EPRI
Electric Power
Research Institute
£EPA
May 1995
EPRI/EPA1995 Joint Symposium
on Stationary Combustion
NOX Control
Book 3: Thursday, May 18,1995
Sessions 6A, 6B, 7A, 7B
Sponsored by
Electric Power Research Institute
Generation Group
Air Quality Control Program
U.S. Environmental Protection Agency
Air and Energy Engineering Research Laboratory
Combustion Research Branch
May 16-19, 1995
Hyatt Regency Crown Center
Kansas City, Missouri
-------
EPRI/EPA1995 Joint Symposium on Stationary
Combustion NOX Control
Book 3: Thursday, May 18,1995
Sessions 6A, 6B, 7A, 7B
May 16-19, 1995
Hyatt Regency Crown Center
Kansas City, Missouri
Prepared by
ELECTRIC POWER RESEARCH INSTITUTE
Co-Chairs
A. Facchiano, EPRI
A. Miller, EPA
Sponsored by
Electric Power Research Institute
Generation Group
U.S. Environmental Protection Agency
Air and Energy Engineering Research Laboratory
-------
Session 6A
Coal Combustion
-------
Commercialization of Low NOX Cell Burner
(LNCB®) Technology
Edward Mali
Thomas Laursen
Jane Piepho
The Babcock & Wilcox Company
20 South Van Buren Avenue
Barberton, Ohio, U.S.A. 44203-0351
Presented at the EPRI/EPA 1995 Joint Symposium on Stationary Combustion
NOX Control, May 16-19, 1995 in Kansas City, Missouri.
-1-
-------
Abstract
Standard, tightly-spaced cell burners were developed by Babcock &
Wilcox during the 1960s in response to economic demands for highly
efficient burner designs. However, the downside of this 1960s'
design is the production of elevated levels of nitrogen oxides (NOX)
emissions which negatively impact the environment.
Cell-fired units have been designated as Phase II, Group II boilers
under Title IV, Acid Rain Control, of the Clean Air Act Amendments
of 1990 for NOX control. This paper will discuss one technology
developed under the auspices of the U.S. Department of Energy's
Clean Coal Technology program for pulverized coal, cell-fired units
- namely, the Low NOX Cell burner (LNCB®) technology.
The body of this paper will describe the development of Low NOX Cell
burner technology and examine six follow-on commercial contracts.
The purpose of the paper is to identify similarities and
differences in design, fuels, costs and performance results when
compared against the Clean Coal Technology prototype.
Introduction
In 1986 Congress established the Department of Energy's Clean Coal
Technology (DOE-CCT) program to encourage industry to find new ways
to burn coal more efficiently and with significantly less
pollution. These 50% cost share programs had a primary goal of
accelerating the time table from research and development to
commercialization. This paper discusses the 1989 Round III program
"Full Scale Demonstration of Low-NOx Cell Burner Retrofit" (Project
DE-FC22-POP90545) and six follow-on commercial contracts using this
technology.
Low NOX Cell Burner (LNCB®) Technology
Background
The Clean Air Act Amendments (CAAA) were passed by the U.S.
Congress in 1990. Title IV called for reduced NOX levels from all
coal-fired utility boilers greater than 25 MW. Title IV has two
phases for compliance. The cell-fired units are Phase II and must
be in compliance by January 1, 2000. The regulations for Phase II
will be promulgated on January 1, 1997. There are 39 originally
designed, cell-fired units (27,000 MWe) which must be retrofitted
with low NOX technology, NOX averaged with other coal fired units
within an owner's/operator's system, or retired from service.
One method for reducing NOX is to modify the boiler's combustion
system. The traditional and more costly approach for standard cell
burners is to increase burner spacing with extensive pressure part
modifications, and then install conventional, internally-staged low
NOX burners. Increased spacing between burners essentially
reconfigures the boiler from cell-fired to wall-fired. This
approach can be cost effective if the waterwalls in the burner
region of the furnace are approaching end of life and need
replacement and if the utility intends to average NOX across a
-2-
-------
X
number of units.
Goals and Solutions
The objective was to achieve at least a 50% reduction in NO_
emissions without pressure part modifications and without degrading
the boiler's performance or life. The new retrofit technology also
had to be less expensive and easier to install than the traditional
NOX reduction method involving pressure part reconfiguration.
B&W, in association with EPRI, developed the state-of-the-art Low
NOX Cell burner (LNCB®), designed to limit the formation of both
fuel and thermal NOX. The LNCB design replaces the two circular
burners in a standard cell with a single, high-heat input S-type
circular burner and a close-coupled overfire air port or NOX port
(see Figure 1). The key design feature is controlled mixing of the
fuel and air streams to delay combustion and limit NOX formation.
The LNCB concept was designed by B&W to fit directly into existing
cell burner furnace wall openings (see Figure 2) . Only local
changes in coal piping at the burner front are required to
accommodate the S-burner's increased coal-firing capabilities.
Under the DOE-CCT program, Babcock & Wilcox (B&W) conducted a
demonstration on The Dayton Power & Light Company's (DP&L) J. M.
Stuart Station Unit 4, a 605 MWe B&W pulverized coal-fired universal
pressure boiler (co-owned by Columbus and Southern Power Company
and Cincinnati Gas & Electric Company). This project also was
jointly funded by the Ohio Coal Development Office, B&W, the
Electric Power Research Institute (EPRI), DP&L, Centerior Energy
Corporation, Duke Power Company, New England Power Company,
Tennessee Valley Authority and Allegheny Power System.
DOE-CCT Demonstration: Dayton Power & Light
At Stuart Station the burners are arranged in six columns and two
rows of two-high LNCB® burners. The heat input per LNCB® burner is
220 x 106 Btu/hr at maximum continuous rating (MCR) with six mills
in service and 265 x 106 Btu/hr at MCR with five mills in service.
The scope included provisions for future scanners. New lighters
were provided under a separate contract. The fuel burned is an
Eastern bituminous coal and has a fixed carbon to volatile matter
ratio (FC/VM) of 1.44 and a nitrogen content of 1.23% (see Table
1) . During initial implementation of this patented technology,
sophisticated three-dimensional computer analysis programs were
used to optimize the overall LNCB configuration and combustion
performance. This included minimizing CO and H2S concentrations by
inverting a selected number of S-burner and NOX port couplets (see
Figure 3) . Field tests confirmed that this final configuration
produced the desired retrofit performance.
The result: Stuart Station Unit No. 4's baseline NOX level of 1.17
lb/106 Btu had been reduced by greater than 50% and routinely
operates on a system demand basis at less than 0.58 lb/106 Btu. A
secondary benefit is a cleaner furnace and convection pass which
has reduced annual maintenance costs.
-3-
-------
One drawback to the design of the deeply-staged system is the
elevated potential for corrosion on the furnace walls as a result
of localized reducing atmosphere. One commercial remedy for
localized corrosion is a field applied chromized coating. Typical
costs average $95/ft2 and should last a minimum of five years. Many
utilities already routinely use commercial coatings to protect
furnace walls.
Integration of B&W's LNCB technology into the existing pressure
part openings of cell-fired boilers offers many benefits. The
demonstration project at Stuart Station achieved NOX reductions in
excess of 50% with no degradation in combustion efficiency and
minimal change in carbon monoxide emissions. The twin features of
minimum burner piping modifications and no boiler pressure part
changes provide benefits of lower cost and shorter schedule than
other options.
First Commercial Contract: Allegheny Power System
The first commercial contract for LNCB® technology was awarded to
Babcock & Wilcox by Allegheny Power System (APS) for Hatfield's
Ferry Unit 2, a 555 MW universal pressure, once-through boiler
burning Eastern bituminous coal with a fixed carbon/volatile matter
ratio of 1.38 and 1.56% nitrogen in the fuel (see Table 2). This
opposed-wall-fired unit consisted of five columns and two rows of
two-high LNCB® burners. The original 40 burners were replaced with
20 LNCB® burner sets. Each burner is capable of 252 x 106 Btu/hr
with all five mills in service and MCR on the unit or 315 x 106
Btu/hr at MCR with four mills in service.
Again, mathematical modeling was used to optimize the arrangement
of burners and NOX ports (see Figure 4) . The five column arrangement
and the angle setting on the spin vanes and louvers are almost
identical to the DP&L design. However, at Hatfield, the impeller
position is moved back from the front of the nozzle.
Results and Follow-on Contracts: With a NOX baseline of 1.17 lb/106
Btu, the guarantee level was established at 50% reduction or 0.58
lb/106 Btu. This condition was met. Based on the performance of
Hatfield's Ferry 2, APS contracted with B&W to provide the same
design for Units 1 and 3 at this site. These units share the same
design, fuel, and burner/NOX port arrangement as Unit 2.
Unit 1 was installed in October and November 1994. At this writing,
Unit 3 is being installed and will start up in April 1995.
Cost Savings: APS enjoyed some significant cost savings on burner
auxiliaries by reusing existing drives and scanners. In addition on
Unit 3, the original cell burners had been upgraded to convertible
S-type burners for mechanical reliability in the mid-80s. Using a
large portion of the hardware already in place, B&W was able to
field modify 20 S-burners from a 15%" I.D. coal nozzle to a 21"
I.D. nozzle for additional savings to APS.
-4-
-------
Second Commercial Customer: Detroit Edison Company
Detroit Edison Company was the second utility to award Babcock &
Wilcox a contract for LNCB® technology at Monroe Unit 1. This unit
is one of four identical 780 MW universal pressure, once-through
boilers with 28 sets of two high burners arranged in eight columns
on opposite walls and symmetrically about a furnace division wall.
The fuel burned in the Monroe units is a blend of 45% low sulfur
sub-bituminous, 47% mid-sulfur Eastern bituminous and 8% low sulfur
Southern bituminous (see Tables 3 and 4).
Design Differences: While Hatfield's Ferry was similar in design
and fuel type to Stuart #4, the Monroe #1 unit represented a
significant change in furnace design and fuel type. The division
wall in the original Monroe furnace design changed the horizontal
spacing of the bottom row of burners and eliminated the need for a
second set of burners about the two middle columns. Using the
existing openings, the mathematical modelers determined that the
most effective design to reduce NOX and minimize CO formation and
corrosion along the sidewalls was to invert the burner/NOx port of
each outside cell on the lower row. This variation from the earlier
inverted design means that on the lower set of LNCB® burners,
burners oppose burners and NOX ports oppose NOX ports (see Figures
5 and 6).
The blended fuel with a significant portion (45%) of Western sub-
bituminous, low sulfur coal requires a higher fuel flow and
consequently a higher air flow to reach MCR on the unit. Therefore,
the two original 15%" I.D. coal pipes were joined with a Y section
to form a 23" I.D. pipe and coal nozzle. The burners are rated at
251 x 106 Btu/hr with all seven mills in service and 293 x 106
Btu/hr with six mills at MCR conditions.
Results and Follow-on Contracts: Baseline NOX conditions with a
blended fuel were 0.93 lb/106 Btu at 730 MW. Babcock & Wilcox
guaranteed a post-retrofit NOX level of 0.52 lb/106 Btu with CO
levels not to exceed 150 ppm. Initial performance tests were
analyzed at two loads, seven mills at 750 MW yielding NOX levels of
0.52 lb/106 Btu and at 650 MW, yielding 0.48 lb/106 Btu. CO level
guarantees were also met. While the % reduction from baseline
conditions i.e. 44% was not as large as the demonstration program
goal of 50%, the final absolute NOX numbers were lower i.e. .52 vs.
.58 lb/106 Btu. The success at Monroe #1 lead to additional orders
on Monroe #3 and #4.
Cost Comparison of Design Variations
Variations in scope and material costs will range from a low
installed cost of $5.5/kW to $8.5/kW for units with convertible S-
burners to a high of $7.0/kW to $10/kW for units where extended
scope includes new drives, scanners and lighters as well as field
applied coatings for corrosion protection.
-5-
-------
NO, Results Compared Against DP&L Demonstration
Application-specific burner zone heat release rates, furnace
configuration, and coal type (e.g., volatility, fixed carbon level,
bituminous versus sub-bituminous, nitrogen content, and oxygen
content, etc.) will impact expected NOX results. Boilers firing
fuels similar to DP&L's Stuart Station are expected to experience
at least a 50% NOX reduction when retrofitted with LNCB® technology.
The results at Hatfield's Ferry confirm this.
Units with higher burner zone heat release rates than J. M. Stuart
Station will generally have higher baseline NOX levels. For these
units, LNCB technology has the potential to reduce NOX emissions by
50% from original design, baseline levels, but not necessarily to
absolute levels as low as those attained at DP&L Stuart Station.
Conversely, units like Monroe, with lower baseline NOX levels,
burning sub-bituminous coals or blends may not achieve a 50% NOX
reduction but will have lower absolute NOX values than similarly
configured units burning bituminous fuels (see Table 5).
LNCB Receives R&D 100 Award
The Low NOX Cell burner was so successful, cost effective and
innovative, that it was selected to receive the coveted 1994 R&D
100 Award. This award, sponsored by R&D Magazine, recognizes
products or inventions introduced for commercial use during the
preceding year that have had a meaningful and profound impact on
technology or industry. With that in mind, the panel of judges
cited LNCB technology to be "one of the most technologically
significant new products of the year."
Conclusion
B&W has installed or received orders for its LNCB® technology on
more than 5750 megawatts of capacity in the United States.
Regardless of boiler configuration or fuel characteristics, the
advantages of the LNCB technology for utilities include:
No pressure part modifications
Short outage schedule
Same number of original lighters
Elimination of 50% of the original scanners
Potential reuse of existing drives, lighters and scanners
Cleaner furnace and convection surface
NOX reductions of 50% or absolute values below 0.6 lb/106 Btu
Low installed cost
The success of this design on a variety of fuels and boiler
configurations makes it a prudent choice for establishing levels of
NOX reduction for Phase II, Group II cell-fired boilers.
-6-
-------
Acknowledgements
The authors wish to thank the following government agencies,
private entities and utilities whose sponsorship made these
projects possible:
U.S. Department of Energy
Ohio Coal Development Office (Ohio Department of Development)
Electric Power Research Institute
The Babcock & Wilcox Company
The Dayton Power & Light Company
Allegheny Power System
Centerior Energy Corporation
Duke Power Company
New England Power Company
Tennessee Valley Authority
Cincinnati Gas & Electric Company
Columbus and Southern Power Company
Detroit Edison Company
Disclaimer
Reference herein to any specific commercial product, process, or
service by tradename, trademark, manufacture, or otherwise, does
not necessarily constitute or imply its endorsement,
recommendation, or favoring by the DOE. The views and opinions of
the authors expressed herein do not necessarily state or reflect
those of the DOE.
References
1. Laursen, T.A. and Piepho, J.M., "Low NOX Options for Group II
Cell-Fired Boilers," presented at the Air & Waste Management
Association's Symposium on "Acid Rain & Electric Utilities:
Permits, Allowances, Monitoring and Meteorology," Tempe,
Arizona, January 1995.
2. Yagiela, A.S., Laursen, T.A., Bellanca, C.P., et.al., "Results
of Babcock & Wilcox's Clean Coal Technology Combustion
Modification Projects: Coal Reburning for Cyclone Boiler NOX
Control and Low NOX Cell Burner Demonstrations," presented at
the Second Annual Clean Coal Technology Conference, Atlanta,
Georgia, September 1993.
3. Piepho, J., Cioffi, P., LaRue, A., and Waanders, P., "Seven
Different Low NOX Strategies Move from Demonstration to
Commercial Status," presented at Power-Gen 1992, Orlando,
Florida, November 1992.
-7-
-------
TABLE 1 - DAYTON POWER & LIGHT, STUART #4, FUEL
Ultimate Analysis
(as received)
Carbon
Hydrogen
Oxygen
Nitrogen
Sulfur
Moisture
Ash
Total
%
66.52
4.49
7.51
1.23
1.00
5.55
13.70
100.00
Proximate Analysis
(as received)
Moisture
Fixed Carbon
Volatile
Matter
Ash
Total
%
5.55
47.61
33.14
13.70
100.00
FC/VM = 1.44
HHV = 11,880 Btu/lb
TABLE 2 - ALL!
Ultimate
(as received)
Carbon
Hydrogen
Oxygen
Nitrogen
Chlorine
Sulfur
Moisture
Ash
Total
3GHENY POWER SYST
Analysis
%
72.79
4.85
4.95
1.56
0.11
2.16
4.90
8.68
100.00
EM, HATFIELD'S F
Proximate
(as received)
Moisture
Fixed Carbon
Volatile
Matter
Ash
Total
•po /T7M
HHV - 13,
ERRY #2, FUEL*
Analysis
%
4.90
50.13
36.29
8.68
100.00
10 o
.JO
083 Btu/lb
Typical analysis used during guarantee testing.
-8-
-------
TABLE 3 - DETROIT EDISON, MONROE #1, BLENDED FUEL
Ultimate Analysis
Carbon
Hydrogen
Oxygen
Nitrogen
Chlorine
Sulfur
Moisture
Ash
Total
45%
Western
51.96
3.71
12.44
0.70
0.00
0.30
26.28
4.61
100.00
47%
Eastern
72.76
4.88
6.09
1.41
0.09
1.30
6.38
7.09
100.00
8%
Southern
69.86
4.30
8.60
1.22
0.09
0.90
6.05
8.98
100.00
Blend
63.17
4.31
9.15
1.07
0.05
0.82
15.31
6.12
100.00
TABLE 4 - DETROIT EDISON, MONROE #1, BLENDED FUEL
Proximate
Analysis
Fixed
Carbon
Volatile
Matter
Moisture
Ash
FC/VM:
Total
Btu/lb
45% Western
Proximate
36.61
32.50
26.28
4.61
1.13
100.00
9,055
47% Eastern
Proximate
50.46
36.07
6.3
7.09
1.40
100.00
13,090
8% Southern
Proximate
49.65
35.32
6.05
8.98
1.41
100.00
12,620
Blend
Proximate
44.16
34.40
15.31
6.13
1.28
100.00
11,237
-9-
-------
TABLE 5 - COMPARISON OF RESULTS
Site
HHV, Btu/lb
FC/VM
N2
Baseline NOX
Retrofit NOX
% Reduction
DP&L, Stuart
11,880
1.44
1.23%
1.17
<0.58
50%
APS,
Hatfield's Ferry
13,083
1.38
1.56%
1.17
<0.58
50%
DECO, Monroe
(blended
analysis)
11,237
1.28
1.07%
0.93
<0.52
44%
-10-
-------
Air Measuring Device
Louver Damper
Sliding Air
Damper Drive
Louver Damper
Adjustment
Ceramic Lined
Segmented Elbow
Figure 1 - B&W's Low NOX Cell Burner.
Figure 2 - The LNCB was designed to fit directly into existing cell burner furnace wall openings.
-11-
-------
Dayton Power & Light
Stuart #4 605 MW
Scope of Supply:
24 Low NOX Cell Burners
• NOX Port
O Burner
Figure 3 - Demonstration site furnace configuration.
Allegheny Power System
Hatfield's Ferry #2 555 MW
Scope of Supply:
20 Low NOX Cell Burners
20 Reused existing lighters,
scanners and drives
• NOX Port
O Burner
Figure 4 First commercial contract furnace configuration.
-12-
-------
Detroit Edison Company
Monroe #1 780 MW
Scope of Supply:
28 Low NOX Cell Burners
New lighters, scanners
and drives
• NOX Port
O Burner
Figure 5 - Second commercial contract furnace configuration.
-13-
-------
Furnace Wall CO Concentration
Left Side Wall
Front Wall Right Side Wall
Gas Temperature Near Division Wall
Cell Burners LNCB's
Rear Wall
DegF
2900
2800
2700
2600
2500
2400
2300
2200
2100
2000
Figure 6 - DECO LCNB®base case.
-14-
-------
ALTERNATIVE SOLUTIONS FOR
REDUCING NOX EMISSIONS
FROM CELL BURNER BOILERS
by
C. A. Penterson
Senior Staff Engineer, DB Riley
and
S. A. Vierstra, PE
Senior Engineer, American Electric Power Service Company
Abstract
An economical solution for reducing NOX emissions from coal fired utility boilers equipped
with pre-New Source Performance Standard (pre-NSPS) cell burners is being developed.
Since over 12% of the U.S. generating capacity is produced from cell burner boilers, this
development is very significant for the utility industry.
Cell burners, manufactured in the 1950's and 1960's, rapidly mix the pulverized coal and
combustion air resulting in highly turbulent and efficient combustion. Typical NOX emissions
from cell burner boilers average 1.0 to 1.8 lb/106 Btu.
This paper presents the results of retrofitting American Electric Power's Muskingum River
Unit 5, a 600 MWe supercritical cell burner boiler, with Riley low NOX CCV™ burners.
Initial results demonstrated the ability to reduce NOX emissions greater than 50% without the
requirement for overfire air, off stoichiometric firing, burner respacing, mill system and coal
piping changes or pressure part modifications. Long term operation, which recently included
switching coals, has resulted in some deterioration in NOX performance. Results of the initial
optimization testing are presented along with plans for reoptimizing unit operation to restore
the low NOX levels.
Plans for developing Riley's "next generation" low NOX pulverized coal burner (CCV n) will
also be presented. The intent is to potentially utilize this advanced design on other cell
burner units in American Electric Power's system for improved NOX reduction performance.
*DB Riley Corporation 1995
The Company reserves the right to make technical and mechanical changes or revisions resulting from improvements developed by its research
and development work, or availability of new materials in connection with the design of its equipment, or improvements in manufacturing and
construction procedures and engineering standards.
-------
Introduction
American Electric Power (AEP) based in Columbus, Ohio has over 24,000 MWe of installed
electrical generating capacity. Over 5600 MWe of this electrical capacity comes from six (6)
coal fired boilers equipped with pre-NSPS cell burners making AEP one of the largest
owners of these boiler types. The six (6) units range in nominal size from 600-1300 MWe
generating capacity.
In 1993, AEP implemented a program to find a viable solution for reducing NOX emissions
from cell burner boilers. Specific objectives were to:
• reduce NOX emissions by at least 50%
• reduce NOX emissions using low NOX burners that would not require air staging via
overfire air (OFA) or off-stoichiometric firing
• develop a technical and economical low NOX technology for application to all of
AEP's cell burner boilers
• reduce NOX emissions without degradation in boiler performance
A review of available technology in the industry indicated that NOX reduction was possible
by, (1) installing modified cell burners that replace the upper nozzle with an OF A port or by,
(2) respacing and installing low NOX burners to resemble a more conventional wall fired
installation. These options were abandoned because of concerns for lower furnace corrosion,
the long term effects of deep staging on overall boiler operation, and the significant cost
associated with pressure part modifications to rearrange the burners.
The technical approach desired by AEP was to install low NOX burners without OFA as a
direct replacement for all the cell burners. Riley Stoker Corporation (Riley) was selected by
AEP to implement this approach. The significant factors influencing this decision were:
• With the Riley Controlled Combustion Venturi (CCV®) burners we did not anticipate
any requirement for overfire air or air staging for reducing NOX by 50% thus
minimizing lower furnace waterwall corrosion due to reducing atmospheres.
• The CCV® burners were simply a "plug in" replacement for the pre-NSPS cell
burners and would not require costly pressure part changes.
• The CCV® burners would not require any changes to the existing milling system and
coal piping configuration.
• The CCV® burners would not require respacing the burners despite the high furnace
heat release rates typical of these types of boilers.
-------
Unit Description
AEP and its subsidiary, Ohio Power, elected to utilize Muskingum River Unit 5 (MRS),
shown in Figure 1, as the demonstration site for the Riley burner technology. This unit,
originally designed and manufactured by Babcock and Wilcox (B&W) in the early 1960's, is
located in Beverly, Ohio. The boiler is a supercritical design producing superheated steam at
a rate of 4,035,000 Ib/hr, 3800 psig operating pressure and 1000°F operating temperature.
The electrical generating capacity is a nominal 600 MWe.
As shown in Figure 2, pulverized coal was previously burned using twenty (20) B&W two-
nozzle cell burners and ten (10) standard circular burners arranged in an opposed fired
configuration. Figure 3 shows a schematic of the original 2 nozzle cell burner. The unit has
five (5) B&W MPS size 89 pulverizers which feed ten (10) coal nozzles each. The furnace
dimensions are approximately 63' W x 39' D. MRS burned a high volatile bituminous coal
from Central Ohio. Typically, uncontrolled NOX levels from cell burner type boilers average
from 1.0 to 1.8 lb/106 Btu. The uncontrolled NOX level measured at Muskingum River Unit
5 was 1.2 lb/106 Btu. The objective of this project was to reduce NOX emissions to below
0.6 lb/106 Btu with minimal impact on boiler performance and, in particular, flyash unburned
carbon (UBC) levels.
Figure 1
Muskingum River Unit 5 (on right)
-------
Coal Impeller
Positioner Drives
Pulverized Coa
and
Primary Air
Figure 3
Original 2-Nozzle Cell Burner
Figure 2
Muskingum River Unit 5
Low NOX Combustion Retrofit
In the spring of 1993, AEP contracted with Riley for the design, fabrication and installation
of fifty (50) low NOX CCV® burners for Ohio Power's Muskingum River, Unit 5. Each
burner was rated for 107 MMBtu/hr heat input at MCR conditions. The basic CCV® burner
design is shown in Figure 4. The key element of the burner design is the patented (U.S.
Patent No. 4,479,442) venturi coal nozzle and low swirl coal spreader located in the center
of the burner (1). The venturi nozzle concentrates the fuel and air in the center of the coal
nozzle creating a very fuel rich mixture. As this mixture passes over the coal spreader, the
blades divide the coal stream into four (4) distinct streams which then enter the furnace in a
gradual helical pattern producing very gradual mixing of the coal and secondary air.
Secondary air is introduced to the furnace through the air register, supported off the burner
front plate, and subsequently through the burner barrel and over the secondary air diverter.
-------
Devolatilization of the coal in the fuel rich mixture occurs at the burner exit in an oxygen
lean primary combustion zone, resulting in lower fuel NOX conversion. Peak flame
temperature is reduced by the gradual mixing, thus suppressing the thermal NOX formation.
Shroud Drive With Roller Assembly
Air Flow Measurement Device
By Brandt Instruments
(Front Plate Mounted)
Register Turning Vanes
Secondary Air Shroud
Burner Front Plate
Venturi Coal Nozzle
Low Swirl
Coal Spreodi
U.S. Potent 4.479.442
Figure 4
Riley Low NOX CCV® Burner
The resulting coal flame produced by the CCV® burner is longer, narrower and more tubular
in shape than the flames produced by the highly turbulent original B&W cell burners. The
CCV flame is also typically well attached to the coal nozzle tip resulting from recirculation
eddies formed on the backside of the secondary air diverter. For good NOX control, it is
very important to have a well attached coal flame (3). As discussed later, detached coal
flames on the upper row of standard circular burners may be limiting the full NOX reduction
potential of this retrofit.
The furnace depth of 39' is sufficient to accommodate the longer low NOX coal flames
without concern for waterwall flame impingement. However, our most significant concern
on this project was that, with the very close burner spacing (4'-2") between the two (2)
nozzles in a cell burner (Figure 5), the coal flames would interact with each other causing
more turbulent combustion than desired and minimal reduction in NOX emissions.
The new low NOX burners were installed in the fall of 1993 during a six (6) week outage.
The unit was also equipped with a 33 point sampling grid at the economizer outlet for
measuring gas emissions during the post retrofit performance testing. Start-up of MR5
commenced in December, 1993. However, difficulties in trying to utilize the original
mechanically atomized oil igniters, which were shared between the burner nozzles within the
cell, forced a delay in testing the CCV® low NOX burner configuration until June of 1994.
-------
SHROUD flCTUftTOR -
SECONDARY AIR SHROUD
EXISTING CELL THROflT
OIL IGNITOR
RILEY MODEL 90 REGISTER
Figure 5
Riley Low NOX CCV® Cell Burner
Optimization Test Results
Post retrofit optimization testing was performed in early summer, 1994. Our primary
concern regarding flame interaction impacting NOX with the close burner spacing quickly
diminished after observing the CCV® cell burner flame shape. The CCV® cell burner flames
were all well attached , narrow and tubular in shape. The flames were distinctly independent
from each other with little or no flame interaction between adjacent flames.
The upper row of non-cell configured burners also produced a long narrow flame but were
detached by various amounts. The detachment was felt to be the result of:
• windbox air maldistribution
• insufficient diverging throat depth for adequate burner recirculation.
Regardless, the optimization testing continued.
As summarized in Figure 6, NOX was reduced to below 0.6 lb/106 Btu for a 52% reduction
from pre-retrofit levels. CO emissions remained extremely low and consistent with baseline
operation. Flyash LOI averaged 2.23% as compared to 2.7% prior to the retrofit while the
actual % of unburned carbon in the ash averaged < 1% as compared to 1.5% prior to the
retrofit. As shown in Figures 6 and 7, NOX decreased with reduction in load and with
normal increases in excess air for maintaining steam temperature.
-------
Parameter
Test No.
Date
Gross Generation, MW
Excess Air, %
Superheat Outlet Temperature, °F
Reheater 1 Outlet Temperature, °F
Reheater 2 Outlet Temperature, °F
Econ. Outlet Gas Temperature, °F
AH Outlet Gas Temperature, °F
Boiler Efficiency, % (Approx.)
NOX @ Econ. Outlet by CLA(1), 106 Bta
NOX @ Stack by CEMS, Ib/ltf Btu
CO @ Econ. Outlet, ppm
FlyashLOI, %
Carbon in Ash. %
Pre-Retrofit
Baseline
2
'9/16/93
600
25
1000
1023
1024
690
341
87.85
~ 1.2p>
0
2.7
1.5
Post Retrofit
Optimization
9
7/6/94
607
22
994
1010
1003
735
353
87.63
0.56
0.59
1
2.23
0.78
11
7/7/94
454
33
996
970
960
676
328
-
0.53
1
1.89
0.67
10
TH/94
368
40
993
937
917
648
325
-
0.51
1
1.70
0.95
(1) Chemillurranescent Analysis
(2) Measured during previous boiler testing.
Figure 6
Summary of Test Results
1.2
CO
m 0.8
CO
Q 0.6
00
CO
LU
X
O
0.4
0.2
0'—
300
BASELINE NOx
NOx (STACK).
NOx (ECON OUT)
FLYASH LOI
BASELINE UBC
FLYASH UBC
10
CO
0
400 500 600 700
BOILER LOAD, MW
Figure 7
The Effect of Load on NOX and LOI
-------
While the Muskingum River Unit 5 burner arrangement benefits somewhat in terms of NO*
control (baseline and post-retrofit) from a row of circular burners above the two levels of cell
burners (Figure 2), full load is achievable with this row of burners out of service. This
firing condition was also tested to simulate the furnace thermal environment of AEP's cell
fired 1300 MWe units. The NOX emissions under this condition increased at full load to 0.63
lbs/106 Btu, less than the 0.66 lbs/106 Btu predicted, which supports that a nearly 50%
reduction in NOX from the existing 1300 MW units is possible with "plug in" CCV® burners.
It was also observed throughout this testing that there was no significant change in boiler
thermal performance as a result of installing low NOX burners. Figure 8 shows a typical fuel
analysis of the coal burned during the optimization testing along with pulverized coal fineness
results.
Coal Source
Central Ohio
E. Kentucky
Proximate
Moisture %
Volatile Matter, %
Fixed Carbon, %
Ash, %
6.6
39.1
42.5
11.8
6.8
33.54
47.82
11.84
Ultimate
Carbon, %
Hydrogen, %
Oxygen, %
Nitrogen, %
Sulfur, %
68.3
5.0
8.41
0.99
4.7
74.0
4.79
6.20
1.48
0.70
Higher Heating Value, Btu/lb
11,660
12,153
Hardgrove Grindability
50
45
Coal Fineness
% 50 Mesh
% 1 00 Mesh
% 200 Mesh
99.5
95.5
79.1
98.4 99
60 +
Figure 8
Fuel Analysis and Coal Fineness Results
-------
Long Term Operating Results
Long term operation following the optimization testing has resulted in NOX levels
deteriorating. Data collected from the CEMS equipment indicated NOX emissions had
increased to an average 0.69 lb/106 Btu up until mid December, 1994. Then, a coal change
to a lower sulfur E. Kentucky coal occurred. NOX emissions increased further to an average
0.75 lb/106 Btu and also became more erratic. Figure 8 shows a comparison of the fuel
analysis.
The most significant factor affecting the performance is felt to be due to changes in mill
operation as a result of changes in grindability, moisture content and heating value. Primary
air flow quantity was reduced with the new fuel's higher heating value which in turn affects
venturi nozzle velocities, burner mixing and NOX emissions. Other factors may include
boiler seasoning, excess air and burner air distribution.
Plans are to reoptimize the unit operation with particular emphasis on burner air distribution,
proper flame attachment and mill operation. The objective is to reduce the NOX level back
down to below 0.6 lb/106 Btu. Long term parametric testing is planned which will include
evaluating specific variables such as primary air flow, excess air, burner air distribution and
various burner adjustments.
In parallel with this reoptimization effort, Riley will be utilizing mathematical modelling
(Fluent analysis) to evaluate the flame detachment concern for the upper circular burners.
Control of NOX is difficult with detached coal flames. Potential solutions will be developed.
Advanced Low NOX Coal Burners
Riley is currently implementing a program to develop its "next generation" low NOX coal
burner design. The program will evaluate the effects of dual secondary air streams, nozzle
and spreader configurations on controlling NOX emissions. Testing, which is scheduled to
commence in early summer 1995, will be conducted on prototype burners installed in the 100
MMBtu/hr coal burner test facility at Riley Research. All previous testing of CCV burners
have been conducted here for direct comparisons of the test results.
Included in this program will be the installation of DB-Riley's dynamic SLS classifier for
studying the effects of various pulverized coal particle size distributions on emissions
performance and flyash unburned carbon levels. The dynamic classifier will be integrated
with the existing Atrita milling system for direct coal firing. An evaluation of the feasibility
of incorporating dynamic classifiers on Atrita milling systems will be made.
The program will also focus on testing three (3) different coals of various volatile contents
and reactivity. Our field experience has shown a significant impact of coal quality on
combustion efficiency (unburned carbon) and to a lesser extent on NOX. The effects of
various coal qualities on burner performance will be determined during this test program.
-------
Summary
DB-Riley and AEPSC have retrofitted low NOX burner technology on a coal fired 600 MWe
cell burner supercritical unit. Of particular significance is the ability to retrofit with this
"plug in" technology without pressure part modifications, burner respacing and off
stoichiometric firing using overfire air ports.
Test data demonstrated the ability to achieve 50% NOX reduction at MR5 (< 0.6 lb/106 Btu)
without degradation in boiler performance during post retrofit optimization testing. Long
term operation has shown some deterioration in NOX to levels exceeding 0.7 lb/106 Btu.
Efforts to reoptimize the unit operation as well as an evaluation of design upgrades are
continuing.
Advanced burner designs are being tested at Riley's Research facilities for improved NOX
control and combustion efficiency. A total systems approach including the burner, dynamic
classifier and various coals of different quality and characteristics is being evaluated to meet
future requirements. The feasibility of using dynamic classifiers for improved fineness and
combustion efficiency would require an economic analysis on a site specific basis. It is
anticipated future requirements will include stricter air quality standards as well as special
requirements for ash disposal and utilization (3).
Acknowledgements
Special thanks are extended to Ohio Power and AEPSC for their continued support
throughout this low NOX retrofit project. Also, thanks are extended to Riley's Service
Department for their dedicated efforts during the start-up and testing of MR5.
References
1. Penterson, C., "Development of an Economical Low NOX Firing System for Coal Fired
Steam Generators". Presented at the 1982 ASME Joint Power Generation Conference,
Denver, October 1982.
2. Lisauskas, R., Penterson, C., "Applications and Further Enhancement of the Low NOX
CCV® Burner". Presented at the Canadian Electric Association Conference, Toronto,
Canada, March 1994.
3. Proceedings from DOE Conference on Unburned Carbonaceous Material on Utility Fly
Ash, Pittsburgh, Pennsylvania, February 1995.
-------
CO-FIRING COAL-WATER SLURRY IN LOW-NOx BURNERS:
Experience at Penelec's Seward Station
Todd Sommer, Robert Ashworth,
Blair Folsom, and Todd Melick
Energy and Environmental Research Corporation (EER)
P.O. Box 153, 1345 N. Main Street
Orrville, OH 44667
Joseph Battista
Pennsylvania Electric Company/GPU GENCO
1001 Broad Street
Johnstown, PA 15907
Abstract
The electric utility industry is facing significant challenges for meeting environmental
regulations for its coal fired generation units, and at the same time must stay cost competitive
in an industry that is being deregulated. An emerging technology that can reduce
environmental equipment costs, lower fuel costs, and help to mitigate the environmental
hazard of coal fine impoundments is the recovery and use of coal fines (< 100 mesh) in the
form of a coal-water slurry (CWS). The are a number of issues to be addressed in the
commercialization of this technology: coal fines recovery and beneficiation; CWS
formulation and transportation; and storage, handling, and firing by the combustion
equipment owner. This paper deals primarily with the utilization of coal-water slurry in
utility boilers. Experience on a 32MWe wall-fired boiler at the Pennsylvania Electric
Company (Penelec) Seward Station indicates that CWS can be stored, handled, and co-fired
with pulverized coal with only minimal modifications to the low-NOx coal burners and no
modifications to the boiler. There was no degradation in combustion performance and NOx
emissions were reduced approximately 20% when firing up to 35% of the total boiler heat
input as coal-water slurry. Penelec contracted to purchase 2,500,000 gallons of CWS and
are currently co-firing the fuel as it is produced and delivered. CWS fuel made from coal
cleaning plant waste fines and from coal fines recovered from existing impoundments is
expected to be available for less than $1.00/MMBtu.
-------
Introduction
Emissions from coal-fired central-station power plants continue to play a significant role in
the quality of the environment. The preparation and combustion of coal in a pulverized coal
fired boiler raises several environmental concerns, namely the release of:
• fine coal particles lost from the preparation process
• sulfur dioxide due to pyritic and organic sulfur in the coal,
• nitrogen oxides due to fuel bound nitrogen and to high combustion
temperatures,
• particulate matter due to the ash content of coal and unburned carbon,
• hazardous air pollutants (HAPs) present in the coal.
In general, the higher the ash, sulfur, and nitrogen content of the coal, the higher the
emissions per million Btu of coal fired. The recent revisions to the Clean Air Act
Amendments mandate reductions in NOX and SOX emissions and future legislation is expected
to address reduction requirements for both fine particulate and hazardous air pollutants.
In addition to the emissions from the plant sites themselves, there are concerns with the fine
coal waste streams from coal processing facilities operated to provide fuel to numerous
power plants. Also of significant concern are the two billion tons of coal fines deposited
over the years in pond impoundments of various sizes all over the country.
An innovative approach is needed to apply technology that addresses each concern in a cost
effective, environmentally acceptable manner that may be implemented synergistically. The
approach should be applicable to a variety of coal-fired combustion systems and power
cycles. Further, the commercialization of the integrated technologies should consider both
domestic and international applications.
Conventional washing removes some impurities from coal and helps in reducing sulfur and
particulate emissions. Deep physical cleaning, to remove additional impurities, has not been
commercially practiced because it requires energy intensive grinding of the coal to a very
fine size so that impurities can be liberated. The resulting fine coal product is difficult to
handle and transport. Adding water to these fines to produce a coal water slurry fuel has
been considered impractical due to the cost of fine coal preparation, slurry formulation
additive cost and "potential" boiler impacts.
-------
It is important to note that the coal-water slurry discussed in this paper is significantly
different than the coal-water slurries which were the subject of so much attention in the early
1980's.1"8 The objective of the work at that time was to develop a coal-based fuel that would
displace oil and natural gas in boilers designed to fire only oil or natural gas. Efforts were
focused on forulating slurries with very high solids loading (65-75%). A number of issues
arose as attempts were made to demonstrate the feasibility of the technology:
• slurry formulation required significant amounts of grinding and expensive additives
• slurry rheology very sensitive to coal properties
• slurry characteristics changed during storage
• flame stability difficult to achieve
• difficulty with numerous impacts on boilers due to the presence of coal ash
Eventually, the technology was abandoned due to the unfavorable economics created by the
high costs incurred to address these issues and the relatively low cost differential between
coal and the premium oil and natural gas fuels.
Slurry formulation developments, along with the identification of available fine coal streams,
have eliminated the need for the grinding and the additives. This approach results in CWS
fuels which have lower concentrations of solids (coal), in the range of 50-55%, but are
significantly lower in cost than conventional CWS fuels developed in the past.
To encourage the further development of this technology, Penelec is proceeding with a multi-
phase program to demonstrate the technical and economic viability of using waste coal fines
to reduce operating costs. EER has supplied the technical expertise to store, handle, control,
and use the CWS in their Seward Station Unit #14. Recently, EPRI's Upgraded Coal
Interest Group (UCIG) has provided financial support for additional studies of fine coal
resources, their cleanability, and use. In addition, the UCIG is assisting in the transfer of
this technology throughout the utility industry.
Coal Fines Utilization
The history of coal cleaning practices in the United States involving the recovery of fine
clean coal has always been conspicuously inadequate. Because any cleaning
(demineralization) and inclusion of fines into the main product frequently marginalized or
degraded the market value of the coal due to the poor removal of impurities and/or moisture,
-------
there has always been a tendency to discard all or much of the fine clean coal. Frequently,
the disposal is in the form of a slurry into waste impoundments either directly or as poorly or
partially cleaned refuse. It is not unusual for discarded fine slurry coal to contain more
mineral impurities than the original raw coal and, even if dewatered, to contain an
unmarketable moisture content in the range of 20 to 40 percent. The problem of excess
moisture content is completely mitigated by the direct combustion of the fines in slurry form.
In a commercial coal prep plant, coarse coal, usually plus 28 mesh, is cleaned and sold. The
-28 mesh coal may be recovered in conventional flotation cells or as raw coal through
thickening and filtration circuits. It can then be mixed with the clean product. If fine coal is
to be widely used as a fuel, it must be cleaned and then recovered in a usable form. Coal-
water slurry is an obvious solution to the fines handling problems. If CWS is produced, then
the fuel consumer must be willing to purchase this coal and to modify combustion equipment
to burn the slurry. The incentive for the consumer, primarily the coal burning utility
companies, will be a clean fuel at low cost. In some instances, the CWS will be a source of
compliance coal or low sulfur coal for blending.
The Pennsylvania Electric Company (Penelec) began looking for alternative methods of
utilizing the < 100 Mesh coal fines stream from the Homer City Coal Cleaning Facility
located adjacent to the Homer City Generating Station in Homer City, Pennsylvania. The
current practice involves dewatering the fines stream and further reducing the moisture
content in large, coal-fired dryers. The fines are then added to the remainder of the coal
stream into the plant. The Station experiences a number of storage and handling difficulties,
however, as a result of the presence of those fines. In addition, the coal dryers consume
50,000 tons of coal and are responsible for almost 1500 tons of SO2 emissions each year.
One of the most promising alternatives appeared to be the use of the waste coal fines stream
in its existing form - a coal and water mixture. Subsequent laboratory and pilot scale work
indicated that the mixture could be concentrated to approximately 50% coal by weight
without additives or additional adjustment to the particle size distribution. A series of small
scale combustion tests at Perm State University verified that the mixture could be co-fired
with pulverized coal with minimal changes to the flame characteristics.9 Penelec
management felt that there would be minimal risk in proceeding with a larger scale
demonstration in a small utility boiler.
-------
Application to Utility Boiler
Seward Station Boiler #14 was selected as a candidate host unit for the demonstration. The
unit was already scheduled for a low-NOx burner retrofit and Penelec simply added the CWS
co-firing requirement to the low-NOx firing system specification.
Boiler Description
Penelec's Seward Station Boiler #14 is a B&W Sterling type, front wall fired, dry bottom
furnace that began operation in 1950. The boiler is normally operated at 360,000 Ib/hr
steam at 675 psig and 835 °F. Each boiler superheater is a pendant type with steam
temperature controlled by a spray attemperator. Figure 1 illustrates a side view of the boiler
and shows the principle dimensions. The burners are configured in two elevations of three
burners per row. This 2 by 3 burner array is closely spaced with a horizontal spacing of 6
feet and a vertical spacing between the rows of just 5 feet. Pulverized coal is supplied to the
burners by two modified B&W "EL" pressurized pulverizers. Each pulverizer has a capacity
of 8-10 tons per hour. EER was awarded a contract in September 1993 to supply 6
FlameMastEER™ Low-NOx burners and associated equipment. Three of the burners can
supply up to 40% of the burner heat input as coal-water slurry.
Burner Description
EER designed, fabricated, and delivered 6 Model 4AF FlameMastEER Low-NOx burners in
April 1994. Each burner is rated at 91 MMBtu/hr which enables the unit to achieve full load
operation with one burner out-of-service. The burner throats were increased from 31 to 34
inches in diameter with new refractory throat tiles and tube openings. Each burner was
equipped with a #2 oil lighter. The lower three burners were supplied with air atomized
CWS atomizers that can provide up to 40% heat input to each burner.
A sectional assembly view of the FlameMastEER™ Burner is shown in Figure 2. The key
functional features of the design include:
• Combustion air supply through separate secondary (SA) and tertiary (TA)
passages
• Variable SA/TA flow distribution
-------
• Variable SA and TA swirl
• Low pulverized coal velocity
• Flameholder attached to the coal nozzle
The air distribution between the secondary and tertiary zones, along with the adjustment of
swirl of both flows, represents the operating variables for control of NOX emissions. The
low primary air/coal velocity and flameholder are designed to provide good flame stability
and acceptable flame characteristics for a wide range of operating conditions and fuel
characteristics. The flameholder establishes local recirculation zones and promotes local
mixing between the coal and the secondary air. This leads to a rapid devolatilization of the
coal and liberation of fuel nitrogen in a low excess air environment resulting in reduced NOX
formation.
Mechanically, the burner has been designed to minimize the number of moving parts. Those
parts which do move, slide axially, eliminating complex linkages and gears. The secondary
and tertiary swirl control vanes move within conical passages of the burner. As the swirlers
moved toward the narrow end of the cone more air passes through the vanes increasing the
amount of swirl. As the swirler is moved in the opposite direction, the air follows the path
of least resistance and by-passes the vanes, resulting in less swirl. The amount of
combustion air entering each burner is controlled by a sliding ring damper. Similarly, the
split between secondary and tertiary air is controlled by a second ring damper. During
normal operation, the tertiary air ring damper is actuated automatically while the ring damper
controlling the SA/TA flow split is set during commissioning and is manually operable if
necessary. These moving parts are located farthest away from the high heat flux
environment in order to reduce warpage and binding. The parts of the burner that are
subjected to a high heat flux are fabricated from a high strength heat resistant alloy.
The three CWS-capable burners were supplied with air-atomized slurry atomizers installed
through the secondary air zone. The VEER-jetR was developed by EER for a much smaller
CWS burner designed for retrofit to fire-tube boilers. A non-plugging spray tip that would
provide effective atomization with a low air-to-fuel ratio was required. The tip has been
demonstrated to be very effective during the Seward operation. Not once did the tips require
cleaning during the initial slurry testing, even though the quality of the slurry was sometimes
poor.
-------
Coal fineness and distribution to the burners also affects low-NOx burner performance. Due
to the low volatile content of the fuel normally fired at the Seward Station, HER specified the
coal fineness to be no more than 0.5% >50 mesh and at least 75% <200 mesh. HER also
recommends that, within multiple burner arrays, each burner should be supplied with equal
amounts of fuel and air. EER included an adjustable orifice coal flow balancing damper,
marketed as the PC FlowmastEER™, with each burner. This device, shown in Figure 3,
allows for the adjustment of pulverized coal flow to within +/- 5% between burners. The
FlowmastEER™ dampers were installed just above the pulverizer due to limited space along
the coal piping.
Coal-Water Slurry System Description
The first CWS handling and feed system installed at Seward, shown Figure 4, consisted of an
air operated diaphragm pump, a transfer line, a 4,000 gallon stationary tanker trailer, a
slurry recirculation line fed from another diaphragm pump, and a high pressure slurry
positive displacement Moyno pump. The system was designed and constructed to be a
temporary installation to enable the CWS test program to be conducted. Due to the success
of the CWS demonstration, Penelec later installed a permanent coal water slurry system that
is similar to the first system, but now utilizes two 9,000 gallon tanks with agitators. This
system is being used to co-fire 2,500,000 gallons of slurry over a six-month period which
began in March of this year.
Instrumentation in the piping system consists of pressure gauges before and after the
strainers, a pressure switch to protect the slurry pump from overpressure, local pressure
gauges at the atomizers, low pressure switches on the atomizing air, a main fuel trip valve,
automatic shutoff valves for the slurry and water purge system, a pressure transmitter for the
air header, and an orifice plate and transmitter for the atomizing air flow. The slurry flow
was determined from a mass flow meter that also provides totalization, density, and
temperature.
A Bailey INFI-90 control system, installed simultaneously with the new burners, includes
both Burner Management and Combustion Control system (CCS) functions. The INFI-90 is
a distributed control system that provides integrated modulating control, sequential control
and the capability of the operator to monitor and interface with the various instruments and
devices associated with the Low NOX Burner and coal-water slurry systems.
-------
The control of most plant systems remains unchanged when switching to CWS co-firing.
The CCS calculates a Btu-corrected slurry flow to compensate for changes in slurry density
during firing. The approximate heating value is then subtracted from the Fuel Master
demand signal controlling the pulverizer coal feeder, reducing the pulverized coal heat input
by an equal amount. This control scheme prevents over-firing of the boiler when co-firing
CWS and, at the same time, uses the slurry pump to maintain a constant heat input instead of
modulating the pulverizer feeder to make up for changes in slurry heating value. This
resulted in a more stable boiler operation due to a more constant fuel input. Primary air
flow remains unchanged during CWS co-firing.
The slurry atomizing air system operates at a constant pressure and does not have a
modulating control loop. Atomizing air flow and pressure are measured and transmitted to
the CCS for display on the operator screen. Control of the various slurry valves and flush
water valves are handled by the Burner Management System, to ensure proper sequential
operation. Typical operation of the atomizers is with 0.15-0.25 Ib atomizing air/1.0 Ib of
slurry.
Operating Results
The testing conducted at the Seward Station was focused on identifying any potential
operating problems which could be attributed to the low-NOx burners, due to their CWS
firing capability, or to the slurry fuel itself.
Low-NOx Burner Performance
Prior to any CWS co-firing, the low-NOx burners were optimized and guarantee performance
tests were completed. The coal flow to each burner was balanced using the International
Standards Organization (ISO) technique using a Rotorprobe™ sampling instrument. Initial
measurements on Boiler #14 "A" pulverizer (lower burner elevation) indicated that one
burner was receiving 17% more coal than the other two. After 3 iterations of
FlowmastEER™ damper adjustments, the coal flow balance was measured to be within 3 %.
HER also rechecked the "A" pulverizer two months later and the coal flow balance was
within 4%.
-------
Prior to the burner retrofit, the average Boiler #14 baseline NOX emissions were 0.90
Ib/MMBtu with coal volatile and nitrogen contents of 25% and 1.3%, respectively. The
FlameMastEER™ Burner emission guarantees were 0.5 Ib/MMBtu NOX, less than 100 ppm
CO with 15-20% excess air in the furnace. At the same time, carbon in the fly ash was not
to increase above baseline levels. Requirements of the guarantee were that the coal volatile
content would not be less than 24% and that the nitrogen content would not exceed 1.3%.
Results of the guarantee tests indicated that all guarantees were satisfied with a post-retrofit
NOX level of 0.47 Ib/MMBtu.
CWS Co-firing Performance
The primary objective of the CWS test program was to evaluate the feasibility of co-firing
low solids (40-50% by weight) CWS with pulverized coal in the Seward boiler. The results
of the study were necessary to determine if a large scale CWS co-firing demonstration
project is feasible. Penelec's ultimate objective is to utilize the coal fines stream from the
cleaning plant in Units #1 and #2 (660 MWe wall-fired boilers) at the Homer City Station.
At the completion of the initial test program approximately 125,000 gallons of CWS were
produced at the Homer City cleaning facility, trucked to Seward Station, and burned in
Boiler #14. A total of approximately 150 hours of operation were accumulated on the system
components.
The test program was conducted in such a way that direct comparisons between boiler
operation, with and without slurry, were possible. A typical test burn involved three hours
of data collection firing coal under steady operating conditions to determine the baseline
performance. That was followed by three hours of testing with slurry co-firing. This
procedure was designed to compensate for the normal variability which occurs in the
operation of Boiler #14. Bunkered coal was used as the base fuel and consisted of the
normal Seward coal supply.
The boiler control system measures and records all pertinent boiler operating data which is
then down loaded onto a diskette at the end of each test. Continuous emission monitors were
utilized as the primary data collection method for SO2, NOX, CO, and O2. Flue gas samples
were extracted just upstream of the air heater. At selected times, manual isokinetic samples
were taken for particulate, SO2, and NOX using standard EPA methods.
-------
Initial testing evaluated the CWS firing system with respect to materials handling and CWS
flow characteristics. The boiler functioned in a normal manner without any boiler trips, mill
trips, opacity excursions, or loss of flame. Although a very cautious test plan had been
proposed, within one hour of beginning the first co-firing test, CWS was successfully burned
in all three burners.
After the initial demonstration that the CWS fuel could be fired safely and without causing
boiler upsets, parametric tests were conducted to evaluate the effect of slurry input, excess
air level, atomizer operating conditions, and burner variables. With the majority of the
testing conducted with less than 35% heat input from CWS, the burner and boiler
performance is still dominated by the pulverized coal combustion.
One noticeable impact was the reduction of NOX emissions during CWS co-firing. Figure 5
illustrates the emissions during boiler operation with and without slurry co-firing. The data
points shown include a range of CWS heat input conditions from 10-35%. It was apparent
that the NOX reduction was not limited to reductions caused by the water in the slurry
stream. At the conclusion of each test, the slurry tank was cleaned with water which was
subsequently injected into the boiler. For an equivalent amount of water, the NOX emission
level was always lower during slurry co-firing. The reader should be aware that EER has
demonstrated on pilot scale facilities that CWS can be a very effective reburn fuel. For
equivalent levels of CWS firing demonstrated at Penelec, significantly higher reductions
(>50%) of NOx emissions could be achieved by utilizing the CWS as a reburn fuel.10
The only operating problem encountered during the test program was with the slurry
stability. Several slurry batches were delivered with solids contents less than 40%. It was
discovered that solids were settling in the storage tank at the production facility. Agitation in
the storage tank solved the problem. Mining labor negotiations resulted in coal supplies with
significant variability during the test period also causing some problems in the production
process. In spite of the low solids content and variability of the CWS quality, the slurry was
burned at Seward without difficulty.
Throughout the testing, boiler performance and operations were unaffected by co-firing
CWS. Steam temperature, flow, and pressure, opacity, attemperation flow, and other
performance indicators were essentially constant. Boiler efficiency dropped by 1-2% as
expected from the losses caused by the water being injected. Carbon loss in the flyash
increased by a small amount when co-firing the slurry. It appears that this is due to the
coarser size distribution of the CWS compared to the pulverized coal that it displaces. The
-------
CWS fired was typically 3-5% >50 mesh material. This issue could be addressed in the
production process as necessary.
Penelec determined from the results of this performance testing that, at least for relatively
short-term tests, coal-water slurry co-firing was a viable alternative for reducing their cost of
power generation. The next step in the demonstration program is to verify that same
conclusion when CWS fuel is co-fired on a continuous basis for periods of at least one week.
Penelec contracted EER to modify the remaining three low-NOx burners on Unit #14 for
CWS co-firing capability. Penelec has also awarded a CWS supply contract to Washington
Energy Processing, Coraopolis, PA to deliver up to 2,500,000 gallons of coal slurry to the
Seward Station over a three to six-month period. The fuel is currently being fired in all
burners on Unit #14. The results of the long-term testing will be reported at future symposia
as they become available.
Economics
The main reason that coal is the primary fuel used for production of electricity worldwide is
its low cost. Any process for removal of pollutants, whether coal cleaning or post
combustion scrubbing, will add to the cost of producing electricity. This extra cost, which
will be passed on to consumers, has to be justified based on the environmental regulation
requirements. The pollution control steps or coal cleaning processes will usually increase the
cost of coal. Assuming that coal would still be used, and the choice has to be made between
the coal cleaning and post combustion scrubbing, the pre-combustion cleaning, in the long
term, may provide a more economical alternative.
EER has performed several preliminary evaluations of applying CWS fuel to utility boilers.
In addition, both Penelec and EER have worked closely with Mr. Ed Zawadzki, a fuel
technology consultant, and the EPRI UCIG to evaluate the potential for recovering and
utilizing coal fines. Based on this experience, the authors believe that there is considerable
potential for coal-water slurry fuels to be produced in sufficient quantities for utility
application at less than $1.00/MMBtu.
The fixed cost to retrofit the slurry co-firing equipment on the very small Penelec unit was
approximately $25/kw. EER has estimated the cost for CWS co-firing capability on larger
units (500-700MWe) to be in the range of $5/kw.
-------
Variable costs are dominated by the fuel cost differential between the CWS fuel and the coal
it displaces. The costs to operate the atomizing air compressors and the reduction in boiler
efficiency are the only significant operating costs to consider. No additional operations
personnel are required. Normal maintenance of the slurry handling equipment is required.
HER also completed another evaluation in the past year looking at the benefit of combining
advanced cleaning methods with the coal slurry technology. Based on a cursory evaluation
of the individual characteristics of the coal cleaning techniques and the slurry co-
firing/reburning methods to be investigated, the reduction of pollutants; NOX, SOX, HAPs,
and PM10 were estimated. Table 1 summarizes the technical and economic results of this
evaluation. With the use of advanced physical cleaning of coal to make a slurry for use as a
co-firing fuel, there would be a net savings rather than a cost for NOX reduction. Also, the
chemical cleaning of coal may look relatively expensive, but under the current Federal
government tax codes, IRS §29, a tax credit of $0.97/MM Bru may be obtained, which when
applied also yields a payback for the chemical cleaning of coal, not a cost.
TABLE 1. ESTIMATED POLLUTANT REDUCTIONS AND COST
Case:
Pollutant:
Baseline
Physically Clean CWS Co-Firing
Physically Clean CWS Reburning
Adv Phys Clean CWS Reburning
Adv Phys + Chem Clean CWS
Reburn
Adv Phys + Chem Clean CWS
Rebum
(w/IRS §29 tax credit)
Percent of Baseline
NOX
100
70
40
40
30
30
sox
100
100
100
88
81
81
HAP
s
100
100
100
88
80
80
PM10
100
100
100
88
85
85
Cost, $/ton
NOX
-
(420)
52
52
44
44
SOX
-
-
326
531
(217)
Total
-
(420)
52
378
575
(173)
-------
Note: The analysis is based on applying the technologies to a 700 MWe wall-fired boiler that yields a baseline
emission rate of 0.8 Ib NO,/MM Btu of coal fired. The baseline coal for the slurry is assumed to contain 10%
ash, 4 % sulfur with a 50-50 mix of pyritic and organic sulfur, and a heat input of coal water slurry of 25 %.
HAPs reduction is shown as mercury; other heavy metals such as selenium, cobalt, arsenic etc. will also be
reduced. The physical and chemical cleaning applies only to the CWS used for returning. In the physical plus
chemical cleaning case, all of the added cost was placed as a cost of SO2 reduction.
Conclusions
There has been a considerable amount of work completed over the past several years to
evaluate the potential for utilizing coal fines as a low cost boiler fuel. Significant quantities
of fines are produced and disposed of every day by operating coal cleaning plants. In
addition, large quantities of fines are stored in impoundments throughout the country. Coal
fines can be recovered economically from both sources, converted to a coal-water slurry fuel,
and can be made available at a very attractive price for supplementing existing fuel supplies.
Pennsylvania Electric Company has proceeded with a demonstration of the utilization of a
relatively low solids content (50%) coal-water slurry fuel on a small utility boiler located at
their Seward Station. Low-NOx burners, some with CWS firing capability, were retrofit and
a slurry fuel handling system installed to evaluate the technology. The 32 MWe boiler has
fired several hundred thousand gallons of CWS fuel without any detrimental boiler impacts
whatsoever. In addition, during CWS co-firing tests, NOX emissions have been reduced
approximately 20%. Penelec has contracted for an additional 2,500,000 gallons of coal-
water slurry and are continuing their evaluation with long-term slurry firing currently in
progress.
HER is prepared to offer complete coal-water slurry firing systems, with performance
guarantees, for retrofit to most utility and large industrial boilers.
Acknowledgements
The authors would like to acknowledge the Penelec Seward Station operations and
maintenance staff and Mr. John Kline, Station Engineer, whose cooperation and support
made the introduction of this new fuel a relatively easy task.
-------
References
1. T.M. Sommer, et al, "Development of a High Solids Coal-Water Mixture for Application
as a Boiler Fuel", Presented at the ASME Joint Power Conference, St. Louis, MO.
(September, 1981).
2. T.M. Sommer, et al, "Rheology and Combustion Characteristics of Coal-Water
Mixtures". Presented at the American Flame Research Committee meeting, Chicago, IL.
(October, 1981).
3. T.M. Sommer, et al, "Commercialization of Coal-Water Slurries", Presented at the
Industrial Coal Utilization Conference, Cincinnati, OH (April, 1982).
4. G.A. Farthing, T.M. Sommer, S.J. Vecci, "Rheology and Combustion Characteristics of
Coal-Water Mixtures", Presented at the Fourth International Symposium on Coal Slurry
Combustion, Orlando, FL (May, 1982).
5. R. Payne, S.L. Chen, W. Richter, "A Procedure for the Evaluation of the Combustion
Performance of Coal-Water Slurries", Presented at the Fifth International Symposium on
Coal Slurry Combustion, Tampa, FL (April, 1983).
6. B.E. Taylor, R.P. Perkins, R.K. Manfred, "EPRI Industrial Coal-Water Slurry
Demonstration", Presented at the Sixth International Symposium on Coal Slurry Combustion,
Orlando, FL (June, 1984).
7. G. Haider, G.A. Clark, A.E. Margulies, "Potential of Micronized Coal-Water Slurry as
an Alternate Fuel in Oil- and Gas-Fired Boilers", Presented at the Seventh International
Symposium on Coal Slurry Fuels Preparation and Utilization, New Orleans, LA (May,
1985).
8. T.M. Sommer, B.E. Taylor, J. Wagner, S.A. Johnson, "Coal-Water Mixture Firing in an
Industrial Package Boiler - A User's Perspective". Presented at the Seventh International
Symposium on Coal Slurry Fuels Preparation and Utilization, New Orleans, LA (May,
1985).
-------
9. B.A. Miller, A.W. Scaroni, J.T. Elston, "An Update on Penn State's Coal-Water Slurry
Demonstration Program", Presented at the Sixteenth International Conference on Coal and
Slurry Technologies, Clearwater, FL (April, 1991).
10. R. Payne, "The Use of Pulverized Coal and Coal-Water Slurry in Reburning NOx
Control", Presented at the Joint Symposium on Stationary Combustion NOx Control, Kansas
City, MO (May, 1995).
-------
137XT
8V-3'
5'-0'
Figure 1. Seward Boiler 14 dimensions
-------
Figure 2. FlamemastEER™ Burner
-------
FULLY OPEN
100% FLOW AREA-NO RESTRICTION
FULLY CLOSED
10% FLOW AREA-MAXIMUM CONTROL
• '•'
Figure 3. PC FlowmastEER™.
-------
RECIRCULATION
TRANSFER PUMP
FROM _
TANKER
MOYNO PUMP
'
j]
*RY1
UMP
r°
LO
•LEX
1 i
)
T
TANK
1
J
STRAINER
7,
-a
ATOMIZING AIR
CWS TO BOTTOM
ROW OF BURNERS
Figure 4. Seward Coal Water Slurry System
-------
0.65
0.6 -
0.55 -
O 0.5
z
0.45 -
0.4
o o o o o o~ oo
5.6
5.8
6.2 6.4 6.6 6.8 7
Excess air (O2 percent at air heater inlet)
7.2
7.4
7.6
Figure 5. NOx Emissions Impact
-------
IN-FURNACE, RETROFIT ULTRA-LOW NOX CONTROL
TECHNOLOGY FOR TANGENTIAL, COAL-FIRED BOILERS:
THE ABB C-E SERVICESTFS 2000™R SYSTEM
T. Buffa
D. Marti
United Illuminating
R. C. LaFlesh
ABB C-E Services, Inc.
Abstract
United Illuminating and ABB C-E Services, Inc. report the first commercial retrofit
installation and performance results from a TFS2000™R firing system. Pre-retrofit and
post-retrofit field trials were conducted to evaluate the impact of the retrofit design on
the boiler emissions and thermal performance. During testing, the retrofitted 390-MWe
utility boiler demonstrated NOx emissions on the order of 0.25 lb/10" Btu, while firing
Eastern bituminous coal over the entire load range, without increase in unburned
carbon (UBC). A potential minimum NOx emission level of 0.16 lb/10^ Btu was
achieved in parametric testing. The effects of the retrofit on boiler emissions, thermal
performance and operating experience are reported.
Introduction
United Illuminating (UI) provides electricity to south-central Connecticut. In 1984, the
electricity produced in the UI system came from an energy mix that was 94% fuel oil
and 6% nuclear. To diversify its fuel base, in that year UI reconverted the Bridgeport
Harbor Station Unit 3 (Figure 1) for coal firing. By 1985, the contribution of oil to UI's
energy mix was reduced to 53%; nuclear was 9%, and coal had provided 37%.
Continuing with its strategy of utilizing diverse fuels, UI shifted its energy mix to 1%
natural gas, 5% hydro, 8% trash-to-energy, 17% oil, 35 % nuclear, and 34% coal by 1992.1
The city of Bridgeport is located in a "Severe" ozone nonattainment area under the 1990
Clean Air Act Amendments (CAAA) Title I. Bridgeport Harbor Station Unit 3 (BHS
Unit 3) is a Phase II unit under CAAA Title IV. The State of Connecticut's Reasonably
Achievable Control Technology (RACT) NOx limitation is 0.38 lb/106 Btu for tangential
coal-fired boilers. With UI's fuel strategy in place, the utility decided to retrofit BHS
Unit 3, its only coal-burning unit, with an aggressive low NOx firing system.
ABB C-E Services invited UI to participate in a research and development project in
which BHS Unit 3 would serve as the first commercial field demonstration of TFS
-------
2000™R technology. Similar technology
had previously demonstrated ultra-low
NOx emissions at the laboratory scale.2
Unit Description
BHS Unit 3 is a Combustion
Engineering, Inc., Controlled
Circulation® steam generator with
radiant reheat cycle and a pressurized
furnace (Figure 2). It was designed in
1965 and commissioned in 1968. The
steam generator is rated at 2,700,000
Ib/hr primary steam flow at maximum
continuous rating (MCR), with a
corresponding reheat flow of 2,387,000
Ib/hr. The MCR design superheat and
reheat outlet steam temperatures are
1005 F. Operating pressure at the
superheater outlet is 2629 psig.
Nominally rated at 390 MWe, the unit
was equipped with a Tilting Tangential
Firing System for firing pulverized coal
from five elevations and oil from four
elevations. During the reconversion to
coal firing in 1984, close-coupled
overfire air was added. BHS Unit 3
operates with Eastern U.S. bituminous
coals from sources in Kentucky. The coal composition is
relatively uniform, with a low sulfur content and low
slagging/fouling potential. Table 1 shows a typical coal
analysis for BHS Unit 3.
BHS Unit 3 is typically operated on automatic load
dispatch, generating steam at MCR on weekdays and at
control load or lower on nights and weekends. Pre-
retrofit NOx emissions under normal operating
conditions were in the range of 0.55-0.60 Ib NOx/106 Btu.
The unit had no history of significant slagging or fouling,
and no history of pressure part failures related to the coal
properties.
TFS 2000™R SYSTEM DESIGN
Figure 1: United Illuminating's Bridgeport
Harbor Station
Moisture
Volatile Matter
Fixed Carbon
Ash
Nitrogen
Sulfur
FC/VM
HHV (Btu/lb)
Hardgrove Index
5.4%
30.1%
57.7%
6.8%
1.4%
0.7%
1.92
13,400
45
Table 1: Typical Coal Analysis
The TFS 2000™R System at BHS Unit 3 is an integrated retrofit design based on the
successful laboratory development of Combustion Engineering, Inc.'s (ABB C-E) TFS
-------
Figure 2: Bridgeport Harbor Station Unit 3, Pre-Retrofit Side Elevation
2000™ system for new boilers.2 The challenge is to provide the most aggressive control
of NOx emissions possible within the constraints of a fixed furnace geometry, without
introducing any radical or negative departures from either design or operating
-------
practices. Previous research and development efforts suggested that the laboratory
results for absolute NOx emissions, and trends for carbon monoxide and unburned
carbon, were consistent with a utility boiler.3 Therefore, the next step in the
commercialization of the TFS 2000™R technology was a field demonstration on a large
utility boiler.
The basic design philosophy of the TFS 2000™R firing system is based on the
integration of four major principles:
1. Firing zone stoichiometry control
2. Pulverized coal fineness control
3. Initial combustion process control
4. Concentric firing
Laboratory testing has indicated that there is an optimum main firing zone
stoichiometry for minimizing NOx emissions.2 However, achieving this level of
stoichiometry can result in high levels of CO and UBC. The TFS 2000™R system
(Figure 3) controls the process of NOx formation and destruction in distinct regions of
the furnace by "staging" the introduction of air through flame attachment coal nozzle
tips and multiple levels of separated overfire air (SOFA) and close-coupled overfire air
(CCOFA). The TFS 2000™R system thereby optimizes the entire stoichiometry history
of the coal particles, to minimize NOx emissions.
Multi-Level
Separated
Overfire Air
Close-Coupled
Overfire Air
CFS™ Air
Nozzle Tips
Flame Attachment
Coal Nozzle Tips
Pulverizer with
Dynamic Classifier
Figure 3: Schematic Diagram of a TFS 2000R Firing System
-------
Pulverized coal fineness is controlled by use of a Dynamic™ classifier. The rotating
classifier vanes more effectively prevent larger coal particles from exiting the
pulverizer, and this helps decrease the UBC levels in the flyash. Finer coal particles can
also enhance fuel-bound nitrogen conversion and its subsequent reduction to
molecular nitrogen under staged firing conditions by allowing rapid ignition near the
coal nozzle tip.
Flame attachment coal nozzle tips are incorporated in the TFS 2000™R system design to
provide early fuel devolatilization within an oxygen-deficient zone. With conventional
firing systems, coal is devolatilized in an oxygen-rich environment, and the fuel
nitrogen released can readily react with the available oxygen to form nitrogen oxide
compounds. With the flame attachment coal nozzle tip, rapid coal devolatilization is
accomplished by establishing a flame front near the exit of the tip. The coal nozzle tip
design is based on existing flame characteristics, coal constituents, and fuel line
transport conditions. Besides the NOx emissions control benefits, establishing coal
ignition early in the combustion process improves flame stability and minimizes
increases in unburned coal levels.
ABB's patented CFS™ concentric firing system air nozzle tips direct some of the
secondary air in the main firing zone away from the fuel streams. Offsetting the air
decreases the local firing zone stoichiometry during the initial combustion stages.
Concentric firing also creates an oxidizing environment near the furnace waterwalls in
and above the main firing zone. This reduces ash deposition quantity and tenacity.
Increased oxygen levels along the waterwalls also decreases the potential for corrosion,
especially with coals having high concentrations of sulfur, iron, or alkali metals.
The specific equipment components selected to achieve these elements of combustion
will vary for different retrofit installations, depending on the design and maintenance
condition of the installed equipment, and on the constructability constraints at the site.
TFS 2000™R System Implementation
The retrofit equipment described below for the field demonstration of TFS 2000™R
technology at BHS Unit 3 was installed in the Fall of 1993. The installation coincided
with a scheduled maintenance outage for the turbine-generator. The outage duration
was 8.5 weeks.
Windboxes
Because the existing main windboxes at BHS Unit 3 were in a deteriorated condition
and the planned outage duration was short, the main windboxes were completely
replaced with new, pre-assembled units. Each new main windbox (Figure 4) contains
one bottom air compartment, four elevations of air/oil compartments with CFS™ air
nozzle tips above and below the oil gun tips, two elevations of CCOFA compartments,
and five elevations of coal compartments with flame attachment coal nozzle tips. New
-------
tilt mechanisms were provided at the
compartments, re-using existing tilt drives.
Secondary air flow to the windbox air registers is
controlled by means of louver dampers equipped
with self-lubricating damper bearing assemblies.
With ABB's flame attachment coal nozzle tips, the
ignition point of the coal occurs closer to the
nozzle tip than it does for conventional coal nozzle
tips. The rapid fuel ignition produces a stable
volatile matter flame and minimizes NOx
production in the fuel-rich stream.
The CFS™ air nozzle tips supplied at BHS Unit 3
are equipped with manually-adjustable horizontal
yaw mechanisms. The yaw adjustment is set so
that a portion of the secondary air is directed away
from the fuel streams toward an imaginary circle
that is concentric with the main firing circle. The
yaw angle is set during commissioning and is not
changed during normal operation of the boiler.
The CCOFA elevation air registers direct a portion
of the secondary air into the furnace at the top of
the main windboxes. Each CCOFA compartment
is equipped with ABB's patented horizontal yaw
adjustment mechanism. The manual yaw
adjustment enables each CCOFA air jet to be
independently directed for effective mixing.
TT
Q
TT
CCOFA
CCOFA
Coal
CFS
Oil
CFS
Coal
CFS
Oil
CFS
Coal
CFS
Oil
CFS
Coal
CFS
Oil
CFS
Coal
Air
Figure 4: Schematic Diagram of TFS
2000R Windboxes at BHS Unit 3
Two new SOFA registers were added above each
of the new main windboxes. Each SOFA register
contains three air compartments with adjustable
horizontal yaw and vertical tilt mechanisms
(Figure 5). During commissioning, the yaw angle
is set to minimize carbon monoxide and UBC emissions. This is a manual adjustment
that is not intended to be varied during operation.
To measure the SOFA air flow, an annular venturi (Figure 6) was installed in each SOFA
air supply duct. ABB's patented annular venturi design requires only about two-thirds
the length of a standard venturi and measures air flow with an accuracy of ±5 percent.
It has a signal-to-noise ratio of approximately 10. Annular venturi are not required
components for a TFS 2000™R system retrofit.
-------
Figure 5: New SOFA Register During
Installation
Figure 6: Annular Venturi for SOFA Ductwork
in Laydown Area
Pulverizer Modifications
Pulverizer modifications to implement TFS
2000™R technology are also site-specific,
and depend greatly on the condition of the
existing pulverizers, as well as the coal to be
fired after the retrofit. BHS Unit 3's five
pulverizers were well-maintained and in
good operating condition prior to the
retrofit. The pulverizers were upgraded to
permit operation at higher fineness levels
without coal flow de-rating. The existing
"spider" fan wheels were replaced by new
high efficiency fans (HEF) utilizing the
existing exhauster casings. In addition, the
existing 600-Hp pulverizer motors were
replaced with new 700-Hp motors. Figure 7
shows one of the new HEF wheels.
In each pulverizer, a new Dynamic™
classifier replaced the existing static
classifier. The Dynamic™ classifier has a
vaned rotor that is supported by two
bearings. It is driven by a 40-Hp motor,
and the speed of rotation is controlled
through an ac variable-speed controller.
Figure 8 is a photograph of one of the
pulverizers during the installation of the
Dynamic™ classifier. The Dynamic™
classifier effectively eliminates large coal
particles (+50-mesh or +70-mesh) and
minimizes the fraction of +100-mesh coal
particles. It allows extensive operational
flexibility, and can be used to compensate
for the effects of pulverizer wear, load
changes, and changes in coal type or
grindability.
Additional Work
Pressure part replacements requiring four
main windbox tube panels and four SOFA
tube panels accompanied the new
windboxes and SOFA registers. Additional
pressure part modifications were made at
BHS Unit 3 to eliminate interferences with
the SOFA register installation.
-------
Figure 7: New HEF Wheel in the Existing
Exhauster Casing
Figure 8: New Dynamic™ Classifier
During Installation
As part of the research and development project, 39 waterwall chordal thermocouples
and 135 convective section thermocouples were installed to provide accurate and
convenient measurements of the boiler's thermal performance under load. In addition,
six waterwall test panels were installed to investigate industry concerns regarding long-
term waterwall tube wastage under substoichiometric firing conditions. These panels
were fabricated of new waterwall tubing and were subjected to ultrasonic thickness
measurement prior to installation. Tubing thickness will be regularly monitored during
future maintenance outages. Figure 9 shows the approximate locations of this test
equipment.
Control system inputs/outputs and logic were added for operation of SOFA dampers
and Dynamic™ classifiers, and to expand the operational flexibility of all windbox
dampers. In addition, UI elected to perform additional back pass modifications, to
upgrade the DCS control system and to add continuous stack emissions monitors and
stack elevator during the outage. These modifications were not required for the new
firing system.
-------
135 Convective Section Thermocouples
Corrosion
Monitoring
Panel
(6 total)
Waterwall
Chordal
Thermocouple
(39 total)
Rear Wall
Right Wall Front Wall
ten Wall
Figure 9: Locations of Test Thermocouples
and Test Panels
NOx Emissions
TFS 2000™R System Performance
Evaluation
Pre-retrofit and post-retrofit field
trials were conducted to evaluate the
impact of the new design on the boiler
emissions and thermal performance.
The focus of the field trials was to
quantify the impact of the new firing
system over the full operating range
of the boiler.
Boiler Emissions Performance
The boiler emissions performance was
characterized through a series of
parametric tests during which certain
operational parameters were varied in
a systematic fashion for several
scenarios of boiler load, staged firing,
and secondary air biasing.
Except where noted, all NOx measurements in this paper were determined via EPA
Method 7E, using a chemiluminescent NOx analyzer, and are reported in units of
Ib NOx/10" Btu. Figure 10 shows the relationship of the measured NOx emissions
from BHS Unit 3 to the calculated stoichiometry at the top coal elevation for both the
pre-retrofit and post-retrofit configurations of the boiler. All measurements were taken
at MCR. The characteristic decrease in NOx emissions with decreasing stoichiometry is
evident. Pre-retrofit NOx testing with the use of CCOFA showed NOx levels in the
range of 0.46 - 0.58 Ib NOx/106 Btu.
Sixty-six post-retrofit tests were conducted while varying the coal fineness and the
degree of staging and mixing, along with a number of operating variables such as
excess air. Post-retrofit NOx emissions as low as 0.20 Ib NOx/10" Btu were achieved
with no increase in the UBC in the flyash.
The two data points labeled "Potential Minimum NOx" (0.18 and 0.16 Ib NOx/106 Btu)
represent short-term (approximately 3 hours) test results. These results were achieved
with carbon monoxide emissions less than 200 ppm and only a two-percentage point
increase in UBC emissions over the pre-retrofit level. It is significant that the potential
minimum NOx results were achieved at a higher stoichiometry than many of the higher post-
retrofit testing results, demonstrating that stoichiometry is not the only variable affecting NOx
emissions.
The post-retrofit test NOx emissions as a function of boiler load are shown in Figure 11.
The secondary air dampers and tilts were controlled to operate the boiler with NOx
-------
0.60
0.40
0.30
0.20
0.10
Pre-Retrofit
A
A
Post-Retrofit Testing •
Potential
Minimum NOx
Stoichiometry at Top Coal Elevation
Figure 10: NOx Emissions vs. Stoichiometry at MCR
s
5
«•.
NOx (lb/10
V.O^J
0.30
0.25
0.20
0.15
0.10
0.05
• Post-Retrofit Testing
•
• . «
*
*
Potential
Minimum NOx
I I I
CL
MCR
Boiler Load (MW)
Figure 11: NOx Emissions vs. Boiler Load
For 14 Units Firing Eastern Bit Coal
0.00
Pre-Retrofit LNCFS LNCFS TFS2000R TFS 2000 R
Average Level I Level III Post-Retrofit Potential
Testing Minimum
Figure 12: Comparison of ABB Retrofit Results
for NOx Emissions
emissions on the order of 0.25 Ib
NOx/106 Btu from MCR through
control load (CL), to minimum load,
with no increase in UBC in the flyash.
Although it is typically expected that
NOx levels will increase dramatically
at low boiler loads because of the
required increase in excess air, at BHS
Unit 3, the post-retrofit NOx emission
at minimum load can be controlled to
less than 0.30 lb/106 Btu.
Figure 12 compares the BHS Unit 3
post-retrofit testing for NOx emissions
to other low NOx retrofit results for
similar coals in tangentially-fired
boilers. The pre-retrofit average NOx
emissions of 0.62 lb/106 Btu for 14
other units firing Eastern bituminous
coals is shown in the first (left) bar.
ABB C-E Services' LNCFS™ firing
systems were applied in these units.4
As shown in Figure 12, LNCFS™
system field results reached a lower
limit for NOx emissions at an average
of 0.36 lb/106 Btu. The BHS Unit 3
field demonstration test results for
NOx emissions are significantly lower.
Limited testing was performed to
evaluate firing system performance
using No. 6 fuel oil, currently used as
an emergency backup fuel to coal at
BHS Unit 3. These brief tests, under
non-optimized conditions, indicated
that NOx levels on the order of 0.12
lb/106 Btu to 0.15 lb/106 Btu can be
obtained from control load through
365 MW, with opacity in the range of
2-7 percent.
Carbon Monoxide Emissions
All carbon monoxide (CO)
measurements reported in this paper
are given in units of parts per million
-------
(ppm) of gas and are corrected to 3% oxygen in the flue gas. The test protocols used are
in accordance with EPA Method 10. Pre-retrofit CO emissions were less than 50 ppm.
During the post-retrofit testing the SOFA yaw angles were varied to demonstrate the
variation of CO emissions with NOx. During the tests documented in Figure 10, at full
load, CO levels of 44 ppm were obtained at NOx emissions of 0.34 lb/10" Bru; CO
emissions of 22 ppm occurred with NOx emissions of 0.24 lb/10^ Btu; and CO
emissions of 178 ppm were found with NOx emissions of 0.16 lb/10^ Btu.
Opacity
Opacity measurements were taken with the plant instrumentation. At BHS Unit 3, the
regulated opacity limit is 20%. The pre-retrofit opacity averaged less than 10%. During
the post-retrofit testing, the opacity remained less than 10% for most tests, and below
the regulated limit under all test conditions. Isokinetic sampling of the flue gas
entering the unit's electrostatic precipitator (ESP) confirmed that there was no
significant change in the flyash (dust) loading entering the ESP. No significant change
in the mass ratio of fly ash-to-bottom ash was observed.
ESP collection efficiency prior to TFS 2000™R retrofit was 99.3 percent. During the
outage for the TFS 2000™R retrofit, routine maintenance was performed on the ESP.
Measured post-retrofit ESP efficiency is on the order of 99.2-99.5 percent.
Boiler Operational Performance
During post-retrofit testing on the BHS Unit 3 boiler, multiple aspects of boiler
operation were investigated to ensure that there were no adverse impacts on boiler
operation related to the changes in the firing system.
Ash and Slag Deposition Patterns
A long-term change in the ash and slag deposition during operation was noted. Post-
retrofit ash deposition has increased in the superheater sections closest to the furnace
outlet, the superheater division panels and superheater platen assemblies (Figure 2).
These ash deposits are friable and easily removed. No other significant changes in ash
accumulation have been observed in the convective sections of the boiler. Slagging has
decreased on about one-third of the furnace wall, in the areas near the CFS™ air
elevations. Although the ash and slag deposition patterns have changed, they are
controllable with the existing sootblowers and wall blowers on the boiler.
The boiler had no history of waterwall corrosion before the retrofit. After
approximately 14 months of post-retrofit operation, no evidence of accelerated
waterwall wastage has been observed.
Coal Fineness
Calibration runs for the Dynamic™ classifier with the "B" pulverizer established the
relationships among coal feed rate, fineness, and classifier rotation speed. Generally, a
-------
I Static
Static (Max)
80 rpm
90 rpm
+50
+70
+100
-200
Figure 13: Comparison of Static and Dynamic Classifier Fineness Results
higher classifier rpm produces greater fineness, and rpm can be decreased as coal feed
rates are decreased. At all coal feed rates, the coal fineness achievable with the
Dynamic™ classifier is finer than with the static classifier, particularly in terms of
decreasing or eliminating the largest +50 and +70-mesh particles. Coal particles in
these size ranges have significant impact on UBC. Figure 13 compares the performance
of the static classifier and the Dynamic™ classifier at BHS Unit 3 with five pulverizers,
each in service at 55,000 Ib coal/h.
Pulverizer performance has met expectations, with the exception of a "rumble"
condition that occurred during testing at high classifier rotation speeds. High fineness
"rumble" can occur with either dynamic or static classifiers on a high-fineness setting.
High fineness "rumble" is an instability, leading to vibrations, that is caused by an
increase in recirculation of fine particles. At BHS Unit 3, the Dynamic™ classifier
rotational speed is currently limited to avoid high fineness "rumble". A shady is in
progress at the ABB Power Plant Laboratories Pulverizer Development Facility in
Windsor, Conn., to develop a methodology for predicting/preventing the onset of high
fineness "rumble".5
Furnace Oxygen Imbalance
The oxygen concentration in the flue gas was measured at the economizer outlet in
accordance with EPA Method 3A. Post-retrofit left/right oxygen imbalance is less than
or equal to the pre-retrofit performance.
Boiler Thermal Performance
Boiler Efficiency
The installation of the TFS 2000™R firing system did not affect the boiler thermal
efficiency (ASME Performance Test Code 4.1). Pre-retrofit and post-retrofit boiler
-------
efficiencies were calculated at MCR and at control load, and the efficiency remained at
91.4 - 91.7 percent, regardless of the NOx emissions level.
Steam Temperature/Flow Control
All post-retrofit operation of the boiler confirms that the superheater and reheater
design outlet steam temperatures can be maintained at loads from MCR through
control load. In addition, the superheater and reheater design pressures and mass flow
rates are maintained at all loads from MCR through control load.
Steam temperature control is accomplished through the use of the adjustable tilts and
the interstage desuperheaters. The windbox tilts continue to operate within their
normal range.
At both the maximum and potential minimum NOx emissions levels, the post-retrofit
reheater desuperheater spray water flows were about the same as the pre-retrofit levels.
Thus, the implementation of TFS 2000™R technology does not adversely impact the
unit's heat rate.
Element Steam Temperature Imbalance
Eight pre-retrofit tests and two post-retrofit tests were analyzed. Two of the pre-retrofit
tests were for normal operation, three were for operation with the top secondary air
dampers closed, and three were for operation with three tilt positions. One post-retrofit
test was conducted with maximum SOFA and acceptable boiler operation, and the
other was at the minimum NOx emission. The (low temperature) superheater rear
pendant outlet steam temperatures, (high temperature) superheater finishing pendant
outlet temperatures, and the high temperature reheater outlet temperatures were
measured and analyzed. As compared to the initial operation of the unit, firing oil, in
1968, there was no significant difference in the element steam temperature profiles
caused by the TFS 2000™R system.
Maximum Local Heat Absorption Rates
The peak waterwall heat absorption rates calculated from readings with the chordal
thermocouples installed in the furnace walls were well below the design values and
confirm that the post-retrofit departure from nucleate boiling (DNB) margin for the
boiler remains within ABB C-E design standards.
Vertical Heat Absorption Profile
The vertical heat absorption profile, as measured through the chordal waterwall
thermocouples is similar under all post-retrofit operating conditions. There is a slight
shift in the furnace vertical heat absorption profile towards the upper furnace under
potential minimum NOx conditions. This shift did not adversely affect boiler waterwall
circulation.
-------
DEC as a Function of NOx Emissions
Significant increases in UBC levels in the flyash have been documented for boilers
retrofitted with earlier low NOx firing systems.4 Pre-retrofit UBC levels at BHS
Unit 3 were in the range of 5.8 - 8.0
percent carbon. For a tangentially-
fired boiler with an Eastern
bituminous coal, this range is about
14
average.
sr 10
Except where noted, the flyash
samples for both the pre-retrofit and
post-retrofit UBC results were
obtained in accordance with EPA
Method 17. Carbon content was
determined directly, not by loss of
ignition (LOI).
\^
• Coarse (65 rpm)
Pre-Retrofit (Static)
O O O
•^^ •" Nominal (80 rpm)
Fine (90 rpm)
I
0.10
0.20
0.30 0.40
NOx (lb/10s Btu)
0.50
UBC levels for post-retrofit operation Figure 14: UBC in Flyash vs. NOx Emissions at MCR
at BHS Unit 3 with three different
fineness levels are given in Figure 14.
For this comparison, boiler load was held constant at MCR. The trend of increasing
UBC with decreasing NOx emissions is evident for the three post-retrofit data sets. The
trends also illustrate that UBC control is dependent upon the particle size of the coal.
NOx emissions as low as 0.20 lb/106 Btu were obtained with no increase above pre-
retrofit levels of UBC in the flyash.
Commercial Operating Experience
The unit has been operating commercially, post-retrofit, firing coal for over 14 months.
The unit operates under load dispatch at MCR on weekdays from about 8:00 am to
11:00 pm. At night and on weekends, the unit load is decreased to as low as 140 MW.
Operators report no significant operational problems, and no indication of accelerated
waterwall wastage or corrosion has been observed.
Daily Average Value for Megawatts, NOx, and UBC
Figure 15 is a plot of the megawatt load and NOx emissions during the period from
January 3,1995 to February 27,1995. The values given for NOx emissions in lb/106 Btu
are as measured on the Continuous Emissions Monitor (CEM) at the stack. The CEM is
RATA certified. NOx emissions during this period of normal operation averaged 0 23
lb/106 Btu.
During the pre-retrofit period from January 2,1993 to September 25,1993, a survey of
weekly ash truck samples (combining flyash and bottom ash) showed an average of 7 6
percent carbon in the ash. For the post-retrofit period from January 2,1994 to February
-------
20--
si
in
c
o
•e
CO
O
10--
0-L
Pre-retrofit
average carbon
in Ash = 7.6%
Si Si Si oj 53
Date
Figure 15: BHS Unit 3 NOx Emissions vs Megawatt Load and Unburned Carbon
m
to
X
O
25,1995, the weekly survey of the ash truck samples averaged 5.8 percent. Figure 15
also shows the weekly ash truck sample carbon content.
Unit Availability and Heat Rate
Table 2 shows the five leading causes per year for lost generation at BHS Unit 3 from
1990 through 1994. Comparisons of the equivalent forced outage rates, which
take into account any capacity reduction as well as shutdown, show that the unit's
overall post-retrofit availability is consistent with its historical performance. Post-
retrofit availability losses caused by economizer leaks and coal pulverizers are higher
than the historical levels. The economizer leaks were definitely not related to the TFS
2000™R installation, and problems have been corrected. The increased availability loss
caused by the coal pulverizers is attributable to problems with the initial design of the
seals for the Dynamic™ classifiers, which have since been corrected. No ongoing
loss of availability attributable to the Dynamic™ classifiers is expected.
Figure 16 shows the year-to-date effective availability factor (EAF), calculated both
including overhauls and excluding overhauls, for the years 1990 through 1994, and
for January through February of 1995. For 1993, the year-to-date figure is taken from
the pre-retrofit period of January through September. For 1994, the year-to-date figure
is entirely post-retrofit experience. The data in Figure 16 confirm that the post-retrofit
availability is consistent with the pre-retrofit experience.
