EPRI
Electric Power
Research Institute
&EPA
May 1995
EPRI/EPA1995 Joint Symposium
on Stationary Combustion
NOX Control
Book 2: Wednesday, May 17,1995
sessions 4 and 5
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 2: Wednesday, May 17,1995
Sessions 4 and 5
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 4
Tuning and Optimization
-------
REDUCING NOx WHILE MAINTAINING BOILER PERFORMANCE
AT TVA'S JOHNSONVILLE STEAM PLANT
USING CONSTRAINED SEQUENTIAL OPTIMIZATION
R. J. Boyle
United Energy Services Corporation
P.O. Box 488
New Johnsonville, TN 37134
J. W. Pech, P.E.
Tennessee Valley Authority
1101 Market Street
Chattanooga, TN 37402
P. D. Patterson
PowerMAX Service of Ultramax Corp.
1251 Kemper Meadow Drive
Cincinnati, OH 45240
Abstract
The Tennessee Valley Authority (TVA) utilizes boiler tuning as an integral part of their
compliance strategy to meet the requirements of the Clean Air Act Amendments of
1990. A desirable tuning solution should not only be effective for minimising NOx
emissions but also should offer opportunities for improved thermal performance
leading to cost savings and a rapid return on investment. Guided by these objectives,
TVA applied the ULTRAMAX Method, which utilizes a new technology, co-developed
by EPRI and Ultramax Corp., called sequential process optimization. This technology
makes possible immediate reductions in NOx while constraining other emission and
thermal performance parameters to avoid adverse effects. The ULTRAMAX Method
consists of a methodology and the supporting computer software that builds on
existing boiler system knowledge and guides the operator through a sequence of
control parameter adjustments to achieve ever-improving levels of performance. It is
utilized during normal operations and can be tailored to boiler systems of virtually any
design. The software can be integrated with a data acquisition system to provide an
operator advisory capability. This paper describes an application of sequential process
optimization technology to six mid-sized, older-vintage coal-fired boilers. The
optimization project took place from September through November 1994. The results
were significant reductions of NOx from the previous baseline conditions at full load,
-------
with little more than two weeks of effort. In addition, heat rate and LOI reductions
were achieved for some units.
Introduction
Seeking to comply with requirements of Title 1 Phase n of the Clean Air Act
Amendments and to achieve cost savings, the Tennessee Valley Authority (TVA) has
been exploring opportunities to improve performance of mid-sized, older-vintage coal-
fired boilers at the organization's Johnsonville Fossil Plant. These boilers, rated at 125
MW, are circa 1950 CE tangential-fired units fueled by pulverized coal.
With limited capital available to replace existing burners on these units with low-NOx
burners, the Field Services Group of the Clean Air Project investigated several options
to achieve emissions reduction while expending the least amount of money. The most
promising option was determined to be the ULTRAMAX Method, which utilizes
sequential process optimization, a technology new to the power industry that was co-
developed by EPRI and Ultramax Corp. Application of the ULTRAMAX Method has
demonstrated a dear capability of minimizing NOx while maintaining heat rate and
causing no adverse effect on other thermal or emissions parameters. The ULTRAMAX
Method has been used successfully to improve process performance of more than a
dozen fossil units. Utilities that have employed the method at their facilities include
Georgia Power, Carolina Power & Light, Ohio Edison, South Carolina Electric & Gas,
Salt River Project, and Long Island Lighting.
Limitations of Parametric Testing
Detailed parametric testing is often used to quantify the effects on NOx emissions of
changes in individual operating conditions. This involves the design of a relatively
large set of runs to form a test matrix of all parameter combinations to be studied.
During a test run, the control setting of only one parameter is adjusted with all others
held constant. After all runs have been completed, the data are analyzed and best
operating conditions identified.
The greatest deficiency of parametric testing is that these procedures do not reflect the
influence on performance of interactions between all parameters. This interaction is
crucial in boiler applications, as optimum combustion conditions depend on a highly
interrelated group of controlled parameters (such as airflows and burner tilts) and
uncontrolled parameters (such as load and fuel quality) that can vary unpredictably.
As a consequence, parametric testing-based operating procedures are useful as general
guidelines for NOx control and overall boiler performance, but they cannot identify
conditions that balance emissions control and thermal performance. Plant operators
thus rely on their knowledge to tune airflow and burner settings, but it is difficult for
-------
experience-based tuning to optimize performance because of the dynamics and
increased complexity of boiler control, as well as individual operator preferences.
Sequential Process Optimization
The ULTRAMAX Method of sequential process optimization consists of procedures
and computer software that utilize data from the boiler unit acquired during normal
operations at parameter settings under operator control. These settings, recommended
by the software, take advantage of minor perturbations from standard settings in order
to learn their effects on emissions and thermal performance. The software then
immediately analyzes the data after each test run to create statistical models and search
for optimal combinations of settings.
The critical element of the ULTRAMAX Method is that each run is devised
independently after detailed analysis of all data from previous runs. For each run, all
control settings are purposefully and simultaneously adjusted; resultant performance
effects are observed; models are created reflecting the influence of all input parameters;
and new settings are advised for the next run, which is likely to obtain improved
performance.
Because data analysis and model refinement occur after each run, settings that may
harm performance can be immediately identified and eliminated from consideration,
which facilitates continuous improvement along a course toward optimum operations.
This characteristic makes ULTRAMAX suitable for normal operating environments,
unlike parametric testing. In addition, the cause-and-effect relations captured in
ULTRAMAX models allow immediate, intelligent response to changes in uncontrolled
variables, such as load, fuel quality, and seasonal temperature.
Online Optimization of Boiler Units
TV A Objectives
For the initial optimization project, TVA selected Units 3 and 4 at Johnsonville Fossil
Plant. TVA's objectives were to: 1) reduce NOx emissions from Units 3 and 4 to a
maximum of 0.45 Ibs/MBtu; 2) maintain or improve performance with regard to heat
rate and other key operating parameters; and 3) demonstrate the usability of
ULTRAMAX as an operations tool for ongoing boiler performance maintenance. These
objectives were to be achieved under acceptable operating conditions and without
violating any operating constraints.
Optimization Approach
Process Formulation. Using ULTRAMAX for boiler optimization requires three
distinct steps: process formulation, sequential optimization, and engineering analysis.
-------
To develop the process formulation, the knowledge of plant operators, engineers, and
others familiar with the boiler and control system was captured to identify and
categorize key input and output parameters, operating constraints, and optimization
goals. Procedures were discussed to identify the responsible personnel and time
periods for control adjustments, stabilizing time, recording of data, input of data, and
review of computer-generated advice. This was accomplished in a team meeting
lasting one day.
Parameters that were adjusted by operators to influence combustion performance,
called controlled inputs, included auxiliary air dampers (upper center, center, lower
center, and lower), burner tilts, and fuel dampers (levels B, C, and D). Constrained
outputs were AH gas in, AH gas out, CO2, CO, opacity, excess O2, boiler efficiency, and
LOI. NOx was the minimizing variable. (See Figure I for a summary of parameters.)
The process formulation, which included standard operating conditions, was input to
the ULTRAMAX software. Data from several test runs at high load were used to
establish baseline performance. The sequential optimization process took place over 15
business days. All tests were performed under the supervision of the TVA staff with
assistance from PowerMAX personnel.
Sequential Optimization. For the first learn/advice cycle, ULTRAMAX analyzed the
baseline data and the process formulation to provide advice for new settings of all
controlled variables. The advice was reviewed by the TVA engineer and operator, who
then implemented new settings for the control parameters being modeled. After the
boiler system responded to this perturbation and reached a steady state, output
parameters measurements were recorded and manually entered into the software for
analysis, model refinement, new advice, and a repetition of the procedures. During
subsequent cycles, the models became increasingly accurate, soon learning enough
about the process to advise new settings outside the plant's recent historical experience.
This facilitated significant NOx reductions on a path toward a true practical optimum.
A key element of the ULTRAMAX search process involves the weighting of more
recent data, which produces an accurate, goal-oriented local model, rather than a global
model. A global model fits all data equally well but is not accurate enough to identify
the refined control adjustments required to discover operating regions of optimal
performance. During initial "learning," the software strikes a balance between advised
settings for best performance and settings for intelligent perturbations, or explorations,
which are intended to provide additional information about cause-and-effect
relationships and further increase model accuracy and robustness.
The project team executed a total of 93 runs for Unit 3 utilizing advice from
ULTRAMAX software, and a total of 95 runs for Unit 4. Figure 2 indicates the changes
in observed outputs from baseline settings versus optimum conditions obtained with
ULTRAMAX. Figures 3 and 4 chart NOx profiles over the course of these runs.
-------
Engineering Analysis. In terms of engineering analysis, the software provided a
detected effects report showing the relative influence of each of the input variables on
each of the output variables of the models. Results of the optimization show that NOx
from Unit 3 is most affected by burner tilts and fuel damper level D. NOx from Unit 4
is most affected by overfire air flow and fuel damper levels B and C.
Conclusion
Under the guidance of ULTRAMAX, NOx emissions from these older coal-fired boilers
were reduced below the compliance regulation level of maximum 0.45 Ibs/MBtu
without need for further modification. NOx from Unit 3 was reduced from 0.46 to 0.40
Ibs/MBtu, a reduction of 13 percent, while gross heat rate was reduced by about 4
percent. NOx from Unit 4 was reduced from 0.54 to 0.43 Ibs/MBtu, a reduction of 20
percent, and gross heat rate was reduced by about 3 percent. All other parameters
remained within their constraints.
Subsequently, TVA used their own internal resources, with training from PowerMAX,
to extend optimization to Units 1,2,5, and 6. This project was accomplished in an
average of one week per unit.
Optimization resulted in reductions in NOx from Units 2 and 6, by 14.6% and 6.7%,
respectively. Emissions from Units 5 and 6 were already within required levels;
however, the project resulted in reductions of LOI for both boilers.
With regard to Unit 1, ULTRAMAX testing determined that airflow could not be
controlled within the windbox, auxiliary air dampers, and fuel air dampers because of
air leaks through windbox cracks and holes. Repairs have recently been completed,
and optimization with ULTRAMAX will begin again shortly.
The net benefit to TVA, including avoidance of retrofitting and average annual fuel
savings over 10 years, is estimated to be $17.4 million.
References
1. Catusus-Servia, J.J., Smoak, R.A., Squires, R.T., et al, "A Test for an Optimization
Method Applied to Controlling NOx Emissions in a PC-Fired Boiler." Joint paper
presented at ISA, POWID/EPRI Controls and Instrumentation Conference,
Phoenix, AZ, June, 1993.
2. Moreno, C.W., "How Modern 'Smarter Not Harder' Technologies Can
Simultaneously Maximize the Combination of Pollution Reduction and Business
Success." Paper presented at Conference on Environmental Commerce, CONEC
'93, Chattanooga, TN, October, 1993.
-------
3. Moreno, C.W., and Yunker, S.B., "Reducing NOx Emissions and Improving Boiler
Efficiency Using Synthetic Intelligence." Paper presented at Conference on Expert
System Applications for the Electric Power Industry, Phoenix, AZ, December, 1993.
4. Moreno, C.W., and Yunker, S.B., "ULTRAMAX: Continuous Process Improvement
Through Sequential Optimization." Paper presented at EPRI, Palo Alto, CA,
June, 1992.
5. Teetz, R., and Patterson, P.O., "Reducing NOx While Maintaining Boiler
Performance Using Constrained Sequential Optimization." Paper presented at
EPRI Fuel Oil Utilization and Management Workshop, Clearwater Beach, FL,
October, 1994.
6. Patterson, P.O., and Yunker, S.B., "Simultaneous Reduction of LOI and NOx
While Maintaining Boiler Performance Using Constrained Sequential
Optimization." Paper presented at US Department of Energy, Unbumed
Carbonaceous Material On Utility Fly Ash Conference, Pittsburgh, PA,
March, 1995.
-------
Decision Diagram
Over Air Flow
Aux Air Upper Center
Aux Air Center
Aux Air Lower Center
Aux Air Lower
Fuel Air B
Fuel Air C
Fuel Air D
Burner Tilt
^
^-
Load ^
TVA Johnsonville
Unit 1-6
125 MW
Tangential
Boiler
NOx Ibs/MBtu
CO
Opacity
SH Temp
LOI
Heat Rate
Boiler Efficiency
AH Temp In
Steam Flow
CO-,
Figure 1
Johnsonville Power Station
Baseline vs. Optimized
Outputs
LOAD MW
NOx Ibs/MBtu
Gross Heat Rate Btu/kWh
Boiler Efficiency %
CO ppm
CO, %
LOI %
Unit3
115
0.46
N/A
87.58
52.22
15.04
5.19
Unit3
Optimized
118
0.40
10,048
87.47
60.43
14.75
N/A
Unit 4
110
0.54
N/A
84.18
22.58
15.04
4.58
Unit 4
Optimized
115
0.39
10,200
86.69
120
14.05
N/A
Figure 2
-------
H
O
Z
0.40
0.35
0.30
11
TVA Johnsonville Unit 3
NOx Profile-Full Load
NOx - Actual
NOx - Moving Average
21
31
41 51
Runs
61
71
81
91
Figure 3
O
z
0.40
0.35
0.30
TVA Johnsonville Unit 4
NOx Profile-Full Load
NOx Actual
NOx - Moving Average (5)
11 21
81
91
Figure 4
-------
GNOCIS
AN UPDATE OF THE GENERIC NOx CONTROL INTELLIGENT SYSTEM
R. Holmes (PowerGen)
I. Mayes (PowerGen)
R. Irons (PowerGen)
J. N. Sorge (Southern Company Services)
J.W. Stallings (EPRI)
EPRI/EPA 1995 Joint Symposium on Stationary Combustion NOx Control
Kansas City, Missouri
May 1995
ABSTRACT
GNOCIS is an on-line enhancement to existing power plant Digital Control Systems
(DCS) designed to reduce NOx emissions while meeting other operational
constraints, such as heat rate and CO emissions. It can also be used to minimize
unburned carbon while meeting a specified NOx limit, or any combination of
emissions/performance variables that can be quantified by a common metric and are
affected by DCS - adjustable parameters. The core of the system is an adaptive
neural network model of the NOx generation characteristics of the boiler. The
software applies an optimizing procedure to identify the best setpoints for the plant.
The recommended setpoints can be either conveyed to the operator via the DCS in an
advisory mode, or implemented automatically in a closed-loop mode. GNOCIS is
designed to run on a stand-alone workstation connected to the DCS via the data
highway. Sensor validation techniques have been incorporated. The goal for
GNOSIS is to deliver 10-35% reductions in NOx from baseline conditions while
maintaining or improving other operational constraints.
Preliminary results are presented for demonstrations at two power plants:
• 500-MW T-fired boiler at PowerGen's Kingsnorth Station
• 250-MW Opposed-fired boiler at Alabama Power Company's Gaston Station
INTRODUCTION
The Clean Air Act Amendments of 1990 (CAAA) require that electric utilities
make significant reductions in nitrogen oxide (NOx) emissions from their fossil-
fired power plants. Similar NOx emission reductions are being required in
Europe through local and European Economic Community (ECC) initiatives.
To date, most efforts to reduce NOx emissions have come in the form of
retrofitting low NOx burners and operational modifications. Although in most
-------
instances NOx reduction targets can be achieved, the increased NOx reductions
can be at the expense of other important operational parameters such as fly ash
carbon-in-ash (CIA) levels, steam temperatures and boiler performance. The
adverse impacts can sometimes be mitigated by further low NOx burner or
combustion optimization; however, these optimizations are typically performed
at one operating condition and cannot fully account for changes in plant
operating condition over extended periods. Further, experience has shown that
NOx emissions, vary with time by as much as +20% around the mean. A system
that reduces the magnitude of these variations and drives the mean down
towards the lower end of this range could help achieve continuous compliance
without additional controls or provide NOx credits for averaging or trading
programs.
GNOCIS
The primary aim of the Generic NOx Control Intelligent System (GNOCIS) is to
develop an on-line enhancement to a power plant's Digital Control System (DCS) to
provide robust control settings which achieve NOx reductions within economic and
operational constraints. GNOCIS can operate on units burning gas, oil or coal and is
available for all combustion firing geometries. It is projected to reduce NOx
emissions by 10-35% from baseline while meeting other site-specific operational
constraints, such as CIA and furnace exit gas temperature.
GNOCIS is built around innovative modeling technology based upon artificial neural
networks and fuzzy logic. This software has been developed by Pavilion
Technologies, Inc., with input from the GNOCIS project team, and is licensed for use
in GNOCIS. The models are built initially from historical data, but are continuously
updated to reflect changes in the plant. The input data for the models is obtained
from the plant DCS. In the advisory mode, the advice is communicated through the
DCS using screen formats that are familiar to the plant operators. Optimum settings
for the combustion equipment, including mills, dampers, and excess air, are
continuously updated. Once the software has been accepted by the plant operators,
this process is automated, and GNOCIS operates as a closed-loop supervisory
controller.
The GNOCIS technology easily allows it to handle the multiple parameters that
impact the NOx/performance tradeoff across the load range of the unit. Analysis
has shown that some 25-50 plant parameters can be significant for conventionally
fired units, and this number may increase substantially for units equipped with LNB
and overfire air systems. The GNOCIS system has the capability of handling
hundreds of inputs in real time.
Demonstrations of GNOCIS are currently in preparation or underway at three power
plants:
• PowerGen's 500-MW tangentially fired Kingsnorth Unit 1
• Alabama Power Company's 250-MW wall-fired Gaston Unit 4
Page 2
-------
• Georgia Power Company's 500-MW wall fired Hammond Unit 4
The developmental projects at Kingsnorth and Gaston are being conducted in an
advisory mode to demonstrate the feasibility of the technology. The full commercial
demonstration at Hammond will be the first closed-loop application. Preliminary
results from Kingsnorth and Gaston are discussed in this paper.
The main contractors for the development of GNOCIS are PowerGen in the United
Kingdom and Southern Company Services (SCS) in the United States. The work
in the United Kingdom is being funded by the Department of Trade and Industry,
PowerGen, and EPRI. In turn, the work in the United States is funded by the
U.S. Department of Energy, The Southern Company, and EPRI. Commercializers
for North America are SCS and Radian Corporation. In its role as commercializer,
Radian is already deeply involved in the demonstrations in the U.S. PowerGen
and one other as yet unnamed organization will be the commercializers in Europe.
The Kingsnorth Trial
Introduction
The primary objective at Kingsnorth is to minimize CIA in the fly ash while
maintaining NOx below the current level of 320 ppm. With the current DCS
configuration, only seven parameters are adjustable by the operators - burner tilts
(ganged together as one setting), excess air, and five mill settings.
This section describes the background to the modeling and the need for trials at
Kingsnorth. Kingsnorth Power Station and its operation is described as are the
additions to the plant's instrumentation and gas and solids analysis necessary for the
implementation of GNOCIS.
Description of Kingsnorth Power Station
The Plant
Kingsnorth Power Station is on a coastal site on the Medway estuary in North Kent,
United Kingdom. It consists of four essentially identical 500-MW tangential-fired
units, which were commissioned between 1970 and 1973.
Coal is supplied to the station by ship. The major source of supply is the Tyne region
in the north east of England. However, the coal supply for much of the GNOCIS test
period was derived from reclaim of the station stockpile, which included fuel
supplies built up over a period of several years from a variety of worldwide coal
sources, in addition to indigenous supplies.
PageS
-------
The boilers were constructed by NEI International Combustion to be capable of
meeting full load on either pulverized coal or residual fuel oil (RFO). This capability
still exists, but local economic factors now dictate that as much as possible of the load
is met by coal. However, the original dual-fuel design meant that the units were
effectively built without a spare mill, the rationale being that any load shortage could
be picked up by RFO firing. Hence, each unit is fitted with only five mills, all of
which are required to make full load on most of the coals supplied to the station.
The furnace has a central vertical dividing wall which forms two identical
combustion chambers, designated A and B. The four burner boxes in each chamber
have independent tilt control. In theory, tilts can be set anywhere between +20 and
-20° from the horizontal. In practice, tilts below -5° are virtually never used due to
potential problems with carbonaceous material buildup in the ash hoppers, which is
a safety hazard.
Each mill fires a single level within the furnace. The mills are designated A through
E, with A mill firing the top row in the furnace and E mill firing the bottom. The
furnace is fitted with a Low NOx Concentric Firing System (LNCFS) with separated
overfire air. Some levels of oil burners (4 and 6) feature 8 burners on a control circuit
and are referred to as "full banks," while other "half banks" (3 and 3A, 5 and 5A) have
diametrically opposed pairs of burners in each half of the furnace.
The furnace is operated in a balanced-draft mode. Superheater temperature control
is achieved by spray, and reheater temperature is maintained by either varying the
tilt on the burner boxes or utilizing a reheater by-pass control.
The percentage of overfire air (OFA) used in the system varies as a function of load.
At full load, the OFA is approximately 15% of the total air flow. OFA damper
positions remain constant as load decreases until, at 400 MW, they begin to close.
As load decreases further, the OFA dampers become nominally closed at 350 MW
and allow only a leakage cooling air flow to pass. Although it is, in theory, possible
to reprogram the damper control software to vary the amount of overfire air, this is a
relatively complex procedure and is outside the scope of normal operator activity on
the plant.
Plant Control Systems
Furnace oxygen is controlled in a closed loop. The controllers automatically adjust
the ID and FD fan speeds to give the desired O2 value at the economizer exit, while
maintaining the required furnace suction.
The mill control system uses mill feeder speeds as the prime control variables. In
fully automatic mode, the goal of the control system is to match the feeder speed of
all mills in service, while maintaining the required load. One or more mills may be
put on manual control, where the feeder speed is fixed at a constant value and the
remaining feeder speeds are again varied to meet the required load.
Page 4
-------
Secondary air damper settings are in closed-loop control. The degree to which the
various dampers are opened is determined both by load and by the mills and oil
guns in service.
Superheater temperature control is on an automatic control loop. Reheater
temperature control is dependent on operator intervention. The operator may
choose, depending on furnace conditions (particularly the degree of slagging and
fouling), either to move the reheat by-pass damper or to alter the tilt on the burner
box or to instigate sootblowing.
Normal Operation
It has been normal routine to operate the plant at a constant excess C>2 set-point of
3.0%, which was judged to represent an acceptable compromise between CIA losses
and dry gas losses. High CIA has traditionally been a problem at the station and
represents the major efficiency loss. At loads below 300 MW, oxygen levels increase
dramatically (to 6-7%) as it becomes necessary to use extra air to (a) maintain
windbox pressure and (b) maintain cooling flow to out-of-service burners.
In general, if all five mills are available, all are used; with coal flows to the bottom
four mills equalized and a reduced flow to the top mill. The latter is expected to give
a disproportionate contribution to CIA losses because of the relatively small
residence time of the coal entering the furnace at that level. If all the mills are not
available, it is the general practice to run all available mills equally balanced and to
meet any shortfall in required load by firing oil. The oil burners to be used are left to
the discretion of the operators.
Control of reheat temperatures by the manipulation of tilts, reheat bypass and
sootblowing is again at the discretion of the operators.
In UK terms, Kingsnorth is relatively remote from the coal fields and has a high coal
delivery cost. Hence, it has a relatively low load factor and tends to cycle on dispatch
for two shifts a day and shut down for the third at all but the periods of highest
demand during the year. During these tests, Unit 1 - the most efficient unit - was
generally off load overnight, at full load during the morning and evening peaks, and
anywhere between minimum load (230 MW) and full load during the rest of the day.
The station has a NOx emission limit of 390 ppm at 6% O2-
Data Acquisition
Digital Control System
The digital control system at Kingsnorth uses an in-house system known as
CUTLASS which is based on DEC PDP11 hardware and Instem I/O equipment. The
system follows the Host-Target model and uses Ethernet communications. There are
10 targets containing data required by the GNOCIS system and to minimize the
Page 5
-------
changes to these targets, an additional target has been introduced. This additional
target handles the acquisition of extra data and the display, trending and archiving
functions for the new GNOCIS displays.
The only change to the existing targets is the introduction of an additional CUTLASS
scheme (software module) to broadcast the data required from each target onto the
Ethernet.
Some additional data is required which is not available from the CUTLASS system.
This supplementary plant data is derived from additional instrumentation located in
a mobile laboratory and from other station instruments. These signals are fed to an
Instem Link-on device which is an 8-channel analog-to-digital converter which can
be driven easily from CUTLASS. These additional signals are scanned every
5 seconds and placed in a CUTLASS global array for onward transmission to the
CUTLASS Gateway PC every 10 seconds.
Results from the GNOCIS model are broadcast on the Ethernet by the CUTLASS
Gateway PC and picked up by a scheme in the additional target once per minute.
These results, together with the data from the existing targets and the supplementary
data are held in a data array for use by the GNOCIS display, trending and archiving
schemes. Figure 1 shows the GNOCIS how NOx optimization screen which the
operators consult for advice.
Instrumentation
For the purpose of these tests additional gas analysis equipment was installed in
a caravan close to the plant's electrostatic precipitators. Gas samples are obtained via
a sintered stainless steel probe inserted into the gas stream at the ID fan outlet and
transferred to the analytical equipment via an unheated PTFE tube. The gas is dried
in a cooler, passed through a pump and, via a filter, into the gas analyzers.
Individual gas sampling trains are provided for the A and B sides of the furnace.
On the B side, the analysis train consists of a Servomex O2 analyzer, two Hartmann
and Braun Uras 3G analyzers for CO and SO2 and a Radas 1G NO meter. The A side
analysis is carried out with a multi-channel Hartmann and Braun Uras 10E gas
analyzer for O2, CO, NO and SO2.
In order to address the station's CIA problem, Unit 1 has been fitted with two Mark
and Wedell 'M&W On-Line Residual Carbon Analyzers. One instrument is mounted
on the A side, the other on the B side. The M&W instruments depend on a
conventional single-point ash sampling system used in combination with a
reflectometer, which measures ash reflectance and hence, by inference, carbon
content.
Page 6
-------
Site Trials
It was originally thought that normal operational changes would provide sufficient
variability to develop and test the models, but in practice this approach was not
adequate, especially for the information associated with the mills and feeders.
A series of parametric tests in November 1994 and December 1994 was planned and
the data from these tests has been used to obtain these results. These trials covered
periods of normal operation and periods when specific parametric tests were carried
out.
Analysis of Site Trials
Data Preprocessing
The data acquisition system was receiving and storing data twenty-four hours a day
throughout the Kingsnorth trials. However, not all data was suitable for use in the
models; in its raw form the data covered periods when the instrumentation was
faulty and when the plant was operating in a regime outside of the GNOCIS
specification (zero and low load).
Data was therefore preprocessed with the following objectives:
• The removal of data corresponding to the unit being off; and
• The removal of data corresponding to operational regimes beyond
GNOCIS' scope
using the extensive preprocessing facilities built into Pavilion's Process Insights
software.
Predictive Modeling
The November/December tests were evaluated in two ways: one to assess the ability
of the models to predict the data and, more importantly, one to assess the ability to
control the data — GNOCIS' main function is to give operational advice, not to be an
accurate predictor of NOx and CIA. Predictive models were constructed, as a first
step, to give an indication for the overall accuracy of the modeling and to highlight
potential difficulties.
The performance of the model was evaluated by selecting up to four periods of
operation, omitting these from the input data and then running the model on this
data. As the model is only a step on the way to the (more important) control model it
was decided that comprehensive statistical testing of the model was not appropriate.
Control Modeling
As discussed in the Predictive Modeling section four periods of operation were
chosen against which the model performance would be assessed. This was done for
Page 7
-------
the control models by taking expert advice on what information should have been
given to an operator to prompt any desired control action. The experts were the
Kingsnorth Efficiency Engineer and a PowerGen combustion expert. This advice was
then compared with control advice suggested by the model. The variables chosen for
the control were oxygen setpoint, burner tilt, and the five feeder speeds. These are
the only parameters that the operator can directly affect.
In order for any optimization procedure to be effectively integrated into the working
environment of a power station, there are a number of requirements that must be
built into the procedure to ensure that its output is acceptable to the plant operators.
One of the features of coal fired plant is that pulverized coal is supplied to the boiler
through a number of mills and coal feeders. Although each feeder is capable of
independent operation, in reality they are controlled as a group with the important
requirement that the total amount of coal delivered can produce the electrical load
demanded. This must be implemented in any optimization procedure.
Another feature of this multiple feeder property is the possibility that a number of
feeders may produce almost identical responses in the output variables and this
presents problems to the optimization process. Furthermore, the coal milling process
is such that the coal throughput is not continuously variable from full output to zero
output but has a lower threshold below which the mill would be taken out of service.
This mixture of discrete and continuous behavior of the control variables is not one
that is commonly encountered in standard optimization problems. It also has to be
noted that the process of removing and returning mills to service is not a task to be
undertaken lightly, and it is not a procedure that operators wish to happen
frequently. In addition to these constraints on the operation of the optimization
algorithm, there are engineering upper and lower bounds on the control variables.
Control Model
In its control models, Process Insights allows correlated inputs provided the user can
specify which are the independent control inputs and which are the dependent
inputs (known as states). Building the control model involves modeling the outputs
as a function of the all the inputs (controls and states) and then modeling the states as
a function of the controls. Once the model has been built, a runtime model is
produced which can be accessed by the Process Insights runtime libraries. These in
turn are called by C or FORTRAN programs that the user has written.
The optimization involved with the control model is complicated due to the variable
number of coal feeders that could be used. The problem was solved by writing extra
"wrapper" code around the basic optimization procedure. This separated the
continuous and discrete aspects of the optimization process.
Page 8
-------
Operational Advice
The small number of control variables and the physics of the problem suggest that
the advice given by the optimizer will not vary greatly, apart from changes in feeder
speed that are needed to comply with the load constraint. However, early
indications at Kingsnorth are that the optimizer does consistently suggest an
operational mode which is unfamiliar to the plant operators, namely in the
distribution of feeder speeds. The load control system currently used at Kingsnorth
tries to balance the coal flow equally between all available feeders whereas, by its
very nature, optimization does the opposite, running the lowest-cost feeders at a
maximum and using the highest-cost feeders to meet the load constraint
The NOx constraint was set at 320 ppm, and the oxygen and burner tilt advice
depended upon whether the NOx constraint was active or not. If the NOx constraint
is inactive, the model recommends that the oxygen be increased to a maximum and
the burner tilt reduced to a minimum. The effect of this advice is to reduce both CIA
and NOx; these are normally regarded as opposing variables.
This is shown in Figures 2 and 3, the bold line corresponding to data obtained while
following the advice and the normal line for data obtained when, for various reasons,
the advice was ignored. The weighted means for the 10-hour period are 415 ppm to
398 ppm NOx and 9.4% to 7.0% CIA. It is important to emphasize that the objective
of the Kingsnorth exercise was to minimize CIA, not NOx, so the NOx reduction
observed is secondary to the CIA results. This indicates that a further reduction in
NOx is possible if this is the prime objective.
When the NOx constraint is active but not exceeded, the same type of advice will be
given. Thus, one set of earlier results have shown a reduction from 7% to 4% CIA
using the advice from GNOCIS, for an overall decrease of over 40%, while at the
same time NOx was lowered from 320 ppm to 290 ppm. These results were achieved
simply by lowering the burner tilts.
An additional test assessing the five mill settings demonstrated the power of
GNOCIS as an optimizer. E mill feeds the lowest burners in the bank; hence, fuel
supplied by this mill has the longest residence time in the furnace. Preferential use of
this mill should produce minimum CIA. GNOCIS suggested that this was not the
case and that the opposite was true, implying that there are some burners with an
incorrect air/fuel ratio being fed by this mill, or that a greater percentage of larger
particles (e.g., >50 mesh) are coming from this mill. In order to demonstrate the
effect of E mill on CIA, it was necessary to change the coal flow of E mill while
keeping constant that of A mill, to avoid the changes due to E mill variations being
masked by those due to A mill. A test was carried out with initial feeder speeds of
A feeder at 1200 rev/min, and B, C, D, and E feeders at 1600 rev/min. This situation
was changed to A feeder at 1200 rev/min; B, C, and D at 1500 rev/min; and E feeder
at 1800 rev/min. The results are shown in the figures 4,5, and 6, depicting E mill
feeder speed, CIA, and NOx, respectively. These clearly show the CIA increasing
Page 9
-------
from 15.2% immediately before the change to 18.8%. The NOx does not show any
obvious change.
The Gas ton Trial
The objective of the Gaston trial is to develop and demonstrate GNOCIS on a
wall-fired unit. NOx is to be minimized while meeting other operational constraints.
A description of the unit and the activities associated with development at this site
follows.
Description of Gaston Unit 4
Plant E. C. Gaston is located near Wilsonville, Alabama, approximately 30 miles
southeast of Birmingham. The site has five pulverized coal units. Units 1 through 4,
250-MW wall-fired units, are owned by the Southern Electric Generating Company, a
wholly owned subsidiary of Alabama Power and Georgia Power. Unit 5, an 880-MW
tangential-fired unit, is owned exclusively by Alabama Power. The site is operated
by Alabama Power.
Combustion System
Gaston Units 1 through 4 are sister units and are similar in most respects. These
units began commercial operation in the early 1960s. Gaston Unit 4 began
commercial operation on July 1,1962. The 250-MW pulverized coal unit, designed
for continuous indoor service, has the capability to fire loads as high as 272 MW. It
uses a balanced-draft, Babcock and Wilcox (B&W) opposed-wall-fired boiler rated at
1,700,000 pounds of steam per hour, at a pressure of 2075 psig, and at a temperature
of 1000°F at the superheater and reheater outlets. The boiler is arranged with nine
burners (3W x 3H) on two opposing walls such that no burner has another burner
directly across from it. Combustion air is supplied to the burners via common wind
boxes on each side of the boiler. Prior to the spring 1994 retrofit of B&W XCL
burners, the boiler was equipped with B&W S-Type burners. The furnace zone is
compartmentalized into three sections. Two division walls hang from the roof of the
boiler and extend downward to just above the top of the bottom ash hopper. In the
furnace, a three-foot-wide gap separates the division walls. This gap shrinks to a
width of 18 inches above the boiler nose.
Each of the six B&W EL-76 ball and race mills supplies pulverized coal to a set of
three burners on one elevation of either the front or rear wall. Fuel is delivered to the
mills by two-speed table feeders. Two forced-draft fans supply all the combustion
air (except tempering air) to the boiler. Six primary air fans draw air from the forced-
draft system to supply hot air to each of the pulverizers. The unit is equipped with
two flue gas recirculation fans. Combustion air is heated with Ljungstrom air
preheaters. Exiting flue gas is treated with a Research-Cottrell hot-side electrostatic
precipitator.
Page 10
-------
Boiler Control System
The boiler control system for Gaston Unit 4 is a Leeds and Northrup Max 1000
distributed digital control system. This system was commissioned during May 1992.
The control system is designed such that the unit is controlled through the CRTs -
there are no bench board mounted controls. The system has over 3200 input/output
(I/O) points when PLC-based I/O is taken into account. The major components of
the DCS are as_ follows:
Graphic Processors (GP) - Thirteen graphic processors are included in the system,
twelve of which are used for operator displays while the other is used for
engineering purposes. The operator console displays, trackballs, and keyboards all
work through the GPs.
Real Time Processor (RTF) - There are a total of five RTFs. These processors provide
the interface between the GPs and the other devices on the data highway. The RTP
also accumulates and stores blocks of data such as alarms, events, and trending for
use by other processors.
Application Processor (AP) - There are two APs in the control system. These systems
perform historical archiving and retrieval, report generation, system configuration,
and special applications. These APs are based on an Intel 80x86 architecture running
a variant of Unix (SunSoft's Interactive Unix V/386). As such there are a number of
useful utilities and services available on these platforms for general program
development.
Data Highway - The system includes a fully redundant fiber optic data highway.
The data highway connects all workstations and distributed processing units.
Distributed Processing Unit (DPU) - The DPU is a multiprocessor unit that provides
data acquisition and control. The DPU is connected to the I/O modules through a
high speed parallel bus and the remainder of the control system through the fiber
optic data highway.
GNOCIS Implementation
At Gaston, the objective was to implement an open-loop, advisory GNOCIS system
with no immediate plans to migrate to closed-loop control. This objective influenced
the design philosophy in a number of respects, primarily:
• Selection and quantity of control variables,
• Increased demand for flexible and informative operator displays, and
• Reduced necessity for stringent recommendation checking and incorporation of
safeguards.
Page 11
-------
Figure 7 shows the informational flow for the GNOCIS implementation at Gaston.
All process data is collected through the DCS and passed on to the GNOCIS host for
calculation of the recommendations. These recommendations are then passed
backed to the operator via the DCS operator displays. If accepted, the operator can
then implement these changes through the DCS operator displays.
Instrumentation
In general, the existing control system provided an excellent platform for the
collection of real-time process data in a format usable by GNOCIS. However, two
short-comings were identified as a result of the initial site surveys.
The most formidable shortcoming was the lack of a suitable continuous emission
monitor (CEM) on this unit. Although equipped with a compliance monitor,
• The extraction point for the CEM was in the stack liner shared by Units 3 and 4,
and
• The CEM system was a dilution-extraction system and did not have a CO
monitor.
Since CO was considered important as a combustion efficiency parameter and it was
unlikely that Unit 3 would be down for considerable periods of time, a temporary
extractive CEM with NOx, CO, and O2 analyzers was installed just upstream of the
confluence of the Unit 3 and Unit 4 flue gas streams.
As with many units of this vintage with low NOx burners burning Eastern
bituminous coal, CIA levels are an important operating consideration at Gaston.
However, due to long measurement delays and the difficulty of obtaining
representative fly ash samples, it is difficult to correlate CIA levels with operating
conditions. Consideration was given to the installation of an on-line CIA monitor,
however, due to (1) uncertainties in the performance of these systems, (2) project cost
constraints, and (3) developmental nature of the program at Gaston, it was decided
to forego the on-line measurement. In lieu of this measurement, correlations were
developed for CIA based on short-term testing and later incorporated into the
combustion models.
Networking / Software / Hardware
A key component of a GNOCIS installation is the interface between GNOCIS and the
DCS which is the source of all process data. Although in some circumstances it may
be feasible to have GNOCIS reside in the DCS proper, generally it is more
advantageous to have GNOCIS reside on a distinct host platform and then establish a
suitable communications link to transfer information between the two systems.
As shown in Figure 8, the GNOCIS host platform is a Windows NT workstation
networked to one of the two application processors in the digital control system
using an Ethernet connection. As mentioned previously, the application processors
Page 12
-------
use a commercial Unix variant as the host operating system which aided in the
development of the communications software by SCS. L&N did not have any
off-the-shelf software which would support the obtaining and transfer of the
information between the application processor and a remote platform. The only
hardware modification to the DCS was the installation of an Ethernet network
adapter card in the application processor. TCP/IP was selected as the network
protocol, and IP addresses were assigned to both the application processor and
GNOCIS platform. Process data is obtained from the DCS on a 10-second cycle and
then forwarded to the GNOCIS platform where one-minute averaging is performed.
Eighty-four points are included in the transfer list at present.
Currently, the plant information network is being used as the transfer medium
as opposed to a dedicated line between the application processor and GNOCIS
platform. This choice has enabled remote development and shakedown of the
system and simplified the task of obtaining historical data from the site. Long-term
plans are to place the GNOCIS platform between the plant network and the
application processor to improve security on the DCS (Figure 9).
As an interim measure in the deployment of GNOCIS, WonderWare running on the
NT platform is being used as the operator interface. This route was taken for several
reasons:
• Development of operator graphics could proceed at SCS offices in Birmingham,
• Possible interference with normal unit operations was avoided,
• Quicker graphic development was possible, and
• Remote, off-site display of graphics and charts was allowed.
As part of a permanent installation, the operator graphics will be migrated to the
L&N DCS.
Model Development
Data collected through the DCS was used to create the combustion models.
Although on the order of a 1000 points are being archived in the DCS, early in the
project, eighty-four process parameters were identified as being possibly important
for combustion modeling purposes. Initial modeling efforts, conducted by Radian
Corporation, have concentrated on data collected during normal unit operation
during October and November 1994 and data collected during several days of short-
term testing during November 1994. These short-term tests, during which the unit
was run at off-design conditions, were designed to augment data available from
normal operation and thereby expand the range over which the combustion model
could make estimates. The collected data was preprocessed to remove bad data and
data collected during transients. In all, 25,000 one-minute patterns (representing in
total approximately 17 days of data) were used to train the models.
Page 13
-------
An initial step in the modeling program was the development of a number of
predictive models. An example of the predictive capabilities of the system is shown
in Figure 10. As shown, NOx emissions predictions based on boiler combustion
parameters (such as primary air flows, excess G>2, feeder speeds, etc.) were relatively
accurate for this time period. As expected, predictions for CO were not as robust.
Although predictive models are useful in a number of circumstances, what is
required of GNOCIS are control models. Considerations in control model
development were:
• Sensitivities of model outputs (such as NOx) to available inputs, and
• Control points readily changeable by the operator.
A summary of the variables included in the final control model are shown in Table 1.
Using the above control model, two strategies were evaluated:
• Efficiency optimization strategy, and
• NOx minimization strategy.
In many instances, such as when NOx emissions are comfortably below the
regulatory limit, improvement of boiler efficiency may be of principle concern. For
the test of this scenario, all steady state loads were considered (i.e. transients were
eliminated from the dataset) and all control variables (from the list shown in Table 1)
were made available for optimization. Based on predicted results, a 0.1 to 0.4 percent
increase in boiler efficiency could be obtained while having only a slight adverse
impact on NOx emissions (generally less than 0.02 Ib/MBtu). Predicted LOI
improved by approximately 1 percent.
For the NOx minimization scenario, only full-load, steady-state operation was
considered. Also, to facilitate interpretation of the results, only mill biasing was
made eligible for optimization, and excess O2 and flue gas recirculation were
clamped to nominal values.
In this mode, predicted NOx emissions were reduced from approximately
0.40 Ib/MBtu to 0.28 Ib/MBtu, while boiler efficiency dropped by approximately
0.1 percent. This was accomplished by moving the mills from a near balanced
to a biased condition (Table 2). Although this bias may not be feasible for actual
long-term operation, this scenario does at least lend hope that opportunities may be
present for significant NOx reductions without overwhelming performance
penalties.
Again, these are predicted results, and although encouraging, they need to be
substantiated with more thorough plant testing which is now underway.
Page 14
-------
Current Activities, Status, and Plans
Work is still in progress at Gaston. A summary of the current status and plans for
Gaston are as follows:
• The preliminary control model has been installed and tested briefly at the site.
• Using data collected during February 1995, the model will be retrained so that it
more closely reflects current operating conditions.
• The GNOCIS "Beta" software now being tested is currently being upgraded with
a more robust optimizer and enhanced operator displays.
• Operator displays will be migrated to the L&N DCS making it more convenient
for operators to obtain advice.
Further testing of GNOCIS at the site is planned for April and May 1995. This testing
should provide a good indication of the actual benefits of a GNOCIS installation at
this site.
Table 1. Gaston Model Variables
Control State Variables
Mill primary air flows Mill throughput
Excess O2 Drum pressure
Gas recirculation Tempering air flow
Output variables Economizer outlet temperature
NOx Reheater outlet temperature
Fly ash loss on ignition (LOI) Air heater outlet temperature
Boiler efficiency
CO
Table 2. Mill Biasing
Mill Mill Location Change in Mill Loading
A Bottom - Rear t
B Top - Front NC
C Middle - Front NC
D Bottom - Front t
E Top - Rear 444,
F Middle - Rear t
Page 15
-------
Low NOx Optiiisation
•20.1
Oiygen X
Burner Tilt A Deg
5.5
11.1
17.!
20.1
1U3
Key: • Current Value • BID Value
• Seconendat ion & Constraint
• Prediction
HOI ppi
20.0
0.0
Carbon in Ash \
20.0
Unit losd [81] 36*
Feeder Speed Reconendat ions 01
(till Fl or irroi key to illict entry point
feeder Speeds RF8 2100
Figure 1
NOx During 10 Hour Periods on Consecutive Days
*20.0
4000 I
390.0
3700
3600
Figure 2
Page 16
-------
Carbon-in-Ash During 10 Hour Periods on Consecutive Days
E
13th March
14th March
300
time/minute
Figure 3
1800
1750
1700
I 1650
•o
S.1600
1550
1500
1450
Feeder E speed v time
10 20 30 40 50 60 70 80 90 100
time minutes
Figure 4
Page 17
-------
Carbon in ash v time
21
20-
19-
18-
17-
315
14
13
12
11
10 20 30 40 50 60 70 80 90 100
time minutes
405
400
395
Q.
Q_
X
O
390
385
380
10
20
Figure 5
NOx v time
-i 1 1 r
30
40 50 60
time minutes
70
Figure 6
80
90 100
Page 18
-------
Process Noise
Setpoints
L&N Max 1000
Actuator Setpoints
Plant
Advise
GNOCIS
L&N Max 1000
Proces
Data
Display
Data Ack
Figure 7. Overview of Gaston Unit 4 GNOCIS Installation
L&N Max 1000 AP
Interactive Unix / 486
Communications
• FTP
- TELNET
• RLOGIN
• Sockets
• Custom Programming
Historical Capabilities
• WORM Drive
• Custom Program to Retrieve
Data
Access to Real-Time Data from
L&N Highway
Ethernet
UDP Service
GNOCIS Platform
Windows NT / Pentium
WINSOCKETS
Use named pipes to converse
with PI
Local graphics (optional)
• WonderWare
Historical archiving (optional)
Figure 8. Gaston 4/ Host Platforms
Page 19
-------
DCS Data Highway
L&N MaxlOOOft Operator:
AP •"--':= Displays^
GNbcisr'!?
HostPlatfpnm
Plant Network
^ 1
H
SoCo WAN
SCS Network
Supports
Services-
Figure 9. Gaston 4/Communications
400
•c
o>
T3
O
a
a.
x"
O
Southern Company Serwces
GNOCIS / Gaston Unit 4
October 1994 / 5 Minute Data
100 150 200 250
NOx, ppm
300
350 400
Figure 10. Gaston 4/NOx Predictive Model
Page 20
-------
COMBUSTION OPTIMIZATION OF LOW NOX
BURNERS AT PEPCO's MORGANTOWN STATION
P. Maines
D. Schnetzler
A. Bilmanis
Potomac Electric Power Company
Morgantown Generating Station
Newburg, Maryland 20664
S. Williams
Potomac Electric Power Company
8711 Westphalia Road
Upper Marlboro, Maryland 20772
M. D'Agostini
D. Eskenazi
T. Schmitt
T. Eldredge
N. Sarunac
E. Levy
Energy Research Center
Lehigh University
Bethlehem, Pennsylvania 18015
E. Petrill
Electric Power Research Institute
Palo Alto, California 94303
Abstract
PEPCO's Morgantown Station, in planning for the upcoming CAA regulations, entered into a
Tailored Collaboration project with EPRI, the purpose of which is to optimize the ABB/CE
LNCFS III firing system at Morgantown to achieve the greatest possible NOX reductions, with
minimum degradation in heat rate.
Controlled parameters of optimization include: distribution of overfire air, burner and SOFA tilt
angles, mill bias, furnace oxygen, windbox pressure, boiler cleanliness and secondary air damper
biasing. To develop the necessary parametric relationships, Lehigh University and PEPCO
conducted tests on the boiler, varying the parameters individually and in combination. From
-------
these tests, the optimal operating conditions were determined and new control algorithms
were developed and programmed into the unit's DCS system. The results of the full load
parametric testing are discussed in this paper.
Introduction
PEPCO's Morgantown Unit 2 has a supercritical Combustion Engineering (CE) tangentially-fired
split furnace boiler with a single reheat turbine. In May 1994, the unit was retrofitted with a CE
LNCFS III firing system and a new Foxboro control system. Prior to the burner conversion in
May 1994, tests were performed which indicated the unit had relatively high full load baseline
NOX emissions levels in the range of 1.0 Ib/MBtu. Because the unit is subject to both Title I and
Title IV NOX regulations, PEPCO initiated an effort to reduce NOX emissions to the lowest
achievable levels with the new burners. This paper describes the development of techniques to
optimize the combustion of the boiler to reduce NOX emissions, while maintaining acceptable
levels of heat rate. Unburned carbon is a particularly critical parameter at this unit because of the
adverse impact of high unburned carbon levels on electrostatic precipitator performance at full
load conditions. The investigation, which is being carried out jointly by EPRI, PEPCO
engineering staff and Lehigh University's Energy Research Center, is a joint PEPCO/EPRI
Tailored Collaboration Project.
Unit Description
The unit has a full load rating of 585 MW. It is equipped with five pulverizers, a balanced draft
furnace and electrostatic precipitators for particulate control. The LNCFS III firing system
consists of low NOX burner buckets, concentric firing air registers, close-coupled overfired air
(CCOFA) and separated overfired air (SOFA) with adjustable yaw. Figure 1 shows an elevation
view of the burners and air registers, indicating the terminology used to designate the five
elevations of burners and the air registers. A plan view of the furnace (also shown in Figure 1)
indicates the numbering system used at the plant for the eight corners in the two sides of the
furnace.
Combustion Optimization Approach
Following the optimization methodology developed previously on a boiler with conventional
burners at PEPCO's Potomac River Station (1), the sequence of steps used at Morgantown Unit 2
is described below:
• Post-retrofit coarse tuning.
• Post-retrofit baseline testing.
• Combustion optimization testing.
Control algorithm development.
• Implementation.
• Fine-tuning.
-------
SOFA
18
17
16
15
14
13
12
11
10
9
8
7
6
5
4
3
2
1
CCOFA
CCOFA
Coal A
Oil/Air
CoalB
Air
Oil/Air
Air
CoalC
Air
Oil/Air
Air
CoalD
Air
Oil/Air
Air
CoalE
Air
Damper Arrangement
Hot Corners: 2, 4, 5, 7
Cold Comers: 1,3,6,8
North
3 6
South
4 5
Furnace Arrangement
Figure 1
Sketch of windbox elevation showing burners and air registers.
Plan view of furnace showing numbering system for the eight corners.
Post-Retrofit Coarse Tuning
This phase consisted of tuning the control system for acceptable start-up, steady-state operation
and compliance with opacity regulations. Operating curves developed by CE involving windbox
pressure, tilts and economizer O2 were used for start-up. The SOFA dampers were closed during
this initial phase due to the unknown effects on ignition points, combustion and opacity.
Post-Retrofit Baseline Testing
Once the unit was started, CE engineering staff adjusted the economizer O2 level, auxiliary air
and SOFA damper settings, SOFA tilt and burner tilt angles to achieve their target windbox to
furnace pressure differentials, NOX emissions levels and boiler performance. The resulting
control curves were programmed into the DCS by PEPCO. Listed hi Table 1 are the values for
NOX emissions and boiler performance guaranteed by CE along with the measured baseline
-------
Table 1
Boiler Vendor Performance Guarantees
and Results of Post-Retrofit Baseline Test
Load
NOX emissions (Ib/MBtu)
CO emissions (pm)
LOI (%)
Efficiency
Guarantee
585 MW
0.66
100
6.9
90.33
CE Test Data
585 MW
0.58
20
5.7
92.09
values obtained during the acceptance test. This condition is referred to in this paper as the post-
retrofit baseline.
Combustion Optimization Testing
Parametric boiler tests were performed at four different load levels to determine which of the
control parameters affects NOX emissions, opacity and boiler performance and to determine the
best combination of settings. Since the unit operates most frequently at full load (585 MW), the
bulk of the testing was performed at that load level. However, additional tests were also
performed at 250, 375 and 450 MW. At each test condition, NOX and CO emissions, opacity,
steam temperatures and other relevant combustion and boiler performance parameters were
measured using data from plant operating instrumentation and the Continuous Emissions
Monitor (CEM). For those test conditions at which data for unburned carbon were needed, full
economizer isokinetic traverses were performed. The fly ash samples were then analyzed in the
laboratory for loss-on-ignition (LOI). NOX was measured at each ESP exit duct and in the stack.
At full load, the major limitation on the ability to reduce NOX is opacity, which is limited to 20
percent over a six-minute average. It was concluded early in this study that the effect of the
boiler settings on precipitator performance would need to be understood and minimized if
significant reductions hi NOX were to be achieved.
The parameters tested were:
• SOFA opening, both average opening and distribution of SOFA settings.
Economizer O2.
Burner and SOFA tilt angles.
Corner to corner damper bias.
-------
• North/South furnace bias.
• Mill loading pattern.
• Boiler windbox pressure.
• SOFA yaw settings.
• CCOFA settings
• Boiler cleanliness
Results
The results of this paper describe the tests and analyses performed at full load conditions
(585 MW).
NOX Emissions
SOFA damper opening was found to have a greater NOX reduction capability than any other
parameter. Figure 2 shows the test data obtained at two economizer O2 levels as average SOFA
damper position was varied. These show that NOX decreases substantially with increasing SOFA
setting and decreasing O2 level. Using a neural network, continuous relationships were
developed between NOX and O2 and SOFA position. The resulting curves showing NOX as a
function of these two variables are given in Figure 3.
The results in Figures 2 and 3 were obtained with all four elevations of SOFA dampers open the
same amount. Tests were also performed in which the SOFA damper settings varied by
elevation. By adjusting the SOFA dampers to permit proportionally more air through the upper
three dampers (#'s 20, 21 and 22), additional reductions in NOX were obtained. For example, at
4.2 percent economizer O2 and an average SOFA position of 65 percent, adjusting the SOFA
dampers to a SOFA bias of 0.11 caused an additional 12 percent reduction in NOX. The SOFA
bias, S, indicates the ratio of the lower to upper SOFA flows as defined by
c Position of SOFA Damper 19
o =
22 (1)
^Position of SOFA Damper i
1-20
The major limitation on reducing NOX at full load conditions are the high LOI levels which occur
at large SOFA openings. High LOI leads to increases in opacity and can result in frequent
opacity excursions. Because of the combined effects of SOFA position and O2 on LOI and
opacity, operating with high O2 levels made it possible to divert more air to the SOFA registers,
thus resulting in overall greater reductions in NOX with minimum opacity levels. The effects of
SOFA setting and O2 level on LOI are shown in Figure 4. With O2 in the 3.4 percent range, LOI
increases sharply as the average SOFA damper position is increased. However, with the
economizer O2 at 4.2 percent, LOI was found to be relatively insensitive to SOFA setting. This
-------
0.48 -r
0.43
35
40
45 50 55 60 65
Avg. SOFA Damper Opening (%)
75
Figure 2
Test data showing effect of
economizer O2 and SOFA setting on NOX.
0.70 ,-
0.65 - -
CO 0.60 -.
in
g 0.55
in
in
E
LJ
O
„ 0.50 -.
0.45 ..
0.40
—1 1 1 1 1 h-
10 20 30 40 50 60
Avg. SOFA Position (%)
70
80
Figure 3
Neural network curve fits of NOX
with SOFA setting and economizer O2 level.
-------
14 -r
12 -.
IT 10 -
O
8 .
-»-3.4% 02
-*-3.8% 02
-»-4.2% O2
1
1 i 1 1 1 1
10 20 30 40 50 60
Avg. SOFA Damper Position (%)
70
Figure 4
Neural network curve fits showing effects of
SOFA setting and economizer O2 level on LOl.
permitted the SOFA dampers to be opened to close to 70 percent without causing opacity
excursions.
Figure 5 shows that opacity is proportional to the LOl level. The opacity levels given in Figure 5
are average values. Although not shown hi this graph, short duration opacity spikes, often with
amplitudes of 30 percent or more, are superimposed on the average opacity. Analysis of the data
shows that the magnitude and frequency of the spikes both increase with increasing LOL The
sensitivity of precipitator performance to fly ash LOl is related to the resistivity of the fly ash.
Fly ash particles containing high carbon levels have resistivity levels which are too low to be
removed efficiently hi the precipitator.
The testing showed that it was best to allow the CCOFA settings to follow the windbox-to-
furnace pressure differential as set by CE. The CCOFA settings were found to have little effect
on opacity or LOL In addition the data showed that CCOFA settings over 40 percent have a
negligible effect on NOX. This is the normal operating range for the auxiliary air dampers. As a
result, the CCOFA dampers were set in the same positions as the auxiliary ah- dampers.
Burner and SOFA tilt angles were tested independently and in combination. With burner tilt
angles held fixed, the SOFA tilt angles were varied from -2 to +15 degrees and this affected NOX
only by about 5 percent. Raising the SOFA tilts increases the separation between the overfire air
and the main fireball. With the SOFA tilts fixed, the burner tilts were varied from -8 to +15
degrees. Figure 6 shows that increasing burner tilt angle tends to increase NOX. Finally in
another series of tests the burner and SOFA tilts were varied in unison between -8 to +8 degrees.
-------
20 -r
16 -
u
n
a
O
3
10 12
L.O.I. (%)
14
16
Figure 5
Variation of average stack opacity with LOT.
0.48 _
0.47 -.
m
E
B 0.46
in
£
(0
w
•g 0.45
til
0.44 ..
0.43
Note-.
Econ. 02 = 3 4%
SOFA Tilts = 4 deg.
Avg. SOFA Opening = 40%
-t-
-1-202468
Burner Tilt Angle (deg.)
10 12
Figure 6
Variation of NOX with burner tilt angle.
SOFA tilt settings were held fixed during this test.
-------
The effect of this on NOX was found to be small (Figure 8). However, the effect on CO was quite
dramatic with the CO increasing exponentially as the SOFA and burner tilt angles approached 14
degrees (Figure 7). As a result of these tests, the SOFA tilt angle was set to one-half the burner
tilt angle, as recommended by CE, with an upper limit on burner tilt angle of+8 degrees.
Corner damper biasing is a technique used to adjust the location of the ignition points in the hot
and cold corners. Due to boiler geometry, corners 2,4, 5 and 7 are hot corners (see Figure 1).
To move the ignition point in a hot corner further away from the burner, the fuel air damper
opening was increased and the auxiliary air damper opening decreased in that corner. The DCS
was used to simultaneously change all fuel air dampers and all auxiliary air dampers in a
particular corner of the boiler. The results show that adjusting corner bias did not greatly affect
NOX, but it was capable of reducing CO and LOI. The corner biases were set permanently for
minimum CO and safe ignition points.
Previous work on the Potomac River boiler showed that the distribution of the coal feed rates to
the different burner elevations can have a large effect on NOX. With excess mill capacity at full
unit load, the coal flow to the mill feeding the top burner elevation can be reduced, creating an
additional overfire air effect with lower NOX. At Morgantown the results showed a significant
reduction in NOX without any negative impact on steam temperatures. However, CO increased
sharply with decreasing mill bias (Figure 9).
Mill bias p is defined here as
o ' Z Z
P = (2)
1
1 .
Yl ~~ ~~ /W ~~ 777
COCU.B ^% cocilJj cooljE
z
E
£ ™coal,i
i-A
where A refers to the top burner elevation and E to the bottom. A negative p value indicates
more fuel through the lower burners. Due to the undesirable effect of mill bias on CO and
unburned carbon, it was found best not to use this as a NOX reduction technique at full load
conditions. However, at lower loads (475 MW and less) mill bias was found to be a very useful
tool for NOX reduction.
Windbox to furnace differential pressure was also tested as a method of NOX reduction. CE had
predicted that with increasing windbox to furnace pressure differential, the NOX would decrease.
The windbox to furnace DP was varied between 2.5 and 4.5 in. w.g. causing a reduction in NOX
from 0.49 to 0.42 Ib/MBtu. At full load conditions, 4.5 in. w.g. is now used as the control set
point as CE had originally recommended.
The LNCFS III is equipped with the ability to adjust the yaw of the SOFA buckets. They can be
rotated either in the same direction as the fireball rotation or in the opposite direction. From the
testing carried out on this boiler, the variations in yaw did not show any significant effect on NOX
-------
250 T
200 ..
E 150 --
a
O
O
SJ100.I-
50 .-
= Avg. SOFA Opening = 40%
, Avg. SOFA Opening = 25%
-8
-404 8 12
SOFA and Burner Tilt Angle (deg.)
16
Figure 7
Stack CO levels varied strongly when the
burner and SOFA tilt angles were changed in unison.
0.52 ,.
m
B
Ui
0.48
0.46 ..
0.44 ..
0.42
Note SOFA Tilt = 1/2 x Burner Tilt
-4-
-15 -10 -505
Burner Tilt Angle (deg.)
10
15
Figure 8
NOX versus burner tilt angle with
SOFA and burner tilt angles varied in unison.
-------
140 _
120 ..
100 -.
D.
Q.
20 ..
-0.25
-0.2 -0.15 -0.1
Mill Bias Parameter
-0.05
Figure 9
At 585 MW, CO increased sharply
as the mill loading pattern was biased downward.
or any other combustion parameter. As a result, the yaw angles were left in the counterflow
position as recommended by CE for minimum CO.
Mill fineness at Morgantown is usually maintained at 75 percent or greater through 200 mesh
and 0.5 percent or less on 50 mesh. The classifier settings were optimized prior to the parametric
test to achieve the best fineness.
Unit Heat Rate
The parametric test data show that as NOX changes with boiler control settings, additional
changes are occurring in parameters such as unburned carbon level, steam temperatures,
desuperheating spray flow rates, flue gas flow rates and fan power. Since all of these parameters
have an impact on the net unit heat rate, an analysis method which accounts for their variations
must be used. EPRI's HEATRT code (2) was used to model Morgantown Unit 2 and to calculate
the heat rate at each set of test conditions. The information required by the code on unburned
carbon level was obtained from the neural network results shown in Figure 4.
Figure 10 shows a plot of heat rate as a function of NOX for full load conditions, where all the
data from the full load tests are represented. These results show that, in general, as NOX is
reduced, heat rate increases, but at any particular level of NOX there is a substantial variation in
heat rate from one set of test conditions to another. Some of that variation is due to scatter in the
data, but a significant part of it represents real differences in heat rate with changing operating
-------
9200 -
9150 .-
s.
fl>
9100 .-
m
I
•**
c
2 9050 . -
Z
9000
•: • • &V:: . ;.
.v*v.v>" > >.
.?*-•:•«•*.
Note: Heat Rates Calculated Using the HEATRT Code
0.35 0.45 0.55 0.65
NO, Emissions (Ib/MBtu)
0.75
Figure 10
As NOX is reduced, unit heat increases. The spread
in heat rate at any one NOX level reflects the effects
of unit operating conditions on heat rate.
conditions. The heat rate data were then used to generate a neural network expressing heat rate
as a function of boiler operating conditions. The results are summarized in Figure 11, which
gives the optimal unit heat rate as a function ofNOx for curves of constant economizer O2. The
results in Figure 11 provide PEPCO with the capability of selecting the optimal set of operating
conditions at each particular NOX level. In this way, not only can the rate of NOX emissions be
traded off against unit heat rate, but in addition the heat rate penalty at a particular NOX level can
be minimized.
Waterwall Cleanliness Effects
Waterwall cleanliness has a significant impact on NOX and steam temperatures. As the boiler
slags, the temperatures in the furnace increase, causing increased thermal NOX and higher reheat
steam temperatures. Morgantown Unit 2 utilizes EPRI's Plant Monitoring Workstation for data
acquisition and for carrying out performance calculations on the boiler and turbine cycle (3).
One of the models in PMW performs on-line calculations of the cleanliness of the waterwall,
superheater, reheater and economizer sections. Heat exchanger section cleanliness is the actual
rate of heat transfer divided by a reference rate of heat transfer. The cleanliness is controlled
through activation of sootblowers.
Figure 12 shows normalized nondimensional NOX versus the waterwall cleanliness factor.
These data show that by maintaining the waterwalls in a sufficiently clean condition, NOX can be
-------
9120 _
Note: Avg. SOFA Position and Distribution Varied
9040
0.45
0.5 0.55
NOx (Ib/MBtu)
0.6
0.65
Figure 11
These results show the minimum
heat rate as a function of NOX and O2 level.
1.05 _
1 ..
X
i 0.95
•O
I
n
o
0.9 ..
0.85 ..
0.8
A A A A
A A A A
8
„ 1/20/95 „ 1/22/95 D 1/23/95 „ 1/25/95
29 30 31 32 33 34 35
WWCF (%)
Figure 12
NOX is a strong function of waterwall cleanliness
factor (WWCF). Normalized NOX is the ratio of NOX
to the NOX with WWCF = 28%.
-------
reduced by as much as 12 percent. An increase in waterwall cleanliness also results in a decrease
in hot reheat steam temperature and an increase in heat rate (Figures 13 and 14). A sootblowing
strategy has been developed for Morgantown Unit 2 to reduce NOX through sootblowing in such
a way as to minimize the impact on heat rate. This strategy relies on results of the type shown in
Figures 12 to 14.
Fuel Quality Effects
One of the results to come out of the combustion optimization testing is the importance of fuel
quality and its effect onNOx and opacity. NOX variations of the order of 0.10 Ib/MBtu have been
observed due to day-to-day changes in coal quality. Fuel quality variations are also resulting in
operating periods during which opacity excursions are more severe.
Control Algorithm Development, Implementation and Fine-Tuning
Based on the results from the parametric tests and the combustion optimization calculations, new
control algorithms have been developed for the boiler settings over the load range. These have
been programmed into the DCS to permit automated operation in a low NOX mode. An auxiliary
controller (referred to as the NOX controller) was also implemented in the DCS to give the
operators the ability to increase or decrease the SOFA settings when encountering problem
situations. For example, if the boiler experiences LOI related opacity problems, the operator can
adjust the NOX controller to lower the percentage of full SOFA opening without needing to set
each damper manually.
The NOX and heat rate levels associated with these control settings are now being evaluated
during extended low NOX tests in which the unit is dispatched as is normal PEPCO practice. The
data are being analyzed to determine how NOX, heat rate and opacity vary from day to day and
with unit load. These results will make it possible to quantify the trade-offs between NOX, heat
rate and opacity and provide PEPCO with a basis for making economic decisions on how best to
operate the unit.
Future Developments
Units 1 and 2 at Morgantown Station are almost identical in design, with Unit 1 being converted
to an LNCFS III firing system in Spring, 1995. Once that unit is started up, it will be necessary
to perform an optimization process similar to the one being completed at Morgantown Unit 2.
While it is reasonable to expect that many of the same relationships will exist between control
settings and NOX, heat rate and opacity, there may very well be significant differences in the
optimal settings, particularly at full load conditions where opacity problems are most severe.
To assist PEPCO in performing the optimization of Unit 1, Lehigh University's Energy Research
Center has been developing an intelligent software package referred to as the NOX Advisor (4).
The NOX Advisor is a PC based software package that integrates an expert system, neural
networks, and a mathematical optimization algorithm. It is intended to be used by plant
operating personnel, providing assistance to them in performing the parametric tests needed for
-------
31 32 33
WWCF (%)
34
35
Figure 13
Hot reheat steam temperature decreases
with increasing waterwall cleanliness.
200 •
150 ••
£"
5,100
» 50
3.
0-
-50
-0.5 0 0.5 1 1.5 2 2.5 3 3.5
Change in WWCF (%)
Figure 14
Because of decreasing steam temperatures, unit
heat rate increases with increasing waterwall cleanliness.
-------
combustion optimization, in evaluating the test data and in utilizing these to determine optimal
operating conditions. The NOX Advisor will also be useful in the future when the operation of
these units will need to be reoptimized periodically as the maintenance condition of the boilers
change or as design modifications are implemented. This version of the NOX Advisor has been
developed to apply to most corner-fired boilers with either conventional or low NOX concentric
firing systems and separated overfire air. Additional versions of the NOX Advisor for wall-fired
units are also being planned.
Summary
The parametric tests performed at Morgantown Unit 2 have identified the key parameters
affecting NOX and heat rate. At full load, these include average SOFA opening, vertical
distribution of SOFA damper settings, economizer O2 level, burner and SOFA tilt angles, mill
loading pattern, windbox pressure and boiler cleanliness. The high levels of unburned carbon
which occur during operation at the lowest NOX levels sometimes result in frequent opacity
excursions at Morgantown Unit 2. Waterwall cleanliness was found to have a substantial effect
on both NOX and heat rate. Finally, normal day-to-day variations in fuel quality cause relatively
large variations in NOX and also seem to affect opacity.
Neural networks were created from the test data to relate NOX, heat rate, LOI and opacity to
boiler operating conditions. The functional relationships obtained from the neural networks are
being used with optimization algorithms to identify the boiler settings which yield the minimum
values of heat rate at specific NOX and opacity levels. These results will provide PEPCO with
the tools it needs to quantify the trade-offs between emissions and performance, and identify the
most cost-effective operating conditions.
The testing and optimization methodologies used at Morgantown are being incorporated into an
intelligent software package, referred to as the NOX Advisor. This code is intended to be used by
plant personnel in optimizing the operation of their boilers.
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. HEATRT: PC Software for Performance Analysis of Pulverized Coal Boilers. Palo Alto,
CA: Electric Power Research Institute Technical Brief, October 1993. TB 101176.
3. Plant Monitoring Workstation, Volume 1 "User's Manual", Schenectady, New York,
Technologies, Inc.
4. J. Pfahler, et al., "NOX Advisor: Intelligent Software for Combustion Optimization "
Presented at 1995 EPRI/EPA Joint Symposium on Stationary Combustion NO
Control, Kansas City, MO (May 1995).
-------
MODULATING CONTROL OF LOW NOX BURNERS
B. L. Smith, Jr.
Burns & McDonnell Engineering Company
4800 E. 63rd Street
Kansas City, MO 64130
and E. D. Kramer
PSI Energy, Inc.
Gibson Generating Station
Owensville, IN 47665
ABSTRACT
Manufacturers of today's wall-fired low NOX burners (LNB) recommend a single-position
secondary air volume control for in-service burners. Air register modulation is avoided. This
paper documents a successful effort to progress beyond conventional practices of contemporary
LNB applications by continuously modulating LNB air registers and overfire air (OFA) registers.
The paper asserts a hypothesis that burner front combustion would be better controlled by
modulating burner air registers to follow burner fuel flow, while OFA responds to NOX. The
authors conducted extensive parametric tests to support the hypothesis. A control system was
subsequently developed and implemented to confirm the results.
INTRODUCTION
The PSI Energy, Inc., Gibson Generating Station is located in southwestern Indiana on the -—
Wabash River. The station is composed of five Foster Wheeler opposed-fired supercritical steam
generators and five 650-MW General Electric turbine generators. The Foster Wheeler steam
generators are rated at 4,588,000 Ibs/hr main steam at 1,005 degrees. This papers addresses NOX
reduction work conducted on Units 1 through 4 which went commercial in 1975 through 1979.
In response to the Clean Air Act Amendment of 1990 (CAAA-90), PSI implemented NOX
reduction modifications to Units 1 through 4 during the years 1991 through 1994. Highlights of
the specific technical details were presented at this conference in 1993.
NOV REDUCTION DESIGN PARAMETERS
'X
The objective of the NOX reduction effort at the Gibson Generating Station was to meet the NOX
compliance limit mandated by the CAAA-90 at the lowest evaluated cost while producing the
least impact on the PSI system. The evaluated cost included the projected impacts on boiler
-------
performance, including but not limited to, thermal efficiency, availability, and maintenance costs.
The project team endeavored to produce a NOX capability of less than .50 Ib/MMBtu with no
increase entries in unburned carbon and no loss in unit operability. The equipment was also
required to have sufficient adjustability to accommodate as many as 45 different fuels.
NOX REDUCTION EQUIPMENT, CONFIGURATIONS, AND CONTRIBUTORS
Each Foster Wheeler boiler is fitted with six Foster Wheeler MB-series mills. The chosen low
NOX burner supplier was Phoenix Combustion, using their Atlas air register and Atlas burner.
Phoenix Combustion also supplied all overfire air equipment. Energy Systems Associates (ESA)
was chosen to supply a computer model describing combustion and heat transfer processes
occurring in the furnace. A majority of the tuning and adjustment was also aided by a sample
and analysis system provided by ESA. All control and monitoring was implemented through a
Westinghouse WDPF control system. Burns & McDonnell Engineering Company provided all
engineering for evaluation, specification, installation, and tuning.
Units 1 and 2 employ a dual-level overfire air system with 16 ports per wall. Units 3 and 4
employ a single-level overfire air system with four ports per wall. The four ports per wall used
on Units 3 and 4 took advantage of existing burner openings made available by alteration of the
former Foster Wheeler interstage air system. Thus, the lower pre-modification NOX levels and
the higher post-modification NOX levels.
NOV REDUCTION RESULTS
Units 1 and 2 NOX emissions was originally 1.30 Ib/MMBtu. Through the use of the Phoenix
Combustion low NOX burners, the Phoenix dual-level overfire air system, and configuration
optimization using the ESA model, NOX emissions at full load were reduced to .32 Ib/MMBtu.
Similarly, Units 3 and 4 NOX emissions were reduced from .80 to .42 Ib/MMBtu. In both
instances, unburned carbon was increased minimally, and unit operability was not noticeably
changed. This overcompliance capability allowed PSI to target a NOX emissions level of
.45 Ib/MMBtu at all loads.
CONVENTIONAL WISDOM
Almost all low NOX burner manufacturers recommend a single position operation for secondary
air volume control. Although not always true, NOX on wall-fired units is usually at its peak at
full load. For that reason, low NOX burners are typically optimized for maximum performance
and NOX reduction at full load. Those positions remain unchanged at lower loads, resulting in
compromised, but adequate performance. Modulation of OF A is usually accomplished in
discrete steps based on unit load. There is typically no modulation of the secondary air inner or
outer registers, or of the overfire air dampers.
-------
THE GIBSON APPROACH
PSI Energy chose to approach Gibson's NOX reduction project from a perspective which more
closely integrates emissions levels with unit operability and efficiency. The Gibson approach
was to intentionally design the NOX reduction revisions for overcompliance. The
overcompliance capability would allow latitude in burner, air register, and overfire air
adjustments. Parametric relationships were then developed for implementation in a control
system to result in adequate NOX compliance with minimal impact on performance.
The advantages of this approach are numerous. The generating station remains responsive to the
CAAA-90 legislated NOX emissions limit of .50 Ib/MMBtu. This is an environmentally
responsible position, taking the attitude that legislators had properly evaluated appropriate
emissions levels and determined that anything under .50 Ib/MMBtu was acceptable. This
position is also environmentally responsible toward non-renewable natural resources. Adjusting
for a NOX emission level of .45 Ib/MMBtu instead of .32 Ib/MMBtu would probably result in
lower unburned carbon, which translates into more economical use of coal as a fuel.
Minimizing the depth of combustion staging (through OF A) also results in less reducing
atmosphere in the lower furnace, less lower furnace corrosion, less tendency for slagging in the
lower furnace, and lower unburned carbon.
WHY CHALLENGE THE CONVENTIONAL WISDOM?
It was the project team's opinion that the conventional approach of not modulating air registers is
optimized only at full load. All other combinations of loads, mills in service, fuel flow, etc. are
then a compromise resulting in less than optimum performance. Operation at lower loads
depends solely on lower windbox pressure to reduce the air flow. This, of course, also results in
less overfire air.
PARAMETRIC DATA
The following data was taken at controlled conditions for development of parametric
relationships for each furnace:
• NOX
• LOI
• Load
• Outer air register position
• Inner air register position
• Burner tip position
• OFA position
• 02
• Quantity of mills in service
-------
The original intent was to develop detailed parametric relationships for each variable. Initial
testing illustrated just how complicated and unnecessary some of those relationships were going
to be. For example, burner tip position was a relatively labor-intensive adjustment that yielded
only minor changes in emissions and LOT. The burner tip position adjustment was provided by
Phoenix Combustion as a means to adjust primary air exit velocity to account for a variety of
different fuels. We also found that the quantity of mills in service did influence emissions, but
not significantly. This is most likely due to the design of the Phoenix Atlas air register, which
required only 8 percent cooling air flow. Thus, the cooling air flow does not significantly impact
burner front stoichiometry or overfire air flow. The relationship between O2 and other
performance parameters proved to be elusive. With the overfire air registers throttled down to
70 percent or less and secondary air registers relatively open, NOX would increase as O2
increased. However, with 100 percent open overfire air registers, NOX would actually be reduced
at higher O2. The latter inverse relationship resulted from the relatively large overfire air
openings on Units 3 and 4. In effect, reducing O2 reduced windbox pressure, which had a much
larger effect on the overfire air registers than on the throttled-down burner air registers.
After initial testing, parametric development work focused on the following six parameters:
• NOx
• LOI
• load
• outer air register position
• inner air register position
• OFA position
PARAMETRIC TESTING
All testing was conducted at full load on one fuel at 3.4 percent O2 with the overfire air closed,
except for specific series of tests conducted to develop relationships identifying the effects of
those particular parameters. A 24-point sample grid was permanently installed immediately after
the economizer and tubed to a sample analysis system in the control equipment room. The
sample analysis system, provided by ESA, included two sample vacuum pumps to
simultaneously pull a sample for analysis and a sample to be subsequently analyzed. This
significantly reduced the amount of time required to purge an old sample out of the sample line
and a new sample into the sample line. The sample system also included the capability to
manifold several sample points together for one analysis. Sampling all 24 points one at a time
required approximately 30 to 45 minutes. Manifolding certain groups of samples could be
accomplished in much less. The analysis system was able to indicate NOX, CO, CO2, SO2, and
O2.
A Cegrit isokinetic ash sampler was installed on one air heater gas inlet duct. The sample jar
was emptied before each test and samples manually retrieved during or at the end of each test.
The unburned carbon was burned off in Gibson's laboratory.
-------
The Phoenix Atlas burner air registers and overfire air registers included pitot tubes for
measurement of inner air flow, outer air flow, and overfire air flow at each register. The project
team specified these devices in hopes of monitoring air flow in engineering units. However, this
proved impractical. Several unsuccessful attempts were made at relating pitot tube d/p to air
flow. At best, these air flow measurements were relative indicators. The burners and air
registers also included seven thermocouples each to monitor temperatures. All indications and
control were accomplished through the Westinghouse WDPF system. The system included
numerous graphics and a graphics printer for documentation.
Most of the parametric tests were run as rapidly as possible. If adjusting an air register or
overfire air setting, stabilization period was minimal, at approximately 15 to 20 minutes. Test
duration was usually determined by the sampling system at about 45 minutes. If adjusting O2
however, a longer stabilization period of possibly 30 to 60 minutes was allowed.
PARAMETRIC RESULTS
The following significant parametric relationships were developed:
• Figure 1 - NOX and LOI vs. outer air register position.
• Figure 2 NOX and LOI vs. inner air register position.
• Figure 3 - NOX and LOI vs. outer/inner position ratio.
• Figure 4 - NOX and LOI vs. OFA position.
• Figure 5 - NOX and LOI vs. percent 02-
• Figure 6 - Outer air vs. mill load for a resultant .45 Ib/MMBtu.
• Figure 7 - OFA vs. air flow for a resultant .45 Ib/MMBtu.
The resultant parametric relationships are illustrated on the attached Figures 1 through 7. It
should be noted that most of the NOX and LOI scales on Figures 1 through 5 exceed
.50 Ib/MMBtu. This is because most of the tests were conducted with overfire air closed in an
effort to eliminate overfire air effects on the performance of the burner itself. The data was taken
solely to identify trends, not to optimize the settings during the test
The outer air register position (Figure 1) was found to be the most influential and fortunately, the
most predictable adjustment at the burner front. Predictably, opening the outer air register
increases NOX and decreases LOI. The same figure indicates a NOX increase as O2 is increased.
The inner air register (Figure 2) adjustment exhibited a similar relationship with NOX but an
inverse relationship with LOI. Hence, the ratio of outer-to-inner air register position became an
area of focus.
Figure 3 indicates an approximation of the outer-to-inner air register position ratio. The
relationship is believed to be representative, if not accurate. It proved difficult to accommodate
the economics of the utility and the needs of the dispatcher while trying to retrieve meaningful
data. The test log did indicate however, that identical tests could be repeated after of a few days,
with repeatable results. Figure 3 indicates that best NOX performance occurs almost
-------
simultaneously with the worst LOI performance. Thus, the effort to choose a compromise setting
focused on optimizing the NOX/LOI tradeoff while finding an area without a steep slope to the
curve. The steeper the slope, the less predictable the performance results will be for any
particular setting. For the Gibson Station, an outer-to-inner ratio of 5 was chosen.
Figure 4 indicates the relationship between overfire air position, and NOX and LOI performance
parameters. The overfire air register design is simply a single-stage version of the burner outer
air register. The tests indicated that little is gained in performance beyond 60 percent open.
Figure 5 indicates the relationship between NOX and LOI, and the economizer exit O2. This data
was taken after fuel and air balancing was completed and air register settings were determined.
Thus, low LOI was achievable down to approximately 2.5 percent O2.
Development of these relationships aided PSI in identifying optimum performance settings. The
final tests focused on using those settings in a closed loop control system for NOX compliance.
The final tests operated the unit at various loads to achieve a NOX emissions value of
.45 Ib/MMBtu. The tests began at full load with overfire air full open and air registers at the
optimized settings. Air registers were then adjusted to achieve the desired NOX setting of
.45 Ib/MMBtu. Unit load was then dropped 50 MW. At the new, lower load, air registers were
adjusted according to the previously developed parametric relationships to match pulverizer fuel
flow. This adjustment raised windbox pressure, which provided additional overfire air. Any
final adjustment to NOX was made using the overfire air. This procedure was repeated from
600 MW down to 250 MW. Figures 6 and 7 indicate the results. Figures 6 and 7 were then
programmed into a NOX control system.
NOX CONTROL CONFIGURATION
The NOX control scheme is composed of two major subloops as shown in Figure 8. The NOX
subloop uses the established parametric relationships to program and position overfire air
dampers. The secondary air subloop similarly positions the air registers, and both loops are
subsequently trimmed by the NOX controller. The system uses the burner air registers to control
burner air flow and velocity, while keeping burner front stoichiometry within reasonable limits.
Maintaining adequate windbox pressure becomes an incidental issue, and is accomplished via air
register positioning only because the two relationships were developed concurrently. The OFA
position characterization curve (Figure 7) was developed as that which resulted in a NOX
emission of approximately .45 Ib/MMBtu. Thus, only minor trimming by the NOX trim
controller should be required. This approach was chosen because it would maintain NOX
compliance while maintaining unit operability and minimum LOI.
System gains are such that the NOX controller (Item 1.2) is an off-line controller with relatively
limited contribution to summer 1.3. The basic OFA positioning is accomplished by function
generator 1.5, which uses total air flow for its programming input. In this manner, the relatively
unreliable NOX analyzer (Item 1.1) can fail and the OFA dampers will still be driven
approximately to the correct position by the function generator.
-------
Total air flow is used as the program base instead of steam flow, firing rate master or fuel flow.
Total air flow incorporates the effect of excess air (O2), which can significantly affect NOX
emission and thus the need for OFA.
Manual loader 1.4 provides an adjustable OFA set point for the trim. It also artificially adjusts
measured total air flow in an attempt to approximate a revised OFA program to match the revised
OFA set point.
Outer air registers are positioned by function generator 2.14. Optimum outer-to-inner air register
position ratio has been determined to be approximately 5:1. Thus, the inner air position is
programmed to be 20 percent of the outer air position program. This arrangement maintains
proper outer/inner air relationship and also maintains similarly matched air/fuel ratio.
The configuration also recognizes that there may be conditions under which this system cannot
maintain NOX at set point because of equipment condition changes or weather conditions. The
NOX trim control loop was incorporated to compare measured NOX to set point and adjust both
overfire air position and air register position, thus affecting windbox pressure and required OFA
position. Multiplier 3.11 and high/low limiter 3.12 work together to allow a variable maximum
trim influence. Multiplier 3.11 ratios the error to establish a very small trim limitation at low
loads and a larger trim limitation at high loads. For example, if measured NOX is higher than set
point, then the NOX trim subloop would attempt to close down on the air registers within limits,
which would raise windbox pressure. The increased windbox pressure would force additional
overfire air and deeper furnace combustion staging. Simultaneously the NOX trim would try to
increase the OFA damper opening.
CONCLUSION
The examples in this paper were derived from Gibson Station Unit No. 3. Similar configurations
have been implemented in Units 1, 2 and 4. As expected, Units 1 and 2, with their modified
overfire air system, required minor revisions. This control system has operated at Gibson for
approximately 2 years in Unit 3 with very good results. The unit maintains NOX at
approximately .45 Ib/MMBtu at all loads and during most upset conditions and ramp rates. The
project proved that modulation of air registers is feasible, responsible and effective.
-------
NOx & LOI vs Outer Air Register Position
25 30 35
Outer Air Register Position (% Open)
40
Figure No. 1
-------
NOx & LOI vs Inner Air Register Position
10
15 20 25
Inner Air Register Position (% Open)
35
Figure No. 2
-------
NOx & LOI vs Outer/Inner Air Ratio
r 3.-4%-G2
i. 3D%.Q2
3456
Outer/Inner Register Ratio
8
Figure No. 3
-------
NOx & LOI vs OFA Position (% Open)
20
40 60
OFA Position (% Open)
100
Figure No. 4
-------
NOx & LOI vs % Oxygen
2.6 2.8 3
Oxygen (%)
3.2
3.4
- 5.4
5.2
3.6
Figure No. 5
-------
45
40
30
w 25
O
a.
£ 20
<
£ 15
Outer Air Position vs Fuel Flow for .45 Ib/MMBtu
20
40 60 80
FUEL FLOW (kpph)
100
120
140
Figure No. 6
-------
100
90
80
3f 70
60
c
o
w 50
LL_
O
«
30
20
10
OFA Position vs Air Flow for .45 Ib/MMBtu
Z
0 10 20 30 40 50 60 70 80 90 100
Air Flow (%)
Figure No. 7
-------
UNIT 3 DUCT NOx
TOTAL AIR FLOW
MILL 3"X* FUEL FLOW
(NOxj
Rl
Ji
1.4
A
\y
NOx TRIM
TO NORTH OFA PORT
HAND/AUTO STATIONS
& DAMPER DRIVES
TO SOUTH OFA PORT
HAND/AUTO STATIONS
k DAMPER DRIVES
TO OTHER BURNERS
ON THIS MILL
o
D-
2.14
^
T]
:T POINT
.45)
A
K
u
/
>
1.2
1_,
1.3
u
•< 0
1 i
i i
^-''•5 AIRFLOW
FUNCTION
(%)
15
0 60 UO
FUEL aOW (KPPH)
FUNCTION 2.14
" —x i
2.17 '—' \
UGHTOFF (20%)
2.19
COOLING (0%)
CONTROL
DRIVE
\2.21
-iG = 1/5
x 12.20
CONTROL
DRIVE
\2.22
OUTER AIR REGISTER
INNER AIR REGISTER
Figure No. 8
-------
ADVANCED INSTRUMENTATION FOR THE B&W LOW EMISSION BOILER
S.A. Johnson, C.L. Senior and M. Khesin
PSI Environmental Instruments Corp.
20 New England Business Center
Andover,MA 01810
and
A. Zadiraka
Babcock & Wilcox Co.
Abstract
The next generation of coal-fired boilers targeted for both the domestic and overseas markets
must meet and maintain very low emission levels in order to be competitive. Cost of electricity
and generation cycle efficiency will also be market drivers.
To meet this challenge and to maximize the use of tired and true coal-fired steam cycle
technology, Babcock & Wilcox (B&W) and the U.S. Department of Energy are partnering to
demonstrate break-through emission control technologies capable of less than 0.1 Ib NOj/MBtu
fired (without the added financial burden of backend NOX cleanup).
The key to the B&W approach is the use of advanced diagnostics to measure and control key
process parameters during deeply staged combustion. Per burner air/fuel ratio control is achieved
by advanced processing of flame scanner signals combined with an advanced burner design. The
resulting computer-controlled system is analogous to automobile engine diagnostic systems
introduced in recent years.
This paper will cover the conceptual design and preliminary tests of LEBS NOX control
diagnostics components in test facilities and existing boilers. Implications of the preliminary
results on future testing and commercial designs will also be discussed.
Introduction
Shortly after the end of this decade, it is anticipated that new generating plants will be required in
the U.S. to meet a growing demand for electricity and replacement of aging plants approaching
the end of their useful service life. These new plants will need to be extremely clean, efficient and
economical. For coal to be the fuel of choice for this new generation of plants, they will have to
address the concerns over acid rain, air toxics, global climate changes, ozone depletion and solid
waste disposal.
-------
At the start of this decade, the coal-fired power plants of the future were expected to be based on
Integrated Gasification Combined cycle (IGCC) and Pressurized Fluidized Bed Combustion
(PFBC) technologies. However, as shown by Figures 1 and 2, advances in emission control
technology and significant improvements in the steam cycle design clearly indicate that pulverized
control technology can be cost and performance competitive with these IGCC and PFBC
technologies for at least the next decade. Driven by emission regulations in Europe and fuel costs
in Japan, advanced pulverized coal fired power plants using ultra-supercritical steam cycles
designed for very low emissions and net plant efficiencies exceeding 40% (HHV) are currently
under construction.
c
o
'
"E
LU
X
O
C\4
O
36%
39%
40%
PC w/FGD + SCR
PFBC
IGCC
C-9387
Figure 1. Current emissions from coal-based technologies.
c
0)
0>
Q.
CD
CD
co
>,
O
IGCC
PFBC
1200/1100/1100
PC Cycle
8-
7-
6-
5-
4-
3-
2-
1 -
o-
t
, Future
Demonstration Plants
Cur
Techr
(
i
•ent <
ology ^^4
**r
fl 100/1 050/1 050
< i 1050/1050/1050
^__^--» 1000/1025/1050
''ToOO/l 025/1 050
' 1050/1050
1050/1000
1000/1000
^^^^^^
L •r<1000/1000
I i i i i 1 1 1 1 1
1500 2400 3600 4500 5000
Main Steam Pressure PSI
Figure 2. Net plant efficiency - coal technologies.
C-9388
-------
Combustion 2000
The U.S. Department of Energy (DOE)'s Pittsburgh Energy Technology Center in late 1990
began a research and development initiative to address the design issues facing new and
replacement coal-fired power plants. This program, named Combustion 2000, involves two
stages of commercialization, the Low Emission Boiler System (LEBS) intended to address the
nearer term and the High Performance Power system (HIPPS) for the longer term.
B&W under contract with DOE, with subcontracts to PSI Environmental Instruments Inc. (PSI)
and Raytheon Engineers & Constructors (RE&C), is one of the three industry teams working on
the LEBS portion of the program. The LEBS program is divided in four phases for execution
over a span of seven years.
Phase I, which began in 1993 and ended in 1994, involved system analysis, R&D planning and
component definition to establish a preliminary commercial generating unit design. Phase n,
which is currently in progress, provided for pilot and subsystem testing to confirm and refine the
conceptual designs developed in Phase I. In Phase m, a design will be developed for the
construction of a Proof-of-Concept (POC) demonstration facility as well as revising the
commercial generating unit design based on the findings from Phase n. Phase IV involves the
construction and operation of the POC facility to prove the readiness of the technology for
commercial application.
The original goals of the LEBS program were no more than 0.2 Ib/MBtu NOX, 0.2 Ib/MBtu SO2
and 0.015 Ib/MBtu particulate with at least 38% net plant efficiency (HHV) without increasing
cost relative to a conventional NSPS plant. As these goals have been refined over the course of
the Phase I activities, they evolved to the current goals of no more than 0.1 Ib/MBtu NOX,
0.1 Ib/MBtu SO2 and 0.015 Ib/MBtu particulate with net plant efficiency approaching 42%
(HHV) while reducing solid wastes and complying with anticipated air toxics regulations at or
below the costs of a conventional NSPS plant.
The B&W design to meet the program's performance goals is a 4500 PSI 1100/1100/1100 F dry
bottom boiler integrating advanced low NOX combustion with deep staging, advanced pulverizers
providing ultra-fine pulverized coal and Limestone Injection Dry Scrubbing (LIDS). A simplified
gas side arrangement of this design is shown in Figure 3.
B&W's activities in Phase I were structured around the four major subsystems; NOX Control, SOX
Control, Boiler and Balance of Plant (BOP). As a result of the Phase I work, the importance of
integrating the controls and sensors activity across the four subsystems in order to achieve the
performance goals was recognized and a fifth subsystem team for Controls and Sensors was
added to the project.
Low NOX Combustion
Low NOX production over the load range, including during load changes, is achieved by the
precise regulation of the combustion conditions. Fundamental to the ability to maintain the
-------
Limestone
Feeder
Stack
Transport
Blower
Solids to I
», , Landfill J
Pulverizer
d)
Recycle
Solids
Tank
Detention
Maker
C-9389
Figure 3. LEBS preliminary plant arrangement.
optimum combustion conditions is the ability to accurately measure and control and individual
burner air and fuel flows to each stage of combustion.
Sensors and final control elements are being developed to provide this capability for each burner
in a multi-burner system. Coupled with advanced burner design and air staging, on-line control of
the individual burner stoichiometries should achieve the minimum possible NOX emissions from
the combustion process balanced against unburned carbon losses.
In addition to incorporating these sensors and final elements, advanced control philosophies are
being developed to operate the advanced low NOX, LIDS and boiler in an integrated manor to
achieve minimum emissions over the load range. These concepts will be tested at the subsystem
level prior to integrating them in the POC design. Since multi-burner operation will not be
available on the subsystem test facility, the subsystem testing will utilize a dynamic model
incorporating combustion dynamics to evaluate the concepts to be included in the POC design.
The LEBS low-NOx combustion system must create a fuel-rich burner zone throughout the load
range of the unit. Initial NOX production is limited by delayed fuel-air mixing inherent in B&W
commercial burner designs. Air-fuel ratios must be held within very close tolerances to assure
flame stability (and scanability) throughout the load range.
NOX reduction is further enhanced by interstage heat removal prior to overfire air injection. The
LEBS boiler provides enough waterwall surface to achieve the desired temperatures at the point
of overfire air addition. The appropriate amount of heat removal is maintained by measuring and
controlling first-stage exit temperature via a smart sootblower system.
The key to making the LEBS NOX control system work on a day-to-day basis will be the ability to
provide on-line balancing of air-fuel ratios. Tests performed at B&W indicate that such balancing
-------
may be feasible even at very low stoichiometric ratios using the Spectrum Diagnostix
SpectraTune™ flame analyzer. The sections that follow describe SpectraTune™ and testing
performed on B&W burners.
SpectraTune™ Concept
Spectrum Diagnostix and PSI Technologies are working with B&W to develop the NOX control
subsystem for the Low Emissions Boiler System (LEBS). The goal of the NOX control system is
to achieve levels of NOX emissions below 0.1 Ib/MBtu and to guarantee NOX below 0.2 Ib/MBtu
throughout the operating range of the unit. This is to be accomplished through advancements in
burner design and staged combustion.
To achieve the NOX reduction goal will require tight control of the burner air to fuel ratio over the
entire load range. It is also necessary to provide even distribution of heat input over the furnace
cross section to avoid deviations in stoichiometry and temperature which result in excessive NOX
emissions or poor carbon burnout. There are several control methods to achieve the above
objectives. One of the most effective is the control of individual burners (or groups of burners)
when the air-to-fuel ratio is maintained and controlled on a burner-by-burner basis. Industry
estimates of the amount of burner imbalance that can be tolerated without increasing either NOX
emissions or unburned carbon range from below 5% to well in excess of 10%. The lower the
NOx the better the burner balancing required.
Balancing burners requires measurement of air and coal flow to individual burners. Though
primary and secondary airflows can be measured with standard flow devices (e.g. venturi meter or
pitot grids), there are no proven systems for on-line monitoring of coal flow rates to individual
burners. An alternative approach involves direct diagnostics of individual burner flames. PSI has
developed and evaluated a new method for monitoring operation of individual burners, based on
computer processing of flame signals from the existing flame monitoring hardware.
Most industrial and utility boilers are presently equipped with flame monitoring devices (flame
scanners) for individual burners. The sole function of these scanners is to determine the presence
of individual burner flames and to achieve reliable flame discrimination (flame detection) between
individual burner flames and the furnace's background fireball for the Burner Management
Systems (BMS). The primary sensor output signal generated in a flame scanner is usually
considered to have two components: intensity and temporal fluctuating frequency (often called
DC and AC). One of them or a combination of both is used for flame detection. The fluctuating
component can be processed via Fast Fourier Transform (FFT) and represented in the form of an
amplitude versus frequency plot which is known as the flame signature. Over the years, numerous
data have been collected to study flame signature characteristics and how they depend on fuel
type, burner design and operating conditions.2'3 These data show that the flame signal, and
particularly its fluctuating (AC) component, is highly sensitive to changes in combustion
conditions or disturbances in the controlled area.
Correlation of flame "flicker" signals with flame quality and emissions can be understood as
follows. In coal flames, the combustion process is dominated by the mixing rate of coal and air
(the chemical kinetics are much faster). Each burner flame consists of a multitude of combustion
recirculation cycles (eddies, swirls, or loops) of various sizes inside and around the flame which
-------
are responsible for mixing the air and the fuel in turbulent diffusion flames. In the tested burner,
the flame consists of an internal recirculation zone of fresh and partially combusted fuel, sur-
rounded by a swirling flow of air. Combustion occurs primarily in the turbulent eddies between
these two flows. These eddies contribute to generating the flame flicker at various frequencies as
the result of turbulent mixing through eddy formation at the edges of the coal and air jet. The
shorter loops and smaller eddies, generate the higher frequencies, and vice versa. The distribution
pattern of fluctuational energy, can be monitored by measuring the average amplitudes in different
selected frequency bands, their derivatives and ratios. The eddies can be correlated with the
flicker frequencies in the following way. Every time a turbulent eddy occurs, it mixes coal (or
pyrolysis products) with air. The amount of fuel and air mixed is controlled by the size of the
eddy. Since combustion kinetics are fast compared to these turbulent mixing times, the fuel and
air are combusted instantly. Therefore, a larger eddy should give a larger emission intensity. The
signature (flicker) plots show the frequency of the eddy versus the size of the eddy. Based on
turbulent flow analysis, we can determine the dominant frequency range(s) of the flame frequency
spectrum (or flicker plot) which are the most sensitive to changes associated with fuel/air mixing
and air-to-fuel ratio.
Figure 4 illustrates the concept. In step 1, the temporal signal is transformed to the frequency
domain. Although each frequency spectrum is a repeatable signature of the flame, certain parts of
the spectrum are more sensitive to changes in key burner operating parameters. Therefore,
SpectraTune™ correlates flame quality factors derived from the spectrum to per-burner air-fuel
ratio and swirl vane settings - parameters that can be controlled by the LEBS operators from a
digital control system located in the control room.
03
C
g>
CO
MAVWWWMMM/W
Time -»
Flame Radiation
CO
Frequency ->
CD
Temporal Frequency Spectrum ^ w
CO O
CLCC
t/3
Control Variable ->
Correlation Between Frequency Spectrum
and Flame Characteristics (NOX, AFR, etc.)
C-8623
Figure 4. SpectraTune™: concept overview.
As part of the LEBS program, PSI has conducted tests to determine the feasibility of using flame
signature scanning for combustion control. Because the burner selected for the LEBS is an
aerodynamically air staged burner, these tests were conducted on a boiler equipped with B&W
-------
DRB-XCL burners. Alabama Power and Plant Gaston agreed to host these tests at Unit 3 which
had been retrofitted with DRB-XCL burners and is well instrumented. Subsequent testing was
performed on a single-burner pilot scale test facility at B&W's Alliance, Ohio research center.
The Alliance tests provided well-controlled staged combustion conditions under which the DRB-
XCL burner could be evaluated. The following sections describe the results of those tests.
Plant Gaston Results
Plant Gaston Unit #3 is a B&W designed boiler rated at 270 MW. It is fired by 18 B&W DRB-
XCL burners arranged three over three on the front and real wall. The unit also contains two
division walls front to back in the burner zone, isolating each burner from its neighbor on the
same elevation.
For these tests, dedicated flame scanners were mounted on existing viewports located on the
backplate of each burner looking into the ignition zone through the inner secondary air annulus.
Test variables included load, inner swirl vane setting, and outer register slide damper position.
SpectraTune™ was responsive to changes in each of these variables.
Figure 5 shows the SpectraTune™ response to damper position. The shape of this response
curve is intuitively consistent with expected changes in secondary air flow as damper position
changes. Similarly, inner swirl vanes could be used to change the flame signatures of adjacent
burners at Gaston. Figure 6 illustrates how changing spin vane setting on burner #2 makes its
signal nearly identical to burner #2.
CD
(f)
C
o
Q.
en
CD
tr
CD
C
6.5 --
6 --
5.5 --
5 --
co 4.5 --
o
CD
Q.
CO
4 --
3.5 --
30% 40% 50% 60% 70%
Secondary Air Damper Position
80%
90%
C-8774
Figure 5. Full scale (270 MW) test results - response to burner adjustment.
The Gaston tests proved that SpectraTune™ can provide repeatable signals that can be used to
locate out-of-time burners. The next logical question is how can that signal be used to optimize
and control the operation of each individual burner in a burner array?
The SpectraTune™ signal will be a function of many variables affecting burner flame, such as air-
to-fuel ratio, swirl(s) intensity, flame stability, etc. Is it possible that various combinations of hese
-------
100
Unbalanced Burner Pair
Balanced Burner Pair
Burner 1
CO
40 60
Frequency (Hz)
40 60
Frequency (Hz)
100
C-6712
Figure 6. Effect of inner swirl on flame signatures.
factors may produce the same result thus preventing us from finding the only optimal solution, or
the only optimal combination of these factors? In theory, in a multivariable system with many
contributing factors there might be a situation when the same flame spectrum corresponds to
different combinations of variables. However, in practical terms, it is probably not important.
Burner adjustment based on frequency spectra can be used in the fine-trim mode, after preliminary
coarse adjustments of fuel and air flows. In addition, each particular type of burner would have
no more than one or two variable(s) to adjust for achieving the fine-tuned status. However, to
provide a definite answer to this question we have to test a single well instrumented burner and to
look at all flame parameters, including O2, CO, NOX and unburned carbon.
Conclusions
Pilot-scale frequency spectra are still being analyzed by B&W and SDX for application to the
LEBS NOX Control Subsystem. To date, evidence suggests that SpectraTune™ signals are
repeatable and responsive to even small changes in air-fuel ratio and turbulence. Pilot-scale data
also show that per-burner NOX optimization may be possible on-line via automatic trimming of the
secondary air slide damper or inner secondary air swirl vanes in order to hold a constant flame
quality (Q) value.
Additional SpectraTune™ tests are planned this summer on the SBS equipped with an advanced
staged burner. These data will be used to relate SpectraTune™ response to very low
stoichiometric ratios and much lower NOX emissions. Closer correlation with air-fuel ratio will be
required in order to operate the burner close to its limits of flame instability.
If successful on the pilot-scale, SpectraTune™ will be proven on a multi-burner array to be tested
on B&W's 100 MBtu/h CEDF facility also located in Alliance, Ohio. These tests will prove the
feasibility of the commercial product, and allow design and pricing to be established. Integration
of SpectraTune™ with the rest of the LEBS control system will also be accomplished.
-------
Acknowledgments
The authors express thanks to the U.S. Department of Energy's Pittsburgh Energy Technology
Center for supporting the B&W LEBS team's efforts.
References
1. McDonald, D.K., Madden, D.A., Rodgers, L.W., Sivy, J.L., "Component Development in
Support of B&W s Advanced Coal-Fired Low-Emission Boiler System" 1994
International Joint Power Generation Conference, Phoenix, AZ, Oct 1994.
2. Khesin, M.J. and Johnson, S.A., "Combustion Control: New Environmental Dimension",
presented at the American power conference, Chicago, IL, April 1993.
3. Makansi, J., "Reducing NOX Emissions in today's Power Plants, Power Magazine,
New York, NY, May 1993.
-------
WALL-FIRED COMBUSTION DEMONSTRATION PROJECT -
ADVANCED DIGITAL CONTROL / OPTIMIZATION PHASE UPDATE
John N. Sorge
J. Scott Allison
Southern Company Services, Inc.
P. O. Box 2625
Birmingham, Alabama 35202
J. G. Noblett
Radian Corporation
P. O. Box 201088
Austin, Texas 78720-1088
Scott M. Smouse
U. S. Department of Energy
Pittsburgh Energy Technology Center
P. O. Box 10940
Pittsburgh, Pennsylvania 15236-0940
EPRI/EPA 1995 Joint Symposium on Stationary NOx Control
Kansas City, Missouri
May 16-19,1995
Abstract
This paper discusses the technical progress of a U. S. Department of Energy
Innovative Clean Coal Technology project demonstrating advanced wall-fired
combustion techniques for the reduction of nitrogen oxide (NOx) emissions from
coal-fired boilers. The primary objective of the demonstration is to determine the
long-term NOx reduction performance of advanced overfire air (AOFA), low NOx
burners (LNB), and advanced digital control/optimization methodologies applied in
a stepwise fashion to a 500 MW boiler. The focus of this paper is on the design and
results to date from the advanced digital control/optimization phase of the project.
-------
Introduction
This paper discusses the technical progress of one of the U. S. Department of Energy's
Innovative Clean Coal Technology (ICCT) projects demonstrating advanced combustion
techniques for the reduction of nitrogen oxide (NOx) emissions from wall-fired boilers. This
demonstration is being conducted on Georgia Power Company's Plant Hammond Unit 4, a 500
MW, pre-NSPS (New Source Performance Standards), wall-fired boiler. Plant Hammond is
located near Rome, Georgia, northwest of Atlanta.
This project is being managed by Southern Company Services, Inc. (SCS) on behalf of the
project co-funders: The Southern Company, the U. S. Department of Energy (DOE), and the
Electric Power Research Institute (EPPJ). In addition to SCS, Southern includes the five electric
operating companies: Alabama Power, Georgia Power, Gulf Power, Mississippi Power, and
Savannah Electric and Power. SCS provides engineering and research services to the Southern
electric system. The ICCT program is a jointly funded effort between DOE and industry to move
the most promising advanced coal-based technologies to the commercial marketplace. The goal
of ICCT projects is the demonstration of commercially feasible, advanced coal-based
technologies that have already reached the "proof-of-concept" stage. The ICCT projects are
jointly funded endeavors between the government and the private sector in which the industrial
participant contributes at least 50 percent of the total project cost. The DOE is participating
through the Office of Clean Coal Technology at the Pittsburgh Energy Technology Center
(PETC).
The primary objective of the demonstration is to determine the long-term NOx reduction
performance of advanced overfire air (AOFA), low NOx burners (LNB), and advanced digital
control/optimization methodologies applied in a stepwise fashion to a 500 MW boiler. Short-
term tests of each technology are also being performed to provide engineering information about
emissions and performance trends [1,2,3,4].
Following a brief unit and technology review, this paper focuses on the design and results to date
from the advanced digital control/optimization phase of the project
Unit and Technology Review
Georgia Power Company's Plant Hammond Unit 4 is a Foster Wheeler Energy Corporation
(FWEC) opposed wall-fired boiler, rated at 500 MW gross, with design steam conditions of 2500
psig and 1000/1000°F superheat/reheat temperatures, respectively. The unit was placed into
commercial operation on December 14, 1970. Prior to the LNB retrofit in 1991, six FWEC
Planetary Roller and Table type mills provided pulverized eastern bituminous coal (12,900
Btu/lb, 33% VM, 53% FC, 72% C, 1.7% S, 1.4% N, 10% ash) to 24 pre-NSPS, Intervane
burners. The burners are arranged in a matrix of 12 burners (4W x 3H) on opposing walls with
each mill supplying coal to four burners per elevation (Figure 1).
During a spring 1991 unit outage, the Intervane burners were replaced with FWEC Controlled
Flow/Split Flame (CF/SF) burners. In the CF/SF burner, secondary combustion air is divided
between inner and outer flow cylinders. A sliding sleeve damper regulates the total secondary air
-------
Airflow
Measurement
AOFA Flow
Control Dampers
Guillotine
Damper
Overfire
Air Ports
Burners
Partition Plates and Secondary Air Duct
Pressure Control Dampers
Secondary Air Duct
Figure 1
Hammond Unit 4 Furnace Layout
flow entering the burner and is used to balance the burner air flow distribution. An adjustable
outer register assembly divides the burner's secondary air into two concentric paths and also
imparts some swirl to the air streams. The secondary air that traverses the inner path, flows
across an adjustable inner register assembly that, by providing a variable pressure drop,
apportions the flow between the inner and outer flow paths. The inner register also controls the
degree of additional swirl imparted to the coal/air mixture in the near throat region. The outer air
flow enters the furnace axially, providing the remaining air necessary to complete combustion.
An axially movable inner sleeve tip provides a means for varying the primary air velocity while
maintaining a constant primary flow. The split flame nozzle segregates the coal/air mixture into
four concentrated streams, each of which forms an individual flame when entering the furnace.
This segregation minimizes mixing between the coal and the primary air, assisting in the staged
combustion process.
As part of this demonstration project, the unit was also retrofit with an Advanced Overfire Air
(AOFA) system. The FWEC design diverts air from the secondary air ductwork and
incorporates four flow control dampers at the corners of the overfire air windbox and four
overfire air ports on both the front and rear furnace walls. Due to budgetary and physical
constraints, FWEC designed an eight port AOFA system more suitable to the project and unit
than the twelve port system originally proposed.
The Unit 4 boiler was designed for pressurized furnace operation but was converted to balanced
draft operation in 1977. The unit is equipped with a coldside ESP and utilizes two regenerative
secondary air preheaters and two regenerative primary air heaters. During the course of the
ICCT demonstration, the unit was retrofitted with six Babcock & Wilcox MPS 75 mills (two
each during the spring 1991, spring 1992, and fall 1993 outages).
-------
Review of Prior Testing
Baseline, AGFA, LNB, and LNB+AOFA test phases have been completed (Table 1). Short-term
and long-term baseline testing was conducted in an "as-found" condition from November 1989
through March 1990. Following retrofit of the AGFA system during a four-week outage in
spring 1990, the AGFA configuration was tested from August 1990 through March 1991. The
FWEC CF/SF low NOx burners were then installed during a seven week outage starting on
March 8, 1991 and continuing to May 5, 1991. Following optimization of the LNBs and
ancillary combustion equipment by FWEC personnel, LNB testing was commenced during July
1991 and continued until January 1992. Testing in the LNB+AOFA configuration was
completed during August 1993. During both the LNB and LNB+AOFA, there were significant
increases (when compared to baseline) in precipitator fly ash loading and gas flow rate and also,
increases in fly ash LOI which adversely impacted stack particulate emissions and forced the unit
to be load limited [5].
Table 1. Project Schedule
Phase
0
1
2
3A
3B
4
5
Description
Pre-Award Negotiations
Baseline Characterization
Advanced Overfire Air Retrofit (AGFA) & Characterization
Low NOx Burner Retrofit (LNB) & Characterization
LNB+AOFA Characterization
Digital Controls/Optimization Retrofit & Characterization
Final Reporting and Disposition
Date
8/89 - 4/90
4/90-3/91
3/91 1/92
1/92 - 8/93
9/93 - 8/95
9/95 - 12/95
Status
Completed
Completed
Completed
Completed
In Progress
Later
A summary of the baseline, AOFA, LNB, and LNB+AOFA long-term NOx emissions data for
Hammond Unit 4 is shown in Figure 2. Baseline testing was performed in an "as-found"
condition. For the AOFA, LNB, and LNB+AOFA test phases, following optimization of the unit
by FWEC personnel, the unit was operated according to FWEC instructions provided in the
design manuals. As shown, the AOFA, LNBs, and LNB+AOFA provide a long-term, full load,
NOx reduction of 24, 48, and 68 percent, respectively. The load-weighted average of NOx
emissions reductions was 14, 48, and 63 percent, respectively, for AOFA, LNBs, and
LNB+AOFA test phases. Although the LNB plus AOFA NOx level represents a 67 percent
reduction from baseline levels, a substantial portion of the incremental change in NOx emissions
between the LNB and LNB+AOFA configurations is the result of operational changes and is not
the result of the AOFA system [6].
The time-weighted average of NOx emissions for the baseline, AOFA, LNB, LNB+AOFA test
phases are shown in Table 2. Since NOx emissions are generally dependent on unit load, the
NOx values shown in this table are influenced by the load dispatch of the unit during the
corresponding test frame. Also shown in this table are the 30 day and annual achievable
emission limits (AEL) as determined during these test periods. The 30-day rolling average AEL
is defined as the value that will be exceeded, on average, no more than one time per ten years.
For the annual average, a compliance level of 95 percent was used in the calculation.
-------
1.4
1.2
I 1
^ 0.8
z 0.6
0.4
Wall-Fired Emission Limit
0.2100
LNB+AOFA
+Other
200
300 400
Load, MW
500
Figure 2
Long-Term NOx Emissions vs. Load Characteristic
600
Table 2
Long-Term NOx Emissions
Unit Configuration -»
Parameter •!•
Number of Daily Avg. Values
Load (MW)
NOx Emissions (Ib/MBtu)
O2 Level (percent at stack)
NOx 30 Day AEL (Ib/MBtu)
NOx Annual AEL (Ib MBtu)
Baseline
Mean
52
407
1.12
5.8
1.24
1.13
RSD,%
9.4
9.5
11.7
-
AOFA
Mean
86
386
0.92
7.3
1.03
0.93
RSD,%
17.9
8.6
12.6
-
LNB
Mean
94
305
0.53
8.4
0.64
0.55
RSD,%
17.7
13.7
7.7
LNB+AOFA
Mean
63
293
0.41
8.73
0.51
0.42
RSD,%
-
23.9
12.9
16.3
-------
Phase 4 - Advanced Controls / Optimization
As a result of the installations of the lowNOx combustion systems at Hammond 4, combustion
optimization has become significantly more difficult than prior to these retrofits. This added
difficulty is a result of several factors including:
• Heightened concern and awareness of combustion conditions as a result of the passage of the
1990 Amendments to the Clean Air Act,
• Increased sensitivity of combustion conditions to process adjustments, and
• Additional complexity and more independent tuning adjustments.
The objective of this scope addition to the project at Plant Hammond is to evaluate and
demonstrate the effectiveness of advance digital control/optimization methodologies as applied
to the NOx abatement technologies installed at this site (LNB and AGFA). The major task for
this project addition include: (1) design and installation of a distributed digital control system
(DCS), (2) instrumentation upgrades, (3) advanced controls/optimization design and
implementation, and (4) characterization of the unit both before and after activation of the
advanced strategies. Major milestones for this phase of the Wall-Fired Project are shown in
Table 3.
Table 3
Advanced Controls / Optimization Major Activities
Milestone
Status
Digital control system design, configuration, and installation
Digital control system startup
Instrumentation upgrades
Advanced controls/optimization design
Characterization of the unit prior to activation of advanced strategies
Characterization of the unit following activation of advanced strategies
Completed
Completed
Completed
In Progress
Scheduled 8/94 - 4/95
Scheduled 5/95 - 895
Combustion optimization is the procedure by which NOx reduction, combustion performance,
and safety are balanced to achieve or approach a predetermined goal. In most instances, the
goals are defined in terms of performance inequality constraints mutually agreed to by the burner
vendor and the utility such as:
• NOx - Reduce to below guarantee value and/or compliance limit.
• Fly ash loss-on-ignition (LOI) Hold below guarantee value and/or state imposed state
utilization limit.
• Boiler performance Maintain above the guarantee value.
These goals may be defined for one or more operating conditions. Only when all constraint
goals are clearly met, will further NOx optimization be performed. Due the complexity of the
-------
combustion process, optimization is formidable unless the goals are lax. Combustion
optimization for the low NOx burners with advanced overfire air is considerably more difficult
than that required for setup of turbulent burners alone. This added difficulty is a result of the
increase in the number of adjustments and sensitivity of these burners to operating conditions
(Table 4).
Table 4
Combustion Tuning Control Points at Hammond 4
Pre-LNB+AOFA Retrofit
Burners
Sleeve registers (24)
Secondary air
Windbox balancing dampers
Mill Biasing
Post-LNB+AOFA Retrofit
Burners
Sleeve registers (24)
Tip Positions (24)
Inner registers (24)
Outer registers (24)
Advanced overfire air
Can-in-can dampers (8)
Flow control dampers (4)
Secondary air
Windbox balancing dampers
Boundary air
Mill Biasing
Generally, optimization requires that the unit be taken out of economic dispatch and run at full-
load for much of the optimization period. After balancing the secondary air flows, the burner
optimization process is accomplished by adjusting the inner registers, outer registers, slide
nozzles, and sleeve dampers while monitoring NOX, 02, and CO at the economizer outlet. When
possible, burner adjustments of the same class (the classes being inner register, outer register,
slide nozzle, or sleeve damper) are moved in unison to a nominal, optimized position. Only
when flow and/or combustion irregularities dictate, are individual dampers adjusted from this
nominal position. The adjustments to the sleeve dampers, inner registers, outer registers, and tip
position are made during the burner optimization process and thereafter remain fixed unless
changes in plant operation or equipment condition dictate further adjustments. The normal
FWEC practice is to supply actuators on the sleeve dampers only. Optimization is performed for
full-load operation and performance is checked at lower loads. Because of the constraints of the
equipment and optimization methodology, the combustion process can be optimized for one
operating condition (load, fuel condition, air distribution, etc.) and therefore is sub-optimal for all
others.
Unlike SO2 emissions which are primarily a function of the sulfur content of the fuel, NOx
emissions are highly dependent on a number of parameters. Nitrogen oxides (NOx) are formed
in combustion processes through the thermal fixation of atmospheric nitrogen in the combustion
air producing "thermal NOx" and the conversion of chemically bound nitrogen hi the fuel
producing "fuel NOx". NOx emissions can theoretically be reduced by lowering: (1) the primary
flame zone O2 level, (2) the time of exposure at high temperatures, (3) the combustion intensity,
-------
and (4) primary flame zone residence time. NOx emission rates are strongly influenced by the
apportionment of the air to the burners and AGFA system.
An example of the interdependencies and conflicting goals which must be considered can be seen
in Figure 3. As shown, as excess air (or equivalently, excess oxygen) decreases, NOx decreases
while LOI increases. High LOI values are indicative of poor combustion and therefore poor
boiler performance. Also, on units which sell their fly ash (Hammond 4 does not at this time), an
increase in fly ash LOI can change the fly ash from a marketable commodity to an undesirable
byproduct. A decision must be made as to what is the optimum operating condition based on
economic and environmental considerations. Similar compromises must also be made when
optimizing boiler efficiency. In this case, the optimum operating condition is clear as long as the
performance index is defined as boiler efficiency and other parameters (such as NOx emissions)
are not considered. Conflicting objectives such as these have been observed on Hammond
Unit 4. As shown in Figure 4, the NOx production rate is an increasing function of the excess
oxygen level while fly ash LOI is a decreasing function. This data was collected during the
short-term low NOx burner tests.
In addition to variations with excess oxygen levels and load, NOX emissions also vary
significantly during long-term operation and it is evident that a number of uncontrolled and
unidentified variables greatly influence NOX production. These influencing variables are
believed to be mill operating conditions (primary air temperatures, air/fuel ratios, flows, grind,
and moisture), secondary air non-uniformity (air register settings, forced draft fan bias, and
windbox pressure differential), coal variability, etc. As shown in Figure 5, NOx long-term
variability at Hammond Unit 4 for the LNB plus AOF A test phase was approximately
0.07 Ib/MBtu at full load, increasing to 0.3 Ib/MBtu at minimum load. As can be seen, there are
significant differences in the NOx emission characteristics although no changes in burner
adjustments or operating procedures were made during this time frame. A potential goal of any
on-line optimization program installed at this site would be to drive NOx emissions down to the
lower percentile and beyond.
o
z
NOx
Optimum Excess Air
Excess Air
Excess Air
Figure 3
Typical Tradeoffs in Boiler Optimization
-------
0.6
S 0.55
.0
UJ
c
o
.
LU
0.45
0.4
Increase O2
5678
Fly Ash LOI, Percent
10
Figure 4
NOx and LOI vs. Excess Oxygen (NOx vs. LOI Tests)
NOx, Ib/MBtu
0.8
0.6
0.4
0.2
95th Percentile
Phase 3B - LNB+AOFA
Complete Data Set
<
4
4
- .
. •
fll
w
— 1
t —t
-L 1 - Mean - -
5th Percentile
100 200 300 400 500 600
Load, MW
Figure 5
Long-Term NOx Emissions During LNB Plus AGFA Test Phase
-------
Generic NOx Control Intelligent System
The methodology to be demonstrated at Hammond is the Generic NOx Control Intelligent
System (GNOCIS) whose development is being funded by a consortium consisting of the Electric
Power Research Institute, PowerGen, The Southern Company, U.K. Department of Trade and
Industry, and U.S. Department of Energy [7]. The objective of the GNOCIS project is to develop
an on-line enhancement to existing digital control systems that will result in reduced NOx
emissions, while meeting other operational constraints on the unit (principally heat rate and other
regulated emissions). The main contractors for the development of GNOCIS are PowerGen and
Southern Company Services. Commercializers for North America are SCS and Radian
Corporation. In its role as commercializer, Radian is already deeply involved in the
demonstrations in the U.S. PowerGen and one other as yet unnamed organization will be the
commericalizer in Europe.
The core of the system is a neural-network model of the NOx generation characteristics of a
boiler, that reflects both short-term and longer-term shifts in boiler emission characteristics. The
software applies an optimizing procedure to identify the best set points for the plant. The
recommended set points are conveyed to the plant operators via the DCS or, at the plants
discretion, the set points can be implemented automatically without operator intervention. The
software incorporates sensor validation techniques and is able to operate during plant transients
(i.e. load ramping, fuel disturbances, and others). Figure 6 shows the major elements of
GNOCIS.
Figure 6
Major Elements of GNOCIS
Following an initial feasibility study in which several promising methodologies were evaluated,
a software package from Pavilion Technologies was selected to fulfill the "core" technology role
in GNOCIS, i.e. to form the basis of the process and control models necessary to perform on-line
optimization. The models are created from data collected from long-term, normal operation,
augmented as necessary by short-term testing.
GNOCIS methodology is now undergoing testing at PowerGen's Kingsnorth Unit 1 (a 500 MW
tangentially-fired unit with an ICL Level 3 Low NOx Concentric Firing System) and Alabama
-------
Power's Gaston Unit 4 (a 250 MW B&W unit with B&W XCL low NOx burners), the results of
which are being reported elsewhere [8].
Customization of GNOCIS at Hammond is now underway. The major activities associated with
the GNOCIS installation at Hammond 4 are:
• Digital Control System Design, Configuration and Installation
• Instrumentation Upgrades
• Pre-Installation Testing
• Model and Optimization Strategy Development
• Post-Installation Testing
These elements are discussed in the following paragraphs.
Digital Control System Design, Configuration, and Installation
An integral part of Phase 4 of the project was the design and installation of a digital control
system (DCS) to be the host of the advanced control/optimization strategies being developed.
Prior to the installation of this DCS, Hammond Unit 4 utilized a pneumatic boiler control system
which would be unsuitable for a closed-loop implementation of GNOCIS, therefore it was
necessary to upgrade this system. SCS Engineering and Georgia Power had overall
responsibility for the following major activities associated with this task:
• Preliminary engineering,
• Procurement,
• Detail engineering,
• Digital control system configuration, and
• Installation and checkout.
In total, the digital control system was configured for 2352 input/output points consisting of 572
analog inputs, 116 analog outputs, 1032 digital inputs, and 632 digital outputs with the balance
being allocated spares. This system is designed such that the I/O is fully distributed and operator
interaction with the digital control system is almost exclusively through the operator display —
there are no benchboard mounted manual/auto stations or switches.
An overview of the digital control system is shown in Figure 7. Based on a competitive
evaluation, a Foxboro I/A system was selected for installation. The milestones in the design,
installation, and startup of the Hammond Unit 4 digital control system are shown in Table 5.
As part of this project, the control room was modified to accept the new Unit 4 digital control
system. Pre-existing Unit 4 benchboards were removed and replaced with a CRT based control
panel. In addition to the upgrades to Unit 4, Georgia Power has upgraded Unit 3 and is also
considering upgrading the digital control systems on Units 1 and 2. Digital control system and
control room modifications for Units 1, 2, and 3 are not a part of the Wall-Fired Project.
-------
The Unit 4 DCS has been interfaced with the other DCS's at the site. Unit 3, Unit 4, and
Electrical DCS systems are connected through a dual-redundant IEEE 802.3 (Ethernet) local area
network (LAN). Through this LAN, the three DCSs are able to share process information and
graphics. If for some reason either the A or B LAN fails, all DCSs can maintain normal
operation. An additional benefit of these LANs are the ability to share costly resources such as
engineering consoles, historical drives, etc. In addition to the inter-DCS network, the Unit 4
DCS (and the others also), are connected through a router to the plant's token-ring PC
engineering and administrative LAN and the corporate wide area network (WAN) (Figure 8).
The latter enables remote access of process data and facilitate software maintenance. A Sun
Sparcstation 5, hosting the GNOCIS software, is connected to this network. The router isolates
the DCS from the plant LAN and company WAN.
Engineering Workstations
Ash System
Controls
Precipitator
Controls
Processing
&
I/O
Streaming Tape
Drive
Figure 7
Hammond Unit 4 DCS Overview
Table 5
DCS Installation Milestones
Date
Milestone
June 1992
August 1992
February 1993
April 1993
June 1993
June 1993
January 1994
February 1994
May 1994
June 1994
Begin preliminary engineering
Issue request for proposals for digital control system
Foxboro I/A system received at SCS
Issue purchase order to Foxboro
Start detail engineering
Begin configuration
Configuration complete
Start checkout
Foxboro I/A system shipped to Plant Hammond for installation
Installation complete
Unit Startup
-------
Southern
Company
WAN
Other Systems Networks
Figure 8
Hammond Plant Network
Instrumentation Upgrades
As a result of data collection requirements of prior phases of the test program at Hammond 4, the
unit had been well instrumented. Of special interest for the GNOCIS installation at this site was
the availability of a continuous emissions monitor (CEM) and on-line carbon-in-ash (CIA)
measurement.
Hammond Unit 4 is equipped with two CEMs. One CEM, installed during 1989 on project
startup, is an extractive system with NOx, SOx, CO, total hydrocarbons, and oxygen analyzers.
The compliance CEM, installed during 1993, is a dilution-extraction system with NOx, SOx,
CO2, and flue gas flow. The outputs from both these systems are input to the DCS and are
available for model training purposes.
Although numerous fly ash samples (both isokinetic and ESP hopper) have been collected at this
site as part of the overall project test program, it was felt that in order to address the need to (1)
obtain sufficient carbon-in-ash training data and (2) verify GNOCIS performance, an on-line
carbon-in-ash monitor was required. Two systems, a Clyde-Sturdevant Ltd. SEKAM and a
CAMRAC Company CAM, have been installed for this purpose. The SEKAM unit will sample
from two locations at the economizer outlet while the CAM unit will sample from a single
location at the precipitator inlet (Figure 9). The CAM system will also be used to perform
mapping studies at the ESP inlet. Consideration is also being given to the evaluation of other
on-line CIA monitors. In addition to their use in GNOCIS, an evaluation of these systems will
be made as to their accuracy, repeatability, reliability and maintenance.
-------
Flue Gas Flue Gas
From Economizer From Economizer
Boiler house
wVjvV
o o I
SEKAM
2 probes (fixec
0 0
Air Heater
« h°°
M/w\
O O 4 O Q
Air Heater
\ Gas Analysis
System Probe
Locations
Platform -
-300F
-300F
CAM
1 probe (movable)
Ports where
ESP Method 17
are conducted
Figure 9
Carbon-in-Ash Analyzers Layout
Time-of-response of the instrument is also a very important consideration for the current
application since it is necessary to form a correlation between carbon-in-ash and current
operating parameters. If the time response of the instrument is long compared to the process
time response, this correlation is generally much more difficult. Factors affecting the time
response are: (1) transport lag of the fly ash to measurement cell, (2) time required to collect
requisite sample size, (3) sample mixing in measurement chamber, (4) measurement time, and
(5) averaging performed in the analyzer. Depending on fly ash loading (and hence load) near the
collection point(s), the response can range from 5 minutes to more than 2 hours.
Pre-lnstallation Testing
One prerequisite of a GNOCIS installation is the availability of substantial and high quality
process data from the host site. At Hammond, this need was amplified in that a goal of the
project is to comprehensively test the performance of GNOCIS. Short-term diagnostic testing
was conducted during August 1994 and March 1995, and more comprehensive performance
testing was undertaken in November 1994. The primary objectives of these tests were to:
• Re-characterize the unit following a number of combustion modifications during the most
recent outage,
• Establish relationships between control variables and measured variables,
• Establish the impact of off-design operational settings, and
• Augment the database used for training of GNOCIS models.
-------
Based on these tests, NOx emissions were found to be approximately 0.43 Ib/MBtu - slightly
higher that 0.40 Ib/MBtu observed during Phase 3B testing ~ with corresponding fly ash loss-on-
ignition levels near 8 percent. This latter value is similar to what had been observed during
Phase 3B testing.
Long-term collection of data to be used for training for this phase has been in progress since
summer 1994. Although this represents a large volume of information and satisfactory GNOCIS
models could potentially be developed using this long-term, normal operating data only, it was
felt that by obtaining process information in off-design conditions, the combustion models
would be more robust. The need to conduct additional testing depends on the variability of data
contained hi the training set. Unfortunately, although having many advantages otherwise, digital
control systems tend to create highly correlated data in which it is difficult to ascertain emission
sensitivities to a number of potential control parameters. One example where this is likely is in
mill loadings. Typically, when in service and in automatic, all mills are constrained to equal fuel
flows and therefore, useless there is some variability, models can not be created based on process
data alone, that can estimate the impact of individual mill flows on important combustion
properties such as NOx emissions. The short-term test suite was planned to artificially create the
off-design operating conditions that may not be seen during normal unit operation.
Model and Optimization Strategy Development
Retrieval of process data from the digital control system is now hi progress and initial modeling
efforts have begun. The first step in the design process is the development of suitable predictive
models. An example of the results from a typical non-linear predictive model of NOx and
carbon-in-ash are shown in Figures 10 and 11, respectively. In this example, the inputs to the
network were coal flows, excess O2, and overfire air flows. The data collected from the DCS and
used hi training was five minute averages. Steps which could have been taken to improve the
prediction capabilities include the addition of more process data and time averaging. Due to the
long response time of the on-line carbon-in-ash devices, especially at reduced loads, the
modeling of this parameter is much more difficult than modeling either NOx or CO emissions.
Although predictive models are useful in a number of circumstances, what is required of
GNOCIS are control models. Considerations hi control model development are sensitivities of
model outputs (such as NOx) to available inputs, and control points readily changeable by the
operator or through the DCS.
Design of the control strategy for Hammond 4 is in the early stages. As a starting point, it is
planned to used the control variables as shown in Table 6. The control variables in the first tier
will be implemented initially, and, if successful, additional variables from the subsequent tiers
will be considered if their inclusion improves the performance of the system significantly.
Software hooks have been designed into the DCS to facilitate the incorporation of these signals
into the control logic.
-------
0.6
0.5 .
= 0.4 .
CD
S 0.3
Z 0.2
0.1
Wall-Fired Project Actual
March 1995
Predicted
March 18, 1995 7:40 - March 21, 1995 18:40
0
5000 5200 5400 5600
Row
5800 6000
Figure 10
NOx Predictive Model
16
14 .
12
| 10 |
o
v 8
Q.
< 6
O
4
2
0
Wall-Rred Project
March 1995
Predicted
Actual
March 18, 1995 7:40 - March 21, 1995 18:40
5000 5200
5400 5600
Row
5800 6000
Figure 11
Carbon-in-Ash Predictive Model
-------
Table 6
Planned Control Variables
Parameter of Interest
First Tier
Overall Furnace Air / Fuel Ratio
Overall Furnace Staging
AGFA Distribution
Mill Biasing
Mills-in-Service
Second Tier
AOFA Distribution
Furnace Secondary Air Distribution
Third Tier
Furnace Secondary Air Distribution
Controlled Parameter
Excess O2 Bias
AOFA Flow (4)
AOFA Flow (4)
Mill Coal Flow (6)
Mill Coal Flow (6)
AOFA Can Dampers (8)
Burner Sleeve Damper s by Banks (8)
Burner Sleeve Damper s (24)
Advisory
Mode
Open-Loop
Y
Y
Y
Y
Y
Y
Y
Y
Supervisory
Mode
Close-Loop
Y
Y
Y
Y
Advise
Y
Y
Y
Using the combustion models thus developed, predictions can be made as to the benefits that can
be obtained by the application of GNOCIS. For example, as shown in Figure 12, predicted CIA
levels near 5 percent were achieved using optimized control setpoints (fuel biasing, excess O2,
overfire air flow rates). The corresponding recommended excess O2 levels are shown in
Figures 13 and 14. Although the recommended setpoints may not be feasible for actual long-
term operation, this scenario does at least lend hope that opportunities may be present for
significant CIA reductions. Again, these are predicted results, and although encouraging, they
need to be substantiated with thorough plant testing.
5000
5200
5400 5600
Row
5800
6000
Figure 12
Control Model - Predicted CIA Output (Preliminary)
-------
6 .
c
0) _.
o 5
0
Q.
CM
O
(A
in
CD
o
x
4 .
3 -
~ 2 .
1
0
5000
Recommended
Actual
5200
5400
5600
5800
6000
Row
Figure 13
Control Model Recommended Right Excess O2 (Preliminary)
c
0)
o
^
01
Q.
CM"
O
in
in
o>
o
x
5000
5200
5400 5600
Row
5800
6000
Figure 14
Control Model Recommended Left Excess O2 (Preliminary)
-------
Post-Installation Testing
Testing of GNOCIS is planned for summer 1995 in both advisory and supervisory modes.
Hammond Unit 4 is currently in the midst of a scheduled outage. Following resumption of unit
operation, now scheduled for May 15,1995, additional data will be collected from the unit to
verify combustion models and re-train if necessary. Testing of GNOCIS in the open-loop
advisory mode is scheduled to commence during June 1995. If this testing is successful, testing
of GNOCIS in the closed-loop, supervisory mode will follow. The test program is scheduled for
completion during August 1995.
Summary
Work is still in progress at Hammond Unit 4. A summary of the current (as of April 15, 1995)
status and plans for this site are as follows:
• Long-term data set collected and it is now being filtered to remove bad and irrelevant data,
• Predictive and control model development is in progress,
• The GNOCIS software will be installed on the Sun Sparcstation 5 during the current outage
and interfaced with the DCS,
• Operator displays will be integrated with the operator consoles, and
• Open- and closed-loop testing of GNOCIS at Hammond 4 is scheduled for summer 1995.
Acknowledgments
The authors wish to gratefully acknowledge the support and dedication of the following
personnel: Mr. Ernie Padgett and Mr. W. C. Dunaway, Georgia Power Company, and Mr. Mike
Nelson and R. J. Kelly, Southern Company Services, for their coordination of the design and
retrofit efforts, and Mr. Jose Perez, Instrumentation Specialist from Spectrum Systems, Inc. We
would also like to recognize the following companies for their outstanding testing and data
analysis efforts: Energy Technology Consultants, Inc., Flame Refractories, Inc., Innovative
Combustion Technologies, Southern Research Institute, and W. S. Pitts Consulting. Finally, the
support from Mr. Art Baldwin, DOE Project Manager, Jeff Stallings, EPRI Project Manager, and
Mark Perakis, EPRI Project Manager, is greatly appreciated.
REFERENCES
1. 500 MW Demonstration of Advanced Wall-Fired Combustion Techniques for the
Reduction of Nitrogen Oxide Emissions from Wall-Fired Boilers - Phase I Baseline
Tests. Southern Company Services, Birmingham, AL: 1991.
-------
2. 500 MW Demonstration of Advanced Wall-Fired Combustion Techniques for the
Reduction of Nitrogen Oxide Emissions from Wall-Fired Boilers - Phase 2 Over fire Air
Tests. Southern Company Services, Birmingham, AL: 1992.
3. 500 MW Demonstration of Advanced Wall-Fired Combustion Techniques for the
Reduction of Nitrogen Oxide Emissions from Wall-Fired Boilers - Phase 3 A Low NOx
Burner Tests. Southern Company Services, Birmingham, AL: 1994.
4 500 MW Demonstration of Advanced Wall-Fired Combustion Techniques for the
Reduction of Nitrogen Oxide Emissions from Wall-Fired Boilers - Phase 3B Low NOx
Burner plus Advanced Overfire Air Tests. Southern Company Services, Birmingham, AL:
1995.
5. 500 MW Demonstration of Advanced Wall-Fired Combustion Techniques for the
Reduction of Nitrogen Oxide (NOx) Emissions from Coal Fired Boilers - Technical
Progress Report - Third Quarter 1991. Southern Company Services Inc., Birmingham,
AL: 1992.
6. Sorge, J., Wilson, S., "500 MW Demonstration of Advanced Wall-Fired Combustion
Techniques for the Reduction of Nitrogen Oxide (NOx) Emissions from Coal Fired
Boilers," Third Annual Clean Coal Technology Conference, September 6-8, 1984,
Chicago, Illinois.
7. Holmes, R., Squires, R., Sorge, J., Chakraborty, R., Mcllvried, T., "Progress Report on
the Development of a Generic NOx Control Intelligent System (GNOCIS)," EPRI 1994
Workshop on NOx Controls for Utility Boilers, May 11-13, 1994, Scottsdale, Arizona.
8 Holmes, R., Mayes, I., Irons, R., Sorge, J. N., Stallings, J. W., "GNOCIS An Update of
the Generic NOx Control Intelligent System," EPRI/EPA 1995 Joint Symposium on
Stationary NOx Control, May 15-19, 1995, Kansas City, Missiouri.
-------
MAINTAINING LOW-NOX EMISSIONS AFTER YOUR BURNER RETROFIT
S.A. Johnson, Spectrum Diagnostix, Inc.
M.J. Khesin, PSI Technologies, a Division of Physical Sciences Inc.
20 New England Business Center
Andover,MA01810
Abstract
To achieve NOX reduction requirements, utilities are retrofitting their coal-fired boilers with either
combustion and post-combustion NOX control systems. For either system, controlling the gas
composition, temperature, and residence time where NOX is reduced determines the success of the
retrofit. Creating the optimum conditions for NOX reduction over a wide range of operating
conditions is difficult due to normal changes in coal and air feeding systems, burners, and furnace
conditions. While such changes used to create only minor operating problems, with low-NOx
systems the result can be NOX emissions outside the compliance limits or costly operational
impacts (e.g. high fly ash carbon or ammonia salt deposition on air heater surfaces).
To maintain low-NOx emission requires an advance in burner and furnace monitoring and control
technology. This paper discusses the results of demonstration tests at New England Power and
Alabama Power of two advanced monitoring techniques: flame signature scanning for burner
balancing and tuning, and continuous gas temperature monitoring for controlling furnace exit and
convective pass temperatures. The combination of these advanced monitoring techniques with
automatic or manual control systems will enable utilities to maintain low-NOx emissions when
using low- NOX burners or SNCR.
Introduction
The year 1995 is here. This is the year that the Clean Air Act Amendments of 1990 (CAAA) were
to have mandated coal-fired utilities to implement strict SO2 and NOX controls on existing boilers
built prior to 1971. Even though delays have occurred, nearly 300 boilers must soon implement
some form of NOX control to comply with Title IV (Acid Rain) of the CAAA, and an estimated
400 more may be subject to further NOX control as a result of Title I (Ozone Non-attainment)
starting in 1996. State regulations are driving NOX retrofits in some areas not affected by the
CAAA.
Typical levels of NOX reduction expected at these units range from 30 to 50%. Most affected
utilities have installed low-NOx burners, overfire airports, selective non-catalytic reduction
systems, or some combination of the three in at least one of their affected units in order to gain
operating experience prior to 1996. The results of these early retrofits have been similar: each
NOX control system can be operated to achieve the target NOX reductions. However, maintaining
-------
low-NOx emissions without incurring increases in unburned carbon, slagging, corrosion, ammonia
emissions, carbon monoxide, or excess air has been a considerable challenge. Conventional
burner or boiler control systems do not provide enough information to optimize NOX emissions in
real time.
Just what are the problems that tend to compromise low-NOx system operation? The main
culprits are:
1. imbalanced air and fuel flow to individual burners
2. changes in boiler thermal performance that affect the gas temperature profile from the
boiler to the stack.
Air-fuel ratio can vary by more than 20% from burner to burner. If just one burner is operating
substoichiometrically, for example, it may produce most of the unburned carbon (UBC) and
thousands of ppm CO. Raising total airflow will reduce CO and UBC, but NOX will increase
because all the other burners will now be firing leaner than they were designed.
Competing requirements for NOX and unburned carbon control have economic effects on boiler
operation. Carbon in the fly ash represents an efficiency loss to the boiler system. Therefore,
additional coal must be burned to produce each kW of power. More importantly, many utility
companies are attempting to sell their fly ash for use as filler, aggregate, or additive for cement or
concrete. Any more than 3 to 5% carbon in the fly ash can adversely affect the strength, porosity,
or color of the end product. Unsalable fly ash is usually hauled away and disposed in landfills at
costs ranging from $8 to $25 per ton.
For a nominal 500 MW boiler, an increase in heat rate of 0.1% would cost about $50K/yr in
additional fuel, but if 20% of its ash is unsalable, the cost would be about $500K/yr. When
burners are tuned, the costs associated with unburned carbon are minimized.
Day-to-day operation (load changes, start-ups, mills out of service, oil gun erosion, and normal
equipment wear) inevitably leads to a de-tuned burner system. Air and fuel imbalances as great as
20% are caused by differences in length or flow restrictions among coal pipes, or by non-uniform
flow paths through the windbox. The symptoms of burner imbalance can be high fly ash carbon,
stack opacity, or back corona in the electrostatic precipitator. Flame standoff or instability can
also occur. The usual response to the problem is to raise burner airflows (unless the unit is fan-
limited at full load). If the problem is being caused by just a few burners starved for air, the result
will be higher NOX and lower boiler efficiency trading off against acceptable burnout. It is no
wonder that low-NOx burner performance degrades after the initial acceptance tests.
The need to maintain environmental performance and efficiency day in and day out over the entire
load range has led to the development of advanced controls systems. These control systems must
maintain thermochemical environment in the furnace to achieve optimal performance. Advanced
control systems to maintain low-NOx conditions include individual control of air and fuel flow to
burners, monitoring of burner aerodynamics, and furnace temperature profile control.
-------
Fuel and Airflow Control
Currently there are no commercial sensors available to measure directly the primary air and coal
flow to each individual burner. Coal and primary air is measured at the inlet of each pulverizer,
but the distribution can be quite different in each coal pipe leading to the furnace. Secondary air
is measured going into the windbox, but the air distribution is affected by minor differences in
flow restrictions within each burner. If staged combustion is used, additional imbalances are
possible between secondary and staging air, while substoichiometric conditions at the burners
increase the level of air/fuel ratio control required to maintain ignition stability for all burners.
Depending on the combustion system, minor changes in air/fuel ratio can cause individual burners
to smoke, have excessively long flames, or produce excess NOX. The easiest adjustment is to
raise total excess air.
Burner secondary airflow can be measured by any one of several commercial devices. Multi-
probe pilot grids are offered by a few burner manufacturers and instrument companies. These
probes are generally used during start-up for balancing flow among burners. Pitot probes
equipped with low velocity head AP transmitters (0 to 0.1 in. of water full span) are available for
increased measurement accuracy required for monitoring continuous burner operation. The
airflow to each burner can be balanced by adjusting slide dampers for radial inlet burners or disks
for axial inlet burners. More accurate flow control can be achieved with butterfly dampers in
individual supply lines for each burner and flow path as is being practiced in Europe.
Eventually, it may also be feasible to control the coal flow to each burner. EPRI is currently
sponsoring a tailored collaboration to develop inexpensive burner pipe coal flow measurement
technologies. Once this measurement is available, however, online adjustment of coal flow would
require flow control devices. Any valves or diverters may require frequent replacement or
recalibration due to wear caused by the abrasiveness of the coal particles. A more practical
approach is to measure the coal flow and adjust the airflow to achieve the desired stoichiometry.
Burner Monitoring
Information from existing flame scanners can be used today to monitor and control low-NOx
combustion systems (with or without monitoring air or fuel flows). PSI Technologies has gained
experience processing burner flame scanner signals from several boilers and correlating the
resulting flame quality factors with burner performance.
Figure 1 shows a typical fluctuational component (flame "flicker") of a flame scanner signal in the
time domain, and the flame frequency power spectrum in the frequency domain which results from
the transformation of the scanner signal. The x-axis of the power spectrum is the frequency which
makes up the fluctuation in the scanner signal, and the y-axis is the radiation power associated
with each frequency. The radiation power associated with each frequency is a measure of
chemical heat released in turbulent eddies of a particular size. Therefore, the flame signature is an
interpretation of the turbulent eddy size or frequency distribution in the ignition zone of the flame.
-------
0.3
O
<
0.25
0.01
O
UJ
10-5
Time (s)
2.56 0
Frequency (Hz)
100.15
C-6910
Figure 1
Transformation of Raw Scanner Signal to a Flame Frequency Power Spectrum
The turbulent eddy size or frequency distribution is a function of characteristic burner dimensions,
air velocities and swirl. Burners of a single design with identical fuel and airflows and swirl will
have identical flame signatures over significant averaging periods. This makes the flame signature
a tool for balancing burner aerodynamics on a multi-burner furnace.
The flame signatures have been shown in field tests to be related to the rate of air-fuel mixing and
air-fuel ratio, both of which are critical to NOX and carbon burnout. For example, Figure 2 shows
the effect of swirl on flame frequency power spectra. Changing the swirl affects primarily high
frequency turbulence. Figure 3 compares the flame frequency power spectra of two burners
before and after burner balancing. Each mill group can be balanced in a matter of minutes.
D2-F (reduced inner sw)
D2-F (increased inner sw)
Frequency
C-6909
Figure 2
The Effect of Secondary Air Swirl on Flame Frequency Power Spectrum
-------
100
'in
§ 80
60
g> 40
Q)
L_
55
75 20
c
D)
CO
-i 1 1 r
Unbalanced Burner Pair
Burner 1
Burner 2
20 40 60 80
Frequency (Hz)
100 S
§100
£>
S 80
1
3. 60
g> 40
£
w 20
75
§> 0
Balanced Burner Pair
/Burner 1
Burner 2
20 40 60 80
Frequency (Hz)
100
C-6712
Figure 3
Example: Burner Balancing
A schematic of the SpectraTune™ system being developed by PSI is shown in Figure 4. The raw
flame output signal from each sensor is transformed into a flame frequency power spectrum and
brought to a CRT display in a dual-channel signal analyzer. The frequency spectra are further
processed to derive flame quality parameters (Q) that can then be related to burner operation
(air/fuel ratio, spin vane settings) and eventually to NOX. Flame quality for each burner can be
compared against a master burner (or a time when burner operation was previously optimized)
and adjustments can be made to airflows, register settings, or swirl vanes to assure that all flames
exhibit similar radiational characteristics. Figure 5 illustrates the procedure. Figure 6 shows data
from a single pilot-scale burner illustrating the correlation that can be achieved with NOX.
-b
\d
u
r , ..r
i i
^~"
i
stments —
i - r .
1 '
I I L. _.
T~ r
J 1- \-
Jr- -1 + -
T ^ T~
1 1 ¥'
1 1 1
i -i r
1 1 i i
— I i i i
"i r ' _i
|6 6 O O O 0 ,
7i n n n
f N
-v~
-H
v y
• —
—
_^
\.
Flame
Scanners
^ \
^ >To Burner
>. 1 Mflnagpmffnt
/ ^ ' System
^— Electric Isolators
Connecting Cables
- SpectraTune™ System
-Terminal Strip
Graphical Output
Figure 4
SpectraTune™ Schematic
-------
c
D)
CO
Time —>
Flame Radiation
X
15
c
g>
CD
Frequency ->
Temporal Frequency Spectrum
0)
C
Correlation Between Frequency Spectrum
and Flame Characteristics (NOX, AFR, etc.)
C-8623
Figure 5
SpectraTune™ Signal Processing Procedure
CD
CO
C
O
Q.
CO
a>
CL
CD
C
13
CO
"o
(D
CL
CO
5 --
4 -
3 -
1 -
High Load
Reduced Load
60
80
100 120 140
Relative NOX (% of Baseline)
160
180
C-8776
Figure 6
SpectraTune™ Response to NOX With a 100 MBtu/h Burner
-------
Thermal Environment Control
In many coal-fired boilers, a portion of the NOX is formed by reaction of nitrogen and oxygen in
the combustion air at temperatures greater than 2732°F (1500°C). Compact boiler designs,
turbulent combustion systems, and fuels that cause heavy lower furnace slagging all contribute to
high peak flame temperatures and significant thermal NOX production.
After a low-NOx burner retrofit, abnormally long flames can also cause an increase in FEGT,
often accompanied by high carbon content of the flyash. Measured FEGT can then be used as
another signal that some burner adjustments are required to shorten the flames.
Staged combustion or reburning is most effective for NOX reduction when the thermal
environment is controlled. Ideally, the fuel-rich flame zone should be kept as hot as possible to
drive fuel-bound nitrogen into the gas phase where it can react to form molecular nitrogen (N2).
However, overfire air should be added far enough from the flame and in large enough quantities
to quench reaction temperatures and prevent reformation of NOX. In pilot studies, it was shown
that some second stage NOX reduction is possible if the gas temperature at the point of overtire
air addition is below 2500°F (1371 °C)2'3.
Furnace vertical temperature profile can be controlled using continuous temperature monitors to
trigger smart sootblowing. Spectrum Diagnostix, Inc. (SDx) employs an optical instrument called
SpectraTemp™ (distributed by Babcock & Wilcox under the trade name GasTemp™), but
acoustic or infrared pyrometers are also available. The advantages of SpectraTemp™ include:
• accuracy
• line-of-sight (not single point) measurements
• no field adjustment required
• no interferences
• wide temperature range (650 to 2900 °F)
• does not require a strong signal
• can be used during startup
Table 1 describes why SpectraTemp™ is more accurate than a conventional IR pyrometer.
Whereas the IR device takes its signal from strong emission lines produced by CO2 or water
vapor, SpectraTemp™ senses much weaker signals produced by ash particles in the visible part
of the spectrum. Ash particles and cold waterwall radiation interfere with the IR signal making
the IR instrument unreliable for measuring true average FEGT.
SpectraTemp™, on the other hand, uses advanced signal processing and an on-board computer to
turn the weak particle radiation into an accurate, line-of-sight temperature reading. Accuracy is
maintained over a wide temperature range by switching detectors when the computer determines
that a temperature out of the range of acceptable accuracy is measured. Thus, SpectraTemp™
can be used to measure and control furnace exit gas temperature (FEGT) throughout the boiler
load range, and even during boiler start-up.
-------
Table 1
Reasons for SpectraTemp™ Advantages
SpectraTemp
,TM
IR Pyrometers
Operates at Shorter Wavelengths
— Knee of the black body curve where
T does not change with emissivity
— No interferences; senses particle
radiation
— Sees all the way across the boiler
— Provides average temperature across
the boiler
Multiple Detectors
—- Each detector accurate within 500 F
window
— Automatically switches detectors
when accuracy decreases
Advanced Signal Processing
— Contains on-board computer
— Amplifies weak signals
Wide Applicability Without Calibration
— Start up
— Load following
— Base load
Can be Used for Process Control
— 4 to 20 mA signal can be used in
sootblower controls (see Diamond
Power)
Backed by Experienced Service People at
Diamond Power, B&W, SDx
Operates at Longer Wavelengths
— Knob on the instrument to adjust
"emissivity"
— Reflective ash causes problems
— Field of view limited to less than 5 ft
— Provides wall region temperature only
One or Two Detectors
— Inaccurate over all temperatures
Minimal Signal Processing
— Emission from CO2, H2O gives a
strong signal so sophisticated
electronics are not required
— Strong signal * accuracy
Not Reliable if Operating Conditions
Change
— Each condition requires recalibration
Not Recommended for Process Control
— Recalibration is impractical as
conditions change
Usually You are on Your Own to Make it
Work
SpectraTemp™ has been used in conjunction with Diamond Power sootblower controls to main-
tain the lowest possible NOX emissions at several boilers.4 Boiler A is a 626 MW supercritical
unit built in the late 1960's. This unit produces about 1.5 Ib/MBtu of NOX at full load with all
mills in service and a furnace exit temperature above 2700°F. This unit has a very high plan area
heat release rate (>2.3 MBtu/h-ft2) and collects furnace waterwall ash deposits even when burning
premium coals. The clean furnace FEGT is usually below 2600°F.
-------
Figure 7 shows NOX emissions for various mill configurations. It can be seen that by removing
mills from service (upper rows or the middle column) and maintaining airflow to the idle burners,
a staged combustion condition is created. As a result, NOV was reduced to about 1.0 Ib/MBtu.
A
Further NOX reduction could be realized by modifying the boiler sootblower schedule to maintain
a constant FEGT. Table 2 shows several sootblower schedules tried at this unit. Figure 8
compares the effects of the new sootblower schedule A with the old one. When FEGT could be
maintained less than 2600°F, the NOX emission was reduced to about 0.8 Ib/MBtu. The total
NOX reduction from 5-mill operation was 45%. About 20% of this decrease is attributable to
reduced furnace temperatures achieved with better sootblowing.
1
\Z(JV
1100
1000
900
800
700
600
500
-
-
-
-
~
B 85% Load O 5 Mills
D Mill 5 Out A 90% Load
0 Mill 4 Out A Mill 2 Out
^-^^ ?'
I
O £-''
-
^r^
i
I.DJ
1.50
1.37
1.23
1.10
0.97
0.83
0.70
2500
2600
2700
FEGT(°F)
Figure 7
Effect of FEGT on NOV Emissions
Table 2
Sootblower Schedules
2800
C-5590
Sootblowers
Operated
5,9,13,19,23,32
8,1,15,27
4,10,12,18,22,30
3,7,16,25
11,14,17,20,27
2,6,21,26,28,31
Schedule A
02:00, 14:00
04:00, 16:00
06:00, 18:00
08:00, 20:00
10:00, 22:00
12:00,24:00
Time of Operation
Schedule B Schedule C
03:00, 21:00 01:00, 07:00, 13:00, 19:00
06:00, 24:00 02:00, 08:00, 14:00, 20:00
09:00, 03:00 03:00, 09:00, 15:00, 21:00
12:00, 06:00 04:00, 10:00, 16:00, 22:00
15:00, 09:00 05:00, 1 1 :00, 17:00, 23:00
18:00, 12:00 06:00, 12:00, 18:00, 24:00
-------
a
LU
2900
2800
2700
2600
2500
Load = 610 MW
_L
3 4
Time (hrs)
5 7
B-84693
a) FEGT continues to rise when blowing two wallblowers an hour
2800
O 2700
CD
LJJ
2600
2500
1811
1756
1700
1644
6
Time (h)
10
12
C-07093
b) using four blowers every 2 hours reduces FEGT
Figure 8
Effect of Sootblower Operation on FEGT
Another way better thermal environment control can lead to lower NOX is when the boiler has
been retrofit with a selective non-catalytic reduction (SNCR) process. With SNCR, either urea or
ammonia is injected into the upper furnace where it reacts with NO to produce N2. This reaction
is extremely temperature sensitive. If the urea injection temperature is greater than 2200°F, the
reagent can oxidize to form more NOX instead of getting rid of the NOX formed in the burner
zone. If the temperature is less than 1800°F, the reaction is incomplete and unreacted ammonia
(NH3) can be emitted with the flue gas. Besides being a hazardous pollutant, ammonia slip can
contribute to costly boiler operating problems. It will react with SO2 at lower temperatures to
form a sulfate fume which can deposit in the air preheater and plug the baskets. Excess ammonia
also adsorbs on flyash particles and can be collected by the electrostatic precipitator. This may
seem advantageous, but the flyash then takes on the characteristic ammonia odor which presents
-------
problems when handling or selling the flyash. In some cases, ammonia slip combines with gas-
phase chlorides to create a visible plume which may cause stack opacity to exceed regulatory
limits.
Continuous temperature measurement and control using smart sootblowing practice can minimize
variation in temperature at the point of urea or ammonia injection. Preventing temperature swings
will provide the following benefits:
1. less ammonia or urea usage to achieve NOX compliance
2. fewer episodes of ammonia slip.
The first benefit may be worth about $100K/yr for a 500 MW boiler if a 10% reduction in reagent
is achieved. If 20% more flyash can be sold, the cost benefit could be as much as $500K/yr.
Preventing shutdowns or load limitations due to air heater pluggage or opcity could be valued at
as much as $630K/day in lost revenue for a 500 MW unit (assume electricity sold at 7 cents/kW-h
and 75% daily load factor).
Currently, SDx is demonstrating control of SNCR injection temperature at a 155 MW coal-fired
boiler. The results of this demonstration will quantify the cost benefits achievable with thermal
history control.
Conclusions
Advanced sensors available today can be used to maintain low-NOx emissions and acceptable
burner and boiler performance after low-NOx retrofits. SDx is working with suppliers of low-
NOX equipment to implement these sensors. The results will be reflected in the operating costs of
coal-fired boilers during the decade of the 90's and beyond.
References
1. Beer, J.M., and Rodgers, L.W., et al, "Development and Testing of a Low-NOx Coal
Combustor for the B&W LEBS Program," American Flame Research Committee
Conference, Maui, HI, October 1994.
2. Sommer, T.M., Johnson, S.A. and Lindstrom, G.D., "Further Development of an
Advanced Low-NOx Combustion System and Its Potential Application to Coal-Fired
Utility Boilers," Transactions of the ASME, Paper No. WA/FU-4, American Society of
Mechanical Engineers, New York, NY, December 1979.
3. Johnson, S.A., Cioffi, P.L., and McElroy, M.W., "Development of an Advanced
Combustion System to Minimize NOX Emissions from Coal-Fired Boilers," presented to
the ASME/IEEE Joint Power Generation Conference, Dallas, TX, September, 1978.
4. Johnson, S.A., Morgan, M.E., Afonso, R.F., Dyas, B., and Carter, H.R., "Optimizing
Sootblower Operation in Response to Changing Coal Quality and Boiler Operation,"
presented at the Engineering Foundation Conference on The Impacts of Ash Deposition
on Coal-Fired Plants. Available through the Engineering Foundation, 345 East 47th
Street, New York, NY, June 1993.
-------
Session 5
Oil and Gas Combustion
-------
THE RADIAN RAPID MIX BURNER™ FOR ULTRA-LOW NOX EMISSIONS
Roger C. Christman
Steven J. Bortz
Dale E. Shore
Radian Corporation
Michael Brecker
Todd Combustion
Abstract
Radian Corporation together with its licensee, Todd Combustion, has developed, demonstrated,
and commercially implemented the Radian-Rapid Mix Burner™ (R-RMB™) that is capable of
producing ultra-low NOX levels for natural gas firing. NOX levels under 10 ppm, simultaneously
with CO levels of a similar magnitude, have been achieved over the load ranges of several
forced-draft industrial boilers. These emission levels have been achieved while maintaining
excellent flame quality and boiler performance.
The paper gives an overview of the RMB™'s design features. A review is provided of its
performance characteristics established during its development phase in a 4 MBtu/hr firetube test
boiler. Data illustrating the burner's sub-10 ppm NOX performance with and without air preheat,
and for circular and rectangular (for tangential firing) burner configurations are presented. Sub-10
ppm NOX data for commercial installations in two 5 MBtu/hr firetube boilers, a 26 MBtu/hr
watertube boiler, and a 130 MBtu/hr watertube boiler are presented. Data reviewing the burner's
performance for oil firing, and plans for its demonstration in a utility boiler are summarized.
Introduction
In response to the growing interest for low cost, practical solutions to the increasingly stringent
NOX emission limitations spawned by the enactment of the 1990 Clean Air Act Amendments,
Radian initiated an internally-funded research and development program with the following
objectives:
• Develop and demonstrate a burner-based solution to ultra-low NOX emissions (<10 ppm)
from gas-fired combustion systems (to compete directly with SCR-based solutions);
• Develop a burner which avoids many of the "classic" low NOX burner operating problems
which are characteristic of current commercial technologies;
• Develop a burner which utilizes reasonable FOR rates and can provide ultra-low NOX
performance for air preheat applications and high furnace heat release rate applications;
-------
Design the burner to be simple, rugged and scaleable to large sizes; and
Integrate and test oil-firing backup capability.
This program, initiated in 1993, is discussed in detail in this paper. The achievement of the
objectives set forth above, required "starting over" with NOX formation and combustion
fundamentals, and has resulted in a commercially available burner which is radically different from
other commercial low-NOx gas fired burners.
The two basic designs discussed in this paper rely upon; (1) rapid mixing of combustion air,
recirculated flue gas (FOR) and fuel gas prior to the ignition point (thereby eliminating prompt
NOX which is a by-product of substoichiometric combustion), (2) burner geometry which
produces an extremely stable flame, and (3) introduction of FOR (or excess air) to dramatically
reduce peak flame temperatures (thereby bringing the production of thermal NOX below 10 ppm).
The two designs differ in that the initial design, the R-RMB™, has a single, high swirl flow field
where fully mixed air, recirculated flue gas and fuel gas exit the burner throat and enter an
expansion quarl. The combination of high swirl and the expanding quart provide a stable ignition
point within the quarl.
The second version of the Rapid Mix Burner, the Dual Rapid Mix Burner (D-RMB™), was
developed to eliminate some of the problems that are inherent in the R-RMB™ design as sizes
exceed about 50 MBtu/hr. In the D-RMB™ design, a small R-RMB™ is surrounded by a second,
annular flow field in which the air, recirculated flue gas and fuel gas are again fully mixed, but are
not swirled by curved vanes. The outer, unswirled portion of the fuel/air/FGR mixture is
continuously ignited by the highly stable inner burner flame as it emerges from the inner quarl.
Thus, the need for a very large expansion quarl is eliminated and the D-RMB™ is small enough to
be installed in existing burner openings without enlargement. The performance characteristics of
the two burner designs are quite similar with <10 ppm NOX and zero or very low CO
concentrations having been commercially demonstrated for both the R-RMB™ and the D-RMB™.
As a result of the development and commercialization efforts of the past two years, Radian and it's
licensee, Todd Combustion, have provided a practical burner-based approach to ultra-low NOX
emissions from gas-fired boilers and dryers. This is significant not only as a low cost alternative
to SCR in new installations, but is also relevant to issues of emissions trading, emissions banking
and the generation of emissions offsets for internal use or sale. As described later, having
achieved commercial status in the industrial boiler arena, the Radian/Todd team is currently
working to demonstrate this technology for very large field erected, superheat boilers of both wall
and T-fired configurations.
Technical Background
In this section, a discussion of the concepts behind the R-RMB™ design, which meet the project
objectives, is presented.
Kinetic calculations indicate that thermal NOX emissions are typically the most important source of
NOX for natural gas flames with the NOX being created through the following reactions:
N + O2 = NO + O n)
N + OH = NO + H 2)
N2 + O = NO + N (3
-------
As the name implies, thermal NOX can be controlled through the control of the peak flame
temperature and if the temperature can be lowered enough, the NOX emissions from a natural gas
flame can be reduced to extremely low values (2800° F = 10 ppm NOX).
Under the appropriate conditions, the formation of prompt NOX can also be important. The kinetic
model used shows that under fuel-rich conditions, particularly when the stoichiometry is under
about 0.6, both HCN and NHs can be formed through reaction of CH with N2 to form HCN and
N. These calculations were conducted using gas and air mixtures with stoichiometries ranging
from 1.0 to 0.4. Below a stoichiometry of 0.5, almost all the NOX formed is prompt NOX. The
rate of prompt NOX formation is also very rapid, being nearly complete in about 1 ms at a
temperature of 2400° F. Although prompt NOX is temperature sensitive, the temperature sensitivity
is not as great as with thermal NOX and under fuel-rich conditions and a temperature of about
2400° F, 20 ppm of prompt NOX is formed.
The relationship between temperature, stoichiometry and NOX form the basis of the RMB™ design
for natural gas or other fuel nitrogen free fuels. The most direct method of achieving ultra low
NOX emissions from a natural gas flame is to both, avoid fuel-rich regions with the corresponding
potential for prompt NOX, and maintain low peak flame temperature to reduce thermal NOX
emissions to the desired level.
The RMB™ generates the desired stoichiometry by rapidly mixing the fuel and oxidant together in
a region near the burner exit. The rapid mixing results in an almost-uniform fuel/air mixture at the
ignition point. The stoichiometry of the mixture can be controlled to minimize prompt NOX
emissions and the flame temperature can be controlled by using FOR or any other source of inert
gases. In effect, the burner acts as a premixed burner with one important distinction. Since the
fuel is added just upstream of the burner throat, the premixed volume is extremely small and
unconfined, avoiding concerns about flashback or explosions. The rapid mixing is achieved using
a fine gas injection grid combined with axial swirl vanes working as a static mixer.
The effect of mixing rate is illustrated in Figure 1. The figure shows the effect of FOR for three
different mixing rates, ranging from slow to rapid mix. As shown in Figure 1, as the mixing rate
increases, the FOR has a greater effect on the NOX emissions. Although a faster mixing rate tends
to produce greater NOX emissions without FOR, for NOX emissions below about 30 ppm, the
rapid mix approach has definite advantages in reducing the FOR rate required to produce a given
NOX level.
The RMB™ can also be operated using liquid fuels by installing conventional atomizers in the
center of the burner. When the burner is operated using liquid fuels that contain fuel nitrogen,
staging of the fuel and air is required to control NOX emissions. The fuel and air mixing can either
be delayed by adjusting the spray angle of the atomizer or by using overfire air ports. FGR can be
used with liquid fuels to control thermal NOX emissions.
RMB™ Development Program
A one-year development program for gas firing consisting of testing in a 4 MBtu/hr firetube boiler
(Radian), 30 Mbtu/hr B&W water tube boiler (Todd Combustion), and a 100 MBtu/hr test facility
(funded by San Diego Gas & Electric Company) was conducted from approximately 6/1/93 to
6/1/94. The gas firing development work included design and test work for the initial version of
the R-RMB™, the second generation rapid mix burner, D-RMB™, and development of a biased
-------
fired configuration for multi-burner applications. Some work has also been conducted toward
applying the RMB™ concept to refinery gas applications and to ultra-low NOX high excess air
applications (dryers, gas turbines, etc.).
The firing density of the firetube boiler is approximately 100,000 Btu/hr/ft3, the Todd water tube
boiler 70,000 Btu/hr/ft3 and the large test furnace 8,000 Btu/hr/ft3. Air preheat is available on both
the firetube and large test facility furnace, but not on the Todd water tube boiler. In an EPRI
funded program, No. 2 and No. 6 oil firing was also conducted in the 4 MBtu/hr Radian firetube
boiler.
TM
R-RMB
The R-RMB™ design is conceptually simple and is shown in Figure 2. The total combustion air
or air/FGR mixture enters through an annular burner throat and passes through a set of axial swirl
vanes where the fuel is added using a grid of gas injectors built into the vanes. The location,
number, and diameter of the gas injectors, in combination with the turbulence generated by the
swirl vanes, provide rapid and complete mixing of the fuel and oxidant within a few inches of the
fuel injection point. This arrangement produces the advantages of premixed combustion without
the negative implications of having a large, confined premixed volume.
The geometry of the burner, with the divergent quarl and swirling flow, combine to generate a very
stable flame which is capable of operating at flame temperatures low enough to generate only a few
ppm NOX. Another advantage of the R-RMB™ is that these ultra-low NOX emissions can be
obtained with negligible (normally less than 1 ppm) CO and hydrocarbon emissions.
NOX emissions from the R-RMB™ firing into the Radian firetube boiler for ambient, 300° F
preheat and 500° F preheat, and as a function of FOR rate is shown in Figure 3. As shown, NOX
emissions without FOR were a function of the air preheat level and ranged from approximately 80
ppm with ambient air to 200 ppm with 500° F preheat. The FGR rate required to produce a given
NOX emissions varied with the air preheat level; but, independent of the preheat level, the NOX
emissions could be reduced to approximately 3 ppm.
Similar results were obtained in the Todd Combustion 25,000 Ib/hr water tube boiler (30 MBtu/hr)
as shown by the ambient combustion air data in Figure 4. NOX emissions without FGR were
higher for the larger burner but when enough FGR was added to reduce the NOX below 20 ppm,
both size burners exhibited a very similar performance.
Similar characteristics were observed with air preheat (Figure 4). Again, the larger burner
produced higher NOX emissions without FGR but once FGR was added, the NOX emissions from
the 4 MBtu/hr and 100 Mbtu/hr burners were almost identical for a given FGR rate.
An important characteristic of the burner, particularly for high excess air and utility (multiple
burner) applications, is the interchangability of FGR and excess air in reducing NOX emissions.
Using only high excess air, NOX emissions (normalized to 3% Oa) can also be reduced to less than
3 ppm.
-------
D-RMB™ (Circular Burner)
The D-RMB™ version of the Rapid Mix Burner is shown in Figure 5. The D-RMB™ consists of
an R-RMB™ surrounded by an non-swirled axial flow field containing a rapid mix injection grid.
Both the swirled and axial flow portions of the burner operate at the same stoichiometry and with
the same FOR rate. The D-RMB™ version of the R-RMB™ would normally be applied for burner
sizes greater than about 50 MBtu/hr.
Figure 6 compares the NOX versus FOR relationship for the D-RMB™ and R-RMB™. Typically,
less FOR was required for the D-RMB™, since the flame length was somewhat longer and
radiation losses and the entrainment of furnace gases contributed to reducing the flame temperature
and NOX emissions. Particularly with 500° F preheat, the reduction in FOR rate achieved using the
D-RMB™ for a 10 ppm NOX level is significant. With the D-RMB™, 27% FOR was required to
achieve 10 ppm NOX, while 40% FOR with the R-RMB™ was required for the same NOX level.
As is the case with the R-RMB™, the D-RMB™ is relatively insensitive to the burner scale.
Figure 7 compares the NOX emissions as a function of the FGR rate for 4 MBtu/hr and
130 MBtu/hr D-RMB™s. The two size burners show almost identical NOX emissions for a given
FGR rate and the measured emissions are significantly less than what is calculated using chemical
kinetics. The lower NOX emissions are related to the importance of external recirculation with the
D-RMB™.
D-RMB™ (Non-Circular Burners)
One important advantage of the D-RMB™ relative to the R-RMB™ is that the outer shape of the
burner need not be circular. This means that the burner can easily be applied to boilers which have
non-circular burner openings (corner fired boilers) without pressure part modifications.
Figure 8 compares the NOX emissions obtained using a rectangular D-RMB™ and a circular
RMB™. Both burners had the same maximum thermal capacity of 4 MBtu/hr. The rectangular
burner had outer dimensions of 4.5" x 8" while the circular burner had a diameter of 7 inches.
Both burners had nearly identical performance verifying that the outer shape of the D-RMB™ need
not be circular so long as uniform and rapid mixing between the air and fuel is achieved.
Biased Firing
As described earlier, the R-RMB™ is capable of producing NOX and CO emissions of less than 9
ppm, even with air preheat. However, when the air preheat level is 500° F or greater, FGR rates in
excess of 30% in combination with 15% excess air are required to achieve these low emission
levels. These high FGR levels will create steam temperature control and other operating problems
in many, if not most, existing boilers with convective superheaters. Since the NOX emissions can
be controlled using the RMB™ equally effectively using excess air or FGR, a multi-burner
D-RMB™ boiler can operate in what is commonly called a biased fired mode of operation to
control NOX emissions. Biased firing means, in a multi-burner furnace, that some burners operate
air rich and others operate fuel rich. Figure 9 shows the performance of the RMB™, calculated
from chemical kinetics, as a function of the burner stoichiometry. The data in Figure 9 show that,
even with air preheat, operating one burner near 80% excess air and another burner at a
stoichiometry of 0.6 should result in NOX emissions from both burners less than 10 ppm. If the
-------
second stage NOX, created when the fuel-rich and fuel-lean flames are mixed, can be maintained
near zero, then overall ultra-low NOX levels can be achieved.
Figure 10 compares the measured results operating a two-burner R-RMB™ installation (Radian
firetube boiler). The lower curve was generated by biased firing with the fuel-lean burner
operating at 94% excess air and the fuel-rich burner operating at 0.63 stoichiometry, maintaining
an overall excess air level of 10%. The upper-most curve represents the same burners both
operating normally ^unbiased) at the same 10% excess air level. The data in Figure 10 demonstrate
that, without FOR, biased firing results in a reduction in NOX emissions from 300 ppm to 20 ppm.
If FOR is used with biased firing, the amount of FOR required to achieve 10 ppm NOX is reduced
from 40% to less than 20%. Although the data shown in Figure 10 are from a two-burner
R-RMB™ configuration, the same performance would be expected from multi-burner D-RMB™
operation.
Oil Firing
No. 2 and No. 6 oil have been tested with the R-RMB™ and D-RMB™ versions of the Radian
rapid mix burner. This work was done in an EPRI-funded program using the Radian firetube
boiler. Electric Power Technologies (EPT) has jointly developed, together with ESEERCO, EPRI
and Con Edison of New York, a technology known as REACH for optimizing fuel oil
combustion. Combining the REACH technology with the R-RMB™ and D-RMB™ provides the
potential for obtaining optimum emission performance for both gas and fuel oil.
The main objective of the project, to verify that the REACH V-Jet atomizers are compatible with
both designs of the RMB™, was achieved with positive results. With both the D-RMB™ and the
R-RMB™, particularly with No. 6 oil, significant NOX reductions were achieved relative to
baseline numbers generated with conventional internal mix or Y-Jet atomizers. For the R-RMB™,
NOX reductions of about 28% were achieved and for the D-RMB™, 35% NOX reduction was
achieved.
The effectiveness of FGR in reducing NOX emissions was dependent on the fuel nitrogen content
of the oil. For No. 2 oil, FGR was very effective in reducing the NOX emissions. FGR also
reduced smoke emissions. For No. 6 oil, NOX emissions typically increased or remained the same
when FGR was added.
On the D-RMB™, using 500° F air preheat, the lowest NOX values achieved, near full load are as
follows:
OilType/(%N) NOX w/o FGR NOX with FGR Atomizer
(ppm) (ppm)
2/(0-03) 78.7 37 50/120BS V-Jet
6/(°-3) 173 178 50/120BS V-Jet
2/6Mix/(0.14) 141 126 50/120BS V-Jet
-------
Commercial Installations
125 HP Firetube boiler (200,000 Btu/hr/ft3 Volumetric Heat Release Rate)
In the time period June through August 1994, two gas-fired 125 hp Cleaver Brooks Boilers were
retrofit with 5 MBtu/hr R-RMB™. These installations were co-funded by Rockwell (the boilers'
owner), Southern California Edison, Southern California Gas, and Radian. Both boilers are
located in Rockwell's Downey, California facility and the retrofit was in response to South Coast
AQMD Rule 1146.1. The burners replaced standard burners which had NOX emissions of
approximately 120 ppm and CO emissions of 20 ppm at full load. The original burners operated in
a on-off mode. During the retrofit, the original burner, fan, and controls were replaced. The
control system utilized separate actuators for the gas, air and FOR supplies. All three actuators
were controlled via a single controller. The front door feature of the Cleaver Brooks boiler was
retained and the new combination FDAnduced FOR fan was installed in the door in the same
location as the original fan. The objective of the project was to demonstrate the ability of the
burner to operate at less than 9 ppm NOX and CO across the load range (5/1 turndown).
Figure 11 shows the resulting NOX levels. The NOX ranged between 6 and 8 ppm over the load
range. The excess O2 varied from about 6% at minimum load to 3% at full load. FOR rates
ranged between 25% and 35%. The flame was approximately 3 feet long.
The two boilers were source tested on June 29,1994 and September 2,1994 per SCAQMD
protocol by a third party contractor. Results are shown below. The CO and NOX numbers are
normalized to 3% O2-
Source Test Data
Load O2 CO NOX
(%) (%) (Ppm) (ppm)
31 3.6 1.05 6.4
55 3.6 0.7 7.6
100 2.9 1.9 8.1
21,500 Ib/hr Watertube Boiler (70,000 Btu/hr/ft* Volumetric Heat Release
Rate)
In November of 1994, a 26 MBtu/hr induced FOR R-RMB™ was retrofit into a 21,500 Ib/hr
Nebraska boiler located at the Wilmington, California, US Borax facility. The burner replaced an
induced FOR burner permitted for 30 ppm NOX and 400 ppm CO. The retrofit consisted of
replacing the existing burner and fan, while the gas train and most control system components
were reused. The objective of the project was to demonstrate the ability of the burner to operate at
less than 9 ppm NOX and CO across the load range (5/1 turndown).
A natural gas only, 26 MBtu/hr R-RMB™ was installed on a Nebraska model NS-B-35 water tube
boiler equipped with an economizer. The burner installation consisted of the R-RMB™, a induced
flue gas recirculation system and windbox baffles to provide uniform air distribution to the burner.
The adjustable fuel valve and air dampers were physically linked to provide control of the air/fuel
ratio over the load range, while the FOR damper had a separate actuator and controller. Figure 12
-------
shows the test results when the boiler was in automatic operation. The FGR rate varied from about
18% at low load to 25% at full load. The NOX varied between 6 and 7 ppm over the load range,
while the CO varied from less than 1 ppm at full load to about 3 ppm at low load. The excess O2
varied from about 4.5% at low load to 3.2% at full load. The flame was about 6 feet long, and the
flame was extremely stable across the 5/1 load range. The burner operates with 6.5 inches draft
loss and 5 psig fuel gas pressure. The new burner has been in operation since November 14,
1994.
On November 7,1994 the boiler was source tested as per SCAQMD protocol. The test results are
shown below. The CO and NOX numbers are normalized to 3% O2.
Source Test Data
Load 02 CO NOX
(%) (%) (ppm) (ppm)
25 3.4 3.0 8.2
53 2.6 0.9 7.1
98 2.7 0.1 6.8
100,000 Ib/hr Watertube Boiler (90,000 Btu/hr/W Volumetric Heat
Release Rate)
In late November of 1994, a single 130 MBtu/hr D-RMB™ was retrofit into a 100,000 Ib/hr
Nebraska boiler at the Morning Star Packing Company facility in Los Banos, California. The
Morning Star retrofit was co-funded by Morning Star, Todd Combustion, and Radian. The burner
replaced a forced FGR 118 MBtu/hr burner permitted for 30 ppm NOX and 400 ppm CO. The
retrofit consisted of replacing the existing burner and fan, while most of the existing control system
components were reused. The objective of the project was to demonstrate the ability of the
D-RMB™ version of the RMB™ to operate at less than 10 ppm NOX and CO across the load
range. Since the NOX emissions of the D-RMB™ are relatively insensitive to the heat release rate,
the increased burner capacity allowed the boiler capacity to be increased from 100,000 to 110,000
Ib/hr.
