EPRI
Electric Power
Research Institute
August 1997
TR-108683-V1
EPRI-DOE-EPA Combined Utility
Air Pollutant Control Symposium
The Mega Symposium
Opening Plenary Session and NOx
EPRI
Electric Power
Research Institute
\
\
Sponsored by
Electric Power Research Institute
U.S. Department of Energy
U.S. Environmental Protection Agency
August 25-29, 1997
Washington Hilton & Towers Hotel
Washington, DC
-------
EPRI-DOE-EPA Combined Utility
Air Pollutant Control Symposium
The Mega Symposium
Opening Plenary Session and NOx
August 25-29, 1997
Washington Hilton & Towers Hotel
Washington, DC
Conference Chairpersons
George Often, EPRI
Lawrence Ruth, U.S. DOE
David Lachapelle, U.S. EPA
Sponsored by
Electric Power Research Institute
U.S. Department of Energy
U.S. Environmental Protection Agency
Prepared by
Electric Power Research Institute
-------
REPORT SUMMARY
This "Mega" Symposium combined several conferences that had been held separately
over the years to provide utilities and other interested parties with comprehensive
information on air pollution control technologies at a single time and place.
Emphasizing field experience, the conference showcased the state-of-the-art in the
measurement and reduction of NOx, SO2, and particulate /air toxic emissions.
Background
This first-ever "Mega" Symposium combines the SO2 Control Symposium, the Joint
Symposium on Stationary Combustion NOX Controls, the Particulate Control
Symposium, and the control technology portions of the EPRI/DOE International
Conference on Managing Hazardous and Particulate Pollutants. The Symposium also
includes sessions on Continuous Emissions Monitors.
Objective
To provide information on the latest developments and operational experience with
state-of-the-art methods for measuring and reducing NOx, SC>2, and particulate/air
toxics emissions from fossil-fueled boilers.
Approach
EPRI, the U.S. Department of Energy, and the U.S. Environmental Protection Agency
cosponsored a "Mega" Symposium in Washington, DC on August 25-29,1997. Over 120
papers were presented with sessions grouped by pollutant, topical area, boiler type,
and/or process.
Key Points
The Symposium proceedings are published in three volumes: Volume I, NOx controls;
Volume II, SO2 Controls and Continuous Emissions Monitors; and Volume III,
Particulates and Air Toxics Controls. Topics covered during formal presentations and
poster sessions include:
• Combustion tuning/optimization
• Low NOx Systems for Coal-, Gas-, and Oil-Fired Boilers
-------
• Selective Catalytic Reduction
• Selective Noncatalytic Reduction
• Cyclones - Combustion NOx Controls
• Full-Scale Flue Gas Desulfurization (FGD) Experience
• FGD Conversions
• FGD Process Improvements
• Dry SO2 Control Processes
• Advanced SO2 Control Processes
• Continuous Emission Monitors
• New Technologies for Particulate Control
• Lab- and Pilot-Scale Research in Mercury Capture by Sorbents
• Mercury Capture by FGD
- High Gas-to-Cloth Ratio Baghouses
• Engineering Studies in Particulate Control
• Postcombustion NOx/SC>2 Reduction
TR-108683-V1-V3
Interest Categories
Air emissions control
Air toxics measurement and control
Emissions monitoring
Fossil steam plant performance optimization
Key Words
Nitrogen oxides Air toxics control
Flue gas desulfurization Particulates
SC>2 control Continuous emission monitoring
-------
CONTENTS
Monday, August 25; 8:00 a.m.
OPENING PLENARY SESSION
• Apply Principles of Industrial Ecology to Manage Emissions
• S02 and NOx Compliance of New England Power CD's Coal Fired Units
• Planning for NOx Emission Regulations via EPRI'S CAT Workstation
• The Cost of Complying with NOx Emission Regulations for Existing Coal Fueled
Boilers
Monday, August 25; 10:00 a.m.; 1:00 p.m.
PLENARY SESSION: Combustion Tuning/Optimization (PC Units)
• The Role of Combustion Diagnostics in Boiler Tuning
• Post Low NOx Burner Retrofit Boiler Tuning Results for a Front Wall-Fired
Boiler
• Fuel System Modifications and Boiler Tuning to Achieve
Early Election NOx Compliance on a 372-MWe Coal-Fired Tangential Boiler
• Experience with Combustion Tuning and Fuel System Modifications to
Inexpensively Reduce NOx Emissions from Eleven Coal-Fired
Tangential Boilers
• Application of an Expert System and Neural Networks For
Optimizing Combustion
• The Emissions, Operational, and Performance Issues of Neural Network
Control Applications for Coal-Fired Electric Utility Boilers
• Emission Solutions Through Optimization
• GNOCIS: A Performance Update on the Generic NOx Control Intelligent System
-------
Monday, August 25; 3:30 p.m.
PARALLEL SESSION A: Low-NOx Systems for Coal-Fired Boilers
(Wall and Tangential)
• Burner Modifications for Cost Effective NOx Control
• NOx Subsystem Evaluation of B&W's Advanced Coal-Fired Low Emission
Boiler System at 100 Million BTU/HR
• Field Demonstration of ABB C-E Services' RSFC™ Wall Burner for Coal
Retrofit Applications
Monday, August 25; 3:30 p.m.
PARALLEL SESSION B: Low-NOx Systems for Coal-Fired Boilers-Group 2 Units
• NOx Reduction without Low NOx Burners for a Riley Dry Bottom Turbo Furnace
• NOx Reduction on a Riley Stoker Dry Bottom Turbo Furnace
• NOx Reduction in Arch-Fired Boilers by Parametric Tuning of Operating
Conditions
Monday, August 25; 3:30 p.m.
PARALLEL SESSION C: Low NOx Systems for Gas/Oil-Fired Boilers
• Development, Test and Industrial Application of Advanced Low NOx Burners
• Ultra Low NOx Operation from a 185 MW Oil and Gas "T" Fired Boiler
• Preliminary Results of Low NOx (< 25 ppm) Burner Retrofit of Pacific Gas &
Electric 345-MW Contra Costa Unit 7
• Ultra-Low NOx Rapid Mix Burner Demonstration at CON Edison's 59th Street
Station
-------
Tuesday, August 26; 8:00 a.m.
PARALLEL SESSION A: Low-NOx Systems for Coal-Fired Boilers
(Wall and Tangential) Continued
• The Integration of Low NOx Control Technologies at the Southern Energy, Inc.
Birchwood Power Facility
• Impact of Coal Quality and Coal Blending on NOx Emissions for Two
Pulverized Coal Fired Units
• Reduced NOx Emissions from Certain Coal Blends for Utility Boilers
• Effect of Low NOx Firing Conditions on Increased Carbon in Ash and Water
Wall Corrosion Rates
• Assess Coal Quality Impacts on NOx and LOI with EPRI's NOx-LOl Predictor
• Field Experience—Reburn NOx Control
• Commercial Demonstration of Methane de-NOx® Reburn Technology on a
Coal-Fired Stoker Boiler
Tuesday, August 26; 8:00 a.m.
PARALLEL SESSION B: Selective Noncatalytic Reduction
• Using Retractable Lances to Maximize SNCR Performance
• SNCR Retrofit Experience on Four Gas and Coal-Fired Boilers in
Tchaikovsky, Russia
• Design and Characterization of a Urea-Based SNCR System for a Utility Boiler
• Enhanced NOxOUT® Control Salem Harbor Unit #3
• Derivation and Application of a Global SNCR Model in Maximizing NOx
Reduction
• In Field Results of SNCR/SCR Hybrid on a Group 1 Boiler in the Ozone
Transport Region
• Stationary Source NOx Control Using Pulse-Corona Induced Plasma
Tuesday, August 26; 8:00 a.m.
PARALLEL SESSION C: Low NOx Systems for Gas/Oil-Fired Boilers
Continued
• Applications of REACH Technology to Reduce NOx and Particulate Matter
Emissions at Oil-Fired Boilers
• Reducing NOx Emissions in a Natural Gas-Fired Utility Boiler Using
Computational Fluid Dynamics
-------
Tuesday, August 26; 1:00 p.m.
PARALLEL SESSION A: Cyclones - Combustion NOx Controls
• Computer Modeling of Cyclone Barrels
• Short-Term NOx Emission Reductions with Combustion Modifications on Low
to Medium Sulfur Coal-Fired Cyclone Boilers
• Combustion Tampering Demonstration on a Cyclone Unit for NOx Control
• Reduction of NOx Emissions with Cyclone Burners by Biasing of
Combustion Air
• Cyclone Boiler Air Staging Demonstration Project Sioux Unit 2
• NOx Control Using Natural Gas Reburn on an Industrial Cyclone Boiler
• Application of Fuel Lean Gas Reburn Technology at Commonwealth Edison's
Joliet Generating Station 9
Tuesday, August 26; 1:00 p.m.
PARALLEL SESSION B: Selective Catalytic Reduction
• Applications of Selective Catalytic Reduction Technology on Coal-Fired
Electric Utility Boilers
• SCR Applications: Addressing Coal Characteristic Concerns
• Selective Catalytic Reduction (SCR) Retrofit at San Diego Gas & Electric
Company South Bay Generating Station
• Engineering and Pilot Scale Assessments of a Low Cost Combined Low-NOx
Burner-SCR System
• Feasibility of Applying Selective Catalytic Reduction (SCR) to Oil-Fired, Simple
and Combined-Cycle Combustion Turbines
• Economic Analysis of Selective Catalytic Reduction Applied to Coal-Fired
Boilers
• SCR for Coal-Fired Boilers: A Survey of Recent Utility Cost Estimates
• SCR for a 460 MW Coal Fueled Unit: Stanton Unit 2 Design, Startup and
Operation
• Selective Catalytic Reduction: Successful Commercial Performance on Two
U.S. Coal-Fired Boilers
• Successful Implementation of Cormetech Catalyst in High Sulfur Coal-Fired
SCR Demonstration Project
-------
Monday, August 25; 8:00 a.m.
Opening Plenary Session
-------
APPLY PRINCIPLES OF INDUSTRIAL
ECOLOGY TO MANAGE EMISSIONS
Jason Makansi, Editor in Chief
Power magazine/The McGraw Hill Companies
11 W 19th St, 2nd Fir
New York, NY 10011-4285
Abstract
Considering a powerplant as more than a fuel-to-electricity station can raise
its economic value to its owner and the community at large and solve
emissions and discharge problems more holistically. The emerging field of
industrial ecology offers a set of organizing principles to do this. This paper
reviews the basis for industrial ecology and specifically its relevance to
solving problems in the removal of particulates, SO2, and NOx from
powerplant flue gas. Highlights from one successful example of industrial
ecology worldwide is provided, along with references to regulatory and
technical developments.
Introduction
Chemists talk of atomic "bonds" between and among elements. Social
scientists speak of social bonds among people, families, communities, and
political entities. The financial and business community use bonds as a
binding agreement for payment of taxes or money owed. Ecologists talk of
bonds in nature, such as how the food chain bonds different species together.
Deregulation and competition is causing the traditional bonds within the
power industry to break down. The reactions which are resulting are
traumatic for some but opportunistic for many. With both directed
(regulated) and non-directed (so-called free market) components, the industry
is also at the same time being re-arranged.
Industrial ecology offers all professionals involved with emissions control,
and non-professionals concerned about the environment, a new set of
organizing principles to manage emissions control with competitive
operation. Importantly, this set of principles, in my estimation, ultimately
will help interest groups focused on environmental control—including
-------
regulators, technology and equipment suppliers, scientists and the R&D
community, economists, and the environmentalist advocacy groups—to
approach problems within a framework that all can understand.
Most importantly, is that the concept of industrial ecology is fully compatible
with the competitive direction of the industry, although it may not appear so.
Some even broaden the phrase to the "business ecosystem," to describe how
companies compete and cooperate—much like organisms and species do—on
many different levels to achieve profitability. James Moore, in The Death of
Competition: Leadership Strategies in the Age of Business Ecosystems,
advocates such an view of the business world^.
What is industrial ecology?
In its simplest form, industrial ecology, which some might construe as a
contradiction in terms, is a set of design and operating principles patterned
after what goes on in nature, where connections among organisms see to it
that raw materials and waste materials are used efficiently. Similarly,
industrial processes should not be considered in a vacuum but relative to
other processes that can feed and support each other, especially in terms of
recycling materials that otherwise "pollute" the surroundings in some way.
As a field of study, industrial ecology is emerging, evidenced by the new
Journal of Industrial Ecology, which began publication this year^. The
publication is a joint effort of Massachusetts Institute of Technology and Yale
University. Many magazine articles have appeared recently, the nation's
laboratories and academic institutions are initiating industrial ecology
programs, and industry appears to be taking note.
But activities embodied by industrial ecology didn't just appear. Industrial
ecology is found in other "paradigms," such as sustainable development, life-
cycle economics, the "industrial park," holistics or holism, integrated
environmental management, industrial symbiosis, and others. Many
powerplant design and engineering concepts that are commercially applied
today are components of industrial ecology. The phrase simply provides a
more convenient, or perhaps more complete, means of not only pursuing
ideas that balance economic progress and environmental impact, but to
communicate these efforts as well.
The phrase has particular meaning with respect to the debate between
"environmentalists" and industry. It pairs a form of the word industry, the
"enemy" of environmentalists, with ecology, perceived by environmentalists
to be the "victim" of industry. Pairing the two words connotes a greater sense
of interaction, even harmony. For a long time, this industry has needed
something like industrial ecology to properly describe its considerable efforts
-------
in protecting the environment and recycle waste materials while still
delivering the electricity everyone demands.
In a more subtle way, the phrase industrial ecology implies, as it should, that
everything, whether "industrial" or "natural" begins and ends in the earth. It
does away with the false notion that things "industrial" are somehow less
than things natural or "ecological."
Just as so-called natural ecological systems are constantly adapting to changing
conditions—higher than normal rainfall, disruptions to the size and number
of species, nutrient levels, sunlight, etc—industrial ecology is a way to
understand that industrial systems constantly adapt, too. And, perhaps most
importantly of all, our surroundings have a keen ability to adapt to the
changes wrought by industrial activity. Industrial ecology acknowledges the
existence of this ability to adapt, though not the magnitude of this capability.
Analogies to other areas. Another way to understand the concept of
industrial ecology is to relate it to the concept of quality. American
manufacturing has been pursuing vigorously for at least a decade the ISO
(international standards organization) 9000 certification process for product
quality. This is also a set of organizing principles, a certain set of procedures a
company or a site follows to ensure the quality of its products for its
customers worldwide. ISO9000 doesn't specify that the product's quality
actually will meet some specific value. Rather, it defines work and
management processes, inspections and oversights, and quality control
programs that certainly raise the chances that product quality specifications
will be met. And it forces all firms to achieve a minimum level of quality
performance—table stakes, if you will, to do business in the worldwide
market.
In the same way, industrial ecology doesn't necessarily define the SO2, NOx,
or flyash emissions standards that a site must meet or the technologies used
to meet them. Rather, it can provide a framework for achieving a more
practical site-specific, local, and/or regional balance among them.
The need for a new set of organizing principles is acute for other reasons.
One, germane to the goals of flyash, NOx, and SO2 reduction, is that, as lower
emissions levels are sought for specific pollutants, the impact on the others
may grow. One perfect example of this is the application of low-NOx burners
to NOx reduction. The impact of low-NOx burners on carbon levels in flyash
is by now well-known. But the economic and ecological impacts from huge
volumes of flyash unfit for recycle is only now being discovered. Another
example is the huge volumes of scrubber byproducts that had to be managed
as a result of tight restrictions on SO2 levels.
-------
-r^p recent regv.l.itery err.phasis en so-called air toxics alone signifies that
environmental regulation and management must now govern smaller and
smaller volumes of potential pollutants. As people become more informed
and aware and society more advanced, less pollution of any form is tolerated.
But a society can also advance in terms of understanding of the relative levels
wi" different pollutants, their true, measurable impact on the environment,
and so on.
History has shown that prescriptive, command and control, single-pollutant
regulatory strategies and subsequent compliance efforts, have often simply
transferred the problem elsewhere or created new environmental
management problems. Often these residual problems were anticipated and
solutions readied, but institutional barriers prevent these solutions from
being implemented. In addition, the one-pollutant regulations are written
such that competition among technologies is not fostered.
The fact that limestone-based flue-gas desulfurization (FGD) captures 90% of
the domestic market—and appears to be capturing a similar share of the
world market—suggests that competition among technologies is lacking, and
incentives to apply potentially better technologies is non-existent. That
switching to low-sulfur coal, which has significant impacts on other
emissions and waste streams, as well as on powerplant efficiency, proved to
be the compliance method of choice for phase I of the Clean Air Act is
another indicator that more holistic approaches to emissions control should
be encouraged.
If low environmental impact, low-cost electricity, and a generation sector that
can attract private investment on a profit basis (not a regulated rate of return
basis) are going to be simultaneously attained, new regulatory regimes are
dearly necessary.
Another good reason for applying industrial ecology takes a business slant. I
believe that the nation's largest coal-fired power stations will become more
than electricity generators in a competitive marketplace. These facilities are
economic anchors in their communities. Because they encompass vast tracts
of land, have rail and other transportation access, and generally have the
necessary industrial infrastructure, they will become centerpieces for new
industrial concepts—perhaps industrial parks—that may best be governed by
industrial ecology-based regulation.
Whatever regulations are passed at the federal level, states and regions will
have to protect indigenous industries and their tax base. So, as one example, if
states in New England do not want to build new power stations, then the
Midwest with its own coal and access to western coal may become an
electricity producing center lor the nation. A fitting analogy here is the
natural gas industry in based in Texas. Texas' gas is transported all over the
-------
country. Midwest and other low-cost sources of coal-based electricity may
experience a similar fate. And states and regions will encourage regulations
:hat allow economic development to coalesce around these facilities. The
principles of industrial ecology could help govern the new bonds that are
likely to form between states dependent on coal for their economies, coal
companies, and electricity generators.
Examples of Industrial Ecology
Industrial ecology is a broad framework. But specific acts of industrial ecology
have been undertaken in the power industry for decades.
Flyash. Coal-fired plants have been recycling flyash into cement and concrete
markets for decades. That's an example of industrial ecology—a "waste"
material from one industrial process is recycled into another. For the most
part, though, this has been a secondary, perhaps tertiary consideration and
priority for most of these plants. When the principles of industrial ecology are
applied, equal emphasis will likely be placed on this recycle stream.
When you consider that plants are searching for new revenue streams in a
competitive business arena, you can quickly understand that the incidental
production of flyash at a coal-fired station becomes the deliberate production
of a raw material for the cement industry which requires that material within
certain quality tolerances, seasonal volume ranges, and so on.
Now take this a step further and think about the "bonds" mentioned earlier
in this paper. Perhaps the bond between generation and transmission
weakens as plants operate as independent profit or business centers. But
perhaps the bond between a large coal-fired plant owner/operator and a
cement company grows. Or the bond between the "new" utility industry and
the cement/concrete industry as a whole. In fact, I would argue that the bonds
among cement producer, coal producer, and power generator are more logical
than that between generator and transmitter. As one symbiotic idea, coal
companies can backhaul flyash to cement producers which may extend the
economic radius of flyash recycle.
For those involved in the design, selection, operation, and maintenance of
electrostatic precipitators, fabric filters, and other devices for collecting flyash,
industrial ecology suggests that you think of your product in terms of (1)
helping your "customer" produce the highest-quality flyash possible and (2)
providing a third-party service for your customer that solves both the
collection and recycle problems. A few emissions control process suppliers are
already using this "third-party" model. Although unsuccessful to date, the
probabilities of success should be greater when the proper motivation and
incentives are in place.
-------
Water. Perhaps the most comprehensive application of industrial ecology
principles at powerplants today involves water. New powerplants being built
today source their water not from clean lakes, streams, and rivers, but from
nearby industrial facilities or the municipal waste water treatment plant.
Some plants report that they take water from polluted streams and outfalls
and in the process of using it and cleaning it up to meet discharge
requirements, return it far cleaner that it was originally. Granted, some of it
may be returned as through a cooling tower vapor, but it is still returned to
the water cycle.
Other examples. Examples abound in the industry of the partial application
of industrial ecology:
•Recovery of powerplant flue gas CO2 for food-grade applications. At least
three powerplants in the US recover CO2 from a small percentage of flue gas
so treated.
•Use of a byproduct of nylon manufacture to enhance the absorption
efficiency in SO2 removal systems.
•Ammonia-based FGD systems which result in valuable fertilizer materials.
This FGD process is now poised to compete with limes tone /lime-based
processes and several suppliers are eyeing the worldwide market.
•Use of FGD gypsum in wallboard manufacturing and/or stabilized FGD
waste in road-building and other civil construction projects.
•Applying an integrated coal gasification combined cycle (IGCC) process and
recycling the slag and sulfur and perhaps even producing methanol or other
chemicals on the front end (what has been called the 'coal refinery')-
•Raising energy crops (fast-growing trees) that can serve as biomass fuel to
sequester carbon and regulate CO2 buildup.
•Combining front-end recycling operations—ferrous, glass, and aluminum.
recovery—with a municipal-refuse waste-to-energy/electricity plant.
Each of these examples of industrial ecology at work has obvious business
analogies that may help powerplants find new revenue streams and bond
with new business partners in the competitive environment.
New regulatory structures needed
Industrial ecology suggests that regulators consider a more holistic approach
to compliance with emissions limits. Is simply stipulating a limit at the
powerplant good enough? How about some form of credit to those who
properly recycle flyash into new products? Perhaps that credit could consist of
a break on the limit specified in the permit for that plant. In other worlds,
encourage a plant to recycle its flyash by being more lenient on the limit that
plant has to meet, or by being lenient on a separate pollutant limit. Another
alternative is to grant a credit of some form to those who use wastes
productively.
-------
<">e reason why so-called "integrated" or multi-pollutant control initiatives
and technologies languish is because regulations are bulk up around single
^ullutants in specific industries. And, not to be neglected, bureaucracies
within regulatory agencies, and product lines within environmental control
companies, are constructed in the same way. Returning again to the concept
of "bonds," it is clear that industrial ecology may help us break some bonds
holding together antiquated ways of regulating industry and developing
products and technologies to meet those regulations.
In recent years, EPA has attempted to pilot regulations based on a site-wide
permit for several pollutants. It was tried in the pulp and paper industry and
has been discussed for the power industry. These initiatives have often been
viewed with suspicion by industry as guises for greater regulation. But they
should be applied—and hopefully accepted—as prescriptions for more
balanced regulation, not more regulation.
Combine with externalities. Today, less discussion seems to emanate from
the industry about the so-called environmental externality—the idea of
placing some monetary value on the damage or environmental impact of
pollution and discharges. The economics for virtually any industrial process
change if externalities are factored in. How that damage is assessed and
valued is a separate complex subject that cannot be addressed here. But clearly
the externality concept plays well with industrial ecology. Externalities would,
for example, help government justify and reestablish taxes and subsidies to
pursue environmental goals.
Externalities, combined with risk assessments, also help us understand the
tradeoffs among different environmental control objectives. Rather than
simply pursue the often destructive and unrealistic goal of "lower is better,"
regulators could in fact support new concepts that, for example, allow one
pollutant to be discharged at a higher level in one location, in exchange for a
lower limit in a region more severely impacted by that particular pollutant.
The idea, after all, is to protect the environment and public health. The goal
of environmental regulation is not to see how low an industry can go with
reducing emissions, or to add burdensome costs onto industry at a time when
global competitiveness can be precarious and fleeting. I think most of us, at
the end of the day, would agree that the protection of public health from a 9-
ppm NOx limit on a gas turbine is vague at best, and probably not worth the
capital costs and efficiency penalties at worst. This is perhaps the most
egregious example of a regulatory limit promulgated solely on the basis that
lower is better.
But without a truly integrated regulatory framework, the principles of
industrial ecology will, in the worst case, never work, and in the best case, not
live up to their true potential.
-------
Fortunately, agencies are beginning to respond. Although details are not
available, at least one report mentions the Interagency Environmental
Technologies Office of the federal government and EPA's Design for the
Environment Program 1 0, both of which seem to be pursuing regulatory
objectives that could support industrial ecology. Also within the Clinton
administration, the President's Council on Sustainable Development, formed
in 1993, issued a report that addresses the concepts of industrial ecology and
eco-industrial parks11.The council now has a task force on eco-industrial
parks.
EPA also has an initiative called Project XL which seeks to grant regulatory
flexibility in exchange for an enforceable commitment by a regulated entity to
achieve better environmental results than would have been attained through
full compliance with current regulations. Project XL is part of the president's
Reinventing Environmental Regulation initiative. Finally, EPA has
launched the Clean Air Power Initiative (CAPI) which specifically addresses
an approach to rationalize and streamline emissions reductions strategies at
power generation facilities. Utilities, in commenting on CAPI, are seeking a
"final" rule or at least some certainty over a specified time frame that
regulations for NOx and SO2 will not change. Although these initiatives may
or may not succeed, or may be viewed by industry as unprogressive, they do
indicate that antiquated regulatory philosophies may be changing.
Putting it all together
An underlying feature of industrial ecology is "community." Apart from
other challenges in using waste materials, the distance that they have to be
transported has always been an economic barrier. Plus, long distance transport
consumes more energy, results in more pollution, etc, so, in economic terms,
the sense of community becomes identical to the practical economic need for
closeness.
Suppose you combine a mine-mouth coal-fired powerplant equipped with an
FGD system that produces gypsum with a wallboard manufacturing plant, a
cement production plant, and plan for mine reclamation, acid-mine drainage,
and so on using ash from the coal. Suppose you also supply other industrial
and commercial facilities with steam and/or hot water, chilled water, potable
water, and other services. If this industrial park anchored by the coal-fired
plant supplies building products to the surrounding community, then
transportation of materials is avoided as are potentially higher costs for
"imported" raw materials or goods and services. Now you have a community
supported by an industrial ecological system, along with the usual social and
political support systems.
-------
To some, this sounds like industrial Utopia, not ecology. To others, it ir.ay be
reminiscent of a page from a communist manifesto. But there is at least one
real-world situation that approximates the description. That is at an eco-
industrial park in Kalundborg, Denmark, which has been written about
extensively in the literature^' 2.
At what is perhaps the rallying point for industrial-ecology enthusiasts
worldwide, a 1500-MW coal-fired district heating plant, the Asnaes station,
anchors a web of industrial and social concerns. The Asnaes plant supplies
steam to heat almost all of the surrounding town of Kalundborg's houses and
commercial buildings, 40% of the heat needed by a nearby refinery's tank
farm, and all the process heat required by a pharmaceutical manufacturing
complex. Gypsum from the plant's FGD system meets two-thirds of the input
needed by the local wallboard-maker. Flyash and clinker from Asnaes are sold
for use in road-building and cement-making. Process gas once flared at the
refinery is burned in the Asnaes boilers, displacing 30,000 tons of coal a year.
The refinery sends the powerplant its cooling water, which is treated and used
and boiler feedwater. Waste water from the refinery is also used to clean
equipment at Asnaes. Other materials are recycled among various industrial
facilities in the town.
Importantly, Kalundborg was not hatched as a grand, centrally planned
experiment in industrial ecology. Rather, it evolved piecemeal over 25 years
and arose from (1) the relative isolation of the town, (2) the social interaction
among the employees and managers at the various facilities, (3) increasing
pressure to sustain profits under greater environmental regulation, and (4) an
atmosphere that apparently fosters constructive negotiation about
regulations, not special interests lobbying their Congressional representatives
at all hours. Danish regulatory processes seem to encourage the parties to
focus on creative solutions rather than fight among each other.
Because of its evolutionary development and implementation, the
experience at Kalundborg may not easily transfer to greenfield developments,
according to some who have studied the situation^. And, of course, Denmark
is relatively small, homogenous country and hardly resembles the US
situation. Also of strategic interest, the manager of Asnaes has reported that
existing economic incentives were sufficient to motivate most of the
relationships between Asnaes and other industrial firms. In other words, no
deliberate institutional mechanism was necessary to promote the exchanges.
Symbiotic initiatives. The United Nations University, Tokyo, Japan, has
launched the zero emissions research initiative (ZERI), funded by the
Ministry of International Trade and Industry. According to one source^, it's
objective is to achieve technological breakthroughs that facilitate
manufacturing without any form of waste. To date, the efforts do not appear
-------
to be forn^ed on t>owerplrmts. but on process industries. Tn. addition.. O?.k
Ridge National Laboratories (ORNL), Oak Ridge (TN) is working with the
City of Chattanooga (TN) to set up a zero emissions industrial park but again
activities specific to electric production have not been reported to date.
Several years ago, at least one researcher under contract to ORNL, Dale
Merrick of Parsons Engineering Science, Oak Ridge (TN), had proposed the
R4C process to handle all types of municipal wastes, which includes a waste-
to-electricity component. R4C certainly has industrial ecology concepts
supporting it even if the phrase is not used within the description of the
process^. At the Research Triangle Institute, Research Triangle Park (NC),
researchers are involved producing a "field book" for developing an eco-
industrial park in the US. EPRI has also recently announced initiatives to
help its clients reshape generation markets with energy partnerships focused
on concepts that certainly could fall under the industrial ecology umbrella.
Lawrence Livermore Laboratories also has industrial ecology programs going
US Examples. Some of the examples of emerging industrial ecology in the
US relevant to electricity production include the Great Plains Gasification
Complex^ and two of the integrated gasification combined cycle (IGCC)
demonstrations taking place under the DOE Clean Coal Technology
Demonstration Program — the IGCC project at Tampa Electric Co's Polk power
station^ and at Cinergy's Wabash River station'7 One plant is a greenfield site,
one is a repowering of an older fossil unit. Although not a power station, but
with directly transferrable experience, Eastman Chemical Co, Kingsport (TN)
is producing industrial chemicals from a coal gasifier unit, and recently
announced successful operation of an advanced process there.
Finally, it deserves mention that the National Academy of Engineering's
National Research Council (NRC), Board on Energy and Environmental
Systems is planning a two-day project planning meeting to define issues and
develop the scope of a prospective NRC study or activity related to eco-
indus trial coal-fired powerplants. The National Academy also has published
valuable reports on industrial ecology^, 13
International focus
An unfortunate paradox of industrial ecology is that the huge developing
economies of India, Brazil, Indonesia, and China, facing enormous present or
neat-term industrial growth can most benefit from industrial ecology because
everything is growing — industry, infrastructure, population — and different
processes to satisfy that growth could feed off each other more sustainably. At
the same time, these countries are growing so fast that planners are unable to
contain that growth within the boundaries of a more integrated approach.
-------
Meanwhile growth has slowed in North America, Europe, and Japan, where
the principles of industrial ecology are being aired.
Although some countries are striving for more integrated regulatory
frameworks, most initiatives are young and require coordinated support from
environmentalists and industrial concerns alike.
Challenges abound
Challenges are many in applying industrial ecology, although it is not my
intent to focus on them here, but a few deserve mention. One is that a waste
material often replaces a naturally occurring one. These natural materials
often have extremely low value because of technical efficiencies built up over
decades in a particular industry (gypsum is always cited as an example here),
low labor costs, and other reasons. Existing industries must protect their own
profits and turf. Economically, it may be impossible for recycled material to
compete. Also, by their nature, "wastes" are more difficult to control in terms
of quality and quantity. Such issues as product quality and inventory
management have dogged supporters of industrial recycling for years.
Alternate disposal methods are still necessary for when markets go sour. The
plant, after all, has to continue making its main product, electricity.
Fluctuations in the prices and supply of recycled materials often make
investors skittish. These effects have been amply demonstrated in recent
years in the municipal-waste recycling markets—newspapers, plastics,
ferrous, aluminum, and glass. Others have noted that industrial wastes are
often poorly characterized, although recently several waste exchanges have
sprung up to address this challenge.
Summary
Industrial ecology is a set of organizing principles, a new way of thinking, a
common frame of reference for parties to the debate to gather around. It is not
a solution to any emissions control problem. It is not, to use a well-worn
word, a panacea. Nor should it be construed as a mandate for all-embracing
centrally planned regulatory initiative. The principles of industrial ecology
can, however, be used to help reconstruct environmental regulation in a
more holistic fashion, develop new revenue streams and business units, and
protect the environment and public health without sacrificing competitive
electricity production economics.
References
1. "A Down-to-Earth Approach to Clean Production." Technology Review.
February/March 1996, p. 49
-------
2. "Biorealism: Reading Nature's Blueprints." Audubon. September/October
1995. p. 70
3. John Ehrenfeld and Nicholas Gertler. "Industrial Ecology in Practice."
Journal of Industrial Ecology. Vol. I, No. 1 p. 67. 1997
4. "Zero Emissions Initiative Yields New Technology." Chemical
Engineering. May 1995. p. 44
5. "Lignite-to-Gas Plant Reveals Numerous Innovations." Power. May/June
1997. p. 80
6. "From IGCC Emerges a Holistic Approach to Coal-Based Plant." Power.
May/June 1997. p. 41
7. Wabash proves out next-generation gasifier, next-century powerplant.''
Power. September/October 1996. p. 34
8. Private communication
9. "The New Biology of Big Business." Business Week, April 15, 1996. p. 19
10. "Engineering Sustainable Development." Mechanical Engineering. May
1996. p. 48
11. "Building on Consensus: A Progress Report on Sustainable America."
President's Council on Sustainable Development. Washington DC.
12. D J Richards and A B Fullerton (eds). Industrial Ecology: US-Japan
Perspectives. National Academy Press. Washington DC. 1994
13. D J Richards and R A Frosch (eds). Corporate Environmental Practices:
Climbing the Learning Curve. National Academy Press, Washington DC. 1994
14. "Primer on Industrial Ecology." Science & Technology Review (a
publication of Lawrence Livermore Laboratories). March 1996. p. 4
-------
SO2 AND NOX COMPLIANCE OF
NEW ENGLAND POWER CO'S COAL FIRED UNITS
Herbert C. Stowe, P.E.
Manager of Power Engineering
25 Research Drive
New England Power Service Company
Westborough, MA 01582
Abstract
New England Power Company (NEPCo) brought six coal units at two fossil generating stations into SO2
and NOX compliance in 1995 and has continued to maintain compliance and improve unit performance.
This paper focuses on the coal fired units and outlines regulations which required SO2 and NOX emission
reductions, discusses the planning, strategic decisions and the multi-functional team approach, and
reviews the operational results.
Coal fired units, environmental enhancements included Low NOX Burners (LNB) with and without
Overfire Air, Gas Co-Firing, Selective Non-Catalytic Reduction (SNCR) Controls, Flue Gas
Conditioning Systems (both Conventional and EPRICON), Precipitator Improvements and Pulverizer
Replacements.
Introduction
New England Power Go's Fossil Generating Stations have had the unique experience of requiring air
quality environmental compliance in SO2 and NOX prior to most of the U.S. and most foreign countries.
Therefore, New England Power Company (NEPCo), the wholesale electric generating company of New
England Electric System (NEES) and New England Power Service Company (NEPSCo), a service
company that provides engineering, construction and environmental services to the affiliated NEES
subsidiaries, have gained considerable experience in strategic planning, technology evaluation and
selection and project implementation, and are continuing to gain system operational experience in
pollution control systems.
This compliance program involved both Brayton Point and Salem Harbor Stations. Each station has
three coal fired units. The unit characteristics for these coal fired units are provided in Table 1.
Regulations
The regulations which resulted in these requirements included the Massachusetts Acid Rain Law and
U.S. Federal Clean Air Act Amendments of 1990.
-------
• Sulfur Dioxide (S02). The Massachusetts Acid Rain Law required environmental compliance with
sulfur dioxide (SO,) regulations starting in 1995. The law required that Brayton Point and Salem
Harbor Station (Massachusetts) units have a compound annual average SO2 emission rate of 1.2
Lbs/MMBTU or less. In effect, the Massachusetts regulations requires early compliance with year
2000 federal acid rain requirements.
• Nitrogen Oxides (NOX). Title 1 of the Clean Air Act Amendments (CAAA) of 1990 required that
areas out of compliance with National Ambient Air Quality Standards (such as ground level ozone)
develop a strategy to obtain compliance. In general, much of the eastern United States and a large
portion of New England is in non-compliance for ground level ozone, a primary cause of summer
smog. The two precursors for ground level ozone are Volatile Organic Compound (VOC) and
Nitrogen Oxides (NOx) emissions. In Massachusetts, utilities contributed less than 1% of the VOC
emissions but about 30% of the NOX emissions. The Federal and State Regulatory Agencies
determined that significant power generation sector NOX reductions would be required to obtain
ozone compliance with the National Ambient Air Quality Standards. One of the steps in the
attainment process was that all fossil fired generating units in Massachusetts should meet
Reasonably Available Control Technologies (RACT) standards forNOx emissions by May 31, 1995.
Compliance Approach
The overall compliance effort occured in two phases. Planning for compliance with the Mass Acid Rain
Regulations began first. The overall SO2 Compliance Strategy was developed in the late 80's and the
projects were installed in the early 90's. NOX Reduction studies and evaluations were also ongoing
during this time frame. In 1992 the Massachusetts RACT Regulations were starting to take shape and
NEES began the focused effort of attaining NOX Compliance by May 31, 1995.
These two compliance efforts were conducted in a similar manner, using the same overall philosophy
directed by in house program teams. Both compliance programs involved four major areas which are
outlined as follows:
• Planning - At the onset NEES' management made it clear that 100% compliance was the corporate
objective. Multimctional corporate teams were established to evaluate the alternatives, develop
comprehensive plans, evaluate plans on base assumptions and possible scenarious and finally setting
strategic direction.
• Approvals - The multi-function team set the strategic direction and the program team developed the
detail implementation plans. NEES management approved the overall plan on a conceptual basis
allowing the program team the flexibility to adapt to continuously changing challenges. Regulatory
approval was enhanced by reviewing the overall program approach up front with our regulators.
• Project Implementation - Multi-department project program teams were established and met on a bi-
monthly basis at the stations to coordinate the many ongoing projects. The program team had a
program manager, a management advisory group and representatives from all the major functions.
Each project was assigned to a project engineer/manager who reported on the project progress at the
coordination meetings. Having all departments represented allowed issues to be addressed in a
-------
- 3
comprehensive fashion. The meetings focused on major issues and assigned responsiblity to resolve
details.
• Operations - Given that the overall objective of the program was successful compliance through
project operation, therefore, there was considerable emphasis on testing, training and operation. For
example, compliance coal testing started more than three years before S02 compliance was required.
SO2 Compliance
The Massachusetts Acid Rain Law required that the SO2 emissions from Massachusetts fossil fired units
not exceed 1.2 Lbs per million BTU's starting January 1,1995 (see Table 2). Strategic planning began
at NEES in the late 80's leading to the most effective strategies to comply. Numerous scenarios were
evaluated using various sensitivity factors to ensure a robust solution. The strategic decision was made
to comply with Massachusetts Acid Rain Law through the use of compliance fuels. Plant improvements
were required to insure success burning of the compliance coals. These improvements included
pulverizer replacements, flue gas conditioning and precipitator enhancements.
• Pulverizer Replacements. The SO2 compliance strategy recommended replacing the Brayton
Point Units 1&2 pulverizers to allow burning lower cost, harder, compliance (low sulfur) coals.
Brayton Point Unit 3 was already equipped with B&W MPS Mills (harder coal capable) which
allowed Brayton Point Station to convert to the more agressive compliance coal specification. This
decision was based on the Fuels Department analysis of extensive sources of the world wide coal
market which determined that significant fuel savings could result due to domestic and international
low sulfur coals. The Pulverizer Replacement Projects were completed in 1992. The existing CE
1960 vintage Bowl Mills were replaced with ABB/CE HPS Exhauster Mills. The pulverizer
replacements opened up new fuel markets for Brayton Point Station which began testing compliance
coals in 1992 and is currently burning a wide variety of compliance coals.
• Fuel Characteristics. Both Brayton Point and Salem Harbor receive coal by ocean going ships.
The Fuel Department cost effectively purchases a wide variety of compliance coals. The sources of
coal are from both domestic (West Virginia, Kentucky and Virginia) and international (Columbia
and Venezuela) coal sources. These coals have a wide variety of properties. Table 3 provides the
range of coal properties burned at these stations.
• Precipitator Enhancements & Flue Gas Conditioning. The low sulfur coals resulted in
anticipated changes in fuel characteristics (resistivity) which can result in less efficient precipitator
collection efficiency. To insure precipitator performance, the Brayton Point coal unit precipitators
control systems were enhanced, the Unit 3 original precipitator was rebuilt, and NEPSCo evaluated
Flue Gas Conditioning Systems. NEPCo evaluated Flue Gas Conditioning industry options and
sulfur trioxides (S03) injection was selected as the system to insuring optimum precipitator
performance. NEPCo worked with EPPJ and installed a new process called EPR1CON on Brayton
Point Units 1&2. The EPR1CON process uses a side stream of flue gas to produce the SO3. The flue
gas side stream is run through a catalyst which converts S02 to SO3 and distributes the SO3 back into
the main flue gas flow. This was a first of a kind installation and has provided very effective control
at Brayton Point Station. A conventional SO3 conditioning system was installed on Brayton Point
-------
- 4
Unit 3 to insure precipitator performance.
• SO2 Compliance Results. Starting in 1995, the Massachusetts Acid Rain Law required the
Brayton Point and Salem Harbor Units to achieve a compound annual average SO, emission rate of
1.2 Ibs per MMBtu or less. Meeting this Massachusetts requirement will put NEPCo SO2 emissions
roughly in balance with the SO2 allowances to be received from the EPA in the year 2000. Brayton
Point and Salem Harbor Station significantly reduced their emissions of sulfur dioxide (SO2) from
historic levels. For both 1995 and 1996 the annual average SO, emission rate was less than 1.1
Lbs/MMBTU's for the coal fired units.
NOX Compliance
Plans for NOX controls of the fossil generating stations started in the late 80's and continue into the 90's.
The 1995 NOX compliance plans encompassed a number of different NOX control strategies. The
Massachusetts RACT Regulations allowed four different methods of compliance. NEES evaluated all
four compliance methods in detail and elected to comply with the Generic Limits. This method
provided the most stringent NOX Limits but provided the most straight forward regulatory approval
process. The Generic RACT Limits for coal fired units are .45 and .38 Lbs/MMBTU for wall fired and
tangential fired boilers respectively. The NEES units with the RACT Limits and other more stringent
Regulatory Limits are provided hi Table 4.
NEES selected a variety of control strategies and manufacturers to suit the specific unit requirements. In
general, there are three basic types of NOX controls which include pre-combustion, combustion and post-
combustion. NEPCo developed considerable expertise in all three methods of NOX control in the
compliance efforts.
• NOX Pre-Combustion Controls. Pre-combustion NOX control primarily consists of fuel
changes. For the three major fuels, the uncontrolled emission rates vary by fuel with natural gas
having less uncontrolled NOX emissions than either coal or oil. NEPCo provided the three coal fired
units at Brayton Point Station with natural gas co-firing to allow enhanced start-ups and operational
flexibility which can be translated into both NOX and SO2 reductions. When cofiring natural gas
which has negligible sulfur, the SO, emission reductions are directly related to the percent of natural
gas fired. NOX reductions on the other hand are an in-furnance emission reduction process which is
more complex and, unit specific. To evaluate the potential for seasonal NOX controls extensive gas
cofiring was conducted on the Brayton Point coal units, and the results are reviewed later in this
paper.
• NOX Combustion Controls. Combustion controls included Low NOX Burners with and without
overfire air. NEPCo has installed Low NOX burners in five of the six units using three different
manufacturers (ABB CE, B&W, and Riley Stoker) with the sixth unit being retrofit with burner
components to achieve partial NOX reduction. The low NOX burners combined with overfire air
achieved greater than 50% reductions.
• Post Combustion Controls. NEPCo was first in the nation to use Selective Non-Cataytic
Reduction (SNCR) controls on coal fired boilers. SNCR which consists of injecting chemicals
-------
5 -
(Urea) into the upper portion of the furnace to react with and to reduce the NOX in the flue gas.
SNCR Systems were installed on the three coal fired units at Salem Harbor Station and when
combined with low NOX burners, they achieved approximately a 70% NOX reduction from the 1990
baseline levels.
Gas Co-Firing. A gas co-firing development project jointly sponsored by the Electric Power
Research Institute (EPRI), The Gas Research Institute (GRI), New England Power Company and
Energy System Associates (ESA) was conducted on the three Brayton Point Coal Fired Units.
Brayton Point Units 1&2 are subcritical units with tangentially fired twin furnaces. The units were
retrofitted with ABB CE Low NOX concentric firing system (LNCFS) Level III with close coupled
overfire air (CCOFA) and three elevations of separate overfire air (SOFA). The Low NOX burner
retrofit included gas ignitor and upper level gas burners. Ignitor co-firing testing provided a 12%
reduction in NOX emissions when firing 10% natural gas.(l) The upper romance gas burner
reburning of NOX emissions was hampered by insufficient natural gas jet penetration into the
furnace. Improvements in the gas penetration are in review and include gas burner modifications
and the use of a gas carrier.
Brayton Point Unit 3 is an opposed wall-fired Babcock & Wilcox (B&W) designed supercritical
boiler. The unit was retrofit with B&W DRB-XCL Low NOX Burners and separated overfire air.
The Low NOX retrofit included gas variable heat input igniters at each burner capable of providing
up to 10% of the units heat input. Brayton Point Unit 3 also had a CO limit of 200 ppm which limits
the reduction in excess oxygen. The gas co-firing was found to have a significant positive benefit on
CO emissions reductions. As little as 2.5% gas co-firing mitigated or eliminated the high CO
peaks.(2) The improved CO emission characteristics allowed reduction in the oxygen levels, which
resulted in NOX reductions. Demonstrated reduction of NOX were achieved in the 5 to 10% range
using gas heat input at or above the 2.5% level. Seasonal NOX reductions will be required in
Massachusetts 1999 and gas co-firing is one of the control options that will be considered for
additional NOX reductions.
NOX Results. Since May 31,1995, the RACT Compliance date, there have been significant NOX
Reductions in all of NEPCo coal fired units. Some of the major improvements include:
• 100% compliance with NOX Emission Limits (enforced on a 24 hour average) for the six coal
fired units.
• Overall NOX Reductions of greater than 65% from Brayton Point and Salem Harbor coal units
1990 Baseline NOX Emission rates.
• Currently 70% NOX Emission Rate Reductions from Brayton Point Unit 3, a supercritical unit.
• 70% NOX Emission Rate Reductions from Salem Harbor Units 1,2 & 3 using the combined NOX
reduction technologies of Low NOX Burners and Selective Non-Catalytic Reduction.
• "No Visible Stack Emissions" at Brayton Point Station. The station has gone from visible
plumes to no visible signs of the coal unit operations.
-------
• No detectable waterwall wastage due to Low NOX Burners after more than two years of service
for all six units.
• Cost effective NOX reductions at approximately $400/Ton of NOX removed.
Summary
The SO2 and NOX Compliance effort has been and continues to be very successful due to the dedication
and commitment of all at NEES to 100% environmental compliance. A summary list of the projects
implemented in this program is provided in Table 4 and graphly displayed in Figure 1.
Acknowledgements
There were numerous NEES engineers, construction workers and managers, outside vendors,
manufacturers and supporting firnis which contributed to the successful SO2 and NOX compliance efforts
at NEES' Brayton Point and Salem Harbor Stations.
Some of the many NEES departments involved in the Clean Air Projects include: Generation
Operations, Power Engineering (Mechanical, Civil, Controls and Electrical Engineering and Design),
R&D, Fuels, Generation Marketing, Load Forecasting, Environmental and Safety, Generation Services
Construction (Mechanical, Electrical, Structural, Environmental Services and Construction
Coordination), Purchasing, Legal and the Station's management, operations and maintenance personnel.
The final acknowledgement is to NEES' executive management which provided the solid commitment
and resources to achieve and maintain compliance and to the Massachusetts Department of
Environmental Protection (DEP) regulators who worked to insure timely implementation of this
program.
References
1. Gas Cofiring Deployment and Validation Project at New England Power Company's Brayton Point
Units 1 and 2, Prepared by: Energy Systems Associates, GPJ - 96/0343, 1996.
2. Gas Cofiring Deployment and Validation Project at New England Power Company's Brayton Point
Unit 3, Prepared by: Energy Systems Associates, GRI - 96/0344, 1996.
-------
Table 1
Brayton Point (BP) & Salem Harbor (SH) Unit Characteristics
Station
/Unit
BP1
BP2
BP3
SH1
SH2
SH3
Net
°&
245
, 245
626
81
78
145
Primary
Fuel
Coal
Coal
Coal
Coal
Coal
Coal
Boiler
Manuf.
CE
CE
B&W
B&W
B&W
B&W
Finnq
Tangential
Tangential
Opposed
Wall
Wall
Wall
No. of
Burners
32
32
40
12
12
16
Table 2
SO2 Compliance Requirements & Actual Emissions
Combined Bravton Point & Salem Harbor Coal Units
1990 Baseline Limit 2.4 Lbs/MMBTU
1995 Regulatory Limit 1.2 Lbs/MMBTU
1995 Actual Emissions 1.09 Lbs/MMBTU
1996 Actual Emissions - 1.07 Lbs/MMBTU
Table 3
Range of Coal Properties
Properties
Moisture (%, As Received)
Fixed Carbon (%)
Ash (%)
Volatile Matter (T)
Nitrogen (%, Dry)
Sulfur (%)
Heating Value (Btu/lb)
Grindability Index (HGI)
Range
6-12
45-54
8-12
26-36
1.2-1.7
.6 -.8
12,000-13,500
42-55
-------
Table 4
NOX Compliance Requirements and Actual Emissions
Station/
Unit
BP1
BP2
BP3
SH 1
SH2
SH3
"ESP
Coal
Coal
Coal
Coal
Coal
Coal
NCX Emissions (Lbs/MMBTU)
1990
Base
.7
.7
1.4
1.0
1.0
1.0
Reaulatorv Limited)
.38
.38
.45
.33
.33
.33
1996
Annual
Average
.30
.32
.38
.30
.30
.30
Percent
Reduced
Baseline
57%
54%
73%
70%
70%
70%
Note-
T. Based on a regulatory agreement Salem Harbor Units 12 & 3 had NOX
Emission Limifs below the RACT Limits of .45 Lbs/MMBTU.
Table 5
SO2 & NOX Compliance Technologies for Brayton Point & Salem Harbor
Unit
BP 1
BP2
BP3
SH1
SH2
SH3
Primary
Fuel
Coal
Coal
Coal
Coal
Coal
Coal
Fuel
Modifications
LSC/GCF
LSC/GCF
LSC/GCF
LSC
LSC
LSC
Furnace Controls
LNB
X
X
X
X
X
QFA
X
X
X
X
SNCR
Control
X
X
X
SO3
Cond.
X
X
X
Other
Controls
(1)
(2)
Key.:
LSC Low Sulfur .(Compliance) Coal
LNB Low NOvBumers
UFA pverfire vMr (Separate)
SNCR Selective Non-Catalytic Reduction
SO3 Cond. Flue Gas Conditioning with SO3 System
Notes:
1 . The original Unit 3 precipitator was rebuilt. The unit has an original precipitator in series
with a new precipitator installed as part of a late 1 970's coal conversion
2. Dunnq normal maintenance the coal nozzles and coal spreaders were replaced with the
updated Riley Low NOX design.
-------
FIGURE I
BRAYTON POINT AND SALEM HARBOR STATIONS
COAL FIRED UNITS "SOZ' AND "NOX" CONTROI .S
BP 162 EPRICON
CONTINUOUS
EMISSION MONITORS
BP3 SO, CONDITIONING
BP 162 GAS BURNERS
BP 1-3
SH 3
OVERFIRE AIR
BP 1-3 GAS CO-FIRING
BP 1-3
SH 163
LOW NOX
BURNERS
PRECIPITATOR PRECIPITATOR
ONLY)
BP 1-3 6 SH 1-3
LOW SULFUR COAL
SH 2
BURNER
MODS
COAL FIRED UNITS
BP I - 245 MW
BP 2- 245 MW
BP 3- 626 MW
BP 162 PULVERIZERS
SH I - 81 MW
SH 2 - 78 MW
SH 3 - 145 MW
-------
Planning for NOX Emission Regulations
via EPRI's CAT Workstation
Jim Schott
Entergy Services, Inc.
Charles Altin
Raytheon Engineers & Constructors, Inc.
As the EPA and the states struggle to address existing and proposed ambient air quality standards
for ozone and particulates, electric utilities must prepare for a very uncertain future. Regulations
could take many forms in the future, ranging from unit-specific emission limits to plant, non-
attainment, state-wide or system-wide bubbles to emissions trading. Utilities are, therefore,
faced with a myriad of choices ranging from the exact value of a system-wide NOX emission
limit to which control technology is appropriate for a particular unit.
To further compound utility planning problems, industry restructuring looms on the horizon. For
some sites, a project which is economical from a system perspective could damage a particular
unit's ability to compete in a restructured industry. For this reason, it is critical to evaluate the
assumptions and scenarios that drive decisions.
Questions of whether units would be retired, what new generation facilities need to be built, and
how current unit loading regimes will change in the future need to be addressed. The
centerpiece of the evaluation is not only what is the lowest cost and most practical way to meet a
new NOX emission limitation, but also what are the key uncertainties and assumptions that drive
decisions.
The issues of NOX transport and its contribution to ozone formation and transport are complex.
These issues have been and are now being addressed in other forums. Therefore, this paper does
not touch on these issues, but does address the impacts of stricter NOX emission standards.
In 1996, Entergy Services embarked on system-wide assessment of NOX control technologies and
their attendant costs. The goal of this initial assessment was to identify the major contributing
units; establish reduction requirements associated with three potential system-wide emission
rates; define cost impacts to meet the potential emission rates; and promote a framework to focus
\i\consult\ent-nox2.doc
07/22/97
-------
resources in the future to refine control technology capabilities, impacts and cost estimates on an
individual unit basis. In essence, this initial assessment would frame the issues and point a
direction for future analysis.
In order to perform this initial assessment and recognizing the large number of units and control
technology combinations, it was necessary to utilize a computer model which would meet
assessment budgets and schedule constraints. To that end, we found EPRTs Air Emissions CAT
Workstation™ to be a perfect fit for the assessment's needs.
Unit Characterization and Setup
The assessment began by characterizing each of the 54 fossil-fuel fired units in Entergy's system.
This characterization involved identifying each unit's projected rate in generation over the next
12 years. Essentially, units were classified as either active or inactive. Although the assessment
concentrated on the active units, the system-wide database developed included the inactive units.
This inclusion of inactive units provided for future flexibility should generation planning evolve
into a different base scenario. For each unit, projected annual capacity factors were established
as well as bi-hourly loading regimes on a four season basis. The primary fuel for each unit was
then designated and unit heat rate curves from 10 to 100 percent were developed. Next, unit NOX
emission rate (Ib/MBtu) versus unit load curves were created from the Acid Rain CEMS data.
With the unit technical data developed, the next step involved identifying the economic
parameters upon which the analyses would be conducted. These parameters were levelized fixed
charge rate, pre-tax discount rate, general escalation rate, replacement capacity and energy
charges, fuel cost and escalation, and the unit's remaining life. The technical and economic data
was inputted to the CAT Workstation.
The CAT Workstation permitted establishing a NOX reduction profile for each control technology
or combination thereof specific to each unit. These reduction profiles were based on the
uncontrolled NOX emission rates over a 10 to 100 percent load range. The reduction profiles
were based on generally accepted technology achievable reductions. Reduction rates were
decreased as load was reduced, and eventually a minimum sustainable NOX emission rate, i.e.
outlet stopper, was used. The relationships between NOX, load and control technology reduction
levels add great complexity to the analysis. These complexities make NOX compliance planning
far more difficult than Title IV Acid Rain SO2 compliance planning.
\i\consult\ent-nox2.doc
07/22/97
-------
Essential to any assessment analysis is the financial impact of control technologies on a unit-
specific basis. This was again established through the use of the CAT Workstation supplemented
with recent implemented project costs. This technique gave assurance that economic decisions
would be realistic. The Workstation's decision making revolved about making the least cost
selection mix of technology and reduction level based on combining average reduction cost
($/ton removed) and annual fuel cost.
With unit characterizations complete, technology reduction levels and costs identified, and the
system-wide database developed, the analyses can then move forward. In doing so, the first task
is to determine the current annual NOX emissions on a unit-specific and system-wide basis. This
is referred to as the "Do Nothing" scenario. This then provides a baseline to evaluate the impact
of proposed limitations on NOX emissions. For example, such proposed system-wide limitations
could be:
Case I
• 0.25 Ib NOx/MBtu for coal-fired units
• 0.20 Ib NOx/MBtu for gas- and oil-fired units
0.20 Ib NOx/MBtu for all fossil-fired units
Case III
• 0.15 Ib NOx/MBtu for all fossil-fired units
These scenarios permit one to determine how the unit/control technology mix changes;
establishes order of magnitude cost impacts; identifies a need to provide for safety factors so that
annual emission limitations are not exceeded; and creates a basis to develop a future work plan
that focuses resources in a most cost-effective manner.
With such an overall plan in place, we found that the CAT Workstation identified many critical
issues which may not have been readily apparent. These issues can be summarized by the
following:
• The NOX versus load curves are developed through use of CEMS data. In using data over a
year's period of time, there can be a significant amount of scatter with variations in the ±20
\i\consult\ent-nox2.doc
07/22/97
-------
to 40 percent range from the mid-point curve. Some units may even have variations
approaching 100 percent. This creates tremendous uncertainty in any assessment and can
almost assure that either excessive emission controls would be applied or that emission
limitations would be exceeded on a routine basis. Therefore, each unit should be fine tuned
and operate with low excess air in order to minimize and stabilize NOX emissions, thus
reducing data scatter and decision uncertainty.
For large units, annual capacity factors can change by up to 20 percentage points over a ten-
year study period, whereas, small units tend to have consistent capacity factors. It should be
noted that generally the more complex and capital-intensive control technologies are applied
to large units since they generate more kilowatt-hours over which to spread the capital
investment. However, significant reductions in capacity factors will increase the marginal
production cost for these units; hence they may be placed lower on the generating dispatch
order, further aggravating the situation. Therefore, wherever possible, large unit capacity
factors should be maintained at a reasonably constant level.
In evaluating the impact of proposed NOX limitations, the time over which the emissions are
averaged becomes critical. It was found that annual emission limitations are easier to achieve
than seasonal limitations due to a wider variety of unit loading regimes. Typically, the
"ozone season" extends from March through October. For many areas of the country, this is
the peak generating season where large units have very high capacity factors and peaking
units are brought on-line. All of this translates into high NOX emission rates. When meeting
annual emission limitations, one gains flexibility in achieving a specific limit if one can take
advantage of relatively low NOX emissions during the spring and fall.
It is important to identify those units which most contribute to the system's overall NOX
emissions. One may find that 20 percent of a system's units contribute more than two-thirds
of the annual NOX emissions. These "heavy hitters" then need to be the focus of most of the
assessment efforts and refined analysis. The CAT Workstation, therefore, helps to focus
engineering resources on critical units.
The actual NOX reductions achievable at a unit will be highly dependent upon boiler
configuration, design parameters and operating practices. Therefore, before committing to a
specific control technology as a function of economic computer modeling, additional
investigation and testing work must be performed to establish actual and reliable achievable
NOX reduction levels. Such testing is usually more easily accomplished for combustion-
related technologies such as burners-out-of-service or biased firing.
\i\consult\ent-nox2.doc
07/22/97
-------
• Although capital cost estimates are within reasonable limits, operating and maintenance costs
must be closely examined. We found it most helpful to convert computer model generated
values into a $/ton removed basis which provided a better way to adjust them to reflect actual
project experiences.
• One must be mindful that there are practical limits to any analysis in terms of the number of
unit/control technology combinations to be studied. These limitations can be computer
program or time/cost driven. Hence, prescreening of unit/technology combinations should be
performed. For example, small, low capacity factor units would not use capital-intensive
technologies.
• One may wish to view the model results in terms of a NOX reduction level and a cost but not
necessarily reflecting a specific control technology. For example, if a 40 percent reduction
was required, then further site-specific and detailed cost estimates would be developed via
additional studies. The study would then focus on which technology (i.e., burners-out-of-
service, biased firing, flue gas recirculation, or selective noncatalytic reduction) can meet or
better the economic value upon which a decision was made in the initial assessment.
In conclusion, the CAT Workstation is a cost-effective method to performing initial screening of
proposed emission limitation and provides focus to further evaluation activities to finalize a
compliance plan.
\i\consult\ent-nox2.doc
07/22/97
-------
THE COST OF COMPLYING WITH NOX EMISSION REGULATIONS
FOR EXISTING COAL FUELED BOILERS
Michael G. Gregory
John R. Cochran
Diane M. Fischer
Mickey G. Harpenau
Black & Veatch
8400 Ward Parkway
Kansas City, MO 64114
Abstract
Utilities are facing many difficult decisions regarding NOX compliance. The challenge is how to
meet these new NOX emission regulations while simultaneously positioning plants for the
competitive marketplace of a deregulated industry.
Complicating these decisions is the uncertainty regarding future NOX regulations. Title IV NOX
emission limits have been finalized and a number of states in the Ozone Transport Region (OTR)
are well along the way of implementing Memorandum of Understanding (MOU) agreements.
However, EPA and state implementation of the Ozone Transport Assessment Group (OTAG)
recommendations have not yet been finalized.
Although the cost of NOX reduction alternatives can be well defined, the development of an
effective strategy to hit the two moving targets of NOX compliance and deregulation must be made
at some risk. Accordingly, determining the optimum compliance strategy is an iterative process.
Black & Veatch has assisted numerous utilities by providing regulatory insight along with
performance, cost, and plant impacts for all major types of NOX reduction equipment on all major
boiler types. This paper summarizes the capital and economic cost results of these coal fueled
boilers investigations. To assist in compliance plan development, current and pending emission
regulations are also summarized.
Compliance Planning
Developing the cost of compliance has two predominant components: strategy development and
technology assessments. Activities associated with strategy development primarily consist of
assessing current and likely future regulatory requirements and subsequent systemwide modeling of
utility compliance costs. The result of strategy development is the establishment of a specific plan
for compliance. Technology assessments consist of establishing the site specific performance and
costs of implementing a range of compliance alternatives. The results of these assessments are used
as inputs to strategy development activities. As such, a number of questions must by addressed
during compliance planning.
-------
Strategy Development:
What known emission limits must be met?
What level of future emission limits may have to be met?
How will compliance costs impact competitiveness in a deregulated industry?
What is the cost to implement various strategies?
What is the most cost effective strategy that minimizes future risk?
Technology Assessment (Site Specific):
What is the performance potential of each alternative?
What are the impacts to plant operation?
What is the cost to install and operate this technology?
Regulatory Environment
Federal, state, and regional environmental agencies and groups are in the process of establishing
regulations to reduce NOX emissions from existing power plants. The following is a brief
discussion of some major regulatory factors presented to illustrate the present and potential
requirements that impact utilities developing a NOX compliance plan.
Federal Regulations: Title IV NOX Reduction Program
In response to Title IV of the 1990 Clean Air Act Amendments, the final rules of 40 CFR 76
detailing NOX emissions reduction requirements for coal fired boilers was promulgated in December
1996. This rule defines the Phase n NOX emission limits for Group 1 and Group 2 boilers. These
requirements are summarized in Table 1. As noted, the emission limits are dictated by the type of
boiler. Each boiler type has a technology basis, which is the EPA-selected technology expected to
achieve the required NOX emissions, although use of these technologies to meet the emission limit is
not specifically required.
The federal regulations allow multiple power plants to perform emissions averaging among all the
units with the same designated representative. Emission averaging consists of comparing the
average emission rate of all the units in the averaging plan to the emission rate those same units
would average if each unit met the limits listed in Table 1. From this comparison, the plant owner
can determine how much systemwide reduction is required and decide how best to achieve these
reductions.
These regulations require all power plant owners to submit a NOX Compliance Plan to the EPA by
January 1, 1998, that discusses the proposed plan for complying with the regulatory requirements
meeting the regulations. All plans must be implemented by January 1, 2000.
Ozone Transport Region (OTR)
Twelve northeastern states and the District of Columbia, which make up the OTR, signed an MOU
in which they agreed to reduce NOX emissions of utility sources from 1990 levels. The MOU
requires that NOX emissions be reduced by 65 percent for states in the Inner Zone and 55 percent
for states in the Outer Zone by the year 1999.1 Additional reductions of 75 percent in both the
Inner and Outer Zones and 55 percent for the Northern Zone are required by the year 2003.'
-------
The OTR will not be a uniform code because the MOU calls for the states to develop their own
State Implementation Plan (SIP). These SIPs will define NOX emission limits, trading programs,
and emission seasons within that state. These seasons typically require further NOX reduction
during the hot weather months when the impact of NOX emissions on ozone formation is highest.
Ozone Transport Assessment Group (OTAG)
OTAG is a group of state agency, industry, and environmental organizations from 37 states,
including states in the Midwest, South and Northeast. OTAG is organized to evaluate the impact of
pollution from upwind states on the air quality of downwind states.
On June 3, 1997, OTAG recommended that the EPA determine an emissions reduction program
which would reduce NOX emissions below that of federal Title IV requirements. While the OTAG
recommendation does not specify an actual emission limit, their findings indicate that NOX
emissions from power generation facilities need to be reduced by 55 to 85 percent or to an emission
limit of 0.15 Ib/MBtu in order to minimize ozone transport. It was also recommended that
12 states, and portions of eight other states, be exempt from any rules developed as a result of the
OTAG findings.2
By not providing quantitative reduction requirements, OTAG left the EPA with the primary
responsibility of developing the emission limits regulations to reduce ozone transport. OTAG's final
report will provide the EPA with the detailed emissions and modeling data developed during the
OTAG evaluation.
Other Environmental Initiatives
Other environmental activities are also occurring which could result in additional reduction
requirements. The Clean Air Power Initiative (CAPI), a program started by the EPA, encompasses
the development of a regulatory strategy which would bring consistency to SO2, NOX, and mercury
emissions for electric power generators. The results of this effort could further reduce NOX
emissions limits. Also, in July 1997 the EPA promulgated final Ambient Air Quality Standards
(subject to congressional approval) which revise the standards for ozone and replace the standard
for PM-10 with new standards for particulate matter less than 2.5 microns. These regulations have
the potential to significantly affect the power industry, especially those facilities in areas with poor
air quality.
Utility Deregulation
Along with the environmental activities, the federal government and state governments are
implementing alternatives for utility system restructuring and deregulation. The goal of most of
these restructuring plans is to reduce the cost of power to the consumer by providing a more
competitive environment. Many states, such as Massachusetts and California, have already initiated
programs to restructure the utility industry in their state. In addition, several bills which address
this issue are currently before Congress. Although each proposal has unique features, they share
the following common characteristics:
• Removing the link between generation of power and transmission of power.
• Allowing consumers to choose their power generation provider.
-------
• Addressing how stranded costs incurred by the utilities will be handled in a deregulated
market.
• Increasing the role of independent power producers.
Deregulation impacts the development of compliance strategies because utilities that develop cost
effective compliance strategies will have an advantage over those utilities that do not optimize their
compliance plans.
NOX Reduction Alternatives
Several alternatives exist for reducing NOX emissions from coal fueled boilers. The following is a
brief overview of the alternatives that have been demonstrated in operation and are commercially
available for coal fueled units. These alternatives are typically described as being either combustion
or post-combustion control techniques.
Combustion Control
Combustion controls suppress NOX formation during the combustion process, and include boiler
tuning, low NOX burners, and natural gas reburn.
Boiler Tuning. Boiler tuning consists of operational modifications utilizing the current boiler and
boiler auxiliary equipment. Boiler tuning can be implemented on most types of existing boilers to
achieve NOX reductions up to 25 percent. Boiler tuning is attractive because capital investment is
either minimal or not required. However, operating costs may increase as a result of more frequent
boiler testing and a greater degree of operator and maintenance personnel attention.
Operational modifications may include coal flow balancing, air flow balancing, optimizing flue gas
recirculation (if applicable), reduced air preheat, reduced excess air, biased optimizing, or burners
out of service. Some tuning contractors attempt to reduce NOX formation through the use of
computational modeling and modification of control setpoints while others provide testing services
in an effort to balance items such as coal and air flow. Many methods have proven successful, but
tuning contractors do not typically guarantee NOX reduction.
Low NOX Burners (LNB). LNBs are available for wall and tangential fired boilers as well as cell
fired boilers.
LNBs for wall and tangential fired boilers control the mixing of fuel and air in a pattern designed to
minimize flame temperatures and quickly dissipate heat. These burners typically reduce the NOX
generated by maintaining a reducing atmosphere at the coal nozzle and diverting additional
combustion air (to complete combustion) to secondary air registers. This minimizes the reaction
time at oxygen-rich, high-temperature conditions. The construction time for an LNB retrofit is
approximately 14 weeks with an outage time of about 8 weeks.
An LNB for a cell burner is somewhat different than those for wall or tangential fired boilers
because the burners are too close to allow installation of the standard multi-register burners. Most
retrofits involve shifting introduction of the coal from the over/under two burner cell to inject all
the coal through the lower burner. The upper burner is then converted into an overfire air (OFA)
-------
port. Another method to retrofit LNBs onto cell fired units is to increase the space between the
close coupled cell burner and install a standard LNB. This alternative requires a relatively complex
installation due to water wall and windbox modifications.
Operational impacts include additional primary air system resistance and the potential for an
increase in unburned carbon. To minimize the increase in unburned carbon, manufacturers typically
suggest coal fineness of 65 to 80 percent through 200-mesh and 98 to 99 percent through 50-mesh.
Retrofit installation of LNBs may impact opacity and paniculate emissions.
Natural Gas Reburn. Natural gas reburn is a common form of fuel staging. The natural gas
rebuming process employs three separate combustion zones to reduce NOX emissions. The first
zone consists of the normal combustion zone in the lower furnace in which 75 to 80 percent of the
total fuel heat input is introduced with about 10 percent excess air (a 1.10 stoichiometric ratio).
The second combustion zone, called the reburn zone, is created above the lower furnace by
operating a row of natural gas burners at a stoichiometric ratio of about 0.80 to 0.95. The
introduction of OFA to complete combustion of the unburned materials in the upper furnace
comprises the third and final zone. The process results in an overall excess air for the boiler of 15
to 20 percent.
Adequate residence time within the furnace for both the additional burning zone and the associated
OFA burnout is required for optimum performance. If adequate residence times are not available,
the effectiveness of the reburn system will be limited. For applications with adequate residence
times, reburn technology has demonstrated reductions in NOX of 40 to 65 percent.
Implementation of gas reburn often requires the installation of new gas supply lines, reburn burners,
and an overfire air system. Integration of reburn systems in existing plants includes interfaces such
as the air heater outlet, penetrations into the boiler for the reburn burners and OFA ports, and the
control system upgrades. Due to the complexity of the reburn operation and the requirement for
accurate control, a digital control system is required.
Natural gas reburn has been demonstrated on several domestic units. However, lack of long-term
experience introduces some risk associated with applying gas reburn technology to utility sized
units. The construction schedule for the installation of a gas reburn system is approximately
18 weeks including about 8 weeks of outage time.
Post-Combustion Control
Post-combustion controls are flue gas treatments that reduce NOX after its formation. Alternatives
consist of selective non-catalytic reduction (SNCR), selective catalytic reduction (SCR), and hybrid
(SNCR followed by catalyst) systems.
SNCR. SNCR systems can use either ammonia or urea as the reagent. These systems rely on
appropriate reagent injection temperature and good reagent/gas mixing to achieve NOX reduction.
The optimum temperature range for injection of ammonia or urea is approximately 1,550 to
2,200°F. Therefore, the optimum temperature occurs in the backpass of the boiler. The location of
this temperature window will change as a function of unit load so multiple injection levels are
-------
usually installed. In addition, effective reagent injection becomes complex and difficult for boilers
larger than 150 MW.
The performance of SNCR systems are a function of boiler arrangement and allowable ammonia
slip. On a given boiler, the ammonia slip will increase as the NOS reduction increases. The
conventional philosophy for coal fueled boilers limits ammonia slip to 5 ppm or less, which leads to
NOX reduction of 20 to 35 percent. However, recent experience indicates that to ensure reliable air
heater operation, ammonia slip values should be limited to 2 ppm or less4.
Installation of SNCR systems is relatively simple. The reagent storage tanks, injection metering
skids, and piping can be installed with the unit online. A short outage of 1 to 2 weeks is required
to finish installation of the injection points on the boiler.
SCFI. With an SCR system, ammonia is injected into the flue gas stream upstream of a catalyst
bed. Ammonia in the presence of the catalyst reacts with both NO and NO2 to form nitrogen and
water vapor. SCR systems have been used on over 50,000 MW of coal fired boilers throughout
Germany and Japan for many years and are now operating on six coal fired units totalling nearly
1,800 MW in the United States.
The ammonia (either aqueous or anhydrous) is received and stored as a liquid and is then
vaporized, diluted with air, and injected into the flue gas. Injection of the ammonia must occur at
catalyst temperatures between 600 and 800° F (site specific). Therefore, catalyst is typically located
between the economizer outlet and the air heater inlet.
Specific plant characteristics will affect the ease by which SCR can be implemented in a retrofit
situation. Boilers which have short duct runs between the economizer and the air heater may
require a large amount of ductwork reconstruction to install SCR. Units with horizontal shaft air
heaters generally present difficult retrofit scenarios resulting in additional costs. In these situations,
the SCR can also be installed downstream of the particulate removal or flue gas desulfurization
system where additional space is often available. However, this arrangement requires duct burners
and a gas-gas reheater to raise the flue gas temperature to the minimum operating temperature of
approximately 600°F.
SCR systems can be designed to keep the ammonia slip reliably below 2 ppm. At these levels, the
problems associated with plugging of air heaters and ammonia deposition on the fly ash are
minimized. The pressure drop resulting from an SCR system retrofit will be between 3 and 6 in
w.g. This often results in ID fan modifications. In most cases, forced draft units must be
converted to balanced draft operation. The outage required to tie-in the SCR reactor ductwork to
the existing ductwork is about 3 to 5 weeks. Total construction time is approximately 18 months.
SNCR/Catalyst Hybrid Systems. Hybrid systems attempt to combine the features of SNCR and
SCR. Reagent is injected in the boiler like a traditional SNCR system. However, rather than trying
to minimize ammonia slip, NOX reduction is maximized, and the resulting ammonia slip is used for
further NOX reduction in the catalyst bed. Typically, the catalyst is installed inside the ductwork to
minimize cost. A variation of the hybrid concept is to install the catalyst material in the
regenerative air heaters. The catalyst allows the SNCR to have the greatest amount of effectiveness
while maintaining ammonia slip as low as 2 ppm.
-------
Performance is based on SNCR effectiveness, catalyst volume, velocity through the catalyst, and
distribution of the ammonia slip within the flue gas stream. Because the catalyst is typically
installed inside the existing ductwork (or air heater), construction must be completed with the unit
out of service, leading to construction outages of 10 to 20 weeks, with a total project construction
time of 30 to 40 weeks.
Cost of NOX Reduction Alternatives
Black & Veatch has performed compliance planning activities for more than 60 coal fueled units
totalling in excess of 17,000 MW over the past 4 years. The cost information presented in this
section summarizes a majority of the results from these investigations. Capital cost, busbar cost, and
cost effectiveness values are provided for each NOX control technology as a function of NOX
reduction. The NOX reduction percentage was selected as the basis of comparison in order to
illustrate and define the situations where each technology is most appropriate. Table 2 describes the
wide range of boiler types and sizes used in developing the costs that are presented in this paper.
As always, costs are very site specific. The costs presented in this paper have been somewhat
normalized to ensure a level of comparability. The results presented should not be used in lieu of
more detailed compliance planning estimates. However, the results presented are representative of
relative cost ranges.
Capital Cost Ranges
Figure 1 presents the capital cost of the most common NOX control alternatives expressed in dollars
per kilowatt ($/kW) on a single unit basis. These capital costs include contingency, owner
indirects, interest, and escalation for a January 2000 startup.
This figure illustrates the differences in both performance and cost between these technologies. For
LNB, gas reburn, SNCR, and hybrid (when based on an in-duct catalyst) the maximum attainable
NOX reduction is defined by the physical arrangement of the boiler. The maximum NOX reduction
of SCR is a design factor, with little impact from boiler characteristics. For example, the
performance of a gas reburn system may be limited in a boiler that does not have the height
required to achieve adequate residence time while the performance of an SCR system is based on
the required NOX removal.
Figure 1 illustrates that LNB and SNCR have a smaller range of capital costs than gas reburn, SCR,
and hybrid systems. Those alternatives whose performance is dictated by boiler arrangement and
baseline NOX emission values have capital costs that are not highly influenced by boiler
arrangement and baseline NOX. The equipment and construction required for installation of an LNB,
gas reburn, or SNCR system are largely independent of NOX reduction percentage and retrofit
difficulty. The apparently large range of capital cost for gas rebum results from including the
pipeline required to supply natural gas to the coal boiler. Otherwise, the range of capital cost
would be quite small.
-------
Conversely, the capital cost of SCR, whose performance is not defined by boiler arrangement, is
greatly influenced by boiler arrangement and associated retrofit factors. As illustrated on Figure 2,
retrofit difficulty and baseline NOX emissions have a significant impact on the capital cost of an
SCR system. Table 3 lists the definitions of retrofit difficulty ranging from nominal to difficult.
Busbar Cost
Development of a NOX compliance strategy must also evaluate the total ownership (annual O&M
and capital) cost of each alternative. Busbar costs for each alternative technology are illustrated on
Figure 3. This information is based on levelized annual costs (capital and operating costs for a
20-year life) and includes, as applicable, differential fuel, reagent (ammonia for SCR and urea for
SNCR), differential power consumption (including fan energy due to increased pressure drop),
catalyst replacement, O&M (material and labor), and fixed charges on capital. NOX reduction
system costs are based on 12-month operation. This information is useful in estimating the total
annual expenditures to operate each technology independent of NOX removal.
Cost Effectiveness
Figure 4 presents the cost effectiveness of the various technologies defined as levelized dollars (as
defined in the previous section) per ton of NOX removed. Essentially, cost effectiveness is the true
measure of economic efficiency.
All control alternatives demonstrate a wide range of cost effectiveness due almost entirely to the
combination of baseline NOX emissions and removal efficiency. This information shows that
focusing on high NOX producing units within a system as part of an averaging strategy may have
the greatest strategic value for minimizing systemwide compliance costs. This concept of averaging
will be further illustrated in the following section of this paper.
Figure 5 illustrates the relationship between cost effectiveness and busbar costs for an SCR system
(90 percent removal) installation on a 500 MW boiler with a nominal retrofit factor. As the cost
effectiveness improves due to higher uncontrolled NOX emissions (more tons of NOX removal), the
busbar cost increases because of higher requirements for ammonia and energy consumption.
Although the cost effectiveness and busbar values would be different, this relationships is consistent
with other NOX reduction systems such as SNCR, hybrid, and gas reburn.
The Value of Systemwide Averaging
There are two options for incorporating NOX control technologies into a compliance strategy. The
determination of which option to use is based on comparing the systemwide emission limit target to
the existing NOX emissions. The outcome of this comparison will help decide which option to use.
Option 1: Unit by unit compliance: install reduction alternatives as required to achieve
compliance for each unit.
Option 2: System averaged compliance: install high-reduction alternative on one or
more units to offset the emissions of other units.
-------
Option 1: Unit by Unit Compliance
This option is based on installing a NOX control technology on all units requiring emission
reductions to meet NOX limits. For Group 1 boilers, this most likely includes installation of LNB.
A large portion of Group 2 boilers do not have such a ''plug-in" alternative available, so control
alternatives such as boiler tuning and the EPA-defined technology basis can be implemented to
achieve compliance with the emission limit. Developing the compliance cost of Option 1 consists
of adding the capital and operating costs of those technologies selected to be installed at each unit.
Option 2: System Averaged Compliance
Title IV, known OTR SIPs, and draft OTAG requirements allow for some degree of systemwide
averaging among units. While the details may vary or are not yet finalized, the option of averaging
multiple units has the potential to cost substantially less than individual compliance as described in
Option 1.
For those generating systems that have both Group 1 and Group 2 boilers, this option is based on
the concept of leveraging by overscrubbing NOX on one or more of the high baseline NOX emitters
to cover the reduction requirement boilers needing less NOX reduction.
To illustrate this option, a fictitious utility (ABC) is presented as an example. ABC has two 500
MW cyclone units, each with a baseline NOX emission of 1.3 Ib/MBtu. The remaining units in the
ABC system consist of lower emitting Group 1 and Group 2 boilers. ABC's objective is to meet
Title IV NOX emission limits. Figure 6 illustrates the number of megawatts of generation from the
remaining units that can be offset if SCR systems designed to achieve a controlled emission limit
of 0.13 Ib/MBtu (90 percent removal) were installed on the two cyclone units.
In this example, the two SCR systems would produce 24,000 tons per year of NOX reduction that
could be used to offset other units within the system. As shown on Figure 6, the offset is great
enough to cover 3,000 MW of wall-fired boilers with uncontrolled NOX emissions of 0.70 Ib/MBtu,
1,400 MW of cell-fired boilers with uncontrolled NOX emissions of 1.2 Ib/MBtu, 3,800 MW of
tangential fired boilers with uncontrolled NOX emissions of 0.6 Ib/MBtu, or any other combination
of boilers requiring 24,000 tons of NOX emission offsets. Therefore, leveraging high-NOx units to
offset the need for numerous burner replacement retrofit projects can be very effective in
developing the best compliance strategy.
An important consideration in developing a system averaged compliance strategy is ensuring that
the emission limit can be met year-round under various generating system operating scenarios.
While using the offsets from the installation of an SCR on a very high NOX producing unit may be
able to cover all other units in the system, this compliance concept will become jeopardized in the
event the controlled plant is taken off line due to an unexpected long-term outage. Installation of a
limited number of LNBs on selected units as part of the systemwide compliance plan would provide
a buffer for this type of unexpected occurrences.
It is also prudent to design the SCR system for additional removal to allow for compensation of
downtime, so if a long-term outage does occur on the high-reduction unit, the SCR can be operated
for higher reduction for the remainder of the year.
-------
Summary
Utility compliance plans for meeting Title IV NOX emission limits must be submitted to the EPA on
January 1. 1998. Every compliance plan will be different, but the components of strategy
development and technology selection will be the key tools in the development of each plan. An
appropriate strategy should be based on meeting emission limits, optimizing capital and operating
expenditures, and positioning the plants to remain competitive in the deregulated market.
The cost of compliance is determined by the selected strategy to meet the NOX emission limit and
the technology selected to implement this strategy. In developing the strategy, site specific factors
will predominate because no single technology provides a "one size fits all" solution. While
numerous strategies can achieve the desired emission limit, thorough, comprehensive planning is
required in the selection of a compliance plan that will provide performance, flexibility, cost
effectiveness, and competitiveness during the upcoming deregulation of the industry.
References
1. Ozone Transport Commission Press Release, September 27, 1994.
2. Utility Boiler Regulations Under the Title TV NOx Emission Reduction Program, presented at
the AWWA 1996 Technical Conference, September 1996, by Stanley Rasmussen, Kevin
Eisenbeis, Andy Byers, Black & Veatch
3. Environmental Reporter, June 6, 1997, Volume 28, No. 6, OTAG Suggests Broad Range of
Controls for Utility Emissions, Exempts 12 States
4. "Seward Unit 5 An SNCR NOX Control Experience," Daniel Kessler, GPU Genco, presented
at the Pennsylvania Electric Association Meeting, Indiana, Pennsylvania, April 8, 1997.
10
-------
Type of Boiler
Table 1
Title IV NOX Emission Limits
Phase II Emissions
Standard
Technology Basis
Group 1 Boilers
Tangential
Dry Bottom Wall
Group 2 Boilers
Cyclone (> 155 MW)
Wet Bottom (> 65 MW)
Cell Burners
Vertical
0.40 Ib/MBtu
0.46 Ib/MBtu
0.86 Ib/MBtu
0.84 Ib/MBtu
0.68 Ib/MBtu
0.80 Ib/MBtu
Low NOX Burners
Low NOX Burners
Gas Reburn or SCR
Gas Reburn or SCR
Plug-In and Non Plug-in
Combustion Controls
Combustion Controls
Table 2
Boiler Types Investigated in NOX Compliance Planning Studies
• Boiler Types.
-Front and Rear Wall Fired PC.
-Tangential PC.
-Cell.
-Cyclone.
-Wet Bottom.
• Boiler Sizes Ranging from 80 to 1,300 MW.
• Uncontrolled NOX Ranging from 0.35 to 2.5 Ib/MBtu.
• Bituminous and Subituminous Fuels.
• Wide Range of Retrofit Difficulties.
11
-------
Table 3
Definition of SCR Retrofit Difficulty Factors
Nominal.
—Clear Access for Construction.
—Unimpeded Reactor and Ductwork Arrangement Possible.
-Adequate ID Fan Capabilities.
-No Electrical System Modifications.
—No SCR Bypass Necessary.
Moderate.
-Clear Access for Construction.
-Difficult Ductwork Transition.
-ID Fan Motor Replacement Required.
—Moderate Electrical System Modifications.
-Includes SCR Bypass.
Moderately Difficult.
-Construction Access Implemented.
—Some Relocations.
-ID Fan Motor and Rotor Replacement.
-Extensive Electrical Modifications.
Difficult.
-Construction Significantly Impeded.
-Extensive Relocations.
—Difficult Ductwork Transition.
-Constrained Reactor Arrangement.
-Complete ID Fan Replacement.
—Extensive Electrical Modifications.
12
-------
120
100
80
60
40
*8*
• » X
1 . .
•
+
+ *
K-i^i *
*
*
&
'
''
1
1
SNCR Systems
•
SCR System
Hybrid
-f-
Gas Refaurn
3K
Low NOx Burner
X
4 20% 40% 60% 80% 100%
Percent NOx Reduction
Capita] Costs Include Escalation, Contingency, Owner Indirects,
and AFUDC for a January 2000 Startup
Figure 1
NOX Compliance Planning Study All-in Capital Costs
500 MW; 90% Removal; 2 ppm Ammonia Slip;
1.5 to 2.0% S Coal; 75% Capacity Factor
.*••"""
J
1 — ^
• • " ""
— • —
-••*-*
h—.^-
^ • ^
-—- •
... —
— •
""
1— — •"
,
DifGcult**
Moderately Difficult
Moderate
Nominal
0.5 1 1.5 2 2.5
Uncontrolled NOx Emission, Ib/MBtu
* Costs Include Contingency, Escalation, Owner Indirects, and IDC
** Docs Not Reflect SCR Construction Situations Necessitating Prolonged
Outages of Greater Than 1 Month
Figure 2
SCR (High Dust) Capital Cost
13
-------
*
• y ;
f
>
5
=
^
>
:
SNCR Systems
SCR System
Hybrid
Gas Rebum
1 *
Low NOx Burner
20% 40% 60% 80% 100%
Percent NOx Reduction
' Annual Costs are Levelized (20 Year) and Include as Applicable
Differential Fuel, Reagent, Power, Catalyst Replacements,
O&M, and Filed Charges on Capital
Figure 3
NOX Compliance Planning Study All-in Busbar Costs
5 4'000 "
oi
3
.
• *
•
**>
X
- 8
i
++ -$
,
jf
i
SNCR Systems
•
SCR System
Hybrid
+
Gas Rebum
Low NOx Burner
0% 20% 40% 60% 80% 100%
Percent NOx Reduction
• Annual Costs are Levelized (20 Year) and Include as Applicable
Differential Fuel, Reagent, Power, Catalyst Replacements,
O&M and Fixed Charges on Capital
Figure 4
NOX Compliance Planning Study All-in Cost Effectiveness
14
-------
500 MW; 90% Removal; 2 ppm Ammonia Slip; Moderate
Retrofit Factor; 1.5 to 2.0% S Coal; 75% Capacity Factor
_ 900-
« 700-
5 ....
V
\
\
fs*
*
',
*'
X
f
.''
*
"-^
.
^•*
- — _
-J t-J OJ W *
D wi 0 Lft c
mills/kWh
Vton
mills/kWh
0.5 1 1.5 2 2.5
Uncontrolled NOx Emission, Ib/MBtu
Costs Include Fixed Charges on Capital and Levelized Annual Operating Costs
Figure 5
SCR Cost Effectiveness
Example SCR Rarcfi t Would Offset Need for Burner Retrofits at
3000 MW of Wall Fired Boilers With Uncontrolled Emissions of 0.7
Ib/MBtu
f7'000
•?
S
£
s
\
\.
\
1
\
S7>
\
j
f
\
i
/
/
i
/
/
\
*~.
"i
^
!
Wall Fired Dry Bottom
(0 46 Ib/MBtu Limit)
Tangential
(0.40 Ib/MBtu Limit)
(0.68 Ib/MBtu Limit)
*
1 1.2
Uncontrolled NOx Emissions, Ib/MBtu
'Installation of 90% SCR on 2 X 500 MW Cyclone Boilers With Uncontrolled NOx
Emissions of 1.3 Ib/MBtu and NOx Emission Limit of 0.86 Ib/MBtu.
Figure 6
System Averaging Example
15
-------
Monday, August 25; 10:00 a.m.; 1:00 p.m.
Plenary Session:
Combustion Tuning/Optimization (PC Units)
-------
The Role of Combustion Diagnostics in Boiler Tuning
R. E. Thompson, F. P. Haumesser
Fossil Energy Research Corp.
Laguna Hills, California 92653
T. A. Davey
Consumers Energy
Essexville, Michigan 48732
A. Hickinbotham
TransAlta Utilities Corp.
Calgary, Alberta T2 P2 Ml
Abstract
Boiler tuning to achieve low NOX emissions, efficient operating O2 levels and acceptable ash
carbon content requires effective combustion diagnostics to quickly identify equipment and
operating constraints. Many coal-fired utility boilers have nonuniform combustion as indicated
by large O2 and NOX gradients at the economizer exit. Common causes of this nonuniformity or
imbalance in O2 and NOX are uneven fuel distribution to the burners, malfunctioning or
misadjusted air registers/dampers, restricted or plugged burner pipes, nonuniform air distribution
to OFA ports, uneven or variable burner tilt position, furnace air inleakage, etc.
Failure to identify and correct the causes of nonuniform combustion can often result in slagging,
fouling and high ash carbon levels in the low O2 regions of the furnace and high NOX, and
reduced efficiency in the high O2 regions. This paper discusses some common causes of
nonuniform combustion, the use of cost-effective diagnostic techniques and experiences from
recent field test programs. In particular, the paper addresses how to distinguish air inleakage
from nonuniform combustion, the importance of proper placement of plant instrumentation,
specific equipment related diagnostics, and issues related to performance guarantees for low NOX
firing systems. The equipment related diagnostic techniques discussed in this paper are an
important complement to optimization software.
Introduction
The importance of combustion diagnostics in boiler tuning has grown significantly as more
utilities install retrofit NOX emission controls on pre-NSPS units and also convert to low sulfur
coal (or coal blends). Although the installation of low NOX burners and/or overfire air (OFA)
systems may satisfy the Phase I requirements of the Clean Air Act Amendments (CAA), many
utilities have realized that achieving reduced emissions and peak performance often requires
optimization of the combustion process. Although periodic calibration/maintenance of O2
-------
analyzers and pulverizers may have been all that was needed in the past, more extensive boiler-
tuning to establish and maintain uniform combustion is now required. Failure to achieve uniform
fuel and air distribution to the burners can aggravate a natural tendency of low NOX burners and
OFA systems to operate at increased ash carbon levels. Local regions of low O2 combustion are
often the source of CO and high ash carbon levels, whereas local high O2 regions produce above
average NOX emissions.
Another boiler tuning related issue concerns the conversion of a boiler originally designed for a
high Btu Eastern bituminous coal to a low sulfur high moisture Western coal. This can result in
a major change in boiler combustion characteristics and an increased sensitivity to equipment
maintenance and operating conditions. Increased coal pipe loadings at high moisture, low mill
outlet temperatures can lead to coal pipe plugging and nonuniform combustion. The tendency of
high calcium content Western coals to deposit a reflective ash often results in high furnace exit
gas temperatures and slagging/fouling. Local fuel rich regions caused by nonuniform
combustion can trigger a slagging and fouling episode resulting in a unit derate or forced outage.
A more recent combustion related problem that is emerging at some Phase I units is accelerated
furnace waterwall tube wastage.1 High waterwall corrosion rates have been most noticeable on
supercritical high heat release rate units burning high sulfur fuels. Corrosion occurs primarily as
the result of reducing atmospheres along the waterwalls, particularly on units with large amounts
of overfire air (OFA). Nonuniform combustion in the burner zone can cause flame impingement
on the walls, fuel rich reducing atmospheres, and ash deposition that traps corrosive species.
Supercritical units are more susceptible to waterwall wastage because of their inherently higher
operating temperatures. Although some of these units previously experienced waterwall wastage
before conversion to low NOX firing, the wastage accelerated with the advent of low NOX deeply
staged combustion which increases the potential for fuel rich reducing atmospheres at the furnace
wall.
As utilities gain experience with retrofit low NOX firing systems and conversion to coals with
combustion characteristics different from the design coal, the importance of combustion
diagnostics and boiler-tuning during baseline testing has also grown. Baseline testing is typically
conducted with the original equipment burners and the normal coal to establish a reference point
of emissions and performance prior to conversion to low NOX firing and/or an alternate coal.
Some utilities previously relied on little more than plant continuous emissions monitor (CEM)
and boiler panel board data from existing instrumentation during pre-retrofit baseline testing.
Although large gradients in O2 and NOX emissions may have been present in the combustion zone
(indicating potential coal and air flow distribution problems related to coal pipe or windbox air
distribution), they went unnoticed. Thus, important indicators of potential future combustion
related problems were ignored with the expectation that the new low NOX equipment would
accommodate any off-design operation inherent in the baseline boiler firing practice. However,
utilities are now realizing, as Phase n baseline testing accelerates, that detailed combustion
diagnostics data are essential in establishing a good baseline from which to compare post-retrofit
warranty emissions and performance. Failure to carefully document pulverizer performance,
burner settings, O2 and NOX gradients, furnace exit gas temperatures, furnace wall atmospheres,
and LOI can lead to potential warranty disputes with equipment vendors concerning the cause of
a possible change in unit performance. For example, if a large increase in LOI occurs following
-------
the installation of an OFA system, is it due solely to the deep staging associated with low NOX
firing? Another possibility is that LOI increases are due to other changes (e.g., pulverizer
fineness, mill bias, air register settings or increased furnace air inleakage) that occurred between
the pre-retrofit baseline and post-retrofit warranty tests. Without detailed combustion diagnostic
test data under comparable boiler operating conditions, it is often difficult to resolve these issues.
In summary, the role of combustion diagnostics in boiler tuning has grown to meet the following
needs:
• Tuning of Phase I units to reduce LOI, optimize NOX, increase efficiency/thermal
performance and reduce waterwall tube wastage.
• Emissions and performance optimization of Phase I units that have failed to meet vendor
warranty requirements and have petitioned for an Alternative Emission Limit (AEL).
• Baseline testing of Phase n units to carefully document existing combustion related
emissions and performance constraints that could impact post-retrofit operation.
• Optimization of emissions and performance on Phase n units that qualified for the "early
election" option for NOX control under the CAA amendments.
• Reduction of slagging and fouling derates or forced outages on boilers having an inherent
sensitivity to variations in coal quality and combustion characteristics.
• Trial test burns of varying blend ratios of low sulfur Western and Eastern bituminous coals to
minimize SO2 emissions and boiler performance impacts.
Combustion Diagnostics and Boiler Tuning
Combustion diagnostics play an important role in boiler tuning because utility boiler emissions
and performance are directly dependent upon the quality and effectiveness of the combustion
process. Achieving low NOX emissions at efficient operating O2 levels with acceptable ash
carbon content requires effective combustion diagnostics to quickly identify equipment and
operating constraints. Some of the most common combustion related problems concern:
• Operation at nonoptimum excess air levels
furnace air inleakage
O2 instrumentation operation/placement problems
• Equipment malfunctions/maintenance issues
air heater partial pluggage or seal leakage
FD/ID fan capacity limitations or windbox pressure constraints
plugged burner pipes
worn coal pipe orifices
worn pulverizer grinding elements
-------
broken air register linkages or motor drives
combustion controls performance problems
• Lack of combustion uniformity
uneven coal flow distribution
uneven air flow distribution
uneven overfire air distribution
A very common problem with balanced draft coal-fired boilers is operation at a nonoptimum
excess air level. In many cases the actual O2 level in the burner zone is significantly less than
that measured at the plant O2 probes because of air inleakage between the furnace and
economizer exit. High LOI or ash carbon content often results. A common source of this
problem is leaks in the furnace roof, casing leaks and leaks at duct expansion joints. If the air
inleakage is more prevalent on one side of the boiler than another, it is very difficult to maintain
balanced combustion conditions because the instrument readings will not accurately reflect the
actual O2 in the burner zone. From a combustion diagnostics perspective, furnace air inleakage
can be characterized by comparing measurements of the O2 concentration profile at the furnace
exit with those at the economizer exit.2
Combustion diagnostic testing to evaluate O2 probe placement or equipment operation problems
is strongly recommended. Even in cases where air infiltration is not an issue, the O2 analyzer
reading may not be representative of the duct average. This is a more common problem with
tilting tangentially-fired units where a large O2 gradient or "split" between furnace halves can
develop for certain burner tilt positions. A large O2 gradient can result not only in above average
NOX emissions from the high O2 regions, but it can lead to operational problems if not rectified
(e.g., high LOI, slagging and fouling in the low O2 regions of the furnace). Many utilities are
reevaluating the location of their O2 probes or are installing multiple O2 analyzers in each duct to
obtain a more representative average. Combustion diagnostic testing at the furnace and
economizer exit can aid in defining the best location for these new analyzers. However,
recalibration of the existing plant O2 analyzers is strongly recommended before conducting any
combustion diagnostic testing.
Many combustion related emissions and performance problems are related to equipment
malfunctions or maintenance related issues that may have been overlooked during periodic boiler
inspections. Some examples include: (1) air heater partial pluggage or seal leakage; (2) FD/ID
fan capacity limitations; (3) plugged burner pipes; (4) worn coal pipe orifices; (5) worn
pulverizer internal or classifier settings; (6) broken air register linkages or motor drives; or
(7) combustion controls problems (e.g., response time, dead band, etc.).
Nonuniform Combustion
Although many equipment and operational related issues can contribute to boiler emissions and
performance problems, a very common source is nonuniform combustion in the burner zone.
Nonuniform combustion makes efficient low O2, low NOX firing difficult because local fuel-rich
zones can result in: (a) incomplete combustion, CO or elevated ash carbon content, and (b) a
tendency toward increased ash deposition with coals that have a marginal slagging index. To
-------
avoid this, the boiler overall operating O2 level is often set at a higher than required average level
resulting in unnecessarily high NOX emissions and reduced boiler efficiency. For example, if one
burner region or furnace corner is operating at a much lower O2 level than the others, it will
dictate the overall average O2 at which the unit is operated, even if the majority of the burners are
capable of operating at a much lower level. Thus, one of the primary goals of boiler tuning is to
establish uniform combustion so that the overall O2 level (and NOX emissions) can be lowered.
Several of the more common causes of nonuniform combustion include:
• Uneven coal flow distribution
coal piping and orifices
riffle box configuration
biased pulverizer coal flow
• Uneven air flow distribution
air register/damper settings
windbox design
air register/actuator malfunction
• Uneven overfire air distribution
Achieving uniform combustion typically involves combustion diagnostics to improve both the
fuel and air-side balance. The general approach to boiler combustion tuning is summarized
below:
• Measure coal fineness, primary air and coal flow distribution
• Optimize mill performance
• Improve coal fineness
• Characterize air inleakage between furnace and economizer exit
• Balance coal flow to individual burners
• Balance air flow
• Adjust secondary air dampers to achieve uniform air/fuel ratio at each burner
• Reduce air infiltration
• Improve instrumentation/placement
• Bias mills between elevations - O2, NOX and LOI optimization
Typically, the most significant improvements in combustion uniformity are achieved by
performing combustion diagnostics on the fuel system to achieve uniform coal flow to all burners
at target fineness levels. Balanced coal flow is typically achieved by measuring the "as found"
primary air and coal flow distribution followed by modification of the individual burner line
orifices to achieve balanced coal flow within ±10% of the mean.
Burner coal pipe diagnostic testing typically involves measurements of both the primary air and
coal flow distribution plus pulverized coal fineness using a RotorProbe™ sampler. The
RotorProbe™ is an approved ISO method of pulverized coal sampling. It has a rotating sample
head that collects coal samples at sixteen radii of the coal pipe (which is very important if any
-------
coal roping is present). A typical example of the improvement in coal flow balance achieved with
coal pipe orifice adjustments is shown in Figure 1 for a 125 MW tangentially-fired unit equipped
with four Raymond Bowl mills. Coal flow deviations of up to 12% were reduced to 5% or less.
The corresponding emission benefits of balancing coal flows are summarized below:
• NOX (corrected to 3% O2) was reduced from 0.411 Ib/MMBtu to 0.346 Ib/MMBtu,
• CO was reduced from 91 to 50 ppm,
• LOI was reduced from 15.7% to 9.7%.
The improvement in combustion uniformity allowed the boiler to operate at reduced O2 levels
benefiting NOX emissions without adversely impacting CO emissions. The LOI reduction was
attributed to a combination of better combustion balance and improved coal fineness through
50 mesh.
Results of a similar combustion diagnostics test program to improve coal flow balance on a 135
MW tangentially-fired unit are shown in Figure 2. Deviations in coal flow of more than 10%
were reduced to 3% or less. The corresponding emission benefits were:
• NOX (corrected to 3% O2) was reduced from 0.483 Ib/MMBtu to 0.423 Ib/MMBtu,
• LOI was reduced from 12.8% to 10.1%.
A 165 MW tangentially fired unit with 16 burners had large deviations in coal flow between
burner pipes ranging from +28% to -22% as shown in Figure 3. Replacement coal pipe orifices
improved the coal flow balance to ±10% for all but one pipe (12%). The associated
improvement in combustion uniformity is illustrated in Figure 4, where the baseline O2 and NOC
profiles at the economizer exit are shown in the top half of the figure. The profiles after coal
flow balancing are shown in the lower portion, where NOC = NO @ 3% O2. (All other test
conditions were directly comparable, except the coal flow distribution.) Because of significant
air inleakage at the ends of the duct, the improvement in combustion uniformity is not very
obvious based on just O2 measurements at the economizer exit, except in the duct region between
15 and 35 feet where the O2 gradient from 2.4 to 3.2% was completely eliminated.
Since NOC emissions are so strongly dependent upon local air/fuel ratio in the burner zone, the
baseline and post orifice replacement NOC emissions profiles in Figure 4 give a much clearer
picture of the benefits of balanced coal flow in terms of combustion uniformity. A NOC
emissions gradient of 40 ppm (or 15%) was evident in the baseline NOC profile but reduced to
20 ppm (or 7%) with new coal pipe orifices. Eliminating the low 02 region (2.4%) in the
baseline O2 profile, normally would allow a further reduction in the average O2 and NOC
emissions for comparable CO and LOI levels. However, these efforts are dependent upon
additional pulverizer testing to improve coal fineness, which currently dictates the unit operating
practice. Secondary air damper tuning to further minimize the NOC gradients in Figure 4 could
also be performed, but typically is not initiated until all pulverizer work has been completed.
-------
Advanced Diagnostic Instrumentation
Fossil Energy Research Corp. has found that advanced combustion diagnostic instrumentation is
essential to identifying and resolving combustion uniformity problems, particularly those
associated with retrofit low-NOx burner and OFA systems. Two instruments have proven to be
particularly valuable:
• a real time multipoint NO, O2, CO combustion diagnostics analyzer (12 channels),
• a portable HOT FOIL® LOI analyzer (for rapid ash LOI determinations).
In a typical installation at a large unit, a grid of 24 sample probes (12 per duct) is installed in the
divided economizer exit duct as shown on the right side of Figure 5. Gas samples are drawn
individually through each probe in the grid to define the gas composition at each location in the
duct. A computer graphics program is used to generate the emissions contour plots previously
shown in Figure 4 to evaluate combustion uniformity. Until recently, generating just one of the
profiles for a 24-point sample grid would require up to a half-day of point-by-point sampling,
data reduction and plotting. However, Fossil Energy Research Corp. has developed a real time
multipoint sampling, data acquisition and graphical display system that continuously monitors
and displays O2, NO and CO profiles. The system, shown on the left side of Figure 5, allows up
to 12 points in a duct to be sampled and analyzed simultaneously. The PC-based data acquisition
and custom software updates the screen display at 10-second intervals, computes averages over
user selectable time intervals and provides storage of the profiles to a hard disk for later recall.
With the use of this custom instrumentation, the test engineer can make air register, mill bias or
fan bias adjustments and immediately see the impact on the O2/NO profiles and the boiler
combustion uniformity. This instrumentation has been very effectively used to optimize NOX
emissions by means of combustion tuning. Fossil Energy Research has found that it can
significantly reduce the time needed to tune a boiler by using two way radio communications
between an equipment operator making air register adjustments on the burner deck and a test
engineer in the truck who can observe the change in the emission profile(s). The test engineer
can direct the burner adjustments until uniform O2 and NOX emissions have been achieved.
Because the NO emissions data are corrected to 3% O2 to compensate for air inleakage, the
difference between the O2 and the NO profiles can be used to distinguish air inleakage from
nonuniform combustion.
The diagnostic approach outlined above was very effectively applied during a test program
conducted on a 375 MW tangentially-fired unit that had very large variations in coal flow to
individual burners. The burner pipe-to-pipe variation in primary air and coal flow, measured
with a RotorProbe™, is summarized in Table 1 for each of the five pulverizers.
Normally, Fossil Energy Research recommends that initial diagnostic testing to improve
combustion uniformity focus on balancing primary air and coal flow distribution to the burners
before tuning the secondary air dampers/registers. However, in this case the real time multipoint
NO, O2, CO combustion diagnostic analyzer was used to adjust secondary air dampers/registers
in a manner that offset the coal flow imbalance. Table 2 summarizes the results of these damper
adjustments. A large O2 and NOC imbalance or "split" between the North and South furnaces
-------
was evident during the full load baseline test with the A mill OOS (i.e., a 0.6% O: difference and
78 ppm (-30%) NOC difference), as shown in Table 2 for Test # 1.
Table 1
Coal and Primary Air Flow Variation at a 375 MW Tangentially-Fired Unit
Pulverizer
Number
3A
3B
3C
3D
3E
Primary Air Flow
Percent Deviation from the Mean
±22.6%
±10.9%
±33.3%
±21.7%
±13.1%
Coal Flow
Percent Deviation from the Mean
±30.0%
±20.0%
±27.2%
±33.6%
±22.8%
Table 2
Results of Damper Adjustments on a 375 MW Tangentially-Fired Unit
Test
No.
7
8
Mill
OOS
A
A
North
O2
%
2.4
2.85
NOC
ppm
229
269
CO
ppm
2
0
South
02
%
3
2.7
NOC
ppm
307
272
CO
ppm
5
5
Delta
02
(S-N)
%
0.6
-0.15
Delta
NOC
(S-N)
ppm
78
3
Test Description
Baseline with A Mill OOS.
Bias North and South Aux.
Air dampers, and South Fuel
dampers.
Secondary air damper adjustments (Test #8) reduced the O2 "split" between furnaces from 0.6%
to .15% and reduced the NOC split from 78 ppm (30%) to 3 ppm (1%). In addition, combustion
uniformity within each furnace was improved, particularly the South furnace, as shown in the
"before" and "after" NOC profiles in Figure 6. A South furnace NOC gradient of 60 ppm (19%)
was reduced to 20 ppm (6%) following a brief period of damper tuning using the multipoint
combustion diagnostics analyzer for real time viewing of uniformity improvements. Similar full
load tests were conducted with other mills out-of-service and the O2 "split" was typically reduced
from 65-85% and the NOC split from 50 to 96%. Based on these preliminary combustion
diagnostic and boiler tuning tests, it was estimated that an average O2 reduction of 0.5% was
feasible corresponding to a NOC reduction of 20%. These predicted benefits are based solely on
boiler tuning as opposed to hardware modifications.
Boiler secondary air tuning was conducted on a 350 MW tangentially-fired unit that also
exhibited a large coal flow non uniformity.2 NOC emissions gradient from 380 to 550 ppm (i.e.,
170 ppm or 37%) was reduced in one day to only 20 ppm or 5% with damper adjustments.
-------
Following these adjustments, the composite O2 level was briefly reduced to the boiler minimum
air stop which prevented further O2 reductions. Except for these controls limitations, it is
anticipated the overall O2 could have been reduced by 0.6 to 0.8% for a substantial gain in boiler
efficiency.
Boiler combustion diagnostics and tuning are much more easy to perform on wall-fired units than
on tangentially-fired units because of the plug flow nature of the flow field through the furnace
compared to the cyclonic flow in a tangentially-fired unit. Baseline combustion diagnostic tests
on a 130 MW wall-fired unit with four mills and 16 burners indicated a substantial full-load O,
and NOC gradient as shown in Figure 7. Although combustion uniformity was difficult to judge
based on the O2 contours because of air inleakage, a large gradient in NOC (140 ppm or 26%) was
evident in the NOC profile. After only two sets of burner secondary air register adjustments, the
NOC gradient was reduced to less than 10 ppm or 2% as shown in Figure 8.
Overfire Air Combustion Diagnostics
It should be noted that the real time multipoint combustion diagnostics analyzer has also been
used to effectively tune OFA systems (which are not inherently well balanced). A 250 MW twin
furnace tangentially-fired boiler was initially emitting substantial CO emissions at high O2 levels
of 4% or greater.3
Using the multipoint combustion diagnostics analyzer to guide the OFA damper adjustments, it
was necessary to open the separated overfire air (SOFA) dampers on the front of the boiler to
three times the opening of the rear dampers in order to achieve uniform balanced combustion that
was comparable to the pre-retrofit emission contours with the original equipment burners. At the
conclusion of the OFA adjustments (conducted prior to the warranty tests), the composite O2 was
reduced by nominally 0.9% at comparable CO levels. The estimated annual fuel cost savings at
an 80% capacity factor was approximately $110,000. Subsequent tests on a similar 250 MW
twin furnace tangentially-fired unit confirmed that the unbalanced OFA flow was not an isolated
incident. A similar OFA tuning test series was conducted to achieve uniform combustion at
design OFA flow rates.
Conclusions
The role of combustion diagnostics in boiler tuning has grown significantly in response to a need
to improve emissions and performance of: (1) Phase I units requiring optimization, (2) baseline
testing on Phase n units, (3) early election and alternative emission limit units, and (4) units
undergoing slagging and fouling episodes (or operating with blends of low sulfur western coals
having substantially different properties than the design coal).
Non-uniform combustion can result in above average O2 levels and NOX emissions and/or local
regions of low O2 combustion with high CO and LOI levels. Common causes of non-uniform
combustion include uneven coal flow distribution, uneven air flow distribution, or uneven
overfire air distribution. The general approach to boiler combustion tuning includes the
following actions:
-------
• Measure coal fineness, primary air and coal flow distribution.
• Optimize mill performance.
• Improve coal fineness.
• Characterize air inleakage between furnace and economizer exit.
• Balance coal flow to individual burners.
• Balance air flow.
• Adjust secondary air dampers to achieve uniform air/fuel ratio at each burner.
• Reduce air infiltration.
• Improve instrumentation/placement.
• Bias mills - O2, NOX, and LOI optimization.
Measurement and balancing of coal flow to the individual burners are recommended as the first
step in achieving combustion uniformity, although an effective balance can often be achieved
with secondary air tuning, particularly if the coal flow imbalance is not large. The use of a real
time combustion diagnostics analyzer allows boiler tuning to be performed very quickly and
cost-effectively compared to older manual methods.
References
1. Jones, C, Malady ofLow-NO., Firing Come Home to Roost, Power, Jan/Feb. 1997, pp 54-60.
2. Thompson, R.E., et a]., Boiler Tuning for NO,, Control as an Alternative to Low
NO^ Burners. ASME Int'l. Joint Power Generation Conf., Phoenix, Arizona, October 1994.
3. Thompson, R.E., et al., Optimization ofLow-NOx Overfire Air Systems to Improve Boiler
Performance", EPRI Heat Rate Improvement Conf., Dallas, Texas, May 1996.
BURNER PIPE NUMBER
Figure 1. Improvement in Coal Flow Balance Due to Coal Pipe Orifice Modifications,
100 MW Wall-Fired Unit
-------
\
1
|
1
F1
1
F4
m
*
i
i
\z
a
S
I
1
Bi
F9 ' F12 ' R1 R4 ' R9 R12
BURNER PIPE NUMBER
Figure 2. Improvement in Coal Flow Balance Due to Coal Pipe Orifice Modifications,
400 MW Turbo-Fired Unit
I 2"
PULVERIZER 1&2 COAL FLOW DEVIATIONS
BURNER PIPE NUMBER
PRE-RETHOFtT
POST-HETROFIT
PULVERIZER 3&4 COAL FLOW DEVIATIONS
M N
BURNER PIPE NUMBER
1 PRE-RETRORT
POST-RETROFIT
Figure 3. Improvement in Coal Flow Balance with Coal Pipe Orifice Modifications,
165 MW Tangentially-Fired Unit
-------
Baseline O2 (%) and NOC (ppm @ 3% O2) Profiles
10
15
20 25
Width, Ft.
30
35
40
20 25
Width, Ft.
35
40
Post Orifice Replacement O2 (%) and NOC (ppm @ 3% O2) Profiles
15 20 25
Width, Ft.
40
20 25
Width, Ft.
40
Figure 4. Improvement in Combustion Uniformity with Coal Flow Balancing,
165 MW Tangentially-Fired Unit
-------
Figure 5. Typical Multipoint NO, O2, CO Analyzer Installation
BEFORE
Front Wall (Feet)
AFTER
Front Wall (Feet)
5 10 15 20 25 30 35
Rear Wall (Feet)
5 10 15 20 25 30 35
Rear Wall (Feet)
NOC (ppmc, dry @ 3% O2)
Figure 6. Improvement in Combustion Uniformity by Secondary Air Damper Tuning
to Offset Coal Flow Imbalance, 375 MW Unit
-------
PRE-TUNING BASELINE O2 CONTOURS
WEST, FT.
35
PRE-TUNING BASELINE NOC CONTOURS
WEST, FT.
10
\
at
T
SSPr"'
^
ieiWmff
/ /
'/
o / / o o en *• > p\
-------
0-
FIRST REGISTER ADJUSTMENT NOC CONTOURS
WEST, FT.
30 35
EAST, FT.
. 10-
g 5H
SECOND REGISTER ADJUSTMENT NOC CONTOURS
WEST, FT.
—I—
15
I
20
EAST, FT.
25
—I—
30
oc
o
Figure 8. Uniform Combustion Achieved with Secondary Air Register Tuning,
ISOMWUnit
-------
POST LOW NOX BURNER RETROFIT BOILER TUNING RESULTS
FOR A FRONT WALL-FIRED BOILER
James J. Ventura, P.E.
Staff Engineer
Union Electric Company
P.O. Box 66149
St. Louis, Missouri 63166-6149
Abstract
In June 1996, Union Electric Company completed a major retrofit of its Meramec Plant Unit #4
boiler. The Foster Wheeler front wall-fired boiler was originally commissioned in 1961. Among
the major modifications, the boiler was retrofitted with Babcock & Wilcox DRB-XCL® low NOX
burners, furnace rear wall overfire air ports, furnace sidewall ports, extensive fuel delivery
equipment modifications, and a Westinghouse WDPF burner management control system.
Post retrofit performance objectives included reducing NOX emissions below 0.45 Ib/MMBtu at
full load (a 58% reduction), CO emissions not to exceed 175 ppm at full load, and unbumed
carbon (UBC) objectives based on a matrix of parameters for two contract coals.
Tuning of the boiler is still in progress. Preliminary tuning results indicate that predicted NOX
performance will be met. However, problems attaining a satisfactory balance between boiler
excess O2 and CO production remain. Due to higher than desired boiler excess O2 levels, current
UBC results are inconclusive.
Background
Union Electric Company's Meramec Plant is located at the confluence of the Meramec and
Mississippi Rivers just south of St. Louis, Missouri. Unit #4 was placed on line in 1961 as a base
loaded 360 MW unit. Figure #1 shows a current cross-section arrangement of the boiler and
pulverized fuel system. The boiler, originally supplied by Foster Wheeler, is a pulverized coal-
fired subcritical drum boiler with a full load steam flow capacity of 2,310,000 pounds per hour at
2200 psig and 1010 °F. Furnace depth is a relatively narrow 28'-6" Three double ended ball
tube mills provide pulverized coal to 18 burners located on the furnace front wall. The burners
are arranged with six burners in each of three horizontal rows. Each mill provides coal to one
level of burners. A rotating table-type feeder (2 per mill) feeds coal to each end of the mills. Coal
is conveyed from a classifier on each end of the mills by an exhauster (2 per mill) and transported
through a ceramic-lined three-way distributor and coal piping to a trio of burners.
-------
Reasons For Retrofit
Compliance with the Clean Air Act Amendments of 1990 was the primary stimulus for retrofit of
the original Unit #4 Foster Wheeler Intervane burners. Prior to the combustion system retrofit,
NOX emissions were on the order of 1.02 Ib/MMBtu. Being a dry-bottom, wall-fired, pulverized
coal unit, Meramec Unit #4 was required to comply with a Phase INOX emission limit of 0.50
Ib/MMBtu. Additional impetus to replace the burners was provided by the general condition of
the boiler furnace. As a result of Union Electric adding more efficient coal-fired units and a
nuclear unit, Meramec Unit #4 has been relegated to cycling duty since the early 1970's. Having
been designed for base load operation, cycling duty had extracted a heavy toll on Unit #4's steam-
cooled Downflow Upflow Radiant Superheater furnace front wall, boiler tieback and buckstay
system, boiler insulation and casing system, and windbox structural integrity. By 1990, studies to
redesign and replace the boiler furnace were already well underway. Accordingly, the furnace
retrofit provided an opportune time to replace the original burners with a low NOX emitting
combustion system.
Low NOX Combustion System Retrofit
After evaluation of separate bids, Babcock & Wilcox (B&W) was ultimately selected to supply
both the furnace retrofit and low NOX combustion system supply scope. The low NOx
combustion system consists of eighteen DRB-XCL® low NO* burners, six overfire air ports, and
two sidewall air ports.
The DRB-XCL® burners are a dual zone burner design that operates on the principle of delayed
combustion.1 The burner diverts air away from the core of the flame, which increases coal
devolatization and subsequently reduces initial NC\ formation. The burners are located on the
new furnace front wall in the same general location as the previous burners with six burners in
each of three horizontal rows. The upper and lower elevation burners were supplied with conical
diffusers and notched end rings for flame stability. The middle elevation of burners were supplied
with distribution cones and conical impellers to minimize flame length and CO production.
A NOX port is located on the right and left furnace sidewall at the lower burner elevation. These
ports were designed to provide secondary air to not only reduce NOX emissions from the lower
elevation burners but also intercept their flames before they reach the furnace rear wall. Six
overfire air ports are located on the furnace rear wall directly opposed to the six upper elevation
burners. These ports were designed to provide secondary air for sufficient mixing to complete
combustion in the upper furnace and help prevent flame impingement at the upper burner
elevation.
-------
Associated Equipment Retrofits
In conjunction with the furnace and burner retrofit, the pulverized fuel delivery system was
redesigned and upgraded. The coal piping from the mill exhausters to the burners was replaced
and rerouted. Ceramic-lined distribution bottles and adjustable orifices were installed to balance
the coal flow equally to each burner. Union Electric Company undertook the responsibility for
completely overhauling the three ball tube mills and replacing the six mill exhausters with larger
capacity exhausters. The controls for Meramec Unit #4 were also upgraded during the boiler
retrofit. A Westinghouse WDPF control system, including a burner management system
compliant with NFPA standards, was installed.
Retrofit work was completed in late June, 1996.
Design Challenges
From the onset of the burner retrofit project, there were several design challenges that required
particular attention. Among the concerns were installation of burners that would produce
extended flame lengths in our relatively shallow depth furnace (28'-6") and the ability of the mills
to produce the required coal fineness. Additionally and of utmost concern was the ability to
achieve proper fuel distribution to the low NOX burners given our double-ended ball tube mill
arrangement. Substantial fuel output variations can occur from mill end-to-mill end with these
types of pulverizers. The basic problem lies in the coal feed rate to the mills. Coal feed is not a
mill output control parameter, but is used instead to control the level of coal in the mill.
Consequently, there is no direct provision to balance coal flow output from each end of the mill.
Accordingly, Union Electric Company assumed responsibility for proper distribution of fuel into
and out of the mills.
Performance Objectives
Union Electric and B&W negotiated performance objectives for the low NOX combustion system
that dovetailed with the performance objectives secured for the boiler retrofit contract.
Combustion system objectives for simultaneous achievement of NOX, carbon monoxide (CO),
excess oxygen (02), and unbumed carbon (UBC) levels were obtained as shown and discussed
below. Objectives were developed for firing either of two similar Illinois bituminous coals or a
representative blend. The contract coals are from the Kerr McGee #5 and Rend Lake mines.
Table #1 in the Appendix lists proximate and ultimate analysis parameters for these two fuels.
-------
MERAMEC #4 BOILER & COMBUSTION SYSTEM PERFORMANCE OBJECTIVES
OBJECTIVE
#1
#2
#3
BOILER PERFORMANCE
MATNST
FLOV
(Ib/hr)
2,650,000
1,520,000
EAM
V
(%)
100%
57%
662,500 - 2,650,000
COMBUSTION SYSTEM
PERFORMANCE
NOx
(Ib/MMBtu)
<0.45
<0.45
<0.50
CO
(PPM)
175/120
175/120
-
02
(+/- .25%)
(Wet Basis)
<3.3%
<4.85%
UBC
(%)
See
Figure
#2
NOTES:
1. Steam flows are allowed to vary +20,000 to -80,000 Ib/hr.
2. CO performance objectives are based on a pipe-to-pipe coal flow balance of +/- 10% or +/-
5%.
3. Performance objective #3 is based on any combination of mills and exhausters in
service.
4. UBC performance objectives depend on coal fired, coal pipe balance, and mill
fineness.
NOX Performance Objectives
In general, the combustion system was supplied to reduce NOX (as measured by EPA Reference
Method 7E) below 0.45 #/MMBtu at 100% boiler main steam flow production. This performance
objective is based on all 18 burners being in-service. Additionally, with any combination of mill
exhausters and burners in-service, NOX is predicted not to exceed 0.50 #/MMBtu from 25% to
100% rated main steam flow production.
Carbon Monoxide (CO) Performance Objectives
CO objectives are contingent upon coal pipe-to-pipe balance. With a coal pipe-to-pipe balance of
+/- 10%, CO production (as measured by EPA Reference Method 10) is predicted not to exceed
175 ppm on a dry volume basis corrected to 3% O2. With a pipe-to-pipe balance of+/- 5%, CO
production is predicted not to exceed 120 ppm on a dry volume basis corrected to 3% O2.
Excess Oxygen (OJ Performance Objectives
Excess Oa concentration in the flue gas at the economizer outlet (as measured by EPA Reference
Method 3A) is predicted to range between 3.3% to 4.85% (+/- .25%) on a wet volume basis at
the 100% to 57% main steam flow conditions, respectively. Additionally, minimum excess O2 is
predicted to be greater than or equal to 3% on a wet volume basis at all conditions.
-------
Unbumed Carbon (UBC) Performance Objectives
Recognizing that average UBC content in the fly ash could vary with mill fineness, ash content of
the contract coals, and coal pipe-to-pipe balance, a set of curves was developed to correlate UBC
objectives with these variables. UBC predictions range from 3.7% to 22.3%, depending upon the
set of variables that apply. Figure #3 exhibits the UBC performance objectives curves that were
developed for the combustion system contract.
Boiler and Combustion System Tuning
Boiler and combustion system tuning commenced in late August, 1996. Initial tuning efforts have
been interrupted and extended in part due to malfunctions of non-B&W supplied equipment, unit
dispatch obligations, a six week test burn of non-contract coal, and additional needed equipment
modifications uncovered during initial tuning activities. Tuning work continues with concentration
on producing a satisfactory balance between boiler O? and CO production. Specific initial tuning
results and problems that have been encountered are as follows:
NOX Tuning Results
NOX emissions from the low NOX combustion system have been well within the performance
targets. As indicated in Figure #3 in the Appendix, emissions during testing have ranged from as
low as 0.27 #/MMBtu to a value generally less than 0.40 #/MMBtu. These values have been
achieved despite operating with economizer exit 62 values much higher than desired or predicted
as discussed below. Therefore, we are optimistic about maintaining NOX emission performance,
when the CO and Oa targets are eventually achieved.
CO and O2 Tuning Results
Attaining a satisfactory balance between CO production and boiler excess O2 has been a challenge
ever since the modified boiler and low NOX combustion system went into service. Several fuel
delivery system components appear to be contributors to this problem. Specifics of the problems
encountered are as follows:
Initial recorded CO levels exceeded 500 ppm (the calibration range of the CO monitor) unless the
boiler O2 was raised to > 5%. Additionally, the CO levels changed erratically from side-to-side in
the furnace and from day-to-day without any apparent reason. Initial efforts concentrated on
verifying the accuracy of the O2 probes and the CO monitor and inspecting the boiler for sources
of air in-leakage. Some minor instrument problems were corrected but no significant air in-
leakage was found. CO levels remained unacceptably high. Adjustments were made to burner
inner and outer secondary air vanes, but no positive effect on the CO levels was apparent. The
possibility that one or two individual burners might need some fine tuning was then investigated.
However, restaging the secondary air about the furnace from burner-to-bumer, level-to-level, etc.
also had little positive effect. Eventually, through trending of various operating parameters on the
-------
distributed control system, it was discovered that CO spikes occurred in unison with the two
speed mill feeders changing speeds. As stated earlier, the end-to-end balance of coal through the
ball mills had been an initial system design concern. However its sensitivity to mill feeder speed
changes and its effect on CO production was not readily apparent until tuning efforts commenced.
Through the ingenuity of the Westinghouse DCS service engineer, we were able to program the
operation of the two speed feeders to simulate variable speed-like operation in an attempt to level
out the feed of coal to the mills. Substantial improvements in the CO levels were obtained.
Concern over feeder motor longevity led us to eventually embark on a program to retrofit the mill
feeders with variable speed drives and new motors.
Although the mill feeder modifications resulted in substantial CO emission improvements, the
levels still either regularly exceed the 175 ppm target or the boiler O2 has to be raised to 5+%.
Figure #4 in the Appendix displays recent test data relating CO production and excess O2.
Qualitatively, the data suggests that substantial amounts of CO are being generated from all
burner levels. However, the middle level of burners (the ones with impellers) and to a lesser
extent, the lower level burners appear to contribute substantially more to high CO production
levels.
High CO production from the middle level of burners (ones with impellers) exemplifies the
complexity thus far trying to tune the combustion system. B&W generally supplies impellers to
reduce CO production at some expense of producing NOX. Accordingly, high CO production
from the middle level of burners was contrary to everyone's expectations and B&W's previous
experience. The impellers were initially set during unit start-up at a position 4 inches retracted
from the edge of the burner nozzle. During recent tuning efforts, the position of the impellers was
varied to determine its effect on CO and NOX production. The results as shown on Figure #5 in
the Appendix indicates that CO production dramatically increased when the impellers were either
retracted further into the burner nozzles or when pushed flush with the end of the burner nozzles.
Accordingly, the initial setting appears to be the optimal position for minimizing CO production
with these particular impellers for our application. Therefore, other factors, such as primary air-
to-fuel ratio and burner level-to-burner level coal flow imbalances, may warrant additional
investigation.
Field observations during the tuning effort have indicated that the mill serving the bottom row of
burners is supplying substantially more fuel to the burners than the other two mills. Measures to
reduce the output from this mill in an effort to improve the burner level-to-burner level imbalance
would reduce unit load capability since the exhausters on the other two mills are at capacity at full
unit load. Thus far, to compensate for these fuel imbalances we have increased the distribution of
excess air to the bottom row of burners.
The primary result of operating the boiler with excess 02 at 5+% in order to minimize CO
emissions to tolerable levels, is an undesirable boiler heat rate penalty. In addition, there is a
genuine concern that the unit will have insufficient forced draft capacity to achieve rated load
during the summer months at this O2 level.
-------
Unbumed Carbon
Limited UBC data has been obtained thus far during the tuning effort due to having to operate the
unit with high boiler excess 62. However, the most recent fly ash samples that have been
analyzed indicate that UBC levels are running at 7% to 9% on essentially Kerr McGee coal. Mill
fineness data corresponding to these UBC levels are averaging 98.72% through the 50 mesh sieve
and 73.7% through the 200 mesh sieve. Accordingly, the mills are very close to demonstrating
the minimum fineness necessary to support UBC performance objectives. Additional UBC data
will be collected once excess 62 and CO has been reduced to acceptable levels.
Boiler/Unit Output
Another initial fuel delivery related incident encountered was that the unit was unable to attain
more than about 90% of rated output. Field observations of mill amperage led us to suspect that
the mills were under performing. It quickly became apparent that more attention would have to be
paid than in the past to maintaining proper mill coal level and ball charge if the unit was to obtain
rated capacity as well as optimum emission performance. Accordingly, we embarked on a
program to optimize the performance of the newly refurbished mills. After some mill optimization
adjustments, the unit was able to achieve full load without any support fuel.
Future Tuning Activities
B&W has been cooperative and actively involved with us trying to tune the low NOX combustion
system to meet the project objectives. Future tuning activities will likely be directed at replacing
the impellers on the middle row of burners with a different design and possibly retrofitting
impellers to the burners on one or more additional elevations. Other fuel delivery system related
factors for which Union Electric has primary responsibility, such as primary air-to-fuel ratio and
end-to-end ball tube mill output, may also warrant additional investigation and corrective action.
Acknowledgments
The author gratefully acknowledges Mr. Kevin Kersting (Power Plant Maintenance &
Engineering), the Meramec Plant operating staff, and Mr. Bill Mariner (B&W Service) for their
cooperation and determined efforts in pursuing the Meramec Unit #4 combustion system tuning
objectives.
-------
REFERENCES
S.A. Bryk, E.G. Lansing, M.P. Meyer, J.C. Morgan, and D.A. Sampson, "Total Furnace
Upgrade of Aging Boiler Significantly Increases Reliability, Reduces NOX Emissions, and
Improves Cycling Operation," Technical Paper BR-1628, presented at the Power-Gen
International '96 Conference, Orlando, Florida (December 1996).
-------
Table 1
ULTIMATE ANALYSIS OF CONTRACT COALS
Constituent
Moisture
Carbon
Hydrogen
Nitrogen
Chlorine
Sulfur
Ash
Oxygen
KERR McGEE
12.19
67.07
4.37
1.36
0.34
1.27
6.3
7.1
REND LAKE
12.0
62.7
4.2
1.4
0.4
1.25
12.0
6.3
PARTIAL PROXIMATE ANALYSIS
Fixed Carbon
Volatiles
49.58
31.93
45
31
-------
10
Figure 1
Meramec Unit #4 Boiler Cross-Section
-------
11
COAL FIRED
0% Kerr McGee
100% Kerr McGee
0% Kerr McGee
100% Kerr McGee
100% Rend Lake
0% Rend Lake
100% Rend Lake
0% Rend Lake
MILL
FINENESS
(%)
70/99.0
70/99.0
80/99. 9
80/99.9
70/99 . 0
70/99.0
80/99.9
80/99.9
UBC PERFORMANCE OBJECTIVES
PIPE-TO-PIPE FUEL BALANCE
+ /- 5%
10 .6
18.8
3 . 7
6.8
10.6
18 . 8
3.7
6.8
+/- 10%
12
22.3
5 .0
9.3
12
22.3
5.0
9.3
S.
- 10--
o
m
o:
25 50 75
KERR McGEE 0-100% / REND LAKE 100-0%
Figure 2
Meramec Unit #4
B&W Carbon In Ash Performance Objectives
100
-------
12
MERAMECUNIT#4
NOx Versus O2
- 0.5 Si.
A.B.C335MW A.B.C235MW A&B 238MW A&B 238MW A&C218MW B&C213MW
MILLS IN-SEFWCE& UNIT LOAD (MW)
1 02(Left) a O2(Right) «. O2 (Predicted) _ NOx (Predicted) g NOx
Figure #3
NOX Versus O2 Tuning Results
-------
13
MERAMECUN1T#4
COMBUSTION SYSTEM TUNING DATA
800
600
400
200
A,B,C335MW A.B.C235MW A&B 238MW A&B 238MW A&C218MW
MILLS IN-SERVICE & UNIT LOAD (MW)
B&C213MW
, O2(Left) 4.02(Right) + CO(Rjght) B CO(Left) ;
Figure #4
CO Versus 02 Tuning Results
-------
14
MERAMECUNITM
CO & NOx VS. IMPELLER POSITION I
1200
1000 -
-8". 5.8%O2
_ 2
•4". 5.8%O2
IMPELLER POSmON AND % O2
0", 5.5%O2
HCO(Left) B CO(Right) Q NOx
Figure #5
CO & NOX Versus Impeller Position
-------
FUEL SYSTEM MODIFICATIONS AND BOILER TUNING TO ACHIEVE
EARLY ELECTION NOX COMPLIANCE ON A 372-MWe COAL-FIRED
TANGENTIAL BOILER
R. Himes
Carnot Technical Services
15991 Red Hill Avenue, Suite 110
Tustin, California 92780-7388
M. Scharnott
R. Hoyum
Minnesota Power
P.O. Box 128
Cohasset, Minnesota 55721
Abstract
Minnesota Power's (MP) coal-fired generation system is comprised of Title IV-Phase II boilers.
Although Phase II boilers are not required to meet emission limits until January 1,2000, the
standards will be more stringent than Phase INOX limits. MP investigated the merits of
foregoing system averaging under a Phase II rule, and pursued an early election option with
current Phase INOX limits.
Camot conducted combustion performance and NOX emissions characterization for MP's
Boswell Unit 3. This engineering diagnostic program included pre- and post-retrofit fuel system
performance testing.
This paper presents the results of applying commercially available fuel system modifications to
successfully achieve NOX compliance on a 372-MWe, coal-fired tangential utility boiler. The
design focused on modifications to the fuel delivery system so that boiler tuning and
optimization could be implemented to achieve compliance with EPA's Phase 1 Early Election
NOX emission standard of 0.45 Ib/MMBtu. Compliance was achieved without installing new
low-NOx burners or overfire air. This retrofit Included an evaluation of pulverizer modifications,
fuel balancing options, and boiler instrumentation and control (I&C) modifications. Specific
pre-retrofit and post-retrofit performance data for the boiler in terms of NOX emissions, slagging
impacts, and efficiency are presented. Pre- and post-retrofit data are also included for pulverizer
A225B290.DOC
-------
performance as a result of installing high-efficiency exhauster wheels and other pulverizer
improvements.
Introduction
Camot was contracted by Minnesota Power Corporation (MP) to assist in the development of a
cost-efficient strategy to ensure compliance with the NOX emissions limits mandated by Title IV
of the 1990 Clean Air Act Amendments (CAAA). The first step in the development of a NOX
Compliance Strategy is to perform baseline evaluations of current boiler performance and NOX
emissions. This paper describes the results of the field evaluations of MP Boswell Unit 3. The
primary test objectives to evaluate existing boiler operations were as follows:
• Quantify and verify the baseline NOX emissions for each boiler.
• Determine the achievable NOX reductions available with existing plant equipment, with a
focus on operating modes which were previously not tested.
• Identify existing hardware which limits further NOX reductions, and identify what the
limiting factors of this equipment are. In particular, focus on mill performance and fuel/air
distribution.
In order to accomplish these objectives, solutions were sought which minimized not only the
capital costs, but also the Operating and Maintenance (O&M) costs. Moreover, combustion
retrofit options were considered which could provide incremental improvements in efficiency or
reduce other O&M costs to offset the cost of compliance.
Development of this low cost NOX reduction compliance strategy began in 1994 and was
completed with the post-retrofit performance evaluation in 1996.
Boswell Unit 3
Boswell Unit 3 is a Combustion Engineering tangential boiler which currently fires sub-
bituminous coal (nominally 1% sulfur). This unit began operation in 1973. Full load rating is
2,472,000 Ib/hr steam evaporation and 365 gross MW generation. Design superheat outlet steam
is 2,619 psi/1005°F and reheat outlet steam is 548 psi/1005°F. The furnace cross sectional
dimensions are 45'-4" wide by 40'-5" deep corresponding to a current full load NHI/PA of 1.9 x
106Btu/hr-ft2.
Steam temperatures are controlled primarily with burner tilt. The unit is balanced draft and
equipped with a back end scrubber for particulate control.
This unit is equipped with five CE-Raymond bowl mills model 863 RS with mill A as the
uppermost and mill E as the lowermost elevation. The boiler burns a PRB coal that averages
10% ash, 25% moisture, 30% volatiles, and HHV of 8,650 Btu/lbM.
A225B290.DOC
-------
Combustion Performance and NOX Emissions Characterization Boswell 3
Test Variables
Baseline (as-found operation) testing was conducted from approximately 45% load to full load in
order to determine the operating adjustments and equipment modifications necessary to reduce
NOX emissions. All tests were conducted at steady state operating conditions. The goal of the
program was to identify the effects of the variables on reducing NOX emissions, such as load,
excess O2, BOOS firing, and mill performance. Another of the underlying project objectives was
to test some different operating modes previously untested specifically for reduced NOX, such as
fuel and air biasing.
Specific items that were of concern during the testing and that were closely monitored by Carnot
included steam temperatures, flame characteristics, combustibles, slagging and fouling, and
boiler efficiency.
Data Collection
On site, Carnot was responsible for selecting operating conditions, within limits acceptable to the
Control Operator, and reviewing data from previous tests to determine subsequent test
conditions. The data collection process is described below.
Operating Data. Operation was closely observed and documented during the test program.
Gaseous Emissions Data. A gaseous emissions sampling matrix was installed by Carnot in
the air preheater inlet ducts. This matrix consisted of stainless steel probes fitted with in-duct
sintered metal filters. A 16 probe matrix was installed on Unit 3. The CEMS was used to
monitor NOX, O2, CO2 and CO levels during all tests.
Loss-on-lgnition. Carnot collected flyash samples from the inlet breaching to the scrubber for
each unit.
Primary Air Velocities. Carnot measured the primary air velocities in the coal pipes using a
"dirty air" pitot probe.
Coal Fineness. MP personnel collected coal samples for coal fineness determinations during
the same test conditions that Carnot was measuring primary air velocities.
Raw Coal Sampling and Analyses. Daily raw coal samples were taken from the inlets to
the feeders of the mills that were on-line. The samples were sent to Commercial Testing &
Engineering located in Lombard, Illinois for the folio whig analyses:
• proximate analysis (total moisture, volatile matter, fixed carbon, ash)
• ultimate analysis (sulfur, carbon, hydrogen, nitrogen, oxygen)
• higher heating value
• hardgrove grindability index (HGI)
A225B290.DOC
-------
Additionally, one sample from Unit 3 was analyzed further for ash fusion and mineral content.
These data were used to evaluate slagging and fouling potential.
Baseline NOX Emissions
Baseline NOX emissions are plotted in Figure 1 for Unit 3. Baseline NOX emissions were
measured across the full range gross loads of 184-366 MW on Unit 3.
NOX emissions were nearly constant from minimum to 90% loads at 0.48 Ib/MMBtu on Unit 3.
Full load peak NOX emissions were 0.56 Ib/MMBtu.
Excess Air Effect on NOX
The effect of varying boiler excess O2 was tested on Unit 3 as graphed in Figure 2. Three
different 02 levels were tested at full load resulting in an emissions sensitivity of 0.04 Ib/MMBtu
NOX per 1% O2 variation. At 90% load and under staged combustion conditions, the sensitivity
was shown to be 0.09 Ib/MMBtu per 1% O2. These results were applied to other test staged
combustion test results in order to normalize emissions to a constant excess O2 for purpose of
comparison. In instances where CO levels were above acceptable limits, the NOX/O2 sensitivity
curve was also used to normalize NOX emissions to a higher O2 where CO emissions could be
anticipated to be within acceptable limits.
The tendency for reduced steam temperatures with lower excess air could be countered with
minor tilt adjustments.
Staged Combustion NOX Reduction
Bulk staging of combustion for reduced NOX operation included the following:
• Burners Out of Service
• Fuel Biasing
• Air Biasing
Burners Out of Service (BOOS). The effectiveness of BOOS was tested on Unit 3 by
opening the A fuel air and/or the AA auxiliary air dampers with the A mill removed from
service. Table 1 summarizes the test results.
Reductions in NOX emissions ranged from 15% to 30% due to the effect of BOOS. These
comparison tests were selected as having similar operating conditions with the exception of the
staged combustion operation. The emissions sensitivity to excess O2 was used to correct
measured emissions to a different O2 level than tested. This was done for Test 3 because
measured CO was 425 ppm, thus an O2 correction of 0.5% was added and the adjusted NOX of
0.33 Ib/MMBtu was used to estimate the achievable reduction.
Fuel Biasing. The impact of biased fuel flow on NOX was tested, consistent with the program
objective of investigating previously untested operating modes. Fuel biasing was implemented
A225B290.DOC
-------
by adjusting the feeder speed hand/auto controller bias. A summary of fuel biasing tests is
contained in Table 2.
Fuel biasing tests on Unit 3 resulted in 11% NOX reduction by itself, and 28% reduction with the
addition of full OF A. With these tests the lower furnace was more fuel rich than the upper;
however, there were no significant changes on boiler operation including steam temperatures.
Burner Tilt Effect on NO,
X
The effect of burner tilt on NOX emissions was tested on Unit 3 as shown in Table 3. Tests 5 at
+13° and 6 at -24° tilts were conducted sequentially on November 4, while Test 11 at 0° tilt was
conducted November 5. All tests, when normalized to a constant excess O2, showed
approximately the same NOX emissions of 0.52 Ib/MMBtu.
Combustion Air Distribution
Unit 3 exhibited O2 maldistribution, with the south side frequently having a 1-2% higher 02
level. Figures 3 and 4 show atypical 02 and NOX contour plots for Unit 3. This maldistribution
was consistent throughout the Unit 3 load range under both baseline and low NOX operating
conditions. The plant probes also indicated the O2 imbalance.
Flyash LOI
All flyash loss-on-ignition (LOI) tests on Boiler 3 resulted in less than 0.4% weight loss after
bum-off of the samples. This was true even for tests with combustion imbalances such the low
O2 with A level BOOS which had average CO levels of 2,950 ppm and only 0.37% LOI.
Slagging and Fouling Properties of Ash
The slagging and fouling properties of the ash are summarized in Table 4. The ash type for all
coals is lignitic with an iron oxide to calcium and magnesium oxide ratio of less than one.
The coal samples from Unit 3 show low slagging potential and tend toward weak slag adherence
to tube metal.
Fouling potential is indicated by the percentage of Na02 ash content for lignitic coal ash. As
shown in Table 4, the coal sampled from Unit 3 indicate very low potential with 0.4% or less
NaO2.
Impacts on Boiler Efficiency
Impacts on boiler efficiency are quantified by various parameters which most affect performance
as shown in Table 5. Included are steam temperatures, attemperation flows, CO emissions, and
air preheater exit temperature. Baseline tests are compared with the minimum NOX emission
tests for each load.
A225B290.DOC
-------
Unit 3 Pulverizer Performance. The coal pulverizer performance for Unit 3 mills C and D
are illustrated in Figures 5 and 6 using Rosin and Rammler plots. All tests were conducted with
a classifier vane setting of 2.5-3.5 and air inlet temperature of approximately 700°F. There were
two conditions which could be limiting the mill capacity:
1. The average as-received coal moisture content of 25% requires either higher air temperature
or increased air flow to maintain output above the dew point temperature.
2. The nameplate mill capacity of 106,000 Ib/hr is for 70% through 200 mesh with 55 HGI coal.
Actual conditions were 70-75% through 200 mesh and 48-51 HGI.
Tests were performed over the range of 42,000-96,000 Ib/hr (39-91% of nameplate capacity).
The design air flow for these mills is 159,000 Ib/hr (2650 Ib/min from curve) from 50%-100%
mill capacity, compared to 152,000-158,000 Ib/hr primary air measured on the high mill loading
tests.
Test 19 on mill D had a relatively low mill loading with coal flow at 42,000 Ib/hr. The measured
primary air flow was 183,000 Ib/hr, indicating an increase from the high mill loading operation.
By design, the primary airflow control should begin ramping down at 50% mill capacity.
Figure 6 shows the coal fineness improved substantially on this test as a result of the higher
air/coal ratio and low loading on the grinding elements.
Figures 7 and 8 show bar graphs of the pipe-by-pipe air flows measured for mills C and D. A
convenient measure of air and/or fuel distribution is the standard deviation divided by the mean
flow (also known as "coefficient of variation"). This value ranged from 2.3% to 9.8% for each
mill for the various tests. A coefficient of variation of 5% is generally achievable. Note that coal
flow per pipe was not measured, and even if air flows are balanced the air/coal ratio could still
vary significantly. The intent of this testing was to get a "first level" indication of the
performance of two representative mills.
Pre-Retrofit Fuel System Performance Tests
The first task of the NOX implementation program for Minnesota Power's Boswell Unit 3 was to
perform pre-retrofit mill characterization tests.
The test objectives included the following:
• Evaluate the coal fineness from each mill (A, B, C, D, E);
• Measure the coal flow distribution through each coal pipe;
• Measure the velocities through each coal pipe to assess primary air flow distribution; and
• Quantify mill power requirements.
These tests involved primary air velocity measurements in the coal pipes and pulverized coal
sampling for fineness analysis. The coal sampling for the mill characterization tests utilized a
A225B290.DOC
-------
RotorProbe system that collects coal samples at an isokinetic rate to ensure that a representative
coal sample is collected.
The results of the fuel system performance tests are summarized in Table 6 for each mill. The
key results derived from the pulverizer tests are:
• Coal Grinding Capacity Limitations. Based upon the design capacity of the RS 863 mill,
the low coal grindability (HGI) measured, and the spillage observed, it is concluded that the
pulverizers are limited in capacity due to reduced coal grindability. The original design of
the RS 863 pulverizer was to pulverize 106,000 Ib/hr of 55 HGI coal as shown in Figure 9.
For reference, the required coal flows for full load operation with four and five mills are
shown on the figure. Actual HGI measurements ranged from 45 to 54. (Per the contract coal
specifications, HGI could be as low as 42). Independent of any air flow limitations, the mill
capacity would be de-rated to 91.7 klb/hr with 45 HGI coal.
• Primary Air Flow Capacity Limitations. Based upon the high fuel moisture content and
the inability to consistently maintain outlet temperatures over 130°F, it is concluded that the
pulverizers are limited in capacity due to inadequate air supply to dry the coal. Other
measures to increase exhauster fan head capability or reduce backpressure on the exhauster
will be required. ABB-CE's high efficiency exhauster wheel (HEEW) should be consisted.
• Coal Fineness. Acceptable coal fineness for these mills is 70% through 200 mesh, 90%
through 100 mesh, and less than 1% on 50 mesh. Mills A, C, D, And E demonstrated
acceptable coal fineness. Mill B fineness was slightly short of these guidelines.
• Primary Air Flow Balance. With the existing equipment features, the primary air flow
balance should have a variance coefficient of 5% or less. Only Mills A and E demonstrated
acceptable primary air balance near 5% variation. Mills B, C, and D have less than
satisfactory balance ranging from 12% to 31% variation.
• Coal Flow Balance. With the existing equipment features, the coal flow balance should
have a variance coefficient of 10% or less. All mills have less than satisfactory coal balance,
with measured variation coefficients of 12% to 32%.
Post-Retrofit Fuel System Performance Test
Performance tests of mills C and D on Boswell Unit 3 were conducted after each of these mills
were retrofit with ABB-CE's high efficiency exhauster wheel (HEEW).
The tests performed on 3C and 3D mills were conducted under good operating conditions. The
coal grindability met or exceeded the specification requirements of 48 HGI. The coal total
moisture content was above the design specification value of 25%. Both mills had been
overhauled within the previous weeks, and the 3C hot air damper was repaired. Hot primary air
(air preheater exit) temperatures ranged from 616-681°F, compared to 650°F design.
A225B290.DOC
-------
The test results for 3C mill at 71% load are as follows:
• coal properties: 49 HGI and 26.4% total moisture
• feeder speed: 71%
• coal flow: 89,713-98,636 Ib/hr (see Discussion of Results below)
• % on 50 mesh fineness: 2.6
• % through 100 mesh fineness: 84.0
• % through 200 mesh fineness: 57.7
• fuel balance standard deviation/mean: 26%
• primary air balance standard deviation/mean: 12%
• mill outlet temperature: 138°F
• mill motor amps: 104
The test results for 3D mill at 75% load are as follows:
• coal properties: 48 HGI and 25.9% total moisture
• feeder speed: 75%
• coal flow: 94,767-104,193 Ib/hr
• % on 50 mesh fineness: 1.3
• % through 100 mesh fineness: 92.3
• % through 200 mesh fineness: 76.7
• fuel balance standard deviation/mean: 12%
• primary air balance standard deviation/mean: 2%
• mill outlet temperature: 140°F
• mill motor amps: 114
The test results for 3D mill at 80% are as follows:
• coal properties: 49 HGI and 27.3% total moisture
• feeder speed: 80%
• coal flow: 101,085-111,139 Ib/hr (see Discussion of Results below)
• % on 50 mesh fineness: 4.6
• % through 100 mesh fineness: 84.9
• % through 200 mesh fineness: 65.5
• fuel balance standard deviation/mean: 17%
• primary air balance standard deviation/mean: 2%
• mill outlet temperature: 142°F
• mill motor amps: 114
Discussion of Results
The primary purpose of the HEEW retrofit was to increase the capacity of the mills. In order to
evaluate capacity, the feeder calibration needs to be known. With the DCS upgrade on Unit 3,
the feeder calibration was changed from 0-1,567 RPM (as specified by Stock originally) to 0-
1,800 RPM. This means that the current feeder speed readings cannot be directly compared to
A225B290.DOC
-------
the pre-DCS feeder speed readings. Since the feeders were calibrated per Stock procedures, it is
reasonable to apply the Stock data sheet calibration factor of 70.198 Ib/hr of coal per RPM.
However, the baseline test data showed a calibration factor of 77.18 Ib/hr of coal per RPM. The
range of coal flows reported above reflects these two methods. It should be noted that the
indicated Ib/hr and TPH of coal flow currently in the DCS underestimate the coal flow relative to
either of these calibration factors.
Fuel and primary air balance exceeded the project targets of 10% and 5%, respectively.
However, these parameters were not guaranteed with the HEEW installation since the HEEW
does not inherently provide any means of controlling fuel line balance.
Coal fineness did not achieve the guaranteed performance of 1% on 50 mesh for any of the tests.
Conclusions
Unit 3 NOX Characteristics
• Baseline NOX emissions for Unit 3 were 0.56 Ib/MMBtu at full load, and 0.48 Ib/MMBtu
from minimum load to 90% load.
• Burners out of service (BOOS) reduced NOX emissions 16-30% at 90% load. BOOS is not
possible at full load due to pulverizer capacity limitations. With adequate mill capacity
similar reductions could be expected.
• The sensitivity of Unit 3 to boiler excess air ranges from 0.04-0.09 Ib/MMBtu per 1% change
inO2.
• Biasing more fuel to the lower elevations was effective in reducing NOX emissions by 11 -
28% at minimum load. Air biasing and burner tilts did not reduce NOX emissions under the
conditions tested.
• The coal fired in Unit 3 has a low potential for slagging and fouling, and future operations
with staged combustion (reducing environments) should not significantly increase this
potential.
• Fly ash loss on ignition (LOT) levels were less than 0.4% under all test conditions, including
heavy staging of fuel and air. Flyash LOI should not be a critical limiting factor to
implementing combustion NOX controls with the characteristics of the coal being utilized on
Unit3.
Unit 3 Fuel/Air Distribution and Pulverizer Performance
• Primary air flow coefficient of variation (standard deviation divided by mean) on mills C and
D ranged from 2.3-9.8% from pipe to pipe. A coefficient of variation of 5% or less is
generally achievable.
A225B290.DOC
-------
• Combustion imbalance was consistently prevalent in the boiler indicated by significant
differences in O2 and NOX emissions in the south and north boiler exit ducts.
• Pulverizer C and D are performing well relative to design. The coal being utilized has higher
moisture and lower grindability than the design properties necessary to achieve 106,000 Ib/hr
nameplate capacity.
• The retrofit of the HEEW clearly made a substantial improvement in mill capacity,
particularly in the drying capability of the mills. Drying capability was the critical limiting
factor prior to the modification because of high fuel moisture content relative to original fuel
specifications. Mill outlet temperature was maintained at approximately 13 5-HOT without
fully opening the hot air damper.
• Mill 3C is not achieving its guaranteed capacity or 50 mesh fineness by any estimates.
• Mill 3D appears to be achieving its guaranteed capacity.
• Mill 3D is not achieving its guaranteed coal fineness.
• The mills are not achieving the required fuel line balance.
• Based upon the increased capacity achieved, it is expected that Boswell 3 will be able to
reliably achieve full load on four mills once all mills are retrofit with the HEEW.
References
1. Coal Fouling and Slagging Parameters, Winegartner, E.G., ed., ASME, 1974. Steam,
Babcock & Wilcox, 40th Edition, 1992.
2. Title IV-Phase IINOX Compliance Strategy for Minnesota Power, Himes, R. and Schamort,
M., Power-Gen '95, Anaheim, California.
Table 1
NOX Reduction Effectiveness of BOOS
MP Boswell Unit 3
Comparison: Baseline, AMIS
A-BOOS, AA dampers open
Comparison: A-MOOS
A-BOOS, AB & A dampen open
Comparison: Af. -MOOS
A-BOOS, AA dampers open
Test
No.
1
3
8
9
17
18
Gross
LoadMW
325
326
337
335
241
239
APHInO2
%dry
4.31
4.84
4.75
4.83
3.51
3.46
Measured NOX
Ib/MMBtu
0.473
0.298
0.467
0.392
0.297
0.251
Excess 62 Normalized NOS Change
NO, Ib/MMBtu
0.33
n/a
n/a
- 30%
- 16%
- 15%
Notes:
(1) "AMIS" means all mills in service.
(2) Test 3 was normalized to an 02 of 5.3% to offset excessive CO measured at 425 ppm.
(3) Neither Test 9 nor Test 18 required normalizing to a different O2 due to acceptable CO levels.
A225B290.DOC 10
-------
Table 2
NOX Reduction Effectiveness of Fuel Biasing
MP Boswell Unit 3
Test Gross Load APHm02 Measured NO, Excess O2 Normalized NO,
No. MW %dry Ib/MMBtu Ib/MMBtu
NO, Change
Comparison: D (lower) bias low
B (upper mill) biased low
Comparison: D (lower) bias low
B bias low + open AA dmprs (BOOS)
19
20
19
21
184
188
184
184
5.82
5.69
5.82
6.39
0.476
0.422
0.476
OJ66
_
n/a
_
0.342
_
- 11%
_
- 28%
Notes: (1) Unit 3 low load Tests 19,20, and 21 had mills A and E removed from service.
Table 3
NOX Reduction Effectiveness of Burner Tilt
MP Boswell Unit 3
Test Gross Load Outlet Temps. Attemp. Sprays APHInO2 Excess 02 Normalized NO., NOX Change
No. MW SH/RH°F SH/RHKMir %dry Measured NOX Ib/MMBtu
Ib/MMBtu
Comparison: 0° Tilt
Up Tilt® +13°
Comparison: 0° Tilt
Down Tilt @ -24°
11
5
11
6
365
366
365
364
1002/1008
997/1015
1002/1008
967/991
18/30
4/0
18/30
4/0
3.32
4.21
3.32
2.97
0.526
0.557
0.526
0.510
_
0.520
_
0.524
_
0%
0%
Table 4
Ash Slagging and Fouling Potential
MPL Boswell Unit 3
Unit
Sample Date
Sample ID/Carnot Test #
Mill Feeder Inlet
Ash Type Determination
Normalized % Acidic
Normalized % Basic
Base to Acid Ratio
Fe203/(CaO+MgO)
Ash Type
Slagging Potential Indicators
T-250 "F
Slagging Potential
Rls = (max HT+4*min IT)/5 °F
Slagging Potential
FT-IT Temp Diff Reducing °F
Oxidizing °F
If <100°F tends toward thin slag buildup and tight slag adherence (undesirable).
If >200°F tends toward thick slag buildup and weak slag adherence (control with sootblowing).
Fouling Potential
Index (NaO2 content)
Type (low/med if Index <3)
(high if Index >3)
3
5-Nov
3B
all
77.1%
22.9%
0.30
0.20
lignitic
2,557
low
2,273
med
150
160
0.40
low
A225B290.DOC
11
-------
Table 5
Efficiency Impact Summary
MP Boswell Unit 3
Test
No.
5
6
1
3
15
18
19
21
Gross
LoadMW
366.3
363.9
324.7
326.3
238.5
239.4
183.6
183.3
Mill Out
of Service
none
none
none
A
E
A3
A,E
A3
Description
Baseline
Minimum NOX
Change
Baseline
Minimum NOX
Change
Baseline
Minimum NC\
Change
Baseline
Minimum NC\
Change
NO,
Ib/mmBtu
0.557
0.510
-8%
0.473
0.298
-37%
0.481
0.251
^»8%
0.476
0.366
-23%
Boiler O2
dry
4.21
2.97
-1.24
4.31
4.84
0.53
4.61
3.46
-1.15
5.82
6.39
0.57
Average
Main
Steam °F
997
967
-30
965
988
23
1,009
995
-14
1,004
1,012
8
Average
Reheat
Steam °F
1,015
991
-24
947
982
35
969
971
2
956
952
-4
SH Spray
Klb/hr
4
4
0
26
25
0
39
28
-11
23
18
-5
RH Spray
KJb/hr
0
0
0
3
0
-3
3
0
-3
0
0
0
COppm
0
0
0
3
425
422
0
60
60
0
0
0
APH Exit
Temp°F
258
264
6
252
238
-14
n/a
n/a
n/a
226
224
_2
Table 6
Fuel System Performance Results Summary
MP Boswell Unit 3
Mill Test
A 3
5
AVE
B 1
2
3
4
AVE
C 1
2
4
AVE
D 1
2
4
AVE
E 1
2
4
AVE
Raw Coal
Feed Ib/hr
81,000
90,000
85,500
88,800
90,000
90,000
97,500
91,575
85,200
87,600
91,500
89,550
75,600
81,600
85,500
83,550
76,800
85,200
85,500
85,350
Primary Air
Flow, Wet Ib/hr
140,175
155,690
147,933
132,635
114,929
118,788
115,889
120,560
137,777
127,586
126,364
130,576
143,657
137,374
128,456
136,496
147,296
130,708
137.512
138.505
Primary Air
Flow, Dry Ib/hr
130,691
142,949
136,820
122,862
104,565
108,424
102,088
109,485
128,402
117,054
112,197
119,218
135,336
127,886
115,003
126,075
138,980
120,655
124,787
128,141
Air/Fuel
Ratio*
1.61
1.56
1.60
1.38
1.16
1.20
1.05
1.20
1.51
1.34
1.23
1.36
1.79
1.57
1.35
1.57
1.81
1.42
1.46
1.56
% thru 200
Sieve
73.1
71.7
72.4
69.6
67.5
695
67.2
68.5
73.0
70.0
762
73.1
70.6
78.0
76.4
75.0
70.9
71.3
73.0
71.7
% thru 1 00
Sieve
92.6
91.4
92.0
90.1
89.0
90.5
89.2
89.7
92.6
88.5
93.8
91.6
91.2
94.1
93.5
92.9
91 4
91.6
91.9
91.6
% on 50
Sieve
0.7
1.3
1 0
1.5
1.6
1.4
1 5
1.5
0.8
0.8
06
0.7
1.0
0.4
0.6
0.7
1.1
0.9
0.9
1.0
"Based on dry air flow, which was calculated.
A225B290.DOC
12
-------
0.55
0.5
0.45
0.4
" 035
1 0.3
O 0.25
Z
0.2
0.15
0 1
0.05
n
—
o
y
/
^
— — Phase ] Liniic
1 1
0 50 100 150 200 250 300 350 400 450 500 550 600
Gicu Lotd, MW
Figure 1
Baseline NOX vs. Load, MP Boswell Unit 3
Ciraot AFII Inlet O2. % dry
Figure 2
NOX vs. Excess 02, MP Boswell Unit 3
s 33 /—iMa.-.--
1 , ' \'' * ^
' '' ' i'''V. s "• -~ !-
o r-r : ~ .'. •• ' . .—; '. ' '' •.''•''.• .^ ^^^
0 S 10 15 20 25 30
Duct Won (ft)
Figure3
Economizer Exit - Excess Oxygen Contours, MP Boswell Unit 3
A225B290.DOC
13
-------
I -
0 5 10 15 20 25 30
Dud WOTi (ft)
Figure 4
Economizer Exit - NO Contours, MP Boswell Unit 3
j «-"
f ""
1 uu
1 JJM
tf
i ««"
c
»«J
(
/
/
Y
/
>
f
t- C-, « «C- *^-<
•** ) •- ' ' o. .JJ bW il
• it n i« in xn a
Figure 5
C Pulverizer Performance, MP Boswell Unit 3
Figure 6
D Pulverizer Performance, MP Boswell Unit 3
A225B290.DOC
14
-------
Figure 7
Primary Air Flow Distribution, MP Boswell Unit 3
Figure 8
Primary Air Flow Distribution, MP Boswell Unit 3
Figure 9
RS863 Pulverizer Capcity vs. HGI, MP Boswell Unit 3
A225B290.DOC
15
-------
Experience with Combustion Tuning and Fuel System Modifications
to InexpensiveiyReduce NOX Emissions
from Eleven Coal-Fired Tangential Boilers
Eric Mazzi, P.E.
Levelton Engineering
#150-12791 Clarke Place
Richmond, B.C. V6V-5H9
and
Sheila Haythomthwaite, P.E.
ADA Technologies
304 Iverness Way South, Suite 365
Englewood, Colorado 80112
Abstract
This paper presents commercially-available, low-cost options for achieving NOX
compliance on pulverized coal-fired tangential boilers which require approximately 5% to
50% reductions. Often reduced NOx and economic benefits (e.g. increased efficiency,
reduced maintenance) can be achieved simultaneously. Actual performance data are
presented based upon direct experience applying combustion tuning and fuel system
modifications on eleven coal-fired, dry-bottom tangential boilers in the U.S. Several low-
cost options are presented, some of which are generally unknown to electric utility
engineers. Frequently combustion tuning efforts fall short of required NOx reduction
goals because it is not known how to inexpensively overcome common limiting factors
(e.g. unburnt carbon, slagging). Important limiting factors are described, including low-
cost methods to overcome these factors.
Introduction
This technical paper focuses on low-cost combustion tuning and fuel system modifications
to reduce NOx emissions and improve performance on dry-bottom, pulverized coal-fired
tangential boilers. This technical results focus on four low-cost NOx reduction techniques
which can be applied to the majority of tangential boilers:
1. Fuel biasing
2. Air Biasing
3. Reduced O2
4. Burners out of service (BOOS)
Although these methods are not ostensibly new and unique, they are often overlooked in
favor of higher cost techniques because common limiting factors are encountered but not
addressed. One purpose of this paper is to highlight these limiting factors and to present
options to overcome them.
-------
Higher cost techniques such as overfire air, low NOx burners, rebura, or post-combustion
controls are not discussed. Furthermore, tuning methods with large efficiency penalties or
potentially high fuel costs are also not included such as air preheat control (e.g. hot air
recirculation or air preheater bypass) or gas co-firing.
Minimizing the primary air-to-coal ratio is a proven technique for reducing NOx
emissions1 by 5-10%, although no data is presented in this paper because most units did
not have adequate control over pulverizer fuel and air flow. Nonetheless it is
recommended that minimized primary air-to-coal ratio be considered a viable, low-cost
NOx reduction technique for those units which have adequate controls and which have an
available margin to reduce coal pipe velocity while maintaining coal particle transport.