-------
Annual Overhaul
Superheater Leak
Reheater Tube Leak
Economizer Tube Leak
Precipitator Repair
Waterwall Leak
Coal Mill
Main Boiler Feed Pump (MBFP)
Casing Leak
Condenser Cleaning
Boiler Circulating Pumps
Lube Oil System
Boiler Circ. PP
Boiler Water Chemistry
Turbine Vibration
Total of 5 Causes
Total Lost for Year
Equivalent Forced Outage Rate (EFOR)
1994 1993 1992 1991 1990
4.4 16.5 4.4 6.2 4.8
1.4
1.9 1-8 1.2
2.6
1.2
1.5 3.8 5.5 1.6 0.7
1.5
1.5
1.5
0.9
1.5
1.9
1.0
1.6
1.6
11.9% 24.4% 14.3% 12.6% 9.3%
17.8% 27.8% 17.9% 15.6% 11.4%
13.4% 11.3% 13.5% 9.4% 6.8%
Table 2: BHS Unit 3 - Five Leading Causes Per Year of Lost Generation
(Retrofit Work Done October - November, 1993)
100 r-
40
20
m| YTD (EAF incl. Overhauls)
| '?•;; ,| YTD (EAF excl. Overhauls)
Dec. Dec. Dec. Sept. Dec. Jan/Feb
1990 1991 1992 1993 1994 1995
(Pre-Retrofit) (Post-Retrofit)
Figure 16: BHS Unit 3 Equivalent Availability Factor Including and Excluding Overhauls
(Post-retrofit start-up was December, 1993)
-------
It is also apparent that the heat rate for BHS Unit 3 has not deteriorated as a result of
the TFS 2000™R retrofit. Figure 17 is a graph of the year-to-date heat rate in
Btu/kW-h for the unit from December of 1987 through December of 1994. The heat rate
post-retrofit (1994) is well within the historical trend of heat rate for the unit.
Heat Rate
Heat Rate
(Btu/kW-h)
12000 r
11000 -
10000 -
9000
8000
7000
6000
87 88 89 90 91
92 93
94
t
TFS 2000™R
Retrofit
Figure 17: BHS Unit 3 Heat Rate Performance History
Equipment Operational Experience and Inspection
During a scheduled maintenance outage inspection conducted in the fall of 1994, the
new firing system condition was assessed as excellent. Tilts were stroked with no
evidence of binding or erratic operation. Eight of twenty coal nozzle tips had slight
warpage in splitter plates that was not affecting performance, and repair was not
required. One oil nozzle tip bluff body diffuser was slightly cracked, and repair was
not required. SOFA-related components were in "as new" condition. The ignitor
system was also in "as new" condition.
Detailed visual inspections were made of the economizer, steam drum, lower waterwall
drums, waterwalls, bottom ash hopper, superheater steam cooled roof and wall
headers, backpass steam cooled outlet headers, low temperature superheater,
superheater division wall panels and pendant platen assemblies, and reheater
-------
assemblies, and all were found to be in a condition typical of pre-retrofit experience.
Service recommendations in the outage report were consistent with pre-retrofit
experience on this unit.
A visual inspection of the Dynamic™ classifer during the 1994 fall outage showed no
visible wear on the rotor assembly. An exhauster inspection indicated no required
maintenance after nine months (approximately 8500 hours) of service.
Conclusions
United Illuminating and ABB C-E Services consider the retrofit of Bridgeport Harbor
Station's Unit 3 to be a commercially and technically successful full-scale demonstration
of TFS 2000™R technology. The boiler thermal performance and efficiency are
unchanged from the pre-retrofit conditions. Although the slagging/fouling patterns
have changed slightly from pre-retrofit, the existing sootblowers and wall blowers are
capable of controlling them.
During testing, the boiler consistently demonstrated NOx emissions on the order of
0.25 lb/10^ Btu over the entire load range, with no increase in unburned carbon in the
flyash. The lowest NOx emissions measured for this boiler during post-retrofit
parametric testing is 0.16 lb/10" Btu. The potential for long-term operation of the
boiler at this level has not been thoroughly investigated. In over 14 months of
commercial operation, operation of the boiler with the TFS 2000™R technology has
caused no significant adverse impact on boiler operation or availability.
Acknowledgements
The authors acknowledge and appreciate the efforts and expertise of all the individuals
from United Illuminating and ABB who were involved in the success of this field
demonstration project. The contributions of D. Gillespie, P. Olson, A. Cortiglio,
R. Diotalevi, D. Cahill, J. Buffa, P. Acimovic, T. Dorazio, W. Derech, V. Piras, and
R. Collette are especially noted. Thanks also to R. Lewis, G. Strich, D. Choi,
P. Stanwicks, T. Kelly, C. Boyle, B. Walsh, D. Gelbar, D. Barlow, J. Dallen, and C. Doherty
for their valuable contributions.
References
1. Personal communication, P. Olson, United Illuminating, 1994.
2. Marion, J.L., Towle, D.P., Kunkel, R.C, and LaFlesh, R.C, Development of ABB C-E's
Tangential Firing System 2000 (TFS 2000™ System), EPRI/EPA1993 Joint Symposium
on Stationary Combustion NOx Control, reprinted as TIS 8603,1993.
3. McCartney, M.S., et. al., Development and Evolution of the ABBCombustion Engineering
Low NOx Concentric Firing System, TIS 8551,1991.
-------
4. Hart, D., Operating Results from ABB C-E Services' LNCFS™ Low NOx Concentric
Firing System Retrofit Installations -1994 Update, TIS 8620.
5. State-of-the-Art Pulverizer Development Facility, Power Perspectives, ABB, September,
1994.
-------
ESTIMATION OF NOX EMISSIONS FROM PULVERIZED
COAL-FIRED UTILITY BOILERS
David I Wildman
Scott M. Smouse
United States Department of Energy
Pittsburgh Energy Technology Center
Pittsburgh, PA 15236-0940
Abstract
The formation of nitrogen oxides (NOX) during pulverized-coal combustion in utility boilers is
governed by many factors, including the boiler's design characteristics and operating
conditions, and coal properties. Presently, no simple, reliable method is publicly available to
estimate NOX emissions from any coal-fired boiler. A neural network back-propagation
algorithm was previously developed using a small data set of boiler design characteristics and
operating conditions, and coal properties for tangentially fired boilers. This initial effort
yielded sufficient confidence in the use of neural network data analysis techniques to expand
the data base to other boiler firing modes. A new neural network-based algorithm has been
developed for all major pulverized coal-firing modes (wall, opposed-wall, cell, and tangential)
that accurately predicts NOX emissions using eleven readily available data inputs. A
sensitivity study was completed for all major input parameters, which yielded results that
agree with conventional wisdom and practical experience. This new algorithm is being used
by others, including the Electric Power Research Institute who has included it in its new
software for making emissions compliance decisions, the Clean Air Technology Workstation.
Introduction
The formation of nitrogen oxides (NOJ during pulverized-coal combustion in utility boilers is
governed by many factors, including the boiler's design characteristics and operating
conditions, and coal properties. Presently, no simple, reliable method is publicly available to
estimate NOX emissions from any coal-fired boiler. The authors previously showed that
existing empirical methods were inadequate for predicting NOX emissions from even a small
data set of tangentially fired utility boilers (1). To provide a simple, short-term, multi-
purpose means to predict NOX emissions from pulverized coal-fired utility boilers, a neural
network back-propagation algorithm was previously applied to a data set for nine tangentially
fired boilers (1). The purpose of this initial study was to assess the ability of a neural
network to reasonably predict NOX emissions for a qualified data set. As previously reported
(1), the algorithm developed by the Pittsburgh Energy Technology Center (PETC) generally
-------
predicted NOX emissions to within ±50 ppm of actual values for the tangentially fired boilers
in the limited initial data set.
The Electric Power Research Institute (EPRI) is developing a variety of software tools to
assist its member utilities with making decisions on how to comply with the Clean Air Act
Amendments of 1990 and the changing power generation marketplace. Sargent & Lundy
Engineers has developed the Clean Air Technology (CAT) Workstation as the primary
compliance decision tool for EPRI's member utilities. The CAT Workstation includes
modules to comparatively evaluate various emissions control technologies. After the 1993
Joint Symposium on Stationary Combustion NOX Control, where the previous work was
presented, EPRI expressed an interest in using PETC's neural network-based algorithm within
the CAT Workstation. However, because the data base upon which the initial algorithm was
developed was limited to tangentially fired boilers, PETC recommended collection of
additional data and development of a new NOX algorithm. Subsequently, a Cooperative
Research and Development Agreement (CRADA) was implemented between PETC and EPRI
to collect additional power plant data, develop a new algorithm, and provide a copy of the
coded algorithm to Sargent & Lundy for inclusion in the CAT Workstation. This effort was
initiated in early 1994 and concluded by mid-year 1994 in order to meet the deadline for
release of the CAT Workstation to EPRI's member utilities.
Technology-based standards, such as Reasonably Available Control Technology (RACT), that
require the installation of NOX control technologies on a boiler-by-boiler basis, are driving the
emissions compliance strategies of U.S. utilities. Carnegie Mellon University's (CMU)
Department of Engineering and Public Policy is examining the potential cost savings
achievable through NOX emissions trading, an alternative market-based strategy to achieve
federal and state goals for NOX emission reduction (2). To estimate the potential cost savings
achievable through inter-utility NOX trading, CMU is using a combinatorial optimization
approach to identify boiler retrofits and operating parameters that yield the most cost-effective
means for NOX abatement (2). PETC's new neural network-based algorithm for predicting
NOX emissions was utilized by CMU to predict emissions from pulverized coal-fired boilers in
the Pittsburgh, Pennsylvania area as part of their efforts. CMU is continuing development of
their trading analysis software for NOX emissions using PETC's algorithm to predict NOX
emissions.
This paper summarizes the development of the new neural network-based algorithm, including
a review of the data base used in the effort and a sensitivity study for the major input
parameters. Planned efforts to significantly expand the existing data base using data and
information being collected by the U.S. Environmental Protection Agency (EPA) from U.S.
utilities, as required by the Clean Air Act Amendments of 1990, is briefly discussed.
Neural Network Approach for Estimating NOX Emissions
Neural networks became popular in the 1980s for analyzing complex, inter-related data sets.
Neural networks do not assume that a relationship is known between process input and output
but rather try to determine the relationship by analyzing data sets of input and output data.
-------
Neural networks, which can be viewed as a non-linear data analysis technique, are
computational systems that use the organizational principles of biological nervous systems.
Computer scientists imitate these principles because biological systems easily outperform all
current approaches for pattern recognition. The primary requirement for developing a neural
network is a complete data set. Also, the proper representation of the input (e.g., a ratio of
two variables may be the best representation) hastens the training process.
Neural networks consist of several simple, highly interconnected data-processing units. These
units approximate the complex system of neurons and electrochemical signals that are used by
the human brain to process information (3). Each of the neurons in the human brain functions
like a tiny computer with limited capabilities. Connected together, these cells form the most
intelligent system known. Neural networks are a relatively new class of computing systems
formed from hundreds of simulated neurons interconnected like the brain's neurons. The
principal aim of the technology is to mimic nature's approach for processing data and
information. The back-propagation algorithm developed independently by at least three
groups (4-6) is the most widely used neural network approach.
Neural networks are being used to solve real problems. For example, they have been shown
to be effective in estimating the fatigue life of mechanical parts (7), performing fault detection
in chemical plants (8), diagnosing automobile malfunctions (9), and recognizing human
speech (10). Neural networks are also being widely applied within the power industry (3).
During the last five years, more than 200 technical papers discussing artificial intelligence
applications in the power industry have been published; more than one-half of these
applications involved the back-propagation algorithm (11). All of the above applications used
noisy data, where the relationships between input-output pairs were complicated and poorly
understood. Neural networks generally excel over traditional data analysis techniques, such
multiple linear regression techniques, in analyzing noisy data sets.
Commercial neural network software, Brainmaker Professional (Version 2.03) from California
Scientific Software, was used in this study. During the network development process, the
user defines the problem and collects the input-output pairs. If the user neglects key input
parameters for a particular application, the predictive capabilities of the network will suffer.
A neural network is not programmed with rules like an expert system, but rather it "learns" in
much the same way that people do, i.e., by example and repetition.
After the input-output pairs are collected, the learning stage commences with the network
associating output data with input data. Each time an input is presented, the network sends
back an answer of what it thinks the output should be. When it is wrong, the network
corrects itself by changing the weight matrices that are used to adjust the input-output
relationships. Brainmaker Professional uses a back-propagation algorithm to adjust these
weight matrices. The training process is repeated until the network derives answers that are
within a user-specified tolerance for all inputs. This learning process can take considerable
time, possibly up to several days or weeks, to complete depending on the complexity of the
problem and the processing speed of the computer used. The final weight matrices are saved
to file and can be used external to Brainmaker Professional to predict outputs for other sets of
input parameters.
-------
Key to the training process in a neural network is the transfer function. Output is determined
by the network's nodes, which combine signals sent to them by lower-layered nodes and
transfer these adjusted signals to higher-layered nodes. Neural network nodes are designed to
process signals with values between 0 and 1; therefore, all input variables must be scaled to
the unit interval. Non-linear transfer functions send signals from node to node within the
network. The default transfer function used by Brainmaker Professional is the sigmoid
function:
1 + exp(-;t)
Figure 1 shows that the sigmoid function maps real numbers into the unit interval. It is often
called a "squashing function" because very large positive values are asymptotically mapped to
1 and very large negative values are asymptotically mapped to zero. The squashing function
forces the network to focus on the bulk of the data and to place less emphasis on the data at
the extreme lower and upper ends of a data set. Hence, network predictions are generally less
accurate at the data extremes.
A network stops training when its answers are within a user-specified tolerance. This
tolerance specification is applied to the data after it has been subjected to the squashing
function. The non-linear nature of the squashing function often causes extreme data values to
deviate more from actual values than non-extreme data. Typically, training is conducted at
progressively smaller tolerance levels until the deviation between predicted and actual values
is acceptable by the user. The value of acceptable deviation between the network answers
and the actual data depends upon the accuracy of the available input-output pairs. For
example, if inherent errors in the available input-output pairs limit their repeatability to 5%,
then training to a tolerance less than 0.1 is unwise. Although decreasing the tolerance will
force the network answers to more closely approximate the training data, the network will not
necessarily perform better on new data sets. Usually, continued training at progressively
smaller tolerances improves the network's performance to a point beyond which it
deteriorates. Once confidence is obtained in the predictive capabilities of the neural network,
it is used to generate answers for input sets with unknown outputs.
Definition and Scope of Study
To expand the initial data base, additional data were collected from a variety of sources,
including several electric utilities, the Utility Air Regulatory Group (UARG), and several
projects within USDOE's Clean Coal Technology Program. A total of 384 data sets from 36
pulverized coal-fired boilers were collected and used in this study. Thousands of other data
sets were also collected that were not usable because they lacked key data inputs. Also, some
data was collected from cyclone-fired utility boilers but was not analyzed because this effort
was limited to pulverized coal-fired boilers. Multiple data sets were available for some
boilers because of variations in operating conditions and coal type. Some boilers had
emissions data from the burners that were originally installed when the unit was built, which
-------
were either conventional or low-NOx burners, and low-NOx burners that were retrofitted at a
later date. Both pre- and post-NSPS (New Source Performance Standards) boilers were
included in the data set. Table 1 summarizes the characteristics of the boilers in the available
data set as a function of firing mode (e.g., wall, opposed-wall, cell, tangential). No data were
available from wall-fired units with low-NOx burners, largely as a result of the limited time
available to produce an algorithm for inclusion in EPRI's CAT Workstation. The new
tangentially fired boiler data consisted of a few data sets from many boilers whereas the data
for the other boiler types consisted of many data sets from a few boilers. The capacity ranges
of the wall-fired boilers, opposed wall-fired boilers, and cell burner-equipped boilers were
limited compared to that of the tangentially-fired boilers because attempts to acquire
additional data over a broader capacity range were unsuccessful during the limited available
time.
A number of possible input parameters, based on boiler design characteristics and operating
conditions, and fuel properties, were considered during development of the new NOX
algorithm. The objective was to use a limited number of simple, readily available parameters
as long as the predicted NOX values agreed reasonably with actual emission values for a range
of boiler types, boiler operating conditions, and coal properties. The starting point for
selection of the new input parameters was the list of parameters used for the original NOX
algorithm. However, data availability precluded use of some of the original inputs and
extension of the data set to additional boiler types required inclusion of new parameters to
adequately describe the various boiler designs.
Table 2 summarizes the parameters that were eventually selected as input to the new neural
network. Furnace height, which is used to calculate the furnace volumetric heat release rate,
was defined in this effort as the distance from the bottom of the ash hopper to the furnace
roof (i.e., the overall boiler height). The more standard definition for boiler height, which is
the distance from the furnace hopper knuckle to the furnace nose, was not used because
detailed furnace dimensions were not available for all units and time limitations did not
permit collection of this data from individual power stations. The year that the boiler first
became operational was used because recent boiler designs attempt to limit NOX formation in
conjunction with low-NOx burners. The year that the current burners were installed was used
as an indicator of the developmental status or maturity of low-NOx burner technology. It
would be too difficult to use individual burner design specifications for different
manufacturers in this simplistic approach. The ratio of boiler load to the boiler's maximum
continuous rating (MCR), along with the plan area and volumetric heat release rates, are used
to reflect a boiler's overall design and operating characteristics. Most utilities provided the
NOX and excess oxygen values at the economizer outlet; therefore, this emissions sampling
location was selected. Other utilities provided NOX and O2 data at the boiler exit or stack.
However, there were fewer data sets with this information. Also, because leakage of air into
the exhaust system of a plant can not be distinguished from operating conditions with higher
excess air, stack emissions data are difficult to use. Therefore, only data sets with the oxygen
content in flue gas at the economizer outlet were used in this study. No input parameter was
used to address the impact of over-fire air usage or mill firing pattern on NOX emissions
because sufficient data were not available for network training. While omission of these
parameters is recognized as a significant weakness because they are known to have a major
-------
impact on NOX emissions levels, the previous effort yielded reasonable NOX emissions
estimates for a wide variety of boilers, boiler operating conditions, and coal types without
inclusion of this information.
Neural Network Results
Figure 2 compares the actual and estimated NOX emission values for the 346 data sets used to
train the network. The network was trained at a tolerance of 0.10, i.e., training was complete
when all of the scaled predicted values (i.e., from the squashing function) agreed to within
10% of scaled actual values. As previously stated, the non-linear nature of the sigmoid
function causes large deviations between actual and estimated values at the extremes of the
data. The greatest difference between an actual NOX value and its corresponding estimated
value is 263 ppm, which occurred for a measured NOX value of 1350 ppm (0% O2 dry
volume basis). The standard error of estimate for the training data was 67 ppm. Training at
a smaller tolerance reduced the maximum deviation and the standard error of estimate but the
network performance degraded on the data that was reserved for testing. As shown in Figure
2, the neural network's maximum estimate is 1150 ppm (0% O2 dry volume basis)
Increasing this value would improve the fit for extremely high NOX emission values but would
degrade the fit of the non-extreme data. It was felt that minimizing the deviations between
actual and estimated NOX emissions for the bulk of the (i.e., non-extreme) data was more
important than minimizing deviations for the extremely high NOX emission values. The
standard error of estimate decreased from 67 ppm to 52 ppm if only the data sets that had an
actual NOX values less than 1150 ppm (0% O2 dry volume basis) were considered.
Figure 3 compares actual and estimated NOX emissions for the 38 data sets reserved for
testing by the neural network. These data sets represent the balance of data sets in the master
data set. The standard error of estimate for the test data was 64 ppm. The greatest difference
between a measured NOX value and its corresponding estimated value was 156 ppm, which
was again for a data set with an extremely high actual NOX emission value.
Sensitivity of Input Variables
Although the prior study yielded a neural network that was very good at estimating NOX
emission values using readily available input data, little effort was placed on assessing the
reasonableness of the data trends predicted using the network algorithm. To assess the
sensitivity of the selected input parameters on NOX emission trends, a series of numerical
experiments were performed using the new algorithm. While these experiments do not fully
evaluate the accuracy of the algorithm because actual data were not available for comparison
in all cases, they do reveal trends that agree with conventional wisdom and practical
experience. Also, they demonstrate how the algorithm can be used in various utility
scenanos, such as low-NOx burner retrofits and boiler operational changes. These sensitivity
analyses are not meant to describe trends that are universally applicable to all pulverized coal-
fired boilers but rather to show that, while the trends are generally reasonable, specific
analyses can yield results that require further explanation. To simplify these analyses, input
-------
parameters were varied individually except when fuel analysis effects were being assessed.
As detailed in Table 3, four different scenarios were analyzed:
1) a pre-NSPS 200 MWC wall-fired boiler built in 1955 that is equipped with the original
burners
2) a pre-NSPS 200 MWe wall-fired boiler built in 1955 that was hypothetically retrofitted
with low NOX burners in 1985
3) a pre-NSPS 180 MWe tangentially fired boiler built in 1967 that is equipped with the
original burners
4) a post-NSPS 250 MW. tangentially fired boiler built in 1984 that is equipped with the
original burners
At 5-year increments from 1975 until 1990, the effect of retrofitting low-NOx burners on NOX
emissions is shown in Figure 4 for the pre-NSPS 200 MWe wall-fired boiler built in 1955 and
the pre-NSPS 180 MWe tangentially fired boiler built in 1967. In this figure and all
subsequent figures, the two dates in parentheses indicate the year that the unit first
commercially operated and the year that the current (i.e., either original or retrofit burners)
were installed. In this analysis, it was assumed that other boiler improvements were
implemented when the burners were retrofitted to reduce the flue gas oxygen content at the
economizer exit from 8.4% to 5.4% for the wall-fired unit and from 5.5% to 3.5% for the
tangentially fired unit. No other input parameter was varied. Figure 4 and subsequent figures
(where data were available) shows the excellent agreement between actual and predicted NOX
emission values. Retrofitting low-NOx burners in 1975 reduces the emissions from the wall-
fired unit and the tangentially fired unit by about 28% and 35% and in 1990 by about 38%
and 38%, respectively. Interestingly, the same 38% reduction was predicted for both units
when low-NOx burners are retrofitted in 1990. It is not known whether this level of NOX
reduction represents a limit imposed by the limited available data set; additional sensitivity
analyses need to be performed for other units to investigate this issue. However, these trends
generally do agree with the observed ability of low-NOx burners to reduce emissions. Also,
essentially no improvement in the ability of low-NOx burners to reduce emissions from the
tangentially fired boiler was predicted over the period from 1975 to 1990; other plants may
show significantly different results.
One of the simplest, most cost-effective means to control NOX emissions from pulverized
coal-fired utility boilers is reducing the operating excess air level. The effect of excess air
level on NOX emissions from the above four units is shown in Figure 5. In these numerical
experiments, the excess air level was varied 2% higher and lower than the actual reported
value at the economizer exit of each boiler. Significant changes in NOX emissions are
predicted as the excess oxygen content is varied for all units except the post-NSPS 250 MWe
tangentially fired unit, which has an actual NOX emission level of only 270 ppm (0% Ox dry
volume basis). These trends generally agree with actual experience in these units and other
field measurements.
Fuel properties can have a significant impact on NOX emissions from pulverized-coal
combustion. Substantial differences in NOX emissions have been observed when the coal
being fired is changed or blended with other coals. The two coals that were fired in the pre-
-------
NSPS 200 MWC wall-fired unit and the pre-NSPS 180 MWe tangentially fired unit were
selected for a numerical blending simulation. As shown in Table 3, the coal burned in the
wall-fired unit had a fixed carbon-to-volatile matter (FC/VM) ratio of 1.42 and the coal
burned in the tangentially fired unit had FC/VM of 1.02. These two coals were numerically
blended at ratios of 1/3, 1/1, and 3/1 to produce simulated coals with compositions between
the two actual coals. The effect of FC/VM ratio on NOX emissions is shown in Figure 6.
Significant changes in NOX emissions are predicted for the two pre-NSPS units whereas the
two units with low-NOx burners are relatively insensitive to fuel properties. Some low-NOx
burner manufacturers claim that their burners are not sensitive to some fuel properties, such as
FC/VM ration, while others report significant sensitivity. Again, other scenarios may show
substantially different results.
Boiler design obviously has a major impact on NOX emissions. Generally, lower furnace heat
release rates will yield lower NOX emissions as a result of lower flame temperatures and
changes in burner/furnace aerodynamics. Figure 7 shows the effect of plan area heat release
rate on NOX emissions. In this analysis, incremental changes, both higher and lower, in the
plan area heat release rate of 5%, 10%, 15%, and 20% from the actual design point for these
boilers were evaluated. Also, the volumetric heat release rate and other input parameters were
held constant. Because volumetric heat release rate was held constant, these analyses reflect a
change in boiler design (e.g., to achieve lower plan area heat release rates, a unit will be
made wider and/or deeper while the unit height is decreased) not a change in boiler firing rate
(i.e., load). In general, the effect of plan area heat release rate was minimal for all four units.
This result seems contrary to practical experience but can be partially explained by the fact it
is unlikely that only plan area heat release rate would be changed if a boiler design was being
changed to achieve lower NOX emissions. This is supported by Figure 8, which shows a very
strong effect of volumetric heat release rate for all units except the post-NSPS 250 MWe
tangentially fired unit, which has very low actual NOX emissions. In this analysis, NOX
emissions were determined for incremental changes in furnace volumetric heat release rate of
5%, 10%, 15%, and 20% higher and lower than the actual design point for each boiler.
Because the two input parameters of plan area and volumetric heat release rate are so strongly
related, it possible that only one of these inputs is required to adequately describe the
combined impact of firing rate and boiler design on NOX emissions.
Future Work
The Clean Air Act Amendments of 1990 require most U.S. coal-fired power plants to install
continuous emissions monitoring (CEM) systems and to electronically report the collected
data to the U.S. Environmental Protection Agency on a quarterly basis. EPA has collected
up to a full year of data from some utilities and several quarters of data from the remaining
utilities. PETC has worked with EPA on various NOX issues, including assessing the
performance and cost of various combustion and post-combustion control technologies.
Discussions have been held with EPA on acquiring the first year's emissions data from all
U.S. coal-fired power plants equipped with CEMs to expand the data base for development of
a new NOX prediction algorithm. While this emissions data base would be unequalled by any
other, EPA is not authorized to collect all the related information that is required to predict
-------
NOX emissions accurately. For example, EPA is not collecting all the boiler design and coal
property data that is used in the current NOX algorithm. Collection of this data would require
an effort that is beyond current resources. However, several options to perform this task are
still being considered.
PETC has also discussed similar efforts with EPRI to develop a neural network-based NOX
algorithm for gas- and oil-fired utility boilers. The Gas Research Institute (GRI) has been
using EPRI's CAT Workstation, which includes PETC's NOX algorithm, as part of their market
projections for natural utilization in coal-fired boilers (12). GRI has expressed an interest an
algorithm to predict NOX emissions from coal-fired boilers using natural gas co-firing or
rebuming.
PETC will continue to expand its current data base with a focus on including more data from
power plants that have retrofitted low-NOx burners. When sufficient data are available, a new
neural network-based NOX algorithm will be developed and supplied to existing and new
users. PETC hopes to initiate some of the above activities later this year, which will likely
result in an improved algorithm for all coal-fired boiler types.
Summary
A neural network has been developed to estimate NOX emissions for wall-fired boilers,
opposed wall-fired boilers, cell burner-equipped boilers, and tangentially fired boilers.
Eleven simple, readily available parameters were selected as input parameters. The data base
used in this effort consisted of boiler design characteristics and operating conditions, and coal
property data for 36 pulverized coal-fired utility boilers. Some of these boilers were either
originally equipped with low-NOx burners or had been retrofitted with low-NOx burners. A
sensitivity analysis of the selected input parameters yielded trends that agree with
conventional wisdom and practical experience. Some of the current and future applications
for PETC's neural network-based NOX algorithm were highlighted; the most notable of the
current applications is the algorithm's inclusion in the Electric Power Research Institute's
Clean Air Technology Workstation, which is the primary tool for making emissions
compliance decisions tool for its member utilities.
Disclaimer
Reference herein to any specific commercial product 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.
-------
References
I. Smouse, S.M., Wildman, D.J., Mcllvned, T.S., and Harding, N.S.., "Estimation of NOX
Emissions from Pulverized Coal-Fired Utility Boilers," EPRI/EPA 1993 Joint
Symposium on Stationary Combustion NOX Emissions Control, 1993.
2. Siegel, S.A., and Kalagnanam, J.R., "The Potential Cost Savings of Implementing an
Inter-Utility NOX Trading Program," International Conference on Acid Rain & Electric
Utilities: Permits, Allowances, Monitoring & Meteorology, Air and Waste Management
Association, 1995.
3. Naser, I, "Exploring Neural Network Technology," EPRI Journal, December, 1992.
4. Rumelhart, D.E., Hinton, G., and Williams, R.J., "Learning Internal Representations by
Error Propagation" Parallel Distributed Processing, Vol. 1, MIT Press, 1986.
5. Parker, D.B., Optimal Algorithms for Adaptive Networks: Second Order Back
Propagation, Second Order Direct Propagation, Second Order Hebbian Learning,"
Proceedings of IEEE First International Conference on Neural Networks, 1987.
6. Werbos, P.J., "Beyond Regression: New Tools for Prediction and Analysis in Behavior
Sciences," PhD Thesis, Harvard University, Cambridge MA, 1974.
7. Troudet, T. and Werrill, W., "A Real Time Neural Net Estimator of Fatigue Life,"
Proceedings of the International Joint Conference on Neural Networks, 1990.
8. Hoskins, J.C., Kaliyur, K.M., and Himmelblau, D.M., "Incipient Fault Detection and
Diagnosis Using Neural Networks," Proceedings of the International Joint Conference on
Neural Networks, 1990.
9. Marko, K.A., Feldkamp, L.A., and Puskorius, G.V., "Automotive Diagnostics Using
Tramable Classifiers: Statistical Testing and Paradigm Selection," Proceedings of the
International Joint Conference on Neural Networks, 1990.
10. Hampshire, J.B., and Waibel, A., "Connectionist Architectures for Multi-speaker
Phoneme Recognition," Advances in Neural Information Processing Systems 2, 1990.
11. Niebur, D., "Artificial Neural Networks for Power Systems - A Literature Survey,"
Proceedings of the EPRI conference on Expert System Applications for the Electric
Power Industry, 1993.
12. Partapas, J., United States Department of Energy Pittsburgh Energy Technology-Gas
Research Institute Meeting, Pittsburgh, PA, March, 1995.
-------
Table 1
Boiler Database Information as a Function of Firing Mode
Firing Mode/
Boiler Information
Number of Boilers
Number of Units
with Low-NOx Burners
Number of Data Sets
Number of Units
with 2 Furnaces
Range of Boiler Capacity
(Gross MWJ
Wall
6
0
77
1
175-200
Opposed
Wall
6
3
96
0
265-688
Cell
3
1
55
0
610-1300
Tangential
21
4
156
9
85-936
-------
Table 2
Input Parameters Selected for NOX Estimation Study
Parameter
Plan Area Heat Release Rate
Furnace Volume Heat Release Rate
Firing Mode
(l=tangential; 2=wall; 3=opposed wall; 4=cell)
Number of Furnaces
Year Unit First Commercially Operated
Year Burners Installed (Original or Retrofit Burners)
Flue Gas Oxygen Content at Economizer Outlet
Load/Maximum Continuous Rating
Coal's Carbon-to-Hydrogen Ratio (C/H)
Coal's Nitrogen Content
Coal's Fixed Carbon-to-Volatile Matter Ratio (FC/VM)
Range of
Values
0.4-3.3
3.4-29.7
1-4
1 or 2
1951-1986
1951-1990
1.8-12.1
0.53-1.12
13.3-21.3
0.8-1.8
1.0-3.3
Units
MMBtu/ft2
KBtu/ft3
—
—
year
year
vol%,
dry basis
MWe/MWe
wt%/wt%
wt%,
dry basis
wt%/wt%
-------
Table 3
Base Input Parameters for Sensitivity Study Cases
Parameter
Plan Area Heat Release Rate
Furnace Volume Heat Release Rate
Firing Mode
(l=tangential; 2=wall; 3=opposed wall; 4=cell)
Number of Furnaces
Year Unit First Commercially Operated
Year Burners Installed (Original or Retrofit Burners)
Flue Gas Oxygen Content at Economizer Outlet
Load/Maximum Continuous Rating
Coal's Carbon-to-Hydrogen Ratio (C/H)
Coal's Nitrogen Content
Coal's Fixed Carbon-to-Volatile Matter Ratio (FC/VM)
Units
MMBtu/ft2
KBtu/ft3
—
—
year
year
vol%,
dry basis
MWe/MWe
wt%/wt%
wt%,
dry basis
wt%/wt%
Unit #1
1.57
16.53
2
1
1955
1955
8.40
1.00
14.98
1.70
1.42
Unit #2
1.57
16.53
2
1
1955
1985
5.40
1.00
14.98
1.70
1.42
Unit #3
1.71
18.03
1
1
1967
1967
5.45
1.00
16.67
1.49
1.29
Unit #4
1.80
14.55
1
I
1984
1984
2.80
1.03
13.37
1.09
1.02
-------
1.0 r
CD
0.8
0.7
0.6
0.1
J 1 i . i
Figure 1. Sigmoid "squashing" function
-------
TO,
C\J
O
E
Q.
Q.
to"
c
0
'to
E
0)
X
O
2;
T3
0)
CB
E
V- •
UJ
1400
1300
1200
1100
1000
900
800
700
600
500
400
300
200
2(
—
-
-
o o ° ^d&flBStAj® cxsooo
o
o
o
0
; 8°^%$1^0
- <$^$$^
$ c$s|P»o
" ^Jp^^^
1 1 , 1 1 1 1 1 . 1 I 1 I 1 1 1 , 1 , 1 , 1 , 1
)0 300 400 500 600 700 800 900 1000 1100 1200 1300 1400
Actual NOx emissions, ppm at 0% 02 (dry)
Figure 2. Actual and estimated NOx emissions (training data)
-------
-I
CM
O
5°
03
E
Q.
Q.
C
g
E
CO
X
O
-O
Q)
E
LJJ
1400
1300
1200
1100
1000
900
800
700
600
500
400
300
200
2C
—
-
-
° 9 <*> °
-
-
o
o
^ o
o
00°
o °
0°
~ OC^CJD8
9) °
I , I , I > I i ! < I i I i I . I . I . I , 1
)0 300 400 500 600 700 800 900 1000 1100 1200 1300 14(
Actual NOx emissions, ppm at 0% O2 (dry)
Figure 3. Actual and estimated NOx emissions (test data)
-------
800
.700
1?
C\J
0 600
o
To
Q. 500
CL
in
o
'in
-| 400
0)
X
O
300
200
19
i-
-
A
8
200 MW
• 8.4% O2
o 8.4% O2
5.4% O2
180MW
wall-fired 1 9557— -
actual data
est. data
estimated data (retrofit)
t-fired1967/-« -
A 5.45% 02 actual data
A 5.45% 0
3 A co/ r\
.4O7o U
-
• — — -^__
^ — -^
•
-
1 1 I I 1 I
2 estimated data
2 estimated data (retrofit)
.
-
•
_
-
-
-
i I
0.9
0.8
-t— •
DQ
0.7 ^
5
0.6 §
in
w
E
Or- Q)
.5 £
O
0.4
0.3
50 1955 1960 1965 1970 1975 1980 1985 1990 1995
Year burners installed
Figure 4. Effect of retrofit burners on NOx emissions
-------
CM
O
800
700
600
200 MW wall-fired 1955/1955
• Actual data
Estimated data
200 MW wall-fired 1955/1985
- Estimated data
180MWt-fired 1967/1967
i Actual data
— Estimated data
250MWt-fired 1984/1984
Actual data
Estimated data
A
0.9
0.8
0.7
E
Q.
Q.
C
o
w
'E
0)
X
O
500
400
300
0
0.6
0.5
0.4
0.3
X
O
200
0
7
8
1 23456
Excess oxygen content in dry flue gas (%)
Figure 5. Effect of excess oxygen in flue gas on NOx emissions
10
11
12
-------
(N
O
I
g
'w
E
0)
X
O
800
700
600
500
400
200 MW wall-fired 1955/1955
• Actual data
Estimated data
200 MW wall-fired 1955/1985
Estimated data
180 MWt-fired 1967/1967
A Actual data
Estimated data
250 MWt-fired 1984/1984
+ Actual data
Estimated data
1.2
1.3 1.4
Fixed carbon-to-volatile matter ratio
Figure 6. Effect of coal composition on NOx emissions
0.9
0.8
0.7
0.6
0.5
0.4
0.3
1.5
w
E
0)
-------
800
700
>;
H.
CVJ
Ofcf\f\
600
o
CO
Q. 500
Q_
CO
c
o
to
-| 400
o>
X
O
Z
300
9nn
•-
-
-
.• ~~
~~ • — .
,
i , , , , i
200 MW wall-fired 1955/1955
• Actual data
Estimated data
200 MW wall-fired 1955/1985
Estimated data
180M Wt-fired 1967/1967
A Actual data
Estimated data
250 M W t-fired 1984/1984
^ Actual data
Estimated data
— i 1 1 , 1 1 i i_
-
-
-
-
-
-
0.9
0.8
13
ffi
0~7 *?:
. / .->
:9
0-6 I"
1
0.5 J
0
0.4
0.3
1.0
1.5
2.0
2.5
3.0
Plan area heat release rate (MMBtu/ft2)
Figure 7. Effect of plan area heat release rate on NOx emissions
-------
CM
O
O
CO
Q.
CL
CO
c
O
'co
E
cu
X
O
800
700
600
500
% 400
300
200 MW wall-fired 1955/1955
• Actual data
Estimated data
200 MW Wall-fired 1955/1985
- Estimated data
180 MWt-fired 1967/1967
4 Actual data
— Estimated data
250 MWt-fired 1984/1984
Actual data
Estimated data
0.9
0.8
0.7
0.6 §
0.5
0.4
0.3
CO
CO
O
200
8
10
12
14
16
18
20
Furnace volumetric heat release rate (KBtu/ft3)
Figure 8. Effect of furnace volumetric heat release rate on NOx emissions
22
-------
Separation and Recycling of Carbon in Ash
David R. Whitlock
Separation Technologies Inc.
10 Kearney Road
Needham MA 02194
(617)455-6600
(617)455-6518 (fax)
Abstract
Separation Technologies has developed a new electrostatic process for separating the unburned
carbon from utility flyash. As NOx emission levels are reduced through burner modifications
and as harder and lower ash coals are burned, the carbon levels in the fly ash increase above the
maximum levels for use in concrete. This paper details the results of a production test program
carried out in mid-1994 to produce low carbon ash for use in concrete. 3000 tons of low carbon
ash were produced at rates of 15-25 tons per hour. The average loss-on ingnition of the fly ash
was reduced from 8.2% down to 2.3% with 83% yield.
Introduction
The carbon level in flyash from utility boilers has been increasing in the recent past as a result of
a number of factors. Not the least of which is the requirement for many utilities to reduce the
amount of NOx emitted. This is usually done through retrofit of special burners and air controls
that have the effect of reducing the excess air and reducing the flame temperature. Both of these
changes increase the level of unburned carbon in the ash.
Additional factors which compound the problem include increased competition among coal
suppliers. With reduced demand for coke, coking coals are now available for combustion.
Sulfur content for SO2 emissions and CaO content for precipitator performance are also
becoming more important. The net result is that many boilers are burning coals with
specifications different than what the boilers were designed for. Two of the more important
specifications are Hardgrove grindability index and volatile content. Harder to grind coals with
lower volatile do not burn as rapidly and require a longer residence time for complete
combustion. This time is not available and so the carbon content of the ash increases. Lower
ash coals can exacerbate the problem because with less ash, the concentration of carbon in the
ash can increase while the total amount stays the same.
Increased carbon in flyash presents a number of problems to the utility. The unburned carbon
represents a loss of fuel value which can amount to a few percent of the total fuel consumption
of the boiler (at 1% loss and $40/t this represents $400,000 for every million tons of coal). The
-------
high carbon ash is also more difficult, and more expensive to dispose of. High carbon ash has a
lower bulk density, so more trucks are needed to carry the same tonnage off site, and at the same
time there is more ash to dispose of, and less fits in the land fill (land fill disposal has a
volumetric limit, not a weight bearing limit). Productive uses of high carbon flyash are quite
limited.
The largest potential use for flyash is as a pozzolanic additive in concrete where the concrete
properties can be improved. When cement hydrates half the calcium is liberated as free lime.
This makes conventional concrete quite alkaline, and susceptible to sulfate degradation if
dissolved sulfate percolates through the concrete, reacts with the lime and precipitates as
gypsum causing swelling and eventually cracking. Flyash in concrete reacts with the free lime
to form cementacious products which produce additional strength in the cured concrete. This
property of pozzolanic materials has been known for many years, and some building codes allow
the substitution of flyash (of certain grades) for a portion of the cement. Other improved
concrete properties include lower water content, lower heat of hydration, lower cost, easier
flowability, and lower permeability.
High carbon flyash is not usable in concrete. The standard ASTM C-618 specifications for
pozzolanic flyash call for less than 6 % loss on ignition. Many specific projects use much lower
specifications. For example the Boston Harbor Central Artery Project has a specification of less
than 3 % loss on ignition.
The proportion of carbon and the properties of the ash result from the composition of the coal
and the conditions it experiences during combustion. Coal is a very inhomogeneous material. It
is formed from assorted vegetation that accumulated in bogs and was protected from decay by
anaerobic, acidic water. The minerals in coal are present in several forms. The organic material
from the original vegetation contains trace minerals as well as carboxyl groups that in low-rank
coal act as cation exchange sites. These sites collect cations from water flowing through the coal
deposit. Some low rank coals have high levels of these ion exchange sites and have trapped
large amounts of alkali metals. As the coal matures and increases in rank, these carboxyl groups
decompose and produce CO2 which escapes. This is accompanied by collapse of pores in the
coal structure and a trapping of those elements with high organic affinities and many other
generally soluble elements. The other major source of ash in coal is from mineral grains that
have been mixed with the coal by the actions of water, wind and earth movements over geologic
time, such as sand and clay.
The dominant method of coal combustion is suspension burning, there are multiple types and
sizes of burners and points of introduction into the furnace. There is also wide and increasing
use of two-stage combustion with sub-stoichiometric firing at the burner with introduction of
additional air to the upper furnace for completion of combustion. All these variations together
with furnace size and configuration for internal recirculation as well as the diversity of coals
being pulverized and burned gives a broad spectrum of range of carbon in the fly ash.
When coal is mined and processed for combustion, the minerals associated with it, along with
extraneous rock and seam partings, are all pulverized down to a typical specification of about
-------
70% less than 74 microns. Pulverization liberates many of the mineral particles. The design,
state of repair, and operating adjustment of a stations pulverizers affects the actual degree of
pulverization achieved. In suspension burning the pulverized coal is dispersed in air and then
exposed to thermal radiation from the flame which thermally cracks and vaporizes organic
constituents in the coal particle leaving char which burns more slowly. It is this char,
particularly from large coal particles, which exits the furnace before completion of combustion.
The mineral grains, exposed to the heat from the flame, melt and form spheres. Pyrite is oxidized
to magnetite.
The residence time in the furnace is too short for most particles to collide and mix. Mineral
particles therefore retain their original oxide content except for being depleted in volatiles. The
liquid silicate spheres are quenched as the combustion products pass through the heat exchange
surfaces and remain in the flyash as glassy spheres.
Lime produced during the hydration of concrete can dissolve some of the amorphous silicate
particles and forms cementacious material that adds strength. This reduces the lime which can
otherwise form calcium sulfate compounds which swell and crack the concrete.
Crystalline material is not soluble in the alkali liberated during the hydration of concrete and so
merely acts as an inert filler. Ash from some combustion processes cannot be used in concrete.
Fluid bed ash is not suitable because of the alkali added for SC>2 control, but also fluid bed
combustion occurs at a lower temperature so the particles are not glassy, but instead are
crystalline. The thermal history of the ash is important in achieving ash that is pozzolanically
active.
For concrete admixture applications, the glassy silicate spheres are the important component.
Unburned carbon adsorbs the air-entraining agents which are added for freeze thaw resistance
and is undesirable. Water, sulfate and alkali are also deleterious and may not exceed 3.0, 5.0
and 1.5% expressed as H2O, SO3 and Na2O (ASTM C618). The low water content needed
precludes a wet process because of the costs of drying and also the potential for pozzolanic
reactions to occur during the wet separation process which would cement particles together,
decrease subsequent pozzolanic activity and tie up alkali with the silicate particles.
Electrostatic Separation and Contact Charging Theory
Electrostatic separators that utilize contact potential differences between materials are also
known as triboelectric separators. Triboelectric (from the Greek tribein, to rub) is a misnomer
because the electrostatic charging occurs as a result of surface contact and rubbing and the
resultant friction is unimportant except to increase the degree and area of surface contact.
All solids are held together by chemical bonds. These bonds result from the energy structure of
electrons in the material. Some electrons are held tightly in bonds, some electrons are free to
move, and some can only move if a sufficient activation energy is provided. The magnitude of
the activation energy for electron migration determines if a material is a metallic conductor, an
-------
insulator, or a semiconductor. The semiconductor industry is based on the electrical properties
that originate at the junction between different material phases.
When two materials are in contact, electrons move (or tunnel) until the energy of electrons in
each material at the interface is equalized. The material with a higher affinity for electrons gains
electrons and charges negative, while the material with the lower affinity loses electrons and
charges positively. A measure of the relative affinity for electrons is the work function, which is
the energy to move an electron from the surface to infinity (99 % of which amounts to 10"8
meters). The work function is dependent on the chemical composition of the surface. Two
surfaces of different work function in contact will exchange electrons until there is a voltage
between them equal to the difference in their work functions.
Charge transfer continues between the two materials until the developing electric field at the
interface equalizes the energy of electrons at the surfaces of the two materials. As the materials
are then moved apart, charge can no longer transfer across the intervening gap, and the two
particles now possess equal but opposite sign charges
The contact exchange of charge is universally observed for all materials, and causes the
electrostatic nuisances that are a problem in some industries. As a practical matter, two surfaces
that are ostensibly the same are still enough different that contact charging will occur. The
extreme sensitivity of the surface potential to surface differences explains why in the
semiconductor industry dopants at parts per billion can have large effects. Single crystal pure
elements can have differences in work function between different crystal faces exceeding one
volt (10). Contact charging of a surface is necessarily an averaging of charge exchange at many
points. A particle exchanges charge with every particle it contacts, and so its net charge
fluctuates around an average which depends upon its neighbors.
The work function of a material is the energy required to remove an electron, and is dependent
upon how tightly electrons are held in chemical bonds. A measure of an element's tendency to
attract electrons is termed electronegativity. Elements with high electronegativity, such as
fluorine, oxygen and sulfur, strongly attract electrons. It has been found that the work function
of a material is related to its electronegativity, and that the work function of a compound can be
estimated from the electronegativity of its atomic constituents and the degree of electron sharing
occurring in the chemical bonds, which allows prediction of work function and allows
estimation of separation characteristics of minerals based on their composition. A Lewis acid is
defined as an electron receptor, and a Lewis base is an electron donor. Observed separations of
minerals are consistent with work function values estimated from electronegativity and bonding.
Oxidized materials tend to charge negatively, as do acidic materials. Coal, being a
hydrocarbon, charges positively with respect to most metal oxides. Oxides such as silica and
sulfides such as pyrite charge negatively with respect to coal, and are readily separable
electrostaticly.
-------
Process Description
Separation Technologies, Inc. (STI) has developed a triboelectric separation process with a new
patented separator geometry that solves scale-up problems and improves electrostatic separation
by 4 or more orders of magnitude, compared to prior technologies. This dramatic improvement
has now allowed the economical dry separation of materials inseparable by any other means.
In the STI separator the material is fed into the thin gap between two parallel plane electrodes
(see Figure 1). The particles are then swept up by a moving open mesh belt and conveyed in
opposite directions. The moving belt sets up a counter current plane couette flow field
independent of the electric field. The particles charge by contact with their neighbors. The
electric field moves opposite charged particles either up or down depending on their sign of
charge. The field need only move the particles a tiny fraction of an inch to move a particle from
a left-moving to a right-moving stream. The belt moves the particles adjacent each electrode
toward opposite ends of the separator. The counter current flow of the separating particles
provides for multiple opportunities for separation and results in excellent separation and
recovery.
The separator is very simple. The belt and the associated rolls are the only moving parts. The
electrodes are stationary and are composed of an appropriately hard material. The belt is an
open mesh polymer. The over all separator length depends on the specific application, but is in
the range of 20 feet plus room for rolls and hoppers. The belt speed is the same order as that of
pneumatic conveying systems. The power consumption of the electrostatic power supplies is
negligible. The power consumption of the belt drive is on the order of 1 kW hr per ton
processed.
The short gap, the very high field, the counter current flow, the vigorous particle-particle
agitation and the self-cleaning action of the belt on the electrodes are the critical features of the
STI separator that give it such good performance. The overall separation performance of the STI
separator is superior to all previous contact-charging separators that have been considered for
commercial use in terms of:
1. Broadness of feed particle size distribution:
The STI separator can process material with a particle size from greater than 250 micron
to less than 5 micron with no degradation in performance.
2. Electrical conductivity of feed particles:
The STI separator can contact charge and triboelectrically separate materials irrespective
of their electrical conductivity, from good insulators (dielectrics) to good conductors
(metals).
3. Low power consumption:
The STI separator has demonstrated at full scale (25 tons per hour) a power consumption
of less than 1 kilowatt hour per ton of feed.
-------
4. Scalability:
The capacity of the STI separator is a linear function of separator width. A separator has
been operated for thousands of tons at rates up to 25 tons per hour.
Laboratory Results
Results from a large number of tests using a several ton per hour laboratory prototype are
summarized in Figure 2 where the carbon content of the two product streams is plotted verses
the feed carbon content. Many samples of ash from many different sources has been tested with
very good results. This includes ash derived from coal from North America, South America,
Europe, Africa, Asia and Australia
Commercial Results:
Results from a 3000 ton production test run in summer of 1994 are shown in Figure 3. This
figure shows the average LOI for the feed and the two products for each truck load of low-
carbon ash produced. The average of the feed, low-carbon and high-carbon over the 3000 tons
of ash product were 8.19%, 2.26%, and 35.9% LOI respectively. This represented an ash yield
of 83%. All the low carbon ash that was produced was used in concrete projects in the Greater
Boston area.
In other tests, thousands of tons of ash from 6 different coal fired units burning several types of
coal from North and South America has been processed on the commercial sized unit with good
results.
The balance of plant bringing ash to the separator and taking the two products away has recently
been upgraded. There is now provision for filling 3 trailers with high and low carbon ash before
moving trailers. Production rate has been increased with over 1000 tons being processed in the
past 10 days. We expect to get to over 1000 tons of production per week in the next few months
as further modifications are done to the balance of plant to increase the feeding capacity.
Combustion tests of the high carbon material have been done, and preliminary indications do
not show any difficulty with reburning the high carbon material.
Implications and Conclusions:
1. Carbon level in flyash can be controlled independent of furnace operations. This means that:
- Operators have greater latitude in controlling boiler operations including excess air, heat
rate, firing rate, turndown, and supplemental firing with out incurring carbon in ash penalty.
- Harder coals with lower volatile content can be used, allowing greater flexibility of fuel
source.
- Reduced need for post combustion NOx treatment.
-------
- Boilers can be operated to reduce NOx emissions still more, generating credits (if
applicable).
2. Carbon level in flyash can be reduced with only post combustion processing. This means:
- No plant downtime to install retrofits.
- No boiler modifications.
- No need to derate.
- No need to switch fuels.
- No pulverize retrofit.
- No loss in efficiency.
3. Flyash can be utilized in concrete instead of being placed in landfills. This means:
- No landfill cost.
- Revenues from flyash sales.
- No potential long term liability associated with a landfill.
4. Utility has a "green" environmentally beneficial solution to a current problem.
- Flyash byproduct is recycled, replacing cement, while producing more durable, longer life
concrete.
- Reduced cement production means less minerals mined, less CO2 emitted.
- Per ton of ash used to replace cement, 2 tons less minerals mined, and 1 ton less CO2
emitted.
- Utility can become one of the largest, if not the largest recycler (on a weight, and percentage
basis) in the community.
- Implementing a non-mandated recycling program can promote community good will.
-------
FEED
Positive Electrode
Negative Electrode
Figure 1
STI Separator
-------
Figure 2
100.00
90.00 --
80.00
o
-Q 70.00
i_
(0
O
60.00
.2 50.00 +
c
.0
*3
"E
40.00 -
30.00 - -
20.00 --
10.00 --
high carbon
low carbon ash
Linear (high carbon)
Linear (low carbon ash)
5.00 10.00 15.00 20.00 25.00 30.00 35.00 40.00
Loss on Ignition (% Carbon) Feed Fly Ash
-------
1 4 7 10 13 16 19 22 25 28 31 34 37 40 43 46 49 52 55 58 61 64 67 70 73 76 79 82 85
Trucks
91 94 97100103106109112115118121
Figure 3
3000 Ton Production Test
-------
NOX ADVISOR: INTELLIGENT
SOFTWARE FOR COMBUSTION OPTIMIZATION
J. Pfahler
D. Eskenazi
E. Levy
N. Sarunac
S. Ahmed
L. Burke
Energy Research Center
Lehigh University
117 ATLSS Drive
Bethlehem, Pennsylvania 18015
S. Williams
Potomac Electric Power Company
8711 Westphalia Road
Upper Marlboro, Maryland 20772
E. Petrill
Electric Power Research Institute
3412 Hillview Avenue
Palo Alto, California 94303
Abstract
Under a Tailored Collaboration Project funded by PEPCO and EPRI, Lehigh University's Energy
Research Center has been developing software for use by plant personnel to optimize a pulverized
coal-fired boiler to reduce NOX emissions and minimize heat rate. Based on expert systems,
neural networks and a mathematical optimization algorithm, the NOX Advisor uses the expert
system to safely guide the plant engineer through a series of parametric boiler tests, gathering a
database to characterize the operation of the boiler over a wide range of conditions. The neural
network develops non-linear mappings between NOX, heat rate and the controllable parameters.
These are then used by the mathematical optimization algorithm to identify optimal operating
conditions. The Advisor can be used to adjust a boiler to reach minimum NOX or identify the
conditions that give minimum heat rate subject to a target NOX value. The Advisor operates off-
line on a PC platform, gathering data automatically from the plant data highway and/or manually
as needed. Results of a field trial at PEPCO's Potomac River Station are described.
-------
Introduction
Studies performed over the last few years on pulverized coal boilers show that NOX emission
levels can usually be reduced rather substantially through adjustments to the boiler operating
parameters. For example, in the case of a corner-fired boiler with conventional burners, NOX is
sensitive to furnace 02 level, mill loading patterns, burner tilt angle, primary air velocity, and fuel-
air and auxiliary-air damper settings. There can be undesirable side effects of reducing NOX.
Generally as the NOX is reduced, unburned carbon increases, and for some boiler designs steam
temperatures drop. This results in increases in heat rate. For units with electrostatic
precipitators, increases in unburned carbon can lead to opacity problems. Furthermore, changes
in burner settings must be made carefully to avoid unsafe operating conditions or conditions that
lead to maintenance problems. Because of the complex relationships between the controllable
parameters and the effects of these on emissions, plant performance, plant safety and maintenance,
the task of identifying the best set of operating conditions at any particular load level can be
difficult. With funding from the Electric Power Research Institute (EPRI) and the Potomac
Electric Power Company (PEPCO), Lehigh University's Energy Research Center is developing
software, referred to as the NOX Advisor, for assisting plant personnel in optimizing the
combustion of pulverized coal boilers to safely reduce NOX emissions and optimize heat rate.
The NOX Advisor combines an expert system, neural networks and an optimization algorithm
along with a graphical user interface into a single program. It guides the plant engineer safely
through a series of parametric boiler tests, collects and models the data, and then determines the
optimal operating conditions. The program can be used to find either the minimum NOX emissions
rate, or the minimum heat rate subject to a target NOX level.
The NOX Advisor's operating strategy is based on experience gained from extensive testing at
PEPCO's power plants. For example, in the case of Potomac River Unit 4, a comer-fired unit
with conventional burners, Lehigh and PEPCO investigators reduced NOX emissions by close to
40 percent using the combustion optimization approach (1). Full load testing at Dickerson Unit 3,
which is also a corner-fired unit with conventional burners, showed that NOX reductions of close
to 30 percent were attainable (2). Using a similar approach, the same investigators have been
optimizing combustion at Morgantown Unit 2, which is equipped with an ABB-CE LNCFS III
low NOX firing system. The results from the Morgantown project are reported in a separate paper
at this conference (3).
The present paper describes a version of the NOX Advisor that was developed to optimize boilers
similar in configuration to those at the Potomac River and Dickerson Generating Stations. These
units have tangentially-fired pulverized coal boilers with conventional burners and no overfire air.
A more general version of the NOX Advisor, which can handle corner-fired boilers with either
conventional or low NOX burners, is under development and will be described in a future paper.
Software Description
The NOX Advisor is written in C++ and runs under Microsoft Windows on an IBM-compatible
PC. The main features of the program are an expert system to direct the parametric boiler testing,
-------
neural networks to model the data, an optimization algorithm to determine the best operating
conditions, and a graphical user interface. These components are shown schematically in
Fioiirp 1
Figure 1.
Plant Archive
Data (PMW)
Boiler Control
Settings
Expert System
S~Neural Network i
1 . i
] Optimization'
I Algorithm (
Personal Computer
Figure 1
Schematic of the NOy Advisor.
Expert System
The expert system is used to guide the plant engineer safely through a series of parametric boiler
tests. Expert systems are rule-based, incorporating engineering knowledge gained from past
experience. An example of a rule is "to reduce NOX, reduce economizer O2." Such a system
allows the program to react to changing conditions in much the same way as would an engineer
experienced in combustion optimization testing. The results of each test are analyzed and used to
recommend the settings for the next test. By following this procedure, the expert system is able
to minimize the total number of tests while still accumulating sufficient data for the neural
network/optimization routines. The expert system was programmed using CLIPS version 6, an
expert system development tool created by the NASA Software Technology Branch at the
Lyndon B. Johnson Space Center (4).
Neural Networks and Optimization
After the testing is complete, the software analyzes the data using neural networks. The resulting
model is then passed to an optimization algorithm that finds the best operating conditions
according to the user's desired objective.
-------
Neural networks are a class of mathematical algorithms that simulate the operation of biological
neurons. The neural network learns the relationships between the operating conditions and
emission and performance parameters by processing the test data. In the training process, a
neural network develops a complex non-linear function that maps the system inputs to the
corresponding outputs. This function is then passed to a mathematical minimization algorithm
that finds the optimal operating conditions.
The neural networks are created and trained using software referred to as Designer Pack by
NeuralWare (5). The optimization algorithm is based on the Nelder and Meade Downhill Simplex
Method (6).
User Interface
The user interface ties together the expert system, neural networks, optimization algorithm, and
data communications into a single program. The expert system and neural network software are
compiled as dynamic link libraries that are used as needed. All communication between the user
and the program is provided for by the interface through a series of self-explanatory dialog
screens. The key advantage of this approach to software development is that the expert system,
neural networks and optimization algorithm are invisible to the user. Therefore, the user needs no
special knowledge to use these advanced technologies.
The interface also provides the NOX Advisor with the ability to acquire data automatically from
the plant's data archives. The Potomac River Generating Station uses EPRI's Plant Monitoring
Workstation (PMW) on a micro-VAX computer to record the plant data. To enable the NOX
Advisor to be tested at Potomac River with automatic data retrieval, a data link was developed
specifically for communicating with PMW through a modem. Similar data links will be created in
the future to allow the NOX Advisor to access other types of data archiving systems. The
interface also allows for the manual input of test data for those parameters not available on the
data highway.
The data collected from each test are stored in a database that is accessible to the user for viewing
tests results and is also used to provide a graphing capability. The user is able to plot the test
results at any time during the testing process. Single or multiple parameters can be displayed as a
function of the test number or any other parameter.
Program Operation
When initiating a test program with the NOX Advisor, the user is asked a series of questions
concerning unit characteristics and the nature of the planned tests. For example, the user will
need to specify the load range to be tested and the objective of the optimization: to minimize the
NOX emissions rate or to minimize heat rate subject to a target NOX emissions rate.
The NOX Advisor then proceeds through a series of parametric tests, recommending operating
conditions important to NOX reduction and unit performance. Based on previous testing
experience with conventional corner-fired boilers, these parameters for full-load are economizer
-------
02 level, burner tilt angle, mill loading pattern, and secondary air damper positions. At the end of
each portion of the parametric test series, the expert system estimates the best setting for the
particular parameter being tested. This new setting is then used as the starting point for the next
test series. Using this method, the NOX emissions rate is gradually reduced as the control settings
are moved towards an optimum. This approach also leads to the collection of more test data near
the optimal conditions, allowing for better predictions by the neural networks.
Upon completion of parametric testing, the test data are passed to the neural network routine for
modeling. Various neural network configurations are evaluated and the most accurate models
are saved. The functions generated by these networks are then used by the optimization algorithm
to calculate the optimal operating conditions. The solution is checked to ensure it is realistic and
that none of the boiler safety limits would be violated. Once the solution is verified, the final set
of optimal conditions are displayed for the user.
Results of Field Trial
The NOX Advisor was tested at PEPCO's Potomac River Unit 5, a 108 MW, subcritical, CE
tangentially-fired four-comer boiler, with a GE single-reheat turbine. All of the boilers at the
Potomac River Station have conventional burners. The testing, conducted at the upper end of the
load range, with four mills in operation, was intended to find the minimum heat rate subject to a
target NOX of 0.42 Ib/MBtu. A total of 56 tests was run over the course of seven days of testing.
For Unit 5, the test parameters were varied in the following sequence:
• Economizer O2.
• Burner tilt angle.
• Combined economizer 02 and burner tilt angle.
• Mill loading pattern.
• Auxiliary air damper positions.
• Fuel air damper positions.
• Combined auxiliary air damper positions and burner tilt angle.
For this unit, the NOX Advisor required 39 input parameters. Half of these parameters were
available on-line from PMW and included unit load, main and hot reheat steam temperatures,
burner tilt angle, and various mill parameters. The remaining parameters were recorded from
instruments in the control room and CEM system. Unburned carbon was entered into the
database after a laboratory analysis of the fly ash for loss on ignition (LOI). The net unit heat rate
was entered manually into the database, after being calculated with EPRI's HEATRT code (6).
An overall progression of the test procedure is shown hi Figure 2, where NOX and heat rate are
plotted as a function of test number. This figure illustrates a key result of the optimization
scheme: NOX decreases as testing proceeds. Vertical dividing lines on the graph indicate when
each control parameter was being varied. This is shown as a different way hi Figure 3. First,
economizer O2 was varied while the other parameters were held constant. The same process was
repeated for burner tilt angle and the other parameters, according to the previously listed testing
sequence.
-------
0.70
0.65
0.60 -•
= 0.55
m
0.50 --
0.45 --
0.40 --
0.35 • •
0.30
Bumertilt/Sec. Air'"
Sec. Air
Bumertilt
\/v
J ^
10
20
30
Test Number
40
50
9200
9150
-- 9100
- - 9050
- - 9000
- • 8950
"Final" ..
5
8900
- - 8850
8800
60
Figure 2
NOX emissions and heat rate as a function of test number
for Unit 5. Tests were conducted at constant main steam flow rate.
* *-» » OO-fr-OOO «-C « * O-fr O-O-O-^-D « « *-« O
9
Fuel Air damper position: Range 3 - 5 q rf 4 4
V' V
ameters
ra
Q.
O
O
O
•o
0)
ra
o
CO
(
X X-X-X-X-X-X-X X X-X-X-X* X-X-X-X-XX-X XX-X *< X-X-X-X-X •»-* X .X, X X- X -X -X-X * X-X-X X X-X.
' Auxilary Air bias Range 0.50 - 0.75
1
X
^ A S %
' A' Top mill loading: Range 13% -25%
i'
X • . Bumertilt angle: Range -10- +20 degrees
W r M
V> X / ' * \ Economizer O2 Range 1.7% -3.0%
^^ / V- __^» 4 1 * • • « / \ - - ^-- ^ . -^ j A-J^*. . . .
^yj ^-Sy^- • • • » • •• •-*• — ^-r' • *»•*• » '•* *-*-v^»->~«-«
3 10 20 30 40 50
Test Number
X
K-K-X*
Ta
*
'V
I"*-*"***
6t
Figure 3
Systematic parametric testing of the control parameters for Unit 5.
Each parameter is scaled for the test range shown on the graph.
-------
In Figures 4 to 7, the detailed variations of each control parameter are shown along with their
impact on NOX and heat rate. A reduction of approximately 1 percent in economizer 02 resulted
in a 0.12 Ib/MBtu reduction in NOX and a 110 Btu/kWh increase in heat rate (Figure 4). Figure 5
shows the results of the parametric tests on burner tilt angle. NOX displays a parabolic shape with
respect to burner tilt angle with a minimum occurring at approximately 0°, while heat rate shows
a linear increase as tilt decreases. The increase in heat rate with decreasing tilt angle was due to
decreases in main steam and hot reheat steam temperatures. The effect of reducing coal flow rate
through the top mill is depicted in Figure 6, where the maximum achievable reduction in flow
yielded a 0.02 Ib/MBtu reduction in NOX. Results are given in terms of percentage of total coal
flow through the top mill. As the flow is reduced through the top mill, the remainder is evenly
distributed among the three lower mills. A value of 25 percent corresponds to even loading of all
four mills.
Figure 7 shows the effects of the vertical distribution of air to the auxiliary air registers. The
auxiliary air damper bias parameter, a measure of how the air is distributed about the centerline of
the boiler, can range in value from -1 to 1. Negative values indicate more air flow to the bottom
half of the boiler, 0 is for even distribution, and positive values indicate more air flow to the top
half of the boiler. In the range tested, the air flow was biased upwards, producing an overfire air
effect, which reduces NOX.
The data in Figures 4 to 7 illustrate how NOX and heat rate changed as the testing proceeded, with
the control parameters being varied one at a time. However, to use these data to determine the
combination of control settings which yields the lowest NOX or the best heat rate, relationships
must be established between NOX and heat rate and all of the control settings. This is done within
the NOX Advisor using neural networks. The relationships developed by the neural networks are
then used to determine the optimal control settings. Figure 8 was created by running the neural
network/optimization routine for different values of target NOX and is a plot of predicted
minimum heat rate as a function of target NOX. This figure can be used to analyze the economic
trade-offs associated with operating at a given target NOX by showing the related heat rate
penalty.
Figures 9 to 12 show the recommended control settings required to achieve a specified target
NOX. These figures show that for this boiler, as the target NOX is reduced, the burner tilt angle is
decreased, the top mill is unloaded more, the auxiliary air flow is biased to the upper half of the
furnace, and air flow to fuel air dampers is reduced. In all cases the economizer O2 is driven to
the lowest acceptable value, which was found to be 1.9 percent.
Table 1 lists the control settings, NOX and heat rate for the baseline test, the final test and the
neural network/optimization routine predictions. The baseline test was a single test conducted at
the normal operating conditions for full load. The settings for the "final test" were selected by the
engineer at the end of the testing sequence based on the test data and his institution of what
-------
0.65
0.60
S 0.55
CQ
0.50
0.45
0.40
1.5
1.7
• NOx
n Heat Rate
NOx curve fit
Heat rate curve fit
1.9
2.1
02, %
2.3
2.5
9100
+ 9080
9060
+ 9040
9020 iE
i "
+ 9000 0-
ra
8980 ^
ra
0)
-r 8960
8940
8920
8900
2.7
Figure 4
Full-load parametric testing of the economizer O2 at Unit 5.
0.65
0.60
0.45
0.40
-10
NOx
Heat Rate
Heat rate curve fit
NOx curve fit
-5 0 5
Burner Tilt angle, degrees
4- 9000
+ 8960
8940 i£
9020
8980
+ 8920
m
8900 £
ra
0)
8880
8860
8840
8820
10
15
Figure 5
Full-load parametric testing of the burner tilt angle at Unit 5.
-------
0.65
0.6
2 0.55
m
X
O
0.5
0.45
0.4
» NOx
n Heat Rate
— —Heat rate curve fit
NOx curve fit
9100
-•9080
9060
9040
9020 :£
3
CD
8980
re
a
-- 8960 Z
8940
-- 8920
8900
12.0 14.0 16.0 18.0 20.0 22.0 24.0
Top Mill Load, % of Total flow
26.0
Figure 6
Full-load parametric testing of the mill loading pattern for Unit 5.
0.65
9180
0.40
0.50 0.55 0.60 0.65 0.70
Auxiliary Air Damper Bias Parameter
8980
0.75
Figure 7
Full-load parametric testing of the vertical distribution of auxiliary air.
-------
9200
9150
,
OJ
&
"Jo
9100
9050
9000
0.40
0.42 0.44 0.46
Target NOx (Ib/MBtu)
0.48
Figure 8
Predicted minimum heat rate as a function of target NOX.
0.50
15
O)
0>
gMO
m
T3
S 0
E
o
o
CD
-5
0.40
0.42 0.44 0.46
Target NOx (Ib/MBtu)
0.48
0.50
Figure 9
Recommended burner tilt angle to achieve a target NOX
-------
25%
0.42
0.44 0.46
Target NOx (Ib/MBtu)
0.48
0.50
Figure 10
Recommended top mill loading to achieve a target NOX.
0.75
CO
s°-
ro
lo,
•o
CD
E
o
g o.
a:
0.50
).40 0.42 0.44 0.46
Target NOx (Ib/MBtu)
0.48
0.50
Figure 11
Recommended auxiliary air bias to achieve a target NOX
-------
0.40 0.42 0.44 0.46
Target NOx (Ib/MBtu)
0.48
0.50
Figure 12
Recommended fuel air damper position to achieve a target NOX.
the control settings should be for minimum NOX. The neural network/optimization routine
predictions are for the cases of minimum NOX and minimum heat rate subject to a target NO of
0.45 Ib/MBtu.
The predicted settings for minimum NOX are similar to those of the "final test" based on
engineering intuition. The NOX Advisor predicted a 0.27 Ib/MBtu reduction from the baseline in
the NOX emissions rate. The unit heat rate was predicted to increase 203 Btu/kWh, due in large
part to a drop in steam temperatures. With the testing being performed at constant main steam
flow rate, the decrease in steam temperatures also resulted in a 5 MW load reduction.
Comparisons between these predictions and test results obtained at an identical unit (Potomac
River Unit 4) show close agreement (see Tables 1 and 2). Further testing is planned at Unit 5 to
determine if the recommended control settings shown in Table 1 actually result in the predicted
NOX and heat rate values.
Benefits of the Parametric Testing Approach
The use of a parametric testing approach yields additional benefits besides gathering the test data
needed for the neural network. As these tests are conducted, the test engineer can view and plot
the data in a variety of ways. This will improve the engineer's understanding of how NO and
performance vary with boiler settings. By doing this the engineer develops the physical insight
needed to overcome constraints on equipment that limit the ability to reduce NO or optimize heat
-------
Table 1
Control settings, NOX Emissions, and Heat Rate
for Baseline and Final Test, and NOX Advisor Predictions
of total coal flow*
Auxiliary Air**
Fuel Air**
NOX, Ib/MBtu
Heat rate, Btu/kWh
Baseline
Final Test
Predicted
Min. NOV
Predicted Min.
Heat Rate, NOX
= 0.45 Ib/MBtu
Economizer O2, %
Burner Tilt Angle, °
Top mill load, %
3.02
20
25
1.86
-1
19.0
1.9
-3
14.6
1.9
+3
17.7
5.5.5.5.5 5-5-2-2-2
5-5-5-5 4-4-4-4
0.69
8959
0.42
9093
5-5-2-2-2
4-4-4-4
0.42
9162
5-4-4-2-2
5-5-5-5
0.45
9100
* Top mill loading is displayed as a % of total coal flow. The remaining coal flow is distributed
evenly among the other three mills.
** Damper position listed from top of boiler. 5 indicates fully open and 1 indicates fully closed.
rate. For example, these constraints might include high unburned carbon, low steam
temperatures, elevated opacity, and furnace to furnace imbalances. By correcting these
deficiencies, the engineer will be able to broaden the operating envelope and achieve lower NOX
and heat rate. This enhanced understanding will also be useful in diagnosing problems as they
arise in the future.
Even more important, using systematic parametric tests increases the safety margin by allowing
the engineer to identify and avoid unsafe settings. Thus, the possibility of undesirable conditions,
such as flame attachment to the burner or an unstable flame, is greatly reduced.
Parametric testing also generates a database covering a wide range of operating conditions. This
database can then be used to perform economic trade-off studies on unit operating strategies.
Even more important, it provides the knowledge needed to re-optimize the boiler with different
conditions and constraints, all without further testing.
-------
Table 2
of total coal flow*
Auxiliary Air**
Fuel Air**
NOX, Ib/MBtu
Control Settings and NOX
Emissions from Unit 4 Test Results
Baseline
Min. NOX
4-4-3-3-2
5-5-5-5
0.62
5.4.3-2-2
3-4-4-4
0.39
NOX = 0.45 Ib/MBtu with
Minimum Heat Rate
Economizer O2, %
Burner tilt angle, °
Top mill load, %
3.1
19
25
1.9
+7
13.8
2.0
+14
25
5-4-3-2-2
4-4-4-4
0.45
* Top mill loading is displayed as a % of total coal flow. The remaining coal flow is distributed
evenly among the other three mills.
**
Damper position listed from top of boiler. 5 indicates fully open and 1 indicates full closed.
Finally, the database allows the NOX Advisor to be used as a training tool. General correlations
giving NOX and heat rate as functions of control settings can be developed and then used by the
Advisor in a training mode. The training mode allows simulation of plant testing for educating
engineers and plant operators on NOX and heat rate reduction techniques.
Summary
The NOX Advisor, an intelligent software system for combustion optimization has been
successfully tested at full load at PEPCO's Potomac River Unit 5. These tests show that the NOX
Advisor is capable of conducting a series of parametric tests, collecting and analyzing the data,
and determining realistic optimal points. At full load conditions, 56 tests were run over the
course of even days of testing. The minimum NOX level found by the Advisor was 0.42 Ib/MBtu,
which is 39 percent below the baseline. These settings increased heat rate by 2.3 percent from the
baseline and reduced load by 5 percent due to reductions in steam temperature. The Advisor can
also determine the boiler settings required to minimize heat rate, subject to a target NOX level.
For example, at 0.45 Ib/MBtu, these yield a 141 Btu/kWh increase in heat rate over the baseline.
-------
The NOX Advisor is currently being generalized to include a wide variety of tangentially-fired
boiler configurations, including boilers with conventional or low NOX firing systems with various
numbers of mills. In addition, planning is underway for a version which can handle wall-fired
boilers.
The NOX Advisor utilizes a parametric testing approach to gather the data needed to perform the
optimization. The use of this method yields some important additional benefits to the utility. By
having the plant engineer involved in the testing, he/she develops a physical understanding of how
NOX and heat rate depend on the control settings. This can help the engineer overcome
constraints on equipment that limit the ability to reduce NOX or heat rate. It provides the physical
insight needed for diagnosing problems which may arise in the future and it increases the safety
margin by allowing the engineer to identify and avoid unsafe settings. This approach also
generates the database needed for performing economic trade-off studies on unit operating
strategies and for re-optimizing the boiler with different conditions and constraints, all without
further testing.
Acknowledgments
This work is sponsored under a Tailored Collaboration Project between PEPCO and EPRI. The
authors are grateful for Dave Cramer's (PEPCO) assistance with the testing at Potomac River
Generating Station.
References
1. E. Levy, et al., "NOX Control and Performance Optimization through Boiler Fine-Tuning,"
Paper presented at 1993 EPRI/EPA Joint Symposium on Stationary Combustion NOX
Control, Miami, FL (May 1993).
2. Law-NOx Operation at Dickerson Station Unit 3 Using the Boiler Tuning Technique: A
Feasibility Study. Bethlehem, PA: Lehigh University, Energy Research Center, October
1993. 93-400-23-38.
3. P. Maines, et al., "Combustion Optimization of Low NOX Burners at PEPCO's
Morgantown Station," Paper presented at 1995 EPRI/EPA Joint Symposium on
Stationary Combustion NOX Control, Kansas City, MO (May 1995).
4. CLIPS User's Guide, Lyndon B. Johnson Space Center, Information Systems Directorate,
Software Technology Branch, May 1993. CLIPS Version 6.0, NASA.
5. T. Ragsdale, et al. Engineering Optimization Methods and Applications. New York,
Wiley and Sons, 1983.
6. HEATRT: PC Software for Performance Analysis of Pulverized Coal Boilers. Palo
Alto, CA: Electric Power Research Institute Technical Brief, October 1992. TB 101176.
-------
Session 6B
Selective Catalytic Reduction
-------
Demonstration of Selective Catalytic Reduction (SCR) Technology for the
Control of Nitrogen Oxides (NOx) Emissions from High Sulfur Coal-Fired
Utility Boilers at Plant Crist SCR Test Facility
W. S. Hinton and J. D. Maxwell A. L. Baldwin
Southern Company Services, Inc. U.S. Department of Energy
Birmingham, Alabama 35209 P.O. Box 10940
Pittsburgh, Pennsylvania 15236
Presented for: EPRI/EPA 1995 Joint Symposium on Stationary Combustion NOx Control
Kansas City, Missouri
May 16-19, 1995
ABSTRACT
This paper describes the status of the Innovative Clean Coal Technology project to demonstrate
SCR technology for reduction of NOx emissions from flue gas of utility boilers burning U.S.
high-sulfur coal. The project is sponsored by the U.S. Department of Energy, managed and co-
funded by Southern Company Services, Inc. on behalf of the Southern Company, and also co-
funded by the Electric Power Research Institute and Ontario Hydro; and is located at Gulf Power
Company's Plant Crist Unit 5 (75 MW tangentially-fired boiler burning U.S. coals that have a
sulfur content near 3.0%), near Pensacola, Florida. The test program is being conducted for
approximately two years to evaluate catalyst deactivation and other SCR operational effects. The
SCR test facility has nine reactors: three 2.5 MW (5000 scfm), and six 0.2 MW (400 scfm). Eight
reactors operate on high-dust flue gas, while the ninth reactor operates on low-dust flue gas using
a slip stream at the exit of the host unit's hot side precipitator. The reactors operate in parallel
with commercially available SCR catalysts obtained from vendors throughout the world.
Long-term performance testing began in July 1993. A general test facility description and the
results from three parametric test sequences and long term test data through December 1994 are
presented in this paper.
INTRODUCTION
Selective catalytic reduction (SCR) is a process in which ammonia is added to the flue gas to
reduce NOx to nitrogen and water over a catalyst. The need within the utility industry for
detailed information on SCR technology has never been greater. The 1990 U.S. Clean Air Act
Amendments (CAAA) create two new nitrogen oxide (NOx) control requirements on fossil fuel-
fired utility boilers. First, Title IV of the CAAA regarding acid rain addresses NOx emission
limits on all coal-fired utility boilers. Second, Title I of the CAAA (attainment of the ambient air
-------
quality standards) calls for certain areas presently not in attainment on ozone to consider NOx
controls to achieve attainment. As a result, renewed focus has been placed on advanced NOx
control technologies such as SCR, which may be required to meet compliance requirements
associated with the CAAA.
SCR technology is in commercial use in Japan and Western Europe on gas-, oil-, and low-sulfur
coal-fired power plants. There are now over 36,000 MW of fossil-fuel-fired SCR capacity in
Japan, including 6,200 MW on coal. There are over 33,000 MW of fossil-fuel-fired SCR capacity
in Western Europe, including 30,500 MW of coal-fired capacity.1
SCR DEMONSTRATION GOALS
Although SCR is widely applied in Japan and Western Europe, numerous technical uncertainties
are associated with applying SCR to U.S. coals. These uncertainties include:
(1) potential catalyst deactivation due to poisoning by trace metal species present in U.S.
coals but not present, or present at much lower concentrations, in fuels from other
countries;
(2) performance of the technology and effects on the balance^of-plant equipment in the
presence of high amounts of SO2 and SO3 (e.g., plugging of downstream equipment
with arnmonia-sulfur compounds); and
(3) performance of a wide variety of SCR catalyst compositions, geometries and
manufacturing methods at typical U.S. high-sulfur coal-fired utility operating conditions.
These uncertainties are being explored by constructing and operating a series of small-scale SCR
reactors and simultaneously exposing different commercially available SCR catalysts to flue gas
derived from the combustion of high-sulfur U.S. coal. First, SCR catalyst performance is being
evaluated for two years under realistic operating conditions found in U.S. pulverized-coal-fired
utility boilers. Deactivation rates for the catalysts exposed to flue gas of high-sulfur U.S. coal are
being documented to determine catalyst life and associated process economics. Second,
parametric tests are being performed during which SCR operating conditions are being adjusted
above and below design values to observe deNOx performance and ammonia slip. The
performance of air preheaters installed downstream of the larger SCR reactors will also be
observed to evaluate the effects of SCR operating conditions on heat transfer and boiler
efficiency. Third, Honeycomb- and plate-type SCR catalysts of various commercial compositions
from the U.S., Japan, and Europe are being evaluated. Tests with these catalysts will expand
knowledge of the performance of SCR catalysts under U.S. utility operating conditions with
high-sulfur coal.
The intent of this project is to demonstrate commercial catalyst performance and to determine
optimum operating conditions and catalyst life for the SCR process. This project will also
-------
demonstrate the technical and economic viability of SCR while reducing NOx emissions by at
least 80% and maintaining acceptable ammonia slip (5 ppm).
SCR TEST FACILITY DESCRIPTION
The SCR demonstration facility is located at Gulf Power Company's Plant Crist in Pensacola,
Florida. The facility treats a flue gas slip-stream from Unit 5, a commercially operating 75-MW
unit, firing U.S. coals with a sulfur content near 3.0%. Unit 5 is a tangentially-fired, dry bottom
boiler with hot- and cold-side electrostatic precipitators (ESPs) for paniculate control. The SCR
test facility consists of nine reactors operating in parallel for side-by-side comparisons of
commercially available SCR catalysts obtained from vendors throughout the world. With all
reactors in operation, the amount of combustion flue gas that can be treated is 17,400 scfin or
12% of Unit 5's capacity (about 8.7 MWe). Table 1 shows typical pilot facility flue gas
constituent concentrations and paniculate loadings.
The process flow diagram for the SCR test facility is shown in Figure 1. There are three large
SCR reactors (2.5 MW, 5000 scfrn) and six smaller SCR reactors (0.2 MW, 400 scfin). Eight of
the nine reactors operate with flue gas containing full paniculate loading (high dust) extracted
from the inlet duct of the hot-side ESP, while one small reactor uses flue gas fed from the ESP
outlet (low dust). Only eight reactors are now being operated, one high-dust small reactor is idle.
Each reactor train has electric duct heaters to control the temperature of the flue gas entering the
reactor and a venturi flow meter to measure the flue gas flow. An economizer bypass line to the
SCR test facility maintains a minimum temperature of 620 °F for flue gas supplied to the test
facility. Anhydrous ammonia is independently metered to a stream of heated dilution air that
injects the ammonia via nozzles into the flue gas stream upstream of each SCR reactor. The flue
gas and ammonia pass through the SCR reactors, which have the capacity to contain up to four
catalyst layers.
For the large reactor trains, the flue gas exits the reactor and enters a pilot-scale air preheater
(APH). The APHs are incorporated hi the project to evaluate the effects of SCR reaction
chemistry on APH deposit formation and the effects of the deposits on APH performance and
operations. All reactor trams, except the low-dust train, have a cyclone downstream of the SCR
reactor to protect the induced draft (ID) fans from particulates. The exhaust for all the SCR
reactors is combined into a single manifold and reinjected into the host boiler's flue gas stream
ahead of the cold-side ESP. The preheated air from the APH on the large reactors is also
combined into a single manifold and returned to the host boiler draft system at the air outlet of the
existing APH. All of the paniculate that is removed from the flue gas with the cyclones is
combined and sent to an ash disposal area.
-------
AIR PREHEATER TESTING PLANS
Air preheater tests include a number of performance measurements aimed at determining the
effects of SCR installations on downstream air preheaters. These tests include pressure drop
tracking, sootblowing efficiency tests, general heat transfer analysis, washing efficiency and
requirements, corrosion analyses, and basket weight loss. By nature, these air preheater tests are
long-term. Included in the project final report will be details of expected washing requirements,
sootblowing requirements and basket replacement requirements for the operation of air preheaters
in conjunction with SCR installations.
CATALYST TESTING PLANS
Six catalyst suppliers are participating in this project, providing eight different catalysts. The
suppliers, corresponding reactor size, and catalyst configuration are listed in Table 2. The two
suppliers from Europe and two from Japan provide one catalyst each. The two U.S. firms are
supplying four of the catalysts. The catalysts being evaluated represent the wide variety of SCR
catalysts being offered commercially and possess different chemical compositions and physical
shapes. Of these eight catalysts, five have a honeycomb geometry while the remaining three are
plate-type catalysts.
After start-up, the baseline performance of each catalyst was determined at design conditions
which are being maintained for the two year test period. Once baseline performance was
established, each reactor was sequenced through a series of parametric tests that varied the
following variables around the SCR process baseline point: NH3/NOx ratio, temperature, and
space velocity. Space velocity is the ratio of flue gas volumetric flow rate to catalyst volume.
With a fixed catalyst volume, variations in flue gas flow rates alter the space velocity around the
design point.
DeNOx efficiency, pressure drop, SO2 oxidation, and ammonia slip are determined at specific
parametric test conditions. After each parametric test matrix has been completed, each reactor is
returned to baseline conditions. This allows for steady-state operation to age the catalyst between
parametric tests. The parametric test matrix is repeated every four months for each reactor train.
The operating parameter ranges examined during the parametric tests and the long-term design
conditions (baseline) are as follows:
-------
Minimum Baseline Maximum
Temperature, (°F) 620 700 750
NH3/NOx molar ratio 0.6 0.8 1.0
Space velocity,
(% of design flow) 60 100 150
Flow rate, (scfrn)
-large reactor 3000 5000 7500
-small reactor 240 400 600
PROJECT SCHEDULE AND STATUS
The demonstration project was organized into three phases. Phase I consisted of permitting,
preparation of the Environmental Monitoring Plan, and preliminary engineering. Phase II
included detailed design engineering, construction, and start-up/shakedown. Detailed design
engineering began in early 1991 and concluded in December, 1992. Construction began at the
end of March 1992 and was completed by the end of February 1993. Start-up/shakedown
concluded in June 1993. Baseline commissioning tests without catalysts were conducted through
June. The loading of all catalysts was completed at the end of June.
The operational phase for process evaluation, Phase HI, commenced in July 1993. The process
evaluation will last for approximately two years and will be followed by preparation of a final
report, which will include catalyst technical performance and process economic projections. The
major milestones on the schedule are shown in Table 3.
PARAMETRIC AND LONG TERM TEST RESULTS
The parametric and long term catalyst testing can be divided into four main categories: 1) test
facility ammonia measurements, 2) test facility sulfur trioxide measurements, 3) test facility
general reactor performance measurements, and 4) catalyst supplier laboratory tests. The
ammonia measurements consist of measurements of intermediate ammonia (usually downstream
of the first catalyst bed), and slip ammonia measurements (at the reactor exit). These ammonia
measurements are used to assess the deNOx performance of the catalyst. Sulfur trioxide
measurements are performed at the inlet and outlet of the reactor to determine the SC>2 oxidation
characteristics of the catalyst. Other parameters are also measured to determine reactor
performance. These parameters include NH3/NOx distributions at the reactor inlet, particulate
and velocity distributions within the reactor, and pressure drops across the catalyst beds. Also
-------
included in the parametric testing are measures of HC1 and N2O concentrations at the inlets and
outlets of the reactors.
Ammonia Measurements
Of particular concern in SCR operation is the relative decline of deNOx activity over time. This
decline is often indicated in the field by increasing ammonia slip exiting the reactor over time.
Figure 2 shows the measured ammonia slip for each catalyst as a function of time exposed to flue
gas. The data show a slight upward trend in ammonia slip for the 7,000 hour operational period.
In most cases however, this upward trend represents less than 1 ppm increase in ammonia slip.
Some uncertainty exists in the exact evaluation of ammonia slip, since most measurements are
near the ammonia detection limit and measurement accuracy is relatively poor (±0.3 ppm) at
these very low ammonia concentrations. In addition, the plot assumes that the NH3/NOx ratio is
constant. In reality, slight variation exist in the NH3/NOx ratio which contribute to the scatter of
the data. This is especially true of early slip measurements where the NH3/NOx ratio was
consistently high due to calibration errors. The most important conclusion to be drawn from the
figure is that ammonia slip has not significantly increased over the 7,000 hour operational period.
In addition to the evaluation of baseline slip over time, other important characteristics such as the
catalyst response to changes in flow rate (space velocity) and temperature are also examined using
intermediate and ammonia slip measurements. These responses can be measured in a variety of
ways but are typically measured in the test facility as changes in first bed NOx reduction or
changes in ammonia slip versus variations in reactor flow or temperature. The first parametric
sequence (preliminary sequence) was an abbreviated parametric test sequence. This test sequence
contained a large number of intermediate ammonia measurements with a relatively small number
of corresponding ammonia slip measurements. For this reason, the first parametric test results
showing flow or temperature dependency on deNOx capability are demonstrated as a function of
first bed NOx reduction (calculated from ammonia measurements).
Figure 3 shows the relative first bed NOx reduction as a function of increased reactor flow rate
for all catalysts that were operational at that time. The data show a general decline in first bed
NOx reduction, as expected. The drop in NOx reduction is mitigated by improvements in mass
transfer with increased flow. If no mass transfer limitations were present under the operational
conditions, a more severe drop in NOx reduction would probably be noted.
Subsequent parametric test sequences focused more on ammonia slip measurements.
Consequently, the effects of flow rate changes on deNOx characteristics are shown as a function of
ammonia slip versus flow rate for the second and third parametric sequences. Figure 4 shows the
measured ammonia slip versus flow rate using data acquired at a 0.8 NH3/NOx ratio and 700 °F
during the second parametric test sequence. As expected, a general increase in ammonia slip is
noted with increasing flow rate. Relatively high slip values for catalyst #1 were measured due to
an error in the setting of the NH3/NOx ratio. For this catalyst, the data represent an NH3/NOx
ratio close to 1.0 rather than the 0.8 NH3/NOx ratio used for the other catalysts.
-------
Figures 5 through 7 show ammonia slip versus flow rate for three NH3/NOx ratios (1.0, 0.8, 0.6
NH3/NOx, respectively) measured during the third parametric test sequence. Figure 5, showing
data for the 1.0 NH3/NOx ratio, is presented to demonstrate a well known effect. The extremely
high ammonia slip values at the 1.0 NH3/NOx ratio indicate that in some cases more ammonia was
being injected than there was available NOx to react, thus forcing the reactor to slip the excess
ammonia. This high NH3/NOx ratio tends to control the ammonia slip and masks potential
changes in slip due to flow rate variations. Figure 6 and 7, using data at a 0.8 and 0.6 NH3/NOx
ratios, show trends more in keeping with the previously noted results from figure 5 (second
parametric test). Again, slight increases in ammonia slip are noted with increasing flow rate.
Figures 8 through 12 show the effects of deNOx capabilities with changes in temperature. As
with the flow rate effect data, the first parametric sequence data shows first bed NOx reduction,
while subsequent measurements focus on changes in ammonia slip. Figure 8 depicts the relative
first bed NOx reduction versus temperature for all catalysts operational at that time. In some
cases, fairly significant increases in first bed NOx reduction are noted. Mass transfer limitations
tend to inhibit improvements in apparent deNOx activity with increasing temperature. The
temperature effect would be nearer to exponential in the absence of mass transfer limitations.