A natural gas only, 130 MBtu/hr D-RMB™, was installed on a Nebraska 100,000 Ib/hr water tube
boiler equipped with an economizer. The burner installation consisted of the D-RMB™ and an
induced FGR forced draft fan. The existing cross limiting control system and gas train was
re-used for the D-RMB™. The D-RMB™ burner operates at 8 inches draft loss and 5 psig fuel
gas pressure at full load. Since the performance of the D-RMB™ is relatively insensitive to the
heat release rate, a 130 MBtu/hr D-RMB™ was used to replace the existing burner that had a
118 MBtu/hr capacity. The larger burner capacity increased the boiler capacity from 100,000 Ib/hr
to 110,000 Ib/hr.
Figure 13 illustrates the test results achieved with the boiler in automatic operation. The FGR rate
varied from 28% in the lower half of the load range to 23% a full load. NOX emissions vary
between 8 ppm and 9 ppm. Over the load range, the CO was less than 1 ppm. Excess oxygen
levels ranged from 4% at minimum load to 3% at full load. The burner turndown was 6/1 and the
flame was less than 10 feet long over the load range.
On December 14, 1994 the boiler was source tested by a third party contractor and the results are
shown below.
-------
Load O2 CO NOX
(%) (%) (ppm) (ppm)
29 3.4 <1 8.7
55 3.4 <1 7.6
100 3.2 <1 8.7
NOX emissions data as a function of FOR rate are summarized for all prototype and commercial
installations using ambient air in Figure 14. For FOR rates of 15% or higher, the NOX emissions
are almost identical for all installations, even through the volumetric heat release rates range from
70,000 to 200,000 Btu/hr/ft3.
Application to Utility Boilers
The RMB™ development was taken on with industrial boilers and multiple-burner electric utility
boilers both being considered as important commercial applications. For many smaller industrial
boilers, the standard R-RMB™ proved to be the most appropriate design. However for retrofit
into utility boilers, the D-RMB™ design eliminated the need for pressure part modifications since
its smaller size (relative to the standard R-RMB™) allows installation into existing waterwall
burner openings. Similarly, the design of the rectangular D-RMB™ allows installation into
tangentially-fired boilers, without waterwall comer pressure part modifications.
For industrial boilers, the RMB™ is considered a fully commercial product, and is being marketed
and implemented as such. For utility boilers, the technology has not yet advanced to this status.
Demonstrations are currently being assembled to prove the RMB™'s viability for utility boiler NOX
control, prior to commercialization. The following paragraphs discuss the use of the RMB™ in
utility boilers, the proposed firing configurations, and anticipated NOX control performance.
There are two fundamental approaches to implementing the D-RMB™ in a multiple-burner utility
boiler; normal firing and biased firing. The consideration of each of these two is dependent upon
the near- and long-term NOX control goals for the specific unit and the utility company. For
normal firing, it is projected that NOX emissions can be maintained at 30 ppm or less over a unit's
load range. For biased firing, the potential exists to maintain NOX under 10 ppm over the load
range.
These projections are the result of a significant amount of data on the burner's performance
characteristics for a variety of configurations, capacity ratings, and furnace designs. These data,
viewed as a whole, illustrate the consistency with which the basic RMB™ design and its various
improved configurations perform. This consistency, along with an understanding of the
fundamental workings of the burner, provides confidence to predict its performance in a range of
applications.
For normal firing (non-biased) on a utility boiler with air preheat of nominally 600° F, it is
projected that NOX levels under 30 ppm can be achieved using no more than 20% FOR. For most
utility boilers, 20% FOR is probably the most that can be tolerated at full load, without creating the
need for extraordinary efforts to accommodate increased furnace pressures, overcome draft losses,
and control superheat and reheat steam temperatures. For some boilers, this maximum could be
less, for others, it could be more. However, given a nominal 20% FGR maximum at full load, as
load was reduced, the total (mass flow) of FGR could remain fairly constant, thereby increasing
-------
the effective percentage FOR. This would provide the means to reduce NOX levels substantially
below 30 ppm, at lower loads. For example, at loads around 60 or 70% of MCR (and below),
FOR rates on the order of 30 to 35% would be feasible, without undue impact on furnace
pressures, draft losses, and steam temperature control. At these FOR rates (and at even higher
rates), flame stability for the RMB™ remains excellent. The end result would be, at such FOR
rates, NOX levels on the order of 10 to 15 ppm, negligible CO emissions, and compact, stable
flames.
The ability to operate a utility boiler with NOX levels between 10 and 15 ppm (at low and moderate
loads) and up to 30 ppm (at high loads) has some important implications relative to compliance. In
the simplest scenario, if the limit calls for a flat 30 ppm, then compliance is straight forward.
However, some regulations call for a mass emission rate, averaged over a stated time. Given a
typical duty cycle of a load-following gas-fired unit, a 24-hour average NOX level may be more on
the order of 15 or 20 ppm. In this case, NOX emissions for the averaging period may be well
within limits, even though the high load case (30 ppm) is above the average.
Another scenario comprises the cases where time averaging is not sufficient to null out the 20 to 30
ppm NOX peaks, or where the regulation indeed calls for ultra-low NOX levels (10 ppm or so) over
the entire load range. The necessary incremental NOX reduction could be provided using a trim
SCR system. Since the starting NOX levels (30 ppm or less at the SCR inlet) and the percentage
reduction necessary with SCR (between zero and 66%, depending upon load) would both be low,
the capital and O&M costs of the trim SCR could be low. Ammonia injection would only be
required for high load periods, reducing the total consumption and reducing the amount of on-site
storage.
Even though combining a 30-ppm RMB™ and trim SCR could provide an overall 10 ppm NOX
compliance level, the more desirable approach would be achieving 10 ppm with burners alone. If
so, then the biased firing RMB™ approach warrants consideration. Ultra-low-NOx RMB™
operation has been shown to be possible by biasing the burner stoichiometry and using
approximately 20% FOR.
If a utility boiler steamside (superheat/reheat temperature control) could tolerate FOR rates of 30 to
40% at high loads, achieving ultra-low-NOx levels without biasing would clearly be feasible based
on existing data. A new utility boiler could be designed to accommodate such FOR rates.
However in a retrofit, adapting to the increased heat transfer in the convective superheaters and
reheater, as well as the increased furnace pressures and draft losses, would likely render the
application impracticable or, at best, very costly.
Fortunately, stoichiometry biasing provides an approach for significantly reducing the total FOR
rate while achieving ultra-low-NOx levels with the RMB™ in a utility boiler. With biasing, some
burners are operated air rich and others fuel rich. The performance characteristics of the RMB™
are such that excess air provides the same flame temperature (reduced thermal NOx) effect as does
FOR. Therefore in a bias-fired mode, those burners operating air rich do not require FOR to
produce single-digit NOX levels. FOR is required only to the fuel-rich burners. The net
requirement is that the total amount of FOR needed is only about half of that needed in a normal
(non-biased) firing mode. Therefore, the feasibility exists for achieving ultra-low-NOx (single
digit) levels with less than 20% FOR by firing the RMB™s in a biased mode (Figure 10).
An example of a biased firing set up is illustrated in Figure 15. The figure depicts a wall-fired
boiler with two burner elevations. The bottom row would be fired at a stoichiometric ratio around
0.7 and the top row would have a stoichiometric ratio around 1.6 or 1.7. The stoichiometry
-------
biasing would be accomplished by diverting about 60% of the total fuel flow to the bottom burners
and about 60% of the total air flow to the top burners. Based on projections using RMB™
development and existing demonstration data, the top burners would produce NOX levels of about
7 ppm, by virtue of the high excess air levels (no FGR necessary for the top burners). FOR
would be supplied to the bottom burners only using partitioning within the ducting and windbox,
as illustrated in the figure. Adding 20% FGR to the bottom burners would reduce their NOX levels
to about 10 ppm. Note that recirculating 20% of the boiler flue gases to half of the burners would
result in those burners operating at an equivalent, effective FGR rate of 40%.
Stoichiometry biasing is not a new idea, as it has already been employed on gas-fired utility
boilers. As a result, there is a basis for projecting its effectiveness with the RMB™ in adding to
the Radian work described earlier. An element important to the effectiveness of biasing is the
control of the stoichiometries within individual burner flames to relatively narrow and discrete
ranges. For utility boilers, some of this control results from having good control of air flow and/or
fuel flow to each burner — this will provide good control of overall burner Stoichiometry.
However, when biasing with conventional (diffusion) burners, the Stoichiometry within the flame
will vary significantly, simply by virtue of the fact that the mixing is by diffusion. On the other
hand, the RMB™ does not depend upon diffusion mixing; the fuel and air are nearly completely
mixed upstream of any combustion. Furthermore, because of the RMB™'s design, the
Stoichiometry upstream and within the flame can be maintained within a very narrow range. As a
result, the effectiveness of biasing using the RMB™ would be superior to that using conventional
burners.
For biasing to be effective, the upper (air rich) and lower (fuel rich) burner elevation streams have
to mix and burn at a temperature low enough so that only small amounts of bulk zone NOX are
formed. In addition, the temperature and residence time has to be enough to assure the complete
burnout of CO. Therefore, the feasibility of successfully carrying out biasing is dependent, not
only on burner characteristics, but also on burner spacing and furnace geometry.
Presently, a program to demonstrate the RMB™ for utility boiler applications is in the planning
stages at Radian and Todd Combustion, with the support of the Electric Power Research Institute
and several other organizations. A three phase program is being envisioned. Phase 1 would
comprise the installation of RMB™s and FGR on a two-burner package boiler and the proof-of-
concept testing of biased firing for utility size burners. A candidate two-burner boiler rated at 176
MBtu/hr heat input has been identified and project implementation is underway. Phase 2 would
consist of an installation and demonstration on a full-scale, wall-fired power plant boiler. The ideal
unit would be in a size range of 50 to 200 MWe. Candidate units have been identified and
discussions with the utilities are underway. Phase 3 would be a demonstration on a similarly-
sized, tangentially-fired unit
Conclusions
• The R-RMB™ provides an effective means of reducing NOX emissions from gas-fired
boilers (and similar devices) to ultra-low levels that, heretofore, were only possible with
catalytic control.
• The RMB™ design is a major departure from that of conventional fuel- or air-staged
low-NOx burners. As a result, the RMB™ avoids many of the side impacts and limitations
of conventional burners. CO and HC emissions are negligible, flame stability is excellent
(even with very high FGR levels), and the flame is compact (thus avoiding impingement).
-------
The RMB™, and its performance characteristics, scale to sizes covering the range of most
commercial, industrial, and utility installations. The design bases and the performance data
for burners sized at 4, 26, 30, 100, and 130 MBtu/hr are consistent.
The burner is a simple, rugged design that is suitable for the industrial and power plant
equipment environment The burner has and requires no field-adjustable variables, other
than a one-time setting of an air shroud for best air distribution and pressure drop.
The burner is a proven product for industrial boilers and similar industrial devices (forced
draft heaters, dryers). The burner is manufactured by Todd Combustion and offered
commercially for new and retrofit applications and is backed by guarantees of ultra-low
NOX and CO emission levels.
The RMB™ is applicable to new and retrofit installations. For retrofit applications, the
burner will fit into existing burner openings, avoiding the need for pressure part
modification.
The RMB™ offers the potential for providing low (30 ppm or less) and ultra-low NOX
levels (10 ppm or less) for gas-fired electric utility boilers; both wall- and tangentially-fired.
Programs to evaluate and demonstrate this potential are being developed.
-------
100
Oil
0
£
Q.
Q.
1: Slow Mixing
2: Nozzle Mix
3: Rapid Mix
Furnace
Wall /T*^^ External
ff Flecirculation
Combustion
Air & Flue Gas
10 20 30 40 50 60 70 80
% FGR/Stack (SCFM/SCFM)
Fuel Gas —t
Internal
Reclrculation
Zone
Figure 1. Measured Effect of Mixing Rate on NOX Emissions Figure 2. Radian Rapid-Mix Burner Operating Principle
1000
1000
15% Excess Air
Ambient Air
300 °F Preheat
500 °F Preheat
15% Excess Air
5 MBtu/Hr Ambient
30 MBtu/Hr Ambient
5 MBtu/Hr 500 °F
—x— 100 MBtu/Hr 500 °F
10 20 30 40 50
% FGR (FGR/Stack Exit)
60
70
10 20 30 40 50
% FGR (FGR/Stack Exit)
60
70
Figure 3. Effect of Air Preheat and FGR Rate on NOX; R-RMB
Figure 4. 5,30 and 100 MBtu/Hr R-RMB Test Results
-------
Combustion
Air & Flue Gas
External
Recirculation
Zone
Combustion
Air & Flue Gas
\
Fuel Gas
Oil—f
Swirl Vanes —
and Gas
Injectors
1000
Internal
Recirculation
Zone
Figure 5. Radian Dual Rapid-Mix Burner Operating Principle
100
8
Q.
a.
$
D-RMB Ambient
R-RMB Ambient
D-RMB 500°F
R-RMB 500°F
0 5 10 15 20 25 30 35 40 45 50 55 60
% FGR
Figure 6. Comparison of D-RMB and R-RMB NOX
Performance 4 MBtu/hr Firetube Boiler
2.0 MBTu/Hr Rect.
3.0 MBtu/Hr Rect.
—x— 2.0 MBtu/Hr Round
3.0 MBtu/Hr Round
Figure 7. Effect of Burner Size D-RMB
(15 % Excess Air) Ambient Air
Figure 8. Comparison of Circular and Rectangular
D-RMB Results (Ambient Air)
-------
10000
Ambient Air
500°F Preheat
0.4 0.6 0.8 1 1.2 1.4
Stoichiometry
1.6
1.8
1000
cvj 100
O
^P
o^
CO
10
Figure 9. Effect of Burner Stoichiometry Theoretical
RMB NOX Emissions w/o FGR
10% Excess Air
1.1,1.1 FGR Both
x 1.73, 0.63 FGR Both
>— 1.8, 0.7 FGR Rich
1.94, 0.63 FGR Rich
10
60
20 30 40 50
% FGR (FGR/Stack Exit)
Figure 10. Biased RMB Test Results With 500 °F Air Preheat
70
-------
NO*
02
—x— NOX Source Test
CM
O
12345
MBtu/hr
Figure 11. 125 HP Firetube Installation
Rockwell Boiler 1
CO < 3 ppm Over Load Range
NOX (ppm) COS A
02
FGR (%)
—x— NOX Source Test
10 15 20
Load MBtu/Hr
Figure 12. 21,500 Lb/Hr Watertube Boiler
U.S. Borax No. 7
DC
O
O
I
x
O
10
9
8
6
5
4
3
2
1
• NOX (ppm) —A— FGR (%)
• 02 (%) —x— NOX Source Test
20 40 60 80 100
Load MBtu/Hr
120
Figure 13. 100,000 Lb/Hr Watertube
Morning Star No. 1
40
36
32
28
24
20 oc
O
u.
16
12
8
4
140
-------
9. ioo -
a1 |
a.
a.
x 10 -
O
1 .
\
^•S^:::::j::::::::::::::::::
XV \
N>e'V
NJxK,
N&V
^
-
*S
^*^&j
— :::;^va
"•{-•-••"•"— -i
— »— — L
^B4-R
— * — LAB 5-D
— x — TODD 30-R
a BORAX 26-R
• R
A — M
^fr*%"*~^,
^^^*^^
^^i
OCKWELL 5-R
S 130-D
::::
_
—
....
10 20 30 40 50 60 70
% FGR
Figure 14. R-RMB and D-RMB FGR Versus NOX Data (Ambient Combustion Air)
Stoich. S 1.6-1.7
Stoich. = 0.7
2Q% FGR
Duct/Windbox/Partition
FGR Fan
Figure 15. Biased Firing Set Up
-------
INTRODUCING EUROPEAN LOW NOX BURNER TECHNOLOGY
TO THE U.S. MARKET
Martin E. Drumm
Dennis Mo ran
Public Service Department
City of Burbank
164 West Magnolia Boulevard
Burbank, California 91503-0631 USA
Bruce C. Sudduth
M.N. Mansour
Farzan Roshdieh
AUS Combustion Company
2042 Business Center Drive
Irvine, California 92715 USA
James B. Champion
Hamworthy Combustion Engineering Limited
Fleets Corner
Poole, Dorset BH17 7LA England
Abstract
Low NOX burners (LNBs) developed by Hamworthy Combustion Engineering Limited
(HAMWORTHY) in the United Kingdom were recently installed in three electric
utility boilers owned and operated by the City of Burbank, California (CITY). The
HAMWORTHY DFL LNBs offered in the United States, under the trade name AUS-
DFL, incorporate the most recent advances made in European LNB technology. The
AUS-DFL is designed to achieve substantial reduction in NOX emissions on both fuel
oil and natural gas firing without the use of a flue gas recirculation (FOR) system or
overfire air (OFA) ports.
The AUS-DFL has a divided flow register design with primary and secondary air
passages. Burner development is based on a progressive refinement of
HAMWORTHY's well established low NOX products and extensive full-scale testing.
Design philosophy focused on establishing an internally staged flame structure with
exceptional flame stability. Fuel injection orientation and the use of strategically
positioned baffles satisfied the design objectives.
00001-R1511-DLR
-1-
-------
In this paper, NOX reductions achieved with the AUS-DFL LNB on natural gas firing
are presented. Retrofitted boilers spanned a range of burner zone heat release rates
(BZHR) from 0.23 MMBtu/hr/ft2 to 0.34 MMBtu/hr/ft2 and, as such, it was possible
to isolate the effect of burner and furnace design on NOX emissions. Correlations
were developed for the dependence of thermal NOX formation on furnace geometry.
The AUS-DFL achieved the lowest NOX emissions level ever obtained in utility boiler
applications without the use of FOR. Full-load NOX emissions ranged from 78 ppm
(@ 3% O2) to 105 ppm (@ 3% Oj) depending upon the BZHR, with low-load NOX
below 45 ppm (@ 3% O2). The NOX emissions reduction from pre-retrofit baseline
conditions ranged from 20 to 70 percent.
Introduction
The adoption of Rule 1135 by California's South Coast Air Quality Management
District requires the CITY to reduce NOX emissions from its power plants by as much
as 92 percent. The Rule specifies a final NOX emissions compliance limit at 0.20 Ib
NO2/MWh net (approximately 18 ppm @ 3% O?) and required compliance to be
achieved by 1999. Intermediate limits for 1993 and 1996 were 1.73 and 0.99 Ib
NO2/MWh, respectively.
The CITY compliance strategy consisted of implementing combustion and post-
combustion NOX emissions control measures on an incremental basis to result in a
single integrated low NOX system. Combustion-based NOX control represented the
first step towards compliance based on cost-effectiveness and operational
considerations. In addition to low NOX, combustion-based NOX emissions control
offered an opportunity to add a burner management system (BMS).
Through a competitive procurement process, AUS Combustion Company (AUSCC)
was selected as the contractor to supply LNBs to three electric utility boilers, ranging
in size from 20 MW to 44 MW. The scope of supply included LNBs, BMS, and
equipment installation and start-up.
AUSCC is jointly owned by Applied Utility Systems, Inc. (AUS) and HAMWORTHY
and provides LNBs and combustion-based NOX control measures to electric utility
power plants, petrochemical furnaces, and industrial boilers. As a Southern
California firm, AUS has acquired extensive expertise in NOX emissions control
required to assist local utilities in meeting stringent Los Angeles emissions limits.
Since 1985, AUS has developed and commercialized a range of combustion and post-
combustion NOX control technologies, including selective non-catalytic reduction
systems with the use of urea, catalytic air heaters, and advanced LNBs.
HAMWORTHY is an industry leader in low NOX combustion technologies with four
00001-R1511-DLR -2~
-------
wholly-owned subsidiaries, including Hamworthy Combustion Systems Limited,
Peabody Engineering Limited, Airoil-Flaregas Limited, and Peabody Engineering
Corporation. HAMWORTHY is one of the largest low NOX combustion companies
worldwide. HAMWORTHY's LNB technology for electric utility boilers is licensed
in the United States and Canada on an exclusive basis to AUSCC.
LNB Design Chosen
As a member of the HAMWORTHY group of combustion companies, AUSCC has
access to several LNBs for utility boiler applications. The merger of several
successful combustion companies both in Europe and the United States under
HAMWORTHY's leadership provided the HAMWORTHY group with access to a
broad range of LNB products, and offered AUSCC the flexibility to match the LNB
selection with the specific end-use application.
The AUS-DFL was selected for the CITY project, based on its ability to achieve low
NOX without the use of FGR or OF A ports. The burner development represents the
progressive refinement of HAMWORTHY's well established, low NOX products, and
it is the product of extensive full-scale testing in HAMWORTHY's own research and
development facility. A schematic of the AUS-DFL is presented in Figure 1. The
burner has a divided flow register design with primary and secondary air passages.
Natural gas is fired through gas pokers, and oil is fired through a center atomizer. A
swirler is used to impart an angular momentum to the primary combustion air, and an
annular vane assembly provides swirl to the secondary combustion air.
Low NOX operation is achieved by optimizing the mixing between fuel and air on both
natural gas and oil firing. Special baffles (view on Arrow B) are placed to extend
from the swirler to the outer periphery of the secondary air passage. The geometry
and the location of these baffles are selected to induce a recirculation field and
produce an inherently stable finger-like flame with fuel-rich stoichiometry. FGR is
internal and self-induced by the burner flow field. Combustion staging is established
by the drilling pattern of gas pokers and orientation of oil atomizer jets relative to the
special baffles. Small gas pokers are used to augment flame stability during gas
firing.
In addition to achieving low NOX without FGR or OFA, the DFL burner displays
exceptional flame stability. The flame is larger in diameter and significantly shorter
in length than flames for conventional burners. The burner is thus most suited for
applications with limited furnace depth. One of the burner's most attractive
characteristics is its insensitivity to excess air on gas firing. The burner shows very
small change in NOX emissions with variation in excess air because of the internal
staging of the flame.
00001-R1511-DLR
-------
Scope of Supply
The low NOX combustion systems were implemented by AUSCC on a turnkey basis.
The scope of supply included LNBs, BMS, equipment installation, and systems start-
up and optimization. A summary of the scope of supply for each boiler is presented
in Table 1.
Engineering services included the application of cold flow and heat transfer
modeling, as appropriate, to ensure satisfactory LNB performance. Unique LNB
interface requirements dictated the custom design of the LNB assemblies to allow
them to fit within the confined space available on the boiler fronts. Detailed drawings
of piping arrangement and electrical instrumentation layouts were required.
In order to provide access to the furnace, removal of the lower right burner as a
single assembly was required. It was necessary to keep the burner removal simple and
easy, in order to allow the boiler maintenance to be performed in a timely manner.
Pre-retrofit burners were assembled into a hinged furnace door that could swing open
with ease. The burner throat was thus the primary acceptable access to the furnace
cavity.
Modifications of the LNBs were implemented to satisfy the CITY requirements. In
addition, in joint effort with the CITY staff, a special cart was designed for burner
removal. This cart greatly reduced the physical effort associated with the burner
removal, and permitted the removal to be accomplished in a few hours' time.
Boiler Description
A summary of the design characteristics of the retrofitted boilers is presented in
Table 2. All boilers are front wall fired with six burners arranged in two rows with
three burners in each row. Olive Unit 1 is force draft fired, while Magnolia Units 3
and 4 are balanced draft fired. The individual burners maximum firing rate ranges
from 58 MMBtu/hr to 107 MMBtu/hr. The combustion air preheat ranges from 520°F
to 550°F. All boilers are subcritical without reheat steam.
The boiler furnace's heat release intensity was characterized by the BZHR. The
BZHR is a parameter that describes the heat flux intensity on the surface of a cube
surrounding the flame zone. The geometry of this cube is determined by considering
the four furnace walls surrounding the flame and two imaginary surfaces
perpendicular to the surface walls; one above and one below the flame zone. The
imaginary surfaces are positioned one-half burner spacing above the top burner
elevation and one-half burner spacing below the bottom burner elevation. Only cooled
surfaces are considered in determining the BZHR. Correction factors are used to
00001-R1511-DLR -4-
-------
account for uncooled surfaces, such as areas covered by refractory, which are
subtracted from the total surface area.
The BZHR is calculated by dividing the total boiler heat input (MMBtu/hr) by the
effective cooled surface area of the cube surrounding the flame zone (ft2) and
reported in MMBtu/hr/ft2. A high BZHR value is an indication of intense heat release
which produces high NOX emissions; a low BZHR should provide correspondingly
lower NOX emissions.
The BZHR for the retrofitted boilers ranged from 0.23 MMBtu/hr/ft2 to 0.34
MMBtu/hr/ft2. The BZHR value appeared to be independent of boiler electrical MW
rating. For example, Magnolia Unit 3 showed a higher BZHR than Olive Unit 1 and
Magnolia Unit 4. This offered the opportunity to decouple NOX formation from boiler
total capacity (MW) and, to an equal degree, from burner firing rate (MMBtu/hr).
Since design guidelines for LNBs were consistently maintained over a broad range of
burner sizes in this retrofit project, it was possible to establish a correlation between
burner design and boiler furnace thermal environment (or BZHR).
LNBs Retrofit Results
NO, Emissions Reduction Achieved
'X
A summary of data obtained during preliminary testing is presented in Figure 2. NOX
emissions reduction achieved was a direct function of baseline NOX emissions.
Reductions as high as 70 percent were achieved on Magnolia Unit 4, with only a 20
percent NOX reduction obtained on Magnolia Unit 3. The reasons for the differences
in baseline NOX emissions between the Magnolia units are discussed below under
Projecting NOX Emissions.
For both Olive Unit 1 and Magnolia Unit 4, the reduction in NOX emissions was
uniform throughout the load range. The reduction achieved on Magnolia Unit 3,
however, was quite variable with no reduction achieved at low loads. This is partially
due to low baseline NOX and also due to an error made in selecting the burner size.
The procurement specification for Magnolia Unit 3 was based on the existing burner
sizes which were higher than required. This resulted in the oversizing of the AUS-
DFL burners. This led to very low register draft loss (RDL) and poor fuel and air
mixing, causing deterioration in NOX and excess O2 performance. At present,
remedies are being implemented to correct the low RDL for Magnolia Unit 3. With
the RDL at original design conditions, it is expected that reduced NOX emissions and
low excess O2 operation will be achieved.
00001-R1511-DLR
-5-
-------
Data presented above were obtained without LNB performance optimization.
Additional reductions in NOX emissions, especially at lower loads, are expected upon
completion of optimization testing. Optimized LNB performance is therefore certain
to reduce boiler emissions throughout the load range.
With the exception of Magnolia Unit 3, the LNBs met and exceeded the specification
requirements. Presented in Table 3 are NOX levels achieved, emission limits
guaranteed by AUSCC, and limits contained in the CITY specifications. In every
case, the achieved NOX emission levels are well within the CITY specifications.
The NOX levels obtained by the AUS-DFL are the lowest ever achieved in a utility
boiler application without the use of FOR and OFA. Taking into consideration the
BZHR of the retrofitted boilers, it is reasonable to state that the AUS-DFL has set a
new industry performance standard for this class of LNBs.
NOX data were only obtained with natural gas firing. The regulations in Southern
California allow oil firing only during natural gas curtailment days. In addition to the
regulatory requirement, the CITY is in the process of modifying its oil storage and
handling facility to comply with state and federal environmental regulations. This
precluded the City's ability to fire oil in the retrofitted boilers.
Effect of Fuel Injection Parameters
Historically, burner optimization objectives emphasized improved flame stability and
control of flame geometry, as well as operation at low excess O2. These objectives
were prompted by economic incentives to improve the combustion efficiency and
enhance flame compactness and, in turn, reduce the cost of steam boilers. The
optimization objectives were mostly met by the optimization of combustion air flow
field.
As the need to reduce NOX emissions emerged, establishing a delicate balance
between satisfying the operational requirements above and providing low NOX
emissions was required. Intense mixing was used to enhance flame stability, reduce
flame volume (provide furnace compactness), and operate at ultra low excess O2
levels. However, intense mixing resulted in high NOX emissions. Therefore, a
compromise was required to satisfy the competing goals of controlling NOX emissions
and maintaining the operational improvements which have been made.
Adopting an evolutionary approach to burner development, HAMWORTHY decided
to build the platform off an existing burner which was optimized to satisfy the
operational requirements established in earlier decades. The HAMWORTHY burners
already offered an industry endorsed track record with regard to flame stability,
00001-R1511-DLR
-------
flame compactness, and low excess O2 operation. The challenge was to retain these
attractive operational features yet achieve low NOX performance.
Since the combustion air flow field was fully optimized to achieve the burner
operational goals discussed above, HAMWORTHY shifted its development focus to
fuel injection parameters. Retaining the combustion air flow field characteristics
provided a degree of assurance that the operational advantages of the HAMWORTHY
burner would not be lost. Building on a strong foundation of exceptional operational
characteristics offered a higher potential of success for reconciling the requirements
of achieving low NOX and maintaining exceptional operational characteristics.
The approach to optimizing fuel injection parameters emphasized combustion staging.
The design objective was to establish an internally staged flame structure, operating
in an overall oxidizing furnace environment. Combustion staging was achieved for
both oil and gas firing by selecting the fuel injector geometry and orientation. As
discussed above, baffles in the fuel injection region were used to augment staging and
enhance flame stability.
The design enhancements of the HAMWORTHY burners to achieve low NOX were
developed in HAMWORTHY's own research and development facility. The
development efforts were performed at the scale of 60 MMBtu/hr. For gas firing, two
poker designs were identified to offer exceptional flame stability and low NOX
emissions. Poker configuration A provided slightly better flame geometry and slightly
higher NOX as compared to poker configuration B. Based on these results, poker
configuration A was selected for installation in the CITY boilers.
Evaluating the poker configuration A in the CITY's Olive Unit 1 was disappointing.
The burner achieved little, if any, reduction in NOX emissions. This led to the
introduction of poker configuration B, which achieved the desired NOX reduction
while maintaining other desirable burner operational characteristics.
The impact of fuel injection parameters on NOX emissions as identified in Olive Unit
1 is presented in Figure 3. As discussed above, poker configuration A offered little
or no NOX reduction from baseline NOX. Poker configuration B, however, resulted in
NOX reduction ranging between 40 and 55 percent. The achieved reduction in NOX
emissions represents solely the impact of poker design changes. The result
emphasized the significance of fuel-air mixing on NOX emissions, and how this design
parameter warrants an increased focus in order to achieve ultra low NOX levels.
While, for proprietary reasons, the design details of poker configurations A and B
cannot be revealed, it is important to note that poker configuration B provided a
higher degree of combustion staging than poker configuration A. In both cases, the
00001-R1511-DLR
-------
flame structure remained the same. However, the rate of mixing between combustion
air and fuel was significantly altered.
No significant difference in NOX emissions between poker configurations A and B was
noticed in the HAMWORTHY research and development facility. This points to a
relationship between boiler BZHR and NOX emissions which is independent of fuel-air
mixing. The impact of BZHR on NOX emissions is discussed in detail below. This
discussion will show that the low BZHR used in the HAMWORTHY test facility
diminished the impact of mixing, making the two poker configurations display
comparable NOX emissions performance.
Effect of Furnace Combustion Staging
Furnace combustion staging is a process modification influencing the overall
stoichiometry of a flame zone, rather than local combustion stoichiometry within the
flame structure. It is implemented with the use of OFA or operation with selected
burners-out-of-service (BOOS) and registers open. Irrespective of the approach used,
furnace combustion staging involves the bypassing of a portion of combustion air
from the burner to other parts of the furnace. The burners are thus operated under
fuel-rich conditions controlling NOX formation. Combustion air bypassed around the
burner is introduced to mix with products of combustion within the boiler furnace and
achieve complete fuel burnout. In general, furnace combustion staging offers 20-40
percent reduction in NOX emissions for conventional burners and LNBs.
The effect of combustion staging on NOX emissions from the AUS-DFL was
investigated for Magnolia Unit 4. NOX data were obtained with five burners-in-service
and one BOOS register closed. The tests were then repeated with the same BOOS, but
with the register open. NOX and CO emissions were measured throughout the load
range. The results of this testing are presented in Figure 4.
The data show that furnace combustion staging produced a very modest reduction in
NOX emissions, averaging significantly less than 10 percent. This result is totally
inconsistent with data obtained on conventional burners and LNBs where combustion
staging consistently offered NOX reductions in excess of 20 percent. The results show
that the AUS-DFL's flame structure is internally staged, and further furnace staging
reaches a point of diminishing return. This result offered valuable insight into the
design characteristics of the AUS-DFL, confirming that the burner flame structure is
indeed internally staged.
00001-R1511-DLR -8-
-------
Projecting NOX Emissions
Total NOX produced in a utility boiler application consists of thermal and fuel bound
NOX. Thermal NOX formation is determined by both combustion stoichiometry and
furnace thermal environment, whereas NOX is predominantly influenced by fuel
nitrogen content and combustion stoichiometry.
This technical paper only discusses issues related to thermal NOX formation. An
analysis was performed to isolate thermal NOX generated in the near burner zone from
NOX produced in the furnace bulk gas. Then an approach to projecting LNBs
performance was developed.
Figure 5(a) presents baseline NOX versus BZHR. The figure shows complete scatter
of the data with no correlation present between NOX emissions level and BZHR. The
data conclusively shows that baseline NOX emissions are not influenced by furnace
thermal environment, but, rather, by burner design considerations. This explains the
significant difference in baseline NOX for Magnolia Unit 3 and NOX levels measured
for Magnolia Unit 4 and Olive Unit 1.
In Figure 5(b), NOX emissions obtained in the same boilers with LNBs were plotted
against BZHR. Plots were prepared for poker configuration B as tested in all three
boilers, and poker configuration A as tested in Olive Unit 1.
By eliminating burner design as a random variable, as is the case for baseline data,
LNBs displayed a strong correlation of NOX emissions to BZHR. As expected,
however, the correlation is burner specific. The slope of the correlation provides an
indication as to the sensitivity of NOX production to boiler thermal environment. The
slope is higher for poker configuration A, indicating a greater dependence of NOX
emissions on boiler thermal environment. A more moderate slope was obtained for
poker configuration B, reflecting the effectiveness of combustion staging as produced
by poker configuration B in controlling NOX formation.
The NOX performance difference between poker configurations A and B diminishes as
the BZHR is reduced. For the HAMWORTHY test facility where the BZHR is
approximately 0.055 MMBtu/hr/ft2, the correlation suggests that NOX emissions from
the two poker designs will be essentially identical. This is consistent with results
obtained in the HAMWORTHY test facility which showed that only a slight
improvement in NOX emissions is offered by poker configuration B. As the furnace
BZHR is increased, the difference in slope for the two correlations results in a
significant (more than double) difference in stack NOX achieved.
00001-R1511-DLR
-9-
-------
Figure 5(c) compares baseline NOX data to data obtained in the same boilers with
LNBs. The comparison clearly shows that the use of LNBs completely eliminates the
scatter in NO, dependence on BZHR. In addition, with LNBs, the BZHR impact on
NOX emissions is significantly reduced, especially with the use of increased
combustion staging (poker configuration B). This result is important for several
reasons. It illustrates that with LNBs, burner design variables influence NOX
emissions from a utility boiler application. Since, in every application, a designer has
significant control over burner design variables and no control over furnace
geometry, an opportunity exists through burner design to maximize NOX reduction. In
addition, by adhering to certain burner design variables, NOX emissions results can be
reproduced in a broad range of boilers.
Conclusions
Based on the results presented in this paper, the following conclusions can be
reached:
• Significant NOX reduction can be achieved with LNBs without requiring
combustion process modifications, such as FOR or OF A;
• Understanding the details of fuel and air mixing is critical to achieve
the desired NOX reduction. Combustion staging internal to the flame
structure is effective in satisfying the competing requirements of
reducing NOX emissions and maintaining acceptable operational
performance;
• NOX emissions from LNBs have a strong dependence on BZHR. Such
dependence is reduced with the use of combustion staging for NOX
control;
• At low BZHR, the impact of burner design variables on NOX emissions
diminishes. Such an impact becomes very pronounced, however, at high
BZHR;
• In LNB applications, burners, more so than furnace design parameters,
influence NOX emission levels. Since, in most applications, burner
design parameters are totally under the control of the designer, NOX
emission projections for new and retrofit boiler applications can be
made with a high level of confidence;
00001-R1511-DLR -10"
-------
• For conventional burners, the correlation of NOX emissions to BZHR is
poor due to the strong and random influence of burner design on
measured NOX levels.
Acknowledgments
The authors wish to acknowledge the technical support and guidance received from
the City of Burbank staff. Mr. Vince Gustafson, mechanical engineering section, and
Mr. Harley Hansen of the power plant office, have provided effective project
management support and coordination. The technical contributions made by Messrs.
Scott Martin, Ernie Chiong and Ralph Hawley are gratefully acknowledged. The
technical support offered by Mr. Larry Gibson of InSight Automation is also hereby
acknowledged. The success of this project would not have been possible without the
overwhelming support offered by the City of Burbank plant staff. Their kind
assistance and support are very much appreciated.
00001-R1511-DLR -11-
-------
TABLE 1. SCOPE OF SUPPLY AND SERVICES
Scope of Services
Unit
Name
Olive
Magnolia
Magnolia
Unit Fuel
No.
1 Gas/Oil
3 Gas/Oil
4 Gas/Oil
MW Design Cold
flow
model
44 x x
20 x x
30 x x
Con- Start-
struc- up
tion
x x
X X
X X
Scope of Supply
LNBs BMS Ignitors Flame
Scanners
x x x x
x x x x
x x x x
-------
TABLE 2. SUMMARY OF DESIGN CHARACTERISTICS OF RETROFITTED BOILER
Steam Specifications Burner Firing
Unit
Name
Olive
Magnolia
Magnolia
Unit MW Firing No. Mlb/ Temp.
No. Config. of hr °F
Burners
Unit 1 44 Front 6 425 955
Wall
Fired
UnitS 20 Front 6 215 890
Wall
Fired
Unit 4 30 Front 6 315 910
Wall
Fired
Configuration
Pres. Comb. BZHR No. Burners
psig Air MMBtu/ of per
Preheat hr/ft2 Rows Row
1,280 540 0.26 2 3
900 520 0.34 2 3
1,000 550 0.23 2 3
-------
TABLE 3. SUMMARY OF NOX EMISSIONS PER PROCUREMENT
SPECIFICATION, VENDOR GUARANTEE AND ACTUAL
ACHIEVED (ppmv, dry @ 3%
Unit
Name
Olive 1
Magnolia 4
Magnolia 3
Procurement
Specifications
130
140
100
AUSCC
Guarantee
105
100
80
Actual
Emissions
105
78
89*
* Not optimized
-------
tMBOK <*»»•£ rM .1
NiPLMi ou . I
(OOFBOLIS .1
SUE OF no. is .«
pea w m is , •
SECTION A-A
Figure 1. Schematic of the AUS-DFL LNB
VIEW ON ARROW B
-------
NOX, ppm (@ 3%
NOX, ppm (@ 3% O2)
200
150
100
50
•*• Baselln*
-•"AUS-DFL
20 40 60 80
Load, percent
Olive Unit 1
100
250
200
150
100
20
40 60 80
Load, percent
Magnolia Unit 4
100
NOX, ppm (@ 3%
NOX Reduction, percent
120
110
100
90
80
70
•*• AUS-DFL
20 40 60 80
Load, percent
Magnolia Unit 3
100
100
80
60
40
20
» Olivs Unit 1
-*- Magnolia Unit 3
•>k- Magnolia Unit 4
20 40 60 80
Load, percent
NO, Reduction, percent
100
Figure 2. NOX Reduction on Gas Firing
without Flue Gas Recirculation
-------
NOX, ppm (@ 3% Oa)
200
150
100
50
* Baseline
+ Poker Conflg. A
* Poker Conflg. B
20 40 60 80
Load, percent
NOX versus Load
100
NOX Reduction, percent
60
40
20
' Poker Conflg. A
' Poker Conflg. B
20 40 60 80
Load, percent
NOX Reduction, percent
100
Figure 3. Effect of Poker Configuration on
NOX Emissions from Gas Firing-Olive Unit 1
-------
150
250
CM
Slide Open
Air Slide Closed
UL BOOS - Poker Config. B
0 20 40 60 80 100
Load, percent
Figure 4. Effect of Furnace Combustion Staging on NO,
Emissions - Poker Configuration B
-------
NO,, ppm (© 3%
300
200
100
Aoilv* Untt 1
• Magnolia Unit 3
T Magnolia Unit 4
0.1 0.2 0.3
BZHR, MMBtu/hr.ft2
0.4
(a) Isolating Burner and Furnace Effects on Correlating
NOX Emissions
NO,, ppm (@ 3% O2)
300
200 --
100 -
0.1 0.2 0.3
BZHR, MMBtu/hr.ft2
(b) LNBs NOX Emissions Correlations
0.4
NO,, ppm (@ 3% O2)
300
200 -
100 -
0 0.1 0.2 0.3
BZHR, MMBtu/hr.ft2
(c) Correlating NOX Emissions to BZHR
Figure 5. Interdependence of Furnace
and Burner Designs in Determining
NOX Emissions
-------
REACH: A LOW-COST APPROACH TO REDUCING
STACK EMISSIONS AND IMPROVING THE
PERFORMANCE OF OIL-FIRED BOILERS
D. V. Giovanni
M. W. McElroy
S. E. Kerho
Electric Power Technologies, Inc.
830 Menlo Ave, Suite 201
MenloPark,CA 94025
Abstract
Improved oil combustion technology, based upon optimization of oil atomizer and
flame stabilizer design, has been retrofit to oil-fired boilers to reduce NOx emissions,
particulate matter emissions, and opacity, and to provide operational and
performance benefits. This technology, referred to as REACH, can be retrofit to wall-
fired and tangential-fired boilers at a cost of less than $0.75/kW, a fraction of the cost
of installing new burners. The technology is compatible with conventional NOx
controls such as overfire air, flue gas recirculation, and low-NOx burners, and can be
combined with these techniques to further reduce NOx emissions. REACH has been
applied to eighty boilers representing over 14,000 MW of generating capacity. This
paper describes REACH technology, its applicability and cost, and the emissions and
performance results achieved in full scale applications.
Background
Improved oil combustion technology, based upon optimization of oil atomizer and
flame stabilizer design, has been developed for retrofit to oil-fired utility boilers.
This technology is referred to as Reduced Emissions and Advanced Combustion
Hardware, or REACH. Over the past several years, REACH has been widely applied
to reduce particulate matter (PM) emissions and opacity, improve unit turndown,
reduce excess air requirements, and solve a variety of boiler performance, reliability,
and operating problems. Recent research and development1'2-3 have led to
refinements in REACH technology that simultaneously reduce NOx and PM
emissions. REACH hardware is now available for retrofit to oil-fired utility and
industrial boilers to cost-effectively control emissions and provide operational and
performance benefits.
REACH has been developed jointly by the Empire State Electric Energy Research
Corporation (ESEERCO), Consolidated Edison Company of New York (Con Edison),
the Electric Power Research Institute (EPRI), and Electric Power Technologies, Inc.
-------
(EFT). Niagara Mohawk Power Corporation also provided funding support and a
host site for REACH prototype testing.
Description of REACH
REACH consists of new flame stabilizers and oil atomizers that constitute the critical
hardware in any oil-fired burner which must be optimized to produce the
combustion conditions necessary for reduced emissions and improved burner
performance.
As depicted in Figure 1, flame stabilizers create a recirculating flow pattern in which
hot combustion products are recirculated back toward the burner, providing a high-
temperature, low-velocity region which promotes faster volatilization and ignition
of the fuel oil droplets. This recirculating flow pattern is referred to as the burner's
internal recirculation zone (IRZ). Controlling the size of the IRZ, the location of the
fuel spray relative to its boundaries, and the amount of air available to the fuel spray
provides an opportunity to control the near-burner air/fuel ratio, which can then be
optimized to control NOx and PM emissions. The technical approach for REACH
was to exploit this opportunity through the use of improved and novel oil atomizer
and flame stabilizer designs.
Internal
Recirculation
Zone
Flame Front
Oil Spray
Combustion Air
Critical Hardware
Flame Stabilizer
Oil Atomizer
Figure 1
Critical Combustion Components of an Oil-Fired Burner
2-
-------
The design philosophy used in developing REACH was to produce technology that:
(1) could be retrofit to the range of burner and boiler designs to maximize
applicability, (2) retain as much as possible of the original burner in order to realize
cost advantages of REACH relative to other retrofit options (e.g., complete burner
change-out), and (3) build upon the successful combustion hardware designs
previously applied by EFT, to capitalize on the operational, performance, and cost
benefits already demonstrated. Consistent with this philosophy, REACH adapts to
the major existing components of a burner. In the majority of retrofit applications,
REACH flame stabilizers and oil atomizers replace the existing stabilizers and
atomizers, while other burner components, including air registers, oil guns, flame
detection equipment, igniters, and control systems are left intact. REACH
technology may also be incorporated into the design of most new low-NOx burners.
Oil Atomization. The oil atomizers used in REACH are custom-designed to adapt
to the existing oil supply conditions and (where applicable) the existing atomizing
steam system. For steam-atomized systems, REACH uses an internal-mix (I-Mix)
atomizer which has been found to produce superior spray quality compared to other
common atomizer designs, such as the Y-Jet and Racer. The generic physical
configuration of the I-Mix atomizer is shown in Figure 2. For mechanically-
atomized burners, the REACH oil atomizer is designed to operate in the range of oil
supply and return pressures customarily used at the site.
Flame Stabilizers. For flame stabilization and aerodynamic control of fuel and
air mixing, REACH uses a compound-curved-vane swirler (CVS) for applications on
both wall- and tangential-fired boilers. A schematic of a CVS is shown in Figure 2.
The CVS provides better performance than conventional diffusers and flat-bladed
swirlers that are commonly in use. The CVS flame stabilizers supplied with
REACH are custom-designed to produce the proper entrainment and swirl of
combustion air at the discharge plane of the burner, and to match the oil spray of the
REACH oil atomizer.
Two Versions of REACH
Two versions of REACH have been defined on the basis of combustion
improvement and emissions reduction objectives. Combustion Performance
REACH (CP-REACH) is designed to reduce PM emissions and opacity and to provide
operational improvements including increased burner turndown, reduced excess air
requirements, improved flame stability, and elimination of flame impingement on
furnace walls. Low-NOx REACH (LN-REACH) is specifically aimed at retrofit
projects where NOx reduction is the major goal. The key difference between CP-
REACH and LN-REACH is the design of the oil atomizer. Boilers equipped with CP-
REACH can be easily converted to LN-REACH.
-3-
-------
BACK PLATE
\
RETAINING NUT
Internal-Mix
Oil Atomizer
(Approx. 1/2 scale)
IGNITOR PENETRATION
SHROUD
SWIRLER VANE
OFLOW
HUB
VIEWA-A
VANE CROSS SECTIONS
(ENLARGED)
AT SHROUD
Compound-
Curved-Vane
Swirler (CVS)
(Approx. 1/10 scale)
CURVED VANES EOUI-SPACED
FLOW
Figure 2
REACH Internal Mix Oil Atomizer and Compound Curved Vane Swirler
CP- REACH . The approach of CP-REACH is to improve the atomization of the oil,
and to produce flow patterns in the burner flame that achieve intense fuel and air
mixing for high combustion efficiency and proper flame shape. In general, REACH
atomizers are designed to produce an oil spray having an average drop size smaller
than typical atomizers.
For best performance over the boiler load range with steam-atomized burners, it is
preferred that the I-Mix atomizer be operated in a constant steam-to-oil differential
pressure mode, and with steam-to-oil mass ratios less than 10 percent at the
maximum burner firing rate. In some retrofit applications, it may be required to
modify steam pressure regulators and valving to provide the desired steam
conditions at the oil gun. For mechanically atomized oil guns, the REACH atomizer
is designed to operate in a spill-return or non-return (simplex) mode, consistent
with the existing oil supply system design (e.g., minimum oil pressure
requirements).
-4-
-------
LN-REACH. The approach of LN-REACH is to incorporate low-NOx oil
atomization with less intensive aerodynamic mixing patterns as compared to CP-
REACH technology. To a major degree, the combustion improvements, reduced
PM emissions, and opacity achieved with CP-REACH can still be realized, while
reducing NOx emissions. This is accomplished by maintaining the same spray
quality (i.e., Sauter Mean Diameter) standards of the CP-REACH atomizer, but
altering the spatial distribution of oil droplets in the oil spray.
While the atomizers used with CP-REACH are designed to produce a uniform or
symmetrical distribution of oil spray around the axis of the atomizer, the LN-
REACH atomizer is designed to produce a non-uniform oil spray, in which the
spray is divided into distinct segments at the base of the flame. The multiple fuel
spray segments, when combined with the air flow patterns of a properly matched
compound-curved-vane swirler, create localized staging of the combustion
process that contributes to the NOx reduction capability of LN-REACH. The
localized-staging flame pattern is designed to produce lower NOx emissions
without the typical increase in particulate matter emissions or flame length. The
atomizer design that produces the lowest NOx emissions in combustion
applications to date is an internal-mix design referred to as the Segmented V-Jet
atomizer (patent applied for).
REACH Design Guidelines
Boiler design and operating parameters that affect the design of REACH hardware
include the following:
• Burner capacity (MBtu/h).
• Oil properties, including Coking Index, vanadium, nitrogen, sulfur, and ash
that affect NOx and particulate formation/burnout rates.
• Burner throat diameter and contour.
• Burner throat velocity distribution.
• Partitioning of combustion air between inner and outer zones for dual
register burners.
• Range of adjustability of combustion air swirl with existing air registers.
• Oil supply conditions and atomizing steam supply conditions (if applicable).
• Configuration of gas firing equipment.
• Burner zone heat release rate (Btu/h-ft2).
• Upper furnace residence time for carbon burnout.
• Configuration of ignition and flame scanning equipment.
• Oil gun design.
• Furnace geometry and potential for flame impingement.
• Other NOx controls such as overfire air, flue gas recirculation to the windbox
and burners-out-of-service.
Generalized guidelines have been developed that provide REACH design
specifications which will be applicable in the majority of retrofit cases, including:
-5-
-------
• Venturi (axial-flow) burners
• Single-register, swirl-stabilized burners
• Dual-register burners
• Tangential-fired burners
The REACH design specifications may be modified to adapt to site-specific
conditions, based upon an engineering evaluation of site-specific boiler design and
operating parameters. In addition, one-time burner adjustments are made in the
field to optimize REACH performance. Such adjustments may include: swirler
axial position (oil gun guide tube position), air swirl, partitioning between air zones,
orientation of gas fuel injectors, oil and atomizing steam differential pressure
settings, and the position and air supply for ignitors. The capability to adjust these
parameters is normally incorporated into the design of existing burners or, if
necessary, can be added during retrofit of REACH.
Field Test Results
CP-REACH technology has been widely applied to utility boilers to improve
combustion performance, and reduce PM emissions and opacity. Recently, LN-
REACH has been installed to provide simultaneous reductions in NOx and PM
emissions, while retaining the advantages of CP-REACH. Examples of CP-REACH
and LN-REACH retrofit projects are described below.
CP-REACH
CP-REACH has been applied to over seventy utility boilers, ranging in size from 15
MW to 850 MW. Specific applications of CP-REACH to tangential-fired, opposed-
wall-fired, and single-wall-fired boilers are described below.
PM Reduction • Various Boilers. The PM reduction potential of CP-REACH is
illustrated in Figure 3 which compares PM emissions data taken with original
equipment hardware to data obtained after retrofit of CP-REACH. Data are
presented for wall- and tangential-fired units, steam and mechanical atomizers, and
a wide range of fuel oils. The reductions in PM emissions ranged from 40-60%, and
are primarily the result of improved combustion efficiency which results in lower
unburned carbon in the particulate.
Opposed-Wall-Fired, 550 MW Boilers. CP-REACH hardware was used to
upgrade the combustion systems on two, 550 MW opposed-wall-fired boilers. The
boilers were equipped with dual-register burners which used bluff-body diffusers for
flame stabilization, and Y-Jet steam-atomizers operated in a racer mode (i.e.,
constant steam pressure). The boilers were equipped with overfire air ports (OFA)
and flue gas recirculation to the windbox (FGR) for NOx control.
Boiler operation when firing the normal 0.7% sulfur oil with original burner
equipment was characterized by high opacity (20-40 percent) during startup and "up"
-6-
-------
load ramps. Opacity exceedances during load ramps were severe, such that the rate
of load increase possible (~ 1 MW/min.) was too slow for load dispatch purposes.
Consequently, the boilers were relegated to hot, stand-by service, and their average
capacity factor was approximately 5 percent. CP-REACH hardware retrofitted to the
two units included new flame stabilizers, internal-mix atomizers, and pressure
gauges for the atomizing steam and oil supply piping to each burner. The atomizing
steam control system was also changed to operate in a constant steam-to-oil
differential pressure mode. After retrofitting CP-REACH, opacity was reduced to 8-
14 percent over the load range, and load ramps were possible without opacity
exceedances. Because of these improvements, the boilers were restored to load
dispatch service and capacity factors have increased significantly.
0.35
Original Atomizers/
Stabilizers
CP- REACH Atomizers/
Swirlers
Unit A
Boiler Type: T-Fired
Capacity, MW: 100
Atomization: Mechanical
Fuel Sulfur: 3%
UnltB
Wall-Fired
230
Steam
3%
Unite
T-Fired
900
Steam
1.5%
Unit D
T-Fired
410
Steam
1%
UnitE
Wall-Fired
600*
Mechanical
0.7%
' Unit Capacity 800 MW, test at 600 MW
Figure 3
Reductions in PM Emissions By Retrofit of CP-REACH on Five Oil-Fired Utility Boilers
-7-
-------
Figure 4 shows NOx emissions vs. load with CP-REACH. (NOx data with the
original burner hardware were not available.) The upper curve is the baseline
condition with OFA ports closed and a minimum FGR rate. The bottom curve
shows the lowest NOx emissions achieved with OFA and FGR. The recommended
REACH operating conditions, represented by the dotted line, were established to
maintain a margin of compliance with the 0.30 Ib/MBtu NOx emissions limit, using
OFA and prescribed FGR rates.
0.7
3
S
5
0.6-
0.5-
0.4-
O 0.3 +
z
0.2-
0.1-
CP-REACH
Baseline
OFA Closed
FGR = Min.
CP-REACH
Recommended
Operation
OFA Open
FGR Varying
NOx Limit -•
CP-REACH
Minimum NOx
OFA Open
FGR = Max,
100
200
300 400
Load, MW
500
600
Figure 4
NOx Emissions Versus Load on a 550 MW, Wall-Fired Boiler
Using CP-REACH with Overfire Air and Flue Gas Recirculation
Single-Wall-Fired, 86 MW Boiler. CP-REACH hardware was retrofitted to
increase burner stability at low loads and to improve boiler turndown. To optimize
the CP-REACH retrofit, the burner air registers were modified to reduce windbox-to-
furnace differential pressure and eliminate FD fan limitations caused by high
pressure drop across the existing burners which restricted maximum achievable
load. The CP-REACH components included new flame stabilizers, and spill/return
-8-
-------
oil atomizers. Following the CP-REACH retrofit, boiler turndown capability was
increased from 2:1 to 4:1, and maximum load was increased from 81 MW to 86 MW.
LN-REACH
LN-REACH offers reductions in NOx emissions in addition to the combustion
improvements possible with CP-REACH. LN-REACH has been applied to wall- and
tangential-fired boilers equipped with steam atomized oil guns. Ten boilers have
been retrofit, ranging in size from 10 MW to 550 MW. Work is in progress to
expand the application of LN-REACH to mechanical-atomized systems. Described
below are LN-REACH applications to steam-atomized existing burners and new,
low-NOx burners. Following these descriptions, the status of mechanically-
atomized LN-REACH hardware is discussed.
Upgrade Of Existing Burners. LN-REACH was applied at three steam send-out,
wall-fired boilers and at a tangential-fired boiler, all equipped with steam-atomized
oil guns. These boilers were equivalent in size to electric generating units ranging
from 15 to 150 MW capacity. CP-REACH was previously installed on these boilers to
improve combustion and opacity. LN-REACH consisted of replacing the CP-REACH
atomizers with low-NOx atomizers. NOx emissions results are summarized in
Table 1. With LN-REACH, opacity levels remained low, typically ranging from 2 to
7 percent. PM emissions with LN-REACH, measured only on Boiler A, were 0.03
Ib/MBtu.
Table 1
Summary of LN-REACH NOx Reductions at One Utility Company
Boiler
A
B
C
D
Boiler
Type
Wall
Wall
Tangential
Wall
Size
(klb/hr)
150
440
550
1,400
NOx (Ib/MBtu)
As Found REACH
0.397
0.420
0.195
0.492
0.225
0.256
0.151
0.375
NOx
Reduction
43%
39%
23%
24%
Application to New Low-NOx Burners — Single-Wall-Fired, 146 MW Boiler.
LN-REACH steam-assisted oil atomizers were installed on new, Primary Gas - Dual
Register Burners (PG-DRB burners)4 to solve a high opacity problem experienced
with these low-NOx burners when operated with overfire air and flue gas
recirculation. Results showed that the LN-REACH atomizers reduced PM and
opacity and maintained existing NOx emission levels, while reducing levels of
overfire air and flue gas recirculation. Although the intent of the LN-REACH
retrofit was not to reduce NOx emissions, the ability to solve the opacity problem
was directly related to the low-NOx characteristics of the atomizer, which produced a
-9-
-------
more favorable tradeoff between NOx, opacity, PM, and excess O2 when operating
with overfire air and flue gas recirculation. Results are summarized in Table 2.
Table 2
Emissions and Performance Improvement with
LN-REACH Atomizers at 146 MW Wall-fired Boiler
NOx, Ib/MBtu
Particulate, Ib/MBtu
Opacity, %
Excess O2, %
O2 imbalance, % O2
Overfire Air, %
FGR to Windbox, %
Atomizing Steam/Oil
Fuel Nitrogen, Wt. %
Oil Gun Coking
Reference
0.19
0.07
18
3.2
1
26
20
>0.2
0.29
Daily
LN-REACH
Atomizer
0.19
0.06
10
3.5
<0.5
20
11
0.14
0.27
None
Application to New Low-NOx Burners—Opposed-Wall-Fired, 550 MW Boilers.
LN-REACH atomizers were installed on three, 550 MW, opposed-wall-fired boilers
recently retrofitted with new, axial-flow, low-NOx burners designed for gas- and oil-
firing. The oil atomizers supplied with the new burners were Y-Jet steam-atomizers
operated in a racer mode (i.e., constant steam pressure). The Y-Jet atomizers were
replaced with LN-REACH atomizers to provide lower NOx emissions and opacity.
The atomizing steam control system was also changed to operate in a constant
steam-to-oil differential pressure mode.
The boilers were equipped with overfire air ports (OFA) and flue gas recirculation to
the windbox (FGR) for NOx control. Figure 5 shows NOx emissions vs. load
following the LN-REACH retrofit. LN-REACH data are presented for three
operating conditions: (1) baseline without OFA or FGR, (2) with OFA and no FGR,
and (3) with OFA and 10% FGR. Also shown for comparison are CP-REACH data
that were obtained on one of the boilers prior to the new burner installation (the CP-
REACH retrofit to the original burners included new flame stabilizers and internal-
mix atomizers). Baseline NOx emissions at full load with LN-REACH were 0.32
Ib/MBtu, as compared to baseline emissions of 0.47 Ib/MBtu with CP-REACH.
Opacity levels were less than 5%. With LN-REACH, for the first time FGR was not
necessary to achieve compliance with the NOx limit of 0.30 Ib/MBtu. At loads up to
480 MW, NOx compliance was achieved without FGR or OFA. Above this load,
compliance was achieved by only opening the OFA ports.
10-
-------
3
s
x
O
Original Burners
CP-REACH
OFA FOR
-Closed Min.
,/Open Min.
New Burners
with LN-REACH
OFA FGR
"Closed Min.
Open Min.
Open 10%
0.2-
0.1-
100
200 300 400
Load, MW
500
600
Figure 5
NOx Emissions vs. Load on a 550 MW, Wall-Fired Boiler Using
LN-REACH with Overfire Air and Flue Gas Recirculation
Low-NOx Mechanical Atomizers. The approach in designing a low-NOx
mechanical atomizer was to produce an oil spray with properties equivalent to the
low-NOx steam atomizer design. To accomplish this, the mechanical atomizer
would require a multiple-exit spray plate. A number of design concepts were
evaluated in a cold flow spray laboratory. Based on spray quality and spray size
distribution criteria, the best designs were selected and prototype atomizers were
designed for combustion tests on a 90 MW wall-fired boiler and a 90 MW tangential-
fired boiler. The primary objectives of the tests were to: (1) confirm that a multiple-
exit design had combustion characteristics (e.g., flame shape and flame stability) that
were compatible with a conventional burner; (2) quantify NOx reduction potential
and impact on stack opacity; and (3) determine where improvements were necessary
such as optimization of spray angle. Secondary objectives were to evaluate the
impact on burner turndown, PM emissions, and combustion efficiency.
-11-
-------
A comparison of NOx emissions between baseline CP-REACH atomizers and the
prototype low-NOx atomizers for both test boilers is presented in Figures 6 and 7. At
an 80 MW reference test condition, the prototype atomizers achieved NOx
reductions of 39% and 21% for the wall-fired and tangential-fired boilers
respectively. CO emissions and opacity with the low-NOx atomizers were
equivalent to baseline levels. At the time of this writing, the PM samples were still
being analyzed.
Projects in Progress
The following LN-REACH retrofit projects are in progress:
Unit MW Boiler Design Oil Atomization Other NOx Controls
A 600 Tangential-fired Mechanical None
B 600 Opposed Wall-fired Mechanical OFA/FGR
C 350 Tangential-fired Mechanical OFA
D 865 Single Wall-fired Steam OFA/FGR
Results from these projects will be available by the end of 1995.
Cost
The cost of CP-REACH and LN-REACH hardware includes two major components:
(1) one-time engineering for analysis of the existing burner aerodynamics and
design of the custom oil combustion hardware, and (2) fabrication and delivery to
the plant of custom-designed atomizers and flame stabilizers (if needed), including
installation instructions. Further, limited startup support to assist with equipment
installation and performance optimization of the combustion equipment is
advisable in most cases. Equipment installation is conducted by the utility company
in conjunction with scheduled maintenance outages. As part of the installation, it
may be necessary to modify the control system from constant atomizing steam
pressure to constant steam-to-oil differential pressure. In most cases, installation
costs will be equivalent to, or less than, the equipment capital cost.
The total cost for design, fabrication and delivery, installation, and startup support
for REACH technology is $0.75/kW, or less. In comparison, the cost for low-NOx
burners is estimated to be 20 to 30 times greater than LN-REACH (i.e., $10-15/kW).
In some instances, more extensive modifications may be warranted as part of the
REACH retrofit, such as upgrading of the fuel oil supply system, ignitors, flame
detectors, and burner assemblies. Modifications of this nature will increase the
initial REACH retrofit costs accordingly. Replacement frequencies for REACH oil
atomizers are projected to be similar to existing designs.
-12
-------
0.8
0.7-
£0.6
2
§ 0.5-
w
I 0.4-
V)
I 0.3-j
UJ
O 0.2-1
0.1-1
0
Baseline
(CP-REACH)
LN-REACH
t i i i i i
10 20
30
40 50 60 70
Unit Load, MW
Figure 6
NOx Comparison of CP-REACH and LN-REACH
Mechanical Atomizers on a 90 MW Wall-Fired Boiler
80 90 100
0.5
3
+*
ffl
0.4-|
T.0.3-
v>
c
o
n o
•F 0.2-
UJ
X
O
Baseline
(CP-REACH)
LN-REACH *•
10 20 30 40 50 60 70 80 90 100
Unit Load, MW
Figure 7
NOx Comparison of CP-REACH and LN-REACH
Mechanical Atomizers on a 90 MW Tangential-Fired Boiler
-13-
-------
Summary
The potential for low-cost atomizer and flame stabilizer retrofits to reduce emissions
and improve performance has been demonstrated with the application of REACH
technology to more than eighty boilers (both wall- and tangential-fired designs,
steam and mechanical atomizers) representing over 14,000 MW of generating
capacity. The majority of these retrofits have involved the application of CP-
REACH and have resulted in improved combustion performance, including
reduced opacity and PM emissions, improved turndown, and reduced excess air
requirements. In some of these retrofits, CP-REACH has also reduced NOx
emissions. LN-REACH has been applied to wall- and tangential-fired boilers
equipped with steam atomized oil guns. The NOx reductions achieved with the
technology have ranged from 20-40% with no increase in opacity levels. Prototype
testing of low-NOx mechanical atomizers has recently been completed with
preliminary results indicating NOx reductions consistent with those seen using
steam-atomized LN-REACH. Additional demonstrations on utility boilers equipped
with steam- and mechanically-atomized oil guns are scheduled in the near future.
References
1. ESEERCO Project EP 90-17, "Development and Evaluation of Combustion
Hardware for Controlling NOx and Particulate Matter" (final report in progress).
Project co-funded by EPRI, Consolidated Edison Company of NY, Niagara
Mohawk Power Corporation, and Electric Power Technologies, Inc.
2. ESEERCO Project EP 85-15, "Heavy Oil Combustion Research." Project co-
funded by EPRI, Consolidated Edison Company of NY, and Electric Power
Technologies, Inc. Documentation includes:
(a) Hersh, S et al. Coke Formation Index: A Measure of Particulate Formation
in Oil Combustion. EPRI Report GS-6714, February 1990.
(b) Kerho, S. E. and D. V. Giovanni. Factors Affecting NOx Emissions in Heavy
Oil Combustion. EPRI Report GS-7353, June 1991.
3. Kerho, S. E. and D. V. Giovanni. Atomizer and Swirler Design for Reduced
NOx and Particulate Emissions. NOx Controls for Utility Boilers Workshop.
July 1992 EPRI Conference, Boston Mass.
4. Kerho, S. E., et al. "Reduced NOx, Particulate, and Opacity on the Kahe Unit 6
Low-NOx Burner System." Proceedings: 1991 Joint Symposium on Stationary
Combustion NOx Control. EPRI Report GS-7447, Volume 2, November 1991.
-14-
-------
DEVELOPMENT AND APPLICATION OF A LOW NOx,
HIGH EFFICIENCY ATOMISER FOR OIL AND EMULSION FUELS
P Baimbridge
M Garwood
A R Jones
PowerGen pic
Power Technology Centre
Ratcliffe on Soar
Nottingham NG11 OEE
England
Abstract
During the 1970's and 1980's the Central Electricity Generating Board (CEGB)
undertook an extensive development programme aimed at improving the combustion
performance of its oil fired utility plant. This work culminated in the development of
the Optimised Pressure Jet Atomiser (1) and the Steam Atomised F-Jet (2).
Subsequently a licence was granted to BP (UK) Limited to develop further the F-jet
atomiser for firing difficult refinery fuels.
On the privatisation of the UK electricity supply industry the patent rights of the F-jet
atomiser were assigned to PowerGen pic. The introduction of NOx emission
legislation for both industrial and utility scale liquid fuel fired plant has prompted
PowerGen pic to develop further the F-jet atomiser to provide a low NOx capability to
reduce NOx abatement costs.
The paper presents an overview of spray rig tests and pilot scale combustion trials
undertaken to develop the F-jet into a high efficiency, low NOx atomiser designated
the Advance F-jet (AFJ). Test results from the application of AFJ atomisers to
industrial and utility plant, firing oil and emulsion fuels, are included.
Burner Spray Rig Trials
The F-jet atomizer, figure 1, was developed by the CEGB to overcome the apparent
shortcoming of the Y-jet type of twin fluid atomizer fitted to a wide range of oil fired
plant. The F-jet concept was further developed by BP (UK) Ltd to provide an atomizer
design with more predictable and improved performance. The BP development was
designated the Advanced Toroidal Atomizer (ATA), figure 2.
-------
The ATA was developed for application on refinery burners and as such had only been
tested over a relatively small size range (up to approximately 9MWt) and with high
steam:oil mass ratios (approximately 0.15:1). Therefore it was decided to undertake
a series of parametric tests to investigate the suitability of the ATA geometry for
power station burners and to identify optimum design guidelines. These parametric
studies were undertaken at 75MW (3) and 45MW(4) thermal ratings.
The tests were carried out on the Power Technology Oil Burner Spray Rig. The rig can
simulate oil flows in excess of 8 m3/hr and oil pressures up to 60 bar. Shell Vitrea 22
oil is used to simulate heavy fuel oil, providing a viscosity under normal operating
conditions of approximately 35 cst. For twin fluid atomisers air is used as the
atomising medium.
The oil is sprayed into a large perspex chamber, maintained under suction by an
exhauster fan, the exhausted air/oil mist being drawn through a small electrostatic
precipitator to reclaim the oil the rig is equipped with pressure and flow measurement
facilities for both oil and air.
A phased doppler anemometer was used to measure both droplet sizes and velocities
approximately 18 inches downstream of the atomizer discharge ports. At each
position of the measurement volume the system collected data from approximately
2000 droplets (increased to 8000 droplets for the second series of tests), and a
traverse was undertaken to assess the whole spray. Measurements were taken at 2
cm intervals across one side of the hollow spray cone.
Three averages were collected for droplet size. These were D10, the number mean
diameter, D20 the area mean diameter, and D30, the volume (or mass) mean
diameter. These are defined as follows:-
D10 = (1/n) If DJ
D20 - [ (1/n) I, (D,)2]*
D30 = [ (1/n) I; (D;)3]*
Where D; is the diameter for the ith droplet.
n is the total number of droplets measured.
The Sauter Mean Diameter (SMD), commonly used to characterise sprays, is given by
(D30)3/(D20)2. However, as lubricating oil and air are used to test purposes the test
results are only used for comparison with each other.
The results of the spray rig tests indicated that:
-------
• Spray droplet sizes are relatively insensitive to mixing chamber volume.
• There is an optimum atomising medium:oil mass ratio, the value of which is
dependent on atomizer loading, ie the optimum mass ratio reduces with
reduced oil throughput and increases with increased oil throughput. Also, with
larger mixing chamber volumes the optimum mass ratio is more sensitive to oil
throughput. For the 45MWt reference atomizer design the optimum air:oil mass
ratio was approximately 0.08:1 under full load conditions.
• There is an optimum ratio of the geometric area of the discharge ports to the
sum of the geometric areas of the oil and steam port. If this ratio is reduced
significantly below the optimum a rapid deterioration in spray quality results.
• Spray cone angles are generally with ± 3 degrees of the geometric cone angle.
• The design discharge port length to diameter ratio is a good compromise
between spray quality and control of the spray cone angle.
The various relationships identified during the burner spray rig tests were used to
compile an atomizer design manual. The resulting atomiser incorporated the best
features from the F-jet and ATA atomizers, and was designated the advanced F-jet
atomizer (AFJ).
Figures 3 shows the values measured for D10, D20 and D30 for a range of atomizer
types designs for Grain PS which has 4 x 660MW units (burner rating 75MWt, heavy
fuel oil) and Ince PS which has 2 x 500MW units (burner rating 38MWt, Orimulsion).
The test operating conditions were:
Atomizer Oil Flow Air Pressure Air Flow
m3/hr bar g Nm3/hr
Grain F-jet ) ) 475
Grain ATA ) 6.6 ) 11.5 505
Grain AFJ ) ) 440
Grain Y-jet ) ) 470
Ince F-jet ) 5.17 ) 12.5 380
Ince AFJ ) ) 315
From examination of figure 3 it can be seen that the F-jet and AFJ provided similar
atomisation quality and were better than the ATA. All of the internal mix atomizers
were much superior to the Y-jet. The performance of the Ince AFJ was much superior
to the Ince F-jet, even though the atomising medium consumption was some 17%
lower.
-------
Based on these results it was concluded that the AFJ provides similar performance to
well designed F-jet, but the available design criteria are much more reliable for the
AFJ. (It should be noted that the AFJ designs tested did not incorporate features
identified during the burner spray rig tests and improved performance would be
expected from current atomizer designs).
Pilot Scale Combustion Trials
To investigate the NOx reduction potential of the F-jet concept pilot scale combustion
trials were undertaken on the 1 MWt Power Technology Combustion Test Facility. The
tests were arranged to investigate the effect of spray cone angle and radial fuel
placement on NOx formation, using a parallel tube air passage fitted with a fixed blade
flame stabilizer located on the oil burner carrier tube.
The relatively small size of the Combustion Test Facility precludes normal particulate
matter (PM) sampling (for subsequent carbon in grits determination) due to the time
necessary to obtain sufficient sample for analyses. Therefore a simplified, less time
consuming semi-quantitative approach was adopted. This entailed passing a constant
flow of flue gas through filter paper for preset time period. The reflectance of the
stain obtained on the filter paper was then measured and recorded. The relationship
between carbon in grits and reflectance was established from extended tests on a
single tip design, and used to estimate carbon in grits levels for other tip designs.
Consequently, quoted carbon in grits levels can only be considered to be indicative.
It was not possible to utilise this technique for Orimulsion tests as the stains obtained
change colour (and hence reflectance) due to absorption of moisture from the
atmosphere. No carbon in grits levels are stated for the Orimulsion firing tests;
although extensive experience on plant has shown that only under very poor
combustion conditions does carbon content become significant.
A transition duct temperature of 1050°C was chosen as a typical operating
temperature and recorded NOx emissions were corrected for the effect of
temperature. The actual correction was 8.4 mg/Nm3 for every 10°C change in
transition duct temperature, which is in line with previous experience.
The NOx emission graphs plotted are the best straight line fits for the data recorded.
The atomizer design used of the combustion trials featured 10 discharge ports.
Atomizer outer bodies with 60°, 80°, 90° and 100° geometric cone angles were
manufactured, together with alternating large/small discharge ports designs with a
90° cone angle. Additionally atomizers with alternating large/small discharge ports
with the small ports drilled on a 60° cone angle were tested.
The nomenclature used to describe atomizer designs is as follows:
-------
70/30-90/60 Geometric area of the large discharge ports is 70% of the total
discharge port area, and the area of the small discharge ports
30% of the total discharge port area. The large discharge ports
are drilled on a cone angle of 90°, and the small discharge ports
on an angle of 60°.
The test results obtained are presented on figures 4 to 8, the main points of interest
being:
• When firing fuel oil NOx emissions increased and carbon in grits decreased as
the cone angle was increased from 60° to 90°. On increasing the cone angle
to 100° both NOx emissions and carbon in dust reduced compared with a 90°
cone angle (figure 4).
• Adopting an alternating large/small discharge port arrangement produced NOx
reductions of up to 40% when firing fuel oil and 45% when Orimulsion firing
(figures 5 and 6). When fuel oil firing carbon in grits increased as the diameter
ratio of the large to small discharge ports increased.
• For fuel oil firing narrowing the cone angle of the small discharge ports
improved the NOx reduction obtained for a 70/30 discharge port arrangement,
but increased NOx emissions for the 90/10 pattern. There was no significant
effect on carbon in grits for the 70/30 design, however for the 90/10 design
the carbon in grits level was very low (figure 7). When Orimulsion firing
reducing the cone angle of the small ports increased NOx emissions for both
the 70/30 and 90/10 designs (figure 8).
Boiler Plant Experience
Advanced F-jets have been tested on a 80 t/hr boiler which provides process steam
at 17.8 bar and 321 °C. Four parallel tube air passages with fixed blade flame
stabilizers are located in a 2 by 2 array on the furnace front wall, supplied with
combustion air at ambient temperature. The original atomizer tips are steam assisted
pressure jets.
The low sulphur oil fired on the boiler has extremely low ash (0.001 % maximum) and
asphaltenes (0.5% maximum) contents. Fuel nitrogen content is typically 0.3% to
0.4%.
Tests were undertaken on the original atomizers and standard, "74-26" and "70-
30/90-60" low NOx advanced F-jet atomizers. The test results are presented on
figure 9.
-------
The "74-26" advanced F-jet atomizers reduced boiler NOx emissions by approximately
7% compared with the original atomizers, over a boiler excess oxygen range of 1%
to 2%. This is a lower NOx reduction than anticipated, possibly due to the relatively
low baseline NOx emissions level.
The "70-30/90-60" AFJs produced NOx emissions of approximately 400 mg/Nm3
independent of excess oxygen level, representing a NOx reduction of approximately
25% at 2% excess oxygen and 10% at 1 % excess oxygen. The flames associated
with the 2 top burners exhibited pronounced dark streaks, indicative of a wet
atomising steam supply. The independence of NOx emissions from excess oxygen
suggests that the minimum NOx emissions achievable from the boiler via atomizer
design modifications is around 400 mg/Nm3.
Under similar operating conditions there was no significant difference in the particulate
matter emissions produced by any of the atomizers. Consideration of the benign
characteristics of the fuel and the particulate emissions recorded suggests that the
particulate emission level is primarily a function of excess air level rather than
atomisation quality. Consequently the full benefits of advanced F-jet atomizers is not
being obtained.
Advanced F-jets have also been fired at Richborough PS which comprises 2x125MW
front wall fired boilers originally designed for coal firing and subsequently modified for
Orimulsion firing. Fifteen burners are arranged in a 3 (vertical) by 5 (horizontal) array
on the furnace front wall. On conversion to Orimulsion the boiler was fitted with F-jet
atomizers, supplied by one of the existing F-jet licensees. More recently Power
Technology designed advanced F-jet atomizers have been installed providing improved
flame stability and appearance combined with a 33% reduction in atomising steam
consumption. NOx emissions were unchanged.
A set of "70/30" advanced F-jets were installed on Unit 2 and were found to have no
impact on boiler NOx emissions as shown on figure 10. However, following
installation of the atomizers the fuel supply pressure to the burners increased by 3
bar, indicating that 2 of the burners were blocked, although with limited flame viewing
facilities this could not be confirmed.
Adjustment of the burner primary air supply pressure resulted in increased NOx
emissions when the PA pressure reduced and reduced NOx emissions when the PA
pressure increased, that is NOx emissions increased with reducing primary zone
stoichiometry. It is believed that the reduction in burner stoichiometry caused by the
blocked burners offset the NOx reduction which would have been obtained by biasing
the radial fuel placement. It is planned to repeat the tests at Richborough in the near
future.
-------
Summary
Combination of the F-jet design concept developed by the CEGB and the ATA
geometry developed by BP (UK) Ltd has resulted in a high efficiency atomizer with
predictable performance characteristics, designated the advanced F-jet atomizer.
Atomizers with thermal ratings from 1MW to 75MW have been successfully fired and
NOx reductions of up to 25% have been demonstrated on industrial sized boiler plant
with no increase in paniculate matter emissions. This has been accomplished with
no other modifications to burner/boiler equipment.
The advanced F-jet atomizer provides a low cost option for either reducing NOx
emissions by up to 25%, reducing particulate emissions, improving boiler efficiency
(reduced excess air) or a combination of these factors, together with improved flame
stability compared with many other atomizer designs.
Acknowledgements
The efforts of the many PowerGen staff involved in test rig and boiler testing are
gratefully acknowledged.
This paper has been presented with the permission of PowerGen pic.
References
1. - "Design Optimisation of a Large Pressure Jet Atomizer for Power Plant", A R
Jones, ICLASS 1982.
2. "The MEL Two Piece Two Fluid (orF-Jet) Atomizer", CEGB TPRD/M/1327/N3,
M Sarjeant, March 1983.
3. "ATA Development Programme: Summary Report", PT/94/220016/M, A J
Read, February 1994.
4. "1994/95 Atomizer Development Programme: Summary Report", M Garwood
(in preparation).
-------
POWERGEN pic POWER TECHNOLOGY DEPARTMENT
Plenum chamber
Atomising
fluid ports
FIGURE
No.
\
THE TWO PIECE TWO FLUID ATOMISER OR F-JET
-------
POWERGEN pic POWER TECHNOLOGY DEPARTMENT
OUTER BODY
INNER BODY
FIGURE
No. .
G.A. ASSEMBLY OF ATA WITH OIL OUTSIDE ARRANGEMENT
P.O. 234
-------
POWERGEN pic, POWER TECHNOLOGY DEPARTMENT
70
60
50
kJ 40
00
o
D^
O
30
20
10
0
Y-JET
GRAIN P.S. '
A
F-JET
ATA
INCE P.S.
D30
D20
D10
AEJ
E-JET
AEJ
FIGURE
No.
3
DROPLET SIZES EOR VARIOUS ATOMIZERS
-------
POWERGEN pic, POWER TECHNOLOGY DEPARTMENT
1300
200
100
1000
900
CD
E
800
c
CD
CD
! 7°°
ro
o
CD
_1_J
o
CD
o
o
X
o
600
500
400
300
200
Q 100° CONE ANGLE
0 90° CONE ANGLE
A 80° CONE ANGLE
0 60° CONE ANGLE
0.5
1.0
EXCESS OXYGEN
100
90
80
70
60
50
40
30
20
10
0
00
h^
C£
O
O
OQ
o
FIGURE
No.
4
NOx EMISSIONS AND CARBON IN GRITS
FOR VARIOUS SPRAY CONE ANGLES
-------
POWERGEN pic, POWER TECHNOLOGY DEPARTMENT
1300.
200
100
1000
r-O
900
I" soo
cf
-------
POWERGEN pic, POWER TECHNOLOGY DEPARTMENT
1300.
1200
1100
1000
900
800
c
CD
f 700
fc 600
O
cn
E
C/0
500
400
00
X
O
300
200
0
STANDARD
90° CONE ANGLE
60/40
70/30
80/20
90/10
ORIMULSION
0.5 1.0
EXCESS OXYGEN %
.5
FIGURE
No.
6
NOx EMISSIONS
FROM DIFFERENT RADIAL FUEL DISTRIBUTIONS
-------
POWERGEN pic, POWER TECHNOLOGY DEPARTMENT
300.
1200
00
1000
^ 900
en
E
-a 800
c
s
° 700
-o
O
CU
^_
i_
o
o
X
O
600
500
400
300 I
200
0 STANDARD 90° CONE ANGLE
A 70/30 - 90/60
0 90/10 - 90/60
70/30,
90/10,
A
0.5
1.0
EXCESS OXYGEN %
1.5*
100
90
80
70
60
50
40
30
20
10
0
C/)
h^
QL
O
O
m
o
FIGURE
No.
7
NOx EMISSIONS AND CARBON IN GRITS FROM 70/30 AND 90/10
ATOMIZERS WITH DIFFERENT SPRAY CONE ANGLES
-------
POWERGEN pic, POWER TECHNOLOGY DEPARTMENT
1300.
1200
1100
1000
900
800
f 700
fc 600
0
cn
E
C/)
500
400
C/)
X
O
300
200
STANDARD
90° CONE ANGLE
70/30 - 90/60
90/10 - 90/60
ORIMULSION
0
0.5
EXCESS OXYGEN %
1.0
1.5
•IGURE N0x EMISSIONS PROM 70/30 AND 90/10 ATOMIZERS
°8 WITH DIFFERENT SPRAY CONE ANGLES
-------
POWERGEN pic, POWER TECHNOLOGY DEPARTMENT
550
540
530
520
510
500
00
49°
00
k2 480
o
00
O
CL
470
460
450
440
430
420
410
400
390
50
40
30
20
10
0
0
0 EXISTING ATOMIZERS
H STANDARD ADVANCED F-JET
A 74-26 AFJ
0 70-30/90-60 AFJ
*
8~0.9 1.0 1.1 1.2 1.3 1.4 1.5 1.6 1.7 1.8 1.9 2.0 2.1 2.2 2.3
BOILER EXCESS OXYGEN (CONTROL ROOM) %
FIGURE
No.
9
NOx AND PARTICULATE MATTER EMISSIONS OBTAINED
FROM 80 t/hr STEAM BOILER WITH VARIOUS ATOMIZER DESIGNS
-------
POWERGEN pic, POWER TECHNOLOGY DEPARTMENT
590
580
570
->; 560
5 «
|550
o
b 450
o
E
^ 540
en ^
E
cn
~z.
c/2
UJ
o 520
510
5001
490
4.
'A1 SIDE : OPEN SYMBOLS
'B1 SIDE : SHADED SYMBOLS
0 STANDARD AFJ (PA = 15 mbar)
A PA = 15 mbar ^
m PA = 20 mbar L LOW NOx AFJ
0 PA = 12 mbar j
-
3
^ ®
ffl ^
% ® ^
*,/' ®
/''' 0
/"'' ° 0
A
A
H
m
0 4.1 4.2 4.3 4.4 4.5 4.6 4.7
I.D. rAN OXYGEN CONCENTRATION (% dry)
FIGURE NOx EMISSIONS AS A FUNCTION OF I.D. FAN
No- OXYGEN CONCENTRATION : RICHBOROUGH UNIT 2
10
-------
AN ASSESSMENT OF BOILER IMPACTS
FROM REDUCED AIR FLOW OPERATION
C. Rossi-Lane
B.L. McDonald
Carnot
15991 Red Hill Avenue, Suite 110
Tustin, California 92680
E. Eddings
Reaction Engineering International
77 West 200 South, Suite 210
Salt Lake City, Utah 84101
G. Norman
Florida Power Corporation
P.O. Box 14042
St. Petersburg, Florida 33733
L. Radak
Southern California Edison Company
2244 Walnut Grove Avenue
P.O. Box 800
Rosemead, California 91770
P. Strangway
Niagara Mohawk Power Corporation
300 Erie Boulevard West
Syracuse, New York 13202
G. Wolff
Consolidated Edison Company of New York
4 Irving Place
New York, New York 10003
K. Zammit
Electric Power Research Institute
P.O. Box 10412
3412 Hillview Avenue
Palo Alto, California 94303
Abstract
Due to an increased number of non-utility generators, in combination with base loaded nuclear
capacity, many utilities have been faced with the need to cycle their fossil fuel-fired boilers well
below 25% of maximum continuous rating (MCR). In order to guarantee adequate boiler
purging, the National Fire Protection Association (NFP A) recommends a minimum air flow level
-------
of 25% of full load air flow during all modes of low load operation. As a consequence, operation
of boilers in this regime presents several adverse impacts to electric utilities, which include:
• lower operating efficiencies that make units less competitive,
• increased NOX emission levels,
• less safe operation because of flame stability problems due to increased air/fuel velocity ratios
at the burner and difficulty for flame proving devices due to flame lift-off, and
• limited turndown ratios.
For most boilers, the ability to operate below 25% of full load air flow at loads below 25% of full
load would lower NOX emissions, increase boiler efficiency, and improve flame stability.
However, it does raise other concerns including adequate boiler purging and steam temperature
control. Furthermore, some units originally designed for coal-firing were converted to gas- and
oil-firing, and the relatively taller furnace design of the coal units may pose some restrictions on
low load/low air flow firing. Flue gas recirculation and variable pressure operation may offer a
viable solution to some of these concerns.
The current paper presents results from an ongoing study that is examining the feasibility and an
approach that would allow the safe low load operation of boilers at reduced air flows. The
current study assessed gas- and oil-fired boilers only; however^ reduced air flow operation is
desired for coal-fired boilers that have deep cycling requirements and may be included later.
Numerical modeling was conducted to quantify reduced ah- flow effects on furnace exit gas
temperature for a gas/oil-fired boiler originally designed for coal-firing. A convective pass heat
transfer model utilized this information to assess potential steam temperature impacts from
reduced air flow operation. Full-scale operational test data was used where available to assess
low air flow impacts on a variety of wall, opposed and tangentially-fired boilers. In combination
with the technical assessment, a cost/benefit analysis was conducted.
Introduction
A number of utilities have expressed interest in revising the requirements of Standard 8502
(formerly 85C) of the National Fire Protection Association (NFPA). Standard 8502 requires
boilers to be operated with a minimum air flow rate of 25% of the full load air flow rate during all
operating modes including furnace purge, start-up, and steady low load conditions. This standard
has caused operating problems for many utilities during operation at minimum load. These
problems include limited turndown, flame instability, increased NOx emissions, and increased heat
rate.
In 1988, Carnot conducted a single field evaluation of reduced air flow operation on a 100 MW
front-wall, gas-fired boiler. Key results of the field evaluation included the following:
• Safe reductions in total combustion air flow from 25% to 13% were achieved with no adverse
impacts on any operating parameters;
Safe operation was demonstrated by closing air registers on the non-firing burners to direct
the combustion air through the firing burners;
• NOx emissions were reduced by over 80%, from 193 ppmc (ppm at 3% O2) by reducing the
air flow; and.
-------
• Main steam temperatures dropped by SOT. The degradation in heat rate associated with this
decrease in steam temperature was more than compensated for by the increase in boiler
efficiency due to the reduced excess 62.
Based on the promising results of the initial field evaluation, several utilities expressed interest in
exploring the feasibility of reduced air flow operation on a wide variety of boiler types. The
current study conducted by Carnot has been sponsored by the following utility groups:
Consolidated Edison Company of New York
Florida Power Corporation
Los Angeles Department of Water and Power
Niagara Mohawk Power Corporation
Southern California Edison Company
Electric Power Research Institute
The objective of the study is to quantify boiler impacts from reduced air flow operation, and to
develop and demonstrate preliminary requirements to achieve safe, reduced air flow operation.
The study is divided into two Phases. Phase I involved an assessment of the benefits to utilities by
reducing the minimum air flow rate. Phase n involves the development of alternative operating
procedures and field evaluations on utility boilers. Phase I is currently underway with evaluations
of gas- and oil-fired boilers complete. This paper presents the results for these evaluations.
Background
Development of the NFPA Standard
NFPA 8502 is the "Standard for Prevention of Furnace Explosions/Implosions in Multiple Burner
Boiler-Furnaces." This standard was developed in response to a noted increase in U.S. boiler-
furnace explosions that occurred in the 1950s.1 The purpose of the standard is to "...establish
minimum standards for the design, installation, operation, and maintenance of boiler-furnaces and
their fuel burning, air supply, and combustion products removal systems. The standard requires
the coordination of operating procedures, control systems, interlocks, and structural design."2
The standard is in a continuous state of review and development which began in 1964.
The NFPA defines an explosion as "... the ignition of an accumulated combustible mixture within
the confined space of the furnace or the associated boiler passes, ducts, and fans that convey the
gases of combustion to the stack."2 The NFPA provides several scenarios which will produce
explosive conditions in a boiler. Each scenario includes a chain of events in which a flammable
mixture accumulates and is subsequently ignited.
Among many other operational guidelines, the standard recommends a minimum purge rate of
25% of the full load air flow under all operating conditions. This includes start-up, shut-down,
and low load conditions. Communications with the NFPA have revealed that the information
supporting the basis for selecting the 25% minimum is not readily available. It can be speculated
that 25% was selected because it offered a very high safety margin in preventing boiler-furnace
explosions. It is likely that the safety margin considered the limited capability of the combustion
monitoring equipment (e.g., flame scanners) which existed at the time the standard was
developed. According to the NFPA, "historical experience has shown a reduction hi furnace
explosions when operating with the minimum airflow (25 percent) value of the standard."3
Although many utilities have adopted the procedures described in the NFPA 8502, including 25%
minimum air flow, the standard developed by the NFPA is to be interpreted as a guideline. In
-------
fact, according to the NFPA, "The standard does not do away with the need for the engineer or
for competent engineering judgment. It is intended that a designer capable of applying more
complete and rigorous analysis to special or unusual problems shall have latitude in the
development of such designs. In such cases, the designer is responsible for demonstrating the
validity of the approach."2 Since the standard is merely a guideline, there is no formal mechanism
for enforcement. However, some utility insurers require that their clients follow the guidelines as
a condition for coverage.
When approached with the idea of reviewing the applicability of the standard in light of today's
boiler operation and possibly modifying the minimum air flow requirement, the NFPA has
responded that they will not consider modification of the minimum air flow guideline until safe,
long-term, full-scale utility demonstrations have been conducted.
Current Boiler Operation
The power production industry was very different during the 1950s and 1960s when the 25%
minimum air flow guideline was established as compared to today. In particular:
• When the NFPA standards were originally developed, air pollution emissions were not
regulated, and were of little concern. Recently, the 1990 Clean Air Act Amendments (CAAA)
have imposed strict limits on the emissions of NOx and acid rain pollutants for all utility
boilers. Furthermore, local air quality regulatory agencies in some states such as California
and New York have implemented even more stringent air quality improvement plans than the
CAAA which add further pressure to the power industry.
• Fuel was relatively plentiful and inexpensive at the time, and, for many utilities, fuel cost was a
direct pass through to the users. So, fuel conservation and operating efficiency were not high
priorities. Today, with increased competition and proposed deregulation of the utility
industry, operating costs/efficiency are often the top priority for utilities.
• Because of limited competition, power demands for a given utility remained fairly constant
during the 1950s and 1960s. Typically, boilers were operated with moderate turndown ratios
of three or four to one. Today, due to increased competition from Non-Utility Generators
(NUGs), seasonal load demands, and voltage requirements for grid power distribution, boilers
are being operated at lower loads for longer periods of time. During off-peak periods, some
boilers are reduced to loads as low as 5 to 15% of full load capacity.
• During the 1950s and 1960s, flame scanner technology was largely based upon unreliable lead
sulfide cells or primitive UV systems. Today, flame scanner systems are much more
sophisticated allowing discrimination between flames from different burners as well as
discrimination between ignitor flames and main burner flames.
• Previously, boiler control relied on hard-wired, analog control systems. In the last few years,
many utilities have replaced these systems completely with micro-processor based control
systems, creating a smooth communication network for improved boiler control.
Operating utility boilers according to NFPA guidelines can result in two or three problems
affecting low-load operation. The two problems which affect most units are thermal losses
created by high excess air levels and high NOX emissions. For some units, a problem of low load
flame instability can accompany operation with high air flow at low loads, thereby creating an
unsafe operating condition.
-------
Program Objectives and Approach
The initial objectives during Phase I of the study were to:
• Determine the potential benefits of reducing the required minimum air flow rate in terms of
improved operating safety, improved boiler performance, reduced NOX emissions, and
operating cost savings.
• Assess the technical feasibility and economic impacts of implementing operating procedures to
allow reduced air flow at low loads.
• Establish the procedures for developing an alternative minimum air flow standard to that of
theNFPA.
• Define the preliminary criteria for implementing reduced air flow operation and the scope of
field evaluations necessary to demonstrate this operation.
Phase I was structured around a group of boilers shown in Table 1. These boilers represent the
principal design types of the primary U.S. boiler manufacturers and are boilers which fire gas
and/or oil fuels. As shown, all of these boilers exhibit turndowns in excess of 4 to 1 with
minimum loads as low as 5% of maximum continuous rating. The cost savings, boiler
performance impacts, and emissions reductions were evaluated for each of these boilers under
reduced air flow conditions.
Table 1
Sample Boiler Candidates for Reduced Air Flow Operation
Primary
Fuel
Gas
Gas
Gas
Gas**
Gas
Oil
Oil
Oil**
Oil
Oil
Oil
Oil**
Design Manufacturer
Wall
Opposed
Tangential
Tangential
Tangential
Wall
Wall
Tangential
Tangential
Tangential
Tangential
Tangential
B&W
FW
CE
CE
CE
B&W
FW
CE
CE
CE
CE
CE
No. of
Units*
4
1
6
1
1
1
1
4
1
1
2
1
Size
MW
200
750
320
375
450
120
850
100
120
220
520
900
M in Load Flue Gas
% MCR Recirculation
10%
7%
6%
20%
13%
10%
18%
10%
10%
10%
5%
20%
yes
yes
yes
no
yes
no
yes
no
no
no
yes
no
*Units of this design owned by the same utility and wanted for reduced air flow operation.
**These units were originally designed for coal firing.
-------
Recommended Procedures to Allow Reduced Air Flow Operation
Objective of the Boiler Purge
The objective of the furnace purge is to prevent the build-up of flammable mixtures which could
lead to an explosion. One of three methods can be used to limit the build-up of flammable
mixtures in the furnace4:
• Provide an adequate source of ignition.
• Limit the introduction of unignited fuel and air so that they do not occupy a significant portion
of the furnace.
• Dilute the unignited fuel and air mixture below flammability limits by a continuous purge
process.
The purpose of the continuous purge rate at 25% of full load air flow is to prevent the build-up of
combustibles in the furnace. According to the British standard on "Purging, Ignition, and Flame
Failure Requirements"5, a purge rate of 50% of full load air flow is recommended for water-tube
boilers used for power generation; however, it also states that the continuous purge may not be
necessary if it can be proved that there is not an explosive mixture in the furnace. In order to
prove a safe condition, the British standard requires that a representative sample of the furnace
gases be analyzed for combustibles and that the sampling equipment be designed such that a
failure of the equipment will result in a safe condition.
Boiler Conditions which Require Purge
Start-up Sequences. Prior to light-off, the boiler must be swept of any residual combustibles.
Then, during light-off, continuous purging must take place so that fuel from burners that are not
lighting do not accumulate and cause an explosion. The purge rate should be maintained through
the boiler warm-up period since the cool furnace may quench some of the flames and cause a loss
of flame from some burners, which would cause a build-up of combustibles in the furnace.
Loss of Flame. After loss of flame, unburned fuel enters the furnace until the tune that the
operator or control system cuts fuel to that burner. To prevent an explosion, the furnace must be
adequately purged of the unburned fuel. The fuel must not be allowed to stagnate in the area of
the boiler where there is insufficient flow such as in a recirculation zone (e.g., the hopper). The
NFPA maintains that the furnace should be purged until it has experienced five volume changes.
If the air flow rate is higher than the purge rate (25% of full load air flow) when the flame is lost,
the air flow may be gradually decreased to the purge rate for a post-firing purge. However, if the
air flow rate is lower than the purge rate when the flame is lost, the air flow must be held constant
at that rate for five minutes. Then, it may be increased to the purge rate for a post-firing purge.2
Recommended Purge Procedures
The basic operating objectives of the current standard include the requirement to maintain air flow
at or above the purge rate during all boiler operations. We do not recommend modification to
this provision other than to include a reference to alternative procedures. The original NFPA
objective will continue to apply to the majority of boilers. Alternative procedures will only be
allowed under specific conditions including:
• Continuous monitoring of CO and excess O2 in multiple sampling locations representative of
furnace conditions.
• Accurate monitoring of fuel and air flow rates at low boiler loads.
-------
• Monitoring of flame stability using adequate flame detectors.
• Boiler operational testing with reduced air flow.
The boiler operational testing requirement will be used to define the specific alternative
procedures allowed for each individual boiler. In particular, it will be used to:
• Determine the minimum acceptable air flow rate and factors limiting further reductions in air
flow, such as "flame stability, CO emissions, steam temperature, air leakage through closed
registers on out-of-service burners, etc.
• Determine and/or verify representative sampling locations to continuously monitor furnace
CO and excess 02.
• Verify satisfactory flame stability, air/fuel distribution, combustion efficiency, and CO
emissions over the full range of boiler operating conditions.
• Verify that the performance of the combustion controls and instrumentation is satisfactory for
reduced air flow operation, or recommend systems requiring improvement such as flame
scanners, fuel flow measurement systems and air flow measurement systems.
• Establish alarm and trip points for the combustion control system for such parameters as low
air flow, high CO emissions, etc.
Impacts of Reducing Air Flow During Steady Operation
Boiler Performance
Approach. During the current study conducted by Carnot, reduced air flow operation impacts
on boiler performance (specifically steam temperatures) were assessed. The approach involved a
detailed analysis of a 100 MW tangentially-fired boiler capable of firing either natural gas or oil.
The boiler was originally designed for coal-firing and was subsequently retrofitted for gas-/oil-
firing. The performance analysis involved three steps:
1. A three-dimensional furnace model was developed by Reaction Engineering International
(REI) using their own code created by Dr. Philip J. Smith of the University of Utah. This
model incorporated the chemical reaction, heat transfer, and fluid dynamics calculations
required to fully characterize the combustion processes within the furnace and to obtain the
furnace exit gas temperature at the inlet of the superheat region.
2. Carnot developed a one-dimensional heat transfer model of the convective pass of the boiler
to evaluate resulting steam temperatures in each section of the boiler based on the data
provided by the furnace model.
3. Trends in steam temperature were used to assess impacts on turbine efficiency and overall
heat rate for both gas and oil firing.
Unit Description. Combustion simulations were performed on a 100 MW tangentially-fired
boiler designed by Combustion Engineering (CE). This unit was originally designed for coal firing
but was converted to oil, natural gas or gas/oil coining capability. The oil and gas burners fire
through three levels of burners placed between the original coal burners. Design superheat steam
-------
conditions are 1000°F and 1492 psig, and design reheat steam conditions are 1000°F and 470
psig. The unit is not equipped with flue gas recirculation but relies on large amounts of excess air
to maintain steam temperatures at low loads. A profile of the unit along with the furnace model
simulation boundaries is shown in Figure 1.
Analysis Matrix. An analysis matrix of fourteen operating conditions is shown in Table 2. As
shown, the boiler was analyzed under both gas- and oil- firing conditions at -25%, 17.5%, and
10% of full load. For the 17.5% and 10% load cases, three air flow levels were analyzed ranging
from the NFPA recommended 25% of full load air flow down to air flow rates corresponding to a
stoichiometric ratio of approximately 1.0 for gas firing and 1.1 for oil firing.
Modeling Assumptions. To fully describe the operating conditions of the boiler, a number of
engineering assumptions were required to provide information not available from the boiler
engineers. These assumptions include the following:
Wall temperatures were assumed constant and the value used was based on the saturation
temperature of the steam at the drum pressure (1450 psig).
The emissivity of the fireside wall was assumed as 0.7 for gas-firing conditions and 0.8 for oil-
firing conditions. (Note that a soot-formation model was also utilized for the fuel oil cases.)
The flue gas outlet boundary of the furnace model lies in the secondary superheat section of
the convective pass. (This was necessary to fully model the nose section.) In an effort to
simplify the model, individual tubes were not modeled in the superheat region. Instead, heat
loss from the gas to the tubes was simulated by assigning an emissivity value of 0.5 at the
outlet of the simulation model. This value was estimated by REI based on previous modeling
experience and was validated by analysis data from CE. Because the superheat steam tubes
are in-line and the main source of flow mixing is in the vertical direction due to the presence
of flow recirculations behind the nose, ignoring the presence of the tubes should not
significantly alter the dominant fluid mechanics.
• A firing angle of 7° off-center was assumed for each of the corner burners.
A burner tilt of+30° was used to maximize steam temperatures for all cases.
Model Validation. Because there was no actual furnace exit gas temperature test data for the
boiler analyzed, the furnace model could not be validated directly. Theoretical predicted
performance data for the boiler at 25% load for gas and oil firing was available from CE. A
comparison of the CE data with data generated by the convective model and the furnace model is
presented in Table 3. As shown, the gas firing analysis temperatures are in very good agreement,
with Carnot's values falling within 10% of those predicted by CE's data. The oil firing analysis
temperatures were in fair agreement, with Carnot's values falling within 10-20% of CE's data.
Furnace Model Results. The furnace simulations conducted by REI were all converged to
the greatest accuracy possible given the level of detail utilized for the simulations. The average
mass balance closure was approximately 3%, and the energy balance closure averaged
approximately 11%.
The furnace modeling indicated the following trends during both gas and oil firing:
• Flame temperature increases as the excess air decreases and the overall stoichiometry
approaches one (1.0).
-------
2ND
STAGE
SUPERHEATER
REHEATER
101'
FURNACE
WIDTH - 32.9
2*
52.7'
1ST
STAGE
SUPERHEATER
ECONOMIZER
RB FURNACE SIMULATION BOUNDARY
Figure 1
100 MW, Tangentially-Fired Gas/Oil Unit Used for Performance Simulations
During Reduced Air Row Operation
-------
Table 2
Boiler Model Analysis Matrix
Case Fuel Type Fuel Flow* Air Flow Stoichiometric Ratio Burner Levels Firing
1
2A
2B
2C
3A
3B
3C
4
5A
5B
5C
6A
6B
6C
gas
gas
gas
gas
gas
gas
gas
oil
oil
oil
oil
oil
oil
oil
26.0
17.5
17.5
17.5
10.0
10.0
10.0
25.0
17.5
17.5
17.5
10.0
10.0
10.0
33.7
25.0
20.0
17.5
25.0
15.0
10.0
25.5
25.0
20.0
17.5
25.0
15.0
10.0
1.28
1.41
1.13
1.00
2.50
1.50
1.00
1.10
1.63
1.30
1.14
2.85
1.71
1.14
top two
top one
top one
top one
top one
top one
top one
top two
top one
top one
top one
top one
top one
top one
*% of Full Load
Table 3
Heat Transfer Model Validation
Fuel Parameter Model Result CE Prediction % Difference
Gas Final Stage Superheater Steam Outlet T
(Case 1) Reheat Steam Inlet T
Reheat Steam Outlet T
Economizer Water Outlet T
Economizer Gas Outlet T
867°F
455°F
756°F
377°F
510°F
875°F
465°F
824°F
404°F
468°F
-1%
-2%
-8%
-7%
9%
Oil Final Stage Superheater Steam Outlet T 798°F 875°F -9%
(Case 4) Reheat Steam Inlet T 392°F 465°F -16%
Reheat Steam Outlet T 653°F 830°F -21%
Economizer Water Outlet T 370°F 407°F -9%
Economizer Gas Outlet T 509°F 468°F 9%
-------
• As overall flow rate decreases, the central fire ball in the furnace becomes more elliptical. At
minimum load and a stoichiometric ratio near one (1.0), the model suggests that the fire ball
breaks down. It is unclear, however, if the lack of the fire ball is due to low mass flow rates
or the low resolution of the computational mesh in this region. Due to the high complexity of
the model and the limited scope of the program, further refinement of the mesh in this region
was not investigated during the current study.
• The modeling suggests that there are three primary recirculation zones existing in the furnace
during low load firing. Schematics of these zones are provided in Figure 2.
1. Hopper Recirculation Zone. A recirculation zone of low intensity forms hi the hopper
region.
2. Furnace Wall Recirculation Zone. There are flow recirculations between the central
swirling flame and the vertical furnace walls. The size and magnitude of these zones
depend on the particular conditions (i.e., excess air and load). However, for all cases, they
do not extend above the nose. As a consequence of these recirculation zones, there is a
strong downward flow of combustion gases along the wall opposite the boiler nose for
most cases.
3. Superheat Recirculation Zone. A recirculation zone exists immediately above the nose in
each case. As the furnace flow spirals upward from the tangential flame in the middle of
the boiler, the high velocity gases are not able to make the sharp turn immediately around
the nose. This causes the flow to separate from the wall at the nose. Gases on the other
side of the nose flow into this low pressure region resulting in a vertical recirculation zone
above the nose that revolves in a clockwise direction. At high flow rates, the momentum
of the higher mass flow rates drives the combustion gases more directly up to the top of
the furnace inhibiting the penetration of the recirculation zone beyond the tip of the nose.
At lower flow rates, the upward flow is less direct and allows the recirculation zone to
penetrate beyond the tip of the nose forming a larger recirculation zone. The average
furnace exit gas temperature (FEGT) at the vertical plane above the nose (which
corresponds to the entrance to the superheater) is shown as a function of load and air flow
rate in Figures 3 and 4 for gas and oil firing, respectively. Although the model shows that
flame temperature increases as air flow decreases, the FEGT at the superheat inlet
decreases with decreasing air flow. This is because the cross-sectional plane above the
nose contains hot combustion gas flowing over the top of the recirculation zone in
addition to recirculated gases that have had a longer residence time in the furnace and
have lost more heat to the furnace walls and the outlet. Thus, it is the recirculation zone
that causes the furnace exit gas temperature to decrease with decreasing air flow.
In comparing Figures 3 and 4, it is clear that there is less variation in FEGT with air flow during
oil firing as compared to gas firing, particularly at 17.5% of full load. This is because there was
less variability in recirculation zone size with each condition during the oil simulations as
compared to the gas simulations.
Convective Pass Model Results. The FEGT data calculated by REI was used as input to
the one-dimensional convective pass model developed by Carnot. The superheat and reheat steam
temperatures calculated with the results of the furnace models during gas and oil firing are
presented in Figures 5 and 6, respectively. As shown, both superheat and reheat steam
temperatures decrease with decreasing air flow. This trend is slightly more pronounced for gas
firing as compared to oil firing. The reduction in steam temperature with air flow is due to 1) the
increase hi flame temperature at low ah" flow which yields higher radiative heat transfer in the
furnace and 2) lower mass flow which reduces the convective heat transfer hi the convective
-------
2ND
STAGE
SUPERHEATER
REHE47ER
FURNACE
1ST
STAGE
SUPERHEATER
ECONOMIZER
FIB FURNACE SIMULATION BOUNDARY
Figure 2
Recirculation Zones Exhibited by the Furnace Model
of a 100 MW, Tangentially-Fired Gas/Oil Unit
-------
1600 -|
fa
0 1400 -
£
•§ 1200 -
u
V
§• 1000 -
£
_ ROO -
0
*• 600 -
•B
«
w 4.00 -
E ooo
fe
0 -
(
) '.
5 1
0 1
^
d
5 2
0 2
3
5
• 2
n J
» 1
3
.6% Load
7.5% Load
0% Load
0 3.
Air Flow, % of Full Load Air Flow
Figure 3
Furnace Exit Gas Temperature as a Function of Full Load Air Flow and Load During Gas Firing
for a 100 MW, Tangentially-Fired Gas/Oil Unit
Based on Boiler Simulations During Low Load, Reduced Air Flow Operation
1OUU -
1 AtV\
ace Exit Gas Temperature, °F
^ o^ oo o to j
5 8 8 8 8 £
e tuu •
200 •
0 -
,
.
,
L
i
1
Sr
"-
. 26%
D17.5
. 10%
]"
Load
%Load
Load
10 15 20
Air Flow, % of Full Load Air Flow
25
30
35
Figure 4
Furnace Exit Gas Temperature as a Function of Full Load Air Flow and Load During Oil Firing
for a 100 MW, Tangentially-Fired Gas/Oil Unit
Based on Boiler Simulations During Low Load, Reduced Air Flow Operation
-------
10 15 20
% of Full Load Air Flow
25
30
35
Figure 5
Superheat Steam Temperature as a Function of Full Load Air Flow and Load for a 100 MW,
Tangentially-Fired Gas/Oil Unit
Based on Boiler Simulations During Low Load, Reduced Air Flow Operations
1600
0
10 15 20
% of Full Load Air Flow
25
30
35
Figure 6
Reheat Steam Temperature as a Function of Full Load Air Flow and Load for a 100 MW,
Tangentially-Fired Gas/Oil Unit
Based on Boiler Simulations During Low Load, Reduced Air Flow Operation
-------
passes of the boiler. To illustrate the variation in flame temperature during gas firing at 17.5%
load, the adiabatic flame temperature increased from 3104°F at 25% air flow to 3713°F at 17.5%
air flow. To illustrate the variation in flame temperature during oil firing at 17.5% load, the
adiabatic flame temperature increased from 2980°F at 25% air flow to 3713°F at 17.5% air flow.
According to the analysis during gas firing, superheat steam temperature decreases by -25 °F per
1% reduction in full load air flow, and reheat steam temperature decreases by ~30°F per 1%
reduction in full load air flow. According to the analysis during oil firing, superheat steam
temperature decreases by ~9°F per 1% reduction in full load air flow, and reheat steam
temperature decreases by ~14°F per 1% reduction in full load air flow.
The convective pass analysis reveals that reduced air flow operation results in severe steam
temperature reduction during low load operation. In fact, during gas firing at 10% load and 10%
air flow, and during oil firing at 10% load with 15% and 10% air flow, the calculations showed
that the gas temperature from the first stage superheater dropped below the drum temperature
revealing that there may be steam temperature limitations when using reduced air flow at these
loads. Note that these limitations may also exist at other loads not analyzed during the program
(i.e., loads between 10% and 17.5% of full load). This result is based on two characteristics of
the boiler:
1. The boiler was originally designed for coal firing and later converted to gas and oil-firing. As
a result, the furnace dimensions are more appropriate for coal firing (i.e., coal furnaces are
taller than gas/oil furnaces to assist in burnout of coal particles). At reduced air flow levels,
there is less mass to carry the heat generated by the combustion process up into the
convective pass of the boiler and there is greater radiative heat transfer due to increased flame
temperature. With the taller furnace, these problems are magnified.
2. The boiler is not equipped with flue gas recirculation. Many gas- and oil-fired boilers utilize
flue gas recirculation to maintain steam temperature at low loads. With flue gas recirculation,
the heat generated within the furnace is more effectively transferred to the convective passes.
The penalty associated with this method of steam temperature control is the cost of the energy
required to operate the flue gas recirculation fan.
Cost/Benefit Analysis
In order to perform a cost/benefit analysis of reduced air flow operation, the impacts of reduced
air flow operation were considered including heat rate savings or penalties, emissions savings and
costs of boiler equipment upgrades to ensure safe operation with reduced air flow.
Net Heat Rate Impact Assessment The detailed analytical modeling of the 100 MW CE
tangentially-fired boiler showed that superheat steam temperatures decreased by approximately
25 °F per 1% decrease in full load air flow and reheat steam temperatures decreased by
approximately 10°F per 1% decrease in full load air flow. These trends were applied to those
boilers not equipped with FGR. To assess the steam temperature penalty associated with units
that are equipped with FGR, Carnot reviewed field test data from a variety units that we have
tested. Based on the test data, we found that superheat steam temperatures decreased by
approximately 5°F per 1% decrease full load air flow at minimum load when flue gas recirculation
was utilized on units that fire natural gas. Due to a lack of test data at low loads on oil-fired
units, it was assumed that superheat steam temperatures decreased by 5°F per 1% decrease in full
load air flow as well.
The decrease in steam temperature results in decreased turbine efficiency. However, less air flow
also results in lower stack losses which increases boiler efficiency. For most units, this boiler
efficiency benefit outweighs the turbine efficiency penalty resulting in a net heat rate benefit.
-------
Units that are equipped with flue gas recirculation and/or were originally designed for gas/oil
firing (as opposed to coal firing) will achieve greater heat rate savings because the steam
temperature impact will be minimal.
The net heat rate savings was quantified in terms of cost savings by conversion to fuel savings.
The following assumptions were made in the heat rate cost savings analysis:
• These units are operated at minimum load with reduced air flow for 2000 hours per year.
• Baseline heat rate at minimum load is 13,000 Btu/kWh.
• The cost of natural gas is $2.892 per MMBtu.
• The cost of fuel oil is $2.511 per MMBtu.
• The turbine cycle heat rate penalty is 0.8% per 50°F decrease in steam temperature.
Except for the 100 MW boiler analyzed with the numerical modeling, the minimum air flow was
assumed to be the air flow level that would result in the same air-to-fuel ratio as at full load (10.3
ft3 air per ft3 fuel for gas firing and 15.59 Ib air per Ib of fuel for oil firing). For the 100 MW CE
tangentially-fired boiler, the minimum air flow at minimum load (10% MCR) was determined
through modeling (10% of full load air flow for gas firing and 16% of full load air flow for oil
firing).
NOX Emissions Impact Assessment. A key benefit of reduced air flow operation is a
reduction in the formation of oxides of nitrogen (NOX). Factors that favor NOX formation are
high flame temperature and local high excess air levels. Thermal NOX is formed by the thermal
breakdown of molecular nitrogen that is present in the combustion air. This nitrogen then reacts
with the available oxygen to form NO and NOX. At loads less than 25% of full load, there is a
large quantity of excess air available for NOx formation.
Based on field test data collected during previous programs conducted by Carnot and others, it
was assumed that NOX is reduced by 10 ppmc for every 1% reduction in full load air flow.
Expected NOX reduction on a mass basis was then calculated assuming a baseline NOX level of
125-400 ppmc at minimum load and 25% of full load air flow depending on the specific unit
design, size and fuel fired. In no case was the NOX concentration expected to be reduced below
30 ppmc for gas firing or below 50 ppmc for oil firing.
NOx reduction was quantified in terms of cost savings potential by assuming a conservative
market value of $3,000 per ton of NOX reduced. This approach assumes that a given utility
would have the opportunity to sell NOX reduction credits on the open market. Currently, there is
a market for NOx in Southern California, and others may develop as the CAAA are implemented.
Safety Impact Assessment. Plant safety is of utmost concern to the utility industry. It is the
intent of this program to increase plant safety with very specific operating procedures and
instrumentation to operate at reduced air flow levels.
While the current NFPA standard recommends that 25% of full load air flow be the minimum air
flow allowed under any circumstance, some boilers that operate at less than 25% of full load
experience a flame stability problem with the current minimum standard. The flame instability
occurs when high air flow causes the velocity of air entering the furnace to blow the flame off the
burner throat. In addition, instability can occur when the flame is quenched by the high volume of
entering fuel and air mixture. Therefore, lowering air flow can eliminate a flame instability
problem and allow still lower loads to be achieved.
-------
To ensure that there is adequate purging of the boiler and that there is no build-up of
combustibles within the furnace and boiler passes, we recommend that reduced air flow operation
include the use of flame scanners, accurate air and fuel flow measurement systems, and
continuous CO and O2 monitors to verify complete combustion. Most units are equipped with
such systems, but, typically, these systems have been designed for full load operation. For
example, flame scanners are often oriented such that they can easily "see" the full load flame, but
as loads are reduced and the flame shape changes, some of these scanners can no longer detect a
flame because of improper orientation. As another example, combustion air flow is typically
measured from the pressure differential across a flow device, such as a venturi. This pressure
differential varies as flow rate squared. Hence, the pressure differential at 25% of full load is only
6.25% of that at full load. If the flow device is designed to avoid excessive pressure drop at high
loads, the differential at very low loads will be too low to measure accurately.
As a result of instrumentation limitations, some units may have to be upgraded with "low load"
instrumentation. The need for this instrumentation can be evaluated only with full-scale testing.
However, Carnot has gathered some preliminary cost data on several systems that may be
required. Cost data for these systems is summarized in Table 4. The costs shown are system
costs only, and do not include installation. Since installation costs are extremely site specific, they
were assumed to be 50% of the hardware cost for each system. For the purposes of calculating
the cost of upgrades, it was assumed that all boilers would require all instrumentation and controls
upgrades. In reality, the upgrades required can only be determined once the current
instrumentation is tested under reduced flow conditions. So, this cost analysis is conservative. In
addition, since most utility boilers are equipped with flame scanners, CEMS and flow meters, it
was assumed that upgrading these systems would not impact plant maintenance and operating
costs.
Table 4
Reduced Air Flow Operation Instrumentation Cost Summary
Instrument Technology No. Required Cost Per Instrument
Flame Detectors
Fuel Flow Meter
Air Flow Meter
CO Analyzer
O2 Analyzer
Signature Scanner
Ultrasonic
Thermal Anemometer
NDIR/Gas Filter Correlation
In-Situ Heated Zirconia Disk
1 per burner
1 per boiler
2 per boiler
1 per boiler
4 per boiler
$4,200
$5,000
$15,000
$10,000
$4,700
Cost/Benefit Analysis Results. The benefit-to-cost ratio was determined using the annual
cost savings associated with heat rate improvements and NOX reduction and the annualized
capital cost of equipment upgrades assuming a capital recovery factor of 15%. Table 5
summarizes the benefits and costs for each boiler investigated during the current reduced air flow
assessment study. As shown, the benefit-to-cost ratio is highly dependent upon boiler design and
minimum load, ranging from 1.4:1 to 15.3:1. Two of the gas-fired units show an increase in fuel
costs. This is because the steam temperature penalty outweighs the boiler efficiency benefit for
these two boilers. However, in all cases, the NOX reduction potential yields an overall benefit.
-------
Table 5
Cost/Benefit Analysis for Selected U.S. Boilers Based on Simulated Reduced Air Flow Operation
Fuel
Gas
Gas
Gas
Gas
Gas
Gas
Gas
Gas
Gas
Oil
Oil
Oil
Oil
Oil
Oil
Oil
Design
Wall
Wall
Opposed
Tangential
Tangential
Tangential
Tangential
Tangential
Tangential
Wall
Wall
Tangential
Tangential
Tangential
Tangential
Tangential
FGR
yes
yes
yes
no
yes
yes
yes
no
yes
no
yes
no
no
no
yes
no
Size
MW
200
200
750
100
320
320
320
375
450
120
850
100
120
220
520
900
Min Load
% MCR
10%
10%
7%
10%
6%
6%
6%
20%
13%
10%
18%
10%
10%
10%
5%
20%
Net Heat
Rate
Savings
4.17%
4.17%
7.34%
-0.63%
9.36%
9.36%
9.36%
-1.11%
2.26%
3.67%
1.04%
2.20%
3.67%
3.67%
14.58%
0.21%
Fuel
Savings
$/yr
$62,710
$62,710
$303,550
($4,737)
$140,759
$140,759
$140,759
($62,597)
$101,960
$28,752
$101,846
$14,363
$28,752
$52,712
$247,486
$24,678
NO,
Savings
Ib/hr
35
35
109
14
42
42
42
48
84
19
123
6
19
35
54
102
NO,
Savings
S/yr
$105,984
$105,984
$328,218
$43,152
$126,033
$126,033
$126,033
$144,723
$250,644
$58,056
$368,697
$17,634
$58,056
$106,434
$161,799
$307,392
Total
Savings
$/yr
$168,694
$168,694
$631,768
$38,415
$266,792
$266,792
$266,792
$82,126
$352,604
$86,808
$470,543
$31,997
$86,808
$159,146
$409,285
$332,070
Equipment
Upgrade
Cost
$224,400
$224,400
$274,800
$148,800
$224,400
$224,400
$224,400
$274,800
$356,700
$148,800
$274,800
$148,800
$148,800
$211,800
$400,800
$476,400
Annualized
Capital
Cost*
$33,660
$33,660
$41,220
$22,320
$33,660
$33,660
$33,660
$41,220
$53,505
$22,320
$41,220
$22,320
$22,320
$31,770
$60,120
$71,460
Bencfit-
to-Cost
5.0 to 1
5.0 to 1
15.3 to 1
1.7 to 1
7.9 to 1
7.9 to 1
7.9 to 1
2.0 to 1
6.6 to 1
3.9tol
11.4 to 1
1.4 to 1
3.9tol
5.0 to 1
6.8 to 1
4.6 to 1
*Based on a capital recovery factor of 15%.
-------
Reduced Air Flow Operation Implementation Plan
The overall goal of the program is to generate a generic plan to facilitate implementation of
reduced air flow operation on all boilers of interest. The implementation plan would include a
four-step evaluation process that would provide a given utility with the information needed to
implement reduced air flow. The evaluation would involve:
• How to estimate the benefits of reduced air flow operation in terms of heat rate and NOX
reductions.
• How to identify areas requiring additional investment (e.g., instrumentation and controls
upgrades).
• An outline of the procedures required to implement safe, reduced air flow operation including
initial testing and the design of boiler operating sequences.
• An outline of the training required to ensure safe, long-term operation at reduced air flow
levels.
The information gathered during the full-scale demonstrations conducted during Phase n of the
current study will be used to design this implementation plan. The implementation plan will
address issues of particular concern to gas-, oil- and/or coal-fired units that are designed for
tangential, wall, or opposed firing. The plan will be refined and expanded as more information is
gathered on a wide variety of units.
There are six primary objectives that are expected to be achieved during the Phase n
demonstrations:
1. Procedures will be designed and evaluated to operate the boiler under steady load conditions
and reduced air flow. At the request of the demonstration site, we may also conduct dynamic
load testing which will involve ramping between 25% of full load air flow and minimum air
flow simultaneously with load.
2. Document reduced air flow operation to demonstrate safe operation between minimum and
25% load.
3. Evaluate the instrumentation and control systems including air and fuel flow meter accuracy
and flame scanner performance. In addition, appropriate locations for continuous CO and 62
sampling will be identified. If existing instrumentation fails to perform satisfactorily (e.g.,
flame scanners), further testing will be postponed until instrumentation upgrades are installed.
4. Evaluate the impact of reduced air flow operation on:
- flame stability,
- boiler performance (i.e., steam temperatures and heat rate), and
NOX emissions.
5. Evaluate the impact of flue gas recirculation at reduced air flows on boiler performance.
-------
Conclusions and Recommendations
The key conclusions and recommendations from Phase I of the program to investigate reduced air
flow at low loads are:
• For most units, reduced air flow operation is very attractive as it can provide significant
efficiency improvements, NOX emissions reductions, and overall improvement in flame
stability.
• Units that were originally designed for coal firing and that have been converted to gas and/or
oil firing may be restricted in the level of air flow reduction they can achieve, particularly if
they are not equipped with flue gas recirculation. The air flow reductions imposed by steam
temperature requirements may be investigated with analytical modeling prior to the
demonstrations on these units.
• Large units that can be turned down to 5 to 10% of full load will benefit the most from
reduced air flow operation.
• Adequate boiler instrumentation and control systems are required to ensure safe operation at
reduced air flow levels. In particular, units should be equipped with flame scanners, accurate
fuel and air flow meters, and combustion monitoring equipment such as back pass continuous
emissions analyzers for CO and excess 62 measurement.
• Full-scale testing is required on each boiler prior to operation at reduced air flow levels to
quantify the unit specific minimum air flow and evaluate the need for equipment upgrades.
• The NFPA will consider modifying their minimum air flow standard only after reduced air
flow has been demonstrated on a wide variety of units over the long term. Therefore, it is
important to conduct as many demonstrations as possible.
References
1. Wayne L. Wingert. "Prevent Furnace Explosions." Power Engineering, September 1967,
pp. 58-61.
2. National Fire Protection Association. NFPA 85C Standard for the Prevention of Furnace
Explosions/Implosions in Multiple Burner Boiler-Furnaces. Quincy, Massachusetts, 1991, p.
8.
3. National Fire Protection Association. 7995 Annual Meeting Report on Comments. Quincy,
Massachusetts, 1995, p. 201.
4. O.W. Durant and E.G. Lansing, "Furnace Explosions and Implosions", presented at the
American Power Conference, 1976.
5. R.E. Dye, "Purging of Marine Boilers - Full-Scale and Model Investigations," Paper No. 13,
Transactions of the Institute of Marine Engineers, Marine Management (Holdings) Ltd., Vol.
93, 1981.
-------
EVALUATION OF THE NOx REDUCTION POTENTIAL OF VARIOUS
NOx CONTROL TECHNIQUES ON A 320MWe OIL/GAS FIRED BOILER
P Baimbridge
M Garwood
PowerGen pic
Power Technology Centre
Ratcliffe-on-Soar
Nottingham NG11 OEE
England
A Facchiano
Electric Power Research Institute
Environmental Control Systems Business Unit
3412 Hillview Avenue
PO Box 10412
Palo Alto, CA 94303
R Pozzi
Azienda Energetica Municipale
Direzione Energia e Ambiente
Via della Signora 12
20122 Milano
Italy
Abstract
Recent NOx emissions legislation in Italy requires that the NOx emissions from oil/gas
fired utility boiler plant be reduced to 200 mg/Nm3 (corrected to 3% oxygen)(0.130
Ib/MMBtu). In the case of Azienda Energetica Municipale (Aem) Milano this must be
achieved by the end of 1997.
Aem Milano operate a 320 MWe opposed oil/gas fired boiler at their Cassano d'Adda
facility. In order to determine the most cost effective approach of achieving a boiler
NOx emission level of 200 mg/Nm3 Aem Milano entered into a tailored collaboration
contract with the Electrical Power Research Institute (EPRI). Under contract to EPRI
the Power Technology Department of PowerGen pic have undertaken studies to
predict the NOx control potential of a wide range of NOx reduction techniques, and
combinations thereof, when applied to Cassano d'Adda Unit 2. Furnace
computational fluid dynamics modelling and limited boiler performance modelling
aspects of the studies were undertaken by the Energy and Environmental Research
Corporation (EERC).
-------
This paper presents an overview of the NOx studies undertaken by Power Technology
and the predicted NOx reduction potential and estimated capital/operating costs of the
various NOx control regimes considered as applied to Cassano d'Adda Unit 2.
Cassano d'Adda Unit 2: Plant Description
Cassano d'Adda Unit 2 is an oil/gas opposed fired 320 MWe boiler manufactured by
Ansaldo. Eighteen burners are arranged in three by three arrays on the furnace front
and rear walls. The boiler operates with a pressurized furnace.
The combustion air is preheated to a temperature of approximately 270°C by two
Lungstrom type airheaters. The boiler is fitted with electrostatic precipitators.
A general arrangement of the boiler is shown in figure 1.
At full load the boiler generates 289 Kg/s of main steam at 540°C and 175 bar, and
231 Kg/s of reheat steam at 540°C and 34 bar. Reheat and superheat steam
temperatures may be controlled by the admission of recycled flue gas through the
furnace hopper at low loads and by the use of attemperation sprays at full boiler load.
Current boiler NOx emissions when firing a fuel oil containing 0.5% fuel bound
nitrogen are as follows:
Boiler Load NOx Emissions Excess Oxygen
MWe mg/Nm3 @ 3% 02 Ib/MMBtu %
320 1050 0.684 1.2
230 765 0.498 1.8
160 580 0.378 2.8
Full load NOx emissions when firing natural gas are 815 mg/Nm3 (corrected to 3%
oxygen, dry)(0.512 Ib/MMBtu) when operating at an excess oxygen level of 1.2%.
NOx Control System Design
The main design features of the various NOx control systems are described in the
following sections. The design of the overfire air and gas reburning systems are
based on the modelling activities undertaken by EERCni.
Recirculated Flue Gas System
A proposed flue gas recirculation system would provide flue gas for injection into the
combustion air supply and/or as a transport medium for the reburn fuel. Plant data
from Cassano Unit 2 indicated that with a recirculated flue gas flowrate of 15% new
FGR fans would be required to provide sufficient FGR pressure.
-------
The secondary air ductwork downstream of the proposed FGR injection point was
considered to be sufficient to promote satisfactory mixing between the combustion
air and the recirculated flue gas, therefore no mixing devices were deemed necessary.
Overfire Air (OFA) System
An overfire air system at Cassano was considered both as a stand alone NOx control
technology or to provide the balance of secondary air necessary for combustion
completion downstream of a reburn NOx reduction zone.
For either case the overfire air supply would be taken from the existing FD balance
duct for ease of access and to reduce costs. Isolating dampers would be incorporated
at the OFA take off points.
For the reburn case, an OFA ring main elevation was selected to avoid relocation of
existing electrical cabinets. OFA flow measurement would be provided using multi
port pitot tubes located in the supply ductwork to the OFA ring main.
A total of 12 OFA ports (6 on the front wall and 6 on the rear wall) with a nominal
port diameter of 300 mm was deemed sufficient for CO burnout for either the reburn
or OFA only cases. The OFA ports would be of simple design (no swirl) fitted with
individual modulating dampers. No OFA port tilt or yaw facility would be included.
An analysis of the estimated pressure losses through the proposed ductwork
arrangement indicated that booster fans would not be necessary.
Gas Reburning (GR) System
A reburn zone stoichiometry of 0.9 was selected (expressed as the ratio of the total
air supplied to the main burner and reburn zones to the stoichiometric air requirements
of the primary and reburn fuels). To achieve this stoichiometry approximately 15%
of the thermal input to the furnace would be provided by the natural gas reburn fuel.
For a reburning system which could utilise up to 5% FGR as an inert carrier medium
four 200 mm diameter reburn nozzles on each of the furnace front and rear walls were
proposed. The OFA ports included with the reburn system are essentially the same
as the stand alone system, except that the OFA nozzles are located at a higher
elevation.
Main burner zone and burn-out zone stoichiometries have been assumed to be 1.06
in both cases. Reburn zone and burn-out zone residence times have been calculated
to be approximately 350 milliseconds and 400 milliseconds, respectively.
Natural gas would be supplied from the existing supply system to a reburn fuel
metering and control station, the required gas supply pressure being within the
capabilities of the existing station equipment.
-------
Low NOx Burners (LNB)
A large variety of low NOx burner designs available from European burner
manufacturers were considered. For the purposes of the Cassano Unit 2 NOx control
studies it has been assumed that the low NOx burner design installed would
incorporate separate primary, secondary and tertiary combustion air streams, with
independent control of secondary and tertiary air swirl level and flowrate. The
required furnace wall aperture which varied considerably between designs, might be
as high as 1350 mm, which would necessitate extensive pressure part modifications
to the furnace walls.
Selective Catalytic Reduction (SCR)
The design requirements of the Cassano Unit 2 SCR system were assumed to be as
follows:
• The system would be located in a high dust environment, that is, upstream of
the dust arrestment plant.
• The system must reduce NOx emissions to 200 mg/Nm3 (0.130 Ib/MMBtu).
• The process reagent would be 25-30% aqueous ammonia solution.
• Ammonia slip must be limited to 3 vppm.
Three SCR system arrangements were considered:
(1) A dedicated SCR reactor chamber between the economizer outlet and the
airheater inlet with associated ductwork modifications.
(2) An "in duct" SCR system utilising the existing economizer outlet ductwork and
incorporating a reduced quantity of catalyst material.
(3) Replacement of a section of the airheater elements with catalyst coated
material (CAT-AH).
For the dedicated SCR reactor, using titanium/vanadium impregnated with oxides of
molybdenum and tungsten catalyst material, the proposed reactor size was defined
as follows:
Catalyst pitch (dependent on type) (mm) 4.9
Specific Area As (m2/m3): 650
Max recommended linear velocity m/sec: 5
-------
Initial catalyst complement: 2 tiers of 2 layers
Maximum catalyst complement: 3 tiers of 2 layers
(including the installation of 1 reserve layer)
Catalyst layer depth 0.55 m
Nominal full load flue gas flow rate (nrrVmin): 14,500 (241.7 m3/sec)
(1 atm = 1.01325 Bar) @ 20°C, 1.01325 Bar
and 1.2% O2
Reactor dimensions were calculated to be as follows:
• height 16.0 m
• width 12.2m
• depth 12.0 m
A nomogram of catalyst performance, relating NOx reduction efficiency, area velocity,
and ammonia slip for the specimen material was used to determine the optimum
operating conditions. Considerations relating to system non-idealities and marginal
variables such as SO2 oxidation rate have been neglected since these would represent
relatively minor amendments to the overall system design and would be less
significant than the variation in the performance specification of catalyst materials.
NOx Reduction Potential of Various NOx Control Technologies
The NOx reduction potential of the various NOx control technologies was assessed
using a variety of techniques:
• Discussions with equipment manufacturers.
• Data correlated from similar retrofit applications.
• Modelling work (undertaken by EERC).
• Correlations from PowerGen's in-house low NOx development programme.
The NOx reduction potentials stated are considered to be realistic/conservative
estimates of the performance which would be achieved in practice. In all cases the
potential NOx reductions have been estimated such that carbon monoxide emissions
are maintained below 200 mg/Nm3 and the flue gas dust burden upstream of the dust
arrestment plant would not exceed 150 mg/Nm3 (corrected to 3% oxygen, dry).
With the above defined basis, the NOx reduction potential of the NOx control
technologies considered were estimated to be as stated below:
-------
Low NOx Atomizers (LNA)/Directional Gas Spuds (DCS)
Industry data conclude that NOx reductions of approximately 15% are achievable by
installing fuel staged atomizer tips without increasing the flue gas solids burden.
Fuel staging when gas firing may be achieved by replacing the existing gas spuds with
directional spuds. However, there is no data in the public domain which quantifies the
NOx reduction which may be achieved by this technique. Increased stability and the
absence of a particulate control problem when gas firing usually allows increased fuel
staging than when oil firing, however this is offset by the fact that staging would only
be effective as a mechanism of lowering thermal NOx via peak flame temperature
reduction (eg no fuel NOx credit). In the absence of any supporting data it has been
assumed that similar NOx redutions achieved by fuel staging in oil flames could be
achieved. This level of 15% is considered to be a conservative estimate.
Flue Gas Recirculation (FGR)
Based on reported data from pilot scale trials121 and boiler plant13"41 the injection of
recirculated flue gas into the combustion air supply can provide NOx reductions of
1.5% for every 1 % of FGR used when oil firing, increasing to 2.5% per 1 % of FGR
for gas firing. (The FGR rate is expressed as a percentage of the furnace
stoichiometric flue gas flow on a mass basis).
The actual amount of FGR which can be utilised on a specific boiler will be limited by
flame stability, boiler vibration, carbon monoxide and/or particulate emissions, and
steam temperature control considerations. For the Cassano Unit 2 studies the
maximum FGR flowrate was limited to 15% as boiler vibration had been experienced
at FGR flowrates above this level in the past.
The NOx reductions estimated for flue gas recirculation for the Cassano Unit 2 studies
were:
Oil firing Gas firing
Existing Burners 20% 37%
Low NOx Burners 10% 37%
Overfire Air (OFA)
The overfire air rate selected for Cassano Unit 2 was 15% to provide a burner zone
stoichiometry of 0.9.
When used in combination with FGR injection into the combustion air supply the NOx
reduction potential of OFA (whilst maintaining satisfactory combustion conditions)
was predicted to be:
-------
Oil firing Gas firing
Existing Burners 20% 35%
Low NOx Burners 16% 32%
Low NOx Burners (LNB)
A total of ten low NOx burner suppliers/manufacturers were contacted and requested
to provide details of the burner design that would be proposed for the Cassano d'Adda
boiler together with the NOx, CO and particulate emissions performance they would
be prepared to guarantee.
Although nine of the burner suppliers/manufacturers provided responses, in many
cases only predicted emissions performance data were supplied.
For oil firing (0.5% fuel nitrogen) predicted NOx emissions from the various low NOx
burner suppliers ranged from 400 mg/Nm3 (0.260 Ib/MMBTU) to 670 mg/Nm3 (0.436
Ib/MMBtu). For gas firing NOx emissions ranging from 150 mg/Nm3 (0.094
Ib/MMBtu) to 500 mg/Nm3 (0.314 Ib/MMBtu) were predicted.
If flue gas recirculation was applied in conjunction with low NOx burners NOx
emissions in the range 360 mg/Nm3 (0.234 Ib/MMBtu) to 500 mg/Nm3 (0.326
Ib/MMBtu) for oil firing and from 150 mg/Nm3 (0.094 Ib/MMBtu) to 350 mg/Nm3
(0.220 Ib/MMBtu) for gas firing were predicted.
Based on an evaluation and interpretation of the responses received the following NOx
emission levels were deemed achievable:
Oil firing Gas firing
mg/Nm3 Ib/MMBtu mg/Nm3 Ib/MMBtu
Low NOx Burners 480 0.312 350 0.220
LNB + FGR 430 0.280 220 0.138
Gas Reburning (GR)
The predicted NOx reduction potential of gas reburning was based on the modelling
work undertaken by EERC. EERC suggested a gas reburning arrangement in which
15% of the total furnace thermal input was supplied by the natural gas reburn fuel,
providing a reburn zone stoichiometry of 0.9. The reburn zone residence time was
approximately 0.4 seconds (plug flow basis).
The NOx reduction performance predicted for gas reburning by EERC included
allowances for non ideal characteristics of the proposed system, based on their
previous experience. Factors considered included flame encroachment into the reburn
zone and imperfect mixing.
-------
Performance predicted by the burner suppliers for the oil firing case differed markedly
to that predicted by EERC. To resolve this descrepency the ratio of reburn zone NOx
inlet concentration and percentage NOx reduction in the reburn zone predicted by
EERC's modelling was maintained. As the reburn zone stoichiometry, residence time
and temperature would be essentially unchanged this approach was considered to be
justified. The NOx inlet concentration to the reburn zone was assumed to be the NOx
emission level at 85% boiler load (270 MWe) without gas reburning applied. A similar
approach was adopted to estimate NOx emissions at part boiler load following the
application of gas reburning.
Selective Non Catalytic Reduction (SNCR)
SNCR was excluded from the analyses at the request of Aem Milano, primarily due
to concerns regarding N20 emissions.
Selective Catalytic Reduction (SCR)
For the SCR system configurations considered the NOx reduction levels achievable for
the oil firing case were estimated to be:
• CAT-AH: 10% (based on a NOx inlet concentration of 250 mg/Nm3)
• In Duct: 55%
• SCR Reactor: 90%
The technical and economic analyses undertaken assumed that an in-duct SCR system
would be utilised for NOx reduction of up to 55%, and a dedicated SCR reactor
system for NOx reductions in excess of 55%.
Post Retrofit NOx Emissions
Predicted post retrofit NOx emissions are presented in Table 1. NOx emissions from
low NOx burners combined with various other NOx control techniques are presented
in Tables 2 and 3 for 0.5% nitrogen fuel oil and natural gas firing, respectively.
Tables 1 and 2 confirm that with a 0.5% nitrogen content fuel oil none of the
combustion modifications considered would be capable of achieving a NOx level of
200 mg/Nm3. A sensitivity analysis undertaken to investigate the effect of main
burner zone operating conditions (excess air level, FGR flowrate, top burner rows out
of service) indicated little impact on the predicted boiler NOx emissions. Therefore,
it was concluded that to obtain the target boiler NOx emissions level a SCR system
would be required, the design of which would depend on the extent of combustion
modifications undertaken.
-------
However, in the future Aem will be required to fire low sulphur content (0.25%) fuel
oils which will have a nitrogen content of approximately 0.2%.
Predicted NOx emissions when firing a 0.2% nitrogen fuel oil are presented in Tables
4 and 5 for the existing burners and low NOx burners respectively. The same data
are presented in graphical form on figures 2 and 3. The effect of fuel nitrogen content
on NOx emissions was estimated from data provided by burner suppliers and the
available published data.
Examination of Tables 4 and 5 shows that the only combined combustion modification
NOx control techniques which may be capable of achieving the target NOx emission
level are low NOx burners combined with recirculated flue gas injection plus gas
reburning. The predicted NOx emission level for this firing arrangement is 207
mg/Nm3.
The gas reburning system modelling undertaken by EERC incorporated allowances for
non-ideal characteristics which accounted for an increase in NOx emissions of
approximately 50 mg/Nm3 over an ideal system. Non-ideal characteristics could be
minimised by sizing the main burners for a maximum boiler load of 270 MWe (to
minimise primary flame encroachment into the reburn zone) and by using recirculated
flue gas as a reburn fuel carrier medium to promote optimised mixing. Additionally it
may be possible to reduce the reburn zone stoichiometry.
It was considered that a NOx control system configured in this manner would have
an excellent chance of achieving the target NOx emissions level of 200 mg/Nm3.
If the existing burners were retained a SCR system would be required to achieve the
target NOx emissions level when gas firing. However, if low NOx burners were
installed in combination with recirculated flue gas injection the boiler NOx emissions
would be approximately 220 mg/Nm3. In practice it is believed that this arrangement
would have a good chance of achieving the target NOx emission level. Additionally,
if OFA or GR were used for NOx control when oil firing the OFA ports could be utilised
to provide further NOx reduction when gas firing.
NOx emissions when gas firing with low NOx burners and various other NOx control
techniques are presented as a function of boiler load on figure 4.
Capital and Operating Costs
On completion of the technical analyses for the various NOx control system Aem
Milan selected a number of firing arrangements for economic analysis151.
Capital costs were estimated using a variety of techniques:
• quotations obtained from equipment suppliers
-------
• output from the Questimate cost estimating software package
• PowerGen's in-house experiences
• published costs of NOx control systems on plant of similar size
• costs estimated by EERC.
Operating costs take account of predicted boiler efficiency levels and power and
auxiliary steam requirements (expressed as additional fuel costs).
SCR reagent costs were calculated from the estimated reagent flow multiplied by the
reagent costs. Catalyst costs are calculated from the estimated catalyst volume
multiplied by the catalyst cost, divided by the estimated catalyst life.
Boiler efficiency penalties associated with the various NOx control system were
calculated by EERC (FGR and FGR + GR) or estimated from published data (SCR).
In the absence of any boiler performance modelling results the efficiency loss to be
attributed to a combination of FGR and OFA was assumed to be the same as for FGR
in isolation. Efficiency losses attributed to FGR plus OFA were assumed to be the
same for the existing burners and low NOx burners. Unburnt carbon losses were
considered to be insignificant and have been neglected. However, an increase of
0.25% oxygen on the operational excess oxygen level was assumed when OFA is
applied (oil firing case only). For the natural gas firing cases it was assumed that
identical excess air levels would be used for both the existing burners and low NOx
burners. To provide operating costs on an annual basis a boiler load factor of 70%
was assumed (full boiler load for 70% of the year).
Costs for fuels, electricity and reagents were provided by Aem Milano.
The accuracy of the cost data are believed to be:
• Capital costs, combustion modifications 10%/ + 30%
• Capital costs, Selective Catalytic Reduction ±30%
• Operating costs, combustion modifications ±20%
• Operating costs, dedicated SCR ±35%
• Operating costs, "in-duct" SCR ±50%
The estimated capital costs and annual operating costs are presented in Table 6.
Selection of Most Cost Effective NOx Control Regime
The economic analyses indicate that the installation of low NOx burners in
combination with FGR injection into the combustion air supply would be extremely
cost effective.
-------
When gas firing this NOx control regime has an excellent chance of achieving the
target NOx emission level of 200 mg/Nm3, representing significant capital and
operating cost savings.
When oil firing with a LNB, FGR, OFA plus SCR NOx control regime the capital cost
saving would be significant compared with the existing burners plus SCR case.
However, annual operating costs would be similar. There would also be a capital cost
saving compared to the LNB, FGR, GR plus SCR NOx control regime. However, in this
case the difference in capital costs would rapidly be paid back by the significantly
lower operating costs associated with the LNB, FGR, GR plus SCR arrangement.
The LNB, FGR, GR plus SCR NOx control regime is considered to be the most cost
effective approach to guarantee achieving the target NOx emissions level when oil
firing. However, there is an excellent chance of achieving a NOx emissions level of
200 mg/Nm3 without recourse to SCR when firing a 0.2% nitrogen content fuel oil.
It was concluded that the best approach would be to install low NOx burners, FGR
injection into the combustion air supply and a gas reburning system initially; if the
target NOx emissions level is not achieved an 'in-duct' or airheater based SCR system
could be added at a later date.
Summary
Based on the results of the technical and economic analyses Aem Milano selected a
NOx control system based on a combination of low NOx burners, recirculated flue gas
injection into the combustion air supply and natural gas reburning. Aem Milano are
currently involved in procurement activities associated with the project. It is
understood that Aem Milano will be requesting a turnkey contract and will be inviting
international bids.
It is programmed that Cassano Unit 2 will be taken out of service in 1997 for erection
of the NOx control equipment and is expected to return to service in 1998.
It should be noted that in general the economics of NOx control will be influenced by
the boiler design (furnace dimensions, ease of access, auxiliary plant capabilities etc),
the absolute NOx emissions level and NOx reduction required, and local fuel and
reagent costs. As such the most cost effective means of achieving the target NOx
emission level of 200 mg/Nm3 on Cassano d'Adda Unit 2 may or may not be
appropriate for boiler plant of a different design at a different location.
Acknowledgements
This paper is presented with the permission of PowerGen pic and Aem Milano
-------
References
"Evaluation of NOx Reduction by Combustion Modification (Low NOx Burners
and Gas Reburning) Level 1. Azienda Energetica Municipale Milano:
Cassano d'Adda Unit 2". Final Report, June 1994. Energy and Environmental
Research Corporation.
"The Reduction of Atmospheric Pollutants During the Burning of Residual Fuel
Oil in Large Boilers". Journal of the Institute of Fuel. March 1978. ATS
Cunningham, P J Jackson.
"Demonstration of Advanced Low NOx Combustion Techniques at the Gas/Oil
Fired Flevo Power Station Unit 1". 1991 Joint Symposium on Stationary
Combustion NOx Control - EPA/APRI. J G Witkamp et al.
"Retrofit NOx Control Guidelines for Gas and Oil Fired Boilers". EPRI TR-
102413. December 1993. B Carmine.
"Aem Milano: Task 2: Economic Analyses of NOx Control Systems for
Cassano d'Adda". PT/94/220065/R. November 1994. P Baimbridge, M
Garwood.
-------
Table 1
Summary of Predicted NOx Emissions from Existing Burners in Combination
with Various NOx Control Techniques: 0.5% Nitrogen Oil and Natural Gas Firing
Oil Firing
Gas Firing
Boiler load (MWe)
320 230 160
320 230 160
Technology NOx Emissions (mg/Nm3)
Existing Burners (EB)
EB
EB,
EB,
EB,
EB,
EB,
+ SCR
LNA,
LNA,
GR
FGR,
LNA,
FGR,
FGR,
GR
FGR,
OFA
OFA + SCR
GR
1050
200
600
200
395
310
295
765
200
390
200
270
225
200
580
200
305
200
220
180
160
815
200
280
200
315
235
225
540
200
210
200
220
165
155
450
200
1
1
1
1
1
55
55
85
35
25
Table 2
Summary of Predicted NOx Emissions from Various
Combinations of NOx Control Techniques: 0.5% Nitrogen Oil Firing
Boiler Load (MWe)
320 230 160
Technology NOx Emissions (mg/Nm3)
Low NOx Burners (LNB)
LNB, FGR
LNB, FGR + SCR
LNB, FGR,OFA
LNB, FGR, OFA + SCR
LNB, GR
LNB, FGR, GR
LNB, FGR, GR + SCR
480
430
200
365
200
235
225
200
315
285
200
240
200
155
150
150
220
200
200
170
170
115
110
110
-------
Table 3
Summary of Predicted NOx Emissions from Various
Combination of NOx Control Techniques when Gas Firing
Technology
Boiler Load (MWe)
320
230
160
NOx Emissions (mg/Nm3
Low NOx Burners (LNB) 350
LNB, FGR 220
LNB, FGR + SCR
LNB, FGR, OFA 150
LNB, FGR, OFA + SCR
LNB, GR 218
LNB, FGR, GR 176
LNB, FGR, GR + SCR
230
145
180
130
Not required
110 85
Not required
145 100
147 145
Not required
Table 4
Summary of Predicted NOx Emissions for Existing Burners in
Combination with Various NOx Control Techniques: 0.2% Nitrogen Fuel Oil
Technology
Boiler Load (MWe)
320
230
160
NOx Emissions (mg/Nm3)
Existing Burners (EB) 960 620 490
EB + SCR 200 200 200
EB, LNA, FGR, OFA 550 355 280
EB, LNA, FGR, OFA + SCR 200 200 200
EB, GR 355 240 195
EB, LNA, FGR, GR 270 185 150
-------
Table 5
Summary of Predicted NOx Emissions from Various
Combinations of NOx Control Techniques: 0.2% Nitrogen Oil Firing
Boiler Load (MWe)
320 230 160
Technology Nox Emissions (mg/IMm3)
Low
LNB,
LNB,
LNB,
LNB,
LNB,
LNB,
LNB,
NOx
FGR
FGR
FGR
FGR
GR
FGR
FGR
Burners
+ SCR
, OFA
(LNB)
, OFA + SCR
, GR
, GR +
SCR
400
357
200
307
200
216
207
260
237
200
205
200
147
140
1
1
1
1
1
95
76
76
50
50
109
1
03
Not required
-------
Table 6
Estimated Capital and Annual Increase in Operating
Costs for Various NOx Control Regimes (Figures in brackets are
annual decreases in operating costs compared to the Baseline Case)
Estimated Capital and Annual
Operating Costs I1>(2) (GBP)
Technology
EB + SCR
EB, LNA, FGR, OFA
+ SCR
LNB, FGR, OFA
LNB, FGR, OFA +
SCR
LNB, FGR, GR
LNB, FGR, GR (FGR
used as reburn fuel
transport medium)
Main
Fuel
Oil
Gas
Oil
Gas
Oil
Gas
Oil
Oil
Oil
Existing
Capital
1 2000000
1 1 300000
12578600
5390600
N/A
N/A
N/A
N/A
N/A
Burners
Operating
1317000
109330
936240
(275540)
N/A
N/A
N/A
N/A
N/A
Low NOx
Capital
N/A
N/A
N/A
N/A
2794400
2794400
6794400
3230500
3394300
Burners
Operating
N/A
N/A
N/A
N/A
274170
(550540)
1014170
(46160)
(15745)
LNB, FGR, GR + Oil N/A N/A 7394300 131755
SCR (FGR used as
reburn fuel
transport medium)
LNB, FGR, GR Oil N/A N/A 3336000 (15745)
(Integrated system,
FGR used as reburn
fuel transport
medium)
(1) Change in operating costs compared to Baseline Case (Existing burners, 0.25%
sulphur oil firing).
(2) Boiler load factor assumed to be 70%.
-------
POWERGEN pic POWER TECHNOLOGY DEPARTMENT
~l I.I I.I I I "I I
II11 [I II II
GA OF CASSANO D'ADDA UNIT 2
P.G. 234
-------
POWERGEN pic, POWER TECHNOLOGY DEPARTMENT
000
EB
EB, LNA, FGR, OFA
EB, GR
EB, LNA, FGR, GR
200
BOILER LOAD MW
FIGURE
No.
2
PREDICTED NOx EMISSIONS FOR EXISTING BURNERS COMBINED
WITH VARIOUS NOx CONTROL TECHNIQUES : 0.2% NITROGEN FUEL
-------
POWERGEN pic, POWER TECHNOLOGY DEPARTMENT
450.
LNB
LNB, FGR
LNB, FGR, OFA
LNB, GR
LNB, FGR, GR
100
200
BOILER LOAD MW
FIGURE
No.
3
PREDICTED NOx EMISSIONS FOR LOW NOx BURNERS COMBINED
WITH VARIOUS NOx CONTROL TECHNIQUES : 0.2% NITROGEN FUEL
-------
POWERGEN pic, POWER TECHNOLOGY DEPARTMENT
LNB
LNB, FGR
LNB, GR
LNB, FGR, GR
LNB, FGR, OFA
200
BOILER LOAD MW
FIGURE
No.
A.
PREDICTED NOx EMISSIONS FOR LOW NOx BURNERS COMBINED
WITH VARIOUS NOx CONTROL TECHNIQUES : NATURAL GAS
-------
REDUCTIONS OF NOx EMISSIONS ON OIL AND GAS FIRING AT
BOWLINE UNIT 1
Alan E. Paschedag
Senior Mechanical Engineer
Bums and Roe Company
Roger A. Martinsen
Project Manager
Burns and Roe Company
Raymond C. O'Sullivan
Manager Power Engineering
Orange and Rockland Utilities, Inc.
Douglas W. Schmidt
Senior Engineer
Orange and Rockland Utilities, Inc.
Daniel V. Giovanni
President
Electric Power Technologies
Anthony V. Conti
Manager Easter Operations
Electric Power Technologies
Introduction
In response to the NYSDEC, Part 227 regulations for the emissions of nitrogen oxides (NOJ,
Orange and Rockland Utilities, Inc. (ORU) and Bums & Roe Company (BRC) evaluated the options
available to reduce the NOX emissions at two oil and gas fired units at Bowline Point Generating
-------
Station. Replacement of all of the existing burners with new low NOX burners and possibly overfire
air ports presents the most costly method of achieving this goal. Therefore, other methods of NOX
reduction were considered including utilizing some form of off-stiochiometric, burners out of service
(BOOS), firing.
It was determined that the stringent emission limits could be met utilizing off-stoichiometric firing
techniques. New oil gun atomizer tips allowing off-stoichiometric firing with mechanical
atomization and swirlers of a new design are replacing the existing atomizers. The new hardware
eliminates the problems of opacity while operating with off-stoichiometric firing.
Unit Description
Bowline Point Unit 1 is a 600 megawatt Combustion Engineering tangentially fired balanced draft
unit firing heavy fuel oil (HFO) and natural gas. The unit is designed for a main steam flow of
4,200,000 Ibs/hr at 2,600 psig and 1005°F. Reheat steam flow is 3,879,000 Ibs/hr at 607 psig and
1005°F. The unit has a flue gas recirculation fan which injects flue gas into the furnace hopper for
reheat steam temperature control. There is no flue gas clean up equipment installed on the back end
of the unit. There are five elevations of burners for each of the four corners. This unit has straight
simplex mechanical atomizing heavy fuel oil guns. The fuel oil guns are designed for 1100 psig oil
pressure at the burner. C-E nozzles are also provided to fire natural gas (see Figure No. 1).
History of NOx Reduction Implementation Program
The New York State Department of Environmental Conservation (NYSDEC) requires every owner
or operator of large boilers to comply with emission limits by May 31, 1995, (see Table No. 1). This
requires Orange and Rockland Utilities to have a NOX Reduction Implementation Program.
The units included in the NOX reduction implementation program are Lovett Units 3, 4 and 5 and
Bowline Point Units 1 and 2. Lovett Unit 3 is a 69 MW with a low capacity factor. The overall
contribution of this unit to the NOX emissions to the overall NP emissions of the utility are
relatively small.
Lovett Units 4 and 5 are wall fired units using coal as a main fuel source. It was concluded that
retrofitting low NOX burners and an overfire air system at both of these units is a necessary part of
the overall plan in reducing NOX.
Bowline Point Units 1 and 2 offered the opportunity to reduce NOX by the required amount at a more
reasonable cost. Results of testing were reviewed as well as available components necessary to
modify the existing burners to allow low NOX firing operation. This resulted in the selection of
modifying the existing burners for low NOX operation and operating with BOOS, if required to meet
the emission targets.
-------
Technical Discussion of Off Stoichiometric Firing
Combustion control of NOX consists of controlling the air/fuel ratio, flame temperature and delaying
mixing in the near-burner zone. This effectively reduces thermal NOX and, to a lesser degree, fuel
NOX. One form of this type of combustion control technology is burners-out-of-service (BOOS)
operation.
BOOS, an off Stoichiometric combustion technology, has been in use for approximately 20 years for
predominantly oil and natural gas fired units. BOOS produces off Stoichiometric combustion by
terminating the fuel flow at selected burners, while increasing the fuel flow to the remaining burners,
to maintain the same total fuel flow, and load, on the unit. Burner air flows are typically left
unchanged, so that the burners-out-of-service are supplying air only. The remaining burners are
operated fuel-rich. For fuel-rich flames, the rate of NOX formation is reduced due to the decrease hi
O2 concentration. The combustion products from the fuel-rich burners lose heat by radiation and
mixing with the bulk gas in the boiler ("internal gas recirculation"). Mixing within the bulk gas
brings in the additional combustion air from the burners to complete the combustion process away
from the primary combustion zone (near the burners). In effect, BOOS tends to lengthen the flames.
Thus, the conditions necessary for NOX formation, excess O2 and high flame temperature, have been
reduced. To optimize NOX reductions by off Stoichiometric combustion, the out of service burners
must be selected to provide the necessary gas mixing between the unburned fuel from the fuel rich
burners and the air from the air-only burners. The optimum BOOS pattern is one which minimizes
NOX emissions, results in low excess Q requirements (for economy), results in an average CO
concentration in the flue gas below some desired maximum (usually 100 ppm) and causes no adverse
boiler operating characteristics (i.e., excessively low or high steam temperatures, vibration, smoke
etc.).
BOOS operation generally results in a deterioration of the unit opacity and particulate levels.
Therefore, the NOX reduction achievable is indirectly related to the resulting opacity of the unit. In
order to maximize the NOX reduction achievable, modifications to the existing burners are necessary.
Until recently, the only significant option was to convert the burners to steam atomization. This
conversion is not only costly to install but incurs operating costs on a continual basis.
Development of an oil atomizers and swirlers designed for mechanical atomization has been jointly
undertaken by EPRJ, Eseerco, Consolidated Edison and Electric Power Technologies (EPT). This
tip proposes to achieve the required NOX reduction, while minimizing the effects on unit opacity.
This tip has been installed on a number of units, for opacity control. The main objective of the
installations to date has not been NOX reduction.
Evaluation of NOx Reduction for Bowline Point Unit 1
Overview of Available Data
Off Stoichiometric firing operation was recommended by BRC for Bowline Point Unit 1 to attain a
-------
reduction of the NOX emission rate required to support the system NO reduction plan. The
anticipated reduction in NOX is approximately 20-30%. To support this mode of operation, the
installation of steam atomized oil guns was recommended. This would allow attainment of lower
NOX emissions while maintaining acceptable levels of opacity and particulate emissions.
Data from a number of sources has been reviewed by BRC to determine the potential NOX reduction
that would result due to off stoichiometric firing operation. A summary of the NOX emissions
resulting from this data has been graphically summarized in Figure No. 2. The sources of the data
are listed below:
Original baseline NOX emissions data
Additional baseline NOX emissions data
ETEC Predicted NOX - BOOS
ETEC Predicted NOX - BOOS + FOR
• FERCO study using a computerized
Industry Wide Experience Database
BOOS Only
FOR Only
BOOS + OFA + FOR
The predictions and data demonstrate that it is highly probable that with off stoichiometric firing,
NOX can be reduced and the unit will operate consistently with NOX emissions below 0.24 Ibs per
million BTU on oil firing.
The baseline emissions test data of July 1993 for Bowline Point Unit 1 was reviewed. Although only
a limited amount of data was collected with BOOS operation, most of the data taken appears to
demonstrate that NOX emissions of less than 0.24 Ibs/MBTU are achievable on oil firing. A summary
of the data used to determine the NOX reduction possible is provided in Table No. 2.
hi the review of the additional emission test data, we note that the fuel oil used during the test had
a nitrogen content of 0.52%. This is considered to be on the high end of the normal range received
and provides additional confidence in the NOX reduction possible with off STOICHIOMETRIC
firing. The possibility that a higher nitrogen content fuel delivery will result in non-compliance
operation is reduced.
It should be noted that a similar analysis for firing on gas has not been performed. Since the main
fuel being fired is No. 6 oil, NOX emissions firing gas are not anticipated to cause a compliance
problem.
The option of adding windbox flue gas recirculation (WBFGR) to the unit was considered. This
would require adding two (2) recirculating flue gas ducts and partitions to the existing hot
combustion air ducts and windbox to supply flue gas to the windbox and hot air only to the auxiliary
air compartments in the windbox (See Figure No. 3). It was determined that the total dynamic head
of the existing flue gas recirculation fan would have to be increased. This would involve adding a
-------
new flue gas recirculation fan, motor, wiring, motor starter etc. To determine the cost effectiveness
we estimated that the capital cost to add WBFGR is approximately $1,600,000 which includes the
cost of installing a new flue gas recirculation fan. Referring to Figure No. 2, we believe that
WBFGR is not required to meet compliance.
Selection of Technical Approach
The installation of the steam atomization system necessary to support steam atomized burners is
costly to install. In addition, the use of steam for atomization incurs a constant operational cost. Due
to these facts, other approaches were sought which would allow the unit to achieve the desired NOX
reduction without these costs.
It was determined that Electric Power Technologies (EFT) offered a modification to the existing
system that would achieve the goal of NOX reduction without steam atomization. The equipment
offered was EPT's REACH technology. The operating experience and technical merits of this
equipment were reviewed.
Evaluation of EPT Technical Approach
EPT Oil Burner Modification Experience
EPT's past experience consists of burner modifications with the objective improving unit operation.
The utilities contacted listed opacity problems and lack of sufficient turndown as the reasons for
installing EPT equipment. EPT has recently developed and demonstrated the low NOX version
atomizer. This project will be the first full scale demonstration of the low NOX mechanical atomizer.
This technology is called Reduced Emission And Combustion Hardware (REACH). Only one
tangentially fired application of the EPT equipment was for the purpose of NOX reduction.
EPT's past experience includes two 146 MW Combustion Engineering tangentially fired units which
were experiencing opacity problems that limited the rate of start-up and load ramping. Opacity was
also a problem during steady load. EPT oil gun atomizer tips and diffuses were installed and the
igniters were modified. The top row of oil burners was relocated and the position used for close
coupled overfire air. The opacity problems were solved and a large margin now exists between the
operating condition and the compliance limit. Maximum ramp rates are achievable for both start-up
an load changes without opacity problems. As a result of the installation of the equipment, the NOX
was reportedly reduced by approximately 50%.
Another utility required that greater turndown, with all burners in service, at four Combustion
Engineering tangentially fired units. All units had a raring of between 80 MW and 90 MW. EPT oil
atomizers, diffuses and extenders were installed at each of the units. The existing unit turndown
allowed the unit to operate at 33% load with all burners in service. The objective was to increase the
turndown and allow the unit to operate at 27% load with all burners in service. The installation
proved successful but the life of the new atomizers is an issue which is being addressed by EPT and
the utility.
-------
A 400 MW Combustion Engineering unit was having problems complying with the compliance
limits for NOX. The project was implemented in two phases. The first phase included the installation
of EPT oil gun atomizer tips, diffusers and extenders. This equipment resulted in a more stable flame
and a NOX reduction of approximately 30%. It should be noted that the NQ during some periods
reached the preretrofit levels. Although this has not been sufficiently analyzed, the NOX was not
corrected for a constant fuel oil nitrogen content. Phase two consisted of increasing the capacity of
the lowest three rows of burners, to achieve full load. The upper burner row was used for close
coupled overfire air. This arrangement resulted in a NOX reduction of approximately 30%.
Technical Approach
The performance of any oil-fired burner is strongly influenced by two critical components: (1) flame
stabilizer, and (2) oil atomizer. Burner aerodynamics, including the shape and size of the internal
recirculation zone (IRZ) depend on the design of the flame stabilizer. An oil atomizer must produce
the correct size oil droplets and distribute these droplets within a spray cone (based on the size of
the IRZ) that promotes desired air-fuel mixing characteristics.
The flame stabilizer and atomizer represent a fraction of the cost of a burner assembly, but are
primarily responsible for performance measure in terms of:
• Light-off reliability without smoking
• Flame delectability
• Flame stability at low firing rates
• Flame shape to avoid impingement with burner throats and water walls
• Excess air requirements to prevent smoking and CO
• Unburned carbon levels and particulate matter mass emissions
• Opacity levels as a function of load and excess air
• NOX emissions
Burner and boiler turndown with all burners in service
Bowline Point Unit 1 has a tangentially-fired boiler, that is, the fuel and combustion air are
introduced into the furnace through four windbox assemblies located in the corners and directed
tangent to an imaginary cylinder at the center of the furnace. Each windbox assembly is
approximately 37 feet high and is divided into eleven (11) horizontal compartments: five (5) fuel
compartments and six (6) auxiliary air compartments. The fuel compartments each contain
equipment (e.g., oil guns, gas nozzles, air nozzles) for delivering the fuel and primary combustion
air. The auxiliary air compartments are located above and below each fuel compartment and deliver
secondary combustion air. The combustion air is proportioned among the fuel and air compartments
through a vertical array of louver dampers. The dampers leading to each fuel compartment are
typically opened 100% when the fuel components (i.e., oil gun or gas nozzle) is in service, and
closed when out of service. The dampers leading to the auxiliary air compartments are operated as
a group. These dampers are automatically positioned to control the windbox-to-furnace differential
pressure.
The air and fuel streams are also adjustable by means of movable burner nozzle tips that may be
-------
tilted + 30. This action effectively raises or lowers the flames (also referred to as the "fireball") in
the boiler as a means to shift heat and control steam temperatures.
The boiler is equipped with forced draft (FD) and induced draft (ID) fans and is operated in a
balanced draft mode. At maximum load, steam generation is limited by feedwater supply or ID fan
draft, and appears to have excess FD fan capability.
Fuel Firing Equipment. The fuel firing equipment consists of five (5) elevations denoted from
top to bottom A,B,C, D and E. Each firing elevation has a retractable oil gun assembly and gas
nozzles above and below the oil gun Atomization is accomplished using once-through, mechanical
atomizers (also referred to as "simplex" atomizers). Fuel oil flow (i.e., firing rate) is controlled by
varying the oil supply pressure to the oil guns. The oil piping system is rated at 1,200 psig, but at
full load (i.e., approximately 600 MW) the supply pressure is typically 1,100 psig. Elevation E may
also be operated using steam-atomized warm-up oil guns having reduced firing capacity.
Flame Stabilizers. At each burner, fame stabilization is accomplished using a combination
swirler and bluff body (also referred to as "extender"). These devices create a lower static pressure
immediately downstream and on the axis of the oil gun. The lower static pressure causes a
recirculation of hot combustion gases back towards the burner. This region of reverse flow is
denoted the internal circulation zone. This reverse flow of hot gases establishes the flame front and
stabilizes the flame. The swirlers and extenders are attached at the furnace end of each fuel
compartment nozzle. The existing swirlers are a flat-blade type and are known to have limited open
flow area and relatively high pressure drop. EPT will replace these swirlers with compound-curve-
blade swirlers having the same outside diameter. However, the EPT swirler will have a different
aerodynamic characteristic and lower pressure drop. The existing extenders are nominally 14 inches
in diameter and symmetrically surround the swirler. EPT will use a modified extender design that
will provide the necessary flame stabilization and help stage the combustion process for lower NOX
emissions.
Atomizers. Bowline Point Unit 1 is currently equipped with 3-piece (including sprayer plate, back
plate, and retaining nut) CE-type simplex mechanical atomizers. The atomizers operate at pressures
up to 1,200 psig.
This type of atomizer and operating mode is characterized by low-to-moderate turndown capability,
wider-angle sprays at low operating pressures, adequate spray quality at high operating pressures,
and poor spray quality at low operating pressures. Oil exits through a single hole on the axis of the
atomizer.
For purposes of improved combustion efficiency and turndown, EPT's atomizer design is of similar
type, but having improved atomization quality over the operating range. In addition, the atomizer
will be sized (i.e., capability, supply flow rate, return flow rate, and delivered flow rate) for the
maximum turndown possible without modification to the oil piping, pumping, and flow control
equipment. New atomizers will fit the existing oil guns.
-------
As a part of this program EPT is also designing and supplying special low-NOx atomizers that will
internally stage the flames (i.e., fuel-rich and air-rich regions within individual flames) through a
non-uniform spray of oil droplets exiting the oil atomizers.
Design Considerations for Minimum Particulate Matter Emissions and Opacity. The
flame stabilizers and atomizers for Bowline Point Unit 1 are primarily designed to promote carbon
burn-out. This is accomplished by:
• Utilizing swirlers and extenders designed for a specific swirl number and entrained
flow area that will produce a strong IRZ over the complete operating range of the
burners.
Utilizing atomizers with a spray angle that, when properly positioned relative to the
swirler, will promote mixing of the fuel and air at desirable fuel-air ratios and
establish a stable flame front.
• Utilizing atomizers that produce the desired size of oil droplets at the normal
operating pressures of the fuel system.
• Adjusting burner parameters (e.g., axial alignment of components, settings of air
dampers) for optimum performance of individual burners.
Defining an optimum excess air versus oil-flow relationship (i.e., excess O2 versus
steam flow curve) for the automatic combustion control system.
• Defining fuel-air and aux-ir damper positions over the load range based on field
testing to optimize combustion conditions and emissions.
Design Considerations for Minimum NOX Emissions. Bowline Point Unit 1 is not equipped
with overfire air (OFA) ports or flue gas recirculation (FOR) to the windbox for purposes of NOX
emissions control. FGR to the hopper, which does not affect NOX emissions, is used for steam
temperature control and will not be utilized in this project.
The control of the top and bottom auxiliary air compartments will be modified for purposes of
minimizing NOX emissions and promoting carbon burnout. The control of the louver dampers
feeding these two compartments will be decoupled from the other intermediate-level auxiliary air
compartment dampers. Combustion air flow will be increased to the top auxiliary air combustion
by manually biasing the corresponding louver dampers to more closed positions. Final positions will
be determined during field testing after the combustion equipment modifications are made.
The technical feasibility, requirements, and potential benefits for de-coupling the automatic tilt drive
mechanism from the top auxiliary air compartments will be evaluated prior to implementation. The
purpose would be an increased separation of the top most air streams to the boiler for purposes of
NOx reduction.
-------
The Segmented V-Jet atomizer produces a spray pattern resulting in an internally-staged flame.
Although the Segmented V-Jet requires steam atomization, a mechanical atomizer is currently being
developed that produces a similar spray pattern to the Segmented V-Jet. In addition to the
conventional simplex atomizer, EPT will design and supply prototypes of a low-NOx mechanical
atomizer for evaluation at Bowline Point Unit 1.
Burner Modification Scope of Supply
The modification of the existing burners involves new firing equipment, component testing, system
tuning and emission testing.
The design of the equipment includes engineering required for the following items:
• Review of existing operating and maintenance data.
• Analysis of burner aerodynamics for various burner settings and flow rates.
Design review of standard EPT swirler to existing fuel compartment.
• Review of the oil supply system to define operating limits and turndown.
• Decoupling top auxiliary air compartment for close coupled overfire air.
• Biasing bottom auxiliary air compartment for sub-stoichiometric flames.
Laboratory testing is also included for the development and quality control of the oil atomizers.
These tests will be conducted during the design phase as the basis for obtaining the final nozzle
configuration for the Bowline Unit #1. Tests will be run after manufacture of the atomizers to insure
that all atomizers have been manufactured within the tolerances for operation required for even
firing. The tests that will be performed are as follows:
• Characterization of the spray quality defined by droplet size and distribution.
• Characterization of the spray angle, pattern and the flow/pressure relationships.
The equipment supplied for this project consists of the folio whig:
• New design swirler assemblies.
• New design extenders.
• Standard design simplex mechanical atomizers.
• Steam atomized warm-up atomizers for one elevation.
• Low NOX simplex mechanical atomizers.
• All related hoses, gauges, attachments, etc..
The scope of work includes all related services such as operating instructions, installation
instructions and support, emission testing, burner management review and recommendations and
other technical support.
Also included is the installation of portable flame cameras and video recorders. These will be used
to record the flame conditions before and after the installation of the equipment modifications.
-------
Guarantees
As would be the case for a new burner retrofit, guarantees were requested and provided. These
guarantees are similar in nature to those for entirely new burners. Guarantees have been given for
NOX, particulate, opacity, turndown and material workmanship. The guarantees offer support that
the technical requirements of the project are reasonable and have a good opportunity for success.
Research and Development of Final Design
Although EPT has installed mechanically atomized oil atomizers at other utilities, the goal of this
project is very specifically NOX reduction. The intent is to obtain the largest NOX reduction possible
from atomizers of this design. For that purpose two sets of atomizers are being supplied. One
represents a standard EPT atomizer which is used to reduce or eliminate operational problems such
as opacity or turndown. The second set of atomizers will be designed with the express purpose of
reducing the NOX to the lowest possible level. Both sets of atomizers will undergo laboratory testing
and quality assurance testing. The NOX levels achieved for both sets of atomizers will be assessed
through actual testing while installed in the Bowline Unit 1. The results of this testing will define
the amount of NOX reduction that the low NOX atomizers provide when compared to the standard
design atomizers. The testing of the low NOX atomizers will provide the best possible NOX reduction
possible with this design. The goal of the project is to meet the NOX emission limits with all burners
in service. Testing will then be performed to determine the lowest NOX emissions possible utilizing
BOOS operation.
Evaluation of EPT Burner Modification
In an effort to evaluate the merits of the burner modification offered by EPT, all available
information was reviewed. This included past EPT experience, current unit emissions data and
industry experience in BOOS operation. The technology was reviewed in a series of meetings
between Orange & Rockland, Bums & Roe and EPT.
The basis for the installation of the EPT equipment currently in operation was as a cure for
operational problems at all but one of the units. Therefore, the main goal was the elimination of these
operational problems and not a reduction in NOX emissions. In each case the goal of reducing the
operational problems was achieved to some degree. This is encouraging because reducing the
tendency for high opacity levels and other operational problems provides more flexibility of unit
operation that can be utilized for the reduction of NOX. It was also noted that a NOX reduction was
achieved at a number of the installations.
Industry data of operation with burners out of service shows that a NOX reduction in the range of
15% to 20% is reasonably expected. This reduction is dependent on the condition of the existing
equipment, baseline levels of NOX, excess oxygen levels and capacity of the existing oil guns.
-------
Conclusion
After all of the data and costs involved had been evaluated, it was determined that the EPT
modification of the existing burners combined with offstoichiometric firing operation provides the
best opportunity for the reduction of NOX levels at Bowline Point Unit 1. The modification of the
existing burners to allow low NOX operation is possible in a much shorter schedule than total
replacement of all burners. The equipment will be sized to allow full load operation with burners out
of service. This will provide additional flexibility for meeting the required NOX reduction.
In short, this approach provides the required NOX reduction, shortest implementation schedule and
most effective cost.
-------
Table 1
New York State
Section 227-2.4 Control Requirements
Effective May 31, 1995, any owner or operator of a large boiler {as defined in paragraph
227-2.2 (b) (7)} must comply with the following NOx emission limit:
NOx RACT (pounds NOx per million BTU)
Boiler Configuration
Fuel Type Tangential Wall Cyclone Stokers
Gas Only
Gas/Oil
Coal Wet Bottom
Coal Dry Bottom
0.20
0.25
1.00
0.38
0.20
0.25
1.00
0.45
na
0.43
0.60
na
na
na
na
0.30
Table No. 2
Bowline Unit 1
Baseline Test Data, Oil Firing
July 1993
Load
MW
Gross
479.9
568.0
569.1
569.3
169.7
269.8
359.4
No. of
Burners in
Service
20 (ABIS)
20
20
20
12
12
16
NOx
Emission
Lbs/MMBTU
0.2562
0.2372
0.2316
0.2290
0.2224
0.1809
0.2144
Opacity
%
9
16
20
23
10
10
9
ABIS All Burners in Service
-------
BOWLINE POINT UNIT 1
General Arrangement of T
Fired (CE) Boiler
-------
0.5
HISTORIC AT DATA FOR O/S +FGK
BOWLINE POINT UNIT 1
FULL LOAD DATA - HEAVY FUEL OIL
TANGENTIAL FIRING - 600 MEGAWATTS
PREDICTED § TEST DATA
0.
I 0.4
D
m
-0.3
LU
Q.
CO
m
CO
z
55°-2
CO
LU
x
O
z
0.1
-------
HOT AIR & FGR
TO AUX AIR
HOT AIR & FGR
TO AUX AIR
BOWLINE POINT UNIT 1
AIR AND FLUE GAS SCHEMATIC DIAGRAM
PROPOSED MODIFICATIONS TO INCORPORATE
FLUE GAS RECIRCULATION
FIGURE NO. 3
------- |