable for evaluating
retrofit feasibility, NOx control cost, and potential NOx reductions for a
specific boiler or a utility generating system. Such an approach could
easily lead to substantial errors relative to a systematic, detailed
engineering and cost analysis of the same boilers.
4. Depending on site-specific constraints and NOx reduction requirements, it
is likely that a combination of NOx reduction techniques will provide the
overall least cost means of achieving those requirements.
Applicability and Value to Industry
The PG&E retrofit analyses involved a single boiler manufacturer's NOx control technology
applied to a few specific boilers. Although the technologies are representative of generic
classes of NOx controls that are offered by other vendors, it is likely that conclusions
regarding technical feasibility and cost would differ if performed by another manufacture
applying its versions of these technologies.
8-53
-------
There are other boiler design types within the U.S. utility industry that are not represented by
the units selected for evaluation in this study. Such boilers, including tangentially-fired
units and cyclone-fired boilers burning gas and oil fuels, can be anticipated to pose
substantially different retrofit constraints. Thus, a comparable feasibility analysis performed
on these units could have different results than those in this study.
Although the technical and cost evaluations may be pertinent to some retrofit situations
encountered elsewhere in the industry, for the reasons enumerated above, feasibility and
engineering/cost analyses specific to each utility company are required. However, the
methodology used in this study is generally applicable across the industry, and can be applied
by other utility companies performing NOx assessments of their generating systems.
The value of this methodology will be further demonstrated as PG&E proceeds towards final
selection and application of NOx controls for their generating system.
PG&E NOx EMISSION FORECAST MODEL
The PG&E NOx Emission Forecast Model determines the NOx emission controls required to meet
specified emission limits and their related cost to PG&E. The costs are calculated both in
terms of capital costs and levelized costs. The model also determines changes in the system
heat rate due to the application of NOx controls. The model will allow PG&E to evaluate various
load and fuel use scenarios with different emission limits imposed. The model calculates annual
NOx emissions using boiler-specific information on operating hours and the loading, combined
with information on boiler specific NOx-versus-load and heat rate-versus-load curves. The model
has the capability to take PG&E's "adjusted load data" (a slightly modified version of the Total
Daily Production, or TDP, files) and produce seasonal, monthly, and annual load profiles and
capacity factors for each boiler. Therefore, although the model calculations are designed
around a system annual operating basis, year-to-year variations in load demand and fuel use may
be accommodated.
A generic version of the model will be made available to EPRI member utilities as part of a
software system now being assembled by EPRI.
CURRENT PG&E ACTIVITIES
PG&E is continuing to develop information on NOx control technologies that might be applicable
to our power plants. We are conducting studies to evaluate NOx control cost and feasibility for
more of the boilers in our system. This information will be used as input to the NOx emission
forecast model to help us develop a cost effective system-wide NOx reduction strategy. Our goal
is to identify a range of NOx reduction strategies that are both cost effective and responsive
to the needs of the communities we serve.
We are also planning to conduct a "proof of concept" test using urea injection on a 345 MW
boiler. Urea will be injected into one-third of the flue gas in the convective pass of the
boiler. The test boiler has two division walls that divide the furnace and flue gas paths into
three flow streams. The results of this test will be used to determine if urea injection has
the potential to provide cost effective NOx reductions on our 345 MW boilers.
8-54
-------
FIGURE 1
ot/nrr
PACIWC 0*5 t HKTtK COMPANY
CONTtA COnA TOWII flANT — UNrtJ NO. « AND 7
AKDOCM, CALIFOINLA
MiTPCf ft WHCO1 ftAAtAMT MHtAT MUtl
CATACTT. u HUM rv MOU* riueeo KJHIHL»'I« cxjrxn nmuti
MAIMUU AllOWAtU WO*UN« HUIML FV . . l.UO H«H nUtUII
UMIAT tnAU TlktftUTUII. I LOCI
8-55
-------
uMACt tut 4-
co
en
05
HOI po«r«-j-
•U«Nf t(_ .
•U*M<*1 - .
NOPPCII
—r-
PO«TS
-•UINCIS
NOI PORTS
EXISTING ARRANGEMENT
NEW TWO-STAGE
COMBUSTION SYSTEM
•URHttf
•UONCMC
•UDHttC
cn
m
oo
o
m
r> m
2—1
5 -n
S^
i
t/>
IN-FUHNACE NOX REDUCTION
-------
00
I
Cxi
EXISTIM5JNEW (EXCEPT
NOTED)
CO
O
JO
m
3
m
53
m »
5S
o
O
CO
EXISTING t LOCATION
OF F.D. FANS
-------
CD
i
cn
CD
PLATFORM
EL. 83' -0-
1
PLATFORM
EL. 63--0-
R_ATFOW
EL. 49*-3*
'
L
SAMPLING CONN.—.
AMHO4IA
DISTRIBUTION ,
\
\
WID \ \
1
,
k
^ 1
\ !
-
"7^
Y ;; i j!|
X
f
rr
i
i
\/..MUJlLUJ
,1
1
1
1 J t=;
-|.
— 1
i
EXISTING
^
s
P
—
~* '
-
-SELECTIVE
CATALYTIC
REACTOR
/-SAMPLING CONN.
j
I
/
I
1
l
EL. «8'-IO
NEW (EXCEPT
— ~ AS
NOTED)
?
A
3
-3/V J
--
t A.M.
T
— t 1
,
'
1
^
1
NEW LOCATION
.
OF EXISTING"
AIR HEATER
!
I
! s
i-
A
EL. 9'-6-
?6'-0-
18' -0' 32' -6"
t PROPOSED
«M COLUMN
LOCATION
EXISTING » LOCATION
OF F.D. FANS
l/>
yo
70
tn
li;
r>
a
-------
FIGURE 5
PREDICTED NOx EMISSIONS FOR
CONTRA COSTA UNIT 6 - FULL LOAD
00
cln
CO
NOx, ppm @3% O2 (dry)
500
400 -
Original
Design
Existing
(FGR+OFA)
TSC
IFNR
SCR
Fuel Oil (0.3 N2)
Natural Gas
-------
Table 1
MAJOR MODIFICATIONS AND EQUIPMENT ITEMS FOR
NOx CONTROL OPTIONS - CONTRA COSTA UNIT 6
Two Stage Combustion
In-Fumace NOx Reduction
SCR (Base Case)
Fans and Ductwork:
- Replace FGR fan rotor.
- New FGR outlet ducts and
dampers.
- OFA ducts and dampen.
- PC ductwork/piping and dampers.
- Replace air heater outlet ducts.
Generally, same items as for TSC.
(Detailed design not performed)
New FD fans, drives, and
foundations.
Increased stiffening on flues and
ducts.
Structural supports, platework,
expansion joints, dampers, turning
vanes, etc. for installation of SCR,
relocated air heater, and new FD
fans.