Figure 9 shows ammonia slip versus temperature data for the second parametric data set.
Relatively little improvement (decrease) is noted in ammonia slip between the 700 °F and 750 °F
conditions. At these temperatures, the kinetic rate has likely increased to a point where mass
transfer limitations have become controlling and no significant improvements are noted with
increased temperature. This is not the case however, for the 620 °F versus 700 °F case
conditions. Over this temperature range, relatively significant improvements in ammonia slip are
realized with increasing temperature. This can be interpreted to mean that kinetic rate is a fairly
significant portion of the overall reaction rate and improvements in this parameter with increasing
temperature result in the realization of ammonia slip improvements.
Figures 10 through 12 show data at 1.0, 0.8 and 0.6 NH3/NOx ratios, respectively for the third
parametric test sequence. Figure 10 shows ammonia slip versus temperature for the 1.0
NH3/NOx ratio and design flow condition. The plot shows fairly significant improvements in
ammonia slip for some catalysts while others show nearly constant ammonia slip across the
temperature range. As with the slip versus flow rate data, high values of NH3/NOx ratio tend to
control the ammonia slip rather than the temperature. No definite conclusions should be made as
to temperature effect using the plot. Figure 1 la shows the effect on ammonia slip with variations
in temperature at the 0.8 NH3/NOx ratio. Relatively high ammonia slip values are noted for
catalyst #1 due to erosion problems with the catalyst resulting in some flue gas channeling.
Subsequent investigation of the catalyst revealed that a manufacturing problem contributed to
weak areas in the catalyst which eroded quickly. The catalyst was replaced at the conclusion of
the third parametric test sequence. Figure 1 Ib is identical to Figure 1 la, but with the catalytst #1
data excluded and the scale changed. This more clearly shows the temperature effect for the
remaining catalysts.
-------
The data in figure lib show that there is relatively little improvement in ammonia slip with
increasing temperature across the entire temperature range. Data for catalyst #7 and catalyst #2
show zero slip at the low and mid temperature conditions. These points are plotted as zero since
the actual slip measurement was below the detection limit of roughly 0.5 ppm. Thus, the points
for these two catalysts at the high temperature condition should not be construed as a significant
increase in ammonia slip with increasing temperature. Figure 12 shows data acquired at a 0.6
NH3/NOx ratio and design flow. This plot is very similar to Figure 1 Ib, again showing little or no
improvement in ammonia slip with increasing temperature.
Another important catalyst response to be considered is the ammonia slip versus NH3/NOx ratio
(i.e. NOx reduction). Information of this type is important because it defines the maximum NOx
reduction that can be achieved within a particular maximum ammonia slip limit. Figure 13 shows
the ammonia slip from each reactor versus the NH3/NOx ratio. This plot is typical for this type of
application showing very sharp increases in ammonia slip as NH3/NOx ratio nears 1.0. As
previously mentioned, NH3/NOx ratios greater than 1.0 force the catalyst to slip ammonia, since
no NOx is available to react with the excess ammonia. The plot clearly shows this effect at
NH3/NOx ratios above 1.0.
Sulfur Dioxide Oxidation Measurements
Another important reactivity characteristic of SCR catalysts in addition to NOx reduction activity,
is their propensity to oxidize sulfur dioxide. This is an important aspect of SCR catalyst since
increased SO3 can exacerbate problems with equipment downstream of the SCR due to increased
formation of acidic deposits. SO2 oxidation is normally considered to be a first order reaction.
Thus, in absence of mass transfer limitations, SO2 oxidation should have a nearly linear
relationship to reactor flow rate and an exponential relationship to temperature. SO2 oxidation is
normally considered to be constant with exposure time based on catalyst supplier historical
experience. Figure 14 shows the baseline S02 oxidation rate as percent of inlet SO2 oxidized to
SO3 as a function of catalyst exposure time. A fair amount of scatter is present in data, but in
general, no significant trends are noted as a function of catalyst exposure time. Nearly all of the
data show oxidation rates of 0.75% or less which was the original design specification for
maximum allowable SO2 oxidation.
Flow rate and temperature effects on SO2 oxidation are shown in Figure 15 and 16 respectively.
This data is from the second parametric test sequence. Figure 15 shows the flow rate effect on
S02 oxidation for 0.8 NH3/NOx and 750 °F conditions. The catalyst #8 data shows relatively
high SO2 oxidation characteristics (in keeping with suppliers quoted rates) and does show a nearly
direct linear effect of flow rate on SO2 oxidation. However, the other catalysts show little or no
effect of flow rate on SO2 oxidation with a relatively constant SO2 oxidation rate across the flow
rate range.
Figure 16 shows the effect of temperature on SO2 oxidation rate. This plot shows that S02
oxidation rate is a much stronger function of temperature than of flow rate, as expected. The
-------
effect across the temperature range is relatively linear for most catalyst with the exception of
catalyst #8 which exhibits a more exponential SO2 oxidation relationship to temperature.
Measurements of SO2 oxidation subsequent to the second parametric test sequence were limited
to only two or three conditions. The SO2 oxidation rate for 0.8 NH3/NOx ratio and standard flow
at medium and high temperature was examined during the third parametric test sequence. These
data are shown in Figure 17. As with Figure 16, the data show a consistent increase in SO2
oxidation rate with respect to temperature.
General Reactor Performance Measurements
In addition to catalyst performance, reactor performance in terms of NH3/NOx distributions,
paniculate distributions, and velocity distributions are critical to the efficient operation of an SCR
unit. These parameters are controlled primarily by reactor design rather than catalyst design.
Other reactor performance measurements such as reactor pressure drop and N20 formation are
primarily catalyst specific parameters.
Typically, each of these general reactor performance parameters are measured as part of each
parametric test sequence. The following discussions outline the general results of these
performance measurements.
Ammonia to NOx Distribution
A balanced NH3/NOx distribution within an SCR reactor is critical to its efficient operation.
Maldistribution of either component can create areas within the reactor where the local NH3/NOx
ratio is very high. This leads to an increase in overall ammonia slip from the reactor and a
reduction in deNOx efficiency. Reactor design criteria are set to minimize this problem. In the
pilot facility, design criteria are primarily set to maintain a smooth distribution of both ammonia
and NOx resulting in an even NH3/NOx ratio across the reactor. In full scale installations, the
ammonia injection grid is tuneable, thus allowing ammonia to be injected in a profile which
matches the NOx profile, thereby creating a smooth NH3/NOx ratio distribution across the
reactor. Although the test facility has this tuning capability, in practice most adjustments that
have been made were aimed at smoothing the ammonia distribution.
The test facility design criteria requires an ammonia distribution within ± 10% of the average
concentration. The NOx distribution within the reactors is normally smooth; therefore this
ammonia distribution criteria controls NH3/NOx ratios within the reactor cross section to within
roughly 10% of the average value.
Ammonia distributions are normally measured in the test facility reactors just upstream of the first
catalyst beds. Results to date have demonstrated that the ammonia distribution within each
reactor is within the ± 10% design criteria originally set.
-------
Velocity Distribution
The flue gas velocity distribution within an SCR unit is important for several reasons. First,
severe maldistributions in velocity make tuning of ammonia injection difficult because the flue gas
velocity maldistributions have the same effect as NOx maldistributions. Further exacerbating the
problem is that velocity maldistributions are often strong functions of reactor flow rate. Thus,
unit load changes would require constant adjustment to the ammonia tuning grid when velocity
maldistributions are present. Another concern with velocity maldistribution is the possible erosion
impact on the SCR catalyst. Localized high velocities can cause premature catalyst erosion in
some areas of the catalyst, thereby reducing the overall life of the catalyst installation.
Measurements of test facility flue gas velocity distributions within the reactors have consistently
shown very smooth distributions throughout the reactor (roughly less than + 5%). The
distribution measurements have proven to be very valuable in identifying problems with the
catalyst such as areas of fouling or channeling which are indicated by a severe localized change in
velocity near the problem area.
Paniculate Distributions
Paniculate distributions are an important parameter in SCR units primarily for the potential
erosion problems that can occur with particulate maldistributions. Localized high particulate
concentrations can accelerate erosion in particular areas, similar to localized high velocities. As
previously mentioned, this increased localized erosion can reduce the overall life of the catalyst
installation.
Particulate distributions measured within the test facility, normally at the reactor exits, have
consistently been very smooth, within ± 5%.
Reactor Pressure Drops
Reactor pressure drops are controlled primarily by pressure drops across the individual catalyst
beds. These catalyst bed pressure drops are dictated, of course, by specific catalyst design.
Pressure drop is a major design concern for a utility boiler mainly due to the energy cost involved
in overcoming the reactor pressure drop. In SCR retrofit situations, reactor pressure drop
becomes critical and often dictates whether or not a fan retrofit or upgrade is required in
conjunction with the SCR retrofit.
Long-term catalyst pressure drops are controlled by the original catalyst design and the
effectiveness of sootblowing and catalyst cleaning over time. Consistent increases in reactor
pressure drop over time indicate that adequate sootblowing/cleaning is not being performed.
10
-------
Figure 18 shows the pressure drops for each of the SCR reactors as a function of time. These
pressure drops represent the summation of pressure drops across all catalyst beds and dummy
bed.
N20 Formation
The possible formation of N2O across SCR reactors was a concern with early SCR experience in
the U.S. Subsequent investigations of the problem revealed that this was a sampling/analysis
anomaly rather than the true formation of N20 across SCR catalysts. Measurements at the test
facility have included evaluation of N2O concentrations to further confirm this finding. All
measurements to date indicate that the test facility catalysts do not form N2O during the course of
normal operation. Table 4 shows N2O concentrations acquired during the second parametric
tests. The table shows reactor exit N2O concentrations for each of the catalysts and gives an
average reactor inlet N2O value as measured in the test facility main inlet plenum (prior to
splitting of the stream to the individual reactors).
Laboratory Tests
Several catalyst characteristics are measured in the laboratory to trace catalyst changes versus flue
gas exposure time. Samples are normally extracted from the test facility catalysts on a quarterly
basis and sent directly to the individual catalyst suppliers for evaluation. These tests normally
examine parameters such as deNOx activity, SO2 oxidation activity, BET surface area, porosity,
and chemical composition.
Of particular interest are laboratory performed deNOx activity measurements. These
measurements are very important because they typically demonstrate activity declines before they
become apparent in test facility tests. The laboratory tests can often identify the source of catalyst
deactivation such as chemical poisoning, loss of surface area, and sintering etc. This type of
deactivation data is critical to estimating parameters such as total catalyst life. Figure 19 shows
relative deNOx activity versus exposure time for three catalysts. This plot is representative of all
catalysts in the test program. The deactivation rates shown are in keeping with catalyst supplier
historical experience. No severe declines in deNOx activity are noted which is an encouraging
finding, since there has been little previous knowledge of catalyst deactivation rates for U.S. coal
applications. This deactivation will of course be tracked throughout the test program and will be
used to estimate ultimate catalyst life.
11
-------
ACKNOWLEDGMENT
We would like to thank the Southern Research Institute on-site team for their significant
contribution in obtaining the data presented here; Mr. Robert Heaphy, Mr. James Garrett, Mr.
Randy Hinton, and Mr. Richard Jacaruso. Sincere thanks also go to Mr. Charles Powell for his
management of test facility operations, making data acquisition possible. We would also like to
thank Mr. James Gibson of Spectrum Systems, Inc. for his diligent efforts in maintaining the SCR
gas analyzer system and for his efforts in assuring data quality. Special thanks go to Mr. Gerald
Bandura (SCS Engineering) and Mr. Danny Mabire (ICS) for their constant efforts in maintaining
the instrumentation and controls for the test facility as well as maintaining and programming the
digital control system. Our sincere thanks also go to Mr. Ken Pathak for his unfailing efforts in
maintaining the huge quantities of data acquired from the test facility.
REFERENCES
1. A. L. Baldwin, J. D. Maxwell, U.S. Department of Energy's and Southern Company
Services's August 24 -September 1, 1991, Visit to European SCR Catalyst Suppliers,
U.S. DOE, Pittsburgh, PA, 1991, p 41-3.
12
-------
Constituent
Table 1
Test Facility Inlet Flue Gas Composition
ESP Inlet
84 MW
43 MW
ESP Outlet
84 MW
43 MW
NOx
S02(ppm)
S03 (ppm)
HCl (ppm)
NH3(ppm)
Particulate (gr/dscf)
325
2340
32
104
<0.4
3.76
401
1780
42
89
<0.4
2.43
332
2030
14
115
<0.4
0.0018
Not Available
1510
20
101
<0.4
BDL*
* Below detection limits
Catalyst Supplier
Table 2
SCR Project Catalyst Suppliers
Reactor Size Catalyst Configuration
Nippon Shokubai Large
Siemens AG Large
W R. Grace Large
W R. Grace Small
Haldor Topsoe Small
Hitachi Zosen Small
Cormetech Small
Cormetech Small
Honeycomb
Plate
Honeycomb
Honeycomb
Plate
Plate
Honeycomb
Honeycomb (low dust)
13
-------
Table 3
Project Schedule
Detailed Engineering 1/92 - 12/92
Construction 3/92 - 2/93
Start-up/Shakedown 1/93 - 6/93
Process Evaluation 7/93 - 6/95
Disposition/Final Report 7/95 - 10/95
Table 4
N2O Concentrations - First Parametric Test
Location (Catalyst, reactor exit) 1234678
N2O, ppmv* 1.8 1.8 1.6 1.6 1.2 1.0 1.7
(dry @ 3% O2)
; Average inlet cone. =2.1 @ main plenum
14
-------
Figure 1
Prototype SCR Demonstration Facility
Process Flow Diagram
HEATER
VENTURI
^ -•
HIGH-DUST/HOT-SIDE ESP INLET
f \
^
i
i
FLUE GAS _s^
SLIP STREAM ^v
. 1 LOW-DUST/
f HOT-SIDE
NH3/AIR
DUMMY LAYER "
LARGE SCR REACTOR
CATALYST
LAYERS
TO
COLD-SIDE
ESP INLET
AIR
HEATER
CYCLONE
"IPOWER
ESP OUTLET
TWO MORE
IDENTICAL
LARGE REACTOR
TRAINS WITH
INDIVIDUAL
BYPASS
'- I PLANT
' I DUCT
-L i
-------
Figure 2
A
4
3.5
I 3
Q.
32.5
_a
55 2
CO
§1.5
1 1
0.5
0
C
mmonia Slip vs. Exposure Time
-
- * D
* m
D * ° *
*
0* *
i , i , i , i , i , i , i
Catalyst 1
Catalyst 2
Catalyst 3
Catalyst 4
D
Catalyst 5
o
Catalyst 6
Catalyst 7
o
Catalyst 8
1,000 2,000 3,000 4,000 5,000 6,000 7,000 8,000
Catalyst Exposure Time (Hours)
Based on standard condition data ._,^TO«
Figure 4
Ammonia Slip vs. Flow Rate
40
60
80 100 120
Flow Rate (% of design)
0.8 NH3/NOX ratio, 700 °F
Second parametric data set
Data for Catalyst 1 Is for 1.0 NH3/NOx
Catalyst 1
160
Catalyst 8
Figure 3
>JOx Reduction Tl
P P - — '
- " - - 3
S!
£ 0.7
re
0)
*0.6
0.5
4
»t Bed NOx Reduction vs. Flow
-^•^.^
_
I , I , I , I , !
Rate
Catalyst 1
Catalyst 2
Catalyst 3
Catalyst 6
Catalyst 7
0
Catalyst 8
0 60 80 100 120 140 160
Reactor Flow Rate (% of Design Flow)
0.6 NHS/NOx ratio, 620 °F
First Parametric Data Set ^^f^M,
Figure 5
60
50
a 40
Q.
55 30
a
o
E 20
10
0
0
1.0 NH,
Ammonia Slip vs. Flow Rate
i
-
n
.
* .
o
°* i , ,* ,?.,.,
Catalyst 1
Catalyst 2
Catalyst 3
*
Catalyst 4
i i
Catalyst 5
Catalyst 6
y
Catalyst 7
9 1 1.1 1.2 1.3 1.4 1.5 1.6
Flow Rate (% of design) a y
S/NOx, 700 °F
Third parametric data set
-------
Figure 6
25
^20
Q.
D.
~£. 15
CO
_re
0 10
1 5
0
0
0.8 NH
Ammonia Slip vs. Flow Rate
-
-
*
• 0
& * 0
T , l , 1 i 1 , 1 c 1* .
9 1 1.1 1.2 1.3 1.4 1.5 1
Flow Rate (% of design)
5/NOx, 700 "F
i
Catalyst 1
Catalyst 2
Catalyst 3
*
Catalyst 4
D
Catalyst 5
0
Catalyst 6
Catalyst 7
O
6Catalyst 8
•
Third parametric data set m_ww.,.
Figure 8
Firs
1.3
O
•-g 1.2
1
51.1
Relative
o
a,<°
t Bed NOx Reduction vs. Tempei
D
*
g
i , i , i . i , i , l , I
•ature
Catalyst 1
Catalyst 2
Catalyst 3
*
Catalyst 6
D
Catalyst 7
0
Catalyst 8
0 620 640 660 680 700 720 740 760
Temperature °F
High flow data at 0.6 NH3/NOx ratio
First parametric data set
Data for Catalyst 3 Is for 1 .0 NH3/NOx ratio ,_
Figure 7
6
5
|4
Q.
I3
O
E 2
1
0
0
0.6 Nt
Ammonia Slip vs. Flow Rate
.
L
• *
• X O
^ a
O
o a
i , I , I , I . I . U
9 1 1.1 1.2 1.3 1.4 1.5 1
Flow Rate (% of design)
HS/NOx, 700 "F
k
Catalyst 1
Catalyst 2
Catalyst 3
*
Catalyst 4
D
Catalyst 5
o
Catalyst 6
Catalyst 7
o
6Catalyst 8
•
Third parametric data set „*„*„.
Figure 9
Ammonia Slip vs. Temperature
Catalyst 2
600
620
640
0.8 NH3/NOx ratio, high flow
Second parametric data set
Catalyst 1 omitted
660 680 700
Temperature (°F)
720
740
760
-------
Figure 1O
Ammonia Slip vs. Temperature
100
•-- 80
Q.
55
E
60
40
20
Catalyst 1
•
Catalyst 2
•
Catalyst 3
*
Catalyst 4
D
Catalyst 5
o
Catalyst 6
600 620 640 660 680 700 720 740 760 780
Temperature (°F)
1.0NH3/NOX, design flow
Third parametric data set
No high temp Catalyst 5 data available
Catalyst?
Catalyst 8
Figure 11b
Ammonia Slip vs. Temperature
600
620 640
0.8 NHS/NOx, design flow
Third parametric data set
Data for Catalyst 1 omitted
660 680 700 720
Temperature (°F)
Catalyst 2
Catalysts
760 780
Figure 11a
Ammonia Slip vs. Temperature
Catalyst 1
_J Catalyst?
600 620 640 660 680 700 720 740 760 780 a
Temperature (°F) catalyst 8
0.8 NH3/NOX, design flow
Third parametric data set
No high temp Catalyst 5 data available
oo
Figure 12
Ammonia Slip vs. Temperature
Q.
B 2
Q.
re
'E
o
E
1.5
0.5
600 620 640 660 680 700 720 740
Temperature (°F)
0.6 NH3/NOx, design flow
Third parametric data set
No high temp Catalyst 5 data available
Catalyst 1
Catalyst 2
760 780
Catalyst 7
Catalyst 8
-------
Figure 13
Ammonia Slip vs. NH3/NOx Ratio
0.6 0.8
NH3/NOx Ratio
1.2
620 "F, high flow
Second parametric data set
Catalyst 1
1.4
Figure 15
SO2 Oxidation vs. Flow Rate
40
Catalyst 1
60
.8 NHS/NOx ratio, 750 °F
Second parametric data set
80 100 120
Flow Rate (% of design)
140
160
Figure 14
SO2 Oxidation vs. Exposure Time
-^ 1.5
c
_o
"•" 1
•a 1
X
O
o
OT 0.5
_
•
X
•_ "
"
X
* J , ' !„ I *, I
uataiyst i
Catalyst 2
*
Catalyst 3
*
Catalyst 4
D
Catalyst S
O
Catalyst 6
X
Catalyst 7
0
Catalyst 8
0 2,000 4,000 6,000 8,000 10,000 •
Exposure Time (Hours)
Data based on design conditions
Linear coir, applied for small flow variations
Figure 16
SO2 Oxidation vs. Temperature
Catalyst 1
600 620 640 660 680 700 720
Temperature (°F)
.8 NH3/NOx ratio, low flow
Second parametric data set
740
760
-------
Figure 17
SO2 Oxidation vs. Temperature
Catalyst 1
710 720 730 740
Temperature (°F)
.8 NH3/NOx ratio, standard flow
Third parametric data set
No high temp Catalyst 5 data available
750
760.
Catalyst 7
Catalyst 8
Figure 19
Laboratory K/Ko
2,000 4,000 6,000 8,000 10,000
Catalyst Sample Exposure Time (Hours)
Typical results from Individual supplier lab analysis
Catalyst 4
12,000
Figure 18
Reactor Pressure Drop vs. Time
10
o'
I 8
0
(A
0)
0 6
£.
^
*r 4
£
3
U)
fl> 2
£
0
-
-O
o-
-a ,* . o..rt a'''\
Q- , / * ° , ,
•»•»**' '* ' ' * *
^\_-^-^^7 ! ^ ^ ' * O' ^Bp— ^ iB— ^S^^ ° > • - ' m^
* *• - - ^? O ••••"*"'*
-••' * ',,''••».- ....-*
S3 /'*"•" '*"
o--"' •<>•••»'
i , , i , , i , , i , i i i i i
uataiyst i
Catalyst 2
*
Catalyst 3
- - * -
Catalyst 4
---B---
Catalyst 5
• -o
Catalyst 6
M
Catalyst 7
O
Catalyst 8
Sep'93 Dec '93 Mar '94 Jun '94 Sep '94 Dec '94 •••-••
Date ^ _^
-------
STATUS OF TESTING OF SCR PILOTS:
A REVIEW OF CURRENT EPRI-SPONSORED RESULTS
Alan J. Mechtenberg
James B. Jarvis
Radian Corporation
Kent D. Zamrnit
Jeffrey W. Stallings
Electric Power Research Institute
G.S. Janik, PhD
New York State Electric & Gas Corporation
Roger Mallory
Tennessee Valley Authority
Maurice W. Guest
Niagara Mohawk Power Corporation
Abstract
This paper provides an update of the ongoing testing at the New York State Electric & Gas
(NYSEG) Kintigh Station SCR Pilot and the final results from the Tennessee Valley Authority
(TVA) Shawnee Station and Niagara Mohawk (NiMo) Oswego Station SCR Pilots. The paper
also previews the report, SCR Design and Operational Recommendations: R&D Lessons
Learned, to be released later in 1995. The report will provide a comprehensive summary of key
pilot results and their implications for full-scale SCR systems.
Recent pilot results have significant implications for full-scale SCR systems on U.S. coal- and
oil-fired utility boilers. At TVA (high dust, high S coal), final results illustrate the effects of
sulfur/ash on catalyst masking and deactivation, and the importance of reactor design features
that counteract ash deposition. Results also show the potential for catalysts that are resistant to
sulfur/ash masking through adjustment of physical catalyst properties. At NiMo (medium S
oil), solids deposition from MgO injection at the host boiler caused significant catalyst
plugging and deactivation of the uppermost catalyst layer. At NYSEG (post-FGD), explor-
atory catalyst testing shows the potential for reducing the operating temperature of post-FGD
SCR, thereby reducing operating costs associated with flue gas reheat. In addition, heat pipe
heat exchanger corrosion tests have significant cost implications for SCR hi this configuration.
In the future, this work will form the basis for the economic evaluation of post-FGD SCR.
-------
The SCR design and operational recommendations report will be organized to allow the user to
access information by topic with tables of SCR issues common to the pilots. The tables will
incorporate short descriptions of the issues, associated results, and recommendations, with
references to the specific report section that contains a full description of key pilot results.
Introduction
Selective Catalytic Reduction (SCR) has been widely demonstrated in Europe and Japan as a
postcombustion NOX control technology. However, most of this experience has been gained
using relatively low-sulfur fuels, typically less than 1.5 percent. By comparison, the application
of SCR in the United States has been much more limited, and to date the experience base is
virtually non-existent for coal- and oil-fired boilers.
In addition to the foreign precedent, regulatory forces stemming from the 1990 Clean Air Act
Amendments will inevitably lead to the more wide-spread use of SCR in the U.S. for both new
and existing units. Although international experience with SCR is significant, it can not be
directly transferred to the U.S. utility industry due to differences in fuel properties, furnace
design, and the design and operation of U.S. powerplants.
Two of the most important differences between foreign and domestic SCR applications are the
sulfur content and trace element concentration of fuels.1 Higher fuel sulfur content corresponds
to higher concentrations of SO2 and SO3 in the flue gas of typical U.S. SCR systems. This can
lead to potential poisoning and more rapid deactivation of the catalyst. In addition, SCR
catalysts have the potential to oxidize SO2 and further elevate the concentration of SO3 in the flue
gas within and downstream of the SCR reactor. This can lead to serious problems with ammo-
nium sulfate and/or bisulfate deposition in the air preheater, changes in the marketability of fly
ash, and potential increases in plume opacity.
A number of compounds present in fly ash of coal- and oil-fired boilers may poison the active
sites of an SCR catalyst. In Germany, for example, arsenic poisoning of SCR catalysts has been
observed. Alkaline metals such as sodium and potassium have been demonstrated to deactivate
SCR catalysts, are were believed to be the cause of reduced catalytic activity in SCR testing at
EPRI's 2.5 MW Arapahoe test facility.2 The relative lack of SCR experience for U.S. applica-
tions, coupled with significant differences in fuel composition represents a potentially serious
situation for any U.S. utility faced with installation and operation of an SCR system on a high
sulfur fuel source.
In response to uncertainties in the cost and feasibility of SCR for the U.S. utility industry, EPRI
has sponsored a multi-pilot plant test program to evaluate the feasibility and cost of SCR as a
function of fuel type and SCR/host boiler configuration. At the present time, pilot units have
been operated at four utility sites. This paper discusses three of those pilots:
-------
1. A high sulfur/high dust SCR unit was operated at the National Center for Emissions Research
located at the Tennessee Valley Authority (TVA) Shawnee Power Station. This unit treated
flue gas from the economizer outlet of a high sulfur coal-fired boiler.
2. A post-FGD unit is operating at EPRI's Environmental Control Technology Center located at
the New York State Electric & Gas Corporation (NYSEG) Kintigh Station. This unit treats
gas from downstream of an FGD system.
3. An SCR unit was operated on a residual oil-fired boiler located at the Niagara Mohawk
(NiMo) Power Corporation Oswego Steam Station. This unit treated flue gas from the
economizer outlet in a configuration similar to the TVA pilot.
Test program objectives for each pilot unit depend on the specific configuration of the pilot and
fuel burned in the host boiler. Testing focused on the impacts of U.S. fuels on the catalyst, the
reactor, and key process subsystems. Additional objectives include the development of testing
and analytical procedures suitable for use on high sulfur SCR units, and identification and
resolution of process control and continuous emissions monitoring issues.
The remainder of this paper contains a description of the distinguishing features of each pilot unit
and their operating characteristics. Testing status and recent results are highlighted for each
pilot, including updates to system performance and key operational issues. Finally, the forth-
coming EPRI report, SCR Design and Operational Recommendations: R&D Lessons Learned is
introduced. This report will contain all of the key design implications from testing at EPRI-
sponsored pilots documented hi a user-friendly format.
Pilot Unit Operating Characteristics
Schematic diagrams of the TVA, NYSEG, and NiMo SCR pilot systems are shown in Figures 1,
2, and 3, respectively. Each pilot represents a 1 MW(e) equivalent SCR reactor divided into two
parallel sections to allow for simultaneous testing of two catalyst types. Each pilot features slip
stream isolation dampers, flue gas heaters, air startup dampers, and ammonia injection control
systems that calculate the required ammonia injection rate from flue gas flow and inlet NOX
signals. Operating conditions and catalyst characteristics are listed for each pilot in Tables 1 and
2, respectively.
The pilot SCR catalysts were designed to maintain certain performance criteria over guaranteed
and design lifetimes of 2 and 4 years, respectively. Performance goals call for 80% NOX
conversion with residual ammonia (slip) levels of less than 5 and 2 ppm at the exits of the
second-to-bottom and bottom catalyst layers, respectively. At TVA (5 catalyst layers), the 5 and
2 ppm slip limits apply to the outlets of the fourth and fifth catalyst beds, and at NYSEG and
NiMo (3 layers), the limits applied to the outlets of the second and third catalyst beds.
-------
TVA Hot Side/High Dust SCR Pilot
The TVA pilot was equipped with features to counter the effects of fly ash on the SCR catalysts.
The catalyst was designed with relatively large cell openings (i.e., a large pitch) typical of a high
dust application. A non-catalytic "dummy" catalyst layer was provided at the top of the reactor
to straighten the gas flow prior to catalyst exposure, and to minimize catalyst erosion. Screens
were installed above each catalyst layer and sootblowers were installed above the first and fourth
layers to help keep the top surface of the catalysts free of solids. Reactor walls and catalyst
baskets were also streamlined to eliminate many of the horizontal surfaces where ash could
accumulate and subsequently fall onto the catalysts.
The original V/Ti catalyst was tested for the entire pilot test duration. Near the end of the test
program, 6 of the 9 test elements in the center of catalyst beds 1 and 3 were replaced with fresh
samples. The vendor theorized that a softer catalyst formulation would be more resistant to
masking that had contributed to deactivation of the original catalyst over time.3 The activity of
the test elements was examined after 7,000 operating hours at the end of the test program.
Results of the activity measurements are highlighted later in this paper.
The original TVA zeolite catalyst experienced substantial deactivation was replaced with a
reformulated design about mid-way through the test program. Prior to replacement, the original
catalyst was removed and water-washed to remove deposits that had plugged the catalyst
channels. The purpose of the experiment was to determine whether catalytic activity could be
restored, and to determine the effects of washing on the catalyst structure. These results are also
discussed later in this paper.
NYSEG Post-FGD SCR Pilot
The distinguishing feature of the NYSEG pilot unit is the equipment used to raise the flue gas
temperature from 120°F at the outlet of the wet FGD system to SCR operating temperatures (600
- 700 °F). Reheat requirements are minimized through the use of a recuperative heat-pipe heat
exchanger. While the heat exchanger recovers heat from the gas exiting the reactor, an additional
185 °F of reheat input is required to maintain a reactor temperature of 650 °F.
Because the cold end of the NYSEG heat exchanger operates below the maximum condensation
temperatures of ammonium sulfate/bisulfate and sulfuric acid, the test program was focused on
evaluating exchanger fouling effects (i.e., heat transfer loss, increase in gas pressure drop, and
corrosion).4 These operating impacts are countered by the smaller reactor size and favorable
catalyst environment (lower flue gas sulfur and ash content) in the post-FGD configuration.
The NYSEG reactor is typical of a post-FGD configuration in that less catalyst is required, and
problems associated with high flue gas sulfur and fly ash content are avoided. No screens,
"dummy" catalyst layers, or reactor sootblowers are required. The NYSEG reactor contains 3
layers, as opposed to 5 in the TVA pilot. The overall catalyst volume is lower than its high dust
-------
counterpart because of the higher surface area-to-volume ratio inherent to a smaller pitch
catalyst. The favorable reactor environment can also lessen the rate of catalyst deactivation, and
further reduce catalyst volume requirements to achieve a given catalyst life. Lower flue gas SOX
and particulate content also reduces the potential for sulfur-related catalyst poisoning.5
NiMo Residual Oil SCR Pilot
The NiMo Oswego Station pilot is a hot side configuration similar to the TVA pilot. The host
boiler is used for load following with frequently changing conditions. Typically, the unit cycles
down to 20% MCR overnight, and hourly load changes of up to 60% MCR were routinely
experienced during pilot testing. This characteristic presented challenges to the pilot process
control system that were addressed during the test program.
Although the particulate content of test flue gas from the NiMo host boiler is considerably less
than that at TVA, problems with catalyst deposition and pluggage were encountered throughout
the test program. Reactor deposits were found to consist of oil ash and magnesium oxide and
magnesium sulfate; the magnesium source being fuel oil additives. Catalyst pluggage was
particularly troublesome for the composite V/Ti catalyst which had a very aggressive catalyst
pitch (3.6 mm) for this service. Plugging countermeasures for the pilot were limited to soot-
blowers above the first catalyst layer and routine catalyst cleaning during system shutdowns.6
A unique ammonia injection system design was evaluated at the NiMo SCR pilot. The TVA and
NYSEG units employed high pressure, vaporized anhydrous ammonia injection systems typical
of past designs in Europe and Japan and in early combustion turbine SCR systems in the U.S. At
the NiMo pilot, direct liquid phase injection of aqueous ammonia was employed, using a twin
fluid atomizing nozzle to facilitate droplet evaporation within the duct upstream of the catalyst.
The benefits of such a system include the avoidance of aqueous ammonia vaporization costs, and
the use of droplet injection to enhance cross-duct penetration and mixing of ammonia with flue
gas upstream of the catalyst.
Pilot Status and Highlights of Testing Results
The following sections detail the status of each EPRI-sponsored pilot, and include brief descrip-
tions of the key pilot performance issues and results. Pilot measurements include flue gas
pressure loss and observations of catalyst pluggage, NOX CEM and manual ammonia measure-
ments to determine on-line system performance, and periodic vendor analysis of catalyst samples
to determine activity changes and changes in chemical and physical catalyst properties. Manual
SO2 and SO3 measurements were also made to determine the potential for SO2 oxidation across
each catalyst over time. In addition, special tests were conducted to examine the effects of SCR
on key pilot unit subsystems (e.g., the effects of ammonia slip and SO2 oxidation on the NYSEG
recuperative heat exchanger).
-------
TVA Hot Side/High Dust SCR Pilot
The TVA pilot was operated between May 1990 and May 1994. During this period, the unit
operated with flue gas flowing for a total of approximately 22,000 hours (about 2.5 years). The
relatively low availability of the pilot unit was the result of both short-term outages caused by
events such as boiler tube leaks or reactor inspections, and longer-term outages caused by
extended boiler outages or failures of certain pilot unit equipment (e.g., the pilot ID fan). Key
pilot results center around the effects of high sulfur and fly ash on catalyst deactivation, and the
counter-measures to ash deposition and catalyst plugging that were evaluated during the program.
Catalyst Activity. Both TVA test catalysts exhibited significant deactivation which was
exacerbated by frequent pilot shutdowns early in the test program. Analysis of ash samples from
the reactor identified a mechanism in which the ash deposits become enriched with sulfur via
interaction with ambient moisture during shutdowns. As the moist acidic deposits reacted with
alkaline ash constituents, hard deposits were formed that permanently plugged a number of
catalyst channels. This mechanism may have also occurred on a smaller scale on the catalyst
surface and within catalyst pores, and contributed to formation of a masking layer and conse-
quential loss of catalyst activity.
The original V/Ti catalyst was tested for the entire pilot operating duration (about 22,000 hours)
Results of catalyst sample activity measurements by the manufacturer are shown in Figure 4.
The measurements were made on small sections of the catalyst sample that did not contain
plugged channels; therefore, results were normalized with data shown in the figure from selected
European experience.7'8'9 In all cases, the samples from bed 1 exhibited a higher activity than
those from bed 3, which may indicate the positive influence of sootblowers located above the
first catalyst layer, but not above the third layer.
Figure 4 also shows the activity curve for replacement V/Ti test elements installed hi the center
of catalyst beds 1 and 3 near the end of the test program. The replacement elements featured
different hardness values than the original catalyst charge. Although the replacement elements
exhibited a lesser rate of deactivation, the positive effects of altering catalyst hardness are not
entirely conclusive because a lower sulfur coal was being fired while the replacement elements
were in place.
The original zeolite catalyst was replaced after approximately 12,000 hours with a reformulated
zeolite design from the same vendor. The original catalyst failed to meets its performance
criteria after only 5,000 hours of operation. Laboratory activity measurement results for the
original and reformulated catalyst samples are shown in Figure 5. When examining the activity
curves, it should be noted that the range of European experience represents traditional V/Ti
catalysts, not zeolites. The reformulated catalyst showed improvement in its baseline activity
and in the rate of deactivation compared to the original catalyst.
-------
Figure 5 also shows the positive effects seen from water-washing of the original zeolite catalyst.
However, subsequent inspections revealed damage to the physical structure of the catalyst, and
swelling of the catalyst contributed to a significant rise in reactor pressure drop. Long-term
effects of these physical changes were not evaluated at the pilot.
Bulk and surface chemical measurements were also made by both catalyst vendors to monitor
changes in the composition of the catalyst and the accumulation of potential catalyst poisons.
Bulk analysis results for the V/Ti catalyst are shown in Figure 6. These results indicate increases
in the concentrations of arsenic, sodium, and potassium with increasing exposure time. Vendor
analysis of catalyst cross sections indicated that these elements had penetrated the catalyst and
were evenly distributed within the interior of the catalyst matrix. Other elements, including
silicon, iron, aluminum, and sulfur were present in high concentrations on a masking layer on the
outer catalyst surface.
The V/Ti catalyst vendor conducted a test in which they essentially sandblasted the masking
layer from the catalyst surface. Subsequent activity tests conducted on the abraded catalyst
sample revealed that the sample activity had increased from a K/K,, value of 0.63 to 0.99 as a
result of the procedure. Because arsenic was found throughout the depth of the catalyst, this
result indicates that the V/Ti catalyst activity loss was the result of masking of the catalyst
surface by sulfated fly ash, and not by arsenic poisoning.
Catalyst Plugging Countermeasures. The TVA pilot represented a severe environment
with respect to potential catalyst plugging with fly ash. Periodic reactor inspections revealed
considerable buildup of solids on the outer catalyst blocks, which resulted from "wall effects" hi
the relatively small reactor. Over the course of the test program, the flue gas pressure loss across
the V/Ti catalyst increased from below 4 inches to over 10 inches of water. Manual counting of
the plugged channels showed that nearly 55 percent of all V/Ti catalyst channels had become
permanently plugged by the conclusion of testing.
A number of strategies were implemented or considered during the test program to limit
increases hi catalyst channel plugging. These include screens, sootblowers, vacuuming, and
moisture avoidance. All strategies employed at the pilot were successful to a certain degree, but
their practicality for full-scale SCR application will vary.
Screens and sootblower above the catalyst beds were installed to help keep the top surface of the
catalyst free of solids. Screens were installed above all 5 catalyst layers, and the optimum mesh
size appeared to be slightly smaller than the catalyst pitch so that any ash that passed through the
screen would also pass through the catalyst. Air sootblowers were installed above the first and
fourth catalyst layers, and their effectiveness was evident through observations of pluggage in
these beds compared to the others. Sootblowers were operated once each 8-hour shift while the
pilot was operating.
-------
After the initial experiences with outage-related catalyst plugging, vacuuming was employed to
remove ash deposits during pilot shutdowns. Early in the test program, step changes in reactor
pressure drop were observed following outages. This is consistent with the aforementioned
sulfur enrichment mechanism that lead to hard, permanent deposits within the catalyst channels.
Pilot operating procedures were later modified to incorporate vacuuming of the top surfaces of
the catalyst at the start of all extended outages. Thus, ash deposits were removed before they
could absorb significant ambient moisture.
In theory, sulfur enrichment of ash deposits could be minimized by preventing moisture from
entering the reactor during outages. This could be accomplished by isolating or bypassing the
reactor, or using a dry air purge or heaters to maintain a margin above the ambient dew point
within the reactor. The design of the pilot, however, did not permit complete isolation of the
reactor to allow for demonstration of this approach.
Other Operating Issues. Several operational issues were encountered during the TVA test
program that provided pilot experience with full scale SCR design implications. These include:
• Ammonia injection nozzle pluggage in the high sulfur/high dust environment.
• Artifact reactions over sampling probe materials during NOX and ammonia sampling.
• Process control issues associated with zeolite catalyst ammonia adsorption/desorption times.
• CEM system maintenance and sample preconditioning issues specific to high dust/high sulfur
SCR systems.
These pilot experiences and resulting implications are beyond the scope of this paper. Detailed
descriptions will be included, along with catalyst performance data and catalyst protection
recommendations in EPRI's forthcoming SCR design and operational recommendations report.
NYSEG Post-FGD SCR Pilot
The NYSEG SCR Pilot began testing in December 1991 and is currently operating. At present,
test catalysts have been exposed to flue gas for approximately 21,000 hours. Key pilot results
include catalyst performance in the relatively clean post-FGD environment and cost issues
associated with flue gas reheat. The reheat system includes a heat recovery exchanger, part of
which operates below the condensation temperatures of ammonium-sulfur compounds and
sulfuric acid. Adverse heat exchanger performance and corrosion effects from fouling were
examined in terms of their potential cost implications for the post-FGD SCR configuration.
Catalyst Activity. Both of the catalyst designs tested at NYSEG have shown very good
resistance to deactivation in the low dust, lower sulfur environment. The original extruded V/Ti
catalyst and the replacement composite V/Ti catalyst showed no measurable activity change in
-------
pilot tests over 13,100 and 12,700 hours, respectively. Therefore, given the same amount of
catalyst and the same performance goals, a post-FGD catalyst would be expected to exhibit a
substantially longer catalyst life than its high sulfur, high dust counterpart.
After being proven over 1.5 operating years, the original extruded catalyst was exchanged with a
new extruded V/Ti formulation designed for higher absolute catalyst activity. In the reformu-
lated design, the physical structure was further optimized to reduce reactant diffusion rate
limitations and increase overall catalyst activity by up to 50%. The top layer of the original
catalyst was moved to the fourth bed in the current arrangement to allow for continued flue gas
exposure beyond 1.5 years. The reformulated catalyst currently occupies reactor levels 1-3.
Baseline testing of the new catalyst has not been completed as of this time.
The original composite V/Ti catalyst exhibited severe deactivation and was replaced after only
5,300 operating hours. Activity changes were found to occur exclusively during pilot shut-
downs, which indicated that ambient moisture had aided in the mobilization and penetration of
catalyst poisons throughout the active catalyst surface layer. Contaminants that penetrated the
catalyst include silicon, sodium, potassium, phosphorus, and sulfur, while calcium and iron were
concentrated at the surface. The source of the contaminants is the fine ash and FGD carryover
solids that are lightly deposited on the catalyst surfaces during operation.
The reformulated composite catalyst has shown no measurable performance decline over
approximately 1.5 years of operation and numerous pilot shutdowns. According to the vendor,
the original catalyst was manufactured using titanium with an incorrect size distribution. Pilot
catalyst performance measurements confirmed that the absolute activity of the replacement
catalyst exceeded that of the fresh, original catalyst. This data appears to confirm that man-
ufacturing error caused, or at least contributed to the failure of the original composite catalyst.
Prior to the exchange of the extruded V/Ti catalyst, a short testing period was dedicated to
catalyst evaluation at temperatures below the typical lower limit of 600 °F. The objective was to
determine the potential for reducing the operating temperature in the post-FGD configuration,
thereby reducing operating costs associated with flue gas reheat. To accomplish this, SCR
catalysts would need to overcome performance effects from kinetic limitations at low tempera-
ture, and from possible fouling due to condensation of ammonium-sulfur compounds on the
catalyst surface and within the catalyst pores. Exploratory testing provided insight to these
effects that could lead to development of specific catalysts for this application.
At a reactor temperature of 550 °F, the extruded V/Ti catalyst exhibited marginally lower, but
steady performance over 1,400 operating hours even though catalyst fouling was detected. The
composite catalyst exhibited a more severe performance decline that varied with changes in the
inlet NOX concentration from the host boiler. This indicated that the degree of catalyst fouling
was varying with the reactor ammonia concentration, which varies in proportion to host boiler
NOX levels at a fixed ammonia-to-NOx ratio. At lower temperatures (i.e, 500 °F) both catalysts
-------
exhibited substantial performance deterioration. After each test period, both catalyst's perfor-
mance was restored to original levels by raising the temperature to baseline (650 °F) conditions.
Heat Exchanger Fouling Effects. Fouling within the recuperative heat pipe heat exchanger
occurs as the gas exiting the SCR reactor is cooled from the normal reactor operating tempera-
tures of 600 - 700 °F to the heat exchanger discharge temperature of 200 - 250 °F, depending on
the reactor temperature and gas flow rate. The heat pipes downstream of the reactor operate at
temperatures approximately 60 °F lower than the surrounding gas temperature. Internal surface
temperatures of approximately one-third of the heat pipes in the cold end are below the sulfuric
acid dewpoint (270 - 280 °F, based on S03 levels of 7 10 ppm). About half of the heat pipes
operate below the maximum temperature threshold for condensation of ammonium-sulfur
compounds (approximately 330 °F, based on 7 ppm SO3 and 1 ppm NH3). Figure 7 shows the
range of heat exchanger fouling observed after operating periods with different levels of
ammonia slip from the SCR reactor.
Figure 8 shows the relative decline in heat transfer and increase in flue gas pressure drop across
the return side of the heat exchanger during operating periods with distinct ammonia slip levels.
In the figure, heat transfer is expressed as a fraction of the design rate to normalize exchanger
efficiency for changes in gas flow and reactor temperature. The figure also shows the effects of
internal water-washing between operating periods. Water-washing was very effective in
dissolving and removing deposits, and consequently in restoring heat exchanger performance and
pressure drop to original conditions.
Figure 8 also illustrates the importance of minimizing ammonia slip from SCR catalysts in the
post-FGD configuration. At the pilot unit, severe heat exchanger performance degradation was
avoided when the average ammonia slip was held to below 2-3 ppm. During these periods, the
total flue gas reheat loss equates to!0-12°Fat pilot design conditions over an approximate 3
month operating period. Longer term testing has shown that much of heat transfer loss occurs
over the first 1-2 months after cleaning, and declines more slowly beyond that time. Over an
approximate 8 month operating period at low (less than 2 ppm) average ammonia slip, the total
flue gas reheat loss was only 12 - 14 °F. Therefore, a full-scale unit may be expected to sustain
operation between annual outages without forced shutdowns for heat exchanger cleaning,
provided a strict limitation on SCR ammonia slip is maintained.
Analysis of deposits proved that the cold end of the exchanger operates in a highly corrosive
environment. Deposits contain ammonia, SO3, heat exchanger metals, and constituents of fine
solids within the pilot flue gas. The relative amounts of ammonia and SO3 vary with the average
catalyst ammonia slip and with the operating temperature in the deposit location. Overall,
deposits are highly acidic, primarily due to the presence of ammonium bisulfate and free sulfuric
acid. Heat exchanger wash-water pH values are typically 1.5 2.0 regardless of the deposit
location or pilot operating conditions.
-------
One of the pilot heat pipes from the cold end of the heat exchanger was removed and destruc-
tively examined after approximately 9,100 hours of pilot operation. The heat pipe was con-
stucted of duplex 2205 steel, with 409 stainless steel fins. The cold end of the heat exchanger is
considered the most severe environment in terms of corrosion effects because it operates below
the sulfuric acid dewpoint on both the supply (upstream of the SCR system) and return (down-
stream of the SCR system) sides of the exchanger. The 409 stainless steel fins exhibited severe
pitting and a maximum 17.5 mils/year corrosion rate. The maximum 2205 duplex steel heat pipe
corrosion rate was 9.7 mils/year, and corrosion was found to be relatively uniform. A higher
corrosion rate (18.1 mils/year) was found in a highly localized area on the supply end of the heat
pipe.
Based on pilot corrosion measurements, the 2205 duplex steel heat pipes would not sustain a full
15 year equipment life in cold-end heat exchanger service. The high localized corrosion rate
occurs where a small pocket of non-condensibles exists within the pipes at the upper end of the
heat pipe incline. Assuming this area could be protected via enameling or by other means, the
next highest corrosion rate (9.7 mils/year) equates to a projected rube life of 6 - 7 years. With the
pilot heat pipe thickness (65 mils), the material would not be acceptable for full-scale service
unless a heat pipe replacement strategy was incorporated into the full-scale design.
Test heat pipes constructed of carbon steel, CORTEN B, and AL6XN were also installed in the
exchanger and destructively examined after 3,300 operating hours. In the most severe test
location (at the cold end) various corrosion phenomena were observed, and none of the materials
were found to be acceptable. Upstream of the cold-end heat pipe module (between modules 1
and 2 as shown in Figure 7), CORTEN B exhibited acceptable corrosion while AL6XN and
carbon steel were marginal. Overall, deposits in the higher temperature areas of the exchanger
(i.e, modules 2 and 3) are much less corrosive because of the lack of free sulfuric acid hi the
higher operating temperature range. For full-scale applications, carbon steel would likely be
acceptable for these areas.
Additional implications from heat exchanger testing were developed during the test program, the
details of which are beyond the scope of this paper. Some of these issues include heat exchanger
water-wash techniques and disposal issues, potential stack particulate emissions impacts, and
other recommendations from corrosion testing. These issues will be detailed in the forthcoming
SCR design and operational recommendations report.
NiMo Residual Oil SCR Pilot
The NiMo pilot unit was operated between October 1991 and October 1993. During this period,
the unit operated with flue gas flowing for approximately 4,800 hours (about 6.5 months). Low
pilot unit availability was caused by low Oswego Station demand, resulting in a low host boiler
capacity factor throughout the test program. Key pilot results include the effects of vanadium
within the fuel on catalyst activity, and catalyst pluggage and deactivation from flue gas solids
from the host boiler.
-------
Catalyst Activity. Activity changes were measured by both catalyst vendors on samples taken
after 2,400 and 4,100 operating hours. The relative activity of the corrugated plate catalyst
increased during both test intervals, and exceeded the original activity by 24% by the end of the
test program. Similar activity increases were measured across all three catalyst layers. The
effect was attributed to deposition of vanadium from the host boiler on the catalyst surface. S02
oxidation rates increased with time over both catalysts, which is a further indication of vanadium
deposition. The measured fuel oil vanadium content varied between 55 and 170 ppm during the
test program.
The activity of the top layer of composite V/Ti catalyst decreased somewhat during the test
program, but no overall performance change was detected via pilot NOX conversion and ammonia
slip measurements at the reactor exit (after 3 layers). After 4,100 hours, the activity of the top
layer declined by roughly 20% based on the average of values for samples taken from the tops of
the first and second layers. Essentially no activity change was seen in the second and third beds
during the course of the test program. Overall, catalyst performance was well within its design
limitations at the conclusion of testing despite significant catalyst pluggage.
Deactivation at the inlet of the composite catalyst was attributed to masking by a thin layer of
solids found on the catalyst surface. Although a significant fraction of catalyst channels were
completely blocked, laboratory activity measurements and calculations were based only on open
catalyst channels within the samples. The catalyst vendor concluded that solids deposition and
consequential activity loss at the top of the reactor was exacerbated by the aggressive catalyst
pitch (3.6 mm). A more conservative catalyst design and additional measures to prevent solids
deposition (i.e., sootblowers above every catalyst level) are advisable for full-scale SCR systems
in similar heavy oil service.
Catalyst Plugging and Deposition. Severe reactor solids deposition and catalyst plugging
was experienced over the course of the test program. Flue gas pressure loss across the plate
catalyst (4.6 mm hydraulic diameter) increased from 3.0 to 3.8 inches water during operation,
which corresponds to overall pluggage of 15-20% of the catalyst channels. The composite
catalyst (3.6 mm pitch), exhibited nearly 50% overall pluggage, resulting in a pressure drop
increase from 5.5 to over 10.5 inches water.
Reactor solids consisted of primarily oil ash, magnesium oxide (MgO), and magnesium sulfate.
Deposits were also found to contain insulation from the host boiler, which may have contributed
to catalyst pluggage early in the test program. MgO is added to the fuel oil and is also injected at
the economizer inlet to control SO3 via reaction to form magnesium sulfate. One mole of MgO
is required to bind every mole of SO3. Analysis of reactor deposits indicates that the overall
additive/injection rate employed by the station equates to nearly two moles of MgO per mole
SO3. Although not proven at the pilot scale, more strict control of MgO usage may reduce solids
deposition and catalyst pluggage effects in full-scale SCR systems for residual oil boilers. In
addition, sootblowers were found to be highly effective in preventing catalyst pluggage in this
service in a detailed evaluation at another EPRI-sponsored pilot.1
10
-------
Other Operating Issues. Operational lessons from the NiMo pilot study include the
demonstration of direct liquid ammonia injection, and process control issues associated with
inconsistent aqueous ammonia concentrations and deep cycling of the host boiler. These issues
will be summarized in EPRI's SCR design and operational recommendations report, a preview of
which follows.
SCR Design and Operational Recommendations Report
Results from all EPRI-sponsored pilots are currently being incorporated into a guidance
document entitled SCR Design and Operational Recommendations: R&D Lessons Learned
(EPRI Report TR-105103). The report will be released later in 1995, and will include results and
design implications from the three pilot studies described in this paper. In addition, the report
will include the results of testing at the advanced SCR pilot system at the Pacific Gas & Electric
Company's Morro Bay Station, and at the multi-pilot SCR system at Southern Company
Service's Plant Crist sponsored under the DOE's Clean Coal Technology Program. Other EPRI-
sponsored SCR studies to be detailed in the report include:
• Results of SCR catalyst research which focused on the examination of SCR poisoning and
deactivation mechanisms.
• Recent updates to SCR cost studies for utility power station applications.
• An evaluation of Hybrid SCR designs.
• SCR procurement guidelines.
The report is being developed to help those responsible for designing, specifying, and trouble-
shooting SCR systems for utility boiler applications. The report will be organized by topic to
combine the related lessons from the various pilot studies. Overall SCR design and operational
issues will be described with pilot implications included where applicable. Tables of important
pilot results will highlight each specific issue and full-scale recommendation, with reference to
the report section or appendix that fully details each key pilot result.
References
1. Technical Feasibility and Cost of Selective Catalytic Reduction (SCR) Nox Control. Palo
Alto, Calif.: Electric Power Research Institute, May 1991. GS-7266.
2. G.H. Shimoto and L.J. Muzio, KVB Inc.; and J.E. Cichanowicz, Electric Power Research
Institute. "Pilot-Scale Evaluation of Selective Catalytic Reduction for Coal-Fired Utility
Boilers," Proceedings: 1982 Joint Symposium on Stationary Combustion NOX Control,
EPA/EPRI, Dallas, TX (1982).
-------
3. C. Huang, J. Hargis, L. Fuller, and R. Mallory, Tennessee Valley Authority; J. Jarvis and B.
Stapper, Radian Corporation; and E. Cichanowicz, Electric Power Research Institute.
"Status of SCR Pilot Plant Tests on High Sulfur Coal at Tennessee Valley Authority's
Shawnee Station." Paper presented at the EPA/EPRJ 1993 Joint Symposium on Stationary
Combustion NOX Control, Miami Beach, FL (May 1993).
4. G. Janik, New York State Electric and Gas; A. Mechtenberg, Radian Corporation; and K.
Zammit and E. Cichanowicz, Electric Power Research Institute. "Status of Post-FGD SCR
Pilot Plant Tests on Medium Sulfur Coal at The New York State Electric & Gas Kintigh
Station." Paper presented at the EPA/EPRI1993 Joint Symposium on Stationary Combus-
tion NOX Control, Miami Beach, FL (May 1993).
5. J. Chen, R. Yang, and E. Cichanowicz, "Poisoning of SCR Catalysts," Proceedings: 1991
Joint Symposium on Stationary Combustion NOX Control, EPA/EPRI, Washington, DC
(1991).
6. M.W. Guest, Niagara Mohawk Power Corp.; T.A. Montgomery and R. L. Hack, Fossil
Energy Research Corp.; and J.E. Cichanowicz and K.D. Zammit, Electric Power Research
Institute. "Status of SCR Pilot Plant Tests on High Sulfur Fuel Oil at Niagara Mohawk's
Oswego Station." Paper presented at the EPA/EPRI 1993 Joint Symposium on Stationary
Combustion NOX Control, Miami Beach, FL (May 1993).
7. Dr. H. Maier, P. Dahl, "Operating Experience with Tail-End and High-Dust Denox-Tech-
niques at the Power Plant of Heilbronn." Paper presented at the EPA/EPRI 1991 Joint
Symposium on Stationary Combustion NOX Control, Washington, DC (March 1991).
8. Dr.-Ing. Peter Necker, "Experience Gained by Neckarwerke from Operation of SCR DeNOx
Units." Paper presented at the EPRI/EPA 1989 Joint Symposium on Stationary Combustion
NOX Control, San Francisco, CA (March 1989).
9. J.K. Beer, L. Balling, "SCR Catalytic Converts Used in the Federal Republic of Germany -
Diverse Applications and Optimized Plant Systems." Information from an Article for
publication in Modern Power Systems. Complete citation unavailable.
10. D. Teixeira, "Results of Catalyst Tests at the PG&E-EPRI ASCR Pilot Plant." Paper
presented at the EPA/EPRI 1995 Joint Symposium on Stationary Combustion NOX Control,
Kansas City, MO (May 1995).
-------
Table 1
Typical SCR Pilot Operating Characteristics
Host Boiler Fuel Type
Pilot Configuration
Total Flue Gas Flow Rate, scfm
Reactor Temperature, °F
Inlet NOX, ppm
Inlet SO2, ppm
Inlet SO3, ppm
Particulate, gr/dscf
TVA Shawnee Station
High (2.5-5.0%) S Coal
Hot Side/High Dust
2100
700
450
2000
20
3.0
NYSEG Kintigh Station
Medium (1.5-2.5%) S Coal
Post-FGD
2000
650
300
150
5
0.0012
NiMo Oswego Station
1. 5% S Residual Oil
Hot Side
2000
700
200-1000
800
23
0.091
-------
Table 2
SCR Pilot Catalyst Characteristics
Catalyst Type
Total Flue Gas Exposure,
hr
Cell Size (Pitch), mm
Wall Thickness, mm
Cell Density, cpsi
Void Fraction, %
Geometric Surface Area,
ft2/ft3
Space Velocity, hr°
Area Velocity, ft/hr0
Linear Velocity, ft/secc
Extruded
V/Ti
21800
5.95
0.80
18
73.7
174
2470
14.1
10.4
TVA Shawnee
Zeolite
11900
7.60
1.68
11
61.0
125
1677
13.4
12.5
Station
Reformulated
Zeolite
7900
5.59
0.86
21
73.0
170
2204
13.0
10.5
Original
Extruded
V/Ti
14700
4.2
0.78
37
64.8
229
4660
20.4
11.8
NYSEG
Kintigh Station
Reformulated" Original
Extruded Composite
V/Ti V/Ti
1600
4.0
0.69
39
66.0
244
4590
18.8
11.6
5300
3.6
0.66
50
67.0
277
9180
33.1
11.4
Replacement"
Composite
V/Ti
12700
3.2
0.30
64
80.0
338
6750
20.4
9.6
NiMo Oswe
Station
Composite
V/Ti
4600
3.6
0.66
50
67.0
277
10260
25.0
11.4
go
Plate
V/Ti
4600
4.6b
1.50
18
66.0
177
4360
25.0
5.4
"Exposure hours current as of March 31, 1995 in continued test program
bHydraulic Diameter
°At 1000 scfin, 60°F, 5 catalyst layers at TVA and 3 layers at both NYSEG and NiMo
-------
Economizer ~ • I .., .fFlue Gasl—
„ TT - ~ mmPers trn n Heater |—
riomUaitQ _^ 1 \ ^
Kconomizer • Damper
r 1
I
; i
NH3/NOX Ratio
Controller
[ Flow
I Meter
Anhydrous _ ffi 1 1
Ammonia . . „. 4
|
1
Control Valve Dilution Air
Heater
t
Dilution
Air
L
i
•i
n
••i
•s
f
NH3 Injection
Nozzles
Venturi
\
|
I SCR Reactor
T
- 5 Catalyst Layers
^ Balancing
J Dampers
r . n
1 £. — » H Flue Cias Return
^U "
v-x Control
10 Fan Damper
Figure 1
TVA Hot Side/High Dust SCR Pilot Unit Process Flowsheet
Heat Recovery Exchanger
lo Stack -*—
From NYSEG
FGD System — | | —
Start-up
Damper
Ammonia
^J^
••" "iir
in ^- *-
— -Br-J Flue Gas ^
"1 Heater 1 \\
r r
*" i
NH3/NOX Ratio
Controller
i
i Flow
t Meter
S, ii
Ammonia Flow J
Control Valve Dilution
Air
Balanc
Damp
1
f
ing s.
ersM
1
rNHg Injection
Nozzles
Venturi
!l
:: SCR Reactor
: : (3 Catalyst Layers)
•
.
j
j Control
y Damper ID Fan
In s — ^
->- D i
u v^x
1
t
Figure 2
NYSEG Post-FGD SCR Pilot Unit Process Flowsheet
-------
Second
Analyzer Set
First Analyzer Set
•Air NOV
00
I
\
\
1
NP
4—
F
I3/NO Ratio
Controller
i
-r
r
"N
NH4OH
Tank
^
/
Dampers
Metering Pump
Catalyst Layers
Control Damper
-f
£
Fan
Exhaust to
Unit5
Isolation
Damper
Figure 3
NiMo Hot Side SCR Pilot Unit Process Flowsheet
o
|
^
*^
">
u
<
a
U
on
1.0
0.9
0.8
0.7
0.6
0.5
0.4
0.3
• First Bed
• Third Bed
a First Bed "Replacement"
o Third Bed "Replacement"
— Range of Selected
European Experience
_L
4000 8000 12000 16000
Exposure Time (hrs)
20000
24000
Figure 4
Relative Changes in TVA V/Ti Catalyst Activity vs. Exposure Time
-------
U
o
•4^
4000
First Bed
First Bed Washed
Third Bed
• First Bed Reformulated
• Third Bed Reformulated
•Range of Selected
European Experience
8000 12000
Exposure Time (hrs)
16000
20000
Figure 5
Relative Changes in TVA Zeolite Catalyst vs. Exposure Time
g
o
o
U
As203
Sulfur/10
Na2O
K2O
Fresh
2nd 3rd' 4th
Catalyst Sample
Figure 6
Bulk TVA V/Ti Catalyst Contaminants
-------
Return Side Inlet NH3 Concentrations:
Return
Side
Supply
Side
Treated Gas
To Stack
235 °F
Inlet Gas
From FGD
120 °F
Average: 8.6 ppm
Maximum: 17.5 ppm
Average: 2.5 ppm
Maximum: 14.4 ppm
Average: 0.8 ppm
Maximum: 6.6 pp
Module 1
Stainless
Steel
Module 2
Carbon
Steel
Module 3
Carbon
Steel
Treated Gas
From SCR
590 °F
Untreated Gas to
Heater & SCR
465 °F
Figure 7
Range of Fouling Within Return Side of NYSEG Pilot Recuperative Heat Exchanger
OJt>
(8
u.
"3
o>
M
96
94
92
90
88
86
84
Heat Transfer
Pressure Drop
0.8 1 I 2.5 1 I 8.5
Average SCR Reactor Ammonia Slip (ppm)
i—2.5—I
2.0
1.5
1.0
0.5
0.0
C
3
sr
n
X
ffi
I-J
o
Figure 8
NYSEG Pilot Heat Exchanger Performance and Return Side Pressure Drop
vs. Time (Heat Exchanger Was Water-Washed Between Operating Periods)
-------
RESULTS OF CATALYST TESTS AT THE
PG&E/EPRI ASCR PILOT PLANT
D. P. Teixeira
F. J. Lone
Pacific Gas and Electric Company
San Ramon, CA
T. C. Fang
T. A. Montgomery
Fossil Energy Research Corporation
Laguna Hills, CA
K. D. Zammit
Electric Power Research Institute
Palo Alto, CA
Abstract
This paper presents results from the PG&E/EPRI advanced selective catalytic reduction (ASCR)
3-MWe slipstream pilot plant located at PG&E's gas/oil-fired Morro Bay Power Plant.
Specifically, results of tests using an "in-duct" ASCR configuration and a catalytic air preheater
(APH) are covered. Results from the in-duct ASCR tests include catalyst performance during
oil-fired operation, APH deposition during oil-fired operation, and the effect of selective
catalytic reduction on visible plume characteristics during oil-fired operation. Initial
performance results for the catalytic APH tests are presented.
Introduction
Regulations requiring major reductions in NOX emissions from fossil-fired power plants in
Northern California have recently been adopted by a number of air quality regulatory agencies.
Table 1 summarizes these regulations.
Table 1
NOX Emission Regulations in the PG&E Service Territory
Species
NOX
NOX
NH3
Fuel
Natural gas
No. 6 oil
Gas/oil
Allowable Emissions
10 ppm at 3% O2 dry
25 ppm at 3% O2 dry
10 ppm at 3% O2 dry
-------
These low NOX levels will almost certainly require the use of some form of selective catalytic
reduction (SCR) technology with ammonia. Although SCR technology has been widely applied
commercially in a number of countries (notably, Japan and Germany) for many years, and more
recently in Southern California, costs for these technologies are still excessive, especially given
the rapidly evolving competitive environment for power generation in California. Furthermore,
from a technical perspective, it is believed that additional, significant cost reduction
opportunities remain to be exploited in both SCR technology and combustion control
technologies. For this reason, PG&E has initiated a significant research effort to realize these
opportunities. Major activities currently underway include a full-scale burner demonstration
(Babcock and Wilcox S-burner) at the Morro Bay Power Plant and the ASCR pilot plant (the
subject of this paper).
ASCR Technology Description
Since most costs associated with SCR are not related to the catalyst itself, but rather to
associated equipment such as .duct modifications, reactor vessels, structural steel, and
earthwork/foundations, PG&E's research has been directed at minimizing, or even eliminating,
these latter expensive components.
In-duct SCR arrangements deploy the catalyst within the fuel gas path between the economizer
outlet and APH inlet. Current in-duct SCR configurations enlarge the ductwork to lower
velocities, accommodate a larger volume of catalyst, and reduce pressure drop. A number of
such designs have been recently used in Southern California utility boiler installations. An even
lower cost SCR technology, "true in-duct" SCR, would utilize the existing ductwork with only
minor modifications. In addition to the in-duct ASCR research activities, another interesting
technology with the potential for eliminating the previously mentioned expensive components is
the catalytic APH. All of these ASCR techniques are actively being investigated by PG&E's
research programs at the ASCR pilot plant test facility.
ASCR Pilot Plant Description
As might be inferred from the above discussion, the principal purpose of the ASCR pilot plant is
to permit a facility which allows investigation of primarily engineering/construction, i.e., cost,
and operations-maintenance, related issues. A brief description of the pilot plant facility follows.
The ASCR pilot plant is located at PG&E's Morro Bay Power Plant. The pilot plant is capable
of drawing a 5,000-scfm slip stream from either the Unit 3 or Unit 4 boilers at Morro Bay. The
facility is designed to operate on a continuous, unattended, 24-hour per day basis, and duplicates
full-scale utility boiler design and operation. The pilot plant simulates boiler components from
the economizer inlet through the outlet of a Ljungstrom regenerative APH, including allowance
for installation of catalytic elements in the APH. Initially, a Haldor Tops0e, Inc. (HTAS),
catalyst was installed in the in-duct catalyst section of the reactor. A catalytic air heater element
provided by a joint effort of ABB Air Preheater, Inc., and Engelhard Company has also been
installed.
2
-------
The test facility consists of the following major components:
. An in-duct SCR reactor.
. A Ljungstrom APH that duplicates the thermal history (metal, air, and gas temperatures vs.
axial distance) of a full-scale PG&E APH.
. Flue gas and air fans, flow measurement Venturis, flue gas heater, ductwork, insulation, and
structural and electric components.
• A computer-driven PLC control system.
. State-of-the-art emissions measurement instrumentation for process control and catalyst
performance evaluation.
• Multiple locations for aqueous or vaporized NH4OH injection.
injection system '
A photograph of the ASCR pilot plant is presented in Figure 1. As shown, the electric heater and
gas fan are located upstream of the economizer and catalyst sections. Venturi flow meters on the
gas and air circuits of the pilot plant provide flow measurement capability. Pressure transducers
and thermal RTDs measure gas pressures and temperatures at several locations throughout the
pilot plant.
A Haldor Tops0e catalyst has been initially installed in the pilot plant using an in-duct
configuration. The reactor can accommodate up to three layers of catalyst; however, only two
layers, or catalyst beds, are currently installed. Each layer is nominally 2 ft high by 4 ft wide,
with a depth of 1.67 ft. The catalyst has a hydraulic diameter (i.e., characteristic dimension
opening size) of 3.3 mm and a nominal wall thickness of 0.5 mm At the design flow rate of
5,000 scfm, the overall space velocity is approximately 26,000 h"1 at the outlet of the fist catalyst
bed and 13,000 h"1 at the outlet of the second.
After exiting the catalyst beds, the flue gas flows through a Ljungstrom APH. This device is a
scaled-down version of the full-scale APHs installed on PG&E boilers. The design was selected
to simulate the flue gas, air, and metal temperature time history of a full-scale APH, specifically^
in relation to heat transfer, element configuration, and construction materials. The ASCR APH
is supplied with a sootblower in the air inlet duct at the cold end of the APH and blows from the
cold air side towards the hot. The sootblower is an oscillating pipe with a single steam nozzle.
The ABB/Engelhard catalyst elements replaced the original hot-end elements with catalytically
coated ones. A high heat and mass transfer geometry was selected. The catalytic elements have
a hydraulic diameter of about 8 mm with a nominal thickness of 0.7 mm. Based on the total
volume of the 16-inch-deep high-temperature elements, the overall space velocity is
approximately 14,400 h"1 for a slipstream flue gas flow rate of 4500 scfm. However, since only
5.5 of the 12 sectors (46%) are exposed to the flue gas at any point in time, the actual operating
space velocity of the catalytic APH is about 3 1,400 h"1 . A high activity vanadia/titania catalyst is
bonded to the surface of a treated metal plate.
-------
ASCR Project Goals
The goals of the ASCR pilot plant project are summarized below. While cost is a major factor in
the research activities, other areas are also being pursued, for example minimization of by-
product environmental impacts associated with SCR technology.
• Cost minimization.
• Technical risk management/risk minimization.
• Permit high reliability/availability at full capacity soon after startup.
• Minimize of operations and maintenance requirements.
• Minimize of adverse environmental impacts associated with SCR technology.
Specific research activities completed or currently underway are summarized below:
• Evaluate advanced catalysts (Haldor Tops0e and ABB/Engelhard).
« Examine APH deposition/fouling during oil-fired operation.
• Assess/avoid visible plumes during oil-fired operation.
• Evaluate the minimum acceptable catalyst size opening for the short periods of oil burning
normally experienced at PG&E.
• Develop advanced flow modeling performance predictive techniques.
• Develop true in-duct ASCR configurations to minimize retrofit costs.
As a result of this research and other ongoing projects, a cost savings goal of $30-50 million has
been targeted.
Results
Advanced flow modeling/predictive performance techniques are covered separately by another
presentation at this conference ("A New Design Tool for SCR Systems" by L. J. Muzio et al.).
True in-duct SCR work is in progress and will not be covered at this time.
Air Preheater Deposition/Removal
The goal of the APH fouling tests was to investigate the effects of an SCR catalyst on APH
operation during oil firing. As is well known, ammonium sulfate and bisulfate salts can form due
to SO 3 present in the flue gas, and NH3 slip emitted by the SCR process at temperatures below
nominally 400°F. These ammonium salts may then deposit in the APH baskets. Deposition of
these salts can lead to increased pressure drops across the APH, and if unchecked, limit the full-
load capability of the unit. Corrosion of the APH could also result.
-------
During this testing, the formation and deposition of ammonium salts from SO3/NH3 reactions
within the APH was evaluated by monitoring the pressure drop across the APH, measuring the
concentrations of various NH4 compounds entering and leaving the APH, and visual inspections.
In addition, the effectiveness of current sootblowing procedures for removing any deposits was
evaluated. SO3 and NH3 concentrations were varied in a controlled manner (up to 20 ppm each
were investigated).
Test results can be summarized as shown in Figure 2. For reference, APH pressure drop is about
3-3 % inches H2O during natural gas operation. As can be seen, the pressure drop began to
increase rapidly during the two periods when the sootblowers were not operated. More
important, operation of the sootblowers per the normal design/procedures permitted rapid
recovery of the pressure drop to normal values. Other tests and inspections verified this result.
The tests conducted showed that the existing air heater/sootblowing design and operating
procedures are readily able to control any deposition, and no modifications are required.
In-duct Catalyst Pluggage during Oil-fired Operation
The goal of this testing was to determine if the 3.3-mm catalyst opening size is acceptable during
the normal, brief (2-4 weeks per year) oil burning periods that PG&E has experienced in recent
years. This is important since a catalyst with a smaller opening results hi lower costs. Tests
were conducted without sootblowers or turning vanes because this would be the preferred lowest
cost approach for the predominant natural gas-fired operation of PG&E plants.
Test results are shown in Figure 3. As can be seen, pressure drop in the catalyst increased
rapidly due to the buildup of particulates from the oil-fired operation: pressure drop increased
from about 3^ inches H2O to 11 inches in less than 500 hours (about 3 weeks). It is estimated
that this increase in pressure drop would result in a 40% derate in full-load capacity—an
obviously unacceptable situation.
Fortunately, these deposits were concentrated at the inlet face to the catalyst and were quite
friable. Subsequent tests with a conventional sootblowing system has shown these deposits to be
easily removed; pressure drop was held at low levels during several recent months of oil-fired
operation.
As a consequence of these tests, it was concluded that the 3.3-mm opening is acceptable for oil
firing as long as adequate sootblowing capability is provided.
It should be emphasized that the deposition observed in not a unique feature of the Haldor
Tops0e catalyst investigated; any catalyst of similar pitch would exhibit the same behavior.
-------
Catalytic APH Technology
It is possible that catalytic APH technology could play an increasingly important role as part of
an overall system to achieve the lowest cost strategy for complying with NOX regulations. The
catalytic APH system selected for evaluation at the ASCR pilot plant is under development by
ABB/Engelhard. As noted previously, only the hot element layer was replaced for this
installation. Greater NOX reduction capability is possible—at least in some cases—by extending
the catalytically active length of the hot element. However, such an approach requires a
complete intermediate and cold element replacement at a significantly increased cost.
Results of the initial tests to characterize the performance of the fresh catalyst at full-load design
and operating conditions are shown in Figure 4. For example, NOX removal and ammonia slip
were 54% and 24 ppm, respectively, at an NH3/NOX ratio of 1.0. This relatively modest
performance is not unexpected given the high space and area velocities associated with the
system. Despite these apparent limitations, such performance, when coupled with other NOX
control technologies, could play an important role in providing the necessary degree of control at
the lowest cost.
Future tests will compare catalyst performance for various operating conditions and evaluate
catalyst deactivation as a function of time.
Visible Plumes
One of the most important objectives of the pilot plant tests was to collect data regarding
possible impacts of the SCR system on plume visibility during oil-fired operation. This issue is
important because PG&E's fossil power plants are often located in densely populated areas; as a
result, operation with visible plumes is unacceptable. Furthermore, the plants do not incorporate
particulate control devices such as precipitators or baghouses.
Since particulate control devices are not available, prevention of visible plumes can only be
effected by controlling the basic chemical constituents leading to the condensed sulfate/bisulfate
species (i.e., SO3 and NH3 slip) and, realistically speaking, only the NH3. NH3 slip, in turn, is
determined by the allowable levels that are covered by the amount of catalyst and, ultimately, by
the bid specification. It is clear that specifying these allowable levels too high can lead to the
formation of plumes; however, specifying the levels too low can result in significantly added
cost for the SCR system.
There is little doubt that SO3 and NH3 condense to form sulfates/bisulfates at temperatures
around 400 °F. However, depending on the physical phenomenon occurring, the resulting
condensed compounds may or may not contribute to plume visibility. If, for example, the
condensation occurs in an environment with many condensation nucleation sites available (e.g.,
particulates that occur in oil firing), it is possible that little or no plume visibility impacts would
occur; in effect, the condensation products formed a thin shell around existing particles.
-------
However, if the particles condense via a homogeneous nucleation process, it is possible that
highly visible submicron particles would result. The resulting fumes, which occur in the
relatively particle-free environments of natural gas combustion, are known to yield, at least in
some situations, particles in the 0.5-micron range. These particles are known to scatter light in
the visible portion of the spectrum that is most sensitive to the perception of the human eye.
Because of the low particle concentrations, small path lengths, the non-steady nature of the tests
involved, and the desire to focus on submicron particle sizes, conventional measurement
techniques, such as impactors, were judged not to be suitable. After evaluating various
alternatives, a methodology was adopted based on optical measurements of the particle
concentrations using the Insitecl optical particle counter system. This technique yields particle
size distribution data down to about 0.2 micron, which covers the portion of the visible spectrum
of interest. In the next step of the methodology, these particle size distribution measurements are
then converted into a visibility parameter using previously developed software techniques
(PLUVUEH).
The key to this approach is to.establish baseline particle size distribution—visibility values based
on the normal operating conditions of the boiler, which are known to have invisible plumes.
This baseline value could then be compared to that resulting from various SCR operating
conditions (e.g., various levels of SO3 andNH3s sootblowing events, etc.). Actually, as shown
below, calculation of the visibility parameter was not necessary since no meaningful changes in
particle emissions characteristics could be observed due to the presence of the SCR system.
Typical results from the plume tests at the ASCR pilot plant are illustrated in Figure 5. This
figure presents the sub-2 micron (i.e., visible portion) particle concentrations as a function of
time for two operating situations. The upper curve shows the baseline case without any NH3 or
additional SO3 injection. The peaks in the particle counts reflect the various boiler sootblower
events as indicated. The lower graph presents the same information, but for the ASCR system
operating (in this example, NH3 = 8 ppm and SO3 = 11 ppm). These levels of SO3 and NH3 are
known to have a potentially large impact on plume characteristics (for example, if the NH3 and
S03 were quantitatively converted to 0.5 micron NH4HSO4 particles, the resulting particle count
would increase by approximately 287,000 particles/cc). Note that the ASCR sootblower system
is inoperative in both cases. As can be seen, within the repeatability of the sootblowing
procedures, no increase in particulates, and hence plume visibility, is observed. Additional tests
were conducted to evaluate the impacts of catalyst, and APH, sootblowing. Similar results were
obtained.
Based on these results, it can be concluded that for the sootblower designs and operating
procedures used by PG&E, levels of SO3 and NH3 up to about 10 ppm are acceptable, and no
visible plumes will be created.
-------
Future Activities
As indicated earlier, future research activities are aimed at two major thrusts: developing/
evaluating true in-duct ASCR, and catalytic APH systems into commercially viable options, hi
addition, PG&E R&D is closely monitoring the progress of so-called "ultra-low NOX burners."
Although major technical barriers must be overcome, success in this area presents major
additional cost savings opportunities.
Conclusions
Based on testing at the ASCR pilot plant test facility, the following conclusions have been
reached:
• At up to 20 ppm of NH3 and SO3 entering the APH, results demonstrate that the current
sootblower design and sootblowing operating procedures are effective in maintaining the
pressure drop across the APH. '
• A greater than two-fold increase in pressure drop was observed across the catalyst beds in the
absence of sootblowers and turning vanes during the relatively short oil burn due to oil-fired
particulate deposition. In a full-scale PG&E application, this increase in pressure drop would
have resulted in a major capacity derating estimated to be on the order of 40%. Subsequent
testing with a sootblowing system permitted operation for several months with no pressure
drop increase.
• Initial results of catalytic APH performance are encouraging. Future tests will evaluate the
long-term activity.
• For the APH/sootblower designs and operating procedures currently in operation, NH3 and
SO3 levels up to 10 ppm will not impact plume visibility.
-------
REFERENCES
1. Holve, D.J., and S.A. Self. 1979. An optical particle-sizing counter for in situ
measurements; parts I and H J. Appl. Optics 18(10):1632-1652.
-------
Figure 1
Morro Bay \SCR Pilot Plant
-------
I 6
CO
I
CO
0
CD
1
0
ASCR Off-Line ,
i!
!!
!!
•i
i!
!!
n w/o Sootblowing
• w/Sootblowing
— Begin NH3 injection
Begin SO3 injection
— Stop SO3 injection
Begin SO3 injection
- - Stop SO3 injection
50 100 150 200 250 300 350 400 450 500 550
Time from Start of Oil Burn (hours)
Figure!
APH Pressure Drop vs. Time
-------
o
C\J
CO
CD
o
CM
T3
CD
m
Begin NH3 injection
1 -
Oh
•I" "
ASCR Off-Line o
0 50 100 150 200 250 300 350 400 450 500 550
Time from Start of Oil Burn (hours)
Figure3
Bed 2 Pressure Drop during Oil-fired Operation
-------
BASELINE TEST
CONDmONS(FULL LOAD1:
- NATURAL GAS FUEL
- GAS FLOW= 4500 SCFM
- AVG INLET N0x= 76 PPMD
@3%O2
- HOT END AIR-GAS
DP= 0.35"H2O
- HOT END AVG METAL
TEMP.= 628 F
- HOT ELEMENT LENGTH
= 16 IN.
-AVE.APH LEAKAGE=17%
- HYDRAULIC DIA.= 8 mm
- NOM. PLATE THICK.= 0.7 mm
- SPACE VEL= 31,4001/hr
0.00
0.20
0.40
0.60
NH3/NOX
0.80
1.00
1.20
INITIAL PERFORMANCE OF ABB/ENGELHARD CATALYTIC AIR PREHEATER, LCAP™
Figure 4
-------
18000
16000 --
8 14000
- 12000
2 10000
§5
« o
BASEUNE PARTICLE COUNTS
NH3 SLIP= 0, ADDED SO3= 0, NO ASCR CATALYST SOOTBLOW
40
60
60
TIME, MINUTES
100
120
140
160
PARTICLE COUNTS WITH NH3 SLIP AND ADDED SO3
NH3 SLIP= 8 PPM, SOS OUT= 11 PPM, NO ASCR CATALYST SOOTBLOW
60
80
TIME, MINUTES
100
120
140
160
Figure 5 :
Effect of NH3 Slip and SO$ on Plume Visibility - No ASCR Catalyst Sootblow
-------
A NEW DESIGN TOOL FOR SCR SYSTEMS
L. J. Muzio, T.D. Martz, T.C. Fang
Fossil Energy Research Corp.
23342 C South Pointe
Laguna Hills, California 92653
D.P. Teixeira
Pacific Gas & Electric Company
2303 Camino Ramon, Suite 200
San Ramon, California 94583
Abstract
A new technique to support the design of SCR systems for natural gas, oil, or coal applications,
and accurately predict the performance (NOX removal and NH3 slip) in commercial applications
is described. The approach takes into account features such as actual duct layout and ammonia
injection grids. The procedure is based on integrating two major efforts: laboratory cold flow
modeling tests; and "ideal" (i.e., uniform ammonia and velocity distribution) catalyst
performance characteristics. A key feature is the inclusion of the NH3 injection process in the
flow model by use of tracer gas techniques. Results of detailed test to validate the design
procedure based on data from a laboratory cold flow model and the PG&E-EPRI 3MW ASCR
pilot plant are presented. Use of the design technique in commercial applications is discussed.
Introduction
To meet more stringent NOX emission regulations, the utility industry is moving in a direction of
utilizing Selective Catalytic Reduction (SCR), in addition to combustion modifications, as a way
of achieving compliance. At the same time, efforts are being made to reduce costs by exploring
"in-duct" SCR arrangements. In-duct SCR arrangements incorporate the SCR reactor basically
along the existing flue gas path from the economizer to the air preheater.
In the extreme, there would be no changes to the ductwork for an "in-duct" SCR and the catalyst
inserted within the existing ductwork, so-called "true in-duct SCR". True induct systems are
likely to be most applicable when used in conjunction with other NOX controls (e.g., low NOX
burners, SNCR, air preheater catalyst) where the SCR NOX reduction requirements is reduced to
the range of 60-80%. In many cases, "in-duct" arrangements will involve enlarging the
ductwork to lower velocities, accommodate a larger volume of catalyst, and reduce pressure
drop. This "in-duct" concept is contrasted to a more traditional retrofit approach associated with
Japanese or German SCR installations, where a large separate SCR reactor is built with fairly
extensive ductwork changes to route the flue gas to and from the SCR reactor.
-------
While the "in-duct" arrangement can potentially reduce retrofit costs, it also poses greater
challenges to the designer. In addition to the type and amount of catalyst, the process designer
must also address velocity uniformity, and NHj/NC^ uniformity across the catalyst face. In the
traditional separate SCR reactor approach, there is usually sufficient ductwork to allow a
uniform velocity to be generated at the ammonia injection grid. A uniform velocity profile at
the ammonia injection grid, greatly simplifies the ammonia/NOx mixing process. Likewise, with
the separate SCR reactor, there will be sufficient space to design ductwork expansions and
turning vanes to meet velocity uniformity requirements at the catalyst face. As a consequence,
previous design approaches specified criteria such as the standard deviation in the velocity
profile, or NH3/NOX profile, at the catalyst face. If these criteria are met, then the SCR system
will perform as designed.
With an "in duct" arrangement, it may not be possible to: (1) provide a uniform velocity at the
ammonia injection grid, or (2) accommodate traditional design guidelines in terms of duct
expansion angles, etc. As a consequence it will be more difficult, if not impossible in some
instances, to meet traditional criteria in terms of a velocity uniformity and/or NH3/NOX
uniformity at the catalyst face. However, this does not mean that an "in duct" arrangement
should be discarded as a viable approach. Rather, it points to the need to develop better
engineering design tools to assess the performance of these systems.
Before discussing the specific objectives of this work, it is of value to briefly discuss the
consequences of non-uniformalities of both, velocity and NH3/NOX across the catalyst face. The
impact of velocity uniformity depends on the type of fuel being fired. If the unit fires only
natural gas, then the primary impact of velocity non-uniformities is varying local space
velocities across the catalyst face. Regions where the velocity is higher than average, correspond
to high space velocity regions; conversely, low velocity regions correspond to locally low space
velocity regions. Since high and low space velocity regions tend to compensate for one another
velocity non-uniformity tends to have only a small effect on SCR NOX removal and NH3 slip.
If the fuel being burned is oil, or coal, where the flue gas is laden with particles, the velocity
impacts for the SCR are different. In these cases, velocity uniformity is more important.
Regions of high velocity can lead to accelerated catalyst erosion. Also, for the case of a
paniculate laden flue gas, it is important to align the particles as parallel as possible with the
catalyst channels, in order to minimize deposition. This can be accomplished using tuning vanes
or other low cost techniques.
The NH/NO,. distribution across the catalyst face has a much greater impact on SCR
performance. While high and low velocity regions tend to compensate one another, this is not
the case with the NH3/NOX distribution. An unbalance in the Nli/NO,, distribution can lead to
increased NH3 slip. In fact, if the NH^/NO,, distribution has significant regions with NH3/NOX
ratios greater than unity, overall performance may not be improved to satisfactory levels by
increasing the amount of catalyst (i.e., decreasing space velocity).
This paper describes a new methodology to support the designed SCR system. The method
integrates the results of cold flow modeling with an SCR process model to quantitatively account
for site specific conditions. The paper will describe the methodology, outline the approach to
-------
cold flow modeling, provide validation of the method, and briefly discuss the use of the
methodology for a full-scale SCR design.
SCR Process Design Methodology
Physical cold flow modeling has been used previously during the design of SCR systems. In
most cases, the cold flow modeling has focused on developing uniform velocities at the catalyst
face and/or at the plane of the ammonia injection grid, and reducing pressure drop. In some
instances, tracer gas techniques have been used to simulate the ammonia injection process. To
provide more input to the SCR design process a methodology has been developed that integrates
cold flow modeling results along with a model of the SCR process to predict performance for a
given specific configuration.
The process model integrates: (1) the cold flow velocity distributions, which define local
variations in space velocity, (2) tracer gas results, which define local variations in NH3/NOX
ratio, and (3) ideal catalyst performance data (i.e., NOX removal and NH3 slip versus NH3/NOX
and space velocity) to calculate overall NOX removal and NH3 slip for a given catalyst and NH3
injection configuration. The model also estimates the pressure drop across the catalyst.
Figure 1 shows the type of ideal catalyst performance data used as an input to the calculations.
It should be emphasized that the model calculates the performance for a specific SCR
configuration relative to this ideal catalyst performance. The model currently incorporates a
generic representation of a titania/vanadia SCR catalyst. For an actual full scale design, a higher
degree of confidence in the absolute performance values would be achieved if data for the
specific catalyst were available (either from the catalyst vendor, or independently-developed test
data).
Cold Flow Modeling Principles
Cold flow modeling is widely used throughout the industry to address a number of flow
problems in ducts, ESP's, furnaces, SNCR systems, etc., and the underlying foundation is fairly
well established. The same modeling principles are used for SCR systems, although the
presence of the catalyst and its pressure drop characteristics impose some additional
requirements.
The following issues need to be addressed in cold flow modeling of SCR systems:
• Overall geometric similarity.
• Dynamic similarity including the catalyst pressure drop.
• Modeling the NH3 injection process.
Each of these is addressed below:
-------
Geometric Similarity
The cold flow SCR models are geometrically scaled (typically 1 to 12). This includes all
ductwork, SCR reactor, NH3 grid, etc. The model should also include all internal duct structures
that would impact the overall flow; typically this includes structures larger than 6-8 inches.
Dynamic Similarity
Cold flow modeling is based on the assumption that the overall flow patterns will be similar
between the model and full scale if the model flow is fully turbulent. As such, it is not necessary
to operate the model at the same Reynolds number as full scale; rather, the cold flow models are
operated at Reynold's numbers greater than 20,000 .
In most cold flow modeling applications, dynamic similarity is entirely met by maintaining the
Reynold's number criteria (Re > 20,000). This is not the case with an SCR system, however,
due to the pressure drop of the catalyst. The catalyst pressure drop results in a redistribution of
the velocity at the catalyst face. The amount of velocity redistribution depends on amount of
pressure drop relative to the dynamic pressure of the flow field (i.e., l/2pV2). This is
characterized by the pressure coefficient (PC).
PC =
(1)
To insure that this is correctly modeled, the cold flow model is designed such that the pressure
drop across the simulated model catalyst is identical to the projected pressure drop across the
full-scale catalyst.
^"model cat ~~ ^"full-scale cat ^ '
Additionally, the model is operated such that the dynamic pressure of the model is identical to
the dynamic pressure of the full-scale system. This means that the model and full-scale pressure
coefficients are equal.
NH3 Injection Process
An important part of the SCR system is the NH3 injection grid and the mixing between the NH3
and the combustion products. The NH3 mixing process depends on four basic parameters:
location of the grid in the flue gas flow path, geometry of the grid (number of injection holes,
size), amount of carrier gas used to transport the ammonia, and the resulting momentum of the
jets leaving the injection holes. Because of density differences between the model and full scale,
it is not possible to simulate all of the above at one time. To simulate the NH3 injection process,
two criteria are adhered to. First, the NH3 injection grid is geometrically scaled. Second, to best
-------
simulate the mixing process, the model ammonia injection grid is operated at the same jet
momentum ratio as full scale:
V2
Pflue gas flue gas ,~-.
Another issue that needs to be addressed is the flow conditions inside the injection grid. Ideally,
each injection lance in the grid should function as a plenum with equal flow from each
individual injection hole along the lance. The degree to which this occurs can be characterized
by the ratio of the average velocity out of an injection orifice to the average velocity at the
entrance to the lance.
V;2
VSR =
vlance (4)
If this ratio is less than 10-15, unequal flow can result from each of the orifices. Alone, this may
or may not be an issue for a particular SCR injection grid. However, it does raise an issue in
terms of modeling the actual injection grid. At full scale, the air/ammonia or flue gas/ammonia
mixture will heat up as it flows from the entrance of a lance to the last injection orifice at the
exit of the lance. The resulting density gradient along the lance will affect the flow rates from
the individual injection orifices. Techniques to account for this phenomenon in cold flow
ammonia injection grid design have been developed and verified.
Test Facilities
The test work to validate the design methodology utilized PG&E's advanced SCR pilot facility
located at the Morro Bay Power Plant, along with a physical cold flow model of the pilot plant.
Only a brief description of the ASCR facility is provided in this paper as it is described in detail
in another paper at this symposium2.
ASCR Pilot Plant
The ASCR pilot plant draws a 5000 scfm slipstream from either Morro Bay Unit 3 or Unit 4
boilers. The design of the pilot plant is a 1 to 11 scale simulation of an in-duct SCR concept for
a typical large PG&E fossil plant. As such, the pilot plant simulates boiler components from the
economizer inlet through the outlet of a Ljungstrom regenerative air preheater.
Figure 2 shows a plan and side view of the reactor. The basic reactor starts just above a
simulated economizer. The slip stream flows through a series of perforated plates that are used
to simulate the full-scale economizer. The flue gas proceeds through a right angle turn then
passes through a relatively high velocity "throat" region. Exiting the throat, the reactor
undergoes an expansion in order to reduce the velocity and accommodate the catalyst. As can be
seen in Figure 2, the reactor expands both vertically and horizontally. In the horizontal direction
-------
the entire expansion occurs along one wall. This is done in order to accommodate the existing
location of the air preheaters in the full scale system (the ASCR pilot plant simulates one duct of
the full scale system). The inside dimensions of the reactor are 24.5 inches high by 45 inches
wide. The actual ASCR reactor section that holds the catalyst is longer than would be for the
full scale system. This was done to accommodate up to three layers of catalyst and to provide
for experimental flexibility in the test program.
The SCR reactor currently incorporates two layers of a Haldor Tops0e SCR catalyst. Each layer
is nominally two feet high by four feet wide with a depth of 1.67 feet (0.5 m). The catalyst cell
opening has a hydraulic diameter of 3.3 mm and a nominal wall thickness of 0.5 mm. At the
design flow rate of 5000 scfm, the two layers of catalyst provide a space velocity of about
13,000 hr1.
Exiting the reactor, the slip stream passes through a Ljungstrom® regenerative air preheater,
similar to full scale and is then returned to either Unit 3 or Unit 4.
For this study aqueous ammonia (NH4OH) is used to supply NH3 to the ASCR pilot plant at the
following locations:
• upstream of the flue gas boost fan,
• downstream of the simulated economizer,
• at the "throat" region,
At the first location upstream of the gas fan, a twin fluid atomizer is used to spray the NH4OH
directly into the slipstream. This injection location provides the highest degree of mixing as the
NH4OH vaporizes and flows through the fan and serves as the "baseline" injection configuration.
For NH3 injection at the other locations, the NH4OH is prevaporized in a heated air stream, and
injected into the reactor, along with the carrier air through a simulated ammonia injection grid.
Cold Flow Model
To support the original design of the ASCR pilot plant and the ongoing test work, a one half
physical scale model of the ASCR pilot plant was designed and fabricated.
The model was constructed mostly from clear acrylic. It encompassed the entire ASCR pilot
plant from the fan outlet to the air heater inlet. This included the circular inlet duct, the
economizer and economizer hopper, catalyst bed, and the inlet transition duct to the air heater.
A schematic of the model is shown in Figure 3. A 5000 scfm blower operating in an induced
draft mode was used to provide air flow to the model.
Validation Test Results
Validation of the design methodology was done through a series of tests at the ASCR pilot plant
and on the cold flow model. This section of the paper will present the validation results in the
following order:
-------
• First, detailed velocity profiles at the exit of the first bed of catalyst (Bed 1) will be presented
and compared to comparable measurements made in the physical cold flow model.
• Next, a number of ammonia injection scenarios will be discussed in terms of the resulting
uniformity of the NH3/NOX ratio across Bed 1. Again, a direct comparison will be made to
comparable tracer gas tests performed in the physical cold flow model.
• Finally, the performance predictions will be presented which integrate the cold flow
modeling data (or NH3/NOX distributions at the pilot plant) with a SCR process model.
One of the primary objectives of this test effort was to validate the cold flow modeling by direct
comparisons to the pilot. To make this comparison as direct as possible, the measurements at the
ASCR pilot plant and cold flow model used the same number, and layout, of sample points to
develop the profiles; 27 points total, 3 vertical locations with 9 points horizontally. While the
number and location of sample points was comparable, basic differences in the two facilities
dictated that some of the measurement techniques differ. These differences will be discussed,
where appropriate.
Detailed Velocity Measurements
During this test series, detailed velocity measurements were made at the exit of catalyst Bed 1 at
the design flow rate of the ASCR pilot plant. At the design flow rate of 5000 scfm the pressure
coefficient in the pilot plant is 30. The physical cold flow model was operated at a flow rate to
yield, as close as practical, a similar pressure coefficient.
In the cold flow model, the velocity measurements were made using a thermal anemometer
which has good resolution and linear response over the velocity range of interest. However, the
thermal anemometer cannot operate at the near 700 °F temperatures of the pilot plant. This
necessitated the use of a l/4"-S-type pitot probe to obtain the detailed velocity profiles under
"hot" flow conditions at the pilot plant. An electronic pressure transducer with a full scale range
of 0.5 inches of H2O was used with the S-type pitot probe.
The detailed velocity profiles from the two facilities are shown in Figure 4, both contour plots
and wireframe diagrams are included. Good agreement is seen between the two profiles. A high
velocity region is seen near the center, with a low velocity region on the right hand side, which
represents the region where there is a horizontal expansion as the gas exits the throat region and
transitions to the catalyst Quantitatively, the standard deviation in the pilot plant velocity
measurements was 39 percent at full load (PC-30). The comparable cold flow model test
(PC=35) yielded a standard deviation in velocity of 27 percent. As will be discussed later, the
slightly higher pressure coefficient, 35 versus 30 should decrease the velocity standard deviation
by about five percentage points.
Figure 5 summarizes the effect of the pressure coefficient on velocity uniformity at the exit of
Bed 1 from a number of tests. Included are data from the physical cold flow model, and pilot
plant (both with hot flue gas and ambient air). As expected, the velocity uniformity increases as
the pressure coefficient increases (i.e., higher pressure drop per unit of velocity head in the
-------
flow). The effect of pressure coefficient on velocity uniformity is quite steep below PC=60.
Since most full scale system will likely operate with pressure coefficient greater than 50-60, the
inherent velocity uniformity should be below 20%. Note, the design pressure coefficient for the
two beds of catalyst at the ASCR pilot plant is 60.
Ammonia Injection
The next portion of the test program addressed ammonia injection and the impact that various
ammonia injection parameters have on the NH/NC^ uniformity at the catalyst face. It should
also be emphasized that the purpose of this test series was not to develop an optimized injection
system for the ASCR configuration, but, rather, to investigate the ammonia injection process and
validate the cold flow modeling techniques. With this goal, the ammonia injection parameters
were chosen to primarily show changes in NH.,/NC)X distribution, in order to validate the cold
flow modeling approach. The parameters that will be addressed include:
• Injection location (ahead of the gas fan, economizer exit, throat)
• Number of injection holes
• Amount of transport air and injection momentum
Before presenting the results, it is worthwhile to briefly review the measurements that were
made and the manner in which the data will be presented. For each test condition, the primary
goal was to determine the NH3/NOX distribution across the catalyst face; all measurements were
made downstream of Bed 1, which represents a space velocity of 26,000 hr"1.
At the pilot plant, the inlet NOX distribution is uniform across the catalyst face. The local
NH3/NOX ratio was determined by measuring the local NOX removal across the catalyst along
with the local NH3 slip. Assuming that the injected ammonia either reacts with NOX (1 mole of
NH3 per mole of NO, and 1.33 moles of NH3 per mole of NO2), or exits the catalyst as slip, the
local NH3/NOX ratio is calculated as
[ANO + 1.33 AN02 + NH3 slJ
local
V local x
(5)
The cold flow model tests were done by injecting a tracer gas into the simulated injection flow
and measuring local tracer gas concentrations at the exit of Bed 1 in the cold flow model. The
normalized tracer gas concentration represents the local NHj/NOx ratio for an overall NH3/NOX
ratio of unity.
Baseline Ammonia Injection. To demonstrate a uniform NH3/NO,. profile across the
catalyst face, a test was conducted injecting the ammonia upstream of the gas fan. The mixing
through the fan provides good mixing between the ammonia and the slipstream. This data set
also serves as a baseline data set where the NH^/NO* ratio is nearly uniform, and the only
impacts on SCR performance are due to velocity non-uniformities. The results of this baseline
-------
test are not shown but the NH3/NOX distribution was quite uniform exhibiting a very small
standard deviation in the NH3/NOX profile of 2 percent.
Economizer Exit Injection: Variation in the Number of Holes. Three injection grids
were investigated at the economizer exit location, which varied the total number of injection
holes (15,50 and 100 holes). Each injection grid consisted of five lances at the economizer exit
which entered the ASCR from the left hand side. For these three tests, the amount of transport
gas and the momentum ratio, J, was held constant. This necessitated decreasing the hole
diameter as the number of holes increased. The results of these tests are shown in Figure 6.
Overall, both the pilot scale and cold flow results show an increase in uniformity as the number
of injection holes was increased from 15 to 50 (i.e., 3 injection holes per lance increased to 10
per lance). A further increase to 100 holes showed no further improvement.
For the 15 hole injector tests the cold flow model results showed a higher overall non-
uniformity in the NH/NC^ ratio than the pilot scale tests. This is evident looking at the detailed
profiles in Figure 6, where the cold flow model exhibits a region of lower NH/NOx ratio on the
left hand side. This is the side where the injectors penetrate the reactor. This general
characteristic is attributed to flow dynamics inside each lance. In the cold flow model, the flow
inside the lance is isothermal, while in the pilot plant the ammonia-air mixture inside the lance
heats up as it flows through the lance. Since these particular lances were designed with a
VSR=6 (see Equation 4) the non-isothermal flow in the pilot plant lance behaves somewhat
differently than the cold flow lances. This effect is particularly apparent for the fifteen hole case
shown at the top of Figure 6. With only three injection holes per lance, any flow maldistribution
effects would be expected to have a larger effect on the tracer gas profiles. For this case, the
vertical contour lines (Figure 6(a)) indicate that a smaller amount of a tracer gas is being injected
on the left hand side near the lance inlet. This local stratification leads to the higher standard
deviation (i.e., 21% versus 14%) in the cold flow profiles, compared to the pilot plant NH3/NOX
profiles.
Throat Injection: Effect Of Momentum Ratio. Next, a 15 hole injection grid was set up at
the "throat" region, just ahead of the duct expansion out to the catalyst dimensions. While this is
a relatively high velocity region, just downstream of a 90° turn, there are some practical reasons
why this could be an attractive location for the ammonia injection grid. In the PG&E units
which utilize FOR for NOX control, the FOR takeoff is generally in the vicinity of the
economizer exit. To avoid injecting NH3 into the FGR stream, it may be desirable to locate the
ammonia injection grid downstream of the FGR takeoff.
The results of these tests are shown in Figure 7 where the momentum ratio was varied from 70
to 150 (corresponding to ratios of injection air to flue gas of 0.5 - 0.75 percent). The contour
plots show that the pilot and cold flow profiles are generally similar. The peaks and valleys are
in the same locations, and there is excellent quantitative agreement in terms of the overall
standard deviation of the NH3/NOX profiles.
-------
SCR Performance Predictions
The velocity profiles and NH3/NOX profiles obtained from the cold flow model and/or pilot plant
can now be used along with the process model to predict SCR performance. By utilizing the
process model, a direct assessment can be made of the impact of velocity and NH3/NOX non-
uniformities on performance (NOX removal and NH3 slip). To accomplish this comparison, the
NH3/NOX profiles (cold flow tracer gas, and ASCR pilot plant) were put into the process model
and the performance calculated. All of the calculations presented in this section were performed
for an initial NOX concentration of 75 ppm (dry @ 3% O2) and a space velocity of 26,000 hr1.
The ideal catalyst performance shown in Figure 1 was used. To reiterate, the model basically
calculates the effect of both velocity and NH/NC^ uniformities on performance, relative to an
ideal performance curve for the catalyst.
Performance predictions based on the pilot scale velocity and NH3/NOX profiles for the baseline
test (NH3 injection upstream of the gas fan) are shown in Figure 8. Recall that the standard
deviation in velocity is 39% (Figure 5), and that the NH3/NOX uniformity is quite good for this
case (2% standard deviation). In Figure 8, three curves are shown for both the predicted NOX
removal and NH3 slip. The solid line represents the ideal performance at a space velocity of
26,000 hr"1 (i.e., perfectly uniform velocity and NH3/NOX). The line through the triangle points
is the baseline prediction corresponding to the 39% standard deviation in velocity and 2%
standard deviation in NH/NOx ratio. For comparison, a calculation was performed using the
velocity non- uniformity (39% standard deviation) but assuming perfectly uniform NH3/NOX (the
curves through the square symbols in Figure 8). Comparing this latter calculation to the ideal
performance illustrates the effect of velocity non-uniformities alone. As can be seen, even with
a large velocity non-uniformity, there is a small impact on performance. At an NHj/NOx ratio of
0.9, the NOX reduction is only decreased by about 2 percentage points and NH3 slip increased by
nominally 1 ppm due to the velocity non-uniformity. A small non-uniformity in the NH3/NOX
ratio (2% standard deviation) has almost the same impact on performance as a 39% standard
deviation in velocity. This clearly illustrates that in the SCR design process, NH3/NOX
uniformity has a far more important impact on performance than velocity non-uniformities.
The performance predictions for the economizer injection configurations (varying the number of
injection holes, from 15, 50 and 100) are shown in Figure 9. Figure 9 shows three separate
figures for the 15, 50 and 100 hole case, respectively. Each of the figures contains curves for the
ideal performance, performance based on the cold flow tracer gas results, and performance based
on the velocity and NH3/NOX profiles from the pilot plant. These results show very good
agreement between the pilot and cold flow model predictions for all three cases. The other
observation that is noteworthy, is the effect of the number of injection holes on the performance.
An improvement in performance can be seen when the number of injection holes was increased
from 15 to 50. No further increase in performance is predicted when the number of holes is
increased to 100.
A further example of the performance prediction methodology is shown in Figure 10 for the case
of injection at the throat location, varying the injection momentum (i.e., amount of transport air).
For these injection scenarios, the NH3/NOX non-uniformities were higher than for the
economizer exit configurations. As a result, particularly for the J=70 case, there is a marked
-------
departure from ideal performance (both in NOX reduction and NH3 slip). As the injection
momentum is increased, the performance is seen to increase. Again, excellent agreement is seen
between the cold flow model and pilot scale.
These results clearly show the utility of using this methodology to help guide the design of SCR
system. With the current methodology, design approaches can be tested, and the impact on
performance quantified. Thus the designers can revisit previous rule-of-thumb design criteria
such as requiring velocity or NH3/NOX uniformity to be below a certain standard deviation, or
requiring a set number of injection holes per unit duct area. This should allow more aggressive
SCR concepts to be confidently pursued, whereas previously they may have been discarded for
violating design criteria which evolved from prior empirical experience.
Full Scale Application
The design methodology described in this paper has been applied in a number of ways. It has
been used to aid in the detailed design of full scale SCR systems and it has also been used as a
screening tool to evaluate various retrofit concepts. For example the procedure was used to
assess a retrofit concept of an "in duct" SCR for a utility boiler. The initial concept utilized a
catalyst space velocity of 11,800 hr"1 and had a target performance goal of 90% NOX removal
with 10 ppm slip.
The cold flow modeling showed a standard deviation in the velocity profile across the catalyst of
10 percent. With a particular ammonia injection grid design the performance shown in Figure
11 was calculated. As can be seen in Figure 11 the predicted performance is under the target of
90% NOX removal with 10 ppm slip. For this particular case, pressure drop was not an issue and
the target performance goals could be met by increasing the depth of the catalyst. This is
illustrated in Table 1 where the model was used to assess the effect of decreasing the space
velocity; reducing the same velocity to 10,000 hr"1 achieves the required performance goals.
Table 1
Full Scale Retrofit: Effect of Increased Catalyst Depth on Performance
Space Velocity
hr-1
11, 800 (design)
11,000
10,500
10,000
ANOX
% @ 10 ppm NH3 slip
87
89
90
91
AP
in. H2O
0.93
0.99
1.04
1.09
-------
Discussions and Conclusions
The paper addressed a new SCR design methodology. The new methodology integrates the
results from physical cold flow modeling with an SCR process model. The procedure easily
allows the effects of velocity and NH3/NOX nonuniformities on SCR performance to be
quantified. With this new methodology, the designer does not need to rely on previous
"rules-of-thumb" engineering criteria. For instance, for a natural gas-fired unit is it really
necessary to engineer the system to have a velocity profile with a standard deviation of 10%?
Likewise, specification on the performance of the ammonia injection grid may be relaxed in
some cases in favor of an overall specification on performance. For instance, injection of
ammonia at the throat location may be dismissed based on prior design criteria. Now its
feasibility can be investigated and quantified. By offering a more quantitative approach for SCR
design it is expected that this will also translate into more aggressive, and therefore lower cost,
designs.
References
1. a. Patterson, R.C., Abrahamsen, R.F., Flow Modeling of Furnaces and Ducts, Combustion,
March 1962.
b. Industrial Gas Cleaning Institute Inc., Gas Flow Model Studies, Publication No. EP-7.
2. Teixeira, D.P., et al, Results of Catalyst Tests at the PG&E - EPRIASCR Pilot Plant,
presented at the 1995 Joint EPRI/EPA Stationary NOX Symposium, Kansas City, Missouri,
May 16-19, 1995. [conferencepaper]
-------
100
SV=10000
SV=15000
SV=20000
SV=25000
SV=35000
SV=50000
NHS/NOx, MOLAR
Figure 1
Ideal Catalyst Performance (Solid: DNOX, Dotted: NH3; Initial NOX = 75ppm)
-------
View
GAS FLOW
ECONOMIZER
SIMULATOR
(b)
Side View
Figure 2
Plan and Side View of the ASCR Reactor
-------
Oulslda Wall
\
1
If)
I
[^1 ' ''
VJi / ' V
^1^' 1 X .
Inside Wall
14 1/4'
12" -M- 12' -M-
6-1/4"
Plan View
88-3/4'
12-1 /4"
i
Turning Vane
Duct Transition
12" Dla.
4 3/81 4'
T UMiii^TT
/ !
(- 1 7-3/8" -»
.
Economizer
Simulator
Catalyst Simulator
6 3/8"
Side View
Figure 3
ASCR Cold Flow Model Schematic
-------
COLD FLOW MODEL
ASCR PILOT PLANT
PC = 35, SD = 2756
2CW
2
-------
.1
1
'£
Q
"E
to
TJ
Q.
.
o
_o
Q)
50
40 -
30 •
20
10
• Cold Flow Model
A Pilot, Cold Flow
O Pilot, Hot Flow
20
40
60
80
100
120
Pressure Coefficient
Figure 5
Effect of Pressure Coefficient on Velocity Uniformity
-------
COLD FLOW MODEL
SD = 21*
(a)
15 HOLES
(3 per lance)
ASCR PILOT PLANT
10 15 20 23 30 35 4O
SD =
(b)
50 HOLES
(10 per lance)
10 15 20 25 30 35 40 45
10 15 20 25 3O 15 40
SD = 7sc
(c)
100 HOLES
(20 per lance)
SD = 11*
Figure 6
NH3/NOX Profiles: Economizer Exit Injection:
(Effect of the Number of Injection Holes)
-------
COLD FLOW MODEL
ASCR PILOT PLANT
20-
15-
10-
SD = 40%
I I I I |
(a)
J = 70
m(air/FG) = 0.5%
SD = 3955
10
15 20 25 30 35 40 45
5 10 15 20 25 30 3540 45
= 325?
(b)
J = 150
m(air/FG) = 0.7%
20-
SD = 30*
40 45
5 10 15 20 25 30 35 40 45
Figure 7
NH3/NOX Profiles: Throat Injection, 15 Injection Holes
(Effect of Injection Momentum)
-------
100
Q.
Q.
CL
CO
CO
:r
<
v.0
X
O
Q
IDEAL
VELOCITY
NONUNIFORMITY
ONLY
VELOCITY AND
NH3/NOX
NONUNIFORMITIE^
0.00 0.10 0.20 0.30 0.40 0.50 0.60 0.70
NHS/NOx, MOLAR
0.80 0.90
1.00
1.10
1.20
Figure 8
Baseline SCR Performance Predictions: NH3 Injection Upstream of Gas Fan
(Velocity Standard Deviation: 39%, NH3/NOX Standard Deviation: 2%)
-------
15 HOLES
0.1 0.2 0.3 0.4 0.5
0.6 0.7 0.8 0.9
NH3/NOX
1.0 1.1 1.2 1.3 1.4 1.5
50 HOLES
0.0 0.1 0.2 0.3 0.4 0.5
0.6 0.7 0.8 0.9
NH3/NOX
1.0 1.1
1.3 1.4 1.5
100 HOLES
0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1.0 1.1 1.2 1.3 1.4 1.5
0.0 0
Figure 9
SCR Performance Predictions: Economizer Exit Injection
Varying the Number of Injection Holes (SV=26,000 hr'1, J=2200)
-------
100
0.00 0.10 0.20 0.30 0.40 0.50 0.60 0.70 0.80 0.90 1.00 1.10 1.20 1.30 1.40 1.50
NH3/NOX
(a)
J = 70
100
"§ 90
0.00 0.10 0.20 0.30 0.40 0.50 0.60 0.70 0.80 0.90 1.00 1.10 1.20 1.30 1.40 1.50
NH3/NOX
(b)
J=150
Figure 10
SCR Performance Predictions: Throat Injection
Varying the Injection Momentum (SV=26,000 hr'1, NO^ = 75ppm)
-------
CO
z
Q
60
50
40
x
O 30
20
10
0.70
•DNOx, ideal
-NH3 slip, ideal
-DNOx, model
-NH3 slip, model
0.80
0.90
NH3/NOx, MOLAR
1.00
1.10
Figure 11
Full Scale Retrofit Performance Predictions
-------
Various Types of SCR Plants
Full-Scale SCR
Compact SCR
In-duct SCR
Air Preheater SCR
Under Consideration of Achievable NOx Removal Rate and Cost-Effectiveness of
Catalyst Use
R- Sigling
A. Klatt
H. Spielmann
Siemens AG Power Generation KWU
P.O. Box 3220
D 91050 Erlangen
Germany
Abstract
SCR technology has proven itself for nitrogen oxide reduction in fossil-fired power plants.
Depending on spatial conditions, different types of SCR plants can be realized in a power plant. In
some individual cases, detailed investigation is required to determine how the requirements of a
specific NOx reduction rate can be fulfilled.
This paper describes the primary differences hi these various SCR plant types based on catalyst
designs, such as achievable NOx removal rates and cost-effectiveness of catalyst use.
-------
Introduction
The technique of selective catalytic reduction (SCR) has proven itself as a secondary measure for
the reduction of nitrogen oxide emissions from fossil-fired power plants due to its capacity and
operating reliability both in the backfitting of existing combustion facilities as well as in new
plants. NOx removal rates of up to 95% with simultaneous adherence to an NHs slip of 5 ppmv or
less can be achieved with this process.
In planning for a new plant or in backfitting with favorable space conditions, the SCR plant is
generally placed between the steam generator and air preheater, using the optimum temperature
range for the catalytic reaction, in a "high-dust" configuration (Pig. 1). The location of the SCR
plant in the flue gas path as well as structural constraints for the amount of catalyst material to be
installed and hence the performance data of the system represent a further type of classification. In
addition to conventional SCR technology - installation of a "full-scale" SCR - newer systems such
as compact SCR, in-duct and air preheater SCR have recently been discussed as concepts or have
already been implemented, particularly for backfit measures (1), (2).
Types of SCR Plants
Common to all types of SCR plants are the components reducing agent storage and evaporation,
reducing agent injection, catalyst, enclosure and instrumentation.
The significant characteristics of the various types of high-dust SCR plants are the intake velocity
of the reactors and the number of catalyst levels (Figure 2).
Full-scale SCR plants for coal-fired power plants are designed for a velocity of approx. 14-17 ft/s
based on the catalytic reactor cross section and on 3 to 5 reactor levels of installed catalyst
material. The pressure drop on complete filling with catalyst is approx. 4 in. water gauge.
The reactor of a compact SCR plant is characterized by an increased gas velocity of 22-25 ft/s.
To limit the pressure drop on the flue-gas side to roughly 4 in. water gauge, the number of the
catalyst levels must be reduced to 2 or max. 3 levels.
In in-duct SCR plants the catalysts in the existing flue gas duct upstream of the ah- preheater are
operated at typical gas velocities of roughly 32-38 fVs. The volume is limited to 1 or max. 2 levels
to ensure that a pressure drop of roughly 4 in. water gauge is not exceeded.
In the air preheater SCR process, catalytically active heat transfer elements are used on the hot
side of the air preheater in 1 or max. 2 layers. A complete replacement of all layers of heat
transfer elements by catalysts is not possible due to the decreasing gas temperature in the flow
direction. On the one hand, the catalyst activity decreases with decreasing temperature, and on the
other there is a danger of ammonium sulfate formation hi the low temperature range (2). The
pressure drop across the air preheater is not influenced by the installation of the catalyst. Heat
transfer properties are only insignificantly impaired.
-------
The installation of plate-type catalysts is possible in all SCR plant types. These converters are
characterized by a high erosion resistance, even at high inlet velocities and ash contents.
Temperature cycle resistance is outstanding, especially when the catalysts are implemented in the
air preheater.
SCR Catalyst Design
In order to represent the principal differences of the aforementioned SCR plant types, the flue gas
data for a fictitious power plant, equipped with a coal-dust fired generator with dry ash removal
were applied as design conditions. For comparison purposes, all SCR plant types which are
described below were designed with this power plant data.
Power plant and flue gas data
Table 1 shows a summary of the relevant data for dimensioning the catalyst.
Table 1:
Fuel fired Ib/hr 150,000
Flue gas flow rate Ib/hr 2,000,000
HHV BTU/lb 13,000
Gas temperature °F 700
Inlet NOx Ib/MMBTU 0.4
H20 Vol. %, wet 8
O2 Vol. %, wet 4
SO2 ppmvd 1000
Particulate loading gr/cf 3.5
Catalyst design conditions
NOx Reduction Efficiency. Inlet NOx concentration and the permissible NOx emissions limit
stipulated by law are the factors on which the required NOx removal rate is based. This removal
rate can be used as a selection criterion for determining which of the various SCR plant types is
best suited. Achievable removal rates of the respective SCR plant types are discussed in the
below.
NH3 slip. In addition to the NOx removal rate, the maximum allowable NH3 slip at given flue gas
conditions is of particular importance in determining the required catalyst volume.
In coal-fired plants, the NH3 slip downstream of the SCR plant, which is gaseous at first, is
increasingly adsorbed on the fly ash on cooling of the flue gases in the air preheater. In
dependence on the type and preparation of the coal fired, a fraction of approx. 70-100% of the
-------
NH3 slip is retained in the fly ash. Recycling of hard coal fly ash in such areas as the flooring or
concrete industry can be limited by bad smells in processing if the NH3 content of the ash exceeds
specific limits (80-150 ppmv).
Based on experience from the operation of SCR plants in Germany, a maximum NH3 slip of 2
ppmv is applied for the design to ensure the recyclability of ash from plants firing high-grad coals.
As the following analysis shows, this value can be realized in full-scale and compact SCR plants at
NOx removal rates > 60% given appropriate dimensioning.
Operating time. The operating time is that time during which the catalyst achieves the required
removal rate. Catalyst activity decreases over time due to aging and catalyst poisons, until it is
finally insufficient for the required NOx removal. A specific deactivation curve based on type of
fuel fired and continuous feedback of information from existing plants is used for dimensioning of
the catalyst. The volume is then dimensioned such that the required removal rate is still achieved
at the end of the guaranteed operating time.
Depending on the type of SCR plant certain operating periods can reasonably be achieved until a
first catalyst exchange or an additional catalyst loading is necessary. These are typically as
follows:
Full-scale SCR 24,000 h
Compact SCR 16,000 h
In-ductSCR 14,000 h
Air preheater SCR 24,000 h (in case of NH3 slip control only)
Catalyst selection. The selection of a suitable catalyst composition and geometry are of
decisive significance for reliable operation of SCR plants. Siemens is the only catalyst
manufacturer offering both honeycomb and plate-type catalysts, and is therefore also experienced
in selecting the optimum catalyst type.
For implementation in ash-containing flue gases and especially at very high flue gas velocities,
plate-type catalysts have proven themselves due to their high erosion resistance and mechanical
stability as well as their low pressure drop in comparison to honeycomb catalysts (3).
The catalyst designs for all of the SCR plant types considered here are based without exception
on a highly active, selective Siemens plate-type catalyst. The specific surface area is 110 ft2/ft3,
corresponding to a pitch of 0.2 inches. The individual catalyst elements have a cross section of
18.2 inches * 18.2 inches with a plate height of 22.4 inches in the flow direction, and can be
installed in all types of SCR plants. However, depending on the requirements, other deviating
catalyst geometries can be implemented, such as those with lower or higher specific surface areas.
This is particularly important if a specific pressure drop is determining for the design.
-------
Inlet flow conditions. For optimum utilization of the catalyst, minimum requirements must be
met for the intake flow to the SCR reactor as regards temperature, flue gas velocity and reactant
concentration. It is advantageous to optimize the flow using model experiments (4).
An inlet flow quality corresponding to the state of the art is assumed for the comparison of the
SCR systems.
Representation of Catalytic Reactor Design Data for Individual SCR Systems
In addition to the linear velocity limits mentioned above, the total number of levels available for
loading with catalyst was limited to obtain similar pressure losses (app. 4 in. water gauge) for
each system with a full catalyst load. This enabled a reasonable comparison of the full-scale,
compact and in-duct SCR systems. Table 2 shows an overview of the SCR plant types examined.
Full-scale SCR
The result of the catalyst design is shown in Table 3. The design yielded a reactor with a total of 4
levels, each of which can be charged with 2 layers of catalyst. With an initial loading of 9310 ft3 of
catalyst on 5 layers and an initial operating time of 24,000 operating hours (3 years), an NOx
removal rate of 80% is achievable. This configuration retains 3 layers free for subsequent
recharging.
Compact SCR
The result of the catalyst design is shown hi Table 4. A reduction of the reactor cross section by
approx. 35% over that of a full-scale plant and limiting the number of available levels to 3 yielded
a compact SCR plant configuration.
This design type is characterized in the selected example by a reactor with a total of three
available levels. For a catalyst initial operating time reduced to 16,000 hours an initial catalyst
charge volume of 6100 ft3 in 5 layers can achieve an NOx removal of 60%. In this case only a
single catalyst layer is available for future recharging with catalyst.
In-duct SCR
The result of the catalyst design is shown in Table 5. The design assumed the existence of two
50% flue gas ducts upstream of the air preheaters. Each of the two flue gas ducts is charged with
1625 ft3 of catalyst on 2 levels of 2 layers each. Due to space limitations and to limit pressure
drop, no free space is planned for recharge catalyst.
Under the very high space velocities (SV > 7000 1/h, flue gas volume flow rate divided by
catalyst volume) typical for this application, it is not possible to adhere to an NHs slip of 2 ppmv
-------
at relevant NOx removal rate over operating times which are reasonable for the utility. An NOx
removal rate of 25% can be achieved with the aforementioned volume, even if the NH3 slip is
increased to 5 ppmv for an initial operating time of 14000 hours.
APH SCR
The result of the catalyst design is shown in Table 6. The design assumed two 50% flue gas ducts
and air preheaters.
Charging the hot end area of each ah- preheater with one catalyst layer, based on the volume
through which the flue gas flows (only roughly hah0 of the installed volume), yields space
velocities of 31,000 1/h. These conditions result in a low NOx removal rate of approx. 5%.
However, an APH SCR system can be used to reduce the NH3 slip from upstream systems such as
compact SCR or in-duct SCR. This enables the achievement of higher NOx removal rates in
upstream systems. For example, if the in-duct SCR plant discussed here is operated in
combination with an APH SCR system and the overall NOx removal rate remains at 25% the
ammonia slip will then be reduced to 2 ppmv.
Similar results are obtained with the compact SCR plant in combination with an APH SCR. The
NOx removal rate increases from 60% to 73% while the ammonia slip is maintained at a level of 2
ppmv.
Catalyst Management Plan
All of the above catalytic reactor design performance data are based on the initial period of
operation. At the end of this period NH3 slip exceeds the permissible limit due to declining
catalyst activity. It is therefore necessary to either install additional catalyst material in the reactor
or to replace a portion of the deactivated material with fresh catalyst. A catalyst management plan
should be established during the planning phase to determine catalyst material requirements.
To compare the SCR systems described with regard to catalyst use, a catalyst management plan is
determined for an operating time of roughly 20 years. Especially in high-dust SCR plants, catalyst
utilization has a significant influence on operating costs. The catalyst replacement strategy is
highly dependent on the specific design of the SCR reactor (availability of reserve levels). Plate-
type catalysts are constructed such that two catalyst layers can be installed on each level. In many
cases the addition or replacement of only a single layer (1/2 level) is sufficient to ensure adherence
to the emission limits until the next or next two inspections.
A replacement strategy which foresees the possibility of charging additional catalyst when the
allowable NH3 slip limit is first reached uses a higher fraction of the catalyst activity associated
with the catalyst charged. In the event that adherence to the required NOx removal rate
necessitates early replacement of portions of the installed catalyst, the relatively high residual
activity (generally roughly C0% of the initial value) of catalyst remains unused.
-------
Therefore, the design of the catalyst as an "environmental protection system" should not only
account for the cost optimization, but should also be viewed with regard to the minimization of
waste materials generated.
Full-scale SCR
The catalyst management plan is shown in Figure 3.
At the end of the third operating year, the half-level which is still empty is filled with additional
catalyst elements, and at later points in time the two remaining half-levels are filled. The complete
charge enables NOx removal to continue for more than 8 years without removal of catalyst. After
this time, further operation of the plant is ensured by the successive replacement of one complete
catalyst level at a time, starting with the most fully exhausted catalyst level in the charge. This
ensures maintenance of the guaranteed removal rate over at least two further operating years after
each recharging or replacement.
Compact SCR
The catalyst management plan is shown in Figure 4.
In this type, the SCR reactor is completely filled by the addition of a half-level of catalyst after
16,000 hours. This enables operation for a further two years. Subsequently, the replacement of at
least one level at each interval is required for continued operation with acceptable service intervals
(a minimum of 2 years).
In-duct SCR
The catalyst management plan is shown in Figure 5.
According to the design, an NH3 slip of 5 ppmv is attained after 14,000 hours operation. It is not
possible to charge additional catalyst, i.e. after approx. 2 years time 50% of the initial catalyst
charge in each of the two ducts must be replaced by new material.
This replacement however enables continued operation for only slightly more than one operating
year.
APH-SCR
It is assumed hi this case that deactivation takes place here as it does in other catalyst systems.
This does not, however, decrease the systems's ability to significantly reduce low concentrations
of NH3. Consequently APH catalyst materials for NH3 reduction can remain hi service for
extended periods without the need for replacement. Cost-effectiveness assessment is based on an
assumed service life of 24,000 h.
-------
As a result, complete replacement of the catalyst material is necessary about once every three
years. This design therefore does not require a catalyst management plan.
Discussion
Careful consideration of all economic aspects reveals that the installation of an SCR plant tailored
to the individual power plant, i. e. smallest possible SCR plant, is not always the optimum
solution.
As the aforementioned designs for catalyst management of the individual SCR plant types have
already shown, the catalyst management plans differ greatly. This leads to corresponding
differences in the expenditure required to maintain the removal rate, in the catalyst requirements
and in the utilization of the catalyst.
Another factor to be considered is the number of unit outages for the addition of catalyst or
catalyst replacement. As it can be seen in the figures for catalyst management plans, the number of
unit outages is much higher for all plant types other than full-scale. The duration of the plant
outage for the addition of catalyst can be taken as approximately equal for the individual SCR
plant types, and is roughly 3 to 5 days .
Figure 6 shows the NOx reductions achievable with the individual SCR plant types, expressed in
tons of NOx over a time frame of 20 years. Full-scale, compact SCR and the combination of
compact SCR with an APH SCR achieve the highest values, corresponding to the higher NOx
reduction rates. As an APH SCR plant is not especially effective as an NOx reduction measure, it
should therefore only be implemented for NH3 slip reduction in combination with other measures
such as compact or in-duct SCR.
Figure 7 shows catalyst consumption including the initially installed volume over a 20 year period.
Compact and full-scale plants have virtually the same material requirements. An in-duct plant has
only slightly lower catalyst material requirements. A combination of compact or in-duct SCR
plants with an APH SCR plant results in the highest catalyst material requirement. It is imperative
that these generalizations be considered hi the light of the NOx reduction requirements for
individual plants in question.
Figure 8 shows specific catalyst consumption (ft3 catalyst/1000 t NOx reduced). The specific
requirements are considerably higher for all other than full-scale SCR plants.
Figure 9 shows the volume of catalyst material to be disposed of over a 20 year period. As is
evident, differences are significant, and the full-scale plant is most advantageous in terms of
disposal requirements.
The comparison of various SCR plant types shows that the catalyst is best utilized in full-scale
SCR plants. High priority should be given to the possibility of later addition of catalyst in planning
an SCR plant. This enables the most cost-effective use of the catalyst. A catalyst management
-------
practice which foresees the possibility of adding catalyst when the ammonia slip begins to exceed
the guaranteed ammonia slip limit utilizes more of the catalyst activity then does the practice of
replacing the catalyst charge.
At first glance, the combination of a compact or in-duct SCR plant with an APH SCR plant is
interesting for improved NOx removal. However, the high requirements over a longer time frame
show, that the associated high specific catalyst requirements are not cost-effective.
Inclusion of the expenditure for plant construction in the comparison does not lie within the scope
of this paper. However, roughly the same expenditures are required for all SCR plant types for the
following components:
NHs storage
NHs injection and control
Plant instrumentation
I&C.
In summary, it can thus be stated that a full-scale SCR system represents the most favorable
solution with regard to the parameters considered. The other SCR plant types discussed, such as
compact and in-duct SCR or combinations with an APH SCR system, could serve reasonable
niche functions under the consideration of special requirements such as
Limited available space
Use in a peaking plant with low yearly operating hours
Relatively short remaining plant service life
Acceptance of lower NOx reduction rates or higher NHs slip values.
-------
Literature:
1. Jan Zmuda. "Hitting the moving target of emissions control." Private Power Executive,
September-October 1993.
2. K. Huttenhofer, J.-K. Beer, H. Smeets, J. van der Kooij. "The DeNOx Air Preheater
Downstream of a Coal-Fired Boiler." EPRI/EPA 1993 Joint Symposium on Stationary
Combustion NOx Control, Miami Beach, Florida.
3. H. Spielmann, K. Huttenhofer. "DeNOx Design Crtiteria and Operating Experience in Coal-
Fired Power Plants." International Joint Power Generation Conference, October 1993, Kansas
City.
4. K. Hauenstein, W. Herr, R Sigling. "Fluid Dynamic Optimization of SCR Plants by Modelling
Demonstrated by the SCR Plant at Keystone Power Station (USA)." EPRI/EPA 1995 Joint
Symposium on Stationary Combustion NOx Control.
-------
SIEMENS
Eco bypass
Boiler
DeNOx reactor
NH3 injection
Reactor bypass
Air preheater
Forced draft fan
Soot
blower
XIXIXIXIX
XIXIXIXIX
Guide vanes
Flow straightener i
Spare layer |
Catalyst layer
i
Catalyst layer
Electrostatic
precipitator
Stack
Figure 1: High-Dust SCR Arrangement
-------
SIEMENS
Flue gas velocity (ft/s)
Number of layers
(initial load)
Number of catalyst
spare layers
Flow direction
indicates one
catalyst layer
indicates one
spare layer
Full-Scale SCR
14-17
5-9
1
Compact SCR
22-25
4-5
1
In-duct SCR
32-38
3-4
0
APH SCR
40-44
1 -2
0
1
Figure 2: Characteristic of different SCR-Types
-------
SIEMENS
Catalyst Volume [ft3]
Spec, surface area [ft^ft3]
Height of one catalyst layer [inch]
Initial catalyst layers
Spare layers
SV value [1/h]
Flue gas velocity [ft/s]
Pressure drop [in. wg]
Guarantee period [h]
NOX reduction efficiency [%]
NH3 reduction efficiency [%]
NH3 slip at the end
of the guarantee period
Full-Scale SCR
9,310
110
22
5
3
2,580
16
2
24,000
80
—
2
Compact SCR
6,100
110
22
5
1
3,940
24
3
16
60
-
2
In-duct SCR
2*1625
110
22
4
0
7,390
31
4
14
25
h
5
0
APH SCR
2*777
110
22
1
0
31,000
39
5
24,000
-
50-70
Table 2: Overview of the SCR plant types examined
-------
SIEMENS
Catalyst Data:
Specific surface area
Pitch
Height of one layer
Catalyst volume
110 ft'/ft3
0.2 inch
22.4 inch
9310 ft3
Reactor Data:
Level 1 with 2 catalyst layers 78.0 tons
Level 2 with 2 catalyst layers 78.0 tons
Level 3 with 1 catalyst layer 53.0 tons
Level 4 with 2 spare layers
Reactor cross section area
Flue gas velocity
Pressure drop (initial)
(final)
Initial guarantee period
NOX reduction efficiency
NH3 slip
SO2 conversion rate
33' * 34'
15.7 ft/s
1.85 inchwg.
3.0 inch wg.
24,000 hours
80 %
< 2 ppmvd
Table 3: Catalyst Design Data Sheet for Full-Scale SCR
-------
SIEMENS
Catalyst Data:
Specific surface area
Pitch
Height of one layer
Catalyst volume
110 fP/fP
0.2 inch
22.4 inch
6100 ft3
Reactor Data:
Level 1 with 2 catalyst layers 51.0 tons
Level 2 with 2 catalyst layers 51.0 tons
Level 3 with 1 catalyst layer 35.0 tons
Reactor cross section area
Flue gas velocity
Pressure drop (initial)
(final)
Initial guarantee period
NOX reduction efficiency
NH3 slip
SO2 conversion rate
25' * 30'
23.9 ft/s
3.1 inchwg.
3.7 inch wg.
16,000 hours
60 %
< 2 ppmvd
Table 4: Catalyst Design Data Sheet for Compact SCR
-------
SIEMENS
Catalyst Data:
Specific surface area
Pitch
Height of one layer
Catalyst volume
110 fP/ft3
0.2 inch
22.4 inch
1625 ft3 (two ducts are
equipped with catalysts)
Reactor Data:
— Level 1 with 2 catalyst layers 17.0 tons
Level 2 with 2 catalyst layers 17.0 tons
Dust cross section area
Flue gas velocity
Pressure drop (initial)
Initial guarantee period
NOX reduction efficiency
NH3slip
SO2 conversion rate
20'* 13'
30.8 ft/s
3.8 inch wg.
hours
14,000
25
<5
<0.5
ppmvd
Table 5: Catalyst Design Data Sheet for In-duct SCR
-------
SIEMENS
Catalyst Data:
Specific surface area
Pitch
Height of one layer
Catalyst volume
flue gas
110 ft2/ft3
0.2 inch
22.4 inch
805 ft3 (two APH are
equipped with catalysts)
Cross section area
Flue gas velocity
Pressure drop (initial)
Initial guarantee period
NH3 reduction efficiency
862 conversion rate
200
39.4
4.7
24,000
50 - 70
<0.2
ft2
ft/3
inch wg.
h (for NH3 control)
Table 6: Catalyst Design Data Sheet for APH SCR
-------
SIEMENS
Catalyst addition
1862 ft3 each
Catalyst replacement
3723 ft3 each
'
'
20, 000 hrs
+12
-Ho
+4
20000
40000 BOOOO BOOOO 100000
Operating Period (hours)
120000
140000
a
a
a.
•rt
I—I
en
I
m
I
z
Figure 3: Catalyst Management Plan for a Full-Scale SCR
-------
SIEMENS
-H
>
-H
-P
U
4J
CD
ro
-p
CD
u
Catalyst addition
1220 ft3
Catalyst replacement
2440 ft3 each
-12
-10
0 20000 40000 60000 80000 100000 120000 140000
Operating Period (hours)
E
a
a
a.
en
i
m
I
Figure 4: Catalyst Management Plan for a Compact SCR
-------
SIEMENS
Catalyst replacement
1625 ft3 each
--12
--10
0 20000 40000 BOOOO BOOOO 100000 120000 140000
Operating Period (hours)
E
a
a
Cfl
i
m
Figure 5: Catalyst Management Plan for an In-duct SCR
-------
SIEMENS
45,000
40,000 -
35,000 -
30,000 -
rjj, 25,000 -
c
^ 20,000 -
15,000
10,000
5,000
0
NOX reduced
-1-
>
4-
•4-
i| L
O
CO
_0)
s
CO
o:
o
CO
•S
(0
Q.
O
O
O
CO
t>
TJ
X
Q.
(0
(0
Q.
E
o
O
ro O
tS Qi
^ O
•9 co
c
40,000 n
Catalyst consumption
Figure 6: NOX Reduced [tons] over an
Operating Period of 20 Years
Figure 7: Catalyst Consumption [ft3]
for an Operating Period of 20 Years
-------
SIEMENS
Specific Catalyst Consumption
3,000
2,500 -
2,000 -
i/r
o
2 1 500 -
O '
o
T—
jfc,
1,000 -
500 -
0 -
o:
o
CO
CD
8
CO
3
u_
-
I
^
01
o
CO
tj
(0
a.
o
O
CC
O
CO
•5
ri
•a g
|S
2.0
Eco
o x
<
•
-
-
-
-
•
-
T.
Q. .
li
ro 0
t3 <£
D O
•o co
_c
Catalyst Volume to be Disposed of
35,000 -T
30,000 -
25,000 -
20,000 -
E
15,000 -
10,000 -
5,000
0 -
•••
1 ^«i O
S'-e
1 c Q
1 o o:
lo o
L.
.
-
.
i
. .
i
_ _
, i
c
.0
a
a:
o
CO
Q.
• ^__
IR
• i ^ -° i
Mil
I MI
i i i
l± QL 01 _i_
•••••
.
.
.
0
t:
o
Q.
a:
o
CO
Q.
s x
O O O -o c Q. .
CO CO CO C § < -Q
SCO D O ry> m (}
a. -a ro ^
-------
OPTIMIZING SCR CATALYST DESIGN
AND PERFORMANCE FOR COAL-FIRED BOILERS
Scot Pritchard
Chris DiFrancesco
Cormetech, Inc.
5000 International Drive
Durham, NC 27712
Shozo Kaneko
Norihisa Kobayashi
Mitsubishi Heavy Industries, Ltd.
Yokohama, Japan
Kohei Suyama
Kozo lida
Mitsubishi Heavy Industries, Ltd.
Nagasaki, Japan
Abstract
The design of an SCR system for a coal-fired application is a challenge due to particulate,
catalyst poisons, and 862 in the flue gas. Our experience shows that coal-fired SCRs are
successful when the system impacts and catalyst deterioration factors are understood and
specific countermeasures are implemented in system and catalyst design.
There are several factors that the system and catalyst designers must consider in order to
assure success. They include: type of boiler, required performance, fuel and ash analysis,
ash loading, type of SCR, inlet conditions, catalyst deactivation mechanisms, and impact
on downstream equipment.
This paper explains each system impact and catalyst deterioration factor that we'have
experienced. For each factor, it presents the specific countermeasures that have resulted
in successful SCR systems. Also, it discusses the design of improved catalysts derived
from our experience. Finally, two case studies illustrate the impact of optimized design.
-------
Introduction
Selective Catalytic Reduction (SCR) is recognized worldwide as the most
effective NOx Control technology for utility boilers and combustion turbines when
substantial NOx reduction of 50% to 95% is required. In addition to its proven
high performance, it has also become an economically viable solution, with
current installed costs, in the United States, estimated at between $20/kw for
natural gas and $50/kw-$70/kw for coal units. The technology has even given
some utilities the capability to achieve lower heat rates by allowing optimization
of burner operation and reduction or omission of flue gas recirculation (FGR).
Mitsubishi Heavy Industries (MHI) pioneered the development and application of
SCR technology starting in the late 70's in Japan, and installed the world's first
High-Dust SCR application on a 175 MW coal fired boiler in 1980. MHI
transferred their technology through licensing agreements to several companies
in Europe and to Cormetech in the United States. Over 319 units including 56
coal fired boilers worldwide operate successfully using this technology.
Figure 1
Catalysts for Coal, Oil, and Gas Applications
Cormetech is a joint equity company of Corning Incorporated, MHI, and
Mitsubishi Chemical Company (MCC). Cormetech designs and manufactures
homogeneous Titanium-Tungsten-Vanadium (Ti-W-V), extruded honeycomb
catalysts for fossil-fuel-fired applications. Figure 1 shows a photograph of
catalysts for coal, oil, and gas applications. Cormetech draws upon the vast
experience database of MHI, the extrusion and catalyst know-how of Corning,
and the catalyst technology of MCC to provide an optimized product to the
market.
Page 1
-------
This paper addresses the methods successfully implemented to design and
optimize catalyst performance in coal fired boiler applications. The basics of
deNOx catalysis are reviewed in preparation for the discussion on system
impact, catalyst deterioration mechanisms, and countermeasures that follows.
Additionally, catalyst design optimization is presented by a discussion of
improved catalysts for coal applications. Two case studies (pulverized coal and
cyclone boilers) comparing conventional and improved catalyst are presented.
Background of SCR Reaction
In coal fired boilers in which SOx is present in the flue gas, two chemical
reactions that occur in the presence of the SCR catalyst are of most importance
to the following discussion.
One reaction, of course, is the deNOx reaction:
4NO + 4NH3+ 02 catalyst ) 4N2+ 6H2O (1)
Figure 2 illustrates this primary reaction mechanism. Starting in the upper left
hand corner of the figure and proceeding clockwise is:
1. Active catalytic site (Me = metal)
2. Adsorption of the ammonia molecule
3. Reaction of NO with NH3 forming N2 and H2O
4. Regeneration of active site by oxidation
The second reaction is oxidation of sulfur dioxide:
SO2 + -02 ""*»* ) SO3 (2)
«
The NOx conversion rate and SO2 oxidation rate will partly depend upon the rate
in which the reactants diffuse through the walls of the porous catalyst to reach
active sites where reaction takes place. Through our experience we have
determined the rate of diffusion of reactants for our catalyst. For deNOx
conversion the rate of chemical reaction is fast relative to the rate of diffusion.
Therefore, the catalyst is effective primarily at the surface of the wall. On the
other hand, the SO2 oxidation chemical reaction rate is slow relative to the rate
of diffusion so that the reactants diffuse into the entire wall. This conclusion is
relevant to the subsequent discussion on system impacts, catalyst deterioration
mechanisms, and countermeasures.
Page 2
-------
NH3
\
H
O O
II I
-Me--O-Me~
Active Site
0 O
--Me-O--Me-
NH3 Adsorbtion on Active Site
4NO + 4NH3 + O2
6H2O
N2 H2O
H
O
H
I
O
-Me-O--Me--
Reqeneration of Active Site
N H
>'
V
O
-Me--O-Me~
Reaction of NO with NH3
Figure 2
SCR Surface Reaction Mechanism
HIGH
ousr
SYSTEM
LOW
DUST
SYSTEM
Figure 3
SCR Configurations
PageS
-------
System Design Impacts & Counter-measures
Figure 3 shows two typical layout options of SCR systems applied to coal fired
boilers. The Low-Dust or Tail-End option shows the installation of the SCR
reactor after the air preheater (APH), Electrostatic Precipitator (ESP), and Flue
Gas Desulfurization (FGD) systems. This option requires the flue gas to be
reheated to acceptable SCR temperatures, typically 550°F - 750°F. Low-Dust
designs have primarily been used on wet bottom boilers which have ash
recirculation, due to concern over catalyst degradation caused by arsenic
poisoning. The High-Dust option locates the SCR reactor between the
economizer exit and the APH inlet. This is more traditional for dry bottom boilers
and less costly, and is now being applied to wet bottom boilers based on
economics, advancements in catalyst resistance to arsenic poisoning, and
limestone addition to the fuel which will be discussed later.
CONTROL SYSTEM
AQUEOUS NJH3
TANK
ECONOMIZER BYPASS
STATIC MIXER
NH3 INJECTION GRID
TURNING VANES
SOOTBLOWERS
FAN
Figure 4
SCR Components
The SCR system consists of many components in addition to the catalyst.
Figure 4 depicts a high-dust SCR system which includes the catalyst reactor,
economizer bypass used for part-load temperature control, static mixer used for
temperature and/or ammonia mixing, turning vanes, ammonia vaporization
system, ammonia injection grid (AIG), sootblowers, and the control system.
Since these components directly influence overall system effectiveness,
Cormetech works closely with system equipment suppliers to ensure proper
design.
Page 4
-------
Operational impacts of an SCR system on coal fired boilers include draft loss,
SO3 formation, and ammonia emission. Pressure drop is caused by the
installation of the catalyst material in the flue gas stream and is typically
minimized by expanding the flue gas path. Typical pressure drop values for
conventional high dust applications range from 0.5 to 1 inch of water per catalyst
layer with 2 - 3 layers installed, with an additional 1 - 2 inches for the other
components of the system.
As shown in equation (2) the SCR catalyst will convert a small amount of SO2 to
S03. This conversion rate is typically less than 0.5% per layer depending upon
the catalyst formulation, deNOx requirements, and operating temperature. If the
SO2 oxidation rate is too high, corrosion and plugging problems will occur in the
APH due to the formation of H2SO4 and NH4HSO4.
As a countermeasure, our catalyst is designed to achieve high deNOx activity
while keeping SO2 oxidation activity low. When considering the catalyst
formulation, a high concentration of vanadium will result in a high deNOx
activity, but will also result in a high SO2 oxidation activity. One countermeasure
to this is a vanadium concentration is selected such that there is sufficient
deNOx activity with an acceptable level of SO2 oxidation.
As a system countermeasure, the cold end surface of a regenerative-type APH
may be enameled and/or sootblowing capabilities may be enhanced.
Ammonia slip may contaminate flyash, and in combination with SO3, cause APH
plugging. Designers can combat the effects by considering:
• 2 ppm to 5 ppm NH3 slip at end-of-life.
• ammonia injection grid with adequate coverage and distribution capability.
• multi-point outlet NOx sampling control grid to assure representative NOx
input value to control system.
• permanent measuring grid for AIG tuning to assure proper distribution of
ammonia.
• flyash ammonia concentration monitoring.
Catalyst Deterioration Mechanisms & Countermeasures
When designing for a given application, designers must carefully review fuel and
ash constituents (see Table A). Under ideal conditions the catalyst will reduce
NOx for an unlimited period of time. However, at actual full scale operating
conditions, catalyst deactivation will occur over time. Table B lists the primary
deterioration mechanisms versus the type of coal fired unit, wet or dry bottom.
Page 5
-------
Ash Moisture, %
Total Sulfur in Coal, %
Trace Material, ppm
Ni
Cr
As
Cl
Ash Analysis, %
SI O2
As2O3
Fe2O3
CaO
MgO
TiO2
MnO
V2O5
Na2O
K2O
P2O6
SOS
6-33
0.6-1.6*
3-40
7-46
1-25
41-1,900
41-71
2-33
2.5-10
2.4-26
0.7 - 49
0.1-1.8
0.02 - 0.2
0.01 - 0.1
0.05-1.6
0.1-4.0
0.06-1.3
1.6-16.5
* Application Experience on Oil up to 5.4% Sulfur
Table A
Typical Fuel/Ash Composition
Boiler Type
Sintering
Accumulation of
Alkaline Metal
Alkaline Earth Metal
Masking
Accumlation of
Arsenic Oxide
Flyash Deposition*
Erosion*
Wet Bottom
Negligible
Small
Large
Large
with Ash
Recirculation
Small
Small
Dry Bottom
Negligible
Small
Large
Moderate
Small
Small
•Highly dependent on proper design of SCR system
Table B
Main Causes of Catalyst Deterioration
Page6
-------
Based on a sound understanding of the deterioration mechanisms, specific
countermeasures are implemented, such as:
• system design evaluation.
• catalyst formulation.
• catalyst volume selection.
• catalyst geometry.
• catalyst macro and micro pore design.
• catalyst characterization testing and predictive tools.
Thermal Sintering
Thermal sintering is the growth of primary catalyst particles resulting in a
reduction of catalyst surface area which reduces catalyst performance. Figure 5
illustrates this mechanism. Thermal stability is maximized with the incorporation
of Tungsten in the catalyst formulation. As a result, sintering is negligible at
normal SCR operating temperatures.
SECONDARY
PARTICLE OF
TiO2
PRIMARY
PARTICLE OF
TI02
FRESH CATALYST
HEAT SINTERED CATALYST
Figure 5
Thermal Sintering
Alkaline Metal (Na, K)
Alkaline metals may directly react with active sites and render them inert'as
shown in Figure 6. Since the deNOx reaction takes place primarily on the
surface, the degree of deactivation depends on the surface concentration of the
alkaline metals. In a water soluble form, these alkaline metals are highly mobile
and will migrate throughout the catalyst material. Since the walls of our catalyst
consist entirely of catalyst material, the surface concentration of alkaline metals
is diluted by this migration, minimizing the deactivation rate.
Page?
-------
H
I
O O
II I
-Me—O—Me—
FRESH CATALYST
Na+ (K+) Na (K)
O O
II I
~Me--O--Me--
AKALINE DETERIORATION
Figure 6
Alkaline Deterioration Mechanism
For the subject catalyst in coal-fired applications, it is our experience that this
type of poisoning has a small impact, since most of the alkaline metals in the
coal ash are not water soluble. The impact is larger in oil fired applications
where the majority of the alkaline metals in the ash are water soluble or when the
thickness of catalyst material is low. For example, given the same concentration
of alkaline metals in the flue gas, this catalyst will have a lower surface poison
concentration than a coated product and, thus have a lower deactivation rate
and longer useful life as shown in Picture 7.
~ 100
O
0)
* 80
«j
o
V
cc
x 60
O
Solid Honeycomb Type
Coated Type
10 20 30
Operation Period ('000 hrs)
Figure 7
Comparison of Durability (Oil Firing)
40
Arsenic (As)
Arsenic poisoning is caused by gaseous arsenic AsaOa in the flue gas. The
As2O3 diffuses into the catalyst and solidifies on both active and non-active sites,
as shown in Figure 8.
PageS
-------
As with alkaline metals, arsenic poisoning is best abated by use of
homogeneous catalyst compositions, which effectively reduce the surface poison
concentration. Since the mechanism of arsenic poisoning is more diffusion-
limited and subject to capillary condensation, optimized catalyst pore structure
also lessens its effects. Further, engineers use accumulation rates determined
by both lab and field tests to ensure a proper catalyst volume is provided for
each specific application.
E3 AS2O5
M Active Material
Support (Tl-W)
O
II
H
1
O
1
AS2O3 O2
° M^>
it i — ^
-Me-O-Me-O-Me-
ti fl "fi
-As-O-As-O-As-
O 0 O
i T i
-Me-O-Me-O-Me-
FRESH
GASEOUS ARSENIC DETERIflHaTlpM
Figure 8
Model of Arsenic Compound Accumulation
A system countermeasure incorporates the use of a fuel additive. As mentioned
previously, wet bottom boilers with 100% ash recirculation present a worst case
scenario for catalyst deterioration caused by arsenic poisoning. In order to
combat the high levels of gaseous arsenic (As2O3) in cyclone boilers, limestone
may be added to the fuel. A typical limestone to fuel ratio is approximately 1:50.
Figures 9a, 9b and 10 show how the addition of limestone effectively reduces
gaseous arsenic at the entrance to the catalyst. The free CaO in the limestone
reacts with arsenic, forming a solid, Ca3(AsO4)2, which will not poison the
catalyst. The impact of adding limestone will be addressed further in the wet
bottom boiler case study.
Arsenic Compound
O Gaseous State
• Solid State
Figure 9a
Arsenic Compounds in Flue Gas
Page 9
-------
Coal +
Limestone
V
^f
<•>:
o ••
A<203 (G) »
3CaO * O2 — >
O3(A*04)2 (S)
O O
o o o
C«CO3— >
JCaCO » C02
0 0
0 0 o
A«2O3 (G) + F*O3
x V_
_ — • _
K • • •-AJ203(G)»0:
•o*.';0.:.".^"
V
rT
•*• -1
•'liliO3(sf
••&<*
DeNOx
Reactor
•*o.:L
V:-'<
• • •
h!
i t
Arsenic Compo
O Gaseous State
• Solid State
• ••• A ••
L»*« P ;t ESP
"•?• H *«
k /V j^
^^xx
i i
Figure 9b
Effect of Limestone Injection
1000
1
2 100
o
i 10
c
Effect of
Limestone
Injection
3 10 30
Arsenic in Coal (nig/kg)
Figure 10
Relationship Between Arsenic in Coal and Gaseous Arsenic
Alkaline Earth Metal (Ca)
Alkaline earth metal poisoning primarily occurs when free CaO in the flyash
Page 10
-------
reacts with SO3 adsorbed on the catalyst surface forming CaSO4. The CaS04
may cause catalyst surface masking, preventing the reactants from diffusing into
the catalyst. Figures 11 and 12 illustrate this mechanism.
CaSO4
CaSO4
FRESH
CALCIUM DETERIORATION
Figure 11
Mechanism of Calcium Deterioration
Level of Co
Catalyst wall
Catalyst surface
Figure 12
Accumulation of Calcium Composition in Catalyst
In order to address masking, particularly in dry bottom boilers where the free
CaO content is nearly double that of a wet bottom boiler (see Figure 13),
characterization of the mechanism and accurate prediction of deactivation rate is
Page 11
-------
employed. Experience shows that the quantity of free CaO in the fly ash governs
the amount of deactivation. Extensive modeling of deactivation from full scale
and laboratory data has been developed providing high confidence in design to
counter this phenomenon.
100-
average
E-* Plants
I Free CaOQ Amorphous Phase Ca
• Free CaOHAmorphous Phase Ca
Wet Bottom Boiler Ash
Dry Bottom Boiler Ash
Figure 13
Free CaO Ratio in Fly Ash
Catalyst Plugging
Catalyst plugging has two primary causes: (a) ammonia salt deposition and (b)
flyash deposition. Proper selection of catalyst pitch and cell opening size
minimizes plugging. Good system design is also a primary prevention tool.
Ammonia salt formation/deposition will not occur provided the SCR inlet
temperature is maintained above the salt formation level. Typically temperature
maintenance is only a problem during part load operation and is elevated by
installing a water- or gas-side economizer bypass in the system. If a gas side
bypass is used, careful attention must be paid to provide adequate temperature
mixing upstream of the AIG. This avoids low temperature streamlines Which can
cause localized salt formation at the AIG and catalyst.
Flyash deposition is minimized through proper flow distribution across the
catalyst face. This is accomplished through engineering of SCR system flue
layout, Computational Fluid Dynamics (CFD) codes, and/or cold flow modeling.
Page 12
-------
The design must limit the number of areas where dust accumulation may occur,
i.e.,
• avoiding flat surfaces such as long leading edges to turning vanes;
• installing ash deflection devices on support beam flanges and sootblower
pipes;
• avoiding flue dead legs between dampers where ash accumulation may
occur in closed position and exhaust onto catalyst when opened, etc.
Erosion
Catalyst erosion is caused by the impingement of flyash on the catalyst face.
Catalyst erosion is a function of gas velocity, ash character, angle of
impingement, and catalyst properties. Cormetech has experience with flyash
loadings as high as 30 g/Nm3 and flue gas velocities up to 6.2 m/s.
Experience has proven that no significant erosion will occur with proper system
design, catalyst material durability, and catalyst edge hardening. Poor flow and
ash distribution at the inlet to the catalyst has been the source of most problems,
and has improved significantly over the entire SCR experience history. Careful
attention must be paid to utilize flow modeling (CFD and cold flow modeling) to
properly simulate flow profiles through the SCR system. A flow rectifier grid
should be installed to straighten flow into the catalyst bed. Since most catalyst
erosion occurs at the catalyst leading edge which is exposed to the direct
impingement of dust particles the catalyst may also be hardened at the entrance
region to provide further protection as shown in Figure 14.
FLY ASH
HARDENED EDGE
CATALYST
Figure 14
Honeycomb Catalyst With Hardened Edge
Although in most cases In-Duct SCR is not attractive due to extremely high
system draft loss and difficulty in control of gas flow and ammonia distribution,
some utility sites are considering the concept due to site plan limitations.
Page 13
-------
Therefore, we will pilot test our hardened edge catalyst at velocities > 12 m/s in
1995.
Catastrophic Failure of SCR Catalyst
Catastrophic failure is defined as sudden and permanent loss of catalytic
performance. Based on our experience, catastrophic failures are extremely rare.
The primary cause is associated with the ignition of ash buildup. The intense
heat of a fire can irreversibly damage any SCR catalyst. Cormetech's ceramic
honeycomb catalyst, unlike plate catalyst with stainless steel mesh substrates,
will not promote oxidation, therefore fires are less likely to spread and are more
easily contained.
Coal Catalyst Developments
Based on our understanding of reaction and deactivation mechanisms we are
improving our coal catalyst. The objective is to increase the deNOx reaction rate
without increasing the SO2 oxidation rate. One method is to modify the pore
structure of the catalyst wall in order to reduce diffusion resistance. The other is
a more novel method and requires further explanation.
Since the deNOx reaction is only effective at the surface, we have developed a
improved catalyst of the same composition as our conventional catalyst with the
exception that the vanadium is preferentially distributed to the surface of the
catalyst wall. In this manner, we can maximize the concentration of vanadium in
the effective fraction of the wall, maximizing deNOx activity without incurring
high SO2 oxidation. Figure 15 shows the performance of this improved catalyst
compared to that of conventional catalyst. Since the product is still made entirely
of catalytic material, resistance to poisons is maintained.
.0
"35
19
O
Surface V2O5 (wt%)
Figure 15
Comparison of Catalyst Performance
Page 14
-------
Development and Qualification of Improved Coal Catalysts
We are developing the improved coal catalysts described above according to the
schedule shown in Figure 16.
1993
Screening of Improved Coal Cataly
1994
1995
its
1996
Figure 16
Improved Coal Catalyst R&D Schedule
We have already completed laboratory performance confirmation and durability
tests. The performance of the improved catalyst is approximately 15% higher
than our conventional catalyst, without impact on SO2 oxidation.
We have tested durability to gaseous arsenic poisoning by an accelerated
method using the apparatus in Figure 17. The test results are shown in Figure
18. The improved catalyst maintains its performance advantage over
conventional catalyst at any arsenic poisoning level.
As2O3 POWDER
THERMOCOUPLE \ SCR CATALYST
EXHAUST
Figure 17
Schematic of Arsenic Testing Apparatus
Page 15
-------
>-
>
o
<
IMPROVED
CONVENTIONAL
Arsenic Load, ppm*hr
Figure 18
Arsenic Durability
Since the degree of deactivation due to surface masking by CaSO4 is dependent
only on the fraction of surface masked we expect that the rate of deactivation is
the same for both conventional and improved catalyst. Therefore, given the
same installed catalyst volume, the improved catalyst will maintain field
performance much longer due to the higher initial activity, as shown in Figure 19.
>
>
o
<
IMPROVED
! 4 6
TIME, years
Figure 19
CaO Durability
Page 16
-------
Before commercialization, the endurance of the catalyst will be confirmed by
operation in an actual plant. Initial endurance tests will start by charging catalyst
samples into commercial SCR plants in Japan. During 1995 catalyst samples
will be charged into a coal unit in the United States and three coal units in
Germany. The performance of the catalyst samples after aging in the
commercial plants will be compared to fresh performance in order to evaluate
endurance.
Impact of Improved Coal Catalysts on SCR Design for Coal Fired Boilers
Whether achieving an increase in performance by optimizing the pore structure
of the catalyst or by the method described above, the impact on SCR design is
significant. In order to illustrate the impact of improved catalysts on the SCR
design, two SCR case studies are described below comparing improved with
conventional catalyst.
Case Study 1: Dry Bottom Boiler
A 250 MW unit is selected as an example unit. Table C shows the design data
for this unit. Figure 20 shows the 10 year catalyst management plan. Since the
activity of the improved catalyst is higher by 15%, the catalyst volume of the
improved catalyst is decreased by approximately 15% from the conventional
catalyst volume.
Design Condition:
Fuel
Plant Output, MW
Rue Gas Row Rate, Nm3/Hr
NOx Inlet, ppmvd @ 3% O2
NOx Outlet, ppmvd @ 3 % O2
NOx Removal Efficiency, %
Ammonia Slip, ppmvd @ 3% O2
Specification:
Type Qf System
Type of Reactor
Type of Catalyst
Number of Reactors
Coal
250
784,000
154
31
80
5
SCR
Vertical Flow-Fixed Bed
Honeycomb
1
Table C
Dry Bottom Boiler
Page 17
-------
Catalyst Volume Ratio
1.0
Add 4th Layer
Improved
Conventional
Add 4th Layer
3456
Operating Years
10
Figure 20
Impact of Improved Catalyst
Dry Bottom Boiler
Case Study 2: Wet Bottom Boiler with Ash Recirculation
A 320 MW unit is selected as an example unit. Table D shows the design data
for this unit. Figure 21 shows the 10 year catalyst management plan. The total
catalyst volume of the improved catalyst for ten years of operation is two-thirds
that of the conventional catalyst. As previously mentioned, limestone injection
will extend the catalyst life. The impact on the catalyst management plan is
shown in Figure 22. With limestone injection, the need for additions and
replacements of catalyst are significantly delayed.
Design Condition:
Fuel
Plant Output, MW
Flue Gas Row Rate, Nm3/Hr
NOx Inlet, ppmvd @ 3% O2
NOx Outlet, ppmvd @ 3 % O2
NOx Removal Efficiency, %
Ammonia Slip, ppmvd @ 3% O2
Flyash Recirculation Rate, %
Specification:
Type of System
Type of Reactor
Type of Catalyst
Number of Reactors
Coal
320
1,125,000
1,652
530
68
5
100
SCR
Vertical Flow-Rxed Bed
Honeycomb
1
Table D
Wet Bottom Boiler
Page 18
-------
Catalyst Volume Ratio
4
3 -
Exchange 2nd Layer
Exchange 1 st Layer :
Improved
Conventional
Add 4th Layer
Add 3rd Layer :
Add 4th Layer
Add 3rd Layer
_j i
01 23456789 10
Operating Years
Figure 21
Impact of Improved Catalyst
Wet Bottom Boiler
Catalyst Volume Ratio
3
Ou
Add 4th Layer
Add 3rd Layer
Improved w/ Lime Inj.
Improved w/o Lime Inj.
Add 4th Layer
Add 3rd Layer
0123456789 10
Operating Years
Figure 22
Impact of Limestone Injection
Wet Bottom Boiler
Page 19
-------
Conclusions
Our experience shows that coal-fired SCRs are successful when proper attention
is paid to system and catalyst design.
Proper system design is crucial to maximizing catalyst performance potential and
achieving the lowest annualized cost.
By exploiting the experience and know-how gained on existing facilities it is
possible to further optimize catalyst design.
Page 20
-------
Fluid Dynamic Optimization of SCR Plants by Modeling
Demonstrated by the SCR Plant Logan Generating Plant
K. Hauenstein
W.Herr
R. Sigling
Siemens AG, Power Generation KWU
P.O. Box 3220
D-91050Erlangen
Germany
Abstract
The SCR process for the reduction of nitrogen oxides has proven itself in practical application in
many fossil-fired power stations.
The concept of integrating a high-dust SCR plant between the boiler and air preheater has often
been hampered by structural constraints. As a result of the limited space available, the plant's
ducting will have in most cases a high degree of complexity. In order to achieve efficient SCR
plant functioning, however, the flow conditions have to be analyzed in detail and to be
optimized. The goal is to obtain uniform flow distribution upstream of the catalysts and a
homogeneous distribution of the reducing agent ammonia, while incurring as small a pressure
drop as possible. To this end, fluid dynamic measures are implemented such as the installation of
guide vanes and static mixing systems.
This paper describes the procedure used to meet the requirements mentioned above. It enables
the changes and modifications to be developed efficiently in terms of time and cost. Experience
shows that the results obtained can indeed be transferredjo and confirmed hi a full-scale plant.
The fluid dynamic solutions derived by modeling and the adaptation of a static mixer to the
geometry of the flue gas ducting are demonstrated using the example of the SCR plant at the
Logan Generating Plant operated by US Generating Company. The results of the measurements
are presented to illustrate the mechanisms of the modifications.
-------
Introduction
The department responsible for heat transfer and fluid dynamics at Siemens/KWU has been
optimizing SCR plants for approx. ten years. The stringent requirements with regard to
homogeneity of the flue gas flow led to the development of special static mixers for the needs of
SCR plants.
The experimental procedure most widely used currently has the advantage over the numerical
method that it is faster and less expensive. The experimental procedure and the design of a
suitable static mixer is explained using the example of the Keystone coal-fired power plant, and
examples are shown of the measures taken and the results achieved.
General Requirements for Flue Gas Flow
It is evident that an optimized design of the SCR system requires proper distribution of flue gas
and ammonia at the catalytic reactor inlet. To this end, a scale model of the SCR system was
designed and constructed from the economizer outlet to the air heater inlet. The effects of
different measures such as guide vanes, mixers and the dust behavior was then studied under
conditions reflecting the most significant boiler load cases. The final SCR duct system, including
positioning of guide vanes, mixers and ammonia injection grid reflects the results of the flow
model test.
Variations over the reactor cross section area as they are considered for catalytic reactor design
and as they are state of the art in many SCR systems are (Figure 1):
• Temperature distribution
The catalyst activity is a function of operating temperature. In reactor areas in which the
temperature deviates from the design temperature, the degree of NOx conversion will deviate
from the design value. Ordinarily, the catalytic reactor design is based on an uneven
temperature distribution of +/-15 K deviation from the mean value.
• Inlet flow
The inlet flow devices should be designed to obtain the most even possible inlet flow in all
areas of the reactor. Only then is the residence time of the flue gases and hence of the NOx
conversion equal in all parts of the catalytic reactor. Catalytic reactor design ordinarily
accounts for an uneven velocity distribution of+/- 15% (deviation from the mean value) over
90% of the cross-sectional area and +/- 20% over the remaining 10% of the cross-sectional
area.
• NHs/NOx Concentration Ratio
The NHs/NOx concentration ratio required for a specific NOx conversion should not deviate
more than +/- 5 % from the theoretical value. Larger deviations can result in excessive
slip concentrations in parts of the SCR reactor.
-------
A further general criterion is that the flow in the flue gas ducting and in the reactor involve the
smallest possible pressure drop. To achieve this, model flow experiments are used to optimize
the location of guide vanes.
Guide vanes for flow direction and other reactor internals must be designed such that no ash
streams arise and that ash deposits are prevented in the downstream zone.
Static Mixer for Homogeneous Flue Gas Flow in SCR Plants
Fig. 2 shows a typical static mixing system with NH3 injection for an SCR plant with economizer
bypass. The flue gas mixer (premixer) heats the main gas flow by mixing with the bypass flow.
In addition, the flue gas mixer equalizes any uneven flue gas velocity distribution.
The downstream NHs mixer (main mixer) homogeneously mixes the NHb which is injected
through lances into the flue gas flow.
The mixers extract energy from the flue gas flow and convert this energy into secondary flow and
turbulence. Optimum adaptation of the mixer to duct geometry of the SCR-plant results in the
extraction only of as much energy as is absolutely necessary, thus minimizing the pressure drop.
Special Conditions for Flow Direction of Flue Gas in the SCR Plant
The Logan coal-fired cogeneration facility, formerly referred to as the Keystone (Regeneration
facility, was constructed in Logan township, New Jersey by Bechtel for the U.S. Generating
Company. Both the steam generator and the SCR unit were supplied by Foster Wheeler Energy
Corporation.
As is shown in Figure 3, the SCR unit is placed between the economizer and the air heater. This
location is chosen because it gives the optimum gas temperature range for commercially
available catalysts. The flue gas flows vertically downward through the catalyst layers.
Significant characteristics of the flue gas flow in the Logan plant are:
• Economizer bypass duct
During part load operation (e.g. 30%), the temperature of the flue gas leaving the economizer
is low and consequently there is a potential for formation of ammonium sulfate/bisulfate in
the catalyst which would have harmful effects on the catalyst. Therefore, an economizer
bypass duct is installed to maintain the flue gas temperature above approximately 320 °C (610
-------
Short mixing paths from NH3 injection point to catalytic reactors
The distance from the ammonia injection grid (AIG) to the first catalyst level is 50 feet. This
corresponds to roughly 1.5 times the hydraulic diameter and is considerably less than in most
of the SCR plants constructed to date.
180° flow deflection from flue gas duct to reactor
The flue gas flow is redirected by 180° just upstream of the catalytic reactors, with a
simultaneous increase in the flow cross-sectional area from that of the duct to that of the
reactor.
Fundamentals of Model Flow Experiments
Flow Model
A 1:20 scale Plexiglas model was built to investigate flow direction, uneven distributions and
other flow aspects. The following components of the SCR plant in the Logan power plant were
modeled:
• Boiler economizer area
• Bypass ducts
• Flue gas duct from the boiler to the SCR reactor with flue gas mixer and NHs injection system
• SCR reactor
• Flow straighteners in the SCR reactor
The pressure drop across the catalytic reactors was simulated using sheets of perforated metal.
Over the course of the investigations, various internal components such as guide vanes and
mixers were tested in this model and their arrangement as well as size was optimized.
The model experiments were performed with air as the flow medium. Significant criteria which
are to be considered and which therefore can be transferred from the model experiments to the
full scale plant, are described in the following.
Similarity Theory
In the case of an isothermal, single-phase, steady-state flow, similarity theory requires the
following to ensure that the model results can be transferred:
• Adherence to the geometrical similarity of the model, i.e. the model must be reduced to a
suitable scale and must properly represent the geometry of all parts of the flow space.
-------
• Adherence to the Reynolds number Re:
Re = - -
v
The following difficulties arise when these criteria are imposed:
The Reynolds number of the model cannot correspond to the Reynolds number of the full-scale
plant, as dn in the model is reduced by the model scale of 20, whereas v is only reduced by a
factor of 4. To compensate for this, the velocity would have to be increased, but this in turn
would endanger the similarity due to compressibility influences. For industrial investigations, the
strict adherence to constajit Re can be relaxed, as the flow conditions in the present highly
turbulent flow (Re > 2* 10 ) are hardly still dependent on the Reynolds number.
The adherence to geometrical similarity (1:20 scale) in general leads to technical difficulties in
manufacturing the internals. For this reason, the model internals are replaced by geometrically
simpler components, which however are similar to the real internals in relation to their fluid
dynamic effect. The important effect of the internals is that they cause a change in the flow
distributed over the duct cross section due to their pressure drop. The components used in the
model must also cause the same effect. This is achieved if the flow resistance of the component
in the original system and in the model are the same, i.e. the friction coefficient (£) in the model
and original must be equal. In most applications, perforated metal sheets are used to simulate the
friction coefficient.
For components which have a clear spatial structure and which offer different resistances in the
three orthogonal directions, special constructions must be used with which the spatial flow
processes can be simulated.
Similar considerations hold for the transfer of the tracer experiments used to verify the
propagation of an NOx or temperature striation from the boiler until upstream of the first catalyst
level or the distribution of the injected NH3. The determining physical processes are here on the
one hand molecular mixing and on the other turbulent mixing of the gas striations in the
surrounding gas flow. For the gases considered in the full-scale plant, as well as in the model,
turbulent mixing is the determining factor, and not molecular diffusion. Mixing is thus purely a
flow phenomenon.
Individual Experiments and Results
Temperature Distribution across the Reactor Cross Section
These experiments are intended to show whether nonhomogeneous temperature distributions
arise upstream of the first catalyst level following the combination of the bypass flows with the
main flow. Figure 4 shows the experimental equipment and instrumentation. A fine mesh
-------
injection grating was used to input the tracer gas homogeneously in the two bypass flows. The
tracer gas was a 4% vol. propane/air mixture which can be detected using a flame ionization
detector. The concentrations determined in this way were recalculated to temperatures, since, as
was already explained, mixing of temperature and concentration are caused exclusively by
turbulence in the highly turbulent model flow.
Following admixture of the bypass streams, the maximum temperature deviations upstream of
the first catalyst level were +50°F/-70°F without the use of the flue gas mixer, and +25°F/-20°F
with flue gas mixer (see Figures 5 and 6).
Velocity Distribution Upstream of First Catalyst Level
A hot wire anemometer was used for the velocity measurements. A thread probe was used to
detect crossflow and backflow. As the Reynolds number was in the highly turbulent range in the
model, the velocity measurements can be directly transferred to the original.
The velocity profile upstream of the first catalyst level was influenced not only by the
arrangement of the guide vanes, as is shown in Figure 7, but also by the flue gas mixer, Figure 8.
In addition to the guide vanes, the mixing system also smoothed the uneven velocity distribution,
which arises from the two flow redirections downstream of the boiler outlet. The flue gas mixer
thus also takes on the function of the guide vanes, as was already be seen from the effect on the
temperature distribution.
Simulation of Concentration Distribution of Injected Ammonia
One of the prerequisites for these experiments was scale modeling of the NH^ injection system.
For the measurements, tracer gas was evenly injected from the nozzles of the injection system
into the main flow. To enable direct transfer of the results of the NHa concentration
measurements to the original system, injection in the model was performed with the same
momentum ratio as in the full-scale system. The momentum ratio was selected as the ratio of the
momentum of the nozzle exit flow to that of the main flow.
Injection in the model was performed such that approximately the same volume flow rate of
tracer gas was injected by all nozzles.
The concentration distribution upstream of the first catalyst level was determined using a flame
ionization detector. The result is shown in Figure 9. Based on the deviation of +8%/-16% in the
concentration distribution determined in the experiment it can be concluded that suitable
regulation of ammonia injection in the full-scale system will present no problems in achieving
the required allowable deviations of +/- 5%.
Following this, an experiment was performed using "selective injection", i.e. injection through
only one nozzle. This should verify that the NH3 concentration can be controlled in selected parts
of the reactor by trimming the individual NH3 nozzles. As is shown in Figure 10, this is possible,
despite the presence of the flue gas mixer upstream of the NH3 injection.
-------
Minimizing Pressure Drop
All of the individual steps for optimization of the arrangement of the guide vanes were also
evaluated with regard to pressure drop.
The pressure drop measurements confirm that an even flow also results in the minimum pressure
drop. The arrangement of the guide vanes in the 180° redirection results in a reduced pressure
drop over that for redirection without fluid dynamic internals due to the evenly distributed
redirection of momentum.
The pressure drop due to the flue gas mixer is roughly 1.3 mbar at full load.
Minimizing the Possibility of Fly Ash Deposits
Fly ash deposits can be minimized by optimization of the flue gas ducting and by the installation
of corresponding guide elements. Therefore, even hi the velocity measurements in the model,
emphasis was placed on the fluid dynamic internals generating no ash deposits which could
impair operation. This was achieved by such measures as avoiding long horizontal sections in
designing the guide vanes. The guide vanes furthest out in the reactor enclosure have a 60°
redirection angle instead of a 90° angle. This shortens the horizontal section of the upstream
edge, thus preventing ash deposits on the guide vanes.
Two processes were investigated in the fly ash experiments. First, the location and amount of
deposits were determined, and then the behavior of ash deposits at increased flow velocity was
determined.
Fly ash from the power plant was used in order to enable simulation of the deposition
mechanism. The experiments were performed using an accelerated procedure hi which the ash
concentration was three tunes higher than that in the full-scale system. This enabled drawing of
conclusions after relatively short experiment times, concerning where fly ash sedimentation can
be anticipated at lower load conditions and whether a limiting condition is attained. The
experiments also yielded indications as to whether deposits which would impair operation could
be anticipated in the full-scale system.
In the part-load case, all significant plant components remained nearly free of deposits, even after
a six-hour experiment with an ash-laden medium. The ash deposits on the flow straightener were
minimal, and the attached deposits were oriented vertically, which indicates vertical
impingement. Increasing the flow velocity did not result in abrupt removal of the ash deposits,
which were gradually eroded away. This provides additional confirmation of the optimum design
of the guide vane configuration as well as of the duct geometry.
-------
Summary
The results obtained in the model flow experiments were implemented in the full-scale system.
These are as follows:
Static flue gas mixer,
Location of the NHs injection,
Locations of the guide vanes and their geometry.
Without model experiments, the need for a flue gas mixer would not necessarily be recognized
immediately. The model experiments made it clear that problems would have arisen in the plant
without this component. The mixer developed by Siemens/KWU is simple and effective, as has
since been verified in other implementations, including Mercer Station.
In addition to the flue gas mixer designed for this specific application, there is also a mixing
system for the injected NH3. However, this system was not required in the case of the Logan
plant.
The model experiments also served to verify that the planned NHs injection design fulfills the
requirements for NHs distribution and control.
Adherence to the specified pressure drop for the overall plant was both verified and made
possible by the model experiments.
Overall, it can be stated that the performance of model experiments is recommended, as the
results are available before construction of the full-scale system. These results can then be used
to make any necessary component modifications without the necessity of first suffering from
negative operating experience. The expenditure for model experiments is thus slight in
comparison with possible later modification work in the full-scale system.
-------
SIEMENS
0
x%
i i
I I
I I
I i
Additional volume A V
0
il=85%
ti=90%
20 (%) (°K) 30
Deviation x
Figure 1: Effects of an Unequality on the Catalyst Volume
-------
SIEMENS
t
Flue Gas
Flow Direction
Figure 2: Siemens Static Mixing System Parmix and Turbomix for improvement of homgeneous Flue Gas
<->:<>+.-;i-tiri-i«-kn fanri Ammonia Infection
-------
SIEMENS
Flue Gas
Flow
Boiler
Economizer
Bypass Duct
Guide Vanes
Reactor width
— 25'-6" —
44'-8';
Static Mixing
System
Dimmy_Layer
~i —i —i —r T T
-Catalyst Layer 1]
Catalyst Layer 2-
Reactor
depth
Sublayer 1
-Sublayer 2
Sublayer 3
' Sublayer 4
(spare layer)
Figure 3: Arrangement of Boiler - Duct - Bypass - Static Mixer - AIG
Guide Vanes and SCR Reactor with Dummy Layer and
Catalyst Layers
-------
SIEMENS
Bypass Flow
Simulation
with Tracer
Gas
Simulation of NH3 Injection
I with Tracer Gas
Air
Fly-Ash
Precipitator
Air Fan
Prandtl-Probe
Dust
Hot Wire
Velocimeter
and
lonization Flame
Detector
Delta-p Sensor
Figure 4: Schematic of Test Model with Instrumentation
Equipment
-------
Flow
Direction
Dimensionless
Temperature
Reactor Depth:
13.411 m
2345678
Reactor Width: 7.737m
Figure 5: Temperature Distribution upstream of First Catalyst Layer without Internals
such as Guide Vanes or Static Mixing System (RMS = 17.7%)
-------
SIEMENS
Flow
Direction
Dimensionless
Temperature
Reactor
13.411 m
Reactor Width: 7.737m
Figure 6: Temperature Distribution upstream of First Catalyst Layer after Installation of
Guide Vanes and Static Mixing System (RMS = 1.6%)
-------
Flow
Direction
Dimensionless
Velocity
Reactor Depth:
13.411 m
B
1234567
Reactor Width: 7.737m
Figure 7: Velocity Distribution upstream of First Catalyst Layer without Internals
such as Guide Vanes or Static Mixer (RMS = 65.3%)
-------
SIEMENS
Flow
Direction
Dimensionless
Velocity
Reactor Depth:
13.411 m
Reactor Width: 7.737m
Figure 8: Velocity Distribution upstream of First Catalyst Layer after Installation of
Guide Vanes and Mixing System (RMS = 8.8%)
-------
SIEMENS
Flow
Direction
Dimensionless
Concentration
D
Reactor Depth:
13.411 m
B
Reactor Width: 7.737m
Figure 9: Concentration Distribution upstream of First Catalyst Layer after Installation of
Guide Vanes and Static Mixing System (RMS = 7.6%)
-------
SIEMENS
Dimensionless
Concentration
-3
-2
-1
Local Injection of Tracer Gas
Duct Depth / Reactor Depth
Flow
Direction
Duct
Reactor
Duct Width /
Reactor Width
Figure 10: Simulation of the Effect of Local Ammonia Injection
-------
Session 7A
Fundamentals and Modeling
-------
PARTICLE SEGREGATION BY SIZE
TOWARDS LOW NOx BURNER DESIGN.
J. Charoensuk, S. M. Godoy and F. C. Lockwood.
Mechanical Engineering Department
Imperial College of Science Technology and Medicine
Exhibition Rd
London SW7 2BX
Abstract
Among the combustion modifications developed in order to reduce the levels of NOx
emissions in Pulverised Coal combustion for power generation, the low NOX burner is
one of the most developed and popular. However, it is still considered that the
maximum potential for NOx reduction via burner systems has not yet been fully
realised. The continuation of the development of low NOx burners continues to be
recommended by the power generation industry.
The flow conditions required for low fuel-NOx formation during the combustion of
pulverised coal particles, must allow for maximum volatile yield in oxygen depleted
atmospheres. This decrease in Oi availability (rich stoichiometry) affects the rates of
combustion and ignition, leading to an increase of carbon in the fly ash.
Previous studies show that small particles release their volatile more rapidly than larger
ones. Particle segregation by size, followed by the placement of particles of differing
sizes in the appropriate regions of the inner recirculation zone of a power station
pulverised coal burner would lead to a more balanced compromise between low NOX
and unburned carbon content in the fly ash.
The present paper explores the effect on NO performance of appropriately
manipulating the distributions of particle sizes and velocities in the burner internal
recirculation zone. It is concluded that such manipulation has the potential to achieve
a significant reduction in NO emissions.
Introduction
Within the context of the environmental constrains imposed on coal utilisation, the
need to reduce NOX emissions from existing pulverised coal fired furnaces in the
power generation industry can be satisfied by number of options. The commercially
available options and their comparative costs are summarised in Table 1.
The current German limit on NOx emissions for coal fired power stations is 120 ppm.
As a comparison, boilers in the UK with low-NOx burners installed operate at NOx
-------
level generally below 350 to 400 ppm, depending on whether they are corner fired or
wall fired. Of the possible choices shown in Table 1, only selective catalytic reduction
(SCR) could be considered to reach the required performance. However this
technique is expensive and it has not been fully proven on flue gases from high sulphur
or high chlorine coals2. Nonetheless, the other and cheaper options can not match
individually the effectiveness of SCR.
Of these options, it is considered that the maximum potential for NOx reduction via
burner systems has not yet been realised. In addition, the capital costs of installing this
type of equipment is generally lower than other NOx reduction measures, hence, the
continuation of the development of low NOx burner techniques is recommended by
the power generation industry2-
Volatile NOx formation can be minimised by promoting devolatilisation in zones of
high temperature and low stoichiometry. Low NOx burners are generally designed to
achieve delayed combustion through the way air and fuel are introduced. The level of
available oxygen is decreased in zones that are critical for NOx formation and the
amount of fuel burned at peak temperature is also decreased. By fuel and air staging,
the coal is devolatilised under conditions of rich stoichiometry promoting the
conversion of fuel nitrogen to molecular nitrogen rather than oxidation to NO. As a
consequence, the reduction in NOx achieved by low NOx burners is typically
accompanied with a small reduction in combustion efficiency reflected on higher
residual carbon content, due to the less intense nature of the combustion.
