EPRI/EPA 1995 Joint Symposium on Stationary Combustion NOx Control: Book 4 - Friday, May 19, 1995


EPRI
Electric Power
Research Institute

&EPA
May 1995
                     EPRI/EPA1995 Joint Symposium
                     on Stationary Combustion
                     NOX Control
                     Book 4: Friday, May 19,1995
                     Sessions 8A and 8B
                     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 4: Friday, May 19,1995
Sessions 8A and 8B
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 8A
Selective Non-Catalytic Reduction

-------
A SUMMARY OF SNCR APPLICATIONS TO TWO COAL-FIRED WET
BOTTOM BOILERS
Richard Himes
Dennis Hubbard
Zac West
Carnot
15991 Red Hill Avenue, Suite 110
Tustin, California 92680

Jeff Stallings
EPRI
3412 Hillview Avenue
Palo Alto, CA 94303
Abstract

In response to NOX reductions mandated under Title I of the 1990 Clean Air Act Amendments
(CAAA), Public Service Electric & Gas and Atlantic Electric of New Jersey evaluated Selective
Non-Catalytic Reduction (SNCR) for NOX control under separate programs at Mercer Station and
B. L. England Station, respectively. Mercer Station is comprised of twin 321 MW Foster
Wheeler coal-fired wet bottom boilers, with natural gas capability up to 100% load. B.L. England
Station has three units, two of which are cyclone boilers of 136 MW and 163 MW. These furnace
designs are of particular interest in that nominally 23,000 MW of cyclone boiler capacity and
6,900 MW of wall- or turbo-fired wet bottom boiler capacity will be faced with NOX reductions to
be mandated under Title IV - Phase n for Group n boilers.

Both stations evaluated Nalco Fuel Tech's SNCR system using a portable test skid, with urea as
the reducing chemical. The Mercer Unit 2 demonstration was performed with a low sulfur coal
(nominally 0.8%), while the B.L. England Unit 1 demonstration utilized a medium sulfur coal
(nominally 2.4%), and also re-injects fly ash back into the cyclones for ultimate collection and
removal as slag. To address concerns over potential Ljungstrom air heater fouling, due to
reactions between ammonia and SOs in the air heater, and fly ash salability at Mercer Station,
both sites targeted no greater than 5-10 ppmv ammonia emissions at the economizer exit. At
Mercer Unit 2, air heater fouling was only experienced during system start-up when the ammonia
emissions at the economizer exit were estimated at levels approaching 60 ppmv. B.L. England
Unit 1, however, experienced frequent fouling of the air heater. NOX reductions achieved at both
sites ranged between 30% - 40% from nominal baseline NOX levels of 1.1 - 1.6 Ib/MMBtu. Each
site is currently undergoing installation of commercial SNCR systems.

-------
Introduction

The 1990 Clean Air Act Amendments (CAAA) will have significant impacts on the electric utility
industry. Title I of the CAAA adds new mandates for bringing all areas of the country into
compliance with the National Ambient Air Quality Standard. Under Title I, utilities with units
located in areas designated as non-attainment for ozone will be required to reduce NOX emissions
from many of their existing units. SNCR is one option for controlling NOX emissions, with the
current paper focusing on its demonstration on two wet bottom boilers in New Jersey. These
results may be of particular interest to utilities with Phase n - Group n designated boilers (wet
bottom boilers, cyclones, and cell burners) faced with pending NOX rules under Title IV of the
CAAA. Currently there are nominally 23,000 MW of cyclone utility boiler capacity, and 6,900
MW of wall- and turbo-fired wet bottom boiler capacity that will be faced with new NOX rules
under Title IV of the 1990 CAAA.

Public Service Electric & Gas (PSE&G) performed a SNCR demonstration program at its Mercer
Generating Station hi the Summer of 1993 to evaluate its feasibility to reduce NOX. The principle
project objectives were to:

• determine if the SNCR technology is applicable while firing coal, gas, and a
combination of the two fuels,
• document the NOX removal effectiveness,
• evaluate the NOX reduction possible with a maximum ammonia slip limit of 5 ppmv at
the economizer exit,
• assess any impacts of residual ammonia on fly ash resale, and
• establish the cost of a commercial, full-scale operation.

A test program was performed by Carnot to establish baseline emission levels and SNCR
performance at five different load conditions, utilizing low sulfur bituminous coal, natural gas, or
a mixture of the two fuels. The selected boiler loads were 80, 135, 200, 220, and 321 MW, with
initial testing conducted under steady state conditions. During the final two weeks of the test
program, the SNCR test skid was set up to follow load with minimal input from operators. This
trial was run to observe the load-following behavior of the injection system, and to collect data on
chemical consumption during 24-hour operations.

Although SNCR has been demonstrated to work on lower temperature boilers, Atlantic Electric
had concerns that at the higher operating temperatures of cyclone boilers, the addition of NOX
reductants would not be effective at reducing NOx, and could actually increase NOX through
oxidation of the nitrogen in the reagent. Recent kinetic modeling done by Nalco Fuel Tech, and
demonstrated on full-scale units, indicates NOX reductions are achievable above 2000°F with
higher baseline NOX levels. A one month test was done hi the spring of 1993 on B.L. England
Unit 1. Reductions in NOX of about 30% from a baseline of 1.1 Ib/MMBtu were achieved at full
load under steady state operating conditions and with optimized injector configurations. At low
load (87 MWe), NOX reductions over 40% at optimized injector configurations were achieved
from a baseline of 1.0 Ib/MMBtu. No increase in air heater pressure drop was observed during

-------
these short term tests, and NH3 emissions at the stack were less than 1 ppmv while post
economizer slips were maintained at less than 5 ppmv. Higher reductions on NOX can be achieved
but at the expense of increased NH3 slip and potential boiler reliability impacts. Subsequently, a
longer term test was carried out to determine the minimum outlet NOX emission rate that could be
achieved under full dispatch conditions. The results of the latter test, which took place from
December 1993 through March 1994, are discussed in this paper.
Unit Descriptions

Mercer Generating Station has two Foster Wheeler continuous slagging, twin-furnace steam
generating units. Units 1 and 2 are identical and rated at 321 MW net each. Each unit's steam
generator has a capacity of 2,060,000 Ib/hr of steam at a superheat outlet pressure of 2456 psig
and an outlet temperature of 1050 F. The reheat capacity is 1,760,000 Ib/hr of steam with an
outlet temperature of 1050 F and pressure of 445 psig. Twelve front wall-mounted burners are
arranged on each furnace in three levels with four burners per level. In addition, there are three
water tube division walls located above the top burner elevation and centered between each
burner row. Mercer Unit 2 has been converted to balanced draft and modifications to the radiant
superheater have been performed to enable the unit to achieve a minimum load of 25% without
adverse effects on boiler operations. A side view detail of the reheat furnace is shown in Figure 1.
Coal is burned as the primary fuel with natural gas used as both a start-up and secondary fuel. A
typical coal analysis is presented hi Table 1. The important distinction between Mercer and B.L.
England coals with respect to the SNCR demonstrations is the coal sulfur content. Mercer fires a
low sulfur coal of nominally 0.8%, while B.L. England fires a medium sulfur coal of nominally 2.6%.

Unit 1 at B.L. England Generating Station was designed by Babcock and Wilcox and is cyclone-
fired and rated at 138 MW gross. The steam generator has a capacity of 980,000 Ibs/hr of steam
at 1,005°F superheat and 1,005°F reheat temperature. Crushed coal with a maximum permitted
sulfur content of 2.8% on a monthly basis and 2.6% on a yearly basis is fired through three
cyclones. Fly ash is re-injected through either of the two bottom cyclones. A sectional view of
the steam generator is shown in Figure 2.

The combustion air to the unit is supplied by two forced draft fans. The fans are designed to each
provide 60% of the combustion air required to operate the unit at full load. However, during this
period the unit was not able to achieve design furnace O2 levels of 2% at full load (138 MW
gross) due to high air heater leakage rates. Steam air heaters are installed hi each forced draft fan
discharge duct to control the average cold end temperature of the Ljungstrom air heater elements.
The combustion air from the steam air heaters is supplied at design temperatures above 100°F to
two horizontal shaft Ljungstrom regenerative type air heaters. The air heaters are equipped with
two water wash nozzles in the combustion air duct, one on the hot end and one on the cold end.
A water wash may be performed with the unit on line at a load of 60 MW. At this reduced load,
only one forced draft fan and 2 cyclones are required for operation. The combustion ah" is
supplied to a common windbox which houses three lO'-O" diameter water cooled cyclone burners
arranged one above two. Each cyclone burner is supplied by a volumetric coal feeder.