Boiler Modifications:
- Partial replacement of sec.
superheater (SSH) tubes.
- Replace SSH attemperator to
increase capacity.
- Compartmentalized windbox.
Major extension of furnace height
(furnace bottom extended
downward) requires
modifications/replacement of
furnace wall panels, structural
supports, and water circuitry.
Compartmentalized windbox.
Reposition air heater toward stack
(install SCR reactor in existing air
heater location).
Modify furnace convection pass
buckstay/support systems.
Combustion Equipment:
- 24 PG-DRB burners with
accessories (installed in existing
furnace openings).
- 12 Dual Register OFA ports
(installed in existing furnace
openings).
- Modified fuel supply valving.
Generally similar equipment
items as for TSC except for
additional row of burners (i.e., 36
PG-DRB burners required).
- None
Other
- Boiler control system modifications
(minimal)
Boiler control system
modifications and
instrumentation expected to be
more extensive than for TSC.
SCR reactor vessel, incl. catalyst.
Ammonia storage, vaporization, and
injection systems.
SCR controls and instrumentation.
Modified underground utilities (due
to interferences).
8-60
-------
Table 2
LOW-NOx COMBUSTION FEASIBILITY STUDY RESULTS
Test Case Description:
PG-DRB Burners
Burner Arrangement
Overfire Air Ports
FGR Rate
Predicted NOx Reduction
at Full Load:
Fuel Oil (0.3%N)
Natural Gas
Increased Furnace Height
Other Considerations
Preliminary Feasibility
Contra Costa Unit 6
TSC
24
2Hx6W
Opposed
12
20%
31%
61%
No
Yes
IFNR
36
3Hx6W
Opposed
12
20%
52%
73%
Yes
No
P-IFNR
36
3Hx6W
Opposed
12
20%
45%
70%
Yes
No
D-IFNR
36
3Hx6W
Opposed
12
20%
58%
75%
No
(1)
No
DM-IFNR
36
3Hx6W
Opposed
12
20%
30%
62%
Yes
No
Moss Landing 6 & 7
TSC
36
3Hx6W
Opposed
12
18%
21%
50%
No
(2)
Yes
P-IFNR
36
3Hx6W
Opposed
12
18%
54%
69%
Yes
(2)
No
Pittsburgh 5 & 6
TSC
24
2Hx6W
Opposed
12
18%
40%
58%
No
(3)
Yes
IFNR
36
3Hx6W
Opposed
12
18%
47%
66%
No
(3)
Yes
CO
I
O)
(1) Load restricted to 55-60% of MCR.
(2) Existing 3-nozzle cell burners require extensive changes in burner arrangement and windbox to accommodate PG-DRB
retrofit. Physical interferences from steam piping and mixing devices along furnace wall complicate retrofit.
(3) Coal-design furnace provides sufficient residence time for combustion staging within existing furnace cavity.
-------
Table 3
COSTS OF TSC AND SCR FOR APPLICATION TO
CONTRA COSTA UNIT 6
Two Stage SCR SCR
Combustion (Base Case) (Alternate)
Capital Cost ($/kW)
Material & Engineering 17.5 30.8 33.7
Installation 12.7 15.6 19.3
Other (1)
TOTAL CAPITAL REQUIREMENT 45.7 72.3 82.5
Levelized Cost (mills/kWh)
Fixed and Variable O&M 0.8 1.1 1.2
Consumables (2) 0.0 1.3 1.3
Carrying Charges (Capital) 23 4.5 5.2
TOTAL LEVELIZED COST 3.7 6.9 7.7
Notes:
(1) Includes contingencies, general facilities, taxes, and pre-production costs.
(2) Includes replacement catalyst and ammonia for SCR.
8-62
-------
ANALYSIS OF MINIMUM COST CONTROL APPROACH
TO ACHIEVE VARYING LEVELS OF NOx EMISSION REDUCTION
FROM THE CONSOLIDATED EDISON CO. OF NY POWER GENERATION SYSTEM
D. Mormile
J. Pirkey
Consolidated Edison Co. of New York
New York, NY
N. Bayard de Volo
L. Larsen
B. Piper
M. Hooper
Energy Technology Consultants, Inc.
Irvine, CA
-------
Analysis of Minimum Cost Control Approach
to Achieve Varying Levels of NOx Emission Reduction
from the Consolidated Edison Co. of NY Power Generation System
D. Mormile
J. Pirkey
Consolidated Edison Co. of New york
New York, NY
N. Bayard de Volo
L. Larsen
B. Piper
M. Hooper
Energy Technology Consultants, Inc.
Irvine, CA
ABSTRACT
Con Edison of New York operates a system of gas and oil fired boilers for
power generation and district heating which is located in New York City. Although
current NOx emissions from these boilers are in the range of NSPS limits, a further
reduction could be mandated as a consequence of a future NOx regulatory strategy to
achieve compliance with ambient ozone standards. In recognition of this
possibility, Con Edison initiated a program in 1989 to determine how NOx emissions
might be best controlled and at what cost.
Tests have been conducted on each unit type/fuel combination to determine
current NOx emission levels and the reduction potential achievable by employing
operationally implemented off-stoichiometric firing. A PC based model of the system
has been formulated which can predict system NOx emissions integrated over any
potential compliance period for the application of any unit specific combination of
NOx control technologies. The model considers capital and operating costs on a unit
specific, control concept design basis and calculates system cost levelized over a
specified period for each case considered.
This paper presents a review of the program status and a preliminary summary
of results obtained to date. The program is not yet completed.
8-65
-------
INTRODUCTION
In 1989, The Consolidated Edison Company of New York, Office of Environmental
Affairs, initiated a program to define cost-effective strategies to contend with
possible future NOx emission regulations. The purpose of the program was threefold:
1) To assess the cost and effectiveness of all viable NOx control
technologies as applied to the Con Edison fossil fuel boilers and to
define the optimum means of achieving any specified level of NOx
emissions.
2) To provide information to assess the economic and emissions impacts of
proposed regulation levels and forms so that Con Edison might formulate
a corporate position relative to rulemaking activities of regulatory
agencies.
3) To identify areas to which Con Edison might best direct internal R&D
funding to nurture the development of NOx control technologies to serve
its future needs.
The program, still in progress, comprises four major tasks: 1) testing of
representative boilers to characterize both the baseline NOx emissions throughout
the Con Edison system and the emissions reductions possible with O.S. firing
techniques; 2) compilation and assessment of information on the control
effectiveness and application costs of all pertinent NOx control technologies;
3) formulation of a PC-based computer model of the Con Edison fossil fuel boiler
system to permit assessment of baseline NOx emissions and the cost and NOx emissions
resulting from application of selected control technologies; and 4) analysis of
optimum NOx control strategies to achieve compliance with a variety of potential
emission requirements, using the results from the previous three tasks.
The testing portion of the program consists of the measurement of NOx
emissions from a selected set of boilers representing the total Con Edison
population of boilers. Each boiler was tested with normal firing procedures over
its firing range (load) and for each fuel (natural gas or residual oil) commonly
burned. The baseline NOx emissions were characterized vs excess 0? level at each
load level tested. Measurement of 02, CO and NOx was made at multiple locations in
the boiler exit ducts using a mobile flue gas analysis laboratory. On some boilers
tests were also performed to define the potential NOx reduction achievable by firing
in an off-stoichiometric (O.S.) mode, consisting of shutting off fuel to selected
burners while leaving their air registers open, thus stratifying the air/fuel mix in
the combustion zone. In all, 21 boilers have been tested, out of a total population
of 31 electric generation and 33 steam sendout boilers.
The compilation and assessment of NOx control technology effectiveness and
costs was accomplished with a combination of public and proprietary NOx emissions
test data for a wide range of control technologies. To the extent possible, the
available data were adjusted to reflect the most likely control effectiveness and
cost of implementation which would occur upon application to specific Con Edison
boilers.
A PC-based, spreadsheet model was composed to calculate the NOx emissions,
electric and steam production, and fuel consumption of each Con Edison boiler for
any specified time period, load schedule, fuel mix and NOx control technology
implementation. A discussion of some features of the program is contained below.
8-66
-------
Some preliminary analyses of optimum NOx control strategies have been
completed using the computer model. The initial results are discussed in the paper.
The purpose of this paper is to present these preliminary results, which may be of
some interest to other utility and regulatory investigators. The authors emphasize
that the analysis is incomplete at present. Additional boiler testing is planned,
refinements are being incorporated into the computer model and the assessment of NOx
control technologies continues to be updated.
CURRENT OPERATION
Con Edison operates a system of 64 fossil-fuel-fired steam boilers located
within the city of New York, ranging in size from 150,000 Ib/hr to over 8 million
Ib/hr steam capacity. Eleven large boilers generate only electricity (173 to 972
MWe each) with condensing turbines. An additional twenty boilers produce
electricity and also send out live, extraction or exhaust steam for commercial
heating use. Thirty-three smaller boilers produce steam only for send-out. The 64
boilers are distributed among thirteen separate plants in the boroughs of Staten
Island, Brooklyn, Queens and Manhattan. Table 1 presents a summary description of
the boilers operated by Con Edison and included in the current analysis. Additional
electric generating plants, partially owned by Con Edison but operated by others,
are not included in this study. Similarly, combustion gas turbines are excluded at
present.
As shown in Table 1, some units burn either gas or oil fuel (or a combination
of both) while the remainder burn exclusively natural gas (60th St) or residual oil
(all of the rest). Boilers with dual-fuel capability are generally restricted to
oil fuel in the months of December through February due to curtailment of gas
supplies. When both fuels are available, current fuel prices generally favor gas
firing. In recent years the relative system-wide fuel mix has been from around 50
to 75% oil on an annual basis (BTU value).
The electric generating boilers represent a spectrum of tangential, face and
opposed fired boilers manufactured by CE, B&W and FW. Most of these were originally
designed for coal firing and therefore represent relatively large furnace volumes
(and consequently, low NOx emissions) for the unit firing capacity. This
characteristic is discussed further below.
The total capacity of Con Edison-operated fossil-fuel electric generation is
approximately 6,700 MWe of which about 5,100 is steam-electric located in New York
City. The remainder comprises gas turbines and shares of steam-electric units
located elsewhere. Figure 1 depicts representative monthly generation and fuel
usage projected for the early 1990's. From the figure it is clear that two annual
peak generation periods occur, one in December/January and the other in July/August.
In 1990 the peak generation days were on January 8 and July 5. As can be seen in
Figure 1 the total actual generation by fossil-fuel steam units is around 40% of the
maximum possible over the year.
From Figure 1 the seasonal shift in fuel mix is clearly seen, with oil
predominating from October through April and gas fuel sharing the load throughout
the summer. This seasonal fuel-mix characteristic has significant implications on
NOx emissions and control strategies.
As mentioned above, the Con Edison boilers were, for the most part, designed
for coal firing and therefore exhibit low NOx emission characteristics. Table 2
shows a comparison between similar classes of boilers (size, design) at Con Edison
and at other utilities with typical gas/oil-design boilers. All data shown are from
test data acquired within the past several years. The Con Edison baseline emissions
8-67
-------
measurements have not been completed. It is clear that the Con Edison boilers have
considerably lower baseline NOx emissions with gas fuel than comparable boilers
elsewhere. With oil fuel the difference is not as clear, although the Con Edison
emissions are among the lower emission levels. The principal implication of the low
initial (baseline) NOx emission levels at Con Edison is that the percentage
reduction in NOx emissions achievable with most NOx control technologies depends to
some degree on the initial NOx level prior to application of the technology.
The baseline NOx emissions shown in Table 2 and used for analysis of potential
NOx reduction capability are derived from short-term, carefully controlled
engineering tests performed with steady-state boiler operation. While these data
are useful for defining the effects of various controllable operating parameters on
NOx emissions, it should be understood that continuous, day-to-day operation of a
unit does not necessarily produce, on average, the same NOx emissions as a short-
term engineering test, even at nominally the same firing conditions. Thus, there is
a degree of uncertainty as to the actual NOx emission to be expected over a longer
time span.
Under Automatic Generation Control (AGC) the load on a unit (firing rate) is
controlled by a central dispatch computer and can cycle continuously over its normal
load range. In this transient mode of operation it is not always possible to
maintain the "optimum" specified firing conditions (excess 02, burner pattern, etc)
vs. load. In order to avoid unsafe conditions as the unit is automatically
controlled over the load range, operators will tend to set a safety margin of excess
02 above the ideal, steady state point at a given load level, and thus the NOx
emission will be increased somewhat. Also, over a longer period of time, boiler
furnace walls may become dirty between soot-blowing periods, burners may deteriorate
slightly and other uncontrollable factors may tend to increase NOx emissions over
the values defined in short-term testing. Figure 3 illustrates the considerable
variability of baseline NOx emissions with AGC control in comparison to the baseline
NOx emissions derived from short-term testing. Thus, in order to maintain NOx
emissions consistently below a specified regulatory limit, the operator would have
to either reduce the average NOx emission well below the limit (so that the peak NOx
emission was still below the limit) or reduce the variability of the NOx emissions
about the average value by maintaining tighter control of excess 02, boiler wall
cleanliness, etc.