A number of fundamental studies on nitrogen release and particle devolatilisation rate,
have reached to the conclusion that the particle size plays an important role in the
mechanisms of devolatilisation3'4'5. Additionally, the amount of NO formed from the
nitrogen released as volatile is also a function of O2 availability, particles residence
time and trajectories, and other parameters that can be addressed as aerodynamic
effects6.
The present paper studies the possibility to further reduce the NOx emissions in low
NOx burners that are already in commercial use by segregating the pulverised fuel
particles by size and by optimising the use of the aerodynamic conditions in the near
burner region.
Background
The flow conditions required for low fuel-NOx formation during the combustion of
pulverised coal particles must allow for maximum volatile yield under oxygen depleted
conditions. The complete nitrogen chemistry involves well over 100 elementary
reactions. The main pathways of gas phase N-chemistry related to combustion may be
supposed to be those illustrated by Figure 1, taken from De Soete, 19937.
One of the objectives of low NOx techniques is to promote the reduction via
hydrocarbons given the appropriate C>2 concentration conditions so that the following
reaction path is emphasized.
-------
NO + CHi -> HCN + Products (1)
HCN— 0->NCO— ^->NHi N° >N2 (2)
This process emphasises the role played by hydrocarbon radicals (CHi), the existence
of which depends on a number of factors such as devolatilisation mechanism, identity
of volatiles, oxygen availability and the aerodynamic effects.
Based on previous studies3'8 wich showed that small particles release their volatile
more rapidly than larger ones, and that lower NOx emissions are produced during the
combustion of pulverised coal of small size grind, when the particles can partially
penetrate the inner recirculation zone (IRZ), it was concluded that particle segregation
by size, followed by placement of different size particles in optimal regions of the IRZ
would lead to a more balanced compromise between low NOX and unburned carbon
content in the fly ash.
as simulated In calculations performed by Romo-Millares9, where two burner
geometries were compared as shown in Figure 2, it was shown that because of their
greater inertia, larger particles penetrate the IRZ further than the smaller particles.
Smaller particles are flung rapidly outwards, by the effect of the secondary air swirl,
towards the zones of greater oxygen availability. These are hot, lean stoichiometric
conditions that will promote the formation of NO. Because staged air burners reduce
the availability of 62 in the inner zone, we want the smaller particles, with their greater
reactivity, to concentrate there, near the centre-line where temperature should remain
at a level that will enhance devolatilisation. The larger ones, with their lower
devolatilisation rates, should be caused to populate the regions where temperatures are
higher. Thus, the NO formed from the volatiles released by the larger particles in the
outer regions of the IRZ will be reduced to N2 when it passed into the centre of the
recirculation zone due to the presence of CHi generated by the devolatilisation under
oxygen depleted conditions of the smaller particles.
In studies performed by Abbas et al.8 comparing the performance of two burner types,
it was found that for the case of a single central orifice burner (SCO), with a flow
pattern that gives longer residence times to the larger particles and allows the small
ones to penetrate the IRZ, also increasing their residence times in this lower
temperature region, the conditions for low NOX are inherent. However, for a single
annular orifice burner (SAO), which represents the geometry of many industrial
burners, this was not the case. The small particles do not penetrate the recirculation
bubble, rather they are reversed towards the outer regions of the recirculation zone, as
shown in Figure 2, while the larger ones spend shorter residence times in this region.
With this geometry, if the smaller particles could be directed into the inner region, the
stoichiometry will be richer, contributing to a decrease in NOX formation.
In order to cause the smaller particles to penetrate the recirculation zone further,
greater initial velocities should be provided. Because we want to maintain the rich
conditions, the amount of primary air should not be increased, the increase in velocity
should be achieved either by reduction of the cross-sectional area, or by the increase in
flow rate by the addition of an inert gas (e.g. EGR). Further penetration of the
particles could also be achieved by the slight penetration of the burner pipe into the
-------
quarl region. This has been achieved by Visser et al10 where the penetration of the
burner gun into the quarl by 1/3 of its depth results in a flow field where the primary
flow penetrates through the recirculation bubble. This is not a desirable situation as
far as low NOx production is concerned, as the residence time of the particles in this
zone is reduced. The longer the residence time in the IRZ, the more efficient
devolatilisation in the 62 depleted zone and the greater the burnout.
The introduction of a flow through the pilot gun in the SAO type burner, should have
a similar effect, i.e. a forward flow that will weaken the reverse flow (coupled with a
desired reduction of temperature) generated by the recirculation of the secondary air.
This effect should ideally be confined to the near burner region inside, or just outside
the quarl, avoiding full penetration of the recirculation bubble (to minimize the carbon
content in the fly ash). This option, with full penetration of the recirculation bubble,
has been studied by Smart and Morgan11 while testing the use of different re-burning
fuels fed through the inner pilot tube of the burner.
Burner Design
Based on the research work performed by IFRF10'11, proven low NOX burner
industrial practices12 and burner manufacturer recommendations13, and the experience
gained in our laboratories on NOx emission studies14 together with the preceding
considerations, it was concluded that, as a first step for the introduction of an
innovation into the existing low NOx burner designs, the possibility for a selective
location of particles into the IRZ should be pursued.
As a starting point the basic design of typical present day low NOx commercial
burners is adopted. These burners make use of fuel staging measures that separate the
coal flow inside the primary tube in 4 jet streams. The ability to seaparate the flow in
such a way, with a low pressure drop and minor erosion, is determined by the
pulverised coal being introduced tangentially into the primary pipe, giving a toroidal
trajectory to the coal particles within the tube wich is subsequentially eliminated by 4
pulverised coal collectors, named FFRs, located inmediately before the burner mouth.
However, the swirling movement, which in the absence of the FFRs would be
imparted to the particles, should inherently classify them by size due to their
centrifugal force, with the larger particles populating the outer regions of the primary
tube.
As the first step in the present design procedure, a geometrical burner modification has
been introduced so that the smaller particles will be encouraged to penetrate the IRZ,
without passing through the recirculating bubble; i.e. to follow a similar trajectories to
those of the larger particles shown in Figure 2.
The following design of a low NOX burner has been undergoing isothermal testing and
computational simulation of its performance under firing conditions. A flow separator
located in the primary line, as shown schematically in Figure 3, leads to a selection of
the smaller particles and an increase of their initial velocity. With the smaller particles
located in the annulus of smaller radius at the burner mouth and with an increased
-------
initial momentum, the aim of further penetration of these particles into the C>2 depleted
region of the IRZ is achieved.
Scaling Criteria
Due to inherent difficulties in conducting full-scale tests and the associated costs,
experiments are often carried out on rigs of significant by smaller scale. The
immediate question is of course: how representative of the full industrial scale burner
is the experimental burner performance, measured at the reduced scale?. In an ideal
scaling exercise, the strict similarity of all the fluid dynamic and thermo-chemical
related processes should be maintained. In practice, compromises are necessary since
of the various physical and chemical processes scale in different ways.
In order to adequately simulate the two phase interaction between the coal particles
and the aerodynamic flow field, the mean particle size has to be reduced by the square
root of the linear scale factor. However, experimentally, this requirement is normally
not practicable, due to the high cost and inconvenience involved in grinding the coal to
the required scaled size and, in particular, since the mean size is altered and
consistency of particle properties across the size distribution cannot be maintained.
In this study, a 1/6 scale isothermal model has been built, making use of the constant
velocity scaling criterion. This maintains the full scale burners velovities at the
reduced scale and ensures that the minimum velocity required for the transport of the
pulverised coal will be reached. To simulate the cyclone separation effect given by the
geometry of the primary tube at the inlet, the main scaling criteria used were the air
inlet velocity into the cyclone and the degree of swirl. In this way, it may be aargued
that the particle separation is be properly simulated.
Results and Discussion
Making use of the primary tube design explained above, preliminary measurements of
particle size and velocities at the exit of the tube have been performed in a reduced
scale isothermal rig. The purpose of these measurements was to assess the
effectiveness of the particle separation by size of the cyclone geometry of the primary
tube coupled with the flow separator -1- of Figure 3.
Polydisperse coal of a size typical of industrial pulverised fuel was introduced into the
primary tube. The experimental conditions are given in Table 2.
Figure 4 shows the particle size distributions of the feed coal and of the samples taken
at the inner and outer tube outlets (i.e. numbers -4- and -5- in Figure 3). As expected
from the design considerations, the coal feed has been segregated by the flow
separator into coal of 75% under 20 (im size in the inner outlet and 75% under 90 (im
in the outer outlet.
-------
The air axial velocities at the exit of the two outlet annuli and the contour velocity
profiles downstream were measured making use of a pilot tube 3mm in diameter. The
data is shown in the shape of a velocity contour in Figure 5.
With the data on initial conditions thus obtained, computer simulations for their effect
on NO emissions were performed. A two dimensional numerical code, 2-D CINAR,
developed by this combustion group for simulation of the flow aerodynamics,
combustion and heat transfer of two-phase flows was employed. The code is well
documented having been used and validated against a variety of experimental data15'16.
For the NO calculations, a simplified chemistry model is employed, which at this
moment does not take into consideration theNO reduction mechanism via CHi.
A comparison was performed between two cases: the "test" case, wich uses the burner
geometry of Figure 3 and the measured boundary conditions of the reduced scale
isothermal experiments, and a hypothetical case, the "base" case, where no particle
segregation occurs and the primary stream has a uniform initial velocity.
In the "test" case, small particles, of a typical 20 (im size, are injected at the inner
outlet position with an initial velocity of 40 m/s. The larger particles, of a typical size
of 70 Jim, are injected at the outer outlet position at 27.0 m/s initial velocity. For the
"base" case, both particle size groups are injected at an initial velocity of 32.0 m/s.
Figure 6 depicts the downstream predictions of the particles trajectories as a function
of initial velocities and size. In the "base" case, the small particles populate the regions
closer to the quarl walls where they meet the secondary air flow at an early stage,
while in the "test" case, these particles populate the central region of the flow, up to
the exit of the quarl and beyond delaying their mixing with the secondary flow.
In Figure 7 the predicted NO profiles are shown. It can be observed that lower levels
of NO are found for the "test" case where the smaller particles have been injected at
greater velocities closer to the axis of the burner. These preliminary results are
indicative of the effectiveness of the burner modifications proposed, towards the
reduction of NO emissions. It must be emphasised here, that these predictions do not
represent a commercial design, they are studies done with the sole purpose of
observing trends, that the modifications sugested in the present paper, will provide. In
addition, as the models employed in the present study do not simulate the reduction
reactions via CHi, the levels of NO found in a practical applications should be much
lower.
Further work is being undertaken on both: experiments, where laser techniques will be
employed to study particles distribution, and computations, modifying the conditions
to simulate better the practical case.
Conclusions
1. Existing burner designs do not match the particle size and velocity distributions in
the inner recirculation zone in a manner conducing to optimizing NO reduction.
2. Improved distributions can be achieved through a simple modification of the
upstream burner aerodynamics.
-------
j. JLUC experimental ana computational work of this paper for one such burner
modification demonstrate that NO emission reduction is indeed procured.
4. Further research should be undertaken to determine an optimised burner design
based on the innovative concept of this paper.
References
1. Hesselman, G. J. and Irons, R. (1992) "NOx reduction in coal combustion- A
general review of the current technology". ETSU, DTI Report No.ROO?.
2. Jones, A. R. (1993) "NOx abatement in the UK -A power generator's perspective".
In Coal Research Forum, NOx: Basic mechanisms of formation and destruction and
their application to emissions control technologies, 20-22 April, London.
3. Costa M.; Godoy, S.; Lockwood, F. C. and Zhou, J. (1994). " Initial stage of
devolatilisation of pulverised coal in a turbulent jet" Combustion and Flame 96.
150-160.
4. Niksa, S., Russel, W. B., and Saville, D. A. (1982) "Time resolved weight loss
kinetics for the rapid devolatilisation of a bituminous coal". Nineteenth Symposium
(International) on Combustion, p. 1151.
5. McLean, W. J., Hardesty, D. R., and Pohl, J. H. (1981)" Direct observation of
devolatilizing pulverized coal particles in a combustion environment". Eighteenth
Symposium (International) on Combustion, p. 1239.
6. Abbas, T.; Costa, M.; Costen, P.; Godoy, S.; Lockwood, F. C.; Ou, J. J.; Romo-
Millares, C. A. and Zhou, J. (1994) "NOx formation and reduction mechanisms in
pulverised coal flames". Fuel, 73, 9, 1423-1435.
7. De Soete, G. G. (1993) "The role of NO and N2O formation/destruction chemistry
in coal combustion control techniques". In Coal Research Forum, NOx: Basic
mechanisms of formation and destruction and their application to emissions control
technologies, 20-22 April, London.
8. Abbas, T.; Costen, P.; Lockwood, F. C. and Romo-Millares, C. A. (1993) "The
effect of particle size on NO formation in a large scale pulverised coal-fired
laboratory furnace: Measurements and modelling". Combustion and Flame 93,
316-326.
9. Romo-Millares, C. (1993). "Mathematical Modelling of Fuel NO Emissions from
PF burners". PhD thesis, Imperial College of Science Technology and Medicine,
University of London.
10. Visser, B. M.; Smart, J. P.; Van de Kamp, W. L. and Weber, R. (1990)
"Measurements and predictions of quarl zone properties of swirling pulverised coal
flames". 23rd Symposium (Int) on Combustion, The Combustion Institute.
11. Smart, J. P. and Morgan, D. J. (1993). " The effectiveness of multi- fuel
reburning in an internally fuel staged burner". The Coal Research Forum, 20-22nd
of April at Imperial College, London.
12. Welbourne, M. C. (1993) "The modelling of wall fired pulverised coal burners for
emissions control". International Combustion Limited, Report No. 25654.
13. De Michele, G.; Benelli, G.; Ligasacchi, S.; Benanti, A.; Tarli, R.; Alberti, M. and
De Santis, R. (1993) "Development and industrial application of an oil and gas low
NOx burner". Report by ENEL and Ansaldo Componenti, Italy.
-------
14.- Abbas, T.; Costen, P. and Lockwood, F. C. (1991) "The influence of the near
burner region aerodynamics on the formation and emission of NOx in a pulverised
coal-fired furnace". Combustion and Flame, 9_1, 346-363.
15.. Lockwood, F.C., Rizvi, S. M. A., Lee, O.K. and Whaley, H. (1984) "Coal
combustion model validation using cylindrical furnace data". 20th International
Combustion Symposium, .513-522.
16. Lockwood, F.C. and Romo-Millares, C. A. (1992)" Mathematical Modelling of
fuel-NO emissions from PF burners" Journal of the Institute of Energy, September
1992, Vol 65 pp. 144-152.
-------
CHi
HCN
o
NCO
O OH
NO
N20
H
N
.NH
NH2
NH3
H OH M
CHi
1
NO
N
FIGURE 1 Main ways of gas phase N-Chemistry, reproduced from De Soete, 19937
0.3
?
O f|l 1 1 T I I t i I T I" I
0.1 0,2 0.3 0.4 0.5 0.6 0.7 0.8 0.9
n n I—i—i—i—t—1—4— i l i i f=F^=3^T } i i i I i ixi i I i i i i I i i i i I i i i i I i i i i I
0.0 0.1 0(2 0.3 0.4 0.5 0.6 0.7 0.8 0.9
Axial position (m)
FIGURE 2 Calculated9 particle trajectories for two burner geometries:
(a) SAO and (b) SCO burners.
-------
ERROR: timeout
OFFENDING COMMAND: timeout
STACK:
1 Flow separator
2 Igratertube
3 Primary tube
4 Inner outlet
5 Onto-outlet
FIGURE 3 Schematic of primary tube modification to ICL burner.
Note: The details of the flow separator have been suppressed
pending patent.
1000
100^
I
£ 10:
Inner outlet
Outer outlet
Polydisperse coal
0 20 40 60 80 100 120
size ( m)
FIGURE 4 Particle size distributions for coal feed and samples from the primary tube
outlet.
10
-------
UNITS: m/s
LU
CD
CD
I |
r ]
czu
cu
[ J
cp
Erlil
ABOVE
20.00-
15.00
10.00
5.00-
250
2.50
5.00 -
10.00-
15.00
uei ow
25.00
?5 00 I
2000
1 5 00
I o.oo 50 -
500
2.50
-2.50p5 ."
5.00
10.00 ~-
1500 o :-
25
50
75
100
125
Figure 5 AXIAL VELOCITY CONTOURS OF LOW NOx BURNER
-------
a. Base case
" 10 0.02S 0 05 0 075 0 1 0.125 0.15 0 175 0.2
b. Test case
39.58nVs
10 0 02& 0 05 O075 0 ] 0 125 015 O 175 0-2
FIGURE 6 Particle trajectories for (a)"base" case and (b)"test" case
-------
a. Base case
I | ABOVE 690.00
I I 613.33-69000
[ig-ifgj 536.67-613.33
^B 460.00 - 536.67
BB 383 33 - 460.00
|H 306.67-383.33
^B 230.00 - 306.67
153.33-230.00
76.67-153.33
0.00- 76.67
BELOW 0.00
b. Test case
I | ABOVE 690.00
| | 613.33-690.00
536.67-613.33
460.00 - 536.67
383.33 - 460.00
306.67 - 383 33
230.00 - 306 67
153.33-230.00
76.67-153.33
000- 76.67
BELOW 0.00
*.0 o.os 0.1
0.15 0.2 0.25 0.3
FIGURE 7 NO contours for (a)"base" case and (b)"test" case.
13
-------
TABLE 1
Cost associated with various NOX reduction techniques1
Technology
Combustion modification
1) Low NOX burners
2) Furnace air staging
3) Combination of low NOx burners
and air staging
4) Re-burn
Flue gas treatment
5) Selective non-catalytic reduction
6) Selective catalytic reduction
Additional
Installation
2
2.4
3.8
6-14
2.2-10.6
14-18
Cost£/kWt
Operation/year
~0
~0
-0
1.0
1.5-2
3 and over
NOX Reduction
%
30-50
up to 50
up to 70
up to 60
40-50
80-90
Table 2
Experimental and computational run conditions
Experimental
Primary air flow rate, kg/s 0.023
Coal feed rate, kg/s 0.001
Axial primary air velocity, m/s, INNER OUTLET 40.0
OUTER OUTLET 25.0
Pulverised coal size, % under 75um 70.0
% under 25 ^m 36.0
Average size, (im 40.0
Cyclone Inlet air velocity, m/s 35
Degree of Swirl, ° 25
Computational
BASE CASE
Axial primary air and particle velocity, m/s 32.0
Polydisperse coal, 4 entry points, 4 size at each entry, urn (1) 15
(2) 33
(3) 50
(4) 105
S econdary air swirl number 1.0
TEST CASE
Axial air and particle velocity, m/s, INNER OUTLET 39.9
OUTER OUTLET 27.0
Particle size urn, INNER OUTLET 20.0
OUTER OUTLET 70.0
Secondary air swirl number 1.0
14
-------
NUMERICAL MODELING INVESTIGATION OF MANAGED NATURAL GAS CO-
FIRING IN LOW-NOX PULVERIZED COAL BURNERS
Hamid Sarv and Keith C. Kaufman
Babcock and Wilcox
Research and Development Division
Alliance, OH 44601
and
Tony Facchiano
Electric Power Research Institute
Environment Control Systems Business Unit
Palo Alto, CA 94303
Abstract
Combustion and emissions performance of natural gas co-firing in an internally-staged and
low-NOx pulverized coal burner was evaluated by numerical modeling. Numerous
permutations in gas velocity, injection angle, injection location, extent of co-firing, and
natural gas/air premixing were simulated for comparison. With 15% gas co-firing, NOX
emissions were 16% lower and the unburned carbon levels were 19% less than the
corresponding values at 100% pulverized coal firing. Since the NOX emission levels for the
baseline low-NOx pulverized coal burner are already about 50% lower than the uncontrolled
values of conventional burners, total NOX reduction with 15% gas use would be close to 60%.
* Author of correspondence
-------
Introduction
Many operating coal-fired utility boilers in the United Stated are in the "ozone non-
attainment" and "acid rain control" areas identified by the Titles I and IV of the 1990 Clean
Air Act amendments. While low-NOx burner retrofits are considered as "reasonable available
control technology" options for reducing the NOX emissions, they may adversely affect the
flame length and carbon conversion in some coal-burning units. Degradations in the
combustion and emissions performance are due to the fact that low-NOx burners operate
differently from rapid mixing conventional burners. In a low-NOx burner, NOX is reduced by
the controlled separation and delayed mixing of the fuel and oxidizer.
Fly ash can be used as a mineral admixture in Portland cement concrete, as long as it meets
the chemical requirement of less than 6% loss on ignition'. High levels of carbon in fly ash
make it unsuitable for recycling and could even pose disposal problems in landfills. Natural
gas co-firing of low-NOx burners offers greater flexibility in overcoming such operational
problems. A recent report on this subject2 reviews the potential benefits of co-firing,
including shorter flame length, higher combustion efficiency, higher turndown ratio, lower
excess air requirement, and lower SO2, CO2, NOX, and particulate emissions. Since natural
gas availability at most coal-fired plants is limited to ignitor use only (< 15% of the combined
thermal load), creative and economical ways must be devised for its effective utilization in
low-NOx pulverized coal burners to "leverage" the desired performance.
So far, co-firing has not been incorporated commercially in wall-fired pulverized coal burners,
and its application has been limited to natural gas injection through gas ignitors or selected
burners. Leveraged natural gas co-firing in low-NOx pulverized coal burners can be
implemented in a number of ways. Independent pilot-scale tests- at the International Flame
Research Foundation 3 (IFRF) and Babcock and Wilcox (B&W) indicate lower NOX emissions
with centerline introduction of natural gas in the coal nozzle. Gas co-firing at IFRF was
accomplished with an externally air-staged burner, while an internally-staged burner was used
at B&W. Part of the NOX reduction by co-firing is due to lower nitrogen content of the
combined fuel. Additionally, injection of natural gas influences the local flame environment
-------
and provides a source of hydrocarbon radicals (CH;; i = 1, 2, and 3) that convert NO to
molecular nitrogen.
This paper discusses a numerical modeling investigation of managed natural gas co-firing in
the internally-staged DRB-XCL® low-NOx pulverized coal (PG) burner 4. Details of the
modeling simulations and predictions of temperature, unburned carbon, and CO, O2, and NOX
emissions for various cases are discussed and compared in the following sections.
Numerical Model Description
Numerical simulations were done with the B&W multi-dimensional COMO (COmbustion
MOdel) code. COMO predicts the detailed distributions of velocity, turbulent kinetic energy
and dissipation rate, particle dispersion, species concentrations, heat fluxes, and temperature in
Cartesian or cylindrical geometries 5"8. It can simulate the effects of operating conditions and
burner and furnace configurations on the combustion and emissions performance. For this
purpose, the computational domain is divided into control volumes and the governing
differential equations are formulated and solved at each location.
Table 1 summarizes the capabilities of COMO. The model utilizes coal-specific chemical
kinetics for devolatilization, char oxidation, and fuel-N transformation. NOX concentrations
are determined with a NO post-processor model that includes the Zeldovich thermal-NOx
reaction mechanism, and global steps for fuel-NOx formation/destruction 9 and
NO/hydrocarbon interaction. The post-processor model is run after the combustion solutions
are completed. In recent years, COMO has been utilized as a design tool at B&W in
burner development programs.
-------
Table 1
Summary of COMO Burner Model Capabilities
Flow
Turbulence
Chemistry
Radiation
Pollution
Special
Features
• 2- or 3- dimensional Cartesian or cylindrical
• Laminar or turbulent flow
• k - e, multiple time scale
• 2 - step : Eddy dissipation model with CO oxidation kinetics
• Sandia National Lab devolatilization and char oxidation rates
• Discrete ordinates model : 8 or 24 fluxes
• Gas properties with wide band models
• Zeldovich theimal-NOx, and global fuel-NOx and NO/hydrocarbon interaction kinetics
• Provisions to use reduced fuel-N mechanism
• Coal, oil, and natural gas combustion
• Particle or droplet trajectories
• Multigrid solution method to enhance convergence
• Local grid embedding
Figure 1 shows a schematic of the low-NOx pulverized coal burner and the main co-firing
configurations. Co-firing concepts which were integrated into the design for subsequent
modeling include:
• centerline injection by a single spud centered in the coal pipe and aligned with the coal
nozzle exit
• peripheral injection with encircling spuds or gas annuli at various radial locations
In this study, each gas spud was modeled as a circular tube with a single exit. The gas
stream directions for individual spuds or annular openings were specified using velocity
components (i.e., axial, radial, and tangential). Two-dimensional (2-D) axisymmetric and
three-dimensional (3-D) cylindrical models were used to simulate the geometrical
representation and operating conditions of the various configurations. These models simulated
the entire furnace enclosure including the internal burner components and flow passages, and
natural gas injection ports. The combustion chamber was modeled as an idealized cylindrical
tunnel furnace, with thermal boundary conditions and a heat loss rate (400 kWJ similar to
existing pilot-scale research facilities. The cylindrical furnace geometry was chosen for better
numerical compatibility with the cylindrical burner. Dispersion and trajectories of coal
particles were modeled using the Lagrangian approach.
-------
Secondary
Air
Swirl
Vanes
Primary Air.
and Coal
fl
Standard DRB-XCL Configuration
Note: For clarity, vanes shown only for baseline configuration
Centerline
Spud
Centerline Injection
fl
Annular Injection
„ Annular
Gas Zone
Figure 1
Typical Burner Configurations for Numerical Modeling Studies of Natural Gas Co-firing
-------
Figure 2 shows the mid-sectional view of the computational grid for the burner/furnace
system. For 3-D simulations, a 30° wedge segment of the furnace and the burner was
modeled. This permitted a complete representation of the burner when periodic boundary
conditions were applied on the circumferential boundary surfaces at.O and 30 degrees. The
2-D axisymmetric grid modeled only the burner exit area and the furnace. Velocity boundary
conditions for the 2-D models were specified using the computed axial, radial, and tangential
velocity profiles of the baseline 3-D cylindrical case. The closeup view in Figure 2 highlights
the implementation of grid embedding in places where high temperature, high velocity, and
high concentration gradients were expected.
In order to have a common basis for comparison, thermal load and burner stoichiometry were
held constant at 5 MBtu/hr (1.5 MW^) and 16% excess air, respectively. Fuel-specific
kinetics information for high volatile "A" (hvA) Pittsburgh #8 and medium volatile (mv)
Lower Freeport bituminous coals was also incorporated in the model. A summary of the
operating conditions is given in Table 2.
-------
Detail of Region A
Embedded Grid Regions
Region A
Base Grid
Furnace Exit
Plane
Figure 2
Typical Numerical Grid Arrangement
-------
Table 2
Input Operating Conditions for Numerical Modeling
As-Fired Coal Analysis
Proximate (weight %) :
Moisture
Volatile Matter
Fixed Carbon
Ash
Ultimate (weight %) :
Carbon
Hydrogen
Sulfur
Oxygen
Nitrogen
Ash
Moisture
Pittsburgh #8 hvA
bituminous
1.06
36.61
55.44
6.89
78.11
5.22
1.12
6.17
1.43
6.89
1.06
Lower Freeport mv
bituminous
4.85
24.90
60.62
9.63
75.26
4.55
1.26
3.00
1.35
9.73
4.85
Pulverized Coal Size Distribution (weight %) :
70.65% < 200 mesh (74 microns)
99.64% < 50 mesh (297 microns)
Natural Gas Analysis (weight %)
CH4 = 93.30
C2H6 = 3.08
C3H8 =0.27
C4H10 - 0.35
CO2 = 0.80
N2 = 2.20
Thermal Load = 5 MBtu/hr (1.5 MWJ
Excess Air = 16 %
Primary Air Temperature = 150°F (339 K)
Secondary Air Temperature = 620°F (600 K)
Natural Gas Temperature = 80°F (300 K)
-------
Results and Discussions
Initial computer runs for 100% PC firing or baseline operation were carried out in both 2-D
and 3-D geometries. Numerous permutations in gas velocity, injection angle, injection
location, extent of co-firing, and natural gas/air premixing were modeled for comparison and
screening. Because of the large number of modeled cases, only selected predictions for
annular injection of natural gas are presented. Other key results for the co-firing of natural
gas with the high volatile Pittsburgh #8 coal are instead summarized below:
• For a given level of natural gas co-firing, injection velocity had a significant impact on
the NOX emissions but a lesser impact on the unburned carbon loss.
• Increasing the extent of co-firing reduced the NOX and unburned carbon levels.
• Injection methods that enhanced natural gas/air mixing and consumed oxygen quickly near
the burner resulted in lower NOX but higher fly ash unburned carbon.
• Natural gas/air mixing for peripheral injection cases was dominated by the secondary air
flow.
• Introduction of natural gas into an inherently low-NOx and oxygen-deficient zone
demonstrated marginal NOX and unburned carbon reductions.
• NOX removal became more effective when natural gas was delivered into flame regions
where the NOX formation rate was high.
• For annular injection of natural gas, NOX emissions and unburned carbon levels were
strongly influenced by the radial placement of the gas stream relative to the coal nozzle.
Mid-plane predictions of the gas velocity fields, temperature contours, and O2 and CO
concentration profiles for the baseline and peripheral co-firing of the high and medium
volatile coals are compared in Figures 3-6. Here, gas co-firing is equivalent to 15% of the
combined thermal load. Burning of natural gas with the fast mixing secondary air alters the
combustion gas flow patterns. Particularly, it distorts the low momentum regions near the
burner exit as illustrated in Figure 3. Figure 4 shows the differences in the near burner
temperature fields among the four cases. With natural gas co-firing, a narrow and high
temperature region extends out into the secondary air stream. The longer high temperature
lobes are indicative of near stoichiometric combustion zones.
-------
High Volatile Coal Baseline
High Volatile Coal Co-firing
20m/s
Medium Volatile Coal Baseline
Medium Volatile Coal Co-firing
FigureS
Near-burner Velocity Fields for the Baseline (100% PC-firing) and
15% Annular Gas Co-firing Predictions
-------
Temperature Contours in Kelvin
High Volatile Coal Baseline
High Volatile Coal Co-firing
Medium Volatile Coal Baseline
Medium Volatile Coal Co-firing
Figure 4
Near-burner Temperature Contours for the Baseline (100% PC-firing) and
15% Annular Gas Co-firing Predictions
-------
Oxygen Concentration Contours in %Volume
High Volatile Coal Baseline
High Volatile Coal Co-firing
Medium Volatile Coal Baseline
Medium Volatile Coal Co-firing
Figure 5
Near-burner Oxygen Concentrations for the Baseline (100% PC-firing) and
15% Annular Gas Co-firing Predictions
-------
Cartoon Monoxide Concentration Contours in %Volume
0.5-7?
High Volatile Coal Baseline
High Volatile Coal Co-firing
Medium Volatile Coal Baseline
Medium Volatile Coal Co-firing
Figure 6
Near-burner Carbon Monoxide Concentrations for the Baseline (100% PC-firing) and
15% Annular Gas Co-firing Predictions
-------
Contours of oxygen concentrations are shown in Figure 5. Relative to the baseline results,
the co-firing cases show a more rapid consumption of oxygen in the upstream section of the
furnace tunnel where natural gas burns. Carbon monoxide concentrations are also generally
high in the vicinity of gas injection and the central primary combustion zones as seen in
Figure 6. As natural gas burns, the ensuing flame shields the substoichiometric primary
combustion zone (where fuel-N species evolve) from oxygen penetration to reduce NOX
formation.
Table 3 lists the normalized furnace exit values of NOX, gas temperature, and unburned
carbon in the fly ash (UBCA) relative to the baseline (100% PC firing of low-NOx burner)
results for two different coals. Flame lengths were also determined qualitatively based on a
pre-specified and arbitrary CO concentration. Tabulated flame lengths are for trend
indications only. Actual flame length depends on additional factors such as soot radiation,
remaining char fraction, and temperature and oxygen distribution within the furnace. As
expected, the baseline NOX and unburned carbon levels for the more reactive high volatile
Pittsburgh #8 coal are lower than values predicted for the medium volatile matter Lower
Freeport coal. Burning 15% natural gas with Pittsburgh #8 coal reduced the NOX
concentration and unburned carbon level by 16 and 19%, respectively. Furnace exit gas
temperature (FEGT) changed by 3% due to co-firing.
Model predictions indicate that a very small fraction of NOX reduction from managed natural
gas co-firing is attributable to fuel-N displacement. The bulk of the decrease is associated
with lower local O2 concentrations and hydrocarbon interactions with NO via the CH; + NO —>
HCN and HCN + NO -» N2 and reactions. Considering the fact that NOX emissions from the
low-NOx PC burner are already about 50% below the uncontrolled levels of conventional
burners, the net NOX reductions with 15% gas use amount to about 53% for the medium
volatile coal and 60% for the high volatile coal.
-------
Table 3
Summary of Key Modeling Results
Coal
hvA
Pittsburgh #8
rav
Lower Freeport
Case Description
100% PC Firing
15% Natural Gas Co-Firing
100% PC Firing
(relative to hvA)
15% Natural Gas Co-Firing
Relative
NOX*
1
0.84
1
(1.06)
0.95
Relative
UBCA*
1
0.81
1
(2.07)
1.04
Relative
FEGT*
1
0.97
1
(0.96)
1.03
Flame
Length**
1
1.04
1
(1-04)
1.02
*
**
Relative exit quantities defined as Xrelative = Xcofire/Xbaseline low.NOx
Flame length defined arbitrarily by contours of 1% CO as a trend indicator. Actual flame
length depends on additional factors.
Conclusions
Implementation of natural gas co-firing in a low-NOx PC burner was studied by numerical
simulation of several concepts. Combustion modeling effectively captured the influence of
operating conditions and hardware configuration. Natural gas introduction into appropriate
locations of the pulverized coal flame showed possible reductions in NOX emissions and
higher carbon utilization. Gas co-firing with a high volatile Pittsburgh #8 coal was more
effective for NOX and unburned carbon reductions in comparison with a medium volatile
Lower Freeport coal. Relative to 100% pulverized Pittsburgh #8 coal firing in a low-NOx
burner, the NOX emissions and unburned carbon levels for 15% natural gas use were 16 and
19% lower, respectively. Since the NOX emission levels for the baseline low-NOx pulverized
coal burner are already about 50% lower than the unregulated values of conventional burners,
total NOX reduction with 15% gas use would be close to 60%. Future plans include further
modeling and optimization of the gas co-firing concept(s), and proof-of-concept verification
tests in a pilot-scale boiler facility.
-------
Acknowledgement
This work was sponsored by the Electric Power Research Institute. We would like to thank
Dr. George Off en of EPRI and Mr. Albert LaRue of B&W for many informative discussions
and helpful suggestions.
References
1. Designation : C 618-93, Standard Specifications for Fly Ash and Raw or Calcined
Natural Pozzolan for Use as a Mineral Admixture in Portland Cement Concrete, ASTM
Annual Book of Standards, 4.02, 1993.
2. Vejtasa, S.A., Biasca, F.E., Giovanni, D.V., and Carr, R.C.,"Gas Cofiring for Coal-Fired
Utility Boilers," Final Report Prepared by SFA Pacific Inc. and Electric Power
Technologies Inc., for Gas Research Institute and Electric Power Research Institute,
November 1992.
3. Smart, J.P., and Morgan, D.J.,"Multi-Fuel Reburning in a Coal Fired Internally Fuel
Staged Burner for NOX Control," Presented at the 1993 Joint Symposium on Stationary
Combustion NOX Control, Bal Harbour, FL, May 1993.
4 LaRue, A.D.,"The XCL Burner - Latest Developments and Operating Experience,
Presented at the 1989 Joint Symposium on Stationary Combustion NOX Control", San
Francisco, CA, March 1989.
5. Fiveland, WA., Oberjohn, W.J., and Cornelius, D.K., "COMO: A Numerical Model for
Predicting Furnace Performance in Axisymmetric Geometries," ASME Paper No. 84-HT-
103, 1984.
6. Fiveland, W.A., and Wessel, R.A./'Numerical Model for Predicting Performance of
Three-Dimensional Pulverized Fuel-Fired Furnaces," J Eng. Gas Turb. & Power 110,
117-126, 1988.
7. Fiveland, W.A., and Wessel, R.A.,"Model for Predicting Formation and Reduction of
Nitric Oxide Pollutants in Three-Dimensional Furnaces Burning Pulverized Fuel,"
Journal of the Institute of Energy 64, 41-54, March 1991.
8. Fiveland, W.A. and Jessee J.P.,"Mathematical Modeling of Pulverized Coal Combustion
in Axisymmetric Geometries," Proceedings of the ASME/EPRI Joint Power Conference,
Phoenix, Arizona, October, 1994.
9. DeSoete, G. G.,"Overall Reaction Rates of NO and N2 Formation from Fuel Nitrogen,"
Fifteenth International Symposium on Combustion, Pittsburgh, PA, The Combustion
Institute, 1093-1102, 1975.
-------
AN INTEGRATED FULL, PILOT AND LABORATORY SCALE STUDY OF THE EFFECT OF COAL
QUALITY ON NOx AND UNBURNT CARBON FORMATION
A R Jones, W H Gibb, R M A Irons and H J Price
PowerGen PLC, Power Technology Centre,
Ratcliffe-on-Soar, Nottingham NG11 OEE, England
J W Stallings and A K Mehta
Electric Power Research Institute
3412 Hillview Avenue, Palo Alto,
California 94304, USA
Abstract
PowerGen and EPRI have co-funded a two year research project to investigate the effect of coal
quality and coal blends on NOx formation and unburnt carbon in low-NOx combustion systems.
The objective of the programme was to develop a better understanding of these effects in relation
to combustion staging and coal blending in order to derive more reliable predictive methods.
Eight bituminous coals covering a wide range of properties and geographical source were
included in the study.
The comprehensive work programme comprised full-scale plant trials on a 500 MWe tangentially
fired boiler fitted with LNCFS backed up by smaller scale testing at the Power Technology Centre
using a 1MW (3.4 MBtu/h) Combustion Test Facility and a laboratory Drop Tube Furnace. Coal
and char samples pertinent to all experimental scales have also been examined at Nottingham
University using a recent development of an automated image analysis technique for coal
petrology and char characterisation.
The overall results of the programme are summarised in this paper. The value of rig testing for the
prediction of NOx emissions from individual coals in full-scale plant has been successfully
demonstrated, through calibration with plant data. As expected, NOx emissions do not correlate
simply with either nitrogen content or with fuel ratio based on laboratory determined volatile
matter; a stronger correlation is found with a function based on both coal volatile matter and
nitrogen content. It has been shown that the sensitivity of NOx emissions to coal quality is
reduced as the degree of combustion staging increases. All coals tend towards a NOx minimum
of 150-200vpm under deep air staging.
The level of unburnt carbon, an often ignored disadvantage of low NOx combustion systems, has
been shown to be highly dependent on coal quality, but is much less sensitive to combustion
staging. A novel coal characterisation method based on grey scale reflectance shows
considerable promise as an index for carbon burnout propensity.
Introduction
Generally throughout the world, emission limits for NOx are becoming more stringent,
necessitating the application of one (or more) of a range of NOx reduction technologies
appropriate to the legislative requirements of the different countries. One of the more favoured
groups of technologies involves the use of combustion staging and includes such concepts as
low-NOx burners, overfire air (OFA) and reburn. However, although this type of technology is
attractive from a capital outlay perspective there is a potential penalty arising from a tendency for
the level of unburnt carbon (UBC) in the flyash to increase. This represents an efficiency loss and,
as there is a carbon-in-ash limit for certain uses for fly ash, could also have a deleterious effect on
ash sales to the construction industry.
-------
NOx emissions and UBC levels are clearly highly dependent on boiler design and operation, but
experience has also shown that coal quality can have a significant effect. Accordingly, the UK
electricity generating company PowerGen has collaborated with EPRI in a research project to
investigate the effect of coal quality on NOx emissions and unburnt carbon. The project
comprised full-scale plant trials supported by smaller scale testing using a 1MW single burner pilot
scale combustor (the Combustion Test Facility, CTF) and a laboratory Drop Tube Furnace (DTF).
The main objective of the project was to attempt to develop a simple, but reliable, technique for
predicting both NOx and UBC levels for a given coal, preferably not involving combustion or the
use of expensive or complex testing equipment. To this end, automated image analysis
techniques for both coal petrology and char characterisation recently developed at Nottingham
University have been applied to the coal and char samples produced at the different scales of
combustion.
Project Outline
The project centred on, and began with, controlled full scale test burns of three coals and a 50:50
blend of two of these coals on one of four 500 MWe boilers at PowerGen's Kingsnorth Power
Station. Extensive flue gas and paniculate analyses and performance data were collected during
the trials and samples of the test coals of sufficient size for the other phases of the project were
withdrawn to Power Technology.
The CTF programme included the characterisation of the three power station coals and the blend,
plus five additional coals and two more blends thereby extending the range of coal qualities
investigated in the project. Finally, the opportunity was taken to investigate the effect of the
particle size distribution of the ground pulverised coal on both NOx emissions and UBC; for this
exercise the three primary coals featuring in the plant trials were used.
The Drop Tube Furnace is becoming more widely accepted as a coal characterisation technique
that can closely approach the high heating rates and high temperatures experienced by coal
particles in a utility boiler, but on much smaller samples and without the need to establish a stable
flame. Characterisation of the devolatilisation is carried out under much more representative
conditions than produced by standard ASTM (D3175-89) or other equivalent international
laboratory methods. In addition, the DTF provides a better method of evaluating the reactivity or
burnout propensity of the chars remaining after devolatilisation and of determining the
denitrification profile of both coals and chars. The suite of eight coals tested on the CTF were fully
characterised on the DTF during this programme.
A major objective of the work programme was to try to calibrate the two pilot scale facilities with
the power plant in order to determine the reliability of such techniques for making predictions of
the propensity of different coals to produce NOx and unburnt carbon. However, it is recognised
that although the CTF, and to a much lesser extent the DTF, provide more realistic conditions
under which to test and evaluate coals and should provide credible performance predictions of
these two combustion bi-products, such techniques are still too complex to be considered as
standard for routine application. Each suffers from a combination of the following disadvantages;
capital and revenue intensive, requires expert operation and data interpretation, needs large coal
samples and must be pre-calibrated with the boiler under consideration. Consequently it was
decided to add to the project an evaluation of the potential of a new image analysis based
technique being pioneered in the Chemical Engineering Department at Nottingham University to
provide accurate performance predictions of unburnt carbon.
The Nottingham technique, which was initially developed under the co-sponsorship of PowerGen
and the Engineering and Physical Sciences Research Council in the UK, combines both
automated and semi-automated petrographic/reflectance analysis of the coals with detailed
automated analysis of the morphology of chars produced both under full and pilot scale
combustion conditions. Should analysis of small samples of the coal in this way prove to correlate
well with combustion data, the technique will provide a non-combustion and less subjective
method of assessing the reactivity of different coals.
-------
Plant Trials at Kingsnorth Power Station
Description of Plant
Kingsnorth Power Station comprises 4 x 500 MWe split furnace, tangentially fired boilers. There
are five elevations of coal burners within each corner arrangement (Figure 1), the eight burners on
each elevation being fed by a Lopulco type mill (A mill feeds the top burner elevation and E mill
the bottom). The full burner configuration is of the LNCFS type with large offset secondary air
nozzles intermixed with the coal and light up oil burners and a separated overfire air arrangement.
Kingsnorth was originally designed for mixed firing (representing a 100% capability on either
residual fuel oil or coal) and is therefore effectively one mill/burner group short of ideal in the now
routine coal only operational mode.
Selection of Coals
The coals for the four separate trials were:
(1) Ashland - a typical US coal that has been a major supply to PowerGen.
(2) El Cerrejon a Colombian coal that has been associated with high carbon-in-ash levels. This
coal also has a low fuel ratio (fixed carbon to volatile matter) and might, therefore, be expected to
give relatively low NOx emissions.
(3) Tyne Blend a typical UK coal that is routinely supplied to Kingsnorth.
(4) A 50:50 blend of El Cerrejon and Tyne Blend - this trial was designed to quantify the effects of
blending.
Test Schedule
Each trial consisted of 2 days testing during which the Unit was operated exclusively with the
chosen coal. Although 5 in service mills are always preferable, the boiler often operates with 4
mills, ideally B,C,D, and E. Previous experience has shown that whether the top mill (A) is in or
out of service affects NOx formation. Accordingly, the operational variables selected for
investigation were excess oxygen, overfire air, and mill grouping as follows:
Day 1 - top mill (A) out-of-service, four tests each of about 1 h duration. Three tests at the normal
overfire air (OFA) setting, ie 18%, covering a range of excess 02 levels, about 1.5, 2.5, and 3.5%.
Fourth test again at about 3.5% excess O2, but with the OFA reduced to about 10%.
Day 2 - as for Day 1, but with A mill in service and the bottom mill (E) out-of-service.
As far as possible, each test was designed to be at full load with operating conditions held
constant and no sootblowing or ashing out. The burner tilts were fixed at +8°.
Measurement and Sampling Schedule
Gas Analysis. The concentrations of NO, SO2, CO, and O2, were continually monitored by
extractive sampling at the ID fan discharges on the A and B sides of the boiler.
Carbon-in-Ash. Representative flyash samples were obtained by multi-point, isokinetic sampling
on each side of the boiler at the air heater inlet to give separate A and B side samples.
Coal and PF Samples. 20 tonne samples of each coal were collected for subsequent testing on
the CTF and DTF.
On each test day, samples of pf were taken from each operating mill.
-------
Results
Test Record and Basic Data. In setting up the boiler to the desired excess O2 concentration for
each test, it was impossible to achieve a balance between the A and B sides. The difference was
generally small, up to about 0.3%, when A mill was out of service but, with A mill in service, varied
from 0.5 to 1.1%.
The overfire air (OFA) percentage was reduced to 10% by shutting the appropriate OFA damper.
No attempt was made to adjust the windbox pressure which was, therefore, higher during these
tests than the equivalent test (3.5% nominal 02) at 18% OFA.
Coal Analysis. The results of the analysis are presented in Table 1.
Pulverised Fuel (pi) • Size Grading. All of the pf samples taken from the operating mills during
each test were size graded. Single point, non-isokinetic sampling is notoriously unreliable and
these results were treated with the usual caution. Despite this reservation, it can be stated with
some confidence that the proportion of coarse material, ie % >300^m, produced by the mills is
quite low, i.e. less than 2%. However, beyond this, it is impossible to say whether there was any
significant difference in pf grading between coals.
Carbon-in-Ash. A summary of the residual carbon-in-ash results for the three individual coals
plotted against excess O2 is presented as Figure 2. The data obtained during the blend trial were
quite scattered, but in general they were found to lie between those obtained in the individual El
Cerrejon and Tyne trials. There were differences between the results for the A and B sides of the
boiler with and without A mill in service, but no consistent patterns could be detected.
Whilst the results for Ashland and Tyne Blend can reasonably be represented by a single curve, El
Cerrejon coal clearly gave higher results. However, for coals with different ash contents, the direct
comparison of carbon-in-ash values is potentially misleading as an indicator of residual
combustible matter or efficiency loss. Accordingly, by taking account of the ash contents of the
different test coals (Table 1), the carbon-in-ash values have been converted into unburnt carbon or
carbon loss values as a percentage of the dry coal. All of the carbon loss data for the individual
coals are plotted against excess 02 in Figure 3. As the ash contents of the Ashland and Tyne
Blend coals were very similar, the results for these coals can again be represented by a single
curve. For El Cerrejon, the ash content was lower (about 10% compared to 13-14%) and,
compared to Figure 2, the results are, therefore, closer to those for Ashland and Tyne Blend.
However, on average, they are still higher by about 50%.
NO Measurements. The relationships between NO and excess oxygen for each individual coal for
the two firing configurations are shown in Figures 4 and 5. No major differences are apparent
between the different coals and, with A mill in service, all the points can reasonably be
represented by a single line (Figure 4). With A mill out of service (Figure 5), the NO emissions
were slightly higher, about 25vpm, when firing Tyne Blend coal compared to Ashland and El
Cerrejon which can again be represented by one line. As expected, the NO emissions are
consistently higher with 'high firing1, i.e. the top burners (A mill group), in service. At 6.5% O2,
equivalent to about 3.2% wet at the air heater inlet, the difference is between 50 and 75vpm,
depending on the coal. There is also a suggestion that NO is slightly more sensitive to O2 with A
mill in service, ie the line is slightly steeper.
With A mill out of service, the decrease in overfire air to about 10% had little or no effect on NO
emissions and so the points have not been plotted on Figure 5. In contrast, however, it can be
seen in Figure 4 that, with A mill in service, the decrease in overfire air from the normal 18% to
10% caused the expected increase in NO emissions. However, the magnitude of the effect
appears to be coal dependent and varies from about 75vpm for Tyne Blend down to 10-30vpm for
Ashland.
-------
Discussion
NO Emissions. For the three coals chosen for the plant trials very little difference in NO emissions
was found under normal operating conditions (18% OFA). Overall, the results for Tyne were
slightly higher than for Ashland and El Cerrejon, but only by about 25vpm with A mill out of service
(Figure 5). Under normal operating conditions with A mill out of service, 18% OFA and 3.5% (wet)
excess O2 measured at the air heater inlet, NO emissions of 330-360vpm (3% 02 dry) were
measured, which is within the 390vpm target for the Station.
In view of the similarities of the nitrogen contents and fuel ratios for the three coals (Table 1),
comparable NOx emission levels would perhaps be expected. However, previous trials of South
African coals on Unit 3 at Kingsnorth PS showed that NO emissions can be very sensitive to coal
quality, with one of the coals giving about 530vpm NO at 5.5% O2 dry (A mill in service), compared
to about 330vpm for these trials under similar conditions. This is illustrated in Figure 6 where the
data obtained from these previous trials are plotted alongside the results of the current study. It
was the test of this coal that revealed that it would not be possible to operate the boilers below the
agreed NOx emission limit on certain coals and which gave impetus to this study. The small
difference between the two sets of Tyne Blend data may be associated with different levels of air
heater inleakage between the tests on Units 1 and 3. It is the furnace O2 concentration that
actually determines the level of NO formation.
As expected, the burner firing pattern had an effect on NO emissions. With the top (A) mill in
service, NO emissions were up to 75vpm higher than when it was out of service. This is in line
with previous findings and is believed to be due to the longer residence time of the pf in the sub-
stoichiometric zone with E mill in service compared to A mill in service.
The investigation of the effect of OFA on NO emissions produced some interesting differences,
depending on mill combination. With A mill in service, a reduction in OFA from a nominal 18%
down to 10% did result in an increase in NO emissions, but the effect was coal sensitive. Tyne
Blend gave the greatest increase, about 75vpm, with El Cerrejon and Ashland some 20-55vpm
lower. In contrast, with A mill out of service, no measurable effect was observed. The effect of mill
combination is probably related, in part at least, to differences in the true proportions of OFA and
combustion air. With A mill out of service, the cooling air passing through A group burners has
been calculated to be equivalent to about 4% OFA. In effect, therefore, the tests with A mill out of
service were conducted with 22% and 14% OFA, compared to 18% and 10% with A mill out in
service. The effect of the proportion of OFA on NO emissions is not linear, but is an asymptotic
curve approaching a minimum attainable NO level or maximum reduction. The 18% OFA
operational norm was chosen as any further increase would simply increase the unburnt carbon
without any significant reduction in NO being achieved. This is fairly generally accepted and has
been confirmed by the tests on the CTF (see below) which indicate that NO emissions are
relatively insensitive to OFA above about 15%. It follows, therefore, that little effect would be
expected between 14 and 22% OFA (A mill out of service), compared to the change between 10
and 18% OFA (A mill in service). The other contributory factor is the lower baseline NO level with
A mill out of service which will make it intrinsically less sensitive to changes in OFA.
The coal sensitivity of the effect of changing OFA with A mill in service is more difficult to explain,
but may relate to the relative amounts of char and volatile nitrogen produced by the different
coals. The similarity in the basic NO levels obtained with the three coals suggests comparable
char nitrogen levels. It is possible, therefore, that, as the coal nitrogen content of Tyne Blend is a
little higher than El Cerrejon and Ashland (Table 1), the amount of volatile nitrogen released would
be greater. Under these conditions, the NO formed when burning Tyne Blend would be more
sensitive to changes in OFA.
Unburnt Carbon. El Cerrejon coal is widely reported to give rise to high carbon-in-ash levels and
this has been found during trials at other PowerGen plant. This is confirmed by these trials and,
after taking the difference in ash contents into account, the mean increase compared to Tyne
Blend and Ashland was about 50% - see Figure 3. It is assumed that this difference is related to
the nature of the char produced by devolatilisation of El Cerrejon coal, ie differences in intrinsic
-------
char reactivity, or, as would seem more likely from the results of other phases of this work, the
porosity of the char.
The relationship between unburnt carbon and excess oxygen found for Tyne Blend (and Ashland)
was in line with that found during recent thorough boiler characterisation tests. Under normal
operating conditions of 3.5% (wet) excess 02 measured at the air heater inlet (equivalent to 6.7%
dry at the ID fan discharge), the unburnt carbon-in-ash value was around 5% - see Figure 2.
Combustion Test Facility Tests
Description
The 1 MWt (3.4 MBtu/h) rated CTF has been described in detail elsewhere1 and so only an outline
description is given here; the general layout of the CTF is shown in Figure 7. The facility is
horizontally fired by a single burner and has been designed with a high degree of fuel flexibility; it
is capable of firing on pulverised coal, residual fuel oil (RFO), Orimulsion and natural gas. The
furnace and high temperature convective pass are refractory lined and water cooled. The basic
design philosophy was to provide an accurate simulation of the time/temperature profile of full-
scale plant, i.e. residence time scaling.
For the tests reported here, the rig was fired with a single low-NOx PF burner with a combustion
air system comprising primary, secondary, tertiary, and furnace overfire air. When overfire air was
being used the flow was adjusted by diverting proportions of the secondary and tertiary air flows
such that the ratio of secondary to tertiary air flow remained constant.
Continuous measurement was made of O2, CO, SO2, and NO in the flue gas.
Test Coals
The additional five coals introduced to broaden the range to be tested at this scale were typical
coals from Australia (Drayton) and South Africa (ATC), two further US coals previously fired on the
Southern Research Institute pilot scale facility in Birmingham, Alabama (Shoal Creek and North
River) under a complementary EPRI supported programme2, and a high rank UK coal
(Leasingthorne). The proximate and ultimate analyses of all eight of these coals are shown in
Table 2. The size of the PF is also shown as the percentage less than
Three 50:50 blends were also tested; Tyne Blend/El Cerrejon (as per the plant trials), Tyne
Blend/Leasingthorne and Ashland/Shoal Creek. Finally, the effect of changing the particle size
distribution of the ground coal for three of the coals, Tyne Blend, Ashland and El Cerrejon has
been assessed.
Test Schedule
Each coal was fired on three separate days to cover a range of OFA levels 0, 1 5, and 25% of
total air flow. On each day the effect of excess oxygen concentration was studied by making
measurements under steady operating conditions at about 4.5, 3.0 and 1 .0% O2. At the beginning
and end of each test day the repeatability of the NO v. O2 characteristic and the CO breakpoint
was checked by ramping the 02 from 4.5 to about 1.0% over approximately 30 minutes.
NO Measurement Results
The NO and 02 concentrations in the flue gas were monitored throughout the experiments to
provide mean NO results for each test condition. Early in the programme the NO2 concentration
in flue gas was also checked and found to be extremely low, typically 3vpm, maximum 7vpm.
Accordingly, this measurement was discontinued and measured NO concentrations are assumed
to equate to NOx emissions.
-------
The NO results obtained are shown in Table 3 and are quoted in volume parts per million (vpm)
on a dry basis, corrected to 3% O2 (measured dry). Throughout these experiments the
relationship between excess O2 and NO was found to be linear. These relationships are shown for
the 8 coals at the zero and 15% OFA settings in Figures 8 and 9: the results at 25% OFA were
only marginally different from those at 15% OFA.
Unburnt Carbon Results
To study carbon burnout, flyash samples were taken from 5 sampling ports from the flame zone
through to the end of the top pass where combustion is reasonably assumed to be complete. The
samples were extracted using a wet sampling system.
The unburnt carbon content or loss-on-ignition values for the final flyash samples obtained from
the final sampling point (gas temperature about 800°C) are presented in Table 4. To correct for
the differences in ash content and derive an equivalent boiler efficiency penalty, the results are
also shown as a carbon loss as a percentage of the original coal, ie ash fraction multiplied by
carbon-in-ash.
Discuss/on
NO Measurements • General Observations. A number of general trends are apparent from the
results presented in Table 3 and Figures 8 and 9. First, over the range tested, the level of NO
formation is linearly dependent on excess O2 concentration, but the slope of the line is variable
depending on the degree of combustion staging. At zero OFA the mean slope for all the test
coals is 46 vpm/% O2, falling to 37 at 15% OFA; at 25% OFA it was 26 vpm/% O2. Second, a
dependence on coal quality is also clearly apparent with a spread in the NO results reducing from
about 70vpm at zero OFA to about SOvpm at 15% OFA (and 35vpm at 25% OFA). Some variation
in the O2 dependence is also apparent. Third, the reduction in NO that can be obtained by the
use of OFA is also clearly apparent, with a mean reduction of 20% at 3% O2 obtained with 15%
OFA. The gains obtained from 25% OFA are more modest with a further 7% reduction relative to
zero OFA. Again, the level of reduction obtained is somewhat coal dependent.
Overall, it appears that the NO level obtainable from any coal tends towards a minimum value in
the range 150 to 200vpm. This minimum will be fixed by the nitrogen content of the char obtained
by devolatilisation of that coal. The NO formed by combustion of the char will not be influenced
by combustion staging.
The Relationship between NO Emissions and Coal Quality. It is generally agreed that >80% of
NO is formed from the fuel nitrogen and it is to be expected, therefore, that NO emissions will be
related to a coal's nitrogen content. It is also recognised that the high temperature volatile matter
release is an important coal property, in that the higher the proportion of the coal nitrogen that is
released during devolatilisation, the lower will be the NO formation by the subsequent oxidation of
the char nitrogen. Standard analysis provides a measure of volatile matter content under
laboratory conditions that is lower than the volatile release under high heating rate conditions.
Although it will be seen in the section below on the DTF that the ratio of high heating rate to
laboratory derived volatile matter covers the range 1.31 to 1.79, if this factor was assumed to be
constant then proximate volatile matter content should also relate to NO formation. This is
conventionally indicated by the fuel ratio which is defined as the ratio of fixed carbon to volatile
matter.
On the above basis, NO emissions would be expected to show some relationship with the product
of coal nitrogen content and fuel ratio. Such a plot for the 8 test coals is shown in Figure 10. A
rough trend is apparent, but there is considerable scatter in the data and there is too great a
concentration of data points at low fuel ratio x nitrogen values and a sparsity at higher values.
In a highly staged combustion system it is probable that the vast majority of the NO is formed by
oxidation of the char nitrogen and the contribution from the volatile nitrogen species approaches
zero. Assuming that (1) fuel ratio is a reasonable guide to volatile release and, (2) the partitioning
-------
of nitrogen between the volatiles and char matches the overall volatile/char ratio, then a high
correlation coefficient would be expected. In practice, though the second assumption is normally
true, the first does not necessarily apply. It is probable that variations in the high heating rate
volatile release relative to the proximate analysis volatile matter content accounts for some of the
scatter in the data shown in Figure 10. (Thus is the subject of further discussion in the section
below on the DTP). The lower correlation coefficient for the 0% OFA results is probably due to a
variable contribution of volatile nitrogen to the higher NO levels under less staged combustion
conditions.
Unburnt Carbon Results. Table 4 shows that for each of the coals the expected trend of
combustibles loss with operational excess oxygen was observed, but there is no apparent effect of
overfire air. Plant experience with overfire air generally would suggest that the deeper the staging,
the higher the residual combustibles, but this was not observed on the test facility. A reasonable
explanation for this is that being a small, well mixed reactor, the CTF still provides the necessary
conditions (i.e. sufficient residence time at high temperature and high local oxygen partial
pressures) for good burnout even under deeply staged conditions.
Marked differences between the coals are apparent. Looking, for example, at the carbon loss
results at 1% excess O2 (similar trends are apparent at other 02 levels), there is about a factor of 2
difference between the least reactive coal, ATC and the more reactive coals. During the
Kingsnorth trials the results (on a carbon loss basis) for El Cerrejon were found to be about 50%
higher that those for Ashland and Tyne Blend. The CTF results for El Cerrejon were generally
higher than those for the other two coals, but not on average by as much as 50%. Again, this
difference could be due to the better control of the fuel/air stoichiometry on the CTF
Comparison of CTF and Plant NO Results
The CTF has been designed to provide a realistic simulation of combustion conditions in real plant
and should, therefore, give similar NO emissions. In this respect the results obtained are very
encouraging. As an example, some comparative results for Ashland are shown in Figure 11. The
plant results are those obtained with A mill out of service and a nominal 15% OFA, equivalent to
about 22% when the air from the standing burner is included. The CTF results included are those
for 0 and 15% OFA.
Agreement between the CTF 0% OFA data and the plant data is excellent. The results obtained
on the CTF at 15% OFA (or 25%, the two sets of data are very similar) were much lower and at
first sight would appear to show poor agreement under notionally similar conditions. However, it is
believed that the 0% OFA case on the CTF actually represents the more accurate simulation of the
LNCFS at Kingsnorth. In this system, pf is simply blown into the furnace from each corner with
offset secondary air and there are no rooted flames. The overall system, therefore, effectively
behaves as a single large burner, with the OFA acting in a similar manner to the tertiary air in the
single rooted flame obtained in the CTF. The greater staging that can be obtained from a rooted
flame with OFA results in the lower NO levels obtained on the CTF with 15% and 25% OFA.
This theory is supported by results obtained with International Combustion's (ICL) recently
developed FAN (flame attached nozzle) burner. In this tangentially fired system the corner flames
are rooted by the inclusion of a flame holder and, therefore, simulate the single low NOx burner in
the CTF. This has been shown on plant to give a reduction in NO formation of about 50-80vpm3;
if applied at Kingsnorth, for Ashland coal this would give a similar result to that obtained on the
CTF at 15% OFA.
As part of a separate study, the S.African ATC coal has also been fired at Kingsnorth and the
results obtained there and on the CTF (0% OFA) are also included in Figure 11. Again agreement
is fairly good. It is also encouraging that the sensitivity to excess oxygen on the CTF, ie the slope
of the NO v. 02 line, is similar to that obtained at Kingsnorth PS.
-------
Drop Tube Furnace Tests
Description of Equipment
The DTP (Figure 12) comprises an electrically heated work tube which is approximately 1.5m in
length with an inside diameter of about 50mm. Coal particles are fed through a screw feeder into
the DTF with a nitrogen purge via a water-cooled feeder probe. The coal devolatilises in a
controlled, high-temperature atmosphere and the resultant char material is collected by a water-
cooled probe. From the collector probe the char particles and waste gases pass through a
cyclone, where the solid particles are removed. When char material is re-fired for reactivity
determinations, it is fed in the same way as coal but the resultant residue is collected using a
millipore filter rather than the cyclone. The separation between the feeder and collector probes
can be varied up to 60cm to provide residence times up to 600ms depending on the temperature
and gas flow. The DTF can be operated at up to 1400°C and air and nitrogen are premixed to
give a known O2 concentration in the reaction zone.
Experimental Programme
The test conditions chosen for the devolatilisation experiments were a temperature of 1300°C and
an atmosphere of 1 % 02 in nitrogen. It is normal practice to test size graded coal for ease of
feeding and to eliminate variations in size distribution between coals. 53-75fjm and 106-125/jm
size cuts were prepared from a nominally <212^m sample of each coal taken from the Combustion
Test Facility feeder, the former being typical of the mean for pulverised fuel. For these size grades
devolatilisation is complete in under 100ms and, at only 1% O2, little change occurs thereafter.
Using a longer residence time gives a more even temperature distribution, ie minimises end
effects, and 200ms was adopted for these tests. The chars were collected and refired at a
temperature of 1300°C and in an atmosphere of 5% O2 in nitrogen, at residences times of 200, 400
and 600ms.
A proximate volatile test was carried out for each coal feed and the coal feed and resultant chars
and residues were analysed for carbon, hydrogen and nitrogen. The loss on ignition (LOI) at
800°C was determined for each coal feed sample.
Results
Devolatilisation. Table 5 gives a summary of the devolatilisation data from the DTF experiments.
The percentage loss of raw coal determined on the DTF ie the high heating rate volatile matter on
a dry basis (VMhhr) and the "Ft" factor, the ratio of the volatile matter content (dry basis) of the feed
coal determined by proximate analysis to VMhhr, are shown for each coal. The combustibles loss,
ie the proportion on non-ash material lost, was determined by the ash tracer method. The
nitrogen loss was calculated from the percentage of nitrogen in the feed coal and resultant char in
the same way as combustibles loss. The relative amounts of nitrogen in the char and volatile
phase were determined from this nitrogen loss.
Figure 13 compares the combustible and nitrogen loss results. There is some scatter in the data
but for devolatilisation the fractional nitrogen loss is on average slightly less than the combustible
loss.
Char Retiring. Table 6 gives a summary of the char refiring data. The average measured NO was
calculated from the NO measured during the char refiring phase. The char conversion was
calculated by the ash tracer method in the same way as the combustible loss for the
devolatilisation stage. To compare all the coals on the same basis the NO figure was converted to
a value representing the NO produced if 100% of the carbon-rich fraction of the char had been
converted and then normalised for a net CV of 29,000 kJkg"1. Figure 13 also includes a
comparison of the combustible and nitrogen loss results under char refiring. The fractional
nitrogen loss is very close to the combustible loss for the char refiring stage.
Figure 14 shows the fractional nitrogen loss as a function of residence time for three of the coals
-------
tested. As expected the fractional nitrogen loss increases with residence time for all coals. ATC is
markedly different from the other coals in that the fractional nitrogen losses are lower and the
curve does not approach an asymptote as residence time increases.
A prediction of the amount of NO produced can be generated for each coal. Figure 15 illustrates
the actual NO produced as a function of the predicted value which assumes full conversion of all
evolved nitrogen to NO. A linear regression of the data gives an NO conversion efficiency of
approximately 50%.
Discussion
The VMhhr values for the suite of coals range from 41% for Shoal Creek (106-125^m) to 61% for El
Cerrejon and Ashland (53-75^/m). The 'R1 Factor for the suite of coals varies between 1.31 and
1.79, with a mean of 1.57. In general, therefore, the proximate volatile content can only be
considered as an approximate guide to the high heating rate volatile release. These 'R' factors
have been used to create a high heating rate fuel ratio and the result has been used to modify the
plot of CTF NOx emissions v. fuel ratio x nitrogen content shown in Figure 10 to produce Figure
16. It can be seen that the correlation is much better, thereby emphasising the limited value of
some data derived under the conditions incorporated into current standard procedures.
Table 5 shows a range of partitioning for the coals tested which reflect the differences in both
percentage nitrogen loss (35-58%) and the absolute levels of nitrogen in the original coal feed
(1.43-1.81%). The hypothesis described above that the level of NOx will be primarily related to the
nitrogen content of the char in a low NOx system is broadly supported by the trend shown in
Figure 17, in which NOx measurements from the CTF trials are plotted as a function of the char
nitrogen g/1 OOg coal from DTF. ATC, which gave the highest NOx levels on the CTF, has a
relatively high char nitrogen g/1 OOg coal and Ashland, which gave a lower NOx reading, has a
much lower char nitrogen g/1 OOg coal. The same trend is evident when char nitrogen g/1 OOg coal
is replaced with NO produced from the DTF (figure 18).
The unburnt carbon, as represented by LOI, shows the same trend for the DTF and CTF
experiments (Figure 19). ATC, which has a high level of unburnt carbon when burnt on the CTF
also gave a high value for the residue collected after char was refired at 600ms. Shoal Creek and
Leasingthorne appear mid-range and the other coals all have relatively low LOIs, data which are
borne out by plant experience.
Application of Image Analysis
At Nottingham University, image analysis programmes have been developed, specifically aimed at
coals for pulverised coal combustion, which can carry out petrographic/refiectance analysis both
automatically and semi-automatically (interactively) as well as semi-automated and automated
analysis of char morphology4. This novel technology, which enables grey scale histograms of the
type shown in Figures 20 and 21 (for the eight coals in this study) to be produced, was
supplemented by the application of Thermogravimetric and Differential Thermal Analyses
(TGA/DTA) to measure intrinsic reactivity. The bar at the lower grey-scales represents liptinite
which is determined in a second automated run using blue light to distinguish it from the mounting
resin used in the block preparation. In the image analysis technique, a Reactives Number (RJ is
defined which represents the percentage of the total reflected light below a certain grey scale
level.
The Shoal Creek histogram shows a very broad and indistinct peak compared with the other coals
and demonstrates that the selection of the cut-off point between reactives and unreactives can be
very critical for some coals. The profile may also be indicative that the coal is a blend, but this is
not as clear as in the Leasingthorne histogram in which two distinct peaks occur in the 60-160 R
range. N
The main objectives of this work were to (1) analyse chars found in fly-ash samples obtained from
-------
plant trials to determine the nature of the char, (2) investigate correlations between the total
reflectance (represented as a standardised grey-scale histogram) of a coal and its overall burnout
in the CTF, and (3) determine correlations between char morphology and intrinsic reactivity of
chars from the CTF and burnout. Comparisons were also made between the char types observed
in the Drop-Tube Furnace (DTF), the CTF and from the plant trials.
The first finding was that the intrinsic reactivity of chars decreases significantly throughout the
combustion process and that, irrespective of the original coal, chars found in fly-ash have similar
intrinsic reactivity values as determined by the TGA/DTA method. This appears to concur with the
observations of other workers and indicates that the reactivity of chars is reduced the longer they
remain in the furnace.
The percentage of material with a value of less than 190 in the grey-scale histogram (the 190
Reactives Number, RN190) correlates well with the percentage combustibles remaining in the char
sample taken furthest from the burner in the CTF An R-square factor of 0.93, showing an
excellent linear correlation, is found for the results at 15% overfire air and 1% O2 excess air (Figure
22), conditions which were found to be fairly representative of plant and at which there was a high
value of residual combustibles. Correlation of percentage combustibles remaining with other more
commonly used factors such as fuel ratio, rank and reactive macerals is in all cases not as good
33 for RN190'
In addition to determining the intrinsic reactivity of the chars, their morphology was also assessed
using an image analysis technique which calculates the pixel point thickness1 of all the char
particles present in an image. Figure 23 shows representations of the type of chars that can be
produced during the combustion process; given a similar intrinsic reactivity, those with an open,
lacy structure would be expected to burn out to completion more readily that the chars of more
solid appearance.
Good correlation was found between the measured morphology of char particles sampled from the
port nearest the burner with both 190 reactives and percentage combustibles remaining. There is
also a similar but not so strong correlation with intrinsic reactivity. The decrease in intrinsic
reactivity as combustion proceeds is confirmed for chars produced in the CTF. Data linked to
morphological changes as combustion proceeds were limited and further work would be required
in this area to produce samples from which sufficient carbon can be isolated for a meaningful
analysis. Char samples produced in the DTF are similar to those found near the burner in the
CTF.
General Discussion
Although each phase and scale of this project has produced a valuable data set which, on their
own, can be used to understand and compare the different behaviours and characteristics of a
range of coals, the integrated nature of the work has enabled a certain amount of calibration
across the scales and approaches to be made in order to establish the credibility of certain
methodologies for NOx and unbumt carbon prediction. There are still gaps in the data, due to
limitations caused by the scope of the study and the choice of the coal suite, but major advances
have been made in defining the usefulness of both established and new laboratory scale methods
for predicting full scale plant behaviour.
Although the combustion arrangement in the 1 MWt pilot scale combustor did not simulate the
tangential firing of the 500 MWe boiler, it was possible to reproduce some of the subtle differences
seen between the NOx and unburnt carbon figures produced on what turned out to be three coals
of quite similar behaviour on the plant. It vindicates the rationale that residence time scaling is
important when investigating the effect of coal quality on the products of combustion, but the
findings also emphasise how difficult it is to simulate the behaviour of a large multi-burner furnace
with a single burner system which is, by comparison, a very well stirred reactor. This is borne out
by much better burnout being achieved on the pilot scale than on plant even when the NOx
emissions are very similar and the coal grind is coarse, and suggests that the pilot scale could
-------
provide more reliable absolute predictions for NOx than for unburned carbon.
It should be stressed, however, that the calibration obtained (i.e. pilot scale operating conditions
which produce good similarity to plant NOx and unburnt carbon results) only holds for the plant in
question and that predictions on the basis of pilot scale data for other plant could only be made
after a similar calibration exercise. Similarly, if the combustion equipment or operational regime on
the calibrated plant were to change significantly, then a re-calibration would be needed.
The expanded range of coals tested on the CTF (and subsequently on the other small scale
facilities) did exhibit some differences both in NOx and, to a lesser extent, unburnt carbon. For
NOx, a rough correlation was found with the product of coal nitrogen content and fuel ratio
especially when the latter incorporated high heating rate volatile matter content, but for unburnt
carbon there was no similar relationship. The dominant mechanism for NOx formation in deeply
staged combustion is understood to be via oxidation of the residual nitrogen in the char and
would suggest that, because of the difference in the rate and quantity of volatile (and hence
nitrogen) release at the high heating rates experienced in a furnace, the incorporation of proximate
volatile matter in the fuel ratio will result in some inaccuracy.
Coals expected to give high unburnt carbon figures, i.e. the higher rank coals such as ATC and
Shoal Creek, did, but there was no pattern that could substantiate anything but the weakest
correlation.
The DTP results indicated an approximately constant conversion rate of 45% of the char bound
nitrogen to NO for the suite of coals tested. It was also observed that the actual NO measured
during the char refiring stage on the DTF correlated reasonably well with NO measured on the
CTF, even though the partitioning of nitrogen between the volatiles and the char varied by a factor
of two. This tends to confirm that volatile nitrogen does indeed play only a very minor role in the
formation of NOx in staged combustion systems. The conclusion is that once calibrated against a
combustion system (either full or small scale) to establish a coefficient linking the measured NO on
that system with the char bound nitrogen a reasonably accurate prediction for the NOx formation
propensity of different coals will be possible.
The situation with respect to LOI prediction from DTF measurements is also quite encouraging in
that a reasonable correlation has been observed with final unburnt carbon figures from the CTF.
However, the limited range of coal qualities tested makes the correlation too reliant on a single
coal with poor burnout characteristics and more data are required.
One of the most encouraging findings of the current work has been the good correlation found
between the RN190 number generated on the Nottingham University automated image analysis
technique and unburnt carbon. It would appear that the level of unreactive material in the coal
that this index represents also correlates reasonably well with the morphology of the chars,
thereby substantiating the observation that the intrinsic reactivity of fully developed chars is
constant, i.e. coal quality independent.
Conclusions
1. The propensity to form NOx in a utility boiler is dominated by plant factors, whereas the levels
of unburnt carbon are controlled by coal quality considerations.
2. For the prediction of NOx emission levels from different coals, the use of a calibrated pilot scale
combustor is the preferred option currently. However, the residual nitrogen in chars formed on a
Drop Tube Furnace which must also be calibrated with the combustion system concerned shows
some promise. There is a need for the correlations already established to be tested on a wider
range of coals.
3. For the prediction of unburnt carbon there is less need to resort to expensive pilot scale
testing. The Drop Tube Furnace has been shown to provide a reasonable indication of burnout
-------
propensity, but automated image analysis of the pattern of reflectance from coal samples is
beginning to show promise as a simpler method of predicting coal reactivity without the need to
heat the coal samples. Again, the range of coals needs to be expanded before confidence in the
concept can be further substantiated.
Acknowledgments
The assistance of the Combustion Test Facility operations team and staff at both Kingsnorth Power
Station and in the Chemical Engineering Department at Nottingham University is gratefully
acknowledged, as is the financial support from EPRI. The paper is published by permission of
PowerGen PLC.
References
1. W H Gibb, A R Jones, R M A Irons, J W Stallings and A K Mehta, The Effect of Coal
Quality on NOx Formation and Unburnt Carbon in PF Fired Boilers: Preliminary Results
from 1MW Combustion Test Facility', presented to the 4th International EPRI Conference
on The Effects of Coal Quality on Power Plants, Charleston, South Carolina (August 1994)
2. L S Monroe, R J Clarkson and J W Stallings, 'Predictions of Full-Scale NOx Emissions and
LOI from Coals Blends in Pilot Combustion Experiments', presented to the 4th International
EPRI Conference on The Effects of Coal Quality on Power Plants, Charleston, South
Carolina (August 1994)
3. J W Allen, "Advances in Comer Firing', presented to the Institution of Chemical Engineers
Conference on International Power Generation and the Environment, London, England
(June 1994)
4. E Lester, The Characterisation of Coals for Combustion', Ph.D Thesis, Nottingham
University, England (1994)
-------
Table 1
Composite PF Sample Analyses
Coal Date
Moisture, % Total
Ash, %ar
%dry
Volatile Matter
%ar
CV, kJ/kg ar Gross
Nett
Sulphur, %ar
Chlorine, %ar
Carbon, %DAF
Hydrogen, %DAF
Nitrogen, %DAF
Oxygen, %DAF
(Diff)
Volatile Matter,
%DAF
CV, kJ/kg DAF
Gross
Fuel Ratio
Si02
AI203
Fe203
CaO
MgO
K20
Na20
Ti02
BaO
Mn304
P205
Ashland
6/10/93
2.0
13.2
13.5
31.4
29,200
28,165
0.92
0.15
83.66
5.27
1.71
8.10
37.0
34,434
1.70
58.74
27.41
7.39
1.70
0.96
2.29
1.51
0.13
0.19
7/10/93
2.1
13.6
13.9
31.3
29,100
28,178
0.86
0.10
84.60
4.65
1.61
8.00
37.1
34,520
1.69
58.99
27.93
6.55
1.35
0.95
2.48
0.17
1.57
0.12
0.17
El Cerrejon
12/10/93
2.6
10.4
10.7
33.8
29,510
28,367
0.81
0.05
82.62
5.68
1.68
9.03
38.9
33,920
1.57
64.30
20.92
7.63
2.21
1.44
2.05
0.36
1.09
0.11
0.05
0.14
25/10/93
2.8
10.3
10.6
35.1
29,180
28,065
0.81
0.04
80.97
5.51
1.69
10.85
40.4
33,579
1.48
62.03
19.38
10.01
3.25
1.99
2.03
0.37
0.95
0.11
0.06
0.15
Tyne Blend
20/10/93
2.4
14.5
14.9
30.1
28,630
27,580
1.50
0.31
83.60
5.39
1.81
7.02
36.2
34,452
1.76
50.30
27.83
11.25
4.24
1.74
1.92
1.57
1.15
0.12
0.08
0.24
21/10/93
2.1
14.3
14.6
29.9
28,850
27,845
1.46
0.31
84.65
5.15
1.78
6.30
35.8
34,510
1.80
51.17
27.93
11.28
3.62
1.43
1.98
1.42
1.16
0.16
0.07
0.21
El Cerrejon/
22/1 1/93
2.9
13.3
13.7
31.4
28,630
27,570
1.05
0.14
82.28
5.35
1.69
9.26
37.5
34,165
1.67
57.71
25.04
9.05
2.76
1.42
2.03
0.84
1.16
0.17
0.06
0.25
Tyne
23/11/93
2.8
12.6
13.0
31.7
28,950
27,876
1,03
0.14
82.69
5.40
1.70
8.83
37.5
34,220
1.67
57.48
24.82
9.56
2.90
1.46
2.00
0.63
1.15
0.17
0.07
0.26
-------
Table 2
Composition of Test Coals
Coal
Moisture, % Total
Ash, %ar
Volatile Matter, %ar
CV, kJ/kg ar Gross
Nett
Sulphur, %ar
Chlorine, %ar
Carbon, %DAF
Hydrogen, %DAF
Nitrogen, %DAF
Oxygen, %DAF
(Diff)
Volatile Matter,
%DAF
Fuel Ratio
CV, kJ/kg DAF
Gross
Ash Comp
Si02
AI203
Fe203
CaO
MgO
K20
Na20
Ti02
BaO
Mn304
P205
Tyne Blend
3.4
12.1
31.2
28,960
27,908
1.18
0.27
83.70
5.21
1.76
7.61
36.9
1.71
34,272
51.33
27.69
10.77
4.14
1.55
2.07
1.24
1.21
0.13
0.08
0.21
ATC
3.2
13.4
28.5
28,100
27,124
0.68
0.07
83.95
4.86
2.01
8.28
34.2
1.93
33,693
50.67
30.82
5.25
7.94
2.06
1.02
0.39
1.85
0.30
0.06
1.56
Drayton
0.1
14.2
33.9
29,300
28,259
0.76
0.04
83.08
5.49
1.76
8.74
39.6
1.53
34,189
68.71
19.02
6.34
2.53
1.01
0.78
0.40
1.21
0.14
0.05
0.90
Cerrejon
3.4
10.0
33.9
29,670
28,539
1.06
0.14
83.05
5.54
1.70
8.32
39.1
1.55
34,261
60.61
22.16
9.52
1.91
1.42
2.37
0.94
1.07
0.11
0.05
0.19
Ashland
2.4
12.7
32.1
29,460
28,387
0.82
0.12
85.49
5.43
1.62
6.35
37.8
1.64
34,700
61.36
27.10
4.93
1.09
0.94
2.67
0.26
1.71
0.18
0.07
0.22
Shoal
Creek
1.7
10.0
25.6
31,550
30,523
0.86
0.06
88.02
5.10
1.90
3.94
29.0
2.45
35,730
51.81
29.32
9.71
2.59
1.39
2.32
1.18
1.66
0.30
0.05
0.69
North River
2.4
10.8
34.8
29,640
28,583
2.09
0.01
81.87
5.25
1.70
8.76
40.1
1.49
34,147
44.10
26.35
18.05
5.88
1.48
2.15
0.62
1.37
0.22
0.05
0.74
Leasing-
thorne
2.7
7.4
29.5
30,960
29,896
1.30
0.04
85.36
5.11
1.71
6.33
32.8
2.05
34,438
50.94
27.99
13.97
2.29
1.12
2.17
0.31
1.21
0.18
0.11
0.24
Size, %-75
66.1
87.8
67.0
66.3
71.6
73.2
75.8
65.1
-------
Table 3
Summary of NO Measurement results, vpm Dry (Corrected to 3% Oxygen Dry)
OFA 02
% % Dry
0 1.0
3.0
4.5
SLOPE
15 1.0
3.0
4.5
SLOPE
25 1.0
3.0
4.5
SLOPE
Tyne
(TY)
220
325
410
54
165
250
305
40
ATC
(AT)
260
355
420
46
205
290
360
42
200
255
300
29
Drayton
(DR)
195
305
380
53
205
250
280
22
235
270
23
El
Cerrejon
(EC)
225
315
370
41
180
235
280
30
165
215
255
27
Ashland
(AS)
215
300
360
41
180
230
265
24
175
215
245
20
Shoal
Creek
(SO
240
340
410
48
220
285
330
31
195
260
315
34
North
River
(NR)
205
290
360
46
160
235
290
37
155
210
245
26
Leasing
-thorne
(LS)
230
360
440
60
175
255
320
41
160
225
285
36
Mean
224
324
394
49
186
254
304
33
175
231
274
28
Min
190
285
360
41
170
235
265
22
165
210
245
20
Max
260
360
440
60
220
290
360
42
200
260
315
36
Note: SLOPE is in vpm per percent oxygen derived from best fit straight line
Table 4
Summary of Unburnt Carbon Results
OFA % 02 % Dry
Loss-on-ignition at 800°C,
0 1.0
3.0
4.5
15 1.0
3.0
4.5
25 1.0
3.0
4.5
Ash
(%Dry)
TYNE
% Dry
5.2
1.4
0.8
3.5
2.1
1.5
3.3
2.0
12.5
ATC
10.5
3.4
2.9
14.6
6.1
3.7
8.5
4.5
3.7
13.8
DRAYTON
7.0
2.6
1.4
4.0
3.4
3.1
3.7
3.0
14.2
EL CERREJ
7.0
2.1
1.7
7.5
3.0
2.3
6.7
4.0
2.5
10.4
ASHLAND
4.9
1.6
0.8
5.0
2.0
1.6
5.1
1.8
1.7
13.0
SHOAL
CREEK
9.7
3.4
1.5
10.2
5.4
2.6
10.6
3.4
2.2
10.2
NORTH
RIVER
8.0
2.8
1.9
4.0
2.6
1.8
5.3
2.5
2.5
1 1.1
LEASING-
THORNE
8.4
2.4
1.3
11.3
3.4
2.3
13.5
2.9
1.7
7.6
Carbon Loss, % Dry
0 1.0
3.0
4.5
15 1.0
3.0
4.5
25 1.0
3.0
4.5
0.65
0.18
0.10
0.44
0.26
0.19
0.41
0.25
1.45
0.47
0.40
2.01
0.84
0.51
1.17
0.62
0.51
0.99
0.37
0.20
0.57
0.48
0.44
0.53
0.43
0.73
0.22
0.18
0.78
0.31
0.24
0.70
0.42
0.26
0.64
0.21
0.10
0.65
0.26
0.21
0.66
0.23
0.22
0.99
0.35
0.15
1.04
0.55
0.27
1.08
0.35
0.22
0.89
0.31
0.21
0.44
0.29
0.20
0.59
0.28
0.28
0.64
0.18
0.10
0.86
0.26
0.17
1.03
0.22
0.13
-------
Table 5
A Summary of the Devolatilisation Data from the Drop Tube Furnace EPRI Contract Work
Devolitisation at 1300°C, 200 ms & 1 % O,
Coal
ATC
Tyne Blend
El Cerrejon
Leasingthorne
Ashland
Shoal Creek
North River
Drayton
Feed Size
/urn
53-75
106-125
53-75
106-125
53-75
106-125
53-75
106-125
53-75
106-125
53-75
106-125
53-75
106-125
53-75
106-125
% Nitrogen
Coal
1.76
1.81
1.62
1.65
1.54
1.59
1.64
1.73
1.43
1.45
1.73
1.65
1.68
1.71
1.47
1.61
Content
Char
1.98
2.02
2.15
1.97
1.91
1.92
2.12
2.07
1.52
1.74
1.82
1.74
1.91
2.37
1.62
1.74
Coal LOI
Coal
88.25
89.02
91.25
92.05
93.20
95.47
94.30
95.87
88.25
91.64
90.56
89.83
89.05
90.53
84.85
89.03
or C + H + N
Char
78.43
80.63
82.13
81.66
82.68
90.88
88.52
92.13
70.03
79.47
82.43
82.70
73.80
77.77
68.98
75.07
% Volatile
Matter
(Dry Basis)
Coal
29.98
31.34
34.21
34.36
37.11
38.43
30.79
32.83
34.00
36.59
26.54
26.62
36.31
36.77
33.20
35.37
%VMhhr
(Dry Basis)
Char
46
43
51
57
61
50
50
48
61
59
46
41
58
57
51
56
"R" Factor
1.52
1.38
1.49
1.65
1.64
1.31
1.64
1.45
1.79
1.62
1.74
1.55
1.60
1.56
1.54
1.58
Combustibles
Loss
0.52
0.49
0.56
0.62
0.65
0.53
0.53
0.50
0.69
0.65
0.51
0.46
0.65
0.63
0.60
0.63
Nitrogen
Loss
0.39
0.37
0.35
0.48
0.51
0.40
0.36
0.37
0.58
0.51
0.43
0.38
0.52
0.41
0.46
0.52
gN/100g Coal
In Volatiles
0.68
0.66
0.57
0.80
0.79
0.63
0.58
0.64
0.84
0.74
0.75
0.63
0.88
0.70
0.68
0.84
In Char
1.08
1.15
1.05
0.85
0.75
0.95
1.05
1.09
0.59
0.71
0.98
1.02
0.80
1.01
0.79
0.77
-------
Table 6
A Summary of the Char Retiring Data for Drop Tube Furnace EPRI Contract Work
Chars refired at 1300°C, 5% O2
Coal
ATC
Tyne Blend
El Cerrejon
Leasingthome
Ashland
Shoal Creek
North River
Drayton
Feed Size
urn
53-75
106-125
53-75
106-125
53-75
106-125
53-75
1O6-125
53-75
106-125
53-75
106-125
53-75
106-125
53-75
106-125
Res Time
ms
200
400
600
200
400
600
200
400
600
200
400
600
200
400
600
200
400
600
200
4OO
600
200
400
600
200
400
600
200
400
600
200
400
600
200
400
600
200
400
600
200
400
600
200
20O
400
400
400
400
600
600
200
200
400
400
600
600
Average
Measured
NO
mg/g coal
4.31
7.16
8.63
4.19
8.15
8 29
7.30
8 48
8.28
4.71
6.79
7.06
5.79
7.54
7.87
5.63
10.23
10.18
6.19
9.55
10.63
6.02
8.74
10 13
4.86
5.54
5.91
5.24
6.18
6.28
9.38
10.09
10.88
7.88
9.34
9.31
5.14
6.66
7.30
5.22
6.37
7.60
5.04
5.12
7.50
7.67
7.82
8.24
8.28
8.47
3 37
3.57
6.73
6 99
8.16
8.31
Char
Conversion
0.40
0.69
0.84
0.32
0.61
0.75
0.78
0.97
0.98
0.49
0.84
0.91
0.80
0.96
0.98
0.60
0.88
0.94
0.69
0.92
0.97
0.44
0.76
0.89
0.55
0.94
0.98
0.63
0.91
0.95
0.78
0.94
0.97
0.56
0.82
0.86
0 81
0.94
0.96
0.67
0.86
0.94
0.65
0.62
0.87
0.83
0.85
0.85
0.94
0.95
0.46
0.40
0.75
0.74
0.85
0.87
Combustibles
Loss
0.71
0.85
0.92
0.65
0.80
0.87
0.90
0.99
0.99
0.80
0.94
0.97
0.93
0.99
0.99
0.81
0.95
0.97
0.86
0.96
0.99
0.72
0.88
0.94
0.86
0.98
0 99
0.87
0.97
0.98
0.89
0.97
0.98
0.76
0.90
0.92
0.94
0.98
0.99
0.88
0.95
0.98
0.86
0.85
0.95
0.93
0.94
0.94
0.98
0.98
0.80
0.78
0.91
0.90
0.95
0.95
Nitrogen
Loss
0.67
0.80
0.91
0.60
0.76
0.88
0.87
0.97
0.97
0.75
0.92
0.95
0.90
0.97
0.99
0.76
0.93
0.97
0.83
0.95
0.98
0.66
0.85
0.92
0.75
0.94
0.96
0.82
0.95
0.97
0.89
0.95
0.98
0.72
0.91
0.92
0.92
0.96
0.97
0.82
0.97
0.81
0.80
0.91
0.89
0.91
0.91
0.95
0.96
0.75
0.73
0.89
0.88
0.93
0.94
NO Produced Average NO
if 100% Char Produced if
Conversion 1OO% Char
mg/g coal Conversion
mg/g coal
10.70 10.44
10.41
10.21
13.11 12.49
13.34
11.02
9.37 8.85
8.75
8.41
9.70 8.51
8.07
7.75
7.27 7.71
7.88
7.99
9.32 10.55
11.56
10.77
8.94 10.10
10.41
10.94
13.63 12.17
11.49
11.39
8.83 6.92
5.89
6.03
8.31 7.24
6.78
6.63
11.98 11.32
10.73
11.25
14.10 12.10
11.40
10.81
6.33 7.00
7.09
7.57
7.85 7.80
7.44
8.12
7.74 8.82
8.31
8.63
9.21
9.22
9.70
8.84
8.93
7.26 8.97
8.93
9.01
9.44
9.56
9.61
Average NO
Produced if
Char
Conversion
100%
(Normalised
for Net CV of
29000 kj/kg)
mg/g coal
11.17
13.35
9 19
8 84
7.84
10.72
9.80
11.81
7.07
7.40
10.76
11.50
7.10
7.92
9.05
9.20
-------
Overfire Air
Elevation
Mill Elevation A
Mill Elevation B
Mill Elevation C
Mill Elevation 0
Mill Elevation E
Overt-" A,r Nozzle
Oil Burner
(Bank 2)
Refractory Plug
Top Offset
Secondary Air Nozzle
PF Burner
Large Offset
Secondary Air Nozzle
Oil Burner
(Banks 3 and 3A)
PF Burner
La/ge Offset
Secondary Air Nozzle
Oil Burner
(Bank 4)
PF Burner
Large Offset
Secondary Air Nozzle
Oil Burner
(Banks 5 and SA)
PF Burner
Small Offset
Secondary Air Nozzle
OH Burner
(Bank 6)
PF Burner
Large Offset
Secondary Air Nozzle
Flue Gaj
Recircuiating Nozzle
Figure 1
Kingsnorth Burner Arrangement
Figure 2 KINGSNORTH PS: the Relationship between Carbon-ln-Ash and Excess O2
25.00
20.00 -
15.00
10-°°
g
X
| ASHLAND
X EL CCHREJON
1 1 TYNE dLEND
EL CEHF1EJON
ASHLANDfrVNE BLEND
4.60
5.00
5.60 6.00 6.50 7.00
Excess Oxygen at the ID Fan Discharge, % dry
7.60
Figure 3 KINGSNORTH PS: the Relationship between Carbon Loss and Excess Oxygen
2.50
2.00
1.50
_ __
u-°1'
| ASHLAND
X f 1 CEnnEJON
[ 1 1VNF BtEND
*
ASHLANOaVNE BLEND
4.50
5.00
r~
8.00
5.50 8.00 6.60 7.00
Excess Oxygen at the ID Fan Discharge, % dry
7.50
-------
Figure 4 KINGSNOHTH PS: the Relationship between NO Emissions and Excess O2
A MILL IN SERVICE
550
500 -
450-
400
350 -
300 -
250
200
4.50
O
OVERFIRE AIR
1O% 18%
O -t- ASHLAND
* X EL CERREJON
• D TYNE BLEND
5.00
5.50
6.00
6.50
7.00
7.50
Excess Oxygen at the ID Fan Discharge, % dry
Figure 5 KINGSNORTH PS: the Relationship between NO Emissions and Excess O2
A MILL OUT OF SERVICE
s"
eo
o
TJ
I
TJ
09
c
o
UJ
o
500 -
450 -
400 -
350 -
300 -
250
200
ASHLAND
X EL CERREJON
Q TYNE BLEND
(J
TYNE BLEND
\
ASHLAND/EL CERREJON
4.50
5.00
5.50
6.00 6.50 7.00
Excess Oxygen at the ID Fan Discharge, % dry
7.50
Figure 6 KINGSNORTH PS NO Emissions: Comparison with Result for a SJVfrican Coal
A MILL IN SERVICE
O
s
S
•a
•o
I
i
o
1
1
O
540
500
460 -
420
380 -
340 -
300
260
4.00
4.50
5.00
5.50
6.00
6.50
Cvrmrtm f"Unj«*ut *b« 4k..
-------
Figure 7 COMBUSTION TEST FACILITY
CONTROL
HOOM
GAS ANALYSIS
HOOM
COOUNOAJHFAN
SECONDARY
AIR HEATER
UQHT FUEL ^
Ff FEEDER °11 PUMP HEAVY FUEL OIL
PUMP AND HEATER
AUXILIARY STEAM
BOILER
-------
FIGURE 6 NO CONCENTRATION V. EXCESS OXYGEN
0V, OVERF1RE AIR
450 -
160
0.0
1.0
2.0
3.0
Excess Oxygen, % dry
-+- TY
-X- AT
-0 DR
-& EC
• AS
-+• sc
5.0 , "x"Nn
I
LS
FIGURE 9 NO CONCENTRATION V. EXCESS OXYGEN
15% OVERFIRE AIR
450
I TY
Excess Oxygen, % dry
Figure 10 The Relationship Between NO Emissions and Fuel Ratio (VM) x Fuel Nitrogen
360
340
320
300
a c
[]
a
a
2.50
3.00 3.50 4.00 4.50 5.00
Fuel Ratio (VM) x Coal Nitrogen (DAF)
* 0%OFA
D 15%OFA
• 25%OFA
f 450
S
£ 400
a
1 350 -
» 300
T3
£ 250
200
FIGURE 11 NO Emissions: Comparison ol Plant and CTF Results
-m- AS(PLANT)
160
0.00 1.00 2.00 3.00 4.00 5.00
Excess Oxygen, % dry
-e- AS(CTF) 0% OFA
-X- AS(CTF) 15% OFA
AT(PLANT)
AT(CTF) 0% OFA
-------
1=-^— Water in
Feeder Probe
Water out
Main Gas Inlet
Preheater Element
^Thermal Lagging
Main Heater
Trim Heater
Ceramic Insulating
Sleeves
Collector Probe
i i
f= -^- Water out / Waste Gas Outlet
*— Water in
Jl L
\
OR
Cyclone Collection
Device
I HID
Millipore Filter Assembly
(for residue collection)
Figure 12
Schematic and Photographic Representation of the Drop Tube Furnace
-------
Figure 13
Comparison Betoeen Fractional Combustibles Loss and Fractional Nitrogen Loss. Fractional Nitrogen Loss as a Function ol Residence Time
(Chare produced 9 1300oC, 1%O2 & 200ms, refired at 1300oC. 5%O2 & 200,400 & 600ms) 6 Chars Retired at HflOnr nnrl K%O9
H n on -'-
C :
o u.ou
fi ;
1 ^
ra n on :
u. ;
000
0.
„,-
00 0.
.^""
10 0.
flfff'
20 0.
,-'"
30 0
„''
-
40 0.
L
*.
60 0
u •
•r
eo o
^
70 0.
, jaji
DTI
BO 0.
0*
eo 1.
8
•3 0.90
1 0.80-
i
1 0.70
n An
00 (
+ *
X
+ a
X
a
n
1 1 ' ' ' | T -T 1 , 1 , T
) 200 400 600 8C
Fractional Combustibles Loss
Coal Chare O Retired Chare — • x-y
+ El Cerrejon
Residence Time /ms
X Leaslngthome
D ATC
Figure 15 DTP - NO Produced as a Function of Predicted NO.
(OevotaHUied Chare Retired at 1300oC, 5UO2 & 200,400 & 600ms)
2.00
1.50
1.00
0.50
0.00
Figure 16 The Relationship Between NO Emissions and Fuel Ratio (VMhhr) x Fuel Nltroget
4 1
4*
$?$
}&&
M
(
ft;
i- x
3
LJIIH Ul CHJSI Til OJfv
+ ATC E 320
? :
X Tuna Rlnnrl o
-------
Figure 17 CTF NOx as a Function of Char Nitrogen / 100g Coal
-------
Figure 20
Grey Scale Histograms
Tyne Blend
Grey Scale
Shoal Creek
.-t-*-t-r-t-.-tNtNrM
Grey Scale
05
Ashland
Grey Scale
North River
i- (N (N CN fN
Grayscale
Figure 21
Grey Scale Histograms
Leasingthorne
Drayton
Grey Scale
25
. 2
- 1.5
. 1
05
El Cerrejon
3 2 S |
Grey Scale
ATC
Grey Scale
-------
Figure 22 plot Of % Combustibles Remaining (CTF Data) vs 190 Reactivity
100
95
en
i»
& 90
u
80
75
0 0.5 1 1.5 2 2.5 3
% Combustibles Remaining (CTF Data)
Reactivity is % less than 190 on coal grey-scale
R-square value is 0.93
Figure 23
-------
COMPARISON OF PILOT-SCALE FURNACE EXPERIMENTS
AND PREDICTIONS
TO FULL-SCALE BOILER PERFORMANCE OF COMPLIANCE COALS
L. S. Monroe
Southern Research Institute
2000 Ninth Avenue South
Birmingham, Alabama 35205
R. J. Clarkson
Southern Company Services, Inc.
Post Office Box 2625
Birmingham, Alabama 35202
J. W. Stallings
Electric Power Research Institute
3412 Hillview Avenue
Palo Alto, California 94303
Abstract
A series of compliance coals have been fired in the Southern Company Services and
Southern Research Institute pilot-scale Combustion Research Facility, with the goal of
predicting NO, emissions, unburned carbon levels, and other operating parameters. The
research was financed by a tailored collaboration between Alabama Power and the
Electric Power Research Institute. The coals included a South American coal, a Powder
River Basin coal, and several local Alabama bituminous coals. They were fired in
conventional and low NOX firing modes, where some of the coals were fired in a
tangential-fired simulation, others as wall-fired, and some in both types of firing. Two of
Alabama Power's boilers were used as the plants simulated in the pilot-scale experiments,
along with a Gulf Power boiler. The results of the NOX emissions and unburned carbon
from the pilot furnace testing are presented and compared to the limited full-scale boiler
data available on these coals.
-------
Introduction
Most utilities are pursuing a strategy of fuel switching to low sulfur coal, in order to meet
the sulfur dioxide (SO2) emission limits of the first phase of the 1990 Clean Air Act
Amendments. However, since this legislation also requires reductions in the level of
emissions of nitrogen oxides (NOJ, utilities must predict the ability of alternative low-
sulfur coals to meet emission limits on NO^, along with the ease of boiler operation of
these low sulfur coals. Typical strategies employed to meet NOX emission limitations,
installation of low NO, burners, are usually designed for a particular coal or coal type, and
may not perform well on other fuels.
Alabama Power, in a tailored-collaboration with the Electric Power Research Institute, has
funded a research program to predict the performance of certain low sulfur coals in their
full-scale plants with low NOX combustion modifications. This work is focused mainly on
NOX emissions and unburned carbon. Since most of the full-scale units that will burn low
sulfur coal did not have low NOX firing modifications at the time this work was started, it
was not possible to test burn these coals in the actual boilers. The predictive ability of the
pilot-scale facility for the important issues of NOX emissions and unburned carbon was to
be used to aid in the coal acquisition process. Also, any insight from the pilot-scale testing
on boiler operations affected by coal quality, such as milling capacity, slagging, fouling, and
particulate control, would also assist in coal procurement.
This paper reports some of the results of this research program. The Alabama Power
plants simulated in this study are Unit 5 of Plant Gaston (880 MW CE tangentially-fired,
retrofitted with LNCFS II) and Unit 3 of Plant Miller (660 MW B&W w/ dual-register
burners), along with Plant Crist Unit 7 (500 MW Foster-Wheeler w/ F-W dual-register
burners) belonging to Gulf Power. Baseline unit operating data have been collected at
each of the three plants and it was used to help in the setup of the simulations. In the
case of Gaston Unit 5, the field data collection effort of the DOE and EPRI Coal Quality
Expert (CQE) during September 1991 was available to help in evaluating the pilot-scale
simulations.
Description of the Pilot-Scale Research Facility
The Southern Company and Southern Research Institute Combustion Research Facility is
located on the Birmingham, Alabama, campus of Southern Research Institute (SRI). The
facility was designed and constructed by SRI under contract to Southern Company Services
(SCS). SCS and SRI jointly own the facility, and SRI operates the facility under contract
to SCS. However, SRI can also perform confidential research for outside parties,
including other utilities, with the payment of a facility usage fee to SCS. The facility is
designed for up to six million Btu per hour firing on natural gas or coal, which is
equivalent to 1.75 MW thermal or about 0.6 MW electric.
The design of the facility was carefully chosen to provide a close simulation of the physical
processes that occur in a full-scale utility boiler. The facility, shown in Figures 1 and 2
consists of a coal crushing and milling area, a coal feeding system, a vertical refractory-
-------
lined furnace, a single up-fired burner, a horizontal convective section pass with air-cooled
tube banks, a series of heat exchangers, an electrostatic precipitator, a pulse jet baghouse,
and a packed column scrubber.
Coal Preparation
The coal handling part of the facility is shown in Figure 1. The coal preparation area
includes covered on-site storage bins, a rotary drum coal crusher, a CE Raymond bowl
mill, and pulverized coal storage. Eight storage bins, each capable of holding up to 25
tons of coal, are located adjacent to the crusher. As-mined coal is loaded into the crusher
with a small front-end loader and is crushed to a size of minus 3/8 inch. From the crusher,
the coal is pneumatically transported to a storage bin located above the coal mill. The
coal mill is a refurbished 1937-vintage CE bowl mill, which has a rated capacity of 2 tons
per hour. This type of mill should give representative milling simulations of the different
air-swept table and roller mills normally used in power plant service. The mill air is
preheated with a dedicated natural gas burner, which also helps inert the mill. The
pulverized coal is captured in a pulse-jet baghouse and transported to an inerted
pulverized fuel storage hopper. In such small-scale pulverized coal firing, it is normal
practice to use indirect firing to help eliminate coal surges which can cause flame stability
problems. The mill circuit can also be used to perform stand-alone milling experiments.
Coal Feeding
The coal feeding system is designed to deliver a constant feed of pulverized coal to the
burner. The storage hopper is equipped with an orbital-motion live bottom to prevent
bridging or rat-holing of the fuel. An Acrison loss-in-weight auger-type feeder weighs a
controlled amount of fuel into the primary air through a rotary airlock valve. This
gravimetric feeder is suspended on a weighing mechanism, and the feed rate is controlled
by varying the auger speed to maintain the desired fuel feed rate.
Radiant Furnace
The furnace is a vertical, up-fired 28 foot high cylinder, with an inner diameter of 3.5 feet.
(See Figure 2). The body of the furnace is built of seven 4 foot tall sections, each being a
water-cooled jacket with 4 inches of refractory lining cast on the fireside. This refractory
lining is used to limit the heat extraction and to ensure the proper simulation of the
radiation environment found inside full-scale furnaces. Small furnaces have much higher
ratios of inside surface area to total volume than full-scale furnaces, and the flame would
be quenched if the entire interior was lined with heat exchange surface. The furnace
diameter and height were chosen to best match the velocities and residence times found in
full-scale units, with a velocity of 20 feet per second and a residence time to the furnace
exit of 1.2 seconds chosen, at full firing rate.
-------
Burner Assembly
The burner is mounted coaxially on the bottom of the furnace and is up-fired using natural
gas, pulverized coal, or any combination of the two. It is equipped with a flow control
system for secondary air flow and a set of registers which impart swirl to the secondary air,
separate from the flow control. The secondary air and the primary air-coal mixture enter
the furnace through a refractory quarl with a 25° half angle. Two cleanout ports are
provided in this section, to allow bottom ash to be periodically removed from the furnace.
A closed-circuit television camera with a control-room monitor allows constant monitoring
of the view of the flame from the top of the furnace. A low NO^ firing system, consisting
of the a generic dual-register burner and an overfire air system, can be installed to
simulate several combinations of low NOX firing.
Convective Sections
The combustion gases exit the vertical furnace through a horizontal convection pass which
is designed to remove a substantial part of the heat from the flue gases, the extraction of
heat is designed in order to simulate the time-temperature profile found in a utility boiler.
A series of three air-cooled tube banks are installed in the convective pass, and the air-
cooling is used to control either the temperature profile of the flue gases or the tube metal
surface temperatures for fouling/ash deposition studies. A cross-flow tubular air preheater
follows the convective tube banks and is used to preheat the primary and secondary air.
Finally, two air-to-flue-gas recuperators are used to cool the flue gas down to a nominal
300 °F before the flue gas enters the pollution control devices, which are described in a
following section.
Computer Data Acquisition and Control System
The facility is controlled and monitored by a combined process control and data
acquisition computer, a Yokogawa Yewpack system. This system performs all the process
control and allows complex feed-forward and calculated variable control. This computer
control also performs the safety monitoring needed for safe operation of combustion
equipment, including flame scanning and interlocks. The Yokogawa computer also has the
software to allow automatic startup and shutdown of the entire facility. Process data
acquisition and storage is accomplished by sending data from the Yokogawa computer to
an IBM-compatible computer, which stores the process data and allows on-line and
historical data examination.
GEM System
A complete extractive continuous emissions monitoring (CEM) system is installed in the
facility, and it is also interfaced to the computer control system. A set of gas analyzers
which analyze the flue gas for concentrations of O^ NO,, SO^ CO^ and CO, receives the
dry flue gas sampled from a set of three extractive lines. Flue gas is sampled from the
facility stack, the ESP/baghouse inlet, and a multi-purpose spare line. The flue gas is dried
before the analyzers by a sample conditioning system which uses an ice bath to condense
water from the sampled gases.
-------
Pollution Control Equipment
The pollution control devices installed in the facility include an electrostatic precipitator, a
pulse-jet baghouse, and a packed column scrubber. The pulse-jet baghouse and the
scrubber were required for the air quality permit of the facility issued by the Jefferson
County Board of Health, and therefore are always on-line. The ESP can be valved into
the system before the pulse-jet baghouse to give a series ESP-baghouse combination.
Coals Studied
The suite of coals used in the study are listed, along with their analyses, in Table 1. The
North River coal, which served as the baseline fuel for Gaston Unit 5, is the only non-
compliance coal in the set. The other coals include local Alabama bituminous compliance
coals, South American coal, and Powder River Basin coal. This set represents a wide
range of coal type, especially regarding NOX emission potential and unburned carbon. The
Shoal Creek coal has a fuel ratio (proximate fixed carbon divided by volatiles) of 2.5, while
the Belle Ayr coal has a fuel ratio of 1.2. The set also includes the Gusare coal from
Venezuela, one of several regional coals notorious for producing large amounts of
unburned carbon in low NOX combustion.
Pilot-Scale Simulation of Full-Scale Boilers
Useful results from any small-scale testing requires the ability to predict performance in
the full-scale equipment from the experimental results. The pilot-scale experiments must
be designed to replicate the controlling mechanisms that occur in the real boiler.
Fortunately, a great deal of previous work on scaling is available to guide the experiments.
Because of the focus of this testing on the NOX emissions and unburned carbon, the
intensity of combustion in the radiant furnace of the full-scale equipment must be matched
in the pilot furnace. In this work, the most common scaling parameter of combustion
intensity, the volumetric heat release ratio (VHRR), from each full-scale plant was
matched in the Combustion Research Facility. The VHRR is computed by dividing the
fuel heat input, as Btu per hour, by the radiant furnace volume to yield Btu per hour per
cubic foot.
The formation of NOX from the combustion of a particular coal is influenced by the flame
temperature, the amount of combustion air, the mixing of air with the fuel in the flame,
and the flame structure. In performing pilot-scale simulations for NOX emissions, each of
these factors are taken into account. The flame temperature, and the subsequent furnace
temperature profile, are fixed by the amount of heat removed from the furnace, given that
the VHRR is matched. It is then a simple matter to match the amount of combustion air
provided to the furnace to provide equivalent oxygen concentrations. Matching the mixing
of fuel and combustion air requires a minimum size of the pilot furnace, greater than 3.5
million Btu per hour firing rate, due to the relative rates of macro- and micro-mixing.
Beyond this limitation, it is impossible to reproduce the large-scale mixing patterns of a
normal boiler in a research furnace, especially since the pilot furnace has only a single
burner. However, the single burner can be configured to match the setting of an average
-------
burner in a multi-burner wall-fired furnace. It is then assumed that inter-burner effects
are smaU, and do not greatly influence the NOX emissions. But any maldistributions of
coal or secondary air are also ignored. For tangentially-fired units, the overall furnace
flame structure is simulated in the pilot furnace, by setting the single burner to produce a
long, slowly swirling flame intended to represent the fireball found inside a single chamber
of a'tangentially-fired furnace. The diameter of the pilot furnace is so small that an
attempt to use tangential fuel injection would result in coal particle deposition on the walls
of the furnace.
The secondary scaling rules that are used include matching burner velocities, coal grind
size, primary and secondary air temperatures, and the time-temperature history through
the furnace. The burner is scaled to deliver the relative amounts of secondary and
primary air, at the same velocities found in the full-scale burner. The time-temperature
history of the pilot furnace should match the full-scale. For this experiment, it was
necessary to replace some of the radiant furnace section refractory to maintain furnace
exit temperatures, which is necessary to produce similar time-temperature histories. In
general, it is possible to control heat extraction in a pilot furnace to match almost any
time-temperature history.
Low NOX firing simulation in the pilot furnace is achieved by the use of a generic dual
register burner for wall-fired units and separated overfire air injection for tangentially-fired
units. The dual-register burner can simulate the common aerodynamics of most
commercial low NOX burners. There are many different designs of fuel gun tip and coal
spreader designs for commercial burners, which could be adapted to the pilot burner.
The final scaling rule used is the most useful and important. A baseline fuel, which has
been fired in the full-scale boiler, is tested in the simulation of the pilot furnace, followed
by the proposed compliance coal. Predictions of compliance fuel performance are then
made relative to this baseline fuel testing. By using a relative measurement technique, not
only the coal quality, but the NOX emissions and unburned carbon, of the two fuels can be
compared directly, while using the best possible pilot simulation of the full-scale furnace.
NOX Emissions and Unbumed Carbon Results
Tangential-Fired Boilers in Conventional Mode
Results of the testing are summarized in Table 2. The first seven test runs were
conducted in a simulation of Alabama Power's Gaston Unit 5. Initially, the baseline fuel
used in the previous CQE test program, North River, was burned in the Combustion
Research Facility. The NOX emissions and unburned carbon results from the full-scale
testing and the CRF are displayed graphically in Figure 3. As can be seen in this figure,
the pilot furnace gave lower NOX emissions than that seen in the full-scale plant data,
although the results are much closer at high furnace exit oxygen. For LOI, the bottom
graph shows that the pilot furnace did an excellent job of simulating the LOI performance
of Gaston 5.
-------
Tangential-Fired Boilers in Low NOX Mode
Further testing on possible compliance coals for Gaston 5, under both conventional and
low NOX modes, are listed in Table 2. These results have been previously reported.1'2'3
The only pilot-scale testing of compliance coals that can be compared to the full-scale at
Gaston 5 is the low NOX testing of the Shoal Creek coal. This coal, produced by
Drummond, is from the Blue Creek seam in west-central Alabama and is very similar to
the Jim Walter #3 and #5 mines. A full-scale test burn of a blend of Jim Walter #3 and
#5 has been performed in Gaston 5 after the installation of the low NOX firing system.
Therefore, the full-scale results on this very similar coal are compared to the CRF testing
in Figure 4. Unfortunately, there is no reliable full-scale information on LOI yet available,
so only the NO^ emission comparison is shown. As can be seen in the graph, the pilot
furnace results correctly predicted the NOX emissions seen in the Gaston 5 system.
Wall-Fired Boilers in Low NOX Operating Mode
Two wall-fired boilers were simulated in the testing program: Alabama Power's Plant
Miller Unit 3 and Gulf Power's Plant Crist Unit 7. The last two lines of Table 2
summarize the pilot testing results for these two wall-fired simulations. The pilot furnace
results are compared to Gulf Power's Plant Crist Unit 7 in Figure 5, for the Gusare coal
from Venezuelan. The pilot results were obtained for five different size grinds of the
same coal3, but the graph only presents the largest and smallest sizes. The coarse size,
which matched the full-scale plant coal grind, was 50 percent through 200 mesh and the
ultra-fine grind sized at 99.9 percent through 200 mesh and 80 percent passing 325 mesh.
The pilot testing results underpredicted the actual NOX emissions and also underpredicted
the LOI as compared to the actual full-scale data. It is a good possibility that the pilot
scale burner, being very well-controlled for fuel and air flows, does not adequately
represent the maldistributions of fuel and air typically seen in full-scale plant.
The comparison of the CRF testing results to Plant Miller Unit 4, a sister unit to #3, is
presented as Figure 6 for the Belle Ayr coal, an AMAX Powder River Basin coal. This
plot compares the NOX emissions from the pilot furnace and the full-scale boiler, but only
shows LOI for the pilot rig. The NOX emissions predicted from the pilot testing agreed
very well with the full-scale data. Since the pilot furnace produced almost unmeasurable
quantities of unburned carbon, it is expected that there would be no problem in the full-
scale. Unfortunately, the LOI data from the full-scale test burn is not available.
Boiler Operation Results
A secondary purpose of the pilot testing program was to also predict the existence and
severity of any boiler operating problems that one of these compliance coals might cause
in a full-scale boiler. These would include such coal quality issues like milling behavior,
slagging and fouling, ESP operation, etc. Several observations of coal behavior were made
in the testing program which mirrored or, in some cases, predicted operational problems in
the full-scale plant. The Shoal Creek coal, being finely-sized and wet from a washing
plant, caused considerable coal stoppages in the coal bunkers in the pilot testing. This
-------
behavior has also been seen in the Gaston 5 test burn of Jim Walter #3 and #5 blend,
which is very similar in nature to the Shoal Creek. The other notable observation in the
pilot testing was the tendency of the ash from the Belle Ayr coal to suddenly deposit in
the convective tube bank sections. As the temperature of the flue gas slowly increased
over the testing period, a critical temperature (which was not measured) would be reached
and rapid deposition of ash would occur. Alabama Power's Plant Miller did experience
convective section fouling while burning this same coal.
Conclusions
Pilot-scale testing of a wide range of coal types in both tangential and wall firing modes
show remarkable similarity to the limited full-scale data available for both NOX emissions
and unburned carbon levels. Because of the database established by this research
program, there is a high degree of confidence that pilot testing in the SCS/SRI
Combustion Research Facility can predict full-scale NO, and unburned carbon levels.
Other coal quality parameters, which will be reported on in the future, can also be
measured which predict behavior in full-scale plants.
Acknowledgements
The authors would like to thank Drummond Coal for providing coal and transportation for
this research program. Alabama Power's Central Analytical Facility is providing the
necessary analytical support for the project. The SRI staff responsible for the operation
and maintenance of the combustion facility are Wim Marchant, Sam O'Neal, Ken O'Neal,
Bill Page, and Dave Smith.
References
1. L. Monroe, R. Clarkson, and J. Stallings, "Pilot-Scale Evaluation of Compliance
Coals for NOX Emissions and Boiler Operations," presented at the EPRI/EPA 1993
Joint Symposium on Stationary Combustion NOX Control, Bal Harbour, Florida
(May 1993).
2. L. Monroe, R. Clarkson, and J. Stallings, "Pilot-Scale Evaluations of Compliance
Coals for NOX Emissions and Unburned Carbon Levels," presented at the EPRI
1994 NOX Workshop, Scottsdale, Arizona (May 1994).
3. L. Monroe, R. Clarkson, and J. Stallings, "Predictions of Full-Scale NOX Emissions
and LOI from Coal and Coal Blends in Pilot Combustion Experiments," presented
at the EPRI Fourth International Conference on The Effects of Coal Quality on
Power Plants, Charleston, South Carolina (August 1994).
-------
Table 1. Pilot Furnace Coal Analyses
Analysis
North River
Cedrum
Shoal Creek
Cedrum-Shoal
Creek Blend
(80/20)
Gusare
(Venezuelan)
Belle Ayr
(PRB)
Proximate, as received
% Moisture
% Volatile
% Fixed Carbon
% Ash
Total
HHV, Btu/lb
MAF HHV, Btu/lb
Fuel Ratio
5.53
34.28
47.95
12.23
100.0
12098
14711
1.40
6.10
31.12
49.07
13.71
100.0
11647
14525
1.58
6.95
24.28
60.18
8.59
100.0
13104
15515
2.48
10.22
27.78
49.52
12.48
100.0
11420
14776
1.78
5.66
33.87
54.37
6.10
100.0
13115
14864
1.61
26.32
31.02
36.54
6.12
100.0
8682
12851
1.18
Ultimate, dry basis
% Sulfur
% Carbon
% Hydrogen
% Nitrogen
% Oxygen (diff.)
% Chlorine
2.72
72.32
4.84
1.67
5.50
0.03
0.65
72.71
4.60
1.73
6.25
0.08
0.77
80.76
4.67
1.78
2.79
0.02
0.69
72.34
4.53
1.74
6.80
0.05
0.60
78.90
5.15
1.49
7.39
0.02
0.59
68.57
4.84
0.87
16.82
0.01
-------
Table 1. Pilot Furnace Coal Analyses (continued)
Ash Analysis
% A12O3
% CaO
% Fe2O3
% MgO
%MnO2
%K2O
%SiO2
%Na2O
%TiO2
North River
25.00
7.35
19.84
1.35
0.05
2.07
38.90
0.60
1.09
Cedrum
22.04
4.62
4.97
1.44
0.06
4.22
62.55
0.73
1.25
Shoal Creek
30.45
2.45
9.58
1.38
0.04
2.76
50.29
1.12
1.52
Cedrum-Shoal
Creek (80/20)
25.59
3.74
5.89
1.44
0.05
3.02
54.58
0.71
1.21
Gusare
(Venezuelan)
23.97
3.25
6.75
3.69
0.06
4.01
56.03
0.69
1.04
Belle Ayr
(PRB)
16.99
20.22
6.18
3.62
0.06
0.39
32.21
2.44
1.37
Ash Fusion Temp.
Reducing, °F
Initial
Softening
Hemispherical
Fluid
Oxidizing, °F
Initial
Softening
Hemispherical
Fluid
Grindability
2096
2289
2320
2490
2365
2427
2480
2587
52.3
2450
2570
2605
2740
2360
2590
2635
2750
47.0
2565
2692
2743
2795
2250
2783
2822
>2824
76.5
2489
2593
2636
2736
2523
2664
2713
2773
54
2205
2315
2355
2500
2260
2300
2360
2525
45.5
2135
2170
2180
2210
2195
2220
2229
2272
50.5
-------
Table 2. Alabama Power/EPRI Pilot Furnace Testing Summary
Date
Sept. 92
Oct. 92
Dec. 92
May. 93
Jun. 93
Jul. 93
Nov. 93
Jan. 94
Jun. 94
Nov. 94
Coal
N. River
Shoal Creek
Shoal Creek
Cedrum
Cednim
Gusare
(Venezuelan)
Cedrum-
Shoal Creek
Cedrum-
Shoal Creek
Gusare
(Venezuelan)
Belle Ayr
(PRB)
Firing Mode
T-fired
Conventional
T-fired
Conventional
T-fired
Low NO,
T-fired
Conventional
T-fired
LowNOx
T-fired
Conventional
T-fired
Conventional
T-fired
Low NO,
Wall-fired
Low NO,
Wall-fired
Low NO,
Plant
Simulated
Gaston #5
Gaston #5
Gaston #5
Gaston #5
Gaston #5
Gaston #5
Gaston #5
Gaston #5
Crist #7
Miller #3
NO, and LOI
3.5% O2 Furn.
Exit
0.48 Ib/MBtu
1.25 %
0.68 Ib/MBtu
2.6 %
0.47 Ib/MBtu
2.3 %
0.58 Ib/MBtu
0.6%
0.37 Ib/MBtu
1.9 %
0.49 Ib/MBtu
10.3 %
0.56 Ib/MBtu
1.0%
0.30 Ib/MBtu
1.9%
0.46 Ib/MBtu
14.4%
0.34 Ib/MBtu
< 0.1 %
Full-Scale
Comparison
0.60 Ib/MBtu1
1.0%
0.50 Ib/MBtu2
NA
0.59 Ib/MBtu3
22 to 41 %
0.33 Ib/MBtu4
NA
1 EPRI/DOE CQE testing at Gaston Unit 5 with furnace exit oxygen of 3.6%.
2 Fuel test burn with Jim Walter #3/#5 blend, a coal from the same seam and very similar to Shoal Creek with furnace
exit oxygen of 3.7%.
3 Fuel test burn at Crist Unit 7 with furnace exit oxygen of 3.2%.
4 Fuel test burn at Miller Unit 4 with furnace exit oxygen of 3.5%.
-------
SCS and SRI Combustion Facility
COAL HANDLING
-INDffiECT TIRED
-ON-SITE STORAGE, 10 EACH 20-TON BDiS
-CRDSHDiG AH) PULVERIZING EQUIPMENT
-GRAVIMETRIC FEED TO FURNACE
Dense Phase
Transporter
Figure 1. Coal handling equipment of the SCS and SRI Combustion Research Facility.
-------
SOUTHERN COMPANY AND SOUTHERN RESEARCH
COMBUSTION RESEARCH FACILITY
Stack
Packed
Column
Scrubber
Electrostatic
Precipitator
Figure 2. Pilot furnace and flue gas handling equipment of the SCS and SRI Combustion Research
Facility.
-------
North River Coal NOX Emissions
Full- and Pilot-Scale Comparisons
o 80°
CO
| 600
OT
o
400
0
w
x
O
200
0
I
f
•v
1
yaqtnr
rith >
Co
— i —
R T(
orth
nvent
— i —
=«*fci»-g
River
Lonal
— i —
. — - —
**
Tange
r- i
i
^•Gh-
^^
TRF
;ntial-
— " '"I
^^
Resul
-Firec
i — i —
-
•*•* _
is ~ ~
— i —
1.0 1.5 2.0 2.5 3.0 3.5 4.0
Furnace Exit 02, % wet
pl.O
-
-0.8
-
-0.6
-0.4
~
-0.2
-0.0
PQ
"b
\
r-H
w"
fl
O
•r-l
CO
CO
a
H
X
0
Jz;
4.5 5.0
4.
3.
3.
0
o
North River Coal Unburned Carbon
Full- and Pilot-Scale Comparisons
a
O
.0
.0
0.0
1
Gastc
with
C
1
1
n 5 '
North
'onvei
i i • • —
1
'estin
Rive
itiona
1
gn^
r ^
1 Tan
1
\
genti<
1
CI
'^
d-Fir
i
*F Re
^^
ed
i
-
-
-
-
suits _
>1
-
1.0 1.5
Figure 3.
2.0 2.5 3.0 3.5 4.0
Furnace Exit 02, % wet
4.5 5.0
NOX emission and unburned carbon comparison of Alabama
Power's Gaston Unit 5 and the SCS and SRI Combustion
Research Furnace for the North River coal, with conventional
tangential firing.
-------
CO
I
OT
o
•c-f
CO
800
600
400
Shoal Creek Coal NOX Emissions
Full— and Pilot—Scale Comparisons
200
x
O
iz;
0
i
1
i
CRF
angei
i
i
Gas
will
. —
Results
itial-Fired
i
;on 5
. J. "W
i a~
. — -~^
with
i
Testi
alter
~^TI
Overf
i
ng
3/5
-— — -•*
ire A]
i
_r
r
r 1.0 «
-0.8
-0.6
-0.4
-0.2
-0.0
CO
O
W
Pi
.2
*w
m
x
O
1.0 1.5 2.0 2.5 3.0 3.5 4.0 4.5 5.0
Furnace Exit 02,
% wet
Figure 4. NOX emission comparison of Alabama Power's Gaston Unit 5
on Jim Walter #3 and #5 coal and the SCS and SRI
Combustion Research Furnace for the Shoal Creek coal, with
low-NOx tangential firing.
-------
Gusare Coal NOX Emissions
Full- and Pilot-Scale Comparison
5^
70
S
pH /-» /-\ r\
fX 600
OH
w
C /< n n
o 400
OT
w
s
X
O
— 1
1.
K!
i— i -
ri
• i-H
£ 1 n
0 _
cC
O
0-
I
x-mT7
CRr
Res
CRI
Tf
i
fTIltr
ultsr
1 Coai
rall-F
— i —
*-Fin
_
*se R<
ired '.
— i —
e
^sults
Dual-
— i —
Cris
Test
n
Regis'
— i —
: 7
ing
_ — '
__— —
,er Bi
— i —
_ •
i— — : — ™*
irner
i
.
_
-
i i i | i i i i i
0 1.5 2.0 2.5 3.0 3.5 4.0 4.5 5.
Furnace Exit 02, % wet
Gusare Coal Unburned Carbon
Full— and Pilot— Scale Comparisons
CRI
— i —
1
1 Coai
CRF T
Wall-
— i —
*se R
^-x
Jltra-
_
— i —
i
3SUlts
^
Fine
~ — . —
Dual-
i —
C
n i
^\
Resul
— - —
^Regij
Jrist '
'estin/
^-^
ts
— —
ster E
y
r
urnei
i
-
0
rJ
r-i.o m
*0
r\ o T— 1
1 1 1 1 1 1 1 1
D O O O C
D ro ^ b» b
^Ox Emissions, Ib/
1.0 1.5 2.0 2.5 3.0 3.5 4.0 4.5 5.0
Furnace Exit 02,
% wet
Figure 5.
NOX emission and unburned carbon comparison of Gulf
Power's Crist Unit 7 and the SCS and SRI Combustion
Research Furnace for the Gusare coal, with dual-register low-
NOX burner firing.
-------
CO
I
co"
800
600
Belle Ayr Coal NOX Emissions
Full— and Pilot—Scale Comparison
400
w
CO
200
0
1
T
1
CR]
Fall-F
i
i1 Res
ired '.
i
ilts___
Dual-
i
JLfill
TH.XIJ
Tes
C
Regisl
i
er-3-
ting
]
,er Bi
i
_
irner
i
—
_
-
1.0 1.5 2.0 2.5 3.0 3.5 4.0
Furnace Exit 02, % wet
Belle Ayr Coal Unburned Carbon
in the Pilot Furnace
CO
fc
ti
• rH
fl
O
In
cd
O
1.0 1.5 2.0 2.5 3.0 3.5 4.0
Furnace Exit 02, % wet
rl.O PQ
^0.8
-0.6
-0.4
-0.0
4.5 5.0
•4r.U
q c
o.O
q r»
o.U
2C
.D
2 A
.U
1 f^
l.D
1 n
1 .U
Oc
.0
n n
i
Wai
1-Fire
i
id Du
i
al-Re
A
i
gister
i i 4
1
Buri
k ,
i
ier
A JL
1
-
-
-
-
-
-
4.5 5.0
Figure 6. NOX emission and unburned carbon comparison of Alabama
Power's Miller Unit 4 and the SCS and SRI Combustion
Research Furnace for the Belle Ayr coal, with dual-register
low-NOx burner firing.
03
O
CO
a
O
• i-H
CO
CO
-0.2 H
X
O
-------
1995 EPRI/EPA Joint Symposium on Stationary Combustion NOX Control
Methodology for the Selection of
Low NOX Firing Alternatives
ComEd Company
J. Herzau, D. Kubik, K. Wanninger
Electric Power Research Institute
E. Petrill, C. Dene, T. Facchiano
Radian Corporation
T. Kosvic, T. Rizk, J. Hollinden
Abstract
Combustion Modification (CM) is an often used technique for reducing NOX formation from
utility boilers. CM can be implemented by changing boiler operation parameters using the
existing firing equipment, or through hardware modifications. Achievable potential NOX
reduction levels and long term operational effects of either approach are boiler specific.
Implementation of either methodology can lead to changes in boiler operating efficiency,
reliability, availability, and maintenance costs. An economic analysis of alternatives must
accurately reflect associated costs. Proper implementation of the selected NOX reduction
technology is essential in preventing the transformation of an emission problem into an
operational problem.
In a joint effort by ComEd, EPRI, and Radian Corporation, a technique has been developed to
facilitate the comparison of the long term cost effectiveness of each means of implementing
combustion modification. The approach begins with Baseline and Exploratory Combustion
Modification (ECM) testing. Baseline tests identify critical operational design parameters. ECM
testing determines the flexibility of the existing furnace to accommodate operational and/or
hardware changes. The resultant test data is integrated into the Radian Furnace Simulation
Model (FSM).
The Radian FSM is state-of-the-art three dimensional, 2 phase flow, fluid dynamic model of the
boiler furnace. The model simulates furnace conditions by drawing extensively from experience
and combustion data. The model computes NOX values and operational rating factors used to
compare CM approaches. These factors may include emissions concentrations, slagging,
waterwall corrosion potential, and flue gas temperature. These calculated factors can then be
used to evaluate long term effects of CM alternatives.
-------
This paper outlines the key features of the FSM and its successful application to a boiler on the
ComEd system.
INTRODUCTION
ComEd operates one of the largest coal-based power systems in the U.S. Earlier
decisions to convert to Powder River Basin, low sulfur coal resulted in only one of
ComEd's 9 coal fired plants being affected by the 1990 Clean Air Act Amendments
Phase I requirements. ComEd may be required to implement boiler hardware or
operational changes on its existing coal fired furnaces by the year 2000. Currently,
ComEd is evaluating NOX compliance options for a matrix of potential emission limit
scenarios. An important aspect of all the possible scenarios will be NOX controls for
ComEd's ten tangentially fired boilers. As such, ComEd needed an approach to
evaluate the impacts of low NOX operations on this class of boilers.
This need evolved into a joint effort by ComEd, EPRI and Radian to employ a three
dimensional fluid dynamic model to evaluate various boiler optimization and hardware
alternatives. ComEd selected Will County unit 3, a twin furnace CE boiler rated at
278 gross megawatts, for evaluation. The unit currently burns Powder River Basin
coal. The unit is equipped with off-set auxiliary air nozzles which lower NOX and
provide an air curtain along the walls. The Unit 3, reheat boiler study is focused on
the development of a specific Furnace Simulation Model (FSM) that fully characterizes
the unit's operating performance. The prime use of the model is to provide information
necessary to evaluate NOX emissions and the potential of immediate and long term
operational problems associated with use of combustion tuning or hardware changes
as the means to affect NOX reductions. The FSM for Will County unit 3 is verified
with existing data.
The FSM couples modern computational fluid dynamics (CFD) capabilities with the
practical boiler knowledge and experience of Radian and ComEd. The model is
designed to run within the PHOENICS Navier-Stokes equation solver developed by
CHAM Limited, London, England. The model predicts the following boiler parameters
and operating conditions:
• furnace temperature profile
• furnace flow field velocity profile
•furnace coal combustion rate profile
• furnace NOX generation profile
• unburned carbon at the furnace exit
• O2 at the furnace exit
• furnace exit gas temperatures
• furnace slagging potential
• furnace tube corrosion potential.
-------
APPROACH
The approach is broken down into four steps: (1) model development, (2) model
validation for baseline conditions, (3) assessment of tuned configuration and (4)
assessment of hardware options. These steps are discussed in detail next.
Model Development. This step involved the gathering of information necessary for
input and verification. Pertinent data included the current burner configuration and
general boiler geometry as well as boiler operating parameters (e.g., excess air levels,
burner tilt angles). Input variables are then modified to reflect geometric and
operational changes.
The model input requirements include burner details sufficient to simulate coal,
primary air, and secondary air flow areas and velocities at each region. Ultimate and
proximate fuel analyses, ash fusion temperature data, mill fineness test results and
fuel air balancing data were provided by ComEd for input into the model.
The result of this phase was a detailed uncalibrated model incorporating the specific
geometry characteristics of Will County Unit 3.
Model Validation - Baseline Conditions. In this step, the previously developed boiler
model is run at specific conditions for which Baseline test data has been acquired.
Comparisons are made of predicted NOX levels, unburned carbon levels, furnace exit
gas temperature and economizer 02 levels to field measured values. Any operating
anomalies noted during the testing (e.g., impinging flame patterns and slag buildup)
are compared with model results for verification.
Combustion relationships reflecting the properties of Powder River Basin coal were
used in this work. These properties include high moisture and volatiles. The setting
of combustion parameters is based upon constructing a separate model of a test
burner for which detailed flame probing data is available for this coal. Parameters are
adjusted to "fine tune" the combustion to obtain the best match with the subscale
data. These parameters are then used in the full 3D boiler model.
Assessment of "Tuned" Configurations. The next step in the model development is
to assess the suitability of various "tuned" firing configurations for long term
operation. The present model data includes flue gas furnace exit temperature, profiles
of excess oxygen, NOX and unburned carbon. The model also provides indices
reflecting possible concerns of slag buildups and tube wastage. Resultant values for
these correlation indices are compared with values established for the "as found"
firing condition (baseline task) to provide a comparison between previous long term
observations of these problems with any future problems that could result from
"tuned" low NOX configurations. Model predictions are also compared to actual field
data taken while operating Will County unit 3 under low NOX firing conditions.
-------
The tuned conditions are examined over a load range with excess air varied to reflect
test results. Mills-out-of-service conditions are also examined as to their influence on
the calculated parameters.
Assessment of Hardware Options. The final step in the model development focuses
on "generic" configurations of various firing hardware changes; specifically close
coupled, separated, and close coupled and separated OFA installation. Model
simulations are made to examine the immediate and long term effects of these
hardware changes. Current efforts are concentrated on close coupled OFA. The
results of these calculations are compared with those from the base validation effort
and those developed in the evaluation of the tuned alternatives.
RESULTS
Model Validation. Combustion parameters obtained from correlation with subscale
data were used to predict NOX and operating factors for validation with a series of
Baseline tests. The results of the comparison are presented in Table 1. The test data
are presented in terms of actual species concentration by "uncorrecting" the reduced
test data for a direct comparison with model output. The model output data are a flux
weighted area average over the exit plane (nose region) of the model.
All the cases were computed with the same combustion and kinetic parameters as
correlated from the subscale model. Predictability of NOX was very good at the full
load conditions.
Indices of the relative changes in unit availability and reliability are also computed.
These are computed in terms of the local gas equivalence ratio. This parameter is
a measure of the fuel richness of both the gas species and the solid particle
environment. Fuel richness affects both corrosion and slagging potential. These
relative indices are:
• Average Equivalence Ratio
• Maximum (Fuel Rich) Equivalence Ratio
• Minimum Equivalence Ratio
• Fuel Rich Area Fraction
• Slag Index
• Low Temperature Corrosion Index
• High Temperature Corrosion Index
The wall patches (computational areas) for which the indices are computed are
presented in Figure 1. To illustrate a practical application of the modeled results,
The index computations are presented for a base case at baseline operation (Case
-------
3) and for a simulation of OFA (Case 10). The relative indices for the two cases
are shown in Tables 2 and 3. A compilation of the slag and corrosion indices are
shown in Boiler Diagram 1. These computations show that, although significant
NOX reduction can be achieved with SOFA, the environment in the vicinity of the
water walls is much more fuel rich, thus conducive to excessive slag and
corrosion.
Boiler Optimization. Comparison of NOx predictions with a second test series of
tuned configuration cases is shown in Table 4. Typically, very good agreement
was found. Where discrepancies were noted, potential flow maldistribution effects
associated with fuel and air biasing are being evaluated.
A summary of the slag and corrosion index values is shown in Boiler Diagram 2.
The simulation results show, for cases where fuel was negatively biased in the
furnace, increased slag and corrosion potential in the lower third part of the boiler
and the right side of the boiler. Even with "concentric air" positioning of the
auxiliary air flow, the model indicates that coal and primary air velocities are high
enough to penetrate the air curtain yielding a fuel rich environment adjacent to the
walls.
The expected impact utilizing LEA by reducing O2 is shown in Figure 2. The results
show that the NOX levels are reduced with reduction in excess oxygen. However,
the wall fuel rich fraction increased with decreased O2, thus increasing the likely
potential for slag and corrosion. High and low temperature corrosion indices
behavein an inverse parabolic relationship to the excess oxygen. The corrosion is
expected to be minimum at excess oxygen of about 3.0%.
Hardware Modification. Potential hardware changes to Unit 3 include Low NOX
Control Concentric Firing System, Level I (LNCFS1) hardware. Boiler Diagram 3
and Figure 3 shows the calculated impact of varying excess O2 on NOX generation
levels and wall slag and corrosion indices. The results of the simulation suggests
that NOX emission can be further reduced with lower O2. However, as was shown
with the optimization cases, high temperature and low temperature corrosion
increases with reduced 02 levels.
ONGOING WORK
Work continues on the study of fuel richness and its impact on the water walls in
low NOX firing situations. Of particular importance is the correlation of absolute
index values to actual observations of furnace slagging/emission conditions.
Proper evaluation and correlation will lead to confidence that long term low NOX
firing modes will be acceptable in terms of changes in heat rate, reliability,
availability, maintainability, and operability.
-------
CONCLUSIONS
Modern computers and CFD analysis techniques have advanced to the point where
the complex processes of combustion in boilers can be described quite well.
Combining computational capability with extensive theoretical and hands-on field
experience in combustion NOX control can result in a combined modeling and
testing approach which satisfies the unit-specific requirements for evaluating low
NOX firing approaches. Application of this model has given confidence that low
cost optimization approaches will work on Will County #3 without significant
deleterious long term effects. Further correlation of absolute index values to actual
observations of furnace slagging/corrosion conditions is critical. Work will continue
to strengthen this confidence.
-------
Table 1: Baseline Configuration
Test Infonnatioa-Basdine Configuration Tests
ro#
3
4
5
6
7
8
10
11
Load(Mw>
278
278
278
260
260
278
190
190
M*t)
Normal
Low
High
Normal
Normal
Normal
Normal
Normal
\ Notes
Tilts - 28 deg.
Tilts + 12 deg.
Damper Var
OFA Simul
MOOS.Dmp
Test
Results
NO2 (ppnw)
243
125
259
280
236
232
111
209
Calculations
NO, (ppmv)
242
133
245
228
221
393
110
364
Table 2: Case 3. Baseline Condition Index Computation
Wall
6
12
10
7
13
11
5
15
9
3
14
16
Avg,
Eq. R.
.690
.815
.774
.799
.695
.798
.710
.580
.765
.762
.703
.858
Max.
Eq. R«
.919
1.230
1.070
.819
.800
.859
.819
.840
.823
.862
.933
1.040
Min*
Eq, R.
.483
.423
.464
.728
.541
.597
.506
.348
.566
.651
.402
.685
Fuel
Frac.
.000
.170
.054
.000
.000
.000
.000
.000
.000
.000
.000
.058
Slag
.00
.00
.00
.00
.00
.00
.00
.00
.00
.00
.00
.00
LT
Com
.000
2.390
.727
.000
.000
.000
.000
.000
.000
.000
.000
26.100
HT
Corr.
.000
1.350
.425
.000
.000
.000
.000
.000
.000
.000
.000
.454
Area,
ft2
706
524
1310
706
524
1310
397
584
1470
397
584
1190
-------
Table 3: Case 10. Separated OFA Index Computations
Watt
6
12
10
7
13
11
5
15
9
3
14
16
Arg.
Eq. IL
1.170
.942
.470
1.130
.938
.458
1.110
.634
.473
1.120
.675
.433
Max.
Eq/R.
1.250
1.260
.753
1.200
1.130
.604
1.140
1.110
.583
1.260
1.200
.697
Min.
Eq. IL
.963
.067
.165
1.020
.077
.174
.875
.021
.222
.942
.020
.245
Fuel
Frac.
.987
.691
.000
1.000
.671
.000
.909
.135
.000
.938
.073
.000
Slag
74.10
.00
.00
20.52
13.00
.00
18.60
3.22
.00
10700.00
288.00
.00
LT
Conr.
12.90
9.97
0.00
2.70
0.12
0.00
11.70
423.00
0.00
12.50
28100.00
0.00
HT
Corr.
10.400
6.140
0.000
10.100
5.640
0.000
8.810
1.180
0.000
9.000
> 100,000
0.000
Area,
ft2
706
524
1310
706
524
1310
397
584
1470
397
584
1190
Table 4: Tuned Configurations
Test Information-Tuned Configuraiton Tests
ID#
35
36a
36b
36d
37b
37e
38a
38b
38c
=====
Ix>ad(Sfw)
278
278
278
278
278
278
278
278
278
—
Oji^v)
Normal
Normal
Normal
Low
Normal
Normal
Normal
Normal
Normal
=====
Notes
Top FDR bias
Same, Aux Air Bias
Same, Aux/FA Bias
Same, Aux/FA Bias
Same, Aux/FA Bias
Top FDR bias, top Aux/FA 100%
Top FDR bias, POOS
Same, Aux 100%
Same, Aux/FA 100%
============================================
Test
Results
NO, (ppmv)
238
190
181
162
190
170
217
209
197
Calculations
NOX (ppm?)
230
229
218
476
231
295
219
232
219
-------
Designation of Wall Patches
Figure 1
-------
Effect of Excess Air on
NOx and Wall Fuel Rich Fraction
300
250
200
1.150
x
O
100
50
— Test NOx
-^CalcNOx
-*-W10 Frf
— W12 Frf
0.6
0.5
0.4 •-
jo
"5
(0
0.3
u_
.c
_o
DC
"5
0.2 if
1
0.1
0123456
Excess Oxygen (% dry)
Effect of Excess Air on
High and Low Temperature Corrosion Indices
o
4
X
CD
-o
c.
~™
53
O
Q.
£
CD
^-
* 2
0 "
To
5
1
0
/•*
— W10 LTCR Ix
-+-W12LTCR Ix
•
t -*-W10HTCRIx
» -"'W12HTCRIX
I
\
\
\
\
\
\
-t \ •
X N '
\ *•
\
\
\ ^s
\ \
\ ^--+-'+
\^
I I
2
o
Q.
.O
1
12345
Excess Oxygen (% dry)
Figure 2
-------
Simulation of LNCFS#1 Hardware
Effect of Excess Oxygen on NOx and Wall Fuel Rich Fraction
300
250
200
1.150
Q.
100
50
Frf
W12 Frf
0.6
0.5
T3
0.4 £
jo
"5
OS
0.3 £
_o
DC
"5
0.2 if
0.1
0123456
Excess Oxygen (% dry)
Simulation of LNCFS#1 Hardware
Effect of Lowered Excess Air Corrosion Indices
9
4
x
3
•o
c
"™
53
O
a.
E
|2
1 2
-J
"5
£
1
Q
— W10 LTCR Ix
+ W12 LTCR Ix
-t
\ -*-W10HTCRIx
\
_. ^ ••'W12HTCHIX
^^^_ \
'^^^" \
\
\
\
\
\
\
\
\
^ i_
\ x
\
N
X
N
•^^i ^^\^ s
^^^\^>.
""^^~«
1 . 1 . 1 , 1 . 1
3
4
X
-------
Diagram 1
WILL COUNTY 3
Slagging and Corrosion Indices for Baseline and Simulated Overfire Air Test
LEFT
REAR
RIGHT
FRONT
(uj)
Cuss.
3
10
(J2)
Casfi.
3
10
©cW
2*]
SsjM
f>
o
0
f}
0
0
s
0
74.1
IT
07
0
IT
IA
10.0
IT
0
12.9
\
HT
0.4
0
HT
1.4
6.1
HT
0
10.4 ^S
©CuJ
^
10
(Tj)
Cass
3
10
(obaas
3
10
|
is
ft
o
s
o
288
S
Q
>IOOO
|
IT
26.,]
0
LT
0
>IOOO
LT
0
12.5
HT
0.5
0
HT
0
>IOOO
HT
0
9.0
/©
(n)
Casfi
3
10
Oc^
3
s-l
^
0
13.0
S
o
20.5
IT
o
9.1
IT
f)
12.7
HT
0
5.6
HT
0
10.1 ^S
©
Case! S LT HT
3000
10 3.2 423 1.2
C^/Cascj S LT HT
3000
10 18.6 11.7 8.8
(&} Wall Number
S = Slagging Index
LT = Low Temperature Corrosion Value
HT = High Temperature Corrsosion Value
Case 3: Baseline, Full Load, NO x = 242 ppm
Case 10: SOFA Variation, NO x = 110 ppm
-------
Diagram 2
WILL COUNTY 3
Slagging and Corrosion Indices for Baseline and Optimization
Using Low 02 or Feeder Bias
LEFT
Case
3
3fiB
36D
ft
0
o
0
IT
0,7
0.2
0
HT
0.4
0,1
0
REAR
RIGHT
Caas
3
36B
36D
ft
ft
ft
0
LT
26.1
0
0
HT
0,5
0
0
FRONT
©
Cast
3
36B
36D
ft
0
o
0
IT
2.4
3.9
6.8
HT
1.4
2.2
4.5
Cass
3
36.B
36D
ft
0
o
486
LT
0
8.4
>IOOO
HT
0
0.1
>IOOO
Case
3
36B
36D
s
o
Q
9.8
IT
o
ft
7.0
HT
o
ft
4.7
riafif
3
36D
ft
ft
8,2
76. 1
IT
ft
0.2
13.7
HT
ft
0,1
12.3
Jbasc
3
3fiB
36D
ft
ft
ft
>IOOO
LT
ft
0
14.4
HT
0
0
11.6
23.5
LT
13.8
HT
^ yasc
3
36B
36D
ft
0
0
23.6
IT
0
o
14.1
HT
o
o
11.7
\£) Wall Number
S •» Slagging Index
LT = Low Temperature Corrosion Value
HT = High Temperature Corrsosion Value
Case 3: Baseline, Full Load, NO x = 242 ppm
Case 36B: Full Load, A/E Feeder = -50%, Aux. Air = 75%, Fuel Air = 46%, NO x - 202 ppm
Case 36D: Full Load, A/E Feeder = -50%, Aux. Air = 75%, Fuel Air = 100%, O 2 = 2.0%, NOX
161 ppm
-------
Diagram 3
WILL COUNTY 3
Slagging and Corrosion Indices for Typical Hardware
LNCFS I at Full Load and Normal, High, and Low O
LEFT
Case
I
IH
IL
s
o
o
0
LT
1.2
1.1
1.7
HT
0.7
O.fi
1.0
REAR
RIGHT
Case
I
IH
1L
s
0
0
0
i/r
33
27,3
45.0
HT
0.7
0,5
0.9
FRONT
Case.
I
IH
IL
s
o
o
0
IT
2,fi
2,3
4.5
HT
1,5
1,3
2.6
Cnsf
I
IH
IL
s
o
o
0
IT
o
1,3
37.7
HT
o
0,1
0.1
©
©
©
©
J() Wall Number
S = Slagging Index
LT = Low Temperature Corrosion Value
HT = High Temperature Corrsosion Value
Case I: LNCFS I Simulation, NO x = 164 ppm
Case IH: LNCFS I High O 2 Simulation, NO x = 273 ppm
Case IL: LNCFS I Low O2 Simulation, NOX = 138 ppm
-------
Development and Testing of the MIT-RSFC
Low-NOx Burner for Coal Combustion
J.M. Beer, LE. Barta, P.P. Lewis, V. Wood
O.Akinyemi, J. Haynes, J.Jimenez and R. Manurung
Combustion Research Facility
Massachusetts Institute of Technology
Cambridge, MA 02139
L.W. Rogers
Babcock & Wilcox
Alliance, OH
Abstract
MIT as a partner with Babcock & Wilcox, a McDermott Company in the
USDOE sponsored Low Emission Boiler System (LEBS) program has
performed an experimental analytical and experimental effort to refine
and define the operational characteristics of the MIT Radially Stratified
Flame Core (RSFC) Burner for the combustion of pulverized coal. The RSFC
has been previously developed to a very promising state for natural gas1
and for heavy fuel oil2 firing. In the present effort the burner was
operated at a nominal firing rate of 5 MM Btu/hr on Illinois #6 coal. NOX
CO and unburned carbon emission data have been collected and analyzed for
a wide range of conditions. Burner performance has been evaluated as a
function of burner settings, firing rate, air equivalence ratio, air preheat,
transport air to coal ratio and excess air. For selected conditions ,
detailed in-flame measurements of gas composition, temperature and
velocity as well as solids concentration and composition have been
performed. These data can be used for detailed analysis and modeling
studies. The performance of the burner was found to be superior to the
expected results. The burner was extremely stable, and relatively
insensitive to operating conditions. NOX levels below 100 ppm with
carbon burnout greater than 99% were achieved.
The present contribution is based on a paper presented to the Joint AFRC-JFRC
Conference in Maui, Hawaii, November 19, 1994.
-------
Introduction
The purpose of this paper is to describe work performed for the U.S.
DOE under subcontract to Babcock & Wilcox on the development of a Low
NOX pulverized coal burner (< 0.15 Ib/MM Btu fired) for the DOE Low
Emission Boiler System (LEBS) program. In this paper we will describe
data and analysis obtained at the MIT Energy Laboratory's Combustion
Research Facility (CRF) on the development of the Low NOX pulverized coal
burner.
For the purposes of this development an existing Low NOX research
burner called the RSFC (Figure 1) and described in patent application No.
771,739 was used. An unique feature of the RSFC burner is its capability
of reducing air-fuel mixing in the near field while promoting it further
downstream of the burner. Radial density stratification, brought about by
the combination of rotating burner air flow and the density increasing
with increasing radial distance near the burner, damps turbulence and
hence reduces air-fuel mixing, but further downstream due to vortex
breakdown the mixing is vigorous, sufficiently, to ensure complete
burnout of the fuel. Standard grind Illinois No. 6 coal, a widely used high
volatile bituminous coal, was fired at about 400 Ib/hr for a variety of
operating conditions and a large body of data collected. Burner
parameters, staging, excess air and turndown were investigated.
The RSFC burner was used in two operating modes. In the first of
these all the combustion air was supplied through the burner. (No overfire
air.) In this mode of operation radial density stratification, necessary for
the internal air staging of the flame, required that increasing fractions of
air be introduced at increasing radial distances through the burner. This
resulted in a minimum NOX emission of 217 ppm (3% 02). A very low value
for a low NOX burner but not sufficiently low to satisfy the LEBS
emissions goal of 120 ppm.
In the second mode of operation less than the stoichiometric amount
of air was passed through the RSFC burner with the rest supplied as
overfire air, downstream of the burner. A hot, fuel rich flame was
established by fast mixing of the pulverized coal-transport air and the
primary burner air flows. The fuel gun was recessed into the primary air
conduit to permit ignition upstream of the burner face. A central
recirculation zone was formed due to the strongly swirling primary air
flow just downstream of the burner exit. For the "with overfire"
operation the RSFC burner was operated with a burner air flow
-------
distribution to give decreasing air flow fractions with increasing radial
distance from the burner axis. The residual combustion air was
introduced as "overfire air" at x/o=10.5 or 13.5, where D is the burner
diameter: 0.3 m.
Very low NOX emissions were obtained even for the mildly staged
flame: 107 ppm for SR=0°92, with continued reduction at deeper staging,
70 ppm for SR=O°82 while maintaining high carbon conversion
efficiencies, >99%. These NOX emissions are low enough to satisfy the
LEBS emissions goal without additional NOX reduction techniques.
When the burner is operated with substoichiometric air flow and
overfire air, a larger fraction of the air can be fast admixed to the
pulverized coal without the risk of providing fuel lean flame zones. Under
these conditions both the volatile evolution and the heat release are fast
and consequently the residence time in the high temperature, density
stratified region of the fuel rich flame is increased resulting in lower NOX
concentrations.
It is noteworthy that the mode of burner operation to give
decreasing fraction of the burner air flow with increasing radial distance
is not suitable for fuel-lean burner conditions, it produces high conversion
of the fuel-N to NO and increases the NO emission.
Theoretical Considerations
The effort of achieving low NOX emission (100-150 ppm) from coal
flames involves the study of many chemical and physical processes taking
place during coal combustion, such as flame ignition, volatile evolution
and combustion, char particle burnout, chemical reactions of N-bearing
species and the physics of turbulent, reacting, particle laden flows.
Among these processes, only a few, can be considered in this paper. Our
attention will be focused on engineering design solutions which are based
on the following generalizations:
the rate of devolatilization and the volatile yield are strong
functions of the flame temperature,
the rate of fuel nitrogen release from coal particles is proportional
to the rate of volatile evolution and char combustion,
the conversion of fuel nitrogen to molecular nitrogen is kinetically
controlled in fuel rich regions of flames, and its rate is a strong function
of local stoichiometry and temperature,
As it is reflected in the above mentioned principles, the fate of fuel
-------
nitrogen is the primary focus of study since fuel nitrogen is known to be
the major source of NOX emission from coal flames. When the objective is
to reach ultra low NOX concentration (e.g., 40-60 ppm) attention also has
to be paid to the processes involving the production and destruction of
thermal and "prompt" NOX. For the purpose of this study it has been
sufficient to focus on the fuel nitrogen processes since prompt and
thermal NO can be controlled to a large extent by the same methods as
fuel NO, e.g., staged combustion.
Early ignition of coal flames is important for establishing a high
temperature core to effect the maximum rate of release of fuel nitrogen
from coal particles into a sub stoichiometric gas phase, and consequently
to lengthen the residence time for fuel nitrogen conversion. The ignition
process is controlled by the rate of volatile matter evolution, which
increases exponentially with increasing temperature.
Gradual addition of oxygen to the evolved volatile matter is also
important for burning the fuel as it becomes available during the ignition
process. With increasing air to coal mass ratio one can generate more
heat, elevating the local gas and particle temperature. Part of this heat
will be fed back to the flame front by flame radiation and gas
recirculation and will enhance the igition process. If, however, too much
oxygen is provided, the mean stoichiometric ratio increases which, in
turn, can increase the amount of fuel nitrogen converted to NOX. It
follows, that there exists an optimal air to coal mass ratio, at which the
evolution of fuel nitrogen is fast due to high temperature, while the
surrounding gas phase is sub-stoichiometric, and favorable for the
conversion of fuel nitrogen to molecular nitrogen. This air to coal ratio is
undoubtedly coal dependent. The RSFC burner is capable of approaching
this optimum coal-air mixing history by the suitable combination of the
radial distributions of the combustion air mass flow and of the swirl
velocity.
In the second mode of operation described above, the well stirred
mixture of transport and primary air with coal creates a nearly uniform
distribution of stoichiometric ratio and temperature, which can be
optimized to obtain the highest rate of fuel nitrogen conversion to
molecular nitrogen. Both the heat release and volatile evolution are fast,
and, consequently, the residence time and temperature in the stratified
region of the fuel rich zone are increased resulting in lower NOX
concentrations.
-------
Parametric Studies
The experiments were conducted with standard grind Illinois #6 coal
(43% vol., 9% ash, 12148 Btu/lb). The Radially Stratified Flame Core
(RSFC) burner (Figure 1) was used in the studies. It has three annuli
(primary, secondary and tertiary) and independent adjustable swirl
generators can be used to set different input conditions such as the
amounts of air flow through the primary, secondary and tertiary annuli
with variable axial and tangential velocity to obtain optimum radial
stratification for a sufficient axial distance or residence time.
Experiments were performed to investigate the effect of burner settings
and burner air distributions on the NOX formation and carbon conversion.
The effect of turndown on exit NOX emission was investigated by
increasing the thermal input from 1.0 MW to 1.8 MW for the same burner
air distribution. The overall and local stoichiometries were changed by
adding air through the burner or the over fire section of the combustion
tunnel. The transport air to coal mass ratio was increased from 1.38 to
1.82, for a constant thermal input. Different flame characteristics were
also investigated by varying the primary, secondary and tertiary air
fractions and swirl numbers, respectively. External staging was achieved
by introducing overfire air into the combustion chamber at two axial
positions, 10.5 and 13.5 burner diameters downstream of the flame. (The
burner diameter is 0.30m.)
The measured exit NOX concentrations are plotted in Figure 2 as a
function of the first stage stoichiometric ratio for the two modes of
operation: unstaged and staged. The uncontrolled non-staged flame
produced about 700 to 900 ppm exit NOX concentration. This was reduced
to 217 ppm in the radially stratified flame. The minimum exit NOX
concentration, 55 ppm NOX at 3% 02, was obtained by a staged RSFC burner
flame with a stoichiometric ratio in the primary stage of 0.6.
A mildly staged flame was chosen as the baseline for investigating
the effect of transport air to coal mass ratio, thermal input, exit 62
concentration, air preheat and the location of the over fire air injection
(first stage residence time) on NOX emission. The details of the baseline
flame are given in the second data column of Table 1.
The heat input was 1.5 MW and the overall and burner stoichiometric
ratio was measured as 1.13 and 0.92, respectively. The total air flow rate
was 1460 Nms/h. The transport air fraction of the total air was 14.7%.
-------
The coal gun was recesses from the flush position by 5 inches.
High primary air flow rate and swirl number were used to promote
early and uniform mixing between coal particles and primary air. The
secondary air flow rate was set at the minimum necessary to cool the hot
primary barrel and to avoid fly ash deposition on the primary and
secondary barrels. The tertiary air fraction was se to 18.5% of the total
air and was physically separated from the initial flame front by the
secondary air annulus.
The over fire air was introduced X/D=13.5 burner diameters
downstream of the flame through an injection bank consisting of two,
triple air nozzles. The inner diameter of each of the six nozzles was 1
inch, producing a high momentum air jet which enhanced mixing between
the over fire air and the combustion products of the first stage.
Measurements of exit concentrations of the main gas components,
O2, CC>2, CO and NOX were carried out in steady state conditions. Ash
samples were also taken in isokinetic conditions at the exit of the
furnace. For the base case, the CO concentration was low, about 33 ppm.
The high CO2 concentration indicates good carbon conversion with low
excess air. The carbon conversion was determined to be 99.52% (from the
carbon content of the fly ash) by using a Perkin-Elmer analyzer. The exit
NOX concentration was 107 ppm at 3% O2.
A parametric study was carried out using the flame described above
as a base case. The effect of transport air to coal mass ratio (T/C) was
determined by increasing the fraction of transport air. The primary or
tertiary air flows were adjusted accordingly to keep a constant burner
stoichiometric ratio. The exit NOX concentrations were measured after
reaching steady state conditions in each case (Fig. 3). With no
compensation on the burner air, the exit NOX concentration increased from
about 105 to 150 ppm, when the T/C was increased from 1.4 to 1.8. The
higher NOX concentrations are due to the higher burner stoichiometric
ratio. This considerable increase of NOX concentration (50%) indicates
that the burner stoichiometric ratio has a very strong impact on NOX
formation. When the tertiary air flow was adjusted (reduced) to
compensate for the additional transport air, the exit NOX concentration
decreased from about 105 ppm to 80-85 ppm. These results indicate that
NOX exit concentrations are very sensitive to the mass flow rate of
tertiary air. On the other hand, when the compensation was made on the
-------
primary air flow, a negligible difference was measured in the exit NOX
concentration. There was little change in the flame shape during these
experiments, even when the primary to transport air mass ratio was
changed significantly, indicating stable flame characteristics.
The available residence time for the fuel nitrogen conversion in the
fuel rich zone is an important design parameter for low NOX boilers. The
effect of the first stage residence time was investigated by injection of
air at two separate axial positions -10.5 and 143.5 burner diameters
downstream of the flame (Fig. 4). By changing the distribution of air
between the injection ports, the mean gas residence time in the first
stage of the combustion chamber is easily varied. The maximum variation
of residence time achievable with this configuration is estimated at about
600 msec.
During these experiments, the over fire air split between the two
inject stations was varied for 5 different cases and the exit gas
composition was determined. Introducing all the over fire air at the
upstream axial location caused the exit NOX concentration to increase
from 110 ppm to 200 ppm. The correlation between the two variables is
linear. This experiment underscores the importance of the mean gas
residence time in the fuel rich zone in obtaining low NOX emission.
The effect of the thermal input was investigated by increasing the
coal flow rates while maintaining a constant (transport + primary
air)/coal mass ratio. The fraction of tertiary air was allowed to change.
The burner SR decreased from 0.92 to 0.82 when the thermal input was
increased from 1 MW to 1.8 MW. The variation of exit NOX as a function of
thermal load and burner stoichiometry is shown in Figure 5. At lower load
(1 MW) the burner stoichiometric ratio was 0.92 and at this condition, the
exit NOX concentration was 94 ppm corrected to 3% 62. At maximum load
(1.8 MW), the NOX concentration decreased to about 70 ppm, but the
stoichiometric ratio was also lower (0.82). Thus, the exit NOX
concentration is relatively insensitive to thermal load within the range of
investigation.
The exit C>2 concentration was also varied between 2.0% and 4.5%,
both at the staged (SFU0.92) and non-staged conditions. In thge non-
staged flame, the exit 62 concentration was varied by increasing the
fraction of tertiary air. For the staged case, the over fire air fraction
was increased, and consequently the burner stoichiometric ratio was
7
-------
constant. The results show an increase of exit NOX concentration from
217 ppm to 310 ppm for the non-staged flame (Fig. 2). No change was
detected for the staged condition. For the latter case, the exit NOX
concentration was found to be between 103 and 107 ppm, (corrected to 3%
O2). The air preheat was also increased at the staged condition (SR=0.92)
from 500 K to 562 K but negligible change (96 to 98 ppm) was detected in
the exit NOX concentration.
Detailed Flame Characteristics
Detailed flame measurements were performed in order to gain
insight into the governing mechanisms of NOX formation. Only a few of
those results will be discussed here. The characteristics of the three
flames studied are given in Table 1. Two of these flames were staged
flames and one was an unstaged flame. The study of staged flames would
help determine the effect of staging on the NOX formation and the
comparison between staged and unstaged flames would help explain the
difference in the exit NOX concentrations for these flames . The flame
designated as W OFA II in Table 1. has been chosen as the baseline staged
flame. The flame has been discussed in the preceding section.
Visual observation of the flames provided valuable qualitative
information about the flame length, the radial flame stratification and the
depth of intrusion of the transport air jet. The visual features of the two
staged cases are similar, the quarl is filled with combustion products.
There is no transport air/coal jet visible on the center line of these
flames.
The non-staged flame W/O OFA, had different attributes. The burner
was operated with a different air distribution for this flame. There was a
smaller amount of primary air, a higher amount of secondary air an
especially high fraction of tertiary air (see Table 1.). The shape of the
flame is different as it does not fill the quarl and the tertiary air is
clearly separated from the flame front (radially stratified). At a distance
of about three burner diameters downstream of the burner the flow
diverges as it approaches a stagnation zone in the flow. This stagnation
zone marks the end of the radially stratified core region. Downstream of
this, vigorous mixing between the tertiary air and the combustion
products occurs.
The different flame shapes indicate that the interaction of the coal
laden transport jet with the primary air jet is very important. For the
8
-------
staged cases, the angular momentum of the primary air is much higer than
that of the non-staged stratified case. It follows that the interaction of
the transport air jet is signifcantly different in the two cases. Both the
visual observation and concentration measurements of solid samples along
the axis support this conclusion. In the staged cases, the angular
momentum of the primary air was sufficiently powerful to create a
stagnation point within the primary barrel, which, in turn, resulted in a
well mixed fuel rich primary air-coal flow to emerge from the burner. In
the non-staged case, on the other hand, stagnation did not occur as early
due to the weak angular momentum of the primary air (low swirl number
and low air fraction).
The exit gas composition was measured for each case and the
results are summarized in Table 1. Low CO concentrations were detected
even for the higher staged flame. The carbon conversion was calculated
from the carbon content of the sampled fly ash. The NOX concentrations
are also given in Table 1. The lowest NOX emission of these flames (70 pm
at 3%O2) was obtained from the staged flame with SR=0.82. This
increased to 107 ppm at 3%O2 for the mildly staged flame and to 217 ppm
@3%C>2 for the unstaged flame.
Gas temperature measurements were made for each flame. The
measurements were made using a water cooled suction pyrometer
equipped with a triple shield and operated at a high gas velocity, so as to
minimize errors due to radiation. Axial temperature profiles taken for the
two staged and one unstaged test flames are shown in Fig. 6. The
advantages of the burner operating mode for staged combustion (W OFA),
i.e., the increased fraction of the combustion air as strongly swirling
primary burner air flow, can be seen by the steep temperature rise near
the burner. This is consistent with high volatile yield and fast decay of
fuel-N content of the solid samples along the distance from the burner
(Fig. 7).
Early ignition and partial combustion in a well stirred zone close to
the burner produces a faster initial decay of the solid carbon as shown in
Fig. 8. The carbon burnout continues along the flame to reach high carbon
burnout efficiency, in excess of 99.5% for both the staged and unstaged
flame.
The NO and NOX concentrations measured along the flame W/O OFA
are shown in Fig. 9. The large differences between NO and NOX levels close
to the burner are in the flame zone where fuel bound nitrogen, HCN,
-------
show up as NOX. As the NO and NOX values merge at around x/D=10 an NOX
level very close to that of the emission from the flame > 217 ppm (3%O2),
is established. Under staged combustion conditions (W OFAII) the NOX
decays faster to an axial distance from the burner of x/D=6 and there is no
further change to that level, 107 ppm (3%O2), to the end of the flame (Fig.
10).
Conclusions
Experimental work was carried out at the MIT Combustion Research
Facility with the MIT Radially Stratified Coal Burner with an Illinois #6
coal. The highlights of the results are:
217 NOX(@3%C>2) concentration with unstaged stratified flame at
1.08 stoichiometric ratio and 99.54% carbon conversion,
107 NOX (@3%C>2) concentration with staged flame at 0.92 first
stage stoichiometric ratio and 99.52% carbon conversion,
70 NOX (@3%O2) concentration with staged flame at 0.82 first staged
stoichiometric ratio and 99.3% carbon conversion.
The parametric study showed relative insensitivity of exit
concentrations to thermal load (between 1 and 1.8 MW) air preheat (510-
560 F) and exit O2 concentrations (2.5-4.5%). The effect of transport air
on exit NOX concentration depended on where the air compensation was
applied at staged conditions.
It is noteworthy that the very low NOX emissions form both the
mildly staged and more deeply staged flames were obtained with a radial
air flow distribution in which more than 50% of the total air flow was
introduced through the transport plus primary air conduits. This has
resulted in the maximum yield of volatile matter, stable ignition and a
steep initial temperature rise, factors which are conducive to the
production of a hot flame core, a prerequisite of low-NOx operation.
The in-flame measurements point to engineering solutions leading to
further improvement of the RSFC burner performance. It seems that
reduction of ignition delay resulting in steep temperature rise near the
burner are the key parameters of low NOX emission.
Acknowledgments
The authors are grateful for the stimulating discussions with the
B&W visiting team, Jennifer Sivy and John Koslosky. They thank Tony
10
-------
Mayne, the DOE-PETC Program Manager, for his help and encouragement
throughout the program. The valuable assistance of the MIT-CRF technical
staff, Don Bash, Bill Mason, Mark James and Richard Gray is gratefully
acknowledged. Thanks are due to Ms. Julie-Marie Andersen for the
preparation of this manuscript. Financial support from the B&W DOE LEBS
Program is acknowledged with thanks.
References
1. Toqan, M.A., J.M. Beer, P. Jansohn, N. Sun, A. Testa, A. Shihadeh, and
J.D. Tear, Low NOX Emission from Radially Stratified Natural Gas-Air
Turbulence Diffusion Flames. Presented at the 24th Symposium
(International) on Combustion, Sydney, NSW, Australia, July 5-10, 1992.
2. Shihadeh, A.M. Toqan, P. Lewis, and J.M. Beer, "Low NOX Emission
from Aerodynamically Staged Oil-Air Turbulent Diffusion Flames,"
Presented at Combustion Institute - Eastern States Meeting, Sept. 1993.
11
-------
TERTIARY AIR
SECONDARY AIR
PRIMARY AIR •*
COAL + TRANSPORT AIR
QUART-
BURNER QUARL
SWIRL GENERATOR
Fig 1. Schematic of the MIT-RSFC Burner (U.S. Patent Pending)
-------
a
o
to
2 a
X!
Fig. 2
1000
100
10
Effect of Stoichiometric Ratio on Exit NOx
AH Experimental Data
Cases of
Special Interest
* STAGED FLAME
D STRATIFIED FLAME
A FUEL-LEAN UNCONTROLLED
FLAME
0.5
0.6 0.7 0.8 0.9
1.0
1.1
1.2
1.3
1.4
1.5
1.6
1.7
First Stage Stoichiometric Ratio
-------
Characteristics of the Three Reference Cases
Heat Input (MW)
Overall SR
First Stage SR
Total Air (Nm3/h)
Transport/Coal Ratio
Transport %
Primary % / Swirl No.
Second. % / Swil No.
Tertiary % / Swirl No.
OFA %
CO (ppm)
CO2 (%)
O2 (%)
NOx (ppm @3% O2 )
Carbon Conv. (%)
wOFAl
1.5
1.21
0.82
1628
1.5
13.5
37.2 72.35
5.3/2.52
11.6/1.66
32.4
45
15.5
3.7
70
99.3
w OFA II
1.5
1.13
0.92
1460
1.5
14.7
44.8/2.35
6.9/2.52
18.5/1.66
15.6
33
15.9
2.0
107
99.52
w/o OFA
1.5
1.08
1.08
1446
1.5
15.1
12.7/1.21
27.2 /1.63
45 /1.66
0
30
16.7
1.5
217
99.54
-------
Fig. 3 Effect of Transport Air to Coal Mass Ratio
on Exit NOx Concentration
With OFA, SR=0.92
No Compensation
Compensation on Tertiary
Compensation on Primary
70
1.5 1.6 1.7
Transport Air to Coal Ratio
-------
(§)
B
ex
O
2;
300
0.0
0 4 Dual Overfire Air Injection
x/D=10.5 and x/D=13.5 (SR=0.92)
0.2 0.4 0.6 0.8
Fraction of OFA injected at x/D=10.5
i.o
-------
Fig. 5 Effect of Thermal Input on Exit NOx
c
o
*FM
-a
9
^
t-»
•a
Thermal Input (MW)
-------
Fig. 6 Axial Profiles of Temperature
1600
O)
J-,
O)
cx
E
O)
H
1500
1400
1300
1200
110
OFA
Injection
a
;?,
• w OFA I (SR=0.82)
A vvOFAII(SR=0.92)
n w/o OFA
10
12
14
16
Axial Distance (x/D burner)
-------
Fig. 7 Axial Profiles of Nitrogen Content of Solid Samples
e
*-»
c
o
-------
o
GO
c
o
90
80
70
60
50
40
30
20
10
0
8
10
12
14
16
18
20
Axial Distance (x/D burner)
Fig. 8 Axial Profile of Carbon Burnout
-------
Fig. 9 Axial NO and NOx Profiles
Without OFA
Cfl
B
01
fi
OJ
y
c
o
O
O
£
c
cd
10000
1000
10
n
~Ch
a
10
12
14
a NO
• NOx
16
18
20
Axial Distance (x/D burner)
-------
Fig. 10 Axial NO and NOx Profiles
With OF A, SR=0.92
10000
E
a,
a*
C/}
C
+-»
-------
DEVELOPMENT AND TEST OF DIFFERENT METHODS TO IMPROVE
THE DESCRIPTION OF NOX EMISSIONS IN STAGED COMBUSTION
A. Brink, P. Kilpinen, M. Hupa
Abo Akademi University
Combustion Chemistry Research Group
FIN-20520 Turku, Finland
L. Kjaldman
Technical Research Centre of Finland
VTT Energy
P.O.BOX 1604
FIN-02044 VTT, Finland
K. Jaaskelainen
Imatran Voima Oy
HN-01019 IVO, Finland
Abstract
Two methods to improve the modeling ofNOx emissions in numerical flow simulation of
combustion are investigated. The models used are a reduced mechanism for nitrogen chemistry in
methane combustion and a new model based on regression analysis of perfectly stirred reactor
simulations using detailed comprehensive reaction kinetics. The applicability of the methods to
numerical flow simulation of practical furnaces, especially in the near burner region, is tested
against experimental data from a pulverized coal fired single burner furnace. The results are also
compared to those obtained using a commonly used description for the overall reaction rate of
NO.
Introduction
Air and fuel staging are effective techniques to reduce the NOX formation. These techniques rely
on the fact that under fuel rich conditions and suitable temperature NO previously formed is
reduced to molecular nitrogen. Consequently, to model staged combustion successfully, the fuel
and nitrogen chemistry must be well described also at fuel rich conditions.
In applications where computational fluid dynamics (CFD) is used to model combustion, an often
used approach is to neglect reaction kinetics and assume that the combustion rate is limited by the
mixing rate. The prediction of the NOX level is, however, dependent on a correct description of the
-------
fuel chemistry. The simplest level of reaction kinetics for fuel oxidation is to assume that the
reaction rate can be described with a single irreversible reaction. The exclusion of the
intermediates CO and H2 formed in the hydrocarbon breakdown leads to an overprediction of the
heat release1 In fuel lean cases, it is possible to obtain a good description of the fuel oxidation
assuming a mixing limited breakdown of the hydrocarbons to CO and H2O, followed by a mixing
or kinetically controlled oxidation of CO. In the literature, more complicated descriptions of the
fuel oxidation process can be found2-3 Several expressions for the global reaction rate of the CO
oxidation have been proposed1.
For simple hydrocarbons, e.g. methane, comprehensive schemes based on elementary reactions has
been proposed. Elementary reactions leading to the formation of NO can also be included in these
schemes. In CFD analysis of practical applications the direct use of such full mechanisms,
consisting of hundreds of elementary reactions, is computationally too demanding. Using the
strategy outlined by Peters and co-workers4-5, it is possible to reduce the comprehensive
mechanism to only a few formal reactions. One such reduced mechanism is the reduced four-step
mechanism for nitrogen chemistry in methane combustion derived for the perfectly stirred reactor
environment6 The use of a reduced mechanism restricts the number of species for which transport
equations must be solved to a number that can be handled in CFD applications.
The effective rates of the homogenous gas phase reactions in combustion are controlled mainly by
the mixing caused by turbulence. For reactions with a rate comparable to or slower than the
mixing rate, it is essential that the turbulence-chemistry interaction is taken into account. Several
methods for this have been put forward. One such model is the eddy dissipation concept7. Here,
the combustion is assumed to take place in isolated regions of the fluid, so called fine structure.
Within the fine structure the reacting components are mixed on a molecular level. In the model, it
is assumed that these parts can be modeled as perfectly stirred reactors.
In pulverized coal combustion, not only the homogenous gas phase reactions are of importance.
For successful modeling, detailed information about the pyrolysis stage and the ignition of the char
particles, among others, are also required. Best results are obtained if empirically determined data
are available. Unfortunately, the detailed composition of the pyrolysis gas is usually unknown and
the necessary data is difficult to obtain. Consequently, approximate expressions describing these
processes are often applied8'12.
The purpose of this work was to improve the description of homogenous fuel and the nitrogen
chemistry used in CFD simulations of staged combustion. Two models are presented. As a first
test case for their applicability to CFD simulations, a 2.5 MW pulverized coal fired single burner
furnace was simulated.
The homogenous gas phase models
A reduced mechanism for nitrogen chemistry
A reliable description of the NOX chemistry is dependent on a good description of the fuel
chemistry, including the heterogeneous processes. For the homogenous chemistry, among the
-------
most complete descriptions possible to incorporate into CFD are reduced mechanisms. The formal
reactions obtained by Glarborg et al. for methane combustion in a perfectly stirred reactor are
given below6,
CH4 + 2H + H20 C0 + 4H2
CO + H2O ^ CO2 + H2
3H2
where M represents any third body. The reaction rates for the formal reactions can be calculated
from the sum of the rates of certain important elementary reactions. Using steady-state balances
for N, NH, NCO and CN, Glarborg et al. obtained the reaction rates for HCN and NO in methane
combustion. The mechanism was developed for high temperatures ( T > 1500 K) and for fuel lean
to moderately fuel rich conditions, 0. 7 < A < 1. 4, where A is the air to fuel ratio.
A simplified model for hydrocarbon oxidation
In the literature, a number of proposals on simplified fuel oxidation descriptions can be found. In
fuel rich regions, a two-step description with only CO as an intermediate leads to an
overprediction of the local temperature. Omitting H2 also affects the predicted air to fuel ratio.
One of the difficulties encountered when including H2 is to decide the proportion between H2 and
H2O formed during the hydrocarbon breakdown. Here, a new approach has been used were the
proportions have been calculated from the global equilibrium13. Using this approach, the oxidation
of a hydrocarbon can be described by
CnHm +(*+a-)02^nCO + (--a-] H2 + a- H2O
Z* T" \_ £-i £i J £
CO + -02 - >CO2
where a is the ratio between the mole fraction of H2O to the sum of the mole fractions of H2O and
H2 obtained from the equilibrium composition. If H2 and H2O are present in the fresh mixture,
these too are assumed to form H2 and H2O in the same proportions. The oxidation rate for CO
given by Howard et al. is used to describe the second step14. The oxidation of H2 at fuel lean
condition is assumed to be mixing limited.
A regression-based NOX model
The models mainly used for calculating the overall reaction rates for the NO formation do not
contain mechanisms accounting for the reduction of NO to molecular nitrogen at fuel rich
conditions15-16. Numerical modeling of the destruction and formation of NO in a perfectly stirred
reactor with the scheme given by de Soete reveled surprisingly large deviations from the results
obtained using the comprehensive mechanism. This can also be seen in Fig 1. Therefore, the
development of a completely new model capable of modeling the NOX chemistry under conditions
encountered in staged combustion and applicable to CFD simulations was motivated.
-------
From the calculations with the full mechanism, an abrupt change in the concentration profile was
obtained for short residence times. This abrupt change was due to the ignition of the system. Much
of the conversion of the nitrogen containing species took place during the ignition. Longer
residence times did not affect the conversion grade to the same extent. Since it was obvious that
the radical chemistry played a significant role, accurate predictions of the radical concentrations
would be needed in the Arrhenius-type rate expressions. Since the fuel oxidation process usually is
described in a very simplified manner, it was realized that a regression based model for the NOX
chemistry would be a way to overcome this problem.
The results obtained from calculations on a perfectly stirred reactor with the comprehensive
mechanism were used as reference data for the regression model. From the simulations with the
comprehensive scheme it was noticed that some simplifications could be made. First, the
conversion grade for the nitrogen containing species included in the model. ,i.e. NO, HCN and
NH3, are to a great extent independent of the incoming concentrations. It was also noticed that the
only important synergistic effect in the reactions could be found between NO and HCN. Using
these approximations it was possible to apply regression analysis to construct expressions for the
conversion of NO to NH3, HCN and N2, and comparable expression for the conversion of HCN
and NH3 separately. A regression expression was included to account for the synergetic effect
noticed. This procedure was repeated for five discrete residence times, i.e. 0.1 ms, 1 ms, 10 ms,
100 ms and Is. For intermediate residence times the conversions can be obtained by means of
interpolation. The regression analysis was done using the backward elimination technique
requiring a 95 % significance for the independent variables retained. As independent variables
were the air to fuel ratio, the temperature, second order terms of these, as well as products tested.
Best results were obtained excluding the point related to non-ignited systems from the reference
set and using a separate regression equation for estimating the ignition time during the prediction
step. An example of the accuracy of the regression based NOX model is shown in Fig 1.
log residence time [ s ]
log residence time [ s ]
Figure 1. To the left, the mole fraction of HCN remaining as a function of residence time. To the
right, the mole fraction of NO formed. The incoming gas mixture contained 360 ppm HCN. The
temperature was 1560 K and the stoichiometric air to fuel ratio, X,was 1.1. The fuel consisted
mainly of methane.
-------
The reference data were calculated varying the temperature from 1300 to 1850 K and the air to
fuel ratio from 0.5 to 1.5. The fuel was assumed to consist of mainly methane. Later, additional
studies were made to investigate the significance of the fuel composition. The findings were
incorporated to the model through a parameter describing the fraction of the fuel consisting of
hydrocarbons.
When using the regression based NOX model, it is important that the conditions respond to those
the model was set up for. This restricts the use of the regression-based NOX model to systems
were the main part of the NOX is formed from fuel-//, and the thermal-M? can be neglected.
The applicability of the chemical schemes to flow simulations
The chemical models presented above have proven to model the gas phase reactions satisfactory
for perfectly stirred reactor conditions6'13. It is difficult to assess the accuracy of models for the
fuel and nitrogen chemistry from a multi-dimensional flow simulation of a pulverized coal
combustion. The near burner fields of temperature and concentrations of the main species depend
strongly on e.g. the pyrolysis and the char ignition which in practical applications usually are
inadequately know. The combustion of the pyrolysis products and the formation of NO from the
nitrogen released from the fuel occur mainly in the near burner region. Consequently, the role of
the present simulation in the assessment of the chemical reaction models is in the first place to see
the applicability of the models to CFD analysis. Key questions are then the influence of the
chemical submodels on the convergence and computing time, and their range of validity with
respect to prevailing conditions in the furnace.
For the flow simulations a computational environment ARDEMUS, developed by VTT and
Imatran Voima Oy for furnace analysis, was used17. The turbulence-chemistry interactions were
modeled using the eddy dissipation concept7. With this model it is possible to describe the
influence on the mean reaction rate from mixing, and from chemical reactions described with a
multi-step reaction mechanism. A multiple-time scale turbulence model was used for the flow
simulations18. A closer description of the submodels used can be found elsewhere13-17. As a first
test case a 2.5 MW furnace located at the International Flame Research Foundation (DFRF) was
used. Extensive measurements made by the IFRF are available as well as a detailed description of
the furnace and the operational parameters19-20.
Results from the simulations
The experience from the present study indicates that the investigated models can be used in CFD
simulations. With the reduced mechanism for nitrogen chemistry in methane combustion, the main
species can be calculated separately from NO and HCN. For the main species a six-fold increase in
the computational time compared to a two-step description of the fuel oxidation was obtained.
The improved fuel oxidation mechanism with H2 and CO as intermediates resulted in a only slight
increase of computational time. In Fig 2 are the oxygen and temperature distributions obtained
with the overall mechanism shown. In Fig 3 is the sensitivity with regard to the pyrolysis models
and ignition model demonstrated. Here the fuel chemistry was described with a two-step
-------
irreversible reaction assuming that CO is the only intermediate. The sensitivity of the near burner
predictions to the description of the pyrolysis has been discussed by Costa et alV.
..-v.*w*(ir,>" •-••"•:" •
.00
.02
.03
.05
.06
.08
.09
.11
.12
.14
.15
.17
.18
.20
.21
.23
342
462
583
704
825
945
1066
1187
1308
1428
1549
1670
1790
1911
2032
2153
Figure 2. To the left, the mass fraction of oxygen in the near burner region obtained with the
global mechanism including H2 and CO. To the right, the temperature distribution in Kelvin.
.00
.02
.03
.05
.06
.08
.09
.11
. 12
.14
.15
.17
.18
.20
.21
.23
.00
.02
.03
.05
.06
.08
.09
.11
.12
.14
.15
.17
.18
.20
.21
.23
Figure 3. To the left, the oxygen distribution in the near burner region obtained using a two-step
irreversible oxidation scheme with CO as the only intermediate. The description suggested by
Solomon et al. for the pyrolysis rate was used8. The particles were assumed to ignite as they reach
1000 K. To the right, the same description of the gaseous fuel chemistry has been used, but the
pyrolysis was described according to Fu et al.9, and Zhang et a/.10. For the ignition, a more
detailed model was also used"'12. The concentrations are expressed as mass fractions. The
boundary of the internal recirculation zone is indicated.
-------
Since NO and HCN can be calculated subsequently to the flow, temperature and species
distribution, the calculation times needed for the nitrogen containing species did not cause any
concern. In the flow simulations, the local conditions were such that the stoichiometry range for
the reduced mechanism was partly exceeded. With the regression-based model, similar problems
were encountered. Especially the local HCN concentrations were outside the range for the
regression-based NOX model. In Fig 4, the NO and HCN levels calculated with the regression-
based model are shown.
4.1E-17
6.0E-04
1.2E-03
1.8E-03
2.4E-03
3.OE-03
3.6E-03
4.2E-03
4.8E-03
5.4E-03
6.OE-03
6.6E-C3
7.2E-03
7.8E-03
8.4E-03
9.OE-03
Figure 4. To the left, the NO distribution, and to the right the HCN distribution obtained with the
regression-based NOX models. The concentrations are expressed as mass fractions. The
combustion of the pyrolysis gases has been modeled with the global model including H2 and CO as
intermediate. The pyrolysis rate has been modeled according to Solomon el a/.8.
The same case was also simulated with the de Soete model. To account for the thermal NO
production, a model for the Zeldovich mechanism was added. Depending on the assumptions for
pyrolysis, char ignition and influence of turbulence on the mean reaction rate, calculated values at
the furnace outlet of 460 ppm, 745 ppm and 1040 ppm by volume in dry gas were obtained. These
values can be compared to the measured value of 820 ppm reported by IFRF. The NO and HCN
distribution obtained using the de Soete model is shown in Fig 5. The oxygen distribution used
with the de Soete model is shown to the right in Fig 3.
-------
l.OOE-20
1.11E-04
2.22E-04
3.34E-04
4.45E-04
5.56E-04
6.67E-04
7.78E-04
8.89E-04
l.OOE-03
1.11E-03
1.22E-03
1.33E-03
1.45E-03
1.56E-03
1.67E-03
2E-03
3.6E-03
3.9E-03
4.3E-03
4.7E-03
5.0E-03
5.4E-03
Figure 5. To the left, the NO mass fractions obtained with the de Soete mechanism. To the right,
the HCN. To account for the thermal NO production, the Zeldovich mechanism has been added.
The fuel oxidation was modelled using a two-step irreversible mechanism including CO as the only
intermediate. Only the near burner region is shown.
Discussion
The aim with this work was to improve the description of homogenous fuel and nitrogen
chemistry used in multidimensional flow simulations of pulverized fuel combustion including
staged combustion. The pulverized coal fired single burner furnace used as a first test case could
not reveal the applicability of the two models to reburn conditions. There was a fuel rich internal
recirculation zone in the near burner region, but the conditions here were not comparable to those
in staged combustion. Both models studied could be used without numerical complications. Since
most of the reactions governing the NOX release occur in the near burner region, the use of the
computationally more demanding reduced mechanism can preferably be restricted to this region.
The influence the description of the pyrolysis stage and the char ignition has on the predicted
distribution of the main species makes it difficult to draw conclusions regarding the accuracy of
the gas phase chemistry from the test case. When compared to a detailed reaction mechanism in a
well defined environment, the chemical models were found reasonably accurate. The local HCN
mass fraction close to the injection point of the pulverized coal was outside the range the
regression-based NOX model has been tested for. It can, however, be anticipated that an effective
conversion of HCN to NO occurs under the prevailing temperatures and air to fuel ratio
encountered in parts of this zone. The result obtained for NO with the de Soete model were in the
near burner region in closer agreement with the measurements made at IFRF. This simulation was,
however, done with a fuel oxidation model inaccurate for rich conditions.
-------
Although the simulations predicted rather low H2 concentrations and in a limited region, the effect
of incorporating H2 was greater than expected. The formation of H2 in the fuel rich near burner
region have influence on heat release, the local stoichiometry, the local absorption coefficient and
temperature, and finally on the behavior of the coal particles and on the flow field. This influence
is enhanced by the internal recirculation zone. Unfortunately, no information about the H2 levels
was available. Regardless of how the fuel chemistry or the heterogeneous processes were
modeled, the temperature and concentrations of the main species predicted at the outlet were
similar in all cases.
Conclusions
This paper shows that the reduced model for nitrogen chemistry in methane combustion as well as
the improved combustion and the new regression-based NOX models can be used in flow
simulations. The effect of the models on convergence was small and the computational times
tolerable. Due to the sensitivity of the results with respect to how pyrolysis and char ignition are
modelled, an improvement in the results from the flow simulation of the pulverized coal
combustion studied was not obtained. The addition of H2 as an intermediate combustion product
had a clear influence on the results for the near burner region. The results are, however,
encouraging, and further studies including staged combustion will be undertaken.
Acknowledgments
Financial support from Imatran Voima Oy, the Finnish National Combustion and Gasification
Research Program LJJEKKI2 and the Academy of Finland is gratefully acknowledged.
References
1. A.Y. Abdulla, D. Bradely, S.B. Chin, and C. Lam, C. "Global reaction schemes for
hydrocarbon oxidation." Oxidation Communications. Vol. 4, Nos .1-4, pp. 113-130 (1983)
2. W.P. Jones and R.P. Lindstedt, R.P. "Global reaction schemes for hydrocarbon
combustion." Combustion and Flame. Vol. 73, pp. 233-249 (1988).
3. D.J. Hautman, F.L Dryer, K.P. Schug, and I. Glassman. "A multiple-step overall kinetic
mechanism for the oxidation of Hydrocarbons." Combust. Sci. Technol. Vol. 25, pp.219-235
(1981)
4. Norbert Peters. Lecture Notes in Physics 241, Numerical and asymptotic analysis of
systematically reduced reaction schemes for hydrocarbon flames, Numerical Simulations of
Combustion Phenomenas, 1985, pp. 90-109.
5. G. Paczko, P.M. Lefdal, and N. Peters. "Reduced reaction schemes for methane, methanol and
propane flames," pp. 739-748, presented at the 21th Symp. (Int.) on Combustion, Munich,
West Germany, (August 1986).
6. P. Glarborg, N.I. Lilleheie, S. Byggst0yl, B.F. Magnussen, P. Kilpinen, and M. Hupa, "A
reduced mechanism for nitrogen chemistry in methane combustion," pp. 889-898, presented at
the 24th Symp. (Int.) on Combustion, Sidney, Australia, (July 1992).
-------
7. B.F. Magnussen, "Modeling of NOX and soot formation by the eddy dissipation concept,"
presented at 1 st Topic Oriented Technical Meeting of IFRF, Amsterdam, The Netherlands,
(October 1989).
8. P.R. Solomon, M.A. Serio, R.M. Carangelo, and J.R. Markham, "Very rapid coal pyrolysis"
Fuel. Vol. 65, pp. 182-193 (1986).
9. W. Fu, Y. Zhang, H. Han, and D. Wang, "A general model for coal devolitilazation." Fuel.
Vol. 68, pp. 505-510(1989).
10. Y.P. Zhang, J. Mou, and W.B. Fu, "Method for estimating final volatile yield of pulverized
coal devolitilization." Fuel. Vol. 69, pp. 401-403 (1990).
11. W.B. Fu, andT.F. Zeng, "A general method for determing chemical kinetic parameters
during ignition of coal particles.'" Combustion and Flame. Vol. 88, pp. 413-424 (1992).
12. W.B Fu, and E.Z. Zhang, "A universal correlation between the heterogeneous ignition
temperatures of coal char particles and coals" Combustion and Flame. Vol. 90, pp. 103-113
(1992).
13. A. Brink, P Kilpinen, M. Hupa, L. Kjaldman, and K. Jaaskelainen, "Mathematical modelling
of NOX emissions from staged combustors with reduced and regression-based nitrogen
chemistry" 26 p., presented at 11th Members Conference of IFRF, Amsterdam, The
Netherlands (May 1995)
14. J.B. Howard, G.C. Williams, and D.H. Fine, "Kinetics of carbon monoxide oxidation in post
flame gases." pp. 975-986, presented at the 14th Symp. (Int.) on Combustion, Pennsylvania
(August 1972).
15. G.G. de Soete, "Overall reaction rates of NO and N2 formation from fuel nitrogen." pp. 1093-
1102, presented at the 15th Symp. (Int.) on Combustion, Tokyo, Japan (August 1974).
16. J.W. Mitchell, and J.M. Tarbell, "A kinetic model of nitric oxide formation during pulverized
coal combustion." AIChE Journal. Vol .28, No. 2, pp. 302-311 (1982).
17. Lars Kjaldman, Numerical simulation of combustion and nitrogen pollutants in furnaces.
Espoo, Finland, Technical Research Centre of Finland, VTT publications 159, 1993, 132 p.
18. S-W. Kim and C-P Chen, C-P "A multiple-time-scale turbulence model based on variable
partitioning of the turbulent kinetic energy spectrum." Numerical Heat Transfer, Part B. Vol.
16, No. 2, pp. 193-211 (1989).
19. R. Weber, J. Dugue, A. Sayre, and B.M. Visser, "Quarl zone flow field and chemistry of
swirling pulverized coal flames: measurements and computations." pp. 1373-1380, presented
at the 24th Symp. (Int.) on Combustion , Sidney, Australia, (July 1992).
20. Measurements and computations of quad-zone flow field and chemistry in a pulverized coal
flame, Umiuden, The Netherlands, International Flame Research Foundation, IFRF DOC NO
F36/y/20, 1992.
21. M. Costa, S. Godoy, F.C. Lockwood, and J. Zhou, "Initial stages of the devolatilization of
pulverized-coal in a turbulent jet" Combustion and Flame. Vol. 96, No 1/2 pp 150-162
(1994).
-------
Session 7B
SCR/Hybrid
-------
SCR PERFORMANCE EVALUATION AND TROUBLESHOOTING ON A
NATURAL GAS-FIRED SOURCE
E. A. Mazzi, P.E.
Carnot
15991 Red Hill Avenue
Tustin, California 92680
and
J. E. Johnson
Applied Energy Services
20885 Placerita Canyon Road
Newhall, California 91321
Abstract
This paper presents an engineering evaluation of SCR performance on a natural gas-fired 110
MW gas turbine/heat recovery steam generation plant. The purpose of the evaluation was to
mitigate excessive ammonia slip from the system. This paper presents results of the engineering
assessment of the system design which culminated in a decision to relocate and redesign the
ammonia injection grid. The program entailed development and execution of field performance
tests for gas flow distribution, process temperature, and NH3/NOX uniformity. An evaluation
conducted by Carnot included: a technical assessment of anhydrous ammonia injection nozzles
in a custom-designed cold flow test apparatus, SCR efficiency at varied NH3/NOX mole ratios,
and assessment of injection grid design. The modifications were implemented by AES, and the
system currently operates within compliance limits of 7 ppm NOX and 20 ppm NH3, corrected
to 15 % O2. This is a "lessons learned" paper including useful insights for SCR design and
troubleshooting, emphasizing design features necessary to control ammonia slip. The results and
methodology are applicable to any SCR system, particularly on gas-fired sources.
Introduction
Applied Energy Services (AES) operates a natural gas-fired 110 MW gas turbine/heat recovery
steam generation (HRSG) plant in Placerita, California. The plant has two HRSGs, each
equipped with a selective catalytic reduction (SCR) system. The combustor utilizes steam
injection to control turbine outlet NOX to approximately 30 ppmc. In July, 1992 the SCR unit
on one of the boilers was shown to require excessive NH3 injection to maintain compliance with
7 ppm @ 15% O2 ("ppmc") NOX emissions from the stack. The NH3 slip was measured at 47
ppmc, while the permitted limit for the plant is 20 ppmc. This paper presents results of the
-------
engineering assessment of the system performance conducted by Carnot which culminated in a
decision to relocate and redesign the ammonia injection grid. The modifications were
implemented by AES, and the system currently operates within compliance limits for both NOX
and NH3 emissions.
Figure 1 is a diagram of the plant showing primary equipment and full load parameters typical
of plant operations at the time the SCR was evaluated. The unit consists two ABB Type 8
turbine/generator sets, and two HRSGs linked with a turbine exit crossover duct. The Steam
Electric (SE) boiler generates superheated steam for the steam turbine, and the Steam Injection
(SI) boiler is used to generate 85% quality steam for enhanced oil field recovery. The
proportion of exhaust gas directed to the SI boiler is varied to match demand for oil field
injection. During the time the SE boiler SCR system experienced problems, operating
requirements dictated that the SE boiler receive 40% more exhaust gas than the SI as shown in
Figure 1.
Figure 2 shows the layout of the anhydrous NH3 injection sparger piping. There were six
discrete cells referred to as North and South top/middle/bottom. The effect of separating the
SCR system into six cells was to create six semi-independent SCR systems. It should be noted
that there was no mechanism for biasing NH3 flow to various zones, thus NH3 distribution was
controlled by piping design and the injection nozzle orifice pressure drop characteristics.
Scope of Problem
The plant started up in 1988 and operated within compliance for both NOX and NH3. After
approximately two years, SCR performance deteriorated to the point where AES decided to
solicit bids for catalyst replacement. New catalyst was then installed within the existing reactor
space by a supplier other than the original manufacturer. Concurrent with this retrofit, attempts
were made to improve the ammonia injection grid (AIG) design, although the modifications were
not thoroughly documented. All cells contained 178 ft3 of NOX catalyst, except the bottom south
cell which was modified to hold 202 ft3. Source test results in mid-1990 showed the system
performing well with 8.4 ppmc NH3 slip from the SE stack. Over the course of the next two
years the NH3 slip became excessive and measurements in mid-1992 showed the average slip to
be 47 ppmc.
The diagnostic program began on a fast track due to regulatory concerns with the following
corrective action approach:
1. Perform "screening" tests on the SCR system to identify gross maldistributions of
NH3/NOX ratios. Concentrations of NH3 and NOX at available test ports were to be
measured. Simultaneous with the NH3 slip investigations, consideration would be given
to removing the CO catalyst as part of any potential design modifications. Removing the
CO catalyst was seen as a method of making space available for additional NOX catalyst
layers and to increase the mixing length between the AIG and the catalyst.
2. Based upon the results of the screening tests, a "detailed" test plan would be developed
and implemented. This was envisioned to entail more detailed mapping of NH3 and NOX
-------
DSEM-072
FUEL
(9270 scfm)
STEAM/FUEL
RATIO = 2.5
(44 MW)
02 =14.8% dry
Q = 285 Kdscfm
H20 = 13.8%
(6.1% w/o STEAM) ^
NO = 26 p/
1160SCFI-
(\
tomvDRY
Xxxxxxx
w? Dfty /\A/\/\/\/
9999999
iNH
3
\,
Q = 335 Kdscfm N
02 =
14.8% DRY
T = 310°F
NOx = 6.5 J
NO = 5.6 A
f S
STEAM (65 Klb/hr) STEAM (61 Klb/hr)
440 psig o 44Q p$jg
amb 575 F FUEL
V ^ ^ ^/ ^ V (9270 scfm)
COMBUSTOR -^olpa coolef COMBUSTOR
2 '21 1
A A
v YCOMPRESSORX V
71 [COMP. 70°F fcOMPl TT
2 1 1 2 279 Kscfm 1 J I G
— ^»J L — • •
r v
Q = 50 tatec/m
.
^
rk c/?oss OL/CT )
Y
STEAM/F
RATIO =
UEL
2.3
(44 MW)
,-^1
!-1
GEN. 1
-i J
02 =14.8% dry
Q = 285 Kdscfm
H20 = 13.5%
(6.1% w/o STEAM)
\
9999999 CU CATALYST >?9999999999
A/H^. /W / A . i » A
\AAAAA /vn3 "VJ- AAAAAAAA
9999999 NOx CATALYST '39999999999
r "*1 SE
— STEAM
J I TURBINE 1 J
V^ T^n VY
^T SE GE/V V^
(21 MW)
0 = 235 Kdscfm
02 =14.8% DRY
""\ /^^\ I = 290° F
E 1 1 SI \
\STACK) \STACKJ
Figure 1. Plant Block Diagram with Approximate Full Load
Operating Parameters atAES, Placerita
-------
DSEM-073
SOUTH SIDE
i
1
14-1/
>
6 sp
18-7
^
6"
\
!<•
Y
<
*"
<
^
VX
&.
/8"
y '^
11-1/2
A
<
-
-^ ^
V
*
t
f
f
I
^ [
-i r
TOP SOUTH f
8-1
>
14"
<
.. r
r :
i :
] :
i ;
] :
] :
1 TOP NORTH
-
-
-
-
-
r
t
f
t
[
t
t
MDDLE SOUTH
\
-, i-
] ^/— 3* SPARGER PIPE TYPICAL
]_ _J
i :
i ;
i r-
] ^
1 MIDDLE NORTH
:
^
-
I
t
__ t
__ t
-i
I
~i v
" [
i BOTTOM SOUTH
QC 0 •
12"
1
V
x±JL
)
AA
r-
]_ J
1 :
i :
i :
] j
, — . ,_
BOTTOM NORTH W
^ < XXXXXXXXXXXXXXXXXXXXAX/*
-
-
.
-4
1
NORTH SIDE
Figure 2. Ammonia/Dilution Air Injection Piping Network
as Viewed from NOx Catalyst
(all dimensions typical)
-------
concentrations, exhaust gas flow distribution measurements, temperature measurements,
catalyst activity core samples, and off line inspections.
3. Using the data from testing and inspections, it was anticipated that a design modification
could be developed and implemented.
There were other regulatory issues and time constraints involved with resolving the slip problem,
however only the technical investigation is described in this paper.
Screening Diagnostic Tests
Because information regarding the design modifications to the SE boiler's SCR system were
sparse, "screening" diagnostic tests were conducted by Camot. The primary objective of the
screening tests was to measure gross cell-to-cell distribution of NOX and NH3 using available
sample ports. Measurements performed included:
• NO, NOX and NH3 concentration stratification upstream of the SCR catalyst.
• NO and NOX concentration at the available test port downstream of the SCR.
• CO concentration upstream and downstream of the CO catalyst.
• Total Hydrocarbons (THC) upstream and downstream of CO catalyst, and at the stack.
(The CO catalyst was designed to oxidize hydrocarbons as well.)
• Stack NO, NOX, CO, and NH3 slip.
The tests were conducted over a two-day tune period, and results are summarized in Table 1.
Figure 3 shows the location and labeling of sample ports utilized for the testing. There were
no available sample ports upstream of the SCR for the middle cells, thus measurements were
made at the top and bottom cells only. The NH3 injection rate was set by the operator to
maintain constant stack NOX emissions just below 7 ppmc. In addition to verifying the previous
source test showing there was a slip problem, the screening tests showed the following results:
• Upstream of the SCR, the north bottom cell was relatively NH3 lean with a NH3/NOX
mole ratio of 0.75, while the other three cells measured were relatively NH3 rich with
an average ratio of 3.0.
i
• Turbine exhaust CO levels were already very low at 0.6 ppmc, with a comfortable
margin relative to the 2 ppmc stack limit. This demonstrated that removing the CO
catalyst could possibly be pursued to make room for additional NOX catalyst and/or
moving the AIG further upstream.
• The CO catalyst provided no benefits to reducing THC emissions. The flame ionization
detection (FID) measurements actually indicated an increase from 1.2 to 2.9 ppm. This
difference is within the scatter of data expected for this instrument's accuracy.
-------
Table 1
Screening Test Results*
Location
GT2 turbine exhaust
upstream of CO catalyst (downstream of cross duct)
upstream of NOx catalyst, bottom north cell (ports A-E)
upstream of NOx catalyst, top north cell (ports F-J)
upstream of NOx catalyst, bottom south cell (ports K-O)
upstream of NOx catalyst, top south cell (ports P-T)
downstream of NOx catalyst, bottom south cell (port U)
stack of SE boiler
Species
NO, ppmv
NOx, ppmv
CO, ppmv
THC, ppm CH4
NH3, ppmv
NO, ppmv
NOx, ppmv
NH3/NOx ratio
CO, ppmv
NH3, ppmv
NO, ppmv
NOx, ppmv
NH3/NOx ratio
CO, ppmv
NH3, ppmv
NO, ppmv
NOx, ppmv
NH3/NOx ratio
CO, ppmv
THC, ppm CH4
NH3, ppmv
NO, ppmv
NOx, ppmv
NH3/NOx ratio
CO, ppmv
NO, ppmv
NOx, ppmv
CO, ppmv
NH3, ppmv
NO, ppmv
NOx, ppmv
CO, ppmv
THC, ppm CH4
Measured
Value
25.7
31.7
0.58
1.2
20.9
24.6
27.7
0.75
0.13
92.6
24.3
28.7
3.23
0.06
86.0
27.0
30.2
2.85
0.09
2.9
90.2
27.8
31.4
2.87
0.08
4.8
5.2
0.08
43.1
5.6
6.5
0.21
2.4
All units are raw concentrations orTa dry basis. 02 measured at approximately 14.8% dry at all locations.
-------
DSEM-074
NH, /DILUTION AIR PIPES
SAMPLE PROBE __ /
UPSTREAM OF r~A-r*iv<:T
CO CATALYST CAT^LYST./
\ ilr
NOx 2" BALL
CATALYST (33-™~* .<..
EXHAUST
FLOW
Figure 3a, SE Boiler Sample Port Locations and Side/Sectional
View Looking at the
South Side as Viewed from Outside
NH, /DILUTION AIR PIPES
\f,~ SAMPLE PROBE
™-£r^T UPSTREAMOF
CO CATALYST
EXHAUST
FLOW
Figure 3b. SE Boiler Sample Port Locations and Side/Sectional
View Looking at the
North Side as Viewed from Outside
-------
Detailed Diagnostic Tests
Building upon the information garnered from the screening test effort, a "detailed test" plan was
developed and implemented by Camot. The basic idea was to obtain flow and NH3 distribution
within a ceU, adding to the gross cell-to-cell data obtained previously. These tests were designed
to utilize the 20 available sample ports located upstream of the NOX catalyst, and the 5 available
ports located downstream of the catalyst as illustrated in Figure 3. To the extent possible, these
tests were to include multiple-point composite traverses in each port for NH3, NOX, velocity, and
temperature. Further measurements of CO and THC emissions were not conducted.
Table 2 contains a summary of the NOX and NH3 measurements collected over a three-day time
period. Refer to Figure 3 for sample port locations and labeling. The SCR inlet NOX was
approximately constant due to constant combustor operating conditions including steam injection
rate. The NH3 injection rate was set by the operators to maintain constant stack NOX emissions
just below 7 ppmc. The SE boiler flue gas flow variations ranged from +4% to -9% from the
average of 5,507 scfs, while the NH3 flow varied from +4% to-13% from the average of 1,107
scfh.
Since measured values of NOX and NH3 were obtained over three days with the variations noted
above, values were normalized to the average SE boiler gas flow and NH3 injection rate. This
normalization, shown in the right columns of Table 2, was performed using equations (1) and
(2) as follows:
(Inlet NOX or Outlet NH3) * (actual scfs / 5,507) * (1,107 / actual scfh NH3) (1)
= Normalized Inlet NOX or Outlet NH3
(Outlet NOJ * (5,507 / actual scfs) * (actual scfh NH3 / 1,107) (2)
= Normalized Outlet NOX
This was done to normalize all data to approximately constant operating conditions including
stack NH3 slip. A more exact method would be to perform a NOX and NH3 mass balance on
each cell including the effects of varied space velocity through each cell and accounting for NO2
effects (i.e. 1.33 moles NH3 per mole NO2). However, all required parameters were not
measured for each cell and applying the above correction was better than none.
The emissions and flow measurements were then used to generate SCR performance parameters
shown in Table 3. As seen in the Table 3 data, different cells operated at different mole ratios
due to cell-to-cell NH3 maldistributions. Figure 4 illustrates the performance of different cells
graphically with NOX removal efficiency versus NH3/NOX mole ratio. A perfect system would
remove 100% of the NOX at a mole ratio of 1.0 (excluding NO2 effects) shown with the upper
curve in Figure 4. The system at AES was designed to operate at a mole ratio of 1.0-1.6,
depending upon space velocity and catalyst age, as shown with the "design" curve in Figure 4.
The "post-mod test" curve shown in this figure is explained later in this paper.
Although the screening tests identified cell-to-cell NH3 maldistribution, these detailed tests
showed that NH3 distribution within each ceU was unacceptable as well. The point by point data
showed NH3 variations by factors ranging from 1.6 to 3.6 in the three cells where measured.
-------
Table 2
Detailed Diagnostic Test Results
Normalized **
Cell/Location
SE (upst CO catly)
SE (upst CO catly)
SE Boiler Exhaust
SE Boiler Exhaust
SE Boiler Exhaust
SE Boiler Exhaust
Bottom South
Bottom South
Middle South
Bottom North
Bottom North
Bottom North
Middle North
Middle North
Top North
Top North
Top North
Bottom North
Top North
Bottom South
Top South
Bottom North
Bottom North
Bottom North
Bottom North
Bottom North
Top North
Top North
Top North
Top North
Top North
Bottom South
Bottom South
Bottom South
Bottom South
Bottom South
Meas
NOx
Port(s) ppm
Inlet 30.8
Inlet 28.3
Stack 6.5
Stack
Stack
Stack
U 5.9
U
V 3.1
X 18.9
X
X 13.9
Y 2.7
Y 5.1
Z 4.3
Z
Z 4.7
A,B,D,E
F,G,H,I,J
K,L,M,N,0
P,Q,S,T
A
B
C
D
E
F
G
H
I
J
K
L
M
N
O
Meas
N02
ppm
5.8
0.1
0.0
0.2
1.3
0.1
0.2
0.1
0.4
NH3
ppm
47.1
58.1
44.4
50.1
4.0
75.9
23.2
82.5
76.2
78.8
65.5
33.2
48.7
19.0
27.7
78.5
134.8
99.8
77.8
111.4
79.9
88.4
109.5
96.4
39.8
Exhaust
Flow
dscfs*
5,600
5,532
5,153
4,978
5,310
5,419
5,515
5,419
5,663
5,314
5,663
5,663
5,663
5,663
5,606
5,667
5,600
5,600
5,700
5,700
5,532
5,532
5,532
5,532
5,685
5,685
5,580
5,580
5,201
5,201
5,685
5,580
5,580
5,201
5,201
Total
NH3
scfh*
942
1,092
1,146
1,146
1,115
1,104
1,083
1,104
1,126
1,125
1,015
1,126
1,015
1,126
1,109
1,036
942
942
1,066
1,066
1,092
1,092
1,092
1,092
1,033
1,033
1,078
1,078
1,120
1,120
1,033
1,078
1,078
1,120
1,120
Stack
NOx
ppm*
6.4
6.4
6.4
6.3
5.9
6.1
6.1
6.4
6.1
6.5
6.4
6.4
6.4
6.4
6.0
6.2
6.3
6.2
6.3
Meas
NOx
ppm
30.8
28.3
6.6
6.2
3.2
19.2
12.7
2.8
4.7
4.4
4.4
Meas
NH3
ppm
41.4
49.3
41.3
49.9
3.7
75.0
26.9
95.9
79.7
82.4
64.9
32.9
48.3
18.8
29.8
84.5
136.5
101.0
70.7
101.2
85.9
89.5
110.8
87.5
36.1
Stack
NOx
ppm
5.5
7.3
7.6
6.8
6.1
6.1
6.6
6.5
6.2
6.1
5.5
5.5
6.1
6.1
6.1
6.1
7.0
6.1
7.0
* Data from Plant Instrumentation
** Normalized to: 1077
5507
average scfh NH3 (measured by plant instrument)
average dry scfs exhaust (measured by plant instrument)
-------
Table 3
SCR Performance Measurements*
(Using Screening and Detailed Test Data)
Cell
Top North
Middle North**
Bottom North
Top South ***
Middle South**
Bottom South
Catalyst
Volume
cubic feet
178.0
178.0
178.0
178.0
178.0
202.0
Average
Measured
Velocity
ft/s
21.1
19.8
18.5
17.0
18.2
19.4
Measured
Temp.
oF
689
690
691
695
690
692
Measured
Exhaust
Flow
dscfs
1,135
1,064
993
909
977
1,040
Space
Velocity
(dry)
1/hr
22,959
21,515
20,075
18,388
19,767
18,530
Inlet
NOx
ppmvd
28.7
30.8
27.7
31.4
30.8
30.2
Inlet
NH3
ppmvd
96
N/A
27
80
N/A
83
Inlet
NH3/NOx
Mole
Ratio
3.3
N/A
1.0
2.5
N/A
2.7
Outlet
NOx
ppmvd
4.4
3.8
16.0
7.0
3.2
6.2
NH3
Slip
ppmvd
75
N/A
4
N/A
N/A
50
NOx
Removal
Efficiency
85%
88%
42%
78%
90%
79%
* All NH3 and NOx measurements normalized to 5507 scfs (60°F) exhaust and 1077 scfh inlet NH3 -- see Table 2 and text).
** There were no inlet sample ports to these middle compartments; inlet NOx same as turbine exhaust; inlet velocity = average of top & bottom
*** The outlet sample port was not accessible; outlet NOx was estimated considering inlet NH3/NOx ratio and space velocity.
-------
Figure 4.
NOx Reduction Efficiency vs. NH3/NOx Mole Ratio
SE Boiler SCR
— Ideal SCR
Design
Detailed Tests
O Post-Mod Tests
0
0.5
1.5 2 2.5
Inlet NH3/NOx Mole Ratio
3.5
-------
Figure 4 also supported this conclusion since cells with mole ratios over 2.5 could not achieve
the removal efficiency expected for a mole ratio of only 1.5. Lastly, it is noted that the true
catalyst efficiency cannot be extracted from this data since the effects of non-uniform NH3 and
NOX mixtures entering the catalyst are not separable from the catalyst activity effects.
Off-line Inspection and AIG Nozzle Replacement
An inspection of the SCR modifications was planned for an upcoming outage scheduled to begin
within one week of completion of the detailed tests. In terms of the SCR performance
evaluation project, the purpose of the outage was to:
1. inspect the AIG to confirm the extent of design modifications made; the system was
converted from using threaded nozzles to a pattern of holes drilled in the sparger pipes;
the details of this modifications were not well documented.
2. make any modifications which appeared technically feasible and cost-effective.
Since the system originally had passed NH3 slip tests, it was tentatively planned to re-install
nozzles of the original design and spacing unless something during initial inspection dictated
otherwise. It should be noted that, besides the AIG modification, two other key changes had
taken place (1) the catalyst had been replaced with a new design by a different manufacturer,
and (2) the catalyst had aged.
An inspection on the first day of the outage by Carnot revealed that the AIG modifications made
were asymmetric. The bottom south cell had 1" pipe tees mounted on the 3" sparger pipes, with
three 1/16" holes spaced vertically in each tee as shown in Figure 5a. All other cells had a
spaced pattern of 1/8" holes drilled directly into the 3" sparger pipes as shown in Figure 5b.
The cell-to-cell AIG design variations explained the cell-to-cell maldistribution of NH3 measured
during the screening tests. The bottom north cell was NH3 lean most likely due to partial
plugging of the 1/16" holes. Although there were more 1/16" holes, the plugging created a
higher resistance than the 1/8" holes drilled in the spargers of the other five cells.
Furthermore, the point-by-point NH3 data shown in Table 2 indicates that this pipe tee design
did not distribute NH3 any better than the 1/8" hole design. Consequently, it was decided to
proceed with re-installing nozzles of the original design and spacing regardless of the catalyst
design changes and aging which had occurred.
Core samples of catalyst were also extracted during this outage by the manufacturer for
laboratory determination of activity. The tests conducted by the manufacturer showed catalyst
activity was nearly the same as the retained fresh samples from the original installation.
-------
DSCF-075
DUCT
WALL
5" TYP 10" TYP
4" PIPE
HEADER
(TYP)
20 PER 3"
SPARGER (TYP)
Figure Sb. AIG Drilling Pattern for All
Except Bottom North
DUCT
WALL
WITH 3 HOLES.
@ 1/16', 10
PER 3"
SPARGER (TYP)
' SPARGER PIPE,
7 PER CELL (TYP)
1" PIPE AND CAP
WITH 1 HOLE @ 1/16",
10 PER 3" SPARGER (TYP)
Figure So. AIG Pipe Tee and Hole Drilling Pattern
for Bottom North Cett Only
-------
Post-Modification Diagnostic Tests and Analyses
During the aforementioned outage, the original design nozzles were re-installed by AES, which
replaced the hole-in-pipe design shown in Figure 5. Tests of the modified system were
conducted by Carnot shortly after the outage and showed the stack NH3 had been reduced to 25
ppmc from the pre-modification level of 47 ppmc.
From the measurements performed both upstream and downstream of the NOX catalyst, system
performance parameters were again estimated as done with the detailed tests. The NOX removal
efficiency versus inlet NH3/NOX ratio is plotted in Figure 4 along with previous results. Re-
installing these nozzles substantially improved the SCR removal efficiency. For example, at an
inlet mole ratio of 1.5 for a given cell the removal efficiency was approximately 78%. With
the hole-in-pipe design, 78% removal efficiency required an inlet mole ratio of 2.5.
Although significantly improved, the stack NH3 slip still exceeded the 20 ppmc regulatory limit.
Single point NH3 measurements at various ports upstream of the catalyst ranged from
approximately 50% to 200% of the port average NH3 concentration. These data and the
manufacturer's catalyst activity tests indicated that non-uniform NH3/NOX ratios entering the
catalyst was still causing excessive slip.
Engineering calculations were performed by Carnot to gain an understanding of the problem.
Four items were addressed:
1. Since the basic objective was to properly entrain the injected ammonia with the exhaust
gases, a calculation of the entrainment of the actual mass flows of jet and bulk (turbine
exhaust) gases was performed. It showed that for a simple hole-in-pipe orifice design,
the minimum distance required to entrain the required amount of bulk gas is
approximately 138 inches, whereas the average available length as designed is 48 inches.
2. A calculation of the heat transfer between the exhaust flow and the air/NH3 inside the
sparger pipe was performed to determine heating (density change) effects on the
distribution of the mixture through the AIG. It helped simplify flow distribution
calculations if the air/NH3 mixture could be considered to be isothermal at the same
temperature as the turbine exhaust gases. This assumption was shown to be acceptable
as the air/NH3 flow would approach the exhaust gas flow temperature within several feet.
3. Proper distribution of air/NH3 through the AIG was calculated by comparing the
frictional and kinetic energy characteristics of the air/NH3 mixture relative to pressure
drop across the nozzle (or orifice). A design criterion to achieve +/- 5% flow
distribution through the pipe header is that the ratio of kinetic energy and (AIG) pipe
frictional losses be less than 10% of the nozzle/orifice pressure drop. Calculations
showed that this ratio was less than 1 %, indicating maldistribution was not due to the
sparger piping design.
4. Entrainment of a jet with bulk flow is enhanced with increased velocity differential
between the jet and bulk flow. To check the possibility that pipe pressure losses could
limit nozzle exit velocity, frictional losses were estimated for the actual AIG design from
-------
the air/NH3 mixing point to the point where the mixture enters the 3" spargers at the
boiler bottom. The typical mixing point pressure was measured at 36 psia, which
exceeded the critical pressure requirement of 28 psia required for sonic velocity. This
meant that an 8 psia pressure drop would be allowable. An analysis of the pipe network
demonstrated that frictional losses were negligible relative to the 8 psia allowable.
In parallel with Carnot's evaluations, an assessment of the modified system was performed by
the catalyst manufacturer using a computational fluid dynamic code. This analysis showed the
system would require a path length of 15 feet from the AIG to the catalyst to achieve proper
NH3 mixing. However, the system only had 3 to 4 feet of space.
Thus the diagnostic tests, engineering calculations, and catalyst manufacturer analyses all
indicated an NH3 distribution problem. Before proceeding with a complete re-design of the AIG,
it was decided to evaluate a low-cost installation of various nozzles which could utilize the
existing sparger piping arrangement.
NH3 Injector Nozzle Evaluation
In order to inexpensively evaluate different nozzle types, a bench-scale cold flow test was
designed and built by Carnot as shown in Figure 6. Only co-flowing designs were to be
evaluated in order to maintain the cost of designing, constructing, conducting and evaluating the
cold flow tests within project time and cost constraints. Similarity of the bench-scale test to the
actual installation would be gained with respect to actual bulk velocities, mixing length,
utilization of actual nozzles (rather than scaled nozzles), and maintenance of critical pressures
to achieve sonic velocity at the choke point for each nozzle.
The bench-scale test consisted of a 20" x 20" plexiglass duct with a blower pulling air at
approximately 20 ft /sec. The duct was sized so as to negate boundary layer effects. A
compressed cylinder of 30,000 ppm CO gas in N2 was injected through test nozzles coflowing
with the bulk air. A high concentration of CO was required to produce a measurable CO
concentration in the cold flow duct. A 20-point traverse measurement of CO concentration was
made 48" downstream of the injection plane to evaluate the relative dispersion effectiveness of
NH3 injector nozzles. Layers of screen material were installed until the bulk air velocity
distribution was within +/- 5% of the average velocity.
Since the nozzle (or orifice) injector pressure drop was the controlling factor for NH3
distribution through the grid, the use of nozzles was preferred over pipe holes in the interest of
long-term maintenance of discharge coefficients (CD). It was anticipated that holes would create
significant variations in the CD due to field drilling imperfections or hole plugging. Four
different nozzles were evaluated as shown in Figure 7:
• 180° deflector head hollow cone (originally designed, and re-installed during the
outage)
• 0.013" slotted axial nozzle - flat fan pattern
• 0.023" slotted axial nozzle - flat fan pattern
• 75 ° deflector nozzle - flat fan pattern
-------
DSCF-076
POINTS IN EACH PORT:
#1 - 2" FROM NEAR WALL
#2 - 6" FROM NEAR WALL
#3 - 10" FROM NEAR WALL
#4 . 14" FROM NEAR WALL
#5 - 18" FROM NEAR WALL
MESHED SCREEN
FLOW STRAIGHTENER
TEST NOZZLES
PORT A
PORTS
PORTCQ
PORTDQ
PORTED)
~ 20 ft/sec \
AIR BULK ^
FLOW
1 48" ^
BLOWER
PORTS FOR TRAVERSE
MEASUREMENTS OF CO'
CONCENTRATION
WITH NDIR ANALYZER
20" x 20" x 8' x 1/8"
THICK PLEXIGLASS
DUCT
30,000
ppmCO
IN
N2
Figure 6. Test Set-up for Cold Flow Nozzle Evaluation
-------
DSEM-077
Figure 7c. 75° Deflector
Flat Fan Nozzle
0.066" ORIFICE
INSIDE BODY
Figure 7h. Flat Fan
Nozzles
(0.013"and 0.023"
slots evaluated)
0.16" X 0.013" SLOT
OR
0.16" x 0.023" SLOT
0.063" GAP
Figure la. Deflector
Nozzels
(orieinal design for
SE boiler SCR: also
reinstalled during
outage)
DEFLECTOR HEAD
0.0138 IN2 ANNULUS
4 SLOTS INSIDE NOZZLE
BODY AT 0.030" x 0.038" EACH
-------
Important dimensions for each nozzle were measured, and the critical pressure to achieve sonic
velocity at the choke point was calculated. A tap and gauge were installed just upstream of the
nozzle connection to measure and maintain the minimum (critical) pressure during each cold
flow test. Since the pressure requirement was high, large CO/N2 gas flows were required and
thus CO traverses had to be made rapidly. In order to assure good readings, initial response
time tests were run to quantify the system response prior to testing nozzles.
The results of the cold flow model are contained in Table 4. As a measure of dispersion, the
standard deviation percentages of mean for downstream CO concentrations were calculated. The
0.013" slotted nozzle produced the best results at 56%, compared to 61-65% for the 180°
deflector nozzle installed in the SE boiler grid. The 0.023" slot and 10° deflector flat fan
nozzles performed the worst. However, the total range of this dispersion parameter was only
56-71 % indicating there were no major differences in dispersion effectiveness within the tested
path length. The data was viewed in different ways, such as ratio of minimum or maximum to
mean and number of points near the mean, but did not reveal any different results as seen in
Table 4.
Although the NH3 slip needed to be reduced only an additional 5 ppmc from 25 to 20 ppmc, it
was decided that new nozzles would not be installed. The results of the cold flow evaluation
indicated that new nozzles would only slightly improve the SCR performance, if at all.
Regulatory time constraints influenced this decision as well.
Conclusions
The results of the screening tests, detailed tests, and post-modification tests performed after
nozzle re-installation indicated NH3 maldistribution both cell-to-cell, and within each cell. It
was never clear why the system could maintain NOX at 7 ppmc with less than 10 ppmc slip
initially, but only 25 ppmc slip after the nozzle re-installation. The possible reasons were that
a combination of process changes such as inlet NOX level or moisture content (both influenced
by combustor steam injection rate for example), plugging of holes in the drilled pipe or injection
nozzle bodies, or minor catalyst aging effects combined to inhibit the SCR performance.
Measurements showed that process temperature, boiler flow distribution, and inlet NOX
concentration distribution were all within design guidelines. However, the engineering
calculations, computer modeling, and cold flow nozzle evaluations collectively indicated that the
4 foot path length was inadequate to achieve proper mixing of NH3 and NOX.
As a result of the evaluation, it was decided that the AIG be completely re-designed and located
further upstream in the SE boiler. In order to make space in a location with proper gas
temperatures, the CO catalyst was removed. This was justifiable based upon initial CO and
hydrocarbon measurements. The new grid was installed several feet further upstream, but still
downstream of the low pressure superheat tubes (see Figure 3). The new grid was designed with
the capability to bias NH3 to different zones. Flow-straightening louvers were also required as
the gas flow became stratified without the CO catalyst. Currently the system operates within
compliance for both NH3 slip and NOX emissions.
-------
Table 4
Cold Flow Nozzle Test Data Summary
port point
A 1
A 2
A 3
A 4
A 5
B 1
B 2
B 3
B 4
B 5
C 1
C 2
C 3
C 4
C 5
D 1
D 2
D 3
D 4
D 5
E 1
E 2
E 3
E 4
E 5
mean
std dev
std dev/mean
Min
Max
Mia/mean
Max/mean
within 25% mean
within 50% mean
180° Deflector
4 @.03" x .038"
ppmv CO
22.5
22.5
10.0
25.0
22.5
27.5
20.0
30.0
22.5
30.0
50.0
70.0
70.0
35.0
50.0
80.0
80.0
95.0
105.0
40.0
35.0
90.0
125.0
70.0
50.0
51.1
31.1
61%
10.0
125.0
20%
245%
4
13
(repeat test)
180° Deflector
4 @.03" x .038"
ppmv CO
7.5
7.5
50.0
67.5
35.0
25.0
65.0
85.0
90.0
55.0
35.0
45.0
125.0
80.0
15.0
30.0
72.5
120.0
85.0
25.0
17.5
35.0
55.0
30.0
17.5
51.0
33.1
65%
7.5
125.0
15%
245%
4
14
Slotted
Flat Fan
.013" slot
ppmv CO
10.0
25.0
25.0
42.5
20.0
10.0
17.5
40.0
55.0
50.0
7.5
17.5
30.0
40.0
20.0
15.0
40.0
40.0
40.0
10.0
15.0
65.0
50.0
30.0
10.0
29.0
16.3
56%
7.5
65.0
26%
224%
4
16
Slotted
Flat Fan
.023" slot
ppmv CO
20.0
25.0
40.0
15.0
15.0
—
—
—
—
—
15.0
80.0
100.0
115.0
105.0
—
-
—
-
~
45.0
100.0
115.0
70.0
—
61.4
40.5
66%
70° Deflector
Flat Fan
.066" orifice
ppmv CO
15.0
7.5
7.5
12.5
20.0
17.5
40.0
22.5
60.0
15.0
12.5
45.0
80.0
72.5
25.0
35.0
85.0
100.0
45.0
27.5
20.0
40.0
67.5
30.0
20.0
36.9
26.3
71%
7.5
100.0
20%
271%
6
12
-------
The SCR investigation described in this paper was appropriate for the unique circumstances
involved, although the methods employed are generally applicable to any gas-fired source.
However when encountering SCR performance problems on other gas-fired sources, the trouble-
shooting approach should be specifically tailored to the combustion device features and SCR
design capabilities. Some suggestions follow:
• Verify with repeated measurements that SCR system is malperforming (e.g. confirm good
slip measurement, confirm problem not related to CEM or other instrument
miscalibration).
• If a performance problem is verified, consider the magnitude of the problem (e.g. NH3
slip is 12 ppm or 50 ppm versus a 10 ppm limit).
• Design investigative tests considering SCR design features such as NH3 distribution
control, available test ports, inlet NOX magnitude and distribution, and catalyst age.
• Design investigative tests considering the gas-fired combustion device features such as
variations in inlet NOX or flow; load ramping; variations in O2, temperature, or moisture
levels; and catalyst poisoning potential.
• Investigative tests should also consider other factors such as regulatory deadlines and
manufacturer guarantees.
• Tests should be designed stepwise considering costs, and targeted towards the most likely
causes of malperformance. This is especially true for troubleshooting situations since the
plant owner is usually interested in compliance and minimum operating costs, not in
conducting a research project.
• Compliance quality confirmation tests should be performed once the system is optimized
or physically modified and tuned.
References
Beer and Chigier. Combustion Aerodynamics. John Wiley & Sons, 1972.
Keffer and Baines. "The Round Turbulent Jet in a Cross-Wind." Journal of Fluid Mechanics.
Vol. 15 (1963).
Perry and Chilton. Chemical Engineers' Handbook. 5th Edition. 1973.
Coulter. Compressible Flow Manual. 1986.
S.M. Cho, A.H. Seltzer, and Z. Tsutsui. Design and Operating Experience of SCR Systems for
; Control in Gas Turbine Systems. 1991.
-------
SELECTIVE CATALYTIC REDUCTION PERFORMANCE PROJECT
AT
PUBLIC SERVICES ELECTRIC AND GAS COMPANY'S
MERCER GENERATING STATION UNIT NO. 2
Albert J. Wallace
Francis X. Gibbons
Public Service Electric and Gas Company
80 Park Plaza
Newark, New Jersey 07101
Everett W. Knell
Robert E. Johnson
Wahlco, Inc.
3600 West Segerstrom Avenue
Santa Ana, California 92704
Ralf Sigling
Siemens Power Generation KWU
P.O. Box 3200
91050 Erlangen, Germany
Thomas M. Jantzen
Dennis E. Hubbard
Carnot
611 West Johnson Avenue
Cheshire, Connecticut 06410
Abstract
PSE&G, Wahlco, and Siemens have successfully demonstrated the technical feasibility of a
Post Combustion NOX Reduction System on a 321 MW net coal-fired utility boiler at
PSE&G's Mercer Station Unit 2. The program featured a selective catalytic reduction (SCR)
reactor in a horizontal flue gas duct and an air heater SCR (AHSCR). Predicted
performance of the In-Duct SCR was confirmed. The AHSCR was shown to be an effective
means to enhance overall system performance as compared to an In-Duct SCR alone.
Selective Non-Catalytic Reduction (SNCR) was also tested in combination with the
SCR/AHSCR system using aqueous ammonia as a common reduction agent. Ammonia based
jet mixing SNCR and the use of an ammonia injection grid was found to be the operation of
choice in the NH3 based SNCR/SCR Hybrid NOX reduction system.
-------
Introduction
Public Service Electric and Gas Company (PSE&G) has completed an assessment of post-
combustion NOX control on a wet-bottom,coal-fired utility boiler. The technologies under
study are In-Duct air heater SCR, and ammonia based SNCR/SCR Hybrid performed at
PSE&G's Mercer Generating Station.
While SCR and, to a limited extent, SNCR have been used on coal-fired boilers, these
processes have not been demonstrated in the U.S.A. on a unit with the same configuration as
the wet-bottom, continuous slagging, pulverized coal furnaces operated at Mercer. These
furnaces produce NOX emissions of approximately 1.8 Ib/mmbtu in which PSE&G working
together with NJ Department of Environmental Protection have determined the need for NOX
Reduction.
PSE&G's interest in NOX control for the Mercer Unit's is the result of our overall
understanding of the role that NOX plays in formation of ground level ozone. When PSE&G
assessed the interim state emission inventory it was determined that PSE&G's power plants
comprised 27% of the total NOX emitted in the State of New Jersey. As a result of this
emission profile, PSE&G established a public commitment to reduce NOX from our power
plants by 60% by 1995 and 80% by the year 2000. Following the SNCR demonstration in
1993, it was concluded that a greater level of NOX reduction would be necessary for New
Jersey to achieve attainment of the ozone standard. This is because of the severity of upwind
transported ozone into New Jersey at levels far in excess of the National Ambient Air
Quality Standard. PSE&G subsequently proposed to EPA and the Ozone Transport
Commission that a regional NOX standard of 0.2 Ib/mmbtu be adopted on power plants
throughout the Ozone Transport Region. The Mercer project demonstrates that emission
reductions of the magnitude proposed can be obtained in a cost-effective manner.
PSE&G determined that installation of a full scale SCR reactor in the congested design and
footprint of the Mercer units was not a viable option. The only location with a flue gas
temperature appropriate for SCR was in a horizontal duct between the economizer outlet and
air preheater. This existing duct was expanded to three times its cross-section to
accommodate sufficient SCR capacity. In addition, the In-Duct SCR system included catalyst
located in the hot-end of the air preheater to reach the .2 Ib/mmbtu goal, a reduction of 88-
89%.
The SCR demonstration was conducted on one of our four flue gas ducts with a horizontal
shaft Ljungstrom air preheater.
The goals of the program were to demonstrate:
First time use of horizontal In-Duct SCR on a coal-fired boiler
First tune air heater/In-Duct SCR combination coal application
First time NH3 SNCR/SCR Hybrid system while firing coal
Minimal balance of plant impacts (i.e., air heater fouling, pressure drop)
Acceptable NH3 slip and effects on fly ash
-2-
-------
Construction of the In-Duct SCR and AHSCR began in January and testing ran from June
through October 1994. The aqueous ammonia SNCR/SCR Hybrid system testing was done
in December 1994 and January 1995.
Description of Mercer Unit No. 2
Mercer Generating station has two Foster Wheeler continuous slagging, twin-furnace steam
generating units. Units Nos. 1 and 2 are identical with a rated capacity of 321 MW net
each. Each unit is designed for 2,060,000 Ib/hr of superheated steam flow at 2456 psig and
1050°F. Reheated steam flow is 1,760,000 Ib/hr at 445 psig and 1050°F. In this twin
furnace design, one furnace generates superheated steam and the other reheated steam. Both
units were originally designed with pressurized furnaces and have been converted to balanced
draft with some modifications to the radiant superheater to allow for operations at 25 % load,
without adverse effects on boiler operations. Figure 1 shows a side view of the reheat
furnace on Unit No. 2.
The Mercer units burn low sulfur Eastern bituminous coal as the primary fuel, with natural
gas available for start-up and as a secondary fuel. Fuel enters each furnace through twelve
front wall mounted burners arranged in three levels of four burners per level. There are
three water tube division walls located above the top burner elevation and centered between
each row. During coal firing, the furnaces are wet-bottomed, continuous slagging by design.
The burners are arranged close to the furnace floor to keep the slag in a molten state at all
loads. While it is possible to introduce coal and gas through the same burner at one time,
general plant practice has been to operate each row of burners on either coal or gas. Fuel
distribution between the furnaces is controlled to maintain the appropriate superheat and
reheat temperatures.
Furnace gases exit the boiler through a convection superheater in one furnace and a
convection reheater in the other. Economizers are installed after both the convection
reheater and superheater; however, the economizer in the superheat furnace has more surface
area. The flue gas splits into two ducts after each economizer to transfer the gases around
the slag tank under each furnace. The flue gases pass through four Ljungstrom horizontal
shaft regenerative air heaters (two per furnace) and are ducted to a multi-chamber
electrostatic precipitator.
Testing Program Description
PSE&G contracted Carnot to conduct a comprehensive test program for evaluation of the
SCR and Hybrid SNCR/SCR systems. The test data included NOX emissions, CO, CO2, O2,
SO2, S03, NH3, and N2O measurements as well as plant operating data at various loads firing
either coal or gas.
Measurements were made at several different locations in order to provide gas composition
data at the inlet, outlet and intermediate locations of the SCR, as well as to determine the
impact of the SCR operation on the remainder of the plant. Figure 2 presents a schematic of
the reheat furnace illustrating the SCR system in duct 22B, the SNCR system, and the
various test locations. A damper on duct 22A allowed for flow balancing.
-3-
-------
Each sample location was equipped with a multi-point probe grid installed by PSE&G prior
to the testing program. The grid consisted of a series of stainless steel probes that
transported the flue gas sample under vacuum to Camot's CEMS system for determination of
NOX, O2, CO, CO2 and N2O. NH3 emissions were determined by extracting a metered
volume of flue gas through a dilute sulfuric acid solution. The NH3 concentration of the
sample was then determined by a colorization process via spectrophotometry.
Performance testing was conducted to establish the performance of the SCR system over the
normal range of loads and operating conditions. Testing was conducted on both coal and
natural gas at four nominal load points, 310, 220, 135, and 80 MW (net).
To determine the effect of NH3/NOX mole ratio on NOX reduction and NH3 slip, several NH3
injection rates were tested for each load and fuel. The NH3 injection rates were selected to
best approximate the following conditions:
• 60 percent NOX reduction
Maximum NOX reduction at:
• 5 ppm NH3 slip at Location D (ahead of air heater SCR)
• 10 ppm NH3 slip at Location D (ahead of air heater SCR)
• 5 ppm NH3 slip at Location E (down stream of the air heater SCR)
Ammonia and SO3 hi the flue gas stream can react to form ammonium sulfate and
ammonium bisulfate which can foul the air heaters. In addition, since PSE&G is able to sell
most of its ash, there was concern that excessive ammonia carryover into the fly ash may
inhibit these sales. Therefore, it was decided that the target ammonia slips would be tested
cautiously while observing the air heater pressure drop.
For each of these conditions composite measurements of NOX and O2 were made at Location
A and Location D to determine the NOX reduction performance of the In-Duct SCR. (NOX
concentration was determined by measuring NO from a composite sample and taking periodic
single point NO2 measurements.) For many of the test points NOX and O2 were also
measured at Locations C and E. This provided data to evaluate the performance of the first
layer of in-duct catalyst and the air heater SCR, respectively.
NH3 measurements were made for each of the above test conditions at Location D.
Additional NH3 measurements were made at Locations C and/or E depending on the test
condition (i.e., measurements were typically made at Location C when the NH3 injection rate
was low and measurements were made at Location E when the injection rate was high). All
of these measurements were composite samples used to determine the NH3 slip, an important
factor hi detennining the performance of the SCR system.
In addition to the NOX reduction and NH3 slip, one of the most important design parameters
to be considered for an SCR system is the conversion of SO2 to SO3 by the catalyst. In order
to determine the percentage of the SO2 converted, the inlet and outlet concentrations of SO2,
SO3, and sulfate in the ash were measured. These measurements were made prior to
injecting NH3 as well as on a periodic basis during NH3 injection at full load coal firing.
-4-
-------
In order to determine the impact of the catalyst on pressure drop through the duct work, the
pressure was monitored at each of the Locations A through E. PSE&G installed differential
pressure cell and transmitters to monitor and record this data continuously on the plant DCS.
The data were used to determine the overall increase in pressure drop at various loads as
well as to determine the increase hi pressure drop over time and to evaluate the impact of
soot blowing on pressure drop.
SCR System
Description
The SCR system was jointly designed by PSE&G, Wahlco, and Siemens and consisted of an
ammonia (NH3) injection system, air soot blowers, in-duct catalyst and catalytic air heater
baskets.
The NH3 injection system consisted of two NH3 vaporization skids and an injection grid
located hi the flue gas duct just downstream of the economizer. Aqueous NH3, at
approximately 27 percent by weight concentration, was vaporized hi a hot air stream
provided by dilution air fans and electric heaters which heat ambient air. The air/NH3
mixture was routed to a fifteen zone ammonia injection grid (AIG) through a series of
throttle (biasing) valves used to match the NH3 flow distribution with the flue gas NOX flow
to produce a uniform NH3/NOX ratio distribution at the catalyst face. As the flue gas exited
the economizer, it passed through a Turbomix* static mixer above the AIG, made a 90° turn,
and entered duct 22B. Guide vanes were located downstream of the 90° bend to straighten
the flow and to balance the velocity distribution at the catalyst face. This reduced catalyst
erosion and pressure drop.
Duct 22B was expanded to approximately three times its original cross sectional area to
incorporate the in-duct catalyst banks. Inside the duct were two separate banks (5 feet in
depth each) of catalyst elements separated by approximately 5 feet of duct space. The
catalyst plates were oriented vertically to minimize fly ash deposition. Metal frames secured
the catalyst in place. Previous coal-fired boiler SCR experience has been limited to vertical
down flow configurations. This is the first horizontal In-Duct SCR™ application on a
pulverized coal unit in the world.
The plate type catalyst was manufactured by Siemens and consisted of a metal substrate
covered by catalyst. The active catalytic materials are Titanium Dioxide and
Molybdenum/Vanadium Oxide.
The success of this SCR system is dependent upon the operation of a complex soot blowing
system. Air operated soot blowers are located ahead of the first and second catalyst banks to
periodically clean the catalyst plates of ash build-up. In addition, horizontal soot blowers are
located on the duct floor (hi front and behind each catalyst bank) to re-entrain any settled fly
ash into the flue gas flow.
The existing horizontal shaft Ljungstrom air heater was modified by replacing the hot-end
baskets with catalytic air heater baskets supplied by Wahlco. The catalyst plates were
-5-
-------
identical to those used in the In-Duct SCR except that they were located hi the rotating wheel
of the air heater. Therefore, slightly less than 50 percent of the air heater catalyst was
exposed to the flue gas at any given time.
Model Study
Siemens performed a physical flow model study to confirm and optimize the design and
check the feasibility of the horizontal In-Duct SCR study for a coal fired application. The
duct cross sectional area was increased threefold to maximize space available for the catalyst.
Due to space limitations, a short transition was required for the duct expansion. This steep
transition created concerns regarding: flow straightening to achieve uniform velocity profiles;
pressure drop increases in addition to the pressure drop over the catalyst; and ash
sedimentation. Due to the unique horizontal flow arrangement of the reactor, no well-
established guidelines exist for addressing these concerns. Finally, ensuring a uniform
NH3/NOX ratio at the catalyst face is necessary for best NOX conversion. The following SCR
design objectives were established as idealized targets which were not necessarily achievable
in all cases, but provided a goal:
• velocity distribution of +/- 15% maximum deviation over 90% of the catalyst face
and +/- 20% over the remainder
• NH3 distribution of +1-5% maximum deviation
• pressure drop increase of 5 i.w.g.
A 1:20 scale model was built from the economizer exit through the air heater inlet. It
included the AIG, economizer hopper, the flue gas duct with a 90° turn into the horizontal
#22B duct, the reactor inlet with guide vanes, and the SCR catalyst banks. With the overall
geometry established by the duct expansion, the strategy involved varying the inlet guide
vane arrangement with the following results:
• velocity distribution of +/- 15% maximum deviation over 90% of the catalyst face
and +34% to -23% over the remainder. RMS1 values ranged from 13% to 15%.
The result was deemed acceptable.
• NH3 distribution of +8% to -13% maximum deviation and RMS values ranging from
5% to 6%. Since the test was performed without bias of the injection grid which
would be done in practice, the result was deemed satisfactory.
• pressure drop increase of 6 i.w.g. which included an allowance for ash deposition.
Ash sedimentation was a concern due to low velocity in the expanded duct of 26 fps at full
load and 7 fps at minimum load. Simulations were made using actual ash from Mercer
Station. No deposition was observed in the catalyst banks but did occur on the duct floor at
Mean Square (RMS)
-6-
-------
the lower loads. Floor deposits were observed to re-entrain when the gas flow was increased
to full load equivalent and it was concluded that sedimentation could be managed with the
provision of sootblowers.
Performance
Table 1 shows the predicted NOX conversion performance of the In-Duct SCR, for coal firing
at the four usual power production levels. At full load with fresh catalyst, the NOX
conversion was predicted to be 88% at 10 ppmvd (at 7 % O2) NH3 slip. After two years,
the full load NOX conversion is projected to drop to 72% at 10 ppm NH3 slip.
The pressure drop from the economizer outlet to the air heater inlet while firing coal at full
load, was predicted to increase from 1 i.w.g. to 7 i.w.g. which included 5 i.w.g. across the
catalyst banks.
Commissioning
Optimization of the NH3/NOX ratio at the In-Duct SCR inlet was paramount during the
commissioning phase of the program. Early, it was determined that the flue gas velocity
profile at the AIG was skewed by a factor of about 3:1 across the duct. This prevented the
target NH3/NOX distribution from being achieved even with some AIG valves wide open and
others nearly closed. Therefore, Siemens recommended addition of a Turbomix* static mixer
which substantially reduced the flow bias to < 2:1.
Figures 3 and 4 compare the normalized AIG flows and the NH3/NOX distributions before
and after installation of the mixer. The range of the relative flow settings was greatly
reduced and the NH3/NOX ratio distribution RMS improved from 6.4% to 5.7%, which
compares well with the model study.
The AIG flow settings, established for full load coal firing which is the most stringent
condition, were used throughout the program for both coal and natural gas fuels, and all
boiler loads.
The pressure drop across the system was unchanged by the mixer. The mixing energy
consumption apparently was offset by reducing the velocity gradient at the 90° bend in the
duct.
In-Duct SCR Test Results
The Performance of the In-Duct SCR was evaluated with respect to NOX reduction and NH3
slip vs NH3/NOX ratio at loads from 321 to 80 net MW firing both coal and natural gas.
Pressure drop, SO2 to SO3 conversion, fly ash effects, ash deposition and various other
operating data were collected.
Figures 5 and 6 show NOX reduction and NH3 slip vs NH3/NOX ratio for all tests. Most tests
were made for full load coal, since this is the most critical case for PSE&G with respect to
the reduction requirement. Performance firing coal at full load was exactly as predicted,
88% with 10 ppm NH3 slip, Figure 5.
-7-
-------
The catalyst was essentially fresh during the six-month program and, therefore, its
performance was at its peak. Since catalyst activity will decline with time, samples are being
taken periodically for lab scale activity analysis. Results after six months show the aging is
proceeding according to the predicted 72% NOX reduction efficiency after 2 years. For a
commercial installation, total system efficiency could be maintained by additional catalyst and
catalyst management subject to pressure drop limitations and SNCR, as discussed later.
The NOX reduction efficiency, firing natural gas at full load, was less than when firing coal.
This is due to the higher gas flow, higher space velocity, inherent to gas tiring stoichiometry
and heat rate. Also, the AIG was not reset for natural gas since it would be unpractical to
do so given the frequent fuel switching practiced.
NOX conversion performance is presented Figures 7 and 8 as outlet NOX in Ib/mmbtu and
NH3 slip for full load coal and gas firing. The 0.2 Ib/mmbtu target was achieved at less than
10 ppm NH3 slip for both fuels; again, the results were with fresh catalyst.
Figure 9 shows performance for both fuels at all loads at an NH3/NOX ratio of about 1.0.
NOX conversion at the lower loads is greater than 90% as predicted. This reflects the lower
gas flow at low load which more than offsets reduced gas temperature. The variability
observed at full load is due to variation hi the baseline NOX.
Overall, the In-Duct SCR was able to achieve NOX reductions hi the range of 85-90% while
maintaining the ammonia slip to less than 10 ppm as measured at the air heater inlet. NOX
reductions in excess of 95% are possible, however, ammonia slip would be greater than 30
ppm.
Air Heater SCR Test Results
The AHSCR raised the SCR system full load coal firing NOX reductions to > 9O% with no
measurable gaseous NH3 slip. NOX reduction above 95% was achieved while still
maintaining gaseous NH3 slip to less than 5 ppm at the air heater outlet. Figure 10 shows
AHSCR NOX conversion vs NH3 slip for full load coal firing together with the previously
shown result for the in-duct SCR alone. The NH3 slip is the gaseous NH3 downstream of
each. The overall performance improvement as a result of the AHSCR is substantial by
comparison to the NOX conversion at 5 ppm slip: 78% for the In-Duct SCR alone and 96%
for the combined system. This is primarily due to the greater performance of the in-duct
SCR of higher NH3 slip. It can operate at this level because the downstream NH3 utilization
of the AHSCR utilizes a portion of the ammonia slip for additional NOX reduction.
Figure 11 shows a similar result for full load natural gas firing where NOX conversion at full
load is unproved from 58% to 94% at 5 ppm NH3 slip. Again, the improvement is mainly
attributed to greater allowable NH3 slip from the In-Duct SCR.
Overall NOX reduction in terms of Ib/mmbtu for the SCR alone and with the AHSCR are
shown in Figures 12 and 13. Again, a significant enhancement of overall system
performance is demonstrated.
-------
The foregoing results are encouraging indeed. But, when firing coal, there are potential
technical limitations that must be taken into account on a case by case basis. Depending on
the disposition of the fly ash, its NH3 content can become an issue when looking at salability
(primarily the concrete industry). During the Mercer project, ash samples were taken by
PSE&G's Research and Testing Lab and resulted hi no repeatable trend. For 5 to 10 PPM
NH3 hi the flue gas, approximately 175 to 300 PPM of ammonia was detected hi the fly ash.
Ammonium bisulfate fouling of the air heater is another operational concern but there was no
indication of this during the entire program. Additionally, established correlations of
NH4HS04 deposition considering NH3 and SO3 concentrations and the AHSCR temperature
profile indicate that it should not be a future problem for Mercer station. This is a site-
specific issue that must be carefully addressed for each application.
Pressure Drop
Figure 14 presents the average full load differential pressure measured across the individual
SCR system components. The data are weekly averages for all periods firing coal at loads
greater than 90% of full load. Although there is some fluctuation, there is no trend in tune.
An effective soot blowing program was instrumental hi minimi/Ing deposits and therefore
pressure drop across the catalyst banks. After testing more frequent soot bio whig cycles, it
was determined that twice per day was sufficient to prevent increased pressure drop.
The average pressure drop from the economizer to the air heater inlet was 3.2 i.w.g. at full
load. This is roughly the net increase of the In-Duct SCR catalyst banks, the AIG, and the
mixer. It is less than predicted mainly due to successful ash deposition management and
some conservatism hi the original estimate.
As the current fans are capable of overcoming the additional pressure drop, there is no
associated capital cost. The additional power consumption per ID fan is approximately 325
BMP at full load.
SO2/SO3 Conversion
One of the important design parameters for an SCR system is the catalytic oxidation of SO2
to SO3. Then: concentrations and sulfates hi the ash were measured at locations B, D, and E
with and without ammonia injection firing coal at full load.
The average conversion of SO2 was 1.7% across the In-Duct SCR. The vast majority of the
SO3 formed, however, was found to react with alkaline constituents hi the fly ash forming
sulfates and, hi this benign, form is unavailable for ammonium bisulfate formation hi the air
heater.
-9-
-------
Hybrid SNCR/SCR System Using Aqueous Ammonia (NH4OH)
Description
One of the purposes of the SNCR component of the SNCR/SCR Hybrid NOX Reduction
System is to supplement the NOX conversion capacity of the downstream SCR and especially
to compensate for its decreasing capacity as the catalyst ages. Another is to provide NH3
slip to the downstream SCR. In general, both the NOX conversion and NH3 slip from the
SNCR system increase together. In principle, all of the NH3 could be supplied as slip from
the SNCR, but, mal-distribution of NOX and NH3 at the catalyst occur. Since SCR
performance is enhanced by uniform spatial distribution of NH3/NOX at the catalyst face, the
use of an AIG to complement the NH3 supply is therefore a better design for high-efficiency
Hybrid system operation.
The objectives were to 1) demonstrate the performance of aqueous ammonia SNCR, 2)
develop and demonstrate overall Nf^OH SNCR/SCR Hybrid System NOX removal of greater
than 90% (this operation simulated the use of SNCR as a catalyst management tool; catalyst
age at two years), and 3) obtain data for design of a commercial system.
The SNCR portion of the NH4OH SNCR/SCR Hybrid NOX Reduction System was tested
using only existing observation ports on the front wall of the upper furnace and openings
made in the rear wall above the economizer. The aqueous ammonia injectors used
superheated steam as an atomizing and delivery medium by jet mixer nozzles.
Testing focused primarily on full load operation while burning coal only. Limited testing at
lower loads was conducted on the front wall to evaluate the effects of cooler flue gas
temperatures in the SNCR injection zone.
Hybrid operation was performed at full load using only the four rear wall injectors. NH3
addition via the AIG was practiced and the proportions of NH3 supplied to the catalyst was
varied from all SNCR NH3 slip to all from the AIG.
Hybrid SNCR/SCR Test Methods
NOX and O2 distribution measurements at Locations B and D with and without ammonia
injection were made to determine SNCR NOX reduction and NH3 slip. When operating in the
Hybrid mode with supplemental NH3 via the AIG, the NOX reductions of each stage were
determined via baseline vs SNCR/SCR injection-on measurements at B, D, and E. Spatial
distributions of NOX and NH3 were measured in four and three point composite samples in
the 16 point (B&D) and 9 point (E) test grids, respectively. Ammonia concentration was
determined by the change in NOX concentration by composites samples from Locations B and
D. Spatial ammonia distributions were also measured by wet chemistry by 4 and 3 point
composites samples at Locations D and E respectively.
Hybrid SNCR/SCR Test Results
The NH4OH SNCR/SCR Hybrid test program was divided into four distinct phases. Since
the optimum temperature window for ammonia-based SNCR is narrow, approximately 1600
-10-
-------
to 1850°F, the first two test phases were designed to evaluate SNCR performance at two
different injection locations. Testing during Phase I evaluated SNCR injection at a location
on the front wall of the boiler at the furnace exit. Phase II evaluated SNCR injection at a
cooler location on the rear wall of the boiler at the same elevation as the front wall location.
Phase El of the test program focused on evaluating the SCR system with injection through
the AIG alone to prove the integrity of the SCR system as it was at the end of the SCR
Demonstration Project. Phase IV evaluated SNCR/SCR Hybrid System performance with
simultaneous SNCR and AIG injection.
A summary of the prominent results of the NHjOH SNCR/SCR Hybrid demonstration
program is provided below:
• An overall NOX removal of 90% and less than 2 ppmvd NH3 slip at location "E" with
40% SNCR removal was demonstrated. These results were attained with the AIG
settings optimized for the fresh catalyst, and indicate that AIG adjustments would not
be necessary as the proportion of required SNCR NOX removal increases as the
catalyst ages.
• NOX removal by the SNCR subsystem ranged up to half of the total. The tests with
Hybrid injection confirm the role of SNCR, for an application such as Mercer with a
high NOX percentage reduction requirement, to be an effective supplementary NOX
removal technique for retrofit situations where space and pressure drop constraints
limit the amount of In-Duct SCR catalyst.
• The SNCR optimization tests showed the rear wall to be the preferred injection
location for the Hybrid system, where both the NOX removals and NH3 slips were
higher than those for injection at the front wall location.
• Rear wall SNCR injection is capable of providing 40% NOX removal with NH3 slip of
150 ppmvd ahead of the catalyst as shown in Figure 15.
• The overall NOX removal performance of the rear wall injectors was found to be best
at a steam-to-flue gas ratio of 0.6 percent and this condition also provided the most
uniform NH3 slip distribution at the catalyst inlet.
• Sootblowing was shown to have a large effect on baseline NOX emissions as well as
NH3 slip with rear wall SNCR injection, indicating a significant effect on furnace exit
temperature.
• The SCR tests with AIG "only" confirmed that the catalyst performance was similar
to that measured during the previous SCR-only demonstration.
The results of the Phase I and Phase n tests indicated that flue gas temperatures at the front
and rear wall injection locations were slightly high and slightly low, respectively, for optimal
SNCR performance with NHfOH. Options for further optimization of the SNCR system
were identified that can achieve greater than 50% SNCR NOX reduction.
-11-
-------
Conclusions
Evaluation of the In-Duct/air heater SCR and Hybrid SNCR/SCR systems for use on
PSE&G's Mercer Generating station has demonstrated sufficient NOX reduction toward
PSE&G's goals. Balance of plant impacts were minimal.
PSE&G's Mercer Duct No. 22B was successfully tested as the first horizontal In-Duct SCR
reactor on a coal-fired unit in the world. The vertical orientation of catalyst plates in a
horizontal duct arrangement utilizing SCR Sootblowers minimized the pressure drop across
the system. There was no measurable ash deposition within the ductwork at all load settings.
Pressure drop was less than predicted and within the range originally specified by PSE&G.
The In-Duct SCR achieved the predicted 88% NOX conversion with < 10 ppm NH3 slip.
The air heater SCR increased the overall system NOX reduction. At 5 ppm NH3 slip, the
overall NOX reduction improved by approximately 20% from = 75% NOX reduction at D to
~ 95 % NOX reduction at E. The air heater SCR therefore played an important role in
ammonia slip elimination, allowing higher levels of ammonia slip from the stationary In-Duct
SCR.
The ability to maintain low NOX emissions and ammonia slip at all loads and with both fuels,
as well as low pressure drop, absence of fouling, and acceptable ash NH3 content was
successfully demonstrated.
The Hybrid SNCR/SCR system tests using aqueous ammonia as a common reagent revealed
that the SNCR process can effectively reduce NOX entering the catalyst system. Operation in
this manner can be performed as a means for managing catalyst life because sufficiently high
SNCR NOX reduction can be attained. In addition, these and earlier tests revealed that the
SNCR process can furnish NH3 to the in-duct catalyst for combined performance equivalent
to an SCR NH3 injection system. However, the most efficient SCR performance is achieved
when using supplemental ammonia from the ammonia injection grid.(1)
The Mercer project demonstrated that emission reductions of the magnitude required to meet
the proposed NOX limit can be obtained in a cost-effective manner. Overall, PSE&G was
able to satisfy its objectives for evaluating these NOX control technologies.
Acknowledgements
The authors acknowledge and thank the following people for their many contributions to the
project: Alex Huhmann, Robert Johnson, Robert Laflesh, James Panacek, Ronald Rauffer,
and Peter Tomas of PSE&G; Kerstin Hauenstein, Wolfgang Herr and Horst Spielman of
Siemens; William Baker, Kevin Dougherty, James Foley, David Oehley, Felix Spokoyny,
and Barry Whyte of Wahlco; Matthew Dugan, Richard Himes, Robert Markoja, Roland Roy,
Paul Stanley, and Zachary West of Carnot; Michael Polacek of Stone & Webster; Lawrence
Muzio and Randall Smith of FERCO; and Thomas Moskal of Diamond Power.
-12-
-------
References Cited
1. A. J. Wallace. F.X. Gibbons, R.O Roy, J.H. O'Leary, and E.W. Knell
"DEMONSTRATION OF POST COMBUSTION NOX CONTROL TECHNOLOGY
ON A PULVERIZED COAL WET BOTTOM UTILITY BOILER," presented at Air
& Waste Management Associations 88th Annual Meeting & Exhibition,
San Antonio, Texas, (June 1995).
2. Turbomix is a registered trademark of Siemens.
3. In-Duct SCR is a trademark of Public Service Electric and Gas Company.
TABLES AND FIGURES
WORST
100%
344
321
1
100%
344
321
2
68%
235
220
3
42%
146
135
4
27%
92
80
Table 1
PSE&G-Mercer 22B Demonstration Predicted Performance
Hybrid NOX Reduction System Basis: Per Duct Except Boiler Load
CASE
BOILER LOAD
Percent (%)
Gross MW
NetMW
UNCONTROLLED NO. EMISSION
LBNOX/MMBTU 1.8 1.65 1.36 1.18 0.55
PPMVD @ 7% O2 1017 931 770 667 311
LBNOX/HR 1427 1283 710 401 117
IN-DUCT SCR PERFORMANCE
%NOX Reduction, Zero Years
NH3 Slip = 10 PPMVD7 84 88 95 98 99
NH3 Slip = 5 PPMVD7 75 82 94 98 98
%NOX Reduction, Two Years
NH3 Slip = 10 PPMVD7 63 72 90 96 99
NH3 Slip = 5 PPMVD7 46 57 88 95 98
-13-
-------
SUPERHEATER
CONDENSER HEADERS
HEATER
OERS
RADIANT SUPERHEATER
FURNACE FRONT, REAR a SIDE WALLS
REI EATER OUTLET
RADIANT SUPERHEAT
HEADERS
WATERWALL HEADERS
I38'-0
MTER INLET
Figure 1 - Side View of Unit No. 2 Reheat Furnace
-14-
-------
PSE&G MERCER IN-DUCT SCR/SNCR PROJECT
SNCR - Selective
Non-Catalytic Reduction
FURNACE #22
STEAM
O1
I
4 INJECTORS
1 ELEVATION
STATIC
MIXING
GRID
PRECIPITATOR
AMMONIA
(NH3)
INJECTION
GRID
AIR HEATER
WITH HOT-END
CATALYST
SECTORS
SCR
CATALYST
BANKS
PUMPING SKID
SCR - Selective Catalytic Reduction
PSE&G Mercer In-Duct SCR System and SNCR Test Locations
Figure 2
-------
South
BEFORE
AFTER
1.97
1.30
0.00
1.29
1.24
0.21
1.24
1.05
0.97
1.22
1.03
0.69
2.45
1.30
0.00
North
South
0.83
0.94
0.99
1.00
O.91
0.94
O.94
1.05
1.12
1..07
1.05
1.04
1.O4
1.O5
1.04
North
Figure 3 Furnace No. 22 Economizer Exit Cross Section -
Normalized AIG Zone Flow Rate Distribution Before
and After Installation of the Static Mixer
BEFORE
L L
BANK tt* 1 TOP (Location B)
0.95
1.06
1.16
1.09
O.97
0.94
1.02
1.08
0.96
0.89
0.89
0.99
0.97
1.03
0.97
1.03
RMS = 6.4%
Measurements : S = Short, L = Long
FLOW IIMTO
PAC3E
8
AFTER
S L L S
BANK #\ TOP (Location B)
0.98
0.93
O.95
0.98
1.O4
1.O7
0.95
0.93
1.16
1.07
0.96
0.95
1.02
1.02
0.98
1.00
RMS = 5.7%
8
7
Figure 4 Normalized NH3/NOx Distribution at SCR Bank No. 1
Before and After Installation of the Static Mixer
-------
Coal Firing 31OMW (net)
(New Catalyst)
100%
50%
0.6
0.8 1
NH3/NOx Mole Ratio
Coal firing 80 MW (net)
100%
§ 90%
8 9
5 g 80%
tt | 70%
° ~ 60%
50%
0
(New Catalyst)
^^
/^
m _— —*
6 0.8 1 1
60
50 "E
40 a
30 .&
20 M
10 i
0
2
NH3/NOx Mole Ratio
-J
I
Figure 5 In-duct SCR Perfromance for Coal Firing
NOx Reduction and NH3 Slip as a Function of NH3/NOx Mole Ratio
Natural Gas Firing 310 MW (net)
100%
| 90%
| 1 80%
a § 70%
x .5
° 60%
50%
0
(New Catalyst)
.^^
i PI i
6 0.8 1 1
60
50 "§
40 a
30 .S-
10°|
0
2
NH3/NOx Mole Ratio
Natural Gas Firing 80 MW (net)
(New Catalyst)
100%
90%
0.8 1
NH3/NOx Mole Ratio
.2
Figure 6 In-duct SCR Perfromance for Natural Gas Firing
NOx Reduction and NH3 Slip as a Function of NH3/NOx Mole Ratio
-------
o NOx Emissions
• NH3 Slip
At Location D
0.7
0.6
0.8 1
NH3/NOx Mole Ratio
1.2
Figure 7 NOx Emissions at In-duct SCR Outlet
Coal Firing 320 MW (net)
NOx Emissions
NH3 Slip
At Location D
0.7
0.6
0.8 1
NH3/NOx Mole Ratio
1.2
Figure 8 NOx Emissions at In-duct SCR Outlet
Natural Gas Firing 320 MW (net)
- 18 -
-------
Coal Firing
Natural Gas Firing
50 100 150 200 250
Gross Unit Load (MW)
300
350
Figure 9 In-duct SCR performance in Most likely Operating Range
NOx Emissions as a Function of Load
-------
System Performance w/AH-SCR
System Performance w/o AH-SCR
2O.O
NH3 Slip {ppmvd @7% O2)
Figure 10 Effect of Catalytic Air Heater on Overall Performance
Coal Firing 310 MW (net)
100% -
» System Performance w/AH-SCR
O System Performance w/o AH-SCR
90% -"
i ys
i! *>* -
•+1 <0
§ «
3 f >
x ^ 70%
o2-
2
60%
60%
0
_^» ^ — — t
• \ /"^ D y —
/ ^
/ /
/
/
/
/
L
a
0 10.O 2O.O 30.0
NH3 Slip (ppm)
Figure 11 Effect of Catalytic Air Heater on Overall Performance
Natural Gas Firing Firing 310 MW (net)
- 20 -
-------
0 35
• System Performance w/AH-SCR
0 30 1
n System Performance w/o AH-SCR ' ,
I °'25
l| 0.20
| 1 0,5 -
Ul ~
X
0 0.10 -
0.05 -
O.OO -
O
' V
\x
\ XVS
" — c
0 10.0 20.0 30.0
NH3 Slip (ppm)
Figure 12 NOx Emissions from In-duct SCR and Hybrid System
Coal Firing 310 MW (net)
O 70 • - ... . .
• System Performance w/AH-SCR
0.60 -
d System Performance w/o AH-SCR
3 0.50 [-
m
1 f O-40
P
P
| 1
'I
\
o ° IV
1 1 0.30 4 A
IS I %?
g 0.20 -
0.10 -
I
0.00 -
o
V °°s"
0 1O.O 20.O 30.0 4O.O BO.O
NH3 SHp (ppm)
Figure 13 NOx Emissions from In-duct SCR and Hybrid System
Natural Gas Firing 310 MW (net)
- 21 -
-------
- Duel 22B Econ out to AH out
- A-H22B
- Duct 22b 1st Layer of Catalyst
- Duct22B2ndUyerofCalaJyst
- Duct 22AA-H (reference)
U
5
a.
s.
a
I
10 --
Week
Figure 14 SCR System and Component Pressure Drop vs. Time
Coal Firing 310 MW (net)
0.5 1 1.5
NH3/NOx Mote Ratio
Figure 15 Rear Wall SNCR performance
Coal Firing 310 MW (net)
- 22 -
-------
EVALUATION OF COMBINED SNCR/SCR FOR NOX ABATEMENT ON A
UTILITY BOILER
Albert J. Wallace
Francis X. Gibbons
Public Service Electric and Gas Company
80 Park Plaza, T-19E
P.O. Box 570
Newark, New Jersey
John. Boyle
John. O'Leary
Nalco Fuel Tech
P.O. Box 3031
Naperville, Illinois 60566
Dennis E. Hubbard
Roland Roy
Carnot
15991 Red Hill Avenue, Suite 110
Tustin, California 92680
Abstract
Public Service Electric and Gas Company has recently completed an evaluation of an SNCR/SCR
Hybrid system on a wet bottom, pulverized coal fired utility boiler. This SNCR/SCR Hybrid
system utilizes urea based SNCR, provided by Nalco Fuel Tech, and an In-Duct SCR system,
supplied by Wahlco Inc., to provide both in-furnace NOX reduction and/or ammonia to feed the
downstream reaction catalyst system. This paper presents the results of the SNCR/SCR Hybrid
tests. Performance data for the In-Duct SCR will be presented in a sperate paper.
-------
Introduction
Public Service Electric and Gas (PSE&G) is evaluating the effectiveness of post-combustion NOX
control technologies on a wet-bottomed, coal-fired utility boiler. The technologies under study are
conventional urea-based SNCR, In-Duct and air heater SCR, and a combination of SNCR and SCR.
While SNCR and, to a limited extent, SCR have been used on coal-fired boilers, these processes
have not been demonstrated on a unit with the same configuration as the wet-bottom, continuous
slagging, pulverized coal furnaces operated at PSE&G's Mercer Generating Station.
PSE&G's interest hi NO, control for this unit is the result of our overall understanding of the role
that NOX plays in the formation of ground level ozone, and that the Mercer units are high NOX
emitters. When PSE&G assessed the interim state emission inventory it was determined that
PSE&G's power plants comprised 27% of the total NOX emitted in the state of New Jersey. As a
result of this emission profile, PSE&G established a public commitment to reduce NOX from our
power plants by 60% by 1995 and 80% by the year 2000. Following the SNCR demonstration in
1993, it was concluded that a greater level of NOX reduction would be necessary for New Jersey to
achieve attainment of the ozone standard. This is because of the severity of upwind transported
ozone into New Jersey at levels far in excess of the National Ambient Air Quality Standard.
PSE&G subsequently proposed to EPA and the Ozone Transport Commission that a regional NOX
standard of 0.2 Ib/mmbtu be adopted on power plants throughout the Ozone Transport Region. The
Mercer project demonstrates that emission reductions of the magnitude proposed can be obtained in
a cost-effective manner.
TJI 1993, PSE&G, in partnership with EPRI, conducted a three month demonstration of urea based
SNCR on Mercer Unit 2. Results of this demonstration were presented at the 1994 EPRI NOX
Control Workshop, May 11-13 hi Scottsdale, Arizona. As a continuing step in the evaluation of
post-combustion NOX control technologies, PSE&G installed a Titanium-Vanadium plate type
catalyst in the south (B) duct of furnace 22 between the economizer outlet and air heater inlet,
complete with an ammonia injection grid and sootblowers. In addition, the 22 B hot side air heater
baskets were replaced with a similar catalyst material. Evaluation of the SCR system began hi June
and was completed hi October, 1994. In order to further evaluate NOX control technologies,
PSE&G set up an additional demonstration program to evaluate a combined SNCR/SCR hybrid
system.
The initial goals of the hybrid system evaluation program were:
• Demonstrate the technology for PSE&G and the industry.
• Reduce the load on the catalyst by maximizing the pre-catalytic NOX reduction
• Identify/quantify NOX concentrations before and after the catalyst
• Identify/quantify NH3 slip and effect on the fly ash.
• Minimize the impact on plant operations.
Evaluation of the urea based SNCR/SCR hybrid system was conducted during the last quarter of
1994.
-------
Unit Description
Mercer Generating station has two Foster Wheeler continuous slagging, twin-furnace steam
generating units. Units 1&2 are identical with a rated capacity of 321 MW net each. Each unit is
designed for 2,060,000 Ib/hr of superheated steam flow at 2456 psig and 1050°F. Reheated steam
flow is 1,760,000 Ib/hr at 445 psig and 1050°F. In this twin furnace design, one furnace generates
superheated steam and the other reheated steam. Both units were originally designed with
pressurized furnaces and have been converted to balanced draft with some modifications to the
radiant superheater to allow for operations at 25%, load without adverse effects on boiler
operations. Figure 1 shows a side view of the reheat furnace on Unit No. 2.
The Mercer units burn low sulfur Eastern bituminous coal as the primary fuel, with natural gas
available for start-up and as a secondary fuel. Fuel enters each furnace through twelve front wall
mounted burners arranged in three levels of four burners per level. There are three water tube
division walls located above the top burner elevation and centered between each row. During coal
firing, the furnaces are wet-bottomed, continuous slagging by design. The burners are arranged
close to the furnace floor to keep the slag in a molten state at all loads. While it is possible to
introduce coal and gas through the same burner at one time, general plant practice has been to
operate each row of burners on either coal or gas. Fuel distribution between the furnaces is
controlled to maintain the appropriate superheat and reheat temperatures.
Furnace gases exit the boiler through a convection superheater in one furnace and a convection
reheater in the other. Economizers are installed after both the convection reheater and superheater;
however, the economizer in the superheat furnace has more surface area. The flue gas splits into
two ducts after each economizer to transfer the gases around the slag tank under each furnace. The
flue gases pass through four Ljungstrom horizontal shaft regenerative air heaters (two per furnace)
and are ducted to a multi-chamber electrostatic precipitator.
Description of Stand Alone SNCR System
For the 1993 SNCR evaluation, only the reheat furnace of unit 2 (furnace 22) was outfitted with
SNCR. Since the basic design of each furnace is the same, the results on one furnace should be the
same for the other. Nalco Fuel Tech's (NFT) mobile demonstration unit was used to perform the
evaluation of SNCR at four injector levels in the furnace with four injectors per level. There were
two SNCR injection control zones setup for this demonstration, each zone had a separate pressure
and water/chemical ratio controls. Urea used during the demonstration was NFT's NOXOUT
purchased from a vendor licensed by NFT.
SNCR NOX reductions were on average of 38% while maintaining ammonia slip of approximately 5
ppm. Based on the results of the SNCR demonstration program, PSE&G has purchased full scale
systems to be installed on both Units 1 and 2 for compliance with New Jersey's NOX RACT
regulations operative on May 23, 1995.
-------
SUPERHEATER
CONDENSER HEADERS
IHEATER
kDERS
RADIANT SUPERHEATER
FURNACE FRONT, REARS SIDE WALLS
WATERWALL HEADERS
OUTLET
ise'-o
MTER INLET
Figure 1 - Side View of Reheat Furnace on Unit 2
-------
In-Duct and Air Heater SCR
As described earlier, the flue gas from each furnace at the economizer exit splits into two separate
ducts. For this technology evaluation project, catalyst was installed in only the south duct (22B).
This would in theory mean that only one half of the reheat furnace's flue gas passed through the
SCR or one quarter of the total unit's flue gas volume (equivalent to = 80 MWn). Certain
imbalances in the flow characteristics required biasing via a damper in the 22A duct to balance the
flows.
The SCR system was jointly designed by PSE&G, Wahlco Inc., and Siemens and consisted of an
ammonia (NH3) injection system, air soot blowers, in-duct catalyst and catalytic air heater baskets.
The NH3 injection system consisted of two NH3 vaporization skids and an injection grid located in
the flue gas duct just downstream of the economizer. Aqueous NH3, at approximately 27 percent by
weight concentration, was vaporized in a hot air stream provided by dilution air fans and electric
heaters which heat ambient air. The air/NH3 mixture was routed to the injection grid through a
series of throttle (biasing) valves used to balance the NH3 distribution to match the flow and NOX
stratification. Downstream of the injection grid, the flue gas made a 90° turn and entered duct 22B.
Guide vanes were located downstream of the 90° bend to straighten the flow and to balance the
velocity distribution at the catalyst face. This reduces catalyst erosion and pressure drop.
Duct #22B was expanded to approximately three times its original cross sectional area to incorporate
the in-duct catalyst banks. Inside are two separate banks (5 feet hi depth each) of catalyst elements
separated by approximately 5 feet of duct space. This is the first horizontal In-Duct SCR™
application on a pulverized coal unit in the world. The catalyst elements were manufactured in
Germany where SCR experience has been only applied to vertical down flow configurations. The
plate type catalyst was manufactured by Siemens and consisted of a metal substrate covered by
catalyst. The active catalytic materials are Titanium Dioxide and Molybdenum/Vanadium Oxide.
The catalyst plates were oriented vertically to minimize fly ash deposition. The plates were
arranged with metal frames to secure the catalyst in place.
The success of this catalytic system is dependent upon the operation of a complex soot blowing
system. Soot blowers are located ahead of the first and second catalyst banks to periodically clean
the catalyst plates of ash build-up. In addition, horizontal soot blowers are located on the duct floor
(in front and behind each catalyst bank) to re-entrain any settled fly ash into the flue gas flow.
The existing horizontal shaft Ljungstrom air heater was modified by replacing the hot-end baskets
with catalytic air heater baskets. The catalyst elements were similar to those used in the in-duct
SCR except that they were located in the rotating wheel of the air heater, and therefore, slightly less
than 50% of the catalyst was exposed to flue gas at any given time.
During the In-Duct SCR optimization phase of the test program, it was determined that the flue gas
velocity profile at the economizer exit was skewed by a factor of about 3:1 across the duct. The
NH3 injection grid did not have sufficient adjustment to accommodate for the high flow bias hi the
economizer exit. Therefore, a flue gas Turbomix" static mixer was designed/installed by
PSE&G/Siemens. The mixer substantially reduced the flow bias which improved the NH3
distribution and hence, SCR performance. The mixer also reduced the pressure drop across the
system apparently by reducing flow separation at the 90° bend in the duct.
-------
Description of SNCR/SCR Hybrid System
The SNCR/SCR Hybrid process consists of utilizing both SNCR and SCR NOX control technologies
simultaneously to achieve a final NOX reduction target. The SNCR portion of the system was
operated in various modes in concert with the In-Duct SCR system at Mercer. The modified SNCR
process utilizes more of the SNCR temperature window providing more effective NOX reduction
with increased reagent utilization, while simultaneously producing process ammonia (via SNCR NH3
slip) to feed the downstream catalyst.
The Nalco Fuel Tech demonstration trailer was installed on Unit 2 with urea injection equipment set
up on the same four furnace levels utilized in the 1993 SNCR demonstration (identified as levels 5,
6, 7. and 8). In the hybrid demonstration, injection was on the "B" side (south half) of Furnace 22
and injectors were located in the convective passes on the south side of Furnace 22. Four additional
ports were installed on the rear of the Boiler at Level 8 (identified as 8R) using standard flow
injectors designed to spray between the support tubes of the down pass. One additional port
location was tested on the south sidewall of the boiler utilizing a Multiple Nozzle Lance (MNL
level). The MNL was a stationary water cooled injector designed to provide a very fine spray of
reagent from multiple nozzles installed along the length of the probe.
Three water tube division walls in the furnace divided gas flow into four straightened flow paths
from side to side. The injectors for the SNCR system were aligned to inject between the division
walls. By injecting through ports only on the south side of the furnace, the effects of the process
were confined primarily to the "B" duct or south side of Furnace #22. The flue gas mixer installed
at the economizer outlet in front of the NH3 injection grid provided mixing of both the NOX and the
ammonia slip.
The additional convection section wall injector (8R) and the MNL ports were designed to take
advantage of the greater NOX reduction capability and reagent utilization available at temperatures
lower than would be used for stand alone SNCR (below 1800° F). Together with the existing
SNCR injection locations, the SNCR/SCR Hybrid injection strategy was designed to provide stages
of chemical injection which yield more in-furnace NOX reduction with attempts to achieve a uniform
ammonia distribution to the catalyst. In some of the tests, reagent feed for catalyst reduction was
provided by deliberately utilizing injection at temperatures below the SNCR window limit to
produce NH3.
During the demonstration, the SNCR/SCR Hybrid system was operated in the following
configurations:
• Maximum SNCR in-furnace NOX reduction. Ammonia feed for the catalyst was provided
by the SCR 's Ammonia Injection Grid (AIG).
• Maximum SNCR in-furnace NOX reduction with ammonia feed for the catalyst provided
by high NH3 slip from the SNCR.
• SNCR operated to provide maximum NH3 feed to the catalyst, with in-furnace NOX
reduction a secondary benefit.
• SNCR Operation to provide NH3 to the catalyst while maximizing overall reagent
utilization.
• Maximum SNCR in-furnace NOX reduction without concern for ammonia slip.
-------
Description Of Test Programs
In order to evaluate this NOX reduction technology, a comprehensive test program was developed.
Carnot was contracted in 1993 to assist PSE&G in the evaluation of the SNCR process and the SCR
and SCR/SNCR Hybrid systems in 1994.
The programs were similar in that baseline NOX emissions as well as CO, CO2, O2, SO2, SO3, NH3,
and N20 data was collected at various loads on coal, gas and combinations of the two fuels.
Figure 2 shows the relative sampling locations.
A multi-point probe grid was installed in the ductwork upstream of the catalyst, between catalyst
banks and downstream of the in-duct catalyst by PSE&G prior to the testing program. The grid
consisted of a series of stainless steel probes that transported the flue gas sample under vacuum to
Carnot's CEMS system for determination of NOX, 02, CO, C02 and N2O. NH3 emissions were
determined by extracting a metered volume of flue gas through a dilute sulfuric acid solution. The
NH3 concentration of the sample was then determined by spectrophotometry.
For each of the test conditions composite measurements of NOX and O2 were made at Location A
and Location D to determine the NOX reduction performance of the In-Duct SCR. NOX concentration
was determined by measuring NO from a composite sample and taking periodic single point N02
measurements. For many of the test points NOX and O2 were also measured at Locations C and E.
This provided data to evaluate the performance of the first layer of in-duct catalyst and the air heater
SCR, respectively.
Urea Based Hybrid SNCR/SCR Performance
The performance of the Hybrid SNCR/SCR system is dependent upon the unit operations. The
primary factors which influence the performance of the hybrid system are the load (catalyst space
velocity, furnace and duct temperatures, and NOX concentration), fuel burned (NOX concentration,
flue gas constituents), and the injection configuration (furnace NOX reduction, ammonia production,
NOX and NH3 stratification). As with the In-Duct SCR and the catalyst air heater, the greatest
demands on the hybrid SNCR/SCR system performance are during full-load, coal-fired operation
due to the high space velocity and high inlet NOX concentration.
During the demonstration, three injection configurations were tested and will be discussed. One
configuration employed the in-furnace injectors used during the previous SNCR demonstration.
These injectors were operated at increased chemical flow rates in an attempt to provide high furnace
NOX reductions and the ammonia necessary for the SCR process. Total chemical utilization was
greater in this operational mode than in traditional SNCR. Another method of operation utilized the
in-furnace SNCR injectors with the chemical flow rate set for maximum NOX reduction and
utilization, typically an NSR of 1.1 to 1.3, with the required NH3 reagent provided by the AIG.
The last operating mode involved injecting urea into the back pass just upstream of the economizer.
Special ports and injectors were used that would not typically be employed with an SNCR system
due to the low temperatures at the location of the chemical release. This configuration resulted in
the highest chemical utilization.
-------
PSE&G MERCER IN-DUCT SCR/SNCR PROJECT
SNCR - Selective Non-Catalytic Reduction
FURNACE #22
CD
AIR HEATER
WITH HOT-END
CATALYST
SECTORS
SCR
CATALYST
BANKS
SCR - Selective Catalytic Reduction
AMMONIA
(NH3)
INJECTION
GRID
PSE&G Mercer In-Duct SCR System and SNCR Test Locations
Figure 2.
-------
Figures 3 and 4 show a summary of the SNCR and SCR NOX reduction as a function of the NSR for
the injection configurations tested during full load coal operation. It can be observed in Figure 3
that the NOX reduction falls onto three curves for the different injection levels. In Figure 4, the
injection levels that deliver chemical to the coolest regions of the furnace, MNL and 8R, provide the
greatest overall NOX reduction with the highest chemical utilization when combined with the full
scale SCR. Injection through these levels provided chemical utilizations in excess of 75 percent and
NOX reductions in excess of 90% when not considering limits on the NH3 slip. The overall NOX
reduction provided by levels 8/8R is less for a given NSR than that of the other high ammonia
production levels due to the lower utilization of level 8 as a result of the higher temperatures at the
point of chemical release.
Figure 3 shows the furnace NOX reduction possible with the different injection configurations. As
observed in this Figure and with Figure 4, the two injection configurations that provided the best
combination of high chemical utilization and high overall NOX reduction (Figure 4) do not
correspond to the best furnace NOX reduction. The MNL/8R configuration provides about 20
percent in-furnace NOX reduction at an NSR of 1.1. The low furnace NOX reduction is a result of
the low temperature at the point of chemical release. The temperature at the release point is well
below the optimum temperature required for the NOX reduction reactions but is high enough for the
efficient production of NH3. This results hi an injection configuration that is very efficient at
producing NH3 to supply the catalyst but is inefficient at reducing NOX prior to the catalyst.
Levels 5/6/7 as shown in Figure 3 provide in-furnace NOX reductions of up to 35 percent at an NSR
of 1.6. Increasing the injection rate beyond this results hi a rapid decrease in the chemical
utilization. During the hybrid tests which used the AIG, the urea was injected hi the furnace at an
NSR of 1.1 to 1.3 (which is similar to the operating conditions determined during the SNCR
evaluation performed in June - September 1993) to provide furnace NOX reductions of approximately
30% and minimize the chemical usage.
Use of levels 5/6/7 and the AIG provided NOX reductions hi excess of 90%, however a greater
amount of chemical is required to achieve NOX reductions similar to MNL/8R. This mode of
operation was set for maximum NOX reduction in furnace with minimal NH3 slip going to the In-
Duct SCR and would best simulate the operation of the Hybrid SNCR/SCR system as a catalyst
management tool.
Although high overall NOX reductions are possible with both the MNL/8R and levels 5/6/7 used
with the AIG, the NH3 slip after the In-Duct SCR is higher for the MNL/8R configuration for a
given NOX reduction. This is a result of the poor NH3/NOX distribution at the catalyst face. Since
the NOX distribution was determined to be uniform at the inlet to the catalyst during baseline testing,
any stratification resulting from the SNCR or SCR NOX reduction process may be seen by observing
the NOX stratification at the outlet of the catalyst as long as the injection rate is sufficiently low so
that no NH3 passes through the catalyst unreacted.
Figure 5 compares the NOX stratification at the outlet of the catalyst for the MNL and 8R injector
configurations. The location of the SNCR injector plays an important role in the NH3 slip
distribution across the In-Duct SCR catalyst face. The influence of the injector location can be
clearly viewed hi the figure, with the MNL providing the highest NOX reduction at the bottom of the
catalyst face (Dl,5) and the 8R injector providing the most reduction at the top of the catalyst face
-------
(D4,8). Reference to Figure 2 shows that this corresponds to tuc piatciuem or me two injection
locations in the back pass of the furnace. As shown in Figure 5, use of the MNL and 8R injectors
together improves the NOX reduction profile, however it is difficult to achieve the level of NH3/NOX
distribution necessary for optimum catalyst efficiency as provided by the AIG. Although it is not
shown in this figure, it should be noted that the NOX reduction associated with the MNL injector is
primarily due to ammonia production and reaction in the in-duct catalyst. The NOX reduction
associated with level 8R is more evenly divided between SNCR and SCR reduction.
Levels 5/6/7 were also tested at high injection rates in an effort to provide high furnace reductions
while producing sufficient NH3 to supply the catalyst. Levels 5/6/7 were chosen due to the high
levels of NOX reduction provided during the SNCR demonstration performed in 1993. Reference to
Figure 4 indicates that these injectors, in addition to levels 7 and 8 which were also located in the
furnace, distribute the chemical in regions that are not conducive to NH3 slip production. At
reduced loads, higher levels of ammonia production were observed, but the chemical utilization was
not sufficient to make in-furnace injection a feasible operating mode.
Several hybrid injection configurations are capable of providing high overall NOX reductions with
chemical utilizations in excess of 50%. To more clearly differentiate between the operating modes,
the configurations with the highest overall NOX reduction are shown in Figure 6. The figure shows
a comparison of the NOX reduction and chemical utilization during full load coal firing for the
different hybrid configurations. The ideal curve which represents the NOX reduction that would be
achieved with 100% chemical utilization and the conventional In-Duct SCR system performance are
included for comparison.
Of the Hybrid SNCR/SCR tests, the MNL/8R injection configuration provided the best performance
at chemical flow rates corresponding to an NSR of approximately 0.8; however, above this injection
rate, the high ammonia stratification at the inlet to the catalyst limits the overall NOX reduction
performance.
It is also interesting, and somewhat deceiving, to observe the performance of levels 5/6/7 used in
conjunction with the AIG. The advantage of this configuration is that furnace NOX reductions in
excess of 30% can be achieved without altering the NH3 distribution to the catalyst. The chemical
utilization for this injection configuration is not fixed, but may approach the In-Duct SCR curve if
the SNCR injection rate is decreased. The system could be operated with little or no SNCR
injection while the catalyst is new and benefit from high chemical utilizations, and then as the
catalyst efficiency decreases overall NOX reduction could be maintained by increasing the SNCR
injection.
Summary
The test programs at Mercer Unit No. 2 demonstrated the ability of SCR and Hybrid SNCR/SCR
systems to provide moderate to high NOX reductions while firing low sulfur coal or natural gas.
The variables affecting the performance of each system, and any operational impacts were
determined.
The following is a summary of the performance of the systems along with any associated boiler
impacts:
10
-------
• Operation of the hybrid system by injection into cooler regions of the back pass provided overall
NOX reductions of up to 85 % with an NH3 slip of less than 10 ppm prior to the air heater
catalyst at full load during coal operation. Overall chemical utilizations of up to 90 percent were
demonstrated, approaching that of the In-Duct SCR operated with the AIG.
• Use of injection levels MNL/8R provided SNCR NOX reductions of 20% during full load coal
firing while providing sufficient NH3 feed for the catalyst with an overall NOX reduction of 80
percent before the air heater.
• Furnace NOX reductions of 35% on average and overall reductions in excess of 90 percent were
achieved with use of the hi furnace injectors and the AIG. The chemical utilization was lower
than during injection into the back pass, averaging 50% at high overall NOX reductions. This
mode of operation provided the best SNCR/SCR Hybrid distribution of ammonia across the
catalyst face.
• The In-Duct SCR operating with the AIG provided higher levels of NOX reduction for a given
NH3 slip than the hybrid system utilizing urea. The poor furnace NOX reduction and NH3
stratification at the inlet to the catalyst reduced the effectiveness of the hybrid system.
• From the In-Duct SCR system testing, the overall NOX reduction was increased by approximately
20% while maintaining the NH3 slip limit by using the catalyst air heater. The NOX reduction
across the air heater was small, averaging 20 ppm, but allowed higher reductions to be achieved
by the in-duct catalyst through operation at higher NH3/NOX mole ratios and at elevated NH3
slip.
• Soot blowing performed twice each day maintained the pressure drop across the horizontal in-
duct catalyst at acceptable levels over the course of the program.
• S02 to SO3 conversion was moderate, averaging 1.7 percent, or approximately 6 ppm. The
impact of the conversion rate was mitigated as the 863 formed reacted with the calcium in the
fly ash minimizing the detrimental effects of SO3.
11
-------
PSE&G Mercer Station SNCR/SCR Hybrid Demonstration
A Furnace Injection
* Furnace/Convection Pass Inj.
• Convection Pass Injection
0.2 0.4
0.8 1
Furnace NSR
1.2 1.4 1.6 1.8
Figure 3. SNCR Reduction Measured at Location B as a function of NSR, Coal Firing 310 MW net
-------
PSE&G Mercer Station SNCR/SCR Hybrid Demonstration
A Furnace Injection
• Furnace/Convection Pass Inj.
• Convection Pass Injection
+ Furnace Injection with AIG
0.2 0.4 0.6
0.8 1 1.2 1.4
Furnace NSR
Figure 4. Overall reduction as a function of NSR, Coal Firing 310 MW net.
-------
PSE&G Mercer Station SNCR/SCR Hybrid Demonstration
0)
oc
en
X
O
XJ
0)
_N
"S
E
o
Z
o
o
Q 0.9
D1.5
D2.6 D3.7
Sample Location
D4.8
Figure 5. NOx Stratification at Outlet of SCR, Coal Firing 310 MW net
-------
PSE&G Mercer Station SNCR/SCR Hybrid Demonstration
Decreasing Chemical Utilization
0.5
1
Furnace NSR
1.5
• 5/6/7, AIG
• MNL/Level 8R
A AIG
• 8/8R
Figure 6. Comparison of NOx Reduction and Chemical Utilization during Full Load Coal Firing
-------
CONCLUSIONS
The SNCR/SCR Hybrid has been successfully demonstrated at PSE&G's Mercer Generating station.
Operation of the SNCR in combination with the In-Duct SCR have been determined to be an
effective means of reducing NOX emissions for compliance with New Jersey CAAA compliance
regulations. By controlling the amount of ammonia slip prior to the air heater to 5-10 ppm, air
heater fouling will not be a major concern.
The SNCR/SCR Hybrid testing has demonstrated the following:
• Deterioration of SCR catalyst performance can be compensated for by use of SNCR without
significant plant operational impact. This in effect will allow PSE&G to defer catalyst
replacement costs, if necessary.
• When ammonia slip is not a concern, due to the presence of catalyst, the SNCR injection profile
can be modified to take advantage of a greater portion of the SNCR temperature window. Thus,
higher in-furnace NOX reductions are possible with higher in-furnace chemical utilization rates
than with traditional SNCR applications. The SNCR operations can be configured to provide
NH3 as a feed for a downstream catalyst, while still providing some in-furnace SNCR NOX
reductions. When considering for this mode of operation, the NH3/NOX distribution across the
catalyst face must be a design criteria in a planning commercial Hybrid SNCR/SCR system.
• Depending on the specific site, a Hybrid SNCR/SCR system can be included in the design of an
In-Duct SCR system. Use of the SNCR/SCR Hybrid could reduce the overall catalyst size
making it fit in existing duckwork or resulting in smaller SCR reactors. Supplemental catalyst
ammonia injection (AIG) may or may not be necessary depending on the SNCR injection
configuration, and the desired level of overall NOX Reduction from the combination.
Acknowledgements
(1) In-Duct SCR is a trademark of Public Service Electric and Gas Company.
(2) Turbomix is a registered trademark of Siemens.
16
-------
EXPERIENCE AND CONSIDERATION OF SNCR-SCR HYBRID SYSTEM
T. Fujino
S. Kaneko
K. Suyama
Mitsubishi Heavy Industries, Ltd.
T. R. von Alten
Cormetech, Inc.
May 16-19, 1995
Kansas City, MO
Abstract
SNCR-SCR Hybrid Systems combine deNOx technologies to overcome the limitations of
SNCR-only applications, such as low deNOx efficiency and high concentration of residual
reagent, e.g., slip ammonia, while offering, in some cases, lower installation costs than SCR-only
applications. Successful application of an SNCR-SCR system must consider:
• Boiler characteristics.
• Required deNOx performance.
• Optimum injection system.
• Combination/interface between SNCR and SCR.
Mitsubishi Heavy Industries (MHI) has applied hybrid systems technology on two full scale
utility boilers. A discussion of this experience is presented herein.
-------
Introduction
Mitsubishi Heavy Industries, Ltd. (MHI) has developed several NOx emission control
technologies since the 1970's (Table 1). MHI's NOx control developments include low
NOx combustion firing systems, NOx reduction by oxidation (wet process) and NOx
reduction by NH3 injection (dry process). MHI has acquired considerable experience
commercializing these technologies.
• Combustion
Modification
• [Wet Process
- Dry Process-
J| Low
H
SCR] L
c
NOx Burner
1 MACT (Reburnina)
M'HTT® ) (HYBRID--SNCF
(NOx + O3-»HNO3)
JPeltet Type Catalyst
f
1
t + SCR)
_J
1 1 Plate Type Catalyst
Honeycomb Type Catalyst
1
/
Year
'70
'80
'85
'90
'95
Table 1
History of Mitsubishi deNOx Technology
This paper discusses the combination of two commercially available technologies used to
control NOx: Selective Non-Catalytic Reduction (SNCR) and Selective Catalytic Reduction
(SCR). Both methods utilize a reagent such as ammonia (NH3). SNCR requires for the
injection of the reagent into the furnace at high temperatures. While this method requires
less equipment and thus requires comparatively less capital investment, its, performance is
generally limited to 20 to 50% NOx reduction and results in high residual NH3. Further,
where SO3 is present, high NH3from SNCR combines with SO3 to form ammonia salts
which may foul downstream equipment. SCR has earned its reputation as a leading NOx
control technology by achieving up to 95% reduction with very low NH3 slip (< 5 ppm) in
full scale units. When SCR is installed downstream of an SNCR system, the catalyst
reduces not only NOx but also un-reacted NH3 from the SNCR thus making the hybrid
system a viable solution. MHI applied its hybrid system to two units in the 1970's and
subsequently patented its technology.
Page
-------
Cormetech, Inc., a licensee of the Mitsubishi SCR technology, engineers, manufactures,
and tests catalyst for SCR applications. Cormetech has participated in two tests of hybrid
systems: a slip stream demonstration on a coal fired unit and a retrofit on a package
boiler. The process and MHI experience are described in this paper.
SNCR Experience
The SNCR Process
The SNCR reaction requires the presence of NH3, NOx, and oxygen within an appropriate
temperature range. The reaction mechanism predicts that NH3 initially reacts with oxygen
at higher flue gas temperature, creating NH2 radicals which in turn react with NO to form
nitrogen and H2O.
Primary Reactions:
4NH3 + O2 -> 4NH2 + 2H2O (1)
NH2 + NO -> N2 + H2O (2)
Net Reaction:
4NO + 4NH3 + O2 -» 4N2 + 6H2O (3)
The main factors influencing the deNOx reduction reaction in SNCR are working gas
temperature and residence time. As shown in Figure 1, under laboratory conditions the
required reaction will occur at gas temperatures of 800°C to 1,100°C, with the most
efficient reactions at 850°C to 1,050°C. Some NH3 is converted to NO, N2, and water
requiring the higher NH3:NOx molar ratio (2:1) than the 1:1 implied in equation 3 above.
SNCR Experience
In most operating commercial boilers, deNOx efficiency with this process will be lower than
the high efficiency shown in laboratory test data because of insufficient residence time at
the appropriate gas temperature. Figure 2 shows the effect of residence time in laboratory
tests. The minimum time forthe complete reaction is at least 0.1 second, preferably more
than 0.4 second. In actual operating conditions, flue gas temperature cools rapidly as gas
contacts heat transfer surfaces such as superheaters, reheaters.
Page 2
-------
r :0.4 SIC
NHi/NOx-2
0* = 2%
—•— NOx rtductlon
—O- NH» reduction
1500
Working gas temperature
Figure 1
Effect of Working Gas Temperature
100
c
o
**
o
3
•o
0»
n
I
2_
x"
O
Z
60
40
20
> NOx reduction %
—O— NHa reduction %
NO :200 PPM
NHa/NOx-2
02 :
0.2 0.4 0.6 0.8 1.0
Residence time(r)
1.2
Figure 2
Effect of Residence Time
PageS
-------
Additionally, actual operating conditions do not meet the ideal conditions of lab tests; the
process of diffusing and mixing flue gas flow with injected NH3 is not ideal. This diffusion
and mixing is critical to deNOx performance. To accomplish sufficient mixing, the system
must provide a suitable distribution of NH3 injection points for the various boiler loads; these
injection points may be located either inside the furnace or on the furnace wall. When
inside the furnace, in the cross-section of high temperature gas flow, the injection nozzles
must be cooled in order to prevent decomposition of the ammonia or overheating of the
nozzles. Figure 3 shows an example of the location and the structure of NH3 injection
nozzles. Figure 4 shows deNOx performance results in field tests. It is clear that the
location of NH3 injection point highly influences deNOx efficiency at each boiler load.
Structure of NHj Injection Nozzle
-Top Wall
-Boiler Tube
No. of Nozxl**
Nozzle F : 3X3-9
Nozzle FF : 3X1-3
Nozzle FM: 4X1-4
Nozzle FH : 3X1-3
Figure 3
Locations of SNCR Nozzles for Mizushima Unit No. 3
If the ammonia injection nozzle is located on the furnace wall, ammonia can be injected
similar to wall sootblower nozzles. However, effective penetration and distribution of
ammonia is difficult with such a simple injection system.
Therefore, the performance of the SNCR system is affected by the characteristics of
the boiler and injection system. The boiler characteristics, such as its type,
performance, structure and heating surface arrangement, operating conditions, etc.,
impact temperature profile and residence time. Based on these characteristics, each
case will present its own unique considerations, such as the location of NH3 injection
nozzles, the nozzle cooling system, NH3 carrier, system, and the gas temperature
fluctuation.
Page 4
-------
60 80 100 0
Boiler Load
1.0 2.0 3.0 4.0
Mole Ratio of NH3/ NOx
Figure 4
Field Results of Mizushima Unit No. 3
SNCR Discussion
SNCR was applied to several units in Japan in the mid-1970's. However most of these are
now out of operation as shown in Table 2. There are two major reasons:
• More stringent environmental requirements, exceeding the capability of SNCR.
• High operating costs due to high consumption of ammonia, blowing medium and
auxiliaries.
In case of fuels containing sulfur, NH3-sulfur compounds such as ammonium bisulfate
may deposit in lower gas temperature zones, including the air preheater, fouling the heat
transfer surface and increasing pressure drop.
Urea can be used as a reagent instead of ammonia, as urea rapidly changes into
ammonia at high temperatures in the furnace. However, use of urea may be more
expensive than ammonia and may generate additional byproducts including carbon
monoxide.
-- *
To summarize the advantages and the disadvantages of SNCR systems:
Advantages:
1. Low gas side pressure drop.
2. Simple equipment and small space requirements.
3. Possible combination with other processes to design most flexible system.
Pages
-------
Unit Name
Mizushima l-#2, MCI
Mizushima l-#2, MCI
Mizushima l-#3, MCI
Mizushima l-#4, MCI
Yokkaichi, #1, MCI
Yokkaichi, YD Boiler, MCI
Chita, #2, Chuba Electric
Yokosuka, #4, Tokyo Electric
Capacity
(Flue Gas Flow,
Nm3/H)
175,000
194,000
540,000
540,000
200,000
40,000
1,000,000
(Unit Capacity
350 MW)
1,000,000
(Unit Capacity
350 MW)
Fuel
L.S. Oil
L.S. Oil
H.S. Oil
H.S. Oil
H.S. Oil
H.S. Oil
L.S. Oil
L.S. Oil
N/R
R
R
R
R
R
R
R
R
Start of
Operation
Aug 1974
Apr 1977
Jun 1975
Oct 1974
May 1976
Sep1975
Feb 1976
Mar 1978
Present
Operation
Out of
operation
Out of
operation
Out of
operation
Out of
operation
Out of
operation
Out of
operation
Modification
to repowering
hybrid system
Hybrid system
Out of
operation
N = New R = Retrofit
Table 2
List of SNCR (MHTT) Applications
JSteam Drum
Boiler Circulating
Water for
Nozzle Cooling
Booster Air Fan
for Carrying NHa
FD Fan
Figure 5
Schematic Diagram of the Hybrid SNCR-SCR System
at Yokosuka Unit No. 4
Page 6
-------
Disadvantages:
1. NOx removal efficiency is limited.
2. Larger consumption of reagent and blowing medium, such as air, than SCR.
3. High residual NH3 and resultant ammonium compounds.
4. High operating cost.
Experience with SNCR-SCR Hybrid System
Features and Experience of the Hybrid
The SNCR-SCR hybrid system compensates for the disadvantages of SNCR by
combining it with SCR. A combined system can achieve higher deNOx performance
and lower residual NH3 concentration. Figure 5 shows a schematic diagram of a
combined SNCR and SCR system applied to an oil fired boiler. An in-duct SCR system
is installed downstream the SNCR between the economizer and the air preheater.
Figure 6 shows the test results on this plant's total deNOx performance at a rated load.
The result achieves the predicted performance and significantly reduces residual NH3.
Based on the successful results, MHI applied for and received a U.S. Patent for its
SNCR-SCR hybrid system—U.S. Patent No. 4, 302, 431.
0.5
1JD
tJS
20 2JS 3.O
Mote Ratio erf NHqMOx
Figure 6
Test Results of Hybrid SNCR-SCR System
at Yokosuka Unit No. 4
Page 7
-------
Discussion
1. Control of NHsJSIip.: If the fuel contains sulfur, the compounds formed by reaction of
the residual NH3 and SO3 in the flue gas may plug and damage the air preheater.
Therefore, the SCR design must consider reducing residual NH3 concentration as well
as further NOx reduction.
2. Adequate Mixing of SNCR Residual NH2: In a hybrid system, a larger amount of
NH3 is injected at the SNCR in order to achieve higher SNCR performance. The
improved performance associated with higher reagent injection increases the
residual NH3 from the SNCR. If the NH3 is not well mixed, the higher residual NH3
may cause plugging of downstream equipment because of higher un-reacted NH3
through the SCR.
3. System Optimization: The NH3 consumption rate in an SNCR system is higher than in
an SCR system. This is due to the decomposition of some injected NH3 by its
oxidation at high temperature. Therefore, engineers must carefully consider whether
the maximum or saturated SNCR performance operation is sufficiently economical and
effective.
4. System Characterization: In a hybrid system, a separate NH3 injection grid can be
added at the inlet of the SCR. In this case, the NH3 injection at the SCR inlet can be
equalized to compensate for the amount of residual NH3 through SNCR. As
mentioned in the section discussing SNCR, NOx and NH3 concentration after the
SNCR varies significantly over the load range. During the trial operation, traverse
measurement tests must be performed to quantify the operational characteristics of the
boiler and SNCR to obtain desirable balance of NOx and NH3 concentration at the
SCR inlet over the entire load range. Figure 7 shows an example of the flue gas and
NH3 injection control curve for SNCR at each boiler load. It is clear that the residual
NH3 is relatively high even with the control of NH3 injection for SNCR. Therefore, the
SCR is required to keep residual NH3 under control.
Figure 7
NOx Reduction Performance on 1,225 Ton/Hour Boiler at Various Loads
PageS
-------
Control of NH3 Injection: If the system is applied to a coal firing boiler, which produces
generally higher NOx concentration than natural gas or oil firing boilers, we may expect
higher residual NH3 from the system. However, due to the presence of sulfur, the
residual NH3 concentration must be kept low to prevent the production of NH3-sulfur
compounds at downstream equipment such as the air preheater. When the imbalance
of NOx or NH3 concentration at SCR inlet is large, there will be concentrated areas of
NH3 at the SCR outlet section, resulting in deposition of NH3-sulfur compounds as
mentioned above. Therefore, operators must inevitably control adjustment of NOx and
NH3 by supplementary injection of NH3 at the SCR inlet.
Conclusion
For moderate levels of NOx reduction, the SNCR-SCR hybrid system is available and
may be feasible under certain conditions. However, as SNCR performance is greatly
influenced by boiler characteristics, including boiler load changes, and as SCR
performance in hybrid systems is limited by installation space, the performance of hybrid
system may carry some elements of uncertainty or complexity.
References
1 T. Sengoku, K. Fukahori and S. Kaneko (MHI), "Non-Catalytic NOx Reduction
System MHTT for Steam Generator" presented at the U.S. - Japan NOx Information
Exchange, Tokyo, Japan (May 25-30, 1981). [conference paper]
Page 9
-------
INDUSTRIAL BOILER RETROFIT FOR NOX CONTROL:
COMBINED SELECTIVE NONCATALYTIC REDUCTION
AND SELECTIVE CATALYTIC REDUCTION
Paul W. Groff
Acurex Environmental Corporation
P.O.Box 13109
Research Triangle Park, North Carolina 27709
Brian K. Gullett
Air and Energy Engineering Research Laboratory
U.S. Environmental Protection Agency
Research Triangle Park, North Carolina 27711
Abstract
A 590 kW (2 MMBtu/hr), oil-fired, three-pass, fire-tube package boiler was retrofit with
a combined selective noncatalytic reduction (SNCR) and selective catalytic reduction (SCR)
system and demonstrated 85% nitrogen oxide (NOX) reduction with less than 6 ppm ammonia
slip. A urea-based SNCR solution was injected in the first pass, reducing NOX and providing
ammonia reagent for the SCR. A catalyst housing was designed to fit between the second and
third passes, where the access doors of the boiler normally attach. The SCR catalyst volume of
0.04 m3 (1.5 ft3) provided a space velocity of 10,000 hr"1 at a pressure drop of less than 1.5 cm
(0.5 in) of water. Numerous runs demonstrated system repeatability and ease of operation.
Introduction
The 1990 Clean Air Act Amendments mandated reduced nitrogen oxide (NOX) emissions
for sources that have the potential to produce more than 22.7 metric tons (25 tons) of NOX per
year. Approximately 54,000 industrial, commercial, and institutional boilers currently in
operation in the United States have been identified as sources that produce more than 25 tons of
NOX per year, presenting a need for low cost, high efficiency NOX removal technologies.1
For NOX control technologies to be applied to existing boilers they must be easy to
operate and must not impair the efficiency of the boiler. Some NOX reduction technologies
include selective catalytic reduction (SCR), selective noncatalytic reduction (SNCR), low NOX
burners with advanced over-fired air (OFA), flue gas recirculation, and natural gas reburn which
requires OFA in all cases.
This work attempted to develop and demonstrate a combined SNCR/SCR system for
retrofit application to an oil-fired package boiler. The objective of the tests was to show that a
typical package boiler can be effectively retrofit, optimized, and operated without using
excessive manpower and controls. Successful demonstration of this technology will define an
option for existing NOX control sources and will likely provide redesign parameters for new
applications. This type of SNCR/SCR system has been demonstrated on gas and coal systems at
pilot2 and full-scale levels. A catalyst housing was designed and attached where the access doors
-------
of the boiler normally attach. A honeycomb-type ceramic catalyst with square cells was installed
in the catalyst housing. The SNCR system consisted of a water-cooled, two-fluid reductant
injector nozzle that removed a large fraction of the NOX and supplied reductant to the
downstream catalyst.
Experimental
The Boiler and Retrofit
The SNCR/SCR system was retrofitted to a 590 kW (2-MMBtu/hr), three-pass, North
American package boiler capable of oil and/or gas firing (see Figure 1). The first pass was the
main fire tube, the second pass consisted of 24 6.4 cm (2.5 in) convective tubes, and the third
pass consisted of 20 6.4 cm (2.5 in) convective tubes. The burner was a forced air burner located
at the front of the boiler that used No. 2 fuel oil at an average rate of 53 L/hr (14 gal/hr) and 173
L/s (367 scfm) of air. Typical baseline operating concentrations were 107 ppm NOX, 2.8%
oxygen (O2), 12.1 percent carbon dioxide (CO2), 0 ppm carbon monoxide (CO), 0.6 ppm nitrous
oxide (N2O), and 174 ppm sulfur dioxide (SO2). Generally, nitric oxide (NO) comprises more
than 95% of the total NOX, and NO was used for the NOX numbers reported. The average gas
flow estimated from the fuel consumption rate and the O2 and CO2 concentrations was 177 L/s
(377 scfm), but a pitot tube velocity traverse 4 m (13 ft) downstream from the boiler measured
208 L/s (441 scfm); the difference is attributed to measurement error and in-leakage. The back
of the boiler has an access plate that has a viewport, a thermocouple port, and a
sampling/injection port, the latter centered axially on the 64 cm (25 in) diameter main fire tube.
The SNCR reagent was injected countercurrent (toward the burner) through the
sampling/injection port. Above and to the either side of the burner are access doors that expose
the downstream ends of the second pass (first convective pass) tubes and the upstream ends of
the third pass (second convective pass) tubes. The doors were removed, and the SCR catalyst
housing was mounted where the doors were. A divider plate was installed to redirect the flow
through the catalyst housing (Figure 2).
Sampling System
The sampling system consisted of four sampling lines: a pre-catalyst ammonia (NH3)
(NH3RES) sampling train, a post-catalyst NH3 (NH3SUP) sampling train, an SO2 sample line, and a
sample line for the continuous emission monitors (CEMs). The CEM sample system drew a slip-
stream from the boiler and then pumped the sample to the CO2, CO, O2, and NO on-line
analyzers as well as a gas chromatograph (GC) set up for N2O measurement. N2O can be a
byproduct of the SNCR NOX reduction system.3 N2O is currently not regulated on the Federal
level as an air toxic; however, it is a contributor to global warming through the greenhouse
effect. The sample gas for the CEMs passed through a Hankison chiller and was pumped
through anhydrous calcium sulfate (CaSO4) for removal of water subsequent to the chiller. The
NH3 measurement was aquired by pulling a slip stream of boiler gas through two 1 L impingers
in series immersed in an ice bath. The first impinger contained 100 mL of 0.025 N sulfuric acid
(H2SO4) solution that captured NH3, and the second impinger was dry and followed by a dry gas
meter. The impinger rinse was measured by an ion selective electrode to obtain the NH3
-------
concentration. The ion selective electrode was calibrated using at least three standards before
each set of samples were analyzed, and a spike made from a separate stock was used to check the
calibration before and after each sample set was analyzed.
SNCR
The system. The SNCR injector, reagent delivery/dilution system, and the SNCR
reagent itself were supplied by Nalco Fuel Tech. The SNCR system consisted of a metering
pump and an air atomized reductant injector nozzle. The reagent was similar to NOXOUT A™, a
Nalco Fuel Tech product that consists of 50% urea (NH2CONH2), approximately 50% water, and
small amounts of anti-sealants and dispersants. The water and urea-based reagent were pumped
to a mixing chamber where the total injected liquid amounted to 0.19 L/min (3 gal/hr). The
water/SNCR reagent mix was delivered to the injector where it was atomized with air. The
normalized stoichiometric ratio (NSR) of reagent nitrogen (N), NH3, to baseline N, NOX, was
controlled by varying the flowrate of SNCR reagent. SNCR reagent feedrates were measured
before and after every data set by pumping SNCR reagent from a burette instead of from the
SNCR reagent reservoir barrel. The feedrates before and after the tests were consistent; however,
the flow was not monitored during testing.
Optimization. SNCR reagent injection controlled all the post-installation optimization by
providing the reductant for the SCR. NH3, supplied by the breakdown of NH2CONH2, reduces
NOX in the presence of the catalyst. The objective when optimizing is to find the injection
condition where SNCR removal is most efficient while ensuring that there is sufficient SNCR
reagent to supply enough residual NH3 (NH3RES) for the catalyst to remove the residual, post-
SNCR NOX (NOXRES). The ratio of these two values, NH3RES/NOXRES, defines the pre-catalyst
stoichiometric ratio (SRRES). Preliminary tests varied the amount of air, the injector insertion
distance into the boiler, the type of nozzle at the tip of the injector, and the total flow of water at
a fixed NSR. Although NOX removal was greater at higher water flowrates, the flow was kept
fixed at 0.19 L/min (3 gal/hr) to maintain sufficient boiler efficiency. The temperature profiles
acquired before the retrofit indicated that the optimum SNCR injection temperature was close to
the end of the main fire tube, but when the SNCR reagent solution and carrier gas were injected
they introduced temperature gradients in the main fire tube that resulted in unexpected
temperature profile results. The temperature profiles with the SNCR injection indicated that the
temperature at the end of the main fire tube was much lower (680 °C) than optimum (about
900°C).4 This can be attributed to the liquid from the SNCR reagent injection adhering to and
then evaporating from the surface of the thermocouple, yielding a gas temperature measurement
that is lower than the actual gas temperature. The optimum injector insertion distance for all
nozzles, determined by the SNCR NOX reduction (XSNCR), was between 0 and 20.3 cm (8 in)
from the back end of the boiler.
Six different nozzles were tested: three nozzles had round-orifice diameters of 0.1
(0.0388), 0.2 (0.0775), and 0.4 cm (0.155 in) that provided a cone spray pattern; one nozzle had
six 0.18 cm (0.07-in) holes evenly distributed 0.28 cm (0.11 in) from the center of the nozzle that
provided a cloud spray pattern; one nozzle had an oval orifice that created a fan spray pattern;
and one had an oval orifice at a 45° angle to provide an angled fan spray pattern. In the boiler,
both fan spray nozzles performed better than the cloud spray, which, in turn, performed better
-------
than the cone spray nozzles. The optimum settings for air pressure and injector distance from the
end of the boiler were different for all the nozzles except the two fan spray nozzles. The angled
fan spray with the angle pointing up was determined to be the best nozzle to use. The removal
was greatest with the injector tip 3.8 cm (1.5 in) from the end of the boiler, and the air at 241 kPa
(35 psi). The factors that went into determining the best nozzle were the combined SNCR and
SCR total NOX reduction (XTOT), NH3SLP, and N2O emmisions.
SCR
The SCR catalyst was a commercially available titanium and vanadium ceramic catalyst
with 7.6 square cells/cm2 operating at a nominal level of 10,000 h"1 space velocity [standard
temperature and pressure (STP)]. The catalyst did not appear to degrade over about 240 hours of
operation; it also did not collect soot despite overnight shutdown and morning re-start for each
test, which could cause temperature changes and increase sooting during startup. Sometimes
NH3 can combine with sulfur trioxide (SO3) to form ammonium sulfate [(NH4)2SO4]. Catalysts
can contribute to this formation by converting SO2 to SO3. No drop in SO2 was observed across
the catalyst and no NH3 salts were evident in the boiler. The pressure drop across the catalyst
was measured by a manometer and by a Magnahelic pressure gauge to be nominally 1 cm (0.4 in)
of water.
Boiler Operation
The boiler was always operated at full load. It was started and run for 3 hours with just
water and air injected through the injector to establish temperature, NH3, and NOX equilibrium.
Then the baseline (no reagent) readings were taken and it was run for 3 more hours with the
SNCR reagent flowing. After 3 hours of equilibration time, NH3RES, NH3SLP, and CEM
measurements/samples were recorded. The system was allowed 1 hour to return to equilibrium
and post-test baseline readings were taken.
Results and Discussion
The SNCR system typically removed 30 to 40% of the NOX. Figure 3 shows XSNCR and
the resultant SRRES for varying NSR values. The data points are all using the angled fan spray
nozzle with the nozzle tip located 1.5 in from the back wall of the boiler. Variation of NSR from
1 to 4 shows little effect on XSNCR and no consistent trend for SR^.
The SCR NOX removal performance versus SR,^ is shown in Figure 4. The catalyst
shows consistent NOX reduction (XSCR) trends with increasing SR^ until SR^ reaches values
above approximately 0.7, where XSCR levels off at about 80%. This is a fairly typical trend for a
catalyst performance The points on Figure 4 are a compilation of both fan spray nozzles.
The ratio of :he decrease in NOX to the decrease in NH3 averaged 1.33/1 across the
catalyst. This is usually a 1/1 ratio. NH3 measurements that are too low (due to NH3 breaking
through the impingers, or due to dry gas meter calibrations returning sample volumes that are
higher than actual) could cause the ratio to seem higher than 1/1.
While NSR has little effect on XSNCR, and due to the wide range of testing conditions
reported, no obvious relationship with SR^ (Figure 3), the appropriate SR^ (> 0.7, from
-------
Figure 4) is probably obtained at NSR values around 2 or higher. This is more clear from Figure
5, which shows the system's total NOX removal, XTOT, versus NSR. This shows that under
optimal conditions this retrofit established NOX reduction of 93%. All ammonia slips were less
than 6 ppm, so there is clearly enough catalyst to remove the residual NH3 from the SNCR
process.
N2O formation increases with increasing NSR values (Figure 6). Under fairly typical
operating conditions of NSR = 2, measured N2O emissions are about 10 ppm. These values have
not been accounted for in prior NOX reduction percentages.
The sometimes large degree of scatter in the data are indicative of the varied SNCR
chemical injection rates. Practical operation of an SNCR/SCR system would involve finding the
optimum conditions for each specific boiler and operating within those conditions. Shakedown,
optimization, and long-term operation of this system are ideally suited for a neural network
feedback/control or fuzzy logic control system.
Conclusion
An SNCR/SCR hybrid system retrofit to an oil-fired package boiler had an apparent
optimum NSR around 2 where 85% NOX reduction could be achieved without exceeding 6 ppm
NH3SUP. N2O formation was generally below 15 ppm. At the optimum NSR, it was nominally 10
ppm.
The SNCR/SCR hybrid system was easily retrofit to our existing oil-fired package boiler.
Although the design of the SCR catalyst housing was specific to this boiler, similar designs could
be developed for other units. Equipping new boilers via minor design changes would probably
be even easier and more effective than retrofitting old ones. Once the system is installed,
optimization time may be about 1 or 2 days. Changes in temperature profiles may make
optimization difficult. Current work is developing a fuzzy logic control system to counter this.
Further testing may be needed to assess long term boiler effects and long term catalyst
durability to determine added costs from reagent use and catalyst replacement.
Acknowledgements
The authors acknowledge Nalco Fuel Tech's contribution of the SNCR injector, reagent,
and reagent delivery system. Special thanks to John E. Hofmann and M. Linda Lin (Nalco Fuel
Tech) for technical advice.
Nomenclature
NH,Pt:o The residual ammonia in the boiler after the SNCR reaction measured before the
jKto
catalyst.
NH3SLIP The residual ammonia in the boiler measured after the catalyst.
NSR The normalized stoichiometric ratio of input nitrogen (from the SNCR reagent) to
the nitrogen in the boiler in the form of NOX.
NOvDCQ The residual NOX in the boiler after the SNCR reaction measured before the
.\K_to "
catalyst.
SCR Selective catalytic reduction.
-------
SNCR Selective noncatalytic reduction.
S^RES The stoichiometric ratio of the post-SNCR residual reagent nitrogen, NH3RES, to
post-SNCR residual nitrogen measured before the catalyst, NOXRES .
XSCR The percent reduction of NOX across the catalyst.
XSNCR The percent reduction of NOX from baseline level to the NOX level measured after
the SNCR reaction before the catalyst.
XTOT The percent reduction of NOX from baseline to the NOX level measured after the
catalyst.
References
1. Alternative Control Techniques Document-NOx Emissions from
Industrial/Commercial/Institutional (ICI) Boilers, U.S. Environmental Protection Agency,
Office of Air and Radiation, Office of Air Quality Planning and Standards, EPA-453/R-
94-022 (NTIS PB94-177177), March 1994.
2. B.K. Gullett, P.W. Groff, M.L. Lin, J.M. Chen, "NOX Removal with Combined Selective
Catalytic Reduction and Selective Noncatalytic Reduction: Pilot-ScaleTest Results," L
Air & Waste Manage. Assoc. Vol. 44. pp. 1188-1194(1994).
3. L.J. Muzio, T.A. Montgomery, G.C. Quartucy, J.A. Cole, J.C. Kramlich, "N2O Formation
in Selective Non-Catalytic NOX Reduction Processes," In: Proceedings: 1991 Joint
Symposium on Stationary Combustion NOX Control, Washington, DC, Vol. 2, U.S.
Environmental Protection Agency, EPA-600/R-92-093b (NTIS PB93-212850, (1992).
4. D.P. Teixeira, L.J. Muzio, T.A. Montgomery, G.C. Quartucy, T.D. Martz, "Widening the
Urea Temperature Window," In: Proceedings: 1991 Joint Symposium on Stationary
Combustion NOX Control, Washington, DC, Vol. 3, U.S. Environmental Protection
Agency, EPA-600/R-92-093c (NTIS PB93-212868), (1992).
5. M.L. Lin, D.V Diep, "NOXOUT PLUS-An Improved SNCR Process for NOX Emmission
Control," presented at AIChE Annual Meeting, Miami Beach, FL (1992).
-------
Stack
Front
Catalyst
housing
Main firing tube
(first pass)
Back
SNCR
injector.
Figure 1. The North American package boiler side view with
SNCR/SCR retrofit outlined by dotted lines.
-------
Side view
Front view
Access door
Catalyst
housing
Catalyst
support
Third pass
tubes
Catalyst
housing
Boiler
wall
Divider plate
Flange
Second pass
tubes
Main firing tube
(first pass)
Packing
material
Third pass
tubes
Catalyst
Boiler
wall
Burner
Figure 2. SCR retrofit diagram.
-------
IUU
£ 80
(E
O
to
X
"of 60
>
O
E
0
cc
x 40
0
z:
DC
2 20
0)
0
i i
i i
•^ D A
35 psi Injector Air
^^H
15 psi Injector Air L
~
D
° S A
mWL S *
[^
•
A
i i
0 1 2
_i
& A .
D
A
D
^ • •
^
35 psi Injector Air
15 psi Iniector Air
A
i i
345
0.8
- 0.6
in
LU
DC
- 0.2
0
NSR
Figure 3. SNCR system performance. Angled fan spray nozzle
with injector tip 1.5 in from the back of the boiler.
-------
100
80
DC
O
X
o
^
0)
cr
>
O
60
40
DC
O 20
C/)
0
0
0.2
0.4
0.6
0.8
SR
RES
1.2
1.4
Figure 4. NO x reduction across catalyst for varying SR
NH3RES is between 10 and 55 ppm.
1.6
RE3
-------
100
xO 80
o^
£
X
75" 60
>
o
0
cc
„ 40
20 -
0
0
NSR
Figure 5. NO x reduction for varying NSR. Angled fan spray nozzle
data only.
-------
25
20
E
CL
O
15
10
0
0
NSR
Figure 6. N 2O reduction across catalyst for varying NSR.
Angled fan spray nozzle data only.
-------
HYBRID SCR
T. Jantzen, P.E.
Carnot
15991 Red Hill Avenue, Suite 110
Tustin, California 92680
K. Zammit
Electric Power Research Institute
3412 Hillview Avenue
Palo Alto, California 94303
Abstract
Hybrid selective catalytic reduction (SCR) systems consist of either a combination of SCR
techniques (i.e. in-duct SCR combined with air heater SCR) or selective noncatalytic
reduction (SNCR) in combination with SCR. These Hybrid SCR systems can offer
substantial benefits in reduced cost and enhanced performance; however, their applicability is
very unit specific. This paper presents the results of a study to document the current
experience and develop a tool by which utilities can determine the applicability of Hybrid
SCR to meet their NOX reduction goals, a guideline for selecting the best configuration, and a
reference for developing the design parameters necessary to implement the technology.
Hybrid SCR systems have been installed and demonstrated on utility boilers. The systems
have included in-duct SCR combined with air heater SCR and SNCR combined with SCR as
well as systems where SNCR was combined with in-duct and air heater SCR. This document
includes a review of the results of these demonstrations as well as comments on the
applicability of those results for other utility systems.
Finally this document provides a reference for the development of design parameters for the
implementation of Hybrid SCR. There are a number of technical and commercial
considerations which must be resolved prior to designing or procuring a Hybrid SCR system.
The boiler operating, temperature and emissions data necessary for the final design are
presented along with the process design variables which must be specified. Procurement
suggestions are included to assist the user in addressing some of the more pertinent
commercial issues.
-------
Introduction and Background
Promulgation of increasingly stringent regulations limiting NOX emissions from utility boilers
has created a need for technologies capable of achieving high levels of NOX reduction.
Selective catalatic reduction (SCR) is capable of achieving the desired reductions; however,
high costs and/or technical limitations caused by unique boiler configurations often make
stand-alone SCR a less than optimum solution. Hybrid SCR systems offer the high levels of
NOX reduction necessary for compliance and can often overcome the limitations caused by
unique boiler configurations.
Description of Hybrid SCR Options
There are four different configurations which fall under the general heading "Hybrid SCR."
Each configuration has its own individual benefits and applicability. The first application
would be the application of selective noncatalytic reduction (SNCR) with in-duct SCR. This
application utilizes the NH3 slip from the SNCR process as part or all of the NH3 for the
SCR reaction. This Hybrid SCR configuration can be further categorized according to the
type of reagent used, urea or NH3, and by whether or not supplemental NH3 is injected
upstream of the SCR. The benefits of each of these options will be discussed below.
The second option is to combine an in-duct SCR with an air heater SCR. Typically a single
NH3 injection grid is installed upstream of the in-duct SCR. NH3 slip from the in-duct SCR
provides the inlet NH3 for the air heater SCR.
The third option is the combination of SNCR with an air heater SCR. As with the first
option this Hybrid SCR configuration can utilize either an NH3 or urea based reagent for the
SNCR and supplemental NH3 injection at the SCR inlet. The utilization of an air heater SCR
differentiates this option from option 1 due to the unique characteristics of the air heater
SCR. The fourth option is the combination of SNCR with both in-duct SCR and air heater
SCR.
Benefits of Hybrid SCR
The application of Hybrid SCR must offer direct benefit over alternative technologies
(typically stand-alone SCR) to justify consideration by the end user. The benefits offered by
Hybrid SCR vary according to the configuration selected as well as many boiler specific
parameters. Below we present a qualitative assessment of several of the benefits. A
quantitative assessment would require a boiler specific evaluation including capital and
operating cost estimates, which are addressed in detail in the EPRI technical report.
Higher Overall NOX Reduction Possible without Extensive Unit Modifications. Any
of the four Hybrid SCR configurations discussed above will offer a higher level of NOX
reduction compared to a stand-alone SCR, if the quantity of catalyst is limited by space
available or by allowable pressure drop. This limitation applies to some degree for most
retrofit applications. If the regulatory limits can be achieved within the limitations of catalyst
volume, there may not be a direct benefit to Hybrid SCR. However, in many cases the
regulatory limits will not be achievable without extensive modifications to the unit. In these
-------
cases, the benefit of Hybrid SCR will be avoiding the cost of unit modifications such as air
heater relocation, duct work, and structural modifications necessary for installing an external
reactor or fan modifications to overcome pressure drop.
Lower System Pressure Drop. For a given quantity of NOX reduction, a Hybrid SCR
system will typically have a lower pressure drop than a stand-alone SCR. If the
configuration utilizes SNCR the lower NOX concentration at the SCR inlet will decrease the
volume of catalyst necessary and therefore the pressure drop. Air heater SCR achieves NOX
reduction with little or no increase in pressure drop compared to the original air heater.
Therefore, if the Hybrid SCR utilizes an air heater SCR to reduce the volume of in-duct
catalyst required, the total system pressure drop will be less than if the entire reduction was
achieved with in-duct SCR. Lower pressure drop will reduce operating costs and may avoid
the need for fan upgrades or replacements on fan limited units. In an extreme case on a
forced draft unit which is pressure limited the pressure drop of an SCR could require the
installation of induced draft (ID) fans. Avoiding the cost associated with this extensive
modification would be a substantial benefit if Hybrid SCR could achieve the same reductions
within the pressure limitations of the boiler.
Safer and Less Costly Chemical Storage. This benefit applies only to Hybrid SCR
systems which utilize urea based SNCR systems and do not use supplemental NH3 injection
at the SCR inlet. NH3 is a hazardous chemical whether stored in the anhydrous or aqueous
form. SNCR systems which utilize a urea based chemical as their reagent would avoid the
need for storage of NH3 on-site. While providing a safer work environment, this would also
save the cost of preparing and implementing risk management prevention plans.
Lower NH3 Slip. With a Hybrid SCR incorporating air heater SCR, it is possible to achieve
the same emissions reductions as an equivalent stand-alone SCR at lower NH3 slip. This
may be critical on systems where the slip must be limited to ultra low values to avoid air
heater fouling and or fly ash contamination. It may be possible to achieve the same
reduction in NH3 slip by installing additional catalyst in the duct; however, the amount
necessary may exceed the volume limitations and the catalyst may not be as effective in the
duct as on the air heater baskets.
The rotation of the air heater may offer two benefits, particularly when ultra low NH3 slip is
desired. The first is that the rotation will reduce the impact of stratification of NH3 and NOX
in the flue gas. Frequently when trying to obtain ultra low NH3 slip, the limiting factor is
the ability to adequately mix the NH3 and NOX such that a consistent NH3 to NOX mole ratio
is achieved at the SCR inlet. Even a slight variation will make it difficult to achieve high or
ultra high NOX reduction at an ultra low NH3 slip. As the air heater rotates, unreacted NH3
can be adsorbed within the catalyst in one location and react with NOX in another location.
This benefit is limited as the rotation will only impact stratification in one of the two axis
within the duct.
The second benefit of the air heater rotation is the desorption of the NH3 into the combustion
air. As the air heater rotates from the flue gas into the combustion air stream, a portion of
the unreacted NH3 will desorb into the air stream. The NH3 in the combustion air stream
-------
will enter the boiler through the burner and the vast majority of it will be oxidized to NOX.
On the surface this may seem like a disadvantage as it will increase the baseline NOX
emissions and require greater reductions to achieve compliance with a given NOX emission
limit. However, the increase is small (typically 5 to 10 ppm) compared to the overall
baseline and the benefit of being able to operate the in-duct SCR at a higher slip will usually
outweigh the increase in baseline NOX.
Operational Flexibility. The Hybrid SCR offers the benefit of operational flexibility
allowing the utility to optimize operation for specific, and frequently changing, operating
conditions. Hybrid SCR which includes SNCR can also be operated as a stand-alone SCR by
shutting down the SNCR injection and injecting NH3 only through the grid upstream of the
catalyst. This will reduce overall system performance; however, under certain operating
conditions this will be acceptable and the operating cost savings can be substantial.
The first scenario where this approach would be applicable is to comply with seasonal NOX
limitations. Future regulations are being considered for Phase n NOX emission reductions hi
the ozone transport region which include lower emission limits during the "ozone season"
(May 1 through September 30) than during the remainder of the year. This regulatory
approach creates an opportunity for Hybrid SCR. If a Hybrid system consisting of SNCR
and SCR is designed to meet the lower NOX emission limit, it is likely that the SCR alone
would meet the higher limit. Compared to a stand-alone SCR, which would have to be
designed to meet the lower limits, the Hybrid SCR would have substantially less catalyst,
reducing the initial capital cost as well as catalyst replacement costs. The fact that the SNCR
system would be used only during the five-month ozone season will minimize the drawback
of the higher reagent consumption.
The second scenario where the Hybrid SCR would add operational flexibility is for load
following units with moderate or low capacity factors. Unless the emissions averaging
period is long (i.e., 30 days), most utilities will design the NOX reduction system to achieve
the regulatory limit at the most stringent design point, typically full load. This is to assure
compliance even if the unit must run at full load for an extended period of time. The
majority of the time, however, the unit will be operating at moderate or low loads, which
typically have less stringent NOX reduction requirements. In this scenario the Hybrid SCR
consisting of SNCR and in-duct SCR would be designed to provide the NOX reduction
requirements at full load. At lower loads the SNCR injection rate would be decreased and
below a specified set point would be shut down completely. The load at which the SNCR
system could be shut down would be dependent on the NOX reduction capability of the SCR
which would improve at lower loads due to the lower inlet NOX and lower space velocity.
As in the first scenario the Hybrid SCR approach would reduce the catalyst volume required,
lowering the initial capital costs as well as the catalyst replacement costs. The high
operating cost associated with higher reagent consumption would only be applicable at higher
unit loads.
Drawbacks to Hybrid SCR
Although the above benefits may appear to make Hybrid SCR a very attractive technology,
there are drawbacks which at times will outweigh the benefits. As with the benefits, the
-------
drawbacks vary according to the selected Hybrid SCR configuration as well as unit specific
parameters. Below we present a qualitative assessment of the potential drawbacks.
System Complexity. Hybrid SCR systems are inherently more complex than a stand-alone
SCR system. This increase in complexity will increase the design costs as well as the
operating and maintenance costs and may make the system more difficult to trouble shoot.
Higher Chemical Costs. The costs of chemicals for a Hybrid system utilizing SNCR will
be greater than the chemical costs for a stand-alone SCR achieving the same emission
reductions. There are two factors that affect the cost of chemicals. The first factor is the
higher chemical consumption required for SNCR systems and applies whether the reagent is
NH3 or urea. A Hybrid system utilizing SNCR may require a normalized mole ratio of 1.2
to 1.5, whereas a stand-alone SCR would only require a 0.9 to 1.0 mole ratio. The second
cost factor applies only to urea based SNCR systems and reflects the differential cost for urea
over NH3. The cost, normalized to moles of nitrogen in the reagent is typically 90 percent
higher for urea vs. aqueous NH3 and 150 percent vs. anhydrous NH3 (based on average
reagent costs from vendor survey conducted as part of EPRI Report TR-103885, "SNCR
Feasibility and Economic Evaluation Guidelines for Fossil-Fired Utility Boilers").
Potentially Higher Capital Costs Associated with Air Heater SCR. The capital cost for
an air heater SCR to achieve a given amount of NOX reduction will typically be higher than a
stand-alone SCR. One reason for this higher cost is that to obtain the same space velocity
the air heater will require more than twice the volume of catalyst. This is due to the fact
that slightly more than 50 percent of the air heater wheel is out of the flue gas path at any
given time. Another factor in the capital cost is the higher cost to manufacturer the catalytic
air heater baskets compared to typically stationary SCR catalyst. This cost comparison is
based on the cost of installing catalyst in an existing duct. In addition to the higher catalyst
costs, the air heater SCR may require modifications to the air heater wheel or the
intermediate and cold end baskets to accommodate catalytic baskets with a greater depth than
the original hot end baskets. If it is necessary to install a external reactor or to make
significant duct work or support modifications, these cost may outweigh the higher costs for
the air heater SCR.
Utility Experience with Hybrid SCR
Three utility Hybrid SCR systems have been installed and tested in the United States as
demonstration projects: San Diego Gas and Electric's Encina Power Plant Unit 2, Southern
California Edison Company's Mandalay Generating Station Unit 2, and Public Service
Electric and Gas's Mercer Generating Station Unit 2. Additional testing has been conducted
on pilot scale reactors and preliminary designs were completed for Houston Lighting and
Power's Webster Station when that utility solicited turnkey proposals for a Hybrid SCR
system. Presented below is a brief system description for the three utility demonstration
projects, as well as performance testing results. A summary of the Webster solicitation is
included presenting average values for predicted performance and for costs. Finally an
overview of some of the pilot scale testing is presented.
-------
San Diego Gas and Electric's Encina Power Plant Unit 2
Encina Unit 2 is a 110 MW Babcock and Wilcox front wall-fired balanced draft boiler
designed to burn natural gas and oil. Prior to the installation of post-combustion NOX
controls the unit maintained compliance with the San Diego Air Pollution District's
(SDAPCD) NOX emission limits of 125 ppm on natural gas and 225 ppm on oil utilizing
burners out of service (BOOS) and hopper flue gas recirculation.
SNCR System. The SNCR system is a urea based system installed and tested extensively
prior to the Hybrid SCR demonstration. The system injects urea solution through one of two
elevations. The first elevation has four injectors located on the front wall. The second
elevation has six injectors, four of which are located on the front wall and two of which are
located on the side walls. Urea is received and stored as a 48 percent aqueous solution and
is diluted as necessary to maintain a constant injection flow of 6.7 gallons per minute
regardless of mole ratio or unit load.
In-duct SCR System. At the economizer exit flue gas is divided into two ducts (designated
north and south) which route the flue gas to two L'jungstrom regenerative air heaters.
Approximately 186 ft3 of Catalyst was installed in each of the two ducts. The catalyst,
which was manufactured by Grace, is a vanadia/titania formulation in a conventional
honeycomb configuration with a 7.4 mm pitch. The in-duct SCR is designed to operate at a
space velocity of approximately 35,800 hr1 and a pressure drop of 3 in. we.
An NH3 injection grid is installed upstream of the in-duct SCR to supplement the NH3 slip
from the SNCR process. The grid initially consisted of eight pairs of vertically stacked pipes
with orifices drilled on both sides to inject NH3 perpendicular to the flue gas flow. The
injection grid was subsequently modified by increasing the number of orifices and was
relocated further upstream to provide greater residence time for unproved mixing.
Air Heater SCR. The existing air heater is a vertical shaft L'jungstrom regenerative air
heater manufactured by Air Preheater Company (APCO). The hot end baskets were replaced
with catalytic baskets. No additional modifications were made to the air heater intermediate
or cold end baskets. The installation of catalytic baskets did not cause any loss in thermal
efficiency or any increase in pressure drop when compared to the original air heater. The air
heater SCR has a space velocity of approximately 22,800 hr1.
Performance Results. The demonstration project was separated into three different phases.
The first phase of testing was to develop an understanding of the overall design of the Hybrid
system, its compatibility with boiler operations, and the impact of key design parameters on
performance. The second phase was designed to apply the knowledge gained in phase one to
maximize the NOX reduction with only minor hardware changes. The objectives of the third
phase were to maximize performance after significant hardware changes were made (redesign
of the NH3 injection grid and installation of an improved air heater SCR). Table 1 presents
the results of the phase three testing.
-------
Table 1
SDG&E Encina Unit 2
Hybrid SCR - Individual and Combined Performance
Phase m Results
In-duct SCR Performance
Air Heater SCR Performance
SNCR Performance (NSR=3.0)
Unit Load
(MW)
22
70
108
% NO, Red
38
50
41
Outlet NO, (ppm)
56
45
53
(NH,
% NO, Red
66
48
36
,/NO.=1.0)
Outlet NO, (ppm)
19
23
34
(NH,/NO, = 1.0)
% NO, Red
57
26
26
Outlet NO, (ppm)
8
17
25
Hybrid SCR Performance
% NO, Red
91
81
72
Outlet NO, (ppm)
8
17
25
Note: Baseline NO, emissions are approximately 90 ppm for all loads.
-------
Southern California Edison's Mandalay Generating Station Unit 2
Mandalay Generating Station Unit 2 is a 215 MW front wall-fired, forced draft Babcock and
Wilcox boiler capable of burning natural gas or fuel oil. The normal NOX emissions on
natural gas at full load are 160 ppm without any post combustion NOX control; lower
emissions (approximately 115 ppm) can be achieved by operating the unit with additional
burners out of service. The unit was retrofitted with SNCR and air heater SCR under two
separate programs during 1991. Both systems were tested independently and in combination
(Hybrid).
Air Heater SCR. Mandalay Unit 2 is equipped with two vertical shaft L'jungstrom
regenerative air heaters originally supplied by APCO. The air heater SCR was designed to
provide the same thermal efficiency as the original air heater and to increase pressure drop
by no more than 20 percent. The installation of air heater SCR on one wheel only caused a
flue gas flow imbalance which affected the assessment of system performance. This makes
an extrapolation to a complete balanced flow system difficult.
An anhydrous NH3 injection system was installed with a grid upstream of the air heater SCR
inlet. The grid consists of seven horizontal pipes with a total of 165 injection nozzles. The
injection grid was designed and used for testing of the air heater SCR as a stand-alone
technology. The injection grid was not used during the Hybrid SCR testing.
SNCR System. A low-energy, urea-based SNCR system was installed on Mandalay Unit 2.
The system consists of three levels of injectors located on the front wall of the furnace and
one level on the rear wall with four injectors on each level. Urea solution was injected
through only the upper elevation two injectors hi the north half of the furnace during
demonstration of the hybrid system because the air heater SCR was on the north duct. The
injection rate was controlled by an automatic control system based on the unit load and set to
ensure an ammonia slip of less than 10 ppm at the outlet of the air heater.
Performance Results. The component systems, urea based SNCR and air heater SCR,
which comprise the hybrid system at SCE Mandalay Station Unit 2 were each designed as
stand alone NOX reduction technologies. The air heater SCR system optimized and tested
extensively as an independent system prior to the testing as a hybrid system. Table 2
presents the results of full load performance tests for the component systems both as stand
alone technologies and as part of the hybrid system. This table indicates the synergy
achieved with hybrid systems as the performance of the individual components increases
when operated in a hybrid configuration.
Figure 1 presents the Hybrid SCR system performance, as well as the performance of the
individual component systems, as a function of load. As can be seen from the figure, NOX
emission reductions of 82 percent were achieved at full load and exceeded 90 percent for
loads of 50 percent or less.
The test results presented in Table 2 are based on 8 BOOS; however, when the boiler was
operated with 6 BOOS (normal operation) the baseline NOX emissions are higher and the
-------
Table 2
SCE Mandalay Station Unit 2
Independent SNCR/SCR versus Hybrid Performance
Full Load Natural Gas Firing
NOX Reduction (%)(Notel)
Technology Independent Performance Hybrid Performance
Urea Based SNCR 16 30
Air Heater 70 74
Total System 75°**2) 82
Note 1: NOX reductions are from a baseline of approximately 115 ppm (8 BOOS) and in all cases NH3 emissions
were less than 10 ppm.
Note 2: Theoretical system performance based on combined independent performance.
benefit of the Hybrid system is greater. With 6 BOOS at 200 MW, Hybrid SCR with urea
injection only increases performance from 65 percent to 81 percent when compared to air
heater SCR only performance.
Public Service Electric and Gas's Mercer Station Unit 2
Mercer Generating Station Unit 2 is a 321 MW Foster Wheeler continuous slagging twin
furnace boiler capable of firing coal and natural gas. Originally designed as a pressured
furnace Unit 2 has been modified to balanced draft. In the twin furnace design one furnace
generates superheated steam and the other generates reheated steam. SNCR was installed
and tested on Unit 2 in 1993. In June through October 1994 an in-duct SCR and an air
heater SCR were installed on one of four ducts for unit 2 and were tested without SNCR.
The final step in the series of demonstrations was to test the combined performance of SNCR
with the SCR. Two separate test programs were conducted in late 1994 and early 1995, one
with a urea based reagent and one with aqueous NH3.
SNCR System. The system was initially designed with three levels of injection. Each level
has four lances located on the rear wall for the lowest of the three levels and on the front
wall for the upper two levels. During the SNCR only testing a fourth level of injection was
added below the original three with two lances on the front wall and one on each of the side
walls. During the Hybrid SCR testing, additional ports were added to inject reagent at
cooler temperatures.
The reagent used during the SNCR only testing was NO.OUT, a proprietary urea based
reagent licensed by Nalco Fuel Tech. Two independent tests were performed during the
Hybrid SCR testing; the first was with NOXOUT and the second was with aqueous NH3.
During the Hybrid testing with NOXOUT no supplemental NH3 was injected at the injection
grid. During the Hybrid SCR testing with aqueous NH3 as the SNCR reagent, supplemental
NH3 was injected at the NH3 injection grid.
-------
120
100
O
co 80
T3
a
v>
O
60 -
40 --
20 --
0
0
—a-
* Baseline w/ 8 BOOS
—-a-— APH inlet (SNCR Reduction)
- - • *- - • APH Outlet
50
100 150
Unit Load (MW)
200
250
Figure 1
SCE Mandalay Unit 2
Hybrid SCR System Performance as a Function of Load
-------
In-duct SCR System. As stated previously the unit consists of two furnaces each with a
separate flue gas path. At the exit of the economizer of each furnace the flue gas is divided
into two separate ducts. The in-duct SCR is installed on one of these ducts and therefore
treats approximately 25 percent of the flue gas from Unit 2. The in-duct SCR consists of
two banks of plate catalyst manufactured by Siemens. The catalyst is a titanium dioxide and
molybdenum/vanadium oxide formulation designed for low S02 to SO3 conversion. The flue
gas flow through the SCR is horizontal, the first time this had been done on a coal-fired
boiler. The catalyst plates are oriented vertically to minimize the build-up of fly ash.
Aqueous NH3 at a concentration of approximately 27 percent by weight is vaporized in a hot
air stream provided by dilution air fans and electric heaters. The air/NH3 mixture is routed
to the injection grid which consisted of 15 horizontal pipes. During the optimization phase
of the SCR testing it was determined that the flue gas flow stratification was greater than
could be accommodated by the NH3 flow biasing valves. Therefore, a flue gas static mixer
was installed upstream of the NH3 injection grid.
Air Heater SCR. One of the four existing L'jungstrom air heaters was retrofit by replacing
the hot end baskets with catalyst coated baskets. The catalyst elements were manufactured
by Siemens using the same catalyst formulation as the in-duct catalyst. The rotation of the
horizontal shaft air heater creates cyclic stress on the catalytic baskets and increases the
concern for loss of catalyst material. Air heater baskets must be designed to avoid
differential expansion between the catalyst material and the metal substrate. However, tests
baskets installed prior to the testing program did not show adverse impact.
Performance Results. As stated previously several testing programs have been conducted
at Mercer Generating Station Unit 2. The first test program was SNCR only and therefore
does not apply to this Hybrid SCR paper. Three additional programs were conducted which
qualify as Hybrid SCR and are therefore discussed below.
The second program was to test the in-duct SCR and the air heater SCR without SNCR. This
six month program optimized the performance and then tested the two stage configuration of
the Hybrid SCR system under a variety of conditions. It is not practical to present all of the
results in this report; however, we have included two figures which summarize the results.
Figure 2 presents the full load NOX reduction as a function of inlet NH3 to NO, mole ratio
for coal firing. Figure 3 presents full load NOX reduction as a function of NH3 slip for coal
firing. The later figure graphically depicts the benefits of the air heater SCR when installed
in conjunction with an in-duct SCR. A typical SCR will be designed and operated to
maintain a specified NH3 slip limit (either a regulatory limit or a process limitation). As can
be seen from these two figures, the performance for the in-duct and air heater Hybrid SCR
configuration is much greater than for the in-duct alone for a given slip limit.
The third and fourth test programs both tested SNCR in combination with the installed SCR
system. The third program utilized NOXOUT as a reagent and the majority of the testing was
conducted without supplemental NH3 injection through the injection grid. The results of this
testing are summarized in Table 3. The fourth program utilized aqueous NH3 as the SNCR
-------
100.0%
95.0% --
90.0% --
-P 85.0% --
4)
fe 80.0%
^
•S 75.0% --
S. 70.0%
65.0% --
60.0% --
55.0% -
50.0%
x
O
0.85
Hybrid Performance
0.90 0.95
NH3/NOx Mole Ratio
1.00
1.05
Figure 2
PSE&G Mercer Station Unit 2
Hybrid System Performance Coal Firing 320 MW
-------
50.0%
0.0
5.0 10.0 15.0
NH3 Slip (ppmvd ®7% 02)
20.0
D In-Duct SCR (w/o AH-SCR)
* Hybrid SCR (w/AH-SCR)
25.0
Figure 3
PSE&G Mercer Unit 2
Performance of In-duct SCR and Hybrid SCR as a Function of NH3 Slip
Coal Firing 320 MW
-------
Table 3
PSE&G Mercer Hybrid Urea SNCR/In-Duct SCR Evaluation
November 1994
Load
(MWn)
310
310
310
80
310
135
80
Fuel
Coal
Coal
Coal
Coal
Gas
Gas
Gas
Furnace
NSR
0.82
1.15
1.84
1.2
1.63
1.14
1.24
Baseline NO,
(ppm ® 1% OJ
840.5
686.1
918.5
357.9
659.7
226.5
148
SNCR NO,
Red.
30
14.2
31.2
14.1
39.2
12.5
11
Inlet NO, ®
In-duct SCR
(ppm ® 7% OJ
588.7
589
631.5
307.3
401
198.3
131.7
In-duct SCR
NO, Red.
77.5
92.6
94.5
88.2
83.3
92.9
93.9
Outlet NO,
(ppm @ 7% Oj)
132.7
43.5
34.7
36.3
66.9
14
8
Hybrid System
NO, Red.
84.2
93.7
96.2
89.9
89.9
93.8
94.6
NHj Slip In-duct @
SCR Outlet
(ppm @ 7% Oj)
9.9
29.2
11.7
1.3
10.7
4.2
3.3
Table 4
PSE&G Mercer Hybrid NH3 SNCR/In-Duct SCR Evaluation
December 1994 - January 1995
Load
(MWn)
310
310
Fuel
Coal
Coal
Furnace
NSR
1.44
1.35
Baseline NO,
(ppm @ 7% OJ
1050
1080
SNCR NO,
Red.
(%)
41
26
Inlet NO, @
In-duct SCR
(ppm @ 7% Oj)
621.9
799
In-duct SCR
NO, Red.
(%)
85.2
87.8
Outlet NO,
(ppm @ 7% Oj)
92.3
97.2
Hybrid System
NO, Red.
(%)
91
91
NHj Slip In-duct @
SCR Outlet
(ppm @ 7% OJ
17
12
-------
reagent and tested with supplemental NH3 injection. Table 4 presents the results of this
testing.
Houston Lighting and Power's Webster Generating Station
In 1994 Houston Lighting and Power (HL&P) in conjunction with EPRI solicited turnkey
proposals for a Hybrid SCR system at Webster Station. Proposals were received, reviewed
and a vendor was selected; however, due to pending regulatory changes the project has been
indefinitely postponed. The system was to include SNCR, in-duct SCR and air heater SCR
for the 380 MW gas-fired boiler. The proposal effort provided a good assessment of Hybrid
SCR performance and costs for the individual component systems as well as the total system.
Due to the proprietary nature of the proposals and need to maintain confidentially we cannot
publish the individual performance projections or costs; however, we have complied a
summary, Table 5, presenting average NOX reduction and cost (in $/kW) by technology.
Table 5
HL&P Webster Generating Station
Hybrid SCR Proposal Summary
SNCR
In-duct
AH-SCR
Complete Hybrid System
Average Cost
($/kW)
5.0
15.0
8.0
28
Average Performance
(% NOX Reduction)
30
60
54
87
Pilot Scale Hybrid SCR Studies
A number of pilot scale studies have been performed to test and demonstrate the feasibility of
Hybrid SCR. These studies have shown very promising results, with SNCR reduction
exceeding 70 percent and total system reductions exceeding 95 percent at NH3 slip values of
less than 5 ppm. These results, however, are under optimum temperature, mixing, and
residence time conditions and are not directly applicable to full scale commercial
applications. The studies did indicate the potential of Hybrid SCR and provided data to
characterize the impact of such variables as inlet NOX and SNCR reductions on overall
system performance.
Hybrid SCR Applicability
As stated previously the applicability of Hybrid SCR is very unit specific. A number of
different factors regarding boiler design, fuel quality, operating modes, and regulatory
limitations all impact the performance and cost of Hybrid SCR. This in turn affects the
-------
benefits and drawbacks of Hybrid SCR when compared to alternative technologies such as
stand-alone SCR. Presented below is a brief description of key variables affecting
applicability. Further details as well as an applicability decision tree are included in the
EPRI report.
NOX Reduction Required
The first issue to be addressed in determining the applicability of Hybrid SCR is to determine
the level of NOX reduction required. If the emission reductions required are low, Hybrid
SCR may not be applicable as other technologies, such as combustion modifications and
SNCR, may achieve compliance at lower costs.
If the NOX reduction requirements are greater than can be achieved with these lower cost
technologies, either a stand-alone SCR or a Hybrid SCR will usually be required. The
selection between stand-alone SCR and Hybrid SCR will depend on the issues addressed
below.
High Baseline NOX Emissions. The performance of an SCR is dependent on the inlet NOX
concentration. For a given space velocity and NH3 slip value, the NOX emission reduction
performance for an SCR is inversely proportional to the inlet NOX (i.e. for a given volume
of catalyst the higher the inlet NOX the lower the reduction possible). Along the same lines,
to achieve the same reduction a unit with higher NOX will require a lower space velocity and
therefore a greater volume of catalyst. Therefore units with high baseline NOX emissions are
candidates for Hybrid SCR, which can reduce the inlet NOX upstream of the SCR.
Limited Allowable Pressure Drop. The installation of catalyst in an in-duct or
conventional SCR will increase the system pressure drop. This pressure drop, which is a
function of the cross sectional area and depth of the catalyst bed, can be as high as 6 to 8 in.
w.c. The allowable increase in pressure drop is often restricted for existing boilers, due to
fan limitations or furnace pressure limitations. If the SCR pressure drop cannot be
accommodated within these limitations, unit load is frequently curtailed.
Hybrid SCR can address pressure drop concerns in two ways. The first is through the use of
SNCR which reduces the inlet NOX to the SCR and thus the required catalyst volume. This
in turn reduces the system pressure drop. The second way that Hybrid SCR can reduce
system pressure is that air heater SCR typically adds little or no additional pressure drop
beyond that of the existing air heater.
Space Limitations. The length of duct work between the economizer and the air heater on
existing units is often very limited. This was done to reduce original construction costs as
well as to minimize the footprint of the unit. In addition to the short duct length, the plant
layout often further restricts the extent of duct modifications which can be made. The cost
of relocating the air heater and or installing an external reactor can often be a substantial
portion of the total cost of the SCR system. Hybrid SCR can avoid the need to relocate the
air heater or install an external reactor by decreasing the volume of catalyst installed in the
-------
duct. The catalyst volume is reduced by the lower inlet NOX due to SNCR and/or because
some of the catalyst is located within the air heater.
Structural Limitations on Boiler Steel. The installation of catalyst either in an existing
duct or hi an external reactor will increase the loading on existing boiler steel and
foundations. In many cases the support steel and or foundations will have to be modified or
a new independent structure will be required. In all cases the existing structure will have to
be analyzed to determine the impact of the additional loading. The complexity of the
analysis as well as the probability of the retrofit requiring substantial modifications to the
boiler steel are dependent on the following:
• Weight of catalyst
• Weight of reactor or duct modifications
• Height of catalyst above grade
• Seismic zone
• Design margin of original structure
Hybrid SCR can reduce these concerns because the Hybrid SCR will require less catalyst and
therefore less weight. Secondly, the Hybrid design will eliminate the need for an external
reactor, allowing the catalyst to be installed in the existing or modified duct. Frequently the
duct is lower in elevation than an external reactor would be. This will reduce the height of
the new load reducing the modifications necessary to the structural steel and the foundations.
The savings realized from lowering the load are more significant in high seismic activity
areas such as California.
NH3 Slip Limit Less than 5 ppm. There are two sources of NH3 slip limits. The first is
regulatory, and these limits have typically been 10 ppm for SCR systems and 20 ppm or
higher for SNCR. The second limit is a process limit to avoid problems downstream of the
SCR. The potential down stream problems include formation of ammonia salts which can
cause pluggage in the air heater and fly ash contamination. For gas-fired boilers there are
typically no process limits. For coal- and oil-fired boilers the process limit is dependent on
the sulfur content of the fuel; and for coal, whether or not the fly ash is sold. Frequently the
process limit for coal- or oil-fired boilers is 5 ppm or less.
The performance of an SCR is directly proportional to the allowable NH3 slip. The
difference between an allowable NH3 slip of 10 ppm and 5 ppm can substantially change the
performance of an SCR. As an example for the in-duct SCR installed as part of the PSE&G
Mercer demonstration the full-load NOX removal during coal firing increased from
approximately 80 percent to 88 percent when the NH3 slip changed from 5 to 10 ppm. The
same NOX reduction can be achieved with a lower slip by increasing the volume of catalyst.
A Hybrid SCR configuration with an air heater SCR can offer greater benefit than simply
adding catalyst in the duct. As discussed in Section 2.0 the air heater SCR will allow
operation at lower NH3 slip values in two different ways. First, the air heater improves
mixing and thus reduces stratification problems which make achieving low NH3 slip values
difficult. Secondly the air heater will transfer some of the NH3 from the flue gas stream to
the combustion air stream. Although this desorption process will increase the baseline NOX
-------
slightly (typically less than 10 ppm) the overall benefit for system performance will more
than outweigh the increase.
Existing SNCR System. The application of Hybrid SCR to a unit which already has an
SNCR will often provide the most cost effective method of achieving compliance with a new
stricter regulation. If the regulation is established with two or more increments of
compliance, a staged implementation may offer substantial savings. For many units which
do not have the unique design parameters discussed above, a stand-alone SCR will be the
most cost effective solution for a single large reduction in NOX emissions. However, if the
required reductions are incremental, the optimum solution may be Hybrid SCR implemented
by installing an SNCR system first and delaying the installation of SCR until the final
regulatory limits are in force. This will reduce the overall present value cost by delaying the
large expenditure of capital necessary for SCR.
Selecting the Optimum Hybrid SCR Configuration
There are several steps in selecting the optimum configuration for Hybrid SCR. These steps
include establishing the unit baseline data and design parameters, performing technology
assessments to project the performance of each technology, and performing an economic
analysis. Each of these steps is addressed below.
Establishing Unit Baseline Data
The first task in selecting the optimum Hybrid SCR configuration is to establish the baseline
data for the unit to be modified. This data will be used to project the performance of the
individual systems and ultimately the combined performance of the Hybrid system. The
more comprehensive the data collection and the more accurate the data the more reliable the
projections will be. The data required includes the following:
• Uncontrolled Emissions
• Temperature Profiles
• Residence Time
• Fan Capabilities
• Space Available for Catalyst
• Air Heater Configuration and Performance
• Unit Operating Mode and Capacity Factor
The EPRI technical report on Hybrid SCR includes detailed description of the baseline data
requirements, the impact of key variables and suggest methods to obtain the data. This level
of detail was beyond the scope of this paper.
Technology Assessment
Once the utility engineer has compiled the above data it can be used to perform a technical
assessment of the individual technologies as well as the overall Hybrid SCR system. The
EPRI technical report includes a guideline for technical assessment which will determine the
-------
general sizing of the system and major components in sufficient detail to estimate the capital
and operating costs, and will provide performance projections. This, along with the
economic assessment included in the report, will allow the utility to select the optimum
configuration for the Hybrid SCR.
Establishing Design Parameters
Once the determination has been made that Hybrid SCR is applicable to a given unit and the
configuration has been selected, the utility will need to establish the design parameters. The
design parameters will be used in the preparation of a performance specification if the utility
chooses to purchase a system from a vendor or will be used to complete the detailed design
if the utility chooses to design and manufacturer the system in-house. For the purposes of
this study it will be assumed that the Hybrid system will be purchased from a vendor
specializing in the design of Hybrid SCR systems. The work completed under the Technical
Assessment will provide the basis for the majority of the design parameters. Below we
outline the design parameters which should be provided to the system designer as part of the
performance specification.
System Performance
The synergy of the individual systems within the Hybrid SCR can provide performance
greater than the sum of the individual systems. As an example the performance of an SNCR
system will be greater if the NH3 slip is increased to provide the inlet feed for an in-duct or
air heater SCR. When specifying the performance the utility has the option of specifying
individual performance or specifying the overall Hybrid system performance. The advantage
of specifying the overall system performance is that this allows the vendor to optimize the
system to provide the greatest reduction in the most cost effective manner for each unit's
design and operational constraints. Different vendors may have technical skills and/or
intellectual properties that allow them to design systems with better performance in one area.
Except in some unique situations (see below) specifying individual performance would
establish an artificial and possibly unnecessary constraint on the system designs. Specifying
system performance only will provide the greatest flexibility and maximize the options
available for the design. The design criteria which are applicable to the entire system are as
follows:
• Baseline Emissions
• NOj Reduction Requirement
• Allowable NH3 Slip
• Boiler Performance Data
• Fuel Composition
• Unit Operational Parameters
• Bid Evaluation Criteria
-------
SNCR Design Parameters
The following design parameters are specific to the design of the SNCR process:
• Boiler Dimensions
• Time and Temperature Profiles
In-duct SCR
The following design parameters are specific to the in-duct SCR process:
• Flue Gas Flow Rates
• Temperature
• Duct Dimensions
• Allowable Pressure Drop
• Catalyst Life
Air Heater SCR
The following parameters are specific to the air heater SCR process:
• Air Heater Configuration and Performance
• Catalytic Basket Life
Specifying Individual Performance
The discussion of design parameters and specifications presented above is based on procuring
a system from a single vendor and specifying the overall performance. As discussed above
there is substantial benefit to this approach which allows the vendors to individually optimize
the performance of the components. However, there are unique cases where the utility may
wish to specify the performance of the individual systems. These cases would be where for
one of several reasons the utility wishes to utilize only one of the systems under certain
operating conditions.
One example where the utility may wish to specify individual performance would be seasonal
NOX limits that are lower in the summer (ozone season) than in the winter. During the
summer the utility may need the entire Hybrid system to achieve compliance. During the
winter, when the higher limits are applicable the system could operate with just the SCR
portion of the Hybrid system. This would eliminate the operating cost for the SNCR system.
Another example would be for a utility which burned two different fuels such as coal and
natural gas. Burning coal would necessitate higher NOX reduction and therefore the entire
Hybrid system. Firing natural gas would only require the SCR portion of the system, again
saving the operating cost of the SNCR system.
A
units
third example would be specifying individual performance for load following or peaking
its which only operate at high loads a small percentage of the time. Due to higher
-------
emissions and higher flow rates the critical design point for most units is full load. It is
likely that a unit designed to achieve a specified limit at full load would be over designed at
most lower loads. The Hybrid system would be operated at high loads and at lower loads
the SNCR would be shut down to reduce operating costs.
In each of these cases it may be prudent to specify the individual performance; however, it
would still be advisable to work with the vendor to optimize the overall hybrid system
performance.
Commercial Considerations
There are a few commercial considerations to be addressed when purchasing a Hybrid SCR.
The first is licensing of technologies and patent issues. It is not possible within the scope of
this report to fully address the broad array of patent and licensing issues regarding SNCR,
SCR, and Hybrid SCR. The user should be aware that there are patents for each technology,
and combination thereof and that different vendors have rights to different aspects of
different technologies. To provide protection from possible future patent or license
infringement law suits it is advisable that the utility obtain indemnification from the vendor
before purchasing a system. The vendor should also be required to prove his right to offer
any technology and consideration should be given to the ability of the vendor to defend the
utility against any patent or license infringement law suits.
Whether to choose a single vendor for the entire Hybrid SCR system or to purchase the
systems individually will depend on the implementation schedule. If the entire Hybrid
system is to be installed at once a single vendor would be best as they will provide the most
comprehensive guarantee. If the systems are to be installed at different times to comply with
increments of compliance, it may be advantageous to purchase them separately; however;
this approach may lead to a situation where the different vendors blame each other for
shortcomings in the overall system performance.
The warrantee should stipulate system performance at the end of the catalyst guarantee life.
This will include NOX reduction, reagent utilization, NH3 slip, and pressure drop. Given the
expected catalyst degradation it would be advisable to obtain a guaranteed performance for
each technology at the start of life as well as the end of life. This will allow the utility to
determine if the performance is acceptable when the system is installed without waiting until
the system drops below the final guaranteed performance before taking corrective action.
-------
SNCR/SCR Hybrid Capabilities and Liniitations
J. Cochran
Black & Veatch
Paper unavailable at time of printing. Please check the late paper
table in the registration area for a copy or contact the speaker
directly.
------- |