Reduced pulverizer air flow can also provide other benefits2 such as improved fineness and
coal balance, reduced wear, reduced proclivity to pulverizer fires, and increased boiler
efficiency. It is emphasized that tangential units often require higher coal pipe velocities
(typically 4,000 fpm or higher) than wall-fired units because of longer pipe runs. Lastly, it
is cautioned that when reduced primary air flow is assessed, individual pipe-to-pipe
balance needs to be verified and maintained, not just average coal pipe velocity for each
pulverizer.
Burner tilt control is frequently touted as a method for reducing NOx emissions on
tangential units as it has been well established that horizontal tilts result in lower NOX.
than upward or downward tilts. Although this has been verified in the experience of the
authors, it is recommended that flexibility to adjust burner tilts be retained in most cases in
order to maintain proper steam temperatures and boiler efficiency. It has been found in
general that overall lower NOx emissions and optimum efficiency can be simultaneously
achieved by tuning with the key methods described above. For example one of the
impacts of BOOS operation is to increase or decrease steam temperatures, depending on
unit-specific factors and load, and design steam temperatures can frequently be maintained
by adjusting burner tilts. All factors being equal, tangential boiler operators should
attempt to maintain horizontal burner tilts for minimum NOX.
Sootblowing and boiler cleanliness is yet another factor which can aid in reducing NOX
emissions. Although this has also been verified in the experience of the authors, it is
recommended in most cases not to constrain sootblowing procedures solely to reduce
NOX emissions. Similar to burner tilt control, flexibility to adjust sootblowing pattern and
frequency is best maintained to respond to changes in fuel properties, unit loading
demands, or to offset performance impacts induced by staged combustion (e.g. to control
furnace slagging or to maintain steam temperatures).
Results of Low-NOx Combustion Tuning on Eleven Boilers
A description of critical characteristics for each of the boilers is shown in Table 1. All
eleven units were dry-bottom, tangential boilers firing a low-sulfur coal (less than 1.4% S).
-------
Table 1
Dry-Bottom, Coal-Fired Tangential Unit Characteristics
Unit
Designation
A
B
C
D
E
F
G
H
I
J
K
Coal Rank and
Characteristics
Bituminous (low S)
Bituminous (low S)
High Volatile Bituminous
(low S)
Bituminous (low S)
High Volatile Bituminous
(low S)
Bituminous (low S)
Bituminous (low S)
Bituminous (low S)
Bituminous (low S)
Subbituminous (low S)
Subbituminous (low S)
Rating
MWe
75
75
75
100
115
125
125
140
230
370
550
Comments
4-corner unit
4-corner unit
4-corner unit
Baseline operation with biased fuel & air
and low O2
4-corner unit
4-corner unit
Baseline has biased fuel & air and low C>2
8-corner unit
8-corner unit
8-comer unit
Equipped with OFA
4-corner unit
Equipped with OFA
4-corner unit
4-corner unit
Equipped with OFA
A summary of the tuning results is included in Table 2. Although many techniques were
tested, only the four principal low-NOx techniques previously introduced are summarized
in this table. As the intent is to provide a results-oriented technical paper, the theory
underlying each technique will not be discussed in detail. Each of the four techniques
presented employ stoichiometry control to reduce NOX formation. Each technique will be
discussed individually highlighting NOx reductions achieved, practical considerations to
implement, and typical limiting factors. Following this discussion, the most common
limiting factors will be addressed.
Fuel Bias ing
As shown in Table 2, fuel biasing resulted in NOX reductions up to 23% and was
successfully applied across the load range. Besides reducing NOx emissions, five of the
eleven boilers tested showed that stack opacity was actually reduced by fuel biasing. This
was possibly due to increased residence time in the furnace and a favorable change in
particle size conducive to improved ESP performance. Fuel biasing is accomplished by
adjusting raw coal feeder speed, usually by biasing down the uppermost pulverizer and
allowing other pulverizers to pick up automatically to maintain load. Typical limiting
factors to implementing fuel biasing include control capability (i.e. feeder controls),
pulverizer capacity, burner tip flame attachment, and small increases in flyash loss on
ignition (LOI) or CO emissions.
-------
Table 2
Summary of Low-NOx Combustion Tuning
Unit
Designation
A
B
C
D
E
F
G
H
I
J
K
Load
100%
70%
50%
100%
70%
50%
100%
100%
80%
60%
100%
100%
70%
50%
100%
70%
50%
100%
70%
100%
100%
90%
50%
100%
Baseline NOX
Emissions
Ib/MMBtu
0.68
0.78
0.79
0.72
0.86
0.64
0.42 (air bias
+ reduced O2)
0.53
0.56
0.83
0.42 (air bias
+ reduced O2)
0.48
0.54
0.54
0.59
0.55
0.53
0.55
0.44
0.60
0.56
0.47
0.48
0.33
Reduced NOX
Ib/MMBtu
(% reduction)
0.58 (15%)
0.43 (44%)
0.66 (16%)
0.59 (18%)
0.37 (57%)
0.45 (30%)
0.42 (0%)
0.45 (15%)
0.36 (36%)
0.42 (49%)
0.40 (5%)
0.41 (15%)
0.40 (26%)
0.48 (15%)
0.51 (14%)
0.43 (22%)
0.47 (11%)
0.36 (35%)
0.39 (11%)
0.51 (18%)
0.53 (5%)
0.30 (36%)
0.37 (23%)
0.30 (9%)
Principal Low-NOx
Technique(s) Applied
fuel bias + reduced O2
BOOS + reduced O2
fuel bias
fuel bias + reduced O2
BOOS + reduced O2
FA/AA Bias + reduced 02
FA/AA Bias + increased O2
reduced 02
BOOS + reduced O2
BOOS + reduced O2
fuel and FA/AA Bias
reduced 02
BOOS + reduced 02
FA/AA bias + reduced O2
fuel bias + reduced O2
BOOS + reduced O2
fuel bias + reduced O2
BOOS + full OFA
BOOS + full OFA
BOOS+ full OFA
FA/AA bias + reduced 02
BOOS + increased O2
fuel bias
fuel bias (unit has OFA)
Limiting Factor(s) to Maintain Low-NOx
or Further Reduce NOX
pulverizer capacity
low windbox pressure
pulverizer capacity
low windbox pressure
boiler efficiency & flyash LOI
pulverizer capacity
boiler efficiency & flyash LOI
opacity
low windbox pressure
pulverizer capacity
low windbox pressure
flyash LOI
low windbox pressure
pulverizer capacity & flyash LOI
pulverizer capacity & slagging
steam temperatures
Target NOX
Ib/MMBtu
(averaging period)*
0.45 (annual)
0.45 (annual)
0.45 (annual)
0.45 (annual)
0.45 (annual)
0.45 (annual)
0.45 (annual)
0.42 (24-hour)
0.42 (24-hour)
0.45 (annual)
0.45 (annual)
* Individual unit compliance emission rate shown; all units could possibly be included in a system average NOx compliance strategy.
-------
Air Biasing
Air Biasing refers to biasing secondary air to the upper burner elevations by manipulating
the auxiliary air (AA) and fuel air (FA) dampers. As shown in Table 2, air biasing
("FA/AA bias") achieved NOX reductions of 5-30% depending on unit characteristics and
load. To implement air biasing, a boiler would ideally have individual control of each
elevation of AA and FA dampers so that lower elevations are biased towards the closed
position and upper elevations are biased more open. The largest NOx reduction effect
results from biasing AA dampers because of the relatively large compartment size, but
biasing FA dampers can provide additional NOx reductions. If the control system has the
capability to control dampers by elevation, which is frequently the case when control
systems are upgraded, then air biasing is easily accomplished. For units without individual
elevation AA controls, partial air biasing can be accomplished by failing open upper
level(s) of AA dampers with the upper elevation burners in service. However, failing open
dampers is recommended only for testing purposes and not for ongoing operation.
Typical limiting factors to implementing air biasing include control system limitations, and
small increases in flyash LOI or CO emissions. Windbox pressure (or windbox-to-furnace
differential) needs to be carefully observed, however it was generally found not to be a
serious limiting factor.
Reduced O2
Reduced O2 is clearly not an original technique for reducing NOx as it has been well
established. However, in some cases surprisingly low O2 levels can be operated in
tangential boilers firing pulverized coal. Boiler O2 as low as 1% (dry, volumetric basis)
can be operated on a long-term basis on some units. In general, reduced boiler O2 was
found to reduce NOX emissions by 0.05 to 0.10 Ib/MMBtu per 1% O2 reduced for the
eleven tangential boilers. Usually reduced O2 is applied in combination with other
methods (e.g. fuel bias plus reduced O2). Typical limiting factors include poor fuel
properties such as high sulfur, high burner zone heat release density, inadequate boiler O2
metering, inadequate combustion air control, and poor fuel and air balance.
Burners Out Of Service (BOOS)
BOOS is accomplished by removing upper burners from service (one pulverizer) and
opening upper level AA (and possible FA) dampers. Dampers can be failed open for
testing purposes, but controls should be automated for ongoing operation. BOOS was
shown to achieve NOX reductions up to 57%, which is the same order of NOX reduction
achievable with higher cost retrofits such as overfire air and low-NOx burners (LNB).
There are two common limiting factors to implementing BOOS: pulverizer capacity at full
load, and low windbox pressure at reduced loads. Other limiting factors may include
those common to all staged combustion NOx controls such as flyash LOI, CO, opacity,
slagging, or fouling. Given the large potential NOx reduction through BOOS, it is
particularly worthwhile evaluate the modifications needed to offset operational impacts.
-------
Applying Combustion Tuning for NOX Compliance
In order to apply combustion tuning for NOX compliance there typically are three steps
involved: conduct baseline tuning, assess NOx strategy, and implement modifications
required to maintain compliance.
1. Conduct Baseline Tuning. The objectives of baseline tuning are to assess NOX
reductions achievable with existing equipment, and to assess limiting factors to
achieving further NOX reductions. Baseline tuning involves detailed testing incuding
many of the following elements:
• measure O2 at economizer outlet by multiple-point sample grid or traverse
• measure O2 at furnace outlet (for balanced draft units only)
• measure NOX and CO with a multiple-point sample grid at economizer outlet
or using stack CEMS
• measurement of flyash LOI at the air preheater or paniculate control device
• measure boiler efficiency and air preheater leakage
» measure furnace exist gas temperature (FEGT)
o detailed recording of control settings and boiler instrumentation readings
<• visual inspection of windbox damper and other critical field device positions
o visual observation flame characteristics
e analysis of raw coal samples collected from a moving stream at the feeders
o isokinetic sampling of coal flow balance and fineness, and dirty air velocity in
coal pipes
• measurement of pulverizer inlet air flow
• measurement of secondary air balance (especially for 8-corner units)
The exact scope of the initial tuning effort is determined based upon the project
objectives, quality of existing plant instrumentation, and availability of previous test
data. Note that not all parameters need to be measured for all tests or at all loads.
With improved tools currently available and experienced personnel, such tuning efforts
can often be completed in 5 days or less for one boiler. Steady load and load ramping
tests may be conducted, depending on the operating profile of the unit. For base-
loaded units, typically only full load tests need to be conducted.
2. Assess NOX Strategy. Assessing NOX strategy may involve a detailed and complex
analysis of technical and non-technical factors, which is too lengthy to be properly
addressed in this paper. For a given unit, this process ultimately results in establishing
a target NOX emission rate across the operating load range and a budget. Key factors
include predicted future operating load profile, predicted generation (i.e. annual
MWH), application of system averaging, regulatory NOX emission rate, and time
averaging requirement ranging from an instantaneous limit to an annual heat input-
weighted average. As a general rule, longer time averaging limits make combustion
tuning more plausible. For example, suppose a unit is tuned to achieve full load NOX
emissions of 0.50 Ib/MMBtu but has a regulatory limit of 0.45 Ib/MMBtu. It may be
able to comply on an annual average basis considering that NOX reductions can be
-------
achieved across the load range. This was the case for several of the eleven boilers
described in this paper where the compliant NOx emission rate could not be achieved
at full load, but NOx could be over-controlled at reduced loads for NOx compliance
calculated on a heat-input weighted basis.
3. Implement Modifications. This is a critical step required to maintaining long-term
compliance. To implement combustion tuning typically involves operating procedure
changes and low-cost modifications designed to overcome critical limiting factors. A
unit-specific package of modifications must be developed to overcome all critical
limiting factors.
The package of modifications necessary to implement combustion tuning will be
different for every unit, and can best be illustrated by example. Suppose a unit can
achieve NOx compliance by operating with air biasing at full load, BOOS at reduced
loads, and reduced Oa at all loads. However testing showed inadequate windbox
damper controls to safely bias secondary air, low windbox pressure at reduced load
when testing with BOOS, and high flyash LOI as a result of reduced O2. In such a
case, a package of modifications may include: modifying the AA damper controls to
enable control adjustment by elevation, modifying windbox dampers to increase
backpressure at low loads, balancing coal pipes to balance fuel and primary air to the
burners, and installing four new boiler 02 probes at the economizer outlet. Typically
these types of modifications would cost less than LNB and, in fact, retrofit of LNB
could require the same set of modifications in addition to the cost of the burners.
Common Limiting Factors and Methods to Overcome
This section includes a description of the most common limiting factors to implementing
combustion tuning for NOx compliance, and descriptions of how to overcome them.
Pulverizer Capacity Limitations
Often the only major obstacle to implementing BOOS at full load is pulverizer capacity.
Pulverizers perform three basic functions: dry, grind, and classify. Capacity may be limited
because of any one of these three interrelated processes.
Pulverizers limited by drying capability will exhibit low discharge temperatures. The options
available to improving capacity can be grouped into three categories as shown below. This
discussion is specific to Raymond Bowl pulverizers, but mostly applies to other pulverizer designs
as well.
1. Optimize Existing Hardware. This involves minimizing backpressure (AP) imposed by
problems such as plugged pipes, plugged riffles, or tight gaps in the pulverizer throat. Suction
pulverizers can have significant amounts of in-leakage which cools the coal/air stream and
overloads the exhauster fan, and in-leakage paths should be repaired. Proper maintenance and
tuning of the air flow controls is also important; this involves the exhauster inlet damper,
-------
barometric damper, and hot air damper on suction pulverizers and the hot and cold air flow
dampers on pressurized pulverizers.
2. Upgrades to Increase Air-Flow Capacity. The most effective option to increase air flow
capacity is to install high efficiency exhauster wheels on suction pulverizers or increased
capacity primary air fans on pressurized. One installation of high efficiency exhauster wheels
was made on RS 863 pulverizers which were temperature limited due to a high moisture coal
of 25% as-fired. Demonstrated increased coal flow capacity was approximately 15% from
91,000 Ib/hr to 105,000 Ib/hr. Improved air flow controls on suction pulverizers will also help
maximize drying capacity by controlling the relative amounts of hot and tempering air. Less
effective upgrades in terms of air flow capacity include reducing backpressure with course-cut
riffles, improved coal pipe orifices, improved static classifiers, retrofit of dynamic classifiers, or
improved vane wheels.
3. Upgrades to Increase Hot Air Temperature. As an alternate to increasing air flow by
increased fan capacity or reduced backpressure, it is technically viable to increase the
temperature of hot air supplied to the pulverizer up to 850 F depending on coal properties.
There are two available methods for achieving this: primary air duct burners and flue gas
recirculation to the primary air. These options can incur higher costs relative to fan or
exhauster wheel upgrades, however added benefits can result due to reduced flow through the
pulverizer as described earlier. Flue gas to primary air would also provide added NOX
reduction due to lower oxygen content in the fuel stream. There are at least two successful
installations of primary air duct burners on Raymond Bowl mills known to the authors.
Pulverizers which are limited by grinding and/or classification also have three categories of
options to increase capacity:
1. Optimize Existing Hardware. This involves inspections and repairs of the pulverizer
which should be conducted prior to investing in upgrades. Plant O&M staff are
generally well familiar with requirements for maintaining mechanical tolerance such as
replacing worn components, proper setting or springs or hydraulics, ring to bowl
clearances, inner cone to inverted cone position, etc.
2. Upgrades to Increase Grinding Power. For pulverizers which are limited by motor
power, the solution is relatively simple which is to rewind or replace the electric
motors. Grinding element upgrades are also available such as increased roll diameters,
ribbed rollers, or bowl extension ring height modifications.
3. Upgrades to Increase Classification Efficiency. Standard equipment Raymond
Bowl pulverizers are reported to have a classification efficiency on the order of 20-
30%, which means that 70-80% of properly sized particles are returned to the grinding
zone. Improving the particle separation efficiency effectively increases the grinding
power of the pulverizer. Improved static classifier designs are available to make
relatively small, but inexpensive, improvements in fineness or capacity. Improved vane
wheels, in addition to reducing backpressure on the exhauster or fan, also are intended
to reduce internal recirculation of coal particles.
For large improvements in classification and capacity, dynamic classifiers are the best
available technology; various designs are reported to have a classification efficiency of
-------
50-70%. One such retrofit on an RPS 703 pulverizer increased capacity by at least
15% while maintaining 200 mesh fineness, improving 100 mesh fineness, and
eliminating the 50 mesh fraction. It should be noted that unlike all other hardware or
control upgrades described in this paper, dynamic classifiers should not be considered
to be "low cost" as their costs ranges in $100,000's for each pulverizer, depending on
size of the unit and design features.
Low Windbox Pressure at Reduced Loads
BOOS is a viable low-NOx operating mode for nearly every tangential boiler which
operates a portion of the time at reduced loads. Often the limiting factor is low windbox
pressure due to reduced restriction in the combustion air path by opening upper level AA
and FA dampers.
One unique solution was applied successfully on an 8-corner coal-fired unit and is
generally applicable to any 4- or 8-corner unit. Typically each windbox damper
compartments has 2 or more damper blades linked together, with one driven blade and one
or more "slave" blades. The modification involved removing the linkage to the slave
blade, and then all slave blades from the AA compartments on each comer were linked
together and tied to a new manual positioner. The same was done for each FA
compartment on each corner. This provided a tunable system to establish proper control
sensitivity for the secondary air and windbox pressure, and hence it enabled BOOS
operation across the load range. An additional benefit was realized because this 8-corner
unit exhibited secondary air imbalances (as is often the case for 8-comer units), and the
new manual positioners were tuned to alleviate this imbalance.
Other solutions to this problem are available such as installing perforated plate in the
windbox or replacing nozzle tips with those having smaller flow area. These options
should be evaluated, however one drawback of both is there would be no flexibility for on-
line adjustment once installed. Perforated plate installations on tangential boilers has also
been shown in some cases to result in increased NOX emissions, possibly due to increased
air through the FA compartment. It is important to note that the damper modification
described above allows for tuning of the relative amounts of AA and FA.
Steam Temperatures
In response to BOOS, fuel biasing, air biasing, or reduced 02 there is usually a change in
the steam temperatures which must be addressed. Fortunately design steam temperatures
can often be maintained on tangential units with only operational changes such as
adjusting burner tilt, changing sootblowing pattern and frequency, or burner pattern
selection. Some coal-fired units also have flue gas recirculation to the furnace bottom
which provides an excellent means of controlling steam temperatures as it was designed to
do. Although generally not preferred due to efficiency impacts, total excess air can be
used as a variable to control steam temperatures by changing the mass flow through the
boiler. One example where this can be applied successfully is operation with BOOS at
-------
low load, where steam temperatures tend to drop below baseline levels. Usually the
amount of excess air can be increased from baseline levels to improve steam temperatures,
while still providing a large decrease in NOX emissions (i.e. the staged combustion NOX
reductions more than offset the effect of increased O2).
Incomplete Combustion
Nearly every low-NOx installation results in undesirable effects due to products of
incomplete combustion. This includes the low-cost tuning methods as well as higher cost
retrofits such as overfire air or LNB. Improved combustion balance is an important
element to offset all of these undesirable effects, and this topic is discussed in the next
section below. The most common effects observed and methods of overcoming them are:
• Flyash LOI, which is the most common measure of unburnt carbon in the flyash, must
be maintained for reasons of efficiency and/or flyash sales. The best solution to
maintaining flyash LOI within acceptable levels is to improve fuel and air balance,
which is discussed in the next section below.
• Slagging and fouling problems can arise and must be addressed. Assuming fuel
properties remain constant, the best approach to controlling slagging and fouling in the
boiler is to improve fuel and air balance, alter sootblowing procedures, and maintain
adequate levels of total excess air. Some units are also constrained in capacity due to
intentional load reductions designed to "de-slag" the boiler.
• CO emissions typically are targeted to be kept below 100 ppm on coal-fired units. It
is important that testing be conducted to measure CO emissions under Sow-NOx
conditions and to ascertain the need for CO monitors. Although CO is often a useful
parameter to monitor, the need can vary drastically on different units depending on
coal properties and combustion conditions. For example, one of the eleven units
tested exhibited CO emissions over 3000 ppm with no measurable increase in flyash
LOI, which indicated that CO monitors would be useful as an operator tool. By
comparison, another unit tested at less than 1% O2 in the upper furnace at full load still
showed less than 25 ppm CO while the flyash LOI levels increased by an order of
magnitude; in this case, CO monitors clearly would not be a useful parameter to
monitor. To offset high CO emissions, is necessary to improve fuel and air balance,
which is discussed in the next section below.
• Stack opacity problems often occur on units with electrostatic precipitators (ESPs)
and may be caused by a number of factors, some of which can be related to low-NOx
techniques described herein. ESP performance can also be adversely affected by
several "non-ideal" effects: rapping, gas flow distribution problems, sneakage (a
portion of the flue gas bypassing the electrified sections through the hopper and high-
voltage feedthroughs), and reentrainment of collected ash into the gas stream. If low-
NOx operation results in increased opacity, it can often be more than offset by
minimizing these "non-ideal" effects on ESP performance by modifying the rapping
schedule, improving gas distribution, or reducing the flue gas flow or temperature. It
is also important to note that reducing flyash LOI can improve ESP performance by
altering the flyash resistivity.
-------
Improving Combustion Balance for Coal-Fired Tangential Units
Improving combustion balance will provide simultaneously reduced NOX emissions and
improved efficiency. For tangential boilers combustion balance can be grouped into three
categories: (1) balanced fuel and air from elevation to elevation, (2) balanced fuel and
primary air from pipe-to-pipe (corner-to-corner), and (3) distribution within the coal pipe
commonly called "coal roping." Additional critical factors discussed below are boiler 62
metering, and procedures to measure and maintain pulverizer performance.
Balanced Fuel and Air by Ele vation
Balanced fuel and air by elevation can be accomplished by having properly calibrated coal
feeders and good control over the air-to-fuel ratio for each pulverizer. Typically
gravimetric feeders are preferred over volumetric, and with either design fuel balance can
be improved via upgraded variable speed drive (VSD) controls. Pulverizer total air flow
control is achieved by metering the inlet air flow to the pulverizer and having good damper
control over the hot and tempering air. Pressurized pulverizers frequently were designed
with these elements and only need to be calibrated and tuned. Suction pulverizers
typically have barometric tempering dampers and no air flow meters; the barometric
damper can be upgraded with a common louver blade design and inlet air flow meters are
available from several manufacturers (pitot, hot wire anemometer, venturi, etc.).
Balanced secondary air from elevation to elevation is important as well. On both 4- and 8-
corner tangential units, the first task necessary to accomplish this is to ensure AA and FA
dampers and drives are well calibrated and maintained. Moreover, it is also highly
recommended that position feedback be installed to provide the operator with an
indication of actual damper positions, or at least to alarm based on position error. Testing
on a large number of tangential units has proven that controller output is insufficient to
assure proper AA and FA damper position, and if position feedback is not available then
each damper drive must be visually inspected on a regular basis.
It should be noted that if coal flow and inlet air flow are metered, then average coal pipe
velocity and dew point temperature can be calculated for on-line indication. These are
two important parameters which would provide useful information to boiler operators.
Coal Pipe and Comer-to-Corner Combustion Balance
Balanced combustion from corner-to-corner is a matter of balancing the secondary air fed
through the windbox, and also balancing the coal pipes (fuel and primary air). Acceptably
balanced secondary air from corner-to-corner can usually be achieved with calibrated AA
and FA dampers and drives as described above, with the exception of 8-corner units. The
secondary air delivery system on 8-corner units is usually designed asymmetrically so that
balanced flow from corner to corner is not assured. On these units, it may be additionally
necessary to measure and/or model air flow and install turning vanes in the ductwork or
windbox. The windbox damper modification described earlier as a method of enabling
reduced load BOOS will also provide corner-to-corner balance control. It is also possible,
-------
albeit expensive, to install on-line flow meters in each AA and FA damper compartment on
tangential units, but if these are installed there needs to be some accompanying method of
controlling the distribution, not just measuring it.
Coal pipe balance is a topic of high interest in the utility industry and there are many
devices available such as adjustable orifices, improved fixed orifices, venturi distributors,
coal balancing valves, and adjustable riffles. Before investing in coal pipe balancing
devices, it is recommended to first ensure that temperature and fineness can be maintained.
Pulverizers which operate at or below the dew point or have inconsistent fineness or
course particles will invariably lead to pipe plugging and malfunction of any devices
installed. Consideration should also be given to the ability to control primary air balance
and coal flow balance separately so that air-to-coal ratio can be maintained in each pipe.
Coal Pipe Distribution (Roping)
Coal roping refers to maldistribution of coal particles over the cross section of individual
coal pipes. It results from having bends in the pipes, however coal pipe bends are an
inevitable condition due to practical design requirements. Roping is exacerbated by poor
coal fineness and operating with wet coal in the pipes (i.e. below the dew point
temperature). Before expending effort to measure and control roping in the coal pipes
directly, it is first recommended to address elevation-to-elevation balance, pipe-to-pipe
balance, fineness, and temperature. Once these parameters are optimized and further
improvements are sought, then coal roping can be measured with SMG-10 or other
advanced tools; however it is recommended that roping be measured after the last pipe
bend (i.e. downstream of the burner elbow). To remedy roping there are devices available
from at least 3 vendors to control roping, but practical considerations need to be
addressed such as piping fit up and available space.
Boiler O2 Monitoring
Accurate boiler C"2 indication should be considered as an essential parameter provided to
boiler operators for reasons of NOx control, energy efficiency, and operating safety. It is
generally advisable that a multiple number of probes be installed at an appropriate location
in the boiler to provide a reliable, representative indication of average O2. A multiple
number of probes will also provide the operator with on-line data to indicate combustion
imbalances (e.g. from side to side). On forced draft units, the economizer exit is often the
best location since air in-leakage is not a problem and the gas temperature is relatively
low. On balanced draft units, air in-leakage needs to be measured (i.e. conduct O2
traverses at the furnace exit and economizer exit as a minimum) and carefully considered
in the selection of the number and location of O2 probes. Generally the nearest practical
location to the furnace exit will provide the most accurate indication, although the higher
temperatures may inevitably lead to higher failure rates for the O2 monitoring system.
-------
Measuring and Maintaining Pulverizer Performance
Because coal pulverizer performance is integral with maintaining NOx emissions and
because they are inherently subject to high rates of wear and performance degradation,
some degree of ongoing performance measurement is required. A general guideline of
minimum requirements for ongoing performance measurement is shown in Table 3.
Table 3
Guideline for Fuel System Measurement Methods and Frequency
Parameter
Fineness
Clean Air
Coal Pipe
Velocities
Dirty Air
Coal Pipe
Velocities
Coal Flow
Raw Coal
Raw Coal
Minimum
Frequency
every 3
months
every year
every year
or when
problem
indicated
every year
every
quarter
every year
Method(s)
ASME sampler
from Coal Pipes is
acceptable if coal
pipe isokinetic
sampling is not
conducted.
pitot tube
Dirty air pitot
Isokinetic
ASTM
ASTM & other
Goal
> 70% 200 mesh
> 99% 50 mesh
+/- 5% balance
pipe to pipe for
each pulverizer
+/- 5%
+/- 10%
ultimate,
proximate, HGI
coal erosiveness
mineral analyses
slagging/fouling
properties3
Comments
Isokinetic sampling not required if
only interested in fineness.
NOT recommend sample @ riffles.
Adjust classifier vanes considering
capacity versus fineness tradeoff.
Re-size fixed orifices to achieve
balance.
Caution: excessive AP may limit air
flow.
Dirty air velocity balance is final
basis for sizing orifices, not clean air.
If fineness and temperature are within
design limits, coal flow balance
achieved by riffles or pulverizer
maintenance & repair
Changes in HGI and moisture
important for pulverizer O&M.
Erosiveness important if high degree
of wear/erosion is experienced.
As shown in the table, fineness is recommended to be measured a minimum of once every
quarter as determined from composite samples collected from the coal pipes. Fineness
samples collected at or near the riffles on suction pulverizers have yielded erroneous
results. Degradation in coal fineness is one good indication of problems and should signal
an off-une inspection and assessment of the pulverizer. It should be noted that if there is a
high percentage of particles retained on 50 mesh sieve (e.g. over 1%), then consideration
should be given to sieving for 30 mesh fineness.
-------
Recommendations for clean air velocity, dirty air velocity, and coal flow sampling shown
in Table 3 represent general guidelines. The requirement for more frequent sampling or
tighter criteria for balanced fuel and air may be reasonable for many boilers considering
unit-specific conditions. For example, a boiler with a propensity for slagging problems
and/or a high market value for flyash sales would justify more frequent testing
requirements and investment in higher cost equipment upgrades. One important
recommendation is that coal pipes be balanced based on dirty air velocities which
represent the as-fired operation, and not to over emphasize clean air balance. Balancing
coal pipes to +1-5% on a clean air basis will generally correlate to good balance in the as-
fired condition. However, this is not always the case since imbalances tend to be
exacerbated by the presence of the entrained coal as shown in the example plotted in
Figure i.
Figure 1
Example of Pulverizer Clean Air vs. Dirty Air Velocities
(8-Corner Tangential Unit)
120%
2345
Corner Number
Summary, Conclusions, and Costs
A summary of low-cost NOX reduction techniques and range of costs is shown in Table 4.
BOOS, fuel biasing, air biasing, and reduced O2 have all demonstrated NOX reduction
potential on the eleven boilers as shown in Table 2, as well as a multitude of other units
throughout the U.S. As described in the introduction, tuning of primary air-to-coal ratio is
a viable method. One additional method available, shown at the bottom of Table 4, is to
convert the upper-most AA compartment to larger nozzle tips. This concept is designed
to maximize the use of this upper compartment in order to provide a partial overfire air
effect. Based on experience on two units, an increase in nozzle area of 30 to 50% may be
achieved with a resulting NOX reduction of 5% or more. There are some mechanical
design issues with burner tilt control which must be addressed, however these obstacles
can be overcome in most cases. Moreover, if nozzle tips are replaced on a regular basis
then this modification could be done for little or no incremental cost.
A range of cost for each option is shown in the second column of Table 4 expressed as
dollars per kilowatt. Generally all of the modifications described in this paper can be
implemented in this range of costs shown, with the exception of dynamic classifiers. Many
-------
units already have the necessary controls in place to implement some of the methods
described, and so costs would only be incurred to conduct field tuning and possibly
operator training. Older units, especially those built prior to 1980, tend to require more of
the low-cost upgrades in order to safely and reliably implement these NOx reduction
techniques. However, it is important to note that many of these upgrades would be
required with higher cost upgrades such as overfire air or LNB were to be installed. One
last important point is that many of the improvements described herein result
simultaneously in NOx reductions, performance improvements, and reduced maintenance.
Table 4
Summary of Low-Cost NOx Reduction Techniques
Technique
BOOS
Reduced
02
FA/AA
Bias
Fuel Bias
Optimize
PA-to-coal
ratio
Convert
upper AA
to larger
nozzle tips
Range of
Cost,
$ per kW
SO-S3
$0-$2
$0-$2
$0-$2
$0-$3
$0-$1
Typical NOx
Reduction,
%
25-50%
5-15%
5-15%
5-10%
5-10%
5%
Usual Limiting Factor(s)
Low WB pressure at
reduced loads.
Pulverizer capacity at foil
load.
Slagging/fouling, opacity,
CO, and flyash LOI.
Inadequate controls or
lack of feedback signal.
Inadequate feeder
controls.
Inadequate feeder or inlet
air flow controls.
Baseline pipe velocity
already at a minimum.
Low coal pipe
temperatures.
Nozzle tilt mechanical
interference.
Approach to Overcome
Limitation(s)
WB damper modifications (for
reduced loads).
Pulverizer or exhauster
upgrades to increase capacity.
Improve boiler O2 metering
and minimize boiler in-leakage.
Fuel system balance.
Secondary air balance.
Modify controls.
Add FA/AA damper feedback.
Increase frequency of feeder
calibration.
Feeder VSD upgrade.
Upgraded feeder controls.
Install pulverizer air flow
meters.
Increase hot air supply
temperature.
Disconnect from tilt and install
fixed nozzles with upward tilt.
References
1. "Fireside Performance Optimization and Emissions Control for Coal-Fired Boilers."
Electric Power Research Institute, 1995.
2. R.P. Storm, "A summary of Experiences Related to Achieving Optimum Pulverizer
Performance and Fuel Line Balance."
3. "Coal Fouling and Slagging Parameters," American Society of Mechanical Engineers,
E.G. Winegartner editor, 1974
-------
APPLICATION OF AN EXPERT SYSTEM AND NEURAL
NETWORKS FOR OPTIMIZING COMBUSTION
S. Williams
D. Cramer
Potomac Electric Power Company
8711 Westphalia Road
Upper Marlboro, MD 20772
E. Levy
N. Sarunac
T. Eldredge
S. Steele
Energy Research Center
Lehigh University
117 ATLSS Drive
Bethlehem, PA 18015
Abstract
Lehigh University's Energy Research Center and the Potomac Electric Power Company have
been developing software for use by plant personnel in tuning a pulverized coal-fired boiler to
reduce NOX and minimize heat rate. The software is based on the interactions of an expert
system, neural networks, and a mathematical optimization algorithm. It uses the expert system to
guide the plant engineer through a series of parametric boiler tests and gather a data base which
characterizes boiler operation over a wide range of conditions. The neural network portion
develops non-linear mapping functions between the outputs of NOX, heat rate, LOI, opacity, and
the controllable boiler input parameters. These mapping functions are then analyzed by the
mathematical optimization algorithm and the optimal boiler operating conditions are identified.
This paper describes the application of the software to a wall-fired boiler. In addition, a
technique is described for using results from the software to develop continuous, on-line closed-
loop combustion control.
-------
Introduction
Over the last several years, the U.S. electric utility industry has become sensitive to the benefits
associated with optimizing boiler operation. The utility might wish to adjust the control settings
of the boiler to minimize NOX or heat rate, or perhaps find the boiler operating conditions which
give the best heat rate subject to meeting a target level of NOX. Additional constraints which
must be satisfied include limits on steam temperatures, restrictions on fly ash unbumed carbon,
and limitations on stack opacity. Numerous papers have been written which show the extent to
which NOX, steam temperatures, heat rate, and LOI can be varied through changes in boiler
operating conditions. Typical controllable parameters include air register settings, furnace O2
levels, mill loading patterns, and burner swirl settings [see for example, 1 to 4].
In recent years, quite a few boiler optimization software packages have been made commercially
available for use by utilities and industrial boiler operators. Most rely on either one of two
techniques: the use of neural networks for on-line data acquisition or the use of sequential
optimization to determine the optimal settings [5. to 9].
Under a collaborative arrangement with the Potomac Electric Power Company, Lehigh
University's Energy Research Center has recently developed software for combustion
optimization which relies on an expert system, neural networks, and mathematical optimization
algorithm. Referred to as Boiler OP, the software is used to guide the plant engineer through a
series of parametric boiler tests. This results in development of a database which characterizes
boiler operations over a wide range of conditions. The software then analyzes the data and
identifies the optimal boiler control settings.
The first versions of Boiler OP were developed for use with tangentially-fired boilers. Results of
the applications of the software to optimization of units at PEPCO's Morgantown and Potomac
River Stations were described in previous papers [10. 11]. A wall-fired version of the software is
now available. This version can be used for both front-wall and opposed-wall fired boilers using
burners with either single or dual registers. This paper describes some of the results obtained by
using the software to analyze parametric test data obtained from a wall-fired boiler. In addition,
an application of the software is described for helping the plant engineer use the DCS to establish
closed-loop combustion control to maintain NOX emissions within a narrow range over the long
term.
Software Description
Figure 1 illustrates how the expert system, neural networks and optimization algorithm are linked
together within Boiler OP. The expert system portion of the code is used to guide the plant
engineer safely through the parametric boiler tests. To accomplish this, the engineer configures
the software to reflect the boiler and burner design, testing objectives, and operating constraints.
The expert system then recommends control settings for the test points. The boiler controls are
adjusted by the boiler operator and the test data are collected by the plant's data acquisition
system. The test data, which are stored in a database for later use, are utilized by the expert
-------
system to determine each successive point in the testing sequence. After the testing is complete,
the database exports the data to the neural network for modeling. The optimization algorithm
then determines the best combination of control settings which meets the test objectives. The
code can be used to determine the boiler settings which produce minimum NOX emissions or to
determine those settings which produce a minimum heat rate subject to a target NOX level.
The comprehensive database generated during the parametric testing makes it possible to carry
out a variety of analyses which can be used to assist the plant in fine-tuning boiler operations
over the long term. These calculation tools make it possible for the operators to review the status
of boiler operations as they affect NOX and unit performance, examine the consequences of
improper control settings, explore alternative control settings and reoptimize the boiler settings
subject to new operating constraints. Once the initial parametric testing is complete, no
additional testing is needed to carry out these calculations. The neural networks and
optimization algorithm within the software use the original database to obtain the desired
answers (see Figure 2).
Analysis of Wall-Fired Boiler Data
Parametric tests were carried out at full-load conditions at a unit with a wall-fired boiler having
low NOX burners with overfire air registers. During these tests, the overfire air, economizer O2,
burner secondary air and swirl settings and the primary air velocities were varied. The data were
then analyzed by the neural networks in the newly developed version of Boiler OP. These tests
were concerned primarily with determining the relative effects of the different control parameters
on NOX and thus, the trends shown here are limited to impacts on NOX.
Figure 3 shows the effects of both economizer O2 level and overfire air setting on NOX. The
effects of overfire air and burner secondary air setting are shown in Figure 4. The impacts of the
burner swirl, primary air bias and overfire air are shown in Figure 5 and, finally, the variation of
NOX with primary air bias for different swirls and overfire air settings is shown in Figure 6. The
primary air bias is an indication of the deviation of the primary air flow from its normal setting.
The positive bias shown here reflects an increase in primary air velocity.
These results show that NOX responded most strongly to overfire air settings and economizer O2
level, while the burner swirl setting and primary air bias have smaller, but significant, effects on
NOX. In this case, NOX was relatively insensitive to secondary air setting.
Using these data, the optimization feature of the software was used to find the boiler control
settings which produce the minimum NOX level at full-load conditions. The results of this
analysis are shown in Table 1.
-------
Table 1
Predicted Full-Load Boiler Control Settings for Minimum NOX
Parameter
02 Level [%]
OFA Damper Position
Secondary Air Position
Secondary Air Bias
Swirl Register Setting
Mill Bias
PA Bias
NOX, [Ib/MBtu]
3.21
92.4
50.9
0.2626
36.1
-0.0156
9.62
0.492
Application To On-Line Closed Loop Control
Boiler OP was developed as off-line advisory software, providing advice to plant engineers and
operators and helping them identify the best boiler control settings. However, recent results have
been obtained at PEPCO's Morgantown and Potomac River Stations which show how the
software can also be used to provide closed loop on-line combustion control.
There are two 600 MW tangentially-fired boilers at Morgantown station, both of which have low
NOX burners with both separated and close coupled overfire air. Following the low NOX burner
conversions, in 1994 and 1995, parametric tests were conducted, to determine the optimal boiler
control settings. These tests were carried out over the load range and provided the data needed to
specify key parameters such as overfire air damper setting, economizer O2 level and burner tilt
angle as functions of load. The control systems (DCS) were than programmed to permit
automatic operation at these optimized settings. Despite the fact the boilers meet the station's
NOX and heat rate objectives most of the time, it was found there are periods during which NOX
deviates from the target level. These deviations occur because of fluctuations in coal quality and
variations in furnace cleanliness, which in some cases cause the NOX to be higher than desired,
and in other cases to be lower.
Data published in the literature show NOX depends on coal composition, varying with volatile
content, amount of fixed carbon and the nitrogen content of the coal. Analysis of coal quality
data from PEPCO suppliers showed the expected variations in coal composition are large enough
to cause significant variations in NOX.
-------
Furnace cleanliness also affects NOX emissions. As slag accumulates on the waterwalls, the
flame temperature increases and this results in increased NOX. Sootblowing tests carried out at
Morgantown Unit 2 showed changes in NOX emissions ranged up to 0.06 Ib/MBtu as the
waterwalls were cleaned [12].
Both Morgantown units are subject to opacity excursions related to combustion conditions.
These units use cold-side electrostatic precipitators for opacity control, and the performance of
these devices is sensitive to the electrical resistivity of the fly ash. Variations in coal
composition and furnace 02 level lead to variations in the amount of fly ash unburned carbon,
which affects fly ash resistivity, which, in turn, impacts opacity.
Using a control logic which they developed from results obtained by analyzing Boiler OP test
data, PEPCO engineers configured the boiler controls to maintain NOX at the target level and
prevent opacity from exceeding the regulatory limit (Figure 7). The control logic depends on
access to NOX and opacity signals from the CEM. It also uses relationships between NO.<, heat
rate and the boiler control settings provided by the software.
Figures 8 and 9, which are for full-load conditions, illustrate the types of results which are
generated by the software using its "reoptimize" capability. The individual points are the results
of calculations performed by the software to indicate the combinations of O2 and SOFA settings
which result in the best heat rates as NOX is increased or decreased around a target value. For
this boiler at these conditions, the software recommended the other boiler control parameters be
maintained at fixed positions. The solid curves, which are curve fits of the predictions, are then
used as input to the DCS to maintain NOX at a fixed level.
A similar strategy was found to work well at Potomac River Station. Each of these units has a
tangentially-fired boiler with conventional burners and a capacity in the range of 100 MW. NOX
was reduced from baseline levels in the 0.65 Ib/MBtu range to 0.45 Ib/MBtu at Potomac River
through combustion optimization. To maintain closer control over NOX and prevent the
fluctuations which were found to arise due to variations in coal quality and furnace cleanliness,
results obtained from Boiler OP were used to configure the boiler controls to maintain NOX at the
0.45 Ib/MBtu target.
Summary
Relationships between parameters such as NOX, opacity, heat rate and the boiler control settings
are complex, making it difficult to optimize combustion without assistance from a computer
code. Through application of an expert system to guide the testing and neural networks and an
optimization algorithm to analyze the data, Boiler OP can be used to determine the optimal
•control settings for a wide a variety of boiler types. The code has been applied to tangentially-
fired boilers, with conventional and low NOX firing systems. A new version of the software was
recently developed for use with wall-fired boilers.
-------
The database generated using this approach can be used by the plant to maintain the boiler in an
optimized state, provided there are no major changes in the maintenance condition of the
equipment or in the coal supply. Recent work at PEPCO has shown how results from the
software can be used by the plant engineer to configure the DCS to provide continuous closed-
loop combustion control to maintain NOX emissions within a narrow range over the long term.
References
1. Levy, E., M. D'Agostini, D. Eskenazi, S. Williams, D. Cramer, E. Petrill, and R. Squires,
"NOX Control and Performance Optimization Through Boiler Fine-Tuning," Proceedings
1993 EPRI/EPA Joint Symposium on Stationary Combustion NOX Control, Miami
Beach, Florida, May 1993.
2. Williams, S., E. Levy, E. Petrill, and R. Squires, "Optimizing Performance and NOX
Emissions Using Improved Practices and Controls at Potomac River Unit 4," Proceedings
1993 Pittsburgh Coal Conference, September 20-24, 1993.
3. Maines, P., E. Levy et al., "Combustion Optimization of Low NOX Burners at PEPCO's
Morgantown Station." Proceedings 1995 EPRI/EPA Joint Symposium on Stationary
Combustion NOX Control, Kansas City, MO (May 1995).
4. D'Agostini, M., R. Walsh, D. Eskenazi, E. Levy, P. Maines, S. Williams, and E. Petrill,
"Tradeoffs Between NOX, Heat Rate and Opacity at Morgantown Unit 2," 1996 EPRI
Heat Rate Conference, Dallas, Texas, May 1996.
5. Boylve, R. J., J. W. Pech, and P. D. Patterson, "Reducing NOX While Maintaining Boiler
Performance At TVA's Johnsonville Steam Plant Using Constrained Sequential
Optimization," EPRI/EPA 1995 Joint Symposium on Stationary Combustion NOX
Control, Kansas City, Missouri, May 1995.
6. Holmes, R. et al., "GNOCIS: An Update of the Generic NOX Control Intelligent
System," EPRI/EPA 1995 Joint Symposium on Stationary Combustion NOX Control,
Kansas City, Missouri, May 1995.
7. Kenien, D., et al., "TOPAZ: Improving Emissions Performance Through Optimization,"
presented to Pennsylvania Electric Association, September 17, 1996.
8. Heimes, F., et al., "Recent Progress With the Intelligent Emission Control System,"
American Flame Research Council International Symposium, Baltimore, Maryland,
September 30, 1996.
9. Neusight - Product Brochure From Pegasus Technologies Corporation, Painesville, Ohio,
1994.
-------
10. Williams, S., and E. Levy, "Application of Boiler OP for Combustion Optimization at
PEPCO," EPRI Workshop on NOX Controls for Utility Boilers, Cincinnati, Ohio, August
6-8,1996.
11. Levy, E., J. Pfahler, J. Miles, F. Keller, P. Lee, S. Woldehanna, and S. Williams,
"Combustion Optimization Using an Expert System and Neural Networks," 1996 Joint
Power Generation Conference, Houston, Texas, October 1996.
12. Pavinski, D., C. Romero, E. Levy, D. Cramer, P. Maines, S. Williams, and E. Petrill, "An
Approach to Boiler Sootblowing for Optimizing NOX and Heat Rate," Sixth International
Joint ISA/POWID/EPPJ Controls and Instrumentation Conference, Baltimore, Maryland,
June 1996.
TESTING AND OPTIMIZATION
• Guide Parametric Testing
• Identify Optimal Control Settings
OPERATOR TOOLS
• Check Penalties
• Ask "What IF' Questions
• Reoptimiie
Figure 1: This diagram shows how Boiler
OP is used to assist a plant engineer in
testing the boiler and determining the best
combination of control settings.
Figure 2: Boiler OP also provides the boiler
operators with assistance in maintaining the
unit in a well-optimized condition.
-------
OFA Damper Opening [%1
SO 50
Secondary AJr Setting (%)
Figure 3: Effects of OFA Damper Opening
and Economizer O, Level on NOX.
Figure 4: Variation of NOX with OFA and
Secondary Air Settings.
OFA -25f*, Primary *tr b!» » 0 %
• OFA - 25 %. Primary air bUi -12%
OFA - 100 V Primary air bias -OX
%, Primary
Figure 5: Effects of Burner Swirl,
OFA and Primary Air Bias on NOX.
Figure 6: Effects of PA Bias, Burner Swirl and
OFA Setting on NOX.
-------
NO. SET
POINT
MEASURED
NO.
CONTROLLER
+A
Q,
TILTS
MILLS ,
CONTROL
LOGIC
V
RECOMMENDED
RELATIONS BETWEEN
PLANT
ENGINEER
BOILER CONTROL
VARIABLES
Figure 7: Illustration of how Boiler OP can
be used to develop closed-loop combustion
control for limiting variations in NOX.
Figure 8: Predicted change in NOX with change in
SOFA setting required to eliminate NOX variations
while maintaining minimum heat rate. Figure 9
shows corresponding change in economizer O2.
-0.05
-0.10 J-
Dolta NO, [Ib/M8tul
X) 0.02 0.04 0.06 0.0f
Figure 9: (See caption for Figure 8).
-------
The Emissions, Operational, and Performance Issues of Neural Network
Control Applications for Coal-Fired Electric Utility Boilers
R.C.Booth
Thomas C. Kosvic
NeelJ.Parikh
Pegasus Technologies, Ltd.
1100 Mentor Avenue
Painesville, OH 44077
Abstract
NOX and unit heat rate are sensitive to the combustion process occurring in a boiler.
Boiler combustion profiles change continuously due to coal quality, boiler loading,
changes in slag/soot deposits, ambient conditions, and the condition of plant equipment.
This paper presents the benefits of applying an on-line, real-time neural network to several
bituminous coal fired utility boilers. The system dynamically adjusts combustion setpoints
to reduce NOX emissions and improve heat rate. The neural network can dynamically
track and optimize operations for a large number of input variables. Neural network
technology augments but does not replace performance monitors or engineering diagnosis.
It is a tool that has been proven to improve boiler operation through dynamic adjustments
that results in reduced power plant emissions and improved boiler performance.
The NeuSIGHT neural network based system has been applied to units with tangential-,
cell-, single wall-, and opposed wall-burner arrangements that have ranged in capacity
from 146 to 800 MW in an advisory mode. Several sites have employed the neural
network system for supervisory (i.e., closed loop) boiler control.
The neural network is applied to the boiler by first modeling the multi-dimensional and
non-linear problem of NOX formation and performance improvement in the furnace. This
is done by operating the boiler in prescribed modes that are designed to encompass a
wider range of parameter variations than will occur in daily operation. This is done to
provide the neural network a large experience base. Once modeled, the neural network
continually retrains itself using current boiler information and performs many 'what if
simulations to optimize setpoints for the current operating and equipment status
conditions. The neural network periodically updates the model, learning from most recent
data, to keep up with changes in operating conditions. Through on-line retraining, the
neural network system optimizes the boiler operation by accommodating equipment
changes due to wear and maintenance outages and adjusting to changes in fuel quality to
widen compliance margins and improve operating flexibility and performance.
-------
The system helped reduce NOX emissions up to 60%, meeting compliance and providing
significant operational benefits. The NeuSIGHT system has also helped to optimize
sootblowing and slag deposition relationships, superheat and reheat temperature control,
pulverizer operation and maintenance, and has improved operational procedures at low
load. The system also improved heat rate up to 2% overall (5% at low load) and reduced
LOI as much as 30% while meeting NOX emission compliance through combustion
optimization alone.
Combustion Optimization
The basis for boiler combustion optimization lies in identifying the relationship of
important fuel/air parameters. Values for these parameters are computed and input to the
boiler control system as modified setpoints that will provide improvements for heat rate,
NOX and combustibles (CO and LOI).
Maintaining reduced NOX can be difficult as operating personnel, fuel, and/or equipment
condition change over time. Long term emission gains require a methodology that can
satisfactorily model the various non-linear interactions and accommodate changing
equipment and coal conditions. An accurate and dynamic plant model requires dozens to
hundreds of process values as inputs in order to keep up with the changing plant
conditions. The key features of the neural network are its ability to empirically model
non-linear data, the ability to continually learn in an on-line mode and modify setpoints
and bias adjustments to keep the unit tuned to the desired emission and performance
levels.
Neural Network Model Development
Depending on the size and complexity of each unit, as many as 350 boiler operational
parameters are collected at prescribed intervals and averaged over a short time period
(typically 30 seconds and 10 minutes, respectively) during a two to four week test period.
The objective of these tests is to vary one operational parameter per test to develop a
history of unit operating data that can be used to train a neural network model for use on
the subject unit. It is desirable to exercise most parameters to values beyond those
encountered in normal operation. This provides the neural network model a wide data
range upon which to train.
Following these tests, the logged data is modeled by Pegasus' NeuSIGHT neural network
program using NOX, heat rate, or other boiler performance measures as model targets.
Many combinations of node arrangement and functional link enhancement terms are
tested. For modeling heat rate, up to 35 hidden nodes are used. The ideal number varies
with the particular input list used. For NOX, it is usually necessary to have at least one
layer of FE (sin/cos type nodes) terms. Without at least one FE layer, the NOX model
accuracy is typically only +/- 20% and misses many peaks and valleys. Additional testing is
performed on the boiler at various loads and operating modes to ensure the model will be
accurate over an extended time period.
-------
Programs are utilized to condition the input data to and from the neural network,
compensating for inherently noisy data. The objective is to tune the system to respond
gradually to changes (slag and ash buildup, pulverizer degradation, instrumentation drift,
etc.), but not be so sensitive that the system may "hunt" between optimum setpoints.
Some of the issues affecting the system's sensitivity include: process/equipment dynamics,
instrumentation response times and time lags between input data. Additional programs are
incorporated to provide gradual transitions for the setpoint and bias adjustments as the
system responds to changes in operating conditions or equipment performance. Limits, in
addition to those imposed by the DCS, are established to ensure safe and stable operation.
Examples include using low CO to bound the O2 loop and high motor current to limit the
pulverizer bias adjustments.
Finally, supervisory control can be implemented. Initially, the loops are put into service
one at a time under the supervision of an engineer. After the desired operating
characteristics are gained, the loop is released to the operators, who are provided three
control options: manual, DCS, or supervisory (closed loop).
The original model is periodically retrained to reflect current operating conditions. A
major concern is to ensure the model is trained on valid data. Algorithms are designed to
monitor the "breadth" of data and to sort incoming data into groups representative of
different modes of plant operation; thereby ensuring a "robust" data set for the neural
network throughout the plant's operating range. Individual inputs are monitored through
pattern matching checks. Bad values are replaced with estimations, based on the patterns
of associated input parameters. If too many inputs are bad or too little data exists for the
current conditions, the neural network reverts to the original digital control system
setpoints until the system has sufficient data to make valid recommendations.
Results From Four Neural Network Applications
The following discussion summarizes the results from four electric utility boilers that
represent a wide range of furnace and burner types.
Unit "A"
Unit "A" is a 1963 vintage Babcock & Wilcox wall fired boiler with gross generation
capacity of 146 MW. Baseline NOX was 1.0 lb/MMBtu. Linear parametric testing, by
plant personnel and others, reduced NO* to 0.78 Ibs/MMBtu at full load under normal
operating conditions. The project goal was to further reduce NOX at least 20% while
maintaining or improving heat rate and flyash loss on ignition (LO1). Since the plant
operating permit includes an annual NOX tonnage limit on each unit, reducing NOX
emissions would improve the capacity factor resulting in increased operating flexibility of
this unit.
The control system interfaced a SUN Unix workstation to a Bailey Controls Net90 digital
control system (DCS) through a Bailey Controls Computer Interface Unit (CIU). The
neural network and data collection/processing system resided on the workstation. New
-------
setpoint values were determined by the neural network periodically within defined
constraints to insure safe operation.
Initial model results identified nine on-line parameters that could be used to reduce NOX an
estimated 15% below current levels. These parameters are:
• O2 setpoint bias (1);
• Mill bias (4), and;
• Pulverizer primary air outlet temperature (4).
Secondary air register positions also affected NOX emissions, but testing was limited since
these are manually adjusted .
NOX Emissions. Two major changes were implemented during an outage, based on
recommendations by the NeuSIGHT system and plant maintenance guidelines. These
were to reposition the secondary air registers and replace the balls and tighten down the
heads of the two lower mills (A and D). These changes reduced NOX by 10% before the
neural network supervisory control was implemented.
< 155
£•
£ 150
Mil IA
(Bottom)
MiMD
MilIC
Mill
MilIB
(Top)
Figure 1
Pulverizer Primary Air Temperature Operating Range
-------
The baseline pulverizer primary air outlet temperatures were maintained at 150 °F.
Currently, the optimum settings vary day to day based on coal quality. Figure 1 illustrates
the primary air temperature range that each mill operates at and the effect of coal moisture
on these temperatures.
These setpoints are constrained between 145 °F and 165 °F under neural network
supervisory control. They are also affected by the excess air setpoint and pulverizer bias
settings.
Implementing supervisory control on excess oxygen, pulverizer primary air outlet
temperatures, and mill biases has maintained the NOX emissions between 0.56 to 0.70
Ibs/MMBtu at high loads. This represents a NOX reduction as high as 28%. Figure 2
shows the impact of specific primary air temperature settings onNOx during a ten hour
operating period.
09/28/95 ZO:O7:51
Figure 2
Impact of Primary Air Temperature with Supervisory Control of NOX
-------
Figure 3 shows the change in NOX that was achieved throughout the normal unit load
range.
0.8 -
0.7
S 0.6 r
s
0.5
0.4
0.3
40 50 60 70 80 90 100 110 120 130 140 150
Load (GMW)
Figure 3
Effect of Supervisory Control on NOX Emissions
Adding supervisory control of mill biasing provided additional gains. The results from
80MW to 120MW indicate larger NOX reductions are feasible due to greater mill biasing
flexibility.
Heat Rate and Loss of Ignition (LOI). Neither heat rate nor loss of ignition (LOI)
values were available on-line. However, they were manually tracked during the project.
Comparing heat rate values for a 30-day baseline period prior to a similar period of
supervisory operation showed that, after correcting for circulating water inlet
temperatures, the average for the test period was 1.2% below the baseline value. This data
is summarized in Table 1. Also included is a 15-day period of supervisory control with mill
bias that resulted in a 4.4% heat rate improvement.
-------
Table 1
Unit "A" Heat Rate
Operating Mode
Baseline
Supervisory Mode
Supervisory w/mill bias
Uncorrected Heat Rate
13,049
12,633
12,206
Corrected Heat Rate
12,794
12,644
12,233
% Improvement
1.2%
4.4%
Limited LOI data suggests a significant reduction in the amount of unburned carbon in the
ash. Table 2 presents typical values recorded prior to this project, followed by the values
taken during the 30-day availability run, which indicates a reduction of 30% to 60%.
Table 2
Unit "A" Ash Analysis
Operating Mode
Baseline
Baseline
Supervisory
Supervisory
Supervisory
Supervisory
LOI (%>
9.7
10.4
3.1
3.4
2.8
6.1
More values are required to statistically verify this dramatic improvement.
Unit "B"
Unit "B" is a 500 MW tangential unit with five levels of burners in a single furnace. The
objective of this program was to determine the feasibility of unit "B" to meet a 0.45
Ib/MMBtu NOX compliance emission limit utilizing a neural network advisory control
system. Three weeks of on-site tests were performed during which the following
parameters were adjusted on an individual basis to provide data for the neural network
model:
• Primary air temperature of each coal mill;
• Coal loading from each mill;
• Excess O2;
• Burner tilt;
• Warm-up oil air damper opening;
• Auxiliary air damper opening, and;
• Active coal air damper opening.
-------
These tests were followed by a brief (two day) period of multi-variable tests to assure the
system's ability to meet the NOX compliance limit.
NO, (Ib/MMBtu)
Top Level F/A Damper
30
Lower Level F/A Damper
Figure 4
Effect of Fuel/Air Damper Position on NOX
NO, (Ib/MMBtu)
Top Mill PA Temperature (°F)
80 100
Lower Mill PA Temperature (°F)
Figure 5
Effect of Mill Primary Air Temperature on NOX
-------
Figures 4 and 5 illustrate the sensitivity of NOX to the top and lower burner secondary
airflow and primary air temperature, respectively.
CO and LOI. Full load tests utilizing a multi-point emissions sample grid showed that
the combustion uniformity (represented by point-to-point NOX emissions) was relatively
unchanged from the baseline. During these tests the average CO increased from 9 to 207
ppm when the optimum low NOX firing mode was employed. However, unburned carbon
in the flyash decreased from 10.9 (baseline) to 7.3 % during low NOX firing.
A/OX Emissions. The preliminary recommendations for low NOX operation after the
conclusion of neural network modeling included:
• Reduce the top elevation mill coal flow whenever feasible;
• Remove the top elevation mill from service whenever load can be sustained
with four or less coal mills in service;
• Open coal air dampers 100% on active burner levels;
• Close manually operated warm-up oil air dampers; and;
• Open the top two levels of auxiliary air dampers 100%.
The third recommendation, above, is counter-intuitive to usual NOX reduction practice for
tangential boilers. However, testing confirmed that this recommendation was valid.
Figure 6 represents the minimum NOX emissions that were achieved for one week of
operation after the preliminary recommendations were given to the boiler operators. The
neural network modeling indicated that NOX compliance could be achieved throughout the
normal operating load range on this unit.
Although NOX emissions from a tangentially fired boiler are usually highest at minimum
load, NOX was reduced 55% from the baseline average at rninimurn load after
NeuSIGHT's recommendations were implemented.
-------
1.0
0.9
08
0.7
0.6
0.5
8 0.4 h
0.3
0.2
01
00
Baseline
Minimum NOX following Pegasus' recommendations
* Compliance Limit
100
300
400
500
600
Load (MW)
Figure 6
NOX Emissions After System Optimization
A NeuSight based advisory or closed loop system would enable this unit to be operated
with an acceptable margin for compliance at all loads. However, in light of the NOX
reduction that was achieved, the utility opted to implement these operating
recommendations but delay the installation of an advisory system until the time when
increased NOX compliance margin is required or until heat rate and LOI considerations
become more important.
Unit "C"
Unit ';C" is an 800 MW boiler with opposed wall burners. The unit is equipped with low
NOX burners and overfire air (OFA). A RS-232 data link connects a SPARCstation 10
based NeuSIGHT process optimization to a Honeywell TDC 3000. The linking software
transmits and receives over 350 points, analog and digital, per minute although the number
of points processed by the system could be increased to more than 2000 points per minute.
10
-------
The insights derived from the neural network modeling indicated that:
• The O2 setpoint should be increased from 3.2 to 4.0%;
• NOX varies directly with excess 02;
• Opening the overfire ports 5 - 10% can actually lower heat rate and NOX
simultaneously (above 10%, the heat rate increased);
• Varying the overfire air port openings reduced NOX by 30%;
• Balancing secondary air flow (i.e. 02 probes +/- 0.5% of average) shifted the
NOX curve from 0.45 - 0.65 Ib/MMBtu to 0.30 - 0.45 Ib/MMBtu, and;
• Balancing air flow reduced the O2 setpoint and improved combustion
conditions virtually eliminating furnace slagging.
Although the initial model recommendations included increasing the excess O2 to 4.0%, a
subsequent calibration of the O2 probes revealed that these probes had been reading low
(approximately 0.8%). After the probes were calibrated the neural network reduced its O2
recommendations, as it retrained, to 3.2%.
Load cycling affects O2 and heat rate:
• When load decreases the control system lags air behind fuel causing high O2;
• At the same time the heat rate will improve as the latent heat in the boiler
requires less fuel to maintain the desired load, and:
• When the unit reaches steady state (even for a few minutes) the O2 will
continue to decrease (due to the lag) and the heat rate will degrade.
Methods to correct this problem included time averaging of the data and the introduction
of lead/lag terms.
Other parameters that the optimization system identified as needing adjustment include:
• Superheat attemperation flow;
• Feedwater inlet temperature;
• Furnace exit temperatures;
• Furnace pressures, and;
• Secondary airflow balance.
The superheat attemperation flow and the furnace exit temperatures appear to have high
penalties: as much as 20 - 50 btu/kwhr. This is likely due to changing furnace and
convective pass cleanliness since the penalty increases over time. A means of
incorporating a soot blowing model that will help minimize these penalties is being
explored.
NOX Emissions. The NeuSight system on this unit has been operating for 18 months.
During this period NOX levels have remained consistently within compliance and boiler
performance has been improved. As part of ongoing modifications to this system, the
11
-------
logic has been adjusted to provide compliance with NOX for the adjusted regulatory limits
(0.7 for 3 hour average and 0.5 for 30 day average) instead of instantaneous NOX values.
Analysis of the year long database from Unit "C" show some significant and interesting
observations. The new NOX requirements will have a very positive effect of improving
heat rate nearly 1.5% while still maintaining a good compliance margin. New logic that
modifies target values to avoid achieving a future average value has been developed and
incorporated into this system.
Figure 7 shows NOX values predicted for various excess oxygen levels and OFA settings at
four different load points.
•S251*xOFAs@70%
-825MwOFAsS50%
60CM2.CFAs@70S.
Excess ft (%)
Figure 7
Effect of Over Fire Air and Excess
on NOX
Heat Rate. Figure 8 shows heat rate values predicted using the final trained data set
developed from the year long database for the same operational parameters.
This indicates that closing the OFA dampers will improve the unit heat rate. Closing the
OFA dampers will nominally shorten the combustion zone which for this boiler improves
heat rate.
The heat rate data also show that O2 control is not always the major heat rate
improvement factor. For this boiler increasing O2 decreased heat rate at most conditions.
12
-------
— - -825 Mw, OFAs@70%
825 Mw, OFAs@50%
- - - 825Mw.OFAs@
750 Mw, OFAs@70%
750 Mw.OFAs@50%
750Mw,OFAi@10%
600M2.0FAs@70&
600 MW,OF.As@50%
Excess Oj (%)
Figure 8
Effect of Over Fire Air and Excess O2 on Heat Rate
Unit "D"
Unit "D" is a B&W boiler that is nominally rated at 650 MW. It has an opposed wall
burner configuration supplied by seven pulverizers. This unit was commissioned in 1983
and was designed to meet a NOX guarantee of 0.65 Ib/MMBtu. The objective of the
neural network program was to reduce NOX to 0.50 Ib/MMBtu or less.
The initial evaluation was performed for normal full load conditions (i.e. 620+ GMW)
where minimizing NOX was most critical. The model results showed that:
• The loading of the top mill (D) has the greatest impact on NOX;
• Based on the model, the middle mills B, C, and E should be biased up to
reduce the loading on top mills D and G;
• The optimized settings for the secondary dampers vary over time, in response
to heat transfer surface cleanliness and mill operation;
• Primary air flow and pressure have weak model correspondence to heat rate
and NOX;
• Reduced NOX emission can be obtained by lowering the primary air
temperatures in the upper mills, particularly mill D;
• The superheat spray flow shows a strong sensitivity for both NOX and heat
rate, reflecting the influence of furnace cleanliness on these parameters.
13
-------
10100 " As found conditions
10000 -
9900 -
9800 -
As found conditions restored
NeuSlGHT recommendations
1.0
10.5 20.0
Data Patterns
29.5
39.0
Figure 9
Effect of NeuSlGHT Recommendations onNOx and Heat Rate
NOX Emissions. Testing showed that NOX was reduced 15% at full load. Although this
was somewhat less than the initial model analysis had indicated, it reflects the fact that
reality is often more constrained than a model.
Heat Rate. Although the model predicted that heat rate could be decreased
approximately 5%, actual testing determined that a 0.75 - 1.25% heat rate improvement
could be achieved. In fact, reduced NOX operation was accompanied by a nominal 1 %
heat rate improvement. This is illustrated in figure 9, above.
Setpoints for supervisory control include mill feeder speed, primary and secondary airflow,
primary air temperature, superheat and reheat spray flow, and excess 62-
14
-------
Conclusions
NOX reduction and heat rate improvements are not contradictory goals under all
circumstances. Heat rate and NOX reductions are possible by optimizing the boiler
combustion process. An additional benefit of improving heat rate is the direct reduction of
SOz and COz emissions. NOX emission levels can be reduced, in some cases, to levels
comparable to low NOX burner retrofits without an LOI penalty for a fraction of the cost.
In all cases, improved combustion can only lead to better performance and increased
operating flexibility.
One advantage of using NeuSIGHT in supervisory control is that the system learns from
the best operators and automatically captures and uses that knowledge through periodic
model retraining. Thus, their experience with the boiler is incorporated directly into boiler
operation for all shifts.
Observing data from various plants shows that sootblowing frequency and location has an
immediate effect on both NOX emissions and heat rate. Utilizing a neural network to
provide guidance in activating the sootblowing sequence can provide NOX and heat rate
benefits while minimizing the costs of sootblowing. This system could, in addition,
quickly learn to modify sootblowing operations to accommodate changing coal conditions
rather than learning by costly changes in performance and/or emissions.
Pulverizer performance has a direct effect on NOX emissions, unit heat rate and LOI. A
neural network-based supervisory control system reacts to the performance of a
pulverizer. If a pulverizer performs poorly, it is biased down automatically by the neural
network to minimize it's impact on performance. A subsystem of the neural network is
under development that will associate "bad" areas of operation with maintenance activities
enabling plant personnel to plan maintenance activities proficiently. This will help increase
the unit availability and capacity factors by reducing unplanned pulverizer outages.
Neural network technology has been shown to provide significant reductions in NOX
emissions and significant increase in unit performance. Future applications of this
technology will provide users of the technology with improved availability, reliability, and
operability, with reduced maintenance expense.
15
-------
EMISSION SOLUTIONS THROUGH OPTIMIZATION
Micheal Lewis
Monte Gottier
Lower Colorado River Authority
2001 Ferguson Road
Marble Falls, Texas 78654
Blake Stapper
Radian International, LLC
8501 North Mopac Boulevard
Austin, Texas 78720
Abstract
Electric utility companies face the challenge of remaining competitive while developing
environmental strategies that address regulatory compliance issues within economic restraints.
Air emission compliance standards have had a significant financial impact throughout the
industry. "Best Available Control Technology" (BACT) solutions are expensive,
implementation is often complicated, and results vary. Alternative solutions could prove to
be more cost effective. This paper documents efforts made by The Lower Colorado River
Authority to use neural net optimization software as a pollution prevention tool to reduce
emissions from a large utility boiler. This project was funded in part through the EPA
Pollution Prevention Incentive for States (PPIS) 1995 Grant Program. The paper details
development of models built to predict NOX and CQ. These models identify and rank
variables that affect the pollutants in order of influence. Control models were developed for
use in determining the optimal set point for each control variable needed to attain the
emission reductions. This report summarizes data gathering methods, test plans, modeling
scenarios, recommendations, and implementation results; and analyzes the resulting
economical benefits. It is the intent of this paper to identify a cost-effective alternative
solution for emissions reduction.
Introduction
In many instances, environmental requirements can be impediments for profitable business.
Industry's concern for the financial impacts of new regulations is perceived by the public as
being insensitive towards the environment. The cost of compliance continues to increase,
while managers are forced to operate with flat or reduced budgets. It is imperative that we
search for solutions that allow us to meet our regulatory obligations with the least economic
impact. Some of the answers may be discovered through the use of advanced software tools
that learn the dynamics of complex processes. An increased understanding of the relationship
-------
between the boiler process and the formation of pollutants could reveal opportunities for
emission reductions. The initial investment required for this is minimal compared to "Best
Available Control Technology " options.
This project was conducted at The Lower Colorado River Authority's, Thomas C. Ferguson
Power Plant located in Horseshoe Bay, Texas. The facility is a 430 MW, natural-gas fired
Combustion Engineering boiler rated at 2,900,000 Ibs/hr steam flow. The objective of the
project was to use an advanced modeling tool to decrease the greenhouse gas emissions from
the boiler. To achieve reductions in carbon dioxide (COj) emissions, the strategy was to
reduce the unit heat rate. A secondary objective of the modeling was to reduce emissions of
oxides of nitrogen (NO,).
Approach
The method used to achieve the project objectives can be divided into three tasks. These are
data collection, modeling and implementation. Data collection includes planning which tests
should be conducted, how the data will be collected, and an evaluation of the data quality
prior to the beginning of the modeling task. Pavilion Technologies' Process Insights™ was
chosen as the software tool for the modeling. Prediction and control models were developed
to identify which input variables were most important to controlling the process. Once the
models were completed, the recommendations were implemented, and the results were
documented.
Data Collection
The quality of data collected will affect every aspect of a project, so it is important that close
attention be given to this task. The data sets need to contain all the process and continuous
emissions monitoring system (CEMS) data in a common format. Data were acquired by
coupling the plant's Honeywell 4000 with a Levi Lamb, Inc. EX4000 interface. The EX4000
software runs in a DOS windows environment, and interfaces with the Honeywell through
a custom designed communications board installed in the Honeywell. The software allows
the user to choose any or all tags and log their values in any time range from one second to
one day. The data values are written in an ASCII format to a series of user defined "roll" files.
The user defines the data interval and number of intervals each roll file is to contain.
The ability to retrieve plant data in sufficient quantity and format is critical. Data were
collected at one minute intervals for 200 tags from the Honeywell and five tags from the
CEMS. The data sets were created by merging ASCII data from the DCS and CEMS in an
Excel spreadsheet. Process Insights readily imports the data sets, and combines the individual
data sets to form one complete data set.
Once the data collection issues were addressed, a test plan was developed. The plan required
the plant to operate through different scenarios at specified load ranges. To build an accurate
model, the data must reflect the complete range of operation for each of the process
-------
variables. The test plan consisted of forty tests, in seven load bins, with a duration of one
hour per test.
There are additional issues concerning implementation of the test plan. There are costs
associated with taking a unit off dispatch control during testing. The load profile may also
require the testing to be conducted at night after peak. This can increase the cost and
complicate the coordination of consultants, engineers, and plant personnel. It can rapidly turn
into a major logistical problem and delay the project.
After the data are collected, they must be reviewed prior to the start of the modeling task.
It is important that the data be an accurate reflection of the process to be modeled. Outliers
like CEMS calibrations, erroneous readings from plant instrumentation, and downtime data
are removed. This will prevent the software from focusing on data that are not consistent with
the normal operation of the boiler.
Modeling
Once the data are collected and reviewed, the model development phase begins. Models are
developed in two stages. In the first, prediction models are prepared to help identify which
variables are most important for predicting the process. In the second stage, control models
are developed in which the process control variables are manipulated to achieve the optimum
conditions. The optimization software selected for this project was a neural net based
software, Process Insights™ from Pavilion Technologies' Inc. This software is designed to
evaluate the data and determine patterns in the process.
Prediction Models. In a prediction model, process variables are used to predict the desired
output variables. The output variables are those parameters that will be optimized as part of
the control model. In addition, any parameters which may require special constraints may be
included as outputs. The objective of the prediction modeling task is to develop a model that
combines the highest accuracy with the minimum number of inputs.
The initial process of determining which inputs to keep, requires personnel with knowledge
of the combustion process and the overall operation of the facility. During this phase, the
number of inputs in the training data set was reduced from 200 variables to 60 variables. The
next step was the training and development of the prediction models. The NO, and CO2
values were identified as outputs, along with the boiler fuel flow rate, net heat rate, and
windbox-to-furnace differential pressure. The other 55 variables were included as inputs.
The software learns the behavior of the process and predicts the outputs. The result of the
predicted values of the outputs are then compared to the actual values. The software ranks
the sensitivity of the outputs to each of the input variables, and a few of the less important
variables are removed to create each succeeding model. This process is repeated until the
inputs are reduced to approximately 10 to 20 variables. The final prediction model used for
this project contained 13 inputs. The sensitivity ranking reports for NOS and C02 are indicated
-------
in Tables 1 and 2, respectively. This final model was verified with a separate validation data
set to confirm the models' ability to accurately predict the outputs.
Table 1
Sensitivity Analysis for CO2 Prediction Model
Rank# Input Name
1 Net Generation
2 Main Steam/ Hot Reheat Differential Temperature
3 Degrees to Saturation
4 C Level Burner Status
5 Flue Gas Combustibles
6 Burner Tilt Position
7 Excess Oxygen
8 B Level Burner Status
9 A Level Burner Status
10 Main Steam Temperature
11 E Level Burner Status
12 D Level Burner Status
13 Throttle Pressure
Table 2
Sensitivity Analysis for NOX Prediction Model
Rank# Input Name
1 Net Generation
2 Degrees to Saturation
3 Excess Oxygen
4 Main Steam/ Hot Reheat Differential Temp. Flue Gas Combustibles
6 Burner Tilt Position
7 A Level Burners
8 Main Steam Temperature
9 E Level Burners
10 B Level Burners
11 C Level Burners
12 D Level Burners
13 Throttle Pressure
-------
Control Models. The control model is developed using the same input and output
variables that were contained in the final prediction model. In the control model, variables
that require optimization or special constraints are designated as outputs. The inputs are
divided into independent and dependent variables. Independent variables consist of those
inputs that are manipulated to control the process, and inputs that are completely external
to the process. Dependent variables consist of inputs whose value changes as a result of a
change in an independent variable, but that cannot be manipulated to significantly affect
the process.
In the case of a utility boiler, outputs may consist of pollutant emission rates, fuel flow
rates and net heat rate. Independent variables include unit load, level of excess oxygen
and burners in service. Dependent variables may be steam temperatures and pressures, or
flue gas temperatures. It should be noted that the classification of a variable as an
independent or a dependent variable may change during the model development process,
because the relationship between the two sets of variables may significantly affect the
accuracy of the control model's recommendations.
The control model accuracy is evaluated differently than that of a prediction model. In
developing a prediction model, the accuracy of each succeeding model may be determined
by loading a validation data set and comparing the predicted and actual values. However,
in a control model, the input variables are being adjusted so that the outputs will achieve
their optimum values. As a result, it is not possible to take an existing data set and run it
through the model to determine accuracy.
The only way to truly evaluate the control model's effectiveness is to implement the
recommendations on the actual process, but it is not practical to do this with each
succeeding control model produced during model development. Thus, it is necessary to
use engineering judgement to determine if the set point changes recommended by the
control model are consistent with boiler optimization practices, and make practical sense.
For instance, if the model suggests that a small change in the level of excess oxygen will
produce a ten percent heat rate improvement, it is obvious that the model is not correct.
However, if the model shows that a decrease in the level of excess oxygen will produce a
small reduction in both heat rate and NOX emissions, it is more accurate because the
recommendation is consistent with boiler optimization experience.
There are a number of ways to affect and improve the model accuracy. Input variables
may be removed or replaced, or reclassified from independent to dependent, or vice versa.
New variables can be developed using data transforms, which may allow the software to
get a different "look" at the data, and possibly identify a pattern that was previously
missed. Finally, the constraints that are placed on the independent variables, and the
optimization scheme for the output variables may be adjusted to yield better results.
The initial control model for the Ferguson station had a total of thirteen inputs and four
outputs. These included net heat rate, NOX emission rate, CO2 mass emission rate and the
windbox-to-furnace differential pressure. Only one of the inputs was classified as a
-------
dependent variable and the rest were independent. From experience, this was not a good
balance, and during model development, a number of the inputs were changed from
independent variables to dependent variables. It also became apparent that it would be
necessary to include the fuel flow rate as an output so that it could be minimized as part of
the overall optimization.
The final control model included these same five outputs. It also had twelve inputs, of
which seven were independent variables and five were dependent. Both burner tilts and
throttle pressure were included as dependent variables, even though they may be used to
control the process. This was because the burner tilt settings were driven by steam
temperature, and the throttle pressure was a function of the load. The model inputs and
outputs are summarized below:
Table 3
List of Control Model Variables
Independent Variables
Excess Oxygen
Net Generation
A Level Burner Status
B Level Burner Status
C Level Burner Status
Dependent Variables
Main Steam Temperature
Burner Tilts Position
Main Steam/Reheat Steam
Differential Temperature
Degrees to Saturation
Throttle Pressure
Output Variables
Net Heat Rate
Fuel Gas Flow Rate
Windbox/Fumace
Differential Pressure
NOX, Ib/hr
CO2,lb/hr
D Level Burner Status
E Level Burner Status
Implementation And Results
The control system at Ferguson is not capable of supporting a closed-loop control model.
It is not possible to track and manipulate the control variables that are necessary for the
models that were developed. For example, the burner control system is separate from the
Honeywell system, and burners must be taken in and out of service manually. As a result,
it is not possible to install the control model and document the resulting improvements in
performance. In this case, the model's recommendations had to be implemented by
changing the way in which the operators configure the boiler at different loads. The
following sections describe how the model's recommendations were implemented by plant
personnel, and the resulting changes in emissions and heat rate.
-------
Implementation
As described previously, it was not possible to install a closed-loop control model on the
Ferguson boiler. Therefore, the operational changes recommended by the model were put
in place through a series of modifications to the control system logic for excess oxygen set
points and burner tilt position, combined with changes in the order that the operators took
burners in and out of service.
The Ferguson Plant is a load control unit routinely operating throughout the entire load range.
To follow the recommended change for excess O2 curve, it was necessary to make the
following changes to the control system logic:
1. The low air flow alarm was moved from 27.5 % to 26 %.
2. The excess O2 flow alarm was moved from 1.0 % to 0.5 %
3. The excess O2 controller logic was modified (Figure 1).
As seen in the figure, the model's recommendations called for a reduction in the excess
oxygen set point. This was especially true at the low loads, where a change in the burner
firing configuration allowed the boiler to operate with much less excess air flow.
Figure 1
8 -I
7 .
6
5 •
4 •
3
2
1 •
0
C
EXCESS 02 RECOMMENDATION
*
i -.
4
N
v
X
k
V
X
X
X
»v
••^
fc-~
) 60 75 100 125 150 175 200 225 250 275 300 325 350 375 400 420
-------
A new sequence for taking burners in and out of service was also recommended as part of the
boiler optimization. The effect of the recommendations was to push the fireball further up
into the furnace, which allow the boiler to operate with lower levels of excess air at lower
loads. The recommendations for the new burner configurations are detailed in Table 4.
Table 4
Recommendation for Burner Operation Starting at Minimum Load
Burners in Service: All D level and V* of C level
As load increases:
Add other'/2 of C level
Add B level
Add A level
Add E level
As Load decreases:
Remove E level
Remove A level
Remove B level
Remove Vi of C level
Since the recommended changes to the burner firing configuration caused the flame to be
pushed higher into the firebox, it was necessary to make some modifications to the burner
tilt positions to maintain steam temperature at acceptable levels. The changes to the
burner tilts were programmed into the control system logic so that it would be possible to
run the boiler in automatic control. The changes that were made to the burner tilt control
logic at various loads are listed in Table 5.
-------
Table 5
Burner Tilt Curve Recommendation
MEGAWATTS TILT POSITION IN DEGREES
70 -15
80 -10
90 -5
100 0
110 5
120-160 10
180-210 5
240 0
260 -5
300 -10
320-400 -12
Results
Once the recommended changes to the boiler operation were implemented, it was possible
to determine their effect on boiler emissions performance. The predicted emissions for
CO2 and NOX were compared to the values from the existing CEMS to determine their
accuracy. Improvements in the boiler heat rate were calculated by comparing the
optimized operation with the baseline heat rate.
The results from the comparison of the emissions predictions showed that the software
was able to accurately predict emissions through the entire load range. Figure 2 presents
a plot of the actual CO2 emissions compared to those predicted by the model. As seen in
the figure, it is difficult to differentiate between the two curves. Figure 3 shows the results
of the predicted NOX emissions versus the actual NOX emissions measured by the plant
CEMS. Again, the two curves agree quite well.
-------
CO2 Prediction Model
500000
450000
400000
350000
300000
250000
200000
150000
100000
50000
0
Fg^
1 230 459 688 917 11461375160418332062229125202749
Figure 2
NOX Prediction Model
1 227 453 679 905 11311357158318092035226124872713
Figure 3
-------
A series of tests were conducted at seven load bins to determine the effect of the
recommended operating set points on the boiler heat rate. In these tests, the boiler was
first operated for a time at the baseline condition. The recommendations were then
implemented to show the optimum performance, and the change in the amount of fuel
required to produce a MWh was documented. The reductions in emissions and fuel costs
as a result of implementation are detailed in Table 6. These numbers are based on a 1996
load profile with an estimated gas price as an example .
Table 6
Table of Results from Model Validation Testing
Load 1996 Hours
Bin in Load Bin
60-70
71-120
121-180
181-270
270-330
331-391+
1428
1173
1047
1940
1794
518
Fuel Reduction
MMBtu*
7100
7900
10600
17800
29600
3000
Fuel Savings
(@$2.37)
$16,800
$18,700
$25,100
$42,200
$70,100
$7,100
NO t Reduction
Ib/hr
12852
-63342
14658
153260
157872
81844
COj Reduction
Ib/hr
3194436
-1414638**
1376805
1879860
2868606
254338
Total
7900
76000
$180,000
357144
8159407
*Fuel Reduction was calculated by taking the fuel flow in MCFH, assuming 1020 Btu/CF to get MMBtu/hr, and
then multiplied by hours to give MMBtu in each load bin.
"Apparent negative value for CO2 is due to slight load increase from baseline test to optimized test condition.
Therefore, overall CO2 increased, although the amount of Cp per MWh actually decreased (as seen in the
reduction in MMBtu).
As seen in the table, the optimized condition resulted in a significant decrease in overall
emissions and fuel costs. Even the negative numbers shown in the second load bin are an
anomaly, because the boiler load drifted up slightly between the baseline and the optimized
tests. Therefore, although the total emissions increased, they actually decreased on a per
MWh basis.
The evaluation of the model's recommendations showed that overall, the CO2 emissions
(and unit heat rate) were decreased by 0.5 percent. The decrease in overall NOX emissions
were determined to be 15 percent. These results clearly show the potential benefits of
using advanced software tools for optimizing boiler operations.
-------
Future Work
The results of the greenhouse gas reduction project at the Ferguson station have
demonstrated the usefulness of neural net models for optimizing the combustion process in
a utility boiler. However, there is a significant amount of unrealized potential for further
improving both the emissions rates and the unit heat rate, which are currently being evaluated.
These future possibilities for optimization include closed-loop control models, improved air
damper controls, and feedwater heater optimization. Implementation of these improvements
are dependent upon an upgrade to the existing distributive control system.
Foxboro I/A
The potential of the neural network models at Ferguson has not been fully realized due to the
lack of a state-of-the-art distributive control system. LCRA is currently upgrading all of its
units to a Foxboro I/A control system, and is planning for implementation at Ferguson no
later than 2001. Once installed, the plant will be able to access and to manipulate all of the
instrumentation that will be required to perform the additional optimization tasks described
in the following paragraphs.
Closed Loop Modeling
The current DCS at Ferguson does not allow the facility to receive the full benefit of the
optimization models. Implementation of the control model's recommendations has been
accomplished at the lowest possible level of technology. That is, the optimum operation of
the boiler, as identified by the model, has been divided into load bins, and used to make
changes to the excess oxygen set points and burner configurations used by the operator. A
higher level of technology would be an open-loop, or advisory, control model, in which the
model would provide real-time recommended set points to the operator's screen. The
operator could then implement those changes as often as possible, which would improve
performance in two ways. The first is due to the fact that the safety margin has to be greater
when the recommendations are not real-time. The second is because it is not necessary to
determine an average optimum condition across the load bin, thus improving performance
within a load bin, and producing a higher degree of optimization.
The highest level of implementation would be to install a closed-loop control system. In this
mode, the optimization model would evaluate the current requirements and operating
constraints for the boiler, and would automatically adjust the control set points to maintain
the optimum performance at every operating condition. Since it is not practical for the
operator to make changes to a number of control set points on a minute-by-minute basis, the
closed-loop control system would allow for full implementation of the recommendations, and
full realization of the potential of the boiler combustion optimization model.
-------
Air Damper Controls
The stoichiometry of the combustion process affects both the boiler emissions, and the boiler
heat rate. Installation of automated air damper controls would balance the airflows within the
boiler, and alter the shape of the fireball to produce the optimum heat release within the
furnace. Automated air damper controls could also be utilized to adjust the combustion
stoichiometry on smaller scales, and produce localized regions of fuel-rich flames, followed
by lean burnout. By staging the combustion in this manner, the boiler NOX emissions could
be minimized. In addition, the Ferguson unit is limited at the low end of the operating range
by the fixed amount of air that passes through the windbox below a certain load. Automated
controls would allow the airflow to be decreased and possibly decrease the minimum load
rating for the boiler, thereby producing significant savings during off-peak operation.
Feedwater Heaters
Feedwater heating is another area in which a series of units expend a significant amount of
energy to achieve the desired temperature. At Ferguson, there are seven feedwater heaters
in all, including two high-pressure heaters and five low-pressure heaters. Heat rate
improvements may be realized by optimizing the level control of individual heaters.
Additional gains may be achieved by optimizing the overall feedwater heating system.
Summary
Deregulation of the utility industry will redefine how electricity is generated for the
foreseeable future. The trend for emissions of criteria pollutants is towards tighter control and
more stringent requirements. The combination of the need to produce electricity more
economically and with reduced emissions is a serious challenge. To meet this challenge and
to be environmentally responsible we must pursue the practical, workable and affordable
solutions.
Advanced control systems coupled with advanced software will be important tools in the
future. The results of this project have demonstrated that neural network modeling, combined
with an understanding of the boiler process, may be used to lower the cost of producing
electricity, while at the same time reducing emissions of greenhouse gases and criteria
pollutants.
Acknowledgments
The authors would like to acknowledge the support of the EPA for co-funding the project.
We would also like to express our thanks to the LCRA project manager, Randy Tobey, who
was instrumental in the success of the project. In addition, we would like to express our
appreciation to Mark Johnson, Ray Treadway, Ferguson Operations, and the LCRA
management. Finally, thanks to Mark Hebets and Trish Tarbell of Radian International for
their help in setting up the computer platform for the modeling, and for their assistance in the
modeling efforts.
-------
GNOCIS
A PERFORMANCE UPDATE ON THE
GENERIC NOX CONTROL INTELLIGENT SYSTEM
John Sorge
Southern Company
600 North 18th Street
Birmingham, AL 35202-2625 USA
john.sorge@scsnet.com
Rick Squires
PowerGen
Ratcliffe-on-Soar
Nottingham NG11 OEE UK
Rick.Squires@powertech.co.uk
George H. Warriner
Radian International, LLC
1979 Lakeside Parkway, Suite 800
Tucker, GA 30084 USA
George_Warriner@radian.com
Jeff StaUings
Electric Power Research Institute
3412 Hillview Ave.
Palo Alto, CA 94304-1395 USA
jstallin@epri.com
Abstract
The Generic NOX Control Intelligent System (GNOCIS) is an on-line enhancement to
existing digital control systems or plant information systems that is targeted at
improving unit performance while meeting or improving other operational constraints
such as NOx emissions and fly ash carbon content. The GNOCIS methodology utilizes
a model of the combustion characteristics of the boiler that includes NOX emissions and
boiler performance. The software applies an optimizing procedure to identify the best
setpoints for the plant, which are implemented automatically without operator
intervention (closed-loop), or, at the plant's discretion, conveyed to the plant operators
for implementation (open-loop). GNOCIS development was funded by a consortium
consisting of the Electric Power Research Institute, PowerGen, Radian International,
Southern Company, UK Department of Trade and Industry, and US Department of
Energy. After a brief review of the GNOCIS technology, a summary of several ongoing
GNOCIS projects will be presented.
Introduction
Deregulation of the industry has forced electric utilities to improve operating
efficiencies of their units in an effort to reduce overall operating cost and become more
competitive. Also, passage of the 1990 Clean Air Act Amendments has challenged US
electric utilities to reduce nitrogen oxide (NO*) emissions and to maintain these low
emission rates during day-to-day operation. Boiler efficiency, fly ash carbon-in-ash
(CIA or LOI), and NOX emissions are strongly influenced by a number of controllable
and non-controllable operating parameters. Due to the combustion complexity and
high coupling of a number of important process parameters associated with boiler
-------
Combustion
Models
Software
•Supervisory
•Communications
•Archiving
•Safety Constraints
combustion — especially for pulverized-coal-fired
units — it is difficult to obtain an optimum or even
acceptable operating point (EPRI, 1993). When
one operating parameter is improved, another is
usually adversely affected. Therefore, continuous
delicate balancing is needed to maintain the
optimum over a wide operating range and for
extended periods. The difficulty in optimization is
compounded on units with low NO* combustion
technologies installed.
Description of GNOCIS
GNOCIS™ (Generic NOx Control Intelligent
System) is an enhancement to digital control
systems (DCS) targeted at improving utility boiler
efficiency and reducing emissions. GNOCIS is
designed to operate on units burning gas, oil, or
coal and is available for all combustion firing
geometries. The major elements of GNOCIS are
shown in Figure 1. The basic elements of GNOCIS are described below.
Optimizer
DCS/DAS Integration
•Operator Graphics
•Configuration Modifications
•Implementation
•Safety Constraints
Unit
Plant
Operators /
Engineers
Figure 1
Major Elements of GNOCIS
Combustion Models. Modeling of the furnace is a critical element of GNOCIS. Since
all optimization techniques make use of models (either local or global) of the process in
developing recommendations, the veracity of the process model is highly important for
the success of the optimization. GNOCIS utilizes neural networks for the combustion
model (Beale, 1990) (NeuralWare, 1993).
The combustion models are usually developed in two steps. The first step is the
development of predictive models of the combustion process. Given the combustion
process, predictive models are created using a subset of the measurable inputs and
outputs of the process. The inputs may consist of both controllable parameters (such
as valve positions) and non-controllable parameters (such as ambient temperature or
fuel quality).
Although predictive models are useful tools, what is required in GNOCIS are control
models. A predictive model is designed to predict outputs given a set of inputs, but a
control model must be designed to work in reverse—to predict inputs given a set of
desired outputs. To predict the inputs effectively, a more complex structure is more
appropriate. This structure is necessary since not all important inputs to the
combustion model are controllable, and if controllable, they may not be independent.
A critical element of the control model design is the selection and assignment of the
various inputs to the controllable (or manipulated), non-controllable, and dependent
-------
classes. In many cases the partitioning is non-intuitive. Also, consideration must be
given to the accessibility of the parameter within the DCS in a closed-loop installation
or the ability of the operator to manipulate the control variables in an open-loop
installation.
The flexibility of the modeling approach utilized in GNOCIS permits rapid
development and modification of the combustion models. Although process variables
utilized are very boiler dependent, variables that have been modeled include NOx, CO,
opacity, LOI, boiler efficiency, heat rate, and furnace temperatures.
Optimizer. Optimization is the process by which a performance index is minimized (or
equivalently, maximized) by the manipulation of one or more independent variables for
which the performance index is a function (Dixon, 1972) (Press, 1988). GNOCIS utilizes
a general, non-linear constrained optimizer with capabilities to handle disjoint feasible
regions. The latter feature enables GNOCIS to make recommendations concerning
operating conditions such as whether a mill should be removed or placed into service.
Several factors were considered in this selection:
• GNOCIS is designed to be part of a supervisory control structure, and therefore
dynamic optimization was not a consideration.
• The combustion process is generally highly non-linear.
Constraints are needed for inputs, outputs, and derived functions.
Typical Optimization Scenario
Maximize boiler efficiency
While
{Maintaining NOx below 0.45 Ib/Mbtu
Maintaining LOI below 5 percent}
Using
{Mill biasing
Excess oxygen
Overfire airflow)
A "typical" optimization scenario that can be
readily configured in GNOCIS may be stated
as shown in the box to the right. Further
constraints could also be placed on the control
variables such that only certain mills are to be
considered in the optimization. Also, plant
staff can easily change the goals through the
operator screens.
Digital Control System/Data Acquisition System Integration. GNOCIS is designed
to be either integrated with a DCS, providing closed-loop optimization, or as an open-
loop advisory system interfaced with a DCS or DAS. Plant data is collected via a DCS
or DAS and passed on to the GNOCIS host platform. Recommendations are then
conveyed to the operator either through recommendation screens on the DCS/DAS or
screens built in the GNOCIS host platform. The operator can then implement the
recommendations, either manually or through the DCS. A closed-loop implementation
-------
is shown in Figure 2. In
this configuration,
GNOCIS gathers unit
operating data via the
DCS, and calculates the
optimum setpoints. The
optimum setpoints can
be implemented by the
operator or
automatically by the
DCS.
Approved
Open-Loop Setpoints
GNOCIS Workstation or PC
Figure 2
GNOCIS Implementation Structure
The recommendations
provided by GNOCIS,
whether open- or closed-
loop, are supervisory in nature and are ideally implemented via the DCS. Therefore,
many facets of a GNOCIS implementation are involved with the modification and
upgrade of the DCS to implement the recommendations.
Operator Graphics. The operator displays are the principal interface to GNOCIS.
These displays (1) convey to the operator the recommendations and predicted benefits
and (2) allow the operator flexibility in setting constraints. An example of a GNOCIS
operator screen is shown in Figure 3. As shown, the operator is presented with the
current operating conditions and two sets of recommendations and predictions. One
set corresponds to the current mills-in-service operating condition. If accepted, the
operator can either implement the recommendations by individually setting the
manipulated parameters to the
targets or have the DCS
automatically implement the
recommendations (Implement
Recommendations).
When damped, the independent
parameter is assumed to be
unavailable for optimization
purposes and is set to the current
operating condition. The
optimization is then performed
with the remaining parameters.
The operator can remove or add
parameters from the optimization
by using this screen (Clamped /Free).
Figure 3
Example of GNOCIS Summary Screen
-------
Since in many instances the mill selection can affect performance and emissions, it is
important to provide recommendations concerning the mills in service. However, due
to many externalities immeasurable by the DCS or best judged by the operator, the mill
configuration should not be automatically implemented. As a compromise, another set
of recommendations is provided as to the optimum mills in service and the
performance/emissions benefits. Given the predicted improvement and the current
state of the plant, the operator can decide whether it is of overall advantage to change
the mills in service. Closed-loop mode, if implemented, can be toggled with Open Loop
by selecting the Close Loop / Open Loop button from this screen.
As discussed previously, the constraints and objective function implementation in
GNOCIS is very flexible. A subset of this functionality is accessible via an operator
graphic. High and low limits can be placed on both the controllable parameters
(manipulated variables) and outputs. Hard constraints (cannot be violated) are used
for the former, whereas soft constraints (can be violated but with a penalty applied to
the objective function) are used for the latter.
Configuration Modifications. In order to obtain the full benefits of GNOCIS,
modifications must usually be made to the DCS configuration, particularly for closed-
loop implementations. However, whether open- or closed-loop, GNOCIS
recommendations are considered supervisory in nature, and in most cases, setpoints or
deviations from design curves are recommended. The level of complexity of the
modifications is dependent on the desired integration of GNOCIS into the DCS and falls
into three broad categories: addition of I/O blocks, implementation of GNOCIS
recommendations, and validity checking (Table 1).
Table 1
Summary of DCS Configuration Modifications
Open-Loop
Operator DCS Closed-Loop
Implemented Implemented
Addition of I/O Blocks ^ •/ •/
Implementation of GNOCIS Recommendations n/a ^ •*
Validity Checking n/a n/a •
GNOCIS Projects
GNOCIS projects and demonstrations are underway at a number of sites (Table 2).
Initial trials of GNOCIS were conducted at PowerGen's Kingsnorth Unit 1 in 1994 and
Alabama Power's Gaston 4 in 1995. Since that time, several other projects have been
initiated. As shown, the projects span most of the major firing geometries, fuel, and
digital control system types. Several of these projects are discussed in the following
paragraphs.
-------
Table 2
GNOCIS Sites / Projects Underway
Unit
Kingsnorth 1
Gaston 4
Kingsnorth 3
Hammond 4
Nelson 4
Cheswick 1
Wansley 1
Branch 3
Gaston 3
Kingston 9
Watson 4
Watson 5
Wansley 2
Company
PowerGen
Alabama Power
PowerGen
Georgia Power
Entergy
Duquesne
Georgia Power
Georgia Power
Alabama Power
TVA
Mississippi Power
Mississippi Power
Georgia Power
Date Inst.
YY QQ Type
94 Q4 OL
95 Q2 CL
94 Q4 OL
96 Q2 CL
97 Q3 CL
97 Q3 CL
97 Q3 CL
97 Q3 CL
97 Q3 CL
97 Q4 CL
97 Q4 MS^CL
97 Q4 MS^CL
97 Q4 MS^CL
Boiler
Type
T
W
T
W
W
T
T
W
W
T
W
W
T
Fuel
C,O
C
C,O
C
G
C,G
C
C
C
C
C
C
C
DCS
Type
Cutlass
L&N Max 1000
Cutlass
Foxboro I/A
L&N Max 1000
West.WDPF
Foxboro I/A
Foxboro I/A
L&N Max 1000
Foxboro I/A
Bailey Infi 90
Bailey Infi 90
Foxboro I/A
Size
MW
500
270
500
500
540
570
865
480
270
200
250
500
865
6310
Inst. Type
OL - Open-loop
CL - Closed-loop
MS - Model study
Boiler type
T - Tangential-fired
W - Wall-fired
C - Cyclone
Fuel
C-
G-
Date
Coal
Natural gas
- Initial testing date
Kingsnorth Units 1 and 3. Kingsnorth Power Station, owned and operated by
PowerGen pic, is on a coastal site on the Medway estuary in Kent, UK. It has four
500 MW tangential-fired units which were commissioned between 1970 and 1973. It is
supplied with coal by sea from a range of domestic and international sources. It was
selected for the trials because (1) it had the worst carbon-in-ash problem of PowerGen's
five 2000 MW stations, (2) it had a suitably modern DCS to which GNOCIS can be
interfaced, and (3) the very wide range of coals it uses makes it a particularly
challenging implementation. GNOCIS was initially installed on Unit 1 as part of the
original developmental program. Due to unavailability of Unit 1, GNOCIS was later
added to Unit 3 to continue the test program.
The boilers were constructed by NEI International Combustion to be capable of meeting
full load on either pulverized coal or residual fuel oil. Each unit is fitted with only five
mills, all of which are required to achieve full load on most (but not all) of the coals
supplied to the station. Each furnace is fitted with a low NOx concentric firing system
with separated overfire air. The coal 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 speeds 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. The station has a NOx emission limit of 390 ppm at 6 percent Oi, set in agreement
with Her Majesty's Inspectorate of Pollution. These emissions are not, however, subject
to statutory continuous monitoring.
-------
Low NOx Optimisation
GNOCIS was used to
optimize NOx emissions and
carbon-in-ash. Controllable
parameters used for the
optimization include feeder
speeds (5), excess oxygen, and
burner tilts (Figure 4). The
actual objective function that
GNOCIS sought to minimize
varied in the trials. However,
the constraints that the
optimizer had to obey were
common. Specifically,
advised values for
controllable inputs should not
be outside their operational
limits, load must not decrease,
and mill speeds should be
between 800 and 2000 rpm or
they should be zero (that is, the mills each have a dead-band). The recommended
control settings that GNOCIS produces are passed back to the DCS, where they pass to
the operator via a display on the unit control panel (Figure 4). The panel shows current
values for the variables, their upper and lower constraints, and the GNOCIS
recommendations.
recommendation/
preolction
Figure 4
Operator Interface at Kingsnorth
Numerous tests were conducted during the course of the trials at this site which lasted
from December 1994 through December 1996. For example, for several tests, GNOCIS
was instructed to produce the best set of inputs which would keep NOx below its
statutory limit and minimize carbon-in-ash. Figure 5 shows the result of one such trial,
where a 4 percentage point reduction in carbon-in-ash was obtained at the small cost of
a 10 ppm rise in NOx (but still well below the statutory limit of 390 ppm at 6 percent
oxygen).
To demonstrate GNOCIS's flexibility and to show that it could cope with other objective
functions, a further test was undertaken. In this, the optimizer attempted to reduce
NOx while containing any increase in carbon-in-ash. Figure 5 shows the success of this
test: NOx fell from 350 ppm to 325 ppm with barely any change in the carbon-in-ash,
which stayed at 12 percent.
GNOCIS is currently available to the operators on both Unit 1 and 3 and is used as
needed to provide operational recommendations.
-------
Minimize LOI
Minimize NOx
300 320 340 360 380 400
Figure 5
Example Results from Kingsnorth
Gaston Unit 4. Alabama Power's Gaston Unit 4, along with Kingsnorth Unit 1, was a
development site for GNOCIS. Gaston Unit 4 is a 270 MW pulverized-coal unit. The
Babcock and Wilcox (B&W) opposed-wall-fired 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. The unit is equipped with B&W XCL low NOX burners and six
B&W EL-76 ball and race mills. Fuel is delivered to the mills by two-speed table
feeders. The unit has two forced-draft fans, six primary air fans, and two flue gas
recirculation fans. Combustion air is heated with Ljungstrom air preheaters. The boiler
control system for Gaston Unit 4 is a Leeds and Northrup MAX 1000 distributed digital
control system.
The original objective at Gaston was to implement an open-loop, advisory system with
no immediate plans to migrate to closed-loop operation. However, during the course of
the project, it was determined that there were significant benefits, both in performance
and ease of use of the system, if upgrades were made to GNOCIS to enable closed-loop
operation. These enhancements also give the operator an easier way to implement
open-loop recommendations.
The informational flow for the GNOCIS implementation at Gaston is similar to that
shown in Figure 2. All process data is collected through the DCS and passed on to the
GNOCIS host for calculation of the recommendations. These recommendations are
then conveyed to the operator via the DCS operator displays, similar to those shown
before, except residing on the MAX operator stations. If acceptable, the operator can
then implement these changes through the DCS operator displays. Also, the operator
has the option of running GNOCIS closed-loop with recommendations automatically
implemented.
-------
Testing of GNOCIS was
conducted during second quarter
1995 and was completed third
quarter 1996. Open-loop tests *
conducted as part of the |
developmental program indicated |
that GNOCIS was able to improve
boiler efficiency by approximately
0.5 percentage points and reduce
LOI by approximately 3
percentage points when this was j
the objective. When used to
minimize NOX/ reductions of '
nearly 15 percent were obtained
(Figure 6). Following completion
of the formal test developmental
program, the site conducted some
intermediate load tests during
December 1996, the results of
which are shown in Figure 7.
GNOCIS is currently operating in
closed-loop mode at this site.
Hammond Unit 4. The GNOCIS
project at Georgia Power's Plant
Hammond was undertaken as part of
a US Department of Energy
Innovative Clean Coal Technology
program being conducted at this site.
The overall project provides a
stepwise evaluation of the following
NOx reduction technologies:
Advanced overfire air (AOFA), Low
NOx burners (LNB), LNB with
Efficiency Improvement
NOx Reduction
0 3 4
0 25
y
iXVA
V\A
^
™
Nsi.
^K
-- ^-
3&0.
v^
«*^
9:1 5 9:30 9:45 1 0:00 1 0:1 5 1 0:30 1 0:45 1 1 :00 11:
Time
Actual
Figure S
Open-Loop Testing at Gaston 4
Figure 7
Closed-Loop Testing at Gaston 4
AOFA, and advanced digital controls and optimization strategies. GNOCIS is being
demonstrated as the advanced control/optimization technology.
Hammond Unit 4 is a Foster Wheeler Energy Corporation (FWEC) opposed wall-fired
boiler, rated at 500 MW. The balanced draft unit is equipped with a coldside ESP and
utilizes two regenerative secondary air preheaters and two regenerative primary air
heaters. Six Babcock & Wilcox MPS 75 mills supply pulverized eastern bituminous coal
to twenty-four FWEC Controlled Flow/Split Flame (CF/SF) low NOX burners. The
unit is also equipped with a FWEC designed Advanced Overfire Air (AOFA) system.
-------
The boiler control system for Hammond 4 is a Foxboro I/A distributed digital control
system.
From project inception, the goal of the GNOCIS installation at Hammond has been to
implement a closed-loop, supervisory system. The Foxboro DCS, installed in 1994,
included configuration enhancements that facilitated incorporation of GNOCIS into the
overall control strategy. As at Gaston, all operator interactions with GNOCIS are
through the DCS operator displays. The GNOCIS host platform at this site is a
Windows NT workstation networked to the DCS.
Testing of GNOCIS at Hammond 4 began during February 1996 and continues to date.
The test program has been hampered by general unavailability of the unit; however,
tests to date have been encouraging. Test 158 is a representative example (Figure 8).
This test was conducted with the unit off economic dispatch and at 480 MW. The
purpose of the test was to evaluate the performance of GNOCIS in regards to boiler
efficiency improvements as GNOCIS was made sequentially less constrained. As
shown, nominal boiler efficiency was near 87.5 percent at the beginning of the testing
and with sequential application of the GNOCIS recommendations, an efficiency of
approximately 88.3 percent was attained. As can be seen in the figure,
recommendations for excess oxygen, AOFA damper, and mill loading were
implemented at approximately 11:15,12:10, and 12:45, respectively. Although not
shown, the recommended AOFA damper position is dependent on whether the mills
are included in the
optimization mix which is
indicative of a non-linear
process.
GNOCIS is currently
operating in closed-loop
mode at the site.
Completion of the test
program is expected
during third quarter 1997.
Test 158
No
I\AA
minal
A.
lrf~S*-l/|
02
i""/*-'
I
AOFA ^
j
*J^*
in l
Mill
/M,A
/Hr1)
A_A_
/f^X *
Figure 8
Hammond / Results of Test 158
Cheswick Unit 1. Duquense Light's Cheswick Station is a 570 MW coal-fired
generating station located in Cheswick, Pennsylvania. The unit also has the capability
to co-fire natural gas at up to about 20 percent of heat input. The unit is equipped with
overfire air and low NOx burners. The unit has installed a LOI monitor and CO
monitors in the boiler exit ducts. Cheswick fires a blend of coals (blend based on coal
sulfur content) and is equipped with an on-line coal analyzer in the coal yard.
-------
The GNOCIS demonstration at Cheswick, which is an EPRI funded tailored
collaboration project, has two objectives. These include NOx reduction and heat rate
improvement. In addition, opacity reduction is being investigated. Two optimization
scenarios are being investigated:
• Find the optimum operating conditions (best heat rate) while maintaining NOx and
opacity in compliance with the current operating permit.
• Find the operating conditions that result in the lowest possible NOx emissions with
acceptable penalties on opacity, heat rate, carbon loss, and furnace conditions.
The important control variables that have been identified through both modeling efforts
and actual testing at the plant include excess oxygen, mill biasing, SOFA damper
positions, and gas co-firing level.
The GNOCIS models' outputs are NOx, heat rate, LOI, CO, and opacity. Sensor
validation models have also been developed for all of the model inputs. These were
deemed necessary to ensure reliable model operation at this facility. Frequent sensor
problems are encountered at the plant with oxygen probes (there are six, located at
various positions across the ductwork) and some temperature sensors. The sensor
validation models also estimate NOx concentrations when the NOx monitor is going
through a calibration cycle. This ensures continuous system optimization. The
GNOCIS models run on a Westinghouse WDPF workstation with operator screens
integrated into the DCS. The initial installation is open-loop, with closed-loop
operation scheduled for late summer 1997. The open-loop installation allows "one-
button" implementation of the GNOCIS setpoint recommendations at the operator's
discretion. Model predictions and initial testing indicate NOx reductions of at least 15
percent at full load and heat rate improvements of 0.5 percentage points.
Future enhancements under consideration for the system at Cheswick are to include
additional inputs from the ESP so that opacity predictions and control can be improved.
GNOCIS will eventually receive data from ESPert, an expert system developed by EPRI
for diagnosing precipitator problems (EPRI, 1994).
Nelson Unit 4. Unit 4 at Entergy's Plant Nelson is a 540 MW gas-fired generating unit
located in Lake Charles, Louisiana. It is a B&W wall-fired unit, with 16 two-burner cells
in an opposed-firing arrangement. The unit is equipped with flue gas recirculation.
There are "NOxports" for overfire air, but they are apparently too small and have
minimal impact on NOx.
-------
The EPRI funded GNOCIS demonstration at Nelson will investigate two optimization
scenarios:
• Find the optimum operating conditions (best heat rate) while maintaining NOx in
compliance with the current operating permit.
• Find the operating conditions that result in the lowest possible NOx emissions with
acceptable penalties on heat rate and furnace conditions.
The important control variables that have been identified through both modeling efforts
and actual testing at the plant include excess oxygen, FGR fan damper position, burners
out of service, and FGR bias between the windbox and the hopper. The GNOCIS
models' outputs are NOx, fuel consumption, CO, and furnace draft.
The GNOCIS models run on a PC running Windows NT. The PC is networked to an
application processor in the unit's L&N MAX 1000 DCS. Operator screens are
integrated in the DCS. The initial installation is open-loop, with closed-loop operation
scheduled for late summer 1997. The open-loop installation allows "one-button"
implementation of the GNOCIS setpoint recommendations at the operator's discretion.
NOx reductions of at least 15 percent at full load and heat rate improvements of 0.3
percent are indicated by model predictions and initial testing. Testing of GNOCIS is
now underway at this site.
Wansley Unit 1. Georgia Power's Plant Wansley, located near Roopville, Georgia, is
the site of two 865 MW pulverized-coal units. Unit 1 and Unit 2 are sister units and are
in most respects identical. The Unit 1 Combustion Engineering (CE) tangential-fired
boiler has seven elevations of burners with eight corners (split-furnace) each supplied
by CE 983 RP roller type pulverizers. The balanced-draft unit has two forced-draft fans
and four induced-draft fans. The unit has been retrofit with an ABB CE LNCFS Level 2
separated overfire air (SOFA) low NOx combustion system installed in 1992. The
boiler control system for the unit utilizes a Foxboro I/A system and a data acquisition
system is used for process data collection and storage.
Kickoff for the EPRI funded GNOCIS project at Wansley was in late January 1997. The
project has the primary objective of further verifying the NOx and performance
improvement potential of GNOCIS. The site is also interested in applying GNOCIS to
control fly ash LOI level. Operating parameters to be considered in the optimization
mix include boiler efficiency, LOI, NOx emissions, and opacity. Control variables are
expected to be excess oxygen, mill coal flows (7), and overfire air flow. Major activities
in the project include: (1) GNOCIS host platform installation, (2) pre-installation testing,
(3) installation of a Mark and Wedell on-line carbon-in-ash analyzer, (4) DCS
configuration changes, (5) model development, and (6) post-retrofit testing. GNOCIS is
-------
being installed to be operable in either open- or
closed-loop modes.
A Windows NT workstation hosting the
GNOCIS software was installed during March
1997. Prior to this installation, process data
used to train the combustion models resided on
the plant information servers and was
transmitted to SCS by site personnel who
subsequently transmitted the data to Radian.
With installation of the workstation, the
pertinent process parameters (approximately
100 points) are archived on the workstation
thus streamlining the transmittal of data to
Radian. Based on prior GNOCIS and
combustion experience, a pre-installation test
plan was developed to provide reasonable
coverage of the potential unit operating
envelope and provide process information in
regions where the unit did not normally
operate. Unit operators conducted the actual
testing while the unit was in economic dispatch.
0.6
0.4
0.2
0
| Actual 1
-~_ . ^^ — M jw, n J\1T\ i^V^
0:00 4:00 8:00 12:00 16:00 20:00 0:00
0:00 4:00 8:00 12:00 16:00 20:00 0:00
Tire
Figure 9
Wansley / Predicted vs. Actual
Since the site was concerned about LOI levels, a Mark and Wedell on-line carbon-in-ash
analyzer has been installed on the unit. The system samples from two ports, located on
the "A" and "B" sides of the furnace. The system began initial operation during the
spring; however, to date it is not considered completely operational, and the vendor
continues to refine the installation. After a review of potential control variables with
the site, marked up DCS functional drawings were submitted to the site for
configuration into the DCS by site I&C personnel with assistance from Southern
Company Services. The control strategies and implementation specifics were guided
by the installations at the earlier GNOCIS sites but customized to reflect the specifics of
the boiler and DCS.
Combustion models, both predictive and control, have been created. An example of the
predictive qualities of one model is shown in Figure 9. The data shown represent one
day out of several weeks of data. As shown, the predictive qualities were quite good
for all variables. The models are now being revised to reflect recent refinements in the
control strategy.
Open-loop testing is scheduled to begin in July 1997, with closed-loop testing following
in August 1997.
-------
Branch Unit 3. Georgia Power's Plant Branch, located near Milledgeville, Georgia is
the site of four pulverized-coal units with a total generation capacity of 1750 MW.
Unit 3 is a 480 MW Babcock and Wilcox (B&W) unit equipped with cell burners
(double). Ten B&W EL76 pulverizers supply eastern bituminous coal to the furnace.
The balanced-draft unit has two forced-draft fans and two induced-draft fans. The
boiler control system for the unit utilizes a Foxboro I/A system, and a data acquisition
system is used for process data collection and storage.
The EPRI funded GNOCIS project at Branch began in January 1997. The project has the
primary objective of further verifying the NOx and performance improvement potential
of GNOCIS. Specific operating parameters to be optimized include NOx, boiler
efficiency, LOI, and opacity. Control variables to be used in the optimization include
excess O2 and mill biasing — a total of 11 parameters. GNOCIS is to be open- and
closed-loop capable on this unit. The major activities to date include (1) installation of
the GNOCIS host platform and archival of process data and (2) DCS configuration
modifications. A CAMRAC on-line carbon-in-ash monitor is being installed on the unit
to support the test program and is scheduled to be operational during July 1997.
Pending unit availability, GNOCIS is scheduled to be operational in open-loop mode in
August with closed-loop capability provided in September.
Gaston Unit 3. Gaston Unit 3 is a sister unit to Gaston 4, having the same boiler and
DCS configuration. This project, funded by EPRI and Southern Company, was initiated
during December 1996. A primary goal of this project is to determine potential cost
savings from installation of GNOCIS on a sister unit. Also, since Unit 3 and 4 share a
common stack CEM, enhancements to the predictive qualities of the combustion model
for each unit will be possible by utilizing process data from each unit. The DCS
configuration changes associated with the GNOCIS installation have been completed
and archiving of plant data is in progress. It is expected that the GNOCIS installation
and testing will be completed by third quarter 1997.
Kingston Unit 9. TVA's Kingston Unit 9 is a 200 MW tangential-fired pulverized-coal
unit retrofitted with a Foxboro I/A digital control system. This site, located near
Kingston, Tennessee, is host to the EPRI I&C Center. The split furnace has four levels
of auxiliary air and is supplied fuel from six pulverizers. Potential control variables
include excess oxygen, mill biasing, auxiliary air damper position, and burner tilts. The
GNOCIS installation for this site is now being defined with startup by fourth quarter
1997.
-------
Summary
A summary of the project and the results to date are as follows:
• GNOCIS has been successfully deployed in both open-loop advisory and close-loop
supervisory modes.
• GNOCIS has been able to provide advice which reduced carbon-in-ash and
improved boiler efficiency.
• GNOCIS provided advice which reduced NOX emissions.
• The advice GNOCIS makes is consistent with good engineering judgment.
• Several projects are underway which will further quantify the benefits and costs
associated with GNOCIS.
Acknowledgments
The authors wish to gratefully acknowledge the following for their guidance, support,
and efforts related to the development and demonstration of GNOCIS: Gary
Fotheringham, Rob Holmes, and Ian Mayes, PowerGen; Jim Noblett and Mark Hebets,
Radian International; Bob Kelly, Kerry Kline, and Mike Slatsky, Southern Company.
Also, the support from Rabindra Chakraborty, ETSU, Dave Crockford, DTT, Scott
Smouse, DOE, and Tom Brown, DOE is greatly appreciated. We would like to thank
our respective organizations for permission to publish this paper. Lastly, we would
like to thank the staff at each of the host sites for their gracious toleration of our
frequent requests and for allowing their units to be host sites for GNOCIS.
References
1. Beale, R. and Jackson, T., Neural Computing: an Introduction. Chichester, John
Wiley & Sons, 1990.
2. Retrofit NOx Controls for Coal-Fired Utility Boilers. Palo Alto CA.: Electric Power
Research Institute, 1993. TR-102906.
3. Neural Computing A Technology Handbook for Professional II/Plus and NeuralWorks
Explorer. NeuralWare, Pittsburgh, PA, 1993.
4. Dixon, L., Nonlinear Optimisation. Crane, Russak, & Company, New York. 1972.
5. Press, W., Flannery, B.Teukolsky, S., Vetterling, W. Numerical Recipes in C,
Cambridge University Press, Cambridge, 1988.
6. ESPert: Electrostatic Precipitator Performance Diagnostic Model. Palo Alto CA.:
Electric Power Research Institute, 1994. TR-104690.
-------
Monday, August 25; 3:30 p.m.
Parallel Session A:
Low-NOx Systems for Coal-Fired Boilers (Wall and Tangential)
-------
BURNER MODIFICATIONS FOR COST EFFECTIVE NOX CONTROL
Todd A. Melick
Energy and Environmental Research Corp
1345 N. Main Street
Orrville, OH 44667
Michael E. Hensley
Louisville Gas and Electric Company
820 West Broadway
Louisville, KY 40232
David A. Gustafson
Jamestown Board of Public Utilities
P.O. Box 700
Jamestown, NY 14702
Abstract
The development of commercial Low NO, Burners has provided Energy and Environmental
Research Corporation (EER) with the expertise to modify existing burner equipment to provide
the controlled fuel/air mixing conditions required for low NO, combustion. This approach
represents a viable alternative to a full burner retrofit for many applications. EER has modified
burners to lower NO, emissions at Louisville Gas & Electric's (LG&E) Cane Run Station and at
Jamestown Board of Public Utilities (JBPU). This paper will discuss the method and results of
these burner modifications.
Introduction
With deregulation of the Utility Industry approaching, many utilities are looking for lower cost
alternatives to satisfy NOX regulations. Justifying new low NO, burners on a boiler that is 30-40
years old and has limited remaining life is also difficult. Performing modifications to the
existing burners provides the utility an option. Modifications are usually 2 to 4 times less
expensive than new low NOX burners.
Units which are the best candidates for burner modifications include:
• Older units where the expense of new burners is difficult to justify over the remaining
boiler life.
• Units operating under a system-wide NO, averaging strategy, where compliance on all
boilers is not essential, and where burner modification offers an economical option for
smaller units.
-------
• Units requiring greater than 55% NOX reduction, where burner modification can provide
an economical NO, reduction and then coupled with Rebum, SNCR, SCR, or other
technologies to provide the overall NOX reduction.
Units with first generation low NO, burners, where only moderate NOX reduction is
required.
• Units with conventional burners, firing sub-bituminous or other highly reactive coals.
The modifications to be performed on each project vary widely according to the type of burner,
the NOX reduction required, and site specific information such as coal, burner area heat release
rate, etc. To perform an initial evaluation, EER requests specific site information. After
completing preliminary calculations the next step is usually a windbox inspection of the existing
burners. Some projects then require a reduced scale isothermal modelling study of the existing
burner in order to determine the exact detailed modifications. Other projects that are similar to
previous jobs or only require a small NOX reduction do not require modelling. The goal of
modelling is to determine the specific modifications required to simulate the burner mixing rates
and exit aerodynamics of EER's commercial Low NOX Burner. The hardware modifications are
usually configured so that the existing burner does not have to be removed from the windbox
which is a major advantage when old boilers contain asbestos.
Low NOX Burner Technology
In 1987, EER participated in a joint development program between EUcraft Power, a Danish
utility, and Burmeister & Wain Energi (EWE), a Danish boiler and combustion system
manufacturer, to develop a reliable, high quality, high performance low NOX burner capable of
meeting mandated 50% emissions reduction goals. The burner developed is shown in Figure 1.
The burners have been installed in both Europe and North America. The burner has some similar
components and NOX performance results when compared to other commercial low NOX burners.
However, the mechanical construction of the burner is unique and has several advantages
compared to other burners. The key functional features of the design include:
• Variable combustion air supply through separate secondary and tertiary passages
• Variable swirl on both the tertiary and secondary air
• Flameholder attached to the coal nozzle
Mechanically, the burner has been designed to minimize the number of moving parts. Those
parts which do move, slide axially, eliminating complex linkages and gears. The secondary and
tertiary swirl control vanes, called rurbolators, move back and forth within conical passages of
the burner. As the turbolatcrs are moved toward the narrow end of the cone more air passes
through the vanes increasing the amount of swirl. As the turbolator is moved in the opposite
direction, the air follows the path of least resistance and by-passes the vanes, resulting in less
swirl. The amount of combustion air entering each burner is controlled by a sliding ring damper.
Similarly, the split between secondary and tertiary (outer zone) air is controlled by a second ring
-------
damper. The parts of the burner which are subjected to a high heat flux are fabricated from a
high strength, heat resistant alloy.
The air distribution between the secondary and tertiary zones and moderating the tertian,' air swirl
to lengthen the flame across the available firing depth represents the main variables for control of
NOX emissions. The low primary air/coal velocity and flameholder are designed to provide good
flame stability and acceptable flame characteristics for a wide range of operating conditions and
fuel characteristics. The flameholder establishes local recirculation zones and promotes local
mixing between the coal and the secondary air. This leads to a rapid devolitalization of the coal
and liberation of fuel nitrogen in a low excess air environment resulting in reduced NOX
formation.
Modelling
EER constructed a six tenths scale model of the "RO" type burner for LG&E. The majority of
the burner was constructed of light weight carbon steel, and modelers clay was utilized to create
several different flameholder configurations. Particular attention was directed at specific
components (coal inlet scroll, coal nozzle rifling, air register) to assure that a baseline flow was
established. LG&E personnel observed some of the flow study as it occurred at EER's
Combustion Test Facilities in El Toro, California.
A combination of flow visualization, velocity mapping, and engineering judgement was used to
examine near field aerodynamics of the current "RO" burner design. The burner was then
modified to create the same flow fields as EER's commercial low NO, burner. The study was
interactive as each of the burner modifications was tested independently and then assembled to
assure that the proper flow fields were being created as the air and fuel exited the burner.
Louisville Gas & Electric
Unit #4. Cane Run Station Boiler #4 is a 170 MWe front wall fired CE boiler that went on line
in 1962. The boiler steam flow is 1,200 Klb/hr. The boiler width is 40'-5" and the depth is 26'-
3.5". The boiler has four CE 683 Raymond bowl mills, each of which supplies pulverized coal to
four (4) CE type "RO" burners. The sixteen burners are rated at 100 MMBtu/hr each. The
burner throats are 38 inches. New refractory was installed but new pressure parts were not
required. The burners are configured in four elevations of four burners per row. The horizontal
and vertical spacing of the burners is 8 to 10 feet. New Forney gas igniters, scanners, and a
Burner Management System were installed during the 10-week late winter 1996 outage. A
Honeywell DCS system was also installed to complete the system. The baseline NOX emissions
were 1.2 Ib/MMBru and the current NOX requirements are 0.5 Ib/MMBtu. EER guaranteed 0.48
Ib/MMBtu NOX emissions with the burner modifications. Installation"was completed by LG&E.
The modified burner is shown in Figure 2.
The burners were started up in April 1996. The flames are very stable and a NOX reduction of
greater than 50% has been achieved across the operating range as shown on Figure 3. The boiler
normally operates with an oxygen content of 4.0% or greater at full load to maintain steam
-------
temperature. The data shown in Figure 3 is for an O2 of 4.5-5.0% at full load. The full load NOX
emissions at 4.0% 0, are 0.45 Ib/MMBtu. The typical coal analysis is shown in Table 1. The
pulverizers and exhausters were rebuilt during the outage and the riffle boxes replaced. During
start-up the full load LOI was 6% which is the same as baseline. However, HER recently tested
three of the four mills with our Rotorprobe™ System to check on fineness and fuel distribution.
Both the fineness and fuel balance were below requirements and that has resulted in.
progressively higher LOI since start-up. During May 1997 the pulverizers averaged 2.5%
retained on 50 mesh, 85.7% through 100 mesh, and 64.5% through 200 mesh. EER has provided
an adjustable orifice device (FlowmastEER) shown in Figure 4 to balance the coal flows to each
of the top eight burners. EER will balance the coal flows in August after the FlowmastEERs are
installed.
Unit #5. Cane Run Station Boiler #5 is a 180 MWe front wall fired Riley Stoker boiler that
went on line in 1966. The boiler steam flow is 1,360 Klb/hr. The boiler width is 47'-3" and the
depth is 27'-0". The boiler has three Foster Wheeler ball mills, each of which supplies pulverized
coal to four burners. The twelve burners are rated at 155 MMBtu/hr each. The original burner
throats were 38 inches. The burners are configured in three elevations of four burners per row.
The horizontal and vertical spacing of the burners is 10 to 11 feet. New Phoenix low NOX
burners were installed in 1994. New Forney gas igniters, scanners, a Burner Management
System, and a Honeywell DCS system were installed during that outage. The baseline NOX
emissions were 0.9 Ib/MMBtu with the old FW Intervene burners. The low NOX burners were
operating at 0.8 Ib/MMBtu but the current NOX requirements are 0.5 Ib/MMBtu.
EER modified the existing burner as shown in Figure 5. The existing coal inlet scroll and air
register were utilized to minimize costs. The modifications consisted of a new coal nozzle with
flame stabilizer, a secondary air sleeve, a core air pipe, waterwall panels forming new burner
throats, and a tertiary air sleeve to attach the existing register to the new burner throat.
Installation was completed by LG&E. The burners were started up in November 1996 and the
most recent 30-day average on NOX emissions was 0.50 Ib/MMBtu. EER also provided an
adjustable orifice device (FlowmastEER) to balance the coal flows to each burner. The
precipitator performance was marginal before the low NOX retrofit. After the coal flows were
balanced to 10 of the 12 burners, the operators noticed a reduction in the number of opacity
excursions above 20%. To minimize opacity concerns and assist in maintaining steam
temperature the boiler is operated at full load with an oxygen content of 4.5-5.0% and this results
in LOI less than 4%. Some recent CEMS NOX data is shown on Figure 6.
Jamestown Board of Public Utilities
The Jamestown Board of Public Utilities (JBPU) operates a 50 MWe coal fired electric
generating station. The plant has four pulverized coal fired boilers supplying steam at 850 PSIG
and 900°F to two 25 MWe turbines. The boilers are rated from 150 to 230 Klb/hr steam flow.
Boiler #12. The boiler was installed by Erie City Iron Works (Zurn Industries) in 1967. It is
rated at 230,000 Ib/hr. The furnace width is 17-0" and the depth is 19'-3" The horizontal and
vertical burner spacing is approximately 6 feet square. The unit has two (2) CE 473 Raymond
-------
bowl mills each of which supplies pulverized coal to two (2) CE type "R" burners. The burner
throat is 31.5 inches and the burners are rated at 62.5 MMBtu/hr. The throat refractory was
patched during the outage. Peabody oil lighters and the Burner Management System are
interconnected to a Westinghouse WDPF Control System. Baseline NOX Emissions are 0.9
Ib/MMBtu and flyash unbumed carbon was 4 to 7%. The current New York State DEC
regulation is .45 Ib/MMBtu. To satisfy these requirements, HER installed burner modifications
and may utilize gas rebuming to meet future requirements. The burner modifications occurred
during the July/August 1996 outage when the superheater was also replaced. HER provided a
turnkey project for this boiler.
The burner modifications were very similar to LG&E. HER also installed a FlowmastEER
adjustable orifice device shown in Figure 4 to balance the coal flows to each burner and the coal
fineness is 75% through 200 mesh. At full load the NOX emissions are maintained below .45
Ib/MMBtu with an oxygen content of 3% as shown on Figure 7. Burner flame shaping can be
accomplished by adjusting the air register. The typical coal analysis is shown in Table 2. The
flyash unbumed carbon from ESP hopper grab samples during the first week of operation
averaged 7 to 10%.
Boiler #11. After the successful start-up of Boiler #12, JBPU awarded EER a similar project
for Boiler #11. The boiler is rated at 165,000 Ib/hr steam flow and is very similar to boilers 9 &
10 described below. EER provided the same burner modifications and the start up was April
1997. The current New York State DEC NOX regulations for this boiler is 0.5 Ib/MMBtu. The
boiler is able to satisfy these regulations while operating at 5% O2 levels. The NOX emissions
verses steam flow is shown on Figure 8.
Boiler #9 & 10. The boilers were installed by CE and then retubed with membrane walls in
1990 by CE. Openings for ten (10) overfire air ports on the side walls were installed, but have
not been utilized. The furnace width is 16'-3.5" and the depth is 17'-8". The horizontal and
vertical burner spacing is approximately 5.5 feet. It is rated at 165,000 Ib/hr. The unit has two
(2) CE 453 Raymond bowl mills, each of which supplies pulverized coal to two (2) ABB type
"RO-IT burners. The burner throat is 28 inches and the burners are rated at 50 MMBtu/hr.
Peabody oil lighters and the Burner Management System are interconnected to a Westinghouse
WDPF Control System. Baseline NOX Emissions are 0.6 Ib/MMBtu. The current New York
State DEC regulations for these boilers is 0.5 Ib/MMBtu. To satisfy these requirements, EER has
installed burner modifications.
Based on the LG&E flow modelling, EER fabricated hardware to modify two of the burners on
Boiler #10. The hardware consisted of a modified secondary air sleeve and coal nozzle
tip/flameholder. JBPU fabricated a device to straighten the pulverized coal flow. JBPU
installed the equipment in December of 1995. By utilizing the existing burner register, the on-
line burner adjustments are limited to only varying the flow distribution between the secondary
and tertiary air zones. There is no adjustment of air swirl. The results were promising so EER
fabricated the remaining hardware for the other six burners which were installed in March 1996.
-------
During the first start up one of the coal nozzles was plugged when operating with minimum
primary air flow. No serious damage occurred and the problem has not recurred since the
minimum exhauster discharge pressure has been raised. With the same boiler air flow curve, the
modified burners were operating at 0.4 Ib/MMBtu. To reduce the unbumed carbon, the plant has
increased the boiler air flow such that they operate at approximately 0.48 Ib/MMBtu. Increasing
the secondary air and decreasing the tertiary air has also helped reduce the unbumed carbon, as
has modifying the sootblower schedule in the lower furnace. The burner flames are very stable
and JBPU has been able to operate below 0.5 Ib/MMBtu. The baseline unbumed carbon was 24
to 30%. The modifications have reduced the unbumed carbon to 18-24% while reducing the NOX
emissions. The unbumed carbon values were quite responsive on Boiler #12 when adjusting the
air swirl and it is speculated that the current swirl level is not optimized.
Future Work
EER will perform burner modifications for Dayton Power & Light on Unit #3 at their Stuart
Station in the Fall of 1997. Unit #3 is a 605 MWe B&W universal pressure opposed wall fired
boiler. The unit is equipped with 24 two-nozzle cell burners. The furnace depth is 39 feet. Cell
burners were developed in the 1950s as high heat input, high efficiency burners. Each cell is
composed of two closely spaced circular burners acting as a single unit. Cell burners generate
very high levels of NOX, ranging from 1.1 to 1.8 Ib/MMBtu. EER has modeled the circular
burner for DP&L and determined the burner modifications required to reduce the NOX emissions.
Conclusion
Modifying existing burners is a viable alternative to new Low NOX Burners as shown by these
examples. EER has modified two different burner designs and is currently modifying two
additional burner designs. The modifications have also been scaled over a large range of heat
inputs from a 330 MMBtu/hr burner on a 913 MWe unit to the 50 MMBtu/hr burners at
Jamestown. Modifying existing burners to satisfy NOX regulations does provide a lower cost
option that should be considered.
-------
Figure 1. FlamemastEER™ Burner
-------
a«n> or*
Figure 2. Louisville Gas & Electric Cane Run 4 Burner Modifications
-------
_
O
1.4
1.2
1.0
0.8
0.6
0.4
0.2
0.0
/• »_•
60 80
100 120 140
Load, MW
• Pre Burner Mod
O Post Burner Mod @ 4.5-5% O2
160 180
Figure 3. Cane Run Unit 4 Pre and Post Low NOx Burner Modification NOx vs. Load
-------
€
Fully Open
100% Flow Area-No Restriction
Fully Closed
40% Flow Area-Maximum Control
Figure 4. PC FlowmastEER™
-------
Figure 5. Louisville Gas & Electric Cane Run 5 Burner Modifications
-------
Post Burner Mod @ 4.5-5% O2
0.0 "-J-
0 20 40 60 80 100 120 140 160 180
Load, MW
Figure 6. Cane Run Unit 5 Post Low NOx Burner Modification NOx vs. Load
-------
m
5
S
1.2
1.0
0.8
0.6
0.4
0.2
0.0
• Pre Burner Mod @ 3.3% O2
O Post Burner Mod @ 3.3% O2
8P° °n
150 155 160 165 170 175 180 185 190 195 200
Steam, Klb/hr
Figure 7. Jamestown Board of Public Utilities Unit 12 Pre and Post Burner Modification NOx vs. Load
-------
1.4
1.2
1.0
m 0.8
jQ
0*0.6
0.4
0.2
0.0
*
o
100 110
o
120 130 140
Steam, Klb/hr
• Pre Burner Mod @ 4.1% 02
O Post Burner Mod @ 5.1 % 02
150 160
Figure 8. Jamestown Board of Public Utilities Unit 11 Pre and Post Burner Modification NOx vs. Load
-------
TABLE I
Louisville Gas & Electric Coal Analysis
Ultimate (%) Typical
Moisture (% by weight) 10.48
Carbon (% by weight) 64.06
Hydrogen (% by weight) 4.58
Nitrogen (% by weight) 1.32
Chlorine (% by weight) 0.04
Sulfur (% by weight) 2.91
Ash (% by weight) 9.43
Oxygen (by difference) 7.23
Proximate (%)
Volatile Matter (% by weight) 36.09
Fixed Carbon (% by weight) 44.00
Higher Heating Value (Btu/lb) 11,600
Table II
Jamestown Coal Analysis
PROXIMATE ANALYSIS ULTIMATE ANALYSIS
As Received Dry Basis As Received Dry Basis
% Moisture 7.44 xxxxx % Moisture 7.44 xxxxx
%Ash 9.14 9.87 % Carbon 70.65 76.33
% Volatile 32.30 34.90 % Hydrogen 4.42 4.77
% Fixed Carbon 51.12 55.23 % Nitrogen 1.39 1.50
100.00 100.00 % Sulfur 1.76 1.90
%Ash 9.14 9.87
Btu/lb 12,433 13,432 % Oxygen (diff) 5.20 5.63
MAF Btu/lb 14,903 100.00 100.00
-------
NOX SUBSYSTEM EVALUATION OF B&W'S ADVANCED COAL-FIRED LOW
EMISSION BOILER SYSTEM AT 100 MILLION BTU/HR
Jennifer L. Sivy and Hamid Sarv
McDermott Technology, Inc., Research and Development Division
Alliance, OH
and
John V. Koslosky
Babcock and Wilcox, Utility, Environment, and Industrial Division
Barberton, OH
Abstract
As part of a U.S. Department of Energy-sponsored Combustion 2000 program titled
"Engineering Development of Advanced Coal-Fired Low Emission Boiler System", Babcock and
Wilcox (B&W) has designed and evaluated a NOX control system. At the heart of the NOX
control system is an advanced low-NOx DRB-4Z™ pulverized coal (PC) burner design. The
burner has undergone comprehensive combustion and emissions performance evaluation. Firing
the unstaged burner with a high volatile bituminous Illinois No. 6 coal at 100 million Btu/hr and
17% excess air (1.17 burner stoichiometry) has achieved the program's minimum environmental
performance requirement of 0.2 Ib NOx/106Btu. Coal type variation tests have shown that NOX
formation increases mainly with decreasing fuel volatile matter (increasing rank). Air staging the
burner at 0.86 stoichiometry with all other conditions being the same reduced the NOX levels to
0.121b/106Btu.
Introduction
Under the sponsorship of the U.S. Department of Energy, Babcock and Wilcox is developing an
advanced coal-fired Low Emission Boiler System (LEBS) for commercial application by the year
2000. NOX control is one of the LEBS subsystems. At the heart of the NOX control system is an
advanced low-NOx DRB-4Z™ pulverized coal burner design. Computer modeling and pilot-
scale testing at 5 million Btu/hr (MBtu/hr) were used extensively to refine the burner design for
scale-up. Following the scale-up, the DRB-4Z™ low-NOx PC burner has undergone further
refinements and performance evaluations '"4 at 100 MBtu/hr. This paper discusses the latest
burner performance results and more specifically the effects of coal rank, PC fineness, and air
staging on NOX, CO, and carbon burnout.
-------
Test Facility
Babcock & Wilcox conducts the large-scale prototype burner tests in its Clean Environment
Development Facility (CEDF) in Alliance, Ohio. Figure 1 shows a schematic of the CEDF
furnace and convective pass section. The CEDF accommodates a single 100 MBtu/hr burner for
firing natural gas, fuel oil, or coal. The inside surface of the furnace is refractory lined to
replicate the thermal environment and flow characteristics of a typical utility boiler Water-
cooled tubes are spaced across the convective pass duct to closely simulate the tube metal and
flue gas time-temperature profile of commercial utility boilers. Deposits on the tubes and the
walls are removed by sootblowers. Raw coal is pulverized by a B&W EL-56 pulverizer
equipped with a dynamically staged, variable speed (DSVS™) classifier to control PC fineness.
Preheated air carries the pulverized coal to a small filterhouse that vents the air and drops the PC
into a storage bin. Pulverized coal flow from the bottom of the storage bin is controlled by a
weigh feeder. The coal is then transported to the burner by heated primary air at the desired air-
to-fuel ratio. Typical primary air temperatures are around 150°F at the burner inlet. Secondary
air is preheated by the flue gas and a gas-fired heater to 600 °F.
• Sight/access port
Fly ash
sampling port
Figure 1
Furnace and Convective Pass Section Schematic of the Test Facility.
-------
For staged combustion, part of the secondary air is directed to two opposed overfire air (OFA)
windboxes located just above the furnace tunnel section. Each windbox houses two OFA
registers equipped with outer spin vanes and a central core air damper control. Damper control
and pressure drop indications across the in-duct orifice plates are used to balance the OFA flow
equally to each side of the furnace.
Gaseous species are sampled continuously from the convective pass section outlet through a
heated sample line. After filtering and drying, CO, CO2, O2, and NOX concentrations are
measured by calibrated analyzers. Fly ash is sampled across the duct via a multi-point probe with
equally-spaced holes and analyzed for loss on ignition (LOT). Previous work 4 has shown that
LOI measurements closely approximate the fly ash unburned carbon levels. Flue gas and fly ash
sampling locations are shown in Figure 1. Computerized data acquisition is used to record
species concentrations, flow rates, temperatures, pressures, and other relevant information for
subsequent analysis.
DRB-4Z™ Low-NOx PC Burner
An unstaged, 100 MBtu/hr, DRB-4Z™ low-NOx PC burner was used for testing. This burner
operates on the principle of controlled mixing of air and fuel to minimize NOX emissions. What
sets the DRB-4Z burner apart from other commercial designs is the implementation of special
proprietary features for greater NOX reduction.
Coal Analyses
Four different coals including a subbituminous Powder River Basin Decker, a high volatile
bituminous Illinois No. 6, a high volatile bituminous Ohio Mahoning, and a medium volatile
bituminous Pennsylvania Middle Kittanning were chosen for testing with the DRB-4Z™ burner.
Table 1 lists the proximate, ultimate, and heating value analyses of the as-received coals. Fixed
carbon-to-volatile matter ratios (FC/VM) for these coals ranged from 1.16 to 2.81. Illinois No. 6
was the reference high volatile bituminous coal selected for this program.
Unstaged Combustion Results
Fuel Type and Excess Air Effects
Coal rank and excess air effects on the DRB-4Z™ burner performance were carried out at a
nominal standard PC fineness of 75% through a 200 mesh screen. Table 2 lists the 16 to 200
sieve (1190 to 74 urn) cut sizes of actual coal samples extracted from the PC-laden stream after
the mill and before the filterhouse.
-------
Table 1
Coal Analysis
Proximate
Fixed Carbon (%)
Volatile Matter (%)
Moisture (%)
Ash (%)
Fixed Carbon/Volatile Matter
Ultimate
Carbon (%)
Hydrogen (%)
Nitrogen (%)
Sulfur (%)
Oxygen (%)
Heating Value (Btu/lb)
Hardsrove Grindabilitv Index
Montana Decker
Subbituminous
37.06
31.93
26.36
4.65
1.16
53.64
3.73
0.88
0.51
10.23
9237
47
Illinois No. 6
High Volatile
Bituminous
44.37
33.53
13.92
8.18
1.32
61.96
4.44
1.17
3.02
7.31
11122
54
Ohio Mahoning 7A
Seam High Volatile
Bituminous
54.32
34.49
3.90
7.29
1.57
74.10
5.21
1.30
1.22
6.98
13292
50
Pennsylvania
Middle Kittanning
Medium Volatile
63.90
22.74
3.26
10.10
2.81
76.61
4.54
1.34
0.81
3.34
13476
86
Table 2
Standard Grind PC Size Distributions for Four Different Coals
Mesh Designation
and Size
Screen # (urn)
16(1190)
30 (595)
50 (297)
70 (210)
100 (149)
140(105)
200 (74)
Subbituminous
Decker
High Volatile
Illinois No. 6
High Volatile
Mahoning 7A
Medium Volatile
Middle Kittanning
Percent Smaller
100.00
100.00
99.60
98.90
97.60
87.00
73.20
100.00
99.98
99.80
99.10
95.50
89.00
73.70
100.00
99.98
99.70
99.20
96.20
90.10
74.20
100.00
99.90
99.30
97.60
91.90
84.30
71.60
Figure 2 shows the NOX and LOI results for each coal at optimum burner settings. In all cases,
raising the excess air converted more fuel-N to NOX, and decreased CO formation and LOI. As
expected, NOX levels increased with increasing fuel-N and decreasing volatile matter contents.
But NOX formation is a stronger function of the fuel factor (FCYVM) than the fuel-N content. For
instance, the medium volatile Middle Kittanning coal forms more NOX than the high volatile
Mahoning coal despite having a similar fuel-N content. LOI levels were higher for the lower
volatile and less reactive Middle Kittanning coal.
-------
o
CO
CL
350
300
250
200
150
100
50
1.05
25
20
15
10
0.
0.
49
42
35
28
0.21
0.07
1.10
1.15
1.20
1.25
1.30
O - Medium Volatile Middle Kittanning
D — High Volatile Mahoning
D..-. High .Volatile.Illinois
A — Subbituminous Decker
1.05 1.10 1.15 1.20 1.25
Burner Stoichiometry
1.30
Figure 2
Coal Type and Excess Air Effects on NOX and LOI for the DRB-4Z™ PC Burner. Nominal
Conditions: 100 MBtu/hr and 75% through 200 Mesh Screen PC Fineness.
All flames were stable and attached at the burner throat. CO emissions from the two high
volatile bituminous coals were higher than the subbituminous and medium volatile bituminous
coals. Relative to full load conditions, part load operation at a fixed burner Stoichiometry
generated lower NOX and higher LOI due to the reduced mixing and cooler furnace environment.
Coal effects on unbumed carbon loss (UBCL), and CO and NOX emissions for 100 MBtu/hr and
17% excess air operation are illustrated in Figure 3. Where available, reproducibility of the
results is shown by error bars, representing one standard deviation. UBCL is calculated from the
LOI measurements and fuel analysis as a measure of the unutilized fuel and combustion
efficiency.
-------
c~
t.
x
o
-J
CJ
m
600
2 400
^ 300
0 200
100
0
2.0
— 1.5
1.0
0.5
0.0
00
Subbituminous
Decker
High volatile High volatile Medium Volatile
Bituminous Bituminous Middle Kittannina
Illinois No. 6 Mahoning
Figure 3
Coal Type Effects on NOX, CO, and UBCL for the DRB-4Z™ PC Burner. Nominal Conditions ;
100 MBtu/hr, 11% Excess Air, and 75% through 200 Mesh Screen PC Fineness.
Coal Fineness Effects
Pulverized coal particle size was also varied from 60 to 90% through a 200 mesh screen for the
Illinois No. 6 coal. Representative size distributions are tabulated in Table 3.
-------
Table 3
Representative PC Fineness Variations for the High Volatile Illinois No. 6
Mesh Designation and Size
Screen # ((am)
16(1190)
30 (595)
50 (297)
70 (210)
100 (149)
140(105)
200 (74)
Coarse
Standard
Fine
Very Fine
Percent Smaller
100.00
99.90
99.20
97.30
89.70
79.10
60.60
100.00
99.98
99.70
99.20
96.20
90.10
74.20
100.00
99.98
99.80
99.40
98.00
94.00
84.30
100.00
100.00
99.98
99.90
99.70
98.00
90.50
Figure 4 compares the full load NOX and LOI data from burning various grind sizes of Illinois
No. 6 coal in the DRB-4Z™ burners at 17% excess air level. Increasing the coal fineness
generally improved the fuel oxidation and decreased the CO and LOI levels without an
appreciable change in NOX emissions. In the CEDF, the DRB^Z™ burner generated about 35%
less NOX than other commercially available low-NOx burners when firing the Illinois No. 6 coal.
And although the LOI levels from burning coarse and standard fineness PC in the DRB-4Z™
burner were higher than the values of other commercially available low-NOx burners, the
difference is reduced significantly for the fine grind size (85% through 200 mesh). Average
NOX, CO, and LOI values for the DRB-4Z™ burner from firing the 85% through 200 mesh PC at
17% excess combustion air were 141 PPMV (0.20 Ib NO2/MBtu), 58 PPMV, and 1.71%,
respectively. Because of these favorable results, the proof-of-concept demonstration phase of
this program will also utilize the 85% through 200 mesh screen coal fineness.
Staged Combustion Results
Air staging tests were carried out by firing the standard fineness Illinois No. 6 coal at a fixed
overall excess air level of 17% (3.2% stack O2) in the same burner used for unstaged combustion.
Both, the burner and OFA register settings were re-optimized at a nominal burner stoichiometry
of 0.90. Burner stoichiometry was then varied from 0.86 to 1.17 by splitting the total secondary
air flow between the burner and the OFA ports. Figure 5 shows the effect of burner
stoichiometry on NOX and LOI at three levels of staging for the DRB-4Z™ burner. NOX
concentrations from operating the DRB-4Z™ burners at 0.86 stoichiometry was 90 PPMV (0.12
Ib NO2/MBtu). Raising the burner stoichiometry from 0.86 to 1.17 increased the NOX emissions
by only 54%. LOI levels decreased as the burner stoichiometry was increased. Increasing the
coal fineness to 85% through 200 mesh is also expected to reduce the LOI levels significantly
when staging.
-------
200
150 --
>
S
tx
100
X
o
50 -
6
5s? 5
c 4
3
2
1
0
0.28
0.21
0.14
o
Z
60
75
85
90
0.07
0.00
•-T
60 75 85 90
PC Fineness (% through 200 mesh screen)
Figure 4
Coal Fineness Effects on NOX and LOI for the DRB-4Z™ PC Burner. Nominal Conditions : 17%
Excess Air and Firing Illinois No. 6 Coal at 100 MBtu/hr.
Although the burner was not sized for staging, it maintained the necessary aerodynamics for
minimizing NOX emissions at all stoichiometries. Staged burners are usually designed with
smaller throats to maintain the proper aerodynamics and mixing patterns for minimizing NOX
and LOI levels. Thus, a DRB-4Z™ burner designed for 0.75 stoichiometry is expected to
achieve the program developmental NOX target of 0.1 Ib/MBtu in the test facility without
resorting to post-combustion NOX control techniques. Figure 6 shows the changes in NOX
emissions with burner stoichiometry for the DRB-4Z™ burner.
-------
200
0.28
Burner Stoichiometry
Figure 5
Air Staging Effects on NOX and LOI for the DRB-4Z™ PC Burner. Nominal Conditions : 17%
Overall Excess Air and Firing 75% through 200 Mesh Screen Illinois No. 6 Coal at 100 MBtu/hr.
-------
o
CD
S
CU
CU
150
140 -
130 -
120
110
100
90
80
75
Minimum Environmental Performance Requirement
Developmental Target
0
0
0
0
0
0
0
0
-I 0
0
0
0.7 0.8 0.9 1.0 1.1
Burner Stoichiometrv
0
1.2
.21
.20
.19
.18
.17
2
-*~i
03
.16 S
\
w
.15 i
X!
.14 ~
.13
.12
.11
10
Figure 6
Burner Stoichiometry Effects on NOX for the DRB-4Z™ PC Burner. Nominal Conditions : 17%
Overall Excess Air and Firing 75% through 200 Mesh Screen Illinois No. 6 Coal at 100 MBtuThr.
Conclusions
A NOX control system was developed and tested for implementation in the proof-of-concept
demonstration phase of the Combustion 2000 program. The DRB-4Z™ low-NOx PC burner is
the centerpiece of the NOX control system. Operating the unstaged burner at 100 MBtu/hr and
17% excess air achieved the program's minimum environmental performance requirement of 0.2
Ib NOx/MBtu with the reference Illinois No. 6 coal. Air staging the burner at 0.86 Stoichiometry
with all other conditions being the same lowered the NOX levels to 0.12 Ib/MBtu. Coal type
variation tests proved that the DRB-4Z
commercially available burners.
TM
burner generates consistently less NOX than other
-------
Acknowledgments
This work was sponsored by the United States Department of Energy (Grant No. DE-AC22-
92PC92160). Thanks are due to many people from the various divisions of B&W who
contributed to the successful accomplishment of the program goals.
References
1. J.L. Sivy, L.W. Rodgers, K.C. Kaufman, and J.V. Koslosky, "Development of an Advanced
Low-NOx Burner in Support of B&W s Advanced Coal-Fired Low Emission Boiler
System," Presented at the American Flame Research Committee Fall International
Symposium, Monterey, CA (October 1995).
2. D.K. McDonald, D.A. Madden, and J.L. Sivy, "The Worldwide Applicability of B&W's
Advanced Coal-Fired Low-Emission Boiler System," Presented at the 13th Annual
International Pittsburgh Coal Conference, Pittsburgh, PA (September 1996).
3. J.L. Sivy, K.C. Kaufman, L.W. Rodgers, and J.V. Koslosky, "A Low-NOx Burner Prototype
Developed for B&W's Advanced Coal-Fired Low-Emission Boiler System," Presented at
the American Flame Research Committee Fall International Symposium, Baltimore, MD
(October 1996).
4. J.L. Sivy, K.C. Kaufman, and D.K. McDonald, "The Development of a Combustion
System for B&W's Advanced Coal-Fired Low-Emission Boiler System," Presented at the
22nd International Technical Conference on Coal Utilization and Fuel Systems, Clearwater,
FL (March 1997).
5. T.J. Flynn, A.D. LaRue, and P.S. Nolan, "Introduction to Babcock and Wilcox's 100
MBtu/hr Clean Environment Development Facility," Presented at the Annual Meeting of
the American Power Conference, Chicago, IL (April 1994).
6. Patent pending.
-------
Field Demonstration of
ABB C-E Services' RSFC™
Wall Burner for
Coal Retrofit Applications
D. Norris
Richmond Power & Light
Richmond, Indiana
G. Bittner
O. Briggs
ABB C-E Services, Inc.
Windsor, Connecticut
N. Nsakala
ABB Power Plant Laboratories
Combustion Engineering, Inc.
Windsor, Connecticut
Abstract
This paper outlines the first commercial application of ABB C-E Services' coal-fired
RSFC™ Radially Stratified Flame Core burner for retrofit in wall-fired boilers. The
successful evolution of the burner from the laboratory prototypes and oil and gas
applications to a commercial coal-fired burner is discussed.
Fundamental design and operating features of the RSFC™ burner are examined, which
allow the six burners currently installed in Richmond (Indiana) Power & Light,
Whitewater Valley Station Unit 1 to surpass all emissions guarantees while firing a
midwestern bituminous coal. NOx reduction performance was accomplished without the
use of additional technology such as flue gas recirculation and/or a separated overfire air
system. The burners have been in service full time since the retrofit installation during
October 1996. Operational and mechanical performance of the burners has been
exceptional with installation and commissioning being completed within a three-week time
frame.
Introduction
Richmond Power & Light (RP&L), Whitewater Valley Station provides electricity to the
city of Richmond, Indiana. In 1996 ABB C-E Services entered into a contract with RP&L
-------
to retrofit six coal-fired RSFC™ low NOx burners in place of the original burners into
their Whitewater Valley Unit 1. Whitewater Valley Station Units 1 and 2 share a common
stack equipped with continuous emission monitoring (CEM) equipment. Under the 1990
Clean Air Act Amendments (CAAA) Title IV, both units at the station are under Phase II
compliance regulations. RP&L decided to pursue the option of early compliance for both
the front wall fired Unit 1, and the tangentially fired Unit 2. As a result, the combined
NOx emissions limit for the two units, as measured by the stack mounted CEM is
0.45 Ib/MBtu. After examining various alternatives, replacement of the original burners
with ABB C-E Services RSFC™ low NOx burners was pursued as the most attractive
means of entering compliance without having to perform any pressure part changes to the
boiler. Figure 1 shows the post retrofit NOx emissions compared to the RP&L permit
limitations.
m
0.70
0.60
0.50
0.40
O 0.30
0.20
1990 CAAA Permit Limit
• RSFC™ Equipped NOx
40.0 50.0 60.0 70.0 80.0
Boiler Load, %
90.0
100.0
Figure 1
RP&L Whitewater Valley Operating Limits
Unit Description
Whitewater Valley Station Unit 1 is a 1950s vintage DB Riley Corporation front wall-
fired, balanced draft, natural circulation steam generator with a superheated steam flow of
-------
325,000 Ibs/hr at 900°F and 900 psig. The six burners are arranged in two elevations of
three burners each, and are supplied pulverized coal by a total of three DB Riley Atrita
#550 pulverizers. Initial coal ignition is accomplished by the use of the original No. 2 oil
mechanical atomized oil guns. A single Ljungstrom® air preheater is used to heat the
secondary air from ambient conditions to a windbox temperature of 620°F at full load.
Nominally rated at 33 MWe, the unit is equipped with an electrostatic precipitator that
was added after the initial start up. As a result of the installation of the precipitator,
Unit 1 is operationally constrained by induced fan draft limitations at full load.
Whitewater Valley Unit 1 fires a midwestern bituminous coal from sources in Indiana and
Kentucky. The coal composition is relatively uniform as summarized in Table 1. All of the
parameters reported in Table 1 (FC/VM and O/N ratios and ash and fuel nitrogen
loadings) fall within the ranges encountered for midwestern bituminous coals. These
values are indicative of a good utility coal, although it must be noted that this coal has a
moderately high sulfur content, characteristic of midwestern bituminous coals; as such, it
has a high SC«2 emission potential.
Typical RP&L Coal Analyses
Quantity
Prox. &Ult. Analv.. Wt.%
Moisture
Ash
Volatile Matter
Fixed Carbon (Diff.)
Hydrogen
Carbon
Sulfur
Nitrogen
Oxygen (Diff.)
HHV, Btu/lb
Ash Loading, Ib/MBtu
N Loading, Ib/MBtu
FC/VM
O/N
Minimum Volatile Matter
As Rec'd
14.1
9.5
31.2
45.1
4.2
62.7
2.5
1.3
5.7
11,330
8.40
1.16
1.45
4.34
Dry
_
11.1
36.4
52.6
4.9
73.0
2.9
1.5
6.6
13,193
—
—
—
DAF*
—
—
40.9
59.1
5.5
82.1
3.2
1.7
7.4
14,838
—
—
—
—
Median Volatile Matter
As Rec'd
14.4
9.6
31.8
44.2
4.1
62.3
2.5
1.2
5.9
11,187
8.60
1.06
1.39
4.99
Dry
—
11.2
37.1
51.6
4.8
72.7
2.9
1.4
6.9
13,061
—
—
—
—
DAF*
—
—
41.8
58.2
5.4
82.0
3.2
1.6
7.8
14,714
—
—
—
—
Maximum Volatile Matter
As Rec'd
13.9
9.8
32.8
43.6
4.3
61.7
2.6
1.2
6.6
11,112
8.77
1.03
1.33
5.75
Dry
—
11.3
38.1
50.6
5.0
71.6
3.0
1.3
7.7
12,907
—
—
—
—
DAF*
—
—
43.0
57.0
5.6
80.3
3.4
1.5
8.7
14,556
—
—
—
—
*DAF = Dry-Ash-Free
RSFC™ Burner Description
The RSFC™ burner has three air register zones to supply three different air annuli at the
burner exit, depicted in Figure 2. The combustion flow field is controlled by means of
mass flow splits between, and swirl number for, each individual air annuli. The swirl in the
-------
primary and tertiary air streams are generated through the use of moveable vane swirlers
An axial fixed vane swirler in the secondary air annulus provides a consistent swirl and
pressure drop over the conditions tested.
Air Register Configuration
Primary (Moveable Vane Register)
• Secondary (Axial Fixed Vane Swirler)
- Tertian' (Tvloveable Vane Register)
Primary Shut-off
Damper
Figure 2
Diagrammatic Representation of the RSFC™ Burner
The workings of the moveable vane damper are hidden from radiant heat that could cause
any part of the air registers or swirl adjustment linkages to overheat Air flow through the
burner when the shut off damper is closed is designed to be a nominal 10% to 15%. Most
of this flow occurs through the secondary air zone where it cools both the primary and
secondary air throats
The modular construction of the swirl block makes the RSFC™ burner very strong and
rigid Each of the blocks used in the construction of the burner stiffens the entire swirler
geometry In addition, many of the parts in the RSFC™ burner have been constructed out
of stainless steel in order to insure that the burner will be functional over the long term
The use of stainless steel protects the burner from heat, corrosion and rust on the critical
moving parts, resulting in a more reliable design with minimal risk of binding during
normal operation.
-------
Photo 1
RSFC™ Coal Burner
Photo 2
Furnace View of the RSFC™ Coal
Burner and Refractory Throat
As shown in Photos 1 and 2, in order to maximize operational flexibility, the air flow and
swirl through the primary and tertiary zones are independently controlled by externally
mounted drive systems. Movement of the Richmond swirl vanes is accomplished through
manual gearbox-driven linkage mechanisms. Although the swirl vane mechanisms can be
easily automated, RP&L decided that the manual arrangement was preferable so boiler
operators could maintain visual contact with the boiler and auxiliary equipment. Mounting
the drive hardware external to the windbox eliminates the potential for bearing
contamination and mechanical binding. If there were ever a problem with a damper
linkage assembly, the burner would not need to be removed to complete the repairs.
The RSFC™ coal nozzle assembly incorporates standard ABB C-E Services design
features to promote flame stability when firing both oil and coal Concentric to the
centerline of the coal nozzle is a coal spreader and guidepipe. The Richmond burner
configuration has an ABB C-E Services high energy arc ignitor and the original
mechanically atomized oil gun located inside the coal spreader guidepipe. A moveable
coal spreader was developed to help minimize the unburned carbon, while maintaining the
low NOx characteristics of the RSFC™ burner. The moveable links for the coal spreader
are located inside the center guidepipe to protect them from erosion and impacting on the
pulverized coal stream.
Another unique feature of the RSFC™ burner is the throat configuration. This design
allows for the optimization of low NOx flame shaping without flame impingement or
attachment to the front wall. The potential for wall slagging and overheating of the burner
components is greatly reduced as a result of the unique design
-------
NOx Emissions Mechanisms
The formation of nitric oxides (NOx) during coal combustion is a complicated process
involving three principal mechanisms, namely: fuel NOx; thermal NOx; and prompt NOx.
The combination of prompt NOx and thermal NOx mechanisms typically contribute a
combined maximum in the range of 30% to 50% to the total NOx formation. Thermal
NOx, described through the "Zeldovich mechanism," results from the oxygen fixation by
atmospheric nitrogen (Zeldovich, 1947). This NOx contribution increases exponentially
with temperature and with the square root of oxygen concentration in the reactant gas
stream. Hence, judicious control of the flame temperature and oxygen concentration
throughout the combustion process constitutes a powerful measure for controlling the
formation of thermal NOx.
The fuel NOx, which results from the oxidation of the fuel nitrogen-bound intermediates,
is the major source of total NOx emissions. Its contribution typically falls in the 50% to
70% range, although in some coal-fired units this contribution can range up to 90%
(Pershing and Wendt, 1979; Levy, et al., 1978). Therefore, the major objective in staged
combustion systems, such as the RSCF™ burner, is to mitigate the formation of fuel NOx.
One of the ways to achieve this is to burn the fuel in such a manner that the fuel nitrogen-
bound intermediates (HCN, NH;, NO, etc.) are preferentially converted to molecular
nitrogen (N2). Here, the volatile matter is allowed to burn in an oxygen-lean environment
(i.e., with a stoichiometry of less than one) near the burner zone. The nitrogen-bound
intermediates released with the volatile matter must compete with carbon and hydrogen
compounds for the limited supply of oxygen. Because the volatile hydrocarbons are
comparatively more reactive, the nitrogen-bound intermediates are left to react with each
other, leading auspiciously to the formation of molecular nitrogen. To be effective, the
slow, heterogeneous char oxidation reaction must occur in an oxygen-rich environment.
Hence, the nitrogen-bound intermediates formed via this route are preferentially oxidized
to NOx; this implies that only a small portion of these nitrogen-bound intermediates can be
converted to molecular nitrogen.
Figure 3 traces the fate of fuel nitrogen during the coal combustion process. It can be
stated, based on the description given above, that the ultimate goal in staged combustion
systems is to maximize the yields of both the volatile matter and fuel nitrogen-bound
intermediates, under a sub-stoichiometric environment, and to convert these intermediates
to molecular nitrogen. One of the reasons the RSFC™ burner is so effective in reducing
NOx emissions is because it optimally applies the principle of combustion staging, as
described in the next section. Another benefit of the RSFC™ burner is that it produces
low NOx through the use of lower stoichiometry in a confined burner or flame region,
rather than in the furnace as a whole. For this reason, waterwall corrosion is not a
concern.
-------
Figure 3
Fate of Fuel Nitrogen During The Coal Combustion Process
RSFC™ Technology
Radially Stratified Flame Core describes the unique flame structure that is at the heart of
the RSFC™ burner design. Many wall-fired burners employ swirling flows to enhance
mixing in the near-burner flow region. The RSFC™ burner is different in that swirling
flow is used to create the opposite effect, namely the delay of mixing in the near-burner
zone. It is this combination of a near-burner, high temperature, fuel-rich core followed by
a downstream, fuel-lean combustion zone that creates the low NOx combustion conditions
generated by the RSFC™ burner
The delay of mixing is achieved through stratification between the pulverized coal and the
surrounding, swirling combustion air. Stratification depends on density differences
between the gases in the flame core and the surrounding, relatively cooler combustion air
and turbulent mixing dampening at the flame/air interface. The fuel enters along the center
line of the burner and is surrounded by strongly swirling air from three separate annuli as
shown in Figure 4. The fuel jet penetrates into the central fuel-rich zone where the
centrifugal forces of the surrounding air eventually cause the fuel jet to stagnate and
recirculate back toward the root of the flame The first flame region, the high temperature
fuel-rich core, allows a large portion of the fuel nitrogen to be released in a low
stoichiometric zone where it is easily converted to molecular nitrogen. The internal
recirculation zone also helps stabilize the flame by providing adequate energy to the root
of the flame. This higher temperature (lower density) fuel rich zone along the center line
of the burner, surrounded by the cooler (higher density), swirling combustion air, creates
the stratification that is characteristic of the RSFC™ burner flame structure. After passing
through this initial stratified, low stoichiometric, combustion zone, the remaining
combustibles then mix with the remainder of the combustion air to complete the
combustion process
-------
The typical low NOx RSFC™ burner flow field is depicted in Figure 4 The concept of
radial stratification originated with the work of Rayleigh (Beer and Chigier, 1970), and
was brousht to practical application by Beer, et al. (1972) Thus phenomena which was
extensively studied dunng more than six years of fundamental study at M.I T 's
Combustion Research Laboratory, was then further developed for commercial applications
by ABB's Power Plant Laboratories This extensive amount of research and development
has now been incorporated into .ABB C-E Sen-ices' RSFC™ burner.
Ill Macro-Mixing [
Primary Air & Fuel Mix
to Create Fuel-Rich
Flame Core
Figure 4
Typical Low NOx RSFC™ Burner Flow Field
Installation of the Coal RSFC™ Burners
The six coal RSFC1"111 burners, high energy' arc (FtEA) ignitors, flame scanner system, oil
guns, actuators and new refractory throats were installed at the Whitewater Valley Station
in ten days, working a single eight-hour shift per day. In addition to the burner
installation, the work scope also included changes to the No 2 fuel oil system piping, and
the addition of air aspirated coal sample ports to the existing coal piping. The outaee
work went very smoothly since the RSFC™ burners were designed as a direct retrofit and
no boiler pressure part replacements or coal piping modifications were required
Initial Start Up Activities
Upon completion of the boiler outage, the new refractory was cured using the oil guns in a
predetermined firing sequence Refractor,- curing was accomplished in an eight-hour
process It is worth noting that the ignition of the oil guns by the HE A ignitors was
always accomplished on the first attempt Initial coal firing commenced immediately after
-------
the refractory curing process was completed. The boiler was continuously in operation
from the end of the October 1996 outage until the boiler's scheduled maintenance outage
at the end of May 1997. An inspection during the May 1997 outage confirmed that the
burners were in excellent condition with no visible signs of erosion or deformation.
Boiler Emissions Performance With The Coal RSFC™ Burner
Pre-retrofit and post-retrofit emissions tests were conducted to evaluate the performance
of the coal RSFC™ burner. The primary objective of these test programs was to quantify
the impact of the new burners over the full operating range of the boiler. The boiler
emissions performance was measured through a series of parametric tests during which
operational parameters were varied in order to quantify the results.
NOx Emissions
All NOx measurements in this paper were determined through the implementation of EPA
Method 7E, using a chemiluminescent NOx analyzer sampling from the airheater gas inlet duct,
and are reported in units of Ibs NOx per 10s Btu. With the RSFC™ burner, NOx emissions
range between 0.31 and 0.45 Ib/MBtu, with the every day operating level averaging
approximately 0.39 Ib/MBtu. Figure 5 shows the relatively flat relationship of the measured
NOx emissions to the furnace outlet oxygen level for the RSFC™ equipped Unit 1. The
RSFC™ burner is able to maintain this flat relationship since the burner optimizes and
lengthens the residence time of the fuel in the fuel rich primary zone. This is accomplished
by stopping axial flow of the fuel and recirculating it back towards the burner front where
it is mixed with the remainder of the combustion air.
-------
Inn
.UU
w n on
o °-90
.OU -1
o"*
co n Tn
v' u./u -
© nfin
-5 n en
£ °-50 i
«= n An
5 n ^o
?*" n on
5 010
Onn
«• — •
*
«
A
— r-^"
»
L. • J *
•
, ^
^
^
• R
Baseline NOx
M
—
SFC™ Equipped NOx |
234567
Oxygen, %
Figure 5
NOx Performance Comparison vs. Furnace
In this zone, the fuel is allowed to burn as hot as possible because there is very little air
and oxygen to drive the thermal NOX formation. Any fuel nitrogen is also released in an
oxygen lean environment where the nitrogen radicals are forced to combine with other
nitrogen radicals to form elemental nitrogen (N2) rather than a nitrogen oxide. It is this
same effect that allows the burner to produce a flat curve of NOx emissions versus boiler
steam load as shown in Figure 6.
-------
RSFC™ Equipped NOx
40
50
60
70
Boiler Load, %
80
90
100
Figure 6
Comparison of NOx Emissions vs. Boiler Load
Carbon Monoxide Emissions
All carbon monoxide (CO) measurements reported in this paper are provided in units of parts
per million (ppm) of gas and are corrected to 3% oxygen in the flue gas. All measurements
were obtained through the use of non-dispersive infrared analyzers sampling from the airheater
gas inlet duct. The test protocols used are in accordance with EPA Method 10. Pre-retrofit
CO emissions averaged 48 ppm at MCR steam load. Post-retrofit testing CO averaged 47
ppm from minimum load to MCR boiler load throughout normal operating conditions.
Atrita Pulverizer Performance
It is important to understand that pulverized coal fineness plays a major role in
determining the overall carbon burnout during pulverized coal combustion. The
distribution of sizes of particles in a pulverized coal sample is commonly expressed as a
plot of the fractions of the weight of the sample contributed by all particles larger (or
smaller) than a succession of sizes. The Rosin-Rammler relationship (Field, et al, 1967),
which is one of the techniques commonly used to characterize pulverized materials, was
applied to the screen analysis data given in Table 2.
-------
The characterized post-outage pulverized coal samples of the three mills all had a coarser
consistency than the pre-outage pulverized coal samples. The post-outage coal fineness
was 62-64 weight % through 200 mesh, with 1.2-2.4 weight % retained on a 50-mesh
screen. Comparatively, the desired coal fineness in low NOx firing applications typically
falls in the 75-85% through 200 mesh, with zero or only a fractional amount of material
retained on a 50-mesh sieve. Hence, the coarse grinds obtained at RP&L Whitewater
Valley Unit 1 compare unfavorably to those desired in a typical low NOx firing
application.
Table 2
Particle Size Distributions of Pulverized Coal Samples
Mill#
1
2
3
Composite
Pre-Outaqe Fineness
-200 Mesh
64.7
73.8
80.0
72.8
-100 Mesh
87.0
92.8
95.6
91.8
+50 Mesh
2.0
1.0
0.4
1.1
Post-Outage Fineness
-200 Mesh
61.6
64.8
64.0
63.5
-100 Mesh
84.8
89.4
88.2
87.5
+50 Mesh
2.4
1.2
1.4
1.7
Another typical impact on low NOx firing is the pulverizer air/fuel ratio. Typically, an air
to fuel ratio of approximately 1.5:1 is desirable to keep the pulverized coal particles in
suspension, without providing an excess of transport air that will impact flame stability or
NOx generation. The Unit 1 Atrita mills operate at an air to fuel ratio of 2.1:1 at high
load, with lower load ratios on the order of 3.8:1. While this level of transport air flow
typically tends to hinder low NOx operation, the post-outage testing shows that high
air/fuel ratios had no effect on RSFC™ burner NOx reduction capability.
Unburned Carbon in the Fly Ash
Fly ash samples were obtained using EPA Method 17 isokinetic sampling techniques from
the airheater gas inlet duct. Each fly ash sample was then analyzed for carbon content.
Pre-outage baseline testing for unbumed carbon content during normal operation through-
out the load range averaged 15.29 % with an average mill fineness of 72.8 % through a
200 mesh sieve, with an average of 1.1 % left on a 50 mesh sieve. Post retrofit testing of
unbumed carbon content during normal operation throughout the load range averaged
16.9 % with an the average mill fineness 9.3% coarser through 200 mesh, and an average
of 1.7 % left on a 50 mesh sieve. As discussed above, the negative change in pulverizer
performance from pre- to post-retrofit most likely has more impact on the change in
-------
unburned carbon than does the RSFC™ burner. After seven months of operation,
Richmond reports that Unit 1 unburned carbon is within 0.5 to 1% of pre-retrofit
unburned carbon levels. The strong recirculation zone created by the RSFC™ burner
greatly enhances coal particle bum out and minimizes the typical low NOx increase in the
unburned carbon content of the fly ash.
Boiler Operational Performance With The Coal RSFC™ Burner
During post-retrofit testing on the Whitewater Valley Unit 1 boiler, multiple aspects of
boiler operation were examined to determine the impact of the RSFC™ burners on boiler
operation. The operational performance issues discussed in this paper have been
confirmed during the nine months of boiler operation since the burners were installed.
Ash and Slag Deposition Patterns
Since the installation of the RSFC™ low NOx coal burners, a long term change in the ash
and slag deposition during operation has been observed. Prior to the burner retrofit, the
original burner combustion resulted in moderate to heavy slag buildup around the burner
throats which RP&L personnel had to manually rod out between two and three times a
week. Since the installation of the RSFC™ burners, there has been no appreciable
slagging in or around the refractory throats. In general, with the new burners, there has
been minimal slag buildup on the furnace walls or convective sections of the boiler.
Furnace Oxygen and Coal Imbalances
During the tuning phase of the postOoutage start-up activities a noticeable coal and O^
imbalance was measured across the width of the furnace. Rather than perform the tedious
task of balancing the coal flow to each burner, the decision was made to use the multiple
operational tuning parameters available within the RSFC™ burners to "tune the burner to
the fuel flow" Through the use of the burner biasing/shut off dampers and the primary
and tertiary swirl vanes, each burner was optimized to the fuel flow through the respective
coal nozzle. As a result of this tuning, the O2 as measured across the width of the
airheater gas inlet duct was balanced to within 0.3% side to side.
Steam Temperature Control
Post-retrofit testing and long term operation has confirmed that the installation of the coal
RSFC™ burners has improved the control of steam temperature for this boiler. Prior to
the retrofit, boDer operators had to perform operational burner adjustments in order to
maintain design steam temperature. With the RSFC™ burners, no operational changes to
the burners are required to maintain design steam temperature.
-------
Acknowledgments
The authors would like to acknowledge the significant contribution of the following key
individuals to the success of this project.
Irv Huffman and Bob Crye of Richmond Power & Light.
Janos M. Beer of Massachusetts Institute of Technology
Angelos Kokkinos, Jim Sutton, Jack Sweeney, Sandy Salazar, David Thornock, Richard
LaFlesh, Jim Mitchell, Jeff Mann, Tom Bienkowski, Al Christie, Mihir Patel, Willy
Feldstein, John Smith, Bob Moriarty and John Lewis of ABB C-E Services.
Allen Pfeffer, Richard Borio, Tony Kaiser, Majed Toqan, Tom Duby, Marty Kozlak, John
Drennen and Julie Nicholson of ABB Power Plant Laboratories.
Jorge Henao and Joan Kraiza ABB Power Plant Controls.
Steve Ediger, Steve Wilcox and Dennis Price of ABB Air Preheater - Concordia, Kansas.
References
Fenimore, C.P., "Studies of Fuel-Nitrogen species in Rich Flame Gases," Proc. 17th
Symposium (International) on Combustion. The Combustion Institute, Pittsburgh, PA,
661 (1979)
Fenimore, C.P., "Formation of Nitric Oxides inPre-mixed Hydrocarbon Flames," Proc.
13th Symposium (International) on Combustion. The Combustion Institute, Pittsburgh,
PA, 373 (1970)
Field, M.A., Gill, D.W., Morgan, B.B. and Hawksley, P.G.W., "The Fineness of
Pulverised Fuel," Combustion of Pulverised Coal BCURA, Leatherhead, England, p.
245, 1967
LaFlesh, R. and Madura, P.E., "Field Demonstration of ABB C-E Services' RSFC™ Wall
Burner for Oil and Gas Retrofit Applications, ABB C-E Services Publication TIS 8645
1996
Levy, J.M., et at., "Conversion of Fuel Nitrogen to Nitrogen Oxides in Fossil Fuel
Combustion: Mechanistic Considerations," M.I.T. Energy Laboratory Report to EPA
under FCR Program, 1978
-------
Pershing, D.W. and Wendt, J.O.L., Relative Contributions of Volatile Nitrogen and Char
Nitrogen to NOx Emissions from Pulverized Coal Flames," Ind. Eng. Chem. Process Des.
Dev., 18, 60(1979)
Zeldovich, Y.A., "Oxidation of Nitrogen in Combustion," Academy of Sciences of USSR,
Institute of Chemical Physics, Moscow-Leningrad, 1947
-------
Monday, August 25; 3:30 p.m.
Parallel Session B:
Low-NOx Systems for Coal-Fired Boilers - Group 2 Units
-------
NOx Reduction without Low NOx Burners for a Riley Dry Bottom
Turbo Furnace
F R. Sokolik
Senior Engineer
Salt River Project
A. G. Gilson
Staff Engineer
Burns & McDonnell
June 15,1997
Abstract
Coronado Generating Station is a coal fired facility consisting of two Riley dry bottom turbo
furnaces. When evaluating technologies for upcoming NOx reduction requirements, traditional
reduction strategies were carefully considered. However, specification, procurement, and
installation of low NOx burners appeared to require extensive engineering, and would be quiis
expensive. Instead, SRP teamed up with Burns & McDonnell to design and implement relatively
simple burner secondary air flow controls. By partitioning the existing burner windbox, separate
control was provided for burner secondary air and for overfire air. Proper sizing and placement
of the partition helped both to better control overfire air flow, and to bring secondary air
velocities closer to accepted norms. The entire project from kick-off, initial engineering,
procurement and installation was conducted in just over two months. Minor tuning and
adjustment of the new control equipment was conducted by SRP personnel after installation.
NOx reduction ranged from 15% at high load to 20% at low loads. Notable improvements were
also seen in flame shape and stability. Combustion was more stable and controllable, with less
dark areas than before the modifications. The associated increase in LOI was minimal.
Introduction
Coronado generating station is a Salt River Project (SRP) facility located in eastern Arizona.
The station consists of two coal fired Riley Turbo-style furnaces each rated for approximately
2,747 kpph steam flow. The units fall under Phase II of the 1990 Clean Air Act Amendment for
NOX. The requirements for NOX reduction presented a particular challenge to the unique style of
the turbo furnace with its downward-opposed square burners. Specification, procurement, and
installation of new burners appeared to be complex and expensive, especially considering the
units' marginal NOX levels of around 0.55 Ib/MMBTU. Instead, SRP retained Bums &
-1-
-------
McDonnell to apply a relatively simple technique which B&McD used previously to reduce
NOX on other turbo furnaces.
Design/Specification
Each furnace consists of 24 opposed burners and overfire air fed by a compartmented windbox.
The secondary air opening around each burner includes 12 directional vanes and two velocity
dampers. The overfire air port includes a 1/3 and 2/3 damper to control air flow out of the port.
Each compartment in the windbox is dedicated to a burner and overfire air port, and has a single
shut-off damper. This common shut-off damper was the key to the NOX reduction technique.
B&McD proposed to partition each compartment in order to separate burner secondary air flow
from overfire air flow. The partition consisted of a steel plate wall about 2/3 back from the
furnace. The shut-off damper would be trimmed to only control air to the burner. The addition
of modulating controls to the two-position cylinder actuator allowed the damper to be throttled
back to control air flow at each burner. Throttling this new secondary air damper at each burner
would increase windbox pressure upstream of the damper and increase overfire air flow. Used
properly, this would reduce burner zone stoichiometry resulting in cooler flames and lower NOX
levels.
Partition Design
One key to successful design lies in the proper trimming of the secondary air damper and proper
air velocities into the furnace. The damper was intended to control air flow to each burner, not
air velocity. Velocity was controlled by the secondary air openings into the furnace. Therefore,
the secondary air damper was sized 25% greater than the secondary air openings into the furnace.
Air velocity is a significant consideration in burner design. Merely si/ring the new secondary air
damper without evaluating the openings into the furnace could result in poor fuel and air mixing,
and defeat any emissions reduction capability produced by the additional overfire air flow.
B&McD's experience indicated that secondary air velocities should run between 120 and 140
fps, and primary air velocities should run between 70 and 90 fps. The as-found configuration of
both Coronado furnaces produced much higher velocities than preferred. This was confirmed by
observations of poor flame conditions in the furnace. Therefore, as part of the program, the
waterwall penetrations were enlarged by the removal of refractory. (The design of the turbo
furnace did not normally require burner throat refractory)
These reduced air velocities from the refractory removal were closer to the targets, and were
carefully evaluated in determining sizing for each secondary air damper. Also evaluated were
target burner zone stoichiometry, current fuel analysis, and excess oxygen levels. A burner zone
stoichiometry of 0.9 was chosen to maximize flame temperature reduction without severe
waterwall corrosion.
The evaluation resulted in a 75% reduction in area of each secondary air damper.
SRj>_PP-l WPD -2-
-------
The windbox partition was constructed of 1/4" steel plate and angle irons. It included bolt-on
panels to provide access to the interior of the burner compartment. An outside contractor
completed the work on Unit 1 during the fall outage in about 3 weeks. Plant personnel added
electro-pneumatic positioners on each burner air damper and wired them back to the control
system. Provisions in the DCS system were added for burner management permissives, cooling
settings, as well as modulation of each burner secondary air damper, should it become necessary.
Windbox Baffles
Pre-outage flow testing showed a significant front-to-rear air flow imbalance. Physical
inspection during the outage indicated that baffles inside the windbox might help with balancing.
The unit had existing balancing dampers for the rear of the windbox only. However, the testing
revealed that even with the dampers wide open, airflow was much higher on the front side of the
windbox than on the rear. Therefore, during the outage, 15" baffles were installed in the
windbox to help balance the secondary air flow from front to rear burners.
Tuning and Results
Tuning was conducted over a period of about 2 weeks, and included measurement of air flows at
each burner and overfire air port after each adjustment was made. Windbox front-to-rear air
flow balance was tuned first. The new baffles forced enough air to the rear of the windbox to
make the existing dampers much more effective. This allowed the balancing dampers to be
throttled between 30 and 50% to balance front and rear air flow.
Next, adjustments were made to each burner secondary air damper. Several combinations of
damper positions were evaluated while simultaneously monitoring combustion conditions and
unit emissions. Air flow traverses at each burner windbox compartment were made after each
new setting. Working with the new secondary air dampers allowed much more control over
combustion conditions. Adjustment of the burner secondary air dampers and primary air flow
provided much better control of flame shape and attachment than before the outage. Low load
flame stability was also much better. The upper furnace area was much clearer, with no
sparklers.
The final setting was determined after about two weeks of testing. For light-off, the dampers
were placed at 80% open. As load is increased, the inboard burners are modulated from 80% to
60% open. The remaining burner secondary air dampers were left at 80% open. Overfire air
was left wide open throughout the load range. The final results showed a 15% reduction in NOX
levels at high load, and a 20% reduction at low loads. Windbox pressure increased from 4.5
inwc to 5.5 inwc. Unburned carbon increased minimally. Flame shape was better defined and
more controllable. Dark areas of the furnace decreased significantly, as did sparklers in the
pendants.
SRP_PP-I WPD
-------
Conclusions
The entire project, from initial design to final tuning, was conducted in about 3 months. Of this
time, 6 weeks was taken up by the outage and tuning. Emissions reductions were enough to
comply with upcoming regulations, but were not so drastic as to cause significant increases in
unbumed carbon or furnace waterwall corrosion rate. Further emissions reduction may be
possible with further throttling of the burner secondary air dampers, but was not necessary for the
Coronado units. The entire cost of the project was approximately 20% of the cost of designing
and installing new burners.
Success of the modification requires a careful evaluation of combustion conditions, fuels,
stoichiometry and existing burner configuration. It also requires a good tuning procedure. This
technique is a good alternative to new burners for Turbo-style furnaces with marginal emissions.
Turbo furnaces with higher emissions could also be candidates for this modification, depending
on the combustion conditions and the physical geometry of the furnace.
References
1. S.C. Stultz and J.B. Kitto (editors). STEAM, its Generation and Use. Barberton, Ohio:
The Babcock and Wilcox Company, 40th Edition, 1992.
SRP PP-] WPD
-------
NOX REDUCTION ON A RILEY STOKER DRY BOTTOM
TURBO FURNACE
THE MEGA SYMPOSIUM
David L. Williamsen
Burns & McDonnell
Engineers, Architects, Consultants
9400 Ward Parkway
Kansas City, Missouri 64114
Michael O. Spoutz
Gainesville Regional Utilities
301 S.E, 4th Avenue
Gainesville, FL 32614
Abstract
Gainesville Regional Utilities of Gainesville, Florida retained the services of Burns &
McDonnell of Kansas City, Missouri to conduct a NOX reduction program. The NOX target was
0.5 Ibs/mmbtu. The unit was at 0.7 Ibs/mmbtu at the start of the program.
The approach taken by Bums & McDonnell was to first get the combustion in order. This was
done by balancing primary air and fuel to the burners through mill testing, improving the quality
of fineness of the fuel, and making adjustments to burner directional vanes. These items alone
reduced the NOX to 0.56 Ibs/mmbtu.
An outage was used to partition the windbox and add secondary air shutoff damper control to
allow individual control of the secondary air to the burner front and force additional air to the
overfire air ports. This was the major modification done regarding the NOX reduction effort.
However other work contributed including classifier modifications, mill reject valve upgrade,
mill ball charge changes, and reducing secondary air velocities at the burner front.
When the physical modifications were completed and the unit brought back on line, boiler tuning
took place to continue with the NOX reduction and combustion improvement. The NOX was
reduced to 0.47 Ibs/mmbtu at full load.
Initial Observations
Initial unit walkdowns were performed to determine existing combustion conditions. The lower
furnace, burner level and upper furnace were all observed. The lower furnace was observed for
areas of unbumed fuel, uniformity of the fire, and flame conditions at the burners. The six
observation doors at this level provide an excellent view of all eighteen burners. The level just
above the burners provides some view of the overfire air ports and a limited view of the burners.
MEGAPAPR.WPD -1-
-------
These doors were used extensively during the actual boiler tuning process, which took place after
modifications were completed. Any incomplete combustion in the comers and along the furnace
walls can be readily seen through the observation doors at this level. The upper furnace doors
were used to observe any fire and/or sparklers present at the pendant level.
Early on, the furnace observations that were made showed a tremendous difference from the
front to the rear burners. The rear burner fires had incomplete combustion at the burners while
all of the front burners had good combustion. The front burners were very uniform in
appearance, with consistent upswept fires. The rear burner fires were erratic with some burners
having raw coal streams extending out considerable distances from the burner throats. Fire and
clouds of unburned fuel were swirling in the lower furnace. Along with this erratic burn, the rear
burners did not show any consistency from burner to burner.
Observations made through upper doors showed some incomplete combustion above the burner
area while the upper furnace had fire present.
Primary Air and Fuel
Based on these furnace observations, the first items addressed were the fuel and primary air
delivery to the burners. Isokinetic coal testing was performed on the three Riley ball tube mills.
This testing revealed good primary air flow distribution, however there was pipe-to-pipe coal
flow imbalance. The coal fineness of the three mills at full mill load was also less than the
desired fineness of 70 percent passing a 200 mesh sieve and 98 percent passing a 50 mesh sieve.
The furnace observations combined with the mill test results dictated the need to review the mill
ball charges. Discussions between B&McD and GRU personnel resulted in the decision to
change out the mill ball charges during the upcoming Spring 1995 outage.
Prior to the outage, the primary air system hot and tempering air dampers were characterized to
allow for proper mill temperature control without influencing flow control. All of the hot and
tempering air dampers were restroked. The mill rating and bypass dampers were stroked to
determine linearity between the rating damper and coal flow. This relationship was satisfactory
and no changes were required. The turndown capability of the mills was also evaluated. Each
mills rating damper was set to maintain a minimum classifier-to-fumace differential. Mill output
was then reduced by opening the mill bypass dampers. A very good turndown of six to one was
achieved while maintaining the minimum classifier-to-furnace differential.
Initial Modifications
Due to time restraints, not all of the recommended modifications for reducing NOX and
improving combustion could take place during the spring 1995 outage. However, some of the
modifications were made, along with some routine maintenance work . For some
recommendations, temporary modifications were substituted until the permanent modifications
could be made during the spring 1996 unit outage.
MEGAPAPR.WPD -2-
-------
The full modifications or changes that took place in 1995 included refurbishing the mill
crusher/dryers, removing a portion of the perforated plate from above and below the burner
igniters to reduce secondary air flow velocities, installing primary air flow measuring probes in
each of the primary air ducts, adding mechanical stops to each burner shutoff damper to provide
cooling air to out of service burners, and changing out the ball charge in each mill.
One of the temporary modifications included setting the upper and lower directional vanes at
fixed downward angles relative to the front and rear furnace walls. Because the vanes had a
tendency to stick in other positions when being adjusted, once the vanes were set they were
welded in place. The permanent modification is to make the directional vanes adjustable and
stable.
Another temporary modification was to place the burner velocity dampers in the fully open
position. The permanent modification is to remove the velocity dampers completely. Because of
the distance these dampers are from the burner front, any effect they have on the secondary air
velocity is virtually lost by the time the air reaches the burner front.
Making the mill reject valves functional was another temporary modification. The permanent
modification is to redesign the valve for better reliability and external indication of performance.
Observations Following Initial Modifications
After the spring 1995 outage when Unit 2 was brought back on line, unit walkdowns were
repeated to observe combustion changes. The fire was definitely lower in the furnace. There
was still some inconsistent burning at the rear burners but the fires were brighter. Swirling coal
and fire were still present in the lower furnace. The NOX level however had dropped to 0.537
Ibs/mmbtu at full unit load.
A shipment of coal received shortly after Unit 2 was brought back on line had a Hardgrove
Grindability Index (HGI) lower than the value that the mills were designed to. Full unit capacity
could not be maintained on this harder to grind coal. Fuel gas had to be used to supplement the
coal to achieve full unit load. After the harder quality coal was no longer being burned, mill
capacity returned.
However, even with the return to a coal that had an HGI within the specified range of these mills,
the coal inventory of the mills had to be increased to obtain full mill capacity. This additional
coal inventory reduced the mill performance (fineness) considerably. The ball charge weight
was increased slightly to bring all three mills closer to the design charge weight. This improved
capacity some but the increased mill inventory was still required. The mill ball charge
distribution needed further evaluation.
The mills use power and sound to control coal levels. The power levels indicated that the new
balls charges were lighter than the old charges. Additional balls were added to two of three mills
and full unit load was achieved without supplemental fuel gas. However the mills were still
operating with a higher than desired mill inventory. The 50 mesh fineness was at desired values
but the 200 mesh fineness was still quite low. During an October 1995 forced unit outage, the
MEGAPAPR.WPD -3-
-------
mills were internally inspected for ball charge level and found to be low. Additional balls were
added to all three mills.
No additional modifications would be made until the next unit outage scheduled for the spring of
1996. The unit continued to operate and maintain full load without the need for supplemental
fuel gas.
Final Modifications
Specifications were prepared by Burns & McDonnell for the work to be done during the spring
1996 outage relating to the NOX reduction program. These specifications were included in the
GRU bid package covering additional plant outage work. The following NOX related outage
items included:
• Partition each burner compartment to separate the secondary air flow to the burner from
the overfire air flow.
• Reduce the entry area to each burner compartment to improve secondary air damper
controllability
• Convert the secondary air shutoff dampers from an open/close operation to modulating
operation by installing positioners on each damper.
• Remove the secondary air velocity damper from each burner compartment. These
dampers have no lasting impact on secondary air velocity due to their location back from
the burner front.
• Replace the upper and lower directional vane operating rods to provide improved
reliability. The existing rods have bends near the vanes to clear the velocity dampers
which cause binding. New straight rods will eliminate the binding and provide easier
adjustment of the vanes.
• Install a dual lift wing in each of the coal nozzles to maintain a homogenous distribution
of the coal flow until it reaches the burner opening.
• Extend the classifier inner shroud below the classifier inlet vanes. This deters coal from
short circuiting directly to the coal pipes, and improves both fineness and distribution.
• Install directional bars on the classifier inlet vanes to direct the larger coal particles
downward and improve classification, therefore improving fineness.
• Redesign the mill reject valves to improve reliability and provide external indication of
the valve operation. The new valves include a counterweight for positive shutoff and are
shaft mounted. The external indication will allow for immediate identification of any
valve problems.
MEGAPAPR.WPD -4-
-------
• Change out the mill ball charge in the west mill to a charge which Includes a higher
percentage of small balls. This charge will provide a 10 percent greater surface area for
the same ball weight.
All of the above tasks were completed during the spring 1996 outage.
Boiler Tuning and Observations Following Final Modifications
Boiler tuning for NOX reduction was performed during a three week period in April 1996. The
first task was to observe furnace conditions at full unit load prior to making any operating
changes. The excess O2 was approximately 4.0 percent and the overture air dampers were 100
percent open. The NOX was at .632 Ibs/mmbtu. The fire at the rear burners was still inconsistent,
with unburned coal extending out and down in the lower furnace area. The rear fires were also
very erratic. The front burners showed good combustion with flames attached to the burner
throats. Furnace temperatures were taken through six observation doors above the burner levels
on both sides of the boiler at the front, center, and rear of the furnace.
The V3 overfire air dampers were then closed and the 2/3 overfire air dampers were left open. All
other conditions remained the same. The NOX climbed to .657 Ibs/mmbtu. No major visual
change was evident.
The V3 overfire air dampers were then opened and the 2/3 overfire air dampers were closed.
Again all other conditions remained the same. The NOX climbed again to .737 Ibs/mmbtu. The
fire appeared to be higher in the furnace.
With conditions still the same, all of the V3 and 2/3 overfire air dampers were then closed and the
NOX rose to .798 Ibs/mmbtu. Although no carryover was occurring in the upper furnace, fire was
observed between the radiant superheater tubes.
The overfire air dampers were then returned to their full open positions.
The next task was to increase the air to the rear burners to improve the combustion at the rear
burners. All of the previous observations of coal streams out from the burners, swirling clouds of
coal, and erratic fires indicated the need for additional air to the rear of the furnace. Also the
overall fire in the lower furnace was not centered but more towards the rear of the furnace.
The new controllers installed during the outage were used to slowly close down on the front
burner secondary air shutoff dampers. The control of the dampers was set up on a per classifier
basis, therefore three burners are operated together during each adjustment. All nine rear
dampers were closed down in 5 percent increments and furnace observations were made after
each change. No noticeable change took place until the dampers were at 80 percent open. At
this point the rear fires began to clear up. The furnace in this area became brighter. The lengths
of the raw coal streams decreased as the additional air being forced to the rear burners was
improving the combustion. The appearance of the fire continued to improve as the dampers were
taken to 70 percent open. The location of the fire had shifted away from the rear wall and was
MEGAPAPR.WPD -5-
-------
now centered in the lower furnace. The swirls of unburned coal and fire climbing the rear walls
had ceased. Furnace temperatures were taken above the burner level through the same six
observation doors. The furnace temperatures increased an average of 65° F.
Closing down on the front burner secondary air shutoff dampers did not have any adverse effects
on the front burners. They continued to show good combustion.
All rear burner secondary air shutoff dampers were then closed down until a change could be
observed above the burner level. These dampers were taken to 75 percent open. The front and
rear secondary air shutoff dampers were then closed down simultaneously at 2 percent
increments. Observations were made at each change in damper position. When the front
dampers reached 56 percent open and the rear dampers reached 61 percent open, furnace
observations were again made. The unit was still at full load and the excess O2 was
approximately 3.7 percent. The NOX was now at .502 Ibs/mmbtu. The furnace temperatures
measured through the same observation doors above the burners remained nearly the same. The
temperatures did however level out more from front to rear. This was apparent in the furnace
observations. The fire is more centered and filling the furnace below the burners. The point
where the front and rear fires converge was very close to the mid furnace area.
The next task was to reduce the coal inventory in the mills to improve coal fineness to the
burners. New peak kW levels were determined for each mill. Operating KW levels were then
selected that were closer to peak values than previously used. This allowed for smaller coal
inventories in each mill.
The mills were tested to determine performance with the new coal inventory levels and the new
ball charge in the west mill. The coal fineness was greatly improved on all mills with the west
mill being the best performer. There was however a front to rear classifier coal flow imbalance
on two of the mills. Results of this could be seen in the individual front and rear pipe fineness
values. Front classifier vanes on both of these mills were opened up to balance these flows.
The excess O2 set point was lowered to 3.0 percent at full load. At full unit load the NOX was
now running at .47 Ibs/mmbtu. This NOX value was able to be maintained as long as the furnace
was kept relatively clean. As the furnace became dirty and the heat absorption decreased, thus
allowing the overall furnace temperature to increase, the NOX also increased. Cleaning furnace
wall surfaces would bring the NOX back in line. However, because of retractable soot blowers in
the upper furnace and the gas recirculating system being out of service, furnace wall cleaning
would decay steam temperatures. Placing these retractable blowers and the gas recirculating
system back in service allowed steam temperatures and NOX to be maintained.
The excess O2 set point was further lowered to 2.8 percent. This was to allow for an additional
margin of NOX control during unit transient conditions. However, to effectively control NOX at
full load, the furnace cleanliness had to be maintained, no matter what the O2 set point.
Comparisons of burner to burner air flows for burner air and overfire air were made using a hot
wire anemometer. Having to adjust three dampers together on the same classifier was not always
the preferred method to balancing air flows. Without individual control of each damper,
MEGAPAPR.WPD -6-
-------
complete balancing was not attainable. Some in field modifications were made to allow control
of some individual dampers. However, after further air flow readings were taken, the need for
individual damper control on all 18 dampers became apparent. It was decided to stop further
temporary modifications and order the material required to permanently install individual damper
control.
The improved combustion at the burners allowed for any mill to be taken out of service during
reduced load operation. In the past, only the east mill, which included the four end burners,
could be taken out of service. Removing either the west or center mill caused burners that were
adjacent to west or center mill burners to trip.
Changing the ball charge in the west mill improved its performance. The west mill will handle a
higher coal capacity than the other two mills while maintaining proper fineness. The new charge
provided approximately a 10 percent increase in surface grinding area for the same total ball
charge weight. The ball charge in the center and east mills were changed to match the more
proficient charge in the west mill.
MEGAPAPR.WPD
-7-
-------
NOX REDUCTION IN ARCH-FIRED BOILERS BY
PARAMETRIC TUNING OF OPERATING CONDITIONS
Luis Canadas, Vicente Cortes
DIQA. ESII. University of Seville. Avda. Reina Mercedes, s/n. 41012 Seville (Spain)
Francisco Rodriguez
INERCO, S.A. Edif. Renta Sevilla, PL?2. 41007 Seville (Spain)
Pedro Otero, Juan F. Gonzalez
ENDESA. Principe de Vergara, 187. 28002 Madrid (Spain)
Abstract
This paper presents the results of a research project whose objectives are to investigate the
parametric sensitivity of boiler operation in relation with NO., generation and heat rate, and to
fine tune operating conditions to minimize NOX emissions. The project is included in the
European Coal an Steel Community Research Programme and was performed in arch-fired
boilers burning low-volatile coals. Research was mainly composed of sets of in-field tests
designed following the factor analysis methodology. Testing procedure included a complete
survey of the experimental units operation with measurement of coal flow and size distribution
to burners, furnace temperatures, in-flame gas composition profiles, and on-line boiler efficiency
and unit heat rate monitoring, amongst others.
Results revealed a strong sensitivity of NOX to operational parameters and deep differences
between boilers of similar technology. NO^ emission reductions greater than 30% have been
documented, exploring non-conventional boiler settings without penalizing or even increasing
unit efficiency.
1. Introduction
A large proportion of the coal produced in the mining regions of the North of Spain are
anthracites and other low-volatile coals. More than 3,500,000 tons/year of this coal (Volatile
matter: 7.2%; N: 1.0%; Ash content: 33.5%) is burnt in the five arch-fired units of Compostilla
Power Station (ENDESA).
Arch-fired furnaces (Figure 1) are designed for the industrial firing of low-volatile coals, as they
provide a solution to problems arising due to the low ignitability and combustibility of these
fuels. The characteristic high temperature levels and residence times of these combustion systems,
as well as the higher char/volatile ratio of low-volatile coals, produce higher levels of NOX
emissions than bituminous coal firing. NOX emissions values in the range of 1800-2000 mg/NmJ
(d.b., 6% OJ have been reported for different Spanish arch-fired boilers, such as Compostilla
Units 4 & 5.
-------
SECONDARY
AIR
Figure 1. Arch-fired boiler.
Although there is no current restriction for NOX
emissions in existing installations of this type, a limit
of 1300 mg/Nm3 (d.b., 6% O2) is thought to be
applied in the future. Commercial solutions for this
oncoming problem with NOX emissions, such as flue
gas treatment using catalytic reduction systems by
ammonia injection, are expensive alternatives. These
techniques should therefore only be used after
exhausting the possibilities of reducing NOX emissions
through the optimisation of the coal combustion
process. Such primary measures should also be
evaluated in terms of their possible effect on the
thermal performance of boilers.
Due to the above factors, a broad-based research
project on the reduction of NOX emissions, by the use
of combustion fine tuning, has been undertaken in the
arch-fired Units 3, 4 & 5 of Compostilla Power
Station. This project, known by the initials RNA, is included in the European Coal and Steel
Community Research Programme.
2. Objectives
The main objective of the RNA Project was to determine the feasibility of approaching, or even
reaching, the 1300 mg/Nm3 NOX limit of possible future application in Compostilla Units 4 &
5, using only combustion adjustments.
Additionally, as NO^ emissions in Compostilla Unit 3 were as low as 1100-1200 mg/Nm3, a
comparative study has been conducted in order to determine the design and performance
parameters that give rise to the different NOX formation behaviour of Compostilla Unit 3, with
respect to that of the twin Units 4 & 5.
3. Tests programme
An extensive programme of in-field combustion tests, at full load and using coal of practically
constant properties, has been conducted during 3 years in two different phases:
Phase I: Trials at the "high NOX emissions" Units 4 & 5.
Phase II: Trials at the "lower NOX emissions" Unit 3. Comparison with Units 4 & 5.
The methodology adopted for these trials was based on the parametric sensitivity approach, and
the following schedule was drawn up for each unit:
a) Boiler instrument checking.
b) Operational modifications testing (parametric sensitivity matrix).
c) Trend confirmation testing.
d) Maximum improvement testing.
-------
In each test, a broad characterisation of boiler performance was made, by measuring or
monitoring the following items, amongst others:
a) Temperature distribution within the furnace (60 measurement points).
b) Gas composition (NO, NOX, O2, CO) above the burner arch (24 measurement points).
c) Gas composition (NO, NOX, O2, CO) and temperature at the outlet of the economizer
(using a grid of 32 measurement points).
d) Emission levels of NO, NOX, O2 and CO.
e) Coal fineness and flowrate to burners.
f) Fuel-oil support.
g) Excess oxygen.
h) Air damper positions.
i) Fly ash carbon content.
j) Coal analysis.
k) Desuperheating spray flows.
1) Boiler efficiency and unit heat rate.
Besides this, modelling of the windboxes of Units 4 & 3 has been performed, in order to
establish the relationships between air dampers openings and actual air flowrates through them.
4. Discussion of experimental findings
4.1 Scope
The operating modifications tested during commercial operation of the boilers consisted of
variations to several different parameters. These included excess air, secondary and tertiary air
dampers openings, fuel-oil support, distribution of active burners, positioning of straightening
vanes, and combustion air temperature. Other factors such as the degree of boiler slagging were
also taken into consideration.
Although results obtained within this Project reveal significant effects of most of the above
mentioned operating modifications1, this paper will only focus on the relevant trends determined
for excess air, secondary and tertiary air dampers openings and fuel-oil support at Compostilla
Units 3, 4 & 5.
4.2 Phase I: Trials at Units 4&5
4.2.1 NOX Emissions. The main findings obtained in Units 4 and 5 of Compostilla P.S. are
shown below, in relation to the operating conditions that brought about a significant fall in NOA
emissions. More specifically, the distribution of air fed into the boiler, which determines the type
of flame obtained, and the excess air are especially important factors in the generation of NOX.
Likewise, the influence of eventual fuel-oil support has been established.
-------
Figure 2 details the results for Compostilla Unit 4 of the most important cases in terms of NOV
emissions. The percentages of change in comparison with reference data are shown in
parentheses. This figure illustrates the possibility of achieving the limit of 1300 mg/Nm3 NOX,
using only primary measures. Results produced in the identically designed Unit 5 arc very
similar.
It must be underlined that the general tendencies shown by the figures recording overall emission
of NOX in the different trials, were fully confirmed by local measurements made directly above
the burners, using probes especially designed for this purpose.
Type of flame (excess air). One of the most important results of the studies undertaken in Units
4 and 5 of Compostilla P.S. was that two types of flame were identified (short and long). These
were caused by differing relative proportions of secondary and tertiary air fed into the boiler
(Figure 1). The qualitative difference between these types of flame arises due to the finding of
a substantial modification in the causal relationship between the formation of NOX and the
operational oxygen level (Figure 3).
In fact, when the ratio between secondary air (S.A.) and tertiary air (T.A.) is low, as occurs with
the air settings used in the past in the above-mentioned boilers (the base condition), an inverse
dependence between concentrations of NOX and O, is observed, i.e., less NOX is formed when
oxygen levels are higher.
This condition is associated with raising of the flame due to a greater flow of tertiary air. This
gives rise to shorter and more intense flames (short flame), where combustion occurs with a less
stratified supply of oxygen, i.e., with a greater initial mix of air and coal. The generation of NOS
is thereby controlled by temperature. This means that when an increase of excess air occurs, the
flame cooling effect prevails over those deriving from higher local concentrations of oxygen,
which brings about a net reduction in the formation of nitrogen oxides.
In this sense, correlations between the average temperature within the furnace and excess air have
therefore been established. Variation coefficients of 40° C for each 1% change in excess oxygen
are obtained. The cooling effect deriving from an increase of 1% in operational oxygen excess
may be evaluated according to the ratios calculated between variations in NOX and the average
temperature in the furnace. A reduction of around 300 mg/Nm3 in NOX is produced, clearly
demonstrating the importance of this factor in the formation of nitrogen oxides in this boiler type.
On the other hand, when the S.A./T.A. ratio is higher, the flame tends to occupy the lower part
of the furnace, and therefore becomes longer, making the combustion process less intense, and
with an increasingly stratified supply of oxygen. NOX generation comes to be controlled by the
influence of local concentrations of oxygen, i.e., by the stoichiometry of the oxidisation/reduction
reactions involved in the formation of nitrogen oxides, to the detriment of the thermal control
associated with variations in excess air.
-------
to
15OO
o
E 1300
01
E
SHORT FLAME
. . LONG FLAME
(100)
Base
Case
More air F.O. support O2(3%) P.O. support
to centre O, (3%)
Figure 2: NOX emission results in outstanding cases (the limit of
1300 mg/Nm3 NOX of possible future application is
represented) (Unit 4)
B>
,
1400
>
,
1200
1000
800
5.5 6
%O2 (w.b.)
Figure 3: NOX/O2 general relationships for short and long flames
(Units 4 & 5)
-------
Thus, in this case (long flame), an increase in excess air gives rise to an increase in the
generation of nitrogen oxides. This is the opposite of the situation when a short flame is formed.
The overall tendencies in the NOX/O2 ratios shown in Figure 3 are supported by the
measurements made at the level of the arch of burners, as shown in Figure 4.
These two flame types, which may be identified at a macroscopic level by the structure of the
temperature profiles measured in the furnace, therefore present a form of behaviour that is
contrary to the basic operating parameter of total feed air. As it is stated below, this fact is
related to the influence of NOX reduction, in each case, over unit heat rate.
Additionally, it was found that the conditions leading to the lowest NO,, generation corresponded
to long flame type situations, attaining values of approximately 1400 mg/Nm3 (d.b., 6% O2). This
is equivalent to a reduction in these emissions of 20% in comparison with the initial basis (1800
mg/Nm3) (Figure 2).
On the contrary, reducing the generation of NOX without changing the flame from the short type,
using increases in the excess of overall operational oxygen or the airflow supplied to the boiler
centre, only gives rise to improvements of approximately 7%.
Fuel-oil support. Important reductions in NOX emissions were found in the two experimental
boilers when fuel-oil support at central burners was employed in long flame type situations
(Figure 5). Minor fuel-oil support (7 tons/h) together with a reduction in the excess of oxygen,
implicated a fall of NOX emissions to around 1200 mg/Nm3 (Figure 2). This is equivalent to a
35% net reduction of this parameter.
These facts may be explained on the basis of the following factors:
An increase in the temperature of the initial zone of the flame, which produces a greater
devolatilisation. The literature on this subject states that the fraction of volatile nitrogen
has a lower degree of conversion to NOX under stratified combustion conditions, such as
those existing in arch boilers.
Fuel-oil combustion consumes the available oxygen in the first zone of the flame,
creating an area rich in reducing substances in which the coal nitrogen and thermal NO
tends to produce molecular nitrogen.
A decrease in the average nitrogen content of the fuel (coal + fuel-oil) due to the lower
content of the fuel-oil (approximately 0.3 - 0.5%), which is markedly lower than that
of the coal used in Compostilla P.S. (1.4 - 1.6%).
Fuel-oil support was found to be most effective in these units when it took place in the central
area of the boiler. This is explicable due to the greater formation of NOX in this region, because
of its higher temperature levels, which make reduction more probable.
-------
Figure 4: NOX/O2 burners arch relationships for short and long flames
(Units 4 &5)
2.200
2.000
I
1.800
1.600
1.400
1.200
1.000
eoo
2,5
us, WITHOUT' P.O.
, WITHOUT P.O.
i
3,5 4 4,5
O2 ( %, d.b.)
5,5
Figure 5: NOX/O2 general relationships for cases with and without
F.O. support (Units 4 & 5, long flame)
-------
Figures 6 and 7 show the correlations obtained at the level of the arch of burners in Unit 4 both
with and without fuel-oil support, depending on type of flame. Thus when fuel-oil support is
being used, the NCyO2 ratio is found to be direct for both types of flame, i.e., the lowest
concentrations of NOX are attained for the lowest levels of excess oxygen.
This fact could be explained by the creation of more strongly reducing zones in the initial areas
of the flame, thereby increasing the relative importance of local stoichiometry in relation with
NOX production. Another possible explanation of this phenomenon could be increased
devolatilisation of fuel nitrogen, this fraction being very sensitive to local levels of oxygen
respecting its conversion into molecular nitrogen.
NOX emissions are found to be lower for short flames as the flow of supporting fuel-oil is
increased. It may be observed that the degree of reduction decreases progressively towards an
asymptotic value (Figure 6).
4.2.2 Collateral effects of reductions in NOr The measurement of NOX emissions must be
accompanied by checking of the effects of modifications on flyash carbon content and, with
greater exactitude, on boiler efficiency and unit heat rate.
In general, the hypothesis that any action aimed at reducing the production of NO,, would
produce a fall in boiler efficiency was widely accepted, this being due to an increase in the
amount of ash unburnt. Although this hypothesis was supported by observations made during
different experimental programs, it was found to be incorrect for Units 4 and 5 of Compostilla
P.S., in the light of the results obtained.
In fact, the change from the usual boiler conditions (as the base case) to the final condition (with
or without fuel-oil support) does not only bring about a significant reduction in NOX emissions
(from 20 to 35%), but is also accompanied by improvements in unit heat rate, in spite of a slight
increase in ash carbon content.
These facts are shown in Table 1, where the most relevant results of trials with a short flame are
presented (when NOX production is optimised by supplying more air to the centre of the boiler)
and final results using long flame with and without fuel-oil support.
Thus, although the levels of unbumt carbon increased about 0.8% under final conditions with
a long flame, in net terms there was a 0.3% increase in boiler efficiency. This improvement in
boiler performance is basically caused by a decrease in the volume of fluegas, due to the lesser
excess air used in the final conditions (as was pointed out above, NOX levels are decreased on
reducing the supply of combustion air for a long flame type).
The use of a lower excess of air also gives rise to a reduction in auxiliary power consumption,
which also has a positive effect on unit heat rate.
-------
° 2.000
o
CO
O
CD
1.750
1.500
1.250
.000
750
500
° Without F.O.
* F.O. 0.58 Tn/h
OF.O. 0.83 Tn/h
O F.O. 1.25 Tn/h
O2 ( %, d.b.)
Figure 6: NOJOZ burners arch relationships for different P.O. support
flowrates (Unit 4, short flame)
•o Without F.O. -X-F.O. 0.5B Tn/h
Figure 7: NOX/O2 burners arch relationships for cases with and without
F.O. support (Unit 4, long flame)
-------
In this context, making a greater use of heat transfer in the water walls, for long flame types,
brings about a drastic reduction in desuperheating spray flowrates, which in turn leads to an
improvement in turbine performance and the supply of higher quality steam to the latter.
Table 1: General data from the most important tests in Unit 4
Case
% O2 (v/v, w.b.)
% NO, (mg/Nm3, 6%
O,, d.b.)
% ash unburnt
Boiler efficiency (%)
Water walls efficiency
(%)
Desuperheating spray
flowrate (Ib/h)
Unit heat rate
(kcal/kWh)
Short flame
(air to centre)
4.28
1675'
3.22
87.67
59.09
101,700
2429.48
Long flame
(final condition,
without P.O.)
2.93
1435
4.07
87.97
67.24
19,300
2408.34
Long flame
(final condition,
with P.O.)
2.99
1223
4.02
87.90
66.34
22,600
2411.05
* 1800 mg/Nm3 in the case of a non-optimised short flame (base case)
This therefore constitutes a clear demonstration of the possibility of attaining reductions in the
emissions of NOX from anthracite burning boilers by around 30%, while also offering substantial
improvements to the economic balance of unit performance (approx. 20 kcal/kWh). The said
results are obtained by abandoning the operational solution traditionally adopted, which involves
combustion conditions that are optimised respecting the production of unbumt.
4.3 Phase II: Trials at Unit 3
4.3.1 Design and operating differences. Compostilla Unit 3 has performed very differently,
with respect to Units 4 & 5, in terms of NOX emissions over time. In this sense, an extensive
study of the design and operating differences of Units 3 & 4 have been performed with the
following main results (Figure 8):
Unit 3 (330 MW) presents a proportionally larger distance between front and rear walls
than Unit 4 (350 MW). This determines a greater specific furnace volume for that boiler.
-------
UNITS
Refractory surf.: 605 m2
S A./TA. = 0.24
V = 20.5(m3/MW)
spec v '
UNIT 4
Refractory suif: 735 m2
SA./TA. = 1.16
V = 19.6 (m3/MW)
spec v '
Figure 8: Furnace design and air distribution patterns for base cases (Units 3 & 4)
-------
Refractory lining surfaces have a lower extent in Unit 3.
Unit 3 typical coal fineness is 95% through 200 mesh, due to the new classifiers installed
in this plant, whilst Unit 4 coal size is around 88% through 200 mesh.
In Unit 3 only 2 adjustable vertical levels of tertiary air are used, whereas Unit 4 has 3
levels of tertiary air. Vertical distance from burners tips to the upper T.A. level is 3.5 m
in Unit 3, and only 1.5 m in Unit 4.
Tertiary and secondary air distribution is substantially different in these units. As it can
be noted in Figure 8, the S.A./T.A. ratio, obtained from windbox modelling, is much
lower in Unit 3 for the base case. Additionally, Unit 4 presents a larger proportion of S.A.
supplied through the vents.
4.3.2 Effects of modifications in air distribution. Tests campaign udertaken in Unit 3 has
produced very important results in order to characterize NOX formation and heat rate fine tuning
in this boiler. Nevertheless, this paper will only emphasize those results which allow a
comparison with the combustion patterns of Unit 4 (similar to those of Unit 5).
In this sense, the most relevant effect determined in Unit 3 is that produced through the
modification of the S.A./TA. ratio. Table 2 and Figures 9 and 10 show comparisons of Units 4
& 3 performance when varying this ratio.
Table 2: Comparison of performance data from most significant tests with variation of
the S.A./T.A ratio (Units 4 & 3)
Case1
U4. Base Case
U4. Long flame
U3. Base Case
U3. SA/T.A.t
U3. S.A/T.A4
S.A.AT.A.2
1.16
3.21
0.24
0.41
0.21
NO,
(mg/Nm3,
6% O2)
1767
1635
1198
1441
1014
Desuperheating J
spray flow
101700
29500
38.90
14.50
42.60
Ash
unbumt
(%)
3.2
3.8
8.3
5.9
9.2
Boiler
effic.
(%)
87.67
87.35
85.60
86.61
85.07
Heat rate
(kcal/kWh)
2429.48
2423.56
2498.42
2463.05
2526.29
1 Oxygen excess: U4: 4,1% ; U3: 4,6%
2 Flowrate ratio from windbox modelling
U4 (Ib/h) ; U3 (%)
-------
%O2 (w.b.)
Figure 9: NOX/O2 general relationships for Units 3 & 4 when modifying
the S.A./T.A. ratio (B.C.: base case; S.F.: short flame;
L.F.: long flame)
6,5 7,5 8,5 9,5
, d.b.)
Figure 10: NOX/O2 burners arch relationships for Units 3 & 4 when
modifying the S.A./T.A. ratio (B.C.: base case; S.F.: short
flame; L.F.: long flame)
-------
The following features could be established from the analysis of these data:
NOX emissions are around 1700 mg/Nm3 for Unit 4 (at 4,1% (X), whereas these values
are significantly lower for Unit 3 (1000-1400 mg/Nm3) for an oxygen excess of 4,6%.
S.A./T.A. ratios are much lower in Unit 3, although substantial variations in air dampers
openings are applied. The combined effect of this factor and the longer distance for T.A.
supply in this unit seems to produce a higher degree of combustion stratification, as it can
be concluded from the lower NOX emissions and higher ash unburnt levels for Unit 3. In
this sense, refractor}' extension and specific furnace volume seem not be the key
parameters to explain Units 3 & 4 differences, as furnace average temperatures are similar
(lower variations than those determined by the slagging influence). Significant effect of
other factors like coal fineness or oxygen excess might also be discarded as they would
determine a contrary trend of NO, and unburnt results.
Increasing the S.A./TA. ratio in both units produces a reduction in desuperheating spray
flows, which positively affects to turbine efficiency and, therefore, to unit heat rate. This
reduction is due to the higher heat exchange in the lower furnace.
Increasing the S.A./T.A. ratio seems to determine a less stratified combustion process in
Unit 3, as NOX emissions are higher and ash unbumt is lower in these conditions. These
relationships are the opposite for Unit 4, most likely due to the very different air supply
patterns of this unit, with T.A. addition much nearer the burners tips. This fact produces
short, intense, and temperature controlled flames when the T.A. flow is high (base case),
whilst, for lower T.A. supplies, flames tend to ocuppy the lower furnace, with an
increasingly stratified supply of oxygen.
Increasing the S.A./T.A. ratio has a positive effect on heat rate for both units, although
this influence is much lower in Unit 4, where the higher unburnt levels in this condition
give rise to a penalization in boiler efficiency.
5. Economic implications
The economic implications of the results obtained in Phase I (Units 4 & 5) are very important.
The current costs of reducing NOX levels (taking into account operating costs as well as
investment write-off) to a level similar to that attained in these units (by around 30%) are
approximately $400 per ton of NOX eliminated. This cost corresponds to using primary measures
according to the most recent evaluations published around the world.
If it were necessary to use secondary measures, this cost would be increased by about $2,000 per
ton of NOX eliminated.
Nevertheless, the expectations arising from the RNA Project indicate that it is possible to attain
the above-mentioned reduction in NOX emissions in anthracite - burning boilers without the need
to make any additional investments, and even with a reduction in operating costs, that for
Compostilla Units 4 & 5 might be evaluated in 200 $/Tn NOX eliminated (1,400,000 $/year).
-------
On the other hand, results obtained in Phase II (Unit 3) show the importance of air distribution
design for NOX emissions control in arch-fired furnaces. Additionally, the significant variability
of NOX emissions and unit heat rate, depending on air supply configuration, permits to operate
this boiler according to different criteria: minimisation of heat rate, minimisation of heat rate for
NOX emissions below 1300 mg/Nm3, minimisation of NO, emissions, etc. These strategies might
be decided on the base of the economic implications, in each case, of heat rate optimisation and
NOj control.
6. Conclusions
The fundamental conclusions arising from the results of the RNA Project are:
They demonstrate the possibility of reducing NO^ emissions in arch-boilers consuming
anthracites by approximately 20 - 35%, using only primary measures. This would
represent for Compostilla Units 4 & 5 the possibility of achieving the 1300 mg/Nm3 NOX
limit, without applying secondary measures.
They demonstrate the compatibility, in some cases, of this said reduction in emissions
with the attainment of noticeable improvements in units heat rate. These improvements
are obtained by moving away from standard operating criteria aimed at reducing ash
carbon contents to a minimum, and using long flame types of combustion and lower
excess air, thereby improving the net performance of boilers and turbines.
They also demonstrate the importance of air distribution design for NOj control and heat
rate improvement, with regard to the operating modifications to be implemented for these
aims.
7. Acknowledgements
The authors gratefully acknowledge the financial support of OCIDE, OCICARBON and ECSC.
They would also like to thank ENDESA for granting permission to publish this paper.
REFERENCES
1. J.F. GONZALEZ, E. MENENDEZ, P. OTERO, L. CANADAS and F. RODRIGUEZ.
"Improving Coal Combustion and Reducing NOX Emissions in Pulverized Coal Boilers"
Proc. of Powergen/97 Europe Conference. Perm Well (1997).
2. A. PLUMED, L. CANADAS, P. OTERO, J.F. GONZALEZ and F. RODRIGUEZ. "NOX
Reduction in Low Volatile Coals Combustion" 8th Int. Conference on Coal Science.
Elsevier Science B.V. (1995).
3. S. MIYAMAE, T.KIGA and K. MAKING. "Application of Low NOX Combustion
Technologies to a Low Volatile Coal Filing Boiler" Proc. of Joint Symposium on
Stationary Combustion NOX Control, pp 3-11/3-123 (1989).
4. T.W. SONNICHSEN and J.E. CICHANOWICZ. "NOxEmissions Characteristics of Arch-
fired Furnaces". Proc. of Joint Symposium on Stat. Combustion NOX Control (1981).
-------
Monday, August 25; 3:30 p.m.
Parallel Session C:
Low-NOx Systems for Gas/Oil-Fired Boilers
-------
DEVELOPMENT, TEST
AND INDUSTRIAL APPLICATION
OF ADVANCED LOW NOx BURNERS
R. De Santis, M Alberti, R. Rizzi
Ansaldo Energia - Legnano - Italy
A. Saponaro, M.Martano
Termosud SpA - Gioia del Colle - Italy
S. Ligasacchi, S. Pasini, E. Tosi
ENEL Ricerca - Polo Termico - Pisa - Italy
Abstract
Over the last decade Ansaldo Energia and ENEL have worked closely to implement power
plant NOx Control Technologies which have subsequently been applied both in Italy and
abroad in over 20,000 MWe. Among the different technologies, which include air and fuel
staging, a major role is played by the use of Low NOx burners. Ansaldo Energia and ENEL
have developed and patented an oil/gas LNB, named TEA (Three-stream ENEL-Ansaldo)
which has shown excellent performance and reliability in over 3,000 MWe. Further
improvements have led to the development, testing and qualification of a new product,
named TEA 2, characterized by lower pressure drop, higher NOx reduction, simplified
mechanical design.
This burner is now the reference product for all ENEL and Ansaldo Low NOx retrofits.
A coal/oil version of this burner, named TEA C, has also been developed and has shown
outstanding NOx reduction capability, in the order of 50-^55% with respect to the original
circular burners emission with simultaneous good control of unburned carbon in fly ash.
The paper outlines the design philosophy of these burners, the qualification tests and the
industrial applications.
Introduction
The use of LNB's as a NOx control option is well established and, in the USA, has been
specified in the legislation. LNB's can be used as the sole NOx control method or as part of
more sophisticated combustion modifications which range from OFA to Rebuming.
ENEL and Ansaldo have long been involved in a cooperation program to develop both LNB's
and combustion systems capable of satisfying the strict Italian legislation limits that, for plants
rated more than 500 MWth, are NOX<200 mg/Nm3 corrected at 3%02 for oil and gas and at 6%
02 for coal.
-------
The first LNB model named TEA 1, suitable for firing oil and gas, was developed by Ansaldo
and ENEL in 1990. Details of the development program can be found in Ref. (1).
Extensive plant experience (see Table 1) confirmed a NOx reduction greater than 50% in
comparison with the pre-retrofit values without significant increase in CO and particulate
emissions, but also showed the possibility of introducing further improvements among which:
• reduction of pressure drop between windbox and furnace to obtain a product requiring
lower ID fan head, and therefore smaller size of the pressure parts openings (compatible
with the pre-retrofit ones);
• simplification of mechanical design, reducing the moveable parts, and also reducing
manufacturing costs;
• additional NOx reduction;
• inherent possibility of adjusting air balancing among the different burners.
An improved version of TEA, meeting the above goals, named TEA MK2, has been
developed and has replaced TEA 1 as the reference product of ENEL for all environmental
retrofits of its plants. Details of the development program and of the burner design are
reported after.
At the same time a new burner for coal and oil, named TEA C, was also developed to meet
the upcoming needs of ENEL for Low NOx retrofits of its coal fired plants. This burner has
taken advantage of the optimization of the aerodynamic structure of TEA MK 2 for oil and
gas, so that secondary and tertiary air registers are common, in their basic design, for both
versions.
Extra effort has been devoted to the design of a coal nozzle capable of providing not only
NOx reduction but also the unburned carbon control. The development and qualification of
this burner is also described.
T.E.A. MK2 - An Improved Version of T.E.A. Low-NOx Burner
General
In improving the design of the oil/ gas burner TEA, the same kind of approach already
employed for the development of the MK1 version was followed. The flame's aerodynamic
structure was improved, with the aim of further staging combustion, obtaining at the same
time a mechanical simplification in order to reduce the construction costs.
The new burner configuration was defined by a theoretical and experimental approach based
on the use of mathematical modeling, fluid-dynamic characterization of isothermal models
and the use of various experimental furnaces to optimize the combustion behavior.
The most modern software and hardware tools were employed for theoretical studies while,
for the experimental tests, the ENEL plants of Livomo and S. Gilla (CA) and the Ansaldo pilot
plant of Gioia del Colle (BA) were utilized.
Base studies - Near Field Aerodynamics and Mathematical Modeling
Based on plant experience some assumptions have been made that suitable near field
aerodynamics would have allowed improvement of the combustion efficiency, increase
burner turn-down and possibly reduce NOX. In the mean time a significant reduction of
pressure losses could be obtained using fixed high efficiency swirlers in lieu of the spin
vanes. The conversion of the tertiary air entrance from radial to axial could allow the use of a
single barrel to regulate the entire flow of combustion air to each burner.
-------
Preliminary full scale (35 MWth) single burner tests on a TEA MK1 model, run in S. Gilla
ENEL test plant, showed that modification of the burner front could achieve substantial
improvement in terms of residual oxygen and even NOx reduction.
A comprehensive program was therefore started to examine the behavior of different burner
exit plane geometries, versus calculated (FLUENT code) and measured (laser velocimetry)
flow field. A good correlation was found between the data obtained with these techniques.
Fig. 1 provides an example of the measured velocities for a given geometry.
The measurements have been performed on the ENEL CAUX-F cold flow test rig (2).
Base studies - Atomization
A significant amount of work has been devoted to atomization studies aimed to select the
best atomizer for burner performance.
Tests have been run on the ENEL test rigs named ISA and ATOMO 1 in Livorno (3). Specific
atomizer designs have been developed among which a "twin hole" atomizer capable of
producing in-flame fuel staging.
A set of atomizers is available for use on the MK 2 for specific application i.e. maximizing
unbumed particulate control.
Tests at S. Gilla on a modified TEA MK 1
A modified TEA 1 burner, incorporating modification of burner front, the change of secondary
and tertiary spin vanes to high efficiency swirlers, and the use of innovative atomizers, has
been extensively tested at S. Gilla Large Burner Test Facility (4) (fig. 2), allowing the
following conclusions to be drawn:
• atomizing quality was a parameter of paramount importance for both NOx reduction and
residual 02 control;
• the expected burner pressure reduction was achievable with the use of high efficiency
swirlers;
• further NOX reduction under oil firing, in respect to TEA 1, could be obtained acting on the
primary air impeller.
These conclusions were the guidelines for the TEA MK 2 design.
TEA MK 2 Prototypical Design
A prototypical design of TEA MK 2 burner has been prepared following these guidelines:
• flow areas and exit geometry for primary, secondary and tertiary air have been maintained
as per model TEA 1 to preserve the same near field aerodynamic profile;
• registers and swirlers have been changed to get the goals of lower pressure drop,
manufacturing simplification and increase of the setting capability of the burner.
In detail, the following changes have been made (see fig. 3):
• use of a single barrel register controlling the entire air flow supplied to the burner;
• change of tertiary air inlet from radial to axial with compound profile and in fixed position;
• modification of secondary air swirlers in fixed compound profile with a second moveable
part to allow some modification of the swirl with a reliable mechanical drive;
• design of a special primary air swirler as defined in previous cold tests on the CAUX-F test
rig.
This prototype has been tested at S. Gilla firing oil and, after some further optimization, was
frozen in design and finally tested at Gioia del Colle firing both oil and gas.
-------
S. Gilla full scale combustion tests
The first prototype of the TEA MK 2 full scale burner has been tested in S. Gilla beginning in
December 1994. The tests have been carried out, mainly firing oil, leaving the conduction of
gas firing tests to Gioia del Colle.
The tests allowed the optimization of burner geometry with particular reference to:
• angle of deflector located between secondary and tertiary air;
• geometry of secondary and tertiary swirlers;
• geometry of primary air impeller.
Fig. 4 shows the comparison of the results obtained at S. Gilla with TEA 1, TEA MK 2, and a
barrel burner having the same thermal capacity.
It can be noted that the capability of NOx reduction obtained with the TEA 1 (about 50% in
respect to barrel burner) has been further improved in the MARK 2 version.
CO emissions of the two versions are similar, even though they are higher for the latter,
confirming that the price to pay for a significant reduction of NOx corresponds to a slight
increase of final 02.
It should be pointed out that the S. Gilla test rig is characterized by a furnace temperature
notably lower than that encountered in power plants, because the tests are performed at a
burner load of about 50 MWth, that is at 50% of the boiler load. For the same reason, the
residence time is also higher.
Measured values of air flow rate vs. burner pressure losses show that, in respect to the
version MK 1, the TEA MK 2 burner is characterized by pressure losses at least 30% lower.
Gioia del Colle combustion tests
The combustion tests were performed at Gioia del Colle utilizing a 48 MWth furnace,
designed to have the same residence time, and partially refractory lined in order to have the
same exit gas temperature of industrial power plants.
These tests aimed to replicate oil firing tests and also to investigate the gas firing conditions.
Oil firing Combustion Tests. During the tests, some differences have been encountered
between the results obtained at S. Gilla, making a further optimization of the burner geometry
necessary. The best results, reported in fig. 5, have been obtained with an atomizing angle of
80° and 90°
Comparing these data with those obtained in S. Gilla, one can find the same trend, with a
substantial equality in terms of NOx emissions, that appear slightly improved in the MK 2
version.
Gas Firing Combustion Tests. The results obtained in gas firing combustion tests are
synthesized in fig. 6. The various curves correspond to different orientations (clockwise,
counterclockwise) of the gas spuds.
The best results in terms of NOx vs. smoke point have been obtained with gas spuds
oriented radially with respect to the burner axis.
The influence of flue gas recirculation (gas mixing) on emissions is shown in fig. 7; one can
see that the decrease of oxygen concentration at the windbox (that is an increase of flue gas
recirculation) leads to a dramatic reduction of NOx emissions, without significant increase of
smoke point.
Orimulsion Combustion Tests. The results obtained during the combustion tests with
Orimulsion are summarized in fig. 8. These tests have demonstrated the full ability of the
burner to operate with this particular kind of fuel with limited NOx emissions (utilizing EPT V-
-------
Jet atomizers, Ref. 5), or with higher emissions, but much lower oxygen concentration,
utilizing PG F-Jet atomizers (Ref. 6).
The burner is therefore qualified and available on the market for this kind of fuel.
Pressure Drops. As a result of further modifications to the burner, the resulting pressure
drops measured at Gioia del Colle were even lower than those encountered at S. Gilla, with a
reduction of 37% in respect to TEA MK 1.
This fact is extremely important because, under the same nominal burner rate, it allows the
reduction of the quarl diameter, allowing in most cases the burner retrofit in the existing
openings without any modification to the pressure parts.
Development of TEA C
General
Low NOx burners currently on the market are valid products, although far from having
exploited all NOx reduction potential included in their concept. NOx values that can be
obtained are in the order of 600-800 mg/Nm3 @ 6% 02 in conditions of boiler operability, i.e.
without impairing other essential values like CO and UBC. These values in Europe are over
the admissible limits whereas in other countries like USA, limits are under revision in view of
attaining significantly lower values than those accepted today, thus requiring the use of
retrofits with more sophisticated technologies like air or fuel staging.
Further development of LNB's is therefore needed in the power industry especially for coal
LNB's which represent the most difficult exercise. To properly address the problem it must be
remembered that the aerodynamic flow conditions needed in a Low NOx coal burner should
be such as to obtain maximum volatile release in flame zones with low oxygen partial
pressure, also to obtain release of organic nitrogen. This lack of oxygen in the primary
combustion zone endangers combustion and ignition velocity thus provoking the increase of
unburned carbon in the ashes collected in the ESP (the Italian maximum values of UBC for
ash recovery-cement industry is 7% expected to be 5% in future regulations).
Published studies and direct experience of both ENEL and Ansaldo, show that the
segregation of coal particles along the burner axis as a function of their granulometry and
their feeding into the most appropriate flame zones allow significant NOx reductions with
simultaneous control of unburned carbon. ENEL and Ansaldo have therefore started a
development program to provide a coal/oil "Low NOX" type burner characterized by a NOx
reduction capability in the order of 50%, compared to circular burners (same order of
magnitude of TEA oil/gas), but also capable of maintaining low UBC levels.
The product design goal has been defined as NOx < 650 mg/Nm3 @ 6% 02 with UBC less
than 7% when burning a medium volatile bituminous coal (i.e. South African). The work
development followed the logic reported below:
• analysis of the products on the market (bibliography; analysis of existing plant data;
performance analysis etc.);
• numerical-analysis (isothermal simulation of coal nozzles);
• cold physical modeling of the coal nozzle using the ENEL CONOS test rig (quantitative
analysis of coal particle distribution with different coal nozzles designs - comparison to
numerical analysis data);
• conceptual design of the new burner;
-------
• small scale and full scale combustion tests (33 MWth at Ansaldo Combustion Centre of
Gioia del Colle);
• long-term performance tests at the ENEL Test Station in S. Gilla (~40 MWth)
Analysis of the Products on the Market
Work began with a review of the U.S. market, which showed a large presence of Low NOx
retrofit applications diversified in type and number of Low NOx burners installed alone or
coupled with other systems like OFA.
This reviewed showed some characteristics common to the different burners designs:
• two external air flows to control the flame shape and consequently the mixing between fuel
and air as the prime factor for NOx control (staged combustion);
• independent controls for swirls and for total air flow;
• use of simplified air registers to improve mechanical reliability and to reduce pressure
drops;
• special coal nozzles which allow segregation of coal and primary air to obtain a localized
staged combustion;
• operational flexibility and possibility of field regulation of the burner critical parameters
(flame shape, combustion efficiency, NOx).
All these aspects have been examined and addressed in the new burner design leading to
the final design choices.
Burner Structure
The burner prototype includes the external geometry of the oil/gas TEA MARK 2 burner and
has a newly developed coal nozzle located in the primary duct.
The secondary and tertiary air registers had already been studied in-depth during the TEA
MK 2 (oil/gas) burner development. The results obtained provided a good understanding of
the problem and the necessary know-how to interpret the experimental results and transform
them into the design of the TEA -C. It was thus possible to rationally reuse the same air
registers of TEA MK2 for TEA-C and only change the different coal nozzles selected as best
candidates from numerical and physical modeling. The burner (fig. 9) became modular
(together with TEA MK2) and standardized, thus providing margins for cost reduction and
retrofit application.
Numerical Analysis
Following the published data and market analysis, an in-depth numerical 3D analysis of the
different coal nozzles was performed using the FLUENT code.
The results constituted the basis for the conceptual design of the new coal nozzle and also
allowed to gain valuable information from the hot full scale tests run both in Gioia del Colle
and in S. Gilla. Both numerical and physical (see next paragraph) modeling provided good
correlation in terms of velocity profiles and coal concentration.
Cold Physical Modeling
It has already been noted how coal segregation at burner level as a function of its fineness is
of paramount importance to obtain NOx reduction and simultaneous unburned control.
Consequently physical modeling has been considered necessary to confirm numerical data.
An "ad hoc" test rig, located at ENEL laboratories in Livorno named CONOS, was developed
and used. The rig was originally designed for calibration of innovative coal flow measuring
-------
devices; it has been redesigned and adapted for the study and characterization of the
different primary ducts, in particular for evaluating the influence of the coal pipe internals.
Fig. 10 reports a typical result in terms of effective coal distribution at the burner exit.
Combustion Tests at Gioia del Colle
Tests at Gioia del Colle can be subdivided into two distinct phases:
• A first phase when different coal nozzles (with related internals) were hot tested using an
existing burner body;
• A second phase when the TEA C prototype was tested to optimize its design.
Coal Tests. The results obtained using the existing burner body have confirmed that the coal
pipe internals as well as the burner exit zone to the combustion chamber are of vital
importance for low NOx and simultaneous UBC control. These tests have shown that a
simple impeller located at the burner exit substantially modifies the burner performance, even
for a low NOx design, and that the secondary and tertiary air registers do not quantifiably
modify said performance.
First tests have been run comparing the response of a typical Low NOx burner in a high-NOx
(low oxygen) configuration obtained inserting at the exit an impeller (configuration A), versus
TEA C, in the same configuration. As can be seen in fig. 11 similar NOx values are reached,
but a definite advantage for TEA C in terms of excess O2. This validated the assumption
made of maintaining for TEA C the same secondary and tertiary air registers of TEA MK 2.
The first Low NOx configuration of TEA-C (configuration B) was intended to provide a coal
segregation from primary air by forming zones with controlled stoichiometry. Results have
immediately confirmed the design assumption showing the wide capability of the burner to
control its performance as a function of coal nozzle settings and coal fineness (fig. 12); with
this configuration parallel NOx and UBC reduction has been obtained increasing coal
fineness.
Continuing the search for a geometry capable of segregating coal and providing the best
compromise between NOx reduction and UBC control, a final configuration (G) has been
implemented incorporating specific coal pipe internals and an innovative nozzle at pipe exit,
able to create maximum coal segregation. This was done with a device whose position in
respect to the exit sections of the coal created by the particular nozzle, determines a specific
local air/coal ratio.
This configuration gave the best results, especially without rotary classifier (fig 13).
Reference conservative values for standard coal fineness are:
• operational excess oxygen - 3.5%;
• NOx - 650 mg/Nm3 @ 6% 02 (about 50% reduction);
• UBC < 7% (referred to South African Coal).
Fuel Oil Tests. No. 6 oil tests have also been performed. The configurations tested were B
and G, the most promising in coal combustion. The results (see fig. 14) are in good
agreement with TEA oil/gas performance and show the influence of type and position of the
atomizers and of primary air flow.
S. Gilla Tests
Based on the results obtained at Gioia del Colle, a prototype TEA-C burner was designed,
fabricated and installed at the test section located at the bottom of the group No. 1 of the
ENEL Power Station of S. Gilla (35 MWth).
-------
The burner has the same dimensions of the one tested in Gioia del Colle and in terms of
secondary and tertiary air registers was identical to the TEA MK 2 oil/gas. Combustion tests
were conducted at 120% of thermal load (40 MWe) to allow boiler operation with only the
bottom burner fired. Thermal load was 1.2 times the one applied in Gioia del Colle tests.
For each burner arrangement NOx, CO, S02 and UBC, as a function of excess oxygen and
coal fineness, have been monitored.
Initially only two configurations have been tested, the reference (A) and the Low NOx (G). In
configuration A, same NOx emissions as measured in Gioia del Colle, have been found.
In configuration G, the burner incorporates the innovative coal nozzle that gave the best
results at Gioia del Colle.
Long term tests on the TEA-C aimed to:
• measure the temperature of the parts exposed to flame;
• assess wear and temperature stress;
• assess time stability of the performance characteristic.
Metal Temperature. Metal temperatures have been assessed on the secondary flame
holder, on one of the coal nozzles and on the deflecting flaps.
At burner' s MCR and firing coal, temperatures did not exceed 400 °C. Same temperatures
were measured with additional 4 tangential burners in service to assure boiler nominal load.
Firing oil at burner' s MCR, the highest temperature measured on the flaps was 600°C. At half
load and with 4 tangential burners in service temperature has been measured at slightly over
800 °C.
Finally with the test burner out of service and with 6 tangential burners in service,
temperature data were similar to those measured at MCR.
Wear of Components. Wear of coal nozzle is recognized as a critical factor and is due to
pure mechanical wear and to synergistic action of oxidation. Due to the TEA C coal nozzle
design, which provides separation of air rich and coal rich flows, the wear potential is low.
This assumption has been confirmed by the tests run in Gioia del Colle where the coal
nozzle was built in carbon steel and did not show any significant wear after more than 500 h
of service. To protect it from the wear risk the coal nozzle is built in high Chromium refractory
cast alloy whereas the deflecting cone, hypothetically exposed to temperatures up to 800 °C
is made by Stellite 6 Alloy.
Time stability of performance. No problems for the start-up and operation stabilization
have been detected either in S. Gilla or in Gioia del Colle.
Conclusions
TEA 2 oil/gas
TEA 2 is a mature product which has replaced TEA 1 as ENEL/Ansaldo reference product for
Low NOx retrofits (LNB's; OFA and Reburning).
Table 2 provides the reference list of plants already equipped with this burner. The large
experience of TEA 1 guarantees that plant performance will be fully satisfactory.
-------
TEAC
TEA C is now the reference product of ENEL/Ansaldo for coal firing applications. TEA C is in
the installation phase in four plants (Table 3) and is expected to be used shortly in at least
another two. Qualification is now based on single burner tests but plant data will be available
starting end 1997.
The TEA C design has also the potential of retrofitting the coal pipe and coal nozzle to coal
pipes of older and unsatisfactorily performing burners. A demonstration is expected at the
ENEL Plant of Bastardo (75 MWe) in early 1998.
References
1. G. De Michele and R. De Santis, "Development and Industrial Application of an Oil and
Gas Low A/Ox Burner," paper presented at the Italian Technology Week, December 1993,
Chicago U.S.A.. [conferencepaper]
2. G. Benelli, G. De Michele, S. Ligasacchi, L. Miglietta, G. Tanzini and A. Zennaro,
"Aerodinamica del Bruciatore a bassa produzione di ossidi di azoto TEA," 45° Convegno
Nazionale ATI (1990), S. Margherita di Pula, Italy, [conference paper]
3. G De Michele, M. Graziadio "Recent developments on liquid fuel atomization at ENEL
C.R.T.N."paper presented at Combustion Institute 1992, Capri, Italy, [conferencepaper]
4. G. De Michele, S. Ligasacchi, S. Pasini, A. Tozzi and G. Trebbi, "Characterization of a
Large HFO-CWF Dual Fuel Burner", 8th Members Conference, Noordwijkerhout, The
Netherlands, May 28-30, 1986. [conference paper]
5. Electric Power Technologies, Inc. "Segmented (Bi-Sector) V-Jet Atomizers for Evaluation
by Enel and Ansaldo in S. Gilla #1 Boiler", March 8, 1994.
6. R. Tombs, A. Jones "PowerGen Pic - A commercial user of Orimulsion" paper presented at
The Proceedings of the 21st International Technical Conference on Coal Utilization & Fuel
Systems, March 18-21, 1996, Clearwater, Florida, U.S.A. [conference paper]
-------
Tab. 1 - TEA 1
PLANT
MONFALCONE83
TURBIGOtfl
CASSANO # 1
PONTI
SULMINCIO81
BRESCIA ti 2
ASSYUT
MEDINA
MONTALTO DI
CASTRO
LA CASELLA H 4
KERATSINI # 8
CLIENT
ENEL
(ITALY)
ENEL
(ITALY)
AEM MILANO
(ITALY)
ASM BRESCIA
(ITALY)
ASM BRESCIA
(ITALY)
EGYPTIAN ELECTRICITY
AUTHORITY
(EGYPT)
SALINE WATER
CONVERSION C.
(SAUDI ARABIA) (2)
ENEL
(ITALY)
ENEL
(ITALY)
POWER
PUBBLIC CO.
(GREECE)
PLANT OUTPUT
(M\Vc)
1x320
2x240
1x75
1x80
1 x230(/h
1x300
2 x 165 kg/8
2x660
1x320
1x160
BOILER/
BURNER
(MWt)
OF - 18x41
FF - 16x42
FF - 6x31
FF - 9x21
FF - 4 x 42
OF - 16x61
FF - 12 x 42
OF - 56x32
OF - 18x41
FF - 8 x 5 1
CONFIGURATION
/
FUEL
LNB (1)
OIL
LNB
OIL
REBURNING
OIL & GAS
LNB
OIL & GAS
LNB
OIL & GAS
LNB
OIL & GAS
LNB
OIL & GAS
LNB+OFA+GM
OIL & GAS
REBURNING
OIL
REBURNING
OIL & GAS
ORIGINAL NOv
(ing/NniJ3%OI)
(6% O2 for toal)
930
600
son
600
800
800
..
--
950
500 (3)
FINAL NOx
(ing/Mm1 3% O2)
(6% O2 for conl)
400
400
(3) 120
(4) 80
(5) 400
(3) 290
(5) 430
(3) 27S
450 (S)
462 (5)
318 (3)
expected
100(4)
prulimEnHry result
400 (5) I.NB only
<200 (5,7)
< 200 (5,7)
<150(3,7)
expected
OPERATION
DATE
1990
1992
1994
1995
1995
1996
1997
1997
1997
1997
THROAT
DIAM.
42.5"
40"
34"
30"
46"
50"
40"
40"
42.5"
42.5"
i) LNB = LOW NOx BURNERS 3) FIRING GAS
OFA = OVERFIRE AIR PORTS 4) FIRING GAS WITH GM
GM = GAS MIXING 5) FIRING OIL
2) DESALINATION PLANT 6) FIRING COAL
7) REBURNING
-------
Tab. 2 - TEA 2
PLANT
CASSANO tt 2
BRESCIA # 1
LACASELLA#3
LACASELLA81-2
SKRMIDEff 1-2-3-4
TAVAZZANO # 5-«
CARREGADOW 5-6
CLIENT
ENEL
(ITALY)
AEM BRESCIA
(ITALY)
ENEL
(ITALY)
ENEL
(ITALY)
ENEL
(ITALY)
ENEL
(ITALY)
CPPE
(PORTUGAL)
PLANT OUTPUT
(MWc)
Is 320
1 Jil75t/h
1x320
2x320
4x320
2x320
2x125
BOILER/
nURNER
(MWt)
OP - 18x40
FF - 4 x 32
OF - 12x44
OF - 12x44
OF - 12x44
OF - 12x44
FF - 6 x 53
CONFIGURATION
/
FUEL
REBURNING
OIL & GAS
LNB (1)
OIL & GAS
REBURNING
OIL
REBURNING
OIL & GAS
REBURNING
OIL & GAS
REBURNING
OIL & GAS
LNB
OIL & GAS
ORIGINAL NOx
(mg/Nni1 3% 0,)
(6% O2 for coal)
800 (5)
650 (3)
750 (5)
950 (5)
950 (5)
950 (5)
950 (5)
950 (5)
FINAL NOx
(mg/Nnr'3%0;,)
(6% O2 for coid)
<190 (5,7)
<170 (3,7)
expected values
460 (5)
2 SO (3)
< 200(5,7)
expected
< 200(5,7)
expected
< 200(5,7)
CXIWcU'd
< 200(5,7)
expected
< 450 (5)
<350 (3)
OPERATION
DATE
1998
1997
1997
1998
1998
1997/1998
1997
THROAT
IMAM.
38"
34"
40"
40"
40"
40"
44"
Tab. 3 - TEA C
PLANT
SULCIS » 3
VADO LIGURE # 4
VADO LIGURE #3
WEST JAVA
CUF.NT
ENEL
(ITALY)
ENEL
(ITALY)
ENEL
(ITALY)
DAYALISTRIK
PRATAMA
(INDONESIA)
PLANT OUTPUT
(MWc)
1x240
1x320
1x320
1x400
BOILER/
BURNER
(MWt)
FF - 24 x 26
OF - 24x35
OF - 30x28
OF - 30 x 44
CONFIGURATION
/
FUEL
LNB
OIL & COAL
LNB
OIL & COAL
LNB
OIL & COAL
LNB COAL
ORIGINAL NOx
(niB/Nm1 3% O2)
(6% O2 for coal)
1400 (6)
1300 (6)
1300 (6)
FINAL NOx
(mg/NmJ3%Oi)
(f>%O2forcoul)
<700 (6)
expected
< 425 (6,7)
expected
<700 («)
expected
<650 (6)
expected
OPERATION
DATE
1997
1998
1998
2000
THROAT
DIAM.
40"
38"
36"
43.5"
1) LNB = LOW NOx BURNERS 3) FIRING GAS
OFA = OVERFIRE AIR PORTS 4) FIRING GAS WITH GM
GM = GAS MIXING 5) FIRING OIL
2) DESALINATION PLANT 6) FIRING COAL
7) REBURNING
-------
-200
-150 -100 -50
distance [mm]
Fig. 1 - Measured velocities for a given geometry
Fig. 2 - Santa Gffla Large Burner Test Facility
TEA MK 1 BURNER TEA MK 2 BURNER
Fig. 3 - Cross section comparison for TEA 1 and 2
_ 800
600
Fig. 4 - Comparison between IS Oi and CO emissions from TEA1, TEA2 FiS- 5 - Comparison between NOi and CO emissions from TEA1 and
and Barrel burners at S. GiDa TEA2 burners at Gioia del Colle
-------
I «
1 350
f 300
-------
1000
o
#
01234567
02[%]
Fig. 11 - ComparisoD between TEA-C and a typical low-iVOx burner
both in configuration A
-180, 200rprr
0, 200
o
D
-60. 200
•mo, 200
Fig. 12 - TEA-C in B configuration - Main results
K* Ton .
0 IIJJ
g 600 -
® 500-
O)
E 300-,
o pnn -
Z ^JU
C
D
A3
1
Orpm
Mrpm
2
J
q
j
^
"
c
tTF
JM
S^
^
32 P/.
r
&
( E
> e
;
15 8
r"
r
[ X-530™^. APMflt* ^-530 mm, AFWSUh O3M """. APM 6 Uh [
| D-S30 mm. APM ^ Uh A"330 """, APM 4 ">
Fig. 13 - TEA-C in G configuration - Main results
Fig. 14 - TEA-C in G configuration - Fuel oil tests
-------
ULTRA LOW NOX OPERATION
FROM A 185MW OIL AND GAS
T FIRED BOILER
JW Allen
PRBeal
International Combustion Ltd
Sinfin Lane, Derby, UK
C J Conboy
JARigert
LILCO
Melville, N.Y. USA
Abstract
Vertical air staging has been used as the principal means of reducing NOx
emissions on an ultra low NOx programme being applied to the (185 MW) oil
and gas T fired No. 4 Unit located at Port Jefferson on Long Island. The
technique, which has been given the acronym TAS (tilted air supply) was
initially aimed at achieving NOx levels of 0.15 lbs/106 Btu, from a baseline of
0.22 lbs/106 Btu, when oil firing over the load range 40-185 MW.
Prior to the site installation the potential of the TAS NOx reduction technique
was demonstrated on the full scale combustion test facility located at the
International Combustion Ltd (1C Ltd) Derby, UK site, under both oil and gas
firing conditions simulating the Port Jefferson No. 4 Unit operations.
The test facility demonstration was successfully translated to the site
installation, meeting the NOx reduction requirements. The possibilities of
further NOx reductions based on increased air staging, the utilisation of
recycled flue gases and re-bum are also considered.
Plant Description
The Port Jefferson No. 4 Unit is a CE designed single drum, single cell, semi-
enclosed tangentially fired boiler with four levels of steam atomised burners.
Originally designed to burn bituminous coal at a MCR of 150 MW, the unit
has operated on No. 6 oil since the 1960's with a MCR of 185 MW. In 1997
the burner box on a sister Unit No. 3 was modified to include close coupled
over fire air (CCOFA) as an initial NOx reducing exercise targeted at NOx
levels around 0.2 lbs/106 Btu. Based on the achievement of the targeted NOx
level Unit 4 modifications were progressed using both CCOFA and TAS
targeted at a 0.15 lbs/106 Btu NOx level.
-------
Flue gas recirculation (FOR) is used for steam temperature control with the
FOR ducted to the bottom ash hopper and the unit is also equipped with an
electrostatic precipitator. A maximum particulate loading of 0.2 lbs/106 Btu is
specified for the precipitator operation, opacity levels are typically in the
6-10% range.
The firing rate on the current fuel oil at 185 MW is 97,200 Ibs/hr, superheat
and re-heat steam temperatures are 1000°F and total steam flow 1,275,000
Ibs/hr.
Continuous emissions monitoring (GEM) is used, as per Phase 1 of the Clean
Air Act Amendments, for measurement of NOx, CO2 and opacity levels, CO,
and (X level measurement is also available.
NOx Reduction Technology
The original and CCOFA modified windbox arrangements are shown in Figure
1. Prior to installation of CCOFA a NOx level survey was carried out over
the boiler load range 60 - 185 MW with results as shown in Table 1.
Table 1
Baseline NOx Emission Survey
Boiler Load
MW
60
90
150
185
NQx Range
ppm (at 3°/d02) Ibs/106 Btu
245 - 265 0.33 - 0.36
210 -260 0.28 - 0.35
230 - 260 0.31 - 0.35
215-245 0.29 - 0.33
Mean NQx
lbs/106 Btu
0.345
0.315
0.330
0.310
Following the provision of CCOFA, on Unit 3, the NOx level, at 185 MW was
measured at 0.22 lbs/106 Btu, which represented an approximate 30% reduction
in NOx compared to the pre-CCOFA modification NOx level.
The arrangement of the windbox and burners in the Port Jefferson No. 4 Unit,
comprising essentially of four plenum chambers, each housing a central burner
with upper and lower air nozzles with individual auxiliary air nozzles between
the plenums, offered a ready adaptation to the TAS system, developed by 1C
Ltd in order to achieve further NOx reductions in T fired boilers. The TAS
system requires axial displacement of the upper and lower air nozzles in the
plenum chamber away from the centrally located oil burner axis. Figure 2
showing NOx reductions achieved by applying TAS to a T fired simulation in
the 1C Ltd combustion test facility demonstrates the potential of this
technology.
-------
Reduced burner and combustion zone stoichiometry is also an essential feature
of low NOx technology, but can be a potential cause of boiler tube
deterioration arising from local reducing atmospheres. To minimise this effect
the auxiliary air nozzles were modified to effect their air flow in the horizontal
plane in order to create an air rich atmosphere adjacent to th boiler tube walls.
This offset air augments the vertical air staging effect of the TAS system and is
therefore a further aid to Nox reduction.
Future requirements on Unit 4 also included natural gas firing up to 100%
MCR load and therefore a gas burner was combined with the central oil burner
housed in each plenum chamber.
This total system was then demonstrated on the 1C Ltd Combustion test facility.
Test Rig Demonstration of the TAS Low NOx System
The 1C Ltd test rig is a full-scale single burner horizontally fired facility. As
such it cannot fully represent the central fireball of the T firing system but can
usefully represent ease of ignition, flame stability and effects of near burner
conditions on emissions utilising a burner cell representative of the T fired
situation, as illustrated in Figure 3. This shows the centrally located fuel gun
with its upper and lower air nozzles, all set horizontally in their plenum
chamber, offset air nozzles are located at the top and bottom of this
arrangement as per the boiler situation. Some leakage air, remote from the
cell, was admitted to the test rig from the windbox in order to represent the
effects of CCOFA A basic steam atomised 'F1 jet oil burner was used in the
test work, Figure 4, and atomiser angles, (A in Figure 4) in the 60° - 90° range
were studied, together with a fuel staged design created by clustering of the
atomiser discharge orifices.
As each pair of TAS nozzles and the burner were fed by combustion air from a
common plenum chamber a variety of oil burner air swirlers were used to
optimise the air flow through the oil burner and to be compatible with the oil
spray characteristics. The gas burner arrangement comprised central spuds
close to the swirler hub to provide good pilot flames and flame stability. The
majority of the gas was supplied through spuds located behind the swirler.
Figure 5 shows the emission results from the optimised burner build firing oil
on the test rig. These showed NOx in the 0.12 - 0.14 lbs/106 Btu range with
CO under control down to 1.5% excess O2. Under similar conditions the NOx
levels when gas firing were in the 0.03 - 0.05 lbs/106 Btu range.
The required turn down ranges of 3.1 were readily demonstrated with this
burner system.
During the development work the effect of oil temperature on NOx levels was
tested with a range of oil temperatures from 230°F - 290°F (110°C - 143°C).
With an early burner build a NOx reduction was indicated as the oil
temperature increased. However, this NOx reduction did not materialise with
the optimised burner build with compatible atomiser, air swirler and fuel
compartment air flow.
-------
The general conclusion from the test work was that the optimised burner
system would achieve the required NOx level of 0.15 lbs/106 Btu on the Port
Jefferson boiler and be compatible with the other operating requirements of less
than 150 ppm CO and maximum particulate levels into the electrostatic
precipitator of 0.2 lbs/106 Btu.
Site Installation
Oil Firing
Figures 3 and 6 show diagrammatic fuel compartment and detailed windbox
arrangements of the site installation respectively. The fuel compartment upper
and lower air nozzles were existing and therefore retained in the low Nox
windbox arrangement, hi a TAS system it is usual to build in a certain degree
of tilt in the fuel compartment air nozzles minimising the differential tilt which
has to be applied by me modified burner box tilting mechanism. In order to
achieve the required tilt without interfering with adjacent components the
burner nozzle geometry was altered sightly but this was shown not to affect the
NOx performance of the system as demonstrated in the combustion test facility.
Based on the rig work the prediction was that site NOx levels would be in the
0.13 - 0.15 lbs/106 Btu range when oil firing, from a baseline NOx level of
0.22 lbs/106 Btu, representing an average 36% reduction in NOx. The site NOx
levels achieved on oil firing are shown in Table 2 and are compared with the
unmodified (baseline) NOx levels over the boiler load range 40 185 MW.
These NOx levels were achieved with CO levels below 10 ppm up to 130 MW
load and 10-15 ppm CO at 100% MCR. Superheat and re-heat temperatures
at full load were reported as 1003°F and 9900F respectively. On this bases the
site NOx levels achieved represent a 1 : 1 site to test facility NOx level factor.
Opacity levels were in the 5 8% range and particulate emissions during a
controlled 100% MCR operating period were measured at 0.13 lbs/106 Btu with
a NOx level recorded at 0.133 lbs/106 Btu, with CO at 36 ppm. All these
figures were within the guarantee and site operating requirements.
-------
Table 2
Site NOx Emissions - Oil Firing
Boiler Load
MW NQx( lbs/106 Btu)
40
60
70
90
100
130
150
180
Baseline
0.35
0.32
0.33
0.31
CCOFA Mod TAS &
CCOFA
0.15
-
0.14
0.12
0.13
-
0.22 0.14
Gas Firing
The test rig results on gas firing are shown in Figure 7 which indicates NOx
levels of 0.04 ± 0.01 lbs/106 Btu, depending upon excess air, OFA and TAS
arrangements. The high CO levels recorded in the tests would not be expected
to be repeated on site and result from the absence of the 'fireball effect' in the
test facility.
Figure 8 shows the gas burner arrangement in the fuel compartment of the
burner test cell, this inner/outer spud configuration was found to give the more
stable gas flame condition during the test rig burner development work.
Site gas firing has not yet been fully optimised but is reported to give NOx
levels in the 0.09 - 0.11 lbs/106 Btu ranges over boiler levels 66-185 MW.
On this basis the NOx figures represent an approximate 2 : 1 relationship
between site and test rig facility NOx levels, compared with an approximate
1 : 1 achieved with oil firing. Ffigher site to rig ratios always result with
increased thermal NOx generation within the boiler system.
Table 3 shows typical NOx results when gas firing.
-------
Tables
Site NOx Emission Gas Firing
Boiler Load MW NQx Ibs/10 ** CO pm 02%
66
91
185
0.09
0.09
0.11
3
4
183
4.9
2.6
0.4
No guarantees were attached to the design for gas firing at Port Jefferson, but
the target should be 0.1 lbs/106 Btu with a maximum 150 ppm CO. This may
be achieved by some fuel biassing within the burn box. Under all conditions
the gas burners are reported to operate satisfactorily with good flame stability.
Further Options for NOx Reductions at Port Jefferson
The application of multiple technologies to achieve NOx reductions is subject
to the law of diminishing returns. However, the results so far indicate that
there is scope for further reductions in NOx at Port Jefferson.
Further optimisation of the CCOFA air staging would probably achieve a
further 5 - 10% reduction of NOx, based on the levels already reported. As
recycled flue gas (RFG) is ducted to the boiler hopper bottom for steam
temperature control this could be usefully redirected with the burner system
either bulk-mixed with the combustion air or specifically targeted around the
fuel jets, a further 10 - 20% NOx reduction could result from this technique.
Depending upon the site fuel economies a permanent co-firing natural gas/fuel
oil requirement could be established and NOx levels would be pro-rata to the
natural gas utilised over the 0.09 - 0.14 x lbs/106 Btu optimises NOx range for
the individual fuels.
Further fuel and air staging within the boiler could probably be used with effect
in achieving significant NOx reductions. These would include biassed firing,
particularly with natural gas. Other techniques which could be considered are
displacing some of the combustion air from the windbqx as separated over fire
air (SOFA). This would allow low stoichiometry conditions (conducive in
NOx reduction) to persist for a longer period within the boiler. The CCOFA
SOFA balance would be used for CO and carbon in paniculate control. An
overall NOx reduction, on the optimised TAS system, of 10 - 20% could be
expected from this technique which would require the SOFA system to handle
20 - 25% of the total combustion air supply to the boiler. A combination of
RFG with SOFA has a potential for 20 - 30% NOx reduction.
A re-bum system in which both fuel and air are displaced from the main burner
box could also be considered. Around 20% fuel (ideally natural gas) would be
deployed in the re-bum system augmenting the low stoichiometry effect
-------
immediately after the main burner system in the boiler. SOFA would be
deployed higher in the boiler to compact combustion. On the optimised low
NOx TAS system re-burn would probably effect a further 30 - 35% reduction
in NOx.
The potential NOx reduction from these techniques is summarised in Table 4.
SOFA and re-bum systems would be more costly as further boiler penetrations
are required to accommodate the displaced air and fuel nozzles. The system fan
capacities and windbox pressures would require consideration before recycled
flue gas is deployed in order to effect NOx reductions.
Based on the above assumptions the indication is that combustion modification
technologies are available which may enable 0.1 lbs/106 Btu NOx to be
approached when oil firing above without recourse to chemical injection or
catalytic techniques. If 0.1 lbs.106 Btu is the aim irrespective of fuel then
co-firing or 100% gas firing depending on fuel economies, is the obvious route.
Table 4
Sirrmary of Further Potential NOx Reduction Options
Option
Potential NOx Level Qhs/lO6 Btu)
Remarks
Co-firing Oil
and Natural
Gas
NOx range 0.14 - 0.09
Pro-rata with natural gas utilisation
Option of co-firing
in same burner or
via separate burners.
Biassed firing with
100% natural gas.
RFG or SOFA
Oil
0.11 -0.125
Gas
0.07 - 0.075
RFG ducted into
windbox air supply,
system pressures
permitting
RFG and
SOFA
0.10 0.11
0.065 - 0.07
Re-burn
0.01 - 0.10
0.055 - 0.065
Natural gas as re-
bum fuel
Conclusions
The (TAS) tilted air supply system, as demonstrated on a full-scale test facility,
indicated a p9tential for 50% reduction in NOx from this in windbox air
staging technique.
-------
Applied to a 185 MW oil and gas T-fire boiler a 36% NOx reduction was
obtained from a system based on air staging incorporating close-coupled over
fire air (CCOFA) when oil firing.
NOx levels, when oil firine, were between 0.13-0.15 lbs/106 Btu over the
boiler load range from 40 - 185 MW (21 100% MCR) from a full load
baseline NOx level of 0.22 lbs/106 Btu. CO levels were below 40 ppm
superheat and reheat temperatures at full load were 1003°F and 990 F
respectively.
Opacity levels were recorded between 5 - 8% with particulate levels of 0.13
lbs/10^ Btu at 100% MCR burner load.
The low NOx system met guaranteed performance levels in all respects.
An optional natural gas firing system has indicated potential NOx levels of 0.09
± 0.01 lbs/106 Btu subject to further confirmation and burner optimisation.
Co-firing of oil and natural gas offers potential for 100% MCR boiler operation
with NOx emissions in the 0.08 - 0.14 lbs/106 Btu range depending on the level
of natural gas utilisation.
NOx reduction techniques based on further air staging (SOFA), targeted
recycled flue gas or re-bum utilising natural gas oner the potential for
operations at or below the 0.1 lbs/106 Btu Nox level.
-------
ADDITION OF CLQSg COUPLED
QVERFIRE &IP
Bncttd Coimnmem
Auxiliary Air
(Typical)
Auxdjafy Air
Snetefl Camnanmem
= - Clan-CauBUa
=! I OV»TO» Air
3 r WHft Honzonni
=1
Fuel Air
^ Auxiliary Air
=1
|h-C-l| Fuel Air
Wnaaaiwan CCOFA
Rotnfit
AuoUary Air
Fuel Air
Auxiliary Air
Brii'ihmj Conoannwnl
Figure 1.
Original ana CCOFA Modined Wmabox Arrangements.
Key ftlomi-ser Load 'MWi %02
• R-Jet 21 4 l 52
• F-Jel :3 lobes) 210 1.37
* F-jetiSlandard) 210 1 67
200
160
120
i 80
I
540
-2
.120
100
60
O
u
40
4 6
TAS(deg-)
10
12
4 5
TAS (deg.)
10
12
Figure 2.
NOx Reaucnon Potential ot TAS Tecnnology
-------
Figure 3.
Test Cell for Oil Firing
Figure 4.
F-jet Atomiser.
-------
Odeg.TAS 10deg.TAS 20deq.TAS
Figure 5.
ULCO Oil Fifine Trials N'Ox and CO Emissions.
i . . . ^i
ra
'S' I- Now noziM to «±
i r
I
Modified Windbox Arrangement \
• New T jet atomisera. oil
burner Gp3 and swiriars.
Figure 6.
Site Low ixiOx \Vmdbox Arraneement.
-------
0 deg. TAS 0 deg. TAS 10 deg. TAS
NoOFA 100% OFA 100% OFA
100
0.0
1.0
2.0 3.0
Oxygen in Flue Gas (%)
4.0
250
200
! | 150
: 0100
50 h
1 0
2.0 3.0 4.0
Oxygen in Flue Gas (%)
5.0
Figure 7.
LILCO Gas Finn? Trials NOx and CO Emissions.
f;
tic
t?^^
r^4
H-
;r
p^ __
^- —
. C^lj fc^^s /
rK*»«>«4ii-r : — : '
'^r "mgg-i- L^ • - i— =
^ 1
1 ^-MOTZJJ
^.^VI-\^>
- ^( ' l^CAT.
1
Figure 8.
Locanon ot Gas Burner rn the Test Cell.
-------
PRELIMINARY RESULTS OF LOW NOX (< 25 ppm)
BURNER RETROFIT OF PACIFIC GAS & ELECTRIC
345-MW CONTRA COSTA UNIT 7
H. T. Allen
Pacific Gas and Electric Company
245 Market Street, Room 1252
San Francisco, CA 94105
V. V. Bland
R. H. Sirois
Stone & Webster Engineering Corporation
7677 East Berry Avenue
Englewood, CO 80111-2137
E. S. Schindler
D. Smolens
TODD Combustion, Inc.
15 Progress Drive, P.O. Box 884
Shelton, CT 06484-0884
Abstract
This paper details the short-term operating results of the low NOX retrofit of Pacific Gas &
Electric Co.'s 345-MW Contra Costa Unit 7, which (at time of writing) entails the lowest NOX
emission rate ever guaranteed (24 ppm) for a low NOX utility boiler retrofit without post-
combustion NOX controls.
In April, 1997, TODD Combustion performed the turnkey supply and installation of twenty-four
(24) low NOX, gas/oil burners utilizing advanced fuel gas injection techniques and an advanced
overfire air port system. Physical three-dimensional modeling was used to optimize the boiler
combustion air and flue gas recirculation systems. Stone & Webster Engineering served as
architect/engineer for the project, which included increasing bulk mixed flue gas recirculation
from 17% to 30%.
The project approach, emissions and boiler performance results are presented in depth.
-------
Unit Data
Contra Costa Power Plant Unit 7 is rated at 345 MWg. The boiler is a Babcock & Wilcox El
Paso style radiant, single reheat, pressurized unit, rated at a main steam flow of 2,160,000 Ib/lir at
2,475 psig and 1,050°F, with reheat steam temperature of 1,000°F.
The unit was originally designed to fire either natural gas or heavy fuel oil through 24
combination circular burners arranged in two rows of six burners on each of the front and rear
walls, in an opposed configuration. Horizontal burner spacing is uneven to allow for two partial
furnace division walls. A single row of overfire air ports is located above each of the six burner
columns on each of the front and rear walls.
Flue gas recirculation is available for control of reheat steam temperatures and NOX emission
level.
The unit, used in daily load following service, was constructed in 1964.
Project Approach
The primary objective of the Contra Costa Unit 7 NOX Retrofit Project was to engineer, procure,
construct, start up and test a NOX control system that will allow the unit to operate within PG&E's
Compliance Plan to meet the requirements of the Bay Area Air Quality Management District's NOX
Regulation 9, Rule 11.
This rule allows systemwide NOX emissions averaging of the affected PG&E Bay Area units. As a
result, affected units do not have a specific NOX limitation but average their emissions with those
from other affected units to meet the systemwide, clock-hour NOX limit. The Compliance Plan that
PG&E has selected requires that Contra Costa Unit 7 be capable of operating at emissions levels in
the 20 to 25 ppmvdc range by 1997. This emission range appeared to be attainable by retrofitting
low NC\ burners and by upgrading the windbox FGR capability to 30% (on a flue gas mass basis).
CO emissions from Contra Costa Unit 7 are also affected by the BAAQMD rule. The rule requires
operation below 400 ppmvdc during steady-state compliance source tests and below 1,000 ppmvdc
during normal operation (i.e. CEMS monitoring).
The NON control system retrofit consists of 24 new low NOX Dynaswirl-LN burners with natural
gas biasing valves combined with the increase in the windbox flue gas recirculation (WFGR)
capability to 30%. The unit's 12 existing overfire air (OFA) ports were also replaced with an
advanced design.
Related project work included:
• Structural strengthening of the existing furnace, back pass, and duct work to accommodate an
increase in FD fan static pressure
-------
• Installation of a new FOR duct run on each side of the unit (complete with platforms, stairs,
ladders, dampers, insulation and lagging) that operate in parallel to the existing ducts that
supply recirculated flue gas to the FOR mixing plenum
• Upgrading the FD and FOR fans and motor drives
• Modifications to the existing DCS and associated burner front hardware for a fully automatic
burner management system (BMS)
• Recertification of the unit's CEMS
• Miscellaneous electrical work
• Asbestos abatement
Flue Gas Recirculation
To satisfy the demands of the low NOX burner retrofit, the FOR system had to be sized to provide
30% FOR to the windbox while providing approximately 2% FOR flow to the unit's furnace
hoppers at the following conditions:
• Maximum continuous boiler rating (MCR)
• All burners in service (ABIS)
• Installation of TODD Combustion Dynaswirl-LN burners with a 3.3 inwc windbox-to-fumace
pressure drop with ABIS
Prior to modifications, the existing FOR system was only capable of supplying approximately 17%
FOR to the windbox. To maximize NOX reduction with the new burners, the capabilities of the
existing FOR fans had to be upgraded to provide higher pressure and flow performance.
At 30% WFGR, modifications to the existing ductwork providing flue gas flow to the FGR
injection plenums were required to obtain acceptable velocities and pressure losses. These
modifications included adding a new duct to each side of the unit (Figure 1). Each new duct has an
approximate cross-sectional flow area of twice that of each of the existing ducts. A damper was
installed in each of the new ducts to control flow in parallel with the flow control dampers in the
existing system.
The pre-retrofit total FGR flow elements located in the inlet duct to each FGR fan were also
removed to reduce system pressure drop. This equipment was used to indicate total FGR flow
supplied to the unit's. The new flow elements that were installed employ a multi-port averaging
pilot tube approach that has a lower unrecoverable flow resistance than the baseline equipment.
In addition, windbox FGR flow may now be determined separately from total flow with the
installation of pressure taps to measure the differential across the FGR mixing airfoils located in the
combustion air ducts. The pressure differential measured by this system is a direct indication of
FGR mass flow to each side of the unit's windbox.
Each of these flow measurement systems was calibrated during tests conducted for low NOX tuning
of the boiler.
-------
PG&E
Contra Costa Unit 7
Combustion Air and Flue Gas
Recirculation Systems
Hopper FGR
C.D.
n
3
C.D.
t
Windbox r-font
Fumace and Hopper
i Windbox Rear =
^3 : C23
m"5
l^Jt
front W.B.
Hooper FGR
C.D.
^m
^iU
WswFGR i 1 1
Oua -!l-j 1 1
I.
C.O. c~=
Air
1 ' !H ' c
, lil
CJ5.
New FGR
•• Due
Aj C.D.
ol
r^js.o. s.o.csi
fi -P
kJ~T3"
Hralnr T
Sleam JH
Air Heater 1)
Seal Air. T
T.V. CoaJig |
35S.O.
n
F.D.Fac— ,|^j
Eoan.
OuUet
fT. Steam
U Air Heater
i -Seal Air
T.V. Cooling
CSS.O,
1
^-«»
- To
S.O. CJ). FGRFan
S.O. = SnufofT Damper
C J) .= Control Damper
Figure 1
-------
Boiler Performance Modeling
Performance predictions were prepared for Contra Costa Unit 7 with varying quantities of FOR
using the Stone & Webster Steam Generator Performance Evaluation Program (SGPEP). The
information used to baseline this mathematical model of the boiler was based principally on the
data and information recorded during pre-retrofit tests conducted on the unit and on boiler design
and performance characteristics existing in Stone & Webster's experience base.
The modeling results show that Unit 7 boiler performance will be affected by large changes in FOR
rates. The following summarizes the important parameters and impacts identified by the SGPEP:
Furnace Exit Gas Temperature
The furnace exit gas temperature (FEGT) at MCR is predicted to drop from 2,550°F with
approximately 10% FGR to 2.475°F with 30% FOR. The FEGT combined with the mass flow of
flue gas leaving the furnace is an indicator of the total energy absorbed by the furnace. For a given
temperature, if furnace flue gas mass flow rate increases, the energy level leaving the furnace also
increases. This results in higher steam temperatures and spray flow rates due to increased heat
transfer in the convection pass regions of the boiler. With the Unit 7 boiler type, it is common for
the FGR rate to significantly affect convective pass performance for rates from zero to
approximately 10%. This effect is considered to be due to heat transfer characteristics that result in
less furnace heat absorption and a significant increase in FEGT. The combined effects of higher
gas temperature and mass flow on convective pass surfaces is responsible for the design impact. At
higher FGR rates, the trend in furnace heat absorption is lessened and the FEGT drops due to the
bulk mixing of combustion products with an increasing quantity of flue gas. The net effect on Unit
7 of increasing flue gas recirculation rates will be increased convective pass tube metal
temperatures even though the FEGT will actually decrease.
Boiler Efficiency
The Unit 7 boiler thermal efficiency is expected to drop linearly between 85.44% with 10% FGR to
85.21% with 30% FGR. This slight reduction in boiler efficiency results from the reduced heat
transfer in the back passes of the boiler. Convective heat transfer (the predominant heat transfer
mechanism in the boiler convective passes) is nonlinear, varying as the 0.6 power of gas mass flow.
Therefore, as FGR rate increases, the heat transfer rate does not increase proportionally, resulting in
higher flue gas temperature leaving the air heater and a higher dry flue gas heat loss value.
Spray Attemperation
The Unit 7 spray water attemperation flow rates in both sections of the superheater and at the inlet
of the reheater will increase with increases in FGR rates. For the FGR range expected following
the modifications, the primary superheater outlet spray flow rate is expected to increase by
approximately 70%. The secondary superheater spray flow requirement will increase by
approximately 40%, and the reheater spray flow requirement will increase by approximately 65%.
-------
An evaluation of the spray system was performed to assess flow/pressure limits and to select
required upgrades to piping, valving and nozzles.
Boiler Pressure Part Metals Analysis
As described above, an increase in FGR rate is expected to affect overall boiler performance. The
temperatures of pressure part components are also affected by changes in steam flow through
tubing due to changes in spray flow requirements and heat transfer rates. Based on the results of
the boiler model which provide overall thermal performance, Stone & Webster calculated critical
temperatures for each superheater and reheater section.
Existing tubing material sections in the primary superheater, secondary superheater, tertiary
superheater and in the reheater were considered marginal. With increased FGR, the primary
superheater tube metal temperatures are expected to increase substantially due to reduced steam
flow in this section as more water is withdrawn to provide spray to the secondary superheater. The
outlet header and link piping material will also be affected at a the 30% FGR flow rate. There are
no substantial impacts to the secondary superheater, tertiary superheater or to the reheater due to the
anticipated change in FGR flow rates.
PG&E has implemented a convective sections tracking and analysis program to monitor the long
term potential effects on these boiler components.
Dynaswirl-LN Burner
The Dynaswirl-LN is a parallel-flow, venturi-register gas/oil burner using fuel and air staging to
achieve low NOX, high efficiency operation with exceptional flame stability. More than 300
Dynaswirl-LNs have been installed in U.S. wall-fired, gas/oil, low NOX utility boiler retrofits
totaling over 6,000 MW.
Air Staging
Primary and secondary airflow enter the venturi register, which tends to equalize distribution and
provides an axial, turbulence-free flow. Primary air, approximately 15% of total flow, exits the
register through a small swirler, which creates a sub-stoichiometric primary combustion zone
with a fixed ignition point that does not vary with load. The low pressure zone formed by the
swirler also recirculates furnace gases within the flame pattern, reducing NOX formation.
Secondary air, approximately 65% of total flow, exits the venturi register around the diffuser as a
parallel flow, mixing further downstream. The remaining 20% of burner airflow, termed tertiary
air, exits through a separate sleeve around the venturi throat, mixing still further downstream for
maximum NOX control.
Primary/secondary airflow is controlled via an "on/off pneumatic airslide, which does not
modulate with load. Tertiary airflow is controlled via a manual airslide, which is fixed in
position during startup.
-------
Fuel Staging
Gaseous fuel is injected with a combination of multiple spuds, or pokers, and a centerfire gun.
The centerfire injector provides a small amount of fuel at the root of the flame. Six pokers
around the periphery of the swirler inject fuel at precise locations within secondary and tertiary
airflow, creating rich and lean zones within the flame. Poker orientation and drilling are keys to
NOX control.
An advanced poker arrangement developed in single-burner testing has been retrofitted to more
than 100 Dynaswirl-LNs in the field, producing additional NOX reductions of 20%. This
proprietary technology was incorporated in the burner design for Contra Costa.
For No. 6 oil firing, the existing mechanically atomized oil guns were reused in the Dynaswirl-
LN burners. The oil guns were lengthened at TODD's manufacturing facility and retrofitted with
advanced design TODD atomizer assemblies.
Figure 2
-------
Overfire Air
The six OFA ports on each of the front and rear walls were replaced with TODD venturi ports.
Air and FOR flow through the ports is controlled via automatic "on/off airslides that do not
modulate with load.
OFA flow is increased from a pre-retrofit maximum of approximately 14% to 27% post-retrofit.
OFA port openings were not increased, and pressure drop remained the same. The new OFA
system increased the momentum of the OFA ports by approximately 1.9 times, and mixing in the
upper furnace was improved.
Airflow Modeling
For all TODD low NOX retrofit projects, a physical three-dimensional model of the combustion
air system is first constructed. When flowed with water containing neutral buoyancy beads
illuminated by a collimated light source, basic flow patterns can be easily visualized. The model
is tested with both with water and air in order to verify results. Corrections can then be made via
baffles and turning vanes. The optimized baffle arrangement can then be installed in the actual
unit during construction, assuring proper airflow in the field.
Physical airflow modeling has three goals: equal airflow distribution among all burners, equal
inlet velocities around the periphery of each burner, and elimination of swirling airflow through
each burner.
First, equal airflow distribution to all burners is key in any low NOX retrofit. When an imbalance
exists, excess O2 must be raised to prevent formation of CO by the burner most starved for air.
This unnecessarily increases NOS formation from the other burners. By equalizing airflow
distribution among all burners, excess O2 can be minimized, which provides maximum NOX
reduction and boiler efficiency. Including FOR distribution in the modeling process further
maximizes NOX control.
Equal inlet velocities and elimination of swirl through each burner are crucial to burner
performance. Because low NOX burners rely on injection of fuel at precise locations within
burner airflow, it is imperative that the proper airflow be present at these locations. Likewise, for
optimum performance, the only swirl present must be that which is created by the burner itself.
No burner can be expected to simultaneously correct imbalances in the draft system and
precisely control fuel/air mixing to minimize NOX formation. Airflow modeling prepares the
airflow for the burner, allowing the burner to precisely control fuel/air mixing for maximum NOX
reduction. This approach has freed burner designers to focus solely on NOX control, increasing
the effectiveness of known control techniques to their maximum extent.
-------
FGR MODEL
In order to insure maximum benefit from the FGR system, a model of the FGR system and
windbox was constructed. The intent of the model was to optimize the windbox FGR
distribution. Equal FGR distribution to each burner has been shown to reduce NOX without
increasing total FGR rate. The relationship between NOX and FGR is a non-linear curve, with the
amount of NOX reduction decreasing as the FGR rate increases. Therefore, while the burner
receiving the least FGR produces more NOX than a burner receiving the average FGR rate, its
increased NOX emissions are NOT offset by the burner receiving the most FGR. The net effect is
an overall increase in NOX emissions. A unit's NOX emissions may therefore be reduced by
taking FGR from the burners receiving higher than average FGR rates and redistributing it to the
burners receiving below average FGR rates. The goal of the FGR model is therefore to distribute
FGR equally to all burners without increasing the FGR system pressure drop. The practical limit
of the model is considered to be distribution within + 2%.
The FGR system model was built at the same 10:1 scale as the burner/windbox model and
included all ductwork from the FGR fan, up to and including the pre-retrofit FGR mixing
section. The FGR model was attached to the burner/windbox model. Methane was used to
simulate flue gas and was introduced into the inlet of the FGR duct. The concentration of
methane at the exit of each burner was then measured to determine the distribution.
Modifications to the mixing slots were made, or baffles were added, with each iteration until the
goal of the model was achieved. The optimum arrangement was then installed during the unit
outage.
OFA Model
A separate physical three-dimensional model was constructed for the OFA system. Methane was
injected in the OFA flow, but not in the burner flow. The mixture was then measured at the
boiler exit, corrected for density and momentum differences, in order to verify proper
burner/OFA mixing. This resulted in recommended OFA port settings for startup.
Installation
The unit outage was scheduled for eight weeks, working one shift per day with no scheduled
overtime. Preoutage activities started weeks before the outage began. The major mobilization
started 2 weeks before the scheduled outage. The unit was shut down and boiler work started on
March 3. The work force consisted of union labor from the Building Trades at a reduced scale as
part of PG&E's NOX Reduction Project Agreement.
The work progressed mainly on schedule, and the outage mechanical work was completed on
April 28. The average work force was 35 people for the burner installation and 70 people for the
balance of plant work.
Checkout of the new BMS control system continued, and the boiler was fired on May 5 for test
testing and refractory curing.
-------
Startup Data
The data presented in this paper is from a limited load startup. Because the Contra Costa power
plant is located on the Sacramento River, where the Striped Bass spawn each spring, the plant
condenser outlet water temperature is limited to 86°F until the bass reach 38 mm, per the
California Department of Fish and Game. Plant load is thus limited to maintain condenser water
temperature at 86°F.
After the outage, the unit started as scheduled on May 15, 1997. Maximum load, which varies
from day to day and with the tide, was often limited to 120 MW, but at times 230 MW could be
reached. During this period, the project team was asked to provide a preliminary tune, develop
preliminary operating curves, and observe the operation of the equipment. The data presented in
this paper is from this time period. Load limitation was not released until July 5, 1997.
Airflow Modeling Validation
Bumer-to-burner airflow distribution in the uncorrected model showed an as-found deviation of
49% between the two worst-case burners (+15% / -34%). In the corrected model, deviation was
reduced to + 1%.
In post-retrofit boiler testing, deviation on the unit was less than 10%, achieving the goal of +
5%. Post-retrofit performance — flame stability, emissions, O, — remains the best proof of
proper airflow within the unit.
AIRFLOW RESULTS AT CONTRA COSTA #7
PERCENT DEVIATION IN AIRFLOW
20 -
15
10
5
0 '*
-5
-10
-15
-20
-25
-30
-35
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24
MODEL'BEFORE'" MODEL AFTER • UNTfAFfER~l=r~ BURNER NUMBER
Figure 3
-------
Turndown
Turndown capability of the burner appears excellent. During startup testing, with no
adjustments, turndown of 30-to-l was demonstrated.
Flame Stability
In addition to quantitative tests to verify airflow distribution, certain field tests are conducted
with the boiler in operation to ensure flame safety and to demonstrate the correct application of
flow control devices indicated by the model. These tests are the minimum gas pressure test and
the maximum airflow test.
The minimum gas pressure test is the minimum burner gas pressure at which a flame remains
stable. It is performed on every burner, and the primary goal is consistency of gas pressure from
burner to burner. At Contra Costa, every burner lost its flame in the range of 0.25 to 0.33 in we
gas pressure with airflow at 25%. This demonstrates consistent air and fuel flow to every burner.
The maximum airflow test ensures that the flame will not blow out even under conditions of
maximum airflow. This provides operations personnel with confidence that under certain worst-
case conditions, the flame will not blow out. The gas pressure is raised to 8 in we. and every
burner is taken to 100% airflow to verify that it does not blow out. This minimum gas pressure
can then be set with confidence that the flames will not blow out under any condition of airflow.
The consistent results again demonstrate that equal airflow is being provided to each burner. All
burners remained stable up to the amp limit of the FD fan motor.
Figure 4
Dvnaswirl-LN Natural Gas Flames at Part Load
-------
FGR Measurement
Accurate measurement of FGR is important because of its potentially large impact on boiler
performance and NOX reduction. The impact on boiler performance stems from changes in heat
transfer that occur in the boiler. Furnace radiation decreases with higher FGR rates because the
peak flame temperature is reduced; the extra gas flow increases heat transfer in the superheater,
reheater and HRA. If the superheat and reheat sprays cannot reduce temperatures sufficiently to
meet maximum metal temperature requirements, then FGR must be limited or the boiler surface
must be modified. Also, if FGR flow cannot be accurately quantified, then the results of the
retrofit are not as useful for the next project. Extra margins must be added, conservative
projections must be made, and both add cost with no value. For maximum NOX reduction, it may
be necessary to add as much FGR as the boiler can tolerate. Accurate measurement of FGR flow
is required for control of the boiler in optimum, low NC\ operation.
FGR Measurement Techniques
A venturi meter; multiple point pitot tube; measurement of windbox/burner O2, CO2 or NOX
concentrations; or simply a calculation from FGR fan amps, fan pressure drop and inlet damper
position can all be utilized to measure FGR. Each of these have problems, however. A venturi
meter takes space, is expensive, has a relatively high recoverable pressure drop, and the upstream
and downstream duct requirements are unacceptable in a unit such as Contra Costa. Orifice
meters have similar problems with higher unrecovered pressure drops.
A multiple point pitot tube has very low unrecovered pressure drop, but in an FGR duct where
very low velocities are desired to minimize draft system losses, care must be taken in the design
of the instrumentation system to avoid signal fluctuation due to internal duct turbulence.
The use of gas species measurements in the burners or windbox combined with combustion
calculations appears to be the most accurate approach. Some types of O2 meters are not capable
of measuring O2 in the 14-19 % range, which is the approximate range needed to measure five to
40% FGR rates. High accuracy O2 instruments may be used if carefully calibrated for this range.
Measurement of NO has been used in the past to calculate windbox FGR rates because NO
instruments can be easily calibrated over the range of expected values in the windbox and
economizer. A problem occurs, however, at very low NO values, when the calculation of FGR
becomes very sensitive to a small change in windbox NO. For example, at 24 ppm of NO in the
economizer flue gas and 7 ppm of NO in the windbox. if the windbox NO increases to 8 ppm,
then the change in FGR rate is 4%. Therefore, FGR can only be measured to +4% using this
method. Some typical preliminary measurements of FGR by dilution for CO,, O2 and NO are as
follows:
SPECIES
CO2
02
NO
TEST#1 FGR%
23
21
22
TEST #2 FGR%
28
26
13
TEST #3 FGR%
22
20
16
-------
By agreement, CO2 has been chosen to document the FOR rate for this project. The FOR rate
will be calculated using all three dilution species, an "S"-type pitot traverse of the FOR duct, the
measurement of a commercial type pitot array, and a calculation using fan amps, fan pressure
rise, a fan curve and damper position. We are committed to accurately determining the FOR rate
and using the information to tune the boiler controls and evaluate the impact on NOX and boiler
performance.
Emissions Data
The following data was recorded during May's preliminary testing:
TEST*
19-1
1 9-6
1 9-7
1 9-8
19-9
20-1
20-2
20-3
20-4
20-5
20-6
20-7
21-1
21-1A
' 21-2
' 21-3
— 21-4
21-b
21 -b
22-1
*
!i
— 4
3
LOAD
65
96
230
230
231
230
230
50
49
50
50
50
49
50
12b
m
12b
12b
12b
12b
12b
128
125
125
129
125
BLRTiT2
2.5
2 2
2.2
b 7
2
4.4
2 2
1 2
1 1
1 3
2.0
2 0
/.1
/.6
/ 2
7 4
/.a
I'.e
7.5
1 2
1 '/
2 2
2 5
2 6
2 5
2
3 6
1 8
1 5
32
a
NOx, FPM
y2
90
53
83
80
72
66
58
42
3/
32
27
38
62
b5
28
26
24
22
32
10
14
16
19
1B
13
Af
69
75
25
•"'CO, PPM
56
9
11
1
11
185
1
^
4
'
3
{
1
1
'
1
4
58
7
3
3
'
203
~ — RTTT!5
10
200 KPPH
0
0
u
21 b
23 1
28
33 8
32 5
56 3
64
65.8
31 .6
56
WB02
1955
NA
18.2
18 0
1/4
16 8
1 7.1
16.9
16,5
16 40
184
OFA%
100 -
0
0
0
0
0
0
0
0
0
0
0
10U
0
0
0
0
0
100
100
100
-------
CONTRA COSTA #7
PRELIMINARY NOx DURING STARTUP
NOx, PPM
150
125
100
75
50
25
0
0 50 100 150 200 250
LOAD, MWg
PRE-RETROFIT POST RETROFIT POST RETROFIT
WITH FOR &OFA WITH FOR & OFA NO FOR OR OFA
300 350
Figure 5
-------
CO, p p m
1,200
1,000
800
600
400
200
0
C O NTRA COSTA #7
NOxVS O2 AND CO
PRELIMINARY RESULTS
t.--^*
N O x, p p m
200
1 50
1 00
50
234
BOILER EXCESS O2, %
NOX345MW NOX170MW C034SMW C0170MW
BASE BASE BASE BASE
N Ox 125
D LN
CO 125
DIN
N Ox 230 M W
DLN
CO 230 M W
DLN
Figure 6
Full Load Projections
The preliminary testing at up to 60% load has shown that the NOX guarantee of 24 ppm is
achievable at foil load. Testing at full load is scheduled for mid-summer, and the results may be
presented verbally in conjunction with this paper.
NON is normally defined as a combination of NO and NO,. The majority of the pollutant NO, is
". The fraction of NO, present varies with a number of operating parameters.
These relationships are being addressed, as part of this project, with an investigation into the
effects of load, burners in service, excess O2 and CO emissions on NO, levels.
-------
ULTRA-LOW NOX RAPID MIX BURNER DEMONSTRATION
AT CON EDISON'S 59th STREET STATION
Steven J. Bortz and Dale E. Shore
Radian International, LLC
Nigel Garrad
TODD Combustion
Jack Pirkey
Consolidated Edison Co. of New York, Inc.
Tony Facchiano
Electric Power Research Institute
Presented at the:
EPRI-DOE-EPA Combined Utility Air Pollution Control Symposium
August 25-29, 1997
Abstract
Radian International, together with its licensee TODD Combustion, has developed, demonstrated,
and commercially implemented the Radian Rapid Mix Burner (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 configurations
and sizes of industrial boilers and similar gas-fired devices. Over the last three years, the RMB has
been deployed and is operating on approximately 25 industrial boilers and furnaces.
Recently with the support of the Electric Power Research Institute, the Consolidated Edison
Company of New York, ESEERCO, NYSERDA, and the San Diego Gas and Electric Company,
the burner was installed and demonstrated in a boiler in an electric utility power plant setting. A
two-burner package boiler with air preheat at Con Edison's 59th Street Station was equipped with
RMBs and a flue gas recirculation system. The objective of the project was to demonstrate the
achievement of sub-15 ppm NOX levels, using staged firing and no more than 20% FGR.
The objectives of the project at 59th Street have been met. This paper gives an overview of the
results.
-------
Introduction
Radian International has developed the Rapid Mix Burner (RMB) which provides ultra-low NOX
levels for gas-fired, forced-draft boilers and other fired devices. The RMB also provides a stable,
compact flame with high combustion efficiency to minimize impacts on CO and other
combustibles.
Over the last three years, these performance criteria have been met on commercial RMB
installations in about 25 industrial boilers and dryers, with burner sizes ranging from 1.5 to 280
MBtu/hr. Performance testing and regulatory compliance testing of these installations have proven
the burner's capability to produce NOX levels under 10 ppm (corrected to 3% O2, dry),
simultaneously with similarly low CO emissions. Some of these installations have used FOR rates
over 30%, some have been in refractory-lined furnaces such that the flame operates at adiabatic
temperatures, some have involved air preheat, some have utilized staged firing, and some have
been multiple burner installations. In the majority of installations, the realized NOX level was 10
ppm or less.
The RMB Design and NOX Control Process
The RMB design is based upon three main criteria. First is the rapid and near-complete mixing of
combustion air, recirculated flue gas, and fuel prior to the ignition point. This mixture is overall
air rich and, since the mixing is near complete, this eliminates prompt NOX formation which is a
by-product of substoichiometric combustion. Second is the burner geometry that produces a very
stable flame. And third is the use of FGR to dramatically reduce peak flame temperatures, thus
bringing the production of thermal NOX to very low levels. In addition, since the fuel and air are
essentially premixed prior to the ignition point, high excess air is as effective as FGR for reducing
thermal NOX. This is a key element in using the burner in certain applications, including utility
boilers.
A schematic illustrating the burner's important design concepts is shown in Figure 1. The burner
consists of two concentric flow paths, or registers. In the inner register, the combustion air or
air/FGR mixture enters through a 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 fixed-vane swirler
design was developed initially at the International Flame Research Foundation and refined by
Radian. 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 outer register contains straight vanes (no swirl in the outer section). These vanes also contain
a built-in grid of gas injectors, as do the inner register swirl vanes. The flame from the inner
burner serves to establish and anchor the flame from the outer register. Both the inner swirled and
the outer axial flow portions of the burner operate at the same stoichiometry and with the same
FGR rate.
-------
The geometry of the burner, with the divergent quarl and swirling flow of the inner register,
combine to generate a very stable flame that can tolerate high flue gas recirculation rates. With
very high FOR, the burner can operate stably at flame temperatures low enough to generate only a
few ppm of thermal NOX.
Another important design element of the RMB is the method of gas fuel injection and the mixing
with combustion air and FOR. As opposed to a diffusion-mix or staged mixing burner, the fbel-air
mixing in the RMB occurs extremely quickly (hence, its name) such that near-perfect mixing is
achieved prior to the flame front. A prime benefit of rapid mixing is that stoichiometry can be
precisely controlled and held constant throughout the flame. This avoids zones of relatively fuel-
rich stoichiometries (common to conventional burners, and especially fuel- and air-staged low-
NOX burners) which result in prompt NOX formation. (Prompt NOX forms very rapidly in a flame
and cannot be avoided in fuel-rich flames.) Thus, the burner is capable of breaking the 20 ppm
NOX barrier which results primarily from prompt NOX formation.
Although the RMB was designed to reach ultra-low NOX levels even when used in applications
with high air preheat temperature, higher FGR rates are required when operated in the normal
firing mode. For example with 600 °F air preheat, the burner achieves NOX levels of 10 to 15 ppm
using normal firing and FGR rates of 40 to 45%. Normal firing is defined as all the fuel and 10 to
15% excess air through the burner; see Figure 2. The burner is designed intentionally to operate
with very high FGR rates and, as such, firing under these conditions can be done without
degrading flame stability. In fact, the burner has been tested at FGR rates as high as 60% with a
stable flame.
FGR rates greater than 20% can be tolerated by many industrial boilers, or by a utility boiler that
was designed for the high mass flows through the convective pass. However from the standpoint
of retrofitting existing utility boilers, the most FGR that can typically be tolerated at high loads is
around 20%. Thus, normal RMB firing with very nigh FGR is not always practical for a utility
boiler where ultra-low NO* levels are needed and FGR must be limited to 20% or less. This
limitation led to the development of the staged firing mode of the RMB.
RMB staged firing is based on the fundamental principle that high excess air through the burner
has the same effect upon flame temperature (and thus upon NOX formation) as does FGR. This is
a normal characteristic of pre-mixed or simulated pre-mix flames. (This characteristic is opposite
to that of diffusion-mix burners where higher excess air increases NOX formation.)
The ability of the RMB to operate at very low NOX emissions using high excess air only has been
well documented in several industrial RMB installations. A 60 MBtu/hr RMB has been in
continuous operation since mid-1995 in a rotary dryer application, producing NOX emissions of 7
ppm (corrected to 3% O2) with 60% excess air and no FGR. It is important to note that, in this
dryer application, the RMB is firing into a refractory lined, adiabatic combustion chamber — with
a NOX level of 7 ppm.
Staged firing was developed to take advantage of the RMB's high excess air NOX characteristic.
With staged firing, the burners operate at high excess air levels (50% or so), yet the overall
furnace stoichiometry is similar to normal operation (10 to 20% excess air). This is accomplished
-------
by re-routing a portion of the fuel away from the burners, but keeping the same total air flow
through the burners. The rerouted fuel is injected just outside the burner envelope and mixes and
bums with the high excess air left over from the burners; see Figure 3.
The operating principal of the RMB staged firing configuration is to burn as much of the fuel as
possible under the high excess air, ultra-low NOX conditions. As a result of the fundamental
principle mentioned above, this is done by operating the burners air rich (under conditions similar
to normal firing, but with higher excess air and with a moderate amount of FOR), where the
maximum theoretical NOX formation is very predictable. Heat is extracted from these air-rich
combustion products and cool furnace recirculation products are entrained. The staged fuel is
then mixed with the air-rich burner products and combustion is completed. The excess air at the
boiler outlet is thus similar to that with normal firing.
In a staged firing application on a utility boiler with air preheat, about 75% of the total fuel input
would be fired equally distributed among the burners. The burners would also have all the air
flow. This would result in the burners operating at about 50% excess air. With 20% FOR added,
the burner NOX level would less than 10 ppm. The burner-NOx contribution to the overall stack
NOX concentration would be 7.5 ppm (75% x 10 = 7.5). The remaining 25% of the fuel would be
fired through the staged fuel injectors (near the burners) with no air. In mixing and burning with
the air-rich burner products, additional NOX would be formed from the staged fuel. For the total
stack NOX concentration to be no more than 15 ppm, the NOX contribution from the staged fuel
combustion would have to be no more than 7.5 ppm. Thus, if the NOX formation from the staged
fuel can be controlled to less than about 30 ppm (7.5/25%), overall NOX emissions would be less
than 15 ppm.
An important item in a staged firing application is the optimum location and orientation of the
staged fuel injectors. The ability to achieve stack NOX emissions of 10 to 15 ppm requires control
of the NOX generated by the staged fuel to 30 ppm or less. Ultra-low NOX emissions from a staged
firing configuration can best be achieved by injecting the fuel into a region of the furnace where
cooler furnace recirculation products can first be entrained into the fuel jet. The most direct
method of controlling the NOX emissions from the staged fuel is to first entrain a large amount of
relatively cool furnace gases into the fuel gas stream before mixing with the high-excess-air,
higher-temperature products from the burners.
Although a given staged fuel injector location may be good for NOX control, the configuration
also must be optimized to ensure adequate CO. Also, any alteration of the heat release distribution
in the furnace must not be detrimental to the boiler's steamside temperature requirements. In
addition to injector location, injection angles and jet penetrations can also be used toward
obtaining the optimum conditions.
Project Goals and Work Scope at 59th Street
Several of Radian and TODD's commercial RMB installations are on boilers with air preheat; and
normal firing and high FGR rates are used to reach ultra-low NO* levels. Other installations
involve staged firing - but without air preheat or FGR - to achieve moderate NOX levels (under
30 ppm). To demonstrate the RMB's usefulness for reaching ultra-low NOX levels on utility
-------
boilers, it was necessary to combine staged firing and FGR on a boiler using air preheat. This need
led to the project at Con Edison's 59th Street Station.
Boiler 118 at the 59th Street Station is a Foster Wheeler D-type package boiler with a rated heat
input of 176 MBtu/hr and a nominal capacity of 150,000 Ib/hr of saturated steam at 550 psig and
450°F. The boiler uses two burners mounted on the front refractory wall. The boiler is equipped
with a regenerative air heater that heats the combustion air to 460 to 500 °F. The boiler has a
forced draft and an induced draft fan for balanced draft operation of the furnace. The pertinent
specifications of the boiler are shown in Table 1.
Table 1
Con Edison 59th Street Boiler 118
Boiler Manufacturer
Boiler Type
Design Steam Flow
Steam Pressure
Steam Temperature
Furnace Dimensions:
Depth
Width
Height
Fuel Gas:
Type
Higher Heating Value
Pressure Available
RMB Gas Burner Design & Operation:
Heat Input
Turndown
Gas Pressure at Burner
Draft Loss
Furnace Operating Pressure (including
FGR at MCR)
Combustion Air Temperature
Flue Gas Temperature at Boiler Outlet
Foster Wheeler
D
150,000 Ib/hr
550 psig
450°F
31ft-2in
7ft-3in
11 ft - 0 in
Natural gas
l.OOOBtu/scf
40 psig
88 MBtu/hr per burner
5 to 1
7 psig
7 inches H2O with 20% FGR, at
rated heat input
0 in wg
500°F
700°F
Two RMBs rated at 88 MBtu/hr each were manufactured by TODD Combustion. A general
layout drawing of the burner is shown in Figure 4. The burners were built to the same design
standards as used for all commercial RMBs.
The burners were installed in the boiler's existing windbox. Prior to the installation, the windbox
was flow modeled by TODD for the purpose of identifying any air flow distribution problems that
could impact burner performance. As a result of this modeling, the windbox was modified with
baffles to assure equal air flow distribution between the two burners, and around the
-------
circumference of each burner. In addition to the windbox baffling, a set of dampers was installed
in the air duct upstream of the windbox to allow the biasing of air between the two burners. This
was provided solely for experimental purposes, and is normally not within the scope (not
required) of commercial projects.
To complement the burners and to provide for staged firing, staged fuel injectors were also
installed on the boiler. Two injectors were installed near each burner, as illustrated in Figure 5.
The injectors consisted of removable pipes that could be inserted through stuffing boxes built into
the burners. These pipes were connected to the burner main gas supply, and were controlled using
manual valving. Several injector nozzles of variable injection angle and hole size were fabricated
to support the optimization of the staged fuel mixing with the main burner flow field.
A new flue gas recirculation system was designed and installed on the boiler, specifically for the
demonstration project. A schematic of the system is shown in Figure 6. The system was
configured with a hot gas fan for drawing flue gas from the boiler generating bank outlet at a
temperature of approximately 700 °F. The FGR fan delivered the flue gas into spargers installed
within the fresh air duct connecting the air heater outlet and the windbox. The FGR supply duct
was split into two legs and individual manual dampers were provided to allow the biasing of FGR
between the two burners, if desired.
The overall goal of the project was to demonstrate the ability of the RMB system to reduce NOX
emissions for natural gas operation to ultra-low levels on a boiler having operating conditions
representative of an electric utility boiler. The specific goals of the project were:
• To demonstrate the ability of the RMB to provide NOX levels under 15 ppm (at 3% O2, dry),
with 450 to 500°F air preheat and no more than 20% FGR to the windbox. The burners
would be operated in a staged firing mode to meet this emissions goal.
• To limit CO emissions to less than 50 ppm, simultaneously with the target NOX levels.
• To determine the lowest NOX levels achievable using normal firing and a nominal amount of
FGR.
The demonstration testing was performed in July, 1997.
Emissions Measurements
NOX and CO emissions, along with excess O2 levels, were measured for the project using an
extractive system and continuous emissions monitoring equipment. The monitoring system was
installed in a fully enclosed, self-contained test trailer. The analyzers consisted of a Thermo-
Electron Chemiluminescent Model 10S NOX analyzer, a Siemens Ultramat non-dispersive infrared
CO analyzer, and a Thermox Model FCA zirconium oxide, microprocessor based ©2 analyzer.
Samples were drawn into the test trailer through polyethylene tubing connected to sample probes
installed in the boiler windbox and boiler outlet flue gas duct. (Windbox O2 concentrations were
measured to determine FGR rates.) The sample gases were flowed through a sample conditioning
-------
system to remove moisture. The test trailer was equipped with certified gases used for calibration
of all instrumentation. The instruments were calibrated periodically throughout each day of
testing.
Results
The burners were put in service with ignitor fire to cure the refractory on July 3, 1997. The main
flames where first established on July 8. On this same day, the fuel, air and FOR inputs were fully
characterized for normal firing between 20% and 80% load (30 klb/hr and 120 klb/hr steam flow).
Load was limited to 80% of the boiler's rating, in order to have sufficient FD and FOR fan
capacity to permit firing at high excess air levels and FOR rates during the testing.
On July 9, the characteristic curves were loaded into the fuel, air and FOR controllers and the
system was setup for automatic operation. The boiler was then ramped up and down in automatic
between 20 and 80% load. During the entire phase of initial characterization and setup work,
there were no instances of high CO emissions, combustion-induced vibration, or any other
problems related to flame stability.
Three additional days were spent testing the boiler, and the demonstration testing was completed
on July 14, 1997.
As part of the initial characterization testing, the manual dampers on the two FOR ducts were
adjusted to balance the FOR flow to both burners. Except at low loads, the FOR was balanced to
within 5% FGR between the upper and lower burner. No adjustment of the air balancing damper
located between the air heater outlet and the windbox was required to balance the air flow
between the two burners. The air flow was within 5% between the two burners.
Testing of normal firing (non-staged) was performed on July 10 and testing with staged firing was
done on July 11 and 14, 1997. The NOX emissions measured for normal and staged firing are
shown in Figure 7, plotted as a function of FGR rate. With normal firing, approximately 28%
FGR was required to reduce NOX emissions to 15 ppm; and 35% FGR was needed to reduce NOX
to 10 ppm. Data are shown for operating at 75, 100, and 120 klb/hr steam flow - for all loads,
the effect of FGR on the NOX emissions was approximately the same. This performance
characteristic is typical of RMB operation in other boilers and demonstrates the insensitivity of the
burner's NOX control capability to firing rate or heat release rate.
Again referring to Figure 7, the use of staged firing resulted in similar NOX levels being achieved,
but with much less FGR. For example, 15 ppm NO* was achieved with about 17% FGR and
staged firing (versus 28% FGR with normal firing). Also of significance was that with staged
firing, the NOX level was reduced to under 40 ppm, without any FGR.
During all characterization activities and testing of both normal and staged firing, the maximum
CO emissions were 5 ppm and typically less than 1 ppm.
-------
The amount of staged fuel used was dependent on the FGR rate and varied from about 22%
staged/total fuel with 20% FGR to 35% staged/total fuel without FGR. During all the staged
firing tests, the main burners were adjusted to a NOX level between 8 and 9 ppm.
NOX emissions as a function of boiler load and FGR are plotted in Figure 8. With normal firing
(i.e., no staging), the NOX level approaches 20 ppm at 80% load with approximately 20% FGR.
By using staged firing, however, it is possible to control NOX emissions to under 15 ppm at all
loads. With an FGR fan sized for full load operation, and with additional tuning of the combustion
system, NOX levels below 15 ppm up to full load would be expected.
Figure 9 compares the NOX levels measured on Boiler 118 at 59th Street with data from the
original 4 MBtu/hr RMB pilot scale development test work done at Radian International. The
pilot scale testing was done on a single-burner firetube boiler at about 4 MBtu/hr. The data from
the two boilers are comparable since the testing for both was done with air preheat levels between
450°F and 500°F. Data are shown for both normal and staged firing. As the figure illustrates, the
results from both boilers are very similar. This comparison is significant since it demonstrates the
insensitivity of the RMB's emissions characteristics to burner size, heat release rate, and burner to
burner interaction.
Conclusions
The objectives of the Con Edison 59th Street Rapid Mix Burner demonstration have been met.
The testing has shown the effectiveness of the Rapid Mix Burner, when operated in the staged
firing mode, for controlling NOX emissions to ultra-low levels, even with air preheat and under
operating conditions acceptable for utility boilers.
• NOX emissions under 15 ppm were achieved using 20% FGR and staged firing.
• Staged firing was shown to be an effective process for reaching, not only ultra-low NOX
levels, but also for reaching moderate NOX levels. NOX emissions were reduced to under 40
ppm without any FGR. This has important implications relative to significantly reducing NOX
emissions on units that either do not have FGR or where the retrofit cost of FGR would be
significant.
• With normal firing (no staged firing), NOX emissions were reduced to less than 15 ppm with
just under 30% FGR.
• For all ultra-low NOX operating conditions, the accompanying CO emission levels were
typically less than 1 ppm, and a maximum of 5 ppm.
References
1. R. Christman, S. Bortz, and D. Shore, "The Radian Rapid Mix Burner for Ultra-Low
NOX Emissions" presented at the EPRI/EPA 1995 Joint Symposium on Stationary
Combustion NOX Control, Kansas City, Missouri (May 1995).
-------
Combustion Fuel
Air & Flue Gas Gas
External
Recirculation
Zone_
Combustion
Air & Flue Gas
!t)
Rapid
Mixing Section
Fuel Gas —t— :.--——-
Oil
Swirl Vanes
and Gas
Injectors
Internal
Recirculation
Zone
Figure 1
Radian Rapid-Mix Burner, Operating Principle
-------
Main Burner
115% Theoretical Air
100% Fuel
40 - 45% FGR
Figure 2
Normal RMB Firing
25% Fuel
115% Theoretical Air
75% Fuel
20% FGR
Figure 3
Staged RMB Firing
10
-------
A» tUOL ASSEMBLY
(MM N I^JU CUtt, Ml)
RDDACTCRY THRQAr
Figure 4
General Layout of RMB for 59th Street Boiler 118
-------
Staged
Fuel
Injectors
(4)
Furnace
Walls -
(approx.)
Burner
Throat
Figure 5
Boiler 118, Location of Staged Fuel Injectors on Front Wall
Air
-vKJ
FD Fan
(Existing)
Flue Gas-
D-^tzz
05
0
£
CL
ID Fan
(Existing)
S S
i-IXI—
FGR Fan
(New)
Boiler
P
P
Figure 6
ConEdison - 59th Street Boiler 118 FOR System - Functional Diagram
12
-------
75 Klb/hr, Normal Firing
100 Klb/hr, Normal Firing
120 Klb/hr Normal Firing
75 Klb/hr, Staged Firing
100 Klb/hr Staged Firing
120 Klb/hr, Staged Firing j
Figure 7
Boiler 118 Effect of FOR and Firing Mode on NOx Emissions with RMB
-------
FOR, Staged Firing
NOx, Normal Firing
Boiler 118 NOx Emissions Versus Load with Normal and Staged Firing
-------
2 MBtu/hr, Normal Firing
-•-90 MBtu/hr, Normal Firing |
A 4 MBtu/hr, Staged Firing
-X-120 MBtu/hr, Staged Firing
% FGR
Figure 9
Comparison of Full and Pilot Scale Results
-------
Tuesday, August 26; 8:00 a.m.
Parallel Session A:
Low-NOx Systems for Coal-Fired Boilers (Wall and Tangential)
Continued
-------
THE INTEGRATION OF LOW NOX CONTROL TECHNOLOGIES
AT THE SOUTHERN ENERGY, INC. BIRCHWOOD POWER FACILITY
J.A. Lauber
Southern Energy, Inc.
M.B. Cohen
R.E. Donais
ABB Combustion Engineering Systems
Combustion Engineering, Inc.
Abstract
The Southern Energy, Inc. (SEI) Birchwood Power Facility represents the first application
worldwide of the TFS 2000™ firing system and selective catalytic reduction (SCR). The
installation of these state-of-the-art NOX control technologies was necessary to meet strict
Commonwealth of Virginia environmental regulations requiring a 0.10 lbs/106 Btu (0.043 g/MJ)
NOX emission rate based upon a 30-day rolling average. The plant successfully completed all
performance and emission testing on September 24, 1996. Commercial operation began
November 14, 1996. Stack NOX emission rates are consistently maintained below
0.10 Ibs/ 106 Btu (0.043 g/MJ).
The paper describes the integration of both in-fumace and post-combustion NOX control
technologies into the overall boiler design. Operational data depicting boiler outlet NOX, stack
NOX and loss on ignition (LOI) are presented across the design load range from 32% to 100%
boiler output. The description, arrangement, design parameters and operation of the NOX control
equipment are discussed. Novel design features include a split economizer, an air heater suitable
for ammonia applications, Dynamic™ classifiers, and a multi-zone secondary air flow control
system utilized for the TFS 2000™ firing system.
Introduction
The Birchwood Power Facility is a 240 MWe coal-fired plant located on State Route 665,
approximately twelve miles east of Fredericksburg, in King George County, Virginia. The
cogeneration unit sells electricity to a local utility and process steam to a large greenhouse
nearby.
The boiler equipment consists of a Controlled Circulation® steam generator provided by
Combustion Engineering, Inc. (ABB CE). The steam generator is an indoor, balanced draft unit
designed for a maximum continuous rating (MCR) of 1,571,567 Ibs/hr (198 kg/s) at a
superheater outlet pressure and temperature of 2465 psig (357.4 kPa) and 1005°F (540.6°C) - see
Figure 1 for plant design conditions. The boiler and auxiliaries are designed for daily cycling
from 32% to 100% load and sliding pressure operation. Full steam temperatures are maintained
-------
down to 50% load. The fuel is a low sulfur, bituminous coal from West Virginia - see Figures 2
and 3. The unit has the additional capability of firing no. 2 fuel oil from startup to 15% of MCR
heat input.
The SEI Birchwood plant, situated 60 miles east of Shenandoah National Park (Class I pristine
area), was designed and constructed to meet some of the most stringent air emission regulations.
As a result, this unit is one of the cleanest coal-fired facilities in the world.
The unit is located within an area that is either in attainment or unclassified for all criteria
pollutants. Since this source has the potential to emit greater than 100 tons per year
(90.7 tonnes/yr) of at least one criteria pollutant, the unit is considered a major source, and
subject to the Prevention of Significant Deterioration (PSD) regulations. The PSD permit ensures
that a new emission source will not cause an area to move into nonattainment status for any
criteria pollutant. One of the primary PSD requirements deals with maintaining emission
increments within acceptable levels for the district. This is defined as approximately 25% of the
difference between the emissions for the area in a baseline year and the emissions that would
produce nonattainment status. The other PSD requirement specifies that Best Available Control
Technology (BACT) be installed for all emission equipment.
To comply with the Commonwealth of Virginia's emission requirements, BACT was chosen for
reducing stack emissions of nitrogen oxides, sulfur oxides and particulates. The primary method
used to control each of these pollutants is as follows:
Nitrogen Oxides TFS 2000™ Firing System in combination with Selective Catalytic
Reduction (SCR)
Sulfur Oxides Spray Dryer Absorber (SDA) utilizing a slurry reagent of calcium
hydroxide, Ca(OH)2
Particulates Fabric Filter (FF) with reverse air cleaning
To ensure that the Birchwood facility would not adversely impact the surrounding area,
mathematical models were used to evaluate the atmospheric dispersion of the emissions from the
plant. The results of these models confirmed that the air quality impact on the neighboring area
was minimal. Subsequently, a permit was issued by The Commonwealth of Virginia Department
of Environmental Quality in the fall of 1992.
The emission limitations specified in the permit for NOX, SO2 and particulates are as follows:
1. Stack nitrogen oxide (NOX) emissions shall not exceed 0.10 lbs/105 Btu (0.043 g/MJ),
220 Ibs/hr (0.028 kg/s) on a 30-day rolling average, and 963.6 tons/yr (874 tonnes/yr).
2. Stack sulfur dioxide emission levels shall not exceed 0.10 lbs/106 Btu (0.043 g/MJ),
220 Ibs/hr (0.028 kg/s) on a 30-day rolling average, and 963.6 tons/yr (874 tonnes/yr). In
addition, sulfur dioxide (SO2) emissions shall be controlled by a lime spray drying system
with a minimum 90% control efficiency in combination with the firing of low sulfur coal.
3. a. Paniculate emissions from the boiler shall be controlled by a fabric filter system rated at
99.9 percent control efficiency.
-------
b. Total Suspended Paniculate (TSP) measured at the stack shall not exceed 0.02 lbs/106 Btu
(0.009 g/MJ), 44.0 Ibs/hr (0.006 kg/s), and 192.7 tons/yr (174.8 tonnes/yr).
c. Inhalable paniculate matter (PM10 = paniculate matter less than 10 microns) measured at
the stack shall not exceed 0.018 lbs/106 Btu (0.008 g/MJ), 39.6 Ibs/hr (0.005 kg/s), and
173.5 tons/yr (157.4 tonnes/yr).
d. Visible emissions from the boiler stack shall not exceed 10 percent opacity, except during
one six minute period in any one hour, where visible emissions shall not exceed 20 percent
opacity.
4. Emission limits for carbon monoxide (CO) and volatile organic compounds (VOCs) shall
not exceed 0.20 lbs/106 Btu (0.086 g/MJ) and 0.010 lbs/106 Btu (0.004 g/MJ), respectively.
5. Various emission limits for inorganic pollutants such as lead, mercury, arsenic, etc. were
also specified in the plant's operating permit.
Description of Plant NOX Equipment Features
Compliance with stringent NOX emission requirements imposed by the site operating permit had
a major impact upon the design of the overall steam generator. An integrated NOX reduction
system comprised of a low NOX TFS 2000™ tangential firing system and selective catalytic
reduction (SCR) formed the basis of the BACT technologies - see Figure 4. In addition, high
performance (HP) pulverizers with Dynamic™ classifiers, a split economizer section and an air
heater suitable for ammonia applications were installed - see Figure 5.
TFS 2000™ Firing System
The TFS 2000™ firing system was the direct result of several years of laboratory development at
Combustion Engineering1. This is an advanced product that evolved from the Company's
standard tangential firing technology. The Birchwood firing system was designed concurrently
with a successful TFS 2000™ R field retrofit demonstration2.
While the firing systems for both plants are similar, and had a common goal to minimize in-
furnace NOX emissions as low as possible, the similarities stop there. The retrofit unit was
challenged by a fast-track schedule, existing pressure part obstructions, economics and a desire to
minimize impacts upon plant control operation. By comparison, the TFS 2000™ firing system
for SEI Birchwood was designed for a new unit application with few restrictions. In addition, the
contractual stack NOX emission guarantee for the Birchwood site was 0.10 lbs/106 Btu
(0.043 g/MJ), while the retrofit demonstration did not include a specific NOX emission guarantee
other than a best effort commitment.
The specific TFS 2000™ system design features that ensure high combustion efficiency and low
NOX generation are:
• Pulverized Coal Fineness Control
• Early Coal Devolatilization Control
• Concentric Firing - Offset Air Firing
• Firing Zone Stoichiometry Control including:
Furnace Bulk Air Staging
Activated Zoned Stoichiometry Controls
-------
Pulverized Coal Fineness Control
The four 863 HP pulverizers at Birchwood are equipped with variable speed Dynamic™
classifiers. Finer coal minimizes combustible losses normally attributable to staged, low NOx
processes. Finer coal can also result in closer coal ignition and lower NOx emissions by
enhancing the fuel bound nitrogen yield, and promoting the reduction to elemental nitrogen.
Additional benefits include fewer large coal particles striking the waterwalls and improved low
load firing stability. During MCR plant acceptance tests, the Birchwood mills pulverized coal to
a fineness of more than 80% passing through a 200 mesh sieve, with 0% on a 50 mesh sieve, and
less than 1.5% on a 100 mesh sieve.
Early Coal Devolatilization Control
The early ignition coal nozzle tip design is another important feature of the TFS 2000 firing
system - see Figure 6. The shear bar and air deflector style tip initiates combustion early and
produces a stable, devolatilizing zone at a point close to the coal nozzle tip. The ignition point,
local stoichiometry, and resultant "prompt NOX." can be better controlled with the windbox fuel
air dampers behind the coal compartments.
Concentric Firing - Offset Air Firing
The concentric firing system (CFS) provides local, horizontally offset secondary air staging in
the TFS 2000 firing system. CFS is unique to tangential firing via the CFS nozzle tips - see
Figure 7. This feature directs a portion of the secondary air away from the main coal stream
toward the furnace waterwalls - see Figure 8. This provides more favorable low NOX conditions
in the pre-fireball region flame envelope by locally delaying secondary air entrainment into the
rapidly expanding fuel/primary air streams. Horizontal air staging is also important when firing
coals with slagging and corrosion characteristics. The oxidizing environment created near the
waterwalls, both within and above the firing zone, helps to minimize these potentials.
The auxiliary compartments at SEI Birchwood contain two CFS nozzles surrounding a center,
non-offset, straight auxiliary air nozzle - see Figure 9. These fully partitioned, three section
auxiliary compartments are sized to permit biasing between the amount of offset air and straight
air flowing around each coal elevation. The control system has an operator adjustable bias feature
impacting the ratio of CFS to straight air flow throughout the boiler load range. With the CFS
bias feature, both NOX and unburned carbon loss performance can be adjusted.
Firing Zone Stoichiometry Control
Furnace Bulk Air Staging
Global or bulk air staging simply means injecting a portion of the combustion air above the main
firing zone. This method isolates air from the devolatilization initial char combustion zone. The
fuel rich chemistry promotes the formation of molecular nitrogen rather than NOX species. In
addition, significantly staged combustion reduces peak flame temperatures, also resulting in
lower thermal NOX.
Two global air staging techniques. Close Coupled Overfire Air (CCOFA), and Separated
Overfire Air (SOFA), were developed through extensive pilot scale testing at ABB CE laboratory
and field demonstrations. This effort has shown that while air flows directed through the SOFA
-------
compartments more effectively minimize NOX, a combination of both CCOFA and SOFA
vide a superior system with respect to the total combustion process.
„ * "irchwood TFS 2000™ system incorporates both global air staging techniques. Two
A compartments are situated at the top of each corner windbox. Above the CCOFA are
eparate, multiple compartment SOFA windboxes. The overfire air flow improves carbon
Burnout and assists with global NOX emission control. The multi-level, low and high SOFA
(,L-bOFA and H-SOFA) provide flexible staging capability throughout the boiler operating
pe - see Figure 9. These SOFA nozzles are each equipped with a manually adjusted
zontal yaw mechanism providing ±15° freedom, and are positioned as necessary to impact
upper furnace mixing.
Active Zoned Stoichiometry Controls
A^ adecluate at high boiler loads, the traditional secondary air damper control system
(SADCS) does not actively attempt to control NOX at low loads. The typical SADCS logic does
not account for increases in excess air during boiler load reductions. Consequently, NOX
emissions trend upward as the boiler load decreases. NOX emission spikes also occur during load
transients and subside as the unit air flow matches fuel flow.
Today s stringent local environmental standards require continuous emission monitoring
throughout the entire boiler operating envelope requiring a more sophisticated approach to
secondary air control. Specifically, the location and amount of OF A flow, regardless of a
particular operating condition, must become an essential part of the total system package. To
address this issue, the windbox damper control for the TFS 2000™ firing system is based upon
an active zoned Stoichiometry control philosophy. This feature, schematically represented in
Figure 10, continuously monitors and strives to maintain the combustion zone Stoichiometry
regardless of the unit excess air. The SADCS logic automatically opens the H-SOFA
compartment dampers, then the L-SOFA dampers, followed by the CCOFA dampers to maintain
programmed zones of Stoichiometry throughout the boiler operating envelope. As a result, the
zoned Stoichiometry control system approach minimizes NOX emissions at any load and
mitigates NOX emission spikes commonly observed during boiler load ramps.
At the Birchwood facility, the air flow through the SOFA elevations is closely monitored and
controlled with an in-situ, multi-point array of total and static pressure sensors in each duct.
Eight SOFA air flow meters are installed in the system. The control system logic sums the SOFA
air flows on an elevation basis, yielding the total L-SOFA and H-SOFA flows. Using the
L-SOFA, H-SOFA flows, coupled with the primary and secondary air flows and the total fuel
flow and boiler outlet O2, the lower furnace Stoichiometry can be determined, monitored,
maintained, and controlled. The system allows independent control of multiple zones by utilizing
ramped functional stoichiometries. f,(x) and f2(x) throughout the boiler operating envelope.
These functions are independent of the specific unit load and outlet excess air levels.
Furthermore, the control system can anticipate transient load induced NOX emission spikes by
monitoring the air and fuel flow demands, while proactively modulating SOFA flows prior to
measuring a change in boiler outlet O2. The feedforward logic has an overload feature which
upon sensing full H-SOFA capacity, progressively spills overfire air into the L-SOFA, and, if
necessary, into the CCOFA, and finally to the main windbox. These features minimize NOX
emissions during load swings by anticipating high excess air levels, avoiding lags or delays
-------
normally associated with extractive sampling systems, and diverting air to the proper OFA
compartments. To date, the plant has operated with SOFA air flows on manual control due to
flow transmitter problems.
The TFS 2000™ control system logic also includes a main firing zone (MFZ) stoichiometry
override protection feature. This ensures that the main windbox auxiliary air dampers maintain
control of the load specified windbox to furnace delta P regardless of the selected MFZ
stoichiometry.
Arrangement - Main Windbox
The Birchwood unit is equipped with four main windboxes, one at each comer. These main
windboxes include four coal elevations, each serviced by a dedicated pulverizer. The coal
compartments each contain an early ignition, shear bar/air deflector style coal nozzle tip, and
admit secondary or fuel air when firing pulverized coal at these elevations - see Figure 6. Dust-
tight, pneumatically operated slide gates located in each coal pipe near the windbox allow
isolation of the piping from the windbox and furnace for maintenance purposes. The fuel
compartment secondary air dampers can be adjusted to regulate air flow to match the local
volatile matter requirements, and/or shaping and positioning the ignition point.
Beneath the bottom coal elevation are two elevations of auxiliary air compartments. These two
air compartments assist with lower furnace carbon burnout while providing operational
flexibility throughout the entire boiler load range.
Sandwiched between these four coal elevations are three oversized auxiliary air elevations. The
auxiliary air arrangements consist of two concentric firing system (CFS) fixed-offset, auxiliary
air elevations surrounding either a center straight air/oil or straight air compartment. These three
compartments are oversized to effectively allow biasing of air flow between the center straight
air compartments and the outboard CFS style compartments.
The corner windbox assembly arrangement is schematically shown in Figure 9. All windbox
nozzle tips can be tilted ±30° from horizontal.
Oil Firing Equipment
Oil firing equipment consists of two retractable oil gun elevations equipped with a high energy
arc (HEA) ignition system. The oil guns are designed for remote operation and are used to
lightoff the unit, while gradually raising temperature and pressure at a controlled rate. The oil
guns are a source of ignition energy for adjacent coal elevations, and provide stabilization of
associated coal compartments at low loads. The oil equipment can maintain up to 15% MCR
load. The high energy arc (HEA) ignitor located in each oil firing compartment ignites the oil by
directing a high energy electrical discharge to the atomized oil spray/air stream.
Flame Scanner System
Scanners are provided for flame supervision and supply "flame" or "no flame" logic signals to
the FSSS* furnace supervisory safeguard system. Infra-red type scanners are located in each oil
gun compartment for discnminating flame indication when lighting off each oil gun. Additional
flame scanners are located in each auxiliary air compartment and the top end air compartment for
furnace fireball indication when firing pulverized coal.
-------
w. an9etflent - Separated Overfire Air (SOFA) Box Assemblies
re§isters t> § 200°™ firing system, two separate, multiple compartment SOFA windbox
of three sen* C°n& are strategically located above the main windboxes. Each SOFA box consists
fine-tuning theN°FA nozzle elevati°ns. These individual elevations provide the capability of
T>,A Qnu A 6 °x contro1 stoichiometric history throughout the operating load range.
me oUr A nozzle t'
adjustable ± 15° tlle standard ± 30° tilt capability, as well as manual horizontally
for controlling cya^ caPabiuty. The SOFA yaw is intended for fine-tuning upper furnace mixing
hydrocarbons O t' UStilDle emissions such as unburned carbon, carbon monoxide and
11111 tdt function and yaw settings were set during unit commissioning.
Selective Catalytic R^ -
ytlc Reduction System
No other NOX control meth H
catalytic reduction Th' C3n matc^ *e high reduction efficiency achieved by selective
selective non-cataltic^010 Wa$ °h°Sen °Ver °ther Post-combustion methods such as
-ceuc
Japan and Germany Althn.Jt?! ( NCR) based uPon ^ successful SCR operating experience in
U.S. coals, the Commonwealth nfv ^ S°™ initi&1 C°ncem °V£r ^ aPPlicability of SCR ^
' VlrSlnia Department of Air Pollution Control selected SCR as
BACT for the SEI pro'ect
The SCR process uses
j- ,- • i_ _, i
o
4NO(g) + 4NH3(g) + 02(g) — . 4N2
0 nol'u n m apProximate temperature window ranging from 572°F to
752 F (3 J ™ , Q: ^^ 752°F (4°0°C)' Catal^ic Slntenng becomes promment in
lowering the NO reduction performance. Below 572°F (300°C), the susceptibility of ammonium
bisulfate (NH4HS04) formation on catalyst surfaces increases. The lower temperature limitation
is dependent upon the amount of SO3 in the flue gas. For the Birchwood project, an SO3
concentration of 1 1 ppmvd at 3% O2 dry volume is predicted based upon a maximum sulfur
content of 1 .20% by weight in the fuel. At this level, a minimum operating temperature entering
the SCR of 580°F (304.4°C) is recommended. The ammonium bisulfate reaction can be
described as follows:
S03 (g) + NH3 (g) + H20 (g) - * NH4HS04 (1)
SCR systems can be effectively integrated into the overall steam generator equipment. The SCR
unit at Birchwood consists of a high-dust, hot-side reactor located between a split economizer
section upstream of a single air heater - see Figure 1 1 . This location provides optimum flue gas
temperatures ranging from 580°F to 752°F (304.4°C to 400°C) across a load range from 32% to
100% respectively. Below 32% load, the flow of ammonia to the system is terminated.
The Birchwood SCR uses a plate-type catalyst in a single, downflow, multi-layer reactor. The
system was designed for a NOX removal efficiency of 53% and a maximum ammonia slip of
5 ppmvd at 3% 02. The stack outlet NOX is maintained at less than 0.10 lbs/106 Btu
(0.043 g/MJ). The guaranteed catalyst life is 32,000 hours from initial operation or 63 months
from initial gas exposure, whichever occurs earliest.
-------
The anhydrous ammonia system consists of a 10.000 gallon storage tank with two-100% electric
vaporizers. The vaporizer assembly is installed in a natural circulation loop, taking liquid
ammonia from the bottom of the tank and returning vaporized ammonia to the top. Vaporizer
operation is controlled by a pressure switch on the tank. The vapor space at the top of the tank
acts as a reservoir for the supply of vaporous ammonia to the system. A single dilution air skid,
including ammonia flow control and shutoff valves, and two 100% dilution air blowers is located
on a platform near the injection grid.
An ammonia injection grid is situated upstream of the SCR reactor in a vertical run of ductwork -
see Figure 12. At this location, a diluted mixture of ammonia in air (6% ammonia by volume in
air) is uniformly injected into the flue gas stream through a network of pipes and nozzles.
Multiple flow control zones ensure that a uniform distribution of ammonia can be achieved
before entering the first catalyst layer. External flow control valves, installed in the piping
between the manifold and the grid, enables the operator to bias the ammonia flow for off-design
conditions.
The SCR system was designed based upon achieving the following inlet flow conditions to the
first catalyst layer:
1. Maximum flue gas temperature variations from the mean value of ± 18°F (±10°C).
2. Maximum deviation of flue gas velocity from the mean value of ± 10% for 90% of the
cross-sectional area, and ± 20% for the remaining 10% of the reactor duct area.
3. Deviations in the molar ratio of ammonia to inlet NOX less than ± 5% from the mean value.
A V,2th physical flow model confirmed that the design and location of the ammonia grid and gas
mixer achieved these performance requirements.
One of the major concerns when operating the SCR down to 32% load is maintaining adequate
temperatures to the reactor. This is important for avoiding ammonium bisulfate formation on
catalyst surfaces and minimizing ammonia slip exiting the process. For the Birchwood facility,
this was achieved through an external gas bypass around the secondary economizer. As is
common with this arrangement, gas temperatures approaching 1000°F (537.8°C) are mixed with
lower temperatures exiting the economizer to achieve the desired thermal conditions to the SCR
system. A unique bulk flue gas mixer directly upstream of the ammonia injection grid was
necessary to minimize gas temperature imbalances - see Figure 13.
Maintaining sufficient gas temperatures to the SCR at low loads required the installation of an
economizer bypass system. The bypass duct is designed with a shutoff slide gate and a louver
control damper. As shown in Figure 14, these dampers are opened at approximately 125 MWe to
provide adequate temperatures to the grid before ammonia is introduced into the process. The
minimum allowable temperature for ammonia injection is 580°F (304.4°C).
As flue gas is bypassed around the economizer at lower loads, the boiler efficiency is decreased
due to the higher gas temperatures exiting the air heater. A novel approach to improving boiler
-------
efficiency during this operation is the split economizer design. Feedwater enters a cased
(primary) economizer located in the ductwork exiting the SCR reactor chamber - see Figure 11.
The feedwater flows through a connecting link at the exit of this primary section to the secondary
economizer located in the boiler backpass. This patented design has a twofold benefit of
providing higher relative boiler efficiencies and improved catalyst performance across the load
range. With this arrangement, temperatures at any operating point can be optimized to enhance
catalytic reactivity performance. Improvements in boiler efficiencies ranging from 0.5% to over
1.0% were realized as compared to those achieved with a single backpass economizer design -
see Figure 15.
The SCR system was designed with a 30% reactor bypass with dampers at the inlet and outlet of
the reactor chamber - see Figure 16. During cold startup (catalyst temperature below 300°F or
148.9°C), the inlet control damper and outlet shutoff dampers are closed. When the flue gas
temperature exiting the boiler economizer approaches 400°F (204.4°C), the reactor bypass slide
gate closes and the total gas flow passes over the catalyst. Ammonia injection does not occur
until temperatures approaching 580°F (304.4°C) are reached. The main advantage of this
arrangement is to protect the catalyst from unbumed hydrocarbons, ash deposits, and low flue
gas temperatures during startup and low load operation. Also, during hot restarts, the chamber
can be isolated until temperatures exiting the boiler better match those internal to the reactor.
The basic function of the ammonia injection controls is to supply the proper flow of ammonia to
the system. This is based upon the NOX concentration at the inlet and outlet of the reactor, the gas
flow entering, and the molar ratio of NH3/NOX. After a flue gas sample is analyzed for both inlet
and outlet NOX, 4-20 milliamp signals from the plant's distributed control system (DCS) are used
to properly position the ammonia flow control valve. If the outlet NOX concentration is still
higher than the desired setpoint, the valve position is adjusted to increase the flow of ammonia to
the system. Manual bypass valves are used in the event that the automatic valve fails or requires
maintenance. A simplified schematic of the control system is shown in Figure 17.
Monitoring flue gas temperature across the SCR system is critical for preventing the formation of
ammonium bisulfate deposits on catalyst surfaces. At the Birchwood facility, gas temperature is
monitored at several locations including the economizer outlet, the ammonia injection grid inlet,
the top of the reactor chamber and the bottom of the last catalyst layer. The primary purpose for
each of these monitoring planes is described in Figure 18. In general, the permissive for injecting
ammonia will not occur until a minimum temperature of 580°F (304.4°C) is obtained at each of
these locations.
Another feature that is a part of the overall NOX system design is the air heater. The air heater
installed at SEI was specifically designed with materials and equipment for maintaining
availability, achieving high thermal performance, and minimizing maintenance. The design
addresses the potential impact of ammonium bisulfate deposits on air heater surfaces. The
formation temperature of ammonium bisulfate is dependent upon the concentrations of NH3 and
SO3 entering the air heater. Typical formation temperatures range from 400 to 450°F (204.4°C to
232.2°C). Consequently, the intermediate and cold end layers are the primary areas of concern.
-------
The air heater for Birchwood has enameled heating surface for both the intermediate and cold
end layers. Four stationary water washing devices (two each on the hot and cold end) and two
soot blowers (one each on the hot and cold end) allow the operator to clean air heater surfaces.
Details of the element depth, gauge, material, configuration and equipment are described and
compared with a standard air heater design in Figure 19.
Plant Operation
The steam generating unit passed all performance and emission tests in September, 1996.
Commercial operation of the plant occurred on November 14, 1996. From the initial first fires
right through the last controlled ramping tests, the TFS 2000™ firing system performed
successfully. Very little control system field tuning was required. As a result, more time was
available for tuning elsewhere, and the entire commissioning process from the firing system
through the SCR equipment was uneventful. For example, most of the TFS 2000™ firing system
preset damper ramps for the air and fuel streams were left as initially programmed. Also, the
stoichiometric flow ramps initially set for the SOFA boxes were left as originally designed.
Results of initial performance testing showed the carbon in the flyash maintained below 5% over
load - see Figure 20.
The Birchwood facility is cycled from full load down to minimum load (approximately 72 MWe)
on a daily basis. To assist in this plant operation, CETOPS™ Total On-Line Performance System
including OTIS™ On-Line Thermal Information System and BSCA™ Boiler Stress and
Conditioner Analyzer modules were installed. The OTIS™ program was expanded as part of a
research and development effort to evaluate furnace heat absorption rates during operation with
the TFS 2000™ firing system. The system records furnace temperatures from chordal
thermocouples, as well as monitors the operation of the firing system. Specifically, the OTIS™
system data package includes individual windbox damper positions, overfire air flows, windbox
to furnace differential pressures, SCR inlet NOX emissions, SCR inlet O2 and unit load. The
system permits the remote observation of the TFS 2000™ firing system during normal load
dispatch routines. In addition, the OTIS™ system has proven to be a valuable tool in evaluating
operating conditions throughout boiler transients. The system provides actual plant operating
information.
Since commercial operation, the plant has experienced approximately forty startups - 85% hot
starts. Ramp rates on the order of 5 MW/minute have not caused stack emissions to exceed
permitted requirements. As shown in Figure 21, the average inlet NOX loading to the SCR is
0.19 lbs/106 Btu (0.082 g/MJ). NOX emissions leaving the SCR system are controlled below
0.10 lbs/106 Btu (0.043 g/MJ). The SCRNOX reduction efficiency ranges between 60% to 65%
over the load range - see Figure 22. Ammonia consumption ranges between 80 Ibs/hr (0.01 kg/s)
to 50 Ibs/hr (0.006 kg/s) over the load range. Ammonia consumption versus load is shown in
Figure 23. Due to the continuous cycling requirements of the plant, the ammonia controls have
operated primarily in manual mode.
During commissioning, ash extracted from the air heater outlet hoppers was analyzed for
ammonia. The ammonia in the ash ranged from 200 to 250 ppmw. Ammonia slip is not
10
-------
continuously monitored due to the poor experience of these analyzers in a dust-laden
environment.
The SCR system is meeting the site's emission requirements under all operating conditions. The
plant did experience ammonia leaks in the piping connections around the ammonia storage tank.
These were a result of misalignment of threaded pipe connections between the tank and electric
vaporizers. During the spring outage, the tank was drained and all piping on the skid welded.
Threading occurring at the valves were re-cut and a compatible joint compound applied. In the
future, it is recommended to maximize welded connections for all ammonia piping.
The plant experienced its first planned outage on April 26 through May 5, 1997. All SCR system
components were inspected and sample catalyst plates pulled for activity testing. A visual
inspection of the reactor internals revealed no significant ash buildup on catalyst surfaces. A
coating of ash was seen on soot blower piping and ash accumulation plates between the modules.
The ash had a grayish color with a consistency of talc. The air heater baskets showed no
indication of ammonium bisulfate deposits. At this stage of the catalyst life, ammonium bisulfate
formation was not expected to be a problem.
Conclusion
To date, the SEI Birchwood Power Facility has successfully maintained emission levels below
0.10 lbs/106 Btu (0.043 g/MJ) on a 30-day rolling average, while maintaining acceptable levels
of unbumed carbon and ammonia in the flyash. This has been accomplished by an integrated
NOX reduction system consisting of a TFS 2000™ staged tangential firing system and selective
catalytic reduction (SCR). The success of the project is a direct result of the conservative design
of the furnace, pressure parts (i.e., split economizer), pulverizers, firing system, plant controls, air
heater, ductwork, dampers and SCR to ensure that emission requirements are met under all
operating conditions. It is anticipated that these reliable technologies will be called upon to meet
even lower regulatory NOX emissions in the future.
References
1. J. L. Marion, D.P. Towle, R.C. Kunkel, R.C. LaFlesh, "Development of ABB CE's TFS
2000™ Tangential Firing System". EPRI/EPA 1993 Joint Symposium on Stationary
Combustion NOX Control, reprinted as TIS 8603, 1993.
2. T. Buffa, D. Marti, R.C. LaFlesh, "In-Furnace, Retrofit Ultra-Low NOX Control
Technology for Tangential, Coal-Fired Boilers: The ABB C-E Services TFS 2000™ R
System", EPRI/EPA 1995 Joint Symposium on Stationary NOX Control, reprinted as
TIS 8623, 1995.
11
-------
\ car FrcdCTicVsturc. V A
irojlaiion* Radiant Rditu Siciin Generator
MCR STEAM FLOW 1
SHOPP-ESSLRH TEMP 2
REHEATER STEAM FLO* I
RKO FRES5LR£'TEMP -
COMMERCIAL OPER-VnO1-; DATE
Figure 1: Plant Design Parameters
Figure 4: Low NOX System Strategy
Volatile Matter (VM)
Fixed Carbon (FC)
Ash
Moisture
Nitrogen
Sulfur
HHV. Btu/lb
Kcal/kg
Tvpical
33.50
50.00
9.50
7.00
1.29
0.91
12.321
6.845
Ranee
FC/VM<2.0
5-20
3- 15
1 .80 max.
0.3 - 1 2
12.000- 13.000
6.667 - 7.222
FIRING SYSTEM TFS 2000™ Tangential Firing System
PULVERIZERS Four HP 863 With DYNAMIC™ Classifiers
ECONOMIZER Primary (Cased) and Secondary-
Split Economizer Design
SCR SYSTEM Single Vertical Downflow Reactor with
Plate Type Catalyst
AIR HEATER Single Air Heater - Enameled Surface
Figure 2: Typical Coal Analysis
(As received, % by weight)
Figure 5: Plant NOX Design Features
Fe203
CaO
Na^O
Chlorine
14 % max.
8 % max.
0.12 0.39%
< 0.25 %
Top & Bottom
Air Qefleoor
Figure 3: Major Constituents in the Ash
(As received, % by weight)
Figure 6: Coal Nozzle Tip
12
-------
Figure 7: Concentric Air Nozzle Tip
(Fixed Offset)
Figure 10: Zoned Stoichiometry Control
Schematic
Figure 8: Tangential Firing Pattern with
CFS
Figure 11: Schematic of SCR System
Arrangement
Figure 9: Schematic Diagram of TFS 2000™
Windbox Arrangement at SEI Birchwood
Figure 12: Ammonia Injection Grid Inlet
Manifold and Piping
13
-------
Figure 13: Static Mixer
Figure 16: Damper Schematic For SCR
System
AjrFlo" InldNOv Oullcl NCK Oulln NOv
1 1.
Figure 14: Economizer Outlet Gas
Temperature Vs. Boiler Load
Figure 17: Ammonia Injection Control
Schematic
0 50 100 ISO 200 250 300
I BOILER LOAD. MW RATING
CAS TEMPERATURE MONITORING
L ECONOMIZER OUTLET
AMMONIA IKJEcnoN CHI
OL SCRINLET- MOMTOBCSC (HIGH TEMPERATVRE)
rv SCR OUTLET . PERMlssrvT FORAMMONtA CVIECT1ON
Figure 15: Relative Increase in Boiler
Efficiency Vs. Boiler Load Using a Split
Economizer Arrangement
Figure 18: SCR Temperature Controls
14
-------
SE1 BIRCH WOOD STANDARD
\t yn.ii. foyre. crrat CAier yjjji. co\n
Figure 19: Comparison of the SEI
Birchwood Air Heater with a Standard Air
Heater Design
Figure 22: SCR NOX Reduction
Efficiency Vs. Boiler Load
Figure 20: Carbon-in-Flyash Vs. Boiler
Load
Figure 23: Ammonia Consumption Vs.
Boiler Load
Figure 21: NOX Emissions Vs. Boiler
Load
15
-------
EPRI-DOE-EPA COMBINED UTILITY AIR POLLUTANT CONTROL SYMPOSIUM
"The Mega Symposium"
August 25-29, 1997; Washington, DC
IMPACT OF COAL QUALITY AND COAL BLENDING ON NOX EMISSIONS
FOR TWO PULVERIZED COAL FIRED UNITS
C.M. Rozendaal
J.G. Witkamp
KEMA
Utrechtseweg 310
6812 AR Arnhem, the Netherlands
H.N. van Vliet
EZH
P.O. Box 909
2270 AX Voorburg, the Netherlands
A.M.C. Vissers
EPZ
P.O. Box 158
4930 AD Geertruidenberg, the Netherlands
Abstract
In the Netherlands almost 45% of the electricity is generated with pulverized coal. Coals are
imported from countries all over the world. Since the quality of the imported coals varies to a
large extent, most of the coals are blended at a central blending facility to ensure that the
specifications of the blend meet the design specifications. The impact of coal quality has been
investigated in a 520 MWe tangentially fired unit equipped with low-NOx burners and overfire
air ports. Twelve coal blends have been fired under identical circumstances. NOX emissions
and burnout characteristics have been successfully correlated with coal quality parameters and
excess air ratio. In a 600 MWe tangentially fired unit equipped with Pollution Minimum
burners and overfire air ports the impact of coal blending has been investigated. A comparison
of plant performance has been made for two different blending methods: blending the coals
prior to pulverization (normal blending scheme) and blending the coals in the furnace by firing
different coals at separate burner levels. The impact of flue gas recirculation through the
burners on NOX emissions has also been investigated with this unit.
-------
Introduction
Since 1971 NC\ and SO, emissions have been under debate in licensing procedures for new
and existing power plants. In the Netherlands a General Administrative Order on the emissions
of NOV, SO, and dust from large combustion installations came into force in 1987. This
Administrative Order was revised in 1992. For new coal fired power plants, with a license
granted at or later than January 1, 1990, the NON and SO, emissions were limited to a value of
200 mg/m03 (at 6% O2). From October 15, 1992 onwards the particulate emission was limited
to 20 mg/m03 In 1989 the NOX and SO2 emission by the Dutch utilities amounted to 74,000
and 45,000 tons/year, respectively. For existing large combustion plants belonging to a utility,
a special agreement was made in 1990 between the government and the utilities. It was agreed
to reduce NOX emissions to 35,000 tons/year and SO2 emissions to 18,000 tons/year by the
year 2000. This agreement offered the opportunity to apply measures for NOX reduction at
those units where they are most cost-effective.
In the Netherlands there are five coal fired power stations with a total of seven coal fired
units. Five units are tangentially fired and two are horizontally opposed wall fired. The total
capacity of the coal fired units is 3900 MWe. Presently all Dutch coal fired power plants are
equipped with flue gas desulphurization units and low-NOx combustion techniques. At two
units selective catalytic reduction is applied as well, for a further reduction of NOX emissions.
During the past few years several programs have been performed to assess the effect of burner
and boiler settings on NOX emission, amount of unburned carbon in the fly ash and boiler
efficiency and to optimize the combustion conditions. It is expected that the requirements of
the year 2000 can be met with the techniques applied.
For coal firing the Netherlands are completely dependent on imported coals. Coals are
imported from countries all over the world, with an emphasis on the USA, Colombia,
Indonesia. Australia, Poland, South Africa, Russia and China. Presently, in order to reduce fuel
costs and thereby the cost of electricity generation, an increasing amount of coal is bought on
the spot market, which may be a coal of poor quality by a high ash content or a low calorific
value. Blending of coals is therefore becoming an increasingly important instrument for
improving the combustion behavior of the coals, meeting emissions limits and reducing costs.
In the Netherlands coal blending has been practised for many years on the power plant level
and since 1992 on a national level with the commissioning of blending silos in the harbor of
Rotterdam, which supplies most of the power plants, including the Maasvlakte and the Amer
Power Stations, with coal blends. Because of the large variety in coal composition and the
increasing amount of extreme coals it is important to assess the effect of coal composition on
NOX emission, unburned carbon in the fly ash and boiler efficiency. During the past few years
several programs were performed to collect information on a wide range of coal compositions
and on the effect of blending of coals that are substantially different in composition with the
aim to minimize fuel costs by the combustion of low-quality coals in coal blends. In this paper
the results are reported of firing 12 different coal blends at the Maasvlakte Power Station
under well controlled conditions and the blending of a high-volatile with a low-volatile coal at
the Amer Power Station.
-------
Experience with correlations on NOX and burnout
In the past several researchers tried to correlate NOX emission with coal properties. An
example is the work by Pohl et al.1 in 1983. One of the important statements in this work was
the distinction between conventional combustion and low-NOx combustion. Pohl derived an
expression for NOX emission as a function of nitrogen content, the amount of volatile matter
and the amount of fixed combustibles. For conventional combustion an increase of NOX
emission with an increase of the amount of volatiles was found, whereas for low-NOx combus-
tion the opposite happened: a decrease of NOX emission with an increase of the amount of
volatiles. This finding reveals the most important feature of low-NOx combustion: the more
nitrogen is released with the volatiles under well-controlled (substoichiometric) conditions, the
lower the NOX emission. Several other researchers2"6 established the importance of parameters
like the volatile content and the Fuel Ratio (fixed combustibles divided by volatile matter). A
relation which is often quoted is the relation derived by Nakazawa et al.6 for Matsushima
Power Station in Japan. Matsushima Power Station has tangentially fired boilers. The
relationship is expressed by:
NOX 100-(N 0.8) + A-(FR 2) + 250 (1)
with: NOX NOX emission (ppm, as measured)
N Nitrogen content of the coal (weight-%, dry and ash free)
A A =80 for FR< 1.6
A = 50forFR> 1.6
FR Fuel Ratio
For the burnout of coal the volatile matter content of the coal and the Fuel Ratio appear to be
important parameters too. This is stated in works by Takahashi et al.7 and Nakata et al.3.
Nakata established the following relations for the burnout for a swirl burner:
Uc -0.25 -FR + 99.978 (2)
and for a parallel flow burner:
Uc -2.95 -FR -r 102.29 (3)
The burnout can be converted into the amount of unburned carbon in the fly ash by the
following formula:
UBC _ c- _
(100 - Uc)-(100 • Ash) + 100 -Ash
with: Uc Burnout (weight-%)
UBC Percentage unburned carbon in fly ash (weight-%)
Ash Ash content of coal (weight-%, dry)
-------
Coal blending experiences
A major issue with coal blending is the question whether coal properties and combustion
characteristics are additive or non-additive8'9 The interactions between the components of the
individual coals in the blend are not very well understood, which makes the evaluation of
blends combustion very complex. Despite this complexity, however, only on a number of
occasions serious problems were encountered with severe slagging of the burners and fouling
of the superheaters in the Netherlands.
Coal quality and blend quality have a significant impact on power plant performance, with
respect to efficiency, emissions, fly ash quality, slagging and fouling. Maintenance and
availability are also influenced by the quality of the feedstock. Therefore, coal properties are
determined in order to assess whether a specific coal or coal blend may be fired in a particular
unit. The properties are determined using laboratory methods specified in national and
international standards. Most of the properties of a blend are calculated as the weighted
average of the values determined for the individual coals in the blend. Additive coal properties
are the heating value, the proximate and ultimate analysis and the chlorine content. Coal
properties which are probably non-additive are for example the Free Swelling Index, ash
fusion temperatures and grindability.
Description of the boilers
Maasvlakte Power Station consists of two identical units (520 MWe each), which are
tangentially fired and have been built according the low-NC\ technique developed by
Combustion Engineering10, available in the design stage in 1982. The Maasvlakte units have
been designed to supply 30% of the total combustion air through close coupled overfire air
ports. In comparison with the traditional design, the volume of the boilers has been increased,
together with the vertical burner distances, which lowers the average flame temperature,
resulting in a decreased formation of thermal NOX. Another advantage of such a widebody
furnace is the ability to fire a large variety of coals. After a number of years of operation the
static classifiers in the pulverizers were replaced with rotating classifiers in order to decrease
the amount of large particles in the pulverized coal to improve burnout and to reduce NOX
emissions. The burner system of Maasvlakte consists of five burner levels: four levels are
sufficient for full-load. As a result of the special agreement between the government and the
utilities the Maasvlakte Power Station has to meet the emission limit of 390 mg/m^ (yearly
average) instead of the previous limit of 750 mg/m03.
Amer Power Station consists of two units of which unit 9 (600 MWC) is a modern tangentially
pulverized coal fired unit, which was commissioned in 1993. Due to its design as a widebody
furnace and its use of PM burners and 25% (close coupled) overfire air, the unit is operated
below the NO, emission limit of 400 mg/rn^. The mills are Babcock mills with rotating
classifiers. The Pollution Minimum burner has been developed by Mitsubishi Heavy Industries
(MHI). The burner system of Amer 9 consists of six levels of PM burners, one level of which
is spare. Each burner consists of a burner box with three compartments, as shown in Figure 1.
-------
AUX.AIR
WEAK BURNER
Figure 1
PM burner
The pulverized coal transported by primary air is separated in a fuel-rich and a fuel-lean flow,
which are injected separately into the furnace. The fuel jets are surrounded by secondary
combustion air necessary for ignition. Below and above the fuel-rich burner, gas burners and
SGR ports for flue gas recirculation are located. The third compartment is used to inject
auxiliary combustion air into the furnace. The principle of NOX reduction with this burner
system has been described elsewhere''
Objectives of test programs
In the Netherlands the coal fired power plants are facing a number of difficulties with respect
to power plant operation. Each power plant has to deal with three major issues:
• variation of coal properties
• stringent emission limits
• disposal of fly ash.
The quality of coal is changing frequently. For instance, since the commissioning in 1993 unit
9 of the Amer Power Station has fired more than fifty different coals in more than 175 blends.
The emissions of NOX are strongly related to coal properties. In low-NOx units the combustion
of low-volatile coals generally results in higher NOX emissions than high-volatile coals. In the
Netherlands disposal of fly ash is not allowed. Fly ash is used for application in building
materials as concrete and cement. Most of the applications require an unburned carbon content
in the fly ash below 5%. Especially in early low-NOx units like the Maasvlakte units it is
difficult to meet both the emissions limit and the quality requirements of the fly ash.
Moreover, power station operators also have to deal with a changing quality of the feedstock
as a consequence of the coal purchase policy.
-------
In this respect a number of research programs have been performed to obtain a better
understanding of the coal quality impact on power plant performance. The objective of the test
program performed at the Maasvlakte Power Station was to correlate coal quality with full-
scale NOX emissions and unburned carbon values to be able to predict these before coals are
purchased and the coal blends are composed and delivered to Maasvlakte. The objective of the
test programme at the Amer Power Station was to obtain information on the impact of coal
blending and the blending method on NOX emission and burnout. Two different blending
schemes were investigated; blending the coals prior to feeding the coal to the boilers (the
standard procedure) and firing the coals separately at different burner levels.
Results obtained with Maasvlakte trials
Prediction formulas for NOX and unburned carbon were already used at the Maasvlakte Power
Station before the pulverizers were retrofitted and equipped with rotating classifiers. The NOX
prediction was based on a formula derived from operational data of a Japanese Power Station .
The coefficients of this equation were adjusted by using Maasvlakte data which had been
collected during normal operation13 The formula derived was a function of nitrogen content
and Fuel Ratio. The prediction showed useful results although the accuracy was sometimes
rather poor. After the retrofit of the pulverizers and the consequently different operational
settings of the boiler it was decided to perform new measurements with twelve coal blends of
varying compositions. All measurements were performed under identical circumstances with
respect to load, burners out of service, burner tilt and speed of rotating classifiers. Table 1
shows the proximate analysis of the coal blends investigated.
Table 1
Analysis of the Coals Investigated at the Maasvlakte Power Station
Blend
A
B
C
D
E
F
G
H
I
J
K
L
Moisture
(wt-%)
11.0
11.8
8.9
13.6
8.1
13.5
13.9
11.8
8.5
7.5
9.0
9.0
Ash
(wt-%)
9.0
8.6
9.0
7.3
11.0
8.4
7.8
9.5
11.0
9.9
14.1
12.3
VM
(wt-%)
29.9
28.7
30.5
31.3
24.5
30.8
29.9
28.6
29.7
29.2
25.8
26.4
HHV
(MJ/kg)
29.06
25.83
27.17
25.65
27.90
25.38
25.85
26.17
26.81
27.96
25.58
26.62
LHV
(MJ/kg)
24.97
24.75
26.10
24.49
26.91
24.23
24.69
25.08
25.76
26.92
24.61
25.61
S
(wt-%)
0.32
0.39
0.47
0.41
0.41
0.41
0.48
0.47
0.61
0.66
0.57
0.55
N
(wt-%)
1.52
1.46
1.13
0.92
1.24
1.17
1.17
1.19
1.20
1.54
1.55
1.64
FR
1.68
1.77
1.69
1.53
2.30
1.54
1.62
1.75
1.71
1.83
1.98
1.98
The test runs were performed at two or three excess air levels to ensure that all coal blends
would be compared on the same basis. The reason for this is that the oxygen content of the
-------
flue gases is difficult to control, but is of great influence on NOX emission. By interpolation of
the NOX curves at a certain oxygen content, NOX emissions are comparable. Figure 2 shows
the NOX emissions for the coal blends investigated.
600
200L—
1.0 1.5 2.0 2.5 3.0 3.5 4.0 4.5 5.0 5.5 6.0
O2 in flue gases (wet) [vol.-%]
— Blend A -4-Blend B -r- Blend C * Blend D —Blend E -«- Blend F
-*-Blend G —Blend H *• Blend I -*-Blend J -rBlend K SBlend L
Figure 2
NOX Emissions as a Function of Oxygen Content and Coal Quality
In this illustration the interpolation line at three percent oxygen content is also drawn. At this
oxygen level NOX emissions of all blends have been correlated with coal parameters.
The dependency of NOX on oxygen content (i.e. excess air level) is comparable for all coal
blends. An increase of the oxygen concentration in the flue gases by 1% results in an increase
of NOX emission by approximately 28 Dig/m^. The difficulty to operate in compliance with the
emission regulations becomes clear from this illustration. The special agreement between the
government and the utilities implied an emission limit of 390 mg/m03 (yearly average). With
most of the coal blends this is only possible with for instance, deep air staging.
NOX has been correlated by means of the least-squares method with volatile matter content,
Fuel Ratio, nitrogen content and with parameters composed of e.g. Heating Value, particle size
and ash content. These correlation exercises demonstrated that the volatile matter content
showed the best correlation and a simple prediction formula for NOX emission could be
derived:
1000 - 17.4-VMrt
(5)
with: NOX NOX emission (mg/m03, dry @ 6% O,)
VMd Volatile Matter (wt.-%, dry base)
-------
Figure 3 shows the predicted and measured NOX emissions as a function of volatile matter
content. Although some scatter is present, the prediction correlates well with the measured
values.
600
30 35 40
Volatile Matter [wt.-%] (dry base)
Measured — Prediction
Figure 3
Predicted and Measured NOX Emissions
A combination of nitrogen and volatile matter resulted in a small but statistically insignificant
improvement. A comparable program was carried out at unit 9 of the Amer Power Station.
That program showed similar results. NON emissions correlated very well with volatile matter,
although the coefficients were different from the ones presented for the Maasvlakte Power
Station.
The unburned carbon in the fly ash was also correlated with coal parameters, such as ash
content, volatile matter and Fuel Ratio. It appeared, however, that no simple correlation could
be found. Even the ash content did not correlate with unburned carbon. Burnout, however,
showed a correlation with Fuel Ratio that fitted reasonably well. The following prediction
formula could be derived:
Uc -0.9-FR + 100,83 (6)
The accuracy of this prediction is less than that for the prediction of NOX emission. The reason
might be the more complicated fly ash sampling in comparison with the measurement of the
NOX concentration. The relation is, however, in agreement with those found by Nakata et al.3
Recent experiences with the combustion of low-quality coals show that incorporation of the
grindability in to the formula (by means of HGI) may improve the prediction accuracy.
-------
Role of char nitrogen (test rig results)
An improved correlation for the NOX emission is probably possible when the nitrogen content
of the char can be included. In low-NOx units the conversion of char-nitrogen into NOX is
probably the major contributor for NOX. Information on this topic can be derived from KEMA
experiments performed in the KEMA l-MWth coal fired test rig14 Combustion tests were
performed with a series of 12 coals. The nitrogen content in the volatiles and in the remaining
char was determined by pyrolysis in a drop tube furnace at a temperature of 1200 °C.
The burner of the l-MWth test rig is an internally air staged burner (designed by the
International Flame Research Foundation) and experiments have been performed without air
staging and with 30% of the combustion air added as staging air downstream in the furnace.
The results of the combustion tests are presented in Figure 4. The experiments without staging
are referred to as 120/0; the experiments with 30% staging air are referred to as 90/30.
Conversion ratios of nitrogen in the coal to NOX have been calculated for all experiments.
Figure 4 shows the conversion ratios as a function of the percentage of fuel-N released with
the volatile matter. A coal with more nitrogen in the volatile matter will, in general, have "a
lower conversion of fuel-N to NOX. This indicates a lower conversion of volatile nitrogen into
NOX than the conversion of char nitrogen into NOX for both the 120/0 and the 90/30
experiments. The slope of the regression line for the 90/30 experiments in Figure 4 clearly
exceeds the slope of the 120/0 regression line. This indicates that the conversion of volatile
nitrogen to NOX for the in-furnace air staging experiments is much lower than the conversion
of volatile nitrogen for the 120/0 experiments and is probably close to zero.
Conversion ratio of coal-N to NOX (%), 100 ppm th
30
25
28 30 32 34 36 38 40 42 44 46 48 50 52 54 56
Relative amount of nitrogen in VM (%)
•*-N conv 90/30 -*~N conv 120/0
Figure 4
Conversion Ratios of Fuel-N into NOX versus the Amount of Nitrogen in the Volatiles
as a Function of the Original Amount of Nitrogen in the Coal
-------
Assuming that the conversion of volatile nitrogen into NOX equals zero for the in-furnace air
staging experiments, it is possible to calculate the char nitrogen conversions into NOX. A
thermal NO,, of 100 ppm is assumed. These calculated conversion ratios of char-bound
nitrogen into NOX is about 20% for most of the coals, which may also be the limiting factor
for NOX reduction by combustion techniques.
Results of the test program at unit 9 of the Amer Power Station
Tests performed
A survey of the blending configurations which were tested at the Amer Power Station, is
presented in Figure 5.
100% L
10
20
30
40
50
60
100% H
L+H
L+H
L+H
L+H
L+H
50% L
50% H
H
L+H
L+H
L+H
L
50% L
50% H
RUN 5
40% L
60% H
RUN 6
60% L
40% H
RUN 7
Figure 5
Overview of the Blend Tests (L = Low-Volatile, H = High-Volatile)
Normally, the five lower burner levels are used for full-load operation and burner level 10 is
not operated. However, due to problems with burner level 40, it was decided for these
experiments to perform all of them with burner level 40 out of service. The numbers in the
illustration refer to the burner levels.
Seven runs were performed. The first run was carried out with a reference coal of a usual
composition. The second and third runs were the baseline tests with the combustion of the
low-volatile and high-volatile coals, respectively. In run 4 a blend of the low-volatile and
-------
high-volatile coal was fired. The blend consisted of 50% of the low-volatile coal and 50% of
the high-volatile coal (on a mass basis). The combustion modes of runs 5, 6 and 7 are
illustrated in Figure 5.
Coal characteristics
Table 2 contains the characteristics of the coals investigated. The first run was carried out with
a reference coal from Poland. The low-volatile coal (run 2) was from South Africa. The high-
volatile coal was from Colombia.
Table 2
Proximate Analysis of the Coals
Run *•
Analysis T
LHV MJ/kg
Moisture %(ar)
Ash %(ar)
Volatiles %(ar)
N %(ar)
S %(ar)
Cl %(ar)
HGI
1 (REF)
24.87
10.3
12.3
30.0
1.1
0.75
0.14
50
2(L)
25.00
7.7
14.8
24.2
1.6
0.49
0.00
50
3(H)
25.31
11.3
8.4
34.9
1.2
0.61
0.00
47
Test results
Effect of excess air on NOX emission and UBC.
In order to be able the compare the test results at the same amount of excess air, experiments
were performed to assess the effect of excess air on NOX emissions and the amount of
unburned carbon in the fly ash (UBC). The results are given in Figure 6 for the three coals
and the blend of the high-volatile and low-volatile coal. The results refer to a situation with
the overfire air ports completely opened. The boiler control keeps the amount of overfire air
constant for a variation in excess air. Therefore, only the stoichiometry at the burners varies
with an increase or decrease in excess air. As a consequence, NOX emissions increase rapidly
when excess air is increased. The low-volatile coal exhibits much higher NOX emissions than
the high-volatile coal. The nitrogen in the char reacts to NO when the overfire air is mixed
with the combustion products. The amount of nitrogen in the char of the low-volatile coal is
probably significantly higher than for the high-volatile coal, resulting in higher NOX emissions
for the low-volatile coal.
-------
02 m flue gases (dry) tvol.-%J
O2 in flue gases (dry) [vol.-%]
• REF «. L + H -i LH
Figure 6
Influence of Excess Air Ratio on NOX Emissions and Unburned Carbon
Figure 6 also illustrates the impact of excess air ratio on the unburned carbon content (UBC)
of the fly ash. The unburned carbon content is in all cases below 5 percent. With three of the
four coals the UBC is even less than 3 percent. From Figure 6 it can be seen that no clear
trend exists between the oxygen content of the flue gases and the unburned carbon content.
One would expect an improvement of the burnout at higher oxygen concentrations. This is
only the case for the high-volatile coal. These unexpected results are probably caused by the
relatively large residence time in the furnace resulting in an already high overall burnout,
thereby reducing the impact of excess air. It is assumed that in this range of oxygen
concentrations the burnout (or unburned carbon content) is more or less unaffected by the
excess air ratio.
Influence of separate gas recirculation.
In order to test the effectiveness of flue gas recirculation by the Separate Gas Recirculation
technique of MHI on NOX reduction, some tests were carried out by replacing the recirculating
flue gases with secondary combustion air, without changing the stoichiometry at the burners.
Figure 7 shows the results with the reference coal. The experiments with replacing the
recirculation gas by air are compared with the experiments with excess air variation. The
experiments without recirculation gas were performed at two excess air ratios. As is shown in
Figure 7 NOX emissions are nearly the same with or without recirculation gas. This was also
observed with the other two coals and the blend. The unburned carbon content of the fly ash
slightly increased with this reference coal. With the other coals the unburned carbon content
remained the same. It was concluded that the replacement of SGR with secondary air does not
change the burnout significantly.
-------
§
E
_E
200
^^.
^-^^^^
g
e
= 3
C
O
•e
« 5
3
e.
|
0
^^___
— • a
02 in flue gases (dry) [vol.-%]
• With SGR
«. With SGR (extra) ^ Without SCR
O2 in flue gases (dry) [vol.-%]
SCR «-with SQR (extra) * without SGR
Figure 7
Impact of Recirculation Gas on NOX Emissions and Unburned Carbon
The heat rate, however, was improved by replacing the recirculated gas at the burners by air.
Boiler efficiency increased to some extent due to reduced heat loss of the flue gases and a
decrease of the amount of attemperation water in the reheat cycle. The increase in nett
efficiency of the unit was about 0.3% (absolute). The nett unit efficiency will rise to 42.6% in
case the flue gas recirculation through the burners is switched off.
Coal blending.
Figure 8 shows the NOX emissions as a function of the blending ratio. The data on the left and
right axes represent the situations with 100% high-volatile coal (100% H) and 100% low-
volatile coal (100% L), respectively.
NOy (mg/m03) @ 6 % O2
800 --
700; r
600
40% L 50% L 60% L
60% H 50% H 40% H
Blend ratio
100% L
I —02 = 4.2% ~»-O2 = 5.0%
Figure 8
Blend Ratio versus NO.. Emissions for two Excess Air Ratios
-------
Two tests are performed (runs 4 and 5) with a blend ratio of 50/50 (on a mass basis). Run 4
(stockyard blend) is indicated in Figure 8 by number 4. The data in this illustration represent
an O, concentration of 4.2 and 5.0% in the flue gases measured before the air heater. The
overfire air ports were fully opened. The O2 concentrations of 4.2 and 5.0% are higher than
the levels which are normally applied for boiler operation, resulting in this particular case in
fairly high NOX emissions, but the data collected at lower oxygen levels were less reliable than
the ones presented in the illustration. It can be seen that there is an almost linear relationship
between the blend ratio and the NOX emission. It is striking that there is almost no difference
in NOX emission between blending before firing (run 4) and blending by firing the different
coals at different burner levels. The results suggest that NOX emissions of coal blends may be
determined by a weighting average of the NOX emissions of the single coals.
The effect of blending on the carbon conversion efficiency is shown in Figure 9. In order to
correct for the ash content, the burnout (= % of carbon conversion) is presented, which can be
calculated on the basis of the UBC data. Because of the weak correlation of oxygen content
and burnout, the values for burnout have been averaged over the range of tested oxygen
concentrations. The burnout is in all four blend cases slightly better than the average values
based on the burnout of the single coals. It can be concluded that with this particular unit
blending does not lead to a lower burnout, as is sometimes suggested in literature for firing
mixtures of high-volatile and low-volatile coal15 The (small) positive impact may be attributed
to some extent to the input of the low-volatile coals on the lowest burner levels, resulting in
an increase of the residence time in the furnace for these low-volatile coal particles. However,
the results of run 4 were also slightly better than the calculated average, suggesting a general
positive effect of blending for this particular unit.
Burnout (%)
100 —
99
100% H
40% L 50% L 60% L
60% H50% H40% H
Blend ratio
100% L
Figure 9
Blend Ratio versus Burnout
-------
Future programmes
The correlation of NOX emissions and coal quality parameters will be continued for other
Dutch coal fired power plants. Moreover, the work on char nitrogen distribution will be
continued and a correlation with full-scale data will be performed. The ultimate goal of this
and other coal research in the Netherlands is to reduce fuel costs by the combustion of low-
quality coals, which will also be investigated by means of field tests. This work will be
supported by economic calculations performed with EPRI's Coal Quality Impact Model16
Conclusions
From the results of the field tests at the Maasvlakte Power Station it appears that NOX
emission and burnout can be predicted fairly well on the basis of relatively simple coal
properties. The best correlation could be achieved for NOX emission. In order to obtain a high
accuracy of the burnout prediction, special attention must be paid to fly ash sampling. For the
evaluation of the coal properties it appeared to be very important that the experiments were
performed under very stable and well-controlled combustion conditions. An improvement of
NOX predictions appears to be possible in case the nitrogen distribution between volatiles and
char is taken into account as well.
With respect to the blending experiments at unit 9 of the Amer Power Station, it can be
concluded that the NOX emission of a coal blend is the weighted average of the NOX emissions
of the single coals. The burnout of the blend is slightly better than the weighted average of the
single coals for this particular unit.
Flue gas recirculation for NOX reduction appears to be ineffective in combination with PM
burner technology and deep air staging.
Acknowledgment
The work on char nitrogen conversion was funded by The European Commission within the
Joule II program Clean Coal Technology R&D "Atmospheric combustion of pulverized coal
and coal based blends for power generation under contract no. JOU2-CT93-0380. The research
performed at the Maasvlakte Power Station and the work on char nitrogen conversion was
undertaken by order of the Electricity Generating Companies of the Netherlands. Novem is
gratefully acknowledged for funding the work on the blend trials at the Amer Power Station.
References
1. J.H. Pohl, S.L. Chen, M.P. Heap, and D.W. Pershing. "Correlation of NOX Emission
with Basic Physical and Chemical Characteristics of Coal," Proceedings of the 1982
Joint Symposium on Stationary Combustion NOX Control, Dallas (July 1983).
2. H. Kremer, R. Mechenbier, and S. Wirtz. "Nitrogen Oxides Formation and Destruction
During Nonstaged and Staged Pulverized-Coal Combustion," International Symposium
on Coal Combustion, Beijing, China (September 1987).
-------
3. Methods for predicting NO^ emissions in coal combustion and unburned carbon in fly
ash Effects of coal properties. T. Nakata, M. Sato, H. Makino, and T. Ninomiya.
CRIEPI, 1987. EW87003.
4. J.R. Gibbins, T.J. Beeley, C.K. Man, and K.J. Pendlebury. "Coal Testing for NOX and
Char Burnout Prediction in Pulverised Fuel Combustion," 19lh International Technical
Conference on Coal Utilization and Fuel Systems, Florida, (March 1994).
5. T. Yamada, S. Kanbara, H. Tominaga, and H. Matsuoka. "The Effect of Coal Quality
on Nitrogen Oxides Formation during Pulverized Coal Combustion1' International
Conference on Coal Science, Tokyo, Japan (October 1989).
6. T. Nakazawa, and T. Kawamura. "Operating Experiences of Matsushima Power Station
No. 1 and No. 2 Coal Fired Boilers." Thermal and Nuclear Power. Vol. 33,
March/April (1982).
7. K. Takahashi, T. Watanabe, S. Miyamae, T. Kiga, and H. Ikebe. "Experimentally
Determined Reactivity Index for Assessing Unreacted Carbon in Pulverized Coal
Boilers," International Symposium on Coal Combustion, Beijing, China, (September
1987).
8. Coal blending for power stations. A.M. Carpenter. IEA Coal Research, July 1995.
ISBN 92-9029-256-3. IEACR/81.
9. P. T0rslev. "Coal Blending and Coal Homogenisation Facilities," Indo-European
Seminar on Clean Coal Technology and Power Plant Upgrading, New Delhi, India
(January 1997).
10. R.J. Colette. "1985 Update on NOX Emission Control Technology at Combustion
Engineering," Proceedings of the 1985 Joint Symposium on Stationary Combustion
NOX Control, Boston (March 1985).
11. T. Namiki. "Application of a Pulverized-Coal-Fired Low-NOx PM Burner for Steam
Generation," Proceedings of the 1985 Joint Symposium on Stationary Combustion NOX
Control, Boston (March 1985).
12. C.H. Gast. "Environmental Impact of NOX and Technologies of Reduction of NOX
Emissions of Power Plants (Primary and Secondary Measures)," Indo-European
Seminar on Clean Coal Technology and Power Plant Upgrading, New Delhi, India
(January 1997).
13. C.M. Rozendaal, C.H. Gast, and H.N. van Vliet. "The Diabolic Triangle in Modern
Low-NOx Coal Firing," 11th International Conference on Power Stations, Luik, Belgium
(September 1993).
14. A. Hoogendoorn, C.M. Rozendaal, D. Boersma, and E.W. Stevenson. "Combustion
behaviour and pollutant emission of coal blends. Drop tube, test rig and full scale
experiments," KEMA report no. 33467-FPP 96-4710, 1996. Report prepared on behalf
of the Joule II extension program (JOU2-CT93-0380) and KEMA R&D contract 1994-
1996 (production sector).
15. J. Jacobs, and J. Zelkowski. "Perspektiven fur NOx-arme Kraftwerksfeuerungen." VGB
Kraftwerkstechnik 67. Vol. 8, p. 803 (August 1987).
16. G.S. Stallard, J.A. Arroyo, and A.K. Mehta. "Coal Quality Impacts on Steam Generator
Design," Fourth International Conference on The Effects of Coal Quality on Power
Plants, Charleston, South Carolina (August 1994).
-------
REDUCED NOX EMISSIONS FROM CERTAIN COAL BLENDS
FOR UTILITY BOILERS
L. S. Monroe
Southern Research Institute
2000 Ninth Avenue South
Birmingham, Alabama 35205
R. J. Clarkson
Southern Company Services, Inc.
Post Office Box 2625
Birmingham, Alabama 35202
J. W. Stallings
Electric Power Research Institute
3412 Hillview Avenue
Palo Alto, California 94303
Abstract
Recent pilot furnace testing of coal blends fired through a generic wall-type low NO* burner have
produced lower NOX emissions than either coal produced alone, unlike most coal blends. This
research originally was supported by a consortium of coal-related companies in Alabama, and has
been augmented by a tailored collaboration between Alabama Power and EPRI. The experiments
were carried out in the Southern Company Services and Southern Research Combustion Research
Facility. The results of these experiments, along with their implications for full-scale use, are
presented. Possible explanations for these surprising results include changed flame aerodynamics
produced by the non-uniform pulverization of the coal mixtures or flame staging caused by the
difference in the volatile content of the two fuels.
-------
Introduction
More stringent environmental regulations and a more competitive business climate are changing
the market conditions facing both coal users and producers. As a major producer and consumer
of coal, the industrial sector of Alabama would benefit from a coordinated program which
demonstrates the applicability of wide ranges of blends of Alabama coal with other fuels.
Therefore, through Southern Company Services, a consortium of coal users and producers,
namely Alabama Power Company and several coal suppliers that have Alabama coal interests, was
formed in order to increase the marketability and competitiveness of Alabama coals. This
consortium, named the Alabama Fuel Development Consortium (AFDC), consists of Alabama
Power Company, U. S. Steel Mining Company, Pittsburg and Midway Coal Mining Company,
and the Alabama Coal Association (represented by Drummond Coal Company and Jim Walter
Resources).
The AFDC was formed in 1995, with the initial meeting held on May 5, 1995. The primary
technical goals of the AFDC are: to broaden the domestic and international use of fuels produced
by Alabama-based companies, to improve the fuel efficiency of energy intensive industries, and to
further develop fuel-related technical expertise in the state. The initial task of the research funded
by the AFDC was to explore the coal quality effects of blending Alabama coals with other coals,
both domestic and international. The low-sulfur Alabama coals typically used for utility fuels have
medium to low volatile content, which result in marginal NOX emission performance. If
competing western U.S. or international coals could be blended with these coals, the less desirable
characteristics of each of the blended coals could be mitigated.
This paper describes a series of experiments where Alabama coals were blended with other coals
and fired in a research pilot-scale furnace. Initially, a baseline Alabama coal was fired, followed
by the addition of 25% by weight increments of a second coal, ending with the second coal alone.
Measurements of NOX emissions, unbumed carbon in the fly ash, grinding behavior, ash
deposition, SO2 emissions, and the ability of an electrostatic precipitator to clean the particles
from the flue gas was measured during the combustion experiments. Additionally, the stability of
the coals and blends at low burner firing rates was measured.
Description of the Pilot-Scale Research Facility
The Southern Company and Southern Research Institute Combustion Research Facility is located
on the Birmingham, Alabama, campus of Southern Research. This pilot-scale simulator of coal-
fired utility boilers was designed and constructed and is operated by Southern Research under
contract to Southern Company Services. The facility is jointly owned, however, Southern
Research is free to use the facility to perform confidential research for other parties, including
other utilities, with the payment of a facility usage fee to Southern Company Services.
The facility is described in detail elsewhere and, for brevity, will not be repeated here1. Basically,
the facility consists of all the necessary equipment to simulate full-scale plant operations from coal
handling and pulverization, through a low NOX burner, to fly ash collection. As is common
practice for research furnaces of this size, the burner is indirectly fired, that is, the coal is
-------
pulverized and stored in a fuel bunker and then resuspended into the primary air line, using a
weigh feeder. The pulverizer used in the facility is a refurbished CE Raymond bowl mill, Series
352. This mill, with a 35 inch bowl and 2 rolls, is rated at 2 tons per hour.
The furnace and pollution control parts of the facility are shown in Figure 1. The furnace is fired
vertically up through a single burner mounted on the floor of the furnace. The radiant furnace is
built from a series of refractory-lined, water-cooled sections, each with an inner diameter of 3.5
feet for a total furnace height of 28 feet. The furnace exit leads to a horizontal flue gas duct,
which contains banks of air-cooled tubes, intended to simulate superheater tubes. After passing
through this horizontal section, the flue gas continues down through a vertical duct with a series
of air-cooled heat exchangers which serves to preheat the combustion air and cool the flue gas to
a nominal 300T.
FDFaa
Scrubber
Burner
Figure 1. The Southern Company and Southern Research Institute Combustion
Research Facility
The flue gas is then routed to a cold-side electrostatic precipitator, a pulse-jet baghouse, and a
packed-column counter-flow scrubber, in series. The pilot ESP has three fields and a specific
collection area of 150 ft2/1000 actual cubic feet per minute flue gas. The pulse-jet baghouse is
-------
required by the facility's air pollution permit and is used to ensure that the paniculate emissions
are controlled. The scrubber is also required by the permit, and uses aqueous sodium hydroxide
to capture sulfur oxides.
Coals Studied
The Alabama coals of interest are generally from the Blue Creek Basin in the western part of the
state. These coals are compliance fuels for sulfur content and usually have volatile content in the
range of 20 to 25%. As with most lower volatile coals, these coals are marginal in meeting the
NOX emissions requirements of the 1990 Clean Air Act Amendments. Blending of these coals
Table 1
Coal and Coal Blend Analyses for the
Jim Walter #3 and Caballo Rojo Coals
I Walter #3 75%JW#3 50% JW #3 25%JW#3 Caballo
25% CR 50% CR 75% CR Rojo
Ultimate Analyses, as fired
Water, %
Carbon, %
Hydrogen, %
Nitrogen, %
Sulfur, %
Ash,%
Oxygen, %
Total, %
JW #3 fraction,
wt. %
Proximate Analyses,
Water, %
Ash,%
Volatiles, %
Fixed Carbon, %
Total, %
0.67
79.77
4.09
1.43
0.55
10.45
3.05
100.0
100.0
as fired
0.67
10.45
21.54
67.34
100.0
5.08
72.01
3.94
1.30
0.51
10.45
6.73
100.0
72.8
5.08
10.45
24.70
59.77
100.0
9.52
66.05
3.92
1.14
0.48
10.53
8.38
100.0
49.7
9.52
10.53
27.62
52.34
100.0
14.57
58.94
3.77
0.99
0.43
8.58
12.73
100.0
26.2
14.57
8.58
32.30
44.55
100.0
22.61
51.61
3.37
0.84
0.37
6.52
14.69
100.0
0.0
22.61
6.52 .
43.90
26.98
100.0
Heating Values
As Fired, Bru/lb
MAF,Bru/lb
JW #3 fraction,
wt. %
Grindability
Hardgrove Index
13844
15576
100.0
85
12523
14826
78.9
75
11364
14213
63.2
62
10361
13481
36.2
56
9057
12779
0.0
52
-------
with other potential fuels may mitigate this potential NOX emissions problem. The two Alabama
coals used in this study are Jim Walter #3, mined by Jim Walter Resources, and Shoal Creek,
mined by Drummond Coal Company.
The coals picked to blend with these Alabama coals were Caballo Rojo, a Powder River Basin
coal, and Mina Pribbenow, a Colombian coal. Both of these coals were mined by Drummond, but
recently the Caballo Rojo rights were sold by Drummond. The coal blend combinations were
chosen to be: blends of Jim Walter #3 with Caballo Rojo and Shoal Creek blended with Mina
Pribbenow. The testing was designed to look at the range of blending possibilities and involved
testing each coal as a baseline and then the two coals blended at nominally 25, 50, and 75 percent
by weight.
The analyses of the Jim Walter #3, the Caballo Rojo, and the blends are presented in Table 1. As
can be seen in the table, the blends vary smoothly from the Jim Walter #3 through the increasing
fractions of the Caballo Rojo. As evidenced by the calculation of blend proportion from the
analyses, the blends were close to the desired fractions.
The analyses of the Shoal Creek, the Mina Pribbenow, and the blends are likewise presented in
Table 2. As above, the analyses of the blended coals shows a smooth variation as more Mina
Pribbenow is added to the baseline Shoal Creek coal. The fraction of each coal, as estimated from
the analyses agrees closely with the desired fraction.
Description of Pilot-Scale Experiments
The pilot-scale experiments were designed to test a suite of coal quality-related parameters.
These parameters were prioritized as follows:
1. NOX emissions and loss-on-ignition (LOT) of fly ash.
2. Fly ash characterization for electrostatic precipitator performance evaluation.
3. Furnace temperature distribution.
4. Flame stability measurements.
5. Pulverizer operation.
6. Ash behavior evaluations (fouling and slagging).
The scaling and operation of the pilot-scale facility in simulating full-scale plants, along with
comparisons of pilot-scale data to full-scale plants, has been presented previously2 Briefly, the
simulation of full-scale units in this furnace has been quite successful for predicting NOX
emissions. For unburned carbon, the pilot furnace results are typically lower than the full-scale,
but show the same trends. The well-controlled particle size of the coal fuel, along with the use of
a single, well-controlled burner in the pilot furnace, are expected to give better coal conversion
than a full-scale boiler with less control over air and fuel distributions.
-------
Table 2
Coal and Coal Blend Analyses for the
Shoal Creek and Mina Pribbenow Coals
Shoal Creek
75% SC
25% MP
50% SC
50% MP
25% SC
75% MP
Mina
Pribbenow
Ultimate Analyses, as fired
Water, %
Carbon, %
Hydrogen, %
Nitrogen, %
Sulfur, %
Ash,%
Oxygen, %
Total, %
SC fraction, wt. %
Proximate Analyses,
Water, %
Ash,%
Volatiles, %
Fixed Carbon, %
Total, %
1.78
75.75
4.49
1.65
0.68
12.11
3.57
100.0
100.0
as fired
1.73
12.11
25.45
60.58
100.0
3.32
74.39
4.53
1.58
0.59
9.81
5.79
100.0
71.2
3.32
9.81
28.52
58.36
100.0
5.06
72.32
4.58
1.47
0.51
8.24
7.83
100.0
49.7
5.06
8.24
30.57
56.14
100.0
6.29
71.08
4.68
1.36
0.42
6.30
9.89
100.0
27.9
6.29
6.30
34.28
53.14
100.0
8.00
68.81
4.52
1.28
0.35
5.05
12.01
100.0
0.0
8.00
5.05
36.28
50.68
100.0
Heating Values
As Fired, Btu/lb
MAF, Btu/lb
SC fraction, wt. %
13371
15526
100.0
13014
14980
77.5
12694
14640
51.8
12430
14220
26.4
12000
13798
0.0
Grindability
Hardgrove Index
75
66
55
53
45
The testing occurred in two sessions: the Jim Walter #3 and Caballo Rojo coals and blends were
tested in August of 1995; while the Shoal Creek and Mina Pribbenow coals and blends were
tested in December, 1995. For all of these tests, the furnace was equipped with the dual-register
burner which simulates low NOX wall-firing. The experimental sequence for each of these set of
coals and coal blends was identical. The pilot furnace was brought up to thermal equilibrium by
firing natural gas. Once the furnace was heated, the Alabama coal was introduced to the furnace,
and, over the space of 12 hours, the fuel feed smoothly switched from gas to coal. After a few
more hours of stabilization, the testing program would begin. After the testing of the Alabama
coal (either Jim Walter #3 or Shoal Creek) was completed, the fuel feed would be switched to the
-------
blend of 75% Alabama coal and 25% other coal. After testing, the process was repeated for the
50/50 and 25/75 blends, followed by the pure other coal (Caballo Rojo or Mina Pribbenow for
these tests).
For each of the coals or coal blends, the initial testing was designed to explore the response of the
fuel, specifically for NOX emissions and for unburned carbon, to variations in furnace excess air.
The total air supplied to the furnace was first increased, then decreased, in order to obtain NOX
data from the emissions monitoring system, with the goal of furnace exit oxygen concentrations of
3.5% for the baseline, followed by 4.5% and 2.5%. During these stable periods of operation,
EPA Method 17 samples were taken at the electrostatic precipitator entrance. These samples
provide the mass loading of fly ash entering the ESP, but, more importantly, they yield samples
which can be analyzed for either LOI or unburned carbon. Also, during this time period, the flue
gas entering the ESP is sampled for fly ash particle size using cascade impactors and the gas
temperature in the upper half of the radiant furnace is measured. Additionally, the fly ash
resistivity was measured in situ during these stable testing periods. The electrical performance of
the pilot ESP was recorded during these tests, to compare to ESP computer model predictions of
actual operation.
Following the NOX and LOI testing, the stability of each fuel at lower burner firing rates was
tested. The firing rate of the burner was decreased on a fixed time schedule, and the firing rate at
which the flame extinguished was noted as the burner stability limit. The testing sequence was
repeated to help gain confidence in this limit. The furnace exit oxygen increased as the firing rate
dropped, similar to a full-scale plant. The primary air was lowered proportionally until the
minimum flow to prevent coal particle fallout was reached, and was then held at that level.
Similarly, the secondary air was decreased until the minimum flow was reached and then held.
Following this procedure, the pilot-scale burner has the same limitations on air flow as the full-
scale burners in practice.
The behavior of the various fuels was compared through the pulverizer. As with most small
research furnaces, this facility is indirectly fired, that is, the coal is ground in a pulverizer and then
stored in a pulverized fuel bunker and then injected into the primary air stream. Although there
are safety considerations in storing pulverized fuel, the benefits of more stable operation and the
use of a large, more realistic, mill outweigh these problems. Therefore, the pulverizer is isolated
from the burner and can be used to measure capacity and the adjustments to produce a certain
size of fuel. In these experiments, it was desired to test each fuel at the same fineness, so the mill
was adjusted to produce a grind which met a standard grind of 70 % (± 2%) by weight through
200 mesh. For most of the coals, it was possible to produce this grind, but, due to a problem with
our sieve shaker, the 75% Jim Walter #3 blended with 25% Caballo Rojo resulted in a grind that
was slightly finer than desired, about 77% through 200 mesh. The details of the grinding behavior
of the Shoal Creek and Mina Pribbenow blends are presented in the accompanying paper3
Finally, the behavior of the ash from the coals and blends was observed in the furnace. The
facility has a series of air-cooled tubes in the convective section which mimic superheater tubes in
a full-scale boiler. The buildup of ash deposits on the tubes is measured by the aerodynamic
resistance of the flue gas passing through the tubes and by the difficulty in removing deposits from
-------
the tube bank. Unfortunately, the facility does not have a slagging panel to accurately measure
the behavior of a fuel for slagging. Therefore, slagging behavior is generally assessed by the
nature of the ash collected as bottom ash and the rate at which it collects on the hot refractory
walls of the furnace.
NOX Emissions and Unburned Carbon Results
The measured NOX emissions from the testing of the Jim Walter #3 and Caballo Rojo coals and
blends are presented in Figure 2. The measured NOX at increasing furnace exit oxygen levels is
shown as a function of the amount of Jim Walter #3 coal present. The NOX emissions are shown
in units of both parts per million (at 3% oxygen dilution) and in pounds per million Btu. As can
be seen from the figure, the NOX generally decreases as the amount of Caballo Rojo, a Powder
River Basin coal, is added. The NOX emissions at about 75% Jim Walter #3 seem to be higher
DUU -
;
o"
fr? 600-
-4-3
a
g
a.
0.4.™ l
I i
m
en
6
w 200
M
0
S;
n-
1
__— — •
1 '
r
r
1
——A
—&•
1 '
\
\
^
^
i
0
D
"
1
2.E
1 '
i% I
3.5% J
4 5% H
|
V
>
!
^ T
^l
T
^-^
-\,
1
'urn
urn
'urn
AJ .
e -
^S— ^
1 '
ace
ace
ace_
1
02-
Oa-
0"
&
_
-
•«
1 — ~^
-
~
-1.0
-
-0.8
-
-0.6
-0.4
-
-0.2
-n n
m
O
O
'Z
100 90 80 70 60 50 40 30 20 10 0
J. Walter #3 Weight Fraction, %
Figure 2. NO, emissions from Jim Walter #3 and Caballo Rojo coals and coal
blends.
than expected. As mentioned above, the pulverized coal grind in this test was slightly finer than
the other tests performed. It has been generally true that finer coal particle size produces higher
NOX emissions in this furnace.
-------
The unburned carbon from the residual fly ash is presented as Figure 3. (We generally use
unburned carbon in our analyses, as it tends to be more precise when describing the efficiency of
carbon conversion. Unburned carbon, if there are no sulfates present, is typically about 0.5 to 1%
100 90 80 70 60 50 40 30 20 10
J. Walter #3 Weight Fraction, %
Figure 3. Unburned carbon in fly ash from Jim Walter #3 and Caballo
Rojo coals and coal blends.
by weight less than an LOI measurement.) As above, the unburned carbon is shown for three
levels of furnace exit oxygen for the increasing amount of Caballo Rojo in the fuel input. As
expected, the amount of unburned carbon in the fly ash decreases as more Caballo Rojo is added
to the fuel. The fuel blend with about 25% Caballo Rojo shows an unburned carbon level lower
than a general linear trend would show. As mentioned above, this result is consistent with a finer
grind of the fuel, which would cause increased NOX and decreased unburned carbon.
With the exception of the 25% Caballo Rojo, these blends show a linear trend of both NOX
emissions and unburned carbon as one fuel is gradually displaced by the other coal. Conventional
wisdom is to estimate the performance of coal blends as a imaginary fuel that appears to be the
weighted average of the two parent fuels, and this blend seems to behave in that manner.
-------
The NO* emissions measured in the testing of the Shoal Creek and Mina Pribbenow coals and
their blends are presented in Figure 4. These results are surprising in that they show NOX
emissions from most of the blends are lower than either coal alone. It was expected that the
blends would follow the linear blending rule, as was seen in the blends described above. The NOX
800
2.5% Furnace
I i |
3.5% Furnace
100 90 80 70 60 50 40 30 20 10
Shoal Creek Weight Fraction, %
-1.0
-0.8
-0.6
-0.4
-0.2
-0.0
m
*o
no
a
o
'5
en
Figure 4. N0r emissions from Shoal Creek and Mina Pribbenow coals and coal
blends.
reduction is most apparent at the 50% blend point, but is also observed at the 25% and 75% Mina
Pribbenow values. A repeat experiment of the blends in the 80 to 90 percent Shoal Creek range,
as shown in Figure 4 as solid symbols, seems to confirm the NOX emissions data. The addition of
a small amount of Mina Pribbenow coal, which moves the blend from 85% to 70% Shoal Creek,
seems to make the NOX reduction start. It seems that 10% Mina Pribbenow coal does not affect
the NOX emissions, but the bulk of the reduction occurs with a blend of only 30% Mina
Pribbenow coal.
The unbumed carbon results, presented as Figure 5, show an increase with the addition of 25%
Mina Pribbenow, followed by a gradual decrease to levels that are lower than the Shoal Creek
coal alone. Due to the difficulties in comparing data in the Jim Walter #3 and Caballo Rojo blends
caused by particle size fluctuations, a great deal of effort was used to ensure similar particle sizes
-------
for all five test fuels in the Shoal Creek and Mina Pribbenow campaign. The repeat experiment of
the blends between 80 and 90 percent Shoal Creek are consistent with the earlier data points. All
of the unburned carbon results are consistent with the observed NO* emissions values, in that the
unburned carbon increases as the NOX emissions are decreased.
The NOX results of the Shoal Creek and Mina Pribbenow coal blend tests were unusual, and were
difficult to understand. However, there is a report in the literature from Germany that shows the
same sort of behavior for a blend of two coals4. In the reported experiments, a blend of two
coals, with an unspecified blend ratio, was fired in a pilot-scale furnace at increasing overfire air
flows. Similar to the data here, the blend produced lower NOX emissions than either coal alone.
i ' i ' i i i 'i ii *i ' r
100 90 80 70 60 50 40 30 20 10 0
Figure 5.
Shoal Creek Weight Fraction, %
Unburned carbon in the fly ash from Shoal Creek and
Mina Pribbenow coals and coal blends.
Unfortunately, there is not enough detail given to help understand the results, with such factors as
blend ratio, neat coal analyses, blended fuel analyses, and unburned carbon or LOI not reported.
This NO* reduction is thought to be produced by a particle size segregation of the two coals from
pulverization. Previous work at Pennsylvania State University has shown that in grinding of a
coal blend, the coal with the lower Hardgrove Index will be found in the larger sizes at a fraction
higher than the bulk blend composition5. Conversely, the coal with the higher Hardgrove Index
-------
will then be found to predominate in the smaller particle sizes. Indeed, a vitrinite reflectivity
analysis of sieved fractions of the three Shoal Creek and Mina Pribbenow blends shows the Shoal
Creek dominating the smallest size and the Mina Pribbenow in the largest sizes. This is illustrated
graphically in Figure 6.
The NOX reduction seen in these blends is likely caused by internal staging of the flame by
aerodynamics of the different coal particles. The higher volatile Mina Pribbenow coal,
concentrated in the larger particles, will be injected into the flame with higher momentum. These
particles, due to the higher momentum, will penetrate farther into the furnace, all the time
releasing volatiles to the general cloud of smaller, burning Shoal Creek coal particles. This
100
a
a
a
75% SC Blend
50% SC Blend
A 25% SC Blend
100
100
Coal Particle Size, micrometers
Figure 6.
Fractions of each coal for different sizes in the Shoal Creek
and Mina Pribbenow blends.
volatile release would provide the hydrocarbon radicals necessary to reduce NOX formed by the
Shoal Creek coal particles back to N2. It may be possible to produce a blend of pulverized fuel by
mixing together two coals pulverized alone, so that the effects of flame aerodynamics and coal
milling can be separated. Then, a coal with high volatiles that may have better burnout
characteristics, can be placed into a blend in the larger sizes, with a low volatile coal in the smaller
pulverized coal fractions.
-------
Conclusions
A series of experiments was performed in a pilot-scale combustion research facility, intended to
evaluate fuels from lower volatile Alabama coals blended with a Powder River Basin and a South
American coal. The NO* emissions and unburned carbon in fly ash for the Jim Walter #3 and
Caballo Rojo coal blends showed nearly linear behavior, with the exception of the 25% by weight
Caballo Rojo which had a finer coal particle size.
The results from the firing of blends produced by adding Mina Pribbenow coal, a Colombian coal,
to Alabama's Shoal Creek coal, were surprising. The NOX emissions produced in the combustion
of the blends were lower than for either coal alone. This behavior has been seen by other
researchers with other coals, and it was observed that pulverization separates the coals into a
coarser fraction of the harder Mina Pribbenow coal and a finer fraction of the soft Shoal Creek.
The authors speculate that concentration of the more volatile coal in the larger sizes of the
pulverized fuel produces an internal staging of the low NOX burner flame, where the volatiles from
the Mina Pribbenow are supplied to the burning cloud of smaller Shoal Creek coal particles. The
unbumed carbon results were consistent with the NOX emissions, where the blends produced more
unburned carbon than either coal alone. The coal blends were harder to pulverize than the Shoal
Creek coal, but this was expected due to the harder nature of the Mina Pribbenow coal.
Acknowledgments
The authors would like to thank the companies and their representatives which make up the
Alabama Fuel Development Consortium: Mr. Ken Mooney of Alabama Power Company, Mr.
Keith Jansen of U. S. Steel Mining Company, Mr. Ray Iverson of Pittsburg and Midway Coal
Mining Company, and the Alabama Coal Association represented by Mr. Randy Tisdale and Mr.
John McClellan of Drummond Coal Company and Mr. Tony Przyblek and Mr. Bruce Hamilton of
Jim Walter Resources. The Electric Power Research Institute joined the program as sponsors
through a tailored-collaboration. Alabama Power Company's General Services Laboratory is
providing the necessary analytical- support for the work, with Mr. Herman Maddox as the APCO
coordinator. The Southern Research Institute staff responsible for the operation and maintenance
of the combustion facility are Wim Marchant, Sam O'Neal, and Bill Page.
References
1. L. S. Monroe, R. J. Clarkson, and J. W. S tailings, "Predictions of Full-Scale NOX
Emissions and LOI from Coal and Coal Blends in Pilot Combustion Experiments,"
presented at the Fourth International Conference on The Effects of Coal Quality on Power
Plants, Charleston, SC (August, 1994).
2. L. S. Monroe, R. J. Clarkson, and J. W. Stallings, "Comparison of Pilot-Scale Furnace
Experiments and Predictions to Full-Scale Boiler Performance of Compliance Coals,"
presented at the EPRI/EPA Joint Symposium on Stationary Combustion NOX Control,
Kansas City, Missouri (May, 1995).
-------
3. L. S. Monroe, R. J. Clarkson, J. W. Stallings, and A. K. Mehta, "Experimental and
Modeling Results from Pulverization of Coals and Coal Blends," presented at the EPRI
Effects of Coal Quality on Power Plants Conference, Kansas City, Missouri (May, 1997).
4. C. M. Rosendall, C. H. Gast, H. N. van Vliet, and K. R. G. Hein, "Coal Blending and its
Effects on NOX and Unburned Carbon in Large Scale Power Plants in The Netherlands,"
presented at DW-Kolloquium "Kohlenstaub-Aufbereitung: ein Beitrag zur NOX-
minderung," Essen (December, 1992).
5. H. Cho and P. T. Luckie, "Grinding Behavior of Coal Blends in a Standard Ball-and-Race
Mill," Energy & Fuels, Vol. 9, p. 59 (1995).
6. M. Cloke, Data Report to Southern Research Institute, June, 1996.
-------
Effect of Low NOX Firing Conditions on
Increased Carbon in Ash and Water wall Corrosion Rates
E. Eddings
K. Davis
M. Heap
J. Valentine
Reaction Engineering International
A. Facchiano
Electric Power Research Institute
R. Hardman
Southern Company Services
N. Grigas
GENCO
Abstract
Operating experience with low NOx firing systems suggests that, although NOx emissions
may be within acceptable limits, these firing systems can create conditions that lead to
unacceptable levels of carbon in the fly ash and high water wall corrosion rates. A reacting
computational fluids dynamics code which includes the physics and chemistry of particle
transport and deposition, coal devolatilization and char oxidation, and both fuel and
thermal NO formation has been used to determine how NOx emissions can be maintained
at the same low levels without attendant operational problems.
Simulations have been made for both corner and wall fired boilers retrofitted with low
NOx firing systems for both the pre- and post- retrofit designs. Detailed simulations have
been made for conditions within the coal nozzle to ensure that input conditions for the
complete boiler model are known. This is particularly important when considering the near
field of low NOx firing systems. The paper demonstrates that the effects of char
deactivation must be taken into account if accurate predictions of carbon burnout are
required in full scale systems. Comparisons are made of local conditions on the water wall
in the firing zone that can cause increased water wall corrosion rates. These local
conditions include reducing gas composition, net heat flux, particle deposition rates and
the temperature and composition of the particles being deposited as a function of location.
-------
Introduction
The ability of computer models to simulate coal combustion in real systems has
improved considerably over the last decade, not only because of improvements in the
physical and chemical models, but also due to computer hardware capabilities that have
improved dramatically over the same period. Although several industries have made
extensive use of computer simulations, only recently has the utility industry begun to
accept their value. This is due in part to the fact that the models can now simulate real
hardware rather than just simple test combustors. Computer simulation reduces the
cost of hardware development by providing insight and giving an indication of the
limiting performance if the equipment were to be operated under ideal conditions.
Simulations do not eliminate the need for experimental programs prior to
demonstrations, but they can provide valuable insight into complex processes.
Many coal-fired units retrofitted with low NOx firing systems are experiencing
problems due to increased carbon in the fly ash or increased water wall wastage.
Although the magnitude of the increase in unburned carbon is system and coal
dependent, it appears that any significant reduction in the emissions of nitrogen oxides
is always accompanied by an increase in the amount of unburned carbon in the fly ash.
Low NOx firing systems are designed to increase the residence time of coal particles in a
fuel rich environment. Thus it might be expected that there is insufficient time at
temperature to ensure complete carbon oxidation in a low NOX firing system-
Low NOx firing systems can create conditions on the water walls that exacerbate
wastage rates. It is thought that fireside corrosion of water wall tubes in coal fired
boilers is primarily caused by high CO conditions close to the wall, causing aggressive
sulfidizing conditions. The deleterious effect of reducing conditions, including the
presence of species such as HzS, is through the formation of iron sulfide (FeS) lamellae
in the magnetite (FesO-i) on the tube surfaces. These sulfide lamellae are less protective
than magnetite causing an increase in the rate of metal loss. The presence of unburned
coal particles in the wall deposits has been connected with accelerated corrosion
because of the generation, locally, of more extreme reducing conditions. Coal particles
may also be important in transporting species such as sulfur to the walls, although the
pyritic component is thought to arrive as segregated molten spheres of FeS. High levels
of heat flux are also thought to enhance corrosion rates. They may do so by promoting
slagging which restricts oxygen ingress to the scale and hence makes the conditions at
the rube surface more sulfidizing.
Simulation of pre- and post-retrofit designs have been made for two boilers fitted with
low NOX firing systems. This paper concentrates on the conditions that might lead to
increases in water wall wastage and unburned carbon.
-------
Simulations
Model Description
The reacting computational code GLACIER was used in this study. GLACIER is based
on software developed over the last fifteen years by Smith and co-workers [1-4].
Particular emphasis has been placed on simulating coal combustion and pollutant
formation. The fluid dynamic properties of the gases and particulate matter flowing
through a combustion chamber are computed to determine the convective and diffusive
mixing of energy and relevant chemical species. The flow and chemical reactions are
influenced by the radiative heat transfer that is generated by the combustion process.
The radiative heat transfer influences the local devolatilization and heterogeneous
combustion of the particles, the local gas temperature and thus the local fluid dynamics.
Computations include full mass, momentum and energy coupling between the gas and
particles as well as full coupling between turbulent fluid flow, chemical reactions and
radiative and convective heat transfer. Two submodels are particularly relevant to this
paper: particle transport and char burnout.
Particle Transport
GLACIER models particle transport in turbulent flows based on an extension of
stochastic process modeling and turbulence theory. This model provides excellent
resolution of particle-phase transport, reactions and deposition while retaining
computational efficiency.
In this approach, the particle mechanics are solved by following the mean path or
trajectory for a discretized group or cloud of particles in a Lagrangian frame of
reference. Particle mass and momentum sources are converted from a Lagrangian to an
Eulerian reference frame where they are coupled with gas phase fluid mechanics. The
dispersion of the particle cloud is based on statistics gathered from the turbulent flow
field. The dispersion model requires no adjustable parameters beyond specification of
the turbulent flow field itself and is independent of any particular turbulence model.
The description of the turbulent flow field required by the model is in terms of the
mean square velocities and Lagrangian correlation functions, available in terms of
parameters from the k-e gas turbulence model and particle properties. All particles are
assumed to be spherical.
This model has been evaluated by comparison to exact solutions, the most accurate
alternative models, and experimental data. It reproduced the exact solutions both
numerically and as analytical functions. These analytical solutions have been used as
calibrations for some of the most accurate models of particle transport in turbulent
flows. In a comparison to turbulent dispersion data collected under nine different flow
-------
conditions, the model predicted the experimental data of all nine cases -within the
experimental error [5].
The particle transport model allows for particles to stick (deposit) or bounce off water
walls and steam tubes depending on the local particle temperature, size, etc. As various
particles are removed from the flow domain, the cloud statistics are adjusted
accordingly to reflect the change in particle momentum, dispersion, etc.
Char Burnout Modeling
GLACIER includes a char oxidation model with a mean oxidation rate and a new
Carbon Burnout Kinetic (CBK) model which has been recently added. CBK is a coal
general formulation designed to evaluate the total extent of carbon burnout and flyash
carbon content for pulverized coal particles with known temperature/oxygen histories.
The CBK model comprises the following four main components:
. The single-film char oxidation model of Mitchell and coworkers [6,7]. In the current
version of CBK, the original published correlations have been modified to
accommodate thermal annealing and statistical kinetic submodel integration and
have been extended to include finite kinetics for lignites.
. A submodel of statistical variations in reactivity and char density among single
pulverized fuel particles [8].
• A submodel of thermal char deactivation, or thermal annealing, adapted from the
model proposed by Suuberg [9]. In this submodel the char reactivity is a variable
whose value depends on temperature and time. CBK incorporates the thermal
annealing model in full, differential form capable of accommodating a wide range of
non-isothermal temperature histories. As discussed in detail elsewhere [10], the
thermal annealing submodel is defined by three parameters determined by
evaluation of a collection of thermal deactivation data from relevant literature,
including recent data generated in collaboration with Imperial College [11].
• A physical property submodel describing diameter/ density changes during
combustion, fragmentation, and mineral inhibition in the late stages of combustion.
The comprehensive CBK model was used in a series of one-dimensional reacting flow
simulations to better understand the roles of the various individual phenomena
described in the code on the overall prediction of carbon burnout. The thermal
annealing submodel was found to have a large effect on reactivities and burnout levels
at combustion temperatures and times typical of pulverized-coal fired boilers. The
submodel of statistical kinetics and densities has a small effect throughout most of
burnout, but contributes significantly to the long tails observed in laboratory burnout
curves. Finally, the submodel of carbon/mineral interactions contributes substantially
to the long burnout tails. None of these individual component submodels can be
deleted from CBK without some loss of general predictive capability. For boiler
applications however, coarse binning can be used in the submodel of statistical kinetics
and densities without introducing significant error. The CBK model was successfully fit
-------
to six sets of long-residence-time data taken in the heated-wall reactor at Sandia
National Laboratories. Based on global model predictions along with the kinetic data
presented by Mitchell et al, [6], the single-particle temperature predictions in the
statistical kinetic model and the successful prediction of low reactivities in residual
carbon samples [12], it can be claimed that CBK describes all of the significant features
of the Sandia char combustion database.
In order to prepare this model for integration into a comprehensive CFD code, careful
sensitivity analyses were carried out in an effort to reduce the computational demands
of the thermal annealing submodel. Numerous test cases were examined to find the
optimal limits of integration and step size, considering both accuracy and speed.
In order to facilitate an evaluation of the usefulness of CBK in a reacting, multi-phase,
CFD code, CBK and GLACIER have been modified to accept/pro vide appropriate data
for transferal between the two codes. Fuel specific quantities (GLACIER inputs) are
shared and ensemble averaged values for oxygen concentration, gas temperature, and
radiative environment temperature for statistical particle "clouds" (GLACIER
calculation results) are input to CBK. CBK has been modified to compile results for
each particle cloud and to determine and output total unburned carbon in flyash.
Boilers Simulated
The lower furnace of two boilers, Hammond Unit 4 and Keystone Unit 2, have been
modeled.
Keystone Unit 2 is a Combustion Engineering, Inc. Combined Circulation Balanced
Draft Divided Furnace Steam Generator rated at 850 MW consisting of two identical
tangential fired furnaces. Each of four burner stacks consists of eight levels of coal
burners alternating with air inlets. In 1994, the unit was retrofitted with an ABB C-E
Services, Inc. Low NOx Concentric Firing System (LNCFS) Level III. The LNCFS Level
III system utilizes both Close Coupled Over Fire Air (CCOFA) and Separated Overfire
Air (SOFA). The low NOx retrofit has put Keystone Unit 2 in compliance with Federal
Government NOx emission regulations, but at the expense of increases in both water
wall tube corrosion and LOI. Corrosion rates of up to 50-70 mil/year have been
observed in certain regions of the front and back waterwalls at CCOFA and SOFA inlet
port elevations. One furnace (Furnace "B") was modeled up to the nose with a 660,000
node Computational Fluid Dynamics (CFD) grid. ABB measured coal flow rates are
used in the model with sufficient combustion air for an excess O2 level of 3.5% at the
superheater entrance. Combustion air is distributed according to inlet areas. Water
wall fluid temperatures are used with a thermal resistance model to control wall heat
transfer.
-------
Georgia Power's Hammond Unit 4 is a 500 MWe opposed wall-fired facility with 24
burners. Prior to low-NOx retrofit, the unit included Foster-Wheeler's Intervane
burners. After retrofitting, the unit included Foster Wheeler's Controlled-Flow/Split-
Flame burner, advanced overfire air (an independent windbox stream for improved
penetration/control) through 8 directly opposed ports, and 4 underfire air ports. This
furnace was simulated with a symmetry plane along the vertical center from front to
rear with more than 480,000 computational nodes. Burner velocity and particle
distribution inputs as well as near burner mesh resolution have been guided by detailed
coal pipe and complete individual burner simulations.
Coals
Table 1 provides details of the coal composition and particle size distribution used in
the simulations.
Table 1. Coal properties
Coal Proximate Analysis
Composition
Fixed Carbon
Volatile Matter
Ash
Moisture
Keystone
46.6 %
35.0%
12.2%
6.2 %
Hammond
52.7 %
33.5 %
9.8 %
4.3 %
Coal Ultimate Analysis
Sj
jecies
C
H
O
N
S
Keystone
69.3 %
4.6 %
4.6 %
1.3%
1.8%
Hammond
72.4 %
4.7 %
5.7 %
1.4%
1.7%
Keystone Particle Size Distribution
Particle Diameter
6 mm
18 mm
31 mm
43 mm
60 mm
85 mm
120mm
175 mm
250mm
Mass Fraction
0.17
0.15
0.11
0.10
0.15
0.10
0.12
0.05
0.05
-------
Water Wall Corrosion
GLACIER does not predict corrosion rates but, if a computer model is to provide
insight into the cause of accelerated water wall corrosion under low NOx firing
conditions and allow an evaluation of potential solutions, it must provide an accurate
picture of conditions close to the wall including:
• Local gas concentrations including carbon monoxide, hydrogen sulfide and sulfur
dioxide.
• Heat flux to the wall, because this can affect deposit formation and slagging.
• Particle deposition rates at the wall.
• Carbon and sulfur content of the particles being deposited at the wall.
• Deposition rates of FeS particles present in the coal stream.
Measured corrosion rates in the Keystone boiler are highest on the front and rear walls,
above the close coupled over fire air, in the region of the separated over fire air, and
lower on the side and center walls. Comparison of the pre- and post-retrofit simulations
indicate that: 1) in the regions of highest corrosion, heat fluxes are higher in the post-
than the pre-retrofit conditions; 2) the regions of highest corrosion do not correlate with
highest H2S concentrations close to the wall, but they do correlate to regions with steep
gradients in oxygen concentration both along and perpendicular to the wall, and 3) on
both the front and rear walls the regions of high corrosion correlate closely with the
fraction of unburned carbon in the wall deposit. Figure 1 compares the fraction of
unburned carbon in the wall deposit on the rear wall for the pre- and post-retrofit
simulations. Shown also are measured wastage rates. The simulations suggest that high
corrosion rates in excess of 50 mils/year occur in regions of the water wall were there
are high gradients in gas concentrations. In these simulations the gas cell closest to the
wall (whose scale is on the order of 2.5 inches) may be oxidizing but one or two cells
into the furnace there may be very high concentrations of tbS. This suggests that
intermittency (variation from oxidizing to reducing conditions) may be a key factor
accelerating corrosion rates. The simulations also suggest that high concentrations of
unburned carbon in the wall deposit accelerate corrosion rates perhaps because of the
creation of highly localized reducing conditions but perhaps they are also indicative of
the presence of FeS deposition.
Figure 2 compares the burnout history for one particle cloud starting from the same
location in the fuel injector for both the pre- and post-retrofit cases. Figures 2a and b
show two particle sizes entering from the burners on the front wall at level six near the
center wall. The smaller particle does not burnout in the post-retrofit case. Figure 2c
presents the same information but for a particle cloud entering from the front side
nozzle at the same level. This particle cloud burns out very rapidly in both the pre- and
post-retrofit cases.
-------
Carbon Burnout
Simulation of the Hammond furnace required detailed modeling of the coal injector,
and this detailed modeling provided improved inputs and guidance in terms of mesh
resolution. These simulations were performed on geometries with and without the
tangential inlet section. In order to obtain adequate resolution for the anti-roping bars,
the following results from the exit of the tangential entry were mapped onto the inlet of
a more detailed simulation of the convergent section of the coal pipe:
• Tangential velocity at the inlet to the convergent section was determined from the
results of the model of the tangential inlet was used.
• Particle loading computed based on the mass density and velocity distributions
from the results of the model of the tangential inlet.
Each computational node from the previous mesh was taken as the location for a
particle cloud starting location (-700 positions). Figure 3 illustrates the effects of the
coal nozzle on various size particles. The gray surface in the center is the movable
sleeve in the center of the coal pipe. The cross sectional planes are colored to illustrate
variations in particle mass density and the black lines/circles display a few typical
particle cloud trajectories. (The line is the statistical mean position, while the circles
represent the variance about that mean.) The larger particles from every starting
location quickly reach the wall and pass axially through the coal pipe in the valleys
between anti-roping bars. The smaller particles tend to follow the flow and rarely hit
the anti-roping bars. The intermediate size particles, which include the majority of the
total mass, impact the roping bars, but do not remain in the valleys. This tends to more
evenly distribute these particles. Another effect of this design is to segregate the
intermediate and large particles into the ellipses, as opposed to the annulus. Since the
exit area of the annulus and the ellipses are similar, the proportion of smaller particles
exiting the ellipses/annulus are also similar. Another feature of this coal pipe, despite
the presence of the anti-roping bars, is the maldistribution in the angular direction.
The colors in the image represent the log of mass density and show significant
variability in the exit plane.
Accurate simulation of coal devolatilization and char oxidation in a boiler enable one to
estimate carbon burnout levels in a furnace. Simulations of the Hammond unit were
performed to compare the results of conventional and more complex (CBK) char
oxidation models. A summary of the results for two simulations (pre- and post- low
NOx retrofit) with the two approaches to modeling carbon burnout are presented in
Table 2.
-------
Table 2 Burnout at the Nose for Several Simulations.
Pre-retrofit Post-retrofit
GLACIER 96.4% 95.1%
GLACIER withCBK 95.1% 89.3%
As expected the burnout for the post-retrofit simulation is lower irrespective of the char
oxidation model used. The results with CBK however indicate two differences between
the models. First, the burnout is lower with CBK. Second, the change from pre- to post-
retrofit is significantly greater than with the less complex model. Test data from the
actual boiler indicates a fairly large difference in burnout should exist.
Figure 4 presents the burnout for these four calculations as a function of particle size.
The noticeable difference between the pre- and post- retrofit cases is clear for all particle
sizes. However, these data also point out an interesting feature of the size dependence.
In the pre-retrofit case it is clear that much of the unburned carbon results from the
largest size fractions. However, in the post-retrofit case the dependence of burnout on
initial size is much less non-linear, suggesting that removing the largest fraction with
improved size classification may be less effective for a post-retrofit situation.
The importance of the relationship between char reactivity and particle
temperature/oxygen concentration history can be seen by following identical particles
in pre- and post- retrofit simulations. The effect of the low NOx burners and staged air
addition is clear. The oxygen is completely consumed in the near burner region and
consequently the normalized reaction rate drops to zero. The particle has completely
burned in the pre-retrofit case before its post-retrofit equivalent encounters the overfire
air and begins to oxidize. In addition, during this time the post-retrofit particle can be
thermally annealed further reducing its capability to achieve high levels of burnout.
GLACIER tracks statistical representations of particle trajectories or particle clouds that
are uniquely identified by their starting location and particle size. Information is
therefore available to determine the burnout of particles issuing from each individual
burner in the Hammond unit. Figure 5 illustrates these results for each burner on both
•walls of the pre- and post- retrofit cases.
The pre-retrofit case displays the obvious trend considering the effects of residence time
in the boiler, i.e., the upper rows tend to result in more unburned carbon, with no
unburned carbon resulting from the bottom row. Front/rear and inside/outside
-------
differences are small. The post-retrofit simulation however displays more complicated
results:
• The inner columns of burners produce minimal unburned carbon.
• While the bottom row again produces no unburned carbon, the middle row results
in the highest levels.
• The rear wall results in significantly higher unburned carbon levels than the front
wall.
These results suggest that the particle paths and oxygen concentration/temperature
fields are complicated by the presence of the low NOx burners and overfire air.
CONCLUSION
REFERENCES
1, Smoot, L.D. and P.J. Smith, Coal Combustion and Gasification, Plenum Press,
New York, NY, 1985.
2. Smith, P.J. and T.H. Fletcher, "A Study of Two Chemical Reaction Models in
Turbulent Coal Combustion," Combust. Sri. Technol, Vol. 58, p. 59,1988.
3. Smith, P.J., 3-D Turbulent Particle Dispersion Submodel Development
(Quarterly Progress Report #4), Department of Energy, Pittsburgh Energy
Technology Center, Pittsburgh, PA., 1992.
4. Adams, B.R. and P.J. Smith, "Three-dimensional Discrete-ordinates Modeling of
Radiative Transfer in a Geometrically Complex Furnace," Combust. Sci. Technol.,
Vol 88, p. 293,1993.
5. Jain, S., "Three-Dimensional Simulation of Turbulent Particle Dispersion
Applications," PhD Dissertation, Dept. of Chem. & Fuels Eng, Univ. of Utah, 1996.
6. Mitchell, R. E., R. H. Hurt, L. L. Baxter, and D. R. Hardesry, Compilation of Sandia
Coal Char Combustion Data and Kinetic Analyses: Milestone Report" SAND92-8208,
1992.
7. Hurt, R. H. and R. E. Mitchell (1992) "Unified High-Temperature Char Combustion
Kinetics for a Suite of Coals of Various Rank" Twenty-Fourth Symp. (Int.) on
Combustion, The Combustion Institute, Pittsburgh, p. 1243
8. Hurt, R.H., Lunden, M., Brehob, E.G. and Maloney, D.J. (1996) "Statistical Kinetics
for Pulverized Coal Combustion," accepted for publication, 26th International
Symposium on Combustion.
-------
9. Suuberg, E. M. (1991) in Fundamental Issues in Control of Carbon Gasification
Reactivity, p. 269.
10. Davis, K.A. et al. (1995) "Optimized Fuel Injector Design for Maximum In-Furnace
NOx Reduction and Minimum Unburned Carbon," DOE Quarterly Report, Oct-
Dec., Project No. DE-AC22-PC95103.
ll.Beeley, T. et al. (1996). "Transient High-Temperature Thermal Deactivation of
Monomaceral-Rich Coal Chars," 26th International Symposium on Combustion.
12. Hurt, R.H., Davis, K.A., Yang, N.Y.C., and Headley, T.R., and Mitchell, G.D.
"Residual Carbon from Pulverized Coal Fired Boilers 2: Morphology and
Physicochemical Properties, Fuel, 74:91297 (1995).
-------
Fraction of Unburned
Carbon in Deposit
1.0
a) Measured b) Pre-retrofit carbon c) Post-retrofit carbon
corrosion deposition deposition
0.5
Figure 1. Measured back wall corrosion and the fraction of unburned carbon in back
wall deposit.
-------
1-
£0.6
M 0.4-
0.2-
— Pre Retrofit
Post Retrofit
0.5
1 1.5
Residence Time (s)
a) 43 micron coal particle entering from the front center level 6 coal nozzle.
i'o.s-
ojO.6-
go.4H
o.2-
Pre Retrofit
Post Retrofit
0.5 1 1.5
Residence Time (s)
b) 85 micron coal particle entering from the front center level 6 coal nozzle.
Pre Retrofit
—- Post Retrofit
0.5 1 1.5
Residence Time (s)
a) 43 micron coal particle entering from the front side level 6 coal nozzle.
Figure 2. Fractional char remaining as a function of residence time.
-------
*-."•
Figure 3. Simulated burnout as a function of particle size for pre- and post-retrofit
boilers using conventional and CBK char oxidation model
Hammond preretro
(CBK)
Hammond preretro
(GLACIER)
Hammond retro (CBK)
Hammond retro
(GLACIER)
Particle Size (microns)
16.5
26.5
40.4
86.7
122.3
169.3
278.6
Figure 4. Simulated burnout as a function of particle size for pre- and post-
retrofit boilers using conventional and CBK char oxidation model
-------
top row:
mid row:
bot row:
Preretrofit
rear wall front wall
1.00 1.00 1.00 1.00
.96 .99 .99.96
.96 .99 .99 .96
1.00 1.00 1.00 1.0
top row:
mid row:
bot row:
Postretrofit
rear wall front wall
.90)(.99)(.99)(.90
.80 1.00 1.00 .80
.1.00) (1.00) (1.00) (1.00)
.95 .99 .99 .95
.89 1.00 1.00 .89
1.00 1.00 1.00 1.0
Figure 5. Burnout of coal particles from each burner in the pre-and post- retrofit
simulations.
-------
Assess Coal Quality Impacts on NOX and LOI with
EPRI's NOV-LOI Predictor
X
S. Niksa, A. Kornfeld A. Mehta, J. Stallings
SRI International Electric Power Research Institute
L. Muzio, T. Fang W. Gibb
Fossil Energy Research Corp. PowerGen
R. Hurt, J. Sun M. Cloke, E. Lester
Brown University Nottingham University
Abstract
EPRI is developing a software tool that allows utilities to predict coal quality impacts on NOX
emissions and LOI. The model requires only standard coal analysis as inputs (proximate and
ultimate analysis, Hargrove grindability index). Given the coal properties for the baseline coal
and alternate coal, along with baseline boiler performance data (NOX and LOI), the model
predicts changes in NOX emissions and LOI due to a switch to the alternate coal. In addition to
the LOI predictions using the standard coal properties, the software also contains an LOI
prediction based on a gray scale reflectance test of a coal sample.
A beta version of the NOX model has been undergoing evaluation by utility companies for about a
year. The LOI part of the model has recently been incorporated into the software and beta tests
initiated.
The paper describes the software, approaches used to model NOX and LOI, how the software can
be used, and recent results.
Introduction
In response to the 1990 Clean Air Act Amendments, many utilities are implementing low NOX
burners to reduce NOX. At the same time, utilities are switching coals either for SO2 compliance
or for economic reasons. With low NOX combustion systems, coal quality has become a more
important consideration. These combustion systems have, in general, made the combustion
process more sensitive to coal quality in terms of combustion efficiency and ash carbon content
(or loss on ignition, LOI). Similarly, for units that are just meeting NOX compliance, a change in
coal characteristics can potentially impact NOX compliance.
To support utilities making quick decisions on coal purchases, EPRI is developing a NOX-LOI
Predictor. This is a Windows® based tool that has the potential of large cost savings to a utility
-------
by avoiding coals that are inappropriate from either a NOX or LOI basis. The software is intended
as a simple tool to quickly screen the performance of a number of coals relative to the current
coal being burned in a particular boiler.
It should be pointed out that both NOX production and LOI involve many complex processes that
are dependent not only on coal characteristics, but also on the boiler and burner design and
operating parameters. It was not the intent of the NOX-LOI Predictor to predict NOX and LOI
given coal, boiler and operating characteristics. Rather the NOX-LOI Predictor starts with the
known performance of a unit in terms of measured NOX emissions and LOI for the coal currently
being used. Given this known baseline performance, the program then predicts how the
performance will change in terms of NOX and LOI when an alternate coal is burned under the
same operating conditions. Based on the user's past experience with particular furnaces, the
initial estimate from the NOX-LOI Predictor for baseline operating conditions can be adjusted for
varying operating conditions (i.e., O2, load, burner tilts, etc.).
The NOX portion of the model has been under evaluation by selected utilities for about a year.
The LOI part has recently been incorporated and evaluation of the LOI predictions have just
begun. This paper will provide an overview of the methodology used for the NOX and LOI
predictions and present recent results obtained from the model.
Structure of the NOX Prediction
Many previous studies have attempted to relate NOX emissions using conventional coal
properties. While these previous efforts have shown qualitative trends, they have not been able
to make quantitative predictions of coal quality effects. A number of factors contributed to this
inability to develop predictive correlations. First, NOX emissions depend both on operating
parameters and fuel properties, and it can be difficult to separate these two effects. Secondly,
most of the prior efforts have relied on the use of conventional fuel properties, specifically, the
ASTM proximate and ultimate analysis. While the ASTM proximate analysis provides a general
indication of differences in fuel properties, it is widely known that the ASTM proximate analysis
does not accurately describe the processes coal particles undergo in large flame. Under the high
heating rate conditions experienced by a pulverized coal particle in a utility furnace, the volatile
yield and nitrogen release with the volatiles can be substantially different than indicated by the
ASTM proximate analysis. Both of these quantities can have a major impact on NOX formation
processes.
The current structure of the NOX prediction portion of the software is still correlation based.
However, at the inception of this project it was felt that, to improve the predictive capability of
the coal property effects on NOX, it would be necessary to use coal properties for high heating
rate conditions. Initially, there was thought given to using test results from a high heating rate
device such as a drop tube furnace1 or an electrically heated grid2 as an input to the NOX model.
While this was technically feasible, having to perform a nonconventional test on a coal was not
attractive from the standpoint of quickly screening a large number of potential coals.
-------
To eliminate the need for a special coal test, it was decided to incorporate a coal devolatilization
model into the NOX Predictor. There have been three phenomenological coal network models
that have been recently under development3'4'5 These three models represent devolatilization as a
depolymerization process that disintegrates the coal's macromolecular structure into smaller
volatile fragments with subsequent reintegration of larger intermediates into char. For the current
application, the FLASHCHAIN model developed by Niksa3 was chosen for incorporation into
the NOX Predictor. FLASHCHAIN was chosen for two primary reasons. First, it was at the
highest stage of development at the start of the project with the capability of accurately predicting
not only volatile yield, but also volatile nitrogen evolution. Secondly, FLASHCHAIN requires
that no special tests be conducted on the coal; the only coal-specific information needed by
FLASHCHAIN is the proximate and ultimate analysis of the coal. FLASHCHAIN also resolves
the partitioning among char-N, tar-N and HCN.
Prior to proceeding with FLASHCHAIN for this application, it was subjected to blind
evaluations against the total weight loss and char nitrogen contents from nine (9) coals that were
measured in a drop tube furnace1. In these cases only the coal properties and thermal histories
were available in advance to enable the FLASHCHAIN simulations, yet the predictions matched
the measured values within experimental uncertainties in all but one case.
In the calculation process, FLASHCHAIN has one adjustable parameter. This parameter is used
to match the FLASHCHAIN prediction with the ASTM volatiles under a low heating rate
condition. With this parameter fixed, FLASHCHAIN then calculates devolatilization for
pulverized coal flame conditions. Full FLASHCHAIN simulations of the coals' behavior under
flame conditions require only a few seconds on modern personal computers.
As indicated above, the NOX Predictor portion of the program uses a correlation-based
methodology. Figure 1 shows the structure of the NOX-LOI predictor. The user enters the
proximate and ultimate analyses of the currently fired coal and the coals to be screened. In
addition, the NOX levels with the baseline coal are also entered along with basic information on
the unit (i.e., firing configuration, low-NOx burners, overfire air, pre- or post-NSPS, etc.). After
the coal and boiler information are input, the program processes each coal through the
FLASHCHAIN model which calculates volatile gas yield, volatile tar yield, tar nitrogen yield and
HCN production.
The FLASHCHAIN results are then used along with a series of correlations to predict the
expected change in NOX emissions for the alternate coal. The correlations were developed using
numerous sets of pilot scale data available in the literature1'7 The correlations relate the
measured pilot-scale NOX emissions to the FLASHCHAIN parameters (volatile yield, tar
nitrogen, HCN and char nitrogen). A challenge was to then be able to relate the fuel property
effects correlated at pilot scale to a specific full scale boiler. Since all utility boilers currently
monitor NOX, the measured baseline NOX was used to relate the pilot scale correlations to a
specific full scale unit. This was done by defining a reference NOX level that the baseline coal
would produce at the pilot scale. In essence, the baseline NOX level is used to calibrate the
correlation.
-------
Alternate Coal
Properties
(Coal B)
prox, ultimate,
LOI Model
Grinding Model
Char Density Model
Char Oxidation
Char Annealing
Ash Inhibition
HHV, HGI
Current Coal
(Coal A)
Alternate Coal
(Coal B)
x
Gray Scale
Measurement
Correlation of Gray
and LOI
I/
Predicted NO.,
for Alternate
CoalB
Predicted LOI
for Alternate
CoalB
Figure 1. Structure of the NO-LOI Predictor
Structure of the LOI Prediction
Currently there are two methodologies incorporated into the software for predicting LOI. One is
phenomenologically based, and the second is based on a petrographic gray scale reflectance
measurement of a coal sample8
Phenomenologically Based LOI Predictor
While the NOX prediction is based on correlation, the LOI prediction model incorporated into the
NOX-LOI Predictor is phenomenologically based. The primary elements of the LOI prediction
model include:
coal grinding submodel
char formation submodel (FLASHCHAIN)
char density submodel
-------
• char oxidation submodel (CBK)
single-film char oxidation kinetics
thermal annealing
ash inhibition
Figure 1 shows the structure of the LOI Predictor. The grinding model estimates the coal grind
size for the alternate coal given the Hargrove Grindability Index (HGI) of the base coal and
alternate coal, and the grind size of the base coal (% <200 mesh and % >50 mesh). The model
uses this data to calculate a Rosin-Rammler slope and estimate of the top size for the base coal
(d-top size > d of 99.99% of mass). The grinding model then calculates the pulverizer throughput
for the alternate coal using the heating values of the coals. A new fineness, in terms of the
percent through 200 mesh is then calculated using full scale pulverizer correlations. These
correlations are in the form of percent passing 200 mesh as a function of HGI and throughput.
The alternate coal size distribution is then calculated using the new 200 mesh fineness and
assuming a constant top size for the pulverizer. If the user prefers, the LOI prediction can also be
performed by holding the coal size distribution constant.
The next step in the calculation is coal devolatilization and char formation. This part of the
calculation is done using the FLASHCHAIN model discussed above. FLASHCHAIN is used to
calculate the fraction of the coal that ends up as char. A rank dependant correlation is used to
determine the initial char density.
Char oxidation is calculated using a single-film model of char oxidation kinetics that also
includes provisions for thermal annealing of the char throughout the oxidation process (CBK)
and ash inhibition.
A key feature of the char oxidation model is its ability to describe long residence time char
oxidation; this is important for predicting LOI since we are dealing with the last fraction of the
carbon left to bum. A single global char oxidation model will basically predict zero LOI;
whereas the CBK model, which includes char annealing and ash inhibition, can simulate the late
stages of char oxidation9. The char oxidation is assumed to occur as it travels in plug flow
through the furnace. In the LOI predictor, the user specifies the oxygen concentration and the
model has an adjustable parameter to account for incomplete mixing in the furnace.
Similar to the NOX prediction, the LOI prediction scales the results relative to the baseline LOI
value. Internal calibration parameters are used to match the LOI from the baseline coal and the
model then uses these parameters to calculate the LOI for the alternate coal.
Reactives Index Based LOI Prediction
Previous attempts to relate combustion behavior to maceral content10'11 and rank12 alone have met
with limited success, except where the range of coals studied has been limited. Maceral content
alone does not take into account either the variation of the vitrinite macerals with rank or the
variation of the reflectance and, therefore, reactivity of inertinites10 Similarly, rank alone does
not account for the variations in maceral content that can occur, especially where the coals are
-------
from different geological origins. The Reactives Index was developed to overcome these
problems and to provide an automated technique which removes the subjectivity of manual
maceral analysis.
The Reactives Index is based on producing a gray-scale histogram of the coal sample, viewed
under a microscope, using image analysis. This idea was first proposed by Riepe and Steller13
and has been further developed at Nottingham University to enable the analysis of the liptinite in
coal14, where there is a close similarity between the gray-scale of the liptinite and the resin used
to mount the coal sample. In essence, the analysis of the sample is carried out in both white light
and blue light to produce a histogram similar to that shown in Figure 2. In Figure 2, the gray
scale histogram of the whole coal is given starting with the synthetic liptinite column on the left
hand side. The thicker line in the figure represents the cumulative curve over the gray-scale
range. A cut-off is made at a gray-scale of 190, and it is assumed that all material with a gray-
scale greater than 190 is "unreactive" in the combustion process. The Reactive Index is defined
as the mass percentage of coal with a gray-scale reflectance value greater than 190 on a
standardized scale. Note that in this paper "Reactives Index" and "% 190 unreactives" are used
interchangeably.
Reactivity
Threshold
[190]
CM CO O •*
2 2 CJ CNJ
Grayscale
Figure 2. A Typical Reactivity (Gray Scale Histogram) Assessment Profile
The image-analysis technique used is simple and rapid to apply, and also removes subjectivity
from the analysis. Including block preparation, polishing and analysis, a result can be obtained in
about one hour. The histogram gray-scale is standardized using a light source of known
intensity, any variations in the gray-scale of the vitrinite due to rank and variations in inertinite
due to reflectance are shown in the results. Thus, a high-rank vitrinite will give a peak at a
higher gray-scale inferring a lower reactivity. Similarly, some inertinites are shown to be more
-------
reactive as their position in the histogram is at a lower gray-scale. In this way, it is possible to
infer that some inertinite material is potentially more reactive than some vitrinite material.
A relationship between the Reactives Index and combustibles remaining were established for 17
coals burned in PowerGen's 1 MW Combustion Test Facility (CTF). These coals covered a wide
range of geological origin, rank and maceral composition. This correlation is shown in Figure 3.
Figure 3 shows the relationship between the Reactives Index and combustibles remaining, which
is the percentage of dry ash-free coal which is not burned (determined using an ash tracer
technique). The combustibles remaining can be related to the more normally measured loss on
ignition (LOI):
Combustibles remaining % = 100
A = Ash content of coal, % dry
LOI = Loss-on-ignition, % dry
A x LOI
(100-A) (100-LOI)
(1)
Pocahontas
Gusare
1 1.5 2
Combustibles Remaining, %
Figure 3. Relationship between the Reactives Index and Combustibles Remaining8
The correlation between the parameters shown in Figure 3 exhibits two clear outliers. The first is
Gusare coal which performed much worse than would be predicted from the 190 Unreactives.
This was partly due to the large quantity of oversize material, 4.1% >212mm, which was found
in the pulverized coal from this coal, compared with a norm of less than 1% >212mm for most of
the other coals. The other outlier was Pocahontas which has a much higher rank than all of the
other coals considered. It has been reported mat this low volatile content coal can be burned in a
-------
station designed for normal high volatile content bituminous coal resulting in only a small
increase in "carbon loss"15 This finding is in line with the CTF test results obtained in this study
which indicate that higher rank vitrinite is more reactive than what the Reactives Index would
predict. Pocahontas is, however, not a coal that would normally be burned in a power station.
Having established a relationship between the Reactives Index and combustibles remaining, the
Reactives Index can then be used to calculate the effect of a coal change from a baseline to
alternate coal on LOI. As with the NOX prediction and phenomenological LOI prediction, the
Reactives Index technique also uses the baseline coals Reactives Index and LOI to essentially
calibrate the algorithm. For example, suppose that burning a baseline coal at a specific power
plant under standard conditions gives an LOI of 8%; using the known ash content, say 10% dry,
this is equivalent to a combustibles loss of 0.936% Assuming the baseline coal has a Reactives
Index of 7%, a plant-specific "calibration line'^ relating Reactives Index to combustibles loss can
be drawn through the origin and the point for the baseline coal. Then for any alternative
"unknown" coal supply, given the Reactives Index, the combustibles loss can be read of this line
and the equivalent LOI value calculated from the dry ash content of the alternative coal.
It should also be noted that the Reactives Index technique was developed to assess the bumout
characteristics of bituminous coals and is not easy to apply to low rank coals. Low rank coals
have inertinite type structures that have the same coloration as vitrinite. Low rank coals also
have low reflectance, making it difficult to separate the main vitrinite peak from the resin peak.
Preliminary Results
Beta tests for the NOX-LOI Predictor are currently underway. The NOX Predictor part of the
model has undergone tests for approximately one year, while the LOI prediction has recently
been incorporated into the program and testing just begun. The following subsections will
present preliminary results from both the NOX and LOI prediction portions of the program.
NOX Predictions
Before presenting the NOX prediction results, it should be reiterated that when validating the
predictions, NOX data is needed from the baseline and alternate coal for the same firing
conditions. Depending on the boiler type, this would include:
• load
• o,
• mills in service
• level of overfire air
• burner tilts
• burner damper settings
If test data is not available under identical conditions, then it is recommended that the results be
adjusted to make the comparison on as common a basis as feasible. For instance, a difference in
the flue gas O2 concentration of 1% can change NOX on a tangentially-fired boiler and wall-fired
boiler by 30-50 ppm and 60-90 ppm, respectively. Likewise, a change in burner tilt can impact
-------
NOX on the order of a 1-2 ppm per degree of tilt change. In many cases, small changes in
operating parameters may be equal to or grater than the fuel parameter effects.
To illustrate the use of the NOX Predictor, the program was used to predict the expected changes
in NOX during a test bum of a Power River Basin (PRB) coal at Public Service Company of
Colorado's Arapahoe Unit 4. This boiler has been the site of a DOE Clean Coal Technology HI
program, and has undergone extensive testing with the baseline coal. A two week PRB test burn
was conducted at the unit during which no changes were made to the combustion hardware or
pulverizers. Arapahoe Unit 4 is a 100 MW roof-fired boiler that was retrofit with Babcock &
Wilcox XCL burners and overfire air as part of the DOE clean coal project. The baseline coal is
a Colorado bituminous coal. Table 1 shows the NOX predictions along with the actual NOX data
obtained during the PRB test burn.
Table 1
NOX Predictions for PS Co Arapahoe Unit 4
Base Coal PRB Coal
Load
%MCR
60%
80%
100%
Actual NOX
ppmc
300
260
292
Actual NOX
ppmc
205
185
219
Predicted NOX
ppmc
206
181
201
As can be seen in Table 1, the predicted NOX levels for the PRB coal are in good agreement with
the measured values, even though the substantially different composition of the PRB and
baseline coals are responsible for large differences in the respective NOX levels.
The NOX prediction portion of the model was also used to predict the NOX at Dairyland Power's
Genoa 3 Unit. This T-fired unit burns a blend of bituminous and PRB coal. The results of these
predictions are shown in Figure 4. In this figure, the filled symbols are CEM NOX data for
varying PRB blends and a load of nominally 70% MCR. The dotted line in Figure 4 is a least
square fit to the CEM data. The open symbols with the solid line are the NOX predictions. For
the NOX predictions, the 40% PRB blend was taken as the baseline coal in the NOX Predictor
program. The NOX predictions for this unit are in good agreement with the measurements. It
should be noted that for this case, little operating data was available: load, CEM NOX, CEM CO2
and the blend ratio. Some of the variation in the data points plotted in Figure 4 could be due to
variations in other operating parameters (i.e., burner tilts, mill patterns, etc).
As part of the Beta tests, PowerGen has evaluated the NOX predictions for T-fired units using
data from their Fiddlers Ferry Power Station. Fiddlers Ferry comprises four 500 MW (electrical)
tangentially fired boilers fitted with an ICL low-NOx concentric firing system. The coal bums
-------
0.3
20 40 60
Coal Blend, % PRB Coal
Figure 4. NOX Predictions for a PRB Coal Blend
were carried out on Units 2 and 4. These units were entered into the NO, Predictor as separate
boilers. Six blends of three coals (UK coal, US coal and Colombian coal) were tested. One coal
blend from each unit was used as a baseline to predict NOX emissions from the other two blends
burned on that unit. NOX data are only available at a few excess oxygen levels which are not
identical for the different coals. Values for NOX emissions from the six coal blends at the same
excess oxygen level are required to compare the measured and predicted expected values. To
accommodate this, values for NOX at 6% and 7% excess oxygen have been extrapolated from a
graph of NOX versus excess O2 levels. The test data and predicted NOX levels are plotted in
Figure 5. The predicted and measured NOX are in good agreement.
The NOX predictions, shown in Figure 5, illustrate another feature of the NOX Predictor. The
NOX-LOI Predictor predicts the NOX for the alternate coal for the same operating conditions as
the baseline coal. However, for numerous reasons, it may not be possible to operate the boiler at
the same set of boiler parameters as for the base coal. For instance, it may be necessary to
operate at a higher excess O2 level. The sensitivity of the alternate coal NO to O2 can be
predicted if the NO versus O2 behavior is known for the base coal. This is illustrated in Figure 5
where PowerGen used the NOX emissions at two O2 levels to predict the NOX versus O2 curve for
the alternate coal.
Figure 6 summarizes the predictions that have been made for both T-fired and wall-fired units
over a range of NOX levels from 200 to 900 ppm. The predicted capability of the correlations is
quite good.
Since the NOX predictions are correlation based, improved accuracy is anticipated as the Beta
testing proceeds and more data becomes available to refine the correlations.
-------
375
350
^
•o
CM
0 300
8?
CD
-275
TJ
1.250
Q.
z 225
200
175
U K/ U S Coal Blend
o
*
o o
° ° n •-""
CJ f . • U
i--'"
n
) 5.5 6 6.5 7
O2, % dry (stack
(a)
o
a Blend:0:100
o Blend:25:75
• • - Baseline (0'100)
» Predicted (25:75)
7.5 B
)
a
5
375
350
•g?25
"^
(N
O300
5?
CD
-275
T3
E
Q_
§225
200
175
375
350
1525
-o
IN
°300
CD,
TJ
r§50
CL
§225
200
175
4
U K/Colombian Coal Blend
..{
.-•&
|- '
0 0
• n
a
o
5 5 5.5 6 6.5
02, % dry (stack;
o
I
D Blend: 0:100
o Blend:25:75
f... Baseline(O'IOO)
« Predicted (25:75)
7.5 8 8
5
U K/ U S Coal Blend
° n .- ''
i--""
D
j 5.5 6 6.5 7
O2, % dry (stack)
0 Blend:0:100
o Blend:50:50
« Predicted (50:50)
7.5 8
8.5
(b)
350
CSOO
6?
CO
j>*
"D
feso
Q.
d
^25
2001
175
U K/Colomblan Coal Blend
t. -P
o n--
o
o .-•
*
D
5.5 6 6.5 7
O.,, % dry (stack)
o
D
n Blend: 0:100
0 Blend:50:50
« Predicted (50:50)
7.5 8 8.5
(c) " (d)
Figure 5. Comparison of Predicted and Measured NOX at PowerGen's Fiddlers Ferry Power Station (500 MW T-fired)
-------
1000
400 600
Actual NOx, ppm
1000
Figure 6. Summary of Tangentially-fired NOX Predictions
LOI Predictions
The LOI predictions have only recently been added to the program and only preliminary
comparisons have been made. As discussed above, the program contains two methods to predict
LOI; a model and a petrographic (gray-scale Reactives Index) coal analysis. Predictions to date
are only based on the phenomenological model.
The results to date from the phenomenological LOI prediction model are plotted in Figure 7. The
program gives the user two choices in making the predictions. The user can either choose to
have the alternate coal's particle size be the same as the baseline coal ("constant coal size"), or
allow the program to modify the coal particle size based on the Hargrove Grindability Index
(HGI) ("estimated coal size"). These two options were included to cover two scenarios. The
constant coal size assumes that the utility would adjust the mill parameters for the alternate coal
to obtain the same grind as with the baseline coal. The estimated particle size basically assumes
that no changes are made to the pulverizers and a new particle size is estimated based on the
change in HGI and throughput between the baseline and alternate coal. As can be seen in
Figure 7, the initial LOI prediction results are encouraging. In all but a couple of the cases, the
predicted trends are the same as the actual measurements and the magnitude of the changes are
in reasonable agreement. Better precision is anticipated as the database expands.
-------
Open: Constant Coal Size
Closed: Estimated Coal Size
10 15
Measured LOI, %
Figure 7. Preliminary Results from the LOI Predictor
Conclusions
A Windows® based software program has been developed to quickly screen coal quality effects
on NOX and LOI. The NOX predictions have been validated for T-fired units. The wall-fired
correlations have recently been revised and are undergoing further testing. The LOI portion of
the model has just been included and initial testing begun. The LOI predictions utilized two
approaches: (1) a char oxidation based model and (2) a petrographic coal test. The initial LOI
test results with the char oxidation based model are also very encouraging.
References
1. A. Jones, et al., An Integrated Full, Pilot, and Laboratory Scale Study of the Effect of Coal
Quality NOX and Unhurried Carbon Formation, EPRJ/EPA 1995 Joint Symposium on
Stationary Combustion NOX Control, Book 3, May 1995.
2. J. R. Gibbins, et al., Implications of nitrogen release from coals at elevated temperatures for
NO'; formation during PF combustion, Coal Science, Elsevier Science, 1995.
3. S. Niksa, Assess Coal Quality Impacts on Your Personal Computer, 1996 International
AFRC Symposium, Baltimore, Maryland, 1996.
-------
4. P. R. Solomon, et al, Fuel. 72:469, 1993.
5. Fletcher, et al.. Energy Fuels. 6:14, 1992.
6. W. C. and TomitaXu, Fuel. (66(5):632, 1987).
7. L. S. Monroe, et al.. Comparison of Pilot-Scale Furnace Experiments and Predictions to
Full-Scale Boiler Performance of Compliance Coals, EPRI/EPA 1995 Joint Symposium on
Stationary Combustion NOX Control, Book 3, May 1995.
8. W. H. Gibb, et al., The Application of Image Analysis Techniques for the Prediction of
Carbon Burnout in PF-Fired Boilers, Conference on the Effects of Coal Quality on Power
Plants, Kansas City, Missouri, May 20-22, 1997.
9. Niksa, S., et al., Assess Coal Quality Impacts on NOX and LOI with EPPJ's NOX-LOI
Predictor, Conference on the Effects of Coal Quality on Power Plants, Kansas City,
Missouri, May 20-22, 1997.
10. M. Cloke and E. Lester, Characterization of Coals for Combustion Using Petrographic
Analysis: A Review, Fuel, Volume 73(3), p. 315, 1983.
11. G. K. Lee and H. Whaley, Modifications of Combustion and Flyash Characteristics by
Blending, Journal of the Institute of Energy, p. 190, December 1983.
12. D. Kleesattel, S. A. Benson, M. L. Jones and D. P. McCollor, A Petrographic Examination of
Chars Produced by the Rapid Pyrolysis of Low Rank Coals, Extended Abstracts of the Joint
Conference of the Australia/New Zealand/Japan Combustion Institute, p. 27, 1987.
13. W. Riepe and M. Steller, Characterization of Coal and Coal Blends by Automated Image
Analysis, Fuel, Volume 63, p. 313, 1984.
14. M. Cloke, E. Lester, M. Allen and N. J. Miles, Automated Maceral Analysis Using
Fluorescence Microscopy and Image Analysis, Fuel. Volume 74, p. 659, 1995.
15. R. F. Afonso, B. Dyas, G. C. Dusatko and R. Glickert, New England Power Brayton Point
Unit 2 - Testing Low Volatile Bituminous Coal in a Tangentially Fired Boiler, presented at
the 3rd International EPRI Conference on the Effects of Coal Quality on Power Plants, San
Diego, California, August 1992.
-------
FIELD EXPERIENCE -- REBURN NO* CONTROL
Blair A. Folsom
Donald A. Engelhardt
Roy Payne
Todd M. Sommer
Energy and Environmental Research Corporation (EER)
18 Mason
Irvine, California 92618
Sandy S. Chang
Dennis T. O'Dea
New York State Electric and Gas
Corporate Drive, Kirkwood Industrial Park
Binghamton, New York 13902
Lloyd E. Riggs
Robert G. Rock
Eastman Kodak Company
Kodak Park Site
Rochester, New York 14652
Abstract
This paper presents the results of five full scale applications of rebum technology to control NOX
emissions. With rebum, a hydrocarbon fuel is injected above the burners to produce a slightly
fuel rich zone where NOX is reduced. Overfire air injection completes the combustion process.
The applications include boilers of wall, tangential, and cyclone firing configurations firing gas
and coal both as the primary fuel and as the reburn fuel with baseline NOX ranging from 0.13 to
1.4 lb/106 Btu. NOX was reduced by 58-77% over this range of applications. Additional
information is presented on new developments in reburn including Advanced Reburn which
integrates reburn with nitrogen agent (ammonia or urea) injection, the use of Orimulsion™ as a
reburn fuel and a reburn application involving multiple reburn fuels (gas, oil and coal).
Introduction
The Clean Air Act Amendment of 1990 (CAAA), established the framework for NOX emission
regulations to mitigate ozone non attainment areas and acid rain. Over the last seven years, EPA
has developed most of the specific NOX regulations authorized by the CAAA. For applications
where low NOX burners can meet the NOX reduction requirements, they are generally the
-------
technology of choice based on cosl. However, now that many of the specific CAAA regulations
have been finalized, it is clear that in many applications low NO, burners alone will not be
capable of meeting the requirements due to unavailability of low NOX burners (such as for
cyclones), performance problems with low NOX burners (such as carbon loss or tube wall
wastage), or insufficient NOX reduction to meet CAAA requirements.
This paper presents recent field experience with rebum, a combustion modification NOX control
technology. Rebum can be applied to virtually any unit firing any fuel. NOX reduction is
typically in the range of 50-70%. Rebum can be integrated with nitrogen agent injection and
other technologies for even greater NOX reduction.
Reburn NOX Control Technology
The concept of NOX reduction via reactions with hydrocarbon fuels has been recognized for some
time'. Over the last seventeen years, HER has developed a considerable rebum data base
consisting of extensive pilot scale tests, a number of full scale applications (discussed in this and
previous papers)2 and a design methodology which can be used to apply reburn and project
performance for a wide range of applications3. Based on this extensive experience, EER offers
reburn as a commercial NOX control product with commercial guarantees.
Basic Reburn Process
Rebum is a NOX control technology whereby NOX is reduced by reaction with hydrocarbon fuel
fragments' The reburn process is illustrated in Figure 1 for a front wall fired boiler. In applying
rebum, no physical changes to the main burners or cyclones are required. The burners are simply
turned down and operated with the lowest excess air commensurate with acceptable lower
furnace performance considering such factors as flame stability, carbon loss, slag tapping and ash
deposition. Rebum fuel is injected above the main combustion zone to produce a slightly fuel
rich rebum zone where most of the NOX reduction occurs. Maximum NOX reduction
performance is typically achieved with the rebum zone operating in the range of 90 percent
theoretical air. Above the reburn zone, overfire air is injected to complete combustion.
Reburn Design Considerations
Due to the substantial design differences among existing boilers and furnaces and NOX control
requirements which vary with local requirements, reburn must be custom designed to match site
specific factors. EER's reburn design methodology utilizes both analytical and physical models
to design the optimum configuration based on site specific factors and to project performance3.
Firing Configuration. Because reburn does not require modifications to the main combustion
system, it can be applied to virtually any combustion system. This paper discusses applications
to boilers of wall, tangential, and cyclone firing configurations. Reburn can also be applied to
industrial furnaces such as glass furnaces4 and steel reheating furnaces. Applying reburn to hot
furnaces with high baseline NOX levels is particularly attractive since both factors speed up the
-------
NO, reduction reactions. This reduces the amount of reburn fuel and the size of the reburn zone
required to achieve a specific NOX goal.
Main and Reburn Fuel Characteristics. Since reburn involves no physical changes to the
main combustion system, it can be applied to furnaces fired with any fuel (coal, oil, gas,
Orimulsion™ etc.). Except for potential costs of the reburn fuel itself, gas is the preferred reburn
fuel. It produces the greatest NOX reduction per unit rebum fuel injected, has no ash or sulfur and
requires no pulverization or atomization. The disadvantage of gas is that its cost generally
exceeds that of other boiler fuels. The cost and availability of gas are the key factors which
encourage consideration of other reburn fuels. For example, on coal fired units, the use of
pulverized coal as the reburn fuel avoids any cost penalty of the reburn fuel over the main fuel.
Other fuels of interest include oil on oil-fired units and Orimulsion™ which will be discussed
further below.
Furnace Volume. There must be sufficient space above the burners or cyclones to install the
reburn components and to produce adequate residence time in the rebum and burnout zones. By
designing the reburn fuel and overfire air injectors for rapid mixing, space requirements are in
the range typically available on full scale utility boilers. EER has designed gas rebum systems
for numerous boilers and has yet to find a commercial system where the residence time was
inadequate. As presented below, EER has achieved NOX reduction as high as 70% in a cyclone
application with effective rebum zone residence time of only 0.25 seconds. While such
applications are feasible, with longer residence times the amount of reburn fuel required to
achieve a specific NOX control goal is reduced and carbon loss is improved, particularly with coal
reburn.
Reburn Fuel Injectors. The reburn fuel injectors should be located close to the upper firing
elevation but leaving enough space above the burners to achieve essentially complete combustion
in the flames from the main burners prior to introduction of the rebum fuel. For maximum NOX
reduction, the rebum fuel should be injected so as to penetrate across the furnace depth and mix
rapidly with the furnace gases. Since the amount of reburn fuel injected is small compared to the
furnace gas flowrate, achieving penetration and rapid mixing is a challenge, especially for larger
furnaces. Penetration and mixing can be enhanced by increasing the momentum of the injected
stream via a carrier gas or by high velocity injection.
In designing the reburn fuel injectors, it is desirable to minimize (and ideally eliminate) any
oxygen introduced with the reburn fuel. This oxygen must be consumed by additional rebum
fuel to achieve the desired rebum zone stoichiometry. For gas reburn, EER's first generation
systems used flue gas as the carrier. The momentum of the flue gas stream provided good
penetration and mixing and the low oxygen content only slightly increased the rebum fuel
requirement. EER's second generation systems utilize the pressure available in the gas pipeline
to produce high velocity rebum jets, totally eliminating additional oxygen and thus minimizing
the reburn fuel requirement. EER has applied this high pressure injection technique to two units
(wall- and tangentially-fired) as discussed below.
-------
For coal as the rebum fuel, a gaseous carrier is required for pneumatic transport. Flue gas is
preferred over air to minimize the reburn fuel quantity as discussed above.
Overfire Air Ports. Most of the primary fuel char oxidation occurs in the oxygen rich primary
combustion zone. The burnout zone completes combustion of the rebum fuel. For gas rebum,
this is primarily CO oxidation. For fuels which contain fixed carbon, such as coal and
Orimulsion™, this includes CO as well as carbon in the flyash.
The overfire air ports must be located to balance the NOX reduction performance of the rebum
zone with the combustion efficiency of the burnout zone. Generally this trade off is optimized
by locating the overfire air ports substantially higher in the furnace than for conventional overfire
air applications but well below the furnace exit.
As with the gas injectors, the overfire air ports need to be designed for rapid and complete
mixing. EER utilizes dual concentric zone overfire air ports with swirl control. This
arrangement allows the injection velocity to be controlled independent of flowrate, a key
advantage where the rebum fuel is varied to control NOX emission level.
Flame Sensing and Controls EER's rebum control approach is to integrate the reburn
system with the normal boiler controls for fully automated operation. Depending on the NOX
control goal, the rebum fuel injection can be fixed or varied in response to boiler operating
conditions and/or NOX emissions. The fuel injection controls include both permissives and trips
to ensure safe operation. Since the gas injection does not produce a visible flame, conventional
scanners are ineffective. Instead, furnace temperature is used as the primary permissive/trip for
rebum system operation. This approach has been fully effective in five rebum applications and
has been reviewed and approved by Factory Mutual and Hartford Steam for those specific
applications. As an active member of the National Fire Protection Association (NFPA), EER has
taken a lead role in proposing modifications to the current safety codes which will provide
guidelines for future applications.
Retrofit Applications
EER has retrofitted reburn to five coal fired boilers as listed in Table 1. The discussion below
describes how the reburn systems were integrated into the boilers. The subsequent section
discusses the reburn performance including both NOX reduction and boiler impacts.
Gas rebum was installed on Illinois Power's Hennepin Unit 1; a tangentially-fired boiler with
three elevations of tilting burners. Figure 2 is a side sectional view of the boiler showing the
location of the gas injectors and overfire air ports. The Hennepin furnace has a relatively large
space between the upper row of burners and the furnace arch. This allowed the gas rebum
system to be designed with a generous rebum zone. The rebum fuel was injected along with flue
gas through comer nozzles near the top of the windbox. The overfire air ports were located on
the furnace walls near the corners below the arch. Single passage overfire air ports were utilized
without swirl. No overfire air booster fan was required.
-------
Gas reburn was installed on City Water, Light and Power's Lakeside Unit 7, a cyclone-fired unit
with two cyclone burners discharging into an open furnace. Figure 3 shows how the gas reburn
components were integrated into the Lakeside furnace. This reburn application was challenging.
The diverging furnace produced a highly stratified flow field with the combustion products
moving in a jet up the rear wall creating a large recirculation zone. The available residence time
for the reburn zone in this high speed flow was limited to 0.25 seconds.
The gas injectors were positioned along the rear wall and side walls just above the furnace stud
line. Although the penetration distance was short, the short residence time dictated rapid mixing;
flue gas was used as a carrier. The overfire air was injected from the real wall in the upper
furnace through single passage ports. To minimize entrainment into the recirculation zone, high
injection velocities were used. The high combustion air pressure in this cyclone unit was
adequate without a booster fan.
Gas rebum was installed on Public Service of Colorado's Cherokee Station; a front wall fired unit
with a partial division wall. It was integrated with low NOX burners which were installed at the
same time. Figure 4 shows how the gas reburn components were integrated into the Cherokee
furnace. In the initial design, the reburn fuel was injected with a flue gas carrier via ports on the
front and rear furnace walls above the top burner row. Single passage overfire air ports were
positioned on the front wall just below the arch. This configuration was subsequently upgraded
to EER's second generation reburn system. The gas injectors were modified to eliminate flue gas
and to use the pressure in the gas pipeline to penetrate and mix the gas across the furnace. The
overfire air ports were replaced with a dual concentric zone design for additional control at low
gas injection rates.
Gas reburn was installed on New York State Electric and Gas1 Greenidge Unit 6; a tangentially
fired unit similar to Hennepin but somewhat larger. Figure 5 shows the location of the gas and
overfire air injectors. This was the second application of EER's second generation rebum
components. Comparison of the Hennepin and Greenidge side sectional drawings shows the
similarity of the rebum design. Figure 6 is an isometric view of the exterior of the boiler
showing the arrangement of the rebum components.
A micronized coal rebum system was installed on Eastman Kodak's Boiler 15; a cyclone-fired
unit. Figure 7 shows the locations of the reburn and overfire air injectors in the furnace and
Figure 8 is an isometric showing the location of the components. In contrast to the Lakeside
furnace, the Kodak cyclone furnaces discharge through screen tubes before entering the main
furnace. The rebum coal was injected into the main furnace just above the screen tubes.
Micronized coal was produced by two Fuller Micromills, high speed impact mills with separate
classifiers. The coal fineness was nominally 85% through 325 mesh. To minimize the oxygen
introduced into the rebum zone, flue gas was used as the carrier gas. A portion of the flue gas
was supplied to the mills and the rest was supplied to the injectors to optimize penetration and
mixing. While the nominal full load rebum fuel requirement could be met with one Micromill,
two Micromills were supplied to enhance fineness under normal operating conditions and to
allow for operation with a mill out for maintenance. To achieve this flexible operation, the
outputs of both mills were combined and then split into the individual pipes supplying each
-------
reburn fuel injector. EER's dual concentric overfirc air ports were utilized without an overfire air
booster fan due to the high secondary air pressure available in this unit.
Performance Summary
The NOX control performance of the gas reburn systems applied to the Hennepin, Lakeside and
Cherokee units has been discussed in detail in a previous paper2 Figures 9 and 10 show the NOX
emissions for the two newest reburn applications, gas reburn at Greenidge and coal rebum at
Kodak, respectively.
NOX emission reduction at the Greenidge unit is required under both Title 1 and Title 4 of the
CAAA. The most stringent requirement is under Title 1 where NYSEG must meet a system-
wide daily NOX emission cap. The baseline NOX emission from the Greenidge tangentially-fired
unit was 0.62 lb/106 Btu. As shown in Figure 9, NOX emissions decreased as the reburn gas
injection rate increased which is typical for all reburn systems. EER's performance guarantee for
this commercial system was 0.30 lb/106 Btu and was achieved at a gas firing rate of about 13%.
NOX emissions decreased further to as low as 0.22 lb/106 Btu at 23% gas which corresponds to
65% reduction from baseline. As discussed below, EER is upgrading the gas rebum system to
advanced gas reburn by integrating an ammonia injection system. This is expected to reduce
NOX to about 0.15 lb/106 Btu while decreasing the gas firing rate, thus improving cost
effectiveness.
As with Greenidge, Kodak's NOX control requirements stem from both Title 1 and Title 4 of the
CAAA with the Title 1 requirements being the more stringent. The baseline NOX emissions from
the Kodak cyclone fired unit were 1.4 lb/106 Btu. The NOX decrease as the coal injection rate
was increased is shown in Figure 10. EER's performance guarantee of 0.6 lb/106 Btu was
achieved at about 13% reburn fuel. Minimum NOX emissions of 0.45 lb/106 Btu have been
measured to date in short term tests. Additional testing is now in progress and will culminate in
a 51-day compliance test by the end of 1997.
Figure 11 compares the NOX control results for the five reburn installations. Two of the
installations, Cherokee and Hennepin were tested with both gas and coal primary fuels. NOX
emissions (lb/106 Btu) are plotted as a function of the NOX reduction (%). In this format, each
reburn application is represented by a line showing the full range of possible NOX control.
Baseline emissions correspond to 0% NOX reduction and for 100% NOX reduction the NOX
emission is 0.00 lb/106 Btu, by definition. The points on the line show the reburn performance
during short term tests and, in the case of three units, one year tests. Each application is labeled
with the site, the firing configuration, and the primary fuel fired through the burners or cyclones.
These data represent a wide range of applications:
• Firing configuration Cyclone, Wall, and Tangential
• Capacity 33 to 158 MW net
• Baseline NOX 0.13 to 1.4 lb/106 Btu
• Primary fuel Coal and gas
-------
• Reburn fuel Coal and gas
All of the short term data are in the range of 58-77% reduction. During the long term tests, NOX
reductions were about 10% less. Close examination of Figure 11 will reveal that the differences
in NOX control performance cannot be related to the variables listed above. The similarity of
NOX reduction for this wide range of applications is a result of tuning the design and operation of
the rebum system to meet site specific factors.
Based on the results of these five applications, rebum generally has minimal impacts on boiler
performance, but the results are site specific. Typical experience is summarized below:
• CO Emissions CO emissions depend primarily on the design of the overfire air
subsystem. EER's overfire air design provides for rapid mixing of the overfire air across
the furnace cross section and has been successful in maintaining CO emissions acceptably
low, typically comparable to baseline and less than 100 ppm.
• Carbon Loss Carbon in the ash may increase 1-2 percentage points for gas rebum due to
the low temperature at the point of final overfire air addition which leads to lower char
oxidation rates. With coal reburn, carbon loss is more of a problem since the rebum coal
is fired under oxygen deficient conditions and much of the char oxidation must occur in
the upper furnace region after addition of overfire air.
• Thermal Effects The redistribution of fuel and air in the furnace with reburn can alter the
furnace heat absorption profile resulting in some changes in thermal performance. In
most cases, design point steam temperatures have been achieved within the capability of
the burner tilt or attemperation systems. Boiler efficiency is typically reduced by 1-2%
due to the hydrogen content of gas for gas reburn and carbon loss for coal reburn.
• Ash Deposition No significant ash deposition problems have occurred in any of the
reburn retrofits. In some cases, eyebrows have formed around the rebum fuel injectors
and/or overfire air ports. EER now offers a self-cleaning port option (patent pending) to
handle any such ash deposition problems. This involves integration of a sootblower with
the fuel injectors and/or overfire air ports as may be required.
• Durability The potential for tube wastage, particularly in the rebum zone, was a
significant concern on early retrofits. No increased tube wastage has been measured on
any of EER's rebum installations several of which have included extensive destructive
and nondestructive tests5.
Costs
The retrofit costs for reburn systems are highly site specific. Preliminary engineering must be
completed prior to an accurate cost estimate. The typical retrofit cost for gas reburn applied to a
medium size (300 MW) boiler is about 10-15 $/kW. This represents a considerable decrease
compared to previous published costs and is a result of experience and the cost savings of EER's
-------
second generation system. This is the total cost including equipment, installation, and owner
costs. Equipment alone is typically about half. The cost increases if flue gas is used to transport
the gas or if an overfire air booster fan is required. Cost decreases (on a $/kW basis) for larger
units. For coal rebum, the typical retrofit costs are about 10 $/kW higher.
Current Activities
Additional Utility Reburn Applications
Work is in progress on six additional reburn retrofits to utility boilers at three sites as follows:
• Tennessee Valley Authority Allen Station These three 330 MW cyclone-fired units are
subject to the new CAAA Title 4 NOX limit of 0.86 lb/106 Btu. HER is providing gas
reburn systems to achieve this NOX level with minimum gas requirement.
• Baltimore Gas and Electric Crane Station These two 200 MW cyclone-fired units are
within the Northeast Ozone Transport Region (NEOTR) and thus must comply with
Title 1 NOX reduction requirements as well as the Title 4 requirements discussed above.
EER is providing gas reburn systems designed to maximize NOX reduction.
• Ladyzhin Station. Ukraine This 300 MW opposed wall-fired wet bottom unit fires a
mixture of high ash bituminous and brown coals. The reburn system is multi-fuel with
capability to handle gas, oil, and pulverized coal. NOX reduction of over 50% is expected
with pulverized coal rebum with greater NOX reduction for oil and gas.
Reburn Technology Development
The development of reburn technology is continuing. The improvements involve integration of
reburn with other technologies and alternate reburn fuels.
• Advanced Rebum The integration of reburn with injection of a nitrogen agent (N-Agent),
such as ammonia or urea, is termed Advanced Rebum. The N-Agent injection can be
integrated in a number of ways depending on the boiler configuration including injection
into the reburn zone, with the overfire air, and downstream of the overfire air. EER
developed this technology in the 1980s through pilot scale and is currently conducting the
first full scale demonstration at NYSEG's Greenidge Station. To date, NOX has been
reduced to about 0.2 lb/106 Btu firing about 10% gas. Testing is still in progress with the
goal of reaching 0.15 lb/106 Btu. Development is continuing as part of a DOE project
discussed in another paper at this Conference6
• Orimulsion™ Reburn Orimulsion™ is an emulsified bitumen fuel produced in
Venezuela. It has firing characteristics similar to fuel oil. EER has tested Orimulsion™
as a rebum fuel and found its NOX reduction performance to be superior to pulverized
coal and nearly as good as natural gas. In Orimulsion™ reburn, the Orimulsion™ is
atomized similar to fuel oil and injected along with a carrier gas to promote mixing and
-------
dispersion. HER is currently conducting a full-scale demonstration of Orimulsion™
rebum on a US utility boiler.
Conclusions
In conclusion, reburn has been applied successfully to five boilers covering the range of 33 to
158 MW net capacity with tangential, wall, and cyclone firing configurations operating with coal
and gas both as the primary fuels and as the reburn fuels. NOX emissions reductions of 58-77%
have been achieved. No significant operational or durability problems have been encountered.
These results demonstrate the potential for reburn to meet a wide range of CAAA requirements.
Work is in progress on 1,700 MW of additional reburn retrofits.
Acknowledgments
The authors wish to acknowledge the support of the five utility and industrial plants where
reburn systems were installed: City Water, Light and Power, Eastman Kodak, Illinois Power,
New York State Electric and Gas, and Public Service of Colorado. The support of the operating
staff at these plants was outstanding. In addition, the following organizations provided funding
support for work discussed in this paper: US Department of Energy (Agreement Nos. DE-FC22-
87PC79796 and DE-FC22-90PC90547), Gas Research Institute (Contract Nos. 5087-254-149,
5090-254-1994 and 5096-290-3652), State of Illinois Department of Energy and Natural
Resources, Electric Power Research Institute (Agreement No. RP2916-23), and Colorado
Interstate Gas.
References
1. J. O. L. Wendt et. al., "Reduction of Sulfur Trioxide and Nitrogen Oxides by Secondary
Fuel Injection," Fourteenth Symposium (International) on Combustion, pp. 897-904,
(1973).
2. B. A. Folsom, et. al., "Three Gas Reburning Field Evaluations: Final Results and Long
Term Performance," presented at the EPRI/EPA 1995 Joint Symposium on Stationary
Combustion NOX Control, Kansas City, Missouri (May 16-19, 1995).
3. D. K. Moyeda, et. al., "Experimental/Modeling Studies of the Use of Coal-Based
Reburning Fuels for NO, Control," presented at the Twelfth International Pittsburgh Coal
Conference, Pittsburgh, Pennsylvania (September 11-15, 1995).
4. R. Koppang, et. al., "Glass Furnace NOX Control with Gas Rebum," presented at the 56th
Conference on Glass Problems, Urbana-Champaign, Illinois (October 24-25, 1995).
5. B. A. Folsom, et. al., "Gas Reburning for High Efficiency NOX Control, Boiler Durability
Assessment," Air & Waste Management Association 89lh Annual Meeting, Nashville,
Tennessee (June 23-28, 1996).
6. V. Zamansky, et. al., "Combined Reburning/N-Agent Injection Systems for Over 90%
NOX Control," presented at EPRI-DOE-EPA Combined Utility Air Pollutant Control
Symposium, Washington DC (August 25-29, 1997).
-------
Table 1. Full Scale Reburn Application Summary Table
Unit No.
Firing Configuration
Boiler Manufacturer
Full Load, MW(net)
Burners
Burner Type
Primary Fuel
Coal
% Sulfer in Coal
Rebum Fuel
Illinois Power,
Hennepin Station
1
Tangential
CE
71
12 (corners)
Tilting
Coal
Illinois Bit
2.8
Gas
City Water Light
& Power,
Lakeside
7
Cyclone
B&W
33
2
Cyclone
Coal
Illinois Bit
3.6
Gas
Public Service of
Colorado,
Cherokee
3
Front
B&W
158
16
Low-NOx
Coal
Colorado Bit
0.4
Gas
New York State
Electric & Gas,
Greenidge
6
Tangential
CE
104
16 (corners)
Tilting
Coal
Pittsburgh Seam
1.8
Gas
Kodak Park
15
Cyclone
B&W
40
2
Cyclone
Coal
Federal
2.25
Coal
Zone
Burnout
Reburning
Main
Combustion
Conditions
Normal
Excess Air
Slightly
Fuel Rich
Low
Excess Air
NOV Reduction
No
Change
HXCY
Reactions
Reduced Load
Reduced Excess Air
Figure 1. Typical Gas Reburning Installation on a Wall Fired Boiler
-------
Overlire Air Injectors
IV /
Reburn Injectors
II \ /
Main Burners —
Figure 2. Hennepin: Side Sectional View Showing Gas Reburning Components
Figure 3. Lakeside: Side Sectional View Showing Gas Reburning Components
-------
•! n'i
v^iT:^"^:1 * ? L-^' t^J—
Rebum Injectors
Main Burners
Figure 4. Cherokee: Side Sectional View Showing Gas Reburning Components
a
H
Reburn Injectors
Main Burners
Figure 5. Greenidge: Side Sectional View Showing Gas Reburning Components
-------
GR OVERFIRE
AIR DUCT
GR OVERFIRE AIR
INJECTORS
(4 CORNERS)
GAS INJECTORS
(4 CORNERS)
Figure 6. Greenidge: Isometric View of Gas Reburning Installation
(— Reburn Injectors")—. j 1
MIR H
Figure 7. Kodak: Side Sectional View Showing Micronized Coal Reburning Components
-------
Overfire Air
Injector
Overfire Air
Supply
Cyclone
Injection
Air/FGR
Coal Reburn
Injector
Figure 8. Kodak: Isometric View of Micronized Coal Reburning Installation
CQ
W
o'
z
0.8
0.7
0.6
0.5
0.4
0.3
0.2
0.1
0
Pre-Retrofit Baseline
Gas Reburn Data
NOy Guarantee | - -«
0.30lb/IO°Ttu
10 15
Gas Heat Input (% of Total)
20
25
Figure 9. Greenidgc: Gas Reburning NO Data
-------
5 10 15
Reburning Fuel Heat Input (% of total)
Figure 10. Kodak: Micronized Coal Reburning NO Data
NOX
(lb/106Btu)
NOX Reduction (%)
Figure 11. Full Scale Reburning NOX Control Overview
-------
EPRI-DOE-EPA Combined Utility Air Pollutant Control Symposium
The "MEGA" Symposium
Washington, DC, August 25-29,1997
COMMERCIAL DEMONSTRATION OF METHANE de-NOX®
REBURN TECHNOLOGY ON A COAL-FIRED STOKER BOILER
losif Rabovitser, Institute of Gas Technology,
Isaac Chan, Gas Research Institute,
David Hall, Cogentrix of Richmond,
Tim Loviska, Detroit Stoker Company
Abstract
The Institute of Gas Technology (IGT), with support from the gas industry has developed
die patented (1,2) METHANE de-NOX® reburning process for stokers to reduce NOX
emissions to levels set by current EPA regulations. In contrast to conventional reburning,
where the reburn fuel is injected above the combustion zone to create a fuel-rich rebum
zone, with METHANE de-NOA , natural gas is injected directly into the combustion
zone above the grate; this results in a reduction of NOX formed in the coal bed and also
limits its formation through decomposition of NOX precursors to form molecular nitrogen
rather than nitrogen oxides.
Earlier the METHANE de-NOX® process was field tested at the Olmsted County Waste-
to-Energy facility in Rochester, Minnesota, and at two incineration plants in Japan.
Compared to baseline levels, about 60% NOX reduction and an increase in boiler
efficiency were achieved.
Since July 1996, IGT, Detroit Stoker Company, and Cogentrix have successfully
demonstrated the METHANE de-NOX® technology in long-term operation on a 360
MMBtu/h coal-fired stoker boiler at a 240 MW cogeneration plant in Richmond,
Virginia. Baseline and retrofit tests, CFD modeling, and long-term testing and
demonstration were conducted. NOX reduction of 60% using 8% natural gas heat input
was achieved over the regular boiler load range of 40 to 100%. The boiler is in
compliance for both NOX and CO regulations without the use of selective noncatalitic
reduction (SNCR) which was the NOX control technology for this boiler. Compared to
SNCR the rebum system provides: 1) the same or better NOX reduction; 2) an increase in
boiler efficiency of about 2%; 3) C02 reduction; 4) elimination of ammonia slip; 5)
improved boiler start-up, load change, maintenance and operation; 6) annualized plant
cost benefit of about $1.5 to 2.5 million depending on plant capacity factor.
-------
Background
Thousands of stoker boilers in the United States firing coal, municipal solid waste
(MSW), biomass, and refuse derived fuel (RDF) have gaseous and participate
emissions, erosion, slagging, and fouling. Environmental regulations are becoming
more restrictive (1990 Clean Air Act Amendments) and ultimately will require
industries to retrofit many existing solid fuel-burning operations, such as stoker boilers,
with technologies that will allow them to meet new emission requirements for NOX,
CO, total hydrocarbons (THC), sulfur dioxide (SO2), particulate, and other pollutants.
Cogentrix, an independent power producer headquartered in Charlotte, North Carolina,
owns and operates 10 cogeneration facilities in North Carolina, Virginia, and
Pennsylvania. The largest and newest facility is located hi Richmond, Virginia. It is a
240 gross MW coal-fired facility serving Virginia Power and a local fiber and chemical
plant. The Richmond plant contains eight spreader stoker coal fired boilers, rated at
285,000 Ib/h each. The boilers are ABB VU40 type, burning low sulfur coal from
Eastern Kentucky. Each of the eight boilers at the Richmond plant was originally
equipped with a urea based SNCR system (NOxOUT). Since initial plant start-up in the
Spring of 1992 there have been problems related to the operation of the SNCR system.
Each boiler is also equipped with a spray-dryer flue-gas desulfurization (FGD) system
with a pulse-jet fabric filter downstream for particulate removal.
The plant is a fully dispatchable facility and operates through automatic generation
control (AGC) from the Virginia Power operating center. AGC allows the plant electric
output to be controlled remotely by the Virginia power system operator. Boiler loads are
rarely constant, as plant electric output is varied to match system demand. At low system
demand levels, electrical production ceases; however, steam is still required by the fiber
and chemical plant. During these periods, the boilers are operated without the benefit of
the regenerative feedwater heating cycle used when producing electricity. The challenges
in applying downstream NOx control to a unit that varies widely in load has been
previously published (3,4).
As indicated, a urea-based SNCR system is used for NOx control. Because the SNCR
reaction is limited to a temperature window of about 1600 to 1950 F, multiple levels of
injectors are required for complete furnace coverage. Initially, water dilution of the urea
was controlled by a relief valve on the discharge of the water dilution pump. Water and
urea were then pumped to the boiler local-zone control panels, where steam was used for
atomization, and the mixture injected into the furnace. Urea solution is metered to
maintain NOx emissions, as measured by continuous emissions monitors (CEM), at a
target value of 0.29 Ib/MMBtu. No instrumentation is provided to monitor ammonia slip
in the boiler.
The SNCR is put into service at about 30% load (88,500 Ib/h steam). At this load, only
one level of injectors is in service. The system requires manual rotation of the injectors
-------
as load increases. At still higher loads (above 70%). a second level of injectors is
manually inserted into the furnace. As load decreases, the second level of injectors is
manually withdrawn from the furnace. As load further decreases, the injectors on the first
level are manually rotated to their original position.
Initial operation of the SNCR system began during plant startup in the Spring of 1992.
By September of 1992, fabric filter (baghouse) differential pressures gradually increased
from a normal range of 4.0 to 6.5 inches water to over 12 inches water. This increase
restricted boiler air flow to the point where rated boiler load was no longer attainable.
Analysis of the bags revealed dense agglomeration of hygroscopic salts, which adhered
stubbornly to the filter surface.
In addition to the buildup of material on the filter bags, the first boiler inspection in
March, 1993, revealed hard deposits plugging the tubular air heaters and plating the
economizers.
Operational experience at similar Cogentrix plants provided some insight into the
problem at the Richmond plant. A plant similar to Richmond, with the same type of the
coal-fired stoker boilers but without an SNCR system, had been on line for about seven
years and never experienced any of the fabric-filter problems plaguing at the Richmond
plant. Also tubular air heaters and economizers experienced erosion, but not pluggage.
Plant personnel suspected that high levels of ammonia slip from the SNCR system might
be occurring. With permission from the state of Virginia, the SNCR system was removed
from two of the eight boilers. With the SNCR system operating, fabric-filter differential
pressure would begin to increase within two weeks, and within four weeks the pressure
would approach nine to ten inches water at rated load. After six weeks, the boiler would
have to be taken out of service and the filter bags washed.
Boilers operating without SNCR, displayed no change in the differential pressure of the
fabric filter. During that period, the boilers were operated at about 89% of rated capacity.
When urea injection commenced again, differential pressure had begun to increase
within one week.
Through late summer and early fall of 1993, more problems surfaced with the boiler
tubular air heaters and economizers. Exit gas temperatures increased 20 to 30 degrees F,
and pressure differential across the air heaters also increased, suggesting a worsening of
the deposition in these areas.
In October 1994, IGT as the technology developer together with Detroit Stoker Company,
the manufacturing partner, and Cogentrix of Richmond, the host site, were awarded a
contract by the Gas Research Institute to retrofit a boiler with the patented METHANE
de-NOX® rebum process. The ultimate goal of the project was to validate and deploy in
a full-scale utility boiler a concept that has been proven to be technically and
-------
economically feasible but required farther experience with full-scale systems to ensure
operational practicality and broad utility and industrial acceptance in order that such
technology is selected for long term use. Cofunding of this project was provided by IGT
Sustaining Membership Program (SMP) as well as several large gas/utility companies.
In-kind funding was provided by Detroit Stoker Company and Cogentrix of Richmond.
The City of Richmond supplied the natural gas line to the plant at no cost to the project.
METHANE de-NOX® Concept for Solid Fuel-Fired Stokers
Conventional rebuming is a process in which additional fuel (e.g., natural gas) is injected
into the products of complete combustion so as to eliminate the excess oxygen and
provide hydrocarbon radicals. These radicals react with NOX as the source of oxygen,
thereby reducing it to molecular nitrogen. The reburn process requires relatively high
temperature and sufficient residence time. At the end of the rebum zone, overfire air is
injected to create a burnout zone. This process has been successfully applied, using
natural gas as a rebuming fuel, to pulverized coal and cyclonic coal-fired combustors (5).
Over 50% of NOX reduction was demonstrated during long term operation. If
conventional rebuming is applied to the stoker, then the combustion volume will be
divided into four zones: two combustion zones (coal bed and combustion zone above the
bed hi the lower furnace), rebum and burnout zones. In this case at least three levels of
injection are required, two for overfire air and one for natural gas as reburning fuel. The
overall height of coal-fired stokers is usually not very large. For example, the elevation
of Cogentrix boilers from the grate to platen superheater is only about 35 feet. It is
practically impossible to install and ensure effective processing of two combustion zones,
rebum and burnout zones in such furnace. The total residence time of combustion
products in the furnace is about 2.5 seconds, but the rebum zone alone requires about
1 second for effective mixing and reacting of rebum fuel with the combustion products.
The METHANE de-NOX® rebuming process, developed by IGT with support from the
gas industry and GRI, for stoker fired combustors is shown in Figure 1. In this case,
natural gas as a rebuming fuel is injected just above the coal bed. Compared to
conventional reburning the METHANE de-NOX® process requires only three zones
instead of four, and two levels of injection instead of three. Injected natural gas not only
reduces NOX formed in the coal bed, but also limits its formation because a significant
portion of the NOX precursors are decomposed and react to form molecular nitrogen. A
reduction in the number of zones and injection levels provides sufficient residence tune
for rebuming and burnout. The METHANE de-NOX® process was evaluated on a pilot-
scale combustor at Riley Stoker's facility (6), and twice field tested during 1991 at the
Olmsted County Municipal Waste Combustion facility in Rochester, Minnesota (7), and
during 1995 and 1996 at the incineration plants in Japan (8). The tests at Olmsted County
achieved an average reduction of 60% of NOX and 50% of CO compared to baseline
levels when injecting 13% natural gas. In Japan, 55% of NOX reduction was achieved
with 7% of natural gas. A significant increase in boiler efficiency (about 3%) was also
demonstrated at the Olmsted Country 100 ton/d MSW boiler.
-------
The major advantages of METHANE de-NOX° compared to the existing SNCR are:
• The same or better NOX reduction in the range of 60 to 70%
• Increase in boiler efficiency by 1 to 2%
• Annualized cost benefit of about $340,000 for a 3 60 MMBtu/h boiler
• Additional environmental benefits due to the reduction of CO2, SOX, particulate,
and hydrocarbons, and the elimination of ammonia slip
• Improvement in plant operation for start-up and load changing regimes
• Exclusion of boiler outages for baghouse, economizer and air heater cleaning
• Elimination of chemicals such as urea, ammonia, etc., for combustion product
treatment
Baseline Testing and Modeling
Baseline, full and partial load testing were conducted on one of the eight stoker units at
the Richmond site. During baseline tests, the SNCR system was shut down per a prior
agreement between Cogenrrix and the Virginia Department of Environmental Quality.
The goals of the baseline tests were to: obtain baseline data for METHANE de-NOX®
system design, determine the location of NOX formation in the furnace, estimate the
ratio of fuel bound/thermal NOX, evaluate furnace gas composition and temperature
profiles, determine the effect of major operating parameters on emissions and furnace
temperatures, and define the effect of high excess air on superheated steam temperature
and boiler capacity. The variable parameters of the baseline tests were: combustion
zone stoichiometry, total excess air, undergrate and overfire air split, undergrate air
distribution, and coal heat input. A total of 16 tests were conducted as part of the
baseline test plan. Two loads were tested, 60 and 100%, total excess air was varied
with oxygen concentrations in flue gas from 1 to 8%. Undergrate to overfire air ratio
was varied from 70/30 to 90/10. Undergrate air distribution was varied via positioning
of four dampers on each side of the stoker unit. Baseline test measurements included
basic boiler performance, spreader stoker performance, furnace gas compositions and
temperatures, coal and ash sample analyses, and emissions data. In-furnace data were
obtained at three elevations using water-cooled probes. Data were obtained above the
grate, in the middle of the furnace, and at the exit of the furnace. The main finding of
baseline tests was that it should be possible to reduce NOX emissions by at least 50%
using METHANE de-NOX" reburning technology. Certain results of baseline testing
were presented at the AFRC 96 (9).
A typical output of the computational fluid dynamic (CFD) model (developed by
B&W) of the boiler furnace at Richmond utilizing METHANE de-NOX® rebum is
shown in Figure 2. Calculated average NOX concentrations as a function of elevation
are shown which indicate that total NOX concentration decreases rapidly from about
425 ppm to about 150 ppm, (@ 3% 0^), there is about 65% reduction. The model
indicates that most of the NOX is from the fuel bound and only about 20% is thermal
NOX. In the figure, an elevation of 118 feet corresponds to the grate level.
-------
Retrofit Testing
During the spring and summer of 1996, the METHANE de-NOX® reburn system was
installed on the stoker boiler and prepared for continuous operation. The rebum system
consists of a natural gas supply system, flue gas recirculation system(FGR) and also
includes a modification of one row of overfire air nozzles without any retrofit of the
boiler waterwall tubes.
Retrofit testing was conducted at the Richmond site in July and August 1996. The SNCR
system was shut off per a prior agreement between Cogentrix and the Virginia
Department of Environmental Quality, as it was for baseline testing. The major goals of
the retrofit tests were to: validate and deploy the METHANE de-NOX® rebum process
for a coal-fired stoker boiler, and determine the operating parameters for the reburn
system and retrofitted boiler for continuous operation. The variable parameters of the
retrofit tests were: combustion zone stoichiometry, total heat input, natural gas input and
distribution, FOR input and distribution, total excess air, and undergrate/overfire air split.
Nineteen retrofit tests were conducted according to the test plan and matrix. Four loads
were tested, 40, 60, 80, and 100%; natural gas input was varied from 5 to 25% with
different distribution between the injection levels. The FOR flow was in the range of 10
to 35% and also with different distribution. The total excess air was varied with O2
concentration in the flue gas at the furnace exit from 2 to 5%, and undergrate/overfire air
ratio was in the range of 15 to 32% at different boiler loads. The retrofit test
measurements included the same volume of parameters as for baseline test, but in
addition flue gas composition and temperature were measured at the boiler exit.
The main retrofit test result is that the METHANE de-NOX® reburn process was able to
reduce NOx up to 70% and allowed operation of the coal-fired stoker boiler in
compliance with the Virginia Department of Environmental Quality NOx and CO
regulations without any urea injection.
Selected retrofit test results are shown in Figures 3 to 7. The effect of primary
combustion zone stoichiometry in the coal bed on the NOX in the center above the grate
for two loads, 60 and 100%, is presented in Figure 3, and on natural gas input required to
maintain NOX at the boiler exit in compliance is shown in Figure 4. The primary
combustion air/coal ratio has a significant effect on NOX above the coal bed, and with
load increase, the NOX level above the grate increases also, Figure 3. Larger natural gas
input allows keeping higher undergrate air to coal ratio which assures effective operation
of the grate. Figure 4. The NOX distribution versus furnace elevation at 100% load for
different natural gas inputs is given in Figure 5. The reburning zone is located
somewhere between the 120 and 130 feet elevations, and after the reburning zone the
NOX concentration is nearly unchanged.
-------
The effect of natural gas input on NOX emissions is shown in Figure 6. When the natural
gas portion in the total heat input is increased, NOX at the furnace exit decreases, and this
effect is more pronounced at the higher load. The required FOR flow for different boiler
loads is presented in Figure 7. At full load only 9% FOR is required for rebum process,
and the portion of undergrate FOR is about 2%. As the boiler load decreases the
percentage of undergrate FOR increases to replace the extra undergrate air used for
regular grate operation at partial loads.
In March 1997 the METHANE de-NOX® rebum system was modified. The FOR
withdrawal was switched from the exit of the dust collector to the ID fan discharge. The
capital cost of the modified system is significantly lower compared to the original system.
The modified rebum system was extensively tested and the results were compared with
the retrofit tests in August 1996.
During all retrofit tests, the boiler operating data were continuously collected through the
boiler's distributed control system. The boiler operating parameters and performance for
the tests in August 1996 (Aug'96) and March 1997 (Mar'97) and for four loads, 40, 60,
80, and 100%, are shown in the Table 1. With about 6 to 8% natural gas injection, NOx
was maintained in the range of 0.28-0.29 Ib/MMBtu and CO emission was less than 0.09
Ib/MMBtu. The Virginia state regulations are: NOx < 0.30 Ib/MMBtu and CO < 0.20
Ib/MMBtu. The O2 at the furnace exit was about 3% at full load and increased to 5% at
40% load. These oxygen levels are much lower than boiler design oxygen
concentrations, especially at partial load. As a result of this, stack losses were
significantly decreased and boiler thermal efficiency increased by about 2%.
For all retrofit tests, the boiler thermal efficiency was in the range 88.7 89.8 % with no
decline at partial loads. The improvement of boiler performance is an important feature
of the METHANE de-NOX® rebum system.
Carbon in bottom ash was from 5 to 10 % for Aug'96 and from 7 to 9 % for Mar'97 tests.
But carbon in fly ash was significantly lower for the Mar'97 compared to Aug'96 tests,
10 to 15 % compared to 20 to 35 %.
Pressure drops for the secondary and primary economizers, and air heater were practically
the same for tests in Aug'96 and Mar'97. The increase in baghouse pressure drop is
because of the FOR flowing through the baghouse in the modified reburn system.
After completion of retrofit testing, the METHANE de-NOX® rebum system was put in
permanent long-term operation.
-------
Long-Term Testing
Since July 1996, the METHANE de-NOX® rebum system has being continuously
operated at Cogentrix plant in Richmond, Virginia. During this period, the SNCR system
has not been employed, and NOX and CO emissions have been within the limits of the air
permit issued by the Virginia Department of Environmental Quality.
From August 1996 to July 1997 long-term testing of the METHANE de-NOX® reburn
system was conducted. The major goals of these tests were to : validate the rebum
technology in long-term permanent operation including load swings and different upset
conditions; determine the combination of boiler operating parameters which provides the
best boiler performance; collect essential data to design the commercial rebum system for
the other seven boilers at Cogentrix in Richmond; and confirm operational practicality
and cost-effectiveness of the METHANE de-NOX® reburn technology for other potential
customers nationwide.
A special procedure for data collection and processing was developed. Total of 168
operating parameters during each test were collected through the boiler I & C system.
The tests were conducted at stable load and load swings in the range of 40 to 100 %,
during boiler regular maintenance, and at the most of the upset conditions.
Based on the tests results, the best combination of operating parameters was selected and
set up in the boiler control system for permanent operation. Table 2 shows selected boiler
operating parameters and performance for four months during the one year period, from
August 1996 to July 1997. The data present average parameters at the maximum load for
the month. NOX and CO are within the required limits, and 02 is about 3 % at full load
and about 4 % at 80 % load. Boiler thermal efficiency was in the range 88.4 - 89.0 %
which is for about 2 % higher than baseline efficiency. Pressure drops through the boiler
firesides and baghouse were stable. No fouling has occurred in the economizer, air
heater, dust collector, or baghouse. During this period, no expenses have been incurred
for urea, maintenance of SNCR equipment, or cleaning of boiler firesides and baghouse.
Implementation of the METHANE de-NOX® has provided improvement of the boiler
long-term reliability and significant reduction in operating costs.
Due to favorable results of retrofit and long-term testing, Cogentrix decided to install the
METHANE de-NOX® rebum system on the other seven boilers, one boiler to be
retrofitted in 1997, and the remaining six - during 1998-99.
-------
Conclusions
Based on the results of retrofit test and one year continuous operation, the METHANE
de-NOX® reburn technology is an attractive alternative to SNCR for emissions reduction
in coal-fired stoker boilers and provides improvement in boiler operating performance
and thermal efficiency. IGT has licensed to Detroit Stoker Company the IGT's reburning
technology for deployment nationwide. At the present time, companies are allowed to
take credit for operating a boiler below the permitted emission level and this presents
another motivation to reduce NOX emissions using METHANE de-NOX®
Acknowledgments
Many sponsors played important roles in the development of the METHANE de-NOX"
technology for coal-fired stoker boilers. The authors wish to acknowledge the City of
Richmond for supplying the natural gas line at no cost to the project, and the financial
support of Columbia Gas Distribution Companies, Consumers Power, Indiana Gas
Company, Inc., North Carolina Natural Gas Corporation, and Texas Gas Transmission
Corporation. Personnel of Cogentrix warrant special mention for allowing interruption of
commercial operations to enthusiastically support and vigorously assist in the
implementation of the retrofit.
References
1. U.S. Patent No. 5,205,227, April 27, 1993.
2. U.S. Patent No. 5,020,456, June 4, 1991.
3. "SNCR Experience with Coal Fired Boilers and Fabric Filters", Hall, David and
Bonner, Thomas. Paper presented at the Institute of Clean Air in Washington, DC, 1994.
4. "NOx Control System at Cogeneration Plant Plagues Downstream Components"
Power, December, 1994, Page 42, Vol. 138, No. 12
5. "Gas Reburning GRI Program Overview," Pratapas, J. M. International Workshop on
Gas Rebum Technology, Malmo, Sweden, February 7-9, 1995.
6. "Pilot-Scale Assessment of Natural Gas Reburning Technology for NOx Reduction
From MSW Combustion Systems", Abbasi, H.A., Khinkis, M.J., Itse, D.C., Penterson,
C.A., Wakamura, Y. and Linz, D.G. Presented at the International Conference on
Municipal Waste Combustion, Hollywood, Florida, April 11-14,1989.
-------
7. "Field Evaluation of METHANE de-NOX® at Olmsted Waste-To-Energy Facility,"
Biljetina, R., Abbasi, H.A., Cousino, M.E. and Dunnette, R. Paper presented at the 7th
Annual Waste-To-Energy Symposium, Minneapolis, Minnesota, January 28-30, 1992.
8. "Demonstration of Rebuming Technology on MSW Furnace", Wakamura, Y. Paper
presented at the American Flame Research Conference, Baltimore, Maryland, September
30-October 2, 1996.
9. "Validation of METHANE de-NOX® Rebum Process for Coal-Fired Spreader Stoker
Boilers", Rabovitser, I., Chan, I., Loviska, T., Hall, D. Paper presented at the American
Flame Research Conference, Baltimore, Maryland, September 30 - October 2, 1996.
Table 1. Boiler operating parameters and performance
for August 1996 and March 1997
LOAD, %
O2, %, furnace
CO, Ib/MMBtu
NOx, Ib/MMBtu
NG, % input
Bottom ash carbon, %
Carbon in flyash (DC2), %
Boiler thermal efficiency, %
Pressure drop (inch H2O) for:
Secondary Economizer
Primary Economizer and Air
Heater
Baghouse
40
4.58/5.29
0.06/0.06
0.28 / 0.29
8.3/8.0
9.8/7.4
31.3/15.0
88.7 / 89.6
0.9/0.7
2.0/1.5
2.2 / 3.4
60
3.97/4.94
0.05/0.13
0.29 / 0.28
7.3/6.3
5.0/7.5
21.9/10.8
89.0 / 89.8
1.0/1.0
2.3 / 2.2
3.5/4.8
80
3.41 /4.11
0.04 / NA
0.28 / 0.27
7.6/6.5
5.6/9.2
19.8/12.5
89.5/88.8
1.5/1.2
3.4/2.8
4.6/5.4
100
2.89/3.23
0.07 / NA
0.29 / 0.29
8.3/6.3
7.7/8.4
35.4/10.2
88.9/89.0
1.9/1.7
4.5/4.1
6.7/8.2
Note: a/b; a - data for August 1996, b - data for March 1997
10
-------
Table 2. Boiler operating parameters and performance
for August'96 through July'97
Month
Boiler steam flow, klbs/h
O2, %, furnace
CO, Ib/MMBtu
NOx, Ib/MMBtu
NG, % input
Boiler thermal efficiency, %
Pressure drop (inch H2O) for:
Secondary Economizer
Primary Economizer and Air
Heater
Baghouse
Aug'96
290.0
2.9
0.07
0.29
8.3
88.9
1.9
4.5
6.7
Oct'96
213.1
4.1
0.08
0.30
4.2
88.4
1.7
4.0
4.8
MarW
257.0
3.5
0.08
0.29
6.3
89.0
1.7
4.2
8.2
Jul'97
299.8
2.5
0.14
0.28
4.0
88.8
1.7
4.2
6.4
NOX Regulation: 0.30 Ib/MMBtu
CO Regulation: 0.20 Ib/MMBtu
\OEGA7~97.DOC
11
-------
Burn outone
NOx and
MOx Precursor
Reduction Zone
Natural Gas/FQR
NOx Formation
Zone
Undergrate Air/FGR
Figure 1. Spreader stoker boiler with
SM
METHANE de-NOX reburn
YG1vEGA_97.ppt
-------
118 120 122 124 126 128 130 132 134 136 138 140 142 144 146 148 150 152
Elevation (feet)
. Combined Fuel & Thermal
i Fuel Only A Thermal Only
Figure 2. Average NOx concentration versus elevation
m
x
O
r 0.4
468
Oxygen, %
10
12
Figure 3. Effect of oxygen on NOx above the grate
at elevation 122'
YC/MEGA_97.ppt
-------
NATURAL GAS INPUT, %
._. fri ->• -^ M K) 00
0 U1 o en o 01 o
*•
^
x-'
^^^
,-s
S
HOx = 0.22-0.25 Ib/MMBtu
10 11 12 13 14 1
15
UNDERGRATE AIR TO COAL RATIO
Figure 4. Effect of natural gas input on grate
stoichiometry
Natural gas input = 6%
0.8
0.6
0.4
0.2
0
__[ * Natural gas input = 8%
0.8
0.6
0.4
0.2
0
• Natural gas input = 17%
122
136.5
ELEVATION, feet
151
Figure 5. NOx versus elevation for 100% load
YC/MEGA_97.ppt
-------
0.6 -•
+J
£ 0.4 4-
x
O
z
0.2 -•
State Regulations
—i—
0 8 15
Natural Gas Heat Input, %
Figure 6. NOx reduction by METHANE de-NOX reburning
40
35
30
g20
u.
15
10
40
Total FGR
Undergrate FGR
60 80
Steam flow, %
100
Figure 7. FGR flow versus boiler load
YC/MEGA_97.ppt
-------
Tuesday, August 26; 8:00 a.m.
Parallel Session B:
Selective Noncatalytic Reduction
-------
USING RETRACTABLE LANCES TO MAXIMIZE
SNCR PERFORMANCE
T. Hunt L. J. Muzio, R. A. Smith
Public Service Company of Colorado Fossil Energy Research Corp.
Denver, Colorado 80207 Laguna Hills, California 92653
J. L. Hebb J. Stallings
U. S. Department of Energy Electric Power Research Institute
Pittsburgh, Pennsylvania 15236 Palo Alto, California 94303
F. Ghoreishi J. H. Booher
NOELL, Inc. Diamond Power Specialty Company
Herndon, Virginia 22071 Lancaster, Ohio 43130
Abstract
Public Service Company of Colorado, the U.S. Department of Energy, and the Electric Power
Research Institute sponsored a demonstration of the Integrated Dry NO./SO2 emissions Control
System. This project demonstrated 70% NOX and SO2 reduction by integrating up to five
different emission control systems. The project was conducted on a 100 MWe coal-fired boiler
located in Denver, Colorado.
One important component of the project is the selective non-catalytic reduction (SNCR) system
used to increase NOX reduction beyond that possible with combustion modifications alone.
Initial testing of the SNCR system demonstrated up to 45% NOX reduction while controlling
ammonia slip to 10 ppm at full load; low load performance was limited to only 11% NOX
reduction at an equivalent ammonia slip. To improve low load performance, NOELL's
Advanced Retractable Injection Lance (ARIL) was installed. The lance is unique as it may be
rotated to inject the liquid urea solution into the correct flue gas temperature range to maximize
NOX reduction. This lance greatly improved low load performance and obtained up to 50% NOX
reduction at 50 MWe. The ARIL lances met all performance expectations but experienced
permanent bending due to the very high flue gas temperatures at the injection location. This was
remedied by adding cooling slots. A second retractable lance design, provided by Diamond
Power Specialty Company (DPSC), was also tested. This lance was designed to use evaporative
cooling to provide additional cooling and reduce lance bending. The DPSC lance solved many of
the operational issues, but obtained slightly lower NOX reduction at equivalent ammonia slips.
While the lances operated differently, the information gathered during this demonstration
provides the data necessary to help select the appropriate location and lance design to maximize
NOX removal based on site-specific factors of both lance designs.
-------
Introduction
Public Service Company of Colorado (PSCC) was selected by DOE for a CTT-HI project in
December 1989 to demonstrate an Integrated Dry NOX/SO, Emissions Control System. The
Electric Power Research Institute also cofunded the project. The demonstration project took
place at PSCC's Arapahoe Unit 4, a 100 MWe top-fired unit which fires a low sulfur (0.4%)
Colorado bituminous coal as its main fuel, but also has 100% natural gas capability.
The Integrated Dry NO,/SO: Emissions Control System combines five major control
technologies to form an integrated system to control both NOX and SO2 emissions. The system
uses low-NOx burners, overfire air, and urea injection to reduce NOX emissions, and dry sorbent
injection using either sodium- or calcium-based reagents with (or without) humidification to
control SO, emissions. The goal of the project was to reduce NOX and SO2 emissions by up to
70%. The combustion modifications were expected to reduce NOX by 50%, and the SNCR
system was expected to increase the total NOX reduction to 70%. Dry Sorbent Injection was
expected to provide 50% removal of the SO2 emissions while using calcium-based reagents.
Because sodium is much more reactive than calcium, it was expected to provide SO2 removals of
up to 70%.
Prior publications presented results of the performance of the individual technologies and
Reference 1 provides an overall summary of the project. This paper will present recent results
focused on improving the load performance of the SNCR system.
SNCR System
The purpose of the SNCR system at Arapahoe was two-fold. First, to further reduce the final
NOX emissions obtained with the combustion modification so that the goal of 70% NOX removal
could be achieved. Second, the SNCR system is an important part of the integrated system
interacting synergistically with the dry sodium injection system. During this program, it was
shown that when both systems are used simultaneously, both NO2 emissions from the sodium
system and NH3 slip from the SNCR system are reduced.
When the SNCR system was originally designed and installed, it incorporated two levels of wall
injectors with 10 injectors at each level. These two separate levels were intended to provide load
following capability. The locations of these two levels were based on flue gas temperature
measurements made with the original combustion system. However, the retrofit low-NOx
combustion system resulted in a decrease in the furnace exit gas temperature of nominally 200 °F.
This decrease in temperature moved the cooler injection level out of the SNCR temperature
window. With only one operational injection level, the load-following performance of the
system was compromised. For a 10 ppm NH3 slip limit, low load NOX reductions were limited to
11%.
Two approaches were pursued to improve the low load performance of the SNCR system. First,
short-term testing showed ammonia to be more effective than urea at low loads. Although
ammonia was more effective than urea, it remained desirable to store urea due to safety concerns.
A system was installed that allows on-line conversion of urea into ammonia compounds. The
-------
on-line conversion system improved low load performance, but the improvement was not as
large as desired at the lowest load (60 MWe).
More recently, NOELL, Inc. (the original supplier of the SNCR system) suggested an additional
injection location in a higher temperature region of the furnace. Because no unit outages were
planned, the only option for incorporating an additional injection level was to utilize two existing
(but unused) sootblower ports in conjunction with NOELL's Advanced Retractable Injection
Lances (ARILs). This location was chosen because the ports existed, not because the
temperatures were ideal for SNCR.
The SNCR system uses NOELL's proprietary dual-fluid injection nozzles to distribute the urea
uniformly into the boiler. A centrifugal compressor is used to supply a large volume of medium-
pressure air to the injection nozzles. The large quantity of air helps to atomize the urea solution
as well as provide energy to rapidly mix the atomized solution with the combustion products.
Figure 1 shows the location of the new ARIL lances relative to the two original SNCR injection
locations. Level 2 is the location that became unusable as a result of the flue gas temperature
decrease after the low-NOx combustion system retrofit. The ARIL system consists of two
retractable lances and two retractable lance drive mechanisms. Each lance is nominally 4 inches
in diameter and approximately 20 feet in length. Each lance has a single row of nine injection
nozzles spaced on two-foot centers. A single division wall separates the Arapahoe Unit 4
furnace into east and west halves, each with a width of approximately 20 feet. When each lance
is inserted, the first and last nozzles are nominally one foot away from the division and outside
walls, respectively.
Each injection nozzle is composed of a fixed air orifice (nominally one-inch in diameter), and
a replaceable liquid orifice. The liquid orifices are designed for easy removal and cleaning. This
ability to change nozzles also allows adjustments in the chemical injection pattern along the
length of the lance in order to compensate for any significant maldistributions of flue gas
velocity, temperature, or baseline NOX concentration.
Two separate internal liquid piping circuits are used to direct the chemical to the individual
injection nozzles in each lance. The four nozzles near the tip of the lance are supplied by one
circuit, and the remaining five are supplied by the other. This provides the ability to bias the
chemical flow between the "inside" and "outside" halves of each side of the furnace in order to
compensate for various coal mill out-of-service patterns. Each lance is also supplied with a pair
of internal thermocouples for detecting inside metal temperatures at the tip of the lance.
The retractable lance drive mechanisms were supplied by Diamond Power Specialty Co. (DPSC).
The drives are Model K 525's which have been modified for the liquid and air supply parts.
Both remote (automatic) and/or local (manual) insertion and retraction operations are
accomplished with the standard IK electric motor and gearbox drive system. A local control
panel is provided on each ARIL lance drive mechanism. Each panel contains a programmable
logic controller for the lance install/retract sequencing and safety interlocks. Each lance can be
rotated either manually at the panel, or automatically by the control system during load-following
operation. One of the key features of the ARIL lance system is its ability to rotate the lances. As
-------
will be discussed, this feature provides a high degree of flexibility in optimizing SNCR
performance by varying the flue gas temperature at the injection location by simply rotating the
lance.
In addition to NOELL's ARIL lances, an alternate lance design, supplied by Diamond Power
Specialty Company, was also evaluated. This alternate lance design represented a simplification
to the original ARIL design. The liquid solution is injected through a single pressure atomizer
located in the air supply pipe ahead of the lance. This eliminates the internal liquid piping, and
spraying at the lance inlet provides evaporative cooling to help cool the lance. The DPSC lance
design also eliminated a telescopic device used in the ARIL lance design. The telescope worked
well, but contains seals that require replacement due to normal abrasion that occurs during the
insert and retract process. In addition, the design prevents air and liquid from being injected in
the local region around the boiler when the lances are retracted.
SNCR Lance Performance Results
The recent test work has focused on the performance of the SNCR lances, both the NOELL
ARIL lances and the alternate DPSC lance, in order to enhance low load performance of the
SNCR system. The majority of the testing was performed with the NOELL ARIL lances. As
such, the results will focus on the ARIL test results along with a comparison of performance
between the NOELL ARIL and DPSC lances.
ARIL Lances
Prior to incorporating the ARIL lances into the SNCR control system, a series of parametric tests
was conducted to define the optimum injection angle at each load. As shown in Figure 1,
each lance can rotate to inject urea into a different region of the furnace in order to follow the
SNCR temperature window as the boiler load changes. The minimum injection angle is 22°
(0° corresponds to injection vertically downward), at which point the chemical is injected
parallel to the tube wall located below the lances. Smaller injection angles are not used to avoid
direct liquid impingement on these tubes. An injection angle of 90° corresponds to injection
straight across the furnace toward the front wall, and an angle greater than 90° results in injection
of the solution in a direction up toward the roof-mounted burners.
While the primary focus of the parametric tests was to define the injection angle versus load, the
tests also investigated the effects of:
• coal mill out-of-service patterns
• coal mill biasing
• biasing the urea flow along the length of the lances
• independent adjustment of the injection angles for each lance
The results of these tests are described below.
-------
Effect of Lance Angle
One of the primary attributes of the ARIL lance system is the inherent flexibility of accessing the
optimum flue gas temperature location by simply rotating the lance. Figure 2 shows the effect of
varying the lance injection angle at loads of 43 and 50 MWe. All of the tests shown in these
figures were performed at a N/NOX ratio of 1.0, with two mills in service. At 43 MWe, varying
the injection angle had little effect on NOX removal, and the maximum removal occurred at an
angle of 35 degrees (Figure 3a). However, Figure 2a shows that the lance angle had a large effect
on NH3 slip; decreasing from 46 ppm at an angle of 22° to under 5 ppm at an angle of 135°
This overall behavior at 43 MWe suggests that, on average, injection is occurring just on the high
side of the SNCR temperature window. In fact, the optimum temperature, in terms of NOX
removal, appears to correspond to an angle of 35°. However, since it is desirable to maintain the
NH3 slip less than 10 ppm, an injection angle of 90° is a more appropriate operating angle at this
load.
At a slightly higher load of 50 MWe (Figure 2b), the effect of lance injection angle was markedly
different. At this load, where the average flue gas temperatures were higher, injection angle had
little effect on NH3 slip. However, at the higher temperature, lance angle had a large effect on
NOX removal. The relative insensitivity of the NH3 slip and large sensitivity of the NOX removal
to lance angle suggests that at 50 MWe, chemical injection is occurring far on the high side of the
SNCR temperature window for injection angles ranging from 22° to 135°.
The results at 43 and 50 MWe shown in Figure 2 illustrate how varying lance angle can be used
to optimize the SNCR performance over the load range. As the load increases, the preferred
injection angle will decrease. Again, the minimum angle is 22°, where the chemical is injected
parallel to the tube sheet located below the lances.
Performance over the Load Range
The SNCR performance using the ARIL lances over the load range from 43 to 80 MWe is shown
in Figure 3. Note that for this particular lance location, the flue gas temperatures are too high for
the lances to be effective at loads greater than 80 MWe. As the load increases, the preferred
lance angle decreases in order to inject the urea into a lower temperature region.
As discussed above, at 43 MWe with an angle of 90°, injection occurred on average just on the
high temperature side of the window. At N/NOX = 1, NOX removals were 35% with less than 10
ppm NH3 slip. At 50 MWe, a 45 ° injection angle was on average at a better location in the
SNCR window, with NOX removals of 40% and NH3 slip less of 5 ppm at N/NOX = 1. As the
load increased to 60 MWe, a decrease in lance angle to 34° resulted in SNCR performance
similar to a load of 43 MWe. At higher loads of 70 and 80 MWe, injection was clearly occurring
on the high side of the temperature window. Note that the NH3 slip at 80 MWe was higher than
the slip at 70 MWe even though the chemical was injected into a region of higher overall
temperature. This effect was a result of temperature stratification in the furnace, and the way in
which the stratification varies with different coal mill patterns. This effect is discussed in more
detail below. However, the data in Figures 2 and 3 clearly show that the lances have markedly
improved the low-load performance of the SNCR system.
-------
Effect of Boiler Operation on SNCR Performance
As mentioned above, local changes in temperature due to variations in boiler operating
parameters (excess O2, mill pattern, mill biases, etc.) can have a major impact on SNCR
performance. This is particularly true at Arapahoe Unit 4 where the 12 burners are located on the
roof of the furnace. Each of the four coal mills feeds three burners, two burners on one side of
the furnace and a single burner on the other side of the furnace. Since the furnace has a division
wall, there is an imbalance in heat release across the furnace, and a corresponding variation in
flue gas temperature, when only three mills are in service. These temperature variations impact
the performance of both the wall injectors and the ARIL lances. The effect will be illustrated by
looking at the performance of the ARIL lances with varying mill out-of-service patterns. During
normal operation, Arapahoe Unit 4 operates with four mills in service over the load range from
80 to 110 MWe (although the unit can operate up to 100 MWe with only 3 mills). From 60 to 80
MWe, the unit typically operates with three mills in service. Below 60 MWe, the unit is usually
operated with only two mills in service. Figure 4 shows the overall impact of various mill out-
of-service patterns on SNCR performance at 60 MWe. As can be seen, NOX removals varied
from 30% to 52% (@ N/NOX = 1.5) depending on which particular mill was out-of-service.
Comparably, the NH3 slip varied from under 5 ppm to over 30 ppm with different mill-in-service
patterns. This behavior made overall optimization of the SNCR system quite challenging.
In addition to the temperature variations that occur with the various mill out-of-service patterns,
day-to-day variations can occur as a result of changes in the performance of the individual mills,
or changes in any other variables which affect the flue gas temperature distribution. Three
operational changes were investigated to deal with these types of temperature variations.
• varying the urea flow along the length of each lance
• independently varying each lance angle
• biasing the in-service coal mills
Varying the urea flow between the two liquid zones in each lance provided minor improvements
in the performance of the SNCR system. Independently varying the lance angles as a function of
the mill-in-service pattern also provided minor improvements. Unfortunately the implementation
of either of these strategies would significantly complicate the automatic control system. On the
other hand, biasing the in-service coal mills, which is relatively easy to implement, resulted in
major improvements in the performance of the SNCR system. Arapahoe Unit 4 is equipped with
four O2 monitors at the economizer exit. Biasing the coal mills to provide a balanced O2
distribution at this location is a fairly simple exercise for the boiler control operator. Figure 5
shows the improvements in SNCR performance that can be achieved by biasing the coal mills.
These tests were performed at a load of 60 MWe with both lances at an injection angle of 22°
and A mill out-of-service. The "biased" condition in Figure 5 corresponds to a negative 10%
bias on B mill and D mill, and a positive 10% bias on C mill. This has the net effect of moving
coal from the east side to the west side of the furnace to compensate for A mill being out-of-
service. Biasing the mills increased NOX removals from nominally 27% to 42% at an NH3 slip
limit of 10 ppm.
-------
Alternate DPSC Lance
While the NOX removal performance of the ARIL lances has been good, their location in the
furnace has resulted in some operational problems. At this particular location in the furnace, the
lances are exposed to a large differential heating between the top and bottom surfaces. The top
surface receives a high radiant load from the burners, while the bottom of the lance radiantly
communicates with the relatively cold tube wall immediately below. This uneven heating pattern
causes a great amount of thermal expansion along the upper surface, and the lances bend
downward toward the tubes. Within 30 minutes of insertion, the tip of each lance would drop by
approximately 12 to 18 inches. Within less than six weeks of operation, the lances became
permanently bent, making insertion and retraction difficult. This was partially addressed by
adding additional cooling slots at the end of the lance.
An alternate lance design supplied by Diamond Power Specialty Company (DPSC) was also
evaluated. As mentioned previously, this design sprays the urea solution through a single
atomizer at the entrance to the lance. This provides evaporative cooling to supplement the air
cooling. The evaporative cooling was expected to help minimize the lance bending discussed
above.
Because of both system and time constraints, the test program was abbreviated compared to the
parametric tests with the ARIL lances. The DPSC lance exhibited the same general
characteristics as the ARIL lances in terms of load following capability. In presenting the test
results for the DPSC, an emphasis will be made on comparing the performance of the DPSC and
ARIL lances. One of the primary differences in the two lance designs is the manner in which the
liquid is introduced. The ARIL lance has an internal liquid circuit that distributes liquid to each
injection nozzle along the lance. On the other hand, the DPSC lance uses a single atomizer to
spray liquid into the lance's air stream at the entrance to the lance. This simplifies the lance
design and provides evaporative cooling in the lance. However, with the single atomizer, there
was some impingement on the lance wall. This resulted in a portion of the liquid flowing down
the center of the lance. A portion of this liquid literally dripped off the edge of the injection hole
and a portion was entrained by the injection air and re-atomized. This re-atomized portion of the
liquid should have a larger drop size distribution than the sprays from the ARIL lances. With a
larger drop size distribution, it was expected that longer vaporization times would be needed
along with higher temperatures. Thus, for comparable performance, it was expected that the
DPSC lance would require a larger injection angle than the ARIL lance. Further, the portion of
the liquid that drips toward the screen tubes should not result in much NOX reduction; rather this
liquid should be a source of NH3 slip.
The ARIL and DPSC lance performances are compared in Figures 6 and 7 for loads of 60 MWe
and 70 MWe, respectively. For both the DPSC and ARIL lance tests, the C mill was out-of-
service and the total liquid flow rate was 4 gpm. At 60 MWe and an injection angle of 22
degrees, the NOX reduction with the ARIL lance was a little lower than the DPSC lance at
injection angles of 34-65 degrees. However, the NH3 slip with the DPSC lance was higher than
the ARIL lance except at an angle of 65 ° As expected, the DPSC lance required a higher
injection angle of 45 to 65 degrees to produce the same NH3 slip characteristics as the ARIL
lance at 22 degrees. This supports the arguments above in terms of the coarser overall
-------
atomization with the DPSC lance, and the need to inject into a higher temperature region (i.e.,
larger injection angle).
Similar results are seen in Figure 7 for a boiler load of 70 MWe; except the NOX reduction was
higher with the ARIL lance at 22 degrees than the DPSC lance at 45 degrees. At an injection
angle of 22 degrees, the NH3 slip with the ARIL lances was comparable to the DPSC lance at an
injection angle of 65 degrees.
The above comparisons were done at a common N/NO ratio and illustrate the general process
temperature characteristics of the two different lances. For automatic operation, the real question
is what NOX reduction can be achieved at a specified NH3 slip limit (i.e., 10 ppm) and at what
urea injection rate (i.e., N/NOX ratio). This comparison is made in Table 1 using the data from
Figures 7 and 8.
Table 1
Achievable NOX Reduction at a 10 ppm NH3 Slip Limit
(C Mill OOS, Total Liquid Flow Rate: 4 gpm)
Load
60 MWe
ARIL (22 degrees)
DPSC (34 degrees)
DPSC (45 degrees)
DPSC (65 degrees)
ANOX
(%)
36
30
36
42
N/NO
(molar)
1.75
1.1
1.25
2.8
70 MWe
ANOX
(%)
43
32
35
N/NO
(molar)
2.6
1.8
2.7
At 60 MWe, Table 1 indicates that the DPSC lance can achieve comparable NOX reduction at an
angle of 45 degrees, compared to the ARIL lance at an injection angle of 22 degrees, and at a
lower N/NO ratio. At the higher load of 70 MWe, the ARIL lance can achieve 43% reduction at
N/NO=2.7 compared to 35% for the DPSC lance operating at an angle of 65 degrees.
Overall, the NOX reduction and NH3 slip performance of both lances was quite good, and either
lance design enhances the low load performance of the SNCR system. In general, the DPSC
lance requires the urea to be injected into a higher temperature (i.e., larger injection angle) than
the ARIL lance. Because of this difference in temperature characteristic, the relative
performance depends on loads. At 60 MWe, the DPSC lance has a slight advantage being able to
match the NOX reduction of the ARIL lance, but at a lower N/NOX ratio. However, at 70 MWe,
the ARIL lance can achieve a higher NOX reduction. Again, the relative performances of the two
lance designs will also be dependent on the coal mill pattern, which could not be investigated
during the DPSC lance test period because C mill was out of service for maintenance.
-------
Overall SNCR System Performance
The parametric tests were conducted to determine at which loads the lances should be used, as
well as the optimum injection angle for each of these loads. Based on the parametric tests, the
control system has been set up to operate with the Level 1 wall injectors at loads above 80 MWe.
Below 80 MWe, the ARIL lances are used. Figure 8 compares the NOX removal over the load
range for injection at the two locations with an NH3 slip limit of 10 ppm. It is evident that the
installation of the ARIL lances has improved low-load performance of the SNCR system.
Currently, NOX removals of more than 30% are achievable over the load range with less than 10
ppm NH3 slip. The minimum NOX removal of 30% occurs at 80 MWe, which corresponds to the
point where the temperature becomes too high for the ARIL lances and too low for the Level 1
injectors. Comparable performance can be expected from the DPSC lances. However, some
additional parametric tests, or accumulated long-term data, are needed to define optimum
conditions at some of the lower loads.
Conclusions
Public Service Company of Colorado, in cooperation with the U.S. Department of Energy and
the Electric Power Research Institute, has installed the Integrated Dry NOX/SO2 Emissions
Control System. Conclusions associated with the recent work to enhance the low load
performance of the SNCR system include:
• With the addition of retractable lances to the SNCR system (either the NOELL ARIL
lances or alternate DPSC lances), low load performance of the system urea-based SNCR
system was improved. NOX removals of 30 to 52% with an ammonia slip limit of 10 ppm
are achievable over the load range. This increases total system NOX reduction to greater
than 80% at full load, significantly exceeding the project goal of 70%.
• The ability to follow the temperature window by rotating the lances has been
demonstrated and also proved to be an important feature in optimizing the performance of
the SNCR system.
• The DPSC lance design overcame some of the design shortcomings of the ARIL lance.
Specifically, (1) eliminating the telescope for the air supply, (2) minimizing air and liquid
spraying outside of the boiler during insertion and retraction, and (3) minimizing bending.
• For comparable performance, the DPSC lances required a higher injection angle (i.e.,
higher temperature) than the ARIL lances. This was attributed to a larger drop size from
the DPSC lances.
• Atomization was not complete with the DPSC lances, resulting in a portion of the liquid
exiting the injection holes as a liquid stream.
• Mill-out-of-service pattern can have a major impact on NOX reductions and NH3 slip with
the lances. This was most easily accommodated by having the operators bias the in-
service mills to provide a more uniform temperature distribution across the furnace.
-------
References
1. T. Hunt, et al., Performance of the Integrated Dry NO/SO2 Emissions Control System,
Fourth Annual Clean Coal Technology Conference, (September 1995) Denver, CO.
Disclaimer
This paper was prepared pursuant to a Cooperative Agreement partially funded by the U.S.
Department of Energy, and neither Public Service Company of Colorado, any of its
subcontractors, the U.S. Department of Energy, nor any person acting on behalf of either:
(a) Makes any warranty or representation, express or implied, with respect to the accuracy,
completeness, or usefulness of the information contained in this paper, or that the use of
any information, apparatus, method or process disclosed in this paper may not infringe
privately-owned rights: or
(b) Assumes any liabilities with respect to the use of, or for damages resulting from the use
of, any information, apparatus, method or process disclosed in this paper.
Reference herein to any specific commercial product, process, or service by trade name,
trademark, manufacturer, or otherwise, does not necessarily constitute or imply its endorsement,
recommendation, or favoring by the U.S. Department of Energy. The views and opinions of
authors expressed herein do not necessarily state or reflect those of the U.S. Department
of Energy.
Figure 1. SNCR Injection Locations
-------
50
D-
S; 40
30
0 20
-a
o>
DC
10
n NOx Fteduction
• NH3 slip
20 40 60 80 100
Injection Angle, degrees
120
140
(a) Boiler Load: 43 MWe
50
Q.
D_
CO
Z
•o
o>
CC
40
30
20
10
D NOx Reduction
• NH3 slip
20 40 60 80 100
Injection Angle, degrees
120
140
(b) Boiler Load: 50 MWe
Figure 2. Effect of Injection Angle on NOX Removal and NH3 slip
(Loads: 43 and 50 MWe, N/NOX = 1.0)
-------
(a) NOX Removal
(b) NH3 Slip
Figures. ARIL Lance Performance Over the Load Range: 43 to 80 MWe
-------
Figure 4. Effect of Mill-in-Service Pattern on
ARIL Lance Performance at 60 MWe (22° Injection Angle)
Open: NOX Removal
Closed: NHg-slip
Figure 5. Effect of Coal Mill Bias on ARIL Lance Performance at 60 MWe
(A Mill DOS, 22° Injection Angle)
-------
D DPSC(34)
a DPSC(45)
o DPSC(65)
...o.--ARIL(22)
Closed: NOX Removal
Open: NH3slip
(): Injection Angle
Figure 6. Comparison of the DPSC and ARIL Lance Performance at 60 MWe
(C Mill OOS, Total Liquid Flow Rate: 4 gpm)
D.
cc
-• DPSC<45)
A DPSC(65)
-•-•- ARL(22)
Closed : NO,, Removal
Open: NH3 slip
() : Injection Angle
N/NO,
Figure 7. Comparison of the DPSC and ARIL Lance Performance at 70 MWe
(C Mill OOS, Total Liquid Flow Rate: 4 gpm)
-------
60
SO •
§ 30
o>
oc
d
•Z- 20
10
60 80 100
Load (MWe)
120
Figure 8. NOX Removal as a Function of Load for an NH3 Slip Limit of 10 ppm
-------
SNCR RETROFIT EXPERIENCE ON FOUR GAS- AND COAL-FIRED
BOILERS IN TCHAIKOVSKY, RUSSIA
H. Krigmont, D. Oehley F. Spokoyny, J. Hogsett
Allied Environmental Technologies, Inc. WAHLCO, Inc.
Huntington Beach, California 92647 Santa Ana, California 92704
G. Quartucy*, L. Muzio E. Eddings
Fossil Energy Research Corp. Reaction Engineering International
Laguna Hills, California 92653 Salt Lake City, Utah 84101
Abstract
Allied Environmental Technologies (ALENTEC) has recently installed SNCR systems on four
gas- and coal-fired boilers. These boilers are owned by PERMENERGO, and are located 800
miles east of Moscow in Tchaikovsky, Russia. These sister units have a maximum continuous
rating of 420 tonnes steam per hour (882,000 pounds per hour).
The SNCR systems were designed and fabricated in the United States by WAHLCO, Inc. The
designs were based on HVT and gaseous emissions measurements made on site at the
initiation of the project. The design was subsequently fine-tuned by computational fluid
dynamics (CFD) modeling performed by Reaction Engineering International. The client dictated
that only existing penetrations could be used for injection ports, which limited the final SNCR
system design flexibility.
This paper presents the SNCR system design specific, modeling predictions and field test
results. Also discussed are the logistics and challenges of performing complex engineering
projects in the former Soviet Union.
Overview
Allied Environmental Technologies, Inc. (ALENTEC) was awarded a contract to install urea-
based Selective Non-Catalytic Reduction (SNCR) systems on four gas- and coal-fired boilers
located in Russia. These boilers are owned by PERMENERGO and are located 800 miles east
of Moscow in Tchaikovsky, Russia. The project will comprise four tasks, as follows, upon
completion:
• Task 1 — Furnace temperature distribution measurements
• Task 2 — Computation Fluid Dynamics (CFD) modeling
* Corresponding Author
-------
• Task 3 — System Design, fabrication and installation
• Task 4 — Start-up and optimization testing
To date, Tasks 1 through 3 have been completed, as well as the initial portion of Task 4.
SNCR System Description
Tchaikovsky Units 1 through 4 are sister units having a maximum continuous rating of 420
tonnes steam per hour (882,000 pounds per hour). The units are opposed-wall-fired units, each
having one row of six burners on both the front and back furnace walls. The boilers are a
balanced draft design. The units fire natural gas as their primary fuel and coal as a back-up.
Baseline full load NOX emissions varied from 139 to 152 ppmc (ppm, dry at 6% O2) when firing
natural gas. When firing coal, baseline NOX emissions were 558 ppmc at a load of 398 TPH.
The SNCR systems installed on Tchaikovsky Units 1 and 2 are of the "high energy" type. These
systems utilize low pressure steam as the carrier fluid. The design steam flow is 4,400 kg/hour at
full load, which represents approximately 1.5 percent of the total combustion products flow.
Each unit incorporates three (3) injection levels on the front wall, in addition to two (2) levels of
side wall injectors. Figure 1 shows a schematic view of the boiler and Figure 2 shows the
location of the injectors at each level.
A nominal 50% urea solution is prepared on-site from bulk urea solids. After mixing, the
solution is pumped to a storage tank in the boiler house. Each of the four individual SNCR
systems is supplied by this tank. Urea flow is set according to a flow versus load curve generated
during the start-up/optimization testing.
Because the client dictated that only existing penetrations could be used for injection ports, the
final system design flexibility was limited. To overcome these limitations, the final system
design incorporated two individual injectors in each available front wall port. The design of
these injectors allows them to be independently rotated in order to provide the best possible
mixing between the injected urea and the combustion products.
Test Results
The test results presented below include the HVT tests, the CFD tests and the field start-up and
optimization tests. Each of these test series is discussed below.
High Velocity Thermocouple (HVT) Tests
The field testing tasks of this program required the measurement of both temperature and
gaseous emissions. Temperature measurements were made using a high velocity thermocouple
(HVT) probe. This HVT was of a standard water-cooled design utilizing a single radiation
shield. Suction power was provided by an air-powered vacuum eductor. The HVT probe also
incorporated the capability of providing a flue gas sample from the aspirated thermocouple
-------
location. During the temperature measurement task, gaseous emissions of NO, CO and O2 were
measured using a NOVA Model 2000 portable combustion analyzer. This analyzer utilizes
electrochemical "fuel cell" type sensors to measure species concentrations.
The HVT tests were conducted over a nominal load range of 210-400 tones/hour (TPH) when
firing natural gas. The nominal test load range for coal firing was 300-400 TPH. Each unit was
operated in a "normal" firing configuration during the HVT testing; no attempts were made to
balance fuel and/or air flows, or to otherwise alter unit operation.
The desired temperature range for urea injection is nominally 930C to 1150C, with the maximum
NOX reduction performance achieved at a temperature of 10IOC J. However, operation at below-
optimum temperatures results in high NH3 slip levels. For this reason, it is desirable to operate at
or above the optimum injection temperature. Thus, the desired injection temperature range for
this project is between 1010 and 1150C.
Figure 3 shows average furnace gas temperatures plotted versus load. Data are included for both
levels D and E from Units 1 and 2 while firing natural gas. Average temperatures at Level E
varied from 1023C to 802C, while average temperatures at Level D varied from 1152C to 865C
over the load range tested while firing natural gas. The difference between maximum and
minimum temperatures at a given load ranged from 118C to 283C, and varied with the
measurement location and load. The degree of stratification can also be characterized by
calculating the standard deviation of measurements made at a given level and load. For natural
gas firing, the standard deviations ranged between 32C and 67C. These data indicate that
Level D may provide the preferred temperature zone from loads of about 270-370 TPH. At loads
in excess of 370 TPH, Level E appears to be in the preferred temperature region.
Figure 3 also shows average temperature plotted versus load for the coal firing tests performed
on Unit 4. These test data are divided into two groups; tests performed on September 13 and 14
and tests performed September 18 through 20. Temperatures at Level E averaged 870C and
956C at a nominal load of 350 TPH for the two test series. The corresponding temperatures at
Level D averaged 1003C and 1074C. Temperature stratification was more pronounced when
firing coal, ranging from 45C to 133C.
The data show that temperatures increased during this five day time period between the two sets
of measurements. Coal was initially fired in Unit 4 during the first week of September, allowing
about one week of seasoning before the initial testing. The second set of coal tests was
performed five days after the initial coal tests were completed. During this time period, average
gas temperatures increased by about 65C at Level D to 85C at Level E. It appears that the
measured increases in gas temperatures were due to the gradual build-up of ash deposits on the
heat transfer surfaces, since soot blowers are not used (or needed) on the Tchaikovsky boilers.
These variations in temperature with time may adversely impact SNCR performance if coal is
fired for extended periods of time.
A typical temperature contour plot for the gas-fired tests is shown in Figure 4. This shows a
characteristic saddle-shaped profile for these units. The profile showed two temperature peaks
-------
near the furnace center of 1070 to 1090C at a load of 370 TPH. From the center, gas
temperatures dropped as either furnace side wall was approached.
Emissions measurements made during the HVT testing showed that NO emissions varied from
113 ppmc to 152 ppmc across the load range at Level E when firing natural gas. (Note that in this
paper, ppmc is equal to ppm, dry, corrected to 6% O2., which is the measurement standard at the
utility site.) When firing coal, NOX emissions at Level E varied from 290 ppmc to 558 ppmc. CO
emissions at Level E were less than 40 ppm when firing natural gas and less than 90 ppm when
firing coal.
Computational Fluid Dynamics Modeling
Computational fluid dynamics (CFD) modeling was performed by Reaction Engineering
International (REI) to evaluate reagent mixing and SNCR performance. REI developed a
reduced mechanism for gas-phase SNCR chemistry, which was used to quantify NOX reductions.
The reduced mechanism of seven (7) reactions and individual rate constants were developed so
that the mechanism could be incorporated into a CFD code2 This model accurately describes
the SNCR chemistry as indicated by comparison of process performance relative to predictions
obtained using a complete chemical mechanism. The SNCR submodel was incorporated into
GLACIER, a computer code which solves the governing fluid mechanics and reaction equations
in an Eulerian framework. Reference 3 provides further details regarding this model.
Prior to completing the system design, four cases were modeled for full boiler load as follows:
• Case 0 — Urea injection at Level E, front wall
• Case 1 — Urea injection at Levels D and E, front wall
• Case 2 — Urea injection at Levels D and E, front and side walls
• Case 3 — Case 2 with dual injectors in front wall ports and different side wall
injection locations
Table 1 provides the performance estimates for each of these cases. These data show that Case 3
presented the most flexibility in terms of NOX reduction and NH3 slip.
The CFD model was divided into two parts; a lower furnace model and an upper furnace model.
The initial modeling of the lower furnace showed that thermal boundary conditions had to be
varied between units in order to replicate the field test results. The boundary conditions modified
included the furnace wall emissivity. In one case, the furnace wall emissivity corresponded to
that expected for a gas-fires boiler, while the other case required a furnace wall emissivity that
corresponded to a dirty wall environment that could have resulted from previous coal firing.
This work provided good agreement between the field measurements and the CFD lower furnace
model.
The lower furnace model conditions were then used as the inlet conditions for the upper furnace
model. The initial upper furnace work (Cases 0-2) showed that over half of the gas at the furnace
outlet plane (FOP) had molar N/NO ratios less than 0.5 or greater than 3.5. This led to the
-------
Table 1
CFD Modeling Results
Case
0
1
2
3
Injection
Level E, 12 front
wall injectors
Levels D&E,
12 front wall injectors
Levels D&E; 12
front, 6 side injectors
Levels D&E; 16
front, 6 side injectors
NSR
0.8
0.8
1.2
1.1
Outlet NOX
ppm @
6%O2
113
130
108
112
NOX
Reduction,
28
17
31
29
NH3 Slip
ppm @ 3%
02
9
7
17
14
Case 3 configuration, in which the injectors were redistributed along the front and side walls
using existing ports. The Case 3 configuration provided improved reagent mixing, as measured
by the increase in the number of areas at the FOP having molar N/NO ratios between 0.5 and 3.5.
Figure 5 shows NOX reduction and ammonia slip distributions in the outlet plane for Case 3.
These data, presented as contour plots, show that NOX reductions were predicted to be highest at
the center of the boiler, and lower toward the edges with an increase right at the side walls due to
the effects of the side wall injection. The corresponding NH3 slip data show peaks near the edges
of the boiler which correlate to low temperature regions in the furnace.
Another way to look at the CFD results is to plot NOX reduction and NH3 slip versus the N/NO
ratio in each cell at the exit plane. This is shown in Figure 6 for Case 3. The degree of vertical
scatter indicates the amount of temperature stratification present, while the range of N/NO ratios
provides an indication of the ''mixedness" at the exit plane. Note that the N/NO ratio ranges
from 0 to about 3.5 at the exit plane. The NOX reduction values fall in a fairly narrow band when
plotted versus N/NO ratio. However, the NH3 slip falls into two general bands. One group
shows NH3 slip less than 10 ppm for N/NO ratios up to 3.5, indicating a high temperature region.
Conversely, there are a number of points showing high NH3 slip levels, corresponding to
temperatures on the low side of the SNCR temperature window. This figure shows that nearly
the same NOX reductions could be achieved with two different ammonia slip levels. This
behavior may be characteristic of the two-level injection scheme, where the lower level provides
lower NH3 slip levels due to the higher gas temperatures encountered in that injection zone.
Field Tests
To date, tests have been performed on Units 1 and 2 while firing natural gas. Plans call for
completion of the Unit 1 gas testing, as well as coal testing on another unit in the Fall of 1997.
-------
Logistical problems have resulted in the test program proceeding slower than originally planned.
Table 2 shows the analyzers and analysis methods used during the field testing.
Table 2
Emission Measurement Methods
Specie Analyzer Analysis Method
NO Siemens Ultramat 5E Nondispersive infrared
CO Fuji ZRH Nondispersive infrared
CO2 Fuji ZRH Nondispersive infrared
O2 Teledyne 326 Electrochemical fuel cell
NH3 Wet Chemical Ion specific electrode
Initial testing began on Unit 2 in the Spring of 1997. However, the wrong injectors were
installed at the center ports on the front wall due to a misunderstanding between plant personnel
and the on-site engineer. Due to this derivation from the system design, performance on this unit
was not as expected, so the testing efforts were redirected to Unit 1. The Unit 1 installation
included double injectors in all front wall ports as the final design had specified. Unit 1 data are
limited, since only three days of testing were permitted due to a condensed test schedule resulting
from an unscheduled unit outage.
The early Unit 2 testing was limited by the inability to accurately control urea flows to the in-
service injectors. Figure 7 shows NOX reduction and NH3 slip plotted versus the molar N/NO
ratio. These data show that NOX reductions in excess of 35 percent were achieved at a molar
N/NO ratio of 2.0. The corresponding NH3 slip at these conditions was in the range of 25 to
32 ppm. The NOX reduction profiles showed that the reductions were highest near the furnace
side walls, while they were measurably lower in the center of the furnace. These data suggest
that the NOX removals may have been compromised by the installation of the single injectors on
the front wall of Unit 2.
The Unit 1 tests were performed at a load of 360 TPH, which is 86% of the rated load of 420
TPH. NOX removals on Unit 1 ranged from 39 to 47% with molar N/NO ratios of 1.5 to 2.1, as
shown in Figure 8, when injecting at Level D. Tests conducted with urea injection at Level E
showed NOX reductions of 42% at a molar N/NO ratio of 2. Injection utilizing both Levels D
and E resulted in NOX reductions of 42 to 46% at a molar N/NO ratio of 2.0, depending on the
specific injector pattern used.
Ammonia (NH3) slip levels are also plotted versus molar N/NO ratio in Figure 8. At a molar
N/NO ratio of 2, NH3 emissions varied from 32 ppm when injection at Level D to nearly 70 ppm
when injection at Level E. The high ammonia slip encountered when injection at Level E may
indicate that the E level is too cold for injection at the 360 TPH load. When injection at both
Levels D and E, ammonia slip varied from 46 to 65 ppm at a molar N/NO ratio of 2.
-------
NOX reduction profiles were obtained by measuring both pre- and post-injection NOX emissions
at each of the 12 sample points at the economizer exit sample grid. Figure 9 shows NOX
reduction profiles for a test performed using Level D injection at a molar N/NO ratio of 1.5.
The data show that the achievable NOX reductions were highest on the outside walls and dropped
significantly to the boiler center. For example, injection at Level D resulted in NOX reductions
at the walls averaging 51%, while NOX reductions at the four innermost points averaged
only 24%.
Comparison of CFD and Field Test Results
The results of the Case 3 CFD modeling and the Unit 1 field test are shown together in Figure 10.
Both NOX reduction and NH3 slip are plotted versus N/NO ratio. These data show that the CFD
modeling are field test results correlated well on an overall basis. A subsequent comparison of
the CFD modeling and field test results shows that there are some differences in the details. It is
hoped that the testing scheduled for Fall 1997 will provide additional data which can resolve
these differences. In the interim, however, it appears that the CFD modeling provided a good
approximation of the overall field test results.
Project Logistics
As discussed previously, this project involved the design, fabrication and installation of four
SNCR systems in Tchaikovsky, Russia. The design and fabrication of the SNCR skids was
performed in Santa Ana, California by WAHLCO, Inc. These skids were fabricated so that two
systems were installed in a single container. Each container contained the required pumps, liquid
metering systems, control valves and piping for two SNCR systems. The SNCR injectors and
associated mixing equipment were also designed and fabricated in California. ALENTEC
designed the urea mixing and storage systems, and supervised their construction at the plant site.
The plant was responsible for connecting the containers to the urea storage and water supply
systems, and installing the piping from the container to the individual injection levels on each
boiler.
Language was not the barrier originally thought, since the plant provided an experienced
translator for the duration of the project. The primary delays in the project involved difficulties
in getting the SNCR system through customs in Moscow. Once on site, the plant staff did an
admirable job installing the SNCR systems. The primary on-site problem was a lack of spare
parts. Spare parts were a week or more away, at best, due to the relatively remote location of the
plant.
Acknowledgements
The authors wish to acknowledge the hospitality extended by Mr. Potapov, the Plant Manager.
His cooperation allowed all parts of the project to proceed smoothly. Mr. Alexi Grebnev, the
Chief Plant Engineer, also provided invaluable assistance in the installation and start-up of the
SNCR systems. Finally, Mr. Nicolai Oschepkov did an excellent job of translating, which
-------
allowed the project to proceed smoothly. All members of the plant staff should also be
commended for their efforts in completing a successful retrofit.
References
1. Muzio, LJ. and Arand, J.K., Homogeneous Gas Phase Decomposition of Oxides of
Nitrogen,, EPRI Report No. FP-253, August 1976.
2. Brouwer, J., et al., A Model for Prediction of Selective Noncatalytic Reduction of Nitrogen
Oxides by Ammonia, Urea and Cynauric Acid with Mixing Limitations in the Presence of
CO, Twenty-Sixth Symposium (International) on Combustion, The Combustion Institute, in
press.
3. Eddings, E.G. et al., Modeling Urea-Based SNCR in a Gas-Fired Utility Boiler, presented at
the U.S.D.O.E. Conference on Selective Catalytic and Non-Catalytic Reduction for NOX
Control, Pittsburgh, Pennsylvania, May 15-16, 1997.
-------
Figure 1. Tchaikovsky Units 1-4 Boiler Schematic Drawing
S6 S5
\ 9 9
1 4
^ *"&
l OO WT^
E1 E2 E3 E4
99 99 P9 99
1^.* M>^ ^V( ^*^
D1 D2 D3 D4
H- 4 P-^4 h- 4 ^4
C1 82 03 C4
S5 S6
? 9 /
>
Figure 2. SNCR Injector Schematic
-------
o
1200
1100
1000
900
800 l-Q
Gas Firing
D Level E
O Level D
Coal firing-Sept 13/14
A Level E
*• Level D
Coal firing-Sept18-20
T Level E
•* Level D
200 250 300 350 400 450
Load
Tonnes/hour
Figure 3. Furnace Gas Temperature versus Load. Tchaikovsky Units 1, 2 and 4
RearWall
Front Wall
Figure 4. Measured Temperature Contour Plot, Tchaikovsky Unit 2,
370 TPH, Natural Gas Fuel
-------
Front Wall, meters
NOV Reduction Profile
0
o o
-co it
Front Wall, meters
NH3 Slip Distribution Profile
Figure 5. NOX Reduction and NH3 Slip Distribution Profiles.
CFD Modeling, Full Load, Natural Gas, Case 3
-------
0.0 0.5 1.0 1.5 2.0 2.5 3.0 3.5 4.0
Molar tWO Ratio
Figure 6. NOX Reduction and NH3 Slip versus Molar N/NO Ratio.
CFD Modeling, Full Load, Natural Gas, Case 3
<+u
30
5?
Reduction,
ro
O
X
O
z
10
n [
D NOX Reduction Cl5
• NH3 Slip Q E
n n DEJ
D tfl Q
a D a n
n D D '
n
*
n
jj..^_ .. J — . — i — , — . — , — . i .... i .... i ....
0.0 0.5 1.0 1.5 2.0
Molar N/NO Ratio
2.5
Figure 7. NOX Removal and NH3 Slip versus Molar N/NO Ratio.
Tchaikovsky Unit 2, 400 TPH, Natural Gas
-------
50
40
o 30
cc
X 20
10
NOx Reduction
D Level D
Q Levels D,E
D Level E
NH3 Slip
A Level D
> Levels D,E
< Level E
n
100
80
60 |
3
40 5
-20
0.0
0.5
c
o
t)
13
-D
O>
CC
1.0 1.5
Molar N/NO Ratio
2.0
2.5
Figure 8. NOX Removal and NH3 Slip versus Molar N/NO Ratio.
Tchaikovsky Unit 1, 360 TPH, Natural Gas
1 2
3 4
Rear
Front DePth
5 6
7 8
9 10
11 12
Sample Port
Figure 9. NOX Reduction Profile. Tchaikovsky Unit 1, Level D Injection
360 TPH, Natural Gas, N/NO = 2.0
-------
c I
O Q.
cc
x
O
2:
60
50 -
40
30
Unit 1 Data
•
A
deNOx
NH3
CFD Model
O deNOx
A NH3
0.5
1.0 1.5
Molar N/NO Ratio
2.0
2.5
Figure 10. Comparison of CFD and Field Test Results
-------
DESIGN AND CHARACTERIZATION OF A
UREA-BASED SNCR SYSTEM FOR A UTILITY BOILER
Victor Ciarlante, P.E.
Research-Cottrell
Branchburg, NJ 08876
Carlos E. Romero, Ph.D.
Energy Research Center
Lehigh University
Bethlehem, PA 18015
Abstract
As part of an overall compliance plan to reduce NOX emissions, Delmarva Power has installed a
urea-based Selective Non-Catalytic Reduction (SNCR) system on their Edge Moor Unit #3
boiler. The unit is a tangentially fired pulverized coal boiler, nominally rated at 84 MWe. The
baseline NOX for Edge Moor #3 was 0.54 Ib/MMBTU. Under Phase I of the 1990 Clean Air Act
Amendments (CAAA) and, as stipulated by Delaware law, the NOX emission limit was 0.38
Ib/MMBTU.
Research-Cottrell provided design engineering and start-up services for the SNCR system, while
Lehigh University's Energy Research Center conducted further system characterization. Start-up
testing demonstrated that the system can achieve compliance levels across the load range of 45%
to 100% MCR while maintaining ammonia slip below 15 ppmv. Characterization testing
demonstrated the capacity of the system for NOX reductions for the range of boiler conditions
typical of full load operation. A theoretical model was used to help in the interpretation of
characterization test results and for fine-tuning of the system.
Introduction
Edge Moor Unit #3 (EM3) is a Combustion Engineering tangentially-fired, balanced draft boiler
with a nominal rating of 84 MWC that was first placed in service in 1954. The unit fires eastern
Presented at the EPRI-DOE-EPA Combined Utility Air Pollutant Control Symposium, Washington, DC,
August 1997
-------
bituminous coal as its primary fuel, but has the capability of firing oil #6 and natural gas backup
fuels, especially during the ozone season. A side view of the boiler is shown in Figure 1.
Delnurv* Power
Edge Moor Unit »3
Figure 1
Although Delaware's Reasonably Available Control Technology (RACT) requirements for
control of NOX called for installation of combustion control, primarily Low-NOx Burners (LNB),
Deknarva Power decided to install a SNCR system on a demonstration basis to evaluate its
performance in preparation for more stringent emission limits required for CAAA's Phase II.
Urea-based SNCR is a post-combustion process that reduces NOX by injecting a controlled
amount of an aqueous urea reagent into the boiler. Once the urea evaporates and comes into
contact with the NOX in the flue gas, gas-phase chemical reactions take place that convert the
nitrogen oxides into harmless molecular nitrogen, water and carbon dioxide. A narrow
temperature range (or "window") of about 500°F is required for an effective NOX reduction.
-------
The design of an SNCR system takes into account several factors, including boiler conditions,
process requirements and site specific constraints. Boiler conditions that impact SNCR design
include fuel properties, load range, capacity factor, furnace geometry, and conditions of the flue
gas, such as temperature and gas velocity profiles and chemical constituents. All of these
conditions are considered during the System Design Phase, which is a necessary prerequisite to
obtaining an effective design. In this phase, a combination of Computational Fluid Dynamics
(CFD) and Chemical Kinetics Models is used, incorporating specialized droplet evaporation and
chemical reaction routines. The system is further designed to allow some system control tuning
during the Start-up Phase.
Theoretically, NOX reduction in excess of 60% is achievable under ideal conditions of residence
time and flue gas temperatures in the range of 1600 to 2100°F. Higher NOX reductions are
limited by un-reacted NH3 (ammonia slip) and they vary depending on boiler types and fuels
[1,2]. hi a full scale boiler or furnace, the reduction of NOX by urea injection is dependent on a
number of boiler specific parameters [3]. In particular, these include the stoichiometry of the
chemical process (baseline NOX and reagent injection ratios), carbon monoxide and excess O2
levels, and the temperature and velocity fields immediately downstream of the point of injection,
since these parameters determine mixing requirements, droplet evaporation rates, and the actual
temperature history and residence time available for reaction. Hence, it is important to
characterize the SNCR system for the typical range of boiler operating conditions.
This paper describes the System Design Phase and system provided and it discusses the results of
the Start-up Phase testing and further system characterization. The purpose of the
Characterization Phase was to obtain information which could be used for fine-tuning the SNCR
system and to determine the sensitivity of the SNCR system performance to changes in key
boiler parameters. A computer model of the SNCR chemical process was validated with
experimental data from this boiler. The model was used to complement the understanding of the
experimental results.
SNCR System Design
Process Modeling
Process simulation is based on combustion and SNCR process computer models. Boiler
dimensions, boiler configuration, and fuel composition analysis are some of the inputs used for
the setup of the models. Temperature measurements of EM3's flue gas were taken at various
boiler depths and elevations to provide validation process data for the model results.
Simultaneous readings of NOX, CO and O2 levels at some of these elevations were also obtained.
NOX reduction model results were generated for four boiler operating conditions:
Case 1: 84 MWe Firing Coal - NOX baseline = 0.59 Ib/MMBTU, O2 = 4.1% by vol.,
CO=100ppmv.
-------
Case 2: 81 MWe Firing Coal and #6 Oil - NO, baseline = 0.61 Ib/MMBTU, O2 =
4.6% by vol., CO = 140 ppmv.
Case 3: 66 MWe Firing Coal - NOX baseline = 0.59 Ib/MMBTU, O2 = 6.3% by vol.,
CO = 80 ppmv.
Case 4: 39 MWe Firing Coal - NOX baseline = 0.68 Ib/MMBTU, O2 = 7.1 % by vol.,
CO = 40 ppmv.
FLOW DIAGRAM
AIR COMPRESSOR
STORAGE TANK
Delraarva Power
Edge Moor Unit #3
- -n
- -< i
- -i-]
- -; •
• *i
- ••
REORC-LOOP
TO FUTURE
' ~ ~ + UNIT #4 REAGENT
FROM FUTURE ' ™'•* I
UNIT B4 REAGENT I — . — . . '
STORAGE TANK CJRC MODULE
1
DIST MODULE
D1ST.. MODULE
DIST MODULE
TO EACH ZONE 3
INJECTOR (5)
TO EACH ZONE 2
INJECTOR (6)
TO EACH ZONE I
INJECTOR (4)
[RETRACT ABLE]
Research-Cottrell, Inc.
Figure 2
Temperature/emissions mapping and process modeling, along with subsequent engineering
began in November 1994. Modeling results were used to predict the performance of the SNCR
process and identify an optimum temperature range for reagent injection. The results of the
-------
modeling effort also provided a basis for selecting an injection strategy that provided the best
opportunity for maximum NOX reduction and niinimuni ammonia slip. Three levels or "zones"
of injection were chosen, with each zone located at a particular boiler elevation, allowing the
system to track the desired temperature window at different boiler loads. For each case, the
target NOX was set at 0.38 Ib/MMBTU and ammonia slip was limited to 15 ppmv.
Equipment Design and System Control
The major system components, as shown in Figure 2, include one truck unloading station, one
reagent storage tank, one circulation module and recirculation loop, three metering modules,
three distribution modules, one air compressor, and fifteen injectors. A 50% by weight aqueous
urea solution is delivered by truck and loaded into the storage tank. At this concentration, the
reagent must be kept at a temperature above 60°F to prevent crystallization. Most of the heat
required is supplied by an in-line heater, which is integral to the circulation module. A
recirculation loop also helps to prevent dead legs and maintains the reagent well mixed in the
storage tank. The equipment was delivered to the plant in September 1995 and installation (by
Others) occurred during a scheduled four-week outage beginning in October and extending into
November 1995.
At the metering modules, the urea is regulated and then diluted with water to allow greater
coverage across the boiler. Two of the metering modules provide independent control of each
injection zone and operate normally in manual or automatic mode; the third is on stand by. The
distribution modules regulate the flow of diluted urea and air to each injector. Flow rates and
pressures are set manually during start-up. The injectors are dual fluid type, whereby low
pressure service quality air (< 100 psig), supplied by a dedicated air compressor, is used to
atomize the diluted urea and cool the injectors.
Zones 2 and 3 contain six and five wall-mounted injectors, respectively, and are intended for
operation at mid- and high-boiler load (Cases 1,2 and 3). Zone 1 has four retractable injectors
for use at low-load (Case 4). These injectors are automatically retracted from the boiler via a
pneumatic mechanism during higher loads to protect and extend the life of the injectors. Several
new boiler ports were added to accommodate insertion of the injectors.
The amount of urea sent to the injectors is regulated by means of a de-driven pump, which
receives a signal from a local control panel during manual mode or from the existing DCS when
the remote automatic mode is selected. Based on boiler parameters, i.e., fuel and steam flow, the
urea pumping rate is set to a predefined value to meet the required target NOX emission. Then, to
adjust and minimize chemical consumption, a NOX signal from the CEM's is used as a feed
forward signal to initiate a PID loop.
-------
Unique Project Design Highlights
Due to the demonstration nature of the project and in an attempt to fully understand and
maximize the benefits of the technology, Delmarva Power took a more experimental approach in
the design of the system. As a result, several unique benefits were derived:
• Vendor Supplied A/E Tasks - Delmarva used Research-Cottrell for several A/E key tasks
related to the SNCR system. These included location of the unloading station and storage
tank and rest of equipment and interface points, determination of power feed tie-ins,
identification of suitable ammonia monitors, and updating of existing drawings.
• Modifications of Existing CEM's - In recognition of the possible plugging of CEM
sample probes by ammonia slip from the SNCR system, KVB (the supplier of the CEM
system for this boiler) re-engineered the CEM's to assure proper operation in this
situation.
• Application of Mechanically Attached Fittings - Victaulic Pressfit fittings were installed
on all field interconnect air, water, and diluted urea piping instead of welded fittings to
reduce construction costs.
SNCR System Start-Up Testing
As part of the start-up effort, undertaken in December 1995, Research-Cottrell tested the SNCR
system at the four cases identified earlier. For each case, a suitable chemical pumping rate was
chosen based on process modeling results to achieve the desired NOX reduction and ammonia
slip. Then, by varying zone of injection, injector orientation, spray patterns, and urea flow, the
system was tuned so as to minimize reagent consumption. Figures 3, 4 and 5 show the results of
this system tuning for the low-, mid- and high-load cases, respectively, while burning coal (a
small proportion of landfill gas was co-fired).
For the low-load case (39 MWe), urea flows in excess of 60 gph are required to reduce NOX
below the target level (0.38 Ib/MMBTU). Higher urea flows are required for the
mid- and high-loads (66 and 88 MWC respectively). Urea flows as high as 110 and 115 gph,
respectively, are needed to achieve target NOX levels at these loads. The similarity in urea
requirements for these two loads is due to the differences in baseline NOX resulting from the test
conditions, which resulted in 37% NOX reduction at mid-load and 44% NOX reduction at high-
load. On an equal NOX baseline basis, the urea requirements for high loads should be higher than
for mid loads.
-------
SNCR PERFORMANCE
(LOW LOAD-COAL)
Urea Row (gph)
• 6.0 g
Figure 3
SNCR PERFORMANCE
(MID LOAD - COAL)
Figure 4
-------
SNCR PERFORMANCE
(HIGH LOAD-COAL)
Urea Row (gph|
Figure 5
Given the values of NSR's (the ratio of the actual mole ratio of urea to uncontrolled NOX divided
by the Stochiometric ratio for theoretically 100% NOX reduction and 100% chemical utilization)
used, ammonia slip is not a concern at low-loads. Ammonia levels in the stack start to increase
at NSR's higher than 1.20. However, for all NSR's tested, the ammonia slip was below the 15
ppmv limit.
The results of the Start-up Phase indicate the importance of understanding the theoretical and
physical performance of the SNCR system in terms of NOX reduction and ammonia slip, since
this understanding can be used to permit the system to operate within the prescribed ranges of
NOX reduction under off-design conditions. SNCR system characterization testing followed the
testing done under the Start-up Phase. To help in the understanding of the test results, a
computer code developed by Lehigh University and Research-Cottrell was used. This code
(referred to as the chemical kinetics model, CKM) is a homogeneous gas phase computer model
of the SNCR chemical process and it can be used for the prediction of flue gas NOX reduction and
NH3 slip.
SNCR Computer Model
The SNCR numerical model describes an ideal, one-dimensional plug flow, with the temperature
history approximated by a linear profile. The model further assumes the reagent has already
been atomized and the droplets are fully evaporated. Urea decomposition was modeled as an
instantaneous one-step breakdown to NH3 and HNCO. The physics of this problem is defined by
a set of ordinary differential equations. The chemistry used is a detailed chemical kinetics
-------
scheme composed of 105 reversible reactions and 24 species. This scheme includes the
ammonia and isocyanic acid oxidation reactions and the wet-CO oxidation reactions.
Figure 6 shows the comparison between reduced NO^ measurements and the model results for
this unit operated at full load conditions. The data corresponds to urea single zone injection.
Input to the theoretical model included CEM measurements of NOX, CO, C02, and O,; urea
injection flow; and temperature information obtained with an optical temperature probe. No
measurable ammonia slip occurred with the temperatures and NSR's used in these
measurements.
Chemical Kinetics Model (CKM) Validation
Delmarva Power; Edge Moor Unit #3
Full Load; 100% Coal; 2.6% O2; 10 deg. BTA;
2340 F Release Temp.; Single Zone Inj.
Figure 6
SNCR System Characterization
In order to characterize the performance of the system under typical ranges of boiler operating
conditions and to obtain more performance data which could be used for system fine-tuning,
Delmarva initiated an ongoing data gathering program. As part of this effort Lehigh University's
Energy Research Center performed theoretical calculations and carried out field tests to
characterize the factors affecting performance of the SNCR system. Since Delmarva intends to
operate the unit's SNCR capability only during the summer months, the system characterization
was performed after the first summer of operation (1996). Further system characterization is
expected to continue in the following summers. Future characterization tests will evaluate the
long term impacts of the SNCR system on boiler operation, such as possible pluggage of
downstream heat exchange surfaces and CEM's due to ammonium bisulfate, fly ash
contamination due to ammonia slip and any changes in unit heat rate.
-------
Injection Zones Characterization
Testing for the Characterization Phase was done at full load. At full load the temperature
distribution in the boiler is such that only injectors in Zone 2 and 3 are effective because they are
located at elevations where gas temperatures are in the proper range for significant NOX
reduction. Testing at different urea flow rates for each individual zone and for both zones in
combination indicated that the reduction of NOS emissions and production of ammonia slip per
unit of urea flow is higher for Zone 3 (located at a higher elevation) than for Zone 2.
Figure 7 shows the percent NOX reduction and NH3 slip versus the fraction of the total urea flow
injected into Zone 2 for a total reagent flow of 80 gph. The graph shows that the NOX reduction
and ammonia slip increases as the fraction of urea introduced through Zone 2 is lowered from 1.0
to 0.4. Additional test results and computer simulation results showed that the NOS reduction
eventually decreases if the fraction of Zone 2 urea flow is lowered beneath an optimal level, the
exact value of which is dependent upon operating conditions in the boiler (Figure 8). This is the
consequence of the lower gas temperatures associated with the downstream zone (Zone 3).
Furthermore, for both zones, the NOX reduction is more efficient during the early stages of urea
injection. That is, the slope of the NOX versus urea curve is steeper at lower urea flow rates,
NOX Reduction and NH3 Slip vs Urea Flow Distribution
80 GPH Total Reagent Row
Full Load; 100% Coal; 3.0% O^ 6 deg BTA
40
£
| 35
3
n
K 30
O
z
25
20
"*" ^ ^ * " " " • - , NOx
^^^ <
^^-^
NH3 ^ ^
•x.
•s.
X,
02 0.4 06 0.8
Zone 2/Total Urea Flow Ratio
9
E
a
_a
a.
6 to
I
z
>
3 <
• 0
Figure 7
flattening out as higher rates of reagent are introduced. The optimum division of flow between
zones would depend on the needed NOX reduction and the required ammonia slip limit.
10
-------
CKM Predicted NOx Reduction vs Urea Flow Split
Edge Moor S3; Full Load; 100% Coal
80 gph Urea
60 gph
40 gph
Fraction of Urea Row to Zona 2
Figure 8
Numerical model results indicate that a partition of about 40% urea to Zone 2 and 60 percent to
Zone 3 maximizes NOX reduction at full load (Figure 8).
Affect of Furnace Parameters on SNCR Performance
Additional SNCR system characterization was performed at full load to determine the sensitivity
of the SNCR process to furnace conditions represented by two key boiler parameters: burner tilt
angle and furnace O2. Figure 9 shows the effect of lowering burner tilt at fixed economizer O2,
from +10 degrees to 0 degrees (typical burner tilt range for full load to maintain steam
temperatures), on the reduced NOX level versus urea flow rate curve. Urea injection was limited
to Zone 2. The results indicate that the trending in SNCR system performance is similar for both
conditions, however, better performance is obtained at lower burner tilt angles. For all tests
ammonia slip was less than 5 ppmv.
The computer code was used to interpret the experimental results. Lower burner tilt angles are
associated with lower baseline NOX levels and lower gas temperatures. Furnace gas temperature
measurements in the vicinity of the injection zone, carried out with the optical probe, indicated
an increase in average (line-of-sight averages across the furnace plane) gas temperature of
11
-------
approximately 80°F by raising the burner tilt angle 10 degrees. Figure 10 shows the results of
the computer simulation in terms of NOX reduction and ammonia slip for different values of NSR
and gas temperature. This system was designed to operate on the "right side" of the NOX
Reduced NOx Level vs Urea Flow Rate
Edge Moor*3; Full Load; 100% Coal
Effect of Burner Till
a Flow Rate (gpri)
Figure 9
Effect of NSR on NOx Reduction and
Ammonia Slip vs. Temperature Curve
100
350
0
1400 1600 1800 2000 2200 2400
Gas Temperature [F]
Figure 10
12
-------
reduction versus temperature curve (which is the region where ammonia slip is negligible). On
this side of the curve, better performance is achieved at relatively lower temperatures.
Additionally, for a constant urea flow rate, a lower baseline NOX will translate in a higher NSR
which also results in better SNCR system performance.
Figure 11 shows the effect of two levels of furnace oxygen (indicated by economizer O2 levels of
3.2 and 2.6%) on reduced NOX for fixed burner tilt angle (+10°). This range of O2 is typical of
the variation of O2, at full load. Again, only urea injection to Zone 2 was considered. The
trending in SNCR system performance is similar for both O2 levels in terms of the amount of
urea required to achieve a particular NO^ level.
Reduced NOx Level vs Urea Flow Rate
Edge Moor #3; Full Load; 100% Coal
Effect of Furnace 02
-, 0.55
| 0.50
I 0.45
5
2 0.40
0.35
0.30
Urea Flow Rate (gph)
Figure 11
The results indicate that the SNCR system performance remains essentially unaffected by O2
variations at full load. Although the baseline NOX was reduced by lowering O2, the decrease of
oxygen in the furnace has an effect on the furnace gas temperature distribution which offsets the
NOX reduction. The CKM was used to investigate the effect of reducing O2 in the SNCR
chemical process. Figure 12 shows the sensitivity of the NOX reduction versus gas temperature
curve to furnace O2. This indicates excess O2 is a second order variable. An increase in O2, from
0.8 to 6.6%, did not impact the NOX reduction performance at temperatures below 2000°F. At
temperatures greater than 2000°F, the change in 02 considered (3.2 to 2.6%) has a negligible
impact on the SNCR chemical process. For this set of tests, ammonia slip was also less than
5 ppmv.
13
-------
Effect of O2 on NOX Reduction vs.
Temperature Curve
100
80
NOx 60
Red.
(%) 40
20
1400 1600 1800 2000 2200 2400
Gas Temperature [F]
Figure 12
Summary
Installation and start-up tests of an SNCR system at Delmarva Power's 84 MWe Edge Moor Unit
#3 have demonstrated the capability of the system to reduce NOX to required Delaware RACT
compliance levels for tangentially-fired utility boilers. For coal-firing and a load range of 45% to
100% MCR, ammonia slip remained below 15 ppmv. Start-up and Characterization tests
provided useful information to be used for fine-tuning the urea-based SNCR NOX control system.
A theoretical model was used to help in the interpretation of characterization test results and for
analysis of several parameters which impact process performance. Results of this analysis and
testing demonstrated the capacity of the system to achieve NC\ reductions for the range of boiler
conditions typical of full load operation.
14
-------
Acknowledgments
The work presented in this paper was funded by Delmarva Power. The authors would like to
acknowledge the cooperation of the Delmarva's 1995 Clean Air Compliance Project Team, led
by D. Skedzielewski and M. Zoccola, the engineering effort of R. Casill, J. Staudt, S. Kozma and
B. Patel at Research-Cottrell, and the design-phase process modeling assistance by Nalco Fuel
Tech. Lehigh University gratefully acknowledges the technical direction provided by
M. D'Agostini and E. Levy of the Energy Research Center.
References
[1] Teixeira, D., Lin, C., Muzio, L., and Jones, D. Joint EPPJ/EPA Symposium on
Stationary NO* Control, Bal Harbor, FL (May 1993).
[2] Joal, M., Lauridsen, T., and Dam-Johansen, K., Environmental Progr. 11(4), p.296-301
(1990).
[3] Romero, C.E. and Ciarlante, V., Sensitivity of the SNCR Process to Furnace Process
Variables, presented at the First Annual DOE Conference on SCR & SNCR, Pittsburgh,
Pennsylvania (May 1997).
15
-------
Enhanced NOxOUT® Control
Salem Harbor
Unit #3
EPRI-DOE-EPA Combined Utility Air Pollutant Control Symposium
August 25-29, 1997
Washington Hilton & Towers
Washington, DC
Prepared by:
R. Afonso; A. Sload; D. Miles
New England Power Company
S. Johnson
Quinapoxet Engineering
J.H.O'Leary
Nalco Fuel Tech
Abstract
The NOxOUT System installed on Unit #3 at New England Power's Salem Harbor station has
undergone a controls upgrade. The changes resulted in significantly lower reagent
consumption and lower ammonia slip. This paper discusses this latest control approach and the
benefits realized.
In order to implement the control changes, new hardware and new software were installed with
minimal intrusion in the plant routine operation. The new hardware includes state-of-the art
ammonia and temperature monitors, a pressure control on the fluid injectors and computer
upgrade. New software takes advantage of furnace temperature enhancing the original steam
feed-forward signal, and takes advantage of ammonia measurements enhancing in-furnace
chemical distribution. Furnace temperature is also utilized to minimize ammonia spiking during
rapid unit transients and to guide operators on improved sootblower usage.
This successful upgrade has provided NEP's operators the ability to operate with significantly
lower reagent consumption (a savings of over 50%).
-------
1. INTRODUCTION
New England Power (NEP) elected in 1993 to reduce NOx emissions at their
Salem Harbor Station, Units #1, #2, and #3 through the installation of Nalco Fuel
Tech's (NFT) NOxOUT SNCR technology. These units are permitted on the
basis of individually maintaining a 24-hour calendar day average NOx level of
0.33 Ibs/MMBtu. The SNCR equipment installation and start-up on Unit #3
occurred in the third quarter of 1993 and was followed by installation and start-
up of low-NOx burners with separated overfire air. Combined, these
technologies have demonstrated consistent compliance with mandated
controlled NOx level while firing coal from a wide variety of sources.
In 1994, NEP investigated the applicability of continuous temperature and
ammonia emission monitors for the ultimate purpose of integrating these signals
into the SNCR control logic to improve overall SNCR system performance and
lower operating cost. Spectrum Diagnostix, Inc. (SDx), a BOVAR company,
manufactures the temperature and ammonia monitor equipment. Testing by SDx
with a SpectraTemp® continuous FEGT monitor and a SpectraScan continuous
ammonia monitors on units #1 yielded the following conclusions:
• Reagent was injected at the upper end of the temperature window for
NOx reduction at full load resulting in low utilization (15 to 22%)
especially with a dirty furnace (at the tail end of a sootblowing cycle)
• At low loads, reagent was injected at lower temperatures with better
residence times resulting in utilization which was higher (30 to 35%)
• Ammonia slip was prevalent during load transients where
temperatures tended to be low and the system control was slow to call
for decreases in reagent flow
• Furnace sootblowing decreased FEGT and improved reagent
utilization at all loads
• Keeping FEGT below 2020°F at full load increased reagent utilization
to 26%
• Keeping FEGT below 1880°F at 70% load increased reagent
utilization to 42%
NEP agreed to a commercial upgrade of system control for Salem Harbor Unit
#3 provided a payback based on reagent savings would be demonstrated. The
principal objectives of the enhanced NOxOUT System control were to allow
Salem Harbor Unit #3 to meet its NOx emission limit of 0.33 Ib/MMBtu while
reducing the consumption of NOxOUT reagent.
NEP organized a team to execute the upgrade and demonstrate improvements.
Four test series were conducted as shown in Table 1.
-------
Table 1. Purpose of Each Test Series
Test Series
Baseline Tests
Control System
Optimization
Control System
Demonstration
Final System Tuning
Purpose
Determine as-found NOxOUT consumption, NH3 slip,
and flyash ammonia content relative to boiler
operation.
Tune the new system control to minimize reagent
consumption in response to temperature and NH3
Quantify improvements in NOxOUT consumption, NH3
slip, and flyash contamination as well as control system
reliability over a one-month period
Incorporate any further system improvements mutually
approved by NFT and NEP as a result of the
demonstration.
Nalco Fuel Tech (NFT) designed modifications to the original NOxOUT System
and worked with NEP to install the new equipment and software. Quinapoxet
Solutions worked with NEP and NFT to plan the program, collect and analyze
data, and interpret results. The program objectives were to quantify chemical
consumption and document boiler and NOxOUT System performance before
(baseline) and after (demonstration) instruments and controls were upgraded,
especially cost of NOx control. The program included 24 days of baseline
observing performance of original system control and 35 days of demonstration.
2. BASELINE TEST
Salem Harbor Unit 3 is a front fired unit rated at 155 MW. The unit has been
equipped with four levels of NOxOUT injectors as well as low-NOx burners and
overfire air to control NOx emissions below 0.33 Ib/MMBtu. A complete division
wall front to back divides the furnace into two chambers. Since low-NOx burners
were installed, the unit has had difficulty achieving designed reheat steam
temperature due to lower furnace exit gas temperature. Furnace sootblowing
was historically performed when primary superheated steam temperatures
became high enough to cause tube overheating (primary steam temperature
approaching 950°F). Prior to the test program, furnace sootblowers were
operated only a few times per week.
2.1 Baseline Reagen t Consumption
Historical data taken during 1995 and 1996 indicated that reagent consumption
varied significantly from day to day, averaging 70 to 200 GPH. Average reagent
consumption for Unit #3 is about 125 GPH. Reagent consumption on the first
day of baseline testing was about 112 GPH averaged over the 24-hour period.
On the second day, the operators blew all 35 furnace sootblowers, and reagent
-------
consumption at full load immediately dropped to about 62 GPH. As a result,
operators changed their furnace sootblowing practices during the rest of the
baseline test in order to better control NOx.
Table 2. Historical and baseline sootblower operations
Prior to 30 September
Used IR sootblowers 1-2 times per
week
Blew all 36 sootblowers in rapid
succession
Purpose of sootblowing was to reduce
primary superheat temperature or
improve Low-NOx burner performance
Baseline
Used IR sootblowers nearly every day
Usually blew 8 sootblowers at a time
Purpose of sootblowing was to reduce
reagent consumption
Historical data was used to determine reagent consumption with the original
control system. During the baseline, on days with active sootblowing, the
reagent consumption averaged around 70 to 80 GPH while on days with no
sootblowing, reagent consumption was 100 to 110 GPH. Historically
consumption averaged 125 GPH.
The plant began collecting baseline data on 30 September 1996 at 3:00 p.m.
Figure 1 shows hourly average load, NOx emissions, and reagent consumption
for the first day and a half of baseline testing. Full load reagent consumption
started at 120 GPH, but steadily climbed to nearly 160 GPH by mid-morning on 1
October 1996. At 80% load, reagent consumption dropped to a range between
65 and 75 GPH. Baseline testing changed focus after 9:50 a.m. on 1 October
1996. All 36 furnace IR sootblowers were blown in rapid succession. Removing
slag from the furnace walls had three immediate effects:
1. Reagent flow decreased immediately to 65 GPH, the minimum flow
allowed by the original system control. Reagent flow could have gone
even lower as evidenced by a subsequent decrease in NOx from 0.31
to 0.29 Ib/MMBtu.
2. Superheat attemperator spray flows decreased from 52,000 to 13,500
Ib/hr, while primary SH steam temperature decreased from 908°F to
850°F Secondary SH steam temperature remained at 1000°F
3. North side reheat steam temperature which is normally 1000°F ranged
from 983°F to 996°F for the next 8 hours after sootblowing. The
south side reheat steam temperature remained at 1000°F
-------
These results confirm a loss of reheat temperature with IR sootblowing and a
major reduction in reagent consumption. Also notice how loss of air staging of
the low NOx burners affects reagent flow.
40 --
20 --
Reduced OFA ^
l-^ Began IR
I Sboffilowing
1
-• 0.15 -
-• o.i
-• 0.05
-t—I-
Hour
Figure 1. Baseline data, 1 October 1996
No furnace wall sootblowing was performed for the next 4 days. As shown in
Figure 2, reagent consumption increased gradually over time. Twenty-four
hours after sootblowing, reagent flows were up to 100 GPH at full load. These
data offer great promise for improved NOx control cost-effectiveness with
minimal impact on boiler performance. Extensive furnace wall sootblowing
caused slight decrease in reheat steam temperature for a few hours, while the
reduction in reagent consumption persisted for days.
-------
100
60
40
20
NOx
GPH~
Uncontrolled NOx=0.43 Ib/MMBtu
0 -I — 1
0.25 3
£
o, I
A
0.15 -
0.1
0.05
0
Hour
Figure 2. Baseline data, 2 October 1996
The next IR sootblowing occurred on 6 October 1996. As indicated on Figure 3,
IR blowers 1 through 8 (lower burner elevation) were used during a load
increase. About 2 hours later, !R blowers 9 through 16 (upper burner elevation)
were blown after the unit had reached full load. Reheater long retractable
blowers were actuated in conjunction with each IR sootblowing cycle in order to
maximize heat transfer to the reheat steam and thus minimize any decrease in
reheat steam temperature.
160
100 --•
20
BlewIR 1-8
NOx'
GPH
BlewIR 9-16 (No affect
"dn'RH" steam" Tempe'raTufe")"
H 1 1 1 1-
0.4
0.35
0.3
0.25
02
0.15
0.1
0.05
0
Hour
Figure 3. Baseline data, 6 October 1996 (Sunday)
-------
The results of this particular sootblowing test was:
1. Reagent usage was reduced to 70 GPH.
2. No effect on RH steam temperature.
From 6 October 1996 on, the Salem Harbor operators began to use sootblowing
to minimize reagent consumption. Furnace sootblowing was performed 25 times
over the next 18 days of baseline testing (as compared to 1 to 2 times per week
prior to baseline testing). As a result, reagent consumption remained low. For
the 24 days of baseline testing, the average daily chemical consumption was
about 2050 gallons per day (85 GPH).
2.2 Baseline Furnace Temperature
On 9 October 1996, one SpectraTemp continuous FEGT monitor was installed
on the north side of the furnace and connected to the NOxOUT data acquisition
system. The original system control was translated into the new system control,
so temperature was not yet integrated actively into the control scheme, however,
operators could now monitor north-side temperature and relate IR sootblowing to
temperature as well as NOxOUT System performance. Table 3 shows typical
furnace temperature and 02 profiles obtained by the plant. Higher temperatures
and lower 02 concentrations occur on the south side of the furnace both indicate
higher fuel flow to the south side burners.
Table 3. Unit #3 HVT Test (24 May 1996)
South North
Probe
Length
%O2
Temp,
°F
2
2.5
1990
4
2.6
2014
6
3.3
2022
8
3.8
2026
10
4.0
2039
12
3.1
2058
14
2.3
2041
16
2.5
1936
10
4.9
1915
8
6.0
1915
6
6.7
1880
4
7.3
1865
2
7.0
1900
Figure 4 shows the trend of instantaneous FEGT measurements with total steam
flow during the baseline tests. It can be seen that full load FEGT ranged from
1950°F to 2100°F with all mills in service. The time between sootblowing events
was the primary factor affecting the exit gas temperature. Baseline data were
reviewed to identify relationships between the furnace exit temperatures and
load, mills in service, and sootblowing. Transient performance was also studied.
Figure 4 shows a linear decrease in FEGT with load down to a total steam flow
of about 750,000 Ib/hr steam flow. Further load reduction requires removing a
mill (usually Mill 4 which feeds the bottom row of burners). At this transition
point, the furnace exit gas temperature increases dramatically.
-------
2100
2000 --
1900 -
63
f
1800 "
1700 --
1600
^ $•
1 4 Out Of R **/
vice ' • /
400
500
600
900
1000
1100
700 800
Steam Flow (Klb/HR)
Figure 4. Relationship between FEGT and Steam Flow, 18 to 24 October, 1996
Table 4. Baseline sootblower observations (high load)
Main Steam
(Klb/Hr)
1040-1045
1045-1050
1037-1041
1025-1027
1027-1030
1026-1037
1042-1049
1037-1045
1026-1040
Auxiliary
(Klb/Hr)
7.5
12.4
12.5
12.5
27
28
0
0
0
Total Steam
(Klb/Hr)
1048-1053
1057-1062
1050-1054
1038-1040
1054-1057
1054-1065
1042-1049
1037-1045
1026-1040
FEGT, °F
2006-2052
1985-2029
2088-2098
2039-2047
2060-2074
2041-2067
1951-1988
1976-2015
2020-2067
Comments
Clean Furnace
Clean Furnace
No Sootblowing for 24 Hr
No Sootblowing for 35 Hr
No Sootblowing for 48 Hr
No Sootblowing for 60 Hr
Clean Furnace
Clean Furnace
No Sootblowing all Day
Note: Each data entry above represents a minimum 2 hours operation at 154 to
156MWg
2.3 Effects of Mill Configurations
When a mill is removed from service, the excess air level tends to be higher than
when firing all four mills. This is because air flow to any burners taken out of
service is kept at a minimum cooling flow. Higher windbox-to furnace differential
-------
pressure exist when a mill is out of service and air flows must be controlled
between the overfire air system and firing burners. Minimum boiler load during
baseline tests was about 70 MW (460,000 Ib/hr steam flow) with the upper three
mills in service.
The unit also operated for 3 days with Mill 2 (top row of burners) out of service
during baseline testing. Load was held relatively constant during top-mill-out
operation, ranging from 125 to 145 MW. Oil guns were inserted into the top
burners and lighted off for a few occasions when loads above 150 Mw were
required. Figure 5 shows hourly average load, NOx, FEGT, and reagent
consumption during top-mill-out operation. Note that NOx emissions and
reagent consumption were very low.
2300
2100 -- m ft-- 60
700
25
.10 -8 -6 -4 -2 0
10 12 14 16
Hour
Figure 5. Baseline data, hourly averages, 14 and 15 October, 1996 (top mill
OOS)
Figure 5 shows excellent NOx control. NOx remained well below setpoint most
times. Reagent flow was at minimum and NOx emissions averaged less the 0.3
Ib/MMBtu for the 3-day period (0.27 Ib/MMBTu on 13 and 14 October 1996; 0.30
Ib/MMBtu on 15 October 1996).
2.4 Transient Load Operation
The first generation NOxOUT System control was optimized without benefit of
more sootblowing performed during this baseline study. The control of reagent
flow was tuned for higher temperatures causing lower reagent utilization. NOx
-------
was therefore generally over-controlled during the baseline test period because
reagent flows characterized against the steam flow were too high.
Figure 6 shows a baseline load reduction transient where the unit goes from 150
to 70 MW (1000 to 480 Klb/hr steam flow) in about 1 hour. As load decreases,
furnace temperature also decreases. NOx emissions decrease from about 0.29
to 0.26 Ib/MMBtu indicating higher than desirable reagent flows throughout the
transient. NH3 slip increased from 12 ppm to about 40 ppm during a flush of the
upper level in service. NH3 slip dropped once the upper level of injection is
taken out of service. Approximately five minutes later, when Mill 4 is taken out of
service, temperature increases from 1890°F to 2000°F and the upper level of
injectors go back into service. As load is reduced further, NH3 slip climbs to 20
to 30 ppmv range.
220
Hour
Figures. Rapid load decrease, 20 October 1996. NOx is over-controlled.
Similar problems occurred when load was increased. This first generation
NOxOUT System control also had room for improvement in the area of tracking
load transients, especially at load points where the system control switches
reagent injection level. An instability occurred at a steam flow of 800,000 Ib/hr
(about 75% load), as shown in Figure 7 The original system control calls for a
shift in injector level at this load. During baseline tests, operators attempted to
hold 75% load for several hours during which time steam flow crossed 800,000
10
-------
Ib/hr a total of five times in 90 minutes. Each time that upper injectors went into
service, the system control called for a reagent flow increase (this due to the
minimum flow required for two metering modules). Since temperature was
constant at about 1900°F, the additional reagent flow reduced NOx and caused
a 30 to 40 ppmv ammonia spike. The system control was slow to correct for
these changes based on trimming the flow rate in response to NOx emissions.
This was improved upon by including furnace exit temperature in the reagent
flow control feedforward signal.
Hour
Figure 7. Instability occurred when going to second injector level
3. SYSTEM CONTROL CHANGES
NFT incorporated mechanical, instrument, and software changes into the
NOxOUT System control in order to reduce reagent consumption. These
changes were implemented without affecting unit operation (no loss of unit
availability).
The original NOxOUT System included four injection levels, two upper levels
which were serviced by one metering module, and two lower levels which were
serviced independently by a second metering module. Two adjacent injector
levels operate at all but the lowest load on Unit #3. The metering modules ran in
11
-------
parallel only when the middle two zones were in service. Also, each metering
module regulated liquid discharge pressure to injector levels with one (local
mechanical) pressure setting which was constant for all load ranges.
Wall injectors utilized at Salem Harbor rely on droplet trajectory into the furnace
for proper reagent distribution. These injectors are twin fluid injectors with air as
atomizing medium. Stored NOxOUT reagent is 50% by weight aqueous urea.
Water is used as a carrier of the reagent. Metering modules proportion the
proper amount of stored reagent and water, mix the two and forward them to
injectors at the proper liquid pressure for reagent distribution. At a given
atomizing pressure, liquid pressure at each injector may be lowered to provide
finer droplets with less furnace penetration and more upstream release of the
NOx-reducing reagent, or liquid pressure may be increased to provide coarser
droplets with more furnace penetration and more downstream release of the
NOx-reducing reagent. The original system provided one pressure setting
throughout the load range.
The original NOxOUT System utilized an Allen-Bradley PLC 500 series
controller with a computer interface for operators. The computer was
programmed using FactoryLink® software on a DOS computer. NFT designs the
NOxOUT System utilizing a modular approach. Unit #3 at Salem Harbor has a
dedicated Circulation Module which heats and circulates stored reagent from
storage to two Metering Modules and back to storage. The Metering Modules
proportion and mix reagent and water and forward the liquid at proper pressure
to Distribution Modules. Each wall injector is supplied with regulated air and
liquid from the Distribution Modules. The system incorporates motor operated
valves in field piping properly interconnecting the various modules.
Operators monitor and control the system by a set of active computer screens
specifically designed by NFT for this purpose using the FactoryLink program. In
the fully automated mode, the system starts and stops (injector levels are
brought into and out of service, pump and valves operate sequentially, reagent
feed rates set and track) automatically throughout the normal operating range of
boiler load as dictated by a set of characterized tables. Alarm functions,
safety/trip functions, and trending of important NOxOUT System data are also
performed. The operators modified NOx setpoints and/or reagent feed rates in
the normal course of automatic operation.
The NOxOUT Process is designed to minimally influence or dictate the operation
of the boiler. The number of mills in service determine which characterized table
is active in the automatic mode (there are four separately characterized 16-
segment tables: <2 mill, 2 mill, 3 mill, and 4 mill). Within each characterized 16-
segment table, boiler steam flow (turbine plus auxiliary flow) is used to determine
indirectly the furnace condition and select the default NOx setpoint, injector
levels in service, maximum reagent flow for each Metering Module, and minimum
12
-------
reagent flow for each Metering Module (this is often referred to as the feed-
forward signal). Within flow ranges, reagent feed rate is controlled up or down
by comparing actual NOx to active NOx setpoint utilizing a system
proportional/integral (PI) loop controller programmed in the Allen-Bradley (this is
often referred to as the feed-back signal). Tuning of these control features relies
on repeatable furnace temperature and velocity profiles with feed-forward
conditions. This repeatability is affected by sootblowing, fuel characteristics, mill
condition, burner and overfire air operation, excess air and cycle efficiency
among other things.
3.1 Mechanical Changes
Low-NOx burners with overfire air were installed and the uppermost level of
injectors taken permanently out of service. This left three injection levels. NFT
increased the NOxOUT System operating flexibility by piping a new remote-set
pressure regulator on each of two metering modules. Injector distribution piping
was modified by utilizing original upper injector zone valve to alternatively supply
the middle injection level such that two operating levels of injectors could be
supplied separately by parallel operating metering modules over the broadest
load range possible.
The original NOxOUT System control allowed only one liquid pressure setting
from each metering module to cover the entire load range through levels that the
metering module supplied. The modified system allowed control of spray
patterns at each level in service (with the upgraded pressure controls) and level-
to-level biasing of reagent treatment over a wider range of parallel metering
module operation (with the upgraded zone piping).
3.2 Added Instrumentation
SDx supplied two continuous optical temperature monitors (SpectraTemp) and
two online gas phase ammonia slip monitors (SpectraScan) used as new inputs
in the modified NOxOUT System control.
SpectraTemp detects radiation, primarily at visible wavelengths, emitted by the
ash particles passing through a narrow field of view. Ash particles are typically
smaller than SOpim diameter and thermally equilibrate with surrounding gas in
microseconds, accurately reflecting local gas temperature.
SpectraScan is a tunable diode laser-based instrument. The wavelength of
energy emitted by the laser is repeatedly and rapidly altered through one
absorption line for ammonia. When the laser is tuned to be off of the absorption
line, the transmitted power is higher than when it is on the line. Measurement of
relative amplitudes of off-line to on-line transmission yields a precise value of
ammonia gas concentration along the measurement path. The laser wavelength
13
-------
is scanned across the absorption line at a frequency of 1 MHz, causing the
detector to output a signal that contains a 2 MHz amplitude modulation. This
signal is processed to determine concentration of gas phase ammonia.
Two new temperature monitors were mounted at the furnace exit plane, one on
the north side and one on the south. NFT installed a pneumatically operated
knifegate isolation valve interlocked with instrument cooling air supply for
instrument protection. Two new ammonia monitors were installed in the ducts at
the economizer outlet, one on the north duct and one on the south. NEP
installed the vacuum system for these monitors.
3.3 Software Changes
The Allen-Bradley and FactoryLink software were modified and updated. New
software is designed to be operated in the same way as the original program but
with additional features activated and tuned in small, progressive increments.
This permitted commissioning updated system control with minimum impact on
operations. The unit was available for normal operation throughout installation
and start-up.
Segmented tables were replaced with characterized curves (four point and ten
point) based on a selectable feed-forward signal (steam flow and/or furnace exit
temperature) defining default NOx setpoint, injector level(s) in service, reagent
flow rates, and metering module default liquid discharge pressures.
Temperature inputs are also used in an innovative manner to allow for response
to rapid furnace transients (formerly resulting in over-treatment of NOx and
ammonia spikes). Droplet trajectory control of twin-fluid wall injectors is possible
by altering either atomizing air pressure or liquid pressure to injectors in service.
Air pressure is set at the distribution module and manually maintained within a
narrow operating range. Decreasing liquid pressure with fixed air pressure
causes a finer droplet pattern which does not penetrate and travel as far
downstream in furnace gases. By manipulating liquid pressure, precise and
immediate control of treatment zone distribution is possible. This is used as a
means of tracking rapid furnace transients. Should a sudden precipitous
reduction occur in furnace temperature (such as is encountered when a mill
trips, coal feed bridges, etc.), liquid pressure to injectors is immediately reduced,
raising temperature where reagent is released and minimizing liquid feed during
the short-lived transient. Change in temperature inputs (a derivative control) is
used to track these transients and automatically move liquid pressure as
required.
Ammonia inputs are used to bias reagent feed between two parallel operating
levels. This is why reconfiguration of interconnecting piping to injection levels
was important. A new system proportional/integral/derivative control was
14
-------
programmed into the Allen-Bradley. This control compares ammonia slip
setpoint to average ammonia measured and adjusts a small (up to approximately
5%) amount of reagent to the more downstream (cooler) zone of treatment. This
feature maximizes reagent utilization at a given furnace condition within
acceptable slip limits.
Other NOxOUT System control features were also added. Daily running NOx
average emission is calculated and utilized in two control features to minimize
reagent consumption by avoiding unnecessary over-treatment during a 24-hour
compliance period. The first feature takes advantage of the fact that less
reagent consumption is required at low load to achieve NOx than at high load.
The furnace is cooler at low load and therefore lower injector level(s) are
operated (this provides more residence time at lower temperatures for NOx
reduction reactions to occur attendant with higher reagent utilization and less
unwanted ammonia slip). This fact was taken into consideration in the original
system control by characterizing lower setpoints at lower loads. In the original
control, NOx setpoints were characterized below compliance level at all loads to
assure daily compliance with room for operating comfort regardless of load
profile. The updated system control tracks the running daily average thus taking
into account load profile and automatically relieves the setpoint at high load
operation while maintaining room for operating comfort. The second feature
takes advantage of the fact that, since NOx is kept below compliance level on
average throughout the day, at some point late in the day, compliance could be
met even without reagent flow. Once this compliance level is realized, the new
feature minimizes reagent flow.
Table 5. Software changes initiated prior to the demonstration period.
Original System Control
Initial flow rate selected by look-up
table based on steam flow and mills in
service.
Look-up table values for chemical flow
established previously for hot-furnace
operation.
Injector liquid pressure constant
Injector level selected by look-up
tables based on steam flow and mills in
service.
Upgraded System Control
Initial flow rate calculated based on
steam flow and mills in service (85-
90%)as well as temperature (10-15%).
10-point curves calculate reagent flow.
Values reduced after reviewing
sootblower performance data.
Injector pressure characterized with
load and varies with rate of FEGT
change. Rapid temperature decrease
results in reduced pressure (thus
reducing both chemical flow and
droplet size).
Option to let temperature drive injector
level was installed but not used.
15
-------
No automatic flow bias to different
injector elevations.
NOx setpoint selected automatically
based on steam flow and mill in
service. Operator override possible.
System response time set to avoid
instabilities given original inputs.
Bias reagent flow from higher elevation
(cooler) to lower elevation (hotter) if
NH3 values exceed setting
Same as before, but with a new screen
showing daily average NOx and
projected average if current setpoint is
maintained. Automated optional high
load setpoint and end-of-day trim
reduce reagent flow to minimum
values.
System response time shortened to
take advantage of new inputs.
4. DEMONSTRATION TEST
Demonstration tests started on 24 October 1996 and were completed on 27
November 1996, a total of 35 days. The first question to be answered is "How
boiler operation during the demonstration tests compare to the 24 days of
baseline testing?" Average daily boiler operating parameters for each test
period are compared in Table 6.
Table 6. Comparison of Boiler Operating Conditions
Boiler Operating
Parameter
Load, MW
Capacity Factor, %
CO, ppmv
Aux. Steam, Klb/hr
SH Spray, Klb/hr
RH Spray, Klb/hr
RH Temperature (S), °F
RH Temperature (N), °F
SSH Temperature (S), °F
SSH Temperature (N), °F
PSH Temperature (S), °F
PSH Temperature (N), °F
IR Sootblows/Day
NOx, Ib/MMBtu
Daily Average During
Baseline
144.2
93
35.6
8.7
25.4
0.5
997
990
1000
998
879
878
8.6
0.301
Daily Average During
Demonstration
140.9
91
40.5
13.1
15.0
0.4
993
979
999
994
870
861
23.0
0.316
Average daily load was about 3 MW higher during baseline, but this difference is
not considered significant. The capacity factor for Unit #3 was comparable for
both periods. Table 6 shows that steam cycle efficiency was negatively affected
16
-------
during the demonstration. Evidence for less efficient operation is the decrease
in steam temperatures. There are three reasons for lower steam temperatures:
1. The demonstration program included a few days of firing wet
coal.
2. The last 3 weeks of the demonstration period, Unit #3 operated
without one of its feedwater heaters.
3. Furnace IR sootblowing increased by nearly a factor of 3 during
the demonstration tests.
Of the three factors, the sootblowing increase had the largest effect on NOxOUT
System control. Table 7 shows the load profiles for each period. This table
shows that the unit operated at full load a little more frequently during the
demonstration test period, but also spent more time at minimum load. Since
reagent consumption is the highest at full load, the load profiles are not
expected to cause a measurable difference in NOxOUT System performance
between baseline and demonstration periods.
Table 7 Comparison of Boiler Load Profiles
LOAD RANGE
150-1 57 MW
140-1 49.9 MW
130-1 39.9 MW
120-1 29.9 MW
100-11 9.9 MW
70-99.9 MW
Total:
% TIME OPERATING
BASELINE
54.1
16.2
14.7
08.3
04.4
02.3
100.0
DEMONSTRATION
56.7
12.5
10.5
07.2
06.7
06.4
100.0
5. Conclusions
The demonstration quantified the amount of reagent that can be saved as a
result of using sootblowers more often and implementing new features designed
into the NOxOUT System controls. Figure 8 shows daily reagent consumption
for both baseline and demonstration periods as well as resulting yearly cost
associated with reagent usage. Baseline reagent consumption without benefit of
increased furnace sootblowing was about 2500 gal/day (an aveage of 104 GPH).
It could be argued that historical reagent usage is higher than 2500 gal/day;
consumption of 3000 gal/day is consistent with the Unit #3 share based on MW
rating.
17
-------
Increased sootblowing frequency allowed reagent consumption to be reduced to
1500 to 2000 gal/day before new system control features were implemented.
Average reagent consumption was around 1920 gal/day (80 GPH) during the
baseline, a reduction of at least 23% from the as-found NOxOUT System
performance and 36% from historical values. Low furnace temperatures and
system control improvements held reagent consumption between 650 and 1500
gal/day (27 to 62 GPH), a total reduction of 40 to 74%.
3000
2500 --
•.= 2000 --
c
5
1500 --
1000 --
S 500 -
- Historical Consumption Level
High FEGT
960
800 J
O
^
640 g
480 ft
O ,o
-- 160 £
10 20 30 40
Baseline and Demonstration Day
50
60
Figure 8. Daily reagent Consumption and Cost Estimate
Cost savings are achievable with minimal cost impact on boiler performance, but
the increase in sootblowing resulted in a decrease in reheat steam temperature,
especially on the north side of the unit. Occasional IR sootblowing for much of
the baseline accompanied by blowing the reheater using retractable IK blowers
prevented sudden decreases in reheat steam temperature.
Once NOxOUT System control modifications were implemented, response to
furnace transients was improved. Figure 9 shows data from a rapid load
reduction from about 130 MW to 70 MW within 30 minutes. Even though FEGT
spiked when Mill 4 was removed from service, NOx held setpoint value of 0.315
Ib/MMBtu except during a 15 minute span where even a reagent flow rate of 25
GPH was too high. These operating results are a vast improvement over
baseline results from a similar transient shown previously in Figure 10.
18
-------
200
Hour
Figure 9. Load reduction transient after control improvements, 23 & 24 October
1996.
The demonstration period showed additional savings resulting from less reagent
consumption. Average daily chemical consumption rates were controlled to the
30 to 60 GPH range for a total reduction in reagent consumption of 40% to 70%.
Thus, about half the benefit was due to more frequent sootblowing and half was
due to changes in the software and hardware. Reagent cost savings as a result
of the NOxOUT System control upgrade realized on any particular day is very
much dependent on the load profile. On average, the demonstrated savings
were in the range of 25% over the 30-day period studied. This savings is
estimated to be attributable as follows:
• Responsive feed forward signal upon load reductions
• Better reagent distribution control
• Relief of high load NOx setpoint
• End-of-day trim
Total:
5%
5%
10%
5%
25%
19
-------
Figure 10 shows an estimated stoichiometric ratio as a function of load and
reagent flow. It can be seen that at demonstrated reagent consumption rates,
the NSR stays around 0.6 for all loads. Figure 11 shows data in mills/KWh.
60 80 100
Reagent Flow (GPH)
Figure 10. Normalized stoichiometric ratio (NSR) for Salem Harbor Unit #3
0 20 40 60 80 100
Reagent Flow (GPH)
120
140
160
Figure 11. Reagent costs in mills/KWh
20
-------
Wei Chen, June 24, 1997 for DOE, EPRI, EPA Mega Symposium on August 26, 1997
Derivation and Application of A
Global SNCR Model in Maximizing NOX Reduction
Wei Chen, Nicole Phyfe,
Majed Toqan, Galen Richards, David Sloan, Mike Rini
ABB Product Development and Technology
Combustion Engineering, Inc.
2000 Day Hill Rd
Windsor, CT 06095
ABSTRACT
Selective Non-Catalytic Reduction (SNCR) is one of the methods used in NOX reduction
for utility boilers. It has been found that if the SNCR is not well designed, operated, or
installed, ammonia slips can be generated from SNCR. Although significant research has
been conducted, no universal tool exists to optimize SNCR processes.
At Laboratories (PD&T) a global SNCR chemical kinetic scheme was used. The global
HNCO kinetics was derived from about 327 elementary steps. The mechanism includes
thermal NOX, prompt NOX, fuel NOX, SNCR by ammonia, SNCR by urea, and rebuming.
A good agreement with the experimental data was found in calculating NOX reduction by
the mechanism. The main characteristics of SNCR process such as ammonia slip,
temperature window effects, and NOX generation were also demonstrated in the calculation.
The global SNCR mechanism was applied to identify the keys in optimizing SNCR
processes.
INTRODUCTION
Selective Non-Catalytic NOX Reduction (SNCR) is a process in which a chemical reagent is
injected into the post-flame region to reduce nitric oxides (NOX) emissions. The ammonia-
based SNCR process was first invented by Lyon (1978). EXXON and other companies
subsequently developed a similar process using urea (H2NCONH2) as the reagent. SNCR
processes have been widely implemented by industry as a low-cost NOX abatement
strategy. SNCR can be used in a stand-alone mode or in conjunction with other NO,.
control processes such as the SCR process.
Chemical kinetics and fluid dynamics are two key factors that influence the efficiency of the
SNCR process. The mixing rate of reagent with the exhaust gases plays an important role
in providing the molecular contact of ammonia or urea with NOX, and depending on the
local temperature and species concentrations; the chemical kinetics of SNCR results in a
certain amount of NO reduction. Therefore, the performance of SNCR processes strongly
relies on the amount of reagent injected, the location and number of injector nozzles and the
operating conditions of a boiler, such as load, fuel type, and excess-air level. The optimum
performance level of an SNCR process will vary with operating conditions. Computer
modeling of SNCR processes can be used to assist in the design of an SNCR system, and
ultimately, to provide information to optimize SNCR operation for a given injection
configuration.
-------
Wei Chen, June 24, 1997 for DOE, EPRI, EPA Mega Symposium on August 26, 1997
A global SNCR chemical kinetic scheme is developed in this paper and implemented to a
pilot scale facility (Figure 1). The global SNCR kinetic model was validated and applied to
the PD&T's Boiler Simulated Facility (BSF) (Figure 2). The global kinetic rate constants
of the ammonia-based SNCR process were obtained from the literature; three options of
ammonia global chemistry were available (De Soete 1973, Duo et al. 1992). For a urea-
based SNCR process, the decomposition of urea into ammonia and cyanuric acid (HNCO)
must also be considered. In other words, the global pathway of the urea-based SNCR
process must integrate ammonia chemistry with urea decomposition and HNCO kinetics in
an appropriate manner. In this paper, global HNCO kinetic rate constants were correlated
from detailed HNCO chemistry simulations by the CHEMKJN® program, a computer
program utilized to calculate complex chemical kinetics (Kee et al. 1993). The global rate
expression for the NO formation pathway is
and the global rate expression for the NO reduction pathway is
where XHNCO , X0i , XN^, andXNO are molar fractions of species HNCO, O,, N2, and NO,
respectively. The sensitivity of the SNCR submodel to aerodynamic patterns, species
concentrations, and temperature was evaluated via a commercial CFD software package.
The SNCR kinetic submodel was applied in a post-processing mode.
The velocity and temperature calculation were made by using the k-e turbulence model.
Species concentrations and reaction terms were calculated by employing the Magnussen
Eddy Breakup Turbulence Model. Since the probability density function model doesn't
account for the temperature effect, using the PDF (Probability Density Function) turbulence
model produced very poor representation of SNCR kinetics-turbulence interactions. Hence
the results represented in this paper have been obtained without using the turbulence model.
Three SNCR case predictions were compared with the test data from PD&T's Boiler
Simulation Facility (BSF) corresponding to the aqueous urea mass in the injected reagent
streams concentrations of 0.59%, 1.03%, and 1.44%. The three normalized stoichiometric
ratios (NSR) of urea to untreated NOX in the flue gas are 0.7, 1.21, and 1.7, respectively.
The NOX predictions were within 16% of the test data, and a similar agreement was
observed in predictions of ammonia slip.
The SNCR global kinetic model was derived by the methodology developed by Chen et al.
(1996), Chen et al. (1996), and Chen (1994). The comparisons of predictions and test data
show that the global kinetics and correlated rate constants are appropriate. In the
computational results, the strong dependence of SNCR processes on fluid dynamics has
been illustrated and it indicates the potential application of the SNCR model to assist in the
design and operation of the process. The rate constants of HNCO are correlated from
limited simulations of HNCO chemistry. It has been observed experimentally that some
major species such as H2O, CO2, and CO play a significant role in the SNCR processes
under certain conditions. An assessment of the significance of those species should be
studied systematically by using the CHEMKIN® program and other computer programs.
The turbulence model in the commercial code does not include the temperature window,
and the development of a turbulence model is also needed. The application and evaluation
of the SNCR model is certainly helpful to gain more confidence in the SNCR model.
-------
Wei Chen, June 24, 1997 for DOE, EPRI, EPA Mega Symposium on August 26, 1997
BACKGROUND
Chemistry schemes
Integration of chemical kinetics within a CFD computational code has proven to be a
difficult task for many years. Primarily two types of chemical kinetics have been used: (1),
elementary chemical kinetics, and (2), empirical kinetics by data correlation. The
elementary chemical kinetics approach is more comprehensive and fundamental, and it may
include hundreds of reaction steps depending on which chemical process is considered.
The full SNCR mechanism includes more than 200 elementary steps and 50 species,
consequently, to obtain species evolution profiles within a reasonable time, the calculation
of reactions must be made by stand-alone computer codes such as CHEMKIN®. It is
unrealistic to apply the elementary chemistry to CFD computations because the number of
elementary steps and species requires substantial computer resources. Although some
strategies, such as reduced mechanisms, have been used, empirical correlations are
typically considered to be acceptable for CFD computations in industrial applications.
Typically, empirical correlation has been generated from test data gathered from perfect
stirred reactors (PSR), or laminar premixed reactors. The generation and application of
empirical correlations has always been favored in industry because of the simplicity of
directly extracting the correlation data from the test facilities. The accuracy of empirical
correlation has oftentimes been questioned because the results may include some fluid
dynamic or mixing effects. A combination of elementary kinetics and empirical correlation,
as proposed in this paper, may overcome some of the disadvantages of both schemes.
SNCR elementary kinetics
In the last decade, significant progress has been achieved in understanding NOX and SNCR
chemistry. In NOX chemistry, there are numerous intermediate species and radicals that can
significantly affect the product distribution. Miller and Bowman (1989) compiled about
200 elementary steps for nitrogen chemistry in conjunction with O, H, OH, and simple
hydrocarbon reaction steps. NOX is formed in combustion predominantly via three distinct
mechanisms: thermal-NO, prompt-NO, and fuel-NO. When an appreciable amount of fuel
nitrogen exists as for example in coal combustion, fuel-NO is dominant in NO formation.
The fact that fuel nitrogen does not convert to NO completely indicates that the reaction
rates of competitive reaction steps are comparable in NOX formation. Therefore, reaction
conditions may be controlled to favor low NOX formation or even NOX reduction.
Several reaction pathways have been identified that contribute to NOX reduction. One of
these involves the conversion of NOX to nitrogenous intermediates, which under certain
conditions produce N2, rather than NOX. Because of the high activation energy required to
oxidize N2, a portion of the intermediate nitrogenous species is reduced preferably to N2.
In Selective Non-Catalytic NOX Reduction (SNCR), ammonia or urea is introduced to
reduce NOX chemically to N2 by amine radicals (NHj) and cyanuric nitrogen radicals
(HNCO) (Miller and Bowman 1989, Glarborg, et al. 1994).
At temperatures above 600 K, ammonia dissociates into amine free radicals by the
following reaction steps (Miller and Bowman 1989)
NH, -> NH2 + H (3)
NH2 -> NH + H (4)
-------
Wei Chen, June 24, 1997 for DOE, EPRI, EPA Mega Symposium on August 26, 1997
and the conversion of NHn to final products NO and N, the major routes from elementary
chemical steps have been identified as follow
HNO + OH -> NO + H2O (5)
and
NH2 + NO -> NNH + OH (6)
or
NH,+ NO -> N2 + H +OH (7)
In the temperature range of interest (900 K-1350 K), the rates of reactions (5), (6), and (7)
are of the same order of magnitude, and reactions (5) and (6) are competitive steps in the
formation or reduction of NO. At low temperatures, the reduction steps (6) and (7) have
slightly higher reaction rates, and the overall outcome favors NO reduction When the
temperature is above 1300 K, NO formation reaction (5) is dominant (Miller and Fisk
1987, Miller and Bowman 1989). The competition of the formation and reduction rates of
NOX results in a bell-shaped effect of temperature on NOX reduction. At lower
temperatures, the reduction reactions (6) and (7) are dominant and NO reduction rate
increases with temperature; at about 1250 K, formation rates from reaction (5) become
significant, and NO reduction rate decreases. Hence, a maximum NO reduction at about
1250 K is observed.
The urea-based reagent dissociates to the primary reactants, i.e., ammonia (NH3) and
isocyanic acid (HNCO). The chemistry of the urea-based SNCR process has temperature
characteristics that are similar to those of the NH3-based SNCR process. A slightly-wider
temperature window has been observed in the case of the urea-based process (Teixeira and
Muzio 1987). The NH? derived from the urea follows the same reaction pathways as
given previously in reactions (5), (6), and (7). About 70% of the total HNCO released
from urea is postulated to form NCO, then to N2 and NOX by reactions (8)-(l 1), (Lyon and
Cole 1990, and Caton and Siebbers 1990)
HNCO + OH -> NCO +H20 (8)
NCO + O -> NO + CO (9)
NCO + NO-> N2O + CO (10)
N,O + M->N2 + O+ M M=third body species (11)
Since NCO reacts with both NO and O, to form N2 and NO, respectively, HNCO
ultimately contributes to the reduction and formation of NOX depending upon reaction
conditions such as concentrations and temperature. Caton and Siebbers (1990) reported
that the temperature window for NOX reduction of HNCO is from 820 to 1340 K; Jodal et
al. (1990) reported that temperature window of urea is from 1023 K to 1373 K, and,
furthermore, found a significant amount of N2O (up to 85 ppm) in the products. Some
elementary HNCO steps were also added by Lyon and Cole (1990) and Glarborg et al.
(1990), to the SNCR mechanism and the simulations showed significant improvement in
predictions with experimental data steps (Lyon and Cole 1990, Glarborg et al. 1990).
In recent years, significant progress has been made in understanding ammonia and urea
chemistry. It is now well-known that ammonia and urea SNCR processes have similar
characteristics, such as a defined temperature window and a characteristic ammonia slip,
although they have different reaction sequences. Urea may have a slightly wider
temperature window, with a temperature range about 50 K higher.
-------
Wei Chen, June 24, 1997 for DOE, EPRI, EPA Mega Symposium on August 26, 1997
Ammonia global kinetics
An empirical global rate for SNCR kinetics should represent the basic characteristics of an
SNCR process, which include
1 . Temperature window effects
2. NOX and oxygen concentration effects
3. Reagent concentration effects
4. Ammonia slip
5. Effects of additives
Normally, these characteristics, combined with an analysis of fundamental chemistry,
constitutes the basis of an empirical or global kinetic rate expression formulation. De Soete
(1973) conducted premixed flow tests by adding ammonia into premixed flames; H2, CH4
and C2H2 were used as fuel and NO was also added in some tests. Through his tests, De
Soete hypothesized two competing reaction steps in ammonia global reactions, and the rate
constants were correlated from his measurements (De Soete 1973)
rNHj-NO = kpiH,-NO*-NH,X-O2 (13)
After years of research in ammonia chemistry, another set of global NH3-NO reaction rates
were developed specifically for an ammonia-based SNCR process by Duo et al. (1992).
The correlation was made from premixed reacting flow data conducted by Duo et al.
(1992), in which the constant reactor temperature was controlled between 1 140 K and 1335
K at inlet concentrations of 5.16xlO'3 mol/m3 of NO, 8.45xlO'3 mol/m3 of NH3 and 0.405
mol/m3 of O2. The volume fractions of reactants were varied by adjusting the mass flow
rates of reactants at the inlets. Unfortunately, although the urea-based SNCR process had
been used for years, no global BQSTCO correlation was found in the literature.
In summary, several global rate expressions and rate constants for ammonia-based SNCR
processes have been identified in the literature. However, rate expressions and rate
constants for the global HNCO pathway have not been found and the development of the
global HNCO pathway is needed to complete the SNCR model.
GLOBAL HNCO KINETICS
Two approaches are available to obtain a global HNCO rate constant: (1) correlate the rate
constants from experimental data, or (2) to correlate the rate constants from computed
species profiles using the CHEMKIN® program with elementary chemical reaction steps.
The later option was chosen in this case simply because of the excessive time and cost
required to collect experimental data. Global kinetic rate constants are typically correlated
from perfect stirred reactor data/calculation or premixed reacting flow reactor
data/calculations to minimize the mixing effects. The CHEMKIN® library of software
programs was developed for this purpose and is appropriate to be used for those
simulations (Kee et al. 1993).
Formulation of rate expressions
To correlate a global rate constant, a postulated or presumed rate expression must be
formulated, and then the dependence of the rate constant on species concentrations can be
established. Based on the study of NOX and SNCR elementary chemistry elucidated above,
-------
Wei Chen, June 24, 1997 for DOE, EPRI, EPA Mega Symposium on August 26, 1997
the formulation of urea and HNCO rate expressions is somewhat intuitive. Basically, three
rate expressions are hypothesized; the urea dissociation step, the HNCO oxidation step,
and the HNCO reduction steps. The urea dissociation rate is an initiation rate for the
reactions, and following the dissociation of urea, there are four parallel reactions, two
global NH3 pathways (i.e., reduction and oxidation), and two HNCO pathways. The
rates from either De Soete (1973) or Duo et al. (1990) may be used for the global NH3
reactions. Two parallel reaction rate constants of HNCO are correlated later in this section.
Dissociation rate When heated, urea dissociates into various products at different
temperature levels. At temperatures below 80 C, urea forms cyanuric acid and carbon
dioxide at a relatively slow rate:
6(H,N-CO-NH2) -> C3N3(NH2)3 + 3CO2 (14)
When the temperature is increased above 80 C, ammonia is released quickly via the
expression:
3(H,N-CO-NH2) -> C2N3(OH)3 + NH3 (15a)
At temperatures above 200 C, the dissociation of urea proceeds quickly and only HNCO
and NH3 are produced:
H2N-CO-NH2 -> HNCO + NH3 (15b)
Since the temperature is well above 200 C in the post-combustion region, only the
dissociation rate for reaction (15b) is considered in this effort, and it can be formulated as
follow,
^a=kureaCtt,ea (16)
where C^ is the molar concentration of urea in the gaseous phase.
HNCO->NO reaction rate HNCO formed from dissociation of urea reacts with
oxygen to form NO,, as described in reactions (8) and (9).
By assuming that oxygen dissociation and HNCO to NCO are fast reactions, the global
rate expression of HNCO oxidation to NO is readily available from reaction steps (8) and
(9)
where Q is the molar concentration of species I. Alternatively, Eq. (17) can be expressed
in terms of mole fractions
HNCO->N2 reaction rate From reactions (8), (10), and (11), again with a
similar assumption made for the oxidation step, the global HNCO reduction rate expression
can be written as follows:
rHNCO-N2 = * UNCO- fl, CHNCOCNO (19)
or write in a mole fraction form
-------
Wei Chen, June 24, 1997 for DOE, EPRI, EPA Mega Symposium on August 26, 1997
HNCO-N2 =
Correlation of rate constants
In the present method, the global rate constants were correlated by utilizing the species
concentrations and first derivatives of the species concentrations with time (i.e., the
effective rates) from CHEMKIN® simulations (Figure 3).
Dissociation rate constants The rate constants of urea dissociation were
correlated from the experimental data published by Siebbers and Caton (1992).
HNCO->N2 reaction rate constants The global reduction rate of HNCO to
N2 was correlated from the CHEMKIN® simulations of elementary kinetics. The
simulations were conducted at a temperature from 700 K to 1700 K, 0.0 to 2500 ppm of
CO, 0.0 to 20% of O2 concentrations, and 1E-5 to 0.1 sec of residence time. The NO level
and HNCO level were maintained at 400 ppm and 480 ppm, respectively. The species
profiles for the global reduction rate constants were obtained by assuming the similarity of
ammonia global kinetics and HNCO global kinetics (Figures 3 and 4).
HNCO->NO reaction rate constants The species profiles for the global oxidation
rate constants were obtained by assuming the similarity of ammonia global kinetics and
HNCO global kinetics.
VALIDATION AND APPLICATION
Several 2-D and 3-D cases of BSF tests were simulated at PD&T in 1992 (Rini et al. 1993)
in which the flow effects and reagent distributions were investigated (Figure 5). In their
study, the HFOP inlet flow mal-distribution had a significant impact on the injection
process, and a massive recirculation zone over the arch appeared to capture a large amount
of the reagents injected. However, the reagent injected near the top of the upper furnace
was characterized by a short residence time and was quickly exhausted from the domain
(Rini et al. 1993). The simulations from 3-D non-isothermal cases exhibited results
similar to the 2-D cases. The chemistry modeling effort consisted of an "idealized" 90-step
mechanism solved by CHEMKIN®. No mixing coupling between the flow modeling and
chemistry modeling was reported. (The use of a 90-step mechanism within the context of a
CFD code should be considered to be computationally prohibitive).
The flow field investigation by Rini et al. (1993) emphasized the significance of the
coupling of the SNCR chemistry and the flow field. In this context, the development of the
global SNCR kinetic model may be considered to be a continuation of the 1993 effort. In
the present study, the global SNCR model was applied to the BSF upper furnace. The
global kinetic model was tuned in order to promote greater predictive confidence in the
application of the SNCR model to industrial problems..
The BSF upper furnace case has 42x72x120 cells and the species profiles computed from
the BSF lower furnace case were patched at the inlet cross section of the BSF upper
furnace (Figure 6). The temperature was corrected accordingly to match the measurements
from BSF tests because the SNCR kinetic model was very sensitive to local temperature.
Three NSRs (normalized stoichiometric ratio) were computed and used as the baseline
cases to tune the SNCR model. The definition of NSR was given by Rini et al. (1993) as
NSR = inlet urea mole fraction/inlet NO mole fraction (21)
-------
Wei Chen, June 24, 1997 for DOE. EPRJ, EPA Mega Symposium on August 26, 1997
Three NSR levels were tested 0.7, 1.21, and 1.7, corresponding to mass fractions of urea
as 0.59%, 1.03% and 1.44%, respectively (Figure 7). A mass weighted average
temperature at the inlet cross section was constant in the three cases set at 1630 K, and the
oxygen concentration was maintained at 3.2%. Although the predictions gave a reasonable
trend, some discrepancies existed when the global reaction rate constants from this study
were used. After a careful review of the SNCR model integrated in the commercial CFD
program, a tuning of kinetic rate constants was made to match the test data. The
predictions by the original and tuned rate constants, and the experimental data are tabulated
below
NSR
0.71
1.21
1.7
NOf (ppmv)
(Original rate constants)
162
103
89.7
NOt (ppmv)
(Tuned rate constants)
134.8
92.3
83.4
NO, (ppm)
(Experimental data)
115
85
75
In Figure 7, comparisons of NOX and NH3 are illustrated, and the error of NOX predictions
is from 16% to 9%. Comparisons show a good trend in the agreement of ammonia slip
and NOX reduction dependence on the amount of reagent injected. NOX predictions showed
satisfactory levels while ammonia was under-predicted (Figure 7). A parametric CFD
calculation was conducted to assess the sensitivity of the global SNCR kinetic model
(Figures 8, 9, 10, 11). Furthermore, fluctuation of the measurements of ammonia slip was
observed under the same test conditions. Under the same test conditions, the spikes of
ammonia concentration were observed up to 53 ppm (Rini et al. 1993).
SUMMARY
The SNCR processes have been commercially applied in the abatement of NOX emissions
of boilers. However, the performance of an SNCR system relies on many factors, as
indicated in the literature. The motivation to understand those factors has two-fold: 1. to
reduce by-products such as NH, and N2O, and 2. to achieve the maximum NOX reduction.
Computer modeling of an SNCR process can assist the understanding thereof at a relatively
lower cost.
The key of modeling an SNCR system is the formulation of a chemical kinetic model. In
this project, a global chemical kinetic model was developed to include both NH3 and urea
reagents. The global rate constants of HNCO were correlated from simulations of
elementary nitrogenous chemical steps. The global SNCR model was integrated into the
commercial CFD computer program as a post-processor, and the coupling of transport
effects and kinetic effects can be investigated. A substantial effort has been made to debug
and test the SNCR model and a tuning of the global rate constants of HNCO was made to
'best-fit' the measurements on PD&T's BSF. The comparisons of predictions and data
show that the SNCR model appropriately reflects the characteristics of a SNCR system
with regard to temperature window and ammonia slip.
As with all mathematics models, limitations exist in the SNCR model and further
development of the SNCR model is needed. The global rate constants of the SNCR model
are correlated under certain reaction conditions. The tests and simulations for NH3 and
HNCO rate constants were done at a high NOX level and oxidized environment. Although
those rate constants can be further evaluated by the method used, it is suggested that the
SNCR model be used when NOX level is not very low (>100 ppm) and in an oxidized
environment.
-------
Wei Chen, June 24, 1997 for DOE, EPRI, EPA Mega Symposium on August 26, 1997
REFERENCES
Caton, J.A. and D.L. Siebbers. "Effects of Hydrogen Addition on the Removal Nitric
Oxide by Cyanuric Acid", Twenty-Third Symposium (International) on Combustion, The
Combustion Institute, 1990.
Chen, W. L.D. Smoot, T.H. Fletcher, and R.D. Boardman, "A Computational Method for
Determining Global Fuel-NO Rate Expressions. Part 1", Energy & Fuel, 1996, 10.
Chen, W. L.D. Smoot, S. Hill, T.H. Fletcher, "A Global Rebuming Rate Part 2", Energy
& Fuel, 1996, 10.
Chen, W., "A Global Reaction Rate for Nitric Oxide Reburning", Ph.D. Dissertation,
Brigham Young University, Provo Utah, 1994.
De Soete, G.C., "Mechanisms of Nitric Oxides from Ammonia and Amines in
Hydrocarbon Flames", (in French), Review of the Petroleum Institute of France, xxvii, 1
(Jan.-Feb.), 1973.
Duo, W., K.Dam-Johansen, and K. Ostergaard, "Kinetics of the Gas-Phase Reaction
Between Nitric Oxide Ammonia and Oxygen", The Canadian Journal of Chemical
Engineering, Vol. 70, October, 1992.
Glarborg, P., P. Kristensen, S.H. Jensen, and K. Dam-Johansen, " A Flow Reactor
Study of HNCO Oxidation Chemistry", unidentified.
Jodal, M., C.Nielsen, T. Hulgaard, and K. Dim-Johansen. "Pilot-scale Experiments with
Ammonia and Urea as Reductants in Selective Non-catalytic Reduction of Nitric Oxide",
Twenty-Third Symposium (International) on Combustion, The Combustion Institute,
1990.
Kee, R.J., J.F. Grcar, M.D. Smooke, J.A. Miller, "A Fortran Program for Modeling
Steady Laminar One-Dimensional Premixed Flames", SANDIA REPORT, SAND 85-
8240.UC-401, Reprinted September, 1993.
Lyon, R.K. and J.A. Cole. "A Reexamination of the Rapre NOX Process", Combustion
and Flame, Vol. 82, 1990.
Miller, J.A., and C.T. Bowman, "Mechanism and Modeling of Nitrogen Chemistry in
Combustion", Prog. Energy Combustion Science, Vol.15, 1989.
Miller, J.A., and G.A. Fisk. "Combustion Chemistry", C&EN, August 31, 1987.
Rini, M., J.A. Nicholson, O.K. Anderson, N.Y. Nsakala, R.L. Patel, and D.R.
Raymond. "An Investigation of Selective Non-Catalytic Reduction (SNCR) of NOX for
Coal Firing", Final Report, KDL-92-FSD-2, December 1993.
Teixeira, D., and L. Muzio. "N2O Emissions from SNCR Processes", NOX Emissions
from Stationary Combustion Systems, EPRI Power and Energy Conference, 1987.
Siebbers, D.L., and J.A. Caton. "Removal of Nitric Oxide from Exhaust Gas with
Cyanuric Acid", Combustion and Flame, Vol. 79, 1990.
-------
1. Global ammonia
mechanism
2. Global HNCO
mechanism
(derived from this research)
• use 1 if NH3 as reagent
•use 1&2 if urea as reagent
Output NOx, ammonia
slip, and N2, etc.
• Thermal NOx model
•Fuel NOx model
•Prompt NOx model
•Rebuming mechanism
Figure 1 Global SNCR Kinetic Model
ISOTHERMAl ROW MODELING
• Modify Existing BSF Flow Model
• Rebuild wmabowj
• Create SNCR Cavity
• Measure Velocity Proflte Entering
ana Wimmin SNCT Cavffy
• lOOANOTTCMCR
L
SPRAY NOZZLE TESTING
• Measure BSF Urea Spray Nozzle
• Droplet Size & Distribution
• OropW Vetoctty
.Sprery Angle
• Mass Flow DkSrtbuUon
1
CFDMOOEL
J
CH
• Set Up Mode! of SUCK Cavity
• Predict ISFATF-BSF Flow Fields
• Vefldate
• Predict BSF Row Re lets - Hot
• Shakedown Spray Sub-Routine
BSF PREDICTIONS
* Ba»0n« Urea Infection Configuration
• Validate
• Thre* Other Indian Configurations
• VtSdate
CHEMK1N SIMULATION
of SNCR CHEMISTRY
• Generate SNCR simulations
• Identify the significant pathways
• Correlate the global kinetics
VALIDATED SNCR
PERFORMANCE PREDICTION
METHODOLOGY
Figure 2 Derivation and Validation of Global
SNCR Kinetic Model
-------
- Experimental NO Reduction by Ammonia
- Simulated NO Reduction by HNCO
- Experimental NO Formation by Ammonia
— Simulated NO Formation by HNCO
Figure 3 Correlation of HNCO rate Constants
NHj+O
Primary SNCR Reaction Sequences
NH2+NO +- N2+H2O
CYANURIC ACID
(HNCO).)
NCO + H,O
NCO + NO *- N,O + CO
N2O + M —>~ N2 + O + M
N2O + OH —>- N2 + HO2
N2O + H —»- N2 + H
Figure 4 SNCR Pathways
-------
Boiler Simulation Facility (BSF)
Figure 5 Boiler Simulated Facility (BSF)
-------
BSF Upper Furnace Model Domain
Side Elevation
f
Plan View
Figure 6 BSF Computational Model
Injector
BSF Validation Results
Controlled NO (Dry)
85F Experimental!
-Model
NHl Slip
0.5 1
Cavity Temperature: 1588°F
Inlet NO: 175ppm
Inlet 02: 2.7 % (mole basis)
Figure 7 Global SNCR Model Validation
-------
Parametric Study of NSR
80%
70%
60%
g 50%
= 40%
£
5 30%
20%
10%
0% .
NO Reduction
NH, Slip
Cavity Temperature: 1588°F
Inlet NO: 175ppm
Inlet O2: 2.7 % (mole basis)
Figure 8 Parametric Study of NSR
Parametric Study of Cavity Temperature
NO Reduction
70%
60%
50%
40%
30%
20%
10%
0%
1200 1400 1600 1600 2000 2200 2100
Temperature I'F)
NH^SIIp
2
0
1200 1400 1600 1800 2000 2200 2400
Temperature (-F)
NSR:
Inlet NO:
Inlet O,:
1.21
175 ppm
2.7 % (mole basis)
Figure 9 Parametric Study of Cavity Temperature
-------
Parametric Study of Inlet O2
NO Reduction
70%
60%
jj 50%
I ">''•
* *»
20%
10%
0.0% 1.0% 2.0% 3.0% '
lnlBtO,(mole %J
NH, Slip
4,0% 5.0%
0.0% 1.0% 20% 30% 4.0% 5.0%
taint 0, (moll V,l
NSR: 1.21
Cavity Temperature: 1668°F
Inlet NO: 175 ppm
Figure 10 Parametric Study of Inlet 02
Parametric Study of Inlet NO
aor.
70%
60%
20%
10%
NO Reduction
0 50 100 ISO 200 S50 300 350
Inlet TO (ppm)
NH,Sllp
0 50 100 150 200 2SO 300 350
InlatNO(ppm)
NSR: 1.21
Cavity Temperature: 1668 °F
Inlet 02: 2.7 % (mole basis)
Figure 11 Parametric Study of Inlet NO
-------
In Field Results of SNCR/SCR Hybrid on a
Group 1 Boiler in the Ozone Transport Region
J. Urbas
GPU GENCO
1001 Broad Street
Johnstown, Pennsylvania 15907
J. Boyle
Nalco Fuel Tech
P.O. Box 3031
Naperville, Illinois 60566
Abstract
A full scale SNCR/SCR Hybrid system, NOxOUT CASCADE, has been designed and is being
installed at the GPU GENCO Seward Station, Unit #15 boiler. The Seward Station hybrid
system is a combination of a redesigned existing SNCR with a new downstream SCR.
Significant improvements in chemical utilization and overall NOx reduction have been seen in
preliminary testing of the SNCR when ammonia slip was permitted to increase above normal
operational limits.
The integrated system was designed using advanced computational fluid dynamics and cold flow
modeling techniques. The units two air pre-heater ducts were retrofit with different types of
catalyst, honeycomb in one and plate in the other. Reactor and duct internals were designed to
compensate for an existing ash loading imbalance, temperature and velocity variation, and a
difference in the SCR pressure drop between the two ducts.
Introduction
The Clean Air Act Amendments of 1990 have given rise to a wave of technology development
that anticipates meeting clean air challenges. In the first half of this decade, the U.S. witnessed
the retrofit of low NOx burners on coal, oil, and gas-fired boilers. Additionally, there were new
developments in air staging technologies, gas reburn demonstrations under the Clean Coal
Technology Program, in-field applications of SNCR retrofit on various types of utility boilers,
and even a retrofit application of SCR on a cyclone coal-fired boiler. Industry observers predict
large costs will be borne by major sources to meet the air quality goals in some Phase II
provisions of the Act. In preparation for "life beyond Phase I," field development is now being
focused on effective combinations of NOx controls. Potentially, two or more available means of
-------
NOx control can be compatibly combined to reduce NOx wherein the end result is more cost
effective than the sum of its parts. Hybrid combinations of SNCR and SCR are a particularly
flexible method for effecting moderate to deep reductions of NOx at cost ranges typically below
those of a fully-engineered SCR retrofit.
This paper presents a discussion of the implementation of a SNCR/SCR hybrid at the GPU
GENCO Seward Station and the expected increase in NOx reduction performance and chemical
utilization. The design was based on minimizing the total life cycle cost while achieving the
required control.
Cascade Methodology and Theory
NOxOUT CASCADE® is a combination of a redesigned SNCR and downstream SCR,
hybridized to provide improvements in chemical utilization and overall NOx reduction. The two
NOx reduction technologies each provide process strengths which make the hybrid combination
more flexible and effective than the sum of its parts.
Selective Non-Catalytic Reduction (SNCR) is typically applied in the furnace, where relatively
high temperatures serve to initiate the breakdown of urea to form the transient species which lead
to effective NOx reduction. The technology is limited to temperatures high enough to insure
very low ammonia breakthrough. At very high furnace temperatures, however, performance can
be lessened by competing reactions which either consume effective chemical or lead to NOx
formation. Modified SNCR takes advantage of a downstream "ammonia sink" by injecting
chemical in cooler regions where NOx reduction and chemical utilization improve dramatically.
Selective Catalytic Reduction (SCR) is typically performed in much cooler flue gas passes where
the oxidation potential of nitrogen species is minimized. The catalytic surface provides sites
which permit the ammonia and NOx to react at nearly perfect utilization. The extent of NOx
reduction is limited by the local ammonia to NOx ratio, the flue gas temperature, and the size of
the catalyst reactor. The catalyst size is limited by the available space, an increase in pressure
drop, the oxidation of S02 to SO3, and the cost of the precious metal components.
NOxOUT CASCADE® utilizes lower temperature SNCR injection to provide substantially
improved NOx reduction performance while generating somewhat higher ammonia slip. The
ammonia slip feeds a small SCR reactor which removes the slip and reduces NOx while limiting
the costs associated with a larger catalyst. For example, a CASCADE system which achieves
65% overall NOx reduction ( 50% reduction with SNCR and an additional 30% SCR reduction)
requires less than one third the catalyst required for 65% SCR reduction. The smaller catalyst
converts proportionally less SO2 to SO3 and decreases the pressure drop by the same fraction.
System Design
Hybrid SNCR/SCR NOx reduction systems can be engineered in several forms. Clearly, it is
possible to install a commercial SNCR system for furnace reductions of NOx, and install a
-------
commercial SCR system downstream of the economizer on the same unit for removal of the
remaining NOx, and enjoy deep levels of NOx reduction with the combined system. For the
purpose of semantic clarity, one might consider the foregoing system "combined SNCR/SCR"
while reserving the "hybrid" description for units which utilize the ammonia slip from the
SNCR process as the sole NOx reductant entering the downstream SCR.
Hybridized SNCR/SCR can assume several configurations depending upon the level of overall
NOx reduction desired and unit configuration. Both factors combined lead to differences in
catalyst dimensions and, therefore, catalyst contributions to the total capital requirement.
Various configurations for consideration would be SNCR with:
• catalytic air heater
• "in-duct" SCR-existing duct dimensions
• "in-duct" SCR-expanded duct dimensions
• reactor-housed SCR
• combination of "in-duct" SCRs with catalytic air heater
Prior literature1 surveyed the above combined technologies listing benefits and potential
drawbacks of combining the technologies. It primarily reported from a technological feasibility
viewpoint where a specific requirement for SCR is presumed. It is important to view the
potential application of hybridized SNCR/SCR from an economic standpoint, particularly in the
case where combustion modifications have already been employed.
Besides assuming several physical configurations, hybrid SNCR/SCR can be operated in
different ways. Among the many considerations for the choice of designated hybrid operation
are:
• What is the desired level of NOx reduction?
• What are the NH3 slip and SO2 oxidation constraints?
• What volume catalyst can fit in the existing ductwork where face velocity will be within
catalyst manufacturer specifications?
• What level of additional pressure drop is tolerable by the present fan?
• Are NOx reduction requirements incremental?
• What structural steel/ductwork changes must be made to support the catalyst?
• What is the expected/guaranteed life of the catalyst?
• What deviation from ideal reductant distribution is tolerable for the NOx limit?
It is obvious that total capital requirement for the catalyst retrofit will increase as the catalyst size
and retrofit complexity increase. The key to minimizing life cycle NOx reduction costs is to find
the appropriate balance between annualized capital charges and operating costs for the remaining
life of the system. The challenge for SCR retrofit is to minimize the capital requirement. The
challenge for SNCR use is minimization of reagent required. Designing hybrid SNCR/SCR
systems suggests optimization of these costs over the life cycle for a specific level of NOx
reduction
-------
Chemical Utilization
In post-combustion NOx control processes, NOx reduction is achieved at a given Normalized
Stoichiometric Ratio, or NSR. Simply put. NSR refers to the ratio of chemical reductant applied
to the amount of NOx existing in the flue gas. With SCR, ammonia is typically the reductant and
is typically applied at an NSR of one for deep reductions. In other words, one mole of NH3 is
applied per mole of NOx. If only a 75% NOx reduction was required, the NH3 NSR would be
approximately 0.75. In non-catalytic systems, the reductant is applied in broader ranges of NSR
because of relatively lower NOx reduction efficiency compared to catalytic systems. In
commercial practice, NSRs range from 0.6-2.0. When urea is used for SNCR systems, an NSR
of 1.0 means 0.5 mole urea is applied for 1.0 mole NOx because urea has two nitrogen moieties
for reaction with NOx.
Chemical utilization is a quantification of NOx reduction efficiency expressed by:
Utilization = NOx Reduction [%] / NSR
In other words, if each Ib-mole of injected urea or ammonia reduces NOx to the theoretical
maximum amount2, utilization is 100%. One hundred percent chemical utilization is approached
in SCR systems, but in SNCR system values range from 30-60%. In commercial post-
combustion NOx control systems, maximizing utilization, all other things being equal, minimizes
life cycle operating costs.
Figure 1 schematically depicts the enabling effect of downstream catalyst (down-sized or
otherwise) on SNCR performance in a hybrid system. SNCR NOx reduction occurs in a defined
temperature window, roughly bell-shaped with maximum SNCR NOx reduction occurring at the
top, or plateau of the bell. In a commercial "stand-alone" SNCR system, performance is
optimized by operating at the "right side of the slope" in the temperature window curve3 (in
Area A). In this region, the hot side of the performance maximum, ammonia slip is very low or
non-existent. This is often an operating constraint imposed by the source owner. In contrast, the
SNCR component of the hybrid system operates best at the plateau which is lower temperature.
In this region (Area B), SNCR NOx reduction is higher and some ammonia slip is produced.
The ammonia slip is available to reduce NOx in a catalyst system downstream. When operated
in this manner, SNCR NOx reduction is maximized (compared to its stand-alone performance)
and additional NOx reduction occurs in the catalyst portion, fueled by the generated ammonia
slip.
Hybrid systems can be designed to operate in the cooler zone (Area C - the "left side of the
slope") which will produce more ammonia slip than the other regions. In this scenario, SNCR
NOx reduction is less than maximal and SCR NOx reduction increases until limited by catalyst
space velocity. Overall system NOx reductions beyond 75% would typically require this type of
operation and require catalyst reactor dimensions that would not be possible to fit in existing duct
space.
-------
Hybrid systems can be designed to maximize SNCR performance while "existing duct" SCR
controls the ammonia slip (Area B). Reagent utilization for NOx reduction can increase
dramatically compared to stand-alone SNCR because of the reasons stated above. Therefore,
reagent cost per unit of NOx reduced is lower with the hybrid system than with stand-alone
SNCR. Current operators of SNCR systems consider these questions in the design stage for
prospective hybrid systems:
• What is the expected additional reduction of NOx for a constant urea (reagent) flow?
• What is the expected reagent flow reduction for constant NOx reduction?
Field Testing
The NOxOUT SNCR/SCR Hybrid process was tested at Public Service Electric and Gas, Mercer
Station4. The unit, which had an existing SNCR system, was partially retrofitted with an
expanded duct catalyst as part of a study of SCR, combined SNCR-SCR, and Hybrid
SNCR/SCR. In this preliminary work it was shown that deeper than design reductions in NOx
were possible through modification of the SNCR system with less than design chemical (urea)
flow rates. This was achieved by decreasing the effective chemical release temperature in the
furnace.
The by-product of this temperature shift, excessive ammonia slip, was utilized in the SCR reactor
where further NOx reduction was achieved and ammonia slip levels were reduced to within
acceptable limits. Although the SCR reactor was large enough to provide greater than 85% NOx
reduction on its own, it was shown that ammonia and NOx distributions to the catalyst were
sufficiently uniform to allow for a substantial reduction in catalyst volume without adversely
affecting the process.
The next logical step in the development of SNCR/SCR hybridization is full-scale application to
a utility boiler with a small catalyst used primarily for ammonia slip control.
GPU GENCO - Seward Station
A Hybrid SNCR/SCR system has been designed for GPU GENCO, Seward Station, Boiler #15.
This unit is a Combustion Engineering, coal burning, tangentially fired boiler rated at 148 MW
gross electrical output, Figure 2. Current minimum load is 106 MWg, but it may become
necessary to operate at loads as low as 74 MWg (50% MCR).
A commercial NOxOUT® SNCR has been installed at Seward Station. The system provided the
required NOx reduction from a 1990 baseline of approximately 0.78 lb/106 Btu to 0.45 lb/106 Btu
with less than 5 ppmv slip. High concentrations of SO2, and therefore SO3, as well as cool air
pre-heater exit temperatures combined to make this installation particularly sensitive to
-------
ammonium salt formation. It is currently being operated at reduced efficiency ( approximately
0.5 lb/106 Btu ) to produce less than 2 ppmv ammonia slip. As much as 75% of the chemical is
injected into the furnace where utilization is relatively low. The remaining chemical is injected
behind the super heater tubes, above the arch, through multi-nozzle lances which provide
excellent chemical distribution and extremely high chemical utilization.
This unit is an excellent candidate for hybrid SNCR/SCR reduction. Using the cooler zone alone
in limited testing, deep reductions in NOx have been possible with decreased chemical flow and
reasonable ammonia slip ( at or below 20 ppmv). A small in-duct SCR reactor was necessary to
remove 90% of the ammonia slip and provide additional NOx reduction. SCR experience on
coal fired units has been limited but recent independent testing has shown that new catalyst
formulations are able to withstand the harsh environment.
Seward Station Cascade Design
Design has been completed for the Seward Station Cascade commercial demonstration.
Available space for two small reactor vessels, one in each of the right and left side ducts, was
located between the economizer hoppers and the air pre-heater inlets. The duct design was
completed to provide the proper average inlet and outlet conditions as specified by the catalyst
vendors selected. Two independent catalyst vendors were selected.
A static mixing grid and turning vanes have been recommended to decrease the known and
predicted gas and solid flow imbalances in the unit. Design of these duct internals was
completed using both computational fluid dynamic (CFD) and cold-flow models. CFD
techniques were used to model the high temperature gases, Figures 3 and 4. The virtual
environment permits non-intrusive measurement and an evaluation of a wide variety of duct
configurations. Cold-flow modeling, however, can more closely approximate the actual
geometry and provides important measurements necessary for scale-up. Both modeling tools
have provided valuable insight into the design of constrained-space reactor vessels.
The catalytic rate of SO3 generation is particularly important in this case because of the current
air pre-heater sensitivities to ammonium salt formation. The catalyst vendors have specified
minimum operating temperatures above which ammonium salt formation and deposition on the
catalyst face will be avoided. Flue gas mixing and turning vanes have been designed to reduce
temperature variations and eliminate localized cool spots.
Maximum performance in a full-scale SCR requires uniform ammonia to NOx ratios across the
face of the catalyst. The ammonia slip to the SCR in a CASCADE system, however, will be
significantly lower than the NOx at all points in the flow. Performance degradation due to
variations in NH3 concentration will, therefore, be greatly reduced. More importantly, any
ammonia slip at or below design maximums will be significantly reduced. Both Cold-flow and
CFD modeling have shown that the gases concentrations are well mixed at the entrance to the
catalyst.
-------
Seward Station Cascade Expected Performance
Public Service Electric & Gas (PSE&G) and Nalco Fuel Tech concluded an evaluation of
combined SNCR/SCR NOx reduction as part of a demonstration of post combustion NOx
control on the Mercer Station pulverized coal, wet bottom utility boiler in 1993. This Hybrid
system utilized the urea based NOxOUT® SNCR process to provide both in furnace NOx
reduction and sufficient ammonia to feed a downstream reaction catalyst bed. The Hybrid urea
process achieved precatalytic reduction of approximately 50 % firing full load coal and a
maximum of 67 % firing gas. These results have shown significant increases in chemical
utilization, from 32 % to 63 % for full load firing coal, as compared to stand alone SNCR
The Cascade system at Seward Station is expected to provide overall NOx reduction of at least
55%, to 0.35 lb/10A6 Btu, from the 1990 baseline at less than 2 ppmv ammonia slip. The
primary injection zone will be significantly cooler and the chemical utilization is expected to
increase dramatically from the current SNCR system. Overall chemical flow is not expected to
increase.
Expected performance data is summarized in Table 1. NOxOUT® SNCR performance was
initially designed to achieve 42% reduction with an NSR of 1.3 and a resulting chemical
utilization of 33%. Based on preliminary testing, NOxOUT CASCADE® performance is
expected to increase to at least 55% reduction with an NSR of 1.2, a decrease in chemical flow,
and a resulting overall chemical utilization in excess of 45%.
Pending the results of complete testing later this year, which will verify the performance
estimates of the catalyst vendors, it will be possible to achieve 65% overall NOx reduction with
the addition of catalyst to the reactor vessel. This new design may also require additional
convective pass chemical injection, but the total chemical flow rate is not expected to increase
significantly.
Beyond Phase I
Phase II of NOx controls in the United States currently refers to "beyond RACT" controls in
ozone nonattainment areas or transport regions, as well as to the statutory acid rain provisions.
While the acid rain provisions require that NOx limits be promulgated for remaining utility
boilers (from Phase I) by January 1,2000, other requirements are anticipated by May of 1999 for
units which must reduce NOx for ozone-related reasons. The "Phase II" requirements will be
moderate NOx reductions5 beyond "RACT" (largely low NOx burners or other combustion
modifications), and they will only be required during "ozone season" - five months out of the
year. According to the referenced Memorandum of Understanding6, more controls may be
required in Phase III in 2003.
-------
Such a part-time control requirement, on units already employing primary controls which reduce
NOx to 0.38-0.50 lb/106Btu, complicates the consideration of minimizing life cycle control costs.
Life-cycle Costs
The use of hybrid SNCR/SCR systems permits "tailoring" NOx reduction and life-cycle cost to
the potentially complex future requirements of NOx reduction for ozone mitigation. The total
life cycle cost of the modified SNCR/SCR NOx reduction process is a function of chemical
utilization and catalyst size and capital requirement. Very high NOx reductions, of perhaps 90%,
require a substantial catalyst volume. This system cannot be placed in existing duct dimensions
and will always require, at the very least, major modifications. A modified SNCR/SCR system,
providing between 50-60% precatalytic reduction, would require between 75-80% further NOx
reduction to achieve 90% overall. This would still demand 88% of the original catalyst volume.
Similarly, for an overall NOx reduction of 75%, a stand-alone SCR system requires
approximately 88% of the original high reduction catalytic volume. (These design computations
are graphed in Figure 5.)
A modified SNCR/SCR process would conceptually be effective for approximately 75% overall
NOx reduction. Precatalytic SNCR reduction of 50-60% requires only 38-50% SCR reduction,
and no more than half of the original catalyst volume as designed for 90% reduction. This is also
only 57% of the catalyst volume required for stand-alone SCR targeted at 75% overall reduction.
An "in-duct" catalyst may be used on a site-specific basis to fulfill this half-sized volume
requirement.
Prior work at the plant site in development of the commercial-scale SNCR system which exists
there indicated that 42% NOx reduction was achieved within the 5 ppmv NH3 slip constraint. To
achieve this level of reduction with a permanent, commercial system required approximately
$14/kW capital. Design NSR for urea reagent is approximately 1.3 for the 42% reduction.
Equivalently, this is 33% utilization at full load with substantial improvements at lower load.
By contrast, the full-scale SCR installed for the PSE&G demonstration of SCR technology was
capable of achieving 90% NOx reduction and more for the several-month investigation. Installed
capital cost for the retrofit was reported to be $90/kW.7
The field demonstration of that hybrid SNCR/SCR system verified that on a coal-fired unit, the
SNCR-related cost performance can be improved substantially. This installation of in-duct
(existing duct) catalyst on a pulverized coal-fired unit provides a basis of broad applicability to
the various types of boilers within this population.
Conclusions
1. A Hybrid SNCR/SCR system has been designed for a full scale retrofit of a tangentially
fired coal boiler in the Ozone Transport Region.
-------
2. Two types of catalyst have been incorporated into the design to provide insight into the
achievable performance of various small in-duct reactors.
3. Extensive CFD and cold flow modeling has been completed to provide the required
temperature, velocity, and ash distribution profiles required by the catalyst vendors.
4. Chemical utilization and NOx reduction are expected to increase dramatically as
compared to stand-alone SNCR.
5. Hybrid SNCR/SCR is capable of addressing the control requirements for many coal-fired
boilers in the Ozone Transport Region with up to 75% NOx reduction.
References
1. Jantzen and K. Zammit, "Hybrid SCR," presented at EPRI/EPA NOx Symposium, Kansas
City, 1995.
2. For NH3, this is Ib-mole NOx; for urea the value is two moles of NOx.
3. U.S. Patent 4,780,289, issued 1988.
4. Wallace, A. J., Huhmann, A., Boyle, J. M., Albanese, V. M., "Evaluation of Hybrid
SNCR/SCR for NOx Abatement on a Utility Boiler," Power-Gen '95, Anaheim, CA,
December, 1995.
5. Ozone Transport Commission Memorandum of Understanding, Newport, RI, September 27,
1994.
6. Ibid.
7. Presentation to Ozone Transport Assessment Group, Strategy and Controls Subgroup,
Washington, DC, August 24,1995.
-------
80%
Effective SNCR Temperature Window
60% ~
§
40%
1
"i
V
U
'C1 - Hybrid
SNCR, Slip
NH3to
Catalyst
20% -
'B' - Hybrid
SNCR Maximum
0%
Increased SNCR
Temperature Window
'A' - Stand-Alone
SNCR
1290
1470
1650 1830
Temperature [°F]
2010
2190
2370
Figure 1. Hybrid SNCR/SCR yields improved performance.
Total reagent utilization approaches 100 %
-------
Side Sectional Temperature and Velocity Profiles
GPU-GENCO Sevard Station.
Temperature (r)
I
HOC
1221
1533
1704
2IS7
Figure 2. Side Sectional Temperature Profile of GPU-GENCO Sevard Station.
Unit #15 Temperature and Velocities at 148 MWgf Full Load.
-------
Figure 3. GPU Gr.NCO - Steward Station
CASCADE* Duct Velocity Contours
-------
Figure 4. Cold rlov Modeling of GPU GENCO Sahara Station SCR Dusts.
-------
o
I
O
Pi
U
CO
i
100%
90%
80%
70%
60%
50%
75% Overall Reduction
o 65% Overall Reduction
55% Overall Reduction
40%
30%
20%
10%
10%
20%
30% 40% 50%
Percent SNCR Reduction
60%
70%
80%
Figure 5. NOxOUT CASCADE Fraction of Full-Scale SCR Catalyst
Volume Needed to Achieve the Specified Overall NOx Reductions
-------
Table 1.
GPU-GENCO Seward Station CASCADE®
Unit #15
GHI
Flue Gas Flow
NOx Before NOxOUT®
Case=>
[10A6 Btu/hr]
[ SCFH - wet ]
[ ppmvdc ]
[lb/10A6Btu]
[Ib/hr]
147 MWg
Coal
1457.0
19,387,898
554
0.780
1136.4
106 MWg
Coal
1096.0
15,381,054
554
0.780
854.8
74 MWg
Coal
800.0
11,560,650
554
0.780
624.0
Modified NOxOUT® System
NOx After NOxOUT®
NOxOUT® Reduction
Chemical Utilization
NSR
[ ppmvdc ]
[lb/10A6Btu]
[%]
256
0.360
53.8%
45.0%
1.20
256
0.360
53.8%
45.0%
1.20
256
0.360
53.8%
45.0%
1.20
SCR System
Final NOx
Overall Reduction
SCR Reduction
Overall Utilization
NH3 at Catalyst Entrance
Final Ammonia Slip
Space Velocity
Catalyst Volume
Actual Gas Temperature
Face Velocity
[ ppmvdc ]
[%]
[%]
[%]
[ ppmvdc ]
[ ppmvdc ]
[1/hr]
[ft3]
[m3]
[°F]
[ft/s]
240
56.7%
6.3%
47.4%
18
2
13071
1483.2
42.0
600
18.4
232
58.2%
9.3%
48.6%
26
2
10370
1483.3
42.0
600
14.6
217
60.9%
15.3%
50.9%
41
2
7794
1483.2
42.0
600
11.0
-------
STATIONARY SOURCE NOX CONTROL USING PULSE-CORONA
INDUCED PLASMA
S. Haythomthwaite
G. Anderson
M. Durham
D. Rugg
ADA Technologies
Englewood, CO 80112
J. Wander
Armstrong Laboratory, Environics Directorate (AL/EQ)
Tyndall AFB, FL
Abstract
ADA Technologies designed, built, laboratory tested, and installed an innovative 300 acfin pilot-
scale nonthermal plasma NOX control system on a stationary Air Force jet engine testing facility.
The Air Force sponsored this program in response to EPA's designation of these units as
stationary sources. This application to gas turbine NOX control is challenging because of low
temperatures, large volume (up to 4,000,000 acfm), low NOX concentrations (4 to 35 ppmv), and
the sensitivity of the engines to backpressure. ADA's pilot system uses pulse-corona discharge
technology to treat flue gases. The laboratory and field testing showed that some NOX removal
was achieved initially, possibly by reduction of NO to N2, and further NO was oxidized to NO2
or higher oxidation states. A scrubber removes NO2 from the flue gas after it passes through the
corona discharge region. Overall NOX removals on the order of 50% were achieved.
Introduction
Jet engine test cells (JETCs) are structures designed to hold jet engines, provide a uniform
environment, and suppress noise during static operational tests following maintenance or
overhaul. These test facilities are used by the Air Force, Navy, and Army, as well as civilian
airline companies and jet engine manufacturers. Jet exhaust contains nitrogen oxide (NOJ, soot
particles, carbon monoxide (CO), and hydrocarbons (HC). Control of emissions from these
facilities presents a real challenge because the emission is intermittent and because the engine is
only minimally tolerant of back pressure (or flow resistance).
The control of acid rain precursors, S02 and NOX, has been the subject of regulations for a
number of years. The 1990 Clean Air Act Amendments have specifically targeted significant
additional control of NOX emissions. Jet engines are classified by EPA as stationary sources
only when operating in a JETC. A typical Air Force JETC emits approximately 10 to 20
tons/year of NOX emissions. EPA is considering imposing regulations in the near future that
would require decreasing net emissions of NOX from these static-firing test facilities.
There are a variety of designs for JETCs but the most common design incorporates an air-
augmentor tube downstream of the engine exhaust to cool the exhaust gases. As the exhaust
-------
gases flow into the augmentor tube, ambient air is drawn from outside into the tube to mix with
the exhaust gases. In addition to cooling the gases, the augmentor tube also dissipates a portion
of the kinetic energy of the exhaust blast.
Conventional NOX control technologies are not well-suited to the jet engine test cell application
because of its rapid and extreme changes in flue gas compositions and flow rates, high dilution
(low concentration), low temperature, high flow rates, and sensitivity to back pressure. These
characteristics of the exhaust gases preclude using any conventional back-end NOX control
technology such as Selective Catalytic Reduction (SCR) and Selective NonCatalytic Reduction
(SNCR). Combustion modifications, or front-end emission control, are also precluded in a
highly tuned process such as a jet engine.
The work described in this paper is the result of a two-phase Small Business Innovation Research
(SBIR) program sponsored by Armstrong Laboratory. The Air Force was interested in
investigating the technical and economic feasibility of an efficient emissions control strategy for
jet engine test cells which does not impact the performance of the jet engines. The primary target
emission was NOX, however the additional benefits of using the same control device for other
emissions was also of interest. The overall purpose of this SBIR program was to develop a cost-
effective technology for control of NOX from jet engine test cells using the pulse-corona-induced
plasma process (PCIP). The Phase I program demonstrated that the PCIP process was
technically feasible on the laboratory scale and capable of 90% NOX removal under simulated jet
engine test cell conditions.
The Phase II program was designed to provide a sound basis for projecting the economics of a
full-scale application of PCIP technology on a jet engine test cell. This was accomplished by
developing the data necessary to design a full-scale system based on actual field test results of a
sub-scale system. The sub-scale system tested in Phase n was representative of the geometric
configuration typical of a full-scale module. Tests defined operating conditions, power
requirements, emissions control capability, and waste characterization.
An overview of Air Force research to control NOX emissions from JETCs (1) summarizes a
variety of technologies to achieve this goal. Modifications to the combustor are precluded based
on performance impacts or potential physical damage. Armstrong Laboratory has sponsored
seven independent projects (apart from this nonthermal plasma program) which address exhaust
gas treatment options. These include exhaust-gas reburning, noncatalytic reduction with
additives, metal-based catalytic reduction, photocatalytic decomposition, electrocatalysts, and a
solid sorbent bed of magnesium oxide coated on vermiculite.
Technical Approach
This two-phase program advanced the technology from the laboratory to a sub-scale field
demonstration on an Air Force JETC. Initial work in Phase I consisted of proof-of-concept
testing to demonstrate that ADA's pulse-corona-induced plasma (PCIP) system design could
destroy NOX in a simulated JETC-exhaust gas. Phase II was designed to further the development
of this technology by scaling up the design, establishing system operating conditions in the
laboratory on a bench-scale jet engine, and then transferring this knowledge to the field tests.
In Phase I corona reaction cells were constructed that treated up to about 10 acfrn. Gas input to
the cell was provided by an exhaust gas simulation system. Measurements were made by real-
time continuous emissions monitoring analyzers, including ADA's photodiode array analyzer.
-------
Initial investigations in Phase II defined the parameters of JETC operation and investigated the
reaction products of the PCIP system. JETCs have intermittent operation, and as mentioned
above, have an extremely time-dependent and wide range of operating conditions. They are also
sensitive to back pressure and any other performance impacts. Laboratory work at ADA
Technologies was directed towards characterizing the chemistry of the process and determining
what additives could potentially be combined with the PCIP system to promote the NOX control
process.
The chemistry of the process required better definition to ensure that PCIP was applicable to the
flue gas stream of interest, without producing an undesirable secondary emission. ADA
performed laboratory experiments to determine what products were generated by the pulser and
to evaluate whether a chemical additive would promote the NOX control chemistry.
This laboratory work was incorporated into the design, construction, laboratory testing,
modification, and field testing of a subscale PCEP system. The subscale unit treated a slipstream
of exhaust from an Air Force jet engine test cell. Ideally, the entire subscale test program would
have been conducted on a slipstream from a test cell. However, since the test cell is not operated
at steady-state conditions for extended periods of time, testing of the subscale system was
performed at ADA's laboratory using flue gases generated by a small turbojet engine. These
tests were designed to provide the data necessary to design a full-scale commercial system and
accurately estimate the costs of the system.
It is very difficult to reliably scale up an electrostatic device from one geometric configuration to
another of different size or shape. This does not mean that it is not possible to obtain meaningful
information from subscale testing. It just requires that care be taken in design of the subscale
system to account for such critical dimensions as tube diameter, electrode geometry, electrode
spacing, length of treatment zone, and gas velocity.
The subscale test system design was based around one full-scale corona tube, treating up to 500
acfrn of exhaust. By not changing any critical dimensions, one may easily extend the results
obtained on a one-tube system to a device with many identical components in parallel.
Initial screening tests were conducted using a bench-scale turbojet engine firing Air Force JP-8
fuel. This engine was run at steady state conditions for long periods in order to make the
necessary measurements to characterize the plasma process. These tests provided a data base of
information that was very valuable in designing the full-scale system. In addition, the PCIP
subscale system design was modified based on these test results.
Once sufficient tests at ADA's laboratory were conducted to characterize the process, the
subscale system was interfaced to a slipstream of an operating jet engine test cell at Nellis Air
Force Base. A series of tests were then performed to evaluate the performance of the plasma
system under actual JETC operating conditions.
Pulse-Corona-Induced Plasma Technology
The PCIP process uses a plasma to destroy NOX at relative low temperatures (i.e., < 200 °F). The
process involves application of a very sharp-rising, narrow-pulse high voltage to a corona system
to produce intense streamer coronas, which bridge across the electrode gap. Streamer corona is
the avalanche of electrons generated near a high-voltage wire due to intense gradients in the
-------
electric field. Many characteristics of the process make it ideal for the jet engine test cell
application.
Nonthermal Plasmas. Nonequilibrium plasmas produce chemically active radicals that react
with pollutant molecules to oxidize or reduce the pollutants to more benign or easily collectible
forms. The plasma can be generated by discharge reactors or electron beams. The goal is for
electrons in the gas stream to attain a high temperature (high energy), while ion and molecular
temperatures remain low. In a plasma system it is only the electrons that can produce chemically
active radicals (N-, O-, O2", O2*, O3, OH, etc.). The various plasma excitation processes are
designed to apply the energy directly to accelerating electrons. This selective heating of
electrons is produced by using a very-high-frequency excitation in the MHz and GHz range, or
by extremely short pulses of high voltage.
Research efforts in the late 1970s and early 1980s resulted in the development of the Pulse-
Corona-Induced Plasma Chemical Process (PPCP), which differed from other plasma processes
since it could be applied at normal temperature and pressure conditions. The sharp-rising,
narrow-pulse, high-voltage corona system produces intense streamer coronas, which bridge
across the electrode gap. The corona electrode serves as a stable trigger element of streamer
corona.
With this process, all the energy is used to accelerate only electrons because the duration of the
high electric field is too short to accelerate the ions, which have a much greater mass. The rise in
the ion/molecule temperature through electron bombardment is minimized by providing
sufficient time between pulses to allow cooling of the ions which have been partially heated
through electron collision. The pulse frequency commonly used is in the range of 50 to 500 Hz.
This results in a highly nonequilibrium plasma characterized by very high electron temperatures
and low ion/molecule temperatures. The PPCP process has been proven by Masuda, among
others, to be an effective mechanism for the treatment of NOX, SO2, Hg vapor, volatile organic
compounds, odors, and other hazardous and toxic vapors (2,3,4,5,6).
Previous testing has been performed using laboratory setups and small pilot-scale devices by
researchers. Masuda found that positive corona was much more effective than negative corona.
He concluded that this observation was more a function of the shape of the corona than the
quantity of energy produced. Negative corona forms in individual "tufts" whereas positive
corona is more continuous between the wire and the outer cylinder so that the entire reaction
chamber can be fully utilized for radical formation.
Keping, et al, (7) and testing at ADA Technologies (8) have confirmed the findings of Masuda
that near-complete destruction of NOX occurs at very high field strengths and that positive corona
was 10 times more effective than negative corona. In addition, Keping presented several key
findings regarding NOX destruction:
• NOX removal efficiency is inversely proportional to initial NO, concentration, and greater
than 99% removal can be obtained at concentrations less than 200 ppm.
• NOX removal efficiency increases with increased pulse frequency.
Laboratory reactor experiments also demonstrated that NOX removal was inversely proportional
to temperature and that the best NOX removal occurred at temperatures below 50 °C. All of these
-------
findings indicate that the approach is a good one for NOX treatment of a dilute, low-temperature
gas stream such as JETC exhaust.
Features of the Corona-Discharge Reactor. The PPCP process is a subset of corona-
discharge reactor (CDR) technology. CDR can be viewed as a reaction system in which
electrical energy is delivered to a gas stream to initiate or enhance the rate of beneficial chemical
reactions. Such reactors are also referred to as electrical-discharge tubes or reactors in the
literature and they have been employed in the past to prepare a variety of chemical compounds at
laboratory and small commercial scale.
Other process systems that are similar to this concept include the electrostatic precipitator (ESP),
which is employed to remove particulate matter from flue gas streams, and the ozone generator,
which is used to produce chemically reactive ozone in an air or oxygen feed stream. In practice,
the operation of an ESP typically results in both the collection of particulate matter and the
generation of varying amounts of ozone, and so the ESP may be viewed as both a particulate
emissions control device and a low-efficiency ozonizer.
Both ESPs and ozone generators have been developed to the point where systems are available in
scales ranging from laboratory-scale to full-industrial-scale units. In ESPs, this amounts to
systems capable of handling hundreds of thousands of cubic feet per minute of flue gas, while in
ozone generators, output capacities can be measured in tons per day of ozone in the largest units.
Whether we compare the ESP or the ozone generator to the proposed CDR, the fundamental
similarity resides in the fact that all three systems utilize electrical energy input to promote
chemical reactions.
Table 1 shows several of the primary variables that impact the performance of the CDR and that
are thus candidates for evaluation in optimizing the operation of the reactor to meet specific
goals (in this case, the destruction of NOX in flue gas).
Table 1
Variables Impacting CDR Performance
Applied Voltage Pressure
Frequency and Duration of Applied Current Temperature
Electrode/Ground Configuration Gas Flow Rate
Chamber size Gas Composition
Equipment Description
Figure 1 shows the final layout of the subscale system as configured at ADA's laboratory after
extensive testing and modification. The subsections below describe the major components of the
system.
-------
Gas Turbine
Enclosure
Gas Turbine
Controls
Figure 1
Subscale System as Installed at ADA's Laboratory
Reaction Cell and Pulsed Power Supply
Figure 2 shows an electrical schematic of the pulse-corona generator, which includes a DC high-
voltage (HV) power supply, a pulse generator and a tubular corona-discharge reactor (CDR). A
slipstream of the exhaust gas from the jet engine was forced through the CDR, in which the
plasma was formed by applying a HV pulse to the corona electrode. The exhaust gas
composition, temperature, and flow rate were measured before and after the gas passed through
the tube to determine the effects of the plasma. Various currents and voltages were also
measured to determine the electrical requirements of the system.
The DC HV power supply was a Hipotronics 220-V AC model 8150-65, which has rated
maximum outputs of 150 kV, 65mA, and 10 kW. It consisted of an HV oil-filled tank and a
rack-mounted control panel. The oil-filled tank contained all the HV components such as the
transformer, capacitors, resistor to measure output voltage, and diode strings which can be
reversed to change output polarity. The control panel contained a motor-driven variable-voltage
transformer that controlled the primary voltage on the HV transformer in the tank and, therefore,
the HV output to the pulse generator. The power off/on switches, circuit breakers, HV off/on
switches, external interlock circuits and other safety features were in the control panel. Also, the
DC analog meters in the panel measure the DC HV output voltage (Vdc) and current (!„,.) as
shown in Figure 2. The high-voltage resistor, Rv, in the voltage-measuring circuit was actually
located in the oil tank. The current-sampling resistor, R,, was in the ground return side of the HV
circuit. A data logger was connected in parallel with the two analog meters to automatically
-------
measure HV output voltage (VD) and current (ID). This instrument was of very limited use
because of the large transients encountered in HV pulse generation.
Figure 2
Electrical Schematic of Pulser and Corona Discharge
The pulse generator was built by Ion Physics and was 4 feet wide, 5 feet deep and 12 feet high.
Although the unit was quite large due to the large spacing required for high voltage in air, it
contained essentially three components. The resistor, R,,, in series with the capacitor, Cp,
consisted of three 1.71-megaohm resistors hi series and it limited the capacitor charging
current, ic. Cp consisted of two 400-pF capacitors in parallel and had a maximum voltage rating
of 80 kV. The charging rate for the capacitor can be either increased or decreased by removing
or adding resistors or capacitors in appropriate configurations. The spark gap consisted of two
movable blocks of carbon enclosed in a Lucite™1 chamber filled with a CO, atmosphere. The
spacing between the carbon blocks was adjusted by an external motor, this spacing determines
the voltage at which the spark gap breaks down. A voltage divider was connected across the
capacitor Cp. The output of the voltage divider was connected to an oscilloscope (as shown in
Figure 2) in an attempt to measure the capacitor-voltage waveform vc. Since the voltage divider
was not frequency-compensated, the peak and minimum values of the sawtooth-shaped
waveform could only be estimated. However, the pulse period could be determined from the
waveform but this period was not constant from pulse to pulse because of variations in the
breakdown of the spark gap.
The output of the spark gap was connected to the corona electrode by a '/4-inch rod through an
HV insulator. The corona electrode is inside an 11.75 inch ID steel tube. The tube was carefully
grounded to reduce transient signals that occur on all grounds when the spark gap breaks down or
-------
there is sparking from corona electrode to tube. The corona electrode extended 7.5 feet along the
tube which determined the active volume for the pulse-corona. It is centered by the rod through
the HV insulator at the top and a Teflon"1 bar at the lower end. The corona electrode could be
removed and replaced by electrodes of various diameters and surfaces to determine the effects of
the electrode on pulse-corona generation. Gas flowed into the bottom of the corona chamber and
out the top. The treatment time of the gas could be varied by changing gas velocity.
Instrumentation
Exhaust gas composition measurements were made using Thermo Environmental NOX analyzers
and a Servomex O2 and CO combined analyzer. Type K thermocouples were used throughout
the system for temperature measurements. These parameters were logged continuously during
tests at Nellis AFB on both a Campbell data logger and a Hotmux input box (for temperature
logging by computer). This continuous logging was essential given the rapid changes in flue-gas
composition, flow rate, and temperature that the system underwent during on-site tests. Tedlarm
bag samples for hydrocarbons were analyzed for total methane and total nonmethane
hydrocarbons, and condensation nuclei particle counter measurements of particle size were made
on selected tests. Pressure drop through the CDR was measured with a Magnehelic pressure
gage, and flow rate was determined using pitot traverses and with a hot-wire anemometer.
Jet Engines Tested
The engine which ADA used for testing in our laboratory was an SR-30 turbine manufactured by
Turbine Technologies, Ltd. The turbine is capable of generating 32 pounds of thrust at an
exhaust mass flow of 0.84 Ib/sec. A single-stage radial compressor feeds a reverse-flow annular
combustor can, and the hot gases expand through an axial-vane turbine. This engine provided
exhaust gas flow rates that exceeded our requirements, and a slipstream of the exhaust was
treated in the CDR. The temperature and composition of the gas were representative of JETC
applications, except that the thermal NOX production was low because of the lower temperature
attained in a small-scale turbine. NO was added to the exhaust stream to overcome this and
provide an entirely representative exhaust stream for treatment.
Engines tested at Nellis Air Force Base were all from F-15 and F-16 aircraft. We tested three
engine models: F100-PW-100, -220, and -229. Each of these engines has an afterburner, and is
typically tested at loads from idle through afterburner operation.
Scrubber Design
The initial subscale system design included the option to inject gases or liquids into the flue gas
upstream of the corona-discharge reactor. Since the goal was an economic evaluation of the
technology, chemical reactions which would promote NOX removal and reduce the power costs
associated with the CDR were desirable. Liquid spray prior to the corona was tested and
eliminated because of interference with corona generation and inadequate residence time for the
spray to distribute through the flue gases. The residence time is short because the droplets, as
particles, are precipitated out in the corona-discharge tube.
The final configuration of the spray system, as shown in Figure 1, includes a spray/scrubber
downstream from the CDR. This configuration included a mist-eliminating material, a high-
-------
pressure pump and nozzle injector, and recycling of the spray fluid. Several spray liquor
compositions were tested based on the results of laboratory tests.
An alternative scrubber configuration was a packed-tower design which treated a small fraction
of the total flow through the CDR. This "scrubber tower" test apparatus bubbled gas through
100 to 200 mL of solution. The tower was a 3-inch diameter, 7-inch high cylinder. Evaporative
cooler pad material was layered in the container to provide extended surface area for contact of
liquid with flue gas.
Results and Discussion
Laboratory testing of the system yielded setpoints for electrode design, polarity (positive),
capacitance, peak corona voltage (which was controllable by varying the spark gap distance),
and power supply voltage (kV). Laboratory test variables also included flue-gas flow rate and
NOX concentration, dilution of flue gas with ambient air, turbine load (which affects flue-gas
composition), gaseous and liquid injectants, and scrubbing using the spray tower or solid
sorbents. From these laboratory results, we calculated the eV/molecule NO removed, and
obtained target removal rates for comparison with field tests. These variables were selected in
order to characterize the performance of the system under conditions representative of Air Force
hush houses.
Variables included in the Air Force JETC field tests were: flue-gas flow rate, spray injection and
the scrubber tower, power supply kV, engine load, and three engine types. The measurements
made during these tests are discussed below with the test results.
Testing at ADA's laboratory yielded energy consumptions of 40 to 100 eV/molecule once the
system was optimized. The energy consumption calculated is for conversion of NO to higher
oxides (+4, +5), not necessarily elimination of NOX. We found that about 90% of the NO
entering the pulser would oxidize. The balance was presumably reduced to elemental nitrogen.
We then focused our work on eliminating the N*4 and 1ST5 through chemical scrubbing. Although
the project was not set up to engineer a scrubbing system, we did have enough success with the
configurations tested to prove the concept. The economics of the full-scale system were based
on a spray-tower configuration, and costing was done based on this type of equipment at JETC
scale. Exhaust gas flow rates from a JETC are as high as four million cubic feet per minute,
which is the same order of magnitude as a utility power plant.
PCIP System Design Improvements
Four electrodes were tested. The 7.5-foot-long corona electrodes were installed in the CDR and
tested individually. They were: a smooth 1/8-inch-diameterrod, a coarsely threaded %-inch-
diameter rod, a smooth '/---inch-diameter tube, and a constructed electrode consisting of 1.5-inch-
diameter washers spaced 0.7 inch apart by smooth spacers. Statistical analysis of all data
generated was performed and the electrode design was found to have only a small contribution to
the effectiveness of the system at oxidizing NO. However, the range of electrical conditions that
could be realized without significant sparkover varied for the different electrodes. We selected
the !/4-inch diameter threaded rod as the best electrode, and used this design for the majority of
laboratory tests and for all field tests. Positive polarity corona was found to be much more
effective than negative polarity.
-------
Tests were conducted both on turbine flue gas and on ambient air. Treating turbine flue gas was
found to be much more effective. This may be due to the influence of hydrocarbons in the flue
gas, which has been observed by others to enhance cold plasma's capability to remove NO (9).
Products of the Corona-Discharge Reactor
Initial test results showed that conversion of NO to higher oxides of nitrogen (TST4 and N"5) was
successful using ADA's wire-in-cylinder design. The conversion is a function of energy input,
and at the concentration levels for the jet engine test cell application, 100% conversion is within
the capability of this equipment.
The pulser converts a small percentage of NOX to species other than NO2 or NO3". This small
percentage is seen by our analyzer as a reduction in total NOX. Reductive as well as oxidative
radicals are present in the corona, so the other species to which NO is converted may be N2O5, or
N2. The chemiluminescent NOX analyzer measures total oxides of nitrogen, including NO, and
NO3', separately from NO. The pulser converts most of the NO into these higher oxides.
Elimination of Higher Oxides of Nitrogen. Laboratory testing was directed toward
defining an optimal scrubbing solution. The test procedure was to inject NO2 through a solution
in a bubbler to determine the solution's scrubbing capability. We simultaneously measured the
pH and the NO3" concentration in the solution as well as the inlet and outlet NO, concentration.
Several reagents were tested to determine their scrubbing capability. The three successful
candidates were sodium hydroxide, hydrogen peroxide, and sodium thiosulfate. For example, the
laboratory tests of various peroxide solutions yielded 37 to 40% NO2 removal when bubbling
NO, through the solution. Once these were demonstrated in the laboratory, they were then
combined with the spray- and scrubber-tower designs to determine their success in directly
treating the flue gas stream after treatment by the CDR.
Once oxidation of NO to higher oxides was achieved by the pulser, tests focused on elimination
of the NO2. Tests were typically run with a baseline (inlet) value of 25 ppm NOX in the flue gas,
which is representative of Air Force jet engine test cell conditions. The pulser operation was set
up to convert about 90% of the NO to NO2. The results which were critical to the direction of the
program are discussed below. Several candidate scrubbing techniques demonstrated high
removal rates (>80%) of NOX in the laboratory when combined with CDR exhaust treatment.
Gaseous Injection and Sorbent Beds
Ammonia injection removed NOX when used in conjunction with the pulser; the disadvantage
was that high excess ammonia was required, resulting in gaseous ammonia emission equal or
greater than the NOX emission. Storage and handling of ammonia is also a serious issue for Air
Force bases.
Injection of hydrocarbons into the flue gas was tried on a small scale and did not have major
impact on pulser operation nor on effectiveness of NO-to-NO, conversion. The hydrocarbons
tested were methane and propane, injected into ambient air (fan operation) rather than turbine
flue gases. Concentration ratios of hydrocarbon to NOX ranged from 2:1 to 8:1. However, it is
likely that the unbumed hydrocarbons in the turbine flue gases did improve CDR performance.
This was observed as an influence of turbine load on the test results (lower load corresponding to
higher unbumed hydrocarbons).
10
-------
A sorbent bed was also tested in the sample line to the CEMS. This was a packed cylinder of
material through which flue gas passed prior to measurement by the NOX analyzer. This
experiment successfully removed about 6 ppm (32%) of the NO2 with vermiculite in the sorbent
bed. The residence time of the flue gas in the bed was approximately seven seconds. Pressure
drop was high through the packed sorbent bed, on the order of 20 inches of H2O.
Liquid Scrubbers
The CEM scrubber tower was used to test the successful NO2-scrubbing chemicals on actual flue
gas.
Sodium Thiosulfate: NajSjC^ at 0.1 M concentration in water was the most successful scrubber.
It resulted in removal of as much as 80% of the NOX when used in combination with the pulser.
A proposed mechanism for scrubbing of N02 by sodium thiosulfate is summarized below:
Na^SA + - O2 + NO2 -» 2 Na* + - S4O6" + NO;
•L* 2,
A weaker solution of sodium thiosulfate (0.01 M) was also tested successfully. This resulted in
75% NOX removal. The pH of sodium thiosulfate solution was very close to neutral.
Sodium Hydroxide: NaOH at 0.1 Mwas almost as effective at removing NOX. 70 to 75%
removal was observed. Weaker solutions of 0.01 and 0.05 M were tested also, resulting in 56
and 64% NOX removal, respectively. The pH of the most-dilute NaOH solution was above 11.
Hydrogen Peroxide: 3% H2O2 resulted in greater than 60% NOX removal.
The laboratory tests with solution injected directly into the flue gas showed limited success.
Under conditions which included low flue-gas flow rate and relatively high NO2 concentration,
the spray removed on the order of 10% of the NO2 from the gas stream. Normal laboratory test
conditions, which most closely match a full-scale system, included approximately 25 ppm of
NOX in 300 to 400 acfm flue gas. When flow rates were cut to about a third, and NO2
concentrations were increased to 34 to 55 ppm, the effect of the spray was confirmed (i.e., 10%
N02 removal). These changes effectively decreased the gas-to-liquid ratio, and demonstrated
that an engineered spray system is technically feasible. Based on these results, we selected a
spray tower design for the cost estimate of the full-scale system.
Field Results from Testing at Nellis Air Force Base
A typical firing of a jet engine in a test cell involves operating the engine at several thrust levels
from idle to full load with afterburning. The load on the engine may be held for several seconds
to several minutes before proceeding to the next load. There is a prescribed series of operating
conditions which each engine is put through to obtain data for an Air Force "Acceptance Test."
Figure 3 shows typical emissions from an Air Force jet engine during an Acceptance Test. The
engines tested were Pratt & Whitney engines from F-15 and F-16 aircraft.
-------
220 Acceptance Test 1003C
-NOxout
- NO2 out
• NOxorNOin
-513MO-
M24
Time
Figure 3
Typical NOX Emissions During Acceptance Run and Controlled Emissions
The slipstream was drawn through five 2-inch diameter ports in the 4-inch diameter probe
tubing, through a fan, and then into the corona-discharge pipe. The flue gas then passed through
mist eliminator material to prevent liquid from reaching the corona region, and then through the
spray section. Measurements made during testing used the same equipment as during laboratory
testing, and included the parameters listed below:
PCIP Inlet and Outlet: NO, NOX, CO, O2, particle sizing, hydrocarbons, temperature
PCIP Operation: Supply kV and mA, pulse rate, pressure drop, flow rate
Jet Engine Operation: Load, fuel flow, N2 rpm (high-pressure compressor speed)
In 10 days of on-site testing, 29 tests were conducted. These tests characterized the flue gas from
three engine types, the F100-PW-100, -220, and -229. Each of these engines is routinely tested
in the hush house and is used interchangeably in F-15 and F-16 aircraft. Testing was conducted
during standard hush house operation, called "acceptance" tests, during troubleshooting of
engines, and during periods of extended operation at idle, military, or afterburner loads. These
extended periods of operation were requested by ADA and enabled us to observe the effects of
varying such operating conditions of our system as flow rate, pulser power, and spray injection.
Based on these results, we examined NO oxidation by the pulser and spray/scrubber effectiveness
for NO2 removal, and we were able to target operating conditions for acceptance tests.
A sample of data from one of the final on-site tests is shown in Figure 3. This test run
represented typical acceptance test conditions. NOX removal was on the order of 50% for this
test. Flue-gas flow rate, pulser power, and scrubber liquid composition were set based on
previous test results. Table 2 shows a summary of operating parameters and calculated values
from this test run. These data were a result of optimizing the system over several days at Nellis
to determine gaseous flow rate, pulser power level, and scrubber configuration.
12
-------
Table 2
Summary of PCIP System Operation
Engine
Load
Idle
Military
Afterburner
Inlet NOX
ppm
4
27
36
ppm NO,
Converted
by Pulser
3
17
20
eV/molecule
55
55
Scrubber
effectiveness
%
100
100
100
NOX
Emitted
ppm
1
10
16
Hydrocarbons were measured simultaneously at the pulser inlet and outlet during an engine
startup burning "pickling oil," a preservative used in long-term engine storage. These startups
are quite smoky, since oil must be burned out of the fdel lines for 30 to 50 seconds. The results
of a single Tedlar011 bag pair taken under this condition showed that the CDR system (without
scrubber) was removing nonmethane hydrocarbons. Methane in both inlet and outlet samples
was 2.1 ppmv, but nonmethane hydrocarbons were 10.3 ppmv at the inlet, and <1 ppmv at the
outlet of the CDR. Hydrocarbon bag samples were also taken at steady load conditions on JP-8
fuel. These were taken on two different days, and results varied somewhat. Bags sampled on
September 30 indicated that nonmethane hydrocarbons were decreased by the pulser by one to
three ppm from inlet levels of two to five ppm. Methane hydrocarbons were again unchanged.
Bags sampled on October 2, however, indicated that inlet and outlet nonmethane hydrocarbons
were consistently 0.5 ppm to 1.0, not varying with load. Overall, the data indicate that the CDR
removes nonmethane hydrocarbons from the exhaust stream. They also indicate that volatile
hydrocarbons present in the exhaust are quite low, and vary with operating conditions.
The fast response time and durability of our system did prove to be effective in the JETC
application. We successfully performed sufficient testing at Nellis to develop full-scale
economic projections, as described in the section below.
Conclusions and Economics of Full-Scale Implementation
The system consists of two major components: the CDR, which oxidizes NO to higher oxides of
nitrogen, and the scrubber, which chemically eliminates the N"4and 1ST5 The design is targeted at
four million actual cubic feet per minute flue gas flow, and utilizes data obtained from the Nellis
field test results to determine removal efficiencies and operating costs.
The assumptions used to specify the design of the corona reactor section is based upon a reactor
residence time of 1.3 seconds, a reactor height of 30 feet, 25 ppm NO oxidized by the pulser, 20
ppm NO, removed, and scrubbing with a 0.01 M sodium thiosulfate solution. The gas exiting the
scrubber section is targeted to contain 10 ppm of NO and 5 ppm of N02. The captured NO2 will
exit the scrubber as a dilute nitrate salt.
A mass balance of the system is shown in Figure 4. Table 3 shows a summary of the capital and
operating costs for the system. The energy needed per NO molecule oxidized (55 eV) and the
required oxidation of 20 ppm yields a high-voltage-power requirement of 14 MW. This pulsed
high-voltage power can be obtained from 28 individual 500-kW supplies.
13
-------
THIOSULFATE
SOLUTION
4 E 6 ACFM
35ppmNO
lOppmNO
5 ppm NO2
,UTE
NITRIC ACID
Figure 4
Full-Scale Mass Balance for PCIP and Scrubber System
Table 3
Projected Full-Scale Capital and Operating Costs Summary
Equipment Capital Cost
Capital Recovery Cost
Operation Hours per Week
O&M Yearly Cost
Annualized Cost
$/lb NO, Removed
$109,200,000
14,196,000
10 50
523,600 2,618,000
$14,719,600 $16,814,000
$56 to $343 $11 to $69
A spray tower design for NO2 scrubbing was chosen over a packed design to minimize pressure
drop. To minimize total height (essential near a flight line) and to make use of the corona reactor
outside walls, the scrubber will wrap around the corona reactor. To maintain the concentration of
nitrate in solution at 5%, a flow of 10 gpm will be withdrawn. Since sodium thiosulfate will be
simultaneously depleted, it will be replaced at a rate of 12 Ib/hr.
The annualized cost of the system is mainly composed of capital recovery, which is principally a
function of the overall capital cost. The largest single cost item is power supply ($28,000,000),
and this cost is dependent upon the energy requirement for the oxidation of NO.
The cost per pound of NO,, removed is also shown in Table 3. Since the bulk of the annualized
cost is capital, longer operating hours quickly translate into a more-cost-effective system.
However, as typical JETC operating periods are an hour or two, 50 hours/week is an
unrealistically high number. For comparison, NO., offset values are on the order of $10,000/ton,
or $5/lb.
The range of costs per pound of NOX is attributable to different assumptions on the varying
operating conditions of JETCs. Both concentration of NOX and flue-gas flow rate vary
enormously, making an integral average difficult to estimate without continuous measurement.
For an emission rate of 20 tons NOX per year, the cost is estimated to be $343/lb NOX. For the
maximum emission rate possible in 10 hours of weekly operation (continual operation at full
14
-------
load, which is not normal JETC operation), the cost drops to $56/lb NOX. The actual cost for a
typical JETC would likely fall between these two estimates.
Other Applications ofCDR Technology
While the JETC application is not an ideal application of this technology, clearly some other
types of gases will be treatable using this technology. The results we obtained in the field were
very consistent with those from the laboratory, an indication that the design we selected is
scaleable. This is encouraging, and we are continuing to examine other potential applications
such as destruction of air toxics, and simultaneous destruction of more than one pollutant.
Acknowledgments
The authors would like to thank Dennis Helfritch of Environmental Elements Corporation for his
help on this program, including analysis and writeups of laboratory and economic data. We
would also like to thank the hush house crew at Nellis Air Force Base.
References
(1) Kimm, L. T., E.R. Allen, and J.D. Wander, "Control of NOx Emissions from Jet Engine Test
Cells," Air & Waste Management Association, 88th Annual Meeting, 1995.
(2) Masuda, S. et al. "Novel Plasma Chemical Technologies - PPCP and SPCP for Control of
Gaseous Pollutants and Air Toxics," 1993.
(3) Masuda, S. and H. Nakao, "Control of NO by Positive and Negative Pulse Corona
Discharges," IEEE-IAS Annual Conf, Denver, CO, 1986.
(4) Clements, J.S., A. Mizuno, W.C. Finney, and R.H. Davis, "Combined Removal of NOX and
SO, from Simulated Flue Gas using Pulsed Streamer Corona," EEEE Trans, on IAS, 1989,
Vol. 25, No. 1, pp. 62-69.
(5) Dinelli, G.L. Civitano, and M. Rea, "Industrial Experiments on Pulse Corona Simultaneous
Removal of NOX and SO2 from Flue Gas by Means of Impulse Energization," IEEE-IAS
Annual Conf., Pittsburgh, Pa., 1988.
(6) Yamamoto, T., P.A. Lawless, K. Ramanathan, D.S. Ensor, G.H. Ramsey, and N. Plaks,
"Application of Corona-Induced Plasma Reactors to Decomposition of Volatile Organic
Compounds," 8th Symposium on the Transfer and Utilization of Particulate Control
Technology, March 20-23,1990, San Diego, California.
(7) Keping, et al., "Removal of NOX and SO2 by Bipolar Corona," The 4th International
Conference on Electrostatic Precipitation, Sept 14-17,1990, Beijing, China.
(8) Durham, M.D., et al, "Application of Pulse-Corona-Induced Plasmas for Control of NOX
from Jet Engine Test Cells," AL/EQ-TR-1994-0020, Armstrong Laboratory, Tyndall AFB,
June 1994.
(9) Vogtlin, G.E., and B.M. Penetrante, "Pulsed Corona discharge for Removal of NOX from
Flue Gas," in Nonthermal Plasma Techniques for Pollution Control. B.M. Penetrante and
S.E. Schultheis, eds, Springer-Verlag 1993.
15
-------
Tuesday, August 26; 8:00 a.m.
Parallel Session C:
Low NOx Systems for Gas/Oil-Fired Boilers
Continued
-------
APPLICATIONS OF REACH TECHNOLOGY TO REDUCE NOX AND
PARTICULATE MATTER EMISSIONS AT OIL-FIRED BOILERS
Dan V. Giovanni
Robert C. Can-
Electric Power Technologies, Inc.
830 Menlo Avenue, Suite 201
Menlo Park CA 94025
Abstract
Reduced Emissions and Advanced Combustion Hardware (REACH) Technology has
been retrofitted to 115 gas- and oil-fired boilers of different designs and capacities to
reduce NOx and particulate matter (PM) emissions. The total installed capacity using
REACH is presently 17,000-MWe worldwide. Combustion and emissions performance
equivalent to that available for new, low-NOx burners has been achieved at costs
between $0.25 and $l/kW depending on the unit size. Two versions of REACH are
commercially available: (1) Combustion Performance REACH (CP-REACH) for solving
a variety of site-specific boiler performance, maintenance, and operating problems
related to poor combustion conditions, and (2) Low-NOx REACH (LN-REACH) which
provides simultaneous reductions in NOx and PM emissions, while retaining the
performance and operating advantages of CP-REACH. This paper describes recent LN-
REACH applications using Segmented V-Jet atomizers (patented) at Collins Unit 4 of
ComEd, a 550-MW, opposed-wall fired boiler, and at Kahe Unit 6 of Hawaiian Electric
Company (HECO), a 140-MW, single-wall-fired boiler. Also described is a CP-REACH
application to reduce PM emissions at a 365 metric tons/hr (approximately 100-MW),
tangential-fired boiler at the Montova Plant of Frene in Italy.
Results at Collins Unit 4 showed that LN-REACH reduced NOx emissions from 0.55
to 0.30 Ib/MBru (883 to 481 mg/Nm3) at 500-MW without overfire air (OFA) and flue
gas recirculation (FGR). The application of OFA and 7% FGR further reduced NOx
emissions to 0.21 Ib/MBtu (338 mg/Nm3) with 10% opacity. At Kahe Unit 6, NOx
emissions at full load were reduced 13% below levels achieved with a combination of
Ist-generation LN-REACH, OFA and FGR, i.e., from 0.22 to 0.19 Ib/MBtu (353 to 306
mg/Nm3) with 10% opacity. For the tangential-fired boiler, CP-REACH reduced PM
emissions from 0.11 to 0.04 Ib/MBtu (200 to 60 mg/Nm3) without an increase in NOx
emissions (which were approximately 0.45 Ib/MBtu, or 725 mg/Nm3).
-------
Introduction
Reduced Emissions and Advanced Combustion Hardware (REACH) Technology
was developed jointly by Electric Power Technologies, Inc., (EFT), the Electric Power
Research Institute (EPRI), the Empire State Electric Energy Research Corporation in
New York (ESEERCO), and Consolidated Edison Company of New York for retrofit
to existing burners to solve a variety of site-specific boiler problems related to poor
combustion conditions, including high stack opacity, high unburned carbon and
NOx emissions, acidic stack fallout, flame impingement, poor boiler turndown, and
high excess oxygen^.
The main elements of REACH are oil atomizers and flame stabilizers. The design
philosophy for REACH was that it: (1) can be retrofit to the existing range of burner
and boiler designs to maximize applicability, and (2) retain as much as possible of
the original burner to realize cost advantages relative to other retrofit options (e.g.,
complete burner replacement). Consistent with this philosophy, REACH adapts to
the major existing components of a burner.
Oil Atomization
REACH oil atomizers are custom-designed to adapt to the existing oil supply
conditions (i.e., pressure, temperature, and capacity)'2'. For steam-atomized systems
REACH uses internal-mix (I-Mix) atomizers, which produce superior spray quality
compared to other common atomizer designs.1
For reducing NOx emissions in steam-atomized systems a novel atomizer design
was developed — the Segmented V-Jet atomizer (patented) — which divides the oil
spray into distinct segments at the base of the flame. For mechanically-atomized
burners, REACH oil atomizers can be designed to operate at supply pressures from
200 to 1,300 psig. Special low-NOx mechanical atomizers that produce oil spray
characteristics conditions similar to the Segmented V-Jet atomizer are also available.
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<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
quality of the oil spray is characterized by Sauter Mean Diameter (SMD), which is defined as
the diameter of a hypothetical droplet that has the same surface-to-volume ratio as that of the total
spray. SMD is most often used to characterize oil sprays because it is relevant to evaporation and
combustion of oil droplets.
-------
entrainment and swirl of combustion air at the discharge plane of the burner, and to
match the oil spray of the REACH oil atomizer.
REACH Technology
Two versions of REACH have been applied in a number of commercial
applications. 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. Detailed descriptions of these technologies and
commercial applications have been published elsewhere^-6).
Adapting REACH to existing boilers typically requires the custom design of
retrofittable oil atomizers and flame stabilizers. The major components of the
burner are retained, and significant changes to air registers, burner auxiliary
equipment, pumping and heating equipment, or combustion controls are not
required. In some instances, the burners have been changed from mechanical to
steam atomization. EPT has designed and supplied REACH for more than 115
boilers totaling 17,000-MWe of generating capacity. In a typical retrofit project,
REACH atomizers and flame stabilizers can typically be supplied within 6-8 weeks,
and installed during a 3-5 day boiler outage. Alternatively, REACH technology may
be incorporated in the design of new burners.
REACH technology has been licensed from EPRI by EPT, COEN Company, and
Ansaldo Energia.
This paper describes results from three recent REACH applications. Two are LN-
REACH projects: (1) Collins Unit 4 of ComEd, a 550-MW, opposed-wall fired boiler;
and (2) Kahe Unit 6 of Hawaiian Electric Company (HECO), a 140-MW, single-wall-
fired boiler. The third is a CP-REACH application at a 365 metric tons/hr
(approximately 100-MW), tangential-fired boiler at the Montova Unit B6 of Frene in
Italy.
Boiler Descriptions
Descriptions of the three boilers which were retrofitted with REACH are provided
below.
-------
Collins Unit 4 of ComEd
Collins Unit 4 is an opposed-wall-fired boiler manufactured by the Babcock &
Wilcox Company and operated by ComEd. The unit has a gross generating capacity
of 550-MW, and is equipped with windbox flue gas recirculation (FGR) and overfire
air (OFA). The boiler was originally designed to burn No. 6 fuel oil (only), and is
equipped with 28 dual-register burners. The burner arrangement is two elevations
of seven burners each on the front and rear walls. A dividing plate in the windbox
separates the two burner elevations. However, there are no dampers to bias
combustion air between the burner elevations. There are seven OFA ports on each
firing wall, with one port above each burner column. Combustion air for the OFA
ports is supplied from the windbox for the top burner elevation. When 100% open,
the OFA system is designed to divert approximately 11% of the total combustion air
flow to the OFA ports. The unit was converted to gas and oil operation by EPT in
1996W>.
CP-REACH was retrofitted by EPT in 1991 to improve oil atomization and
combustion performance. The retrofit included: (1) replacement of Y-jet atomizers
with internal-mix atomizers, (2) replacement of diffusers with compound-curve-
vane swirlers for flame stabilization, and (3) conversion of the atomization steam
system from constant steam pressure at 150 psig to constant steam-to-oil differential
pressure of 10 psid over the load range. As part of a gas conversion project
performed in 1996, EPT retrofitted LN-REACH flame stabilizers and low-NOx,
Segmented V-Jet oil atomizers. Results with the Segmented V-Jet oil atomizers are
presented in this paper.
Kane Unit 6 of Hawaiian Electric Company
Kahe Unit 6 is a single-wall-fired boiler manufactured by the Babcock & Wilcox
Company and operated by HECO. The unit has a gross generating capacity of 146-
MW, and is equipped with windbox FGR and OFA. The OFA system was designed
to divert up to 30% of the total combustion air to six OFA ports located on the front
and rear boiler walls (three ports per wall). The boiler burns No. 6 fuel oil (only),
and is equipped with nine PG-DRB burners in a 3 x 3 array on the front wall(7).
The original oil atomizers supplied by the boiler manufacturer were capable of
meeting the NOx emissions limit of 0.23 lb/MBru (370 mg/Nm3) when used with
maximum FGR and OFA. However, there was no operating margin with the 20%
opacity limit to allow for variability in operation or oil properties. First-generation,
LN-REACH atomizers were retrofitted by EPT in 1990 to improve oil atomization
and combustion performance. Results showed that the NOx emissions limit could
be met with reduced levels of OFA and FGR and with significantly lower opacity (10-
13%). In the spring of 1997, EFT retrofitted low-NOx, Segmented V-Jet oil atomizers.
The objective was to increase the margin of compliance with the NOx and opacity
limits. Results with the Segmented V-Jet oil atomizers are presented in this paper.
-------
Mantova Unit B6 of Frene in Italy
Mantova Unit B6 is a tangential-fired boiler manufactured by Franco Tosi and
operated by Frene of the ENI Group in Italy. The unit has a steam generating
capacity of 365 metric tons/hr (approximately 100-MW), and is not equipped with
windbox FGR. However, the boiler does have a small, close-coupled OFA
compartment above each burner in the top elevation. The boiler is capable of
burning No. 6 fuel oil and natural gas, and has four burner elevations and 16
burners. Each burner has single fuel-air compartment and auxiliary-air
compartments immediately above and below the fuel-air compartment. The
auxiliary-air compartments between the 2nd/3rd and 3rd/4th elevations are bricked
shut. To avoid excessive opacity (smoking) the unit is normally operated with the
fuel-air compartment dampers 100% open, and the functional auxiliary-air
compartment dampers virtually dosed (i.e., 10% open for cooling).
In 1997 EPT installed CP-REACH flame stabilizers and internal-mix oil atomizers to
reduce PM emissions. The CP-REACH retrofit included: (1) replacement of Y-jet
atomizers with internal-mix atomizers, (2) replacement of diffusers with
compound-curve-vane swirlers and extenders for flame stabilization, and (3)
conversion of the atomization steam system from constant steam pressure at 150
psig to constant steam-to-oil differential pressure of 10 psid over the load range, and
(4) reactivation of the close-coupled OFA. To increase swirler flow entrainment,
extender assemblies (i.e., bluff-body rings) were attached to the exit of the fuel-air
nozzles to increase airflow turbulence and promote the formation of a strong
internal recirculation zone.
REACH Characterization Tests and Measurement Methods
The REACH installations at the three boilers were commercial applications.
Consequently, tests to optimize combustion and emissions performance
emphasized development of operational guidelines over the load range for the
plant operators, instead of parametric tests to characterize the sensitivity of NOx and
PM emissions to FGR, OFA, excess O2, etc.
At Collins Unit 4, a multi-point extractive system was used to obtain flue gas
samples for analysis of NOx, CO, CO2, and 62. Samples were extracted from probes
installed in the "A" (north) and the "B" (south) flue gas ducts immediately
upstream of the air heaters. Gaseous samples were obtained from probes installed
in four ports in each duct. Three sampling probes were installed in each port (long-,
intermediate-, and short-length probes). Thus, samples were obtained from 24
points in the flue gas duct, ensuring a representative sample of the boiler exhaust
gases. Twenty-four sample lines were strung from the probes to a mobile emissions
monitoring laboratory that was located at the ground level adjacent to the unit. The
gas samples were conditioned to remove water, and were then directed to emission
monitors.
-------
NOx, CO, CO2, and O2 were measured using EPA continuous monitoring procedures,
and EPA certified calibration gases. O2 was measured with a Servomex Model 570A
instrument, CO2 with a ACS Model 3300 instrument, CO with a TECO Model 48
instrument, and NOx with a TECO Model 10A chemiluminescence instrument.
Opacity was measured with the plant continuous emissions monitor system (CEMS).
At Kahe Unit 6 and Montova Unit B6, the plant CEMS were used for measurement
of NOx, excess O2, and opacity.
Results
The objectives of the LN-REACH retrofits at Collins Unit 4 and Kahe Unit 6 were to
reduce NOx emissions without increasing PM emissions and opacity. The CP-
REACH retrofit at Montova Unit B6 was intended to reduce PM emissions without
increasing NOx emissions. Results are presented below.
Collins Unit 4
As described above, CP-REACH was retrofitted at Collins Unit 4 in 1991 to improve
oil atomization and combustion performance. The CP-REACH retrofit eliminated
high opacity which caused derating of the unit.
As part of the gas conversion project at Collins Unit 4, EPT supplied new LN-
REACH flame stabilizers and replaced the CP-REACH internal-mix atomizers with
Segmented V-Jet atomizers. Figure 1 compares NOx emissions vs. load for CP-
REACH and LN-REACH, and also shows results from parametric tests to
characterize the effects of OFA and FGR on NOx emissions. For CP-REACH, NOx
emissions without OFA and FGR varied from 0.35 Ib/MBtu at 300-MW to
approximately 0.6 Ib/MBtu at 540-MW. The downward-pointing arrows at 525-MW
show the beneficial effects of: (1) opening the OFA ports, which reduced NOx
emissions to approximately 0.46 Ib/MBtu, and (2) combining OFA with
approximately 7% FGR (using one GR fan), which further decreased NOx emissions
to 0.31 Ib/MBtu. NOx emissions less than 0.30 Ib/MBtu could be achieved with CP-
REACH at full load by operating two GR fans to deliver 10-12% FGR.
With LN-REACH, NOx emissions without OFA and FGR varied from 0.18 Ib/MBtu
at 300-MW to approximately 0.30 Ib/MBtu at 500-MW (the maximum load possible
during the tests due to boiler feed pump problems). The downward-pointing arrows
at 500-MW show that OFA reduced NOx emissions to 0.26 Ib/MBtu, and OFA
combined with approximately 7% FGR (one GR fan) further decreased NOx
emissions to 0.21 Ib/MBtu. Opacity was less than 10% over the load range.
Emissions compliance tests performed with LN-REACH at a later date showed NOx
emissions of 0.25 Ib/MBtu at 550-MW with the OFA ports open and 7% FGR.
Opacity was 10-13%.
-------
+*
m
in
o
w
tn
Ul
x
O
0.1
100
200
300 400
Load, MW
500
600
Figure 1. NOx emissions vs. load for Collins Unit 4 with CP-REACH (open symbols)
and LN-REACH with Segmented V-Jet atomizers (closed symbols) The solid lines
represent baseline NOx emissions without FGR and OFA. The singular data points
show NOx reductions achieved at 540-MW for CP-REACH and 500-MW for LN-REACH
with OFA (squares) and OFA + 7% FGR (triangles). The Segmented V-Jet atomizers
alone (baseline) reduced NOx emissions by 45%. Adding OFA and FGR further reduced
NOx emissions by 32%.
-------
The reductions in NOx emissions with LN-REACH were substantial. At 500-MW
without OFA and FGR, LN-REACH delivered a 45% reduction in NOx emissions
compared to CP-REACH. With OFA and 7% FGR, LN-REACH achieved a 32%
reduction in NOx emissions compared to CP-REACH. A major advantage of LN-
REACH in this application was that NOx emissions compliance was achieved
without OFA and FGR below 490-MW. At higher loads, plant operators had the
flexibility to choose either OFA or FGR to meet the NOx emissions limit depending
upon steam temperature or other operational considerations. In any case, there was
sufficient NOx and opacity compliance margin available to allow for variations in
fuel composition (e.g., fuel nitrogen) and boiler operation (e.g., only one GR fan).
The experience at Collins Unit 4 showed that for boilers not equipped with OFA and
FGR, LN-REACH can achieve significant reductions in NOx emissions at very low
cost, i.e., about $0.25/kW.
Kahe Unit 6
LN-REACH Segmented V-Jet atomizers were installed in Kahe Unit 6 in April 1997.
Figure 2 compares NOx emissions vs. load for the Segmented V-Jets and the first-
generation LN-REACH atomizers installed in 1990. The data were obtained with
approximately 15% OFA and 12% FGR. Results showed that the Segmented V-Jet
atomizers reduced NOx emissions at maximum load from 0.22 to 0.19 Ib/MBtu
(approximately 13%). At lower loads, the reduction in NOx emissions by using the
Segmented V-Jet atomizers was approximately 20%. As was the case at Collins Unit
4, the lower NOx emissions with LN-REACH allowed plant operators to meet the
NOx and opacity limits with less than the maximum OFA and FGR, thus providing
a margin to allow for variations in fuel composition and boiler operation.
Montova Unit B6
CP-REACH flame stabilizers, extenders, and internal-mix atomizers were installed
in Montova Unit B6 in January 1997, and tests to characterize particulate matter
(PM) and NOx emissions were performed in February 1997. As summarized in
Figure 3, PM emissions with CP-REACH were reduced by 70% to 56 mg/Nm3 (0.036
Ib/MBtu) from 200 mg/Nm3 (0.11 Ib/MBtu). NOx emissions remained essentially
unchanged at 715-735 mg/Nm3.2
2The NOx and PM emissions reported for the original operating condition were "typical values"
provided by Frene based on historical data, and were not results from specific tests performed as part of
the REACH retrofit.
-------
0.30
0.25
0.20
3
m
2
o
w
.2
LU
X
O
0.15
0.10
0.05
I I
Residual Fuel Oil
Fuel Nitrogen = 0.35%
NOx Limit
1 st-Generation
LN-REACH
LN-REACH
Seg. V-Jets
20 40 60 80 100 120 140 160 180 200
Load, MW
Figure 2. NOx emissions vs. load for Kahe Unit 6 with Ist-generation LN-REACH (open
symbols) and LN-REACH with Segmented V-Jet atomizers (closed symbols). The data
were collected with 15% OFA and 12% FGR to the windbox. The Segmented V-Jets
reduced NOx emissions by 13% compared to the Ist-generation LN-REACH atomizers.
Opacity at full load was 11-13% for the Ist-generation LN-REACH compared to 8-10%
with the Segmented V-Jets.
-------
0.15
0.10
m
2
£>
2
CD
v+
Original
CP-REACH
CP-REACH w/OFA
Figure 3. PM and NOx emissions at Montova Unit B6 with the original combustion
equipment (left) and CP-REACH. The CP-REACH data shown on the far right were
obtained with the dampers 50% open on the close-coupled OFA compartments above the
top burner elevation. Prior to installation of CP-REACH it was not feasible to open the
OFA compartment dampers due to high opacity.
-------
Figure 3 also shows the effect on NOx emissions of opening the auxiliary-air
compartments above the top burner elevation. As discussed previously, to avoid
high opacity the boiler was normally operated with the fuel-air compartment
dampers 100% open and the functional auxiliary air compartments 10% open (for
cooling) on all the burners. As shown in the figure, setting the auxiliary air
compartments above the top burner elevation to 50% open reduced NOx emissions
by approximately 9% from 0.45 to 0.41 Ib/MBtu (725 to 660 mg/Nm3) with a very
slight increase in PM emissions from 56 to 69 mg/Nm3 (0.036 to 0.045 Ib/MBtu).
This result was not surprising, and suggests that further biasing of combustion air to
the auxiliary-air compartments in the top elevation or application of LN-REACH at
Montova Unit B6 would be effective in lowering NOx emissions.
Conclusions
LN-REACH applications using Segmented V-Jet atomizers (patented) at Collins Unit
4, a 550-MW, opposed-wall fired boiler, and Kahe Unit 6, a 140-MW, single-wall-
fired boiler, showed significant reductions in NOx emissions without increases in
opacity or PM emissions. At Collins Unit 4, LN-REACH reduced NOx emissions at
500-MW from 0.55 to 0.30 Ib/MBtu (883 to 481 mg/Nm3) without OFA and FGR.
The application of OFA and 7% FGR further reduced NOx emissions to 0.21
Ib/MBtu (338 mg/Nm3) with 10% opacity. At Kahe Unit 6, NOx emissions were
reduced 13% below levels achieved with a combination of Ist-generation LN-
REACH, OFA and FGR, i.e., from 0.22 to 0.19 Ib/MBtu (354 to 306 mg/Nm3) with
10% opacity. At both plants the lower NOx emissions with LN-REACH allowed
plant operators to meet the NOx and opacity emissions limits over the load range
with less than the maximum OFA and FGR, and thus providing a margin to allow
for variations in fuel composition and boiler operation.
CP-REACH flame stabilizers, extenders, and internal-mix atomizers were installed
in Montova Unit B6 of Frene in Italy in January 1997. Compared to emissions with
the original combustion equipment, PM emissions with CP-REACH were reduced
by 70% to 56 mg/Nm3 (0.036 Ib/MBtu) from 200 mg/Nm3 (0.11 Ib/MBtu). NOx
emissions remained essentially unchanged at 715-735 mg/Nm3.
-------
References
1. Giovanni, D.V., M.W. McElroy, and S.E. Kerho, "REACH: A Low-cost Approach
to Reducing Stack Emissions and Improving the Performance of Oil-fired
Boilers," EPRI/EPA Joint Symposium on Stationary Combustion NOx Control,
Kansas City, Missouri (May 1995).
2. Kerho, S. E., and D. V. Giovanni, "Atomizer and Swirler Design for Reduced
NOx and Particulate Emissions," EPRI Workshop on NOx Controls for Utility
Boilers, Boston, Massachusetts (July 1992).
3. Giovanni, D.V., R.C. Carr and S.E. Kerho, "Reduction in NOx Emissions by
Retrofit of Low-NOx Atomizers on a 550 MW Oil-fired Boiler," Third
International Conference on Combustion Technologies for a Clean
Environment, Portugal (July 1995).
4. Carr, R. C, Marco Alberti, and Christopher J. Nagel, "Retrofit of Gas Combustion
Equipment and Low-NOx Oil Atomizers at a 550-MW Oil-Fired Utility Boiler,"
International Joint Power Generation Conference, Houston, Texas (October 1996).
5. Conti, A.V., S.E. Kerho and J. Lucente, "Low Cost Retrofit Combustion Hardware
for Emissions Control on Industrial Boilers," ASME/EPRI International Joint
Power Generation Conference & Exposition, Minneapolis, Minnesota, (October
1995).
6. Conti, A.V. and J. Lucente, "Reduction of NOx Emissions on Oil Firing at
Bowline Point Unit No. 2," ASME/EPRI International Joint Power Generation
Conference & Exposition, Minneapolis, Minnesota (October 1995).
7. Kerho, Stephen, E., Dan V. Giovanni, et al, "Reduced NOx , Particulate, and
Opacity on the Kahe Unit 6 Low-NOx Burner System," 1991 EPA/EPRI Joint
Symposium on Stationary Combustion for NOx Control, Washington D.C.,
March 1991.
-------
REDUCING NOx EMISSIONS IN A NATURAL GAS-FIRED
UTILITY BOILER USING COMPUTATIONAL FLUID DYNAMICS
G. H. Richards, W. Chen and M. A. Toqan
ABB Power Plant Laboratories
Combustion Engineering, Inc.
Windsor, CT 06095-0500
K.Horn
Pacific Gas & Electric
San Francisco, CA 94177
V. Blande and R. H. Sirois
Stone and Webster Engineering Corporation
Englewood, CO 80111-2137
Abstract
The Delta Power Plant, Pittsburg Unit #7 was simulated using computational fluid dynamics
with the goal of reducing NOx emissions and improving combustion efficiency. Pittsburg Unit #7
is a 700 Mw boiler burning natural gas with a tangential firing system, separated overfire air, and
flue gas recirculation (FOR). Under standard operating conditions at MCR, typical NOx and CO
emissions at 3% O2 were 75 ppm and 65 ppm, respectively. Baseline test data obtained from
the unit at loads of 650,450, and 170 Mw was used to calibrate the numerical model. The model
was then used to investigate the effect of operational parameters, e.g., the quantity and location
of FOR, the amount of SOFA, etc., on the boiler emissions of NOx and CO. Model results
suggest that a 50% reduction in NOx emissions is possible without exceeding the CO target or
negatively impacting boiler performance. A physical cold flow model was generated to study and
improve the mixing of the FGR and the combustion air entering the windbox. Post modification
testing of Unit 7 achieved NOx emissions of less than 40 ppm.
Introduction
A project was initiated in 1996 by Pacific Gas and Electric with the objective of reducing NOx
emissions and improving combustion efficiency of the Delta Power Plant, Pittsburg Unit 7.
Pittsburg Unit 7 is a 700 Mw boiler burning natural gas with a tangential firing system, separated
overfire air, and flue gas recirculation (FGR). Under standard operating conditions at MCR,
typical NOx and CO emissions at 3% O2 were 75 ppm and 65 ppm, respectively. The project
-------
goal was to reduce NOx below 50 ppm without increasing CO emissions or negatively impacting
boiler performance and to do so with the minimum possible capital cost. To achieve this goal, a
program was developed that included baseline testing in Unit 7 and both physical flow and
computational fluid dynamics modeling. Stone and Webster Engineering Corporation was a
consultant on the project and oversaw the baseline testing as well as the computational fluid
dynamics CFD (performed at ABB Power Plant Laboratories) and physical flow (performed at
NELS in Buffalo, NY) modeling efforts.
Unit 7 is a tangentially fired, Combined Circulation® unit capable of producing 5,360,400 Ibs/hr
of steam at 3,818 psig and 1,005 °F and reheat 4,510,240 Ibs/hr of steam from 585 °F to 1005 °F.
Natural gas is fired through 5 burner elevations, each containing 4 gas nozzles. Combustion air is
supplied through two separate ducts by forced draft fans. In each duct, the combustion air
passes through an air heater, after which approximately 12% of the air is diverted for the
separated overfire air (SOFA). The remaining combustion air is mixed with flue gas (FOR) before
being injected into the windbox. Unit 7 typically operates with 30% FOR and 2% excess
oxygen.
Unit 7 baseline testing, performed at 650, 450, and 170 MW suggested flow rate and gas
composition imbalances in the furnace. Velocity traverses were performed in the air ducts to
determine the quantity of overfire air and FGR to each side of the furnace and the flow rate of
combustion air/FGR entering each comer of the furnace. Flow imbalances of up to 20% were
measured from comer to comer. Gas species measurements in each of the burner compartments
revealed that the FGR was not uniformly mixing with the combustion air before entering the
windboxes.
A physical cold flow model of Unit 7 was constructed to simulate the mixing of the combustion
air and the FGR. The model extended from the outlet flanges of the FD and FGR fans through
the furnace to the cavity above the boiler's primary superheater and reheater sections. The
model was used to identify potential modifications that would optimize the unit's air/FGR flow
balance to improve combustion performance and allow reduction of NOx emissions. A numerical
model of Pittsburg Unit 7 was generated using a commercially available CFD code and ABB
Combustion Engineering, Inc.'s (ABB CE) proprietary NOx model. The numerical model was
used to investigate the effect of potential operating condition modifications on NOx and CO
emissions. The model was used to estimate the potential NOx reduction that could be obtained
by properly mixing the FGR and the combustion air and to determine the effect of other potential
modifications, including: FGR biasing between burner elevations, fuel biasing, and quantity of
separated overfire air (SOFA).
This paper describes both the physical cold flow model and the numerical model of Pittsburg
Unit 7, compares the numerical model predictions to the baseline test measurements, and
presents the results of the parametric study to determine the effect of potential operating
variables on emissions. Results are also presented from the field testing after the implementation
of the improved FGR mixing system.
-------
Physical Cold Flow Modeling
A physical cold flow simulation model (CFM) of Unit 7 was constructed and tested to simulate
the existing combustion air/FGR flow balance through the unit's burners and SOFA ports. The
principle objective of this model study was to identify potential modifications that would
optimize the unit's air/FGR flow balance to improve combustion performance and allow
reduction of NOx emissions without significantly increasing draft system pressure drop.
This objective appeared reasonable, based on test data obtained from Unit 7 that revealed a
maldistribution of FOR concentration in the unit's windbox at the four burner comers (Figure 1).
Additional measurements in the combustion air/FGR mixture flowing to each of the unit's twenty
individual burners (five burners per comer) also confirmed the poor distributions presented in the
figure. These data indicated that proportionally higher recirculated flue gas concentrations were
present near the bottom of the front burner windboxes resulting in higher FGR rates in the
bottom burners compared to the middle and upper burners in these comers. The FGR gradient
was substantially different in the rear burner windboxes showing much lower FGR flow rates to
the center of each windbox compared to either the upper or lower region. The poor mixing of the
FGR and the combustion air led to measured oxygen concentrations ranging from 14.2 to 20.8%
on a dry basis. The CFM was used to confirm and further evaluate these findings and to develop
draft system flow management modifications for the full scale unit
The physical model used in this work was constructed primarily of clear plastic and sheet metal
and was built to a 1/9 scale. The model extended from the outlet flanges of the FD and FGR fans
through the furnace to the cavity above the boiler's primary superheater and reheater sections.
Combustion air and recirculated flue gas were simulated in the model by ambient air and heated
air, respectively. Flow visualization techniques were used in the model to aid in identification of
gross imbalances while velocity and static pressure measurements were used to document
improvements resulting from iterative draft system modifications. These measurements were
also used to confirm that the specification requirements regarding flow uniformity and acceptable
pressure drop increases were met.
Tests of the CFM confirmed field measured FGR concentration distributions and validated the
model's baseline simulation accuracy. Subsequent additional tests identified key factors affecting
FGR distribution and allowed development of recommended modifications that could be applied
to the full scale unit's draft system. These modifications consisted principally of minor
alterations to the FGR supply ducts and nozzles to improve injection velocity and distribution
of recirculated flue gas as it entered the combustion air ducts upstream of the burners. The
results of these modifications to the full scale unit will be discussed later in this paper.
Numerical Model Description
-------
A numerical model of Pittsburg Unit 7 was generated to identify the potential reduction in NOx
emissions as a result of improving the mixing of the FOR and the combustion air and to evaluate
increasing the quantity of FOR and the effect of the amount of SOFA on NOx reduction. The
model was used to determine the potential NOx reduction of the different concepts and their
effect on CO emissions and superheater tube metal temperatures. The simulations were
performed using a commercially available CFD code. NOx predictions were made using a
proprietary ABB CE NOx model that was incorporated into the CFD code as a post processor.
The CFD code is a general purpose computer code for modeling fluid flow, heat transfer, and
combustion for a user specified geometry. It uses a control volume-based technique to solve the
differential equations governing the reacting flow problem and uses the SIMPLE (Semi-Implicit
Method for Pressure-Linked Equations) algorithm (1) to solve for the pressure-velocity coupling.
Eulerian transport equations are solved for pressure, each of the components of velocity (U, V,
and W), turbulence, enthalpy, radiation, and gas species. The k-e turbulence model (2) was used
for all simulations performed in this study. Conservation equations were solved for both the
product and reactant species (natural gas, O2, CO, CO2, and H20), while N2 was calculated by
difference. A generalized finite rate formulation was used to model the complex chemical
reactions. A 2-step reaction mechanism was used with the natural gas reacting with O2 in the
first step to form CO, which was subsequently oxidized to CO2. For each reaction, both the
Arrhenius rate and the mixing rate from the eddy breakup model of Magnussen and Hjertager (3)
are calculated. The reaction rate is assumed to be the limiting (slowest) rate, which is then used
to calculate the source terms for the species and enthalpy equations.
Unit 7 was simulated using a body-fitted grid of approximately 250,000 cells to resolve the
furnace and firing system geometry. The unit was simulated up through the vertical outlet plane
of the furnace, including the superheater division panels, the superheater platen, and the
superheater pendant. The grid was extended at the furnace outlet to prevent flow recirculation
which would prevent overall convergence of the model. The windbox was modeled with 4 gas
nozzles per burner compartment, resulting in 20 gas nozzles per comer. The free area of each gas
nozzle was maintained in the model, although the aspect ratio of the nozzles was decreased to
minimize the number of grid cells required to model the windbox. The species composition of the
combustion air/FGR mixture was assumed to be constant for each burner compartment. For the
baseline test simulations, each burner compartment was assumed to have the measured gas
species composition.
The superheater division panels were modeled as walls with a specified steam temperature,
thermal resistance, and emissivity. The superheater tube banks were modeled as anisotropic
porous media for the flow calculations. Inertial resistance factors (loss coefficients per unit
length) were calculated for each of the superheater tube banks as a function of tube spacing. Heat
was extracted in the superheater tube banks by specifying an average steam temperature and
calculating local heat transfer coefficients. The furnace walls were modeled with a thermal
resistance boundary condition. An average waterwall fluid temperature was specified, as were
-------
the wall emissivity and the thermal resistance. The thermal resistance was set to account for the
tube metal resistance as well as the soot that may be present on the boiler walls.
The ABB CE NOx model was incorporated into the commercial code as a post processor using
user defined subroutines. The NOx model uses global reaction rates for thermal NO, prompt
NO, and fuel NO (4). The model also accounts for NOx destruction through the rebuming
mechanism (5). The dominant mechanism for NOx formation in gas combustion is thermal NO.
The thermal NO was calculated using a partial equilibrium assumption for the atomic 0 with a
rate that was calculated from studies performed at ABB using the CHEMKIN (6) code. The
turbulent mixing process adds fluctuations in both temperature and gas phase species
concentrations. As the temperature and species concentrations have a nonlinear contribution to
NO formation, significant errors may result if the fluctuations are ignored when predicting the
rate of NOx formation. A probability density function (PDF) method (7-8) was used to account
for the turbulent fluctuations using a beta function for the shape of the PDF.
The CFD simulations were run on an 8-processor workstation and typically required 2 days of
cpu time to converge each case. Once the solution was obtained to the combustion problem, the
NOx model was run as a post processor.
Numerical Model Calibration/Validation
The baseline field tests were used to calibrate the numerical model and verify that the model was
accurately simulating the observed combustion behavior in the furnace. Baseline tests were
performed in Unit 7 at 3 loads, nominally 650,450, and 170 MW. The data obtained for each
test condition included boiler emissions, windbox flow rates and gas species compositions, and
high velocity temperature probe measurements made near the horizontal furnace outlet plane.
The 650 MW case was used to calibrate the numerical model. The calibration included
determining the appropriate waterwall thermal resistance and setting the standard deviation of the
PDF in the NOx model to reasonably match the measured NOx values. ABB CE's Reheat Boiler
program (9), a 1-dimensional boiler simulation model, was used to calculate the expected
horizontal furnace outlet temperature (at the boiler nose) for each of the 3 baseline test cases.
The Reheat Boiler program performs a heat balance around the boiler using the measured water
and steam flows, temperatures, etc. where available, and using estimates of unmeasured input
parameters based on historical experience. The thermal resistance of the furnace walls was
adjusted in the numerical model for the 650 MW simulation in order to match the predicted
furnace outlet temperature to the value calculated from the Reheat Boiler program.
The calibration procedure was needed to account for both unknown wall boundary conditions
and approximations made in both the physical models in the code and the numerical
representation of the furnace (e.g., the lack of a good soot model and the simplification of the gas
burners and the tube banks which can not be resolved with a reasonable grid size). The 2 reduced
-------
load cases were then simulated using the same physical properties and wall resistance as the 650
MW case. The results for the reduced load cases were in good agreement with experimental
measurements, suggesting that the parameters used in the numerical model were adequate.
Baseline Test Input Data
The total natural gas and combustion air flow rates are listed in Table 1, along with the average
% FGR and SOFA for each of the 3 baseline cases. The % FOR is defined as the quantity of
recirculated flue gas divided by the total air and fuel flow. The % excess oxygen reported in the
table is on a dry basis. Table 2 presents the measured air flow rates for each comer of the
furnace. Note the significant imbalance in air flow distribution between the north and south sides
of the boiler. The SOFA also showed a north to south bias (front to rear). The gas species
compositions measured in each burner compartment indicated that the FGR was not uniformly
mixing with the combustion air before entering the windboxes. There was a bias of FGR to the
bottom and top elevations and to corners 1 and 2 (see Figure 1). The gas compositions measured
at each burner compartment were used as input to the numerical model.
Table 1. Baseline Test Data: Furnace Flow Rates
Load
(MW)
650
450
170
Natural Gas
[SCFHx E-3]
6307
4356
1850
Comb. Air
[SCFHx E-6]
70.3
52.0
28.2
%FGR
31
44
47
%SOFA
12
14
14
% Oxygen
1.9
3.5
7.3
Table 2. Baseline Test Data: Individual Windbox Flow Rates
Load
(MW)
650
450
170
Corner 1 - NE
[SCFHx E-6]
22.4
18.6
9.0
Corner 2 - NW
[SCFHx E-6]
24.8
21.7
10.2
Corner 3 - SE
[SCFHx E-6]
30.2
27.7
11.7
Corner 4 - SW
[SCFHx E-6]
25.8
26.8
11.8
Results of Baseline Simulations
Table 3 shows the gas temperatures predicted by the numerical model (CFD) and ABB's Reheat
Boiler Program. The 2 temperatures were within 50°F of each other for all loads indicating that
the numerical model was accurately predicting the heat extraction in the radiant section of the
-------
boiler. To check the models, high velocity temperature probe measurements were made through
ports at the 119' level, approximately 4.5' above the horizontal furnace outlet plane. Figures 2-3
compare the measured gas temperatures to the model predictions for each of the 3 baseline test
cases. Ports 1 and 2 are on the west (right) side of the furnace. Port 1 is between the platen and
pendant superheater sections and Port 2 is on the north side of the furnace between the
superheater division panels. As shown in the figures, the model shows good correlation with the
measured trends in outlet temperature as a function of load. Exact agreement between the model
and the measurements was not expected as the measurements were made at an elevation in the
furnace where heat is being extracted from the gas by the superheater sections. The numerical
model made some simplifying assumptions about the superheater sections as it was not feasible
or necessary to model them in detail. As most of the NOx generation occurs in the lower furnace,
the approximations made in the convective heat transfer regions of the boiler had little effect on
the gaseous emissions predicted by the model. Figures 2-3 do, however, demonstrate that the
numerical model is predicting reasonable temperature trends as a function of load.
Table 3. Baseline Model Results: Horizontal furnace outlet temperature predictions (°F)
Load (MW)
650
450
170
Reheat Boiler Program
2466
2174
1681
Numerical Model
2465
2220
1671
After the baseline combustion simulations were converged, the NOx model was run on each case.
As shown in Table 4, the NOx model was predicting from 4-8 ppm lower NOx on a dry basis
than was measured for the baseline cases. However, the trend in NOx emissions as a function of
load was accurately predicted by the model. Note that these were as measured values and were
not corrected to 3% oxygen. The numerical model predicted outlet CO concentrations that were
significantly higher than the measured values. The high CO concentrations predicted by the
model are due, in part, to the fact that the model ends at the vertical outlet plane of the furnace
where the gas temperature is on the order of 1700 °F for the 650 MW test. Significant CO
oxidation can still occur at this temperature. Inaccuracies in the CO oxidation rate and the
Magnussen mixing constants may also contribute to the high levels of CO as the predicted CO is
very sensitive to these parameters. However, it was felt that the model would adequately predict
the trends in CO emissions as a function of operating conditions. All of the CO predictions
reported in this document were normalized by the factor required to match the baseline 650 MW
prediction to the measured value. As shown in Table 4, the model still overpredicts the CO
emissions at the lower loads where it was measured to be 1 ppm.
An examination of the CO emissions in the furnace simulations shows high CO levels coming
from the SW corner of the furnace due to low oxygen levels in that comer. This results in high
temperatures and CO levels along the back wall of the furnace. Poor mixing of the SOFA also
occurs in that corner as the SOFA nozzles are on the side walls of the furnace and not in the
corners. High levels of CO are predicted to occur at both side walls of the furnace as the gas
-------
enters the convective pass, consistent with field observations. As will be shown in the following
section, a more uniform mixing of the FGR will reduce the regions of high stoichiometry in the
windbox and subsequently reduce the NOx emissions.
Table 4. Baseline Test Results: NOx and CO predicted vs. measured
Load (MW)
650
450
170
NOx -dry PPM
Measured Predicted
65 61
28 23
22 14
CO - dry PPM
Measured Predicted*
56 56
1 18
1 15
Normalized to 650 MW measured value.
Effect of Operating Conditions on Predicted NOx Emissions
A full load case (700 MW) was generated with the numerical model by scaling up the 650 MW
baseline test case. This full load case was used as the basis for performing an optimization study
to determine which variables had the largest effect on NOx emissions in an attempt to optimize
the boiler for NOx emissions without adversely affecting the combustion efficiency or causing
overheating of superheater tubes. This section reports the model results for several variables that
were studied including: air flow distribution, quantity of FGR, FGR biasing, and quantity of
SOFA.
Combustion Air Flow Distribution
Under standard operation, Unit 7 has significant flow imbalances in both the main windbox and
the SOFA. Poor mixing of the FGR also causes large variations in the oxygen concentration from
burner to burner. Three simulations were performed to understand the relative importance of
balancing the windbox flow rates vs. improving the FGR mixing. One case used the composition
distribution as measured in 650 MW baseline test, but with uniform mass flow rates through each
windbox (Uniform Flow). The second case had uniform gas species compositions (30% FGR),
but the mass flow rates as measured in 650 MW baseline test (Uniform Composition). The third
case had both uniform flow rates and compositions.
Table 5 lists the predicted NOx and CO emissions for the three cases described above in addition
to the full load case. Note that predicted NOx emissions are essentially half the baseline values
when a uniform FGR mixing is assumed. By reducing the high oxygen concentration (20 %) that
some of the burners were experiencing under the baseline conditions, the NOx emissions were
dramatically reduced. The oxygen concentration of the uniformly mixed cases was approximately
17.5%, corresponding to 30% FGR. Note that there was little change in the predicted NOx
-------
concentration as a function of the mass splits between the comers. Correcting the 15-20%
imbalance in the mass flow rates of combustion air/FGR to the windboxes would have less effect
on NOx emissions than improving the mixing of the FOR and the combustion air. The local
oxygen concentration at the burner seemed to determine NOx emissions. The CO emissions, on
the other hand, were impacted by both the composition and the mass flow distribution in the
furnace. Optimum CO emissions were obtained with both uniform composition and flow as the
very fuel rich regions were eliminated.
Table 5. Numerical model predictions for the air flow distribution cases
Full Load
Uniform Flow
Uniform Composition
Uniform Flow and
Composition
NOx
[ppm dry]
69
83
34
33
CO
[ppm dry]
72
90
106
59
HFOT
[°F]
2501
2490
2488
2481
Platen SH
[°F]
2121
2107
2103
2094
Pendant SH
[°F]
1803
1794
1788
1781
FGR Quantity
Table 6 presents the model results when the quantity of FGR was varied from 15-45 %. All of
the tests assumed that both the mass flow rates and the gas species compositions were uniform
in the windbox. The 30% FGR case is the Uniform Flow and Composition case listed in the
previous table. Decreasing the FGR to 15% causes a dramatic increase hi NOx emissions due to
the increased local oxygen concentration and gas temperatures. The NOx emissions drop as the
quantity of FGR is increased, while the CO emissions increase as shown in Table 6.
Table 6. Numerical model predictions for the % FGR cases
15%
30%
35%
45%
FGR
FGR
FGR
FGR
NOx
[ppm dry]
115
33
24
21
CO
[ppm dry]
43
59
58
88
HFOT
[°F]
2546
2481
2460
2404
Platen SH
[°F]
2110
2094
2076
2062
Pendant SH
f°F]
1788
1781
1781
1772
The furnace outlet and superheater temperatures also decrease with increasing amounts of FGR.
This decrease in temperature is due to the increased mass of recirculated gas that must be heated
-------
and does not imply that there will be less problems associated with overheating in the
superheater sections. In fact, the predicted quantity of heat absorbed by the waterwalls below
the furnace nose decreases by 19% as the FGR is increased from 30 to 45%. Increasing FGR
lowers the gas temperature and reduces the radiative heat transfer to the waterwalls. Decreasing
the FOR to 15% causes an increase in lower furnace absorption of 16%. Note that these values
may not accurately reflect the quantitative changes that will occur in the boiler due to inaccuracies
in the radiation model, especially due to the presence of soot. However, the trend of decreased
lower furnace absorption with increasing FOR is real.
FGR Biasing
The numerical model was also applied to a series of test conditions where the quantity of FGR
was varied as a function of elevation. The gas velocity of each burner compartment was assumed
to be constant while the species composition was varied as a function of elevation. Table 7 lists
the test conditions that were investigated, along with the model predictions. As illustrated in the
table, biasing the FGR in the numerical model was not beneficial from an emissions standpoint.
As the temperature in the boiler is determined by the gas phase stoichiometry, the temperature
profile in the furnace can be manipulated by biasing the FGR. Moving more of the FGR to the
top burner elevations helps to lower the peak gas temperatures in the furnace which are typically
near the top of the windbox. The gas temperatures at the bottom elevations are increased,
resulting hi a more uniform temperature profile in the furnace as the bottom elevations are
typically cooler due to the increased waterwall surface of the hopper, etc.
Table 7. Numerical model predictions for the FGR biasing cases
FGR Biasing Scheme
Large bias to top
Large bias to bottom
Small bias to top
Uniform distribution
NOx
[ppm dry]
71
44
37
33
CO
[ppm dry]
185
250
76
59
HFOT
[°F]
2467
2521
2478
2481
Platen SH
[°F]
2083
2128
2089
2094
Pendant SH
[°F]
1779
1806
1781
1781
Even though the peak gas temperatures in the boiler were decreased with FGR biasing, the NOx
emissions were not improved. The predicted NOx emissions were more sensitive to the burner
stoichiometry than to the temperature. Under uniform flow and composition conditions the
average gas phase stoichiometry in the lower windbox is just slightly fuel rich. Decreasing the
FGR to 25% increased the local burner stoichiometry sufficiently to increase NOx due to an
increase in local gas stoichiometry near the burner. The effect was more dramatic with increased
FGR biasing as illustrated in Table 7. Even though the temperature profile was more uniform
with 15% FGR in the bottom elevations, the concentration of NOx increased as there was a fuel
-------
lean core in the center of the furnace with a longer residence time for NOx formation. CO
emissions also increased with increased biasing of the FOR.
SOFA Quantity
The predicted NOx emissions decrease slightly with increasing SOFA as illustrated in Table 8.
The additional SOFA was uniformly removed from the windbox, maintaining the same quantity
of FOR (30%). The model predicts that the CO emissions increase significantly as the quantity
of overfire air is increased. The increase hi CO is a result of poor mixing of the overfire air. Note
that increasing the quantity of SOFA would require significant modifications to the overfire air
system, resulting in a significant capital cost.
Table 8. Numerical model predictions for the % SOFA cases, assuming uniform inputs
%SOFA
12%
15 %
20%
NOx
[ppm
dry]
33
30
26
CO
[ppm
dry]
59
114
404
HFOT
2481
2478
2501
Platen SH
2094
2096
2098
Pendant SH
1781
1785
1790
In summary, the CFD study suggested that the most significant reduction in NOx emissions
could be obtained by simply obtaining a uniform distribution of FGR within the windbox. If a
uniform distribution of FGR could not be obtained, the model suggested that it was beneficial to
have the higher quantities of FGR in the bottom burners. The model also showed that the
potential reductions in NOx emissions did not justify the additional cost required to balance the
mass flow rates to each corner of the furnace and to increase the quantity of SOFA. The
improved FGR distribution should also decrease CO emissions and have little effect on the
superheater tube metal temperatures.
Results of Flow Management Modifications on Unit 7
Implementation of the FGR mixing system modifications that were developed through
investigation of the physical cold flow model resulted in substantial improvement in the unit's
windbox and burner FGR flow distributions. These improvements contributed to more balanced
combustion and lower NOx emissions on the unit. Figure 4 presents windbox FGR distribution
data taken after flow management modifications were installed. A comparison of Figure 4
distributions with those presented in Figure 1 clearly demonstrates a significant improvement in
FGR concentration gradient.
-------
This enhanced balance of FOR and combustion air to each of the unit's twenty burners resulted
in a significant reduction in full load NOx emissions. Figure 5 presents Unit 7 NOx emissions as
a function of FOR rate for pre- and post-modifications operation. Note that some of the scatter
in the data is a result of variations in the percent of excess air. However, the data show that for
relatively low FOR rates, pre- and post-modification emissions are comparable; but as FOR rate
increases, NOx emissions are observed to be much lower after the draft system modifications
were installed compared to baseline operation. These results suggest that as the unit's flue gas
recirculation rate increased prior to these draft system modifications, the combustion air/FGR
imbalance became worse. This effect substantially diminished the NOx reduction benefit that
may now be derived from operating at higher FOR rates. (Please note that following draft
system modifications, it was not possible to test Unit 7 at lower loads due to utility system load
requirements.) The reductions in NOx emissions with the improved FGR distribution and the
quantity of overfire air are consistent with the CFD predictions.
Conclusions
Reductions in NOx emissions of almost 50% were achieved in PG&E's Pittsburg Unit 7 as a
result of an optimization project that combined field testing, CFD modeling, and physical cold
flow modeling. Field testing revealed a maldistribution of FGR in the combustion air to the
windbox as a function of both comer and elevation and flow rate imbalances of up to 20% in both
the combustion air and the SOFA. CFD models of the unit accurately simulated 3 baseline test
cases, performed at 650, 450, and 170 MW. The CFD model was then used to quantify the
potential reduction in NOx emissions that could be expected if the FGR and the combustion air
were uniformly mixed (approximately 50%). The CFD model was also used to investigate the
effects of biasing the FGR as a function of elevation, balancing the mass flow rates to each
windbox and SOFA compartment, and increasing the quantity of SOFA and FGR. The NOx
model was used to identify the primary boiler input parameters which have large effects on NOx
reduction. Predictions showed that increasing the quantity of FGR could decrease NOx
emissions, but that both CO emissions and superheater tube metal temperatures may increase.
The potential decrease in NOx emissions obtained by increasing the quantity of SOFA did not
justify the capital cost of modifying the SOFA system.
The physical flow model was used to develop draft system flow management modifications for
the full scale unit to improve the mixing of the FGR and the combustion air. The modifications
led to more uniform FGR distributions in the windboxes, and ultimately reductions in NOx
emissions of approximately 50%. This project demonstrates the potential for significant
reductions in NOx emissions in utility boilers by simply tuning the unit. Field testing,
computational fluid dynamics, and physical cold flow modeling can be used together as a means
to understand and improve the performance of a utility boiler.
-------
References
1. Patankar, S.V., Numerical Heat Transfer and Fluid Flow, Hemisphere Publishing Corp.,
Washington D.C., 1980.
2. Launder, B.E. and Spalding, D.B., Lectures in Mathematical Models ofTurbulence, Academic
Press, London, England, 1972.
3. Magnussen, B.F. and Hjertager, B.W. "On Mathematical Models of Turbulent Combustion
with Special Emphasis on Soot Formation and Combustion," 16th Symposium (Int.) on
Combustion, The Combustion Institute, Pirtsburg, PA, 1976.
4. Miller, J.A. and C. Bowman, "Mechanism and Modeling of Nitrogen Chemistry in
Combustion," Prog. Energy Combust. Sci. Vol 15, p. 287 (1989).
5. Chen, W., L.D. Smoot, L.D., S.C. Hill, T.H. Fletcher, "A Global Rate Expression for Nitric
Oxide Rebuming. Part 2," Energy & Fuels Vol. 10, p. 1046, (1996).
6. Kee, R.J., J.F. Grcar, M.D. Smooke, J.A. Miller, "A FORTRAN Program for Modeling
Steady Laminar One-Dimensional Premixed Flames", SANDIA REPORT, SAND 85-8240.UC-
401, Reprinted September, 1993.
7. Williams, FA., Turbulent Mixing in Nonreactive and Reactive Flows, Plenum Press, New
York, NY, 1975.
8. Smoot, L.D. and P.J. Smith, Coal Combustion and Gasification, Plenum Press, New York,
NY, 1985.
9. Miemiec, L.S., G.F. Lexa, and A.A. Levasseur, "Effects of Coal Cleaning on Boiler
Performance," ASME paper 87-JPGC-FACT-7, Presented at Power Gen. 1987.
-------
Corner 2 Wlndbox FGR Dist
o 10 AI 30 * si eo
Comer 1 Wlndbox FGR Dist
FOR run, %
Corner 4 Wlndbox FGR Dist
FGR Ran, S
SO 00
Comer 3 Windbox FGR Dist
FGR Ran. %
10 31 3D «
Figure 1. Measured FGR distribution for 650 MW baseline test.
£ 1 800
| 65O MW - Predicted;
I • 65O MW - Measured
i — 4SO MW - Predicted
I A *SO ww - M«»«u
, - - 170 MW - Predicted
• 170 MW - Measured
Figure 2. Comparison of predicted and measured gas temperatures (°F) for Port 1.
-------
• 65O MW - Measured
—— 450 MW - Predicted
4 45O MW • Measured
—— 17O MW - Predicted
• 17Q MW - Measured
Dletance from Front Furnace Wall (ft)
Figure 3. Comparison of predicted and measured gas temperatures (°F) for Port 2.
Comar 2 FGR Distribution
Comer 1 FGR Distribution
Corner 4 FGR Distribution
Corner 3 FGR Distribution
Figure 4. Measured FGR distribution for 650 MW post-modification test.
-------
<5
= 40
,-t-
•
FGR RATE ("/,)
Rgure 3 - Pittsburg Unit 7 Pre- and Post-Retrofit NO, Emissions as a Function of FGR Rate
Figure 5. Measured NOx emissions as a function of FGR quantity.
-------
Tuesday, August 26; 1:00 p.m.
Parallel Session A:
Cyclones - Combustion NOx Controls
-------
COMPUTER MODELING OF CYCLONE BARRELS
B. Adams, M. Heap, P. Smith
Reaction Engineering International
A. Facchiano
Electric Power Research Institute
C. Melland
Basin Electric
K. Stuckmeyer
Union Electric
S. Vierstra
American Electric Power
Abstract
Computer simulations of cyclone barrels and an opposed wall cyclone-fired furnace have been
used to further the understanding of cyclone barrel combustion and to evaluate the NOx
reduction potential of modifications to cyclone barrel operation. This paper describes the cyclone
barrel model and the simulations that were carried out to determine cost effective nitric oxide
control options both for the barrel and the furnace. The simulations indicate that the conditions
within the cyclone barrel are far from well mixed. This observation is supported by gas
temperature and species concentration measurements in two cyclone barrel types. Comparison of
model predictions with two data sets of temperature and gas species concentrations indicated
very good qualitative agreement and acceptable quantitative agreement between the model
predictions and measured properties. NOx formed in the furnace section of opposed wall
cyclone-fired boilers can be significant and is dependent on the thermal environment in the lower
furnace. The simulations indicated the potential reductions that could be achieved through the
application of several techniques including: fuel switching, flue gas recirculation, water injection,
barrel biasing and barrel staging.
Introduction
There are 105 operating, cyclone-equipped utility boilers burning a range of fuels including high
sulfur bituminous coals, lignites and blends. These boilers represent approximately 14 percent of
pre-New Source Performance Standards (NSPS) coal-fired generating capacity and they
contribute approximately 21 percent of the total NOx emitted. Although the majority of cyclone
units are 20-30 years old, utilities plan to operate many of these units for at least an additional
10-20 years. Low NOx firing systems typical of pulverized coal fired boilers may not be
applicable to cyclone-fired boilers because of their unique design. Selective catalytic reduction,
-------
selective non-catalytic reduction and rebuming technologies can reduce NOx emissions from
cyclone-fired boilers, but at high capital and/or operating costs.
The EPRI Cyclone NOx Control Interest Group (CNCIG) decided that a numerical simulation
tool could be used to assess the potential NOx reductions resulting from modifications to
cyclone barrel operation. The approach selected was to extend the capabilities of an existing
reacting computational fluid dynamics code, GLACIER, to include the features that are necessary
to describe the physical and chemical phenomena controlling combustion in a cyclone barrel
typical of those used by CNCIG members. Originally the program envisioned three components:
. Construct a model to simulate a single cyclone barrel at the Sioux Station of the Union
Electric Company.
. Evaluate the model for a baseline case and determine the sensitivity of the results to both
model and input parameters.
• Use the model to explore cost effective options to control NOx formation within the barrel.
During the study, the importance of NO formation within the furnace was recognized and a
fourth component was added:
Integrate the barrel model with a furnace model to assess NOx control options for those
cyclone-fired boilers which produce significant levels of NO in the furnace.
While the cyclone barrel model was being developed data were obtained' on the conditions
within cyclone barrels. Gas species and temperature data probes were inserted through the site
port of several cyclone barrels at the Sioux Plant" These data indicate that conditions within the
barrel are not well mixed and there are very fuel rich regions with very high carbon monoxide
concentrations. Further validation of the model was obtained by comparing predictions with
measurements for a lignite fired cyclone barrel at the Leland Olds Station of Basin Electric. This
paper compares these data with the results obtained from the numerical simulations. Also,
several modifications to barrel operation were simulated and the impact on NO formation in the
barrel as well as other aspects of barrel operation such as slag freezing and corrosion potential are
addressed. The paper ends by presenting results from simulations of the lower furnace using a
combined barrel-furnace model of the Sioux Plant.
Model Background
GLACIER is based on software developed over the last fifteen years by Smith and co-workers3"6
Particular emphasis has been placed on simulating coal combustion and pollutant formation.
GLACIER includes full mass, momentum and energy coupling between the gas and particles as
well as full coupling between turbulent fluid flow, chemical reactions, radiative and convective
heat transfer, and finite-rate NO formation.
When predicting the mean concentrations of NO in coal-fired combustors, the effects of both
turbulence and finite-rate chemistry are important. GLACIER can reliably compute the relative
-------
effects of fuel, thermal, and prompt NO as a function of local stoichiometry and temperature. A
reburning mechanism is also included through reactions involving NO and fuel species.
Prior to this study GLACIER had been used extensively for the prediction of pulverized coal
flames. Several modifications were made to enable accurate simulation of cyclone combustion.
These were primarily associated with improved particle transport and deposition models that
enabled particle flow in the burner and their subsequent deposition and reaction in the slag layer
to be simulated more accurately. In addition four basic assumptions were made:
1. When a particle was deposited in the slag layer its burning time was small compared to the
slag flowing velocity and all the coal off-gases were released at that location.
2. The refractory coated walls were adiabatic until they reached a temperature equal to the
calculated T8o for the coal being fired.
3. Coal particles did not stick to the wall but experienced an elastic collision if the wall
temperature or the particle temperature was less than the calculated Tiso-
4. Even the largest particles were isothermal, this means that no volatiles evolved before the
particle was completely dry.
Computational Results
The modified GLACIER computer model was used to simulate a single barrel among the
cyclones in service at the Union Electric Sioux Plant. The ten foot diameter barrel is fitted with a
radial burner and fires a 70/30 blend of Powder River Basin/Illinois No. 6 coal. The model used
twenty four particle starting locations with eight discrete particle sizes for each coal type. Table 1
summarizes the input conditions for the Sioux cyclone.
The flow field in the cyclone barrel is dominated by the secondary air flow which enters
tangentially as do the coal, primary and tertiary air streams in the radial burner. All three streams
rotate in the same direction but the primary air stream is directed at the point at which the vertical
coal chute enters the burner. The coal is delivered with a very small amount of air and there is a
small air flow along the axis of the burner from the tertiary air duct. There are two recirculation
zones within the barrel, one at the exit of the burner and the other along the wall of the cyclone
close to the burner end wall. The general flow is tangential until the gases accelerate axially as
they exit the divergent throat. Since all the secondary air enters through a single duct it is not
surprising that the flow field is not symmetric. Figure 1 shows the computed radial temperature
distribution at three radial planes corresponding to locations: a) close to the burner; b) in the
region where the secondary air enters the barrel, and c) close to the divergent throat. The gas
temperature field illustrates the highly stratified flow that exist throughout the barrel, even as the
gases exit the divergent throat they are not completely mixed. The secondary air flow rolls
around the wall creating a scroll effect as it entrains the coal gases from the wall. The high
temperature zones occur on the boundary of the air stream as it mixes and reacts with the rich
coal gases.
-------
Table 1
Summary of Input Conditions for the Sioux Cyclone Barrel
COAL
Composition (wt%)
Moisture
Ash
C,H,O,N,S
Feed Rate
Firing Rate
T250 Slagging Temp
Particle Size
70/30 PRB/lllinois #6 Blend
24.6%
6.6%
68.8%
22.5 tons/hr
413MMBTU/hr
2388 F
16-3000 |im
AIR
Barrel Stoichiometry
Total Feed Rate
Air Temperature
Pri/tert/sec flow split
S.R.=1.16 (3%O2)
178.3 tons/hr
615 F
12/3/85%
Figure 1 also shows that there is ignition close to the burner exit. Apparently, this occurs due to
two separate phenomena. The coal feed contains some fines. In this simulation the coal field was
characterized by discrete "bins" for both coal types and it should be remembered that the Powder
River Basin Coal has a much higher moisture content than the Illinois coal. Only the smallest
particles dry and devolatilize close to the burner, but these small particles are essential for the
creation of heat release close to the burner. The second phenomenon is associated with the
deposition of coal particles on the divergent burner end wall of the cyclone. Some coal particles
impinge on the end wall which is hot enough to have running slag, therefore in this model they
stick and can react. Sensitivity studies were carried to study these phenomena. Either removing
the small particles or preventing deposition on the end wall by changing the burner flow field
resulted in the heat release being pushed to the back of the cyclone.
Examination of the computed results for the baseline case indicates the following:
• Only the smallest coal particles burn in suspension because the large particles do not dry
before they are deposited on the walls. This a consequence of the isothermal particle
assumption and probably underestimates the volatile release before the particles are
deposited.
• Because the flow within the cyclone is segregated there are both fuel rich and fuel lean
regions throughout the barrel. There are extremely rich regions close to walls, i.e.,
equivalence ratios in excess of 4. The location of these regions are determined by the particle
deposition pattern.
-------
. The paths taken by the particles in the cyclone depend on the particle size and their starting
location in the coal chute. Coal particles that are not diverted radially as they exit the burner
tend to get caught up in the axial flow and corkscrew through the divergent exit.
. Nitric oxide appears to be formed in four regions within the barrel:
— as the coal fines burn close to the burner exit, NO is formed primarily by the fuel NO
mechanism;
— in the region close to the wall NO is formed from the fuel nitrogen that evolves with the
coal gases even though some of locations are extremely fuel rich;
— as the fuel rich products from the coal contact the combustion air stream gas temperatures
increase and thermal NO is formed as the gases exit the barrel;
— there is a region around the reentrant throat where both thermal and fuel NO are formed.
The same model was used to simulate a cyclone barrel burning lignite at the Leland Olds Station
(LOS). Lignite cyclones are nearly identical to the cyclones burning bituminous and sub-
bituminous coals, but there are differences. The burner of the lignite cyclone is a scroll burner
whereas the bituminous and sub-bituminous cyclones use a radial or vortex burner. The
difference in burner design comes from the necessity to use two stages of drying for the lignite
coals before it enters the cyclone burner. The two-stage drying configuration normally removes
approximately 10% of the moisture. One of the lignite-fired cyclone barrels at the LOS was
simulated. The overall firing rate was 330 MMBtu/hr with 10% excess air. The air flow split
between the primary, secondary and tertiary streams was 16.6/81.9/1.5 percent. Measurements
indicated that the coal composition varied by size fraction and the particle size distribution varied
with location in the coal delivery pipe. These variations were included in the model inputs.
Figure 2 presents a contour plot of the predicted gas temperatures in the LOS cyclone on a
vertical plane through the barrel axis. Peak temperatures within the lignite-fired cyclone are
below those predicted for the barrel fired with the coal blend but the predictions again indicate
that conditions within the barrel are not well mixed.
Comparison with Data
Sioux Cyclone Barrel
Figures 3,4 and 5 compare the measured data with the computed gas temperature, oxygen, CO
and NO concentration. The probe path angled down from the top of the burner to the bottom of
the divergent exit and the data are shown as a function of probe insertion distance along this path.
Data were obtained from several barrels operating at the same nominal conditions over a period
of several days. Figure 3 compares measured temperature and oxygen concentration with the
computed values. Two data sets are shown for the measured oxygen concentration corresponding
-------
to measurements taken on different days but in the same barrel. Figure 4 and 5 compare CO and
NO, respectively. Considering that the model indicates very steep gradients in both temperature
and species concentration throughout the barrel the agreement between the measured data and
computed results is very good. The qualitative trends are in very good agreement and the
quantitative agreement is acceptable. Several observations concerning the data comparison can
be made:
1. The two data sets for oxygen probably reflect a slight change in operating conditions that
affected the location of the ignition zone near the burner. In the computer simulation it was
relatively easy to move the heat release to the back of the cyclone but the presence of coal
fines and/or deposition of coal particles on the front wall influenced early ignition.
2. The measured CO concentrations are higher than those predicted by the model but the
measurements support the concept that there are very rich regions within the barrel creating
naturally staged conditions.
3. Two measured data sets are presented in Figure 5 to show that the measured dip in NO that
occurred at the eleven foot insertion point was not an anomaly. However, even when the
reburning mechanism was included the model did not duplicate this dip in NO concentration.
Very little effort was expended to improve the agreement between the data and the predictions
because it was felt that considering the complexity of the cyclone combustion process, and the
difficulty of ensuring that the cyclones operate at a fixed and known conditions that the level of
agreement was more than sufficient to validate the model.
Leland Olds Cyclone Barrel
Figures 6 and 7 compare data taken through cyclone measurement ports 1 and 2, respectively. As
with the Sioux barrel, both the data and the predictions indicate that conditions within the barrel
are far from well mixed. The temperature contours in the vertical axial plane give an indication
of how unmixed the gases are even as they pass through the reentrant throat. This is illustrated
also by the gas species concentrations. It should be remembered that this discussion is limited to
a comparison of measurement and predictions in one vertical plane. The flow in the barrel is not
symmetric and there are large gradients in temperature and species concentration in the radial
planes. Considering the conditions within the barrel and the accuracy of probe positioning which
the qualitative agreement between data and predictions is very good.
Port 1 allowed measurements to be made along the barrel axis and data are available for two
cyclones. With the exception of CO concentration the measured data follows the same general
trends. The oxygen concentration decreases from the burner and both the temperature and NO
concentration increase with distance from the burner. However in cyclone #9 the oxygen
concentration decreases to zero while the corresponding measured CO concentrations are very
high. Cyclone #8 did not exhibit the same behavior. Oxygen concentrations were higher and the
CO level was uniformly low along the axis. It should be noted that the cyclone with the higher
measured gas temperature and oxygen concentration, cyclone #8, does not give higher NO
values.
-------
Sensitivity studies carried out with the cyclone model suggested one possible reason for the
variation between cyclones and also between measurements and predictions. The simulations
indicate that there is a recirculation zone at the burner whose size and strength is dependent on
the tertiary air flow. Figure 8 shows predicted flow fields in the region near the burner exit for
low tertiary air flow. Increasing the tertiary air flow reduces the size and strength of the
recirculation which in turn reduces the amount of hot gases that flow back towards the burner.
This results in delayed ignition and higher oxygen concentrations on the burner axis near to the
burner. The coal distribution is controlled by the primary air flow and the variation in tertiary air
has almost no impact on the majority of the coal particle dispersion. Only the smallest particles,
which do ignite close to the burner are affected. Considering that the tertiary air flow accounts for
only 1.5% of the total air flow it is reasonable to expect that there will be cyclone to cyclone
variations.
Both the predicted temperature and the NO concentrations are sensitive to the coal moisture
content and this could account for the differences in measurements and predictions shown in
Figures 6 and 7. It should be remembered that these simulations are restricted to the barrel. NO
can be formed in the lower furnace as the combustion products exit the cyclone. Furnace NO will
be less significant with lignite than with bituminous coals because of the lower adiabatic
temperature. The predicted NO formed in the barrel is 0.89 IbsNOx / MMBtu.
NOX Control Options for Cyclone-Fired Boilers
The cyclone barrel model was used to evaluate the impact of relatively simple operational and
design changes on NO formation within the barrel. In most cases the changes that were evaluated
were considered to be extreme limit cases in order to quickly eliminate options that were not
promising.
Operational Changes
Three operational changes were considered from the baseline operation: a) change the coal grind
to a finer distribution typical of an advanced crusher design, b) operate at high excess air level,
and c) switch from the blended coal to all Illinois coal. The final option was included not because
the NO formation was expected to reduce but to give an idea of the models ability to predict the
effect of fuel switching. Figure 9 compares the NO emitted from the barrel for these cases. None
of these changes had a significant effect on the amount on NO formed in the barrel.
Design Changes
The design variations investigated were:
• Two variations in the air split between the primary/tertiary/air streams flow. The low primary
air case (LP) had a 10/2.5/87.5% split and the high primary air case (HP) had a 14/3.5/82.5%
split. The baseline flow split was 12/3/85%.
-------
. The secondary duct was divided into three equal sections and air was transferred from the
section nearest the burner to the section nearest the exit, in the baseline case the flow was
uniform throughout the secondary duct. Two biased secondary air flow cases were run:
a) -15%/normal/+15% (SpllS) and b) -75%/normal/+75% (Spl75).
. A case (VSpl) was run with an imaginary flow diverter placed along the length of the
secondary duct to divert one third of the flow towards the cyclone axis.
• The geometry of the coal chute was changed in one case (Chute), the length was doubled and
width halved.
The computed results for these design changes are presented in Figure 10. None of these design
changes had a major influence on the NO formed in the cyclone. The flow diverter design did
reduce NO but resulted in significantly poorer cyclone performance in terms of slag tapping and
particle carryover.
Staging, FGR and Water Injection
Figure 11 shows the influence of barrel stoichiometric ratio on nitric oxide production in the
barrel. NO emissions decrease with decreasing stoichiometric ratio. Figure 12 shows the effect of
adding flue gas recirculation (FGR) to the secondary air. NO decreases with increasing FGR.
Also shown in Figure 12 is one prediction for liquid water injection along the barrel axis. The
amount of water added was thermally equivalent to 5% FGR and the predicted emissions are
within 5 ppm. Examination of the computations indicated that 20% FGR almost eliminated
thermal NO formation. The model predicts that greater than fifty percent of the NO that is
produced in the barrel is produced by the thermal mechanism accounting for the effectiveness of
FGR.
Operational Impacts
Figure 13 shows the influence of staging and FGR on several factors that were used to monitor
cyclone performance other than NO. The percentage of wall area with a temperature less than the
calculated T2so was used to give an indication of problems associated with slag freezing.
Although 20% FGR gave a significant reduction in NO, the wall surface area that was frozen
doubled over the baseline condition. Corrosion is also a potential problem when cyclones are
staged. Two parameters that are associated with production of iron sulfide were used to assess
corrosion potential: (1) the wall surface area covered by gas with a stoichiometric ratio less than
0.6 and (2) the wall surface area with a temperature greater that 2550 °F. If these indicators are
acceptable then neither staging nor FGR appears to increase the potential for corrosion.
Furnace NO
X
The discussion so far has dealt exclusively with the NO that is formed within the cyclone barrel.
In all the cases considered to date the heat loss through the barrel wall was between 1.5 and 5%
of the heat input. Also, the exhaust gases contain significant levels of CO that will react as the
-------
combustion gases mix in the main furnace. Since the heat loss in the barrel is small thermal NO
formation will continue in the furnace and the amount produced will depend on the thermal
conditions experienced by these gases in the furnace. To evaluate the potential importance of
furnace NO a model of the Sioux furnace was constructed that coupled the barrel outlet
conditions to the ten barrel opposed wall-fired furnace. The unmixedness at the barrel exit was
interpolated onto the less refined computational grid required to model the complete furnace
which contained more than 502,000 computational cells.
The Sioux furnace operates normally with FGR addition through gas recirculation ducts and gas
tempering ports. Figure 14 compares predicted furnace emissions for four NOx control options -
FGR, water injection, staging and biasing - with the baseline case. In the staged case all the
barrels were operated at a stoichiometric ratio of 0.9 and the remaining combustion air added
through the gas recirculation ports. In the two biased cases the upper cyclones were operated at
SR = 1.4 and the lower cyclones at SR = 0.9 and 0.95, respectively. Consequently for the 0.9/1.4
case there was also a reduction in overall excess air. It can be seen that staging could be very
effective in reducing NO but any of the other techniques could also be used to obtain modest
reductions in NO emissions.
Conclusion
A model of NOx formation in the barrel of a coal-fired cyclone has been developed and used to
explore control options that involve simple variations in barrel operating conditions. Comparison
of model predictions with a data set of measured temperature and gas species concentrations
indicated very good qualitative agreement and acceptable quantitative agreement. Model results
show the cyclone to be poorly mixed and very stratified with large regions which are either very
fuel or air rich.
Because a cyclone barrel operates with very little heat loss, thermal NO is a significant fraction
of the NO formed in the barrel and, since it is necessary to combust almost all of the carbon in
the barrel, exit gas temperature is almost unaffected by the simple design and operational
changes considered to date. Therefore, thermal NO formation which occurs primarily as the
combustion products exit the barrel also is not affected. Injection of recirculated flue gas reduces
the temperature and can have a major impact on NO emissions from the barrel but the reduced
temperatures could have a detrimental affect on cyclone operability because the slag may freeze
or become difficult to tap due to significant increases in viscosity. Staging produces less
significant NO reduction due to the existence of very fuel rich regions in the cyclone barrel
which continue to produce NO regardless of the degree of staging. Staging and FGR both show a
potential for reducing overall NO levels in the boiler due to lower cyclone exit temperatures.
Based on the current criteria, use of staging and FGR do not seem to increase the corrosion
potential within the cyclone.
Results of coupling the cyclone exit gases with a furnace model indicated NOx formed in the
furnace of opposed wall-fired cyclone boilers can be significant. Staging and biasing appear to be
viable NO emission control options but the effects of staging on the water wall corrosion need to
be evaluated. Water addition could be a low cost alternative if minimal reductions are required.
-------
References
1 Private communication from Carnot. December 1995.
2. Stuckmeyer. K. Adams. B.. Heap. M. and Smith. P.. "Computer Modeling of a Cyclone
Barrel." EPRI NOx Controls for Utility Boilers Conference. Aug. 6-8. 1996. Cincinnati. OR
3. Smoot. L.D. and P.J. Smith. Coal Combustion and Gasification. Plenum Press. New York.
NY. 1985.
4. Smith. P.J. and T.H. Fletcher, "A Study of Two Chemical Reaction Models in Turbulent
Coal Combustion." Combust. Sci. Technol.. Vol. 58. p. 59. 1988.
5. Baxter. L.L. and P.J. Smith. "Turbulent Dispersion of Panicles: The STP Model." Energy
and Fuels. Vol. 7. pp. 852-859. 1993.
6. Adams, B.R. and P.J. Smith. "Three-dimensional Discrete-ordmates Modeling of Radiative
Transfer in a Geometrically Complex Furnace," Combust. Sci. Technol.. Vol 88. p. 293,
1993.
Figure 1. Radial Temperature Distribution at Three Axial Locations Illustrating the Stratified
Conditions in the Barrel: a) near burner outlet, b) secondary air inlet, c) near cyclone outlet.
-------
Figure 2. Predicted Gas Temperature Contours in Leland Olds Station (LOS) Cyclone.
e e 10
Distance into Cyclone (ft)
-B- Field 02 - 127/95
-«- Field 02- 12/^95
-*- Predicted 02
E 15-
cp
§
§10-
o
s-
6 8 10
Distance into Cyclone (ft)
Reid CO Data
Predicted CO
he I
Figure 3. Comparison of O2 and
Temperature Data for Sioux Cyclone.
Figure 4. Comparison of O2 and
CO Data for Sioux Cyclone.
-------
1000
Distance into Cyclone (ft)
Figure 5. Comparison of NO Data for Sioux Cyclone.
900- -
800-
700-
600-
500-
400-
300-
200-
100-
0--
•- Predicted 02
«- Field 02 Data Cyc«9
^— Field O2 Data CycttS
3500-
Distance into Cyclone (ft)
Predicted NO
Reid NO Data CycJr9
Field NO Data CycKS
6 8 10 12
Distance into Cyclone (ft)
3000-
£ 2500-
1 2000-
5 :
I 1500-
(| 1000-
500-
0-
80000-
70000-
60000-
_ 50000-
I
— 40000 -
O
0 30000-
20000-
10000-
0
•- Predicted Temp
e- Field Temp Daia Cyc#9
rt- Field Temp Data Cyc#8
6 8 10 12
Distance into Cyclone (ft)
14
Predicted CO
Field CO Data Cyc#9
•6— Field CO Data Cyc#8
6 8 10 12
Dislance into Cyclone (tt)
Figure 6. Port 1 Data Comparisons for Leland Olds Cyclone.
-------
10 12 14 16
Distance into Cyclone (tt)
6 8 10 12 14
Distance into Cyclone (ft)
6 8 10 12 14 15
Distance into Cyclone (ft)
6 8 10 12
Distance into Cyclone (ft)
Figure 7. Port 2 Data Comparisons for Leland Olds Cyclone.
Figure 8. Burner Flow Patterns with -1.5% Tertiary Air Flow for LOS Cyclone.
-------
baseline
LP
HP
Spl15
SpIVS
Vspl
Chute
Figure 9. NO Results for Sioux
Cyclone Operational Changes.
Figure 10. NO Results for Sioux
Cyclone Design Changes.
1.00-
0.95-
0.90-
_ 0.85-
£
5 0.80-
"x 0.75-
0
z 0.70-
n
0.65-
0.60-
0.55-
0 50"
•
»
•»
•
i ' i ' i ' i ' i • i
bOU-
600 -!
> ^550r
f 500-
X
o
n \
&400r
^
O
2 300-
250-
200~
1 • FGR
® Water
ra
P
1
1
0.6 0.7 0.8 0.9 1 1.1 1.2 1.3 1.4
Barrel Stoichiometric Ratio
0 2 4 6 8 10 12 14 16 18 20
% FGR
Figure 11. NO Results versus
Sioux Barrel Stoichiometry.
Figure 12. NO Results for Sioux Cases
Involving FGR and Water Injection.
-------
%Twall2550 F and gas stoichiometric ratio<0.6).
Bias .9/1.4
Baseline
0.2 0.4
0.6 0.8 1
Ib NOx/MMBtu
1.2 1.4
Figure 14. Sioux Furnace NOx Emissions for Various NOx Reduction Strategies.
-------
SHORT-TERM NOX EMISSION REDUCTIONS WITH COMBUSTION
MODIFICATIONS ON LOW TO MEDIUM SULFUR COAL-FIRED
CYCLONE BOILERS
R. Himes
D. Hubbard
Carnot Technical Services
15991 Red Hill Avenue, Suite 110
Tustin, California 92780-7388
W.M. DePriest
Sargent & Lundy
55 East Monroe
Chicago, Illinois 60603-5780
B. Stone
Associated Electric Cooperative
2814 South Golden
Springfield, Missouri 65801-0754
D. Stopek
Illinois Power
500 South 27th Street
Decatur, Illinois 62525
Abstract
In anticipation of Title IV - Phase IINOX limit mandates for cyclone boilers, several utilities
initiated a Cyclone NOX Control Interest Group (CNCIG) in 1994 under the co-sponsorship of
the Electric Power Research Institute's (EPRI's) NOX Target. Due to the limited number of
demonstrated NOX control technologies available to cyclone boilers, CNCIG established an
engineering approach to explore, evaluate, and demonstrate promising alternatives for cyclone
NOX control. The current paper presents preliminary results from an ongoing investigation of
combustion modification approaches to NOX reduction implemented through biasing of
A165B287.DOC
-------
combustion air at full load, and simulating overfire air operation at reduced load. An overview
of the following areas is presented:
• cyclone boiler operating characteristics and limitations
• issues associated with sub-stoichiometric cyclone barrel operation,
• test programs developed to address these issues,
• test data gathered to date by two utilities from nominal 600 MW cyclone boilers, and
• remaining issues to be addressed by ongoing test activities.
Biased cyclone operation resultant in sub-stoichiometric conditions, though promising in
achieving nominal 15% - 20% NOX reductions at full load, and 35% - 50% NOX reductions at
minimum load through a simulated overfire air firing configuration, can pose significant near-
term and long-term risk to the boiler. Prior to long term implementation of combustion
modifications as part of an overall NOX compliance strategy, it is essential that these risks be
quantified so as not to impair the long term operability and reliability of the boilers.
Background
Cyclone boilers were first developed in the late 1940s to take advantage of the inherent
characteristics of certain fuels that were problematic when fired in pulverized coal design boilers.
These coals exhibited low ash fusion temperatures which caused slagging on furnace walls when
burned in a typical pulverized coal boiler. Furthermore, it was not economical to design larger
furnace volumes that would allow the slag to cool sufficiently before impinging on the furnace
walls. Babcock & Wilcox (B&W) developed the cyclone boiler in order to take advantage of the
slagging tendency of the fuel by burning the coal at a high heat release rate in the cyclones. In
cyclone boilers, combustion occurs within a water-cooled horizontal cylinder attached to the
outside of the furnace. Coal is gravity fed into the radial design burner tangentially in the same
rotation as the primary air (Figure 1). By appropriate partitioning of the primary and tertiary air
flows, the flame location relative to the burner can be adjusted. The main combustion air enters
through the secondary air damper which extends nominally 60% of the length of the cyclone
barrel. Firing of high moisture subbituminous coals has prompted some modifications to the
base design, with blanking plates installed over the first third of the secondary air duct, and
reduction of the tertiary air to the point where it only provides cooling of the burner front. These
changes are all focused at increasing the residence time of the coal for drying, devolatilization,
and ignition in the near burner region.
By burning the fuel at high turbulence and temperature, the coal ash is removed as molten slag,
thereby reducing the fly ash content of the flue gas and required furnace size. Overall air-rich
conditions were proscribed for the combustion chamber to ensure that the molten slag did not
corrode the cyclone boiler tubes. However, under these air-rich conditions, the high turbulence
and temperature present in cyclones produce relatively high levels of NOX emissions. Although
NOX formation can be minimized by reducing temperature and available oxygen during the
combustion process, this is difficult to do in cyclone boilers without promoting corrosion, or
freezing of the slag layer, as well as increasing levels of fly ash and unburned carbon. Although
A165B287.DOC
-------
existing cyclone boilers were grandfathered under the original Clean Air Act of 1970. B&W
stopped production due to the inability of cyclone boilers to achieve New Source Performance
Standards for NOX emissions at a cost comparable to pulverized coal-fired boilers.
Secondary Air Gas Burners
Coal Deslaggmg
Oil Burner
Replace.
Wear Liners
1 If
/• 1*3
Re-entrani Throar 1 JJt
Siag Tap Opening
Figure 1
Schematic of Radial Burner Design Cyclone Barrel (Steam, 1992)
Title IV of the 1990 Clean Air Act Amendments (CAAA) targeted reduction of NOX and SO:
emissions from the coal-fired boiler population grandfathered under the initial Clean Air Act.
NOX emission targets for tangential and wall-fired boilers were based upon a hardware
modification that retrofit low NOX burner technology. The design characteristics of cyclone
A165B287.DOC
-------
boilers, however, are not conducive to the application of standard low NOX burner technology.
In recognition of the limited NOX control options for cyclone boilers, several Department of
Energy (DOE) co-sponsored Clean Coal Technology programs demonstrated the application of
both gas and coal reburning on cyclone units. In addition, Title I of the 1990 CAAA has resulted
in the implementation of commercial selective non-catalytic reduction (SNCR) and selective
catalytic reduction (SCR) systems for NOX control on cyclone boilers located in the ozone
transport region in the Northeast. The relative cost of these technologies, however, has been
demonstrated to be relatively high, and as emission control strategies represent added costs to the
generation of power, there are currently strong incentives to minimize the costs of compliance.
Toward this end, several utilities are investigating the NOX reduction potential of combustion
modifications, in conjunction with the associated operational impacts and limitations, which may
include slag tapping impairment and accelerated tube wall corrosion.
Coal-Fired Cyclone Boiler NOX Emissions
Unlike most pulverized coal-fired boilers, in which fuel NOX contributions predominate, cyclone
boiler NOX emissions appear to stem from both fuel and thermal NOX in near equivalent
proportions. The combustion intensity per unit volume in a 10-foot diameter cyclone combustor
is typically 450.000 Btu/hr-ft3, whereas the typical heat release in a pulverized coal-fired unit is
20,000 Btu/hr-ft3. In addition, the heat absorption through the cyclone combustor walls is
relatively low (< 10%), primarily because of the small surface area, and the insulating effect of
the liquid slag layer. A majority of the crushed coal is burned on the wall in a cyclone, as
opposed to suspension burning in a pulverized coal-fired boiler. The high combustion intensities
and long residence times at peak temperatures in a cyclone are conducive to slag flow, but also
contribute to high thermal NOX formation and fuel bound nitrogen conversion. Assessments of
baseline coal-fired cyclone boiler NOX emissions have indicated that they can range from 0.85 -
2.7 Ib/MBtu. This broad range in the cyclone boiler population baseline emissions is indicative
that a number of factors contribute to NOX formation, among which include:
• fuel type (bituminous, subbituminous, coal blend, or lignite),
• unit size and configuration (screened furnace, open furnace with target wall, and opposed
wall)
For example, bituminous coal-fired cyclones trend toward higher NOX emissions with increased
unit size (Figure 2). Contrary to this trend, however, are cyclones fired with a high moisture, and
more friable coal (subbituminous or lignite), which exhibit an apparent insensitiviry to NOX
emissions that is likely due to the reduced peak flame temperatures attainable within the cyclone
barrel. The moisture acts as a diluent to reduce the peak flame temperature, while the coal
friability can result in a larger fraction of coal fines. As the coal fine fraction increases, a greater
proportion of the coal is burned in suspension as opposed to at the wall of the cyclone barrel,
thereby reducing the heat release within the cyclone barrel. The resultant reduced peak flame
temperatures limit the thermal NOX contribution to the total NOX.
A165B287DOC
-------
Demonstrated NOX control technologies for cyclone boilers to date have taken the tact of letting
the NOX form within the cyclone in order to minimize impacts to the cyclone operation. The flue
gas is subsequently made reducing through rebuming fuel addition, or treated with ammonia or
urea, to react with the NOX to form molecular nitrogen. An alternative approach is to limit the
oxygen availability within the cyclone barrel so as to limit the formation of nitrogen oxides.
Although this can be accomplished on larger cyclones by biasing combustion air from the lower
tier cyclones to the upper tier cyclones, or ultimately through incorporation of an overfire air
system, it is limited by the ability of the cyclone to maintain tapping of the slag flow from the
barrel and lower furnace, as well as by potential accelerated corrosion of tube wall surfaces from
hydrogen sulfide or pyritic iron. The following discusses issues associated with biased cyclone
firing operation and preliminary results obtained from two nominal 600 MW units firing an
Illinois bituminous coal and a Western subbituminous coal.
f
m
Subbituminous/Ligi ite Fired Cyclones
400 em
Unit Capacity (MW)
Figure 2
Comparison of Full Load NOX Emissions from
Coal-Fired Cyclone Boilers as a Function of Coal Type
Cyclone Boiler Operating Limitations
The primary parameters/issues involved in defining the cyclone boiler process operating
envelope for combustion modifications implemented through either biased firing at full load or
simulated OFA at low load include:
A165B287.DOC
-------
• coal properties (heating value, moisture content, ash properties)
• first stage stoichiometry
- fan operating limits under bias conditions
• slag tapping
• tube wall corrosion
• carbon carryover/utilization
Coal Properties
The coal properties define the operating limits that will likely be encountered first. Illinois
bituminous coals are relatively high in sulfur and iron, and thus present the greatest risk from
molten iron, iron sulfide, and/or hydrogen sulfide attack of water wall tubes. The relatively high
ash fusion temperatures typically provide sufficient margin relative to slag tapping
considerations. Western subbituminous coals, on the other hand, have moisture contents ranging
between 25% - 30%. which result in combustion temperatures several hundred degrees
Fahrenheit cooler and associated slag tapping constraints. A comparison of the coal properties
fired at Associated Electric Cooperative's New Madrid Station and Illinois Power's Baldwin
Station is presented in Table 1.
Table 1
Comparison of Coal Properties at New Madrid and Baldwin Stations
Associated Electric
Cooperative New Madrid
Proximate (as Received)
Moisture
Ash
Volatiles
Fixed Carbon
Heating Value
Ultimate (% dry)
Ash
Hydrogen
Carbon
Nitrogen
Sulfur
Oxygen
Mineral Analysis
Ferric Oxide
Sulfur Trioxide
Base/ Acid Ratio
Slag Viscosity (T250, F)
27.24%
4.63%
33.23%
34.90%
8,689
6.37%
4.98%
70.93%
1.00%
0.32%
16.40%
5.47%
11.27%
0.707
2,200
Illinois Power
Baldwin
14.86%
11.33%
10,541
13.31%
4.95%
68.76%
1.34%
3.39%
8.25%
13.52%
2.99%
0.330
2,500
A165B287.DOC
-------
First Stage Stoichiometry
Cyclone biasing is implemented by reducing the secondary air on lower level cyclones, and
increasing the air flow by the same amount on the upper level cyclones so as to maintain the
same boiler excess oxygen content. The extent to which cyclone biasing can be implemented,
however, is limited by the amount of air that can be introduced through the upper level cyclones,
and the windbox/fumace pressure differential capability of the forced draft fans. In addition, the
slag tapping characteristics in both the overall fuel rich and oxidizing cyclones can limit the
degree of biasing implemented. Although similar in base design, New Madrid was capable of
achieving a lower cyclone level Stoichiometry at full load of nominally 90%, whereas Baldwin
was limited to 100% theoretical air. As the NOX reduction potential is proportional to the long
term Stoichiometry that is achievable in the lower level cyclones and furnace, it should be noted
that results from one unit are not necessarily applicable to other units.
Slag Tapping
Continuous slag tapping is critical to the long term operation of the cyclone. Once the slag flow
becomes impaired, the barrel will fill with slag up to the reentrant throat, resulting in the cyclone
being taken out-of-service. The boiler may also need to be taken out of service if the buildup
occurs in multiple cyclone and cannot be resolved through operational changes. Current utility
practice for curing slag tapping issues is empirically based. Slag tapping problems encountered
with Western coals have typically been addressed by temporarily firing the problematic cyclone
with a higher heat content fuel, or blend. The greater heat release per mass of fuel results in
increased combustion temperatures, which in conjunction with potential benefits from the
blending of a low and high fusion slag, frequently restores proper tapping. Problems tapping
slag with bituminous coals, on the other hand, typically incorporate modification of cyclone
airflows to increase the combustion temperature and reduce the slag viscosity. Modification of
cyclone airflows to restore proper slag tapping can compromise NOX reduction benefits from
biased cyclone operation. Thus, short term results obtained under controlled conditions represent
a higher bound on the NOX reduction potential, with actual long term performance contingent
upon required system operation to address upset and transient conditions.
Water Wall Corrosion
The principle causes of water wall corrosion and forced outages in cyclone boilers are the (1)
formation of liquid iron, (2) formation of liquid FeS, and/or (3) sulfidation attack by H2S rich
flue gas. Each of these compounds is due to the occurrence of locally reducing conditions.
Based on numerical modeling studies and physical measurements within cyclones reported by
Adams, et. al. (1997), it has been shown that the cyclone barrels exhibit highly stratified regions
of fuel rich and oxidizing zones, irrespective of the overall Stoichiometry over a range of 0.9 -
1.15 Stoichiometry (Figure 3). To substantiate potential impacts from changes in iron formation,
biasing tests at Baldwin incorporated collection of slag samples, with subsequent analyses of the
slag composition. To address lower furnace corrosion potential from sulfidation attack,
hydrogen sulfide (H2S) analyses were conducted on flue gas samples extracted from the lower
furnace.
A165B287.DOC
-------
. 20%
20%
Staged, S.R.=0.95
20%
Baseline, S.R.= 1.15
Figure 3
Numerical Modeling Prediction of Cyclone Barrel Oxygen Contours Along the Midplane
(Glacier Code-Reaction Engineering International, 1997)
A165B287.DOC
-------
An important aspect that can only be addressed through longer term operation of the boiler, is
exposure of metal surfaces to alternating reducing and oxidizing environments, as well as
temporary excursions in reducing compounds during upset conditions not typically observed
during short term tests under controlled conditions. For example, during tests at Baldwin
Station, continuous operation of the cyclones under full load biased firing conditions resulted in a
spike in the hydrogen sulfide concentration in excess of full scale (>200 ppm). The spike was
attributable to a lower level cyclone being lost due to coal feed plugging with resultant increases
in the coal fed to the remaining 13 cyclones. Operationally, this resulted in further reductions in
the cyclone barrel stoichiometry since the fuel feed system controls responded to maintain load
faster than the airflow controls to the individual cyclone barrels. The reducing conditions in the
lower furnace were thus increased temporarily with concomitant increases in reducing
compounds such as hydrogen sulfide.
Carbon Carryover/Utilization
Finally, increased mass flow through the upper level cyclones, or incomplete combustion within
the furnace during full load biasing can result in increased carbon carryover into the particulate
collection device. Due to the reduced furnace size of cyclone boilers, there is less residence time
available to complete the suspension burning of coal particles escaping the cyclone barrels.
Also, because of the significant removal of ash as slag, only 40% - 50% of the ash is captured as
fly ash. Thus, any increase in carbon can represent a significant increase in the unburned carbon
content. For example, assuming 60% capture of ash as slag, a 10% unburned carbon level in the
fly ash would be equivalent to a 4% unburned carbon level from a pulverized coal boiler in
which all of the ash appears as fly ash. Although the economic impact from the unburned carbon
levels may be moderate, increased carbon levels in the fly ash can lead to precipitator fires and
increased forced outage time. In addition, units with marginal precipitators could experience
increased opacity due to the increase in ash resistivity and reduced collection efficiency.
Approach
A test approach was developed so as to collect sufficient unit operating information to:
• evaluate potential operational and combustion modification approaches for reducing NOX
emissions,
• assess current potential operating limitations due to slag tapping and furnace wall corrosion,
or equipment limitations due to fan cyclone biasing capabilities, and
• establish short term NOX versus load emission profiles under baseline and modified operating
procedures for incorporation into emissions averaging scenarios.
Please note that the initial short term tests were targeted at providing proof-of-concept for any
NOX reduction potential, as well as to identify any short term operating limitations. Based on the
results from these tests, longer term tests would then be considered to evaluate longer term
operating limitations and NOX reduction performance over a range of coal, weather, and unit
conditions. Also to be factored into the evaluation of the combustion modification approach for
A165B287.DOC
-------
achieving NOX reductions are the incremental costs associated with possible long term corrosion
of water wall panels, as well as other adverse operational issues.
In order to establish individual cyclone stoichiometries, it is necessary to have primary/tertiary
and secondary air flows monitored, as well as the coal flow. Both New Madrid Unit 2 and
Baldwin Unit 1 have gravimetric feeders and distributed control systems that monitor the
differential pressures across the bellmouth and primary/tertiary air ducts for each of the cyclone
barrels. The airflows were then tabulated based upon calibrated K-factors for each cyclone barrel
and the following equation:
Mass Flow Rate (Ib/hr) = K * [Pressure Diff (i.w.c.) * Density (Ib/ft3)]
The cyclone barrel stoichiometries were then tabulated based upon the measured air coal flows,
and the calculated theoretical air/coal ratio determined from the coal analysis.
So as not to jeopardize unit operability, initial tests focused on single cyclone tests in which the
stoichiometry was modified over a range of conditions. These tests provided an indication of the
maximum stoichiometry achievable under a given windbox to furnace pressure differential and
maximum percent open damper settings. Based upon these airflows, fuel rich stoichiometries
were calculated that would maintain a constant overall boiler excess oxygen level. Observations
were then performed over fuel rich to fuel lean conditions regarding slag tapping characteristics,
flame quality, and temperature. Although cyclone barrel measurements were performed at New
Madrid Station, results confirmed numerical modeling information in which a relatively unmixed
flow field was found to exist within the cyclone barrel. As a result, cyclone barrel measurements
indicated fuel rich and fuel lean regions within the barrel, irrespective of overall operating
conditions. Thus, measurements within the cyclone barrel were not conducive to providing
insights regarding changes to species composition as the line of sight access available from the
view port passed through highly stratified flow fields.
Based upon successful operation of individual cyclone barrels over the range of stoichiometries
of interest, the entire unit was biased at full load for a limited period of time. All biasing was
accomplished through the available bias controls, with the degree of bias limited by the amount
of air that could be introduced through the top level cyclones. A simulated overfire air test was
also conducted at a minimum load, in which the top level of cyclones were out-of-service, which
served as overfire air ports for introduction of burnout air. Comparisons between baseline and
biased operation were made for the following parameters:
• cyclone barrel temperature,
• lower furnace hydrogen sulfide concentrations,
• oxygen, CO, and NOX variation at the economizer outlet.
• unbumed carbon in ash
A165B287.DOC 10
-------
Unit Descriptions
New Madrid Unit 2
New Madrid Unit 2 became operational in 1977 with a design nameplate rating of 4,365,000
Ibs/hr primary steam generation at 1,005 F and 2,400 psig throttle pressure. Reheat steam
generation is 3,995,000 at 1,005 F and 635 psig. The unit is equipped with fourteen ten-foot
diameter cyclone barrels having radial design burners. The cyclone barrels are staggered three
over four, with seven on each side of the furnace.
Although initially designed to be fired with Illinois coal, New Madrid Generating Station has
fired Western subbituminous coal since mid-1994. As part of the fuel switch to low-sulfur
Western coal, blanking plates were installed in the front of the secondary air inlets to each of the
cyclone barrels. The purpose of the blanking plates is to increase retention of the coal within the
cyclone barrel.
Flue gas recirculation fans were originally supplied for steam temperature control, but were
removed prior to 1984, with the furnace penetrations being tubed over. Additional heat transfer
surface, sootblowers, and water blowers were added to enable full load operation by controlling
the furnace exit gas temperature (FEGT) within limits with Western coal. Convective pass
plugging with the low-sulfur Western coals has been experienced, with changes being made in
the unit control logic to limit a calculated FEGT to 2,450 F.
Fly ash is disposed of dry in a 70 acre pond. The electrostatic precipitators have ample capacity
with permit limits for opacity of 40%, and current CEMS opacity readings of nominally 15%
under baseline full load operating conditions.
Baldwin Unit 1
Baldwin Unit 1 became operational in 1970 with a design nameplate rating of 4,200,000 Ibs/hr
primary steam generation at 1,005 F and 2,620 psig throttle pressure. Reheat steam generation is
3,787,000 Ibs/hr at 1,005 F and 549 psig. Similar to New Madrid, the unit is equipped with
fourteen ten-foot diameter cyclone barrels having radial design burners. The cyclone barrels are
staggered three over four, with seven on each side of the furnace. The unit is capable of
achieving full load with 13 of 14 cyclones in service enabling feeder and coal pipe maintenance
while on-line. A side-view schematic is presented in Figure 4.
The unit was designed to be fired with Illinois coal, although it has purchased a washed coal to
remove pyritic sulfur since 1973. Steam temperature control is provided by attemperation and
flue gas recirculation. The unit is balanced draft, and since the incorporation of a baffle plate at
the economizer outlet to increase ash retention in the hopper, the unit is limited in load to near
595 MW due to induced draft fan limitations.
Fly ash is disposed of locally in a man-made pond. The electrostatic precipitators have marginal
capacity with permit limits for opacity of 30%, and current CEMS readings of nominally 20%
AI65B287.DOC 11
-------
under baseline operation. ESP limitations are important, as increased carbon content in the ash
can exacerbate particulate collection performance and result in increased opacity.
Figure 3
Side View Schematic, Illinois Power Baldwin Unit 1
New Madrid Unit 2 Cyclone Bias Results
Baseline and biased cyclone tests were conducted at 585 MW, 445 MW, and 350 MW in order
to document operating and emissions profiles under controlled conditions over the load range.
Based upon the three loads tested (Figure 5), the baseline NOX exhibited a sensitivity of 0.57%
delta NOX per percent reduction in maximum continuous rating. This NOX sensitivity to load is
A165B287.DOC
12
-------
NOX reductions of nominally 18% were achieved at lull load when the lower level cyclones were
biased to 90% stoichiometric levels, with the balance of air introduced through the upper tier
cyclones. As indicated in Table 2, however, NOX reductions were found to increase from 18% at
585 MW to 35% at 350 MW with all lower level cyclones operated at 90% stoichiometry and a
varying number of cyclones operating in the upper level. The increased NOX reductions at lower
load, for a given reduction in lower tier cyclone stoichiometry, is attributed to the fewer number
of cyclones operating under fuel lean conditions in the upper level cyclones. All six upper level
cyclones were in service during full load operation of 585 MW, while only two cyclones in the
upper level were operated at 445 MW, and no upper level cyclones were in service at the low
load condition of 350 MW. The increased stoichiometry within the fuel lean cyclones increased
the partial pressure of oxygen available for fuel nitrogen conversion and thermal NOX formation.
Thus, some of the NOX reductions achieved through reduced operating stoichiometries in the
lower level cyclones was offset by increased NOX formation in the upper level cyclones. The low
load test approximates the percentage NOX reduction achievable if all cyclones were able to be
operated under reduced stoichiometry conditions of 90%.
Table 2
Summary of NOX Reductions with Cyclone Biasing as a Function of Load at New Madrid
Load Lower Tier Cyclone
(MW) Stoichiometry
(# Operating Cyclones)
585
445
350
.89 (8)
.87 (8)
.88 (8)
Upper Tier Cyclone NOX Reduction Relative
Stoichiometry to Baseline
(# Operating Cyclones) (%)
1.49(6)
1.44(2)
Not Applicable (0)
18%
27%
35%
To begin the process of assessing potential long term operating impacts and costs attributed to
biased cyclone operation, lower furnace hydrogen sulfide, economizer outlet carbon monoxide,
and ash LOI measurements were performed and compared with baseline values. These results
are summarized in Table 3. Hydrogen sulfide measurements in the lower furnace under biased
operation indicated a ten fold increase from baseline levels of 2 - 3 ppm to 20 - 30 ppm. Carbon
monoxide concentrations at the economizer outlet were not found to change significantly,
thereby indicating complete combustion within the furnace, even under biased operation. In all
cases, CO measurements across the 24-point sampling grid were less than 50 ppmv. Contour
plots of the excess oxygen levels at the economizer outlet were also indicative of relatively
uniform combustion within the furnace, in spite of the potential for combustion air stratification
due to biasing.
A165B287.DOC 14
-------
consistent with other coal-fired cyclone boilers of similar size. It should be noted that the data
collected was obtained with a different number of operating cyclones, although each cyclone in
service was fired at similar heat input rates per cyclone. Thus, heat release rates within operating
cyclones were constant over the three loads tested, while heat release within the furnace volume
decreased. The reductions in NOX observed as the load was reduced is thus likely representative
of lower thermal NOX formation in the furnace.
1.5
1.4
1.3
1.2
1.1
Baseline Data (1.08-1.19 Stoichiometry)
..lower Tjer.Cy.ctones.g).0,9 Stoicfiiometjy
300 350 400 450 500
Load (MW)
550
600
Figure 5
Short Term NOY versus Load Curves for New Madrid Unit 2
NOX reductions of nominally 18% were achieved at full load when the lower level cyclones were
biased to 90% stoichiometric levels, with the balance of air introduced through the upper tier
cyclones. As indicated in Table 2, however, NOX reductions were found to increase from 18% at
585 MW to 35% at 350 MW with all lower level cyclones operated at 90% Stoichiometry and a
varying number of cyclones operating in the upper level. The increased NOX reductions at lower
load, for a given reduction in lower tier cyclone Stoichiometry, is attributed to the fewer number
of cyclones operating under fuel lean conditions in the upper level cyclones. All six upper level
cyclones were in service during full load operation of 585 MW, while only two cyclones in the
A165B287.DOC
13
-------
upper level were operated at 445 MW, and no upper level cyclones were in service at the low
load condition of 350 MW. The increased stoichiometry within the fuel lean cyclones increased
the partial pressure of oxygen available for fuel nitrogen conversion and thermal NOX formation.
Thus, some of the NOX reductions achieved through reduced operating stoichiometries in the
lower level cyclones was offset by increased NOX formation in the upper level cyclones. The low
load test approximates the percentage NOX reduction achievable if all cyclones were able to be
operated under reduced stoichiometry conditions of 90%.
Table 2
Summary of NOX Reductions with Cyclone Biasing as a Function of Load at New Madrid
Load Lower Tier Cyclone
(MW) Stoichiometry
(# Operating Cyclones)
585
445
350
.89 (8)
.87 (8)
.88 (8)
Upper Tier Cyclone NO, Reduction Relative
Stoichiometry to Baseline
(# Operating Cyclones) (%)
1.49(6)
1.44(2)
Not Applicable (0)
18%
27%
35%
To begin the process of assessing potential long term operating impacts and costs attributed to
biased cyclone operation, lower furnace hydrogen sulfide, economizer outlet carbon monoxide,
and ash LOI measurements were performed and compared with baseline values. These results
are summarized in Table 3. Hydrogen sulfide measurements in the lower furnace under biased
operation indicated a ten fold increase from baseline levels of 2 - 3 ppm to 20 - 30 ppm. Carbon
monoxide concentrations at the economizer outlet were not found to change significantly,
thereby indicating complete combustion within the furnace, even under biased operation. In all
cases, CO measurements across the 24-point sampling grid were less than 50 ppmv Contour
plots of the excess oxygen levels at the economizer outlet were also indicative of relatively
uniform combustion within the furnace, in spite of the potential for combustion air stratification
due to biasing.
Table 3
Summary of Biased Firing Impacts on H2S, CO, and LOI at New Madrid
Load Lower Furnace H2S Economizer Exit CO ESP Hopper LOI
(MW) (ppmv) (ppmv) (weight %)
585
350
22.2-28.4
25.3-28.6
19
22
13%
4%
A165B287.DOC 14
-------
Baldwin Unit 1 Cyclone Bias Results
Baseline and biased cyclone tests were conducted in June and July at 590 MW and 340 MW in
order to document operating and emissions profiles under controlled conditions over the load
range. Complete analysis of the results was not completed for incorporation into the current
paper, although NOX reductions at full load (595 MW) were on the order of 17%, while under
simulated overfire air conditions at 340 MW load achieved 50% NOX reduction. Hydrogen
sulfide measurements in the lower furnace ranged from 30 to 90 ppm while the lower level
cyclones were operated at a stoichiometry of 90%. Additional analyses are examining slag
composition for changes in the different forms of iron relative to baseline and biased firing
operation. It is important to note that these tests, although of 8 - 10 hour duration, are still short
term in nature, and conducted under controlled conditions as the units were not under load
following operation.
Summary and Future Plans
Although cyclone biasing at full load indicated 15% - 20% NOX reductions were achievable, the
tests were conducted over relatively short periods of time, and under controlled conditions. Long
term tests are planned for New Madrid Unit 2 to determine a more representative NOX reduction
potential under normal load dispatch conditions, as well as operating impacts from slag tapping
and potential accelerated water wall wastage. Pending a complete analysis of the results from the
short term evaluation of biased firing and simulated overfire air, Baldwin Station will evaluate
the prudency of proceeding with a longer term demonstration with a relatively high sulfur and
iron coal. Simulated overfire air tests at reduced load indicate an increased NOX reduction
potential on the order of 35% - 50%, albeit tempered by similar long term operating concerns
expressed above for biasing. It is premature to speculate on the operating and maintenance cost
impacts associated with combustion modifications until longer duration tests are completed. At
this point in time, indications justify continued investigations, with evaluations proceeding.
Acknowledgements
The authors wish to thank all of the New Madrid and the Baldwin staff for all their assistance
during the testing. Also, we aknowledge the expert efforts of CONSOL who conducted lab
analysis at Baldwin. Illinois Power further thanks both the State of Illinois and EPRI for
cofunding the test program at Baldwin.
A165B287.DOC 15
-------
COMBUSTION TEMPERING DEMONSTRATION
ON A CYCLONE UNIT FOR NOX CONTROL
Peter D. Marx, P.E.
RJM Corporation
Ten Roberts Lane
Ridgefield, CT 06877
and
Christopher J. Barton
Public Service of New Hampshire
97 River Road
Bow, NH 03304
Abstract
Cyclone furnace NOX reductions up to 22% with 0.5% heat rate penalty were achieved using a
recently developed process called Combustion Tempering. Test data indicates higher NOX
reductions are probable. A detailed test program was performed on Public Service Company of
New Hampshire's Merrimack Station Unit 1 for the purpose of controlling NOX emissions using
RJM's patented technology "Combustion Tempering." Combustion tempering consists of
targeting cooling streams to specific areas in the cyclone for the reduction of NOX • In a cyclone
furnace, thermal NOX is a large portion of the total NOX generated. Thermal NOX is formed at
elevated temperatures coincident with O2 interaction. CFD modeling results identified specific
zones of thermal NOX production within the cyclone that comprise over 60% of the total NOX
produced by the cyclone. These zones are small on a volume basis and require little cooling
medium to significantly reduce the zone temperature. Model results indicated that a limited
amount of cooling in a single targeted zone would result in a 25% reduction in NOX. Short term
testing confirmed the correlation with model results. Cooling the largest single zone during the
test period produced NOX reductions ranging from 18% to 22%. There was no operational
impact on cyclone operation and no apparent increase in coal flow during the testing period.
Projected costs for a commercial system on this unit at a 22% NOX reduction are $13 I/ton of
NOX removed based on S2.50/KW installed cost. Model results indicate cooling multiple zones
will increase the NOX reduction to 35%.
-------
Background
Public Service of New Hampshire's Merrimack Station is located in the Town of Bow, New
Hampshire on the Merrimack River. Merrimack Unit 1 is a drum type Babcock & Wilcox
cyclone fired boiler designed for 815,000 Ib/hr of steam at 1800 psig and 1005°F main steam and
1005 °F reheat steam. The boiler has three cyclones arranged one over two. The boiler was built
in 1960 and designed for firing coal. A side view of the boiler is shown in Figure 1.
Title I of the 1990 Clean Air Act requires that Unit 1 make a significant reduction in the
emission of nitrous oxide (NOX). Regulations imposed by the State of New Hampshire instituted
a NOX limit of 0.92 Ib/mmBtu. Uncontrolled NOX emissions on this unit range between 1.3 and
1.5 Ib/mmBtu depending upon coal blend and furnace cleanliness. Public Service of New
Hampshire performed an evaluation of the unit and installed an ammonia based SNCR System
for the reduction of NOX. Operation of SNCR technology proved somewhat successful but is
costly to operate.
Public Service of New Hampshire and RJM Corporation agreed to cost share the development
and demonstration of Combustion Tempering as a NOX control technology for Unit 1. The focus
was to reduce NOX emissions at an operational cost significantly less than the SNCR System.
An SNCR System has injection rates ranging from 1% - 2.5% of flue gas flow. A parameter of
the program was to maintain combustion tempering flows below that of the SNCR System. In
addition, the technology must not impact the cyclone combustion process and slag tapping
characteristics.
Technology Development Program
The development of combustion tempering technology occurred over the past two years. Step 1
focused on modeling one of the three cyclones on Unit 1. The general cyclone operating
conditions used as model inputs were as follows:
Table 1
Cyclone Operating Parameters
Coal Flow, Ib/hr 30,000
Secondary Airflow, Ib/hr 240,000
Primary Airflow, Ib/hr 60,000
Seal Airflow, Ib/hr 6,000
Water Circuit Temperature, °F 600
Secondary Air Temperature, °F 550
Primary Air Temperature, °F 550
Water Circuit Pressure, psig 2,500
-------
Public Service Company of New Hampshire
Merrimack Steam Plant - Unit No. 1
Bow, New Hampshire
Fisure 1
-------
The coal used in the model was a blend of the coals used in Unit 1 and the proximate analysis is
listed in Table 2.
Table 2
Coal Analysis
Coal Type
% Moisture
% Ash
% Volatile Matter
% Fixed Carbon
HHV, Btu/lb
% Sulfur
High Sulfur
6.30
6.23
37.11
50.36
13,340
2.53
Low Sulfur
14.0
5.0
34.0
47.0
11,500
0.5
50/50 Blend
10.15
5.62
35.56
48.68
12,240
1.52
CFD Modeling
RJM Corporation used Fluent CFD and NOX software to model the cyclone combustion process
and identify the cyclone NOX profile. A baseline model was developed that reflected the
combustion characteristics of the Unit 1 cyclone. The geometry of the model is shown in
Figure 2. Multiple high NOX areas were identified within the cyclone as shown below.
FRDNT
Slo~t
EMU
NOx Production Zones
NOx Production Zones
Over 60% of the total NOX occurs in a targetable number of areas. Each of the targetable zones
were reduced 100°F to represent the impact of a cooling media on that zone. Total heat loss due
to the cooling is 0.5%. The model predicted a 25% NOX emissions reduction within a single
-------
CDAL INLET
30"
RADIAL
BURNER
34"
PRIMARY
AIR INLET
SECONDARY AIR INLET SLOT
87"
108"
CYCLONE GEOMETRY
FIGURE 2
44"
54" RE-ENTRANT
THROAT
-------
combustion tempered zone. Because of the various modeling assumptions involved (coal
volatilization and burnout characteristics, coal sizing and distribution, coal composition,
turbulence modeling, combustion rate constants, etc.) the model NOX results were used in
comparing trends and in quantifying the relative effectiveness of the combustion tempering
technology rather than predicting absolute NOX emission levels.
Water was injected into the model at the two major NOX locations at various droplet sizes.
Table 3 illustrates a summary of the model results.
Table 3
Summary of Model Results
NOX Production Zones Water Spray Droplet Size NOX Reduction
(gpm) (microns) (%)
Radial Burner Exit 4 30 10
Radial Burner Exit 4 50 17
Radial Burner Exit 4 70 15
Secondary Slot 4 30 21
Secondary Slot 4 50 25
Secondary Slot 4 70 17
Based on modeling results, an evaluation was made of the possible Combustion Tempering
approaches and a test program was developed.
Field Testing
A month long field test was conducted on all three cyclones of Unit 1. The focus of testing was
to determine the following:
• Confirm the location of the largest controllable cyclone NOX zones.
• Determine the NOX reductions achievable on a per cyclone basis with optimum cooling
media distribution.
• Measure the total NOX reduction achievable cooling a portion of all three cyclones.
Verify the correlation between CFD modeling and field test data.
• Observe the effect on combustion process and cyclone slag tapping characteristics.
Determine the effect of Combustion Tempering on fly ash LOI.
-------
Water was sprayed into various locations of Merrimack Unit No. 1's three cyclones to determine
the effectiveness for NOX control. Existing ports were utilized which limited the access to the
optimum zones. The majority of the water spray tests were directed into the two zones identified
by the cyclone CFD model. Limited testing was performed on other zones in an attempt to
identify the size and locations of the NOX production zones in the cyclone. The unit's
Continuous Emission Monitor (CEM) was used to measure NOX emissions. Testing initially
focused on spraying water into zones in one cyclone to find the location that provided the peak
NOX reduction. Three probes were used to spray into the peak zone of each cyclone to determine
the overall unit NOX reduction.
Three air atomized probes were developed that would produce a consistent water droplet size.
Atomizing air pressure could be controlled to vary the droplet size from the probes. A
relationship between droplet size and NOX reduction was established and compared against the
model results.
Testing results varied due to coal blending and the furnace soot blowing cycle. Multiple tests
were run to determine the average conditions.
Coal Blending
The unit blends various low, medium, and high sulfur coals depending on price and SO2 cap
requirements. The coal used during the test period was a 50/50 blend of medium sulfur and high
sulfur coal. Coal blending is done in the yard on a "blade by blade" basis during the bunkering
period. The blended coal is not considered a homogeneous mixture. The medium sulfur coal,
when burned on its own, produces higher baseline NOX. The shift in the baseline NOX trend is
due to differing percentages of blended coal burned throughout the day.
Unit Soot Blowing
Soot blowing is performed every four hours for a one hour and fifteen minute period. At the end
of the period the furnace is in a "clean" condition and the baseline NOX is lowered anywhere
from 3% to 5% from the "dirty" condition. RJM Corporation has conducted in-cyclone NOX
measurements on similar front fired cyclone units and found that 90% of the total NOX produced
is from the cyclone in the "clean" condition. As the furnace walls become sooted, the NOX
generation within the furnace increases and the percent reduction achieved from combustion
tempering decreases.
Test Results
Conclusions from field testing are summarized as follows:
• A total NOX reduction of 18% to 22% was achieved when cooling a portion of a single
controllable NOX zone in each cyclone. Results varied as unit soot blowing cycles
affected boiler cleanliness.
-------
The heat rate penalty was 0.5% at a 22% NOX reduction.
The combustion tempering flow at 22% NOX reduction was equivalent to .5% of the total
flue gas flow, significantly below that of the SNCR System.
All three cyclones added proportionally to the total NOX output of the unit.
The NOX reduction achieved from one cyclone was additive to reductions achieved from
the other cyclones.
Controllable zones of NOX production existed in the cyclone that matched the CFD
modeling results.
NOX reductions achieved in one zone was additive to a second zone within the same
cyclone.
Testing also demonstrated that higher NOX reductions were probable at higher
Combustion Tempering rates by targeting multiple combustion tempering zones.
Slag tapping was unaffected by the Combustion Tempering process.
Fly ash LOI was unchanged by the Combustion Tempering process.
Visual observations of the cyclone tapping were made throughout the test period and no impact
was observed. Testing was conducted for 8 to 10 hour periods during the month long tests. At
all times the unit was run at full load operation with air and fuel inlet conditions that matched the
model input conditions. Multiple tests were run at the baseline condition followed by tempering
tests to verify the quantity of the NOX reduction. Figure 3 illustrates the average NOX reductions
achieved when tempering one NOX zone in all three cyclones and the modeling projectedNOX
reductions with cooling multiple locations in each of the three cyclones.
Test Series I
The first series of tests were conducted by spraying water through a center axis port in the
cyclone burner at three different insertion depths. These were 50", 60" and 70" as measured
from the probe tip to the end of the view port at the radial burner where the probe was inserted.
Table 4 summarizes the unit NOX reduction achieved when spraying into one of the three
cyclones.
-------
40%
35%
30%
MERRIMACK UNIT 1 COMBUSTION TEMPERING
TEMPERING FLOW vs NOx REDUCTION
o
Q
x
i
b
z
25%
20%
15%
10%
FUTURE
(MULTIPLE ZONES PER CYCLONE)
ACTUAL
(ONE ZONE PER CYCLONE)
o%
10
12
14
16
TEMPERING WATER FLOW
GPM
FIGURE 3
-------
Table 4
Combustion Tempering through Center Axis Port in One Cyclone
Insertion Depth Water Spray Droplet Size NOX Reduction
(") (gpm) (microns) (%)
50 4 50 2
50 4 70 2
50 4 30 3
60 4 50 3
70 4 50 2
70 12 300 10
Test Series II
The probe was then bent to allow water injection into the NOX production zone predicted by the
CFD model at the radial burner outlet. Controlling the spray of the water into target zones
produced up to three times the NOX reduction as non directed spray. Table 5 is a summary of
these test results.
Table 5
Targeted Combustion Tempering in One Cyclone
Insertion Depth Water Spray Droplet Size NOX Reduction
(") (gpm) (microns) (%)
50 4 50 6
60 4 50 6
70 4 30 6
70 4 50 7
70 4 70 8
70 4 90 8
Though the results looked promising, after two hours of use the spray nozzle was severely caked
with coal and required cleaning. This location was tested on the other two cyclones with similar
results. Due to the constant cleaning requirement of the spray nozzle, testing was redirected to
the secondary air slot region.
Test Series III
Additional testing was conducted along the secondary air slot of each of the three cyclones. A
probe was inserted through the ignitor viewing port and water was sprayed at various zones
along the slot. Different probe tips were used that produced different spray types and angles.
A summary of a set of tests of one cyclone is listed in Table 6.
-------
Table 6
Combustion Tempering at the Secondary Air Slot in One Cyclone
Insertion Depth
(")
40
50
60
70
80
80
80
Water Spray
4
4
4
5
5
5
5
Droplet Size
(microns)
50
50
50
50
50
70
80
NOX Reduction
5
6
6
7
8
7
6
The insertion depth was measured from the atomizer tip to the view port face. Higher flow rates
were required at the longer insertion depths to prevent the test probe from overheating. After
several hours of testing the probe was removed and inspected for damage. The probe was clean
and no visual damage could be detected. A downward fan spray pattern was found to work the
best as other patterns tested did not provide adequate coverage which limited the NOX reduction.
Each of the tests were repeated on the other two cyclones with similar results. A summary of the
NOX reductions achieved for each of the zones tested in cyclone A are shown in Figure 4.
Three Cyclone Testing
One probe was inserted into each of the three cyclones to determine the maximum achievable
NOX reduction. All probes were located along the secondary slot. Due to mechanical
interferences between the probe and piping outside B cyclone, it was not possible to insert the
probe at the optimum location. See Table 7.
Table 7
Combustion Tempering at the Secondary Air Slot in All Three Cyclones
Insertion Depth (")
Cyclone
ABC
80
80
80
80
50
50
50
50
80
80
80
80
Water Spray
(gpm)
16
16
16
16
Droplet Size
(microns)
50
50
50
50
NOX Reduction
21.5
22
18
20
-------
7. NDx REDUCTIDNS WHEN
SPRAYING WATER AT VARIOUS LOCATION
ERDNT
Secondary Slot
END
4 GPM 4 GPM 4.8 GPM 6 GPM
57. - 77. 57. - 77. 77. - 107. 77. - 97.
RADIAL
BURNER
4 GPM
87. - 11X
4 GPM 4 GPM 4 GPM
27. 37. 2X
CYCLONE A
EIGURE 4
-------
As previously noted the total reduction would vary based on the unit soot blowing cycle and the
coal blend on a given day. Total NOX reductions varied between 18% to 22% for the testing
period. The reductions were stable, repeatable and additive to the reduction achieved with the
SNCR System.
Multi-Zone Testing
Multiple zones were tested in a single cyclone to determine if the reductions were additive.
Probes were inserted to spray water into two separate zones along the secondary air slot (Sec),
one at the 50" insertion depth and the other at the 80" insertion depth. A third probe was
inserted through the radial burner (Rad) at the 50" insertion depth and bent to spray a targeted
zone. Data was taken with various combinations to determine the effect as shown in Table 8.
Table 8
Targeted Multiple Zones in One Cyclone
No. of Probes & Location Tempering Flow Rate NOX Reduction
(gpm) (%)
1 Rad 4 7
1 Rad, 1 Sec 8 12
1 Rad, 2 Sec 14 15
2 Sec 10 10
1 Sec 4 5
It is suspected that some spray overlap occurs when spraying into multiple zones that decreases
the Combustion Tempering efficiency.
Long Term Testing
Following the series of month long tests, a long term test was conducted to determine if spray
hardware could last in the optimum tested spray location. The hardware used for the testing was
permanently installed in October 1996 for a six month test in two of the three cyclones. The
orientation of the spray was altered from the original tested orientation since an optimal access
port location was not able to be installed without a unit outage. This orientation and elimination
of spray in one cyclone limited the achievable NOX reduction to a total of 7% to 10%. Following
the six month test period, the hardware was inspected and found to be worn but totally
functional.
-------
Projected Costs
A commercial Combustion Tempering system has been estimated to cost S2.50/KW for this size
unit. This cost would reduce to an estimated $ 1.30/KW for units in excess of 500 MW. Costs
include modeling, engineering, injection hardware, pumping and metering skid, installation, and
start-up. On this 130 MW unit, costs would be $13I/ton of NOX reduced based on a 22% NOX
reduction.
The existing ammonia based SNCR System annualized costs are approximately $700/ton of NOX
reduced based on a 35% NOX reduction. Combining a urea based SNCR System with
Combustion Tempering to obtain the same targeted 35% NOX reduction would reduce the
annualized cost to $350/ton of NOX reduced. Combining the technologies has the potential of
yielding NOX reductions of 60 - 65% from baseline NOX levels at a cost of $454/ton of NOX
reduced.
-------
REDUCTION OF NOx EMISSIONS WITH CYCLONE BURNERS BY
BIASING OF COMBUSTION AIR
Gerald M. Goblirsch
Northern States Power Company
414 Nicollet Mall
Minneapolis, MN 55401
Darrel G. Knutson
Timothy C. Peterson
Northern States Power Company
Allen S. King Generating Station
210 North 10th Avenue
Bayport, MN 55003
Zigmund J. Frompovicz
Electric Power Technologies, Inc.
830 Menlo Avenue, Suite 201
Menlo Park, CA 94025
Abstract
Northern States Power Company (NSP) has conducted tests to evaluate cyclone air
flow biasing to reduce NOx emissions at Unit 1 of the Allen S. King Generating
Station. King Unit 1 is a 600-MW opposed-wall, cyclone-fired boiler with 12 cyclone
burners and two elevations on each wall (three burners per elevation). The test
program compared NOx emissions with a dirty furnace (significant wall ash
deposits) to a dean, water-washed furnace. At 500-MW with a dirty furnace, NOx
emissions were 1.38 Ib/MBtu with 12 cyclones in service and 2.5 percent excess O2 at
the boiler exit. NOx emissions were 1.23 Ib/MBtu at the same operating conditions
with a clean furnace (an 11 percent reduction in NOx emissions). Tests to evaluate
the effect of secondary air flow biasing were conducted with both a dirty and clean
furnace and with the stoichiometry of the cyclones in the bottom elevation reduced
to 0.90 (fuel rich). Secondary air biasing with a dirty furnace reduced NOx emissions
17 percent from 1.38 to 1.14 Ib/MBtu, whereas NOx emissions with a clean furnace
were lowered by 24 percent from 1.23 to 0.94 Ib/MBtu. Secondary air biasing had no
significant impact on boiler operations. Boiler exit CO remained unchanged, and
-------
there were slight increases in furnace exit gas temperature and carbon in the ash.
Electrostatic precipitator operation and stack opacity were not affected.
Analysis of flue gas samples along the furnace walls below the bottom cyclone
elevation and above the top elevation showed that with and without secondary air
biasing, excess O2 was greater than 1.0 percent, hydrogen sulfide (H2S) was generally
less than 50 ppm, and carbon monoxide (CO) was less than 1,000 ppm. Higher
concentrations of CO and H2S were measured in the center of the furnace in the
proximity of the biased cyclone's reentrant throat. However, the concentrations
measured do not pose a significant concern for increased tube wall wastage. Based
on these positive results, NSP is planning to operate King Unit 1 with secondary air
biasing for an extended period to obtain long-term data on NOx emissions and
waterwall impacts.
Introduction
Northern States Power Company (NSP) is a member of the EPRI Cyclone NOx
Controls Interest Group (CESTCIG). CINCIG has sponsored modeling and
experimental work at other utility companies to evaluate the potential of cyclone
burner secondary air biasing for reducing NOx emissions1 Results from this work
showed that reductions in NOx emissions of greater than 20 percent were possible
with air biasing. Based on these promising results, NSP elected to further evaluate
the secondary air biasing as a NOx reduction technique on its cyclone fired boilers
for potential implementation.
NSP has elected to utilize a system-wide averaging NOx emission plan, instead of
meeting specific NOx emission limits on individual power plants. The use of
secondary air biasing to reduce NOx emissions on NSP's cyclone boilers fits within
the needs of NSP's system-wide NOx emission averaging plan.
Electric Power Technologies, Inc. (EPT) worked with NSP to develop a test program
to evaluate the feasibility of utilizing secondary air flow biasing to reduce NOx
emissions at Unit 1 of the Allen S. King Station. The purpose of the tests was to
determine the NOx emission reductions possible, the amount of lower cyclone air
biasing required, and effects on boiler operations and performance, ash deposition,
opacity, and potential waterwall tube wastage due to production of reducing
combustion conditions by measurement of hydrogen sulfide (H2S) and carbon
monoxide (CO).
-------
King Unit 1 Description
King Unit 1 is a 600-MW opposed-wall, cyclone-fired, supercritical boiler
manufactured by Babcock & Wilcox Company. The unit was commissioned in 1968
and has a boiler nameplate rating of 3,873,000 Ib/hr of steam flow at 1,005°F
superheat temperature and 3,675 psig when burning Illinois bituminous coal. The
design reheat steam temperature is 1,005°F at 676 psig. The unit was originally
supplied with flue gas redrculation and gas tempering. The flue gas retirculation
has been removed, but flue gas tempering is operational to control furnace exit flue
gas temperature (FEGT). Tempering flue gas enters the furnace through 12 access
ports, six per front wall and six per rear wall, located just below the furnace exit. The
furnace has four rows of steam wall sootblowers. There are 12 waterlances installed
and spaced evenly between the steam wall blowers on the 2nd and 4th elevations.
King Unit 1 has twelve, 10-ft cyclones with radial burners. The cyclones are
arranged in two elevations of three cyclones each on the front wall and rear wall (12
cyclones total). Combustion air to the cyclones is supplied through an open,
wraparound windbox. Combustion air enters each cyclone as primary air, tertiary
air, and secondary air.
The original fuel supply for this unit was a high-sulfur, Illinois bituminous coal.
Following promulgation of federal regulations to reduce sulfur dioxide emissions,
the fuel supply at King was changed to blends of low-sulfur, western subbituminous
coals and petroleum coke. Currently, King burns a blend of 70 percent Wyoming
subbituminous coal, 20 percent Montana subbituminous coal, and 10 percent
petroleum coke.
Measurement Methods
Measurements of flue gas composition were performed at the boiler exit and within
the primary furnace. At the boiler exit, a multi-point extractive system was used to
obtain flue gas samples for analysis of NOx, CO, CO2, H2S and excess O2. Samples
were extracted from each of two flue gas ducts upstream of the air heater.
Individual sample lines transported the flue gas samples to a mobile emissions
laboratory located at ground level below the flue gas ducts. The gas samples were
conditioned to remove water and directed to emission monitors. Excess O2, CO,
CO2, NOx, and H2S were measured using EPA continuous monitoring procedures
and EPA certified gases. Excess O2 was measured with a Servomex Model 570A
monitor, CO2 with a ACS Model 3300 instrument, CO with a TECO Model 48 and
48H, NOx with a TECO Model 10A chemiluminescence monitor, and H2S with a
Western Research Model 922 SO2/H2S photometric (UV absorption) analyzer.
Hydrogen sulfide measurements were augmented by using EPA Method 11
(Determination of Hydrogen Sulfide Content of Fuel Gas Stream in Petroleum
Refinery) as a quality control check. Opacity was monitored by the plant continuous
-------
emission monitor system (CEMS). Flyash samples were collected by Cegrit flyash
samplers installed in each boiler outlet duct upstream of the air heater, as well as
from collection hoppers in the electrostatic precipitator (ESP).
Water-cooled probes were used to collect flue gas samples from the furnace and to
measure flue gas temperatures. The temperature measurements were primarily
conducted at the furnace exit to monitor any changes in FEGT resulting from
secondary air biasing. The temperatures were made with a high velocity
thermocouple (HVT) consisting of multiple shields and a Type R thermocouple.
King Unit 1 has two optical pyrometers located at the furnace exit plane, one each on
the sidewalls. The HVT temperature measurements were used to check the
reliability and accuracy of the pyrometers early in the test program. Further, furnace
flue gas temperature, internal cyclone temperature, and cyclone and furnace floor
slag temperature measurements were made with a Raytek Raynger 3i portable
optical pyrometer. Measurements of H2S, O2, and CO were performed along the
waterwall tube surfaces in the furnace to determine the degree of furnace reducing
conditions. The primary sampling locations were in the lower furnace along the
furnace floor, outside the cyclone reentrant throats, and along waterwall surfaces
above the upper cyclone row in the middle furnace. The flue gas sampled from each
furnace location was directed to the mobile emission laboratory, conditioned, and
analyzed.
Calculation of Secondary Air Flow Biasing
Secondary air flow biasing was determined by using the coal flow and total air flow
indicators from the boiler control system for each cyclone. The indications of air
and coal flow are used to calculate an air/fuel ratio for each cyclone which is
displayed on monitors in the control room. During preliminary tests of secondary
air biasing, diverted secondary air from the lower cyclones produced over-range air
flow measurements in the secondary air flow transmitters for the upper cyclones.
Subsequently, the secondary air flow transmitters/meters on the upper cyclones
were respanned to accommodate the increased air flow.
The degree of secondary air flow biasing was calculated from each cyclone air flow
meter, as well as theoretical air and excess air values derived from an average fuel
analysis. King Unit 1 operates at approximately 2.5 percent excess O2 as measured at
the boiler outlet This corresponds to about 15 percent excess air (stoichiometry of
1.15). Using data from the cyclone air flow meters, a uniform baseline secondary air
flow to all cyclones results in an air/fuel ratio of 7.83 Ib air/lb coal. The theoretical
(stoichiometry of 1.0) air flow was calculated to be 6.81 Ib air/lb coal. Secondary air
flow biasing reduced the air/fuel ratio to cydone(s) in the lower elevation to
approximately 6.1 Ib air/lb coal, equivalent to a stoichiometry of 0.90.
-------
Results
Baseline NOx emissions
The King test program was conducted in two separate phases: the first test occurred
prior to the Spring 1997 outage and the second test following the Spring outage.
From a combustion perspective, a primary difference in the two periods was the
condition of the furnace walls. During the outage, the walls were cleaned of
sintered coal ash and areas of encrusted flyash accumulations. These deposits act as
insulators and inhibit convection and radiation heat transfer to the furnace walls,
thereby increasing the flue gas temperature in the lower and middle primary
furnace regions and NOx emissions. Figure 1 illustrates the effect of furnace
cleanliness on NOx emissions at 500-MW with 12 cyclones in service and 2.5 percent
excess O2 at the boiler exit. Prior to the spring outage NOx emissions were measured
to be 1.38 Ib/MBtu with a "dirty" furnace. Following the spring outage, NOx
emissions were reduced by approximately 11 percent to 1.23 Ib/MBtu at the same
operating conditions except with a "clean" furnace. Figure 1 also shows that
baseline NOx emissions were 1.38 Ib/MBtu at 550-MW with 12 cyclones in-service,
2.5 percent excess O2, and with a dean furnace. This result indicates that washing
the waterwalls in the furnace permitted load to be increased by 50-MW with no
change in NOx emissions. Because of the significant effect of furnace cleanliness,
NOx emission data in this paper are referenced to clean or dirty furnace wall
conditions.
Effect of Secondary Air Flow Biasing on NOx Emissions
Air flow biasing was accomplished by reducing secondary air flow to the lower
elevation cyclones and redistributing the air to the upper elevation cyclones. Thus,
the cyclones in the lower elevations were operated fuel-rich (stoichiometry < 1.0)
and the cyclones in the upper elevation air-rich (stoichiometry > 1.0). All of the
biasing tests were with a lower cyclone stoichiometry of approximately 0.90 and a
stoichiometry of 1.40 in the upper cyclones.
Figure 2 summarizes the results with all 12 cyclones in service. At 500-MW with a
dirty-furnace, secondary air flow biasing reduced NOx emissions by 17 percent to 1.14
Ib/MBtu. At 500-MW with a dean furnace, secondary air biasing reduced NOx
emissions by approximately 24 percent to 0.94 Ib/MBtu. Further, at 550-MW with a
dean furnace, secondary air flow biasing lowered NOx emissions by 26 percent to
1.02 Ib/MBtu.
-------
1.6
•*••
m
1.4-
1.2-
1.0-
1.38
1.38
ANOx =
1.23
w
I 0.8-
tn
E
x 0.6-|
O
0.4-
0.2-
Dirty Furnace
500-MW
Clean Furnace
500-MW
Clean Furnace
550-MW
Figure 1. Comparison of baseline NOx emissions at King Unit 1 with uniform
air flow to the cyclone burners (i.e., no air biasing) for a dirty boiler (left) and a
clean boiler (center and right). Data were obtained with 12 cyclones in service
and excess O2 of 2.5 percent.
-------
No Sec. Air Biasing
With Sec. Air Biasing
1.23
1.38
Dirty Furnace
500-MW
Clean Furnace
500-MW
Clean Furnace
550-MW
Figure 2. Effect of secondary air biasing on NOx emissions at King Unit 1 with a dirty furnace (left) and
clean furnace (center and right). Data are with 12 cyclones in service.
-------
Secondary Air Flow Biasing with Cyclones Out-of-Service Effect
With uniform secondary air flow to all cyclones, operating with a cydone(s) out-of-
service poses no air flow difficulties in the operating cyclones. However, when
secondary air is biased from the lower cyclones, all of the diverted combustion air
must pass through the upper cyclones to maintain the same overall boiler air flow.
At high boiler loads, the operating cyclones in the top elevations cannot
accommodate all of the diverted combustion air due to air flow meter over-range
conditions. Therefore, the combustion air not accounted for by the operating
cyclones must be passed through the out-of-service cyclones. Tests were performed
with a dirty boiler to determine the effect of the diverted air on NOx emissions.
Figure 3 compares the effects of: (1) secondary air biasing, and (2) biasing with
cyclones out of service. At 500-MW with one cyclone-out-of-service and
approximately 7 percent of the combustion air flow flowing through the out-of-
service cyclone, NOx emissions were 0.97 Ib/MBtu, or 29 percent below baseline
NOx emissions of 1.38 Ib/MBtu, and 15 percent lower than NOx emissions with
biasing alone (1.14 Ib/MBtu). At 500-MW with two cyclones out-of-service and
approximately 13 percent of the combustion air passing through the out-of-service
cyclones, NOx emissions were essentially the same at 0.98 Ib/MBtu. It is not certain
why NOx emissions were not reduced by operation with two cyclones out of service.
It is possible that the actual air flow flowing through the two cyclones was not
consistent with the calculated values, and thus less air was diverted to the cyclones
than expected, or that the air distribution in the furnace was non-uniform and not
conducive to further NOx reductions.
Effects of Biased Secondary Air Flow on Boiler Operations
The use of biased secondary air flows as a viable operating condition to reduce NOx
emissions depends on secondary effects which may effect boiler performance and
operations. King Unit 1 is prone to severe slagging and fouling ash deposition
which can significantly curtail high load generation. Moreover, flyash produced at
King is sold and any degradation of flyash quality may jeopardize this product
stream. Effects on boiler efficiency, cyclone and furnace bottom slag tapping, ESP
operations, and stack opacity also must be considered.
Overall Boiler Operations. No significant changes in boiler operation were
observed during the air flow biasing test periods. Superheat and reheat steam
temperatures were in line with normal operations, and there was no abnormally
high attemperator spray flows. Sootblowing progressed in the normal cleaning
patterns, and there was no increase in opacity.
-------
No Sec. Air Biasing
With Sec. Air Biasing
w/Sec. Air Biasing & 1 Cyc DOS
w/Sec. Air Biasing & 2 Cyc OOS
Figure 3. Effect of secondary air biasing and cyclones out of service on NOx emissions
at King Unit 1. The data were obtained at 500-MW with a dirty furnace and with one
cyclone out of service (third from left) and two cyclones out of service (right) and the
secondary air dampers open.
-------
Furnace Exit Flue Gas Temperatures (FEGT). The slagging and fouling
experienced at King is mainly due to the high concentrations of alkali compounds
in the coal ash, i.e., calcium and sodium, and is strongly dependent on localized flue
gas temperatures. Flue gas temperatures at the furnace exit, reheat section outlet,
and primary superheater outlet are used by the plant to assess boiler conditions and
are adjusted based on historical operating experience. Baseline FEGT's at 500-MW
measured with a water-cooled HVT probe were 1,983°F and 1,990°F on the north and
south sides of the furnace respectively. Following secondary air flow biasing, both
side traverses showed a slight increase in FEGT. For example, the north side
averaged 2,030°F, an increase of approximately 50°F, and the south side showed an
increase of around 25°F with a temperature of 2,013°F. The FEGTs measured were
within current plant operating guidelines. The plant pyrometers located at the
furnace exit also indicated a slight increase in flue gas temperatures. Additional flue
gas temperature measurements by stationary in-situ thermocouples at the reheat
outlet and primary superheater outlet showed no appreciable change in flue gas
temperatures between balanced and biased secondary air flow operations.
Slag Temperature Measurements. Pyrometer temperature readings of the slag
tapping from the lower cyclones and the slag pooling on the furnace floor indicated
that temperature increased when secondary air was biased away from the cyclones in
the lower elevation. The higher temperatures keeps slag viscosity low and slag
flowing with no observable problems. Some "angel hair" slag was visible at the
floor tap holes, but did not present any operating problems.
Flyash Unburned Carbon. Flyash samples were collected from three locations:
(1) Cegrit samples installed at the air heater inlet ducts; (2) ESP inlet hoppers; and (3)
ESP outlet hoppers. The Cegrit samples are an in-situ flyash sample collected from
the centerline of the flue gas duct. A baseline sample (i.e., no biasing) had an
average carbon content of 8.58 percent. With secondary air biasing, carbon-in-ash
increased slightly to 10.8 percent with 12 cyclones in-service, and decreased slightly
to 7 percent with 11 cyclones in-service. Overall, there was no significant change in
unburned carbon. At the ESP inlet section, carbon-in-ash was approximately 7.9
percent for baseline operation and 13 percent with secondary air biasing. Carbon-in-
ash at the ESP outlet section showed no appreciable difference when operating in
the balanced (2.82 percent carbon) or biased (3.10 percent carbon) air flow
configurations. Although the flyash carbon levels increased during the air flow
biasing tests, they were still within an acceptable range for disposal. A summary of
unburned carbon results is shown below:
-------
Test
Condition
(*cyclones I/S)
Base (12*)
Bias (12)
Bias (11)
Bias (10)
Cegrit
AH Inlet
(% carbon)
8.58
10.84
7.03
ESP
AH Outlet
(% carbon)
7.86
12.89
13.00
!arbon-in-Ash (average)
ESP
Outlet
(% carbon)
2.82
3.05
3.16
Effect of Secondary Air Flow Biasing on Flue Gas Composition
A primary concern with secondary air flow biasing is the potential for producing
localized reducing flue gas regions and the presence of hydrogen sulfide (H2S).
Water-cooled probes were utilized to extract flue gas from various sections of the
primary furnace. The flue gas samples were directed to the mobile emission test
van, conditioned, and analyzed. Extensive flue gas sampling was performed in the
lower furnace below the lower cyclone reentrant throats. The traverse path was
oriented along the furnace centerline and just below the cyclone centerline.
Additional flue gas sampling was performed above the upper cyclone elevations in
the middle furnace region. In this area, special attention was given to the centerwall
areas along the front waterwall, rear waterwall, and sidewalls.
Figure 4 presents O2, CO, and H2S data from flue gas sample traverses with and
without secondary air biasing at three locations in the furnace: (1) lower furnace, (2)
above the cyclones, and (3) in the middle furnace. The data were collected with all
12 cyclones in service and 2.5 percent excess O2 at the furnace exit.
Lower Furnace Test Results. Figure 4 (left) compares O2, CO, and H2S
measurements in the lower furnace at 550-MW. The data show that by reducing the
secondary air to the lower cyclones to achieve a stoichiometry of 0.9, the average O2
decreased from 1.7 to 1.3 percent, CO increased from 0.6 to 3.6 percent, and H2S
increased from 121 to 252 ppm. Along the probe traverse path, flue gas composition
changed as the probe extended farther into the furnace. With no air biasing, the
lowest O2 measured was 0.3 percent, the highest CO was 1.5 percent, and the highest
H2S concentration was 214 ppm. With secondary air biasing, the lowest O2 observed
was 0.1 percent; the highest CO was 7.6 percent; and the highest H2S concentration
was 818 ppm. These values all occurred in the central areas of the lower furnace,
approximately 7-ft. from the waterwall surface and in the proximity of the cyclone
reentrant throat exit and cyclone exit flue gas wash. Along the waterwalls during
both operating conditions O2 ranged from 2.2 to 3.4 percent, CO varied from 0.1 to
0.6 percent, and the H2S concentrations ranged from 11 to 40 ppm.
-------
in
o
'55
w
UJ
O
O
•a
CO
IN
O
w
(0
0)
o
X
111
300
-250
200
O2, %
CO, %
H2S, ppmv
co
m
3
w'
-150 £
•a
100
Lower Furnace
500-MW
Above the Cyclones
550-MW
Middle Furnace
550-MW
Figme 4. Oxygen (O2), carbon monoxide (CO), and hydrogen sulfide (H2S) data measured at three
locations in the furnace at King Unit 1 with and without secondary air biasing. Data were collected
with 12 cyclones in service and 2.5% excess O2 at the furnace exit.
-------
Middle Furnace Test Results. Flue gas sampling traverses were performed at
four different elevations in the middle furnace starting above the upper cyclone
position. In each case, flue gas samples were collected at multiple points along a
short traverse into the furnace, starting as dose to the waterwall as possible to
minimize the effects of outside air infiltration. All of the testing in this furnace
region was performed at a boiler load of 550-MW with all 12 cyclones in-service and
with 2.5 percent excess O2. With secondary air biasing, Figure 4 shows that in the
furnace wall area immediately above the upper cyclone elevation the average O2
increased from 1.55 to 2.5 percent, CO decreased from 0.15 to 0.05 percent (1,500 to 500
ppm), and H2S decreased from 20 ppm to <20 ppm.
Flue gas sampling at a higher furnace wall elevation revealed a similar pattern in
O2, CO, and H2S concentrations as observed at the lower elevations. Without air
biasing, the average O2 was measured at 2.1 percent, average CO was 500 ppm, and
the H2S concentration was 19 ppm. When secondary air flows were biased, the
average O2 decreased slightly to 1.9 percent, CO decreased to approximately 300 ppm,
and H2S levels were <20 ppm.
Summary
A test program was conducted at the King Unit 1 to evaluate the feasibility of
utilizing air flow biasing as a cost-effective method to reduce NOx emissions on
boilers with cyclone burners. The main concerns prior to the testing involved the
degree of NOx emission reduction achievable, the amount of cyclone air biasing
required, the potential of waterwall tube wastage due to the production of H2S and
reducing combustion conditions, and adverse effects on boiler operations and
combustion, ash deposition, and opacity.
Two conditions were evaluated during the program. One testing period was
conducted with a dirty furnace (pre outage) and the second was with a dean furnace
(post outage). At 500-MW with a dirty furnace and balanced combustion air flow
(i.e., no biasing), NOx emissions were 1.38 Ib/MBtu. With a dean furnace without
biasing, NOx emissions were 11 percent less at 1.23 Ib/MBru. Biasing the secondary
air to decrease the stoichiometry of the cydones in the lower elevations to 0.90 with
a dirty furnace reduced NOx emissions by 17 percent to 1.14 Ib/MBtu. For a clean
furnace, biasing reduced NOx emissions by 24 percent to 0.94 Ib/MBtu. The
difference in NOx emissions for a dean and dirty furnace is significant and can be
explained by increased heat transfer to the dean furnace walls, which lowers the gas
temperature in the primary combustion zone. This would indicate that there is a
thermal NOx component that takes place in the furnace outside of the cydone
combustion environment.
Data collected during the test program indicated that operating with secondary air
biased from the lower cyclones presents no significant change in boiler operations.
-------
Cyclone operations, evaluated by slag tapping characteristics, were not adversely
affected in either the lower cyclones with reduced excess air, or the upper cyclones
with higher excess air. Combustion conditions in both cyclone elevations, as
evaluated by plant operations, appeared normal in regards to coal ignition and final
coal combustion. Over the short-term of the testing period, there were no adverse
effects on furnace and convection pass ash deposition.
Flue gas sampling from the furnace waterwall regions confirmed low
concentrations of H2S. As expected, the highest H2S and CO concentrations, and
lowest O2 levels, were measured in the lower furnace immediately outside the fuel
rich cyclone exits. Without secondary air biasing, the lowest O2 measured was 0.3
percent, and the highest CO and H2S were 1.5 percent and 214 ppm, respectively.
With secondary air biasing, the lowest O2 measured was 0.1 percent, and the highest
CO and H2S were 7.6 percent and 818 ppm, respectively. In the furnace region above
the top cyclone elevation without biasing, the average O2 was 1.55 percent, the
average CO was 1,500 ppm, and the average H2S was 20 ppm. With biasing, the
average O2 increased to 2.5 percent, the average CO decreased to 500 ppm, and the
average H2S was < 20 ppm. The H2S levels measured along the furnace walls below
the lower cyclone elevation and above the upper cyclone elevation do not pose a
significant concern for increase tube wall wastage. The environment that is present
in these regions is not highly reducing, having O2 concentrations greater than 1
percent and CO levels lower than 1 percent. The low H2S levels were primarily due
the low overall sulfur and pyritic sulfur levels in the coal. At these conditions, a
moderate tube metal loss less than 10 mils/yr would be expected.
Conclusions
The test program at King Unit 1 showed that significant reductions in NOx
emissions can be realized with secondary air biasing air. There were no detrimental
operating performance effects measured or observed. Based on the promising test
results, NSP is planning to operate King Unit 1 with secondary air flow biased
conditions for a period of six-mqnths starting around September 1997. During this
test period, long term effects on cyclone operation, furnace ash deposition,
convection pass ash deposition, flyash loss on ignition (LOI), boiler operations and
performance, ESP operation, and opacity trends will be monitored. Following this
operating period, Unit 1 will be inspected and boiler wall UT tube thickness
measurements will be performed.
References
1. Himes, Richard, "Short Term NOx Emission Reduction with Combustion
Modification on Low to Medium Sulfur Coal-Fired Cyclone Boilers."
-------
CYCLONE BOILER AIR STAGING DEMONSTRATION PROJECT
SIOUX UNIT 2
Richard C. Smith, P.E.
David E. Boll
Union Electric Company
St. Louis, Missouri
Bradley R Adams, Ph.D.
Reaction Engineering International
Salt Lake City, Utah
Abstract
EPRI's Cyclone NOX Control Interest Group (CNCIG) has been investigating a variety of issues
related to finding low cost, innovative methods to lower NOX emissions from cyclone-fired
boilers. Alternative reduction techniques were desired in order to avoid high capital, O&M, and
fuel costs associated with traditional technologies, such as coal or gas reburning, selective
noncatalytic reduction, and selective catalytic reduction. CNCIG focused on exploring concepts
pertaining to cyclone combustion modifications that have the potential to lower emissions to near
the 0.86 Ib/MMBtu limit proposed under Title IV of the Clean Air Act Amendments of 1990.
Computational fluid dynamics (CFD) modeling and corrosion studies of cyclone and furnace
combustion processes indicated air staging had the potential to lower NOX to the specified limit.
Modeling indicated NO, emissions could be lowered below the 0.86 limit through a combination
of low stoichiometry cyclone operation and injection of secondary air above the top row of
cyclones on a two-level, opposed-fired cyclone boiler, currently burning a blend of Powder River
Basin and Illinois coals.
During April/May, 1997, Union Electric Company installed a temporary overfire air system
(OFA) on Sioux Unit 2. The Sioux 2 boiler has ten cyclones, opposed-fired, arranged in two
rows on the front and rear walls. The system was designed to inject combustion air into the
furnace through existing gas recirculation ports located above the top row of cyclones. The
demonstration was intended to validate predictions of NO* emissions reductions and to
demonstrate short-term operability with the OFA system.
This paper presents background CFD model results and design of the temporary OFA system.
Testing is planned for July 1997 with short-term performance information becoming available in
August, 1997. Preliminary data collected in July show reductions in NO* of 50% from baseline
are feasible and repeatable.
-------
Background
Union Electric Company's Sioux Power Plant consists of two 500 MWe generating units located
on the west bank of the Mississippi River in St. Charles County, approximately 20 miles north of
St. Louis, Missouri. Unit 1 was commissioned in May-1967, and Unit 2 was commissioned in
May-1968.
Turbine-generators on both units were supplied by General Electric, and were designed for 3500
psig 1000°F throttle steam, with 1000°F reheat. Condenser cooling is once through with water
taken from the Mississippi River. The steam generators are supercritical, universal pressure, once
through, cyclone-fired with balanced draft (originally supplied as pressurized steam generators).
The steam generators were provided by Babcock & Wilcox as UP-19 and UP-20. Rated steam
flow is 3,290,000 Ib/hr at 3,625 psig and 1005°F at the superheater outlet with 575 psig and
1005°F at the reheater outlet. The cyclones are 10 feet diameter by 12 feet deep with five each on
the front and rear boiler walls, arranged in two rows. All cyclones are equipped with radial
burners and retractable oil igniters. All cyclones utilize primary, secondary, and tertiary air.
Secondary air is supplied to cyclones from a common windbox. Secondary air flows are
controlled by control and shut off dampers for each cyclone. Primary and tertiary air is provided
to each cyclone through common 24-inch diameter ducts from the windbox. Each steam
generator incorporates a flue gas recirculation system for steam temperature control. The lower
gas recirculation vestibules are equipped with ten ports with five each on front and rear walls.
See Figure 1 for the arrangement of the steam generators.
The steam generators were designed for Illinois No. 6 high sulfur coal, but both Sioux units are
currently fueled with a blend of 70% Powder River Basin (PRB) and 30% Illinois No. 6. The
steam generators can typically achieve net outputs of roughly 430-440 MWe while burning the
blend, but new fine grind crushers are expected to improve the ratio to 80 to 85% PRB. High
percentages of Illinois coal can be burned in the event full unit capability is needed to meet system
demands. A typical fuel analysis for the 70% PRB blend is provided below:
Table 1
Analysis of Fuel Blend
HHV 9,325 Btu/lb
Moisture 24%
Ash 6.2%
Sulfur 1.13%
SO2 2.4 Ib/MMBtu
Basis For The Air Staging Demonstration
Previously, Sioux Unit 2 was selected for a trial demonstration of the POWERMAX sequential
process optimization (SPO) method as part of the EPRICNCIG work. Computational fluid
dynamics (CFD) modeling of the cyclones indicated that biasing or staging cyclones row to row
could achieve substantial reductions in NOX levels without impacting cyclone operability or
Page 2
-------
corrosion while burning specific western fuel blends. POWERMAX was used to aid operation of
the boiler with the lower row of cyclones substoichiometric while the upper row had excess air.
The short-term trial indicated reductions in NOX of about 25% could be achieved at nominal unit
output of 440 MW.
CFD modeling of the cyclone boiler furnace along with the cyclone barrel models indicated that a
significant amount of NOX was being produced in the furnace. Modeling results estimated that the
Sioux cyclones produced 60 to 70% of NOX measured in the stack with 30 to 40% being formed
in the furnace. The cyclone biasing trial combined with the results from these models showed that
measures employed in the furnace could have beneficial effects on NOX. Coupling results of the
barrel and furnace models showed the potential to achieve the greatest NOX reductions was by
lowering cyclone barrel stoichiometry and injecting the balance of the secondary air into the
furnace as overfire air (OFA).
Furnace Model Background
The computer model used to simulate combustion in the Sioux boiler was developed by Reaction
Engineering International and was used to model a series of furnace operating conditions as part
of the EPRICNCIG program. This furnace model was partnered with a model for the coal-fired
cyclone barrel in order to predict combustion in the Sioux Unit 2 boiler under staged-air
conditions. The model included mathematical descriptions for the controlling physics and
chemistry in combustion, including turbulent gas flow, radiative and convective heat transfer,
equilibrium gas species reactions, and finite-rate pollutant formation. The mathematical equations
were applied over a computational grid or mesh, which represented the furnace geometry, in
order to calculate approximately sixty different variables used to describe the combustion process.
The model included coupling between the mathematical descriptions to more accurately represent
the physical interdependences in actual combustion processes. The refinement or resolution of
the computational mesh determines the accuracy to which the simulated physics are resolved. The
furnace model discussed here included more than 500,000 computational points in order to
resolve furnace geometry, flow patterns, temperature profiles, and species concentrations.
Modeling Approach
Computational results from a single cyclone barrel model run at a stoichiometry of 0.90 were used
as inputs for each of the ten cyclone inlets in the Sioux furnace. The mapping of results from the
cyclone exit to the furnace inlets reproduced the flow velocities, temperature, and species
concentration profiles while accounting for correct rotation of each cyclone. Each cyclone was
assumed to be burning the 70/30 blend of PRB and Illinois coal at a firing rate of 413 MMBtu/hr.
OFA was added to the furnace model through flue gas recirculation (FOR) ports located above
the cyclones to bring total furnace stoichiometry to 1.15.
The major assumptions made during modeling of the furnace were: (1) All particles burned out
before entering the furnace. Any volatiles not released in the barrel were distributed in the
combustion products at the cyclone exit according to the off-gas profiles from the barrel model.
(2) The furnace had nonuniform constant temperature walls which varied from the maximum slag
Page 3
-------
temperature in the lower furnace to near water wall temperatures in the upper furnace. (3)
Cyclone flow (velocities, temperature, species concentrations) entering the furnace was identical
at all cyclone exits except for differences due to cyclone rotations. (4) Reductions of NO*
concentrations through rebuming of CO were allowed only at fuel-rich locations in the furnace.
(5) All sulfur chemistry and resulting SO2 and H2S concentrations were based on equilibrium
chemistry.
Results from the OFA modeling simulation were compared with predictions from a baseline
furnace simulation without OFA. In the baseline, all cyclones were modeled at a stoichiometry of
1.15.
Modeling Results and Discussion
Figure 2 compares predicted NO* levels at the center of the furnace for the baseline and OFA
cases. The baseline case shows the higher NOX concentrations observed during normal boiler
operations whereas the OFA case shows significantly reduced NOX. The primary reasons for this
dramatic reduction in NO* were reduced production of thermal NO due to lower temperatures in
the bottom of the furnace and rebuming of NO* in the fuel-rich regions of the furnace.
Figure 3 compares predicted CO levels at the center of the furnace for the baseline and OFA
cases. The OFA case has much higher CO concentrations throughout the furnace due to the
lower part of the furnace being at a stoichiometry of 0.90, i.e., fuel-rich. Only after the air is
added to the furnace through the OFA system do the CO levels decrease.
Figure 4 plots axial velocity contours at the OFA injection height indicating the nonuniform nature
of the flow moving up the furnace. It is this nonuniformity of flow that contributes to the
challenge of designing and implementing an effective OFA system that can provide sufficient
mixing and coverage of overfire air with the fuel-rich combustion products moving up in the
furnace.
Figure 5 shows CO concentrations for the OFA case at several heights in the furnace, ranging
from the OFA ports to the gas tempering ports to a height just below the turn into the convection
pass. The plots show that CO concentrations are reduced in the center of the furnace but remain
relatively high along the furnace walls. As the graphs indicate, there are two high CO pockets,
one each along the front and rear walls. These pockets are created from the flow patterns lower
in the furnace around the flue gas recirculation ports. Basically, the OFA jets do not reach all the
way across the furnace before being turned upward by combustion products leaving the lower
furnace, and thus they are unable to mix with CO near the walls.
Corrosion
Corrosion issues were studied in conjunction with the CFD model work. Model outputs were
used by EPRI to examine formation of corrosive species such as IbS in the lower and upper
furnace. The principal conclusion was that corrosion potential in the cyclones and lower furnace
will be less with the 70% PRB blend than the potential for corrosion with design Illinois coal.
Page 4
-------
The corrosion study concluded that flamespray coatings employed above the stud line and
maintenance of refractory in cyclones and the lower furnace should be adequate to protect against
severe corrosion due to substoichiometric cyclone operation with OF A.
OFA System Design
The purpose of the OFA demonstration was twofold: (1) attempt to validate predictions of NO*
emissions reduction from the CFD modeling effort; and to (2) demonstrate short-term operability
with the OFA system. The design of the OFA was kept as simple as possible to minimize impact
on the Unit 2 planned overhaul outage schedule and to minimize costs.
The OFA air system was designed to stage approximately 20-25% of the normal unit air flow at
16% excess air. The fixed nozzles were initially sized based on previous EPRI research
conducted on front-fired wall units; that research recommended a ratio of the overfire air velocity
to the vertical component of furnace gas flow of 4:1 to 6:1 for adequate penetration (refer to
EPRI TR-102906). Also, one of the early design constraints was not to perform any pressure
part modifications. This resulted in having the overfire air injected into the furnace at angles other
than normal to the furnace walls. Based on the furnace modeling of the preliminary OFA
schemes, it was decided that modification to the FGR port tubes should be made to allow the
overfire air to be injected normal to the walls for better coverage, and that higher velocities would
be required to penetrate the furnace combustion gases. The modeling also confirmed that the
location of the existing FGR ports, both in plan and elevation, provided good air distribution and
mixing as well as sufficient residence time within the furnace. The final nozzle size was fixed to
produce velocities approximately 300 ft/see.
Figure 6 shows the final OFA system configuration as installed. Ten (10) each take-off ducts
were installed in the top of the main windbox, five each on front and rear. The locations of the
take-offs were selected to match the FGR ports as closely as possible while avoiding interferences
with structural steel components. The 36-inch diameter ducts rise vertically and turn 90° to
horizontal to penetrate the vertical walls of the flue gas recirculation vestibule. The ducts then
run through the back of the FGR ports to inject secondary air into the furnace. Limited air flow
instrumentation was provided. Each duct contains a manually operated butterfly damper to
provide shut-offTisolation as well as a limited amount of air flow control. All duct sections that
could be subject to furnace thermal radiation were fabricated from stainless steel. Puff blower
piping and GR port tubing interferences were encountered during installation of the new ducts.
Because the installed OFA system design differed from the scheme included in the CFD furnace
modeling, it has been recommended that the model be modified to replicate the final
configuration. This would allow more direct comparison of the installed OFA system operating
information to model results, and thus increase confidence in future model predictions.
The total installed cost of the OFA system was approximately $400,000. The cost breakdown is
shown below. It should be noted that a system designed for permanent installation and
environmental compliance purposes would likely be significantly more expensive, and costs will
vary based on boiler type, e.g., opposed fired or single wall fired, single row, etc., and other site
PageS
-------
specific factors. Design features and other considerations necessary for long-term continuous
operation of the OFA were not included in the scope of this project, such as automation of the
OFA air flow dampers, flamespray of tubes above the furnace stud line, etc.
Table 2
OFA Demonstration System Cost Estimate
Materials $130,000
Labor 140,000
Engineering 60,000
Testing 70.000
Total $400,000
Initial Operating Results
At the time this paper was written, testing with the OFA system was just beginning. NOX, CO,
and O2 were measured with a 12-point grid installed in the economizer outlet ductwork. Flyash
samples were taken from precipitator hoppers and tested for unburned carbon. Rather than
adjusting the manual dampers during load reductions, the furnace oxygen control was changed to
increase the O2 setting to compensate for the additional air flow.
Preliminary operation indicated that reductions in NOX of approximately 50% from baseline
emissions were feasible and repeatable. Table 3 summarizes typical data obtained early in the trial
operation period. Unit output was limited to 420 MWe due to induced draft fan vibrations even
though 430 to 440 MW are achievable with the current fuel blend.
Table 3
Performance Indicators — Preliminary Operation
Baseline Staged Data
Unit Load, MW 420 420
NOX, ppm 990 492
SO2, ppm 1060 1092
CO, ppm 26 29
02, % 3.0 3.0
This preliminary data suggest a reduction in NOX emissions at the stack from a baseline of 1.22
Ib/MMBtu to approximately 0.55 Ib/MMBtu. Estimates of cyclone stoichiometries and other
impacts on unit performance parameters were not available at the time this paper was written.
To further assess the potential of corrosion caused by operation of the unit under these staged
conditions, furnace gas sampling and analysis of total reduced sulfurs (TRS), including the H2S
species, and CO will be performed by a testing contractor. Samples will be taken from the lower
furnace and at the elevation of the OFA nozzles. Another contractor will be collecting molten
-------
slag samples from the furnace floor just above the slag tank for analysis. The results of this
sampling are not known at the time of this writing.
Conclusions
Preliminary operation of the OF A system appears to support the validity of the CFD modeling
work and predictions of NO* reductions. Additional testing is necessary to confirm these findings.
The furnace model may be modified in the future with the final OFA system design to further
increase confidence in future model predictions.
Long-term operation and testing for corrosion of water wall tubes was not addressed by this
project.
Acknowledgments
The authors would like to acknowledge the efforts of Ken Stuckmeyer, Keith Stuckmeyer, Bob
Fairbanks, and the Sioux Plant staff in making this demonstration a success. The work described
in this paper was conducted under the EPRI Cyclone NO* Control Interest Group program.
Funding was provided by EPRI and Union Electric Company.
References
1. Adams B.R. et al., "Application of Computer Models to Coal-Fired Power Plants," presented
at the EPRI Fifth International Conference of Effects of Coal Quality on Power Plants, Kansas
City, Missouri, May, 1997.
2. Bakker Wate T., "Staged Combustion in Cyclone Boilers Evaluation of Corrosion Potential -
Furnace," confidential report to the EPRI Cyclone NOX Control Interest Group, 1997.
3. Eskinazi D., "Retrofit NOX Controls for Coal-Fired Utility Boilers," EPRI Technical Report
No. TR-102906, September, 1993.
4. Lisauskas R.A., McHale C.E., Afonso R., Eskinazi D., "Development of Overfire Air Design
Guidelines for Front-Fired Boilers," presented at the EPRI 1987 Symposium on Stationary
Combustion NOX Control, EPRI Report CS-5361, Volume 1, August, 1987.
5. Smith R.C., Stuckmeyer K.B., and Melland C., "Combustion-Based Modifications and
Optimization Allow Substantial NOX Reduction on Cyclone Boilers at Drastically Reduced
Cost," EPRI Innovators, No. IN-107206, March, 1997.
6. Stuckmeyer K.B., Adams B.R., Heap M., and Smith P., "Computer Modeling of a Cyclone
Barrel," presented at the EPRI Workshop on NOX Controls for Utility Boilers, Cincinnati,
Ohio, August, 1996.
7. Stuckmeyer K.B. and Boyle R.J., "Optimization of a Cyclone Boiler for NO* Control,"
presented at the EPRI Workshop on NOX Controls for Utility Boilers, Cincinnati, Ohio,
August, 1996.
Page 7
-------
OVBffKE -
AIR DUCTS
-c
-c
(WUTHW-01/T)
J.
GAS TEMPERING
PORTS (3 ON FRONT WALL
AND $ ON REAR WALL)
- GAS RECIRCULATON
PORTS (5 ON EACH PROMT
AND REAR WALL)
' CYCLONES (10)
Figure 1. Sioux Steam Generator Arrangement with OFA System
PageS
-------
800 ppm
800 ppm
500 ppm
600 ppm
700 ppm
500 ppm 800 ppm
600 ppm
400 ppm
500 ppm
800 ppm
400 ppm
Figure 2. CFD Model Results
NOX Concentration Profiles for the Baseline (left) and OFA (right) Cases.
Page 9
-------
lOppm
500 ppm '
1000 ppm -
1500 ppm -
2000 ppm -
2000ppm -
10 ppm
Figure 3. CFD Model Results
CO Concentration Profiles for the Baseline (left) and OFA (right) Cases.
Page 10
-------
24m/s
18m/s
Om/s
24 m/s
18 m/s
6 m/s
18 m/s
6 m/s
Figure 4. CFD Model Results
Axial (upward) Velocity Contours at FGR/OFA Injector Height.
Page 11
-------
>10,000 ppm
10,000 DPJTI
OFA injector height
10.000 ppm •
-2500 ppm \
<2500ppm
L4 ft. above injectorc
27 ft. above injectors
1 ppm
10,000 ppm
2500 ppra
10.000 ppm '
Just bdow GT injectors
Gas tempering injectors
Just below convective pass
Figures. CFD Model Results
CO Concentration Profiles at Various Furnace Heights.
Page 12
-------
GAS
RECIRCULATION
VESTIBULE
Figure 6. OFA System Arrangement - Plan (top) and Elevation (bottom)
Page 13
-------
NOX CONTROL USING NATURAL GAS REBURN ON AN INDUSTRIAL CYCLONE
BOILER
H. Farzan
McDermott Technology, Inc.
(Formerly Babcock & Wilcox, R&D Division)
G. J. Maringo
Babcock & Wilcox Company
C. T. Beard and G. E. Weed
Eastman Kodak Company
John Pratapas
Gas Research Institute
Abstract
Eastman Kodak Company's cyclone boiler (Unit No. 43), located in Rochester, New York, has
been retrofitted with the gas reburn technology developed by the Babcock & Wilcox (B&W)
Company to reduce NOX emissions in order to comply with the New York State regulations
adopted in conformance with the Title I of the Clean Air Act Amendments (CAAA) of 1990. At
the peak load, the ozone nonattainment required NOX reduction from baseline levels necessary to
meet the presumptive limit for cyclone boilers in this regulation is 56%.
Eastman Kodak Company and the Gas Research Institute (GRI) are cosponsoring this project.
Chevron has supplied the natural gas. Equipment installation for the gas reburn system was per-
formed in a September 1995 outage.
Boiler No. 43's maximum continuous rating (MCR) is 550,000 pounds per hour of steam flow
(or approximately equivalent to 60 MWe). Because of the compact boiler design, there is insuffi-
cient furnace residence time to use coal or oil as the reburn fuel, thus making it a prime candidate
for gas rebum. Kodak currently has four cyclone boilers. Based upon successful completion of
this gas reburn project, modification of Kodak's other cyclone boilers to include reburn technol-
ogy is currently being considered.
The paper will describe gas reburn system design, manufacturing, installation, the test results,
and commercial operation of the Kodak's Unit No. 43 boiler with the gas reburn system.
-------
Introduction
The Clean Air Act Amendments of 1990 posed significant challenges to electric utilities to re-
duce both sulfur dioxide (SO2) and oxides of nitrogen (NOX) emissions. The Act mandates an
approximate 3.5 million ton-per-year reduction in SO2 emissions from 111 selected existing util-
ity boilers by January 1, 1995. An additional 5.3 million ton-per-year reduction is also mandated
to occur by January 1. 2000, in order to reach a long-term SO2 emissions cap of 8.9 million tons
per year. Titles I and IV of the Act mandate NOX reduction from stationary sources. Title IV
(acid rain) requires the use of low-NOx burner technology by utilities and Title I (ozone
nonattainment) requires RACT (reasonable, available control technology) to reduce NOxfrom
utility and industry sources. The impact on utilities is that by the year 2000, more than 200,000
MWe of electricity generating capacity, must be retrofitted with low-NOx systems.
The limitations imposed by the act are particularly challenging, especially for NOX emissions to
cyclone-fired boilers. The cyclone furnace consists of a cyclone burner connected to a horizon-
tal water-cooled cylinder, the cyclone barrel. Air and crushed coal are introduced through the cy-
clone burner into the cyclone barrel. The larger coal particles are thrust out to the barrel walls by
the cyclonic motion of combustion air where they are captured and burned in the molten slag
layer that is formed; the finer particles burn in suspension. The mineral matter melts and exits
the cyclone via a tap at the cyclone throat that leads to a water-filled slag tank. The combustion
gases and remaining ash leave the cyclone and enter the main furnace.
Cyclone-fired boilers represent approximately 26,000 MWe of generating capacity in the U.S.,
which is approximately 15% of pre-New Source Performance Standards (NSPS) coal-fired gen-
erating capacity. These units contribute about 21% of NOX emitted by pre-NSPS coal-fired units.
Typical low-NOx burners and staged combustion techniques are not applicable to cyclones be-
cause these techniques rely on developing an oxygen deficient or reducing atmosphere to hamper
NOX formation. A reducing condition in the confines of a cyclone barrel is unacceptable due to
the potential for tube corrosion and severe maintenance problems which could result. Cyclone
operation must occur with excess oxygen in the cyclone barrel, and this condition coupled with
high temperatures and severe turbulence within the cyclone barrel is the reason why cyclones are
disproportionately high generators of NOX.
The reburning technology offers cyclone boiler owners a promising alternative to expensive flue
gas cleanup techniques for NOV emission reduction. Reburning involves the injection of a
supplemental fuel (natural gas, oil, or coal) into the main furnace in order to produce locally re-
ducing conditions that convert NOX produced in the main combustion zone to molecular nitro-
gen, thereby reducing overall NOX emissions.
Eastman Kodak Company has four coal-burning cyclone boilers at their Rochester, NY facilities.
These boilers are subject to Title I compliance (ozone nonattainment). To comply with New
York State requirements, Kodak requested that B&W perform an engineering feasibility study to
determine the best Reasonable Available Control Technology (RACT). B&W concluded that
because of the compact boiler design, gas reburn technology is the most viable approach for
these boilers. Because Kodak's boilers are relatively small, the average furnace residence time is
-------
less than that available in larger cyclone units. Thus, coal or oil rebum technologies could result
in major impact on unbumed combustibles, fouling of convective pass tubes, and precipitator
problems in Kodak's boilers. The final step to determine RACT is to demonstrate the gas reburn
technology in one of these boilers (No. 43 Boiler) in order to determine specific NOX reduction
potential and cost of NOX control for boilers designed with minimum available furnace resident
times.
The uncontrolled NOX level in No. 43 Boiler is 1.37 Ib/MBtu at peak load. The required NOX re-
duction from this baseline level necessary to meet the presumptive limit set by the New York
State regulation, is about 56%. Based on successful completion of this gas reburn project, modi-
fication of the other cyclone boilers with reburn technology is anticipated.
Background / Reburning Process Description
To address the special needs of the cyclone boiler population with respect to NOX reduction,
B&W pursued the rebuming technology. The B&W reburn technology development for cyclone
boilers was performed via: 1) an initial engineering feasibility study (funded by EPRI Project
RP-1402-30), 2) a pilot-scale evaluation co-funded by Electric Power Research Institute (EPRI
RP-2154-11) and the Gas Research Institute (GRI 5087-254-1471), and B&W, and 3) a U.S. De-
partment of Energy's Innovative Clean Coal Technology demonstration at Wisconsin Power and
Light's Nelson Dewey station.1'4
The feasibility study suggested that the majority of cyclone- equipped boilers could potentially
apply this technology in order to reduce NOX emission levels by as much as 50-70%.1 The major
criterion that substantiated this potential was that of sufficient furnace residence within these
boilers, allowing application of the technology. This residence time is required for both the NOX
reduction process in the reburn zone and subsequent combustion completion in the burnout zone
to occur within the boiler. Based upon this conclusion, the next level of confirmation, pilot-scale
evaluation, was justified. The pilot-scale tests evaluated the potential of natural gas, oil, and coal
as the reburning fuel in reducing NOX emissions.2 The pilot-scale data confirmed the results of
the feasibility study and showed that reburning is technically feasible and a potentially commer-
cially viable technology for cyclone boiler owners. Coal was then selected as the reburn fuel to
be used during the Clean Coal Project at Wisconsin Power & Light (WP&L) Company's Nelson
Dewey station. The WP&L reburn demonstration validated the results of both the engineering
feasibility and pilot-scale studies.3'4
Reburning is a process by which NOX produced in the cyclone is reduced (decomposed to mo-
lecular nitrogen) in the main furnace by injection of a secondary fuel. The secondary (or
reburning) fuel creates an oxygen-deficient (reducing) region that accomplishes decomposition of
the NOX. Since reburning is applied while the cyclone operates under normal oxidizing condi-
tions, its effects on cyclone performance can be minimized.
The reburning process employs multiple combustion zones in the furnace, defined as the main
combustion, reburn, and burnout zones as shown in Figure 1. The main combustion zone is op-
erated at a stoichiometry of 1.1 (10% excess air) and combusts the majority of the fuel input (65
to 85% heat input). The balance of fuel (15 to 35%) is introduced above the main combustion
-------
OVERFIRE
AIR PORTS
REBURNING
BURNER
CYCLONES
BURNOUT
ZONE
REBURN
ZONE
NO + NH,
MAIN
COMBUSTION
ZONE
3 - 4% EXCESS 02
0.85 - 0.95
ST01CHIOMETRY
zone (cyclones) in the reburn zone
through reburning burners. These
burners are operated in a similar
fashion to a standard wall-fired
burner, except that they are fired at
extremely low stoichiometries. The
oxygen-deficient combustion gases
from the rebum burners mix with
combustion products from the cy-
clones to obtain a furnace reburning
zone stoichiometry in the range of
0.85 to 0.95, which is needed to
achieve maximum NOX reduction
based on laboratory pilot-scale re-
sults. A sufficient furnace residence
time within the reburn zone is re-
quired for flue gas mixing and NOX
reduction kinetics to occur.
The balance of the required combus-
tion air (totaling 15 to 25% excess air
at the economizer outlet) is intro-
duced through over-fire air (OFA)
ports. B&W's Dual Air Zone Ports are designed with adjustable air velocity controls to enable
optimization of mixing for complete fuel burnout prior to exiting the furnace. As with the reburn
zone, a satisfactory residence time within this burnout zone is required for complete combustion.
Project Description
Project Objectives
The objective of this project at Eastman Kodak Company's No. 43 Boiler was to demonstrate the
long-term application of gas reburning to reduce NOX/SO9 emissions from a coal-fired cyclone
boiler, while maintaining acceptable cyclone boiler operating conditions.
Specific goals of this demonstration project were as follows:
• To maximize NOX emission reduction at peak load. The project goal for NOX emission
is 0.6 Ib/MBtu. while using 28% (or less) natural gas as a percentage of total heat input
to the boiler. This corresponds to a 56% NOX reduction from the baseline level of 1.37
Ib/MBtu.
Figure 1 Reburning process
• To demonstrate boiler operational safety and acceptable turndown with reburn. The
turndown at the baseline conditions (no rebum) for No. 43 Boiler is 55% of maximum
continuous rating (MCR). At this load, boiler slag tap freezing occurs. With the intro-
duction of reburn fuel into the boiler, the boiler turndown needed to be evaluated.
-------
• Minimal impact on combustible losses (less than 0.1 percentage point change in com-
bustion efficiency).
• CO levels of less than or equal to 200 ppm.
• No major impact on boiler tube losses. No. 43 Boiler has not experienced major tube
losses within the main boiler in the past (no reburn). Rebuming is not expected to in-
crease the tube deterioration. The quantitative objective with rebum in service is to
have no adverse impact on current expectations to continue No. 43 Boiler operation
over the next 20 years.
• Obtain NOX removal at a cost not to exceed $3000 per ton of NOX removed.
• The capital cost of the reburn system should not exceed an installed cost of $75/KW
for small cyclone boilers. The capital cost is high due to the small size of the boiler
(economy of scale is not available here, e.g. at 100MWe the cost is $16 to $17 perKW
without controls). Although the cost of instrumentation and controls is site-specific, it
is included in this estimate.
Project Methodology
B&W's methodology for designing and operation of a reburn system at Eastman Kodak Com-
pany is identical to that used on the previous full-scale reburn application^3). This includes using
the previously acquired pilot-scale gas reburn data^2', reviewing past baseline characterization of
No. 43 Boiler, a site-specific engineering study including scale-up of the results using proprietary
B&W numerical models validated with baseline information, and finally full-scale design, instal-
lation, and commercial operation1^. Once the system was in operation, a commercial evaluation,
including revised cost information, was be developed. In order to accomplish these objectives,
the gas reburn project consists of 7 tasks as shown in Figure 2.
Task
Description
1994
S
0
N
1 j Finalize Agreement F^^"?
2
2-1
3
4
Engineering Design
1995
D J
i
Numerical Modeling ••••••
Test Plan Completion
Equipment Fabrication & Installation
5 ! Field Testing
5 Data Interpretation
7 Management & Reporting
8
Reburn Operation
F
-h
M
A
M
J
J
A S
1 I
I
i
0
I*
N
D
1996
J
H
F
M
A
•
•••
i
M
j
J A
I
—;
"I
••^•B*. To March 1997*
O Participation Agreement with Kodak t-f Test Plan
• 4-Week Planned Outage for Reburn Installation • Boiler Evaluation
Figure 2 Project schedule
-------
Project Organization
The project team is as follows:
• Eastman Kodak Company - Host site and co-sponsor
• Gas Research Institute - Co-sponsor
• B&W - Boiler manufacturer and prime contractor
• Chevron U.S.A. - Gas supplier
• Rochester Gas & Electric - Gas distributor
• Acurex - Field monitoring
Project Schedule
A schedule of 24 months was planned for this project, as shown in Figure 2. Equipment installa-
tion for the gas reburn system was scheduled for a September 1995 outage. At the conclusion
testing, the rebum system was optimized and delivered to Kodak for day-to-day commercial op-
eration to March 1997. The boiler went through an outage in March 1997 when the long-term
performance of the reburn system and its effect on boiler tube life was assessed. Currently,
Kodak is using the gas reburn technology as RACT for the Boiler No. 43.
Host Boiler Description and Conceptual Design of the Reburn System
Eastman Kodak Company's No. 43 Boiler was purchased from Babcock & Wilcox (B&W) in
1968. The unit is a two (2) drum Stirling Power Boiler designed for a Maximum Continuous
Rating (MCR) of 550,000 Ib/hr steam flow with a four (4) hour peak rating of 605,000 Ib/hr
steam flow. The boiler is designed with two (2) B&W nine-foot-diameter cyclone furnaces
equipped with B&W radial burners. The cyclones are capable of firing either bituminous coal or
heavy fuel oil. Operating steam pressure and temperature at full load are 1425 psig and 900°F,
respectively, at the superheater outlet with a feedwater temperature of 400°F. The unit is also ca-
pable of 450,000 Ib/hr steam flow, while maintaining full-load steam pressures and temperatures
at a feedwater temperature of 238°F. Figure 3 shows the original boiler sectional side view.
B&W's reburning technology involves customizing the design to each specific site application in
order to optimize performance. Depending on the boiler design and capacity, B&W evaluates
the effectiveness of using natural gas, oil, or pulverized coal as the reburning fuel. One of the
key parameters in this determination is defined as the available furnace residence time criteria.
Smaller capacity boilers (less than about 650,000 Ib/hr steam flow) typically have minimal fur-
nace residence time and this dictates the use of natural gas reburning. Cyclone boilers of this
size contain either one or two cyclone furnaces and make up approximately 18% of all cyclone
firing capacity.
Eastman Kodak Company's No. 43 Boiler is one of these uniquely designed cyclone units. As is
typical of the smaller size cyclone units, No. 43 Boiler contains heat transfer surface sections
routed vertically up through the furnace region. These sections include the cyclone riser and
wingwall tubes. This feature not only helps minimize furnace residence time, but it also creates
reburning design problems with respect to space limitations for physically locating reburn system
components and in-furnace mixing obstructions.
-------
GAS OUTLET STEAM COIL
a[R HEflTEP"
Figure 3 Eastman Kodak Company's No. 43 boiler
The major components of the B&W gas reburn system includes new rebum burners, overfire air
(OFA) ports, ducts and flues to transport air and gas recirculation to the new system components,
air monitors and dampers to control the flow rates, a gas recirculation fan, and controls. B&W S-
type burners are used in the rebum system to provide a stable flame and good mixing characteris-
tics. The burner is operated in a similar fashion to standard wall-fired burner applications (e.g.,
-------
includes a standard flame scanner and gas lighter). Although optional, the reburn system at No.
43 Boiler includes gas recirculation to the burner to maintain maximum mixing flexibility within
the reburn system and thus maximum NOX reduction potential.
Identification of the optimum number, size, and location of the rebum burners and OFA ports is a
critical reburn system design issue. Since the burners and OFA ports require boiler pressure part
openings, physical space limitations are a potential constraint. The design methodology of the
gas reburn system is described in detail elsewhere'5).
Gas Reburning Retrofit at Kodak's Boiler No. 43
Retrofit of the gas rebuming system to Boiler No. 43 at Kodak Park consisted of installation of
the following items:
1. Two (2) B&W S-type gas reburn burners. The burners consist of an inner core zone
that houses the natural gas spuds and an outer air zone that contains adjustable spin
vanes. The core zone includes a manual sliding disk to control flow to this region. In
addition to housing the manually adjustable spin vanes, the outer air zone includes the
retractable B&W CFS gas lighter, the scanner sighting ports, and an observation port.
The lighters contain a high-energy ignition probe and air cylinders for retracting pur-
poses. The lighters are remotely operated by the Burner Management System or can be
operated locally.
2. Four (4) OFA ports to introduce the balance of air flow for complete combustion.
B&W's Dual Air Zone OFA ports were selected in order to control mixing capabilities
from both a penetration and side-to-side mixing standpoint. The OFA ports contain
two zones, an inner zone for penetration and an outer air zone with manually adjustable
spin vanes for side-to-side mixing capability.
3. A mixture of secondary air and gas recirculation is introduced to the individual burner
windbox. The air flow source is from the airheater outlet and is controlled/measured
via an automatic control damper and air flow monitor. The gas recirculation source is
from the economizer outlet and a booster fan is available to provide adequate condi-
tions to mix the secondary air with the gas recirculation. Isolation dampers and a con-
trol damper are available around the fan to control flow, in addition to allowing fan
maintenance to be performed while the boiler is operating. The air and gas recircula-
tion flows were optimized during start-up activities and control curves for each of the
parameters were incorporated into the control system.
4. Natural gas line.
5. Upgrade of the control system.
6. Various flues, ducts, flow control dampers, and monitors.
-------
The isometric view of the system shown in Figure 4 gives the special relationships of each of the
components in the system. Integration of the rebum system with the existing plant consisted of
interfaces with the air heater outlet, flue gas recirculation system, hot air recirculation, penetra-
tion into the boiler, and the control system. The rebum equipment was fabricated at B&W's fa-
cilities and delivered to Kodak for installation. Prior to the boiler outage, all aspects of the
system's erection that did not require a boiler outage were completed. This included installation
of the FGR fan and all flues and ducts up to tie-in points. The remainder of construction, includ-
ing burner and OFA installation, was completed in a four-week outage beginning September 11,
1995. In addition, Kodak had other boiler maintenance activities planned (e.g., rear-wall tube
panels) and the overall outage lasted 9 1/2 weeks.
Results
The focus of this project's testing was to determine the maximum NOX reduction capabilities
while minimizing the gas heat input without adversely impacting plant performance, operation,
and maintenance. In particular, the evaluation was designed to confirm and expand upon the re-
sults of the SBS pilot test program®.
Test Plan Variables
Numerous variables are associated with the reburn system and a day-to-day test matrix was set
up to proceed from one parameter to another during parametric optimization testing. The test
variables included:
• Percent of boiler heat input by natural gas
• Reburn burner stoichiometry
• Reburn zone stoichiometry
• Reburn burner spin vanes adjustments
• OFA port spin vanes/sliding disk adjustment
• Boiler load
• Gas recirculation rates to the reburn burners
Information collected to evaluate performance of the technology are as follows:
• Impact on NOX and CO emission levels by the itemized test variables
• Boiler temperature and cleanliness factor profiles
• Unburned combustibles loss
• Boiler thermal efficiency
• Operational experience
NOX Emissions
Operation of the reburning system reduced NOX emission levels by 35 to 71% from baseline con-
ditions. Figure 5 shows the NOX emission levels versus percent heat input from natural gas intro-
duced through reburn burners at the peak load (605,000 Ib/hr steam flow). NOX is reduced from
a preretrofit baseline value of 1.37 Ib/MBtu to 0.6 Ib/MBtu while using 20% natural gas. NOX
-------
Tubular airheater
gas outlet
Tubular airheater
air inlet
Steam drum
Flue gas
recirculation
fan
Combustion air
Gas-reburn
burner
Overfire air supply
- Overfire air ports
Gas-reburn
burner
Cyclone coal
burners
Babcock & Wilcox Gas Reburn Project
for
No. 43 Boiler - Building 321
Eastman Kodak Company
Rochester, New York
1995
Figure 4 Isometric view of the reburn system
10
-------
reduction ranged from 35 to 56% while varying 10 to 20% natural gas as rebuming fuel. During
these tests, cyclone stoichiometry is maintained as close to 1.1 (8 to 12% excess air) as possible
to minimize potential cyclone operating concerns.
The maximum NOX reduction of 71% was achieved at 27.5% natural gas heat input. Although
lower NOX emissions of 0.4 Ib/MBtu can be achieved at 27.5% gas heat input, the recommended
maximum heat input is 22%. Long- term operation at higher gas heat input requires additional
testing / optimization prior to final acceptance.
Figure 6 shows the NOX emission levels versus percentage of gas heat input at MCR (550,000 lb/
hr steam flow). In addition, this figure identifies the effectiveness of FOR to the reburn burners
for higher NOX reduction. The normal preretrofit NOX emission level at this load was 1.25 lb/
MBtu. Varying the amount of the natural gas rebum from 10 to 20% of the total heat input de-
creased NOX emissions from 0.82 to 0.56 Ib/MBtu. Corresponding NOX reduction from baseline
levels are 34 to 55%. When FOR was introduced to the reburn burners while utilizing the same
amount of the natural gas rebum fuel (10 to 20%), the NOX emissions were 0.72 to 0.5 Ib/MBtu
(42 to 60% NOX reduction). The improvement in NOX reduction during FGR operation is greater
at the lower natural gas heat input (e.g., 10%) and diminishes at higher gas heat input (e.g., 30%).
This is expected because mixing between natural gas and cyclone combustion gases improves
with increasing natural gas heat input.
Additionally, Figure 7 illustrates the effect of various FGR flow rates versus NOX reduction po-
tential. Increasing the FGR flow from 1 to 6% of total boiler flow resulted in improving NOX
emission levels from 4 to 82 ppm. It should be noted that the reburn system design was opti-
mized based on using FGR via numerical modeling and varying FGR rates obviously changes the
mixing potential within the reburn zone. As with any standard natural gas fired burner applica-
tion, one practical limitation on the quantity of FGR that can be introduced is flame scanner indi-
cations. Based on total system optimization, the optimum FGR flow rate at Kodak's boiler 43
was determined at about 4%.
Boiler Turn Down
Based on all the parametric tests, optimized operating conditions were identified and incorpo-
rated into the control system. The optimized control curves include load versus gas heat input,
reburn burner air and FGR flows, and OFA port flow. Figure 8 identifies the NOX versus load
data after these operating conditions were incorporated into the control system. The figure shows
the effect of load on the required amount of natural gas to assure NOX emissions below a targeted
level of 0.6 Ib/MBtu. Reducing load from 605,000 to 350,000 Ib/hr of steam, requires less natu-
ral gas as a percentage of total heat input; 21 and 14%, respectively. The main reasons for lower
gas requirement are the lower baseline NOX levels and higher rebum zone residence time at
350,000 Ib/hr of steam than at the 605,000 Ib/hr of steam condition. The baseline NOX level de-
creased from 1.37 Ib/MBtu at the peak load to about 0.85 Lb/MBtu at 350,000 Ib/hr of steam.
11
-------
Kodak Boiler 43 Gas Reburning NOX Emission Summary
% Gas Heat Input versus NOX Emission Levels
605,000 Iba/hr Steam Flow w/GR to Reburn System
RECOMMENDED MAXIMUM CONTINUOUS GAS HEAT INPUT
10 15 20 25
% Heat Input from Natural Gas
Figure 5 NOx emissions at peak load with gas reburn
Kodak Boiler 43 Gas Reburning NOX Emission Summary
% Gas Heat Input versus NOX Emission Levels
550,000 Ibs/hr Steam Flow
0.4
0.3
0.2
—A— Acurex Test Data w/GR
--- a • - Acurex Test Data wo/GR
: RECOMMENDED MAXIMUM CONTINUOUS GAS HEAT INPUT = 22%
10 15 20 25
% Heat Input from Natural Gas
30
Figure 6 NO^ emissions at MCR with gas reburn
12
-------
Kodak Boiler 43 Gas Reburning NOx Emission Summary
% Gas Recirculation to Burner vs. Change in NOx Emission Levels
550,000 Ibs/hr Steam Flow
P nn r !
Q_ OU '-
-& P :
i t :
> bu r
2 E
0. crn 1-
— r '
r- "~
"o r
•" °n ~ :
5 p
* on -
i ^°E i x
10 " ' jf
= jK
(
• X
X".
• •
•
-X
x •
0123
% Gas Recirculation to
•
/
•X
•
•
• 1 6.4 -16.6% Natural gas |-
• 20.8 - 20.9% Natural gas [
4567
Rebum Burners
Figure 7 Effect of FGR on NO^ emissions
Kodak Boiler 43 Gas Reburning NOX Emission Summary
Boiler Load (Steam Flow) versus NOX Emission Levels
Optimized Control Conditions
0 0
3
g- 0.8
? 0.7
g
8.0.6
en
•o
c
O r\ K
o. 0-5
&
•z.
0.4
0 3
— • — Acure
; 14% Gas 20% Gas
* ^^__19%Gas--^- '^"^*^>~*,
: RECOMMENDED MAXIMUM CONTINUOUS GAS HEAT INPUT
, , i . . t
300 350 400 450 500 550
Boiler Steam Flow (1000's Ibs/hr)
< Test Data
21 % Gas
"-«
= 22%
i i
600 650
Figure 8 NO emissions at optimum conditions with gas reburn
13
-------
Unbumed Combustibles Losses
The effect of the gas reburn technology on unburned combustibles was explored by measuring
carbon content of the fly ash as well as CO emission level. Typical average CO emission levels
can be maintained below 200 ppm. Higher CO emissions have been observed when FOR flow to
the reburn burners was turned off and when boiler economizer outlet oxygen concentrations are
not balanced. During parametric tests several fly ash samples were isokenetically collected and
analyzed for carbon content. Figure 9 shows the comparison of the combustible content in the
fly ash from the baseline and gas rebum conditions over the range of boiler steam flow. The
Baseline LOI levels varies between 6 to 11% over the boiler load range compared to 8 to 11%
during gas reburning operation. These data indicate a minimal impact on unburned combus-
tibles resulted during reburning operation.
Boiler Operations Experience
Smooth transition from baseline to reburn conditions has been Kodak's experience to date. The
majority of the optimization tests were performed at approximately 20% gas in order to achieve
the targeted 0.6 Ib/MBtu of NOX emissions. The boiler temperature profile, FEGT, and convec-
tion pass tube temperatures are within acceptable ranges. Under these optimized conditions, au-
tomatic boiler load control operates efficiently. One feature that was identified revealed that the
boiler required additional oxygen probes at the economizer outlet to consistently measure/control
the boiler oxygen concentrations accurately.
Kodak Boiler 43 Gas Reburning NOx Emission Summary
Boiler Load (Steam Flow) versus Loss on Ignition Levels
Optimized Control Conditions
oj in :
o i
ra 1 n "
ca :
O.
co :
(D
c
o
CO
O
o -
300
ill!
! I i i
^^ | | ; Reburning Test Data
\ X. i i^-^""*""" I*
\ i \J^,--'< Jl '"
\ I V4" .s'
\ ; '.^..'' Baseline Test Data
V..^ j __,.•'•":
i "'T'" ; ''••
; i
i i i !
i ! ! !
i. ! ! ,, !.
350 400 450 500 550 600 650
Boiler Steam Flow (1000's Ibs/hr)
Figure 9 Effect of gas reburn on the unburned combustibles
14
-------
Boiler Tube Corrosion Evaluation
Since gas rebum is accomplished by operating a portion of the furnace in a reducing atmosphere, po-
tential corrosive conditions could exist. During the design of the rebum system, special attention was
given to the design of reburn burners in order to provide sufficient penetration and distribute the gas in
the boiler evenly; thereby reducing the potential for boiler tube wastage. In addition, to determine that
the rebum system was not responsible for furnace corrosion, a furnace tube thickness study was per-
formed during the outage when the reburn system was installed. The same study was performed again
once the rebum system had been operating approximately 15 months.
Boiler tube thickness measurements were taken at four elevations in the furnace. Elevation 1:
reburn burner centerline, elevation 2: approximate midpoint of the reburn zone, elevation 3: top
of the reburn zone (center of the overfire air ports) and elevation 4: burnout zone (approximately
eight feet above the overfire air ports). The elevations for the tube thickness measurements were
identified by instrument survey and should be accurate plus or minus Vi" The tubes were sand-
blasted and then ultrasonically tested for thickness on the crown of the tube as well as left and
right of the crown. The accuracy of the instrumentation is plus or minus 0.005" and the readings
indicate that the tube thicknesses are the same within the accuracy of the measuring equipment.
Therefore, it is concluded that gas rebum has not contributed to any furnace tube wastage.
Kodak will continue monitoring the boiler tubes for a potential long term effect, but no corrosion
has been identified to date.
Economical Evaluation
An economical analysis was performed to evaluate the cost of the reburn technology for the
small cyclone boilers. It should be mentioned that both capital and operating costs are site-spe-
cific. The capital cost depends on the size of the unit, ease of the retrofit, availability of natural
gas at the plant and the control system upgrade requirements. The amount of the gas usage and
the differential gas to coal price determines the operating cost.
The gas reburn system at Boiler 43 is more capital intensive due to the customer's need for a new
control system for the entire boiler. For this first demonstration a FOR fan and associated equip-
ment was installed to fully characterize the reburn technology potential. In addition, certain
cost occurred due to the first-of-a-kind demonstration project. The total capital cost was approxi-
mately $5.9 million and it is high due to the small size of the boiler. The economy of scale is not
available here, e.g. at 200 MWe the cost of equipment is $16 to $17 per KW without controls.
Operating Cost & Levelized Cost
The differential cost of gas and coal is the main component of the operating cost. For the dura-
tion of this project, the gas/coal differential price was 1.74 $/MBtu. A maintenance cost of $
25,000 per year is estimated for the Boiler 43 gas reburn system.
Using the above capital and operating cost, the levelized cost of NOX reduction for Boiler 43 is $
1335/ton of NOX removed. This levelized cost is based on 85% capacity factor, and reducing
NOX from the baseline value of 1.37 to the post retrofit of 0.6 Lb/MBtu. The levelized cost was
calculated over 20 years and a discount rate of 7.13%.
15
-------
It should be mentioned that both capital and operating costs are site specific. The analysis does
not include the additional advantage of the gas reburn in reducing SO2 emissions. Also, larger
units benefit from the economy of scale resulting in reduced capital cost per KW of capacity.
The operating cost evaluation has a strong dependence on gas to coal cost differential and must
be determined on a site specific basis.
Conclusions
Boiler No. 43 reburn system has been optimized and Kodak will continue day-to-day commercial
operation. With the successful completion of the project, Kodak is considering the technology
for implementation on its other cyclone-fired units. B&W's reburning technology is commer-
cially available and is offered for turnkey retrofit projects. B&W also offers commercial guaran-
tees for future installations.
Acknowledgments
The authors extend their appreciation to Joe Hallstrom, Jeff Rogers, Bruno Morabito, and Jim
Riggs for leading the manufacturing of the reburn equipment, performing numerical modeling,
field shakedown, and collecting/evaluating the test data for this project.
References
1. G. J. Maringo et al., "Feasibility of Reburning for Cyclone Boiler NOX Control," EPA/
EPRI Joint Symposium on Stationary Combustion NOX Control, New Orleans, Louisi-
ana, March 23-27, 1987.
2. H. Farzan et al., "Pilot Evaluation of Reburning Cyclone Boiler NO,. Control," EPA/
EPRI Joint Symposium on Stationary Combustion NOX Control, San Francisco, Cali-
fornia, March 6-9, 1989.
3. H. Farzan et al., "Reburning Scale-Up Methodology for NOX Control from Cyclone
Boilers," International Joint Power Generation Conference, San Diego, California, Oc-
tober 6-10, 1991.
4. A. S. Yagiela et al., "Update on Coal Reburning Technology for Reducing NOX in Cy-
clone Boilers," EPA/EPRI Joint Symposium on Stationary Combustion NOX Control,
Washington, D.C., March 25 -28, 1991.
5. H. Farzan et al.. "Gas Reburn Retrofit on an Industrial Cyclone Boiler," EPA/EPRI
Joint Symposium on Stationary Combustion NOX Control, Kansas City, Missouri, May
16- 19, 1995.
6. H. Farzan et al., "NO., Control Using natural Gas Reburn on an Industrial
Cyclone Boiler," International joint Power Generation Conference, Houston, Texas,
October 14-16. 1996.
16
-------
Application Of Fuel Lean Gas Reburn Technology
at Commonwealth Edison's Joliet Generating Station 9
R. W. Glickert
Energy Systems Associates
J. S. Herzau and R. L. Meskimen
Commonwealth Edison Company
J. M. Pratapas
Gas Research Institute
Abstract
This paper presents the preliminary results of the first full-scale application of Fuel Lean Gas
Reburn (FLGR) in a cyclone fired boiler. Fuel Lean Rebum was installed on a 340 MW
cyclone-fired utility boiler at ComEd's Joliet Generating Station Number 9. Fuel Lean
Reburn involves the injection of 3-7% natural gas heat input via turbulent jets into the upper
furnace of fossil-fueled boilers while maintaining an overall fuel lean furnace, thereby
preventing excessive carbon monoxide emission and eliminating the need for downstream
completion air as employed hi conventional gas reburning technology. The controlled
dispersion of natural gas into high NOX regions downstream of the primary combustion zone
is designed to create locally fuel rich zones or eddies in which NOX reburning can take place
while maintaining an overall fuel lean furnace environment. The Fuel Lean Rebum shows the
potential to achieve better than a 40% reduction in NOX emissions using 7% gas heat input or
less. With the elimination of downstream completion air and the use of less gas heat input
than with conventional gas reburning systems, Fuel Lean Reburn systems offer lower capital
and operating costs than conventional reburn systems with a moderate decrease in the NOX
control potential. A Fuel Lean Reburn system was installed on Joliet Unit 6 in the Spring of
1997 with parametric and long-term performance testing taking place in the Summer of 1997.
This project is sponsored by ComEd and the Gas Research Institute.
Introduction
The 1990 Clean Air Act Amendments (CAAA) are requiring utilities to make large reductions
in nitric oxide (NO,,) emissions from their fossil-fired generating units. The conventional
utility approach to reducing NOX emissions has been through the installation of low NOX
burners with separated overfire air. However, cyclone fired boilers which are prevalent in the
Midwest and within ComEd's generating system are a unique firing design for which
conventional low NOX burner technology can not be applied. Cyclone fired boilers represent
about 9% of the US coal-fired generating capacity but account for about 14% of the NOX
produced by coal-fired utility boilers. This is due to the fact that the average uncontrolled
NOX emission rate from cyclone units is typically twice as high as comparative pulverized
-------
coal units based on the rapid fuel/air mixing characteristics inherent in the cyclone
combustion design.
In response to the need for cost-effective NOX emission control options the Gas Research
Institute (GRI) and ComEd are jointly sponsoring a field evaluation of Fuel Lean Reburn
Technology on ComEd's Joliet Generating Station 9, Unit 6. Energy Systems Associates has
developed the process design for the Joliet Unit 6 system and is performing parametric and
optimization testing.
Fuel Lean Reburning
NOX concentrations leaving boilers are a result of chemical kinetic and not thermodynamic
limitations. Figure 1 shows the equilibrium concentration of NO in typical flue gas a
function of temperature and the excess oxygen content. The equilibrium NO concentration is
seen to decrease rapidly with decreasing flue gas temperature. Concentrations of NO under
100 ppm are predicted at flue gas temperatures of 1800 °F for excess oxygen concentrations
of 3% and lower. So if the equilibrium NO concentration at 3% excess oxygen is 100 ppm,
why do NO concentrations greatly exceed this level during normal practice? This is caused
by the quenching of the NO equilibrating chemistry during normal boiler operation. In-
furnace and post-combustion NOX control technologies help to remove these chemical kinetic
constraints to NO equilibration by injecting either natural gas or an amine (NH;) based
compound such as ammonia or urea into the furnace.1
Figure 1: Thermodynamic Equilibrium Nitric Oxide as a Function of
Temperature and Stoichiometry
E
Q-
-Q-
O
SR2=0.95
SR2=1.00 (0% O2)
SR2=1.05 (1 % O2)
SR2=1.1 0 (2% O2)
SR2=1.1 5 (3% 02)
1500 1700 1900 2100 2300 2500
Temperature (deg F)
2700 2900
-------
A schematic of the Fuel Gas Lean Reburn process retrofitted on a cyclone boiler is shown in
Mgure 2. The combustion of coal in the primary furnace creates a high temperature low
excess oxygen content flue gas containing 400 to 1000 ppm on nitric oxide. The injection of
natural gas in the proper temperature window (-1800 to 2400 °F) results in chemical reactions
which reduce NO to molecular nitrogen. The desired injection temperature is as low as
possible consistent with the need to achieve natural gas ignition and burnout. The process
uses high velocity turbulent jets for dispersing the gas into the furnace. The amount of
natural gas is controlled to maintain an overall fuel lean stoichiometry in the upper furnace
Therefore, the gas is consumed in excess'oxygen already present in the flue gas and no
additional overfire air is injected above the gas injection zone for completing burnout
Conventional gas reburning typically requires 15% to 20% gas heat input injected at a higher
temperature than Fuel Lean Gas Reburn to create a uniformly fuel rich reburning zone. In
addition, conventional reburning employs the use of downstream overfire air to complete gas
burnout.
Figure 2. Conceptual Overview of Fuel Lean Gas Reburn Technology
Fuel Injector
5-7% Heal Input
- Flue Gas
NO = .55 to .70
Combustion Products
N0 = 1.0
93-95% Heat Input ^
Localized Natural Gas Eddies
Reduce NO to HCN
HCN
Eddy
The HCN formed further
reacts at the eddy boundary
with free radicals to
produce 'Mj>
-------
Project Overview
A Fuel Lean Reburn demonstration program is currently underway at CoEd's Joliet Station 9
in Joliet, Illinois. This project is being supported by the Gas Research Institute (GRI) and by
ComEd. The goal of this project is to achieve a 45% reduction from the baseline NOX
emissions at full load using a maximum of 7% natural gas heat input and to achieve a 50% to
60% NOX reduction at low load using up to 12% gas heat input. Under all operating
conditions it is necessary to maintain the average carbon monoxide concentration below 200
ppm when corrected to 50% excess air.
This demonstration program has been executed using two distinct project phases. Phase 1
involved proof-of-concept experiments conducted on a pilot scale combustor in combination
with computational fluid dynamic (CFD) modelling of the simulator and full scale process.
Phase 2 has involved the full-scale system engineering, the system retrofit and full-scale
performance testing and optimization including validation of the CFD and advanced
modelling practices.
Proof-of-Concept Experiments
The proof-of-concept experiments were conducted on the Small Boiler Simulator (SBS) test
facility at Babcock and Wilcox Company's Alliance Research Center in Alliance, Ohio. The
SBS is a 6 million Btu/hr capacity experimental boiler capable of simulating residence time
and temperature profiles of a full scale utility boiler. The SBS was fired by a single cyclone
combustor. The water-cooled SBS furnace measures 4.5 feet wide by 6.0 feet deep by 14.0
feet high. When operated at its design capacity, the calculated residence time from the
cyclone exit to the furnace exit is about two seconds.
The experimental procedure involved the injection of natural gas into the upper furnace of the
SBS using gas lances at a flue gas temperatures ranging from 1950 °F to 2300 °F. The
natural gas heat input was varied between 3% and 11% while maintaining overall fuel lean
conditions in the furnace. The pilot scale experiments documented NOX reductions up to 40%
at full load using 7% gas. At reduced loads, NOX reductions up to 58% were measured using
11 % gas. Figure 3 documents the percent NOX reduction achieved at full and reduced load
as a function of the percentage of gas heat input.
The amount of gas heat input and the degree of NOX reduction attainable with the Fuel Lean
Reburn process was found to ultimately be limited by the acceptable CO emission limit.
Figure 4 shows the CO emissions measured as a function of the percentage of excess oxygen
remaining in the stack gas. These results show that as the excess oxygen level was decreased
below 1.5%, the CO threshold was reached and CO levels began to consistently exceed 100
ppm.
-------
Figure 3. NOX Reduction for Lance Injectors at Mid-Furnace and Upper Furnace
at Full Load Reduced Load
o
0)
QL
x
O
60
50
40
30
20
10
KEY:
SL-MF: Single Lance, Mid Furnace
DL-MF: Dual Lance, Mid Furnace
DL-UF: Dual Lance, Upper Furnace
II
o
A°0
« SL-MF, 5.5 MBtu/hr, SR1 =1.14
A SL-MF, 4.3 MBtu/hr, SRI =1.19
H DL-MF, 5.6 MBtu/hr, SR1 =1.12
DL-MF, 4.2 MBtu/hr, SR1 =1.21
T DL-UF, 5.6 MBtu/hr, SR1 =1.1 6
O DL-UF, 5.4 MBtu/hr, SR1 =1.20
A DL-UF, 4.0 MBtu/hr, SR1 =1.23
567
% Natural Gas
10 11
12
Figure 4. CO Emissions Versus Percent Oxygen at the Stack
£
a.
-3
O
O
400
300
200
100
V
o
SL-MF, 5.5 MBtu/hr, SR1 =1.14
SL-MF, 4.3 MBtu/hr, SR1 =1.19
DL-MF, 5.6 MBtu/hr, SR1 =1.12
DL-MF, 4.2 MBtu/hr, SR1 =1.21
DL-UF, 5.6 MBtu/hr, SR1 =1.1 6
DL-UF, 5.4 MBtu/hr, SR1 =1.20
DL-UF, 4.0 MBtu/hr, SR1 =1.23
o n
o
m
a
% Stack O2
-------
Figure 5. Schematic of Joliet Unit 6
Other major conclusions of the pilot scale testing and CFD modelling were that better NOX
reductions were predicted and measured at lower injection temperatures (approximately 2000
°F) than at higher injection temperatures, numerous low volume jets were determined to be
more effective than fewer higher volume jets, natural gas must be introduced into regions
containing the bulk flow for effective NOX reductions but high velocities in the bulk flow
region increased turbulent mixing and combustion rates of the gaseous fuel thereby limiting
the exposure to CH( radicals and working against the Fuel Lean Reburn Process and finally,
that the Fuel Lean Rebum Process can produce strong NOX reductions limited only by CO
levels generated.3
Based on the favorable results of the proof-of-concept experiments and CFD modelling, GRJ
and ComEd agreed to proceed with the full-scale installation and testing.
Unit Description
Joliet Unit 6 is a 327 MW B&W cyclone-fired boiler which was placed in-service in 1959.
The unit currently fires a low-sulfur Western sub-bituminous coal. The boiler consists of a
single divided furnace arranged with a water cooled slag tap furnace; a continuous tube,
horizontal tube superheater interrupted for attemperators; a continuous tube, horizontal
reheater with an economizer; and two regenerative Ljungstrom air heaters. The unit is fired
with nine horizontal cyclone furnaces.
The primary furnace, secondary
furnace, secondary superheater,
primary superheater/reheater and
economizer are arranged successively
one above the other so that the gas
flow is vertically upward through the
unit as shown in Figure 5. The
primary superheater and reheater are
located on opposite sides of the
divided furnace. A diversion air
damper located at the exit from the
economizer provides for the control
of the reheat steam temperature by
diverting flue gas either towards or
away from the reheat side of the
furnace. The boiler is capable of
delivering a maximum of 2.2 million
pounds of steam per hour at 2000 psi
and 1015 °F superheat and 1005 °F
reheat. The primary furnace is
approximately 60 feet wide, 20 feet
deep and 25 feet high. Four cyclones
are located in the front wall or reheat
side and five cyclones are located on
the rear wall (superheat side).
-------
NOX Regulation Affecting Joliet Unit 6
The acid rain provisions of Title IV of the 1990 CAAA will require Joliet 6 to meet an
annual average NOX emission rate of 0.86 Ib/MMBtu beginning in the year 2000. For the
first quarter of 1997, the NOX emission rate for Joliet 6 averaged 0.96 Ib/MMBtu. In addition
to the current regulation enforceable under Title IV, ComEd must address additional NOX
regulations resulting from the ozone attainment provisions required by Title I. The Ozone
Transport Advisory Group (OTAG) is currently working to establish required measures to
reach ozone attainment in the eastern 37 states. The anticipated regulations from this process
will likely be more stringent than current acid rain limits, particularly during the summer time
ozone nonattainment months.
Overview of the FLGR System on Joliet Unit 6
The Fuel Lean Reburn system consists of a total of 36 gas injectors divided equally between
the superheat and reheat side of the furnace. Low pressure steam is supplied to each gas
injector to aide in the gas jet penetration. The steam was provided in this first large scale
demonstration of Fuel Lean Gas Reburn has a hedge to ensure that adequate penetration of
the gas into the bulk gas stream could be achieved. However, it is our expectation that
adequate penetration may be achieved without steam, thereby limiting its necessary use in
future deployments. There are 26 gas injectors located at 208 foot furnace elevation which is
approximately 56 feet below the entrance to the convective section. The remaining 10
injectors are located at the 229 foot furnace elevation, or about 35 feet below the entrance to
the convective section. The gas/steam injectors are designed to operate at about 35 psi and to
maintain a design steam to gas mass ratio of 2:1. The gas system was sized to provide a
maximum full load gas heat input of 12%. This is equivalent to about 350,000 SCF/hr of
natural gas or about 15,000 Ibs/hr of gas. The maximum steam flow is 30,000 Ibs/hr but
under normal operating condition will operate at about 17,500 Ibs/hr. A schematic of the
gas/steam injector is shown in Figure 6. The injectors are designed to provide high velocity
jets of gas spatially into the furnace. The steam is intended to aide in gas penetration and to
influence the gas mixing characteristics. Each injector is equipped with manual isolation and
flow control valves for both the gas and the steam.
Results of Full-Scale Testing
At the time of this writing, full-scale performance testing of the Fuel Lean Gas Reburn
process at Joliet Unit 6 is underway. The test plan calls for optimization of the Fuel Lean
Reburn process at 320 MW which is currently full load, at 270 MW, 210 MW and 150 MW.
To date all testing has been performed at 320 MW.
-------
Figure 6
- ISO UL
noku
TTOT.
w /OHAALY to
r~f
-Jgrn®®*-^
> MOUJ MM ( nwi owict)
UOa OU. I/* ' UM. A«T *0 L»-
AJR tt KrORAU-C PO«tf>. WC.
-------
The normal baseline operating condition at 320 MW is with all nine cyclones in-service and
operating at roughly equivalent levels. Under normal operating conditions the baseline NOX
emission rate is in the range of 0.91 Ibs/MMBtu to 0.98 Ibs/MMBtu with this small variation
being attributable to variability in fuel and operating conditions. The baseline CO level
measured at the economizer outlet typically does not exceed 10 ppm corrected to 50% excess
air (EA).
The parametric testing will included an evaluation of the following parameters on NOX and
CO emissions:
1. The impact of steam co-injection. The objective is to determine if the steam improves
the gas penetration or mixing and adds value to NOX or CO control.
2. The importance of the flue gas temperature at the gas injection location on the fuel
lean reburning kinetics. The local injection temperature can be changed by using
either the upper or lower elevation of injectors. Flue gas temperature measurements
indicate that the lower (208 foot) elevation has an average gas temperature of about
2100 °F at full load whereas the upper elevation has an average gas temperature of
about 2000 °F.
3. The impact of changes to the spatial distribution of gas within the furnace.
4. The impact of varying percentages of gas heat input up to about 12% heat Input.
Figure 7 shows the percentage of NOX reduced as a function of the percent gas heat input to
the unit a full load with the results indicating either steam co-injection or no steam use. The
results show NOX reductions ranging from 8% to 43% using 1.5% to 7% gas heat input
respectively. The preliminary results show no clear benefit to the use of steam co-injection.
This result may mean that adequate jet penetration is achieved without the steam. The
current gas/steam injector design will be evaluated to determine if it can be optimized to
achieve greater NOX reduction.
Figure 8 shows the same set of NOX reduction data organized according to the location of the
gas injector in-service. The lower elevation of injectors represents the hottest potential
average injection temperature at full load while the upper injectors represent the coolest
possible injection temperature at full load. The results show no clear indication that injection
at the cooler elevation is consistent with improved fuel lean reburning kinetics. However, the
known temperature difference between these elevations is only about 100 °F at this load. The
effectiveness of cooler gas temperatures will become more pronounced at lower boiler loads
when gas temperatures in these locations will decrease 100 to 200 °F.
Carbon monoxide emissions tend to become the limiting factor in achieving greater NOX
reductions. The CO results showed very poor correlation to the average excess oxygen level
remaining in the flue gas as well as little correlation to steam co-injection characteristics or
gas injection elevation. Figure 9 shows the CO level in parts-per-million at 50% EA as a
function of the percent NOX reduced. The results show it is possible to achieve up to a 20%
-------
Figure 7
Fuel Lean Gas Reburning at Jollet Unit 6
Full Load 320 MW
50
40
8?
§ 30
u
3
15
-------
NOX reduction with virtually no change in the boiler CO emission level. At 30% NOX
reduction the CO level begins to average about 100 ppm. Based on the need to remain below
200 ppm on average, the results to date indicate about a 38% NOX reduction can be
maintained with acceptable CO. Preliminary results also indicate that the unburned carbon
levels measured in the fly ash have not changed as a result of the Fuel Lean Reburn process
with baseline and reburning conditions both achieving loss-on-ignition (LOI) values of 2%
and lower.
Figure 9
Fuel Lean Gas Reburning at Joliet Unit 6
Full Load 320 MW
300
en
8 200
£
£
m
0
1 100
3.
o
u
0
(
• /
* L
•
•
* *
s/ •
} 10 20 30 40 50
NOx Reduction (%)
The test plan calls for additional testing over a wider boiler load range as well as better
investigation of spatial gas distribution patterns at full load.
An additional benefit of the Fuel Lean Gas Reburn Process at Joliet Unit 6 has been the
potential to extend the maximum generating capacity of the unit. Joliet 6 boiler is currently
permitted to generate as much as 340 MW. However, the forced draft fans limit the
summertime maximum reliable capacity to about 320 MW. Because the Fuel Lean Gas
Reburn Process injects natural gas without combustion air, the displacement of coal heat input
creates additional capacity on the forced draft fans. This has enabled the unit to increase
maximum generating capacity about 10 MW to 330 MW using only 3% gas heat input and
maintain NO., emissions below 0.80 Ibs/MMBtu.
-------
Summary
The Fuel Lean Gas Rebuming installation at Joliet Unit 6 offers a potential option to meet
Title IV NOX regulations for cyclone and wet bottom boilers and Title I regulations. The full
load testing completed to-date has documented NOX reductions of as much as 43% using 7%
natural gas heat input. The full-scale results achieved are consistent with the pilot scale
results and conclusions from CFD modelling of the process. To date the steam co-injection
has not proven beneficial and the expected temperature dependency of Fuel Lean Gas Reburn
kinetics has not been verified. As anticipated from the pilot scale and CFD modelling, CO
emissions prove to be the limiting factor to achieve larger and sustained high levels of NOX
reductions. Additional testing at reduced loads as well as a long-term demonstration test
remain to be completed in the summer/fall of 1997. Other tests of Fuel Lean Gas Rebum are
underway or planned by GRI and ESA on boilers with different firing designs at other
utilities.
References
1. B.P. Breen, H.S. Hura, "Upper Furnace Natural Gas Injection for NOX Reduction in
Utility Boilers", presented at the 13th Annual Pittsburgh Coal Conference, Pittsburgh,
Pennsylvania (September 1996).
2. International Gas Reburn Technology Workshop, Malmo, Sweden, February 1995.
3. B.P. Breen, H.S. Hura, J. A. Urich, "Development and Application of a CFD Model
for Upper Furnace Gas Injection Process Design on a Cyclone Fired Boiler'1,
presented at the GRI Workshop on Numerical Modeling of Natural Gas Injection Into
Large Boilers for NOX Emission Control, Chicago, Illinois (May 1996).
-------
Tuesday, August 26; 1:00 p.m.
Parallel Session B:
Selective Catalytic Reduction
-------
APPLICATIONS OF SELECTIVE CATALYTIC REDUCTION
TECHNOLOGY ON COAL-FIRED ELECTRIC UTILITY BOILERS
Ravi K. Srivastava
Acid Rain Division
U.S. Environmental Protection Agency
501 Third Street, NW
Washington, B.C. 20001
Anne Johnson, Mary Jo Geyer, and Perrin Quarles
Perrin Quarles Associates, Inc.
501 Faulconer Drive, Suite 2D
Charlottesville, VA 22903
Abstract
In January 1996, the U.S. Environmental Protection Agency (EPA) initiated a study to examine
the application and performance of SCR technology on coal-fired electric utility boilers. The
findings of the study indicate that all of the SCR applications surveyed have achieved targeted
NOX emission rates. Many boilers reported average NOX emission rates at or below 0.15
Ibs/rnmBtu. Those boilers reporting emission rates higher than 0.15 Ibs/mmBtu are generally
meeting emission limits set at these higher rates. In general, the operational histories of SCR
installations indicate that NOX reductions are being achieved in a reliable manner.
Introduction
Emissions of nitrogen oxides (NOJ contribute to adverse health and environmental impacts
resulting from formation of tropospheric ozone, acid rain, and fine particulates. In 1995, coal-
fired electric utility boilers in the United States accounted for 5.5 million tons of NOX (about 23
percent of the total national NOX emissions).i>2
Selective catalytic reduction is a NOX control technology that utilizes a catalyst to reduce NOX to
nitrogen and water. Although SCR was developed in the United States, other countries, such as
Japan and Germany, have aggressively implemented this technology on coal-fired utility boilers
over the past fifteen years and have achieved substantially reduced NOX emission rates. In
general, more than 200 installations of SCR systems operating on coal-fired boilers worldwide
have accumulated an experience base of more than 1700 years.3
-------
In light of the broad international experience with use of SCR and the health and environmental
concerns surrounding NOX emissions, EPA initiated a study to assess the performance of SCR
being achieved on coal-fired utility boilers. This paper summarizes the findings of this study;
more detailed information can be found in the final report on the study.4
SCR Technology Description
Selective catalytic reduction is a post-combustion NOX control technology capable of providing
NOX reductions in excess of 90 percent. This technology is widely used in commercial
applications overseas and is experiencing expanded use in U.S.. facilities. The SCR process uses
a catalyst at approximately 300-450 °C to facilitate a heterogeneous reaction between NOX and an
injected reagent, ammonia (NH3), to produce nitrogen and water.
A typical SCR system is comprised of: a storage, delivery, vaporization, and injection system for
the reagent; an SCR reactor housing the catalyst; soot blowers; and additional instrumentation.
Anhydrous or aqueous ammonia are used as reagents. The catalyst is a critical component of an
SCR system and its NOX reduction performance and resistance to deactivation affects the cost
effectiveness of the SCR application.
Three SCR system configurations are available for use with coal-fired boilers. In a high dust
configuration, the SCR reactor is located between the economizer and the air preheater. In this
configuration, the catalyst is exposed to flyash and chemical compounds present in the flue gas
that have the potential to degrade the catalyst mechanically and chemically. However, as
evidenced by the extensive use of this configuration, appropriate design of a high-dust SCR
system can mitigate the mechanical and chemical impacts on the catalyst. In a low dust
configuration, the SCR reactor is located downstream of the electrostatic precipitator (ESP). In
this configuration the potential for degrading effects of flyash on the catalyst is reduced. In a tail
end configuration, the SCR reactor is located downstream of the flue gas desulfurization (FGD)
unit. The tail end configuration is implicitly low dust. However, this configuration may be more
expensive than the high dust configuration due to associated flue gas reheating requirements.
Assessment of SCR Performance on Coal-Fired Utility Boilers
To conduct a comprehensive assessment of SCR performance on coal-fired utility boilers,
information on existing SCR installations and their operation was reviewed in the EPA study.
This information was obtained from coal-fired plants in the U.S., Germany, Sweden, Austria,
Denmark, and Finland. Since SCR systems are typically designed and operated to ensure
compliance with applicable regulatory requirements, information on applicable NOX emission
limits was also obtained from these plants. Several of the European utilities requested that the
names of their plants be kept confidential. To accommodate this request, each of the units in this
paper is identified by a letter and a number. The letter identifiers relate to the countries in which
plants are located and are: US (United States), A (Austria), D (Denmark), F (Finland),
G (Germany), and S (Sweden). The number identifiers relate to the numerical order in which
data was received from units in a country. The information received from SCR installations is
discussed in the following sections.
-------
Information on Units and Associated SCR Systems
Information on 33 units equipped with SCR systems at 21 coal-fired plants in the U.S. and
Europe was acquired and reviewed in the EPA study. These units range in size from 40 MWe to
740 MWe. Thirteen units were constructed with SCR systems and are designated as "new",
while 20 units were retrofitted with SCR systems. The SCR installations include 29 high dust,
two low dust, and two tail-end configured systems. Of the plants reporting boiler information,
thirteen units are tangentially fired, four are cyclone fired, and five are wall fired. Sixteen of the
units are dry bottom boilers and 15 are wet bottom boilers; of the latter, six recirculate their
flyash. As described above, the information collected on boilers and their associated SCR
systems represents a broad variety of SCR applications. In addition, startup dates of these
applications indicate that many of the SCR systems have been in operation for six or more years.
Hence, these applications contribute a significant level of SCR-related operating experience to
the EPA study.
Pertinent Regulations
Summarized in Table 1 are the applicable NOX emission limits for the U.S. and the European
units that provided data for the EPA study. In order to facilitate comparisons between the U.S.
limits expressed in pounds of NOX per million Btus of heat input (Ibs/mmBtu) and European
limits, Ibs/mmBtu equivalents for the latter are also shown in Table 1. Note that in arriving at
the Ibs/mmBtu equivalents for the German and Austrian limits, the standard bituminous coal F-
factor of 9870 dscf/mmBtu given in EPA Reference Method 19 (40 CFR Part 60, Appendix A)
has been used.
As seen in Table 1, in general, coal-fired units in the U.S. and Europe are meeting emission
limits set at or below 0.17 Ibs/mmBtu. Notably, there are units in the U.S. and Germany that are
complying with emission limits set at or below 0.10 Ibs/mmBtu. An exception is a U.S. unit that
is currently complying with a Reasonably Available Control Technology (RACT) limit of 1.4
Ibs/mmBtu.
It is interesting to note that the plant in Sweden has an economic incentive to lower its NOX
emissions. In Sweden, all electric utility plants with more than 10 MWe capacity, that produce
more than 50 GWe, pay a fee on NOX emissions (on a per kg basis). After a one percent
administrative fee is deducted, all remaining revenues are redistributed to the utilities based on
the fraction of total national electrical power generated by each utility.
Analysis of NOX Emissions
Emissions data were received from six units at five plants in the U.S., 18 units at 10 plants in
Germany, three units at one plant in Sweden, four units at three plants in Austria, one unit in
Denmark, and one unit in Finland. Many of these units provided continuous emissions data for a
month or more of operation while other units provided summary data (e.g., annual averages, etc.)
-------
As discussed before, these data represent a broad spectrum of SCR applications.
The NOX emissions data were reported in various units (e.g., Ibs/mmBtu, mg/m3, mg/MJ, etc.).
In order to compare SCR performance between data sets, it was necessary to normalize the
emissions data to a standard basis. The basis chosen in this work was Ibs/mmBtu since U.S.
units report their NOX emissions on this basis. Generally, F-factors were used to convert all
concentration data to emission rates expressed in Ibs/mmBtu. Several plants provided complete
information for the coal(s) used on unit(s) during the period(s) for which emissions data were
provided. For these plants, an F-factor was calculated for each emissions period. Other plants
provided the characteristics for all coals used at the plant, without specifically identifying the
unit(s) where the coal(s) were burned. For each of these plants, the coal data were used to
determine a single plant-specific F-factor. Note that the higher the F-factor, the higher the
calculated NOX emission rate in Ibs/mmBtu. Thus, to avoid underestimation of the controlled
NOX rates achieved with SCR, the highest of the computed F-factors (wherever possible) has
been used in this study.
A few of the German plants did not provide sufficient coal data for determining F-factors but
indicated that the coals used were either bituminous or "hard German coals." Such coals were
considered to be equivalent to U.S. bituminous coals with the associated standard F-factor of
9780 dscf/mmBtu given in EPA Reference Method 19 (40 CFR Part 60, Appendix A). This
standard F-factor was used in converting emissions data from these plants. The standard F-factor
was also used in converting data from plants in Finland and Denmark.
Note that the continuous emissions data received from SCR applications may include exempt
data (i.e., data related to startup, shutdown, or some other excusable event). These exempt data
would have been excluded by plants in their compliance determinations. Because exempt events
for most of the units were not specifically identified in the continuous emissions data that were
received, all reported data were included in the calculations. The only plants providing
continuous data that identified exempt emissions were US-4 (there were none for the reported
period) and D-l (three periods were identified). Data for the three exempt periods identified by
D-l were excluded from emissions analyses.
In contrast to exempt emissions, periods of no reported emissions could be identified in the
continuous emissions data and were excluded from emissions analyses. In many cases it was
possible to identify periods in which the unit was down because the corresponding load (Mwe)
data were zero; data for these periods were also excluded from analyses. Additionally, data for
the periods during which units (S-l :A and A-3) were burning a supplemental fuel (such as gas or
oil) were also excluded from all analyses.
To assess SCR NOX control performance, overall averages, daily averages, and 30-day rolling
averages of emissions were examined. Presented are analyses related to continous emissions
data received from SCR applications; analyses of summary data received can be found in the
final report on the study. As noted previously, the emission averages were computed using data
that may include exempt emissions and using the highest calculated F-factor, wherever possible.
Hence, the averages presented hi the following sections are considered to be conservative.
-------
Overall Averages
Overall average emissions for units that provided continuous data are presented in Figure 1. The
solid lines in this figure represent the regulatory emission limits for each of these units. For the
German and Austrian units, a range of limits based on the applicable F-factors is shown. Note
that the Swedish plant S-1:A provided continous emissions data for the months of October 1995
and January 1996. For this unit, the overall average emission for each of these months is shown
separately in Figure 1.
As seen in Figure 1, in all cases the overall average emissions were lower than the applicable
emission limits. Further, all of the units, except US-6, achieved 0.17 Ibs/mmBtu or lower for
their overall average emissions and 12 of the 19 units achieved 0.15 Ibs/mmBtu or lower. Unit
US-6 is regulated with an emission limit of 1.4 Ibs/mmBni and achieved an overall average
emission rate of 0.91 Ibs/mmBtu.
The German, Austrian, Danish, and Finnish units, achieved overall emission averages ranging
from 0.08 to 0.17 Ibs/mmBtu, and, therefore, are consistently able to achieve NOX emissions rates
at or below the applicable limits. The new German unit (G-4:A) operating under an emission
limit of 100 mg/m3, achieved an overall average of 0.08 Ibs/mmBtu.
The Swedish unit achieved overall emission averages of 0.04 and 0.07 Ibs/mmBtu during two
separate months. The lower average was achieved during variable load conditions experienced
in October 1995, and the higher during maximum load conditions experienced in January 1996.
These emissions were dramatically below the Swedish standard of 80 mg/MJ (0.19 Ibs/mmBtu).
Four United States units (US-1 :A, US-1 :B, US-2, and US-4) were able to achieve overall
emission averages ranging from 0.13 Ibs/mmBtu to 0.16 Ibs/mmBtu. These averages were below
the applicable 0.17 Ibs/mmBtu emission limits in their Prevention of Significant Deterioration
(PSD) permits.
Daily Averages
Daily emission averages were calculated for units that provided continuous emissions data. For
each of these units, except US-6, the mean and the range of the daily averages experienced are
shown in Figure 2. As noted before, unit US-6 is regulated with an emission limit of 1.4
Ibs/mmBtu. This unit achieved daily averages below 1.13 Ibs/mmBtu for the reported period.
Note that for the Swedish Unit S-l :A, the averages for the months of October 1995 and January
1996 are shown separately in Figure 2.
The mean values shown in Figure 2 reflect that the SCR installations are consistently emitting
NOX at rates below the applicable emission limits. As explained before, the data shown in Figure
2 may include emissions generated during exempt periods (e.g., startup, shutdown, equipment
malfunction, etc.). As emissions during such periods can be higher than those occurring during
normal operation, inclusion of such emissions may explain those cases where maximum daily
averages are much higher than the corresponding means.
-------
As seen in Figure 2, the new German unit (G-4:A) complying with the limit of 100 mg/m3,
achieved relatively low daily averages. Further, the Swedish unit S-l :A was able to achieve NOX
emission rates that were substantially below the annual average emission limit of 80 mg/MJ.
Also the four U.S. units achieved 24-hour averages consistently below 0.17 Ibs/mmBtu; three of
the four units consistently emitted at rates of 0.15 Ibs/mmBtu or lower.
Thirty-Day Rolling Averages
Thirty-day rolling averages were calculated for all of the units for which at least 30 days of
continuous data were provided. For each of these units, except US-6, the mean and the range of
the 30-day rolling averages experienced are shown in Figure 3.
The 30-day rolling averages in Figure 3 range from 0.06 to 0.18 Ibs/mmBtu. Note that nine of
the 16 units experienced thirty-day rolling averages at or below 0.15 Ibs/mmBtu. The average
for the new German unit (G-4:A) was 0.08 Ibs/mmBtu and the averages for the five new U.S.
units (US-1: A and B, US-2, US-4, and US-5) ranged between 0.06 to 0.16 Ibs/mmBtu. As noted
previously, unit US-6 is regulated with an emission limit of 1.4 Ibs/mmBtu that is much higher
that the emission limits for the other units. Hence, this unit achieved 30-day rolling averages
below 0.95 Ibs/mmBtu for the reported period.
Emission Averages Using U.S. Bituminous Coal F-Factor
U.S. utilities can calculate emission rates using an F-factor based on their actual coal
composition or on standard F-factors found in EPA Method 19 (40 CFR Part 60, Appendix A).
In order to assess the impact of using calculated F-factors in the analysis of emission rates, a
comparison of calculated versus standard F-factors was performed. Figure 4 depicts a
comparison between the average NOX emission rates determined using the calculated F-factors
(for 20 units that provided sufficient coal data) and those determined using the Method 19 F-
factor for bituminous coal.
As seen in Figure 4, the average NOX emission rates determined using the calculated F-factors are
higher than those determined using the Method 19 F-factor for bituminous coal. Thus by using
calculated F-factors, wherever possible, EPA has been conservative in its conversion of NOX
concentration data to rates expressed in Ibs/mmBtu.
NOX Removal Efficiency
Some units provided continuous, contemporaneous pre- and post-SCR NOX emissions data.
Other plants provided pre-SCR NOX averages. In calculating the NOX removal efficiencies for
the SCR applications, any data points where either pre- or post-SCR emissions were zero were
discarded. Data points were also discarded where the pre-SCR emission rate was found to be
lower than the contemporaneous post-SCR emission rate. All other emissions, including those
occurring during possible exempt periods, were included in the SCR NOX removal efficiency
calculations. Where a range of pre-SCR values was provided, the mean value was used to
calculate the corresponding NOX removal efficiency.
-------
Shown in Figure 5 are the NOX removal efficiencies being achived at the SCR applications.
These efficiencies ranged from 54 percent to 94 percent. The units with relatively lower removal
efficiencies (54-79 percent) were the German and the U.S. units with relatively lower pre-SCR
NOX rates (0.32-0.8 Ibs/mmBtu). Where pre-SCR rates were higher (greater than 0.8
Ibs/mmBtu), the units achieved better than 80 percent removal efficiency.
Four units (G-4: A and the three units at Plant S-l) achieved relatively high NOX removal
efficiencies (85-89 percent) despite low pre-SCR rates (0.40-0.81 Ibs/mmBtu). This
demonstrates that SCR can reduce low levels of uncontrolled emissions effectively. As noted
before, unit G-4: A is required to meet a lower emission limit than the other German units. Plant
S-l appears to be minimizing NOX emissions below the applicable standard to take advantage of
the economic incentive provided in the Swedish regulation. Also of interest is the unit G-9: A,
which is characterized by a high pre-SCR NOX level of 2.03 Ibs/mmBtu. The retrofit SCR
application at this unit reduces NOX emissions by 93.6 percent to meet the applicable emission
limit.
As seen in Figure 5, in general higher NOX removal efficiencies need to be achieved by units with
higher uncontrolled NOX emissions. Further, the results suggest that SCR technology is capable
of reducing a wide range of uncontrolled emissions (including NOX emissions in excess of 2.00
Ibs/mmBtu) to rates at 0.17 Ibs/mmBtu or lower.
Operational Experience
SCR installations were requested to provide information on their SCR-related operational
experience. Many of the SCR systems surveyed have been in operation for six or more years and
have accumulated significant levels of operating experience. Although some plants experienced
problems related to SCR, most have not, and all of the plants that reported problems have
successfully resolved them. The following paragraphs discuss the information on operational
experience received from SCR applications.
Ammonia Slip and Balance-of-Plant Impacts
At SCR installations, ammonia slip results from the reagent that does not participate in NOX
reduction and instead "slips by" the catalyst. Ammonia slip can react with the sulfur trioxide
(SO3) present in the flue gas to form ammonium salts. In high dust applications, these
ammonium salts can increase the potential for air preheater pluggage. Further, excessive
ammonia slip can cause flyash contamination and thereby have an adverse impact on flyash
marketing. Generally, ammonia slip hi an SCR application may be rninimized by ensuring that
adequate amount of catalyst is provided and that injected ammonia is adequately mixed with the
flue gas.
Information on ammonia slip levels were provided by 17 plants for 26 units. Guaranteed slip
levels (at the end of catalyst lifetime) are below 5 ppm for the units that reported this
information. Fourteen units reported actual slip levels being achieved; these levels range from
-------
<0.1 ppm to 5 ppm and seven units reported levels of less than 1 ppm. Most of the units that
reported ammonia slip information have been in operation for five years or more. These data
reflect that ammonia slip levels are being controlled to levels below 5 ppm and that many
installations are achieving much lower ammonia slip levels after significant periods of operation.
Air Preheater Impacts. Plants were requested to provide information on air preheater
washings related to SCR operation. Many of the plants responded by providing historical
information on air preheater washings. Of the 24 units reporting the impact of SCR on air
preheaters, only those with high dust configurations reported the need to conduct washing. The
frequency of air preheater washing varied from once in a six-to-seven-year period to once each
year, except for unit US-6 which has reported many washings of its air preheater since SCR
retrofit in 1995. However, unit US-6 noted that excessive ammonia slip occurring due to
insufficient bypass damper closing was believed to have caused much of the air preheater fouling
that required washing. Initially, Plant US-2 also conducted washings once or twice a month
following the SCR operation. However, with the addition of a layer of catalyst in October 1996,
the need to conduct frequent air preheater washings has been eliminated at this installation.
Considering that annual washing of air preheaters at coal-fired plants is commonly conducted,
the results suggest that all of the responding plants did not experience notable increases in air
preheater washings resulting from normal SCR operation.
Flyash Contamination. Flyash absorption of any excess, unreacted ammonia (NH3) released
by an SCR system into the treated flue gas is a function of the ammonia slip rate, quantity of
flyash, and specific ash characteristics (namely pH, alkali mineral content, and volatile sulfur and
chlorine content). At an elevated pH, ammonia in the ash will be released, possibly leading to
odorous emissions. Such an occurrence can impact the marketability of flyash. Flyash
contamination with ammonia is avoided by designing and operating the SCR system to maintain
low levels of ammonia slip.
Plants were requested to provide information related to flyash disposal at their SCR installations.
The information received reflects that most of the responding plants sell their flyash and,
therefore, flyash contamination is not an issue at these installations. In light of the low ammonia
slip levels (less than 5 ppm) being maintained at the SCR installations, this result is not
unexpected.
Catalyst Replacement
A need for catalyst replacement arises when catalyst becomes deactivated. Five primary causes
of deactivation are: poisoning by arsenic and other chemical poisons, fouling of the surface by
flyash or sulfur-related compounds, plugging of flow channels, erosion, and thermal degradation.
In general, these deactivation mechanisms are countered by using poison-resistant catalysts,
selecting proper catalyst pitches, using appropriate soot blowing cycles, and selecting thermally
stable catalyst formulations.
Historical information received from SCR installations on their catalyst replacement cycles
-------
reflects that, in general, a layer of catalyst was replaced/added after 15,000-56,000 hours (or
approx. two to seven years) of operation. At Plant G-4B, no problems with catalyst performance
were noted after 55,000 operating hours (or approximately six years). These results suggest that
catalysts are performing satisfactorily over relatively long operating periods at all of the
responding SCR installations.
Findings
The following general observations on NOX emissions relate to all responding units with the
exception of unit US-6 which had an unusually high uncontrolled NOX emission rate (2.4 - 2.66
Ibs/mmBtu) and is at an interim stage of NOX emission reductions. Thus, when compared to
emissions from all of the other units in this paper, the controlled NOX emission rate for this unit
is an outlier.
Many of the SCR systems surveyed have been in operation for six or more years and, thus, have
accumulated a significant level of operating experience. In general, more than 200 installations
of SCR systems operating on coal-fired boilers worldwide have accumulated an experience base
of more than 1700 years.
Using SCR, coal-fired power plants in the U.S. and Europe are achieving average NOX emission
rates between 0.04 Ibs/mmBtu and 0.17 Ibs/mmBtu. Of the 19 units (including US-6) submitting
continuous emissions data, 12 had overall averages at or below 0.15 Ibs/mmBtu. Further,
Germany, Sweden, and Austria have units that are achieving daily averages consistently below
0.10 Ibs/mmBtu. The thirty-day rolling averages for the units surveyed ranged from 0.06
Ibs/mmBtu to 0.18 Ibs/mmBtu. Nine of the 16 units experienced thirty-day rolling averages at or
below 0.15 Ibs/mmBtu.
SCR NOX removal efficiency for plants included hi this study varied from 54 percent to 94
percent. As expected, this efficiency appears to depend on the pre-SCR NOX level and the
applicable emission limit. Results reflect that SCR is capable of reducing a wide range of pre-
SCR NOX emissions to rates at 0.17 Ibs/mmBtu or lower.
The Swedish plants are emitting NOX emission rates that are significantly below the applicable
regulatory limit. This suggests that the economic incentive provided in the Swedish regulatory
system has resulted hi NOX reductions hi excess of those that would be available through
compliance with the regulatory limit.
Guaranteed ammonia slip levels are below 5 ppm for the units that reported this information.
Fourteen units reported actual slip levels being achieved; these levels range from <0.1 ppm to <5
ppm and seven units reported levels of less than 1 ppm. Most of the units that reported ammonia
slip information have been in operation for five years or more. These data show that ammonia
slip levels are being controlled to levels below 5 ppm and many units are achieving much lower
ammonia slip levels, even after significant periods of operation.
Of the 24 units reporting the impact of SCR on air preheaters, only those with high dust
-------
configurations reported the need to conduct washing on a regular basis. At these units, the
frequency of washing varied from once in a six-to-seven-year period to once each year.
Considering that annual washing of air preheaters at coal-fired plants is commonly conducted,
the results suggest that no notable impacts on air preheaters resulted from normal SCR operation.
Most of the SCR installations that provided information on flyash disposal reported that they sell
their flyash. This indicates that flyash contamination with ammonia is not an issue at these SCR
installations. In light of the low ammonia slip levels being maintained at SCR installations, this
result is not unexpected.
Several plants provided historical information on their catalyst replacement cycles. This
information indicates that in general a catalyst layer was replaced/added after 15,000-56,000
hours (or approximately two to seven years) of operation. At one plant no problems with catalyst
performance were noted after 55,000 operating hours (or approximately seven years). These
results suggest that catalysts are performing satisfactorily over relatively long periods of time at
all of the responding SCR installations.
References
1. U.S. EPA. Acid Rain Program Emissions Scorecard 1995. EPA 430/R-97-009, March
1997.
2. U.S. EPA. National Air Pollutant Emissions Trends 1900-1994. EPA 454/R-95-011,
October 1995.
3. "Coal Power 2," Worldwide Coal-Fired Power Stations and Their Pollution Control
Systems, A Database by IEA Coal Research, 1997.
4. U.S. EPA, Acid Ram Division. Final Report, Performance of Selective Catalytic
Reduction on Coal-Fired Steam Generating Units, June 25, 1997.
-------
Table 1
NOX Emission Limits for the Units Examined in the Study
Country
United States
Austria
Denmark
Finland
Germany
Sweden
Size, MWt
>500
-
>150
>300
NO, Emission Limit
0.1 7 Ibs/mmBtu,
O.lOlbs/mmBtu,
1.401bs/mmBtu
200 mg/m3 (-0.16 Ibs/mmBtu)
400 mg/MJ (0.93 Ibs/mmBtu)
70 mg/MJ (0.16 Ibs/mmBtu)
200 mg/m3 (-0.16 Ibs/mmBtu),
100 mg/m3 (-0.08 Ibs/mmBtu)
80 mg/MJ (0.19 Ibs/mmBtu)
+ Economic Incentive
-------
0.24
0.16
NO,
(Ibs/
mmBtu)
0.08
0.00
Ta» US. mil (na »hown)'
has an owrall average of
| OL91 end Is subject to a
Unit 0(1.411
Germany Austria Sweden Denmark Finland United States
Figure 1. Overall averages of NOX emissions being achieved at SCR applications.
U.<3D
0.24
0.22
0.20
0.18
0.16
NOx 0.14
(Ibs'mmBtu)
0.10
0.08
0.06
0.04
0.02
nnn
-
-
-
-,
i|
?'
V
-
-
-
•
1
1
1
T 1
1 ' 1
1
1
1
1
1
1
t
II II
1 " Vf
I
Germany
Austria Svieden Denmark Finland Unrted Stales
Figure 2. Daily averages of NOX emissions being achieved at SCR applications.
-------
(Ibs/mmBtu)
0.26
0.24-
0.22-
0.20 -
0.18 -
0.16
0.14
I
0.12
0.10
0.08-.
0.06 -
0.04-
0.02 -
0.00
Germany
Austria Denmark Finland United States
Figure 3. Thirty-day rolling averages of NOX emissions being achieved at SCR applications.
Mean NOx
Emission Rate
(Ibs/mmBtu)
0.050
G-1 G-4.a G-6.a G-6.C G-6.e G-7 G-9.b G-10.b US-1.b A-1
G-3 G-4.b G-6.b G-6.d S-1.a G-9.a G-10.3 US-La US-2 A-2
Plant/Unit
Figure 4. A Comparison of NOX emissions calculated using alternative F-factors.
-------
100
80
60
SCR
Efficiency
40
20
NOx Limit
0.5 1 1.5
Uncontrolled NOx Emissions (Ibs/mmBtu)
2.5
Figure 5. NOX removal efficiency being achieved at SCR applications.
-------
SCR APPLICATIONS:
ADDRESSING COAL CHARACTERISTIC CONCERNS
Sikander Khan, Project Engineer
Gajendra Shroff, Principal Engineer
Jagdish Tarpara, Senior Engineer
Bechtel Power Corporation, Gaithersburg, Maryland1
Ravi Srivastava, Environmental Engineer
US EPA Acid Rain Division, Washington, DC
Abstract
Concerns exist within the U.S. power plant industry regarding the suitability of SCR application
on power plants firing a variety of coals found in this country. Specifically, these concerns relate
to impacts of coal characteristics, such as sulfur, alkali, trace elements, etc., on SCR system
performance.
This paper presents the results of a study conducted by Bechtel (as a subcontractor to THE
CADMUS GROUP, INC.) for the U.S. Environmental Protection Agency to examine the
existing coal-fired SCR experience in light of the wide variability in the properties of coals fired
in U.S. power plants. As compared to the U.S., considerable SCR experience exists abroad.
This paper addresses the key question: Is SCR experience from abroad transferable to U.S.
applications?
Introduction
Selective catalytic reduction (SCR) technology has been applied extensively worldwide for NOX
control on coal-fired installations. In the U.S., there now exist seven coal-fired installations
utilizing SCR for NOX control. Because this application base does not cover the wide variety of
coals burned in U.S. plants, uncertainties seem to exist regarding SCR system design needs in
accommodating the varying characteristics of U.S. coals. In particular, these uncertainties have
been linked to coal characteristics, such as high sulfur and ash contents and high concentrations
of some trace or toxic elements (e.g., arsenic, alkalis).
The subject of this paper is a study that was undertaken to evaluate the current state-of-the-art
SCR technology and its suitability for NOX reduction in domestic power plants firing a wide
variety of U.S. coals(1)' The major areas investigated to meet the study objectives include the
following:
Characteristics of worldwide SCR installations on coal-fired plants
1 Subcontractor to THE CADMUS GROUP, INC., WALTHAM, MA
-------
• Current SCR design practices and features available for coal-fired applications
• Characteristics of coals fired in the SCR-equipped, worldwide coal-fired plants and their
comparison with the characteristics of coals fired in U.S. plants
• SCR experience with fuels other than coal that produce an operating environment comparable
to those found in coal-fired installations
The purpose of the above investigations was to evaluate current SCR experience in relation to the
concerns associated with the technology's capability to accommodate certain characteristics of
coals in general and of U.S. coals in particular. The results of these investigations are
summarized below.
Existing Coal-Fired Applications of SCR
The investigations for this study identified at least 212 worldwide SCR installations on coal-fired
units. The majority of information on these installations was obtained from a database
commercially available from the International Energy Agency . Additional information was
obtained from various other sources1 Of the 212 installations, seven of them are located in the
U.S. and the rest in other countries, notably Germany and Japan.
Tables 1 and 2 summarize the information on these 212 SCR-equipped units. While Table 1
covers all SCR units, Table 2 provides key data on the U.S. installations only. The information
presented in these tables is analyzed below:
• The coals fired in the SCR installations include lignite and bituminous coals.
• The installations range in size from 13 to 1,050 MW. This range spans the broad range of
boiler sizes used hi the utility power plants.
• The design NOX reduction efficiencies for the installations range from 35 to 90 percent. For a
large number of these units, this efficiency is 80 percent or higher. Operating data available
on some installations show that NOX reduction has exceeded 90 percent(j'4)
• SCR has been applied to a variety of boilers, including cyclone-fired, pulverized dry-bottom
(wall- and tangential-fired), pulverized wet-bottom, and vertical-fired type boilers. These
installations span the spectrum of boiler types used in utility power plants.
Since the above boilers use different combustion processes, coals with varying characteristics
must be fired in these boilers. The coals selected for these boilers are based on different
requirements associated with certain characteristics, especially the volatile content and ash
fusion temperatures.
• Information available on the SCR units also shows that these installations are subjected to a
variety of operating modes, which include base-load operation, cycling operation, and ash
recycling (for cyclone-fired boilers)(5'6)
• The SCR installations span a variety of applications, including:
-------
- Different SCR reactor arrangements, including high-dust, low-dust, and tail-end type
arrangements
- Anhydrous and aqueous ammonia reagents
- Honeycomb and plate type catalysts (pellets used on three units)
- New and retrofit installations
• The seven SCR installations in the U.S. also cover a variety of applications, including:
- Dry-bottom wall-fired, dry-bottom tangential-fired, cyclone-fired, and pulverized wet-
bottom type boilers
- Plate and honeycomb type catalysts
— Anhydrous and aqueous ammonia as reagents
- New and retrofit installations
- NOX reduction effectiveness ranging from 47 to 88 percent
In addition to the commercial U.S. SCR installations shown in Table 2, there have been test
demonstrations of SCR on coal fired plants. In one of these demonstrations, tests were
conducted over a period of two years at Plant Crist where SCR effectiveness was examined with
coals containing up to 3 percent sulfur. These tests showed that firing of coals with high-sulfur
content had no adverse impacts on catalysts with proper compositions0'
The overall SCR experience discussed above covers a large number of installations with different
types of boilers subjected to varying operating conditions and firing a variety of coals. Some of
these installations were designed for and have achieved high NOX reduction levels, exceeding 90
percent. In general, this experience strongly supports viability of SCR application on coal-fired
boilers.
Advances in SCR Design Related to Coal-Fired Applications
Flue gas resulting from coal firing, offers a relatively harsh environment for SCR applications.
Based on the extensive European and Japanese SCR experience, the following concerns have
been associated with coal firing:
• Deposition of dust on catalyst surfaces resulting in masking of active sites and catalyst
pluggage
• Erosion of catalyst due to fly ash impingement
• Deterioration and deactivation of catalyst active sites
The above effects and the mechanisms that cause them are now well understood. The coal
properties that can contribute to these problems have been identified. This enhanced know how
has resulted in many SCR technology improvements that address coal-related issues. The
advances in technology include modifications in catalyst composition and physical properties as
well as improved design techniques in other areas of the SCR system
-------
Table 3 summarizes the coal properties important to SCR, potential impacts of these properties,
and countermeasures available to mitigate these impacts. The information presented in Table 3
is analyzed below:
• Coal ash, if present in high quantities, can potentially cause deposits on catalyst surfaces,
pluggage of catalyst gas passages, and erosion of catalyst. The countermeasures that have
been successfully used to combat such problems include use of a proper catalyst pitch (a
measure of gas passage opening), use of sootblowers and screens above catalyst layers, use of
devices to obtain a uniform flow distribution within the SCR reactor, and provision of a
hardened leading edge on catalyst.
• During combustion of coal, the sulfur in coal forms SO3, which can react with other flue gas
components (e.g., CaO) and ammonia (SCR reagent) to form ammonium and calcium salts
that have a variety of impacts. Within the reactor, these salts can cause masking of catalyst
surface and pluggage of catalyst pores. Downstream of the reactor, SO3 can react with
residual ammonia (ammonia slip) to form sticky ammonium bisulfate that can cause
pluggage in air heater surfaces or cause contamination of fly ash collected in particulate
control devices, affecting marketability of this ash. SCR can exacerbate the effects of coal
sulfur, since the catalyst causes additional SO3 to be produced by oxidizing a portion of S02
present in flue gas.
Two of the most effective countermeasures used to combat coal sulfur impacts include
modifications in catalyst formulation to minimize oxidation of SO2 to SO3 and precise
control of ammonia slip to low levels (2 to 5 ppm).. In the new catalyst formulations,
vanadium content is optimized and pore size distribution is designed to maximize small
pores, both of which inhibit the SO2 oxidation rate. A controlled ammonia slip restricts the
maximum amount of ammonium bisulfate that can be formed downstream of the reactor (5
ppm of ammonia slip can convert to a maximum of 5 ppm of ammonium bisulfate), thus
minimi-Ting problems associated with this sticky substance.
Other countermeasures against ammonium salts include maintenance of proper flue gas
temperatures at the reactor inlet by utilization of an economizer bypass and use of
sootblowers to remove any deposits formed on the catalyst surfaces. For high-sulfur coal
applications, proper margins can also be added to catalyst volumes to account for the
poisoning effects of ammonium salts over the catalyst's operating life.
• Alkaline metals present in coal, especially sodium and potassium, can react directly with the
catalyst active sites to reduce the overall number of sites available for adsorption of ammonia
and NOX. However, such catalyst deactivation or poisoning is caused primarily by alkalis in
the water-soluble form. Since the majority of alkalis present in coal are not water soluble,
this poisoning effect is not a major concern with coal-fired SCR applications'7*
A countermeasure applied against the above poisoning risk consists of adding proper margins
in the catalyst volume to account for any potential catalyst deactivation. Research also shows
that, addition of certain metal oxides (e.g., tungsten trioxide) to catalyst formulation, can
actually enhance catalyst activity in the presence of alkaline metals(8'9)
-------
• Certain heavy metals present in coal, such as arsenic, lead, and phosphorus, have been cited
as catalyst poisons. These are present in coal in extremely small concentrations, and,
therefore, are not a major concern. However, in certain German cyclone-fired boiler
installations utilizing ash recycling, catalyst poisoning due to arsenic has been experienced.
This arsenic poisoning has been traced to enhanced arsenic levels hi flue gas, resulting from
ash recycling used at these installations. It has been shown that ash recycling enriches the
arsenic concentration in flue gas by a factor of 10 to 15(5> 8).
Arsenic poisoning is caused by gaseous arsenic trioxide, which diffuses into the catalyst and
solidifies on both active and non-active sites. Making provisions for an increased catalyst
volume to account for the potential loss of catalyst activity with this poisoning is one of the
countermeasures available. Arsenic content hi flue gas can also be reduced by addition of
lime or limestone (sources of CaO) to the coal(5) Another countermeasure is an optimized
catalyst pore structure that limits arsenic diffusion into the catalyst. A more direct
countermeasure available now is the use of an arsenic-resistant catalyst. In this catalyst,
molybdenum oxide is added, which minimizes the loss of catalyst activity caused by the
arsenic absorbed on the catalyst surface*5'8'.
• Chlorine is present hi coal as organic and/or inorganic chlorides. Chlorides can have both
poisoning and promoting effects on catalyst activity. Certain transitional metal chlorides
(e.g., CuCl) actually act as strong catalysts. HC1 and NH4C1 can cause catalyst deactivation
by removing NH3 and/or by reacting with active vanadium oxide to form inactive vanadium
chloride(9) At low flue gas temperatures, NH4C1 deposits may be formed on catalyst active
sites.
Catalyst deactivation from chlorides has not been reported as a significant problem in
worldwide SCR installations. It has also been reported that U.S. coals generally cany lower
chlorine concentrations as compared to foreign coals(10) The measures that can be used to
combat chloride-related effects are the same as those described above for some of the other
coal properties. This includes use of wider-pitch catalysts, sootblowers, and proper flue gas
temperatures at all operating loads. Similarly, use of an increased catalyst volume to cover
potential catalyst deactivation where high chlorides exist is an additional counterrneasure.
• Fluorine, present hi coal as metal fluorides or as a hydrofluoric acid, can have a potential
impact on SCR similar to that associated with chlorine. However, concentrations of fluorine
hi coal are lower than those of chlorine. Fluorine is, therefore, not considered a significant
factor hi terms of catalyst poisoning. The countermeasures against possible fluorine impacts
are the same as those described for chlorine.
The above analyses show that deactivating effects on SCR of contaminants in coal are well
understood and countermeasures are now available to minimize these effects. The SCR suppliers
contacted for this study indicated that an effective SCR design can be offered to match a given
set of coal characteristics and other site specific conditions. These suppliers are offering SCR for
commercial application on coal-fired plants with verifiable performance guarantees.
-------
Applicability of Foreign SCR Experience to U.S. Coal-Fired Applications
A major task of this study was to identify the SCR-related characteristics of coals fired in utility
power plants in the U.S. and compare these to the characteristics of coals fired in worldwide SCR
installations. In addition, SCR experience with other fuels was reviewed to verify the
applicability of this experience to U.S. coals. The information on U.S. coals was obtained from
two databases: one available from Department of Energy, Energy Information Agency
(DOE/EIA) and the other from U.S. Geological Survey (USGS/1''12)
The DOE/EIA database lists key information on coals fired in U.S. utility plants in 1995,
including coal mine location, coal tonnage shipped, and coal sulfur, ash, and heat contents. The
USGS database contains information on approximately 136 coal parameters from numerous coal
mines. This database contains coal sources that fall outside of the range of the coals included in
the DOE/EIA database. The coal sources in the USGS database were therefore reduced to match
the key parameters in the DOE/EIA database, which included coal mine location and
corresponding maximum sulfur and ash contents. While the DOE/EIA database provided the
coal sulfur and ash content information, the modified USGS database was the basis for the
information on remaining coal characteristics, pertinent to U.S. coals fired in utility power plants.
The information on the characteristics of fuels fired in the worldwide SCR installations was
obtained from a variety of sources'2'7'13> 14' It should be noted that this information, especially
on the coal trace element contents, was available only on a limited number of SCR-equipped
units.
To facilitate comparisons between U.S. coals and coals fired in SCR installations, Figures 1
through 12 have been drawn to show the number of plants or coal tonnage fired with varying
concentrations of coal constituents pertinent to SCR. Table 4 shows additional information on
coals fired in SCR installations. In addition, Table 5 shows SCR installations on fuels containing
high sulfur. The comparisons based on the information shown are discussed below:
• For U.S. coals, approximately 96 percent of the plants fire coals with ash content less than 15
percent (Figure 1). The remaining plants fire coals with ash between 15 and 30 percent. This
experience is comparable to the coal ash in SCR installations (Figure 2). Approximately 73
percent of these installations fire coals with ash less than 15 percent The remaining plants
fire coals up to 30 percent ash.
• Approximately 58 percent of U.S. plants fire coals with less than 1.0 percent sulfur (Figure
3). An additional 24 percent fire coals with sulfur ranging between 1.0 and 2.0 percent.
Only about 6 percent of the plants fire coals exceeding 3 percent. On a weight basis,
approximately 85 percent of the coals fired in U.S. plants have sulfur content less than 2
percent (Figure 4). Only 6 percent of coal tonnage fired in U.S. plants has sulfur in excess of
3 percent.
Of SCR installations, approximately 74 percent fire coals with up to 1.0 percent sulfur, 22
percent fire coals with sulfur between 1.0 and 2.0 percent, and 4 percent fire coals with sulfur
between 2.0 and 3.0 percent (Figure 5). These data show that the majority of coals (94
-------
percent) fired in U.S. plants (both on number of plants or weight basis) have less than 3.0
percent sulfur and that the existing SCR experience covers this range of coal sulfur.
SCR experience on high-sulfur applications is also evident from installations firing high-
sulfur fuels other than coal (Table 5). One of these installations fires asphaltum with a 5.4
percent sulfur content, which is greater than the maximum sulfur content of U.S. coals. Such
experience is considered applicable to coal firing, since ashphaltum contains relatively high
amounts of vanadium (200 ppm), which promotes conversion of SO2 to SO3, resulting in a
higher SO3 concentration. This fuel, therefore, generates a harsher environment for SCR,
compared to coal.
• The average arsenic content (89 percent) of the majority of U.S. coals on a weight basis falls
below 20 ppm (Figure 6). The remaining coals have arsenic levels up to 30 ppm. In
comparison, significantly higher arsenic levels of up to approximately 300 ppm have been
reported for coals fired in SCR installations (Table 4).
• For U.S. coals, up to 1,000 ppm of chlorine concentrations have been reported (Figure 7).
The maximum chlorine concentration reported for SCR installation coals is 1,900 ppm,
which is higher than that for U.S. coals (Table 4).
• For U.S. coals, up to 305 ppm of fluorine concentration has been reported (Figure 8).
However, for approximately 97 percent of these coals on a weight basis, fluorine is limited to
150 ppm. The fluorine concentration reported for SCR installation coals at 47 to 270 ppm is
comparable to that for U.S. coals (Table 4).
• A major portion (89 percent) of U.S. coals on a weight basis contains ash with Na2O ranging
between 0 and 2.0 percent (Figure 9). Ash In the remaining coals (11 percent) has 2 to 5
percent of Na2O. The maximum Na2O reported in ash of coals fired in SCR installations is
1.6 percent, which is comparable to the concentration reported for the majority of U.S. coals
(Table 4).
• U.S. coals have ash with K20 ranging from 0 to 2.6 percent (Figure 10). In contrast, ash in
SCR installation coals has higher K2O concentrations, ranging between 0.1 to 5.0 percent
(Table 4).
• For the majority (98 percent) of U.S. coals on a weight basis, P2O5 in ash varies between 0
and 1.0 percent (Figure 11). For less than 2.0 percent of these coals, the reported P2O5 level
is 1.7 percent. Ash in the SCR installation coals is reported to contain between 0 and 1.3
percent of P2O5, which is comparable to the experience with U.S. coals (Table 4).
• The U.S. coals have ash containing between 0 and 25 percent of CaO (Figure 12). The
maximum CaO in the ash of SCR installation coals is comparable at 26 percent (Table 4).
The above comparisons show that the characteristics of coals fired in SCR installations have
properties that are comparable to the properties of coals fired in the U.S. power plants. Even in
the case of Na2O and P2OS where the SCR installation coals have slightly lower concentrations,
the difference is not significant. The experience from existing SCR installations is, therefore,
directly applicable to applications firing U.S. coals.
-------
Conclusions
Based on the investigations in this study, the conclusion reached is that SCR technology has been
successfully applied to a large number of coal-fired boilers and that this experience supports the
feasibility of this technology for application on boilers firing U.S. coals. In addition to this
extensive experience with existing coal-fired applications, the following factors contributed to
this conclusion:
• The issues involved with coal-firing impacts on SCR components, mainly catalyst, are well
understood. The coal constituents, such as sulfur, ash, and arsenic, that can cause
deactivation of catalyst, if present in large amounts, have been identified. The mechanisms
of such deactivation are known and counteracting measures have been developed.
• The characteristics of coals fired in U.S. boilers do not differ significantly from the coals
being burned in existing worldwide SCR installations. Based on a comparison with limited
data available for the coals fired in SCR installations, the key constituents for the majority of
U.S. coals are at levels similar to those for the coals from SCR installations.
• Only 6 percent of U.S. coals has a sulfur content greater than 3 percent. There are SCR
installations firing coals with a sulfur content of 3 percent or slightly higher. In addition,
SCR has been successfully applied to other high-sulfur fuels, one of them being an
asphaltum-buming installation in Japan with a sulfur .content of 5.4 percent (greater than that
of any U.S. coal).
• Some of the important trace elements (arsenic, chlorine, fluorine, etc.) are present in higher
concentrations in the coals being burned in the foreign SCR installations, as compared to
U.S. coals. For some other trace elements (Na2O and P2O5), the concentrations in coals
being burned in foreign SCR installations are somewhat smaller. However, these trace
elements are considered weak poisons for SCR catalysts and the experience to date does not
list them as significant considerations for SCR.
• The SCR suppliers are aggressively marketing this technology for coal-fired applications.
The suppliers contacted for this study consider this technology to be fully commercial and
have shown their willingness to offer guaranteeable performance for SCR systems on boilers
firing any U.S. coals.
In summary, considerable experience has been gained to date from coal-fired SCR installations
in the U.S. and abroad. This experience adequately covers the operating conditions that can be
expected to be present when burning the coals fired in U.S. power plants.
Disclaimer
The views and opinions expressed here are those of the author(s) alone and do not necessarily
represent the policies of the U.S. Environmental Protection Agency.
-------
References
1. "Evaluation of SCR Technology for Application on U.S. Coal-Fired Boilers," Draft Report,
EPA Contract No. 68-D-2-0168, Work Assignment No. 5C-09, June 1997
2. "CoalPower 2," World Coal-Fired Power Stations and Their Pollution Control Systems, A
Database by IEA Coal Research, 1997
3. "Selective Catalytic Reduction (SCR) Controls to Abate NOX Emissions," A White Paper
Prepared by SCR Committee of Institute of Clean Air Companies, Inc., October 1996.
4. Rummenhohl, V., et al., "Relating the German DeNOx Experience to U.S. Power Plants:
Lessons Learned from More than 30,000 MW of DeNOx Retrofits," ASME Joint
International Power Generation Conference, Phoenix, Arizona, October 2-6, 1994.
5. "Papers presented at Kat-Treff USA '97 Seminar," by Siemens, Virginia, June 1997.
6. "SCR Technology Seminar," by Cormatech Inc., Mitsubishi, and STEAG, Durham, North
Carolina, March 25-26, 1997
7. Pritchard, S., et al., "Optimizing SCR Catalyst Design and Performance for Coal-Fired
Boilers," EPRI/EPA Joint Symposium on Stationary Combustion NOX Control, May 1995.
8. Balling, L., et al. of Siemens AG, "Poisoning Mechanisms in Existing SCR Catalytic
Converters and Development of a New Generation for Improvement of the Catalytic
Properties," EPRI/EPA Joint Symposium on Stationary Combustion NOX Control, March
1991.
9. Chen, J. and Chicanowicz, J. E., "Poisoning of SCR Catalysts," EPRI/EPA Joint Symposium
on Stationery Combustion NOX Control, March 1991.
10. "Trace Elements in Coal," Volume I, by Vlado Valkovic, Institute Ruder Boskovic, Zagreb,
Yugoslavia, 1st Edition.
11. "Cost and Quality of Fuels for Electrical Utility Plants, 1995 Tables," Energy Information
Administration, U.S. Department of Energy, July 1996.
12. "Coal Quality Database: Version 1.3," by U.S. Geological Survey, 1994.
13. "BACT Analysis Report on NOX Submitted to State of New Jersey for the Air Permit
Application for Cameys Point Station (Formerly Chambers Works Project), by ENSR
Consulting and Engineering, May 1990.
14. Information Obtained by Perrin Quarles Associates, Inc. from Plants in Germany, Submitted
to Bechtel in February and March, 1997.
-------
Table 1
SCR System Characteristics for Worldwide Installations
SCR Characteristic
SCR Experience1'
Coal type fired
Unit size range, MW
NOX reduction efficiency, %
Number of SCR units:
• With different boiler types(1):
Cyclone-fired
Pulverized-coal, dry bottom
Pulverized-coal, wet bottom
Vertical-fired
Pressurized, fluidized-bed
• In different countries:
Germany
Other European countries
Japan
Taiwan
U.S.
• With different types of SCR:
High dust
Low dust
Tail end
• With different reagents:
Anhydrous
Aqueous
• With different types of catalyst:
Honeycomb
Plate
Pellets
• With different types of installations:
New
Retrofit
Bituminous and lignite
13-1,050
35-90
19
130
44
6
3
121
38
42
4
7
120
30
43
173
13
106
54
3
54
156
Notes:
1. The dry-bottom boilers include tangential-fired, front wall-fired, and opposed wall-fired
boilers.
2. For some units, all of the information shown in this table was not available.
-------
Table 2
SCR Installations in the U.S.
Description
Location
Boiler type
Unit capacity, MW
Type of installation
SCR location
Catalyst type
Reagent: NFlj type
SCR supplier
NOX reduction, %
Birchwood
Virginia
PC-TF
240
New
High-dust
Plate
Anhydrous
ABB
50-70
Carneys
Point
New Jersey
PC-WF
2x140
New
High-dust
Honeycomb
Aqueous
FW
63
Indiantown
Florida
PC-WF
330
New
High-dust
Plate
Aqueous
FW
60
Logan
New Jersey
PC-WF
200
New
High-dust
Plate
Aqueous
FW
63
Mercer
Unit 2
New Jersey
Wet-bottom
321
Retrofit
High-dust
Plate
Aqueous
Wahlco
88
Merrimack
Unit 2
New Hampshire
Cyclone-fired
375
Retrofit
High-dust
Plate
Anhydrous
Noell
65
Stanton
Unit 2
Florida
PC-WF
465
New
High-dust
Plate
Anhydrous
Noell
47
Notes:
1. The legends used are as follows: PC-WF - Pulverized coal wall-fired; PC-TF - Pulverized Coal tangential-fired; FW - Foster
Wheeler
2. At Mercer, the catalyst is an in-duct type. Tests were also conducted on a combination of the in-duct catalyst and catalyzed air
heater baskets achieving a NOX reduction efficiency of 95 percent.
-------
Table 3
SCR-Related Fuel Properties, Their Impacts, And Solutions
Fuel Property
Ash
Sulfur (S03;T
Alkaline Metals
(Na and K)(2)
Alkaline Earth
Metal (Ca)(2)
Heavy Metals
(Asf>
Chlorine/
Fluorine
Potential Impacts
• Catalyst pluggage
• Surface deposits
• Erosion
• Catalyst surface masking
• Catalyst pluggage
• Oxidation of SO2 to SO3
• Air heater pluggage
• Fly ash contamination
Reduction in available catalyst
active sites
Catalyst surface masking
Catalyst surface masking
Catalyst surface masking
Solutions
• Use proper catalyst pitch
• Use sootblowers
• Use screens above catalyst to collect
popcorn ash
• Design SCR reactor for a uniform
flow distribution (use flow
straightening devices)
• Use catalyst with hardened leading
edges
• Use optimized pore structure
(maximize small pores)
• Use sootblowers
• Minimize catalyst's vanadium content
to minimize SO2 oxidation
• Minimize ammonia slip
• Provide proper catalyst volume
•" , Use economizer bypass to maintain
proper gas temperatures
• Provide proper catalyst volume
• Modify catalyst composition (add
tungsten)
• Provide proper catalyst volume
• Use sootblowers
• Provide proper catalyst volume
• Use optimized pore structure
• Use arsenic-resistant catalyst
• Provide proper catalyst volume
• Use sootblowers
• Use economizer bypass to maintain
proper gas temperatures
Notes:
1. During combustion, sulfur in coal oxidizes to S02 and a small quantity of SO3. The potential
impacts listed for sulfur are caused by SO3. The catalyst forms additional SO3 within the
SCR reactor by further oxidation of a small portion of SO2.
2. Other potential catalyst poisons include magnesium, lead, and phosphorus. These are not
considered important for SCR, since their concentrations in coal are small.
-------
Table 4
Properties of Coals Fired in SCR Installations
Coal Analysis
Trace metal, ppm:
Arsenic
Chlorine
Fluorine
Ash analysis, wt. %:
Si02
A1203
Fe203
Ti02
CaO
MgO
Na2O
K2O
P,0<
Range
2 - <301
40 - 1,900
47 - 270
33.9 - 76.7
2.0-33.0
0.83 - 15.4
0.55-1.8
0.11-26
0.07 - 49
.05-1.6
0.1-5.0
0-1.3
TABLES
SCR EXPERIENCE WITH 'HIGH SULFUR'
FUEL-FIRED BOILERS
Facility
Heme IV
(Germany)
Walsum9
(Germany)
Shimoneseki
(Japan)
Asahi Chemical Co.
(Japan)
Industrial Facility
(Japan)
Chubu Electric Power
(Japan)
Higashi Nippon Tetsudo
(Japan)
Takehara
(Japan)
Plant Crist, Gulf Power
(DOE Demons.)
Size
500 MW
450 MW
175 MW
2x80 tph
Steam
2x375
125 MW
250 MW
Slip
Stream
%
Sulfur
>3%
>3%
3%
5.4%
2.62%
2.5%
3.0%
2.5%
3.0%
Fuel
Coal
Coal
Coal
and Oil
Asphal-
tum
Orimul-
sion
Heavy
Oil
Heavy
Oil
Coal
US
Coals
Year
of
Ope
1989
1988
1980
1983
1987
1982
1981
1993
Comments
No corrosion or plugging
observed since 1989.
New Plant: 70% NOX removal.
Only two air heater washings
The facility cofires coal and
heavy oil.
The fuel has about 200 ppm of
vanadium; and 0.6% nitrogen.
The fuel has about 250 ppm of
vanadium; and 28-29% H2O.
The fuel has about 36 ppm of
vanadium.
Contains vanadium.
No corrosion, plugging or other
problems reported.
Two-year tests with coal
containing up to 3.02% sulfur.
-------
FIGURE 1
ASH IN COALS VERSUS NUMBER OF U.S. COAL-FIRED
PLANTS
l.egtnd. •/. Aih:
A=0-5%
B=5-IOV.
C=10-1S%
D = IS20'/.
K=202S%
F=25-30%
Minimum % Ash In Coalj
FIGURES
SULFUR IN COAIS VERSUS NUMBER OF U.S. COAL-FIRED
PLANTS
Ltgtnd. % Sulfur:
A=«1V.
Maximum % Sulfur in Coals
FIGURE 2
ASH IN COALS VERSUS NUMBER OF SCR-EQUIPPED COAL-
FIRED UNITS
um % Ash In Coals Burned Worldwide
FIGURE 4
SULFUR IN COALS VERSUS COAL TONNAGE SUPPLIED
TO U.S. PLANTS
Leecnd. % Sulfur:
A=0-I%
B=l-2%
C=2 3V.
D=3-5%
i % Sulfur In Coal
-------
FIGURE 5
SULFUR IN COALS VERSUS NUMBER OF SCR-EQUIPPED
COAL-FIRED UNITS
Maximum % Sulfur In Coals Burned Worldwide
FIGURE 7
CHLORINE IN COALS VERSUS COAL TONNAGE SUPPLIED
TO U.S. PLANTS
s
•z
H
1
O
Legend: Chlorine, ppi
A=0-I00
8=100-300
C=300-1000
A\crage ppm Chlorine in U.S. Coals
FIGURE 6
ARSENIC IN COALS VERSUS COAL TONNAGE SUPPLIED TO
U.S. PLANTS
l.cecnd: Anenlc. ppm:
A=0-S
B=5-10
C=10-J0
D=20-30
Average ppm Arsenic in U.S. Coals
FIGURE 8
FLUORINE IN COAL VERSUS COAL TONNAGE SUPPLIED TO
U.S. PLANTS
Ltecnd: Fluorine, ppm:
A-0-50
B=SO-100
O100.150
D=1SO-30S
Average ppm Fluorine in U.S. Coals
-------
FIGURE 9
Nii,O IN ASH VERSUS COAL TONNAGE SUPPLIED TO U.S.
PLANTS
Lteencl-Na.O. •/.:
A=0-O.S
B-05 1
C=l-2
D=2 5
Average % NajO in U.S. Coal Ash
FIGURE 11
P,0S IN ASH VERSUS COAL TONNAGE SUPPLIED TO U.S.
PLANTS
(A)
Average % P,O5 in U.S. Coal Ash
FIGURE 10
K,0 IN ASH VERSUS COAL TONNAGE SUPPLIED TO U.S.
PLANTS
Uecnd:
AKI-0.5
B=0.5 I
C=l-2
D=2-2.6
Average % KjO In U.S. Coal Aah
FIGURE 12
% CaO IN ASH VERSUS COAL TONNAGE SUPPLIED TO U.S.
PLANTS
H
•a
Uetnd: C»O. V.:
A=0-5
B=S 15
C=I5-2S
Average % CaO In U.S. Coal Ash
-------
SELECTIVE CATALYTIC REDUCTION (SCR) RETROFIT AT SAN DIEGO
GAS & ELECTRIC COMPANY SOUTH BAY GENERATING STATION
B.R. McLaughlin
Raytheon Engineers & Constructors
5555 Greenwood Plaza Blvd.
Englewood, CO 80111
E.A. Jones, Jr.
San Diego Gas & Electric
4949 Carlsbad Blvd.
Carlsbad, CA 92008-4302
E.G. Lewis
Babcock & Wilcox
P.O. Box 351
Barberton, OH 44203-0351
Abstract
San Diego Gas & Electric (SDG&E) retrofitted South Bay Generating Station Unit 1 with
selective catalytic reduction (SCR) technology to comply with San Diego County Air Pollu-
tion Control District (SDCAPCD) Rule 69 which regulates NOX emissions. South Bay Unit 1
is a 154 MW gas/oil fired B&W radiant boiler of late 1950s vintage. The SCR project also
included an aqueous ammonia storage and delivery system and SCR controls integrated in
a recently installed Distributed Control System (DCS). This paper discusses significant
aspects of the technical and economic evaluation, design, installation, and operational
characteristics of this NOX compliance project. A preview of SDG&E's compliance pro-
grams for South Bay and Encina Generating Station is also presented.
Compliance With Rule 69
NOX emissions reduction Rule 69 was adopted by the San Diego County Air Pollution
Control District (SDCAPCD) Board in January 1994 to meet California Clean Air Act re-
quirements for Best Available Retrofit Control Technology (BARCT) for electrical generat-
ing steam boilers. A Compliance Plan was submitted to the SDCAPCD in July 1994, and
updated in July, 1995 and July, 1996.
On December 12, 1995, SDCAPCD adopted amendments to Rule 69. The amendments
eliminated the individual unit emission rate limits, retained the two aggregate annual NOX
emission caps and added a third cap. The current Rule 69 NOX aggregate annual cap for all
SDG&E units is:
-------
January 1, 1997 to December 31, 2000 2100 tons/year
January 1, 2001 to December 31, 2004 800 tons/year
January 1, 2005 650 tons/year
Rule 69 also requires each unit to be operated at its maximum performance for NOX reduc-
tion. For South Bay Unit 1 with the new SCR, SDCAPCD requires a NOX limit of 20 ppmv
when firing natural gas and 40 ppmv when firing No. 6 fuel oil averaged over each 24 hour
calendar day of operation or portion thereof, excluding startup and shutdown.
Prior to Rule 69, NOX emission limits were governed by SDCAPCD Rule 68 limits that
represented Reasonable Available Control Technology (RACT). Rule 68 NOX limits are as
follows:
Natural Gas Fuel 125 ppm
Oil Fuel 225 ppm
NOX Compliance Strategy
Compliance with San Diego County Air Pollution Control District (SDCAPCD) Rule 69
required an overall systems analysis of the operating units within SDG&E's generation
system. The systems analysis was comprised of the following steps:
Design Basis Development - For each unit, an emissions data base was developed
and target emission levels "were established based on unit capacity factor, availability,
and contribution to the overall system emissions cap.
Screening of NOX Control Technologies - Identify candidate NOX control/removal
technologies and their current status in the industry. NOX control technologies con-
sidered included the following:
a. Combustion Modification, Low NOX Burners, Over Fire Air (OFA), Flue Gas
Recirculation (FGR)
b. Selective Non-Catalytic Reduction
c. Partial Selective Catalytic Reduction
d. Combinations of Above
e. Selective Catalytic Reduction
Technologies were then screened based on performance, constructibility, schedule, and
economics.
-------
Determine Performance and Plant Impacts - For each unit and the selected technol-
ogy, the emission performance and boiler performance impacts were determined.
Process flow diagrams and layout drawings for selected technologies were also devel-
oped to assist in evaluating these impacts and in development of total installed costs.
Economic Comparison - The economic analysis performed was based on the present
worth of revenue requirements (PWRR) methodology. PWRR requires a projection of
all costs associated with each technology over the service life, including capital costs,
operating and maintenance costs (reagent, catalyst replacement, labor, and power)
which are then converted to present worth over the selected life of the installation.
The PWRR of the selected technologies for each unit are shown on Figure 1. Cost
effectiveness was then evaluated on the basis of cost per ton of NOX removed, as
shown on Figure 2. This measure represents the PWRR divided by the tons of NOX
removed over the service life and is designed to show which projects have the most
removal at the lowest evaluated cost. The results are strongly influenced by the
capacity factor (CF); the units with the highest CF generally have the lower control
costs.
Risk Analysis - This analysis considered certain intangibles, such as capability for
higher NOX removal and ease of permitting; cost sensitivities such as catalyst cost
fluctuations, space velocity and projected catalyst life; and risk evaluations with
respect to performance, initial capital cost and levelized cost.
Based on the system-wide analysis, economic evaluation, and an assessment of risk
and cost sensitivities, the decision was made to proceed with SCR on Units 1 and 2 at
South Bay. Unit 1 was targeted for an in-service date of February 1997, with Unit 2 to
follow in 2000.
D CATALYST & MAINT
• NH3
H POWER
D FIXED CM
• FIXED SCR
EN1 EN2 ENS EN4 ENS SB1 SB2 SB3 SB4
UNIT ABBREVIATION
* SCR ONLY AT EN 1-3; SB 1,2 // SCR & LNB AT EN 4,5 // SCR & LNB & OFA & FGR AT SB 3,4
FIGURE 1 PWRR COMPARISON BY UNIT (MINIMUM REQUIRED EQUIPMENT*)
-------
Q
HI
O
HI
o:
X
O
O
EN1 EN2 EN3 EN4 ENS SB1 SB2
UNIT ABBREVIATION
SB3 SB4
'ALL UNITS ON A CONSISTENT TIME BASIS: 20 YEAR SERVICE LIFE.
= PRESENT WORTH OF REV REQ DIVIDED BY TOTAL TONS OF NOx REMOVED OVER LIFE
FIGURE 2 PRESENT WORTH CONTROL COST COMPARISON*
Project Organization
The organization of the project is illustrated in Figure 3. SDG&E acted as the overall
project coordinator, including taking the lead on licensing/permitting issues and interfac-
ing with station operating and maintenance personnel.
Raytheon Engineers and Constructors (Raytheon) in Denver, Colorado, provided support
in the development of the NOX compliance strategy and economic evaluations, design and
engineering services, administration of contracts, and construction management services.
Babcock & Wilcox (B&W) in Barberton, Ohio had turnkey responsibility for the SCR and
Ammonia Systems. Other separate contracts included 1) asbestos abatement, 2) DCS
modifications (Duke Engineering Services), 3) CEMS upgrades (Control Technology, Inc.),
and 4) balance of plant electrical work (Davies Electric).
SCR Process Overview
Selective catalytic reduction (SCR) is a highly effective method of post-combustion NOX
control •when high levels of NOX reduction are required or when low NOX emission levels at
the stack must be attained. SCR is a dry process in which ammonia (NH3) in vapor form is
injected into the flue gas stream upstream of the SCR catalyst layer as shown in Figure 4.
The ammonia acts as a reducing agent, and the NOX contained in the flue gas decomposes
into nitrogen and water vapor in the presence of the SCR catalyst. The catalyst is typically
-------
South Bay
Station
SDG&E
Raytheon
Project Engineer
•Asbestos Abatement
•DCS Integration
•Stack CEMS Mods
Licensing
B&W
•SCR System
•Ammonia System
• Electrical
• Demolition
• Station Tie-ins
FIGURE 3 SCR IMPLEMENTATION
Catalyst
I
Flue I
Gas I
i • >NO,
' • >NOX
^ X — * NH3
^•— »NO,
^ X— » NHs
^ • » NO,
^ X — » NH3
W///////////A
• >H2O
• »N2
•— H20 ca*
• »N:
• »H2O
FIGURE 4 PRINCIPLES OF NO,, REMOVAL PROCESS FOR SCR
a titanium-based material with a micropore surface which provides many active sites for
the NOX reduction to occur. Several reactions between the ammonia and NO and NO2
compounds can occur in the SCR process:
4NO + 4NH3
6NO
O2 ----- > 4N2
6H2O
8NH3 ----- > 7N2
NO + NO
2NH3
2N
12H2O
3H2O
Most SCRs operate within a temperature range of 450 to 840°F (232 to 449°C), with opti-
mum performance for natural gas firing occurring in the range of 650 to 750°F (343 to
399°C). Minimum SCR operating temperature varies based upon fuel type, flue gas compo-
-------
sition (including SO2) and catalyst formulation, and is key to SCR system performance.
The catalyst layer must be located strategically in the boiler gas train for proper tempera-
ture exposure and minimum pressure drop.
South Bay Unit 1
South Bay Unit 1 is a 154 MWe gross, pressure fired, B&W El Paso design, radiant boiler
with natural circulation. The primary fuel is natural gas with No. 6 fuel oil as backup.
There are eight (8) front wall mounted burners arranged in two (2) rows of four (4) burners
each. Flue gas recirculation is taken from the economizer hopper. The back-end arrange-
ment is parallel flow upward to two (2) vertical shaft, Ljungstrom type, regenerative air
preheaters. Rated steam flow is 980,000 Ib/hr at 2,150 psig and 1005° F / 1005° F steam
temperatures.
SCR Design Basis
Specifications for design of the SCR system required meeting performance criteria while
firing either natural gas or No. 6 fuel oil. Key conditions for the SCR design basis and
performance are shown in Table 1.
Table 1
SCR System Design Basis/Performance Conditions
Item Design Conditions
Fuel Natural Gas No. 6 Oil
Flue Gas Parameters
Flow Rate acfm1 646,924 627,406
Temperature °F 709 701
Composition at SCR Inlet
NO,, ppmdv2 150 225
O2 % Volume1- 1.80 2.30
SO2 ppmwv1- — 200
SO3 Increase % — 3.5
H2O % Volume1 18.52 11.81
Composition at SCR Outlet
NOX ppmdv2- 15 33
NH3 ppmdv2 10 2
DeNOx Efficiency % 90 85.3
Maximum Allowable
Pressure Drop Increase in.wg 3 3
Minimum Flue Gas Temp. °F. 500 570
1 Wet basis. 2- Dry volume basis corrected to 3% O2.
-------
Flow Model Testing
SCR performance can be significantly affected by nonuniform flow into the catalyst. There-
fore, catalyst manufacturers normally specify a maximum allowable nonuniformity of flue
gas velocity as a condition of guarantee. Additionally, the project specification limited the
allowable increase in system resistance to 3 in. wg. To satisfy these requirements a 1/10
scale cold flow model study was performed. The ammonia injection grids are immediately
downstream of the gas recirculation (GR) fan take-off and possible recirculation of flue gas
back to the GR fan was a concern. The study simulated four different boiler loads for each
fuel and considered the effects of a possible 60/40 side-to-side maldistribution of flue gas
flow from the boiler. Iterative testing determined the optimum number and location of flow
correction devices necessary for the required uniformity of velocity and also confirmed that
recirculation from the injection grids back to the GR fan would not be a problem. After an
acceptable uniform model flow was established, a tracer gas was used to confirm that the
ammonia injection grid design would provide satisfactory uniform distribution of ammo-
nia into the flue gas stream. As a result of the model test, flow corrective devices consisting
of splitter plates and perforated plates were added to the SCR inlet duct design as shown in
Figure 5.
SCR System Arrangement and Components
The SCR system consists of several subsystems including the SCR reactor and ducting,
catalyst, ammonia vaporization / flow control skid, ammonia distribution manifolds and
injection grids, an aqueous ammonia storage system, flue gas analyzers and a control
module in the DCS. Auxiliary equipment includes atomizing air compressors, dilution air
fans, interconnecting piping, system instrumentation, electrical power distribution equip-
ment and a tie to the existing plant uninterruptable power supply.
From parallel economizer outlets, the flue gas flows through parallel ducts into two low
pressure ammonia injection grids which are designed for optimal mixing of ammonia and
flue gas with minimal system resistance. The ammonia laden flue gas continues on into a
single reactor where it passes through a bed of parallel plate-type catalyst. Within the
catalyst bed the NOX reacts with the ammonia to form nitrogen and water vapor. After
exiting the reactor, the flue gas splits to pass through the parallel regenerative air heaters
before being discharged up the stack.
A simplified arrangement is shown in Figure 5.
SCR Reactor and Catalyst
A single reactor, arranged within the existing boiler structure, houses a single-stage, paral-
lel passage, horizontal flow bed of Babcock-Hitachi, titanium-based, plate-type catalyst.
Other catalysts were also considered. For this arrangement, pressure drop and cross-section
were critical and the plate-type catalyst offered the best solution for meeting performance
-------
requirements in a compact setting while minimizing pressure drop. The catalyst is con-
tained in 28 blocks in a 4 high by 7 wide by 1 deep arrangement that measures approxi-
mately 12 feet high by 44 feet wide by 6 feet deep.
The reactor is self-supporting and is constructed of structurally reinforced carbon steel
casing which is externally insulated. Personnel access doors and a removable catalyst
maintenance panel provide access to the interior. One monorail mounted, 3-ton electric
hoist is used to move catalyst blocks in and out of the reactor, while a second hoist is used
to position blocks within the catalyst bed. Seals are installed around the perimeter of the
catalyst bed to prevent flue gas from by-passing the catalyst.
Ammonia Vaporization and Flow Control
Aqueous ammonia, at 29.4% concentration, is supplied at high pressure to the skid
mounted flow control system by a common plant header. A pressure regulating valve
provides ammonia flow at constant pressure to a pneumatically actuated flow control
10
-------
valve which is controlled by a signal from the DCS. Downstream of the flow control valve,
the ammonia is injected into the vaporizer through a dual-fluid atomizer utilizing com-
pressed air for atomization. Due to a shortage of plant compressed air, atomizing air is
provided by stand-alone compressors dedicated to SCR service. In the vaporizer the liquid
ammonia is mixed with a hot air stream and is thereby vaporized and diluted to a safe
concentration. The hot air used for vaporization and dilution is extracted from the boiler's
hot secondary air ducts and pressure is boosted by one of two skid mounted fans. The
vaporization system is designed to maintain a minimum ammonia / air temperature of
180°F (82°C) to avoid condensation prior to injection into the flue gas. From the vaporizer,
the dilute ammonia/air mixture is conveyed through insulated ducting to the ammonia /
air injection system.
Ammonia/Air Injection System
The ammonia / air injection system is comprised of an external manifold / valve station
and four internal injection grids. The ammonia / air mixture is ducted from the vaporizer
to the manifold / valve station where the total flow is apportioned to 14 independently
controlled branches. Two grids, one horizontal and one vertical, are located in each of the
parallel ducts exiting the economizer hopper. Each of the grids is divided into separate
branches which sub-divide the flue gas duct into separately controllable zones. In this
manner, the total flow of ammonia may be distributed across the flue gas ducts to ensure
optimum SCR performance. A simplified arrangement of the internal grids is shown in
Figure 6.
AMMONIA FROM MANIFOLD
VALVE STATION (UVS)
"T
"T
FIGURE 6 SIMPLIFIED INTER INJECTION GRIDS
-------
Aqueous Ammonia Storage System
The selected reagent for the SCR is industrial grade aqueous ammonia at 29.4 % concentra-
tion. The aqueous ammonia storage and delivery system was designed to handle current
requirements for Unit 1 along with future requirements for Units 2 and 3. The arrangement
provides for a total of three (3) carbon steel storage tanks (two current and one future) at
22,500 gallons each. Total storage capacity is approximately 30 days usage for all three
units. The storage tanks are above ground and are surrounded by a concrete berm. All
ammonia deliveries are by truck. The truck unloading area is outside of, and adjacent to the
tank area. The truck unloading pad is equipped with a catch basin and drain line which
would divert any leakage from a truck into a large sump located within the berm. All
interconnecting piping and the supply pumps are located inside the berm while controls
are located outside. Emergency showers are located both inside and outside the berm. An
ammonia detector is centrally located within the tank area for both local and remote alarm
to the control room.
Two 100% capacity supply pumps in the tank area maintain a constant recirculation feed
loop through the storage tanks. This loop supplies a flow to a single supply line which
feeds a common plant header. The single supply line is equipped with its own electronic
leak detection system. In turn, the common plant header supplies aqueous ammonia to
Units 1, 2 and 3. Flow to each unit is independently controlled from this common header.