-------
                                     »rt« ouricr
            Figure 1
 Side View of the Reheat Furnace
 Public Service Electric and Gas
Mercer Generating Station Unit 2
               Figure 2
    Atlantic City Electric Company
E.L. England Generating Station Unit 1

-------
Table 1
Coal Analysis, Dry Basis
Analysis Item
% Moisture
% Carbon (Dry Basis)
% Hydrogen (Dry Basis)
% Nitrogen (Dry Basis)
% Oxygen (Dry Basis)
% Sulfur (Dry Basis)
% Ash (Dry Basis)
Btu/lb - Heating Value (Dry Basis)
Mercer 2
7.5
86.9
4.26
1.24
1.79
0.81
5.00
14,972
B.L. England 2
8.7
72.2
4.91
1.24
6.10
2.64
12.89
13,109
The combustion gases pass through a single furnace into the convective pass. In the convective
pass, the gases pass over the primary and secondary superheaters as well as a reheat superheater.
An economizer is located after the primary superheater.  The gases pass through the two
Ljungstrom ah- heaters and then the electrostatic precipitator (ESP) before exiting out the stack.

Temperature Profiles

Temperature profile data was obtained by Nalco Fuel Tech (NFT) at Mercer to provide input into
there Computational Fluid Dynamics Model (CFDM). Figure 3(a) plots the average temperature
measured at full load and low load at each port. A high velocity thermocouple, or suction
pyrometer, was used to measure furnace gas temperatures and provide a gas sample. Flue gas
temperatures at nominal full load ranged from 2000 F - 2250 F at the three ports with available
access, which were marginally downstream of the actual injector locations. At minimum load (81
MW), temperatures at these ports were reduced to 1200 F -1550 F. Similar data was taken by
RJM, a Nalco Fuel Tech licensee, at B.L. England Unit 1. As shown in Figure 3(b), a similar
range, albeit higher absolute values were encountered at four locations with available access. Full
load flue gas temperatures ranged from 1750  F  - 2500 F. Minimum load (60 MW) flue gas
temperatures were reduced to  1400 F - 2100  F.

Baseline NOx

Due to relatively small furnace dimensions  and high temperatures necessary to maintain proper
slag flow, baseline coal-fired NOX emissions at Mercer Unit 2 range from nominally 1175 ppmc
(1.6 Ib/MMBtu) at full load to 400  ppmc (0.55 Ib/MMBtu) at 25% maximum continuous rating
(MCR). A summary of the baseline NOX emissions over the entire load range is shown in Figure 4.

Atlantic Electric's coal-fired cyclone boilers at B.L. England station had average baseline NOX
emissions on the order of 800 ppmc (1.1 Ib/MMBtu). Based upon data from the plant Continuous
Emissions Monitoring System (CEMS) installed in November 1993, the variability in NOx  emissions
is quite high (Figure 5). Scatter about the mean emission level was found to be on the order of+/- 250
ppmc, representing a 40% uncertainty hi the baseline NOx value. Attempts to more quantitatively

-------
               ^ffi::pl::±::lf^^                          ^'dS^i
                                              Ms l,« «S-«"P ?* -¥^; 5-^ ^;V^!«
                                (a) Mercer Unit 2
                             (b) B.L. England Unit 1
                                    Figure 3
Average Furnace Exit Gas Temperatures Measured at Full Load and Low Load at Each Port

-------
               a^sb^y *
                                   Figure 4
                         Mercer 2 Baseline NOX Emissions
                                   Figure 5
B.L. England I Baseline NOX Emissions Based on CEMS Data from December 1-12, 1993

-------
define relationships between operating conditions and NOX were not successful. This inability to
correlate average NOX to load and economizer O2 is hypothesized to be a result of the cyclone
and unit specific conditions. During the long-term demonstration, baseline data were recorded
three times daily when the SNCR system was shut off for 5-30 minutes during each period.

SNCR Injection Systems

Temporary SNCR injection systems were used during both demonstrations. The system consisted
of a 21,000 gallon temporary storage tank, Nalco Fuel Tech pumping/metering control trailer,
injector distribution modules and air-atomized injectors. The pumping and metering trailer
included a tank recirculation pump, electric heater and chemical pumping and metering skids. The
trailer was also equipped with a control panel to properly meter the chemical flow rate to the unit
in the proper proportion with the dilution water. The injection trailer was equipped with a water
pump and a static mixer to mix and deliver the diluted reagent to the injectors. The injection skid
was capable of providing independent flow and concentration control to each level of injectors.
The chemical solution was transported to the boiler by two 1-1/2 inch stainless supply pipes.
Each pipe represented a zone in the control scheme of the demonstration trailer. At each level, a
distribution module received the water/chemical mixture and metered the flow to each lance.
Each lance had an indicator for both air and chemical flow.

At Mercer, three levels of injectors were initially specified based on available ports and modeling.
Four additional ports were added during testing. The initial ports were located at elevation 208'
(Level 8) on the front wall, elevation 195' (Level 7) on the front wall, and 186' (Level 6) on the
rear wall. Each wall had four lances. Level 8 was shown to help reduce NOX emissions, but only
with high ammonia slip rates. Upward-angled lances were added at elevation 172' (Level 5) in an
attempt to obtain reduction levels similar to Level 8, but with lower slip rates. Level 5 was
comprised of two lances on the front wall and one on each side wall.

At B.L. England, the injector distribution modules were located on the 6.5 (Level 1) and 8th
(Level 2) floor elevations, and controlled the atomizing air and chemical flow rate to the
individual injectors. A total of 17 injectors were installed on the unit through a combination of
new and existing observation ports. The Level 1 injectors were controlled by two injector
distribution panels, one controlling six injectors along the front wall and one controlling three rear
wall injectors and two side wall injectors. The Level 2 injectors were controlled by a single
distribution panel which fed six injectors along the front wall above the nose of the boiler. These
injection ports were the same as those used in the short-term test and were installed based upon
temperature measurements and the Nalco Fuel Tech CFD and CKM Model recommendations.
Figure 6 is a diagram of the NOXOUT® Process installed on Unit 1.

Control of the SNCR systems were performed by use of a steam flow signal and NOX
concentration feedback signal. The steam flow was characterized in segmented ranges to the
proper level of injection and NOX setpoint with attendant maximum and minimum reagent
pumping rates. These pumping ranges were designed to limit ammonia slips and were based upon
short term tests. A corrected NOX concentration from the plant CEMS was tied into the control

-------
   PLANT ATOMIZING AIR
     MIXED CHEMICAL
       TO LEVEL 2
         MIXED CHEMICAL
           TOLEVEL1
         NALCO FUEL TECH
           FC-103 NO.OUT
         TESTING TRAILER
         NO.OUTA  I
         RETURN
     LEVEL 2 WALL
     DISTRIBUTION
       MODULES
     LEVEL 1 WALL
     DISTRIBUTION
       MODULES
     LEVEL1WALL
     DISTRIBUTION
       MODULES
                   \
  SO% SOLUTION
|1  LQ
OLUTTON)
  uO
                   1
CHEMICAL
PUMPING
 EQUIP.
     (JO
                                                 LEVEL 2 WALL
                                                 INJECTORS (8)
                                  B.L. ENGLAND UN1T1
                                  138 MWg COAL-FIRED
                                    CYCLONE BOILER
                                          LEVEL 1 WALL
                                          INJECTORS (6)
                                                LEVEL 1
                                                 WALL
                                        INJECTORS
                                            (5)
 CONTAINMENT BASIN
                      NO.OUT A SUPPLY
               DILUTION WATER

               PLANT PROCESS & CONTROL SIGNALS
                     (BOILER LOAD ft TRIPS)
                                     Figure 6
                   NOXOUT Process Diagram at B.L. England Unit 1

-------
as feedback for reagent pump rate trim within the limits established by the load. The pump rates
and the injection levels used at the different steam flows were determined as testing proceeded, in
order to attain NOX setpoint while minimizing ammonia slip.
Test Conditions and Results

At Mercer, the SNCR system was tested for NOX reduction at specific load and fuel conditions.
Figure 7 shows a grid of coal-fired test conditions and the amount of NOX reduction achieved. As
indicated the range of coal-fired NOX reductions ranged from 32% to 38% with normalized
stoichiometric injection ratios (NSRs) of 0.8 - 1.2. The principle factor establishing the maximum
NSR was the maintenance of the ammonia slip at or below 5 ppmv. Deliberate attempts to reach
increased slips of 15 ppmv were tried over short time spans at some loads and conditions. Under
100% coal-fired operating conditions, the NSR was increased from 1.2 to 1.83 at 80 MW with
NOX reductions increasing from 32% to 46%. The chemical utilization between these two tests
was reduced from 27% at a NSR of 1.2 to 25% at 1.8, thus the principle limiting factor to
performance is the ammonia slip value which does not impair boiler operating reliability due to air
heater fouling or ash resale.

At B.L. England, the objective for the longer-term test was to demonstrate what reduced NOX
emission rate could be maintained under economic unit dispatch. Some of the specific concerns
that could not be addressed in the conceptual test that needed to be addressed in the longer-term
test were load following and area regulation requirements and the impact of various levels of
resulting ammonia slip on air heater fouling.

As previously mentioned, the longer-term test took place from December 1993 through March
1994. During December, only baseline testing was done with urea injection limited to setting
injection pressures and flows. Urea injection under automatic control was begun in early January,
and the unit was left on economic dispatch for the bulk of the test period. To accomplish the test
goal, and due to an increased frequency of air heater fouling with the unit operating under area
regulation, several injector configurations and NOX setpoints were evaluated. A summary of these
test conditions is presented in Table 2.

The stack NOX emission value that could be maintained was found to be dependent on several
parameters. These include: (1) the ability of the SNCR control system to react to changing unit
conditions and hold a given setpoint; (2) combustion conditions such as fuel-to-air ratios, flue gas
temperatures, CO and SO3; (3) secondary effects such as NH3 slip, ammonia/ash recycle, ambient
conditions and air heater fouling.

The ability to control to a NOX emission rate during the long term demonstration was limited to a
few controllable process variables. SNCR process variables including injector location, style and
nozzle tip orientation, dilution water flow rate and atomization pressure were optimized and fixed
for the test condition. The real time, controllable parameter remaining was reagent injection rate
to either or both injection levels. Reagent injection rates were determined by the controls based
on setpoint, NOx, steam flow (load), and O2. The test period was split into eight operating

-------
periods, characterized by injection configuration, NOX setpoint and boiler conditions with regard
to excess 62, as defined in Table 2. Other factors impacting the operation of the SNCR system
during the long- term demonstration are listed, chronologically, in Table 3.  It should be noted
that limitations imposed by the design of the temporary injection system restricted the ability of
the process to respond rapidly to changes in boiler conditions.  Some of these limitations can be
minimized in a full scale commercial installation.
Fuel/Load
100%Coal<5ppm
NHa Target




100% Coal <15ppm
NH3 Target




80 MW
NOxRed-32%
NH3 Slip - 7.8 ppm
NOxtoN2O-35.0%
NSR-1.2
Utilization - 27%
2Sep93
NOxRed-46%
NH3 Slip - 17.5 ppm
NOxtoN2O-34.4%
NSR-1.83
Utilization - 25%
2Sep93
135 MW
NOxRed-37%
NHs Slip - 2.3 ppm
NOxtoN2O-10.5%
NSR-0.82
Utilization - 45%
24Aug93
Not Tested





200 MW
NOxRed-38%
NH3Slip- 1.4 ppm
NOxtoN2O-7.0%
NSR-0.78
Utilization -44%
18Aug93
NOxRed-51%
NH3 Slip - 3.3 ppm
NOxtoN2O-5.6%
NSR-1.63
Utilization -31%
10Sep93
220 MW
NOxRed-34%
NH3 Slip -4.1 ppm
NOxtoN2O-7.0%
NSR-0.78
Utilization - 44%
18Aug93
Not Tested





321 MW
NOxRed-37%
NHs Slip - 4.0 ppm
NOxtoN2O-7.7%
NSR-1.17
Utilization -31%
17Aiig93
Not Tested





                                        Figure 7
   Mercer 2 NOX Reduction Under 100% Coal-Fired Test Conditions (Steady State Operation)
                                        Table 2
                   A Summary of Test Conditions at B.L. England Unit 1
Period
Date
Injection
Configuration
NOX Setpoint
(Ib/MMBtu)
Boiler Conditions
    1         Jan. 6 - Jan. 21
    2         Jan. 25   Feb. 8
    3        Feb. 9 - Feb. 16
    4        Feb. 17 - Mar. 2
    5        Mar. 2 - Mar. 9
    6        Mar. 9-Mar. 15

    7       Mar.  15-Mar. 26
    8       Mar. 26 - Mar. 29
2 Levels
 1 Level
 1 Level
 1 Level
 1 Level
 1 Level

2 Levels
2 Levels
      0.75
      0.82
      0.75
      0.68
      0.61
 0.61 60-100 MW
0.68 100-132 MW
      0.68
      0.82
 LowOz
 Low O2
 Low O2
Design Cb
Design C>2

Design Cb
Design O2

-------
                                               Table 3
	Chronologicalj.is^ofFactors Impacting B.L. England Unit SNCR Demonstration
 Date          Event
 1/6 -1/25      Setpoint 0.75 Ib/MMBtu, 2 Level Injection
 1/6 - 3/2       Large 62 swings resulting in very low boiler Oj
 1/7           Fluctuating air pressure to injectors
 1/7 - 2/4       Urea flow rate unstable
 1/10          Increase in air heater pressure drop
 1/10 -1/21    Low average Oj resulting from air heater plugging and loads in excess of 130 MW
 1/13          Reduce minimum SNCR operating load to 60 MW
 1/20 -1/25    Air heater sootblowers out of service
 1/21          Upper level minimum flow rate decreased. Lower level water flow rate increased
 1/25          Air heater water wash
 1/25          Setpoint raised to 0.82 Ib/MMBtu. Lower level injectors only
 1/27          Decreased water pressure to injectors
 2/5           Air heater water wash
 2/9           Setpoint changed to 0.75 Ib/MMBtu
 2/16          Setpoint changed to 0.68 Ib/MMBtu
 2/21          Air heater water wash
 2/28 - 3/2      Total flow rate to injectors 1.45 gpm
 3/2 - 3/9       Injector flow rate decreased to less than 1 gpm
 3/2           SNCR system controlling off of west econ. 02
 3/2           Setpoint changed to 0.61 Ib/MMBtu
 3/4           Maximum load limited to 128 MW
 3/4           Operating at design 02 conditions
 3/9           Increased maximum injection rate at steam flows above 635 Klb/hr
 3/9           Increased water injection pressure
 3/9           Setpoint changed to 0.68 Ib/MMBtu at steam flows above 745 Klb/hr
 3/4 - 3/15      Maximum load reduced from 128 MW to 110 MW due to air heater plugging
 3/15          Water pressure increased to injectors; 1 gpm/injector, 30 psig air
 3/15          Setpoint changed to 0.68 Ib/MMBtu. Second level in service
 3/16          "A" air heater water washed
 3/17          "B" air heater water washed
 3/19          Economizer Oj control disconnected
 3/21          Air pressure increased from 30-45 psig on level 2
 3/26          Setpoint changed to 0.82 Ib/MMBtu

-------
Though conditions throughout the test were not always representative of target operating
conditions, the SNCR system was able to demonstrate overall reductions hi NOX. Such
reductions were more significant at lower load, but, on average, indicate that a 30% reduction in
NOX should be achievable under dispatch conditions (Figure 8) while simultaneously maintaining
NH3 slip at a level which minimizes air heater fouling. From an average baseline of 1.1 Ib/MMBtu
during the test, the system could generally average NOX emissions around set points of 0.75-0.82
Ib/MMBtu. During these periods, the chemical usage corresponded to an average NSR of 0.7.
Air Heater Fouling

At Mercer, no problems were experiences during the SNCR demonstration with increased air
heater pressure differential except during one excursion of ammonia emissions during the
optimization phase of the program. During this condition the ammonia slip was approximately 60
ppmv, and exhibited corresponding increases in the air heater differential pressure. When soot
blowers were employed and the slip rate was returned to the 5 ppmv target level, the differential
pressure returned to normal. It should be noted, however, that the bulk of the tests were
performed under steady state operating conditions, and only involved load following during the
last two weeks of the demonstration. Although the results are indicative of limited air heater
fouling problems with a low sulfur coal, the operating experience under transient load following
conditions is limited.

Air heater fouling at B.L. England occurred four times over the course of the long-term
demonstration. A chronology of air heater washes during SNCR operations is shown in Figure 9.
Water washes are typically performed when the furnace pressure approaches 20 in.H2O at full
load, which corresponds to pressure drops of 9 to 10 in. H2O across the air heaters. The average
number of SNCR system operating hours before water washing was required was 274 hours as
compared to 1,038 hours during unit operation without chemical injection. The fouling of the air
heaters resulted from ammonium sulfate and bisulfate formation and possibly increased ash
cohesiveness due to the elevated NH3 levels on the ash. The air heater fouling was likely
compounded by low air temperatures exiting the steam coils during periods of low ambient air
temperature (<20°F) and periods when the steam coil air heater was out of operation.

Sootblowing did not appear to reduce the air heater fouling after an increase in the pressure drop
was noticed. Optimizing the frequency of sootblowing and steam pressure to the sootblowers
may have the potential to improve the time between air heater washes, but is not expected to
improve the ability to remove deposits in the intermediate air heater baskets and deep within the
cold-end air heater baskets. The increased ah- heater pressure drop also exacerbates the existing
problem with limited fan capacity at full load. Increases in fan horsepower and parasitic loss of
power will be a potential added cost and impact of SNCR for NOX control at B.L. England.

-------
      1000
   8
   £
   Q.
   in
   O
   in
   J2
   UJ
   5
                                                           Baseline NOx
                                                       -A-NOxwithSNCR
                                                       -A-NH3Slip
          80
90
100
   110
Load (% MCR)
120
130
140
                                   Figure 8
Summary of NOX Reduction and Ammonia Slip During B.L. England 1 SNCR Demonstration

-------
                                   WASH
                                                W WASH
 	1	'	l""l	I	.....f.............I	...I
1/6     1/15          1/312/6      2/16          3/1         3/16        3/29

                                   DATE
                                  Figure 9

                SNCR Operating Periods with Air Pre-Heater Washes

-------
SNCR Capital and Operating Costs

The costs involved with the procurement, operation, and maintenance of the SNCR system may
be divided into several categories. They are the capital costs incurred for the design and
installation of the system, the chemical costs required for system operation, costs associated with
the operation of the urea heater, injection pumps, and associated equipment, labor costs for
required maintenance, and costs due to boiler efficiency penalties. To determine the cost
associated with each category, the following assumptions and operating parameters were
used.The capital costs for a commercial system at Mercer (321 MW) and B.L. England (138
MW) have been estimated at $3,400,000 ($10.6 $/kW) and $2,070,000 ($15/kW) respectively.
Included is the equipment, engineering and design, installation and start-up services, one year
service contract, as well as the licensing fee.

The boiler heat rate penalties resulting from the vaporization of the aqueous solution decrease the
boiler efficiency by approximately 0.5%. At Mercer, this is based upon 24 injectors discharging 1
gpm/injector, resulting in a total liquid flow of 24 gpm. At B. L. England it was assumed that
multi-level injection would be used for the commercial installation, with each injector discharging
1 gpm/injector on the lower elevation and 0.6 gpm/injector on the upper elevation, resulting hi a
total liquid flow of 15 gpm.

A summary of the cost estimates for the installed SNCR system at each site is presented in Table
4. The analyses are based on the above assumptions, a plant capacity factor of 60%, and an
annual capital recovery rate of 14.67% was used.

Based on this analysis, the total SNCR annualized cost would be $1.4 million per year, or $937
per ton NOX removed. Of this total, 22% is capital investment, 59% is reagent cost, and the
remaining 19% represents maintenance, power requirements, and boiler efficiency penalties.

-------
                                        Table 4
                  Estimated Cost Analysis for Commercial SNCR Systems

Uncontrolled NO*, Ib/hr, (Ib/MMBtu)
Controlled NOx, Ib/hr, (Ib/MMBtu)
NOX Removed, tons/yr
Reagent Use, gal/hr
Reagent Cost(1), $/yr
Water Cost(2), $/yr
Capital Cost, $/kW
Capital Cost
Annualized Capital Investment
Maintenance, $/yr
Power Requirements^, $/yr
Boiler Efficiency Penalty, MMBtu/yr
Boiler Efficiency Penalty(4), $/yr
Capital and O&M Costs, $/yr
Total Cost, $/Ton NOX Removed
Mercer
Unit 2
5,008(1.6)
3,155 (1.0)
4,870
660
2,567,000
13,665
10.6
3,400,000
498,780
178,500
18,600
82,000
137,350
3,414,000
701
B.L. England
Unitl
1,794 (1.3)
1,242 (0.9)
1,451
206
801,225
9,248
15
2,070,000
303,669
85,000
8,000
57,000
95,500
1,360,000
937
(1) NOXOUT HP reagent cost of $0.74/gal.
(2) Based on water cost of $2/1000 gallons
(3) Based on energy cost of $0.03/kWh
(4) Fuel cost of $1.675/MMBtu
Summary And Discussion

At Mercer 2, NOX emissions were reduced by an average of 36% while maintaining the ammonia
slip at or below 5 ppmv, corrected to 7% O2. The NSR corresponding to these operating
conditions ranged from 0.8 - 1.2, with the average over the load range being 1.0.  Chemical
utilization was varied with load and injection temperature, but averaged 36%.  Additional tests
performed at higher ammonia slip levels indicated that increased NOX reductions were feasible.
Under coal-fired conditions, only two loads (80 MW and 200 MW) were investigated in this
regard with NOX reductions ranging from 46% to 51% at a NSR of 1.8 -  1.6 respectively.  The
corresponding ammonia slips were 17 ppmv and 3 ppmv. In general, the low sulfur coal fired at
Mercer Station did not present any operating limitations within the range of conditions tested with
respect to air heater fouling or fly ash resale impacts.

-------
The SNCR demonstration at B.L. England Station, on the other hand, exhibited an increased
sensitivity between coal sulfur level (e.g. flue gas SO3 concentration) and ammonia from the
SNCR process. The following is a summary of performance along with any associated boiler
impacts:

• the system was able to maintain an average reduction of 30% during the test while maintaining
NH3 slip at less than 10 ppmv. For a representative baseline 1.3 Ib/MMBtu, one could expect
to maintain an average emission rate of 0.9 Ib/MMBtu under dispatch conditions as one
approach to minimize ammonia slip and air heater impacts.

• the SNCR system was unable to control to tested NOX setpoints below 0.9 Ib/MMBtu at loads
above 120 MW with stack O2 levels in excess of 9.6% while simultaneously maintaining NH3
slip at less than 10 ppmv.
• load changes induce wide variations in NOx, and NOX changed more rapidly than the
temporary SNCR system could respond: These together limited how closely NOX could be
controlled. Nalco Fuel Tech and RJM both believe that improvements (reductions in response
tune) can be included in a commercial injection and control system.

• over control of NOX at loads below 100 MW will help compensate for the inability to control
NOX at high levels of excess air and load over a 24-hour period.

• air heater fouling from the deposition of ammonium bisulfate increases regardless of setpoint
or injector configuration; an acceptable, and controllable, level of ammonia slip appears to be
less than 10 ppmv but needs to be tested further. Ah- heater fouling during the test periods
was probably exacerbated by low ambient air temperatures and failure of the steam coil air
heater.

• flyash reinjection eliminates problems associated with elevated NH3 content of the fly ash

Based on the test results, none of the injection configurations are capable of controlling NOX
below a 0.90 Ib/MMBtu setpoint without impacting the boiler reliability. All configurations tested
increased the rate of air heater fouling and caused elevated concentrations of NH3 on the fly ash.

The system could not maintain NOX setpoints below 0.82 without sacrificing excess O2, load or
NH3 slip. Upward adjustments of excess O2 and/or conditions of high baseline NOX generally
resulted in periods in which the setpoint could not be maintained. Such instances of high NOX
were observed in seven of the eight periods, with each exhibiting controlled NOX levels
approaching 0.9 Ib/MMBtu.

In comparing these two demonstrations, it is important to note that although the SNCR is a
commercially viable technology and is being installed at both of these stations on a commercial
basis, the results obtained are site specific hi nature. As has been documented in the literature, the
NOX reduction performance potential of SNCR is a function of both SNCR process chemistry and
furnace parameters. The following factors are relevant:

-------
• chemical type,
• amount of reagent,
• injection temperature(s),
• residence time, velocity,
• initial NOX level,
• additive/trace gas species,
• injection/mixing system, and
• fuel effects.

These two demonstrations employed a similar test skid from Nalco Fuel Tech, using urea at
comparable injection rates, on boilers that exhibited similar initial NOX levels and temperature
windows. One significant difference is the coal fuel sulfur level used at the two sites. The sulfur
content and trace metals in the fuel will affect the formation of SO3, which reacts with unreacted
ammonia to form undesirable deposits at low temperatures. The amount of 863 can constrain the
concentration of unreacted ammonia, which can in turn constrain the amount of reagent and
achievable NOX reductions. Based on these two SNCR demonstrations, the achievable NOX
reduction with a high sulfur coal boiler operating is uncertain. The data from these two suggests
that unless SOs interactions with ammonia can be addressed beneficially, NOX reductions may be
limited below typical levels obtained at low sulfur application sites. It is thus important for
regulators to understand the site specificity of the SNCR technology performance when applied to
the vast range of coal-fired utility boiler designs and fuel properties.
Acknowledgments

The authors would like to thank Frank Gibbons, Al Wallace, and Alex Huhmann of Public Service
Electric & Gas, Debra Mellish and Mike Cunningham of Atlantic Electric, Giff Broderick of RJM,
and John O'Leary of Nalco Fuel Tech for their efforts during the actual test programs and their
contributions to the review provided for this paper, and the respective test reports.
References

Gibbons, F.X., et.al., "A Demonstration of Urea-Based SNCRNOX Control on a Utility
Pulverized-Coal, Wet-Bottom Boiler," EPRI Workshop - NOX Controls for Utility Boilers, May
11-13, 1994, Scottsdale, Arizona.

Hubbard, D. G., et.al., "SNCR Demonstration of NOX Control for a Cyclone-Fired Boiler Burning
High Sulfur Coal," AFRC/JFRC Pacific Rim International Conference, October 16-20, 1994,
Maui, Hawaii.

EPRI Report TR-105068, "Long Term SNCR Demonstration at B.L. England Station Unit 1."

EPRI Report TR-105071, "Demonstration of a Urea-Based SNCR Technology on a Pulverized,
Coal-Fired, Wet-Bottom Boiler."

-------
DESIGN OPTIMIZATION OF SNCR DENO. INJECTION LANCES
X
Dale G. Jones, Noell, Inc., Long Beach, CA 90806
Jochen Steinberger, Noell, Inc., Long Beach, CA 90806
Terry Hunt, Public Service Co. of Colorado, Denver, CO 80223
Christopher Barton, Public Service of New Hampshire, Bow, NH 03304
Lawrence J. Muzio, Fossil Energy Research Corp., Laguna Hills, CA 92653
Jeff Stallings, Electric Power Research Institute, Palo Alto, CA 94304
Ron Sherrick, Diamond Power Specialty Co., Lancaster, OH 43130
Abstract

Because of the temperature sensitivity of SNCR processes, load following control to minimize
NH3 slip is a critical issue. Advanced retractable injection lances (ARIL) provide improved
load following control for SNCR DeNOx. Using ARIL technology for SNCR DeNOx, 30% to
50% DeNOx can be achieved while limiting NH3 slip to 10 ppm.

One ARIL lance is located at each boiler sidewall, each supplied with compressed air and
liquid chemical. A row of nozzles on the sidewall of the lance form a "jet curtain" to provide
flue gas cross-sectional coverage. Load following is provided by automatically rotating the
ARIL lances in response to a furnace exit gas temperature (FEGT) or a boiler load signal.
Design optimization is required for most boilers, generally including field testing and cold
flow modeling. Field tests using "stub" lances have verified low NH3 slip at 40% DeNOx or
more. Flow model tests have shown how to optimize the location of the ARIL lances so that
improved flue gas mixing can be achieved without increasing air compressor capacity, which
is typically less than 0.2% of plant output.

ARIL SNCR DeNOx was demonstrated in 1991 at Pacific Gas & Electric Co. Morro Bay Unit
3 (330 MW). Two additional ARIL systems have been installed on coal-fired boilers, one at
Public Service Co. of Colorado, Arapahoe Unit 4, (100 MW, balanced draft), and the other at
Public Service of New Hampshire, Merrimack Unit 1, (120 MW, cyclone-fired). Although
the total installed cost for smaller 100 MW coal-fired plants is about $15/kW, the total
installed cost of ARIL technology for SNCR DeNOx on larger plants (due to economies of
scale) is expected to be in the range of $7/kW to $10/kW.

-------
Introduction

Selective Non-Catalytic NOX Reduction (SNCR) involves the selective reaction of the
decomposition production of ammonia, or urea, with NOX in the presence of oxygen. The
process is extremely temperature sensitive, operating over a narrow temperature window of
nominally 1600 to 2200°F. The basic SNCR DeNOx reaction is

NO + NH = N + HO
2 2
The amine radical (NHj) is formed by direct furnace injection of liquid ammonium hydroxide,
or urea, where the NH3 component loses a hydrogen atom, and any HNCO component reacts
to form amine radicals and other reactants.

Unreacted amine radicals form NH3 slip at the boiler economizer outlet which can react with
SO3 or HC1, etc., to form various salt compounds. Depending on where the salts form
downstream, equipment can become fouled, or plume visibility impacted due to atmospheric
condensation at cooler temperatures. The only way to avoid undesirable ammonium salt
byproducts is to limit the NH3 slip, generally to less than about 10 ppm at the economizer
outlet.

The NH3 slip constraint imposes a limit on the amount of NOX removal that can be achieved
using the SNCR DeNOx process. Because the SNCR DeNOx process is temperature sensitive,
load following to maximize DeNOx while minimizing NH3 slip is a critical issue.

Advanced retractable injection lances (ARIL) provide improved load following control for
SNCR DeNOx. There is one ARIL lance at each boiler sidewall, each supplied with
compressed air and liquid chemical. A row of nozzles on one side of the lance form a "jet
curtain" to provide the flue gas cross-sectional coverage. Load following is provided by
automatically rotating the ARIL lances in response to a furnace exit gas temperature (FEGT)
or a boiler load signal.

Using ARIL technology for SNCR NOX reductions of 30% to 50% can be achieved with a 10
ppm NH3 slip constraint. Lower levels of NH3 slip can generally be realized using
ammonium hydroxide instead of urea, and urea is also more expensive. However, both
chemicals have been demonstrated and can be used for SNCR DeNOx with the ARIL lance
technology.
Design Approach

Most boilers are sufficiently different that there is no standard formula or system
configuration that can be used. Instead, design optimization using a combination of field
testing and boiler cold flow modeling provides assurance of improved SNCR system
performance and minimizes risk.

-------
Field testing includes both high velocity thermocouple (HVT) temperature measurements in
the upper furnace zone(s) of interest and preliminary SNCR field testing using short "stub"
lances. The "stub" lances protrude approximately eight feet into the flue gas and treat a
portion of the flue gas. This allows the local SNCR process performance (DeNOx and NH3
slip) to be measured downstream at the economizer outlet. It has been found that lateral
mixing through the boiler convection tube banks is somewhat limited, and a good
approximation of full-scale performance results can be obtained using this low cost field test
approach.

Boiler cold flow modeling is another important component of the design approach. The cold
flow modeling quantifies the degree of mixing that can be achieved at different boiler
sidewall locations for the ARIL lances. The degree of mixing can be improved by about 50%
if the location is optimized, compared with just locating the ARIL lances at a reasonable
position. Since secondary flows and turbulence are not accurately known a priori, the cold
flow model testing can lead to significant reductions in the electrical consumption for air
compressors. In comparison, simply "plugging in" the ARIL lances and then performing the
required optimization testing at full scale provides a significantly increased risk of having to
relocate the ARIL lance position(s).
Field Testing Results

An SNCR system utilizing wall nozzles was installed as part of the DOE CCT HI integrated
emission control system at PSCC Arapahoe Unit 4 (100 MW coal-fired). As a result of the
low-NOx burner retrofit, the furnace exit gas temperature dropped by almost 200°F. This
drop in temperature compromised the low load performance of the SNCR system with the
existing wall nozzles. To improve SNCR performance at low loads, an ARIL lance system is
being installed using two existing furnace side wall ports. The SNCR injection locations for
this boiler are shown in Figure 1. To support the design of the ARIL system, a series of tests
were conducted using a "stub" lance on one side of the furnace. An eight foot "stub" lance
was manually installed into an observation door near the preferred location for the ARIL
lances. The effect of "stub" lance rotation angle at 60 MW boiler load is shown in Figure 2,
where it is seen that the DeNOx can be varied between 12% and 43% at a fixed NSR =1.0
simply by rotating the ARIL lance. A better indication of ARIL lance performance is
provided in Figure 3, where the wall injectors are compared with the "stub" lance for DeNOx
and NH3 slip performance at 60 MW load. At a 10 ppm NH3 slip limit, it is seen that when
using wall injectors, the DeNOx is limited to about 13%, but with the "stub" lance, the NH3
slip was only about 7 ppm with DeNOx over 50%. The ability to provide NH3 slip control by
automated rotation of the ARIL lance provides an important advantage for load following
operation.

A similar "stub" lance test was conducted to support the full-scale ARIL design for PSNH
Merrimack Unit 1 (120 MW coal-fired cyclone). An eight foot "stub" lance was installed into
the upper sootblower port at the secondary superheater (SSH) tube bank (see Figure 4).
Table 1 shows the NOX removal and NH3 slip performance results obtained at an ammonium
hydroxide NSR value of nominally 1.5:

-------
Table 1

"Stub" Lance Performance at PSNH Merrimack Unit 1
(Load: 120 MW, NO^: 900 ppm, NH^TNO,: 1.5)

"Stub" Lance Angle DeNOx NH3 Slip

90 degrees (straight down) 33% 42 ppm (3.1% of NH3 injected)

67 degrees (into upper furnace) 28% 25 ppm (1.9% of NH3 injected)

45 degrees (more into U.F.) 18% 5 ppm (0.4% of NH3 injected)

0 degrees (toward front wall) 15% not detected
Figure 4 also shows the gradient in NOX removal measured at the economizer exit. As can be
seen, some of the flue gas passed over the top of the stub lance with little NOX removal near
the outer wall at the economizer exit. This clearly shows the need to optimize the SNCR
mixing processes in the temperature window.

During the field test program at Merrimack Unit 1, the vertical temperature gradient at full
load from the nose of the arch to the furnace roof (a distance of 22.7 ft.) was estimated to be
about 900°F. This unexpected result, combined with lateral mixing and DeNOx gradient
effects measured at the economizer exit, result in the expectation that the full-scale ARIL
lances will produce about 50% DeNOx at NSR =1.5. The guaranteed NOX removal at full
load for Merrimack Unit 1 is 31% at NSR = 0.80.

The flue gas temperature for the "stub" lance tests at both sites was approximately 2100°F to
2300°F. The SNCR process seems to provide improved control of NH3 slip and reasonable
NSR requirements for a given amount of NH3 removal when the injection location is at
relatively high flue gas temperatures. This indicates that given droplet evaporation times on
the order of 50 to 100 msec, the main consideration is to be sure that the SNCR chemical is
injected and mixed upstream of the preferred "temperature window", typically 1700°F to
1850°F.
Cold Flow Modeling Tests

The ARIL lance ports at PSCC Arapahoe Unit 4 were existing ports, and there was not choice
in the location, so a cold flow modeling test was not performed.

For the PSNH Merrimack Unit 1 boiler, detailed cold flow modeling tests were conducted,
with six (6) possible ARIL lance locations, as shown in Figure 5. It was initially thought that
Location 2, near the furnace roof, would provide the best results. Figure 6 shows tracer gas

-------
contours for the ARIL lances located at Position 2 (see Figure 5). The tracer gas contours
can be interpreted as local N/NO ratios assuming an overall N/NO ratio of unity. As can be
seen from the cold flow model tracer gas concentration profiles shown in Figure 6, the
mixing achieved at Location 2 averages only about 47% (i.e., the quantity M shown in
Figures 6 and 7 is a statistically determined extent of mixing based on the tracer gas tests).
With the specified air compressors, the bottom half of the flue gas duct is basically not
covered with injection at Location 2. Further flow model testing was done to identify an
optimized location for the ARIL lances (i.e., Location 6). As shown in Figure 7, the mixing
at 90 degrees (straight down) exceed 70%, an improvement of more than 50% in the mixing
for the same air compressor capacity. This improvement makes the difference between
meeting and not meeting SNCR performance guarantees and could not have been predicted
without doing the flow model testing.
Installation Sequence

The ARIL lances are fabricated and assembled for testing in the shop. The assembly includes
a local PLC for automatic control of lance rotation angle as a function of boiler load, furnace
exit gas temperature (FEGT), or other preferred boiler operating parameter. Each ARIL lance
includes a local operating panel, where the plant equipment operator can install or retract the
lance, and/or rotate it to any desired position within the operating envelope. The ARIL lances
are automatically retracted after a flush cycle in the event of (a) boiler load below a minimum
set point, (b) loss of compressed air pressure, or (c) lance tip metal temperature are above a
preset limit.

Advanced lance metallurgy is required for long term operation in hot flue gas. The selected
alloy is capable of short term (i.e., eight hours) operation at 2100°F. At 1400°F, the selected
alloy is about 400 times stronger than 316 stainless steel for resistance to thermal creep
effects, resulting in a lifetime expected to exceed three (3) years. Fortunately, the compressed
air cooling effect is greatest at the boiler sidewall location, where the bending moment is also
the greatest. Lance deformation due to thermal creep effects has been estimated to be less
than 1.5 inches/year.

The ARIL lances were tested with compressed air and water to confirm proper operation of
the system prior to shipping to the site. A videotape of these tests, which includes some
details of the cold flow model testing can be obtained from the authors.

The completed ARIL lance system is field-erected in the same way that new sootblowers
would be installed. The required connections for the compressed air and liquid piping are
made, and the electrical and control wiring connections are made. Startup testing includes
actuating the ARIL lance system over its operating envelope of translation and rotation
functions.

-------
Status of Startup and Testing

At the time of the deadline for this paper, the PSCC Arapahoe Unit 4 ARIL lance system was
being started up, but no SNCR performance data had yet been obtained. It is believed that
some data will be provided at the conference.

The startup of the PSNH Merrimack Unit 1 ARIL lance system is scheduled to begin in the
middle of May, 1995, with SNCR performance data to be available about the first part of
June, 1995. Please contact the authors for further information.
Capital Costs

The ARIL lance retrofit for PSCC Arapahoe Unit 4 (100 MW) was budgeted to be somewhat
less than $4.00/kW, including field installation by PSCC, including control system
modifications and startup testing, but not including about five weeks of performance testing.
The existing air compressor system and liquid chemical injection system was used at
Arapahoe Unit 4, and these items are not included in the price. The project was started in
July, 1994, after the "stub" lance testing was completed.

The ARIL lance retrofit installation at PSNH Merrimack Unit 1 (120 MW) was budgeted to
be somewhat less than $15/kW, including all field testing, flow modeling, engineering design
and installation labor, equipment building, chemical storage tanks and forwarding pumps, 3 x
50% air compressors, 2 x 100% injection pumps, connection piping and wiring, ARIL lances,
centralized PLC control system and software, plus startup and initial system testing. The
project was started in September, 1994, and the initial field testing of the "stub" lances was
completed in October, 1994.

Due to the economy of scale that can be realized when ARIL lances are adapted to larger
units, a more realistic capital cost in the range from about $7/kW to about $10/kW could be
expected for boilers in the size range from about 250 MW to about 750 MW.

-------
SOUTH
                GAS
                FLOW
NORTH
             Location for
             ARIL Lances
                                             SECONDARY
                                            SUPERHEATER
                                                                Full Load
                                                               Wall Injectors
                                Figure 1
                SNCR Injection Locations at PSCC Arapahoe 4

-------
co
1
rx
o
      10 -
                      67.5       56        45        33        22
                        Angle of Injection (degrees up from vertical)
18
                                          Figure 2
       Effect of Injection Angle for Proof-of-Concept SNCR Lance Tests at PSCC Arapahoe 4
                                  (Load: 60 MW, NSR - 1.0)
                    110 MWe, Lower Level
                    100 MWe, Lower Level
                    80 MWe, Lower Level
                    60 MWe, Lower Level
                    60 MWe, Lance
                                   20         30          40
                                        NO Removal (%)
                                          Figure 3
                   Proof-of-Concept SNCR Lance Results at PSCC Arapahoe 4

-------
                              Untreated Flue Gas
Injection Angle "StubTance
                         Passes Over Top of Test Lance
                                    40 :
                            Percent 30
                            DeNOx 20 !
                                    10
                    Figure 4
Proof-of-Concept Lance Tests at PSNH Merrimack Unit 1
   (Load: 120 MW, NOri = 900 ppm, NH^O, = 1.5)

-------
          Injection Lance
            Locations
45
        >'
       90
                                                                   Economizer
                                                                      Exit
                                   Figure 5
           Injection and Measurement Locations, PSHN Merrimack Unit 1
                            (Cold Flow Model Tests)

-------
   90 Degrees




   M = 4456
              NORTH
        a.
  67.5  Degrees




  M =  49*
             NORTH
        b.
                       20.00-
                       15.00-
                       0.00
                         0.00      5.00      10.00      15.00     20.00     25.00
                       20.00-
                       15.00-
                       10.0O-
                                                        SOUTH
                                                                              SOUTH
                                          10.00      15.00     20.00     25.00
   45  Degrees




   M  =  47*



            NORTH










       C.
                       15.00-
10.00-
                        0.00-
                                     Figure 6
                                                       SOUTH
PSNH Merrimack Unit  1 Cold Flow Tracer Gas Contours (Injection Location 2)

-------
90 Degrees   NORTH

M =  7156
            a.
                        20.00-
                        15.00-
10.00-
                         5.00-
                         0.00
                                                       SOUTH
                                                            2ooo     25.00
67.5  Degrees  NQRTH

M  =  635?



            b.
                        20.00-
                                                       SOUTH
                                   5.00      10.00      15.00     20.00     25.00
 112.5 Degrees NORTH

 M = 4756
             C.
                         15.00-
10.00-
                         5.00-
                         O.OO-
                                                       SOUTH
                           0.00       5.00      10.00     15.00      20.00     25.00
                                           Figure 7
      PSNH Menimack Unit 1 Cold Flow Tracer Gas Contours  (Injection Location 6)

-------
INTEGRATED NOX CONTROL AT NEW ENGLAND POWER,
SALEM HARBOR STATION
M.B. Frish, S.A. Johnson, and J.P. Comer
PSI Environmental Instruments Corp.
20 New England Business Center
Andover, MA 01810

and

R.F. Afonso and A. Sload
New England Power Service Co.
25 Research Drive
Westborough, MA 01582
Abstract

Selective non-catalytic reduction (SNCR) is a viable technology for reducing NOX emissions from
coal-fired boilers, especially those older units where large capital expenditures for alternative
technologies may not be justified. However, NOX reduction efficiency of the SNCR process is
maximized when the proper amount of reagent is injected at the proper temperature and dispersed
rapidly enough to avoid ammonia slip.

Early NEP experience at Salem Harbor station indicated that NOX reductions of 60% were
achievable with SNCR. However, less NOX reductions were tolerated to avoid NH3 slip and
subsequent flyash contamination and visible stack plume resulting from excess ammonia.

Preliminary tests by PSI Environmental showed that ammonia slip could be monitored in real time
using their patented SpectraScan™-NH3 instrument, and that furnace exit temperature could be
continuously monitored and controlled using GasTemp™ another PSI Environmental product.
Based on this information, detailed tests were planned to show integrated control over the SNCR
process. A goal of the project was to achieve lower NOX with less reagent! This paper describes
the status of the project.