NOx CONTROL TECHNOLOGIES
The technologies selected for inclusion in the study are those which have been
historically employed on an operational basis for NOx control on gas/oil fired
utility boilers and certain other developing technologies close to
commercialization. Descriptions of these technologies have been well documented in
the published literature and the discussion presented here is confined to pertinent
information relating to NOx control capabilities. Considerable uncertainty exists
as to the control capabilities of most of the candidate control options. The NOx
reduction algorithms employed in the preliminary analysis are current best
estimates. An effort is being conducted as part of the program to refine these
estimates for final analysis.
OFF-STOICHIOMETRIC FIRING fO.S.)
This control option has been effectively employed by a number of utilities to
achieve significant NOx reductions on gas/oil fired boilers. Figure 2 (abstracted
from Ref. 1) shows the results achieved by one utility (Southern California Edison
Co.) employing O.S. firing on a range of boilers firing natural gas fuel. These
results are representative of those demonstrated in other utility systems which
8-68
-------
generally indicate a NOx reduction dependency on initial, uncontrolled NOx level.
The shaded area in the figure depicts the range of NOx reductions demonstrated in
the current Con Edison test program and confirms the dependency of control
capability on initial NOx level. Similar trends have been demonstrated for oil fuel
firing. The Con Edison O.S. test data generally fall in the range of 30% NOx
reduction, which is substantially less than the control capability normally
associated with this technology but is explained by the low baseline NOx levels.
The steady state, short term data acquired in the test program for O.S. firing
have been used in the analysis for the performance of this control option. This
data may substantially overstate the magnitude of NOx reduction that could actually
be achieved during normal AGC operation. Figure 3 shows a comparison between steady
state and AGC test data for one of Con Edison's units in uncontrolled and O.S.
operating modes. The AGC data shows considerable scatter and does not reflect any
NOx reduction benefit for O.S. firing in comparison in the steady state data.
Similar data scatter has been observed for baseline operation. The data scatter is
due primarily to variations in operating excess air and to boiler cleanliness
effects resulting from switching back and forth between natural gas and fuel oil
firing. It may be possible to narrow the data scatter band by improving operating
procedures and air flow control, but differences between steady state and AGC NOx
emissions cannot be eliminated. The implication of these results is that both
baseline and O.S. operating mode NOx emissions should be predicted on the basis of
AGC operation, which is the intent for the final analysis.
LOW NOx BURNERS (LNB)
There are very few installations of LNB's on gas/oil fired utility boilers and
there is little published data reporting NOx control performance. Ref. 2 provides
preliminary data for installation of one such burner design on two gas/oil fired
utility boilers. The test results demonstrated an improvement over that which had
been achieved for O.S. firing in the range of 10-20%. On the basis of these
results, the analysis assumes an NOx control performance for the LNB control
technology of 10% greater NOx reduction than that achieved in the O.S. testing of
the Con Edison units.
UREA INJECTION (UREA)
UREA injection is a developing technology which is likely to have widespread
future application in utility systems for NOx control Versions of this technology
are currently being demonstrated on several boilers in the Southern California
Edison system. NOx reduction data acquired in these programs have been employed for
the present study to formulate a NOx control algorithm. The data have been
extrapolated to lower initial NOx levels than tested by kinetic analysis. The model
thus formulated was used in the analysis and is shown in Fig. 4.
The EXXON Thermal DeNOx technology which is similar to UREA injection except
that the reagent is ammonia, could be employed as an alternative to UREA injection.
For the purposes of this initial study, the UREA technology has been assumed to be
representative of this general category of NOx control approach.
WINDBOX FLUE GAS RECIRCULATION (WFGR)
WFGR has been employed on both new and existing gas/oil fired boilers for NOx
control. The technology has been demonstrated to be a very effective NOx control
option but little data exists in the published literature pertaining to it's control
performance. Reference 2 reports some data for two retrofit installations in the
Southern California Edison system. This data has been utilized to formulate a NOx
8-69
-------
control model for natural gas and fuel oil firing which is shown in Fig. 5. The
nitrogen content of the fuel oil applying to the test data is 0.3% which is
essentially the same as for the Con Edison fuel.
REBURNING
The Con Edison boilers are particularly suitable for the application of the
Reburning technology because of their uncharacteristically large furnaces for
gas/oil fired units. This technology was not considered in the analysis, however,
due to the lack of sufficient data to estimate NOx control performance, particularly
at low initial NOx levels.
SELECTIVE CATALYTIC REDUCTION (SCR)
SCR was assumed to have a NOx reduction capability of 80% for all initial NOx
levels.
SYSTEM NOx MODEL
A PC-based spreadsheet model was written to calculate the NOx emissions and
cost of control for any combination of control technologies for the Con Edison
system, and for each boiler unit individually. The model comprises three
functional areas: data input, calculations and summary.
In the data input area the user enters the conditions defining the specific
case to be evaluated. After the first run, only those data which change from case
to case need to be entered each run. The input data fall into three categories:
general description of the case, NOx control selection, and unit loading schedules.
The general description data include case number and narrative description of the
case conditions. The NOx selection input consists of completing a matrix table of
NOx control technologies for each unit in the system. The final data input consists
of loading schedules for each unit for both short term (1 hour to many days) and
annual periods. The short-term period is intended to provide the total and average
NOx emissions from each unit over a specified duration (e.g. 8 hours, 1 day, 1 week,
etc). The annual period is used to calculated the NOx emissions, generation, fuel
consumption and variable control costs over a year's time. For each time period the
user inputs the hours of operation of each unit, at each of five (5) load levels and
for each fuel used. The specification of hours of operation at each load level is
important since NOx emissions are variable (usually non-linear) with load, and
therefore the load history must be known in order to calculate integrated NOx
emi ssions.
Also located in the data input area, but usually not changed by the user, are
tables of NOx reduction effectiveness and generic costs (capital and O&M) for each
control technology. Capital costs are specified in $/KW and variable O&M costs in
terms of $ per unit of generation or of tons of NOx removed.
The calculation area of the model begins with tables of baseline NOx
emissions, (Ib/mmBtu) vs load for each unit and each fuel fired. Similar tables of
NOx emissions vs load are provided for O.S. firing conditions.
Controlled NOx emissions (in Ib/mmBtu) are calculated sequentially for each
technology specified in the data input area. Thus, each technology effectiveness
(and resulting NOx output) is dependent upon the output NOx level of the preceding
technology. For example, if both LNB's and FGR are selected for a unit, then the
FGR effectiveness at each load level of the unit will depend upon the LNB output NOx
level at the corresponding load. Of course, each technology not selected has no
effect on the NOx level.
8-70
-------
Following the last application of NOx technology to each unit, the final
outlet NOx level is determined at each load level for each fuel. Based upon the
hours of operation at each load level for each fuel specified in the input tables,
the total short-term and annual NOx emissions (Ib/NOx) are integrated for each unit,
along with the total generation (kwh) and thermal input (Btu).
The cost of NOx control is calculated for each unit by summing each cost
element (capital, fixed O&M, variable O&M) for each technology used. The capital
cost for each selected technology is the generic cost ($/kw) times the unit rating
(kw) times a unit-specific multiplier which represents the degree of difficulty of
applying each technology to that unit. Similarly, the variable O&M cost of each
unit is calculated as the sum of each applied technology's variable O&M cost, which
is the product of the generic cost ($/kw or $/ton NOx) times the annual usage (kwh
or tons NOx) times a unit-specific cost multiplier. Fixed annual O&M costs are the
specified generic fixed O&M costs ($/yr) times a cost multiplier for each unit.
Finally, capital costs are level ized by multiplying the total capital cost for each
unit by a recovery factor representing a specified time period (e.g. 20 years) and
a rate of return (e.g. 10%). Similarly, the total annual O&M costs are levelized
according to standard procedures to account for rising O&M costs over the economic
life of the project, essentially in accordance with the EPRI TAG procedures. The
capital and O&M levelizing factors are entered by the user.
The final function of the spreadsheet model is to compile the emission and
cost results for each unit into a total for the system (including appropriate system
averages, such as Ib/mmBTU NOx emission) and to present the results in a concise
tabular format.
By calculating the-unit specific emissions and costs (and therefore the system
emissions and costs) for a successive series of varied NOx control applications, the
user can determine the lowest-total-cost combination of controls which will result
in total system emissions meeting any specified level for any specified time-
averaging period.
ANALYSIS RESULTS
The Con Edison System NOx model has been constructed and is fully operational,
but preparation of input information has only been partially completed. Selected
analyses have been performed, however, by utilizing that information which has been
developed and by otherwise employing prior information in ETEC's possession and best
estimates. The results of these analyses are reported herein and although they are
subject to some level of uncertainty in terms of magnitude, derived trends and
observations based on these trends are believed to be generally valid.
Figures 6 and 7 show calculated system NOx emissions for 24 hour periods
coinciding with peak generating days in July and December for baseline operation and
for various NOx control strategies. Each plotted data point corresponds to a
specific control strategy consisting of the application of various combinations of
NOx reduction technologies to each unit in the system. Solid symbols denote that
the indicated control combination has been uniformly applied to all units in the
system while open symbols indicate selective utilization. In this latter case, the
letters "Fg" indicate WFGR applications on only gas/oil fired boilers (excluding oil
only units) and a numeral denotes the limited number of unit applications of the
technology identified by the end letter in the sequence (ie OU(5) denotes O.S. on
all units and UREA on 5 units).
8-71
-------
The results apply to actual unit load duration curves for 1990 but the fuel
mix has been altered to reflect maximum gas burning in July and maximum oil burning
in January (ie. dual fuel units burn either all gas or all oil depending on the
month). This allocation of fuels burned approximates that shown in Fig. 1 which is
based on a PROMOD projection. The indicated NOx emissions for each strategy have
been determined by summing the respective integrations over each unit's load
duration curve of the emission rate applying to the fuel burned and the combination
of control technologies installed on the unit.
The baseline (uncontrolled) NOx emissions indicated in the figures have been
determined on the basis of the steady state test data acquired to date and estimates
for as yet untested units. The levels shown understate actual NOx emissions since
they do not reflect the effects of AGC operation, dual fuel firing and boiler
cleanliness in switching between fuels. Each of these factors would tend to
increase unit baseline, and hence system, NOx emissions. The reduced emission
levels shown to be achievable by the application of the various strategies are also
overstated in this regard since they are based on the baseline emissions. Aside
from this factor, the achievable reductions have been determined employing
potentially overly optimistic estimates of the NOx control capabilities of the
individual control options, as pointed out previously. As a consequence of the
above factors, the results as shown are probably too low and the rate of decline in
achievable emissions with increasing control cost is too steep.
The analysis results shown in Figure 6 and 7 are primarily of interest to Con
Edison. It is possible, however, to draw certain observations based on the
indicated trends that may be of more general interest to other gas/oil utilities and
these are discussed below.
OPTIMUM NOx CONTROL STRATEGIES
The purpose of the analysis was to determine the minimum control cost to
achieve varying levels of NOx emission reduction. This cost would be represented by
a curve defining the locus of minimum control cost strategies for achieving
successively reduced levels of NOx emission. Defining such a curve by employing the
model is an iterative procedure in which various strategies are analyzed and the
calculated NOx emission levels and costs are compared. This procedure was followed
in the present case and the optimum strategies determined are those shown in Figures
6 and 7 as being the lowest points at any cost level.
The strategies that were analyzed only broadly define the optimum curve since
intermediate steps have not yet been evaluated. For instance, the locus of
strategies between O.S. on all units and O.S. plus UREA on all units would be
defined by the intermediate steps of sequentially adding UREA combined with O.S. to
successive units. Two such intermediate steps are shown in Figures 6 and 7 for
OU(3) and OU(5).
The analysis results indicate that the optimum strategy to achieve a specific
level of NOx emissions would consist of maximizing the system wide utilization of
the lowest cost technologies first before employing on any unit the next most costly
technology. For instance, it would always be more cost effective to employ UREA on
additional units compared to employing the next most costly technology, which in
this case would be WFGR, on any additional unit. This analysis result is summarized
below:
8-72
-------
Strategy for Increasing Order of Control Option Application
Levels of NOx Control All units Successive units
I O.S.
II O.S. + UREA
III O.S. + UREA + WFGR
IV O.S.+ UREA + WFGR + SCR
LNB could be employed as a substitute technology for O.S., providing an added
10% increment in NOx reduction. However, the combination of O.S. plus UREA would
always be more cost effective than the utilization of LNB's. WFGR would be employed
in an optimum strategy only on gas/oil fired boilers since it's control capability
for reduced initial NOx levels is too low for cost effective utilization on oil-only
boilers.
The above ranking order for utilization of control technologies would apply
only to situations in which an emission regulation were expressed as a LB/day
emission limit averaged over a system. Alternative forms of emission limits would
likely result in a different ordering of technologies for optimum employment.
DIMINISHING RETURN
Figures 6 and 7 graphically illustrate the diminishing return of increasing
expenditure to reduce NOx emission from the Con Edison system. This observation is
quantified in the table below which applies to the optimum locus of strategies in
Figure 6.
System NOx Emission Cost of control
Reduction, % Mill/KWH
50 .4
70 1.4
75 1.8
80 4.5
The table values show for instance that an 80% emission reduction would
require a factor of three greater expenditure than a 70% reduction. This trend is
actually understated since the achievable emission reductions shown in the figures
are optimistic as explained previously.
SEASONAL INFLUENCE ON COST OF CONTROL
Figure 8 replots the optimum strategies defined in Figures 6 and 7 in terms of
daily emissions averaged on a LB/MMBTU basis. Peak day NOx emissions are shown to
be higher in January than in July. The reason for this is attributable to higher
baseline NOx emissions in January due to substantially increased oil firing, to
differences in unit loading schedules and to generally reduced NOx control
effectiveness for some of the technologies for oil firing.
The difference in emission rates for the two seasons is particularly
significant if a regulation were passed of a form limiting emissions on a LB/MMBTU
basis. The inset table in the figure shows that the cost of compliance in this
instance would be at least a factor of two greater for January in comparison to
July. The purpose of such a regulation, however, would be to reduce ambient ozone
concentrations, which tend to be most pronounced during the summer months because of
8-73
-------
meteorological conditions favoring their formation. Therefore, a regulation of this
form would result in an additional expenditure that would serve no environmental
purpose. In such a situation, the emission limit should be formulated to cost
effectively achieve it's intended purpose.
CONCLUSIONS
1) A system NOx emissions model of the type described can be a useful tool in
assessing the implications of a potential regulation in advance of it's
promulgation for preparing a utility for the regulatory process.
2) The Con Edison boilers have low uncontrolled baseline NOx emissions because of
their design and low capacity factors. In such instances, it is more difficult
to reduce NOx emissions because of the reduced effectiveness of NOx control
technologies for low initial NOx levels.
3) The process of establishing NOx emission regulations should recognize that
relatively small differences in control limits can have a dramatic effect on the
required cost of control.
4) The form of an emission regulation can inadvertently result in the expenditure
of unnecessary control costs if it does not specifically address it's intended
purpose.
REFERENCES
1) Bagwell, F.A., et.al., "Utility Boiler Operating Modes for Reduced Nitric
Oxide Emissions", JAPCA, November, 1971
2) Bayard de Volo, N., et.al., "NOx Reduction and Operational Performance of
Two Full-Scale Utility Gas/Oil Burner Retrofit Installations", 1991 Joint
Symposium of Stationary Combustion NOx Control, Washington, D.C.,
March 25-28, 1991
8-74
-------
TABLE I
CON EDISON GENERATING UNITS
Plant
Function
POWER
POWER PLUS
STEAM
SENDOUT
STEAM
SENDOUT
Plant
ARTHUR KILL
ASTORIA
RAVENSWOOD
EAST RIVER
(Pwr Only)
59TH St.
WATERSIDE
74TH ST.
HUDSON AVE.
RAVENSWOOD
E.RIVER SO.
59TH ST.
74TH ST.
60TH ST.
Unit
20
30
10
20
30
40
50
10
20
30
50
60
70
110
111
112
113
114
115
41
42
51
52
61
62
80
90
120
121
122
71,72
81,82
100
4 units
10 units
3 units
6 units
6 units
Capacity
MW
345
440
187
173
365
375
375
95
395
900
148
148
180
72
43
43
43
79
79
71
71
71
71
97
97
160
160
64
64
64
187
MLB/HR
275 EA
150 EA
150 EA
150 EA
150 EA
Mfg
B&W
CE
B&W
B&W
B&W
CE
CE
CE
CE
CE
B&W
B&W
FW
B&W
B&W
B&W
B&W
CE
CE
CE
CE
CE
CE
CE
CE
CE
CE
CE
CE
CE
CE
CE
B&W
B&W
FW
FW
FW
FW
Firing
Config.
Face
Corner
Face
Face
Face
Corner
Corner
Corner
Corner
Corner
Opposing
Opposing
Face
Face
Face
Face
Face
Corner
Corner
Corner
Corner
Corner
Corner
Corner
Corner
Corner
Corner
Corner
Corner
Corner
Face
Face
Face
Face
Package
Package
Package
Package
No of Burn.
32
40
22
22
32
32
32
32
32
64
12
12
18
8
5
5
5
8
8
8
8
8
8
8
8
16
16
8
8
8
8
8
16
6
2
2
2
2
Fuel
O-Oil
G-Gas
O
O
G.O
G,0
G,O
G,O
G,O
G,O
G,0
0
G,0
G,O
G,O
0
0
O
O
0
0
G,0
G,0
G,O
G,0
G,0
G,0
G,0
G,O
0
O
0
0
0
0
0
0
O
0
G
8-75
-------
TABLE II
CON EDISON BOILER CURRENT NOx EMISSIONS
AND COMPARISON WITH OTHER UTILITY BOILERS
FIRING
CONFIGURATION
Single Face Fired
Opposed Fired
T Fired
SIZE
MW
175
175
180
187
215
230
345
365
148
225
230
350
480
750
320
395
440
900/2
FULL LOAD, UNCONTROLLED NOx EMISSIONS,
PPM
GAS
OTHER
UTILITY
405
750
520
337
550
360
890
700
1200
335
CON ED
300
175
—
225
275
150
—
—
OIL
OTHER
UTILITY
—
450
250
370
...
250
425
320
750
225
CON ED
250
300
250
325
250
175
200
275
8-76
-------
2,500
00
LU
CO
1,500
g 2,000
O
O
I
LJJ
Z
LJJ
O
=! 1,000
CO
CO
O
500
0
OIL
Maximum Fossil Fuel
Generating Capability for one month
4880 GWH
50 million
40
30
20
10
JAN FEB MAR APR MAY JUN JUL AUG SEP OCT NOV DEC
MONTH
FIG. 1 Projected Con Edison Fossil Fuel Generation
and Fuel Consumption for the early 1990's
CO
E
E
z"
O
CO
z
O
O
_i
LU
=)
-------
Normal
Operation
High
Excess
Air
Furnace
Operation
Fuel-Rich Burner Operation
1.4 Burner Equivalence Ratio
71 Burner % Air
Furnace % Excess 0 „
FIG. 2 Off-stoichiometric Combustion for Natural Gas Firing
8-78
-------
a>
-^i
C£>
OJ
O
E
Q.
Q.
V)
z
O
V)
uy
LU
x
O
300 -
20°
100
0
0
AGC OPERATION
/\ UNCONTROLLED
O OS
STEADY STATE TEST DATA
UNCONTROLLED
O.S.
50
100
MW
150
200
FIG. 3 NOx EMISSIONS BAND DURING AGC OPERATION
ON GAS FUEL FOR ASTORIA UNIT # 10
-------
40
80
240
280
120 160 200
Initial NOx, ppm
Figure 4. NOx Control Effectiveness of UREA Injection versus Initial NOx Level
240
280
120 160 200
Initial NOx, ppm
FIG. 5 NOx Control Effectiveness of 20% WFGR versus Initial NOx Level for
Gas and Oil Fuels
8-80
-------
C\l\i
CD
_l A
O
0
g 150
CO"
0
CO
CO
•5
gj 100
?~
^c
Q
oo *:
<
LJJ
n
c *
x
4
TECHNOLOGIES APPLIED TO SELECTED UNFTS
LFU TO ALL APPLICABLE UNFTS. SCR TO (5) UNITS
B BASELINE, NO CONTROL
O OFF-STOICHIOMETRIC FIRING
L LOW-NOx BURNERS
F FLUE GAS RECIRCULATION
n GAS FIRED UNITS ONLY
9
U UREA INJECTION
S SELECTIVE CATALYTIC REDUCTION
AOU . LF
^ Lu A OUS(2)
^
OFgU A LUS(2)
~ A A°™LFU A
U=gU A
LFUS(5)
A
. LFUS
A
II 1 1 i I I
01 23 45 6789 10 11 1-
LEVELIZED COST OF CONTROL, Mill/kwh
FIG. 6 Optimum System NOx Control Strategy to Achieve Varying Levels
of Emission Reduction for Peak Generating Day in July
-------
<£OU
CO
0 200
O A
O
CO"
§ 150
CO
^
LJJ
co Q 10°
CO ^
LJJ
Q_
o:
I 50
C\J
0
c
JANUARY
k B
A° ^
A
OU(3) .0F
A°U(5)
ALF
— ^ LU(5) A
OUS(2)
. ou A
LEGEND
A LHJ TECHNOLOGIES APPLIED TO ALL APPLICABLE UNITS
A LFUS<5) TECHNOLOGIES APPLIED TO SELECTED UNITS
LFU TO ALL APPLICABLE UNITS, SCR TO (5) UNITS
B BASELINE, NO CONTROL
O OFF-STOICHIOMETRIC FIRING
L LOW-NOx BURNERS
F FLUE GAS RECIRCULATION
g GAS FIRED UNITS ONLY
U UREA INJECTION
S SELECTIVE CATALYTIC REDUCTION
. s
A ALFU A
LFgU m
LFUS
A
ii ii
1 23 45 6 78 9 10 11 1!
LEVELIZED COST OF CONTROL, Mill/kwh
FIG. 7 Optimum System NOx Control Strategy to Achieve Varying Levels
of Emission Reduction for Peak Generating Day in January
-------
0.35
00
do
CO
CD
E
.E
m
_j
CO
CO
CO
LU
LU
Q_
DC
CM
0.30
0.25
POSSIBLE
EMISSION
LIMIT
LB/MMBTU
A 0.13
B 0.08
% ADDmONAL COST OF CONTROL
FOR JANUARY COMPLIANCE
IN COMPARISON TO JULY
100
120
JULY, 1990
JANUARY, 1990
10
2468
LEVELIZED COST OF CONTROL, Mill/kwh
FIG. 8 Added Cost of NOx Control to Comply With LB/MMBTU
Emission Limit in January in comparison to July
12
-------
REDUCED NOx, PARTICULATE, AND OPACITY ON THE
KAHE UNIT 6 LOW-NOx BURNER SYSTEM
Stephen E. Kerho
Dan V. Giovanni
ELECTRIC POWER TECHNOLOGIES, INC.
Menlo Park, California
J. L. B. Yee
HAWAIIAN ELECTRIC COMPANY, INC.
Honolulu, Hawaii
David Eskinazi
ELECTRIC POWER RESEARCH INSTITUTE
Palo Alto, California
-------
REDUCED NOx, PARTICULATE, AND OPACITY ON THE
KAHE UNIT 6 LOW-NOx BURNER SYSTEM
Stephen E. Kerho
Dan V. Giovanni
ELECTRIC POWER TECHNOLOGIES, INC
Menlo Park, California
J. L. B. Yee
HAWAIIAN ELECTRIC COMPANY, INC
Honolulu, Hawaii
David Eskinazi
ELECTRIC POWER RESEARCH INSTITUTE
Palo Alto, California
ABSTRACT
Hawaiian Electric Company (HECO) completed major combustion system
modifications in mid-1988 on Kahe Unit 6, a Babcock & Wilcox (B&W) oil-fired unit
rated at 146 MW. The modifications were undertaken to reduce emissions of NOx and
particulate matter, and to restore operational flexibility that had been restricted with
burner-out-of-service operation previously used for NOx control. Modifications
included installation of the B&W PG-DRB burners, front and rear wall overfire air
(OFA) ports, extensive ductwork for the OFA and flue gas retirculation (FGR) flows,
and upgrading of the automatic burner control system. This installation represented
the first application of this type of low-NOx firing system to a utility boiler in the
United States.
As reported in 1989, the NOx reduction goal of emissions below 0.23 Ib/MBtu was
achieved and particulate emissions were controlled to below 0.1 Ib/MBtu. However
opacity levels increased from pre-retrofit levels of approximately 6% to between 15-
20%. In an attempt to reduce opacity levels and still comply with NOx emission
limits, HECO and the Electric Power Research Institute jointly sponsored a follow-on
Phase 2 performance improvement program conducted by Electric Power
Technologies, Inc to evaluate the potential of new atomizer designs to reduce NOx,
particulate, and opacity. The program demonstrated significantly reduced opacity and
particulate levels while maintaining NOx emissions below 0.23 Ib/MBtu even though
the levels of OFA and FGR were reduced.
8-87
-------
INTRODUCTION
In July 1987, the Hawaiian Electric Company (HECO) contracted with the Babcock &
Wilcox Company (B&W) to retrofit a low NOx combustion system on their 146 MW
(grossT oil-fired Kahe Unit 6. The unit is front wall-fired and burns oil with a
maximum sulfur content of 0.5%. Up to this time, the unit had been operating with
flue gas recirculation (FGR) to the combustion air and burners-out-of-service (BOOS)
in an attempt to satisfy the operating permit requirement for maximum NOx
emissions of 0.23 Ib/MBtu (180 ppm, dry, 3% O2). Typical emissions using these
controls were 0.28 Ib/MBtu NOx (219 ppm) and 0.06 - 0.08 Ib/MBtu particulate matter
(PM). Normal opacity levels were in the 4-6% range, which is below the visible
threshold.
The principal objective of the retrofit was to reduce NOx emissions to below the
regulatory requirement while minimizing particulate matter (PM) emissions.
Additionally it was intended that the retrofit technology would allow a return to all-
burners-in-service operation, thereby improving the operating flexibility of the unit
which had been impaired with BOOS operation. Specifically, a higher turndown was
expected from improved flame stability at low loads (the lowest load for dispatch was
95 MW with BOOS), and a higher reliability in achieving full load was expected with
the ability to accommodate burner maintenance outages without load reduction. The
project was the first installation in the United States of the integrated application of
low-NOx burners, FGR to both the combustion air and directly to the burners, and a
state-of-the-art front and rear wall overfire air (OFA) design to a heavy oil-fired utility
boiler. The combustion system, designated "PG-DRB", is licensed by B&W from
Babcock-Hitachi (BHK) who commercialized the technology in Japan.
The retrofit was successful in meeting the NOx requirement of the operating permit
and in providing the desired improved operating flexibility. However, operating
problems such as undesirable opacity levels led to a follow-on Phase 2 program of
combustion optimization work and equipment modifidation. This paper presents the
results of the follow-on program which was conducted in 1990.
OVERVIEW OF 1987 NOx SYSTEM RETROFIT
Kahe 6 is a radiant reheat type steam-electric unit manufactured by B&W. An
elevation view is presented in Figure 1. For NOx control, the boiler was originally
equipped with nine B&W dual register burners arranged in a 3 X 3 array on one wall,
and flue gas recirculation to the windbox which permitted up to 20% of the flue gas to
8-88
-------
be mixed with combustion air prior to the burners. The retrofit PG-DRB system
consisted of the following elements:
1. PG-DRB burners
2. Dual fluid (steam/oil) atomizers
3. Utilization of existing FGR to the windbox combustion air
4. Primary gas (PG) system which supplies FGR directly to the burners
unmixed with the combustion air
5. Overfire air system
6. Upgraded control system
The PG-DRB burner, shown in Figure 2, consists of an oil atomizer/impeller located
axially in the primary (core) air zone of the burner. The core air is introduced into the
center zone through slots located at the back of the burner. Core air flow is limited to a
maximum of approximately 10% of the total air flow. The flow to this region can be
controlled with a small sliding disk. The core zone is surrounded by the PG zone,
which is enclosed by the inner and outer air zones. Pure gas recirculation is fed
through a perforated plate located at the entrance to the PG zone annulus which helps
to distribute the flow around the periphery of the zone. A butterfly-type valve
provides controllability of the PG flow to individual burners. Air to the inner and
outer air zones is controlled by a single sliding disk. An impact-suction pilot tube grid
is installed prior to the inner and outer air zones to allow measurement of the airflow
in these zones. The pilot grid consists of a manifold which encompasses Ihe burner
with six finger-like extensions into Ihe total air flow zones. These measurements,
togelher wilh air slide position, provide Ihe capability of controlling air flow to the
individual burners. The inner air zone contains gear driven spin vanes, while the
outer zone has fixed spin vanes followed by gear driven spin vanes. The manually
operated gear driven vanes provide the ability lo vary swirl characteristics and Ihus
Ihe resulting flame shape of the burner.
The OFA system was designed to divert up to 30% of the tolal combustion air to six
OFA ports located on the front and rear boiler walls (three ports per wall),
approximately 10 feet above the top burner elevation. Each OFA port is equipped with
damper assemblies and air spin vanes to allow independent control of air quantity,
velocity, and furnace penetration. A schematic showing the port design is provided in
Figure 3. Like the burners, the OFA ports were equipped with flow monitors, allowing
on-line measurement of separate flows through the spin annulus and central core of
each overfire air port. Flow modeling tests using a scale model of the windbox and
furnace were used by B&W to obtain air flow distribution information for the windbox
8-89
-------
and OFA system. The model results were used to establish placement and sizing of the
OFA ports for optimum mixing. The modeling results were the basis for the decision
to use six ports (instead of three) and the recommendation for a nominal 70:30 rear-to-
front wall distribution of overfire air.
Summary of Retrofit Low NOx System Performance Evaluation
The results of the program were presented in detail at the 1989 Symposium (Reference
1) and are summarized below. The retrofit realized its principal goal to reduce NOx
emissions to below 0.23 Ib/MBtu with all burners in service. At 145 MWg, NOx and
PM emissions levels of 0.21 and 0.07 Ib/MBtu respectively were achieved with a stack
opacity of 15%. The fuel nitrogen content was approximately 0.3% (wt). The test was
performed using 10% FGR (defined as the amount of recirculated flue gas divided by
the sum of the total air and fuel flows) to the windbox and 27% of the total air to the
OFA system (split 70% to the rear ports and 30% to the front ports). These acceptance
test results typified the best overall emissions performance achieved and required an
extensive test effort during the commissioning of the equipment to control PM
emissions and opacity. Although the opacity levels noted above are within the
regulatory requirement of <20% for a 6 minute average, they are considered
undesirable because a visible plume results. These results represented an over 75%
reduction in NOx from pre-retrofit levels with all burners in service (ABIS) and
without FGR.
During commissioning, a strong inverse relationship between NOx and PM/opacity
was encountered. Initially, when the combustion equipment was tuned to achieve
NOx levels below 0.23 Ib/MBtu, the corresponding PM emissions were typically 0.13-
0.15 Ib/MBtu and opacity exceeded 20%. The magnitude of this trade-off was
unexpected from previous experience reported by BHK in Japan, where over 10,000
MW of PG-DRB is operational. It appears that this trade-off is a fundamental feature
of the PG-DRB system when fired with heavy oils. Further assessment of the Japanese
experience in the light of these results led to the conclusion that a similar trade-off
exists at Japanese installations, however it is not an issue there because the boilers are
equipped with electrostatic precipitators for participate and opacity control.
Initial Oil Atomizer Selection
In order to reduce PM emissions, a comprehensive program was implemented by
B&W during commissioning to optimize oil atomization with the PG-DRB burner
system. Improved atomization would result in smaller oil droplets which burn out
8-90
-------
more completely, resulting in reduced PM emissions. During the course of the
program, a number of B&W dual fluid (steam/oil) atomizer designs and atomizer
spray cone angles were evaluated. These included the Y-Jet, Racer, modified Racer
(Racer with increased steam rates), T-Jet, and a developmental I-Jet design. These
atomizer types are characterized by their geometry, steam-to-oil mass flow rates, and
the size of the oil droplets produced. The Racer, Y-Jet, and T-Jet designs were flow
characterized using water and air as the working fluids. Drop size distribution
information was obtained using an Aerometrics Phase Doppler Particle Analyzer. The
conversion of water/air data to oil/steam was done using viscosity, surface tension,
and mass ratio corrections which were obtained from the literature. For oil properties
and operating conditions at Kahe, the nominal Sauter Mean Diameter (SMD) of the oil
droplet size distributions were 400, 320, and 235 microns for the Racer, Y-Jet, and T-Jet
respectively. The T-Jet was judged to provide the best performance and was selected by
B&W for continuous operation. The importance of reducing drop size was
demonstrated by the reductions in PM and opacity achieved from the initial levels: PM
emissions were reduced from 0.13 - 0.15 Ib/MBtu to 0.07 Ib/MBtu and opacity levels
from over 20% to 15%.
LONG-TERM OPERATING EXPERIENCE
Operation at Kahe 6 after approximately two years was characterized by a number of
combustion related problems. Although NOx level