In an emergency situation, the ammonia supply system can be remotely shut down from
the control room. Otherwise, essentially all controls are local to the storage tank area.
A schematic of the ammonia storage and supply system is depicted in Figure 7.
ATOUIZINC AIR
FIGURE 7 AMMONIA STORAGE AND SUPPLY SYSTEM SCHEMATIC
-------
Control Systems
Figure 8 illustrates the basic controls setup and the coordination required to integrate the
SCR and Ammonia Systems into the Unit 1 & 2 Distributed Control System (DCS), which is
a Bailey Infi 90, as well as tie-ins to other control systems at the station.
CEMS
Shelter
DCS
Termination
Panel
SCR Interface
Panel
MCC
«— SCR
Instrumentation
«•— 480V
DCS
Cabinet
(I/Os, MFPs)
Ammonia System
Control Panel
Ammonia System
Safety Alarms
Operator
Interface
Station (OIS)
-*- Plant Annunciators
480V
Ammonia
System
Instrumentation
FIGURE 8 CONTROLS INTERFACE/COORDINATION
SCR System - In general, the controls of the SCR System have been established to provide
ammonia flow for a constant NOX reduction with an outlet NOX feedback. The amount of
ammonia introduced is controlled by two loops: 1) a feed forward loop which estimates
ammonia demand based on NOX removal stoichiomerric relationships, and 2) a feedback
loop which compares the outlet NOX reading against the desired outlet NOX and trims or
adds ammonia accordingly. The feed forward and feedback loops are programmed in the
DCS and will automatically control the ammonia flow control valve which is located on the
vaporization skid. This valve can also be controlled directly by the operator.
Logics programmed into the DCS for SCR control include:
• Ammonia/Air Mixture Temperature and Ratio Limits
• Ammonia Flow Control
• Operating Perrnissives
• Dilution Air Fan Control
• System Shutdown Logic
• Atomizing Air Control
As part of the overall SCR retrofit, the stack Continuous Emissions Monitoring Systems
(CEMS) was modified to accommodate the lower NOX emissions at the stack.
-------
Ammonia Storage and Feed System - Instrumentation and controls are provided for the
safe operation of the system. System layout allows for normal operation without requiring
personnel to enter the storage tank containment area. All controls are accessible from
outside the berm and all indicators are visible from the berirt wall. Controls were inte-
grated into the station in the following areas:
• Local Control Panel - pump control, local annunciation, and ammonia pipe leak
detection panel.
• DCS - equipment monitoring and remote trip.
• Unit Control Room - annunciation of safety related alarms.
• Unit 1 Auxiliary Operator Panel - system trouble alarms.
In summary, by coordinating the controls for the SCR and Ammonia Systems into the DCS
and other local panels, the system monitoring and control functions were assimilated into
the responsibilities of the existing station operating and maintenance personnel.
Installation and Startup
Similar to most retrofits at operating power plants, the approach was to complete as much
work as possible prior to the outage and carefully plan the outage work to ensure: 1) the
demolition and subsequent installation work can be performed within the time allotted,
and 2) the boiler can be returned to service successfully with minimal impacts to the opera-
tion of the existing boiler system. Careful coordination between all participants in the
project was needed. This required installation of the ammonia storage and feed system as
well as portions of the SCR electrical and control work prior to the outage. The SCR system
could only be installed during the outage. In summary, the major activities associated with
this approach are as follows:
Pre-Outage: Ammonia Storage and Feed System
• Earthwork/Foundations/Relocation of Buried Utilities
• Install Ammonia Storage Tanks, Pumps and Ammonia Feed Piping
• Process Electrical and I&C Including Tie-ins to Existing Items
• Verify System Integrity on Condensate
• Verify System Integrity on Aqueous Ammonia
• DCS Modifications for Ammonia and SCR System
Outage: SCR System
• Asbestos Removal and Abatement
• Demolition of Ductwork and Steel
• Install New Ductwork and Steel
-------
NATURAL GAS OIL
FIGURE 9 SCR PERFORMANCE TEST RESULTS
Next Step for Rule 69 Compliance
SDG&E is preparing for the competitive utility marketplace which begins in 1998. NOX
removal technologies will continue to be reviewed and evaluated to determine 'which are
most cost effective in meeting the requirements of Rule 69. In 1998, the Rule 69 SDCAPCD
Compliance Plan will be revised based on the evaluated NOX control technologies, competi-
tive environment influences and economics. SDG&E is committed to installing NOX reduc-
tion technology on a yearly basis to ensure meeting the SDCAPCD yearly system NOX
emissions cap.
In order to meet the Rule 69 compliance schedule, SDG&E has developed the preliminary
schedule in Table 2. Compliance activities have started with SCR installation at South Bay
Generating Station Unit 1. Afterwards, Encina Generating Station Units 4 & 5, South Bay
Units 2 & 3 and Encina Units 1, 2, and 3 will follow.
Year
1997
1998
1999
2000
2001
2002
2003
2004
Table 2
Unit
South Bay 1
Encina 4 & 5
Encina 4 & 5
South Bay 2
South Bay 3
Encina 1
Encina 2
Encina 3
Control
SCR
FGRW
LNB/FGRW
SCR
SCR
SNCR
SNCR
SNCR
-------
• Install Catalyst (Using Catalyst Hoist)
• Install all Process Equipment
• Electrical Tie-ins Including I&C Tie-ins to DCS and CEMS Upgrades/Modifications
• Inspect Furnace and all Flue Gas Passage Items
Startup and Testing: Operational Unit
• Startup on Condensate
• Initial Operation on Aqueous Ammonia
• System Check Out and Troubleshooting
• Adjustment of the Ammonia Injection Grid (AIG)
• Boiler Tuning for SCR and Overall NOX Removal Program
• Pre-Test in Preparation for First SCR Performance Test
SCR Performance Tests
The first performance test on the SCR System was concluded in early May, roughly three
months after initial operation of the SCR system. This test was performed while firing
natural gas and fuel oil.
An independent testing agency was contracted to perform the tests which measured the
performance of the following process guarantees:
NOX Outlet
Ammonia Slip
PM10 Increase
CO Increase
Ammonia Consumption
VOC Increase
Opacity
Flue Gas Pressure Drop
SO2 Increase
Auxiliary Power
SCR Minimum Inlet Temperatures
A comparison of inlet/outlet NOX values including ammonia slip is shown on Figure 9.
Summary
In summary, the project was completed within schedule and budget and after nearly six
months of operation, the emission requirements in Rule 69 are being met. Additionally, the
SCR System has been integrated into the plant DCS with minimal impact on operating and
maintenance personnel.
-------
ENGINEERING AND PILOT SCALE ASSESSMENTS OF A LOW COST
COMBINED LOW-NOX BURNER-SCR SYSTEM
D. P. Teixeira
Pacific Gas & Electric, Research & Development
San Ramon, CA
L. J. Muzio, T. C. Fang
Fossil Energy Research Corp.
Laguna Hills, CA
Kent Zammit
Electric Power Research Institute
Palo Alto, CA
Abstract
To date, compliance with strict NOX requirements has often dictated the use of large—and
expensive—SCR technology to achieve the necessary NOX levels. An alternative to these
expensive SCR reactors is the use of a technology strategy which combines several low cost
systems to achieve the necessary degree of control.
This paper describes one attractive combined system which, when applied to a large utility boiler,
has the potential for low overall installed cost. This system consists of low-NOx burners, in
conjunction with a "true in-duct SCR" The design concept minimizes or avoids structural
modifications, earthwork foundations, draft system alterations, and ductwork changes to
minimize the cost of the retrofit.
An engineering assessment of the true in-duct SCR concept was conducted utilizing a new
methodology which integrates physical cold flow modeling and SCR process modeling. The
engineering study addressed catalyst location as well as the location and design of the ammonia
injection system.
The EPRI/PG&E ASCR pilot plant at Morro Bay, California was then converted to a "true
in-duct" configuration, including a catalytic air preheater. The pilot tests provided a direct
validation of the engineering calculations.
Introduction
Faced with more stringent NOX emission regulations, the utility industry is moving in a direction
of utilizing Selective Catalytic Reduction (SCR), in addition to combustion modifications, as a
-------
way of achieving compliance. At the same time, efforts are being made to reduce costs by
exploring in-duct SCR arrangements. In-duct SCR arrangements incorporate the SCR reactor
basically along the existing flue gas path from the economizer to the air preheater.
In many cases, in-duct arrangements will involve enlarging the ductwork to lower velocities in
order to accommodate a larger volume of catalyst and reduce pressure drop. This in-duct concept
is contrasted to a more traditional retrofit approach associated with Japanese or German SCR
installations, where a large separate SCR reactor is built with fairly extensive ductwork changes
to route the flue gas to and from the SCR reactor. In the true in-duct SCR, there would be no
changes to the ductwork and a smaller amount of catalyst is inserted into existing ducts. The
avoidance of extensive structural modifications, foundations/earthwork, separate reactor vessels,
etc., and the smaller catalyst volume greatly reduce the cost of such SCRs.
True in-duct catalysts would experience relatively high velocities and somewhat poor distribution
of ammonia, therefore constraining NOX removal efficiency to about 60 to 80 %. However, when
used in conjunction with other NOX control techniques, such as FOR, low-NOx burners (LNB),
catalytic air preheaters, etc., the desired overall NOX removal can still be achieved and at a
significantly lower cost than a conventional SCR system.
While the in-duct arrangement can reduce retrofit costs, it also poses greater challenges to the
designer. In addition to the type and amount of catalyst, the process designer must also address
velocity and NH3/NOX uniformity across the catalyst face. In the traditional separate SCR reactor
approach, there is usually sufficient ductwork to allow uniform velocity to be generated at the
ammonia injection grid. A uniform velocity profile at the ammonia injection grid greatly
simplifies the ammonia/NOx mixing process. Likewise, with the separate SCR reactor, there will
be sufficient space to design ductwork expansions and turning vanes to meet velocity uniformity
requirements at the catalyst face. As a consequence, previous design approaches specified
criteria such as the standard deviation in the velocity profile, or NH3/NOX profile at the catalyst
face. If these criteria are met, then the SCR system will perform as designed.
With an in-duct arrangement, it may not be possible to: (1) provide a uniform velocity at the
ammonia injection grid, or (2) accommodate traditional design guidelines in terms of duct
expansion angles, etc. As a consequence, it will be more difficult, if not impossible in some
instances, to meet traditional criteria in terms of a velocity uniformity and/or NH3/NOX
uniformity at the catalyst face. However, this does not mean that an in-duct arrangement should
be discarded as a viable approach. Rather, it points to the need to develop better engineering
design tools to assess the performance of these systems.
This presentation describes a new methodology to support the design of SCR systems. The
method integrates cold flow modeling with an SCR process model to quantitatively account for
site-specific conditions. The paper will describe the methodology and a case study where the
methodology was used to design a true in-duct SCR. Pilot scale tests were then conducted in
conjunction with LNBs and a catalytic air preheater.
-------
SCR Process Design Methodology
Physical cold flow modeling has been used previously during the design of SCR systems. In
most cases, the cold flow modeling has focused on developing uniform velocities at the catalyst
face and/or at the plane of the ammonia injection grid, and reducing pressure drop. In some
instances, tracer gas techniques have been used to simulate the ammonia injection process. To
provide more input to the SCR design process, a methodology has been developed that integrates
cold flow modeling results along with a mathematical model of the SCR process to predict
performance for a given specific configuration.
As depicted in Figure 1, the process model integrates:
1. The cold flow velocity distributions, which define local variations in space velocity.
2. Tracer gas results which define local variations in NH3/NOX ratio, and
3. Ideal catalyst performance data (i.e., NOX removal and NH3 slip versus NH3/NOX and space
velocity with uniform velocity and NH3/NOX profiles) to calculate overall NOX removal and
NH3 slip for a given catalyst and NH3 injection configuration.
This integration of physical flow modeling with an SCR process model gives performance
predictions that are unique to a specific application. The model also estimates the pressure drop
across the catalyst. The basic methodology was validated at the PG&E/EPRI ASCR Pilot Plant1
in 1994.
Cold Flow Modeling
• Velocity Profiles
• NH3/NOX
Ideal Catalyst
Performance Data
A NOx = f (SV, T, NHSfNOx)
Site-specific Performance
Predictions
• NOx Removal
• NH3 Slip
• Pressure Drop
Figure 1. SCR Process Model Methodology
-------
Case Study: PG&E Morro Bay SCR Design
The methodology was used to perform a preliminary design for a true in-duct SCR for PG&E's
Morro Bay Unit 3. Morro Bay Unit 3 is a 345 MW gas-fired load-following unit with 24 burners
in an opposed wall-fired configuration. Figure 2 is a side elevation drawing of Morro Bay
Unit 3. For NOX control, the unit was previously retrofitted with FOR, overfire air (not shown),
and LNBs. The LNBs in Morro Bay Unit 3 are Babcock & Wilcox S-Burners, and their
optimization with FOR and OFA's currently provides NOX emission levels of below 40 ppm (dry
@ 3% O2) across the entire load range. The performance target for the true in-duct SCR was a
further reduction to 10 ppm NOX and NH3 slip.
The experimental approach for this case study was to use the PG&E/EPRI ASCR pilot plant as a
test bed for the Morro Bay Unit 3 true in-duct SCR design. This was possible because the ASCR
pilot plant is a 5000 scfm slip stream of Unit 3 (1% at full load), and the pilot SCR can be
evaluated directly with the existing combustion NOX controls on Unit 3.
Figure 3 shows two simplified views of Morro Bay Unit 3, with the potential ammonia injection
grid (AIG) and catalyst locations. Note that space between the economizer and the APH is
limited with a tight 90° turn, the FGR extraction opening, and a short length of duct. The logical
location for the catalyst was at the current location of a damper just upstream of the air preheater
(Catalyst Location 1). Similarly, the logical location for the AIG was either just downstream of
the 90° turn, or just downstream of the FGR take-off duct.
A 1/12 scale plexiglass model of the Morro Bay Unit 3 boiler was used for the cold flow
modeling phase of the design. The important issues addressed during the cold flow modeling
study included:
1. optimum location for the catalyst and AIG,
2. the velocity uniformity at the catalyst,
3. NH3/NOX uniformity at the catalyst, and
4. minimizing NH3 entrainment into the FGR stream.
Because the AIG is potentially located near the FGR intake (see Figure 3), a key design
consideration for this SCR arrangement was how to minimize ammonia entrainment into the
FGR flow. Such ammonia would return to the furnace and increase NOX production, which
would then increase NH3 demand. An engineering calculation showed that with uniform
distribution of ammonia at the FGR intake and full conversion of NH3 into NOX in the boiler,
the furnace NOX can increase by as much as 7 ppm.
Three approaches were considered for restricting ammonia entrainment by the FGR:
1. locating the AIG downstream of the FGR duct (i.e., Location 2 in Figure 3),
2. biasing the individual AIG injectors (i.e., turn off injectors near the FGR intake), and
3. using a baffle to separate the AIG from the FGR flow.
-------
Figure 2. Morro Bay Unit 3
-------
Figure 3. Potential Locations for the AIG and In-Duct Catalyst for Morro Bay Unit 3
Tracer gas studies eliminated approach one as there was a large recirculation zone that carried
NH3 from Location 2 back into the FOR duct. Further, approaches two and three would only be
feasible if the flue gas that entered the FGR duct was localized to the outer portions of the duct.
To assess the origin of the flue gas that enters the FGR duct, a point source of tracer gas was
added at various points across the exit of the 90° turn (AIG Plane 1). With tracer gas
measurements in the FGR duct, the fraction of main flue gas that enters the FGR duct from each
specific point could be calculated. The results of two such tracer gas tests are shown in Figures
4(a) and (b). Figure 4(a) is without turning varies and surprisingly shows that combustion
products entering the FGR duct are drawn from essentially the entire width of the flue gas duct.
This would preclude approaches 2 and 3 to limit NH3 entrainment into the FGR duct. However,
with the installation of the turning vanes (both vertical and horizontal vanes), the FGR is
localized to the outer edges of the duct.
With the turning vanes installed, further cold flow tests showed that Location 1 was the preferred
location for the AIG and catalyst. Also, biasing the AIG injectors was preferred over the flow
baffle to minimize NH3 entrainment into the FGR duct while maximizing NH3 at the catalyst
face.
Table 1 shows the results of the AIG bias tests. Note the tradeoff between minimizing NH3
entering the FGR with NH3/NOX uniformity at the catalyst. Two lances turned off at the outer
edges of the duct was the best case based on cold flow model results (Case 2). Subsequent pilot-
scale data showed a similar trend, but that additional throttling of injectors were needed at the
pilot plant to obtain similar NH3 entrainment fractions (Case 3).
-------
Side Near FGR Duel
(a) No Turning Vanes
o 5
Side Near FGR Duel
Center of Boiler
(b) With Turning Vanes
Figure 4. Origin of the FGR, View of Economizer Exit Elbow from APH
Table 1
AIG Optimization in Both the Cold Flow Model and the Pilot Plant
Physical Cold Flow Model Pilot Plant Results
Case
1
2
3
AIG Bias
None, all lances in service
Two lances turned off
Two lances off, two throttled
NHj/NO,
Uniformity
Std. Dev. %
14%
18%
N/A
Fraction of
NH3
in FGR
40%
9%
N/A
NH,/NOX
Uniformity
Std. Dev. %
14%
N/A
23%
Fraction of
NH3
in FGR
66%
27%
11%
-------
Following the cold flow study, the SCR process model was used to predict catalyst performance
for the Morro Bay system. Based on a 2.5 inches H2O pressure drop limit and a SV of
35,000 hr"1, the model found that a 3.9 mm pitch catalyst (3.3 mm hydraulic diameter) could
achieve 10 ppm NOX and NH3 if the initial NOX level is below 35 ppm. With the encouraging
results from the assessment study, the project proceeded to the pilot scale tests.
Case Study: Pilot-Scale Performance Tests
The PG&E/EPRI ASCR pilot plant was modified according to the true in-duct SCR design from
the cold flow process modeling study. Shown in Figure 5, the modified ASCR reactor is at
nominally 1/8 scale to Morro Bay Unit 3. Note that the pilot reactor models just one half of the
full-scale boiler, and incorporates all major components, including the FGR extraction duct,
turning vanes, and downcomer pipe.
It is also important to point out that the pilot plant featured a catalytic air preheater downstream
of the in-duct catalyst. The catalytic air preheater consisted of a small-scale, size 8HK
Ljungstrom® air preheater with catalyst-coated hot end elements. Figures 6 and 7 show
performance of the APH catalyst alone. As can be seen, load or space velocity has a larger
impact on performance than temperature.
Table 2 shows the test conditions for both the in-duct and APH catalysts for the true in-duct tests.
Overall True In-Duct Performance
The true in-duct catalyst performance was tested for two configurations of the AIG. One
configuration was Case 1 in Table 1 which had all of the injectors flowing equally (14% standard
deviation in NH3/NOX). The second configuration was the optimized AIG (Case 3 in Table 1)
which had the two injectors near the FGR take-off duct turned off and the next two throttled
down (23% standard deviation in NH3/NOX).
Figures 8(a) and (b) show that the in-duct catalyst alone and the combined in-duct and catalytic
air preheater elements both performed quite well. During these tests, the inlet NOX levels varied
from 26 to 38 ppm. The target outlet NOX level of 10 ppm was achieved by the in-duct catalyst at
NH3/NOX ratios of 0.72 - 0.95 with resulting NH3 slip levels of 3 ppm to 9 ppm. With the
combined in-duct and APH catalyst system, the target NOK reductions were achieved at NH3/NOX
ratios of 0.65 - 0.85 with corresponding NH3 slip levels of 1.3 to 3 ppm.
An interesting observation in Figure 8 is the effect that NH3/NOX uniformity had on the
performance of the in-duct catalyst. Despite the rather large difference in NH3/NOX uniformity,
(14% versus 23% standard deviation) there is a very small impact on performance in terms of
either NOX reduction or NH3 slip. In fact, these differences appear to be within experimental
uncertainty. While a general statement can be made that NH3/NOX uniformity has a greater
impact than velocity uniformity, the actual magnitude will depend on the site-specific situation.
-------
37.8"
Turning
Vanes
AIG
Downcorr
Pipe
Catalyst
45.0"
21.4"
16.8"
14.0"
11.8"
- I I
Catalyst
17.5"
AIG
24.5"
Figure 5. Modified Pilot In-Duct Reactor
The process model discussed previously was used to assess whether the results shown in Figure 8
are indeed realistic. The velocity profiles and the NH3/NOX profiles measured at the in-duct
catalyst face for the two AIG cases tested were input into the process model using the catalyst
conditions shown in Table 2. The results of the calculations are shown in Figure 9. In this
figure, the predicted NOX removal and NH3 slip for the two AIG configurations are shown along
with the line drawn through the experimental points plotted in Figure 8(a). Figure 9 supports the
pilot plant performance measurements in that changing the NH3/NOX uniformity by biasing the
injectors (i.e., increasing NH3/NOX nonuniformities from a standard deviation of 14% to 23%)
has little impact on NOX reduction for this particular application. The model predicts that at an
NH3/NOX ratio of 0.9 biasing, the injectors should only result in a one percentage point drop in
NOX removal (from 75% to 74%) and a one ppm increase in NH3/NOX slip (from 5 ppm to 6
ppm). For reference, if the NH3/NOX ratio and velocity were perfectly uniform across the
catalyst, the model predicts that at an NH3/NOX ratio of 0.9, the NOX removal would be 81% and
the NH3 slip 3 ppm.
-------
Hot End Awi. Metal Temperature:
628°F (baseline)
0.0
Figure 6. APH Catalyst Performance as a Function of Temperature
Figure 7. APH Catalyst Performance as a Function of Load
-------
Table 2
Pilot In-Duct and APH SCR Specification and Conditions
Manufacturer
Catalyst Type
Flue Gas Flow
Flue Gas Temperature
Inlet Velocity
AP Limit
Hydraulic Diameter
Pitch
Wall Thickness
Catalyst Dimensions
In-Duct
Haldor TopsgSe A/S
DNX-930
4200 scfin
680°F
27.6 ft/sec
2.5" H2O
3.3 mm
3.9mm
0.6 mm
27.5" x 45" x 19"
APH
ABB/Engelhard
FNC® Coating
4200 scfm
660°F
23.5 ft/sec
8 mm
0.7 mm
12" inner rad,
Space Velocity (SV)
29,100 1/hr
28" outer rad, 16"L
29,300 1/hr m
1. Based on the fraction of the air preheater catalytic elements actually exposed to the
flue gas flow (46%).
The pilot plant performance measurements and the model predictions show that for this specific
SCR arrangement, a fairly large change in NH3 nonuniformity had little impact on performance.
This clearly illustrates the potential shortcomings of using rule of thumb criteria for either
velocity or NH3/NOX uniformities. By employing the current methodology, actual performance
predictions can be made for site-specific SCR arrangements.
Figure 9 also represents a comparison of pilot plant measurements to model predictions on an
absolute basis. The solid line in Figure 9 is the line drawn through the data points in Figure 8
(Note: the NH3 slip levels in Figure 9 have been normalized to an initial NOX level of 35 ppm).
As can be seen, the model predictions, and the actual pilot plant measurements, are in good
agreement. Also, the measured pressure drop across the in-duct catalyst at a flow rate of 4200
scfm (SV=29,100 hf') was 2.02 inches H2O, the model predictions were 2.1 inches H2O.
-------
100 -,
90 -
80 -j
70 -
60 -
50 -
40 -
30 -
20 -
10 -
0
0.0 0.2 0.4 0.6 0.8 1.0
NH3/NOX, molar
(a) True In-Duct Catalyst Only
1.2
1.4
100
0 90-
> o
0.6 0.8
NH3/NOX
(b) Combined In-Duct and APH Catalysts
Figure 8. Pilot Scale Performance Results with Two Injection Bias Conditions
-------
100
80
CC
O
Model:AII hjectors Balanced
A- - - Model:2 Oft, 2 Throttled
Riot Rant Meas.
Closed Symbols: NO, Removal
Open Symbols: NH, slip
0.4 0.6 0.8 1.0 12 1.4 1.6
0.0 02
Figure 9. Comparison of Model Predictions with Pilot Plant Catalyst Measurements
(NH3 Slip Based on an Initial NOX Level of 35 ppm)
The results discussed above showed the performance of the true in-duct SCR and combined true
in-duct plus catalytic air preheater elements in terms of NOX reduction and NH3 slip. The next
step is to interpret these results in terms of the applicability of a true in-duct SCR in helping to
meet future NOX regulations. Since a true in-duct SCR will generally be used with other
combustion NOX controls, the basic question is, "How low must the NOX entering the SCR be
reduced?" To illustrate this interaction, the pilot plant performance results presented in Figure 8
have been generalized and replotted in Figure 10. In Figures 10(a) and (b), the outlet NOX and
NH3 slip levels have been plotted for various inlet NOX levels. With the results generalized in
this way, it is easy to assess the applicability of a true in-duct SCR. For instance, a future NOX
regulation may require 10 ppm outlet NOX with less than 10 ppm NH3 slip. As seen in Figure
10(a), with the in-duct catalyst alone, outlet NOX and NH3 slip levels of 10 ppm can be attained,
provided the inlet NOX levels are less than 40 ppm. For the combined in-duct/APH catalyst
system, the 10 ppm outlet NOX level can be easily achieved with a 40 ppm inlet NOX. In fact,
based on Figures 8(b) and 10(b), the 10 ppm outlet NOX and NH3 slip targets can be achieved
with inlet NOX levels of nominally 80 ppm with the combined in-duct catalyst and APH catalyst
system. It should be noted that the test results shown in Figure 8 represent catalyst performance
with nominally 1700 - 2100 hours of operation. Some margin would need to be included in a full
scale system to accommodate further catalyst aging. Also, the results are based on an allowed
pressure drop of 2 inches H2O across the in-duct SCR. With higher allowed pressure drop, more
catalyst could be utilized resulting in higher performance.
-------
Target Outlet NO, & NHj-slip: 10 ppm
0.2
0.4 0.6 0.8 1
NHa/NO,
(a) True In-Duct Catalyst Only
1.2 1.4
Inlet NOx= 30 ppmc
Inlet NOx= 20 ppmc
0.6 0.8
NH3/NOX
(b) Combined In-Duct and APH Catalysts
Figure 10. Generalized Pilot Scale Performance Results
-------
Summary and Conclusions
This paper has described a successful design and pilot-scale application of a low cost combined
LNB-SCR system. A design methodology was used which incorporates cold flow modeling as a
means to obtain site-specific velocity and ammonia uniformity information as well as resolve
complicated issues such as catalyst and AIG placement and minimizing NH3 entrainment into the
FOR stream. A process model predicted the NOX removal and NH3 slip as well as catalyst
pressure drop, and the predictions were verified at the Morro Bay ASCR pilot plant.
The case study of the design and application of a true in-duct SCR to Morro Bay Unit 3 showed
that:
1. ammonia entrainment into the FOR can be controlled by using turning vanes and biasing the
NH3 injection away from the FGR intakes,
2. increased ammonia non-uniformity due to injector biasing (Std. Dev. 14% to 23%) had little
impact on performance for this particular SCR system,
3. sub-10 ppm NOX and NH3 slip were possible with the Morro Bay Unit 3 true in-duct SCR
design if the inlet NOX remain at 40 ppm or lower,
4. combining a catalytic air preheater with the true in-duct catalyst gives extra operating margin
and allows for higher inlet NOX levels while maintaining sub-10 ppm outlet NOX and NH3.
In conclusion, this study has demonstrated the feasibility of a low-cost LNB-SCR system and the
utility of a methodology as a predictive design tool for cost-effective, high performance SCRs.
References
1. L. J. Muzio, et al., A New Design Tool for SCR Systems, 1995 Joint EPRI/EPA Symposium
on Stationary Combustion NOX Control, Kansas City, May 1995.
-------
FEASIBILITY OF APPLYING SELECTIVE CATALYTIC REDUCTION (SCR)
TO OIL-FIRED, SIMPLE- AND COMBINED-CYCLE COMBUSTION TURBINES
J. B. Jarvis
T. R. Carey
Radian International, LLC
8501 North Mopac Boulevard
Austin, Texas 78759
R. W. Frischmuth
Electric Power Research Institute
3412 Hillview Avenue
Palo Alto, California 94304
M. K. Yuen
Hawaiian Electric Company, Inc.
820 Ward Avenue
Honolulu, Hawaii 96814
Abstract
This paper contains the results of an EPRI Tailored Collaboration project to determine the
technical feasibility and cost of selective catalytic reduction (SCR) as applied to 0.4%-sulfur,
100% oil-fired combustion turbines. SCR has been widely applied on natural gas-fired
turbines; however, there is comparatively little commercial operating experience on oil-fired
combustion turbines, particularly when firing fuel oil containing up to 0.4% sulfur.
The project included two years of pilot testing on an exhaust gas slipstream from the Unit 14
LM2500 combustion turbine at the Maui Electric Company's Maalaea generating station. The
pilot unit included the SCR reactor and a heat-pipe heat exchanger designed to simulate the
operation of a full-scale heat recovery steam generator (HRSG). Two different types of
catalyst were tested. These included a low-temperature conventional SCR catalyst which
would be employed within the HRSG, and a high-temperature zeolite catalyst which would be
employed on simple-cycle units, or alternatively, upstream of the HRSG on combined-cycle
units.
The test results show that the key issues affecting the technical feasibility and cost of SCR
include catalyst performance and life in the combustion turbine exhaust gas environment and
HRSG pressure drop increases resulting from the deposition of ammonium sulfate and bisulfate
salts on heat transfer surfaces. This paper presents these test results and discusses their
influence on SCR costs.
-------
Introduction
New combustion turbine generating capacity has recently been added to the Hawaiian Electric
Company (HECO) generating system, and additional capacity will be needed in the near
future. In accordance with State and Federal regulations, the Prevention of Significant
Deterioration (PSD) permit applications and subsequent permits for these units addressed the
environmental control equipment which has or will be installed to control air pollutant
emissions. An unresolved issue associated with this permitting process was the control of NOX
emissions.
Recent permit decisions covering control of NOX emissions from U.S. mainland natural gas-
fired combustion turbines have been based on the use of dry low-NOx combustion technologies
anoVor selective catalytic reduction (SCR). These permit decisions have resulted in the
widespread application of SCR - mostly within the heat recovery steam generators (HRSG's) of
natural gas-fired, combined-cycle units.
Despite the widespread application of combustion turbine SCR, there is comparatively little
commercial operating experience on oil-fired combustion turbines. Over the past few years,
dual-fuel units with SCR have generally utilized oil-backup for only limited periods of time -
on the order of 400 hours per year or less. In addition, this oil-backup is usually very low-
sulfur fuel oil (less than 0.05% sulfur), which is widely available on the mainland.
The situation in the State of Hawaii is different in several respects. First, large quantities of
natural gas are not available in Hawaii, and all generating equipment in the HECO generating
system is fired exclusively on oil. Second, it is not cost-effective to bum very low-sulfur fuel,
and the fuel that is burned, as authorized by recent air permits, contains up to 0.4% sulfur.
This combination of factors raises concerns about the technical feasibility of SCR since SCR on
0.4%-sulfur, oil-fired combustion turbines has not been demonstrated on the mainland or
elsewhere. Third, there is limited excess generating capacity in Hawaii, and there is no grid to
provide power when demand exceeds the available generating capacity. Thus, HECO is
concerned about the reliability of SCR on 0.4%-sulfur, oil-fired combustion turbines.
Recently installed combustion turbines in the HECO generating system were permitted on the
basis of water injection for NOX control. However, the Hawaii Department of Health included
a permit condition which requires a demonstration program to determine the technical and
economic feasibility of SCR. The outcome of the demonstration program will assist in NO,
BACT determinations for a total of seven existing and planned HECO-system combustion
turbines.
To meet the state permit requirement, a two-year demonstration program was conducted using
an SCR pilot unit located adjacent to Unit 14 at the Maui Electric Company's Maalaea
-------
generating station. The first year of the demonstration program consisted of an evaluation of a
conventional vanadium-based catalyst which would be installed within a relatively low-
temperature region of a combined-cycle unit HRSG. The pilot unit equipment configuration
used during this portion of the demonstration program included a "surrogate HRSG" heat-pipe
heat exchanger designed to simulate the HRSG operating environment downstream of the SCR
reactor. The second year of the demonstration program consisted of an evaluation of a high-
temperature zeolite catalyst which would be employed on simple-cycle turbines, or
alternatively, on combined-cycle units upstream of the HRSG. Both catalysts were selected
based on the results of a bench-scale catalyst evaluation conducted prior to pilot testing.
The remainder of this paper presents an overview of the demonstration program results. First,
descriptions of the host combustion turbine unit and the SCR pilot unit are presented. Then, a
summary of key demonstration program results is presented which addresses the major factors
found to affect the technical and economic feasibility of SCR for HECO-system, oil-fired
combustion turbines. Finally, supplemental information is presented which supports the key
demonstration program conclusions. This information includes the results of an engineering
evaluation to predict full-scale HRSG operating impacts from the pilot-scale test results and
operating data quantifying the performance of an oil-fired, combined-cycle combustion turbine
with SCR located at Tekniska Verken's Garstad generating station in Linkoping, Sweden.
Even though the Garstad unit fires oil with a lower sulfur content than is fired in HECO-
system combustion turbines, it is the only unit in the world with comparable operating
conditions. Recent operating experience at this facility supports several of the key conclusions
from the demonstration program.
Description of the Host Unit and the SCR Pilot Unit
The pilot unit used to collect demonstration program data was located adjacent to Unit 14 at
Maui Electric Company's Maalaea generating station. Unit 14 includes a General Electric
LM2500 combustion turbine which typically operates at full-load (about 20 MW), although
unit output decreases to between 12 and 15 MW at night. Unit 14 is fired with No. 2 fuel oil
containing up to 0.4% sulfur. The exhaust gas SQ concentration at the maximum fuel sulfur
level is about 79 ppm; however, the SOj concentration during the pilot test program ranged
from 20 to 70 ppm, and averaged 40 ppm and 30 ppm during the low- and high-temperature
test periods, respectively.
Unit 14 is equipped with a Zum heat recovery steam generator (HRSG) that feeds a 16-MW
steam turbine. Steam produced from Unit 14 generates about 8 MW while steam from an
identical, parallel combustion turbine and HRSG (Unit 16) generates the remaining power.
HRSG's generally contains multiple tube banks of closely spaced, finned heat transfer tubes.
For the Unit 14 HRSG design, these tube banks include a high-pressure superheater, a split
high-pressure evaporator, a high-pressure economizer, and a low-pressure evaporator. Unit 14
was designed to allow the retrofit installation of SCR, and the split high-pressure evaporator
-------
contains a spool piece at a location where the exhaust gas temperature is appropriate for a low-
temperature SCR catalyst. A diagram of the Unit 14 HRSG is shown in Figure 1.
A diagram of the SCR pilot unit is shown in Figure 2. The pilot unit was designed to accept a
slipstream of turbine exhaust gas from two locations. During the first year of low-temperature
testing, exhaust gas was withdrawn using an isokinetic scope located within the split high-
pressure evaporator. The exhaust gas temperature at this location is typically about 65CPF.
During the second year of high-temperature testing, exhaust gas was taken from upstream of
the HRSG where the exhaust gas temperature is typically between 950 and 1,OOCPF, depending
on unit load.
As shown in Figure 2, the turbine exhaust gas passed through a venturi flow meter, a duct
expansion, and a perforated plate. The exhaust gas then passed through trim heating and
cooling equipment, and an ammonia injection grid. After a 90° turn, the exhaust gas passed
through a second perforated plate, an SCR reactor, and a heat-pipe heat exchanger. Finally,
the exhaust gas passed through an induced-draft fan and was returned to the Unit 14 stack.
The purpose of the two perforated plates was to straighten the gas flow after a flow
disturbance. The second plate also served to mix the ammonia and exhaust gas.
The SCR reactor contained two layers of catalyst. Engelhard's VNX catalyst was installed in
the reactor during low-temperature testing. The VNX catalyst is a conventional vanadium
pentoxide catalyst supported on titanium dioxide, and is suitable for use at temperatures
between 600 and 750°F This "washcoat" catalyst is supported on a ceramic honeycomb
substrate which, for this application, had a 3.175 mm pitch. Engelhard's ZNX catalyst was
installed during high-temperature testing. This catalyst is a zeolite which, according to
Engelhard, can be used at temperature up to 1,10CPF. This catalyst is also a washcoat catalyst,
and the catalyst installed in the pilot unit also had a 3.175 mm pitch.
During low-temperature testing, gas exiting the SCR reactor passed through a pilot-scale, heat-
pipe heat exchanger. This "surrogate HRSG" was designed to simulate the conditions in a full-
scale HRSG downstream of the SCR catalyst where ammonium salts are expected to deposit as
the exhaust gas cools. During high-temperature testing, the surrogate HRSG was removed and
replaced with a spool piece (this unit was not designed to handle the high-temperature exhaust
gas).
Summary of Results and Conclusions: Low-Temperature SCR Evaluation
Low-temperature testing was conducted using Engelhard Corporation's VNX catalyst. The
performance characteristics of the low-temperature catalyst over the one-year demonstration
period are presented in Figure 3. This figure shows NO, removal efficiency versus ammonia
slip as a function of time for data collected at baseline operating conditions (65CPF and a space
velocity of 10,300 hr"1). The ammonia slip data in Figure 3 are presented as percentages of the
-------
nominally 35 ppm inlet NOX concentration to normalize the data for variability in the inlet NOX
concentration.
The results in Figure 3 show that high NO^ removal rates were achieved with the VNX
catalyst. The ammonia slip concentrations were initially very low, but increased slightly over
the one-year test period due to deactivation of the catalyst. Catalyst activity projections based
on these test results, including results obtained at the reactor midpoint, indicate that it should
be possible to obtain 65% NOX removal with a maximum of 10 ppm ammonia slip over an
operating period (catalyst life) of about four years.
The results of SO2 oxidation measurements showed that S02 oxidation over the catalyst
averaged about 2% of the SO2 present in the exhaust gas. At the average SQ concentration of
40 ppm measured during low-temperature testing, 2% SO2 oxidation would increase the
exhaust gas SO3 concentration from about 1 ppm (in the turbine exhaust) to about 1.8 ppm
downstream of the SCR reactor.
Overall, the results show that reasonably good NO,, removal and SO2 oxidation performance
would be expected for the VNX catalyst. Even so, there would be sufficient ammonia slip and
SO3 in the exhaust gas to result in the deposition of ammonium salts such as ammonium sulfate
and bisulfate on heat transfer surfaces within the HRSG, particularly as the catalyst ages and
the ammonia slip begins to rise. Operating impacts associated with ammonium salt deposition
include an increase in the HRSG pressure drop, decreased heat transfer efficiency, and
corrosion of carbon steel heat transfer surfaces. An engineering evaluation based on the
demonstration program results indicates that increased HRSG pressure drop is the most
important of these impacts and has the greatest influence on the operation of the combustion
turbine and SCR-related costs. Specifically, HRSG pressure drop increases of about 3.3 inches
of water per month would be experienced on existing HECO-system units beginning in the
second year of a four-year catalyst life-cycle.
An ammonium salt management strategy would be required to prevent equipment damage
and/or a unit derate associated with an excessive HRSG pressure drop. Strategies evaluated as
part of the demonstration program included periodic water washing of the HRSG, more
frequent catalyst replacement, and HRSG sootblowing. Each of these strategies is discussed
below.
Water washing could be conducted to remove deposited ammonium salts when the HRSG
pressure drop reaches unacceptable levels (dictated by HRSG casing pressure limitations
and/or turbine backpressure limitations). Demonstration program results show that water
washing should remove the soluble ammonium salt deposits, although the effectiveness may
vary depending on the HRSG design and access to affected tube banks. Even if washing is
effective, however, frequent unit outages would be necessary to perform the washing
operation. For existing HECO-system units, calculations based on the demonstration program
results show that HRSG washing would be required at two-month intervals beginning hi the
-------
second year of a four-year catalyst life-cycle. Forced outages associated with the need for
periodic water washing increase the cost to implement SCR on the higher-sulfur, oil-fired
combustion turbines in the HECO generating system.
More frequent catalyst replacement could also be employed to reduce the impacts of
ammonium salt deposition. To implement this strategy, the catalyst would be replaced while
the ammonia slip was still relatively low. This minimizes the deposition rate of ammonium
salts in the HRSG. For existing HECO-system units, annual catalyst replacement would
reduce the need for HRSG washing to once per year (i.e., during an annual outage). However,
relative to the water washing strategy, higher catalyst procurement and disposal costs would
largely offset savings associated with the reduced number of forced outages.
Sootblowers were also considered as a means of preventing the buildup of ammonium salt
deposits and minimizing the frequency of unit outages. However, there is limited operating
data quantifying the effectiveness of sootblowers for removing these types of deposits from
HRSG heat transfer surfaces, and the data which are available indicate that sootblowers are not
effective. Specifically, sootblowers were installed and operated on an oil-fired, combined-
cycle combustion turbine with SCR located at Tekniska Verken's Garstad generating station in
Linkoping, Sweden. Even though this unit fires a lower-sulfur fuel oil than is used hi HECO-
system units, rapid increases in HRSG pressure drop have been experienced at Garstad, and
sootblowers have not been effective in removing these deposits or limiting the rate of pressure
drop increase.
Operating problems associated with ammonium salt deposition increase the cost of SCR.
Table 1 summarizes cost estimates developed as part of the demonstration program for the
various ammonium salt management strategies. For existing HECO-system units, costs
associated with the water washing and annual catalyst replacement strategies more than double
the annualized cost of SCR relative to a hypothetical, similarly-sized natural gas-fired
combustion turbine.
Costs associated with the sootblowing strategy could be somewhat lower if sootblowing
effectively removed ammonium salt deposits (the cost estimates in Table 1 are based on the
premise that sootblowers will completely prevent the deposition and accumulation of
ammonium salts). However, the operating experience at Garstad demonstrates that
sootblowers are not effective. Therefore, the costs listed in Table 1 for the sootblower
management strategy underpredict actual costs, and the costs listed for the water washing and
annual catalyst replacement strategies more accurately represent the actual costs to implement
SCR on HECO-system combustion turbines.
Summary of Results and Conclusions: High-Temperature SCR Evaluation
High-temperature testing was conducted using Engelhard Corporation's ZNX catalyst. The
performance characteristics of the high-temperature catalyst over the one-year demonstration
-------
period are shown in Figure 4. Again, the figure shows NOX removal efficiency versus
ammonia slip (as a percentage of the inlet NOX) as a function of time for data collected at
1,000°F, and space velocities of 20,600 and 10,300 hf1 (these space velocities correspond to
the reactor midpoint and outlet, respectively).
Relative to the low-temperature VNX catalyst, the high-temperature ZNX catalyst performance
shows lower NOX removal capability with significantly higher ammonia slip. Calculations
show that the activity of this catalyst is only about one-third of that for the conventional low-
temperature catalyst, and the catalyst deactivation rate is somewhat greater. The results show
that this catalyst could be used to remove up to 65% of the NOX hi the exhaust gas with less
than 10 ppm ammonia slip. However, the required catalyst volume would be greater than for a
low-temperature application with the same performance requirements, and the catalyst life
would be only about one year.
In addition to poor NOX removal performance, the high-temperature ZNX catalyst oxidized
excessive amounts of the SO2 in the exhaust gas to SO3. At a typical exhaust gas SQj
concentration of 40 ppm, and an ammonia injection rate giving 65% NOX removal, about 20%
of the inlet SO2 was oxidized to SO3 (this corresponds to the production of about 8 ppm SQ).
When the ammonia injection system was turned off, about 50% of the inlet SQ was converted
to SO3 (this corresponds to the production of about 20 ppm SO,).
The demonstration program results indicate that the use of the high-temperature ZNX catalyst
on a simple-cycle combustion turbine would generate environmental issues associated with
sulfuric acid emissions or emissions of ammonium salt particulate. In addition, the cost would
be high due to the large catalyst volume and short catalyst life (costs for a simple-cycle
application are included in Table 1). Further, a combined-cycle application is considered
technically infeasible because of the excessive deposition of ammonium salts in the HRSG
downstream of the SCR catalyst. It is apparent from the test results that further advances in
catalyst technology would be required before high-temperature SCR could be applied to higher-
sulfur, combined-cycle combustion turbines.
Engineering Evaluation of HRSG Operating Impacts
In a full-scale, combined-cycle combustion turbine SCR application, ammonium salts will
condense on the HRSG's finned heat transfer tubes if ammonia and sulfur trioxide (SQ) are
present hi the cooling exhaust gas (ammonium salts will begin to condense from the exhaust
gas when the temperature falls below approximately 450T). Condensed ammonium salts
deposit on the fins and foul the area between the fins. This restricts gas flow over the tubes
and results in an increase hi the HRSG pressure drop and a decrease in heat transfer efficiency.
HRSG pressure drop increases appear to be the more important of these two effects, and will
eventually force a shutdown of the combustion turbine to avoid equipment damage and/or a
unit derate. Figures 5 and 6 show clean and fouled heat transfer tubes, respectively, from the
surrogate HRSG used during the demonstration program.
-------
Ammonia is present in the exhaust gas due to unreacted ammonia (ammonia slip) exiting the
catalyst, and the ammonia concentration would increase over time as the catalyst ages. Thus,
catalyst performance and life will directly influence the deposition rate of ammonium salts in
the HRSG. The S03 concentration in the exhaust gas also affects the ammonium salt
deposition rate. S03 is present in the exhaust gases of sulfur-bearing fuels, and additional SO,
is produced via S02 oxidation as the exhaust gas passes through the catalyst.
Surrogate HRSG performance changes were quantified by monitoring pressure drop increases
and heat transfer decreases over time. These performance characteristics changed over time
due to deposition of ammonium salts such as ammonium bisulfate and ammonium sulfate on
the finned heat transfer tubes. The mass deposition rates of ammonium salts were also
quantified during pilot testing by periodically washing the surrogate HRSG and measuring the
concentrations of ammonia and sulfate in the wash solution. Over the one-year, low-
temperature test program, the ammonium salt deposition rate and corresponding impacts on
surrogate HRSG performance were measured at ammonia slip concentrations ranging from 1.0
to 8.4 ppm.
Mass transfer, heat transfer, and pressure drop models were developed to evaluate the
demonstration program data and to allow projections of full-scale operating impacts. The
specific objective of this evaluation was to determine the required washing frequency and
forced outage schedule for the Unit 14 HRSG over a four-year catalyst life-cycle. To begin
this analysis, a mass transfer model was developed to predict the mass deposition rate of
ammonium salts as a function of temperature and the concentrations of ammonia and SQ. The
model was based on the assumption that the gas phase is at equilibrium with respect to
ammonium bisulfate such that the deposition rate is controlled by diffusion of ammonia and
SO3 to the tube surface. Reasonably good agreement was obtained between the modeling
results and the mass deposition rates measured in the surrogate HRSG.
In addition to the mass transfer model, heat transfer models were developed to provide exhaust
gas and tube temperatures throughout the surrogate HRSG, and pressure drop models were
developed to predict pressure drop changes as a function of gas velocity and the mass of
deposited ammonium salts. Good agreement was obtained between the model predictions and
the surrogate HRSG performance data.
The models developed to evaluate the demonstration program data were modified for
application to a full-scale HRSG and used to predict performance changes in the Unit 14 HRSG
over a four-year catalyst life-cycle. An example of the results of this modeling process are
described below for a case in which the unit is operating at full-load, the NO, removal
requirement is 65%, and the fuel sulfur content is 0.2% (this fuel sulfur content is
representative of the fuel used during the demonstration program, but is only half of the
maximum possible 0.4% fuel sulfur content which could be supplied to HECO-system
combustion turbines).
-------
To begin this analysis, ammonia slip over the four-year catalyst life-cycle was estimated from
the demonstration program data and from information provided by the catalyst vendor. The
results of this analysis are shown in Figure 7. As shown in this figure, ammonia slip is
initially very low and rises with time as the catalyst ages. After four years, the ammonia slip
reaches 10 ppm (consistent with the projected four-year catalyst life). Next, the mass transfer,
heat transfer, and pressure drop models were applied to estimate the ammonium salt deposition
rate, and the corresponding changes in heat transfer efficiency and HRSG pressure drop, on a
month-by-month basis.
The Unit 14 HRSG has a maximum casing pressure rating of 20 inches of water. The "clean"
HRSG pressure drop is about 7 inches of water, and the estimated catalyst pressure drop for
the baseline reactor design is about 5.8 inches of water. Thus, the incremental pressure drop
increase resulting from ammonium salt deposition cannot exceed 7.2 inches of water (i.e., 20 -
7-5.8 = 7.2 inches of water). To provide a safety margin, it was assumed that the unit
would be shut down to remove deposits via water washing when the incremental pressure drop
approached or exceeded a 5.2-inch threshold (i.e., 2 inches of water below the maximum
allowable increase). It was further assumed that the washing process would reduce the
incremental pressure drop to zero by removing all accumulated ammonium salt deposits.
The results of this analysis are shown in Figure 8. This figure shows the incremental HRSG
pressure drop due to ammonium salt deposition for the first 24 months of the four-year catalyst
life-cycle. During the first year, the incremental pressure drop increases slowly because the
ammonia slip from the catalyst is low (the ammonia slip concentration is lower than the SO3
concentration and thus controls the ammonium salt deposition rate). As a result, the first
HRSG wash is not required until after 12 months of operation, and washing reduces the
incremental pressure drop to zero. Subsequent pressure drop increases occur more rapidly as
the ammonia slip reaches and then exceeds the SO, concentration, and the required washing
frequency increases. By the end of the second year, HRSG washing is required at two-month
intervals, and the pressure drop is increasing at a predicted rate of 3.3 inches of water per
month. The pattern at the end of the second year continues through the third and fourth years
of the four-year catalyst life-cycle. The predicted washing frequency does not increase in these
later years because the ammonia slip concentration has risen above the SO, concentration, and
the ammonium salt deposition rate is now controlled by the SOi, concentration in the exhaust
gas.
A forced outage lasting approximately three days would be required to clean the HRSG. For
the HECO generating system, this affects operating costs in two ways. First, less efficient
units would be employed to make up for lost power production during forced combustion
turbine outages. Second, the need for frequent cleaning effectively derates these units, and the
installation of future generating capacity would be required at an accelerated rate.
-------
SCR Operating Experience at the Garstad Generating Station
Tekniska Verken operates an oil-fired, combined-cycle combustion turbine with SCR in their
Garstad generating station in Linkoping, Sweden. Even though this unit is fired with oil
containing less sulfur than is used in HECO-system units, it is the only unit in the world with
comparable operating conditions (i.e., 100% oil-fired with an appreciable fuel sulfur content).
Operating problems have recently been experienced in this unit which are consistent with
predictions developed from the demonstration program results.
The unit at Garstad began operations in October, 1994. During the first two years of
operation, the unit was fired with oil containing an average sulfur content of 0.07%.
Ammonium salt deposits were found in the HRSG after about 1.5 years of operation, but these
deposits had not caused significant operating problems.
During the third year of operation, the unit was fired with oil containing 0.13% sulfur due to
the unavailability of lower-sulfur fuel oil. During a 2.5-month period in the fall of 1996, rapid
increases in the HRSG pressure drop were experienced which forced a reduction in unit load
followed by a forced outage to clean the HRSG.
Operating data covering the third year of operation at Garstad are shown hi Figure 9. This
figure shows the average daily unit load and the HRSG pressure drop (HRSG, catalyst, and
ammonium salt deposits) versus time. Higher-sulfur oil firing began in early October. Over
the next 2.5 months, the HRSG pressure drop increased at a rate of about 4.8 niches of water
per month. In December of 1996, unit load was reduced due to the excessive HRSG pressure
drop, and the HRSG was washed in early January, 1997.
The operating experience at Garstad validates the demonstration program methodology used to
predict HRSG operating impacts. The delayed onset of operating problems and the magnitude
of the pressure drop increases observed at Garstad are consistent with the predictions
developed from the demonstration program results.
The HRSG at the Garstad unit is equipped with five sets of steam-fired sootblowers. During
the first two years of operation, these sootblowers were fired at a frequency of once per week.
When the increases in HRSG pressure drop became apparent, the sootblower operating
frequency was increased to once per day. According to Tekniska Verken, the sootblowers had
no apparent effect on HRSG pressure drop or the rate of pressure drop increase. Since the
sootblowers were ineffective, and their use resulted in reduced unit generating capacity, the
sootblower operating frequency was subsequently reduced to once every three days.
-------
HP Economizer LP Evaporator
Figure 1
Diagram of the Maalaea Unit 14 HRSG
MECO Unit 14
Exhaust
High-Temperature
Take-Off
Low-Temperature
Take-Off
Plate
Cooling Air Injection
Flue Gas Heater
Ammonia Injection
0000
• Air
- Ammonia/Air
Heat Pipe Heat
Exchanger
or Open Duct
Spool Piece
Figure 2
SCR Pilot Unit System Configuration
-------
100
95
35
r so
o
85
80
75
• September 1994
• December 1994
* March 1995
v June 1995
• August 1995
4 6
Ammonia Slip (% of Inlet NOx)
10
Figure 3
Low-Temperature Catalyst Performance at Reactor Outlet (3100 scfrn, 650°F)
100
— 75
50
25
10 20 30 40 50 60 70
Ammonia Slip (% of Reactor Inlet NO,)
80
Figure 4
High-Temperature Catalyst Performance (3100 scfm, 1000°F)
90
-------
Figure 5
"Clean" Hot-End Tubes (taken after cleaning the tubes for the first time)
Figure 6
First Cold-End Tubes After Third Long-Term Test)
-------
•a
a.
a.
4-Year Catalyst Life
at 10 ppm Slip
2 -
234
Operating Time (years)
Figure 7
Projected Effect of Low-Temperature Catalyst Deactivation on Ammonia Slip
o
£
e
a.
O
v>
a:
I
Maximum Allowable Pressure Drop Increase
Washing Threshold Pressure Drop Increase
0 2 4 6 8 10 12 14 16 18 20 22 24
Operating Time (months)
Figure 8
Predicted HRSG Pressure Drop Profile vs. Time
-------
30,
25
= 20
15
£
a.
u
in
10
• Pressure Drop (Load >85%) I
O Daily Average Oil Flow
O
Sc?
"&„
(9
(9 _
Sol
° °i1
Begin high sulfur
(0.13%) fuel oil
combustion
Conduct second
HRSG wash
3-Sep
3-Oct
100
80 5.
8
Q.
60 1
2-Nov
2-Dec 1-Jan
Date (1996-1997)
31-Jan
2-Mar
40
20
0
1-Apr
Figure 9
Garstad HRSG Pressure Drop and Unit Load (Oil Flow) Data
Table 1
Demonstration Program SCR Cost Estimates
Cost Case
Capital Cost
Annualized Cost
Low Temperature SCR
Hypothetical natural gas-fired, 4-year $1,456,000 $482,000
catalyst life.
Oil-fired, 4-year catalyst life:
- Sootblower management strategy $2,996,000 $969,000
Water washing management strategy $2,719,000 $1,212,000
Oil-fired, annual catalyst replacement $2,719,000 $1,253,000
High-Temperature SCR
Oil-fired, simple-cycle, 1-year catalyst life $3,491,000 $1,652,000
Oil-fired, combined-cycle Not feasible Not feasible
-------
ECONOMIC ANALYSIS OF SELECTIVE CATALYTIC REDUCTION
APPLIED TO COAL-FIRED BOILERS
J. F. Cox, E. C. Healy, J. D. Maxwell W. S. Hinton
Southern Company Services, Inc. W.S. Hinton & Associates
P.O. Box 2625 2708 Woodbreeze Drive
Birmingham, Alabama 35202 Cantonment, Florida 32533
J. Stallings A. L. Baldwin
Electric Power Research Institute U.S. Department of Energy
P.O. Box 10412 Federal Energy Technology Center
Palo Alto, California 94303 P.O. Box 10940
Pittsburgh, Pennsylvania 15236
Abstract
This report presents the results of an economic evaluation produced as part of an Innovative
Clean Coal Technology project, which demonstrated selective catalytic reduction (SCR)
technology for reduction of NOX emissions from utility boilers burning U.S. high-sulfur coal.
The project was sponsored by the U.S. Department of Energy (DOE), managed and cofunded by
Southern Company Services, Inc. (SCS), on behalf of Southern Company, and also cofunded by
the Electric Power Research Institute (EPRI) and Ontario Hydro. Six worldwide catalyst
suppliers and major equipment suppliers also participated with technical and financial
contributions to the project. The project was located at Gulf Power Company's Plant Crist
Unit 5 (75-MW tangentially fired boiler) near Pensacola, Florida. The test program was
conducted for approximately two years to evaluate catalyst deactivation and to quantify
operational impacts of SCR technology employed in a high-sulfur flue gas environment. The
SCR test facility included nine parallel reactors equipped with commercially available catalysts
ranging in size from 400 scfm (680 Nm3/hr) to 5000 scfm (8500 Nm3/hr). Results of the
economic evaluation indicate that total capital requirements ($/kW) decrease with increasing unit
size for both new and retrofit SCR applications. However, capital costs are approximately
50 percent higher for retrofit installations when compared with new plant applications. Lower
levelized costs ($/ton) are achieved when SCR is applied to larger, higher utilized units due to
economies of scale and the fact that a greater amount of tons are removed on larger units. In all
SCR applications, higher inlet NOX levels significantly decrease the levelized cost.
Introduction
There are several regulatory and environmental drivers in various stages of consideration that
may increase the likelihood of employing SCR technology hi the future. Recent experience of
-------
applying SCR to new coal-fired installations has created regulatory precedent under New Source
Review, which will affect future best available control technology (BACT) and lowest
achievable emission rate (LAER) determinations for other new units. With one exception, these
new installations are owned and/or operated by independent power producers (IPPs) who report
that adopting SCR technology was necessary to obtain construction and/or operating permits
quickly.
The 1990 Clean Air Act Amendments (CAAA) mandated several NOX control requirements and
regulatory reviews to reduce NOX emissions from utility boilers. Application of SCR to existing
boilers is being considered for units located in areas designated under Title I (nonattainment
provisions) for attainment of the ambient ozone standard. Recent efforts by the Ozone Transport
Assessment Group (OTAG) have focused on NOX reduction strategies on a broader scale,
encompassing 37 states east of and bordering the Mississippi River. Results of the OTAG
review may increase the likelihood for retrofits of SCR technology, particularly if emission
averaging and NOX trading are allowed. Additionally, nationwide reductions in NOX mandated
under Phase II of Title IV (acid rain provisions) may be required. In order to meet these
additional NOX reduction requirements, utilities are given flexibility to select the most suitable
and cost-effective NOX control technologies for their situation.
Application of SCR Technology for a New Unit
The economic evaluation presented in this section is based on the application of a high-dust, hot-
side SCR (located between the boiler economizer outlet and the air preheater inlet) at a new coal-
fired facility. The technical design premises used to prepare the economic analysis were selected
to be representative of actual or anticipated plant configurations and NOX control requirements
currently being permitted or likely to be permitted on new coal-fired boilers in the U. S.
Therefore, defining assumptions were selected in an effort to have broad utility applicability.
250-MW Base Case Unit Description
The base case represents a new, base-loaded
250-MW pulverized coal power plant typical
of the majority of new coal-fired projects
currently under development, construction, or
recently declared in commercial operation.
The 250-MW plant size is consistent with
current and future capacity trends of new
domestic power plants. The plant is located
in a rural area with minimal space
limitations. The fuel is a high-sulfur
bituminous Illinois No. 6 coal having an
analysis shown in Table 1.
Utilizing current generation low-NOx
combustion systems, the boiler is assumed to
Table 1 : Coal Analysis Used
Proximate Analysis DP
Ash
Volatile Matter
Fixed Carbon
Moisture
Total
Ultimate Analysis Dr
Carbon
Hydrogen
Nitrogen
Sulfur
Chlorine
Oxygen
Ash
Water
for Economic Evaluation
i Basis As Received
9.30 %
37.88 %
52.82 %
100.00%
y Basis
74.82 %
5.00 %
1.58%
2.58 %
0.16%
6.56 %
9.30 %
Total 100.00%
Higher Heating Value 13,265 Btu/lb
8.39 %
34.16%
47.65 %
9.80 %
100.00 %
As Received
67.48 %
4.51 %
1 .43 %
2.33 %
0.14%
5.92 %
8.39 %
9.80 %
100.00%
12,500 Btu/lb
-------
produce NOX at an emission rate of 0.35 Ib/MBtu. For purposes of this study, it is assumed that
tangentially fired boilers and wall-fired boilers are interchangeable with respect to thermal
performance and flue gas constituents. The flue gas exits the boiler and enters a single, hot-side
SCR reactor. Flue gas flow is vertically downward through the reactor to the air preheater. The
air preheater is typical of what is commercially offered as a "deNOx" air preheater. The SCR is
designed as a universal reactor capable of accepting either, or both, plate- or honeycomb-type
catalysts. Anhydrous ammonia is used as the reagent. Assumptions used to prepare the material
balance for the base case are shown in Table 2. General design criteria for the SCR are shown in
Table 3. Where applicable, the design criteria reflect operational lessons learned from the test
facility and/or current utility industry trends in post combustion NOX control.
Table 2: 250-MW Base Case Material Balance and Combustion
Calculation Assumptions
Unit capacity (gross)
Capacity factor
Type of installation
Boiler type
Heat input
Coal feed
Gross plant heat rate
Type of air preheaters
Number of air preheaters
Air preheater outlet temperature
Air preheater leakage
Excess air @ boiler outlet
250 MW
65%
New facility
Wall-fired or tangentially fired
2,375 MBtu/hr
190,000 Ib/hr
9,500 Btu/kW-hr
Vertical shaft, Ljungstrom
One
300°F
13%
18%
Table 3: SCR Design Criteria
Type of SCR
Number of SCR reactors
Reactor configuration
Initial catalyst load
Required range of operation
NOX concentration @ SCR inlet
Design NOX reduction
Flue gas temp @ SCR inlet
Flue gas pressure @ SCR inlet
Design ammonia slip
Guaranteed catalyst life
SO2 to SO3 oxidation
Maximum pressure drop
Velocity distribution
Ammonia distribution
Temperature distribution
Hot-side
One
3 catalyst support layers + 1 dummy layer
2 of 3 layers loaded, 1 spare layer
35% to 100% boiler load
0.35 Ib/MBtu
60%
700°F
-5 in. W.G.
5 ppm
2 years (16,000 hours)
0.75% (initial catalyst load)
6 in. W.G. (fully loaded reactor)
AV / Vmem < 10% over 90% of reactor area
AV / Vmean < 20% over remaining 10% area
AC/Cmean<10%
A T < 10°C max deviation from mean
Initial Space Velocity and Catalyst Volume
Space velocity is a process variable that is used in determining the quantity of catalyst required
for a given NOX removal requirement. Space velocity is defined as the volume of flue gas
-------
treated per unit volume of catalyst. The standard convention for expressing flue gas flow rate is
in ft3/hr (m3/hr) corrected to conditions of 32°F (O°C) and 1 atmosphere (1 bar). Catalyst
volume is expressed in corresponding units of ft or m Thus, space velocity can be expressed:
Flue Gas Flow, ft3/hr (m3/hr)
sv= :
Catalyst Volume, ft3 (m3)
For this analysis, a relationship between space velocity and NOX removal has been determined
for both new and retrofit units. The relationship for new units is represented by a least squares
curve fit of space velocities taken from five new coal-fired SCR installations in the U.S. Design
information was assembled from commercial bid evaluations, project specific design criteria, and
publicly available sources. The relationship for retrofit units was developed using a least squares
curve fit of test facility data measured during parametric testing and each catalyst supplier's
proposed space velocity based on the test facility steady state design criteria. The relationship
does not represent a single catalyst supplier's offering, but rather a composite of all catalyst
space velocities. This approach was selected to provide a reasonable method for estimating
space velocity, which is independent of catalyst geometry.
Catalyst Life and Catalyst Management Plan
The term "catalyst life guarantee" is often misinterpreted to mean the performance of the SCR
sharply decreases and the entire volume of catalyst must be replaced after the guarantee period.
This interpretation is not correct. In general, catalyst performance during early project years (or
months) normally exceeds the guarantee values. Over time, the catalyst performance will
deteriorate gradually until the SCR is unable to maintain the required NOX removal while
simultaneously achieving the required ammonia slip. (Most SCR installations operate on a
constant NOX removal to allow continued operation with permitted NOX emissions at the
expense of increased ammonia slip.) Even though the SCR cannot meet guaranteed ammonia
slip versus NOX performance, the catalyst still has considerable activity remaining.
The SCR reactor for this evaluation includes space for three catalyst layers plus a flow
straightener. Initially, only two of the three catalyst layers are loaded. The third empty layer
allows catalyst suppliers to develop optimized management plans to improve catalyst utilization.
A fresh catalyst layer can be added to the reactor after the guarantee period when ammonia slip
begins to exceed the guaranteed limit. The activity of the new catalyst combined with the
residual activity of the existing catalyst restores the performance of the SCR and extends the next
addition or replacement outage beyond the original guarantee interval.
During the demonstration at Plant Crist, catalyst deactivation data were periodically measured by
taking catalyst samples from the test facility reactors and returning the samples to the respective
catalyst supplier. The catalyst suppliers performed a standard protocol of laboratory and bench
scale tests to develop an activity versus time relationship. The base case catalyst management
plan (Figure 1) was derived using these data. A least squares curve fit of catalyst relative activity
(k/ko - ratio of activity to initial activity) over time resulted in a value of approximately 0.80
after 16,000 hours of service.
-------
The catalyst management plan is based on
a 16,000-hour (2-year) catalyst life
guarantee period. After the initial
guarantee period of 2 years, a new layer
of catalyst is added to the reactor spare
layer, taking advantage of the residual
activity in the initial layers to boost the
performance of the SCR. The next
addition of catalyst is required in project
year 6, when one of the initial layers is
replaced. After project year 6, staged
replacement of catalyst layers occurs
approximately every three years.
Economic Premises
s.
JS
2 1
a o
o o
;: c £ c £ £ K
KKKKKKK
Minimum NOx Removal Activity
5 10 15 20 25
Project Year
Figure 1: Catalyst Management Plan
30
Table 4: Economic Evaluation Factors
The base case economic evaluation includes total capital requirement, fixed and variable
operating costs, and levelized costs for a new 250-MW pulverized coal utility boiler. The
economics are presented on both a current dollar basis, which includes the effect of inflation, and
a constant dollar basis, which ignores inflation.
The methodology used to calculate the
economic factors is consistent with guidelines
established by EPRI in their Technical
Assessment Guide (TAG). The economic
parameters assumed for this evaluation are
representative of typical domestic utility
financing (Table 4).
Current Dollar Analysis:
Capital Charge Factor
O&M Cost Levelization Factor
Constant Dollar Analysis:
Capital Charge Factor
O&M Cost Levelization Factor
0.150
1.362
0.116
1.000
Cost Methodology
The capital cost methodology must reflect all utility costs incurred (including incremental costs)
and address a complete scope of supply for a commercial SCR system. For example, the
differential cost of an ID fan for a unit without SCR compared to a unit with SCR is seldom
assessed against the SCR scope of the project. This differential cost, while real to the utility, is
more commonly assessed to either a fan or draft system account that does not fully capture the
incremental economic impact to balance-of-plant systems due to the SCR. The capital cost
estimates prepared for this economic evaluation include incremental cost adders applicable to
new facilities brought about by the addition of SCR.
Fixed O&M costs include estimates of operating labor, maintenance labor, administration or
support labor, and maintenance material. Variable O&M captures the cost of all commodities as
well as costs of expendables such as anhydrous ammonia, catalyst addition/replacement, and
utilities. Variable O&M also includes the boiler efficiency penalty incurred due to increased
APH outlet gas temperature. Because variable O&M costs are dominated by catalyst
replacement, the catalyst management plan is one of the most significant factors affecting overall
-------
costs of SCR technology. Table 5 presents the assumptions used to calculate the fixed and
variable O&M costs for the base case evaluation.
Table 5: Fixed and Variable O&M Assumptions
Anhydrous ammonia cost $250/ton
SCR catalyst cost $400/ft3
SCR catalyst escalation 3.0%
Power cost 30 mills/kWh
ID fan efficiency 75%
SCR draft loss (fully loaded reactor) 3.0 in. W.G.
Ductwork draft loss 0.75 in. W.G.
Ammonia injection grid draft loss 0.75 in. W.G.
Unrecoverable air preheater draft loss 1.0 in. W.G.
Operating labor man-hour rate $23.00/hr
Maintenance factor (% of total process capital) 2.0%
250-MW New Unit Base Case
The 250-MW base case (60 percent NOX removal efficiency) results are summarized in Table 6.
The total capital requirement for a new SCR installation was estimated at $54/kW or
$13,415,000 in 1996 dollars. Total first year O&M is $1,045,000 in 1996 dollars. The initial
catalyst charge accounts for approximately 21 percent of the total process capital for the SCR
installation. From an O&M perspective, the total catalyst added over the project life is
approximately 61 percent of the variable O&M and 43 percent of the total annual O&M cost,
reiterating the fact that O&M costs are dominated by catalysts.
Compared with other NOX reduction alternatives, the higher capital costs of SCR dominate the
levelized cost. For the 250-MW base case, the capital cost is 59 percent of the current dollar
total levelized cost, indicating a major portion of the levelized cost goes toward debt service
(revenue requirement) of the capital investment rather than operating costs.
Table 6: Capital, O&M, and Levelized Cost for New SCR vs. Unit Size
(60 Percent NOX Removal Efficiency)
Unit Size (MW)
Total Capital Requirement
Total Capital Requirement
First Year Fixed Operating Cost
First Year Variable Operating Cost
Current Dollar Analysis
Levelized Cost (mills/kWh)
Levelized Cost (S/ton)
Constant Dollar Analysis
Levelized Cost (mills/kWh)
Levelized Cost ($/ton)
125-MW
$7,602,000
S61/kW
$213,000
$ 367,000
2.89
$2,811
2.09
$2,037
Base Case
250-MW
$13,415,000
$54/kW
$312,000
$ 733,000
2.57
$2,500
1.85
$1,802
700-MW
$31,327,000
$45/kW
$614,000
$2,053,000
2.22
$2,165
1.59
$1,547
-------
Sensitivity Analyses for New Unit Applications
(Effect of Variables on Economics)
Sensitivity analyses were performed to examine the impact of major process variables on the
capital, O&M, and levelized cost of SCR technology applied to a new unit. Results from
sensitivity cases examined as part of this evaluation are discussed in the following sections.
Capital, O&M, and Levelized Cost for New SCR vs. Unit Size
In order to examine the variation in SCR costs with unit size, capital and O&M estimates were
calculated for a 125-MW unit and 700-MW unit (Table 6). To maintain consistency with the
base case unit, an SCR removal efficiency of 60 percent NOX reduction was assumed for each
scenario. Where possible, consistent or identical assumptions were made with regard to the
125-MW and 700-MW units.
When plotted in $/kW vs. unit size, the total capital requirement of the SCR system decreases unit
cost with increasing unit size, indicating significant economy of scale. Total capital requirement
ranges from $61/kW for the 125-MW unit to $45/kW for the 700-MW unit (Figure 2). The
levelized cost decreases with increasing unit size due to larger NOX tonnages removed; however,
the trend is not overly sensitive to unit size (Figure 3).
80
£70
01
£
I50
40
Based on New Unit
with 60% Removal
4,000
§ 3,000
S
ui
O 2,000
•a
Q>
"5
o 1,000
Based on New Unit
with 60% Removal
Current Dollar
Constant Dollar
0 100 200 300 400 500 600 700 800
Unit Size (MW)
Figure 2: Total Capital Requirement vs. Unit Size
100 200 300 400 500 600 700 800
Unit Size (MW)
Figure 3: Levelized Cost vs. Unit Size
Capital, O&M, and Levelized Cost for New SCR vs. Capacity Factor
In many instances SCR costs are determined assuming a capacity factor of 65 percent.
Frequently, however, the capacity factor of the SCR may be less than that of the overall unit, or
the unit may be dispatched at a level lower than anticipated. In practice units equipped with SCR
may only utilize ammonia injection at or above certain loads. At lower loads, many units emit
very low levels of NOX eliminating the need to utilize the SCR. For those units that must meet
daily emissions limits, it may be possible to turn off the ammonia injection once the daily limit
has been achieved. Therefore, in effect, these practices may decrease the SCR capacity factor.
-------
With this in mind, the variation of SCR
costs with capacity factor was determined
(Figure 4). O&M and levelized cost
estimates for capacity factors of 30 and
90 percent are compared with the base
case (65 percent) in Table 7. Results
indicate increasing the capacity factor
beyond 65 percent has a relatively small
impact on levelized cost; however, as the
capacity factor drops below 60 percent,
costs increase rapidly. For example, a 20
percent reduction in capacity factor from
the base case increases the current dollar
levelized cost from $2500/ton to
$345 I/ton (a 38 percent increase).
6,000
•£•5,000
•K 4,000
o
o
I 3,000
•3 2,000
1,000
Based on New Unit
with 60% Removal
0.9
0.2 0.3 0.4 0.5 0.6 0.7 0.8
Capacity Factor (%)
Figure 4: Levelized Cost vs. Capacity Factor
Table 7: O&M, and Levelized Cost for New SCR vs. Capacity Factor
(250-MW Plant and 60 Percent NOX Removal)
Capacity Factor (%)
Base Case
30 65 90
$266,000 $312,000 $346,000
$581,000 $733,000 $842,000
First Year Fixed Operating Cost
First Year Variable Operating Cost
Current Dollar Analysis
Levelized Cost (mills/kWh) 5.13 2.57 1.96
Levelized Cost ($/ton) $4,994 $2,500 $1,907
Constant Dollar Analysis
Levelized Cost (mills/kWh) 3.69 1.85 1.41
Levelized Cost ($/ton) $3,593 $1,802 $1,377
Capital, O&M, and Levelized Cost for New SCR vs. NOX Removal Efficiency
NOX removal cases for 40 percent and 80 percent were calculated to examine the impact of NOX
removal efficiency on levelized cost. For these cases, capital, O&M, and levelized costs versus
NOX removal efficiency for a 250-MW plant size were determined (Table 8 and Figure 5). The
levelized cost decreases with increasing NOX removal percentages and is fairly sensitive to the
percentage removal. Thus, once committed to an SCR, significant levelized cost savings ($/ton)
can be realized for an incremental increase in capital cost. The difference in capital cost between
40 percent and 80 percent removal is $ 1,168,000 or approximately a 9 percent increase in capital
cost over the 40 percent design. However, the corresponding difference in current dollar levelized
cost is $1466/ton, a 42 percent decrease from the 40 percent case. This difference is primarily due
to doubling the number of tons removed by increasing NOX removal from 40 to 80 percent. This
trend in lower levelized cost is also very evident in high NOX emitting boilers where similar NOX
removal designs (as a percentage) yield lower $/ton due to a larger number of tons removed.
-------
Table 8: Capital, O&M, and Levehzed Cost for New SCR vs. NOX Removal Efficiency
(250-MW Plant Size)
Total Capital Requirement
Total Capital Requirement
First Year Fixed Operating Cost
First Year Variable Operating Cost
Current Dollar Analysis
Levelized Cost (mills/kWh)
Levelized Cost ($/ton)
Constant Dollar Analysis
Levelized Cost (mills/kWh)
Levelized Cost ($/ton)
NOX
40%
S 12,974,000
$52/kW
$305,000
$621,000
2.39
$3,502
1.74
S2,536
Removal Efficiency
Base Case
60%
$13,415,000
$54/kW
$312,000
$733,000
2.57
$2,500
1.85
$1,802
80%
$14,142,000
$57/kW
$324,000
$857,000
2.79
2,036
2.00
$1,460
4,000
3,000
O 2,000
TJ
.§
®
» 1,000
Based on New 250 MW Unit
Constant Doll
4,000
o 3,000
to
O 2,000
•o
I
01
o 1,000
Based on New 250 MW Unit
with 60% Removal
Current Dollar
Constant Dollar
30
40
80
50 60 70
NOx Removal (%)
Figure 5: Levelized Cost vs. NOX Removal Efficiency
90
0.2 0.25 0.3 0.35 0.4 0.45 0.5
Inlet NOx (Ib/MBtu)
Figure 6: Levelized Cost vs. Inlet NOX
Levelized Cost for New SCR vs. Inlet NOX Concentration
At many new boiler installations, difficult decisions are made on how to best optimize overall
NOX reduction requirements using a combination of a low-NOx combustion system and SCR.
While maximizing combustion NOX reductions can allow lower SCR variable O&M, it may have
a negative impact on thermal cycle efficiency due to increased unburned carbon in the flyash.
Increased carbon monoxide production may also be a limiting factor during deep staged
combustion. Optimizing the combustion system to minimize unbumed carbon can lead to higher
NOX concentrations entering the SCR and, therefore, higher variable O&M costs to achieve a
permitted outlet NOX emission limit. The relationship between levelized cost ($/ton) and SCR
inlet NOX concentration indicates a significant trend of increasing levelized cost with decreasing
inlet NOX concentration (Figure 6 and Table 9). In this case, fewer tons of NOX are removed by
the SCR, highlighting one of the key differences in levelized cost between a controlled new unit
application and an uncontrolled (or higher NOX emitting) retrofit application.
-------
Table 9: Levelized Cost for New SCR vs. Inlet NOX Concentration
(250-MW Plant Size and 60 Percent NOX Removal Efficiency)
Inlet NOX Concentration (Ib/MBtu)
0.25 0.30 0.35 0.40
Current Dollar Analysis
Levelized Cost (mills/kWh)
Levelized Cost ($/ton)
Constant Dollar Analysis
Levelized Cost (mills/kWh)
Levelized Cost ($/ton)
2.53 2.55
$3,446 $2,894
1.82 1.84
$2,483 $2,086
2.57
$2,500
1.85
$1,802
2.59
$2,205
1.87
$1,590
0.45
2.61
$1,977
1.88
$1,425
Levelized Cost for New SCR vs. Catalyst Relative Activity
For this evaluation, three catalyst management plans were formulated based on a 16,000-hour
(2-year) catalyst life guarantee period by correlating deactivation data from the Plant Crist
demonstration. Because the relative activity data indicate a wide variation in values as well as
the fact that each catalyst supplier extracted an unequal number of catalyst samples at different
time intervals over the test period, three sets of individual catalyst data were identified for further
evaluation. From the demonstration data, catalyst management plans were developed for
sensitivity evaluation based on relative activity (k/ko) data having values of 0.90, 0.80, and 0.70
after 16,000 hours. This range sets the limit for the upper and lower bounds of the relative
activity variation and appears reasonably plausible and equally likely to occur given the scarce
amount of data beyond 8,000 hours and the differences in catalysts.
20
T3
ffi
0.15
Q>
"§10
WKo=0.70 at 2 yre
k/ko=0 80 at 2 yrs
k/ko=0 90 at 2 yrs
Using the relative activity data, catalyst
management plans were developed which
define the catalyst replacement schedule
over the project life. From the management
plan, the total number of catalyst layers
installed or replaced in each year over the
life of the project was determined
(Figure 7).
Knowing the volume of catalyst, as well as
the time which it is added, project cash
flows were developed for each scenario.
The total catalyst volume installed between
the most optimistic (k/ko = 0.90) and most
pessimistic (k/ko = 0.70) management plans varies by 21,664 ftJ (a factor of three times). This
translates to a catalyst cost difference of $377,000 per year, representing a 64-percent difference
in annual O&M dollars (Figure 8). Additionally, if catalyst disposal costs were included, the
cost difference would be more pronounced since the k/ko = 0.70 plan would require the disposal
of three times as much catalyst. The current dollar levelized cost varies only $373/ton or a
25
30
0 5 10 15 20
Replacement Year
Figure 7: Number of Catalyst Layers Installed or Replaced
.3
10
-------
difference of only 14 percent between the two extreme cases (Figure 9 and Table 10). Even
though the k/ko = 0.70 plan adds three times as much catalyst, the catalyst is added in later
project years, which has less effect when performing a present value analysis and levelizing to
calculate the equivalent annual catalyst cost.
800
-700 -
reoo
8
0500
«
I1400
I
0300
a
1200
c
41100
Based on New 250 MW Unit
with 60% Removal
Constant Dol
4,000
o 3,000
in
O 2,000
•O
I
a
5 1,000
Based on New 250 MW Unit
with 60% Removal
Constant Dollar
0.6 0.7 0.8 0.9
Relative Activity (k/ko)
Figure 8: Catalyst Cost vs. Relative Activity
0
0.6 0.7 0.8 0.9 1
Relative Activity (k/ko)
Figure 9: Levelized Cost vs. Relative Activity
Table 10: Levelized Cost for New SCR vs. Catalyst Relative Activity (Catalyst Management Plan)
(250-MW Plant Size and 60 Percent NOX Removal Efficiency)
Total Catalyst Volume Added or Replaced Over
the Life of the Plant (ft3)
Equivalent Annual Current Dollar Catalyst Cost
Equivalent Annual Constant Dollar Catalyst Cost
Current Dollar Analysis
Levelized Cost (mills/kWh)
Levelized Cost ($/ton)
Constant Dollar Analysis
Levelized Cost (mills/kWh)
Levelized Cost ($/ton)
k/ko = .90
10,832
$216,000
$144,000
2.33
$2,269
1.72
$1,671
k/ko = .80
24,372
$450,000
$325,000
2.57
$2,500
1.85
$1,802
k/ko = .70
32,496
$593,000
$433,000
2.71
$2,642
1.93
$1,881
Levelized Cost for New SCR vs. Catalyst Life
In addition to the previous scenarios, three additional catalyst management plans were developed
by varying the year in which the third layer of catalyst is initially added. These plans represent
better than expected (add 3rd layer at year 3) and worse than expected (add 3rd layer at year 1)
results and assume a relative activity of k/ko = 0.80 after 1, 2, and 3 years. Total catalyst volume
and levelized cost for the three different management plans are summarized in Table 11. In the
worst-case scenario (3rd layer added at year 1), 22 layers of catalyst were installed or replaced
over the project life (Figure 10). In contrast, only 8 layers were installed or replaced in the
best-case scenario (3rd layer added at year 3). In terms of levelized cost on a current dollar basis,
the two extreme cases vary by $674/ton (Figure 11).
11
-------
,20
Q.
01 -1C
^ ! 0
i_
O
•O
CD
Table 1 1 : Levelized Cost for New SCR vs. Year to Add 3rd Catalyst Layer
(Catalyst Management Plan)
(250-MW Plant Size and 60 Percent NOX Removal Efficiency)
Total Catalyst Volume Added or Replaced Over
the Life of the Plant (ft3)
Equivalent Annual Current Dollar Catalyst Cost
Equivalent Annual Constant Dollar Catalyst Cost
Current Dollar Analysis
Levelized Cost (mills/kWh)
Levelized Cost ($/ton)
Constant Dollar Analysis
Levelized Cost (mills/kWh)
Levelized Cost (S/ton)
k/ko = .80
after 1 yr
51,452
$965,000
$686,000
3.09
$3,011
2.12
$2,065
k/ko = .80
after 2 yrs
24,372
$450,000
$325,000
2.57
$2,500
1.85
$1,802
k/ko = .80
after 3 yrs
16,248
$285,000
$217,000
2.40
$2,337
1.77
$1,724
k/ko=D 80 at 1 yr
Wko=0 80 at 2 yrs
Wt
-------
Table 12. Capital, O&M, and Levelized Cost for New SCR vs. Catalyst Price
(250-MW Plant Size and 60 Percent NOX Removal Efficiency)
Total Capital Requirement
Total Capital Requirement
First Year Fixed Operating Cost
First Year Variable Operating Cost
Current Dollar Analysis
Levelized Cost (mills/kWh)
Levelized Cost ($/ton)
Constant Dollar Analysis
Levelized Cost (mills/kWh)
Levelized Cost ($/ton)
Catalyst Price ($/ft3)
$350 $400 S450
$13,040,000
$52/kW
$306,000
$677,000
2.46
$2,398
1.78
$1,737
$13,415,000
$54/kW
$312,000
$733,000
2.57
$2,500
1.85
$1,802
$13,777,000
$55/kW
$319,000
$789,000
2.67
$2,602
1.92
$1,867
Application of SCR Technology for a Retrofit Unit
The economic evaluations reported in prior sections of this paper focused on an SCR reactor
installed on a new coal-fired facility. However, the majority of the near-term U.S. SCR market
may be in retrofit applications. The cost of implementing SCR technology is a topic of
considerable debate in the present deliberations by OTAG and in defining the U.S.
Environmental Protection Agency's (EPA) proposed CAAA Title IV NOX emission limits for
Group 2 boilers.
Because of the considerable uncertainty and debate involving SCR retrofit cost for existing
plants, the cost and technical feasibility of retrofitting SCR technology to existing coal-fired
generating units was determined. The results of this study reflect the wide range of retrofit costs
typically encountered due to site-specific issues. It is recognized that the costs summarized are
applicable to dry-bottom, pulverized coal boilers and may or may not be indicative of other
installations due to boiler type, site constraints, and/or inlet NOX levels to the SCR.
Summary of Capital and O&M Costs for Retrofit Units
For a 250-MW plant, the capital cost difference between a new SCR installation and retrofit SCR
installation were evaluated (Table 13). Results of this evaluation indicate capital costs to be
approximately 50 percent higher for retrofit installations when compared with new plant
applications.
Of the units evaluated, capital costs for installing a retrofit SCR range from $59/kW for an
880-MW unit to $87/kW for a 100-MW unit assuming no balanced draft conversions. For a
similar 880-MW unit requiring a balanced draft conversion, the capital requirement is estimated
to be $130/kW. This comparison is highly site specific, however, and actual retrofit costs may
be higher or lower than those presented here.
13
-------
Current dollar levelized costs for an SCR retrofit vary from $1,541/ton to $7419/ton depending
on NOX removal percentage, unit size, inlet NOX concentration, utilization (capacity factor), and
capital and O&M costs. Lower levelized costs are achieved when SCR is applied to larger,
higher utilized units due to economies of scale and the fact that greater amounts of tons are
removed on larger units. All of the units included in the study are equipped with some type of
combustion modifications to lower the NOX concentration prior to the SCR. While some capital
cost savings can be achieved in the SCR by lowering the inlet NOX, the resulting levelized cost is
higher due to reduced tons removed when compared to an SCR retrofit on an uncontrolled unit.
Figure 12 shows a comparison of levelized cost versus NOX removal efficiency for a new and
retrofit SCR installation applied to a 250-MW unit designed for 60 percent NOX removal. The
difference in cost is primarily due to higher capital cost of the retrofit installation, since the inlet
NOX concentration for the retrofit (0.40 Ib/MBtu) and the new unit (0.35 Ib/MBtu) are similar
and approximately the same number of tons of NOX are removed. Compared to the difference in
capital costs, the levelized cost differences between the new SCR installation and the retrofit
SCR installations are much less significant. Nevertheless, technical and economic assessment of
SCR must be based on both the levelized cost and the first cost (capital cost cash flow) of the
proposed installation.
In addition to the low inlet NOX conditions, specific retrofit cost data were extrapolated to high
inlet NOX conditions in an effort to represent many of the boilers in the OTAG region
Table 13: Capital Cost Differences for New and Retrofit SCR Installations
(250-MW Plant Size)
NOX Removal Efficiency
40% 60% 80%
New SCR Installation
Total Capital Requirement $12,974,000 $13,415,000 $14,142,000
Total Capital Requirement $52/kW $54/kW $57/kW
Retrofit SCR Installation
Total Capital Requirement $18,800,000 $20,281,000 $21,403,000
Total Capital Requirement $75/kW $81/kW S86/kW
5,000
o 4,000
O 3,000
•o
0)
» 2,000
1,000
Based on 250 MW Unit
with 60% Removal
Retrofit SCR
(Inlet NOx = 0.40 Ib/MBtu)
5,000
30 40 50 60 70 80
NOx Removal (%)
Figure 12: Levelized Cost vs. NOX Removal
90
0 0.3 0.6 0.9 1.2 1.5 1.8 2.1 2.4 2.7
Inlet NOx (Ib/MBtu)
Figure 13: Levelized Cost vs. High Inlet NOX
14
-------
(Figure 13). The extrapolated data indicate a significant trend of decreasing levelized cost with
increasing inlet NOX concentration, highlighting a key difference in cost effectiveness between
lower emitting boilers (or boilers which have been controlled with combustion modifications
prior to the SCR) and higher emitting, uncontrolled boilers.
Conclusions
The economic analysis of applying SCR to a new 250-MW unit with a 60 percent NOX removal
efficiency resulted in capital and first year O&M costs (in 1996 dollars) of $13,415,000
($54/kW) and $1,045,000, respectively. Levelized costs are $2,500/ton on a current dollar basis
and $l,802/ton on a constant dollar basis.
Once the base case was established, sensitivity analyses were performed to examine the impact
of major process variables on the capital, O&M, and levelized cost of applying SCR technology
to a new unit. Those factors affecting capital costs include unit size, NOX removal efficiency,
and catalyst price with unit size having the greatest impact. For new plant applications, total
capital requirements for a 60 percent NOX removal design range from $45/kW for a 700-MW
unit to $61/kW for a 125-MW unit.
With respect to current dollar levelized costs, several additional factors were examined including
capacity factor, inlet NOX concentration, catalyst relative activity, and catalyst life. Of these,
levelized costs are most sensitive to capacity factor and inlet NOX. The relationship between
levelized cost and SCR inlet NOX concentration demonstrates that SCR installations become
much more costly as the inlet NOX is reduced. Capacity factor shows a similar trend. For
example, a 20 percent reduction in capacity factor from the base case (65 percent) increases the
current dollar levelized cost by 38 percent.
An evaluation of retrofit SCR applications was performed as well. The retrofit applications
evaluated show a range of capital requirements from $59/kW for an 880-MW unit size to
$87/kW for a 100-MW unit size. For units that require a balanced draft conversion, capital
requirements range from $112/kW to $130/kW. In retrofit applications, lower levelized costs are
achieved when SCR is applied to larger, higher utilized units due to economies of scale and the
fact that a greater amount of tons are removed on larger units. Similar to the new unit cases,
higher inlet levels significantly decrease the levelized cost.
Acknowledgments
This project has been a collective effort on the part of many individuals and organizations too
numerous to name individually. In lieu of a detailed listing, we would like to recognize the
following organizations and their employees for their contribution to the success of this project:
U.S. Department of Energy, Southern Company Services, Inc., Electric Power Research Institute,
Ontario Hydro, Gulf Power Company, Southern Research Institute, ABB Air Preheater, Inc.,
Spectrum Systems, Inc., ICS, Inc., Cormetech, Inc., Haldor Topsoe A/S, Hitachi Zosen, Nippon
Shokobai, Siemens, W.R. Grace & Co., W.S. Hinton and Associates, Burns and Roe Services
Coroporation, J.E. Cichanowicz, Inc., and Steag, AG.
15
-------
References
1. Healy, Edward C., J. D. Maxwell, and W. S. Hinton, "Economic Evaluation of Commercial-
Scale SCR Applications for Utility Boilers," supplemental report prepared for The U.S.
Department of Energy, Pittsburgh Energy Technology Center, DE-FC22-90PC89652,
September 1996.
2. Southern Company Services, Inc., "Demonstration of Selective Catalytic Reduction (SCR)
Technology for the Control of Nitrogen Oxide (NOx) Emissions from High-Sulfur, Coal-
Fired Boilers," final project report prepared for The U.S. Department of Energy, Pittsburgh
Energy Technology Center, DE-FC22-90PC89652, October 1996.
3. Healy, Edward C., J. D. Maxwell, W. S. Hinton, and A. L. Baldwin, "Economic Analysis of
Selective Catalytic Reduction Applied to Coal-Fired Boilers for NOX Reduction," presented
at the Air & Waste Management Association's 90th Annual Meeting & Exhibition,
June 8-13, 1997, Toronto, Ontario, Canada.
16
-------
SCR FOR COAL-FIRED BOILERS:
A SURVEY OF RECENT UTILITY COST ESTIMATES
J. E. Cichanowicz
Consultant
236 N. Santa Cruz Ave. Suite 202
Los Gatos, Ca 95030
jecinc@batnet.com
Abstract
Accurate projections of SCR capital cost are critical both for prudent NOx
control rulemaking by federal and local environmental regulators, and to
establish realistic utility compliance plans. Within the last few years, many
utilities have engaged architect/engineers and/or SCR vendors to project SCR
cost for selected units in their system, employing detailed site-specific
assessments. This paper reports results of SCR cost studies conducted by
eleven utility companies, addressing 24 dry-bottom boilers, and 27 Group 2
boilers. The results show significant uncertainty characterizes SCR capital
cost estimates, as a wide range of values is projected for both boiler types.
This paper discusses and evaluates cost trends, demonstrates the impact of
capital cost uncertainty on the cost per ton of NOx removed, and compares
capital cost results with those from a computer algorithm widely used in NOx
control rulemaking.
Introduction
The availability of selective catalytic reduction (SCR) NOx control technology
at reasonable cost is a key consideration in the promulgation of NOx emission
limits by federal and local environmental agencies. The cost of SCR is a topic
of considerable disagreement among the various "stakeholders" participating
in the NOx emissions debate - most significantly the utility industry, federal
and local environmental regulatory agencies, and vendors of SCR technology.
This disagreement is capital cost translates into equal uncertainty regarding
projections for cost per ton of NOx reduced. Specific concerns have been
summarized in written comments submitted by the utility industry (UARG,
1997) to the Environmental Protection Agency's Acid Rain Division (ARD),
and in technical reports submitted by industry for use by the Ozone Transport
Assessment Group (OTAG, 1996a). Rebuttal positions have been formally
issued by the EPA ARD (EPA, 1996a), and OTAG stakeholders that support
SCR-based NOx limits (OTAG, 1996b).
-1-
-------
In December 1996, the EPA ARD issued NOx emissions for Group 2 boilers
based on an evaluated cost per ton of NOx removed, which required EPA to
project SCR capital cost for the national boiler population. In June of 1997,
the OTAG Policy Committee issued general recommendations for the control
of NOx emissions, which may require broad application of SCR from
generating units in the 38 states that comprise OTAG's interest. Both the EPA
ARD and the OTAG Policy Committee base their understanding of SCR cost
on discussions with equipment suppliers, and experience from Europe. EPA
ARD developed a cost algorithm to project SCR capital cost as a function of
generating capacity, and used this algorithm to select the NOx levels proposed
in December 1996 (EPA, 1996b), as well as support OTAG analysis.
During approximately the same time period, many utilities sponsored
detailed studies by architect/engineering firms and/or SCR vendors to
estimate SCR capital and operating cost. Given the significant cost
implications of NOx policy decisions, it is prudent to summarize cost results
derived from these utility-sponsored engineering studies, for discussion and
comparison with other cost sources.
Objective
The objective of this paper is to report the range of SCR capital cost
determined by site-specific engineering studies, estimated by either
architect/engineering firms and/or SCR technology vendors.
Subsequently, the impact of capital cost on the cost per ton of NOx removed is
calculated. The influence of two economic parameters that strongly dictate
SCR cost - capacity factor and capital recovery factor - is also demonstrated.
Approach
Utilities known to have sponsored major NOx planning studies that
employed detailed site assessments were requested to volunteer cost results
for summary and comparison.
Given the competitive climate within the utility industry, disclosure of
factors that affect the cost of generation has become extremely sensitive. With
the exception of two studies entered into the public record as part of CAAA
Section 407 rulemaking (OVEC, 1997, and TECO, 1996), utility companies only
shared cost information on the basis that specific units remain anonymous.
This evaluation considered only engineering studies that employed a site
assessment by the architect/engineer, or SCR vendor. Cost estimates were
required to be developed from specific equipment lists, derived after
considering plant layout, design specifications of the plant and components,
and condition of existing equipment (e.g. flue gas handling components). In
-2-
-------
most cases, the studies employed a general arrangement drawing to identify
the location of the reactor and ancillary equipment, as well as flue gas routing.
Eleven utility companies provided engineering studies for review that
addressed SCR capital cost for selected units in their system. These utilities
are located in the midwest, portions of the northeast, and the mid-Atlantic
states. The data set consists of a total of 24 dry-bottom boilers (e.g. wall- and
tangential-fired), and 27 Group 2 boilers (cyclone, wet-bottom, and cell-fired).
The number of boilers represented is a small fraction of the national
inventory. Within the 38 state OTAG region alone, approximately 700 dry-
bottom boilers exist. The total number of cyclone, wet-bottom, and cell-fired
boilers nationally number approximately 140. No rigorous statistical analysis
is possible with this data set, as the details of sites are unknown.
Description Of Cost Methodology
This section summarizes the cost methodology followed by most studies.
Cost Methodology
A detailed description of SCR cost methodology has been presented in an
earlier paper (Cichanowicz, 1993). This discussion highlights the following
cost elements that are of particular interest in this evaluation: Process
Capital, Installation Charge, Process/Project Contingency, Utility Indirect
Charge, and Allowance for Funds During Construction (AFDC)
Process Capital. Process Capital reflects acquisition cost for equipment
required for both the SCR process, and modifications to the balance-of-plant
or ancillary components. The Process Capital reflects the sum of expenditures
for equipment delivered to the site, but not an installation charge.
Table 1 summarizes the major Process Capital components. The first three
items (SCR catalyst, reactor, reagent storage, and reagent vaporization) are
direct SCR capital requirements, with remaining items denoted as balance-of-
plant components or installation expenditures. A subsequent section of this
paper addresses how costs partition between these two categories.
Installation Charge. The Installation Charge reflects primarily the labor
charge and lease of special equipment required for installation/erection, as
well the upgrade of balance-of-plant equipment or ancillary components.
Process/Project Contingency. These cost elements comprise a "reserve"
fund for unanticipated expenses due to either project-specific or process-
specific issues. Most of the engineering studies used 15-20% (of Process
Capital and Installation Charge) for the sum of both contingency funds.
-3-
-------
Utility Indirect Charge. Utility-incurred costs are comprised of staff
engineering, project management, and facilities such as access roads,
buildings, etc., and is usually 5-10% of Process Capital and Installation Charge.
Allowance for Funds During Construction (AFDC). AFDC is a finance
charge, incurred for time periods when equipment is not employed in power
production. Although not necessarily a significant cost component compared
to the sum of all other components, AFDC represents a real incurred cost, and
is included for completeness. All planning studies reviewed included a
modest charge of nominally 4-5% annually, for a period of usually 1-2 years.
Most results were derived for a 1995-1997 dollar basis, and for generally
similar process conditions (major exceptions are noted). These similarities,
and the desire to observe only gross trends, allow the use of results as directly
reported, thus not corrected for cost year basis and process conditions.
Capital Cost Components
Two cost indices are proposed to further characterize capital cost: (a) the
Process Capital/Installation Charge ratio, and (b) the sum of the catalyst,
reactor, and reagent storage/vaporization components to the Process Capital.
These cost ratios are further described as follows:
Ratio of Process Capital/Installation. Process Capital/Installation Cost
ratio, determined before application of Process/Project Contingencies, Utility
Indirect Costs, and AFDC indicates whether the bulk of direct costs are driven
by capital procurement (ratio >1) or manpower for installation (ratio <1).
It is anticipated a difficult retrofit site with significant obstacles that
complicate access of construction equipment would be characterized by a
relatively low Process Capital/Installation Cost ratio; a site with relatively
unrestricted access would be characterized by Process Capital/Installation Cost
ratio of >1.
Ratio of SCR Process/Process Capital. SCR Process/Process Capital
ratio, determined before application of Process/Project Contingencies, Utility
Indirect Costs, and AFDC indicates whether the bulk of process equipment
acquisition costs are for SCR components, or balance-of-plant equipment to
allow the boiler/plant to accommodate SCR process impacts.
It is anticipated that sites requiring few modifications would be characterized
by a relatively high SCR Process/Process Capital ratio; retrofit sites that
require boiler modifications and upgrades to accommodate the SCR process
would be characterized by a relatively low SCR Process/Process Cost ratio.
-4-
-------
Results
Results from this survey are presented according to two major boiler
categories: dry bottom and Group 2 (cyclone, wet-bottom, and cell-fired).
Ideally, separate cost comparisons would be developed for each of the five
major boiler categories. However, the relatively small number of units and
the desire to observe only general trends allows this simplification. Results
are discussed according to (a) capital cost, and (b) components of capital cost.
Capital Cost
Dry-Bottom Boilers. Figure 1 summarizes SCR capital cost for dry-bottom
boilers, presented as a function of generating capacity. The NOx reduction
efficiency for all units is 80-90%, with residual NH3 a maximum of 5 ppm
(2 ppm for selected sites). With the exception of two units, boiler initial NOx
production rates are approximately equivalent to the Phase 1, Group 1 limits
of 0.45-0.50 Ibs/MBtu, depending on boiler type (e.g. tangential- or wall-fired).
Note several of the data represent multiple units at the same station.
Group 2 Boilers. Figure 2 summarizes SCR capital cost for cyclone, wet-
bottom, and cell-fired boilers, presented as a function of generating capacity.
The wet-bottom boilers, all which feature SCR designed for 50% NOx
removal from approximately 1.1-1.3 Ibs/MBtu, are identified separate from
the cyclone and cell-fired boilers. The SCR NOx reduction efficiency for the
cyclone and cell-fired boilers is 80%, with one case at 50% noted. Except as
indicated, all cyclone/cell-fired boiler NOx production rates are 1.2-1.5
Ibs/MBtu. All costs reflect sufficient catalyst to maintain a residual NH3 level
of at most 5 ppm, throughout the entire operating period.
Capital Cost Components
Figure 3 presents trends in both the Process Capital/Installation Cost and SCR
Process/Process Capital cost ratios, as a function of projected capital cost, for
dry-bottom boilers. As suggested, the highest capital cost sites are
characterized by a Process Capital/Installation Cost ratio of 1-1.25; the lowest
capital cost sites can have values exceeding 2. The SCR Process/Process Capital
ratio ranges from 0.50 for high cost sites, to 0.75 for low cost sites.
Results Discussion
Results presented in this paper are based upon an extremely small sample of
the boilers, compared to the candidates considered to deploy SCR NOx
control. Clearly, caution should be exercised in extrapolating any results or
observations in cost trends from this sample to the national or the OTAG
regional population.
-5-
-------
Average Capital Cost
Dry-Bottom Boilers. The average of capital cost for dry-bottom boilers for
all 24 boilers presented in Figure 1 is $86/kW. The average for units greater
than 175 MW capacity is $75/kW.
Figure 1 demonstrates the wide variation in capital cost depending on site-
specific conditions. If only generating capacity is considered as an indicator of
"average" SCR capital cost, significant variations from the $75/kW average
for units >175 MW are witnessed. Specifically, within the cluster of units at
approximately 550 and 625 MW, any unit can vary in cost by $30-50/kW.
Group 2 Boilers. The average of capital cost for Group 2 boilers for all units
presented in Figure 2, calculated with four different averaging techniques,
ranges from $79-86/kW. The lowest cost ($79/kW) was determined by
eliminating boilers of less than 200 MW capacity, using only 2 boilers at each
of the Kyger and Clifty Creek sites in the average, and eliminating balance-of-
plant upgrades necessary to accommodate SCR at three large cyclones. The
highest cost was determined by employing all boilers in Figure 2 in the
average (all 11 Kyger and Cliffy Creek units, and not eliminating small
boilers), and including balance-of-plant costs for the large cyclones.
For units above 200 MW capacity, if generating capacity alone is used to
project SCR capital cost, significant variations from the nominal $83 /kW
average are witnessed. These variations appear to be $15-50/kW.
Economies of Scale
SCR is generally recognized by most observers to exhibit economies of scale
with respect to capital cost. This trend is dependent upon the assumption
that all other plant and SCR process design factors are maintained equivalent,
as generating capacity increases.
Figure 1 shows that cost per unit capacity decreases as generating capacity
increases from 100 to 200 MW. The average SCR cost for the units at
approximately 600 MW suggests continued cost reduction at larger capacities.
For the Group 2 boilers, the different boiler designs prevent identifying any
trend between cost and generating capacity.
For both boiler categories, the reduction in SCR unit cost with increasing
generating capacity is most pronounced for increases from lowest (-100 MW)
to intermediate capacities (-175-200 MW). SCR capital cost may not exhibit
economies-of-scale anticipated at larger capacities, as the design basis for the
SCR process and host unit changes significantly with increased capacity. An
example is the utilization of two reactors (each of 50% treating capacity) in
place of one reactor (at 100% capacity), to maintain turndown for larger units.
-------
Cost Per Ton Evaluation
The cost of NOx control per ton of NOx removed - sometimes referred to as
cost-effectiveness - is an important cost index. The EPA ARD has issued NOx
regulations for Group 2 boilers based on the "cost-effectiveness" of low NOx
burners on Group 1 boilers compared to the "cost-effectiveness" of candidate
NOx control technologies on Group 2 boilers. Essentially all NOx trading
programs proposed or presently in place employ this cost index. Also, several
states have proposed definitions of Reasonably Available Control Technology
(RACT) depending on the "cost-effectiveness" of NOx reduction achieved by
any given technology. It is instructive to examine the significance of the
uncertainty in capital cost observed in Figures 1 and 2 on the evaluated cost
per ton of NOx removed. Also, the impact on cost-effectiveness of two
economic factors of particular significance for SCR - the capital recovery factor
and generation capacity factor - is addressed.
Capital Cost
Table 2 summarizes NOx control cost per ton provided by SCR, as applied to
(a) dry bottom boilers in a "post-RACT" mode, and (b) Group 2 (cyclone)
boilers. Table 2 also presents the sensitivity of cost per ton to uncertainties in
capital cost, capacity factor (CF), and capital recovery factor (CRF).
Dry-Bottom Boilers. Cost results apply only to the specified conditions of
80% NOx reduction, initial NOx production rates of 0.45-0.50 Ibs/MBtu, 4 year
mean catalyst life, and a final space velocity of 3200 1/h. The generation
capacity factor and annual capital recovery factor are 65% and 0.15,
respectively. For the average SCR cost (as approximated from Figure 1) of
$75/kW, Table 2 shows that SCR NOx control cost is $1600-1768/ton, for boiler
NOx production rates of 0.50 and 0.45 Ibs/MBtu, respectively.
Table 2 also shows the impact of $15/kW variations in capital cost. Increasing
capital cost by $15/kW to $90/kW would increase the $1600-1768/ton cost
range by $220-250/ton. Similarly, decreasing capital cost by $15/kW to
$60/kW would decrease the $1600-1768/ton range by approximately the same.
Group 2 Boilers. Cost results apply only to the specified conditions of 80%
NOx reduction, 4 year average catalyst life, initial NOx production rate of 1.3
Ibs/MBtu, and a final space velocity of 2000 1/h.
For the average SCR cost (as approximated from Figure 1) of $79/kW, Table 2
shows that SCR NOx control cost is $696/ton, for NOx production rates of 1.3
Ibs/MBtu. Adjustments to capital cost by $15/kW impact cost by $80/ton.
-7-
-------
Sensitivity Analysis: Additional Factors
Although the focus of this paper is SCR capital cost, it is prudent to briefly
consider two other factors that can dominate the evaluated cost per ton of
NOx reduced by SCR. The ability to recover capital cost, as determined by unit
capacity factor and financing conditions, is particularly important with SCR,
due to the high capital requirement compared to alternatives. This section
demonstrates how the range of capacity factor and capital recovery factor
impact the evaluated cost per ton.
Capacity Factor. Future projections of capacity factor for the deregulated
industry have received considerable attention recently. Projections for system
capacity factor averages range from 65% to as high as 85%, depending on the
economic conditions presumed for the relevant time period. This differential
of 20 percentage points translates into a considerable difference in cost per ton
of NOx. As shown in Table 2, simply increasing capacity factor from
historical norms of 65% to 85% lowers SCR cost per ton for dry-bottom boilers
by $340 to $380/ton, for boiler NOx productions rates of 0.50 and 0.45
Ibs/MBtu. For Group 2 boilers, the same increase in capacity factor lowers
evaluated cost by approximately $120/ton.
Capital Recovery Factor. Utility planning studies reviewed documented
the range in capital recovery factor employed for cost evaluations. This factor
depends not only on the details of financing capital, but also the secondary
cost of equipment ownership, such as property taxes, insurance, etc. Most
significantly, the term over which the utility intends to operate the facility -
either 10,15, or 20 years exerts a dominant role in determining the capital
recovery factor. The studies reviewed for this paper show the range in capital
recovery factor to be 0.14- 0.167. Within the NOx policy debates, stakeholders
supporting the application of SCR have proposed a capital recovery factor of
0.115, for a 20 year plant life. Accordingly, a sensitivity analysis was conducted
to determine the impact of capital recovery factor to levels as low as 0.115.
Table 2 shows for dry bottom boilers reducing capital recovery factor to 0.115
lowers evaluated cost by $650/ton, for a boiler NOx production rate of 0.50
Ibs/MBtu. For Group 2 boilers, the same variation in capital recovery factor
lowers evaluated cost by $90/ton. Accordingly, the role of capital recovery
factor is significant, and is equal to or greater than the impact of reasonable
changes in capacity factor or capital cost.
Observation
Results from this evaluation highlight how uncertainty in capital cost,
capacity factor, and capital recovery factor impact cost per ton of NOx.
-8-
-------
Depending on the value of capital cost, capacity factor, and capital recovery
factor, the evaluated cost per ton of NOx can vary by almost 50%. Specifically,
Table 2 reports cost per ton for both dry-bottom and cyclone boilers,
employing inputs that based on the studies reviewed for this paper,
discussions with utilities, and economic projections, appear extreme. These
values are $60/kW capital cost, 85% capacity factor, and 0.115 capital recovery
factor. Employing these values for dry-bottom boilers produces a cost of $935-
1029/ton, approximately 55% of that estimated for the "baseline" case. For
cyclone boilers, cost is $424/ton, or 60% of the "baseline"
The cost per ton is further reduced, when employing capital cost estimates
based on the computer algorithm describing capital cost versus generating
capacity, that was derived by EPA ARD. For dry-bottom boilers, using 550
MW as a reference case, this correlation used in OTAG rulemaking projects a
capital requirement of $47/kW, for an SCR process designed for 80% NOx
removal from Phase 1/Group 1 boiler NOx production rates. A generating
capacity of 650 MW is anticipated to require SCR capital cost of $44/kW,
according to this correlation.
These algorithm-derived estimates are 60-65% of the average capital cost at
550 and 650 MW presented in Figure 1. Using these algorithm-derived capital
costs results in estimates of cost per ton of $791-818/ton, half of the baseline
case. Similar trends were noted with Group 2 boilers, where the algorithm
also significantly underpredicted cost for 50% NOx reduction cases.
Summary
Engineering studies submitted by 11 utility companies revealed trends in SCR
capital cost, based on detailed site-specific assessments. For dry bottom
boilers, the projected cost for SCR was $86/kW, and reduced to $75/kW when
boilers less than 175 MW were eliminated. For cyclone boilers, the average
cost was $79-86/kW, depending on how the average was calculated.
This significant capital cost uncertainty translates into equivalent uncertainty
in cost per ton. Economic and technical premises selected from this survey
suggest deploying SCR delivers NOx reduction for $1600-1768/ton, in a post-
LNB application. For cyclone boilers, the cost per ton anticipated for these
conditions is $674/ton.
Both capacity factor and capital recovery factor exert significant impact on cost
per ton. By using values for these inputs that based on the utility site-specific
studies appear to be extreme, evaluated cost can be reduced to 55-62% of the
previously cited values. Further complicating the matter is the apparent
tendency of the SCR capital cost algorithm developed for NOx rulemaking by
EPA ARD to underpredict SCR capital cost, producing estimates
approximately 60-65%% of those inferred from Figure 1. In summary,
-9-
-------
estimates of SCR cost that do not employ a detailed site-specific analysis could
be significantly in error, and generating capacity and capital recovery factor
should be carefully considered to reflect authentic industry experience.
References
Cichanowicz,
1993
EPA, 1996a
EPA, 1996b
OTAG, 1996a
OTAG, 1996b
OVEC, 1997
TECO, 1996
UARG, 1997
Cichanowicz et. al., "Factors Affecting SCR Capital Costs For
Utility Boilers", paper prepared for The Utility Air Regulatory
Group, October, 1993
Investigation of Performance and Cost of NOx Controls As
Applied to Group 2 Boilers, prepared by The Cadmus Group,
August 1996, EPA Contract No. 68-D2-0168, Work
Assignment No. 4C-02.
Distributions Of NOx Emission Control Cost-Effectiveness By
Technology, prepared by ICE Incorporated, October 31,1996,
EPA Contract No. 68-D3-0005, Work Assignment No. 5F-01.
"Electric Utility Nitrogen Oxides Reduction Technology
Options For Application by The Ozone Transport Assessment
Group", January 1996, prepared by UARG for Consideration
by the OTAG Control Technologies & Options Workgroup.
"Electric Utility Nitrogen Oxides Reduction Technology
Options For Application by The Ozone Transport Assessment
Group", April 1996, prepared by Selected States &
Stakeholders for Consideration by the OTAG Control
Technologies & Options Workgroup.
"SCR Capital and Operating Cost Estimate for Kyger Creek
and Clifty Creek", final report prepared for the Ohio Valley
Electric Corporation, May 1997
"Nitrogen Oxide Limitation Study", prepared for Tampa
Electric Company, March 15,1996
Comments Filed On Behalf of the Utility Air Regulatory
Group, In Response to The January 19, 1997 Proposed Rules
Implementing The Second Phase Of The Nitrogen Oxides
Reduction Provisions In Title IV Of The Clean air Act, Docket
No. A-95-28.
-10-
-------
TABLE 1
SUMMARY OF PROCESS CAPITAL COMPONENTS
FOR SCR RETROFIT
Cost Component
Comment
SCR Catalyst, Reactor
Reagent Storage
Reagent
Vaporization/Injection
Sootblowers
Foundations
Structural Steel
Ductwork Modifications
New Ductwork
Process I&C
Fan Modifications
Balanced Draft
Conversion
Electrical
Boiler Modifications
Other (BOP)
Duct Burner, Gas/Gas
Heater
Misc/General
usually the largest cost components
facilities for the unloading, transfer, and
storage for aqueous or anhydrous ammonia
reagent
equipment to vaporize reagent, and
monitor and control injection rate
included in almost all SCR designs and cost
estimates; sometimes not seperately
identified
re-inforcing of existing foundations, or
construction of new foundations depending
on reactor location
re-inforcing of existing structures, or
construction of new structures depending
on reactor location
modifications to existing ductwork to
accommodate SCR equipment
new ductwork for process bypass, reactor
access, etc.
control systems for process operation
improvements to existing fans to increase
flow rate rating, or replacement with new
fans
reinforcement of ductwork structure, and
addition of fans as necessary to convert
from forced to balanced draft.
additional auxiliary power supply for
reagent, blowers, etc. can require an increase
in power delivery capabilities on-site
installation of economizer bypass, removal
or addition of heat absorbing surface area as
necessary to provide correct flue gas
temperature vs. load
modifications to the air heater to improve
tolerance to increased SOS; improvements
to particulate control equipment to tolerate
residual NH3, SOS; etc.
heat exchange equipment necessary for
post-FGD applications
flow modeling, construction management,
demolition charge, etc.
-------
$/kW
140
r f
2x
10
1/-> *-*
00
OA
90
on
OU
7H
/ U
60 -
i-n
50
127
116
2x
FIGURE 1. DRY BOT
SCR CAPITAL COST
n
FOM BOILERS:
VS. CAPACITY
n 2x
-
2x
3x
2x
n
1
3x
100 200 300 400 500 600 700 800 1000
Generating Capacity, MW
-------
$/kW
110
100-
90
80-
70-
60
50-
40-
125+
125+
FIGURE 2. GROUP 2 BOILERS:
SCR CAPITAL COST VS. CAPACITY
125
I
E
Wet-Bottom Boiler, 50% NOx
Cyclone/Cell, 80/90% NOx
BOP Costs
80% „
50% ^
100 200 300 400 500 600
Generating Capacity, MW
700 1000
Note Scale Change
-------
3.16
2.00
1.75
1.50-
1.25-
1.00-
0.75
0.50-
0.25-
FIGURE3. RATIO OF
TOTAL PROCESS CAPITAL/INSTALLATION,
SCR PROCESS/TOTAL PROCESS CAPITAL
TOTAL PROCESS CAPITAL/INSTALLATION E
SCR PROCESS EQUIPMENTH-PC ^
40 50 60 70 80 90 100 110 120 130 140
Capital Cost, $/kW
-------
TABLE 2
COST PER TON EVALUATION
SCR ON POST-LNB DRY-BOTTOM, AND GROUP 2 BOILER
(Baseline Case And Sensitivity Analysis)
Evaluation
Process
Conditions
Economic
~ actors
Cost Per Ton
$/Ton (NOx)
Baseline: SCR Applied
to Post-LNB, Dry-
bottom Boiler
Baseline: SCR Applied
to Group 2 (Cyclone)
Boiler
Sensitivity: Incremental
Capital
(+/- $15/kW)
Sensitivity:
Capacity Factor (65% to
85% increase)
Sensitivity:
Capital Recovery Factor
(0.15 to 0.115 decrease)
SCR /Dry-Bottom
($60/kW, 85% CF, 0.115
CRF
SCR/Group 2
($60/kW, 85% CF, 0.115
CRF
80% NOx reduction/
5 ppm slip, 3200 1/h SV,
4 yr catalyst life
(Baseline)
80% NOx reduction/
5ppmslip,20001/hSV, 4
yr catalyst life,
(Baseline)
Dry-Bottom Baseline
Boiler Case
Group 2 Boiler Baseline
Case
Dry-Bottom Baseline
Boiler Case
Group 2 Boiler Baseline
Case
Dry-Bottom Boiler
Baseline Case
Group 2 Boiler Baseline
Case
Dry-Bottom Case, except as
noted
Group 2 Baseline Case,
except as noted
$75/kW, 65%
CF, 0.15 CRF
$79/kW, 65%
CF, 0.15 CRF
same
same
same, except
CF
same, except
CF
same, except
CRF
same, except
CRF
as noted
as noted
1600 (0.50)
1768 (0.45)
696 (1.3 )
A 220 (0.50)
A 250 (0.45)
A 80 (1.3)
A 340 (0.50)
A 380 (0.45)
A 120 (1.3)
A 650 (0.50)
A 90 (1.3)
935 (0.50)
1029 (0.45)
424 (1.3)
-------
SCR FOR A 460 MW COAL FUELED UNIT:
STANTON UNIT 2 DESIGN, STARTUP, AND OPERATION
John R. Cochran
Black & Veatch
8400 Ward Parkway
Kansas City, MO 64114
Denise Scarlett
Orlando Utilities Commission
500 S. Orange Avenue
Orlando, FL 32802
Robert Johnson
Siemens
18235 Windsor Drive
Stillwell, KS 66085
Abstract
Orlando Utilities Commission's (OUC) Stanton Energy Center consists of two 460 MW (gross)
pulverized coal fueled units each burning eastern bituminous coal. Unit 1 began operation in 1986
and controlled NOX emissions only by low NOX burners to meet an emission limit of 0.60 Ib/MBtu.
Unit 2 was intended to be a virtual replication of Unit 1, but with more advanced Babcock &
Wilcox XCL burners to achieve an emission rate of 0.32 Ib/MBtu. However, the Best Available
Control Technology (BACT) determination for Unit 2 required the use of a post combustion NOX
emission reduction system to meet an emission limit of 0.17 Ib/MBtu. Comprehensive technical
and economic analyses indicated that a high dust, selective catalytic reduction (SCR) system should
be installed on Unit 2. Unit 2 began commercial operation on June 1, 1996.
This paper describes Unit 2's SCR system design basis, configuration, startup, testing, initial
operation, and annual inspection. In addition, SCR system cost effectiveness values are presented.
Licensing Background
Prevention of Significant Deterioration (PSD) permitting for Unit 2 began in 1991. The original
BACT analysis recommended the use of combustion controls to achieve an uncontrolled NOX
emission rate of 0.32 Ib/MBtu. This was recommended on the basis of incremental NOX emission
reduction costs estimate! as exceeding $6,000/ton when using either selective non-catalytic
reduction (SNCR) or SCR technologies. Despite these high estimated costs and a number of
technical concerns, EPA Region IV and the Florida Department of Environmental Regulation
required using a post combustion NOX emission reduction system to achieve emissions of 0.17
Ib/MBtu (30-day rolling average). This system also had to achieve a one-time demonstration of the
capability to achieve 0.10 Ib/MBtu during initial performance testing. In addition, Unit 2 was also
-------
required to limit ammonia slip emissions to less than 5 ppm (volumetric). The permit for Unit 2
did not dictate a NOX emissions control technology.
Technology Selection
Given the flexibility to select the optimum control technology for Unit 2, OUC directed Black &
Veatch to perform technical and economic analyses to select between SCR and SNCR technologies.
Subsequently, comprehensive experience investigations and detailed capital and annual cost
estimates regarding both technologies were performed. Economic and technical analyses both
supported the use of an SCR system for Unit 2. SCR had a differential levelized annual cost of
$2.4 million per year less than SNCR. Because of compelling technical and economic results, OUC
proceeded with a decision to install a high dust SCR system on Unit 2. '
Plant Description and SCR Design Basis
Unit 2 is equipped with the following components.
• A single Babcock & Wilcox sub-critical, pulverized coal steam generator.
One vertical shaft Ljungstrom air heater.
Two cold-side electrostatic precipitators.
A three, 50 percent module wet limestone flue gas desulfurization system.
The unit burns eastern bituminous coal with maximum design sulfur and ash contents oF2.5 percent
and 11 percent, respectively. Table 1 lists typical and range fuel qualities for the Unit 2 design.
Table 2 lists pertinent steam generator performance parameters (full and minimum loads) that had to
be considered in the SCR system design.
The following summarizes the SCR system guarantee requirements:
• Initial performance test (new catalyst) — NOX reduction equal to or greater than 70 percent
while maintaining outlet NOX emissions equal to or less than 0.10 Ib/MBtu.
• Post initial performance test — NOX reduction equal to or greater than 47 percent while
maintaining outlet NOX emissions equal to or less than 0.17 Ib/MBtu (30-day rolling
average basis).
Maximum ammonia slip of 2 ppm (corrected to 3 percent oxygen).
S02 oxidation rate less than 1.0 percent.
• Total system pressure drop of less than 3.42 in wg at full load.
Ammonia usage rate of less than 232 Ib/h at full load.
Power consumption of less than 68.5 kW (dilution fans and soot blowers).
• Minimum catalyst life of 24,000 hours of boiler operation to achieve above guaranteed
performance (exclusive of 1).
A more detailed description of the technology selection analysis is presented in the technical paper "Selective
Catalytic Reduction for a 460 MW Coal Fueled Unit: Overview of a NO,, Reduction System Selection", J.R. Cochran, et
al, Black & Veatch, presented at the 1993 EPRI/EPA Joint Symposium on Stationary NO, Control.
-------
SCR System Configuration
Figures 1 and 2 illustrate the process flow diagram for Unit 2's SCR system. Anhydrous ammonia
was selected as the reagent for the SCR system to minimize NOX reduction costs.2 The SCR system
consists of ammonia receiving and storage; ammonia vaporization and injection; reactor, ductwork,
and sootblowing; catalyst; and controls subsystems.
Ammonia Receiving and Storage
Anhydrous ammonia is received at the plant by truck. Ammonia is directed to one of two 12 foot
diameter, 30,000 gallon tanks. This storage is adequate to meet plant requirements for 30 days.
The tanks are designed for a working pressure of 250 psig. A concrete containment area surrounds
the tanks. In addition, a sun shield and emergency deluge system is located directly above the
storage tanks.
Ammonia Vaporization and Injection
The ammonia vaporization system consists of two, 100 percent capacity electric heated vaporizers.
The vaporizers are located directly adjacent to the ammonia storage tanks in an enclosure.
Vaporized ammonia pressurizes the ammonia storage tanks. Vaporized ammonia is taken from the
top of the tanks and directed through piping to the ammonia mixer. A motorized control valve
modulates the flow of ammonia to the mixer based on NOX emission monitor measurements and set
points.
Ammonia is diluted with air in a stainless steel mixer to a concentration of approximately 6 percent.
Two 20,000 cfm (100 percent capacity) fans provide the dilution air to the mixer. This dilution air
is taken from the secondary air supply. Diluted ammonia is directed from the mixer to a 12 zone
injection grid located in the SCR reactor inlet duct. All piping from the mixer to the point of
ammonia injection is constructed of stainless steel.
Reactor, Ductwork, and Sootblowing
A single reactor treats all flue gas exiting the boiler. The reactor and ductwork are constructed of
1/4 inch plate A36 carbon steel. The reactor is 65 feet wide, 35 feet deep, and 53 feet high. The
reactor is totally enclosed in the boiler building.
Single louver isolation dampers are installed at the inlet and outlet of the reactor. A partial bypass
duct and dampers are installed for use during startup, and shutdown.
2 For a more detailed description of this selection refer to "Aqueous or Anhydrous? What Stanton and Other SCR
Experience Tell Us About Ammonia Selection", M.G. Gregory, et al, Black & Veatch, presented at the 1996 ICAC NO,
Forum.
-------
Flue gas exiting the economizer turns upward 90° The flue gas flows through two stages of static
mixers and the ammonia injection grid in this vertical ductwork. The static mixers are utilized to
mix flue gas and ammonia. Flue gas with ammonia then crosses over to the reactor. A dummy
catalyst layer is utilized to ensure gas flow vectors are vertical prior to entering the first layer of
catalyst, as well as absorbing the energy of initial impact of particulate/fly ash. A physical flow
model study was performed to configure flow distribution devices and to minimize pressure drop.
The initial catalyst charge consists of two layers of catalyst. However, the reactor is configured to
accept two additional catalyst layers for future use as part of the catalyst management plan
described in this paper. A hoist and monorail system was used to load catalyst into the reactor.
The hoist lifts catalyst from grade through a hoistway within the SCR reactor enclosure building to
the appropriate reactor level. The catalyst is then transferred from the hoist to a monorail. Catalyst
is moved into the reactor through a single door per level using a monorail system. Once the
catalyst was placed, seals were installed to minimize leakage around the catalyst.
The reactor is equipped with steam sootblowers. Sootblowers, of a retractable, rake design, are
located directly above each catalyst layer. Sootblowing steam has a pressure of approximately 150
psig. The sootblowers are oriented to blow debris downward through the catalyst.
Catalyst
Siemens' SINOx™ plate type catalyst was selected for Unit 2. This catalyst had been used in
numerous US and European high dust SCR applications. The catalyst has a high durability wiffi~
respect to erosion and corrosion. In addition, this catalyst has a low pressure drop as compared to
ceramic honeycomb catalysts.
Two levels of catalyst are currently installed in the reactor. Each of these levels has a 10 x 11
arrangement of catalyst module frames. Carbon steel, catalyst module frames contain a 2 wide by
4 long by 2 deep arrangement of catalyst elements. Each of these catalyst module frames are fully
loaded with two 500 mm element layers.
The initial two levels of catalyst totals 13,100 ft3 (370 m3) with a space velocity of 3,945 h" ' The
plate type catalyst has a relatively conventional composition of vanadia, titania, and tungsten
supported on a stainless steel expanded mesh. The catalyst has a specific surface area of 328 nr/m3
and an SO2 to SO3 oxidation rate less than 1 percent.
Ultimately the reactor could accept another two levels of catalyst module frames. Future catalyst
increments can be added in single element increments. Therefore, the current four element layers in
the reactor could be increased to a total of eight element layers as catalyst additions become
necessary'. The additional module layers will act as spares to give OUC considerable flexibility in
managing future catalyst additions.
-------
Controls
The control system for the Unit 2 SCR system mainly consists of inlet and outlet, NO, and O,
monitors, as well as logic programmed into the plant's DCIS. Ammonia injection was designed to
be controlled by a feedforward signal of reactor NOX inlet concentration and unit load, and a
feedback trim signal from reactor NOX outlet monitoring.
Startup and Initial Performance
Catalyst was loaded into the reactor in March 1996. Startup of the system occurred during April
and May 1996. No substantial difficulties were encountered with startup or commissioning
activities. The ammonia injection rates were tuned based on monitoring CEM data and sampling
over a 50-point SCR outlet fixed grid matrix. The plant began commercial operation on June 1,
1996.
SCR system performance tests were performed at low-, mid-, and full-load points from June 21
through June 23, 1996. Table 3 summarizes the results of these tests. All guarantee values were
met with the exception of ammonia consumption.
SCR operations have continued to date (July 1997) in full compliance with NOX emission limits
without any interference with normal unit operations. No air heater pluggage or degradation have
occurred. Existing fly ash sales and disposal practices of fixated scrubber solids with flyash have
continued without any effect by the SCR s~ystem. Opacity from Unit 2 generally ranges from 1 to 3
percent, with no visible plume. As compared to Unit 1, OUC has not realized any additional O&M
staffing requirements as a result of the SCR system on Unit 2. The most persistent problem
encountered to date with SCR system operation has been with the emissions monitoring system.
As previously described, the control system for the SCR consists of feedforward signal based on
measured NOX and O2 at the reactor inlet, as well as unit load as the primary control signals for
ammonia feed modulation. Secondary modulation is designed based on feedback of reactor outlet
NOX concentrations to meet setpoint values. This control system setup minimizes the potential for
overinjection of ammonia and the resultant ammonia slip. Any ammonia slip present in the flue gas
exiting the catalyst can ultimately react with SO3 in the flue gas to form ammonium bisulfate and
sulfate salts which may potentially foul the air heater.
To date, OUC has only encountered persistent problems with the Lear Siegler monitoring system
located at the inlet and outlet of the SCR system. The probes have plugged in this high dust
environment In addition, the monitors have had periodic calibration problems. Efforts continue to
resolve these probe pluggage and calibration difficulties. Until monitoring system problems are
resolved, the SCR system is being controlled based on a feedback signal of stack NOX only. The
long lag time between ammonia injection point and NOX measurement results in not only a sluggish
control system, buTalsd~~risks overinjection of ammonia. Fortunately, the relatively large amount of
fresh catalyst installed in the reactor will allow primary feedback control for now. However, as the
catalyst deactivates, this practice will become increasingly risky. OUC expects to return to
feedforward control in the next 2 months.
-------
In an effort to minimize the risk associated with purely feedback control, OUC has changed their
NOX set point practices. Initially, because of uncertainty regarding the SCR system's capability of
meeting a 0.17 Ib/MBtu limit on a 30-day rolling average, OUC set outlet NOX emissions at 0.13
Ib/MBtu. However, the reliability of the system to date has caused OUC to change this set point to
0.16 Ib/MBtu.
Annual Inspection Results
During the annual outage in April 1997, the reactor and catalyst were inspected. At this point, the
catalyst had been in operation (exposed to flue gas) for approximately 7,500 hours. The purpose of
the inspection was to assess fly ash deposits and the effectiveness of sootblowing and to determine
the mechanical condition of the catalyst. The inspection indicated that there were no significant
areas of maldistribution of gas or dust flow. There were no significant deposits on, or pluggage of,
the catalyst (see Figures 3 and 4), which indicated proper sootblowing operation. In addition, no
unusual erosion of the catalyst was evident.
Eight pairs of plate catalyst elements were pulled randomly from the first module layer. Siemens is
in the process of testing the activity (k/k0) of these elements. The expected activity after
7,500 hours of operation is a k/k0 value of 0.9. However, based on inspection and operational
results to date, it is expected that remaining activity will be equivalent or higher than this value.
Prior to inspection, sootblowing had been performed once per day. The results of the inspection
prompted OUC to change operating procedures to a once per week SCR reactor sootblowing cycle.
Until the next annual inspection, the effects of this operational change will be assessed by
differential pressure measurements across the catalyst and air heater.
Catalyst Management Plan
Catalyst additions and replacements represent the most significant O&M cost for an SCR system.
Catalyst additions and replacements can have a cost as high as $310 to $340/ft3 ($11,000 to
$12,000/m3). Accordingly, each element layer addition to the Unit 2 reactor will cost approximately
$1.1 million. Therefore, a major objective of OUC is to minimize future catalyst additions.
Figure 5 presents the original catalyst management plan for Unit 2 developed on the original design
conditions for the unit. The management plan reflects the operation of the unit at a constant NOX
outlet from the SCR reactor. The graph illustrates that the relative activity of the catalyst will
decrease over the initial guarantee period, resulting in a predicted increase in ammonia slip.
Periodic addition of catalyst increases the total activity of the system and lowers ammonia slip to
the initial low levels.
As previously described, the reactor is configured for up to four catalyst module levels (eight
element layers). Currently, four element layers (2 full catalyst module levels) are installed.
Guaranteed deactivation rates require that an additional element layer be added to the reactor after
about 24,000 hours (about 3 years of unit operation). This catalyst will be added to catalyst module
level 3. After about 47,000 hours the second element layer will be added to catalyst module levels.
After about 69,000 hours an element layer will be added to catalyst module level 4, and after about
90,000 hours (more than 11 years of unit operation will have elapsed) another element layer will be
-------
added completely filling the reactor with catalyst. Subsequent catalyst additions will replace
existing catalyst element layers.
Cost Effectiveness
The cost effectiveness of an SCR system (generally reported in $/ton values) is predominately
dependent on the inlet NOX concentration and SCR removal requirements. The higher the inlet NOX
concentration and SCR removal rate the better the cost effectiveness. Therefore, considering low
inlet NOX concentrations (0.32 Ib/MBtu) and relatively low removal rates required (50 percent),
Unit2 $/ton NOX removal costs will be higher than coal fueled units not using low NOX burners.
Table 4 lists the costs associated with the Unit 2 SCR system with an inlet NOX concentration of
0.32 Ib/MBtu and an outlet NOX concentration of 0.16 Ib/MBtu (50 percent removal).
The all-in capital cost of the Unit 2 SCR system was $21.4 million ($47/kW). This cost includes
all direct and indirect costs associated with the project's SCR system including the following
equipment, labor, materials, and services:
• Ammonia receiving and transfer.
Ammonia storage tanks, detection, and deluge equipment.
• Ammonia vaporization and transport.
• Dilution air fans and ammonia mixer.
• Ammonia injection.
• Static mixers.
• Interconnecting ductwork, expansion joints, and turning vanes.
• Louver bypass dampers.
• Reactor, internal supports, and access doors.
Catalyst loading system including all associated hoists and monorails.
Catalyst, module frames, and seals.
• Steam sootblowers.
• Inlet and outlet NOX and O2 monitoring.
• Instrumentation and distributed control system.
Fire protection and safety equipment.
• Foundations and support steel.
Insulation and lagging.
• Complete enclosures for the ammonia vaporizers and reactors.
Piping and valves.
Construction labor, materials, and equipment
Scale model study.
• Engineering, construction management, and owner indirects.
• Startup, tuning, and performance testing.
Unit 2's projected levelized annual O&M costs total $1.2 million (0.41 mills/kWh). These costs
were levelized by multiplying first year costs by a factor of 1.36 (20-year basis). These costs
include ammonia, catalyst additions and replacements, energy (vaporizers, dilution air fans,
sootblowers, and differential ID fan power), and annual testing and tuning. No increase in staffing
requirements have resulted from the use of SCR. The cost listed for catalyst additions and
-------
replacements is based on the projected guarantee basis catalyst management plan. Intermittent
catalyst addition costs at approximately 3 year intervals were individually escalated and discounted.
As such, no catalyst additions are expected for the first 3 to 4 years of unit operation.
Fixed charges on capital were added to annual O&M costs resulting in a total projected levelized
annual cost of $2.9 million (0.98 mills/kWh). This is equivalent to an emission reduction cost
effectiveness of $l,200/ton of NOX removed. This value is relatively high as compared to higher
NO,, emitting sources. SCR cost effectiveness values of as low as $400/ton are possible for high
NOX emitting wet bottom boilers (uncontrolled emissions greater than 1.5 lb/MBtu).3
Summary
The SCR system installed at Stanton Unit 2 has operated for over 8,500 hours without significant
difficulty. The system has provided abundant operational flexibility while reliably meeting a NOX
emission limitation of 0.17 lb/MBtu. The SCR system has not impacted plant operations or
availability. Off-line SCR system inspection did not indicate any fouling or erosion of the catalyst.
In addition, catalyst reactivity testing indicates a high probability for the catalyst to exceed the
guaranteed 24,000 hour life.
Although costs to date have been much lower, the projected life cycle cost for the SCR system is
equivalent to a levelized cost effectiveness of $l,200/ton of NOX removed and an all-in (capital and
operating) busbar increment of 0.98 mills/kWh.
3 "The Cost of Compliance for Group 2 Boilers", J.R. Cochran, et al, Black & Veatch, 1997 EPRI-DOE-EPA Mega
Symposium, August 25, 1997.
-------
Table 1
Unit 2 Design Coal Quality
Parameter
Coal Supply A
Typical, % Range, %
Coal Supply B
Typical, % Range, %
Ultimate Analysis:
Carbon
Hydrogen
Sulfur
Oxygen
Ash
Nitrogen
Chlorine
74.09
4.71
0.77
4.04
10.00
1.26
0.13
73
4.65
0.71
2.38
8.3
1.24
0.02
78
4.90
0.82
-4.04
11.0
- 1.49
-0.15
68.73
4.86
2.50
6.81
7.8
1.14
0.16
66
4.5
2.2
6.0
7.0
1.0
0.12
-70
- 5.1
-2.5
-7.0
8.5
1.3
0.20
Higher Heating
Value, Btu/lb
13,000
12,900 13,150
12,400 12,200 - 12,600
Table 2
Steam Generator Performance Parameters
(Economizer -Outlet)
Parameter
100% Load
25% Load
Uncontrolled NOX, Ib/MBtu 0.32
Design Flue Gas Flow Rate, Ib/h 4,273,000
Flue Gas Temperature, °F 706
Design Economizer Outlet Volumetric
Flow Rate, acfm 2,079,000
Corrected How Rate, dscfm (3% O2) 784,000
Oxygen Concentration, % 3.9
0.32
1,749,000
601
780,000
240,000
8.3
-------
Table 3
SCR System Performance Test Results
Parameter
100% Load
Mid Load
Low Load
Boiler Load, MW
NOX Inlet, Ib/MBtu
NOX Outlet Emission, Ib/MBtu
NOX Reduction, %
NH3 Consumption, Ib/h
NH3 Slip, ppm at 35 O2
Power Consumption, kW
453
0.35
0.10
71
330
<0.1
41
281
0.32
0.16
54
141
<0.1
41
187
0.35
0.14
62
112
0.3
41
Parameter
Table 4
Stanton Unit 2 SCR System Costs
Cost
Total All-in 1996 Capital Cost*
Levelized Annual Costs**:
Ammonia
Catalyst Additions and Replacements
Energy
Annual Testing and Tuning
Levelized Annual O&M Cost
Fixed Charges on Capital
Total Levelized Annual Cost
Cost Effectiveness
$21,400,000
$47/kW
$280,000
$590,000
$260,000
$70.000
$1,200,000
$1,690.000
$2,890,000
mills/kWh
0.09
0.20
0.09
0.02
0.41
0.57
0.98
$l,200/ton
Includes all related SCR equipment and construction; including
foundations, full SCR and vaporizer enclosures, electrical, engineering,
construction management, and owner indirects.
Levelized costs reflect the escalation and present worth discounting of
future expenditures. A levelization factor of 1.36 times first year costs
was used.
10
-------
Injection
Grid(I2Pt)
Dilution Air Fans
(2 x 20,000 cfm)
Remote Near SCR Reactor Secondary Air
Figure 1
Stanton Unit 2 Ammonia Storage and Vaporization System
Initial
Charge of -•
CaiaJysi N
Future
Catalyst
and
Sootblowe
^ "Sootbloiitre
IT
.^cooooooooo
Flue Gas to Air Heater
Vaporized
Stall c
Mixers
Economize
Outlet
Flue Gas
Figure 2
Stanton Unit 2 Catalyst Reactor System
11
-------
Figure 3
Unit 2 Catalyst (Level 1) After 7,500 Hours of Operation
(Note that the debris on the catalyst protective screens
sloughed off of dummy" layer support beams.)
Figure 4
Unit 2 Catalyst Element (level 1) After 7.500 Hours of Operation
and With Protective Screen Removed
12
-------
Figure 3
Unit 2 Catalyst (Level 1) After 7.500 Hours of Operation
(Note that the debris on the catalyst protective screens
sloughed off of dummy layer support beams.)
Figure 4
Unit 2 Catalyst Element (level 1) After 7.500 Hours of Operation
and With Protective Screen Removed
12
-------
_ NHi-Slip (Constant NOx Reduction)
0 20.000
40.000 60,000 80,000
Operating Time (h)
100.000 120.000
Figure 5
Relative Catalyst Activity and Ammonia Slip vs. Catalyst Loading Requirements
13
-------
SELECTIVE CATALYTIC REDUCTION:
SUCCESSFUL COMMERCIAL PERFORMANCE ON TWO
U.S. COAL-FIRED BOILERS
Paul Wagner, P.E.
U S Generating Company
Logan Generating Plant
Route 130 South
Swedesboro, New Jersey 08085-9300
Douglas Bullock, P. E.
U S Generating Company
Indiantown Generating Plant
19140 SWWarfield Blvd.
Indiantown, Florida 34956
RalfSigling
Siemens KWU
Freyeslebenstrasse 1
Erlangen, Germany D-91050
Robert Johnson
Siemens Power Corporation
1007 AlMansell Road
Roswell, Georgia 30076
Abstract
There is valuable operating experience in the United States with Selective Catalytic Reduction
NOx emission control on coal-fired boilers. Several systems have started up within the last
couple of years. Experience on two of these systems will be examined in this paper.
The authors will present and discuss the following relevant topics:
• full-scale SCR system design considerations in context with expected boiler operation;
• SCR system start-up and performance over the first several years of operation;
• how SCR catalyst management and recent catalyst addition are currently providing operating
flexibility for these boilers;
-------
how SCR system and catalyst enable economic and more efficient boiler operation with
reduced LOI levels;
life-cycle costs analysis of SCR operation and catalyst management.
Background
The Logan Generating Plant and the Indiantown Generating Plant are managed and
operated by U. S. Generating Company (US Gen). Each plant uses selective catalytic
reduction (SCR) technology to reduce NOx emissions. They, along with the Carney's
Point Generating Plant, are the first coal burning plants to utilize full-scale SCR
technology in the United States.
USGen is a wholly owned subsidiary of Pacific Gas and Electric Corporation. USGen
currently owns, operates and manages seventeen plants generating up to 3,400 MWe.
These plants sell electricity and process steam to various customers and industrial
firms across the United States.
Logan Generating Plant
The Logan Plant is a 218 MWe net cogeneration plant located in Logan Township,
Gloucester County, New Jersey. The plant provides electricity to Atlantic Electric. For
cogeneration, the plant provides up to 50,000 Ibs. per hour of process steam and 2
MWe to the Monsanto Delaware River Plant. Excess electricity is sold to the energy
market through wheeling agreements. The plant (Figure 1) includes a 2,660 psig
pulverized coal-fired steam generator with the SCR system, a dry scrubber using quick
lime and recycled flyash reagents for sulfur dioxide control, and a reverse air fabric
filter for particulate control. The plant is a zero-discharge facility in which all process
and waste waters are recycled through a lime/soda ash softener and reverse osmosis
technologies. Construction began in April, 1992 and the unit started commercial
operation in September, 1994.
Logan Generating Plant received one of the 1995 "Projects of the Year" award from
Power Engineering/Power Engineering International magazines.
Indiantown Generating Plant
The Indiantown Generating Plant (Figure 2) is a 330 MWe net coal-fired, co-generation
plant located in Indiantown, Florida. The plant sells electricity to Florida Power and
Light and provides up to 125,000 Ibs. per hour of steam to Caulkins Citrus. The steam
generator is rated at 2,400 psig, and the unit includes the SCR system, as well as a dry
scrubber and reverse air fabric filter. The plant uses agricultural runoff for makeup
water and is a zero liquid discharge facility. Softeners and evaporation equipment is
installed to recycle and reuse water internally. Indiantown began commercial operation
in December, 1995.
-------
SCR System Design
The initial design for Logan Plant included a Selective Non-Catalytic Reduction system
for NOx emission control. In the final design review stages, the decision was made by
USGen to eliminate the SNCR system and to install Selective Catalytic Reduction for
post-combustion NOx control. This decision was made for several technical and
commercial reasons. USGen and the boiler manufacturer became increasingly
concerned with the applicability of SNCR for the Logan plant due to the complexity of
the system and the requirements for load cycling. There was a higher degree of
confidence with the SCR system from an operational viewpoint. Combined with
commercial benefits such as improved debt service coverage, shortened construction
schedule, and improved heat rate, USGen concluded that SCR would be a more
effective , and more economical, operating system for controlling NOx emissions.
USGen decided to standardize on SCR technology for the three coal-fired plants at
Logan, Indiantown and Carneys Point. No changes in the air permits were required as
a result of this design change.
Logan Plant
The SCR system at Logan Plant was designed to control boiler NOx emissions to 0.17
Ib/MMBtu, from a NOx loading of 0.27 Ib/MMBtu. Ammonia slip was limited to 5 ppmvd,
corrected to 7% oxygen in the flue gas, at the end of 24,000 operating hours. 141.3m3
of Siemens' SINOx plate catalyst was installed initially in three element layers. Three
element layers were reserved for spares to accommodate future catalyst management
plans.
Indiantown Plant
The SCR reactor at Indiantown is similarly arranged, except that the initial loading of
Siemens plate catalyst ( 161.4 m3) was installed in 2 element layers. A 40% NOx
emission reduction was the initial design requirement. Ammonia slip was also limited to
5 ppmvd, corrected to 7% oxygen. The primary difference, pertaining to the SCR
systems, in the two plants lies in the air permits.
Siemens' SINOx plate catalyst was chosen by the SCR system supplier for installation
at the Logan and Indiantown SCR plants. Siemens has furnished SCR catalyst in over
100 SCR systems on coal and oil-fired boilers in Europe and the United States, and
has compiled over 1,000,000 operating hours of service since 1988. The plate catalyst
is preferred for coal-fired applications. Due to the greater open area compared to
honeycomb structured catalyst, the plate catalyst provides excellent NOx emission
control with a lower pressure drop impact on the unit. It also is extremely resistant to
both fly ash deposition and erosion. Additionally the mechanical design of the plate
catalyst affords greater resistance to thermal and mechanical stresses that are normal
for power boiler operation. The Siemens plate catalyst is very resistant to arsenic
poisoning and can be designed to minimize sulfur oxidation rates. Operating
-------
experience with plate catalyst on a variety of boiler types has demonstrated the ability
to provide consistent performance over varying load conditions.
The following table presents the SCR catalyst design information for the Logan and
Indiantown plants.
PLANT LOGAN INDIANTOWN
NOx inlet loading (Ib/MMBtu) 0.27 0.25
NOx reduction efficiency (%) 63 40
Ammonia slip (ppmvd@ 7% O2) 5 5
Ammonia slip (ppmvd @ act. O2) 6.2 6.2
Initial Operating period (hr) 24,000 24,000
Operation to date
The SCR systems have performed well over the first few years of commercial
operation. NOx and ammonia emissions have been maintained under normal
parameters. Other subsystem deficiencies have affected the overall reliability of the
SCR, but USGen and Siemens have implemented a number of solutions that will
improve long-term operation.
Low-NOx Burners
The low-NOx burners coupled with overfire air systems were designed to minimize NOx
emissions from the boilers, ranging from 0.27 Ibs/MMBtu at Logan to 0.30 Ibs/MMBtu at
Indiantown. To date, furnace NOx emissions have generally exceeded 0.30 Ibs/MMBtu
at both plants. These higher than design levels have forced the SCR system to control
the excess NOx in order to maintain compliance with permit requirements. As a result,
more ammonia has been needed, and this has resulted in higher levels of unreacted
ammonia.
High loss on ignition (LOI) levels in the fly ash have been experienced in both plants,
as a result of the non-performance of the burners. As boiler NOx was lowered by
reducing excess oxygen, the LOI and carbon monoxide (CO) levels increased. The
Logan boiler is quite sensitive to changes in excess oxygen, so the boiler has
experienced fluctuations in both NOx and CO emissions.
The situation at Indiantown has been very similar. The plant has tried tuning the
burners, with the supplier, numerous times in attempts to optimize NOx emissions with
load swings and dispatch requirements. The optimization program was started by
Bechtel Startup and Foster Wheeler Energy, the company that furnished the plant.
Due to contractual requirements, the entire program was conducted outside of USGen's
scope. USGen was not happy with the results of this program, due to poor boiler
stability, ramping capabilities and overall combustion. Due to schedule slippage, this
optimization program was condensed into a 2-3 week period. The results, confirmed by
additional tests, indicated that combustion, and furnace NOx levels, could not be
improved without changes and modifications to other boiler auxiliaries.
-------
At the Logan Plant, a combustion optimization program was implemented to improve
emissions and reduce fly ash LOI. Over forty different tests were conducted to
determine the best settings of register dampers, adjustable tips and overfire air
quantities. USGen was able to determine the following from these series of tests:
1. Minimization of CO spikes was achieved by closing the two lower fire air ports on
the front wall.
2. Biased overfire air flow to the rear wall had the most improvement in flyash LOI,
approximately 3%. However, LOI levels fluctuated between 18 and 28%.
3. Achieved balanced NOx and oxygen emissions at the economizer outlet test grid.
At full load, NOx was in the range of 0.35 to 0.39 Ib/MMBtu.
4. SO2to S03 conversion in the SCR was well below design levels.
5. Ammonia slip above 5 ppmvdc at the air preheater gas inlet was not detected at the
stack. The high levels of slip and the frequency of air preheater washes prompted
the addition of another half-layer of fresh catalyst.
USGen discontinued the program due to loss of support from the boiler manufacturer,
but continued to experience both high boiler outlet NOx and high fly ash LOI.
Unfortunately, the Logan boiler was designed with a short distance from the overfire air
ports to the furnace nose arch, approximately 15 feet. This has resulted in minimal
residence time to completely burn out carbon char. The end result is high fly ash LOI.
Additional test programs are being planned to optimize combustion by concentrating
efforts on fuel and air management.
Impacts of SCR system performance
The SCR system at either plant was not designed to provide the NOx removal
efficiency required with these new operating conditions. While the system controlled
NOx emissions to the required levels at each plant, higher ammonia consumption rates
were needed to maintain the appropriate NH3 : NOx stoichiometry. These higher rates,
and the resultant ammonia slip through the catalyst caused some balance of impacts
that were unacceptable to USGen.
Figure 3 illustrates the projected ammonia slip, and associated catalyst management
plan, for the Logan Plant under design conditions. Separately plotted are several data
points representing measured ammonia slip data at approximately 8,000 hours of
service. It is clear that the measured data is significantly different from the projected.
This is indicative of a critical imbalance among catalyst volume, NOx loadings,
ammonia consumption and ammonia slip control.
The increased slip led to accelerated rates of air preheater fouling due to ammonium
bisulfate formation. Both plants began to monitor air heater differential pressure to
determine boiler operating time between air heater washes, and as an indirect
ammonia slip measure. This information was useful for providing some guidance in
understanding the operation of the SCR system, but clearly it was unsatisfactory for
-------
identifying the primary problem related to the higher than design furnace NOx
emissions.
Ammonia slip in the gas stream is very difficult to measure accurately on a frequent
basis. It is common practice in European installations to assess ammonia slip from the
SCR process by quantifying ammonia by weight in the unit's fly ash. Trends
established by this process are very similar to those presented in the catalyst
management plans. The comparative ease of measuring ammonia in this fashion make
it a low-cost alternative to costly testing and monitoring, and it provides a very
predictable indication of ammonia slip. Both Logan and Indiantown began to use this
measurement in 1996 as a tool for improving system diagnostic capabilities. It is
Siemens' experience that 200 ppm of ammonia by weight in the fly ash is a threshold
for most units, in terms of balance of plant impacts.
Indiantown decided to install additional catalyst because it was unacceptable to shut
the unit down every ten weeks to wash the air heater. During outages to wash the air
heaters so that pressure drop could be lowered, ash sample analyses confirmed that
ammonium bisulfates were the cause of the pluggage. This formation was the result of
excessive ammonia slip through the SCR. Indiantown personnel believed the answer
was in the addition of SCR catalyst, but they were uncertain of the reasons. Support
from the SCR system supplier was lacking, so USGen turned to Siemens for an
understanding of the problem. It was through these discussions that USGen learned
that the design basis for the SCR system was founded on unrealistic low-NOx burner
expectations. Working together, USGen and Siemens have resolved the operating
problems. There has been no need to wash the air heaters in the last nine months of
operation.
Based on the high frequency of air preheater washes, high levels of ammonia in the fly
ash samples and greater than 5 ppmvdc slip while operating near the NOx permit limit,
USGen exercised their option to install fresh catalyst under warranty agreements with
the boiler manufacturer. A half layer of catalyst was procured and installed during the
September, 1996 annual plant outage. The half layer was installed beneath an existing
layer. Following the installation of fresh catalyst, an immediate drop in ammonia in fly
ash levels was noted. Long term improvements were achieved in reducing air
preheater fouling rates due to slip while slowly reducing NOx emissions below permit
levels.
Figure 4 illustrates the relationships among ammonia slip and catalyst volume as
functions of operating temperature and required NOx reduction efficiency. The lines
are iso-ammonia slip and would be specific only for a given unit. Each unit, in other
words, would have a curve appropriate for specific design conditions. This particular
figure was developed for the Logan Plant and shows the ammonia slip design point at
the 24,000 operating hour point. To maintain a 5 ppm ammonia slip with an increased
NOx reduction efficiency requirement, either a reduced operating period must be
accepted or additional catalyst must be installed to maintain the 24,000 hour period.
-------
USGen decided to increase SCR catalyst volume at both plants to restore an operating
balance between efficiency and ammonia slip. The immediate benefits to unit
performance can be seen in Figure 5. Prior to the installation of the new catalyst,
ammonia concentrations in the Logan fly ash were approximately 180 to 210 ppm by
weight. These dramatically dropped to 60 ppm upon startup with the additional volume
of catalyst.
Impacts on long-term performance
USGen is currently evaluating a number of operating options for the Logan and
Indiantown plants, as a result of these changes in SCR operation. At both plants, the
long term catalyst management plans have been altered by the need for higher
reduction efficiencies. This alone has impacted the financial strategies for the plants
by increasing the projected operating costs over a thirty year life cycle. As can be seen
in Figure 6, the frequency of catalyst replacement is higher compared to the original
design. These costs are also affected by higher ammonia consumption rates.
It is also known that the design ammonia slip limit of 5 ppmvd at 7% oxygen is probably
too high for consistent boiler operation. Past performance indicates that the slip should
be limited to 2 or 3 ppmvd to avoid the balance of plant impacts that both Logan and
Indiantown plants have experienced. Of course this magnifies the projected increase in
long-term operating costs by shortening the intervals between catalyst addition or
exchange. USGen and Siemens are evaluating several options to determine the most
cost effective ammonia slip and management plan for each plant.
Benefits of the SCR system
However, USGen has also recognized that the SCR system affords a number of
benefits in terms of overall boiler performance.
There is relatively little impact on operations, in that maintenance requirements are
minimal. USGen has concluded that the cost of SCR is insignificant to the cost of
producing electricity.
Knowing that the SCR system can handle higher NOx loadings than design, USGen
has realized that future operating costs can be mitigated by improvements in boiler
efficiency. In tuning unit burners to reduce NOx emissions from the furnace, USGen
has accepted loss on ignition levels in excess of 30%, and a corresponding boiler
efficiency of approximately 86%. The benefit of the SCR systems is the ability to
accept a higher NOx loading and to allow the boiler to operate at higher efficiencies
with a much lower LOI level. USGen is currently testing the boilers at Logan and
Indiantown in order to determine the optimal operating points, while letting the SCR
system absorb the fluctuations in NOx loadings.
-------
It is expected that long-term costs can be improved by a reduction in maintenance
associated with burner operation and parts, and boiler corrosion.
USGen has concluded that the SCR systems will maintain compliance while controlling
ammonia slip to optimal levels at each of the plants. Proper planning will also help US
Generating choose a catalyst management plan according to the current and future
operating conditions at each plant, and in conjunction with future unit outages.
References
P A. Wagner, G. F Weidinger, W. C. Talbot, D. W. Bullock, "Multiple Coal Plant SCR
Experience-A U. S. Generating Company Perspective," presented at the ICAC Forum
'96, Baltimore, Maryland (March 1996)
D Beckham, "Horses and Carts: The Roadmap to Productive Electric Competition,"
presented at the Cost-Effective Reduction Strategies for Power Companies conference,
Washington, D. C. (November, 1996)
-------
Coal Bargi HBJ! and
g Facility
Urn* / Sirry Prepantfon Ar«a
Bmfici&l R«u»
Figure 1. Logan Generating Plant
Powwfor
Atlantic B»ctric
Unw Pnporadon ATM
Figure 2. Indiantown Generating Plant
Seite 1
-------
n Generating Plant
1.0 -s.
0.9 -
0.8 -
0.7 •
0.6 -
0.5 -
0.4 -
03 -
20.000 Operating 40.000
time(h)
60.000
Figure 3. Relative Activity (k/k,,), Ammonia Slip (ppmvdc) vs. Operating Time
iso-ammonia slip NH3slipeor<5ppm
Figure A. Ammonia Slip and Catalyst Volume vs. Operating Time
Seite 2
-------
Logan Ammonia In Ftyash Trends
(average of len and right sides Q fun load conditions)
alyst Instaled 10/11/66 1
3/2*96 Slliree
1W1MO 12/1« 1/30/ST 3/1VS7
Figure 5. Ammonia in fly ash trends
SCR Designed Outlet NOx @ 0.10 Ib/mmbtu
0 10K 20K 30K «K 50K 60K 70K 80K
•^~ Original Design Boiler operating Hours — — Projected w/ Boiler
NOx @ 0.27 Ib/mbtu NOx © 0.37 Ib/mbtu
O S/96 APH Inlet Test
Figure 6. Logan Catalyst Management Plans
Ammonia Slip (ppmv) Trends
Seite 3
-------
Successful Implementation of Cormetech Catalyst
in High Sulfur Coal-Fired SCR Demonstration Project
Chris E. DiFrancesco
Phone: 919-620-3015
Fax: 919-620-3001
CORMETECH, INC.
5000 International Drive
Durham, NC 27712
Scot G. Pritchard
Phone: 919-620-3019
Fax: 919-620-3001
CORMETECH, INC.
W. Scott Hinton
W. S. HINTON & ASSOCIATES
Phone: 904-478-2400
Fax: 904-484-4640
2708 Woodbreeze Drive
Cantonment, Florida 32533
Abstract
The U.S. Department of Energy (DOE), Electric Power Research Institute (EPRI),
Ontario Hydro, and Southern Company Services (SCS) jointly funded a project under
the Innovative Clean Coal Technologies (ICCT) Program to demonstrate the
capabilities of Selective Catalytic Reduction (SCR) technology on high sulfur U.S.
coal. The demonstration site was at Gulf Power Company's Plant Crist Unit No. 5
(75 MW capacity) near Pensacola, Florida. The demonstration was completed in July
1995.
Cormetech was one of a number of catalyst manufacturers that participated in the
program. Cormetech supplied catalyst for two (2) small-scale SCR reactors, one high
dust and one low dust. The high dust catalyst was in operation for 10,600 hours and
the low dust catalyst was in operation for 5,800 hours. Required performances,
including NOx removal and SO2 oxidation, were maintained during the demonstration
for both reactors. Moreover, the catalysts are projected to have met required
performances well beyond the duration of the demonstration.
The report included herein details the primary field test results performed by SCS and
catalyst test results performed by Cormetech during the test period for the reactors
containing Cormetech catalyst. Specific results and their impact are discussed,
including changes in catalyst performance and properties over time.
-------
I. BACKGROUND
The U.S. Department of Energy (DOE), Electric Power Research Institute (EPRI),
Ontario Hydro, and Southern Company Services (SCS) jointly funded a project under the
Innovative Clean Coal Technologies (ICCT) Program to demonstrate the capabilities of
Selective Catalytic Reduction (SCR) technology on high sulfur U.S. coal. The results of
the project are summarized in Topical Report Number 9, Clean Coal Technology -
Control of Nitrogen Oxide Emissions: Selective Catalytic reduction (SCR) by The U.S.
Department of Energy and Southern Company Services, Inc, May 1997.
The demonstration site was at Gulf Power Company's Plant Crist Unit No. 5 (75 MW
capacity) near Pensacola, Florida. The demonstration facility includes a total of nine (9)
SCR reactors which were run in parallel. Three (3) 2.5 MWe reactors and six (6) 0.2
MWe. All reactors represent high dust applications (upstream of hot-side ESP) except
one 0.2 MWe reactor which was configured as a low dust application (downstream of hot-
side ESP).
The two year demonstration project began in June 1993 and concluded in July 1995.
Corraetech was one of a number of catalyst manufacturers that participated in the
program. Cormetech designs and manufactures honeycomb catalyst of homogeneous
composition for SCR based on licensed technology of Mitsubishi Chemical Corporation
and Mitsubishi Heavy Industries. Developmental catalyst was not employed on this
project. The specific licensed catalyst technology used has been employed world-wide on
a total of 400 units including 75 coal-fired boilers.
Cormetech supplied catalyst for two (2) 0.2 MWe SCR reactors, one high dust and one
low dust. The 0.2 MWe reactors were approximately one (1) square foot in cross-section
and consisted of three (3) and two (2) layers of catalyst respectively. The catalyst for the
low dust reactor was installed in April 1994 as a substitute for another catalyst vendor that
withdrew from the test Therefore, the total number of operating hours was somewhat
less for the low dust catalyst versus the high dust.
SCS managed the project from permitting to engineering and construction, as well as, all
field operation and testing.
Sootblowing was used regularly on all reactors. The 2.5 MWe reactors were equipped
with automatic rake type sootblowers, while the remaining reactors were manually air
blown.
In addition to field tests on each reactor, catalyst samples were pulled and returned to
each respective catalyst manufacturer. Each manufacturer was responsible for testing and
reporting on the state of their catalyst to the project funders.
Pagel
-------
This paper presents specific field and laboratory data for Cormetech's catalyst. The data is
compared to requirements showing that the demonstration was successful. Key flue gas
and ash data for Plant Crist are provided hi the Appendix to define the conditions under
which the SCR was operated.
H. DESIGN CONDITIONS
Performance Design Conditions
Temperature, °F
Superficial Velocity, ft/s
O2, vol % wet
Inlet NOx, ppmv
Inlet SOx, ppmv
Molar Ratio, NH3/NOx
NOx Conversion Target, %
Maximum Allowable NH3 slip, ppmv
Maximum Allowable Pressure Drop, in H2O
Maximum Allowable SO2 Oxidation, %
Number of Full Size Layers
Catalyst Pitch, mm
Catalyst Length, mm
Space Velocity (SV), Hr"1 @ 32°F, 1 atm
700
18.1
3
400
-2000
(-3% S in fuel)
0.8
80
5
4
0.75
3 (High dust)
2 (Low dust)
7.1 (High dust)
3.7 (Low dust)
1000 (High dust)
600 (Low dust)
2776 (High dust)
7033 (Low dust)
-------
m. PERFORMANCE RESULTS
A. High Dust (10,600 hours in operation)
Field Results
Ammonia slip remained < 1 ppmv for the duration of the demonstration. This was well
below the 5 ppmv maximum allowable slip. No change in ammonia slip over time was
detectable (Figure 1).
Ammonia Slip - Field Measurements
Design Conditions
4--
2--
Max Allowable
0 2000 4000 6000 8000 10000
Exposure Time (hrs)
Figure 1
SC>2 oxidation rate remained well below the 0.75% maximum allowable rate for the
duration of the demonstration. Average SO2 oxidation rate was below 0.4%. No change
in SC>2 oxidation rate over time was detectable (Figure 2).
1.5-
>
1 1-
0.5 -
SO2 Oxidation - Field Measurements
Design Conditions
Max Allowable •
ta -----;
" • B !
D 2000 4000 6000 8000 10000
Exposure Time (hrs)
Figure 2
Pressure drop remained below the 4 inches H2O maximum for