Background of Project

In 1992, New England Power Company (NEP), and the Massachusetts Department of
Environmental Protection (DEP) reached an agreement that provided the basis for an innovative
demonstration project targeted to reduce nitrogen oxides (NOX) emissions from Units #1, #2,
and #3 at Salem Harbor Station. The agreement called for the application and demonstration of
Selective Non-Catalytic Reduction (SNCR). The agreement also included the potential

-------
application of a second technology, low-NOx burners, based on the success of the SNCR
demonstration.

SNCR is a technology in which a reagent such as urea or ammonia is injected in the upper part of
the boiler. The reagent reacts with NOX created by fuel combustion, converting it to harmless
nitrogen and water.

The SNCR demonstration involved two major efforts: 1) installing reagent (urea) injection ports
in the upper part of the boiler along with associated systems to control and monitor reagent flow;
and 2) optimizing the performance of the SNCR equipment which included a monitoring and
testing program with comprehensive instrumentation throughout the boiler. The boiler of Unit #2
was selected for initial testing. The SNCR supplier and an independent test contractor combined
to fully explore system performance for a variety of test conditions.

Preliminary Test Results

Spectrum Diagnostix, Inc. (SDx - formerly PSI Environmental Instruments Corp.) tested a pre-
commercial version of its SpectraScan™ ammonia monitor at Salem Harbor Unit #2 in 1994. The
purpose of the test was to prove the accuracy and response time of the instrument while gaining
experience necessary to improve the commercial design. Figure 1 shows ammonia content of the
flue gas measured by SpectraScan™ as compared to the value calculated from wet chemical
analysis of a flue gas grab sample taken at the economizer outlet. In the latter case, flue gas
ammonia was absorbed in a known volume of dilute sulfuric acid and analyzed by ion-specific
electrode. Several grab samples were taken at each sampling time to increase the accuracy of the
wet chemical method.
SNCR System
Shutdown
10-

0
SpectraScan
Calibration Check
I
30
60 90
Time (min)
Figure 1
Ammonia Slip Measurements on SNCR System (High Sulfur Coal, Utility Boiler)

-------
It can be seen that SpectraScan™ was both accurate and responsive. Over a range of 6 to 34 ppm
of ammonia, SpectraScan™ was consistently within a ppm of the reference method value. More
importantly, the instrument revealed the time duration of significant ammonia spikes that could
affect operation of downstream equipment.

Ammonia slip is not always so variable; the data in Figure 1 were taken during initial shakedown
of the SNCR equipment. Figure 2 shows a more typical situation at another coal-fired boiler.
Here ammonia emissions are much more constant at 2 to 4 ppm until the time when an additional
set of urea injectors was activated for a few minutes at 12:00 noon. SpectraScan™ recorded a
spike of ammonia in the flue gas that reached about 65 ppm and then subsided when the urea was
turned off. Instrument response to this process upset was less than one minute. An unintentional
ammonia spike occurred later the same day as operators adjusted the system to operate at one-
third load.
100
2 80
co
CO
E 60
Q.
H 40
o
E
20
•Steam Flow-
FEGT
NH3-
I
350
280 ^
^
ii
210 *
1
140 "-
ca
cu
70 co
10:00
12:00
14:00
Time (h)
16:00
18:00
3000

2500

2000 \T
o

1500 S
a.

1000

G-9369
Figure 2
Ammonia Monitor Response to Spikes in Urea Flow
The initial tests at Salem Harbor provided insight into SNCR operation and revealed opportunities
for more precise process control. Table 1 shows steady-state ammonia slip measurements
averaged over several hours of operation at constant boiler load. First, note the ammonia slip can
be quite variable even though boiler load may be constant. The ammonia concentrations are
shown to within ±1 standard deviation. Comparing February 17th against February 23rd,
ammonia slip was quite different even when boiler loads were the same. A slight difference in flue
gas temperature leaving the air preheater hints that differences in furnace temperature may be
responsible for an increase in ammonia slip. As expected, increasing urea flow from a normalized
stoichiometric ratio (NSR) of 1.3 to 2.0 increased ammonia slip to unacceptable levels. Also,
ammonia slip could be minimized at low load as well as high load.

-------
Table 1
Ammonia Slip Measurements at Salem Harbor #2 - Steady State

Load, MW
Flue Gas O2, %
Air Heater Outlet
Ammonia at Econ
North
South
T, °F North
South
. Outlet, ppm
Normalized Stoichiometric
Ratio (estimated based on
expected NOX)
17 Feb 1994
83
3.8
3.1
265
278
2.1 ±1.4
1.3
18 Feb 1994
84
3.8
3.4
263
284
37.2 ±17.1
2.0
22 Feb 1994
38
7.7
7.5
255
255
10.3 ±5.0
1.3
23 Feb 1994
83
3.6
3.2
265
262
13.3 ±3.3
1.3
However, ammonia slip becomes very difficult to control during load swings. Table 2 shows a
steady increase in ammonia emissions as the Salem Harbor unit was ramped down to 38 MW
(about 45% of maximum rated capacity). During this test, urea flow was held constant during the
load transient and then adjusted to the proper NSR after low load was stabilized. The high
ammonia values underscore the need to match urea flow to NOX emissions as well as shift injector
location to follow temperature changes to prevent ammonia slip during load swings. Note also
that ammonia concentrations decreased after the tubular air heater, indicating that ammonium
bisulfate may have deposited during this transient. Higher NH3 at the air heater outlet at the end
of the load swing could indicate some offgassing from the deposits even after the ammonia slip
had stabilized.
Table 2
Ammonia Slip Measurements at Salem Harbor #2 - During a Load Reduction

Load, MW
Urea Flow, GPH
Injector Location
Air Heater Outlet T, °F North
South
NH3, PPM, @ Economizer Outlet
NH3, PPM, @ Air Heater Outlet
10:20 a.m.
66
202
Upper
260
262
66
25
11:30 a.m.
45
200
Upper
N/A
N/A
104
47
1:50 p.m.
38
213
Upper
255
255
47
91
2:55 p.m.
38
55
Lower
255
255
10
19

-------
Transient ammonia slip can be a real problem when SNCR is applied to cycling boilers. Figure 3
shows a 3-hour period of ammonia measurements at another coal-fired boiler. During this time
period, load ramped from 40 to 70% in the first hour, dropped back down to 40% in the second
hour, and ramped back up to 65% in the third hour. Ammonia slip held steady between 2 and
10 ppm, except for two spikes of about 40 ppm ammonia when load came up to 50%. This load
corresponds to point when the SNCR system sifts from a lower to upper level of injectors. Even
though ammonia quickly stabilized, spikes such as illustrated in Figure 3, if frequent enough,
could lead to equipment plugging or boiler operation problems.

E
Q.
a.
Q.
co
1C

100
80

20
0
Load
MH
Nrlg
-
-
___^'— **
- — x
M
J \
,
JI3:04
50% Load
;
-
-
^--•^""^•^^ s~-^. -
"** — — -s__ x —
^\_s '"""" A
J\
|V_^_;
,',,,,"
100
80

20
0

. — .
o^
CB
O

13:44 14:24 J15:04 15:44 Current
Time (h) '
cno/ ^ 50% Load
50% Load c-9241
Figure 3
Ammonia Slip vs Load for a Cycling Boiler

Preliminary measurements with a continuous optical temperature monitor at Salem Harbor
indicated that changes in furnace temperatures can also limit the effectiveness of SNCR systems.
Figure 4 shows several hours of full load furnace exit gas temperature (FEGT) data at Salem
Harbor. The coal being burned on this particular day produced fluffy ash deposits on boiler
waterwalls above the burner zone. These deposits impeded heat transfer resulting in higher
FEGT. When sootblowers were used to remove the ash deposits, FEGT decreased immediately
by 50 to 150°F; the magnitude of the temperature change depended on how many sootblowers
were used and the amount of ash removed at different locations in the boiler.

Two important conclusions relative to SNCR operation can be derived from these temperature
data:

1. Reagent injection temperature is far from constant at constant load

2. sootblowers (or burner tilts, burner tuning, etc.) can be used to hold injection temperature
more nearly constant.

-------
2700
2600
2500
Q.
2400
2300
_L
J_
16.5 17.5 18.5 19.5 20.5
Time (24 h)
+ Blow Soot
21.5 22.5

B-6791
Figure 4
FEGT Data from Salem Harbor
SNCR Optimization
Current SNCR systems use a combination of open loop look-up table type and closed loop feed-
back control systems to choose the location and amount of reagent to inject in order to control
NOX emissions. These system typically monitor stack NOX and adjust total reagent flow to
maintain compliance; this is the closed loop portion. The injection location is chosen from several
possibilities using a look-up table which correlates unit load and pulverizers in service with the
location of the optimum temperature for reagent injection. This look-up table can be significantly
in error depending on fuel properties and furnace cleanliness among other things. Also, because
there is no direct indication of reagent utilization, significant amounts of ammonia slip (a result of
incomplete reagent utilization) can occur without the operator's knowledge. Ammonia slip can
cause operational problems such as air heater pluggage as well as economic penalties such as
wasted reagent and unsalable flyash. The current control system could be improved by the
addition of a direct injection location gas temperature reading and a continuous, real-time,
ammonia slip measurement.

In late 1994, NEP and SDx initiated a test program to demonstrate the feasibility of this new
SNCR control philosophy. The SDx SpectraTemp™ optical temperature monitor was chosen to
determine temperature in the injection zone. A pre-commercial SpectraScan® ammonia monitor
was installed at the economizer outlet. Data from both these instruments can be analyzed along
with CEM and control room data to show possible savings brought about by tighter process
control.
The tests are being conducted at Unit #2 at Salem Harbor Station. A sectional sideview sketch of
the unit is shown on Figure 5. Unit #2 is an 85-MW, single wall-fired boiler commissioned in
1952. It is equipped with B&W circular burners and a urea injection system for NO control.

-------
-REHEAT INLET
HEADER
INTERMEDIATE
HEADER
Figure 5
Salem Harbor #2

The test plan includes three test series:

1. baseline measurements
2. parametric variations
3. process optimization

During baseline testing, the SpectraTemp™ and SpectraScan® instruments were operated for
several weeks to show that the instruments are accurate and reliable, ammonia measurements
compared favorably against wet chemistry analysis as they had last winter. During this time,
boiler and process operating data were collected to provide as basis for evaluating process
improvements.

Parametric tests are expected to take about one week to complete. Both boiler and SNCR
process variables will be changed systematically to measure the effects on furnace temperatures
and ammonia slip. A test matrix is shown in Table 3. Control room data taken at the beginning
and end of each test period include: load, steam flow and temperature, flue gas 02 concentration,
pulverizer parameters (amps, airflow, inlet/outlet temperatures), attemperation spray flow or valve
position, urea flow rates to each injector, and CEM data (NOX, O2, CO, opacity). For each test,
NSR will be varied in increments of 0.25 and held for about one hour while data are being
collected.

-------
Table 3
Parametric Test Matrix
Test No.
1
2
3
4
5
6
7
8
9
10
11
Load %
>90
>90
>90
>90
>90
<75
<75
<75
<75
<50
<50
Top Row
of Burners
Yes
Yes
Yes
Yes
No
Yes
Yes
No
No
Yes
No
Injectors
Upper
Upper & Middle
Upper
Upper & Middle
Upper & Middle
Lower
Lower & Middle
Lower & Middle
Lower
Lower
Lower
NSR
Min.
1.0
1.0
1.0
1.0
1.0
1.0
1.0
1.0
1.0
1.0
1.0
Max.
1.75
1.75
1.75
1.75
1.75
1.75
1.75
1.75
1.75
1.75
1.75
Furnace
Conditions
Dirty
Dirty
Clean
Clean
Clean
As found
As found
As found
As found
As found
As found
For the process optimization tests, the instruments will be used to maintain desired injection
temperatures and ammonia slip over the normal range of unit load. First, the sootblower schedule
will be modified if necessary to better maintain the injection temperatures shown to minimize NOX
and ammonia slips. Then, if NH3 is below 5 ppm, urea flow will be increased to achieve even
lower NOX. If NH3 is above 5 ppm, urea flow will be decreased to reduce NH3. At reduced
loads or when mills are out of service, injector locations will be changed according to changes in
measured temperatures. The results of these tests can be used by NEP, SDx, and the SNCR
system supplier to engineer an improved control subsystem.

Project Status

As of this writing, the baseline tests are in progress. The SpectraTemp™ instrument and the
electronics console of the ammonia monitor have operated unattended and trouble-free for three
months. The NH3 measurement cell and gas extraction probe have experienced periodic
problems, but have been operating successfully for the last three weeks.

The sampling cell operates at about 0.3 atm pressure to increase accuracy of the measurement,
and above 500°F to assure that all ammonia is in the gas phase. This high temperature, low
pressure environment has been a challenge to all seals. O-ring seals have been upgraded to
Kalrez® to assure temperature resistance. Also, all constricted flow passages (filter, pressure
reducer, etc.) have been relocated inside the probe where they are heated internally by rod heaters
and externally by the hot flue gas. Cold spots that could cause consideration of ammonium
bisulfate aerosol that plug the sample line at these flow restrictions are thus avoided.

-------
Figure 6 shows an example of continuous ammonia measurement made at Salem Harbor #2
during a 24-hour period on 9 April 1995. In this plot, the continuous measurements were
converted to rolling 2-minute averages by the data reduction program. Unit #2 normally cycles
between full load and control load depending on the demand for electricity. It is interesting to
note that the times when ammonia slip exceeds 10 ppm corresponds to when the boiler is not at
full load. Work to achieve more quantitative correlation of ammonia with boiler operation is now
in progress.
60
-o
CD
50
40
a

-------
If 20% more fly ash is sold, the difference on the NEP balance sheet is about $120K/yr. NEP
has a corporate goal to recycle 100% of the fly ash generated by its two coal-fired stations by the
year 2000.

Another consequence of ammonia slip at Salem Harbor is increased risk of ammonium bisulfate
deposits on the tubular air heater. Such deposits could reduce heat transfer effectiveness in the air
heater and eventually lead to pluggage and downtime to water wash the equipment. If only one
day of unscheduled downtime is avoided, about $45K of replacement power could be avoided.

Total savings could be in the range of $200K/yr at Salem Harbor #2 alone. NEP has also installed
SNCR along with low-NOx burners on units #1 and #3 at the station. Better SNCR performance
could allow these units the flexibility to tune their burners for higher NOX and lower unburned
carbon in order to maximize flyash sales. Total savings for all three boilers at Salem Harbor
(325 MW) could be well over $500K/yr. The results of this project will help determine the
viability of enhanced SNCR as a compliance strategy at this station.
-------
COMMERCIAL APPLICATION OF UREA SNCR FOR NOx RACT COMPLIANCE

ON A 112 MWe PULVERIZED COAL BOILER

James E. Staudt, Ph.D.
Robert P. Casill, P.E.
Research-Cottrell

Thomas S. Tsai
Leonard J. Ariagno
Eastern Utilities
Abstract

Montaup Electric Company, a subsidiary of Eastern Utilities Associates, operates a
tangentially-fired pulverized coal boiler at its Somerset, MA generating station. NOx
emissions from the 112 MWe Montaup boiler #8 must be reduced from their
uncontrolled levels in order to comply with the Reasonably Available Control
Technology (RACT) requirements promulgated by the Commonwealth of Massachusetts.
According to RACT, NOx emissions from the unit must not exceed 0.38 Ib./MMBTU
when firing coal or 0.25 Ib./MMBTU when firing oil.

For reduction of NOx emissions from Montaup boiler #8, combustion controls and flue
gas treatment were considered. Montaup Electric chose Selective Non-Catalytic
Reduction (SNCR) technology as its primary means of reducing NOx from baseline
levels to the levels required for RACT compliance. The SNCR technology operates by
injection of controlled amounts of aqueous urea into the furnace to reduce the NOx to
nitrogen, water and carbon dioxide. Total project scope included design engineering,
equipment supply, installation, and system startup. Optimization and startup of the
system were completed in early March. Startup testing demonstrated that the system can
achieve compliance levels across the load range 35% to 100% MCR while maintaining
ammonia slip at about 10 ppm or less. This paper will describe the system provided to
Montaup Electric and will discuss the results of startup and optimization testing.
Submitted to: EPRI/EPA 1995 Joint Symposium on Stationary Combustion NOx
Control, Kansas City, Missouri, May 16-19, 1995.
-------
Background

Montaup Electric Company operates a 112 MWe boiler at its Somerset, MA Generating
Station. Boiler #8 began operation in 1959. It is a natural circulation, reheat, single-
drum radiant boiler with a dry bottom that is rated at 800,000 Ibs/hr steam flow at the
superheater outlet at 1,925 psig and 1,000°F. The furnace has four corners and four
levels of tilting, tangentially fired coal buckets. Between the four coal buckets in each
corner are three oil guns. The fuel for the boiler was changed from coal to oil in the
early 1970's, and in 1984 the boiler was retrofit to operate primarily on coal. A
Continuous Emission Monitoring System (CEMS) was recently installed on the boiler,
and a new Distributed Control System (DCS) is planned for installation in April and
May of this year. Plans are to maintain boiler #8 in operation well beyond the year
2000.

Boiler #8 operates on dispatch typically in a load following mode. During the day and
during other high demand periods the boiler operates at full capacity. In the evening
and on weekends the boiler load is frequently reduced to 35% to 55% MCR.
Maintaining boiler #8 within environmental compliance under this wide range of
operating conditions is, therefore, of critical importance.

In the Commonwealth of Massachusetts, Reasonably Available Control Technology
(RACT) for control of NOx from tangentially-fired utility boilers is defined as
compliance with the following emissions performance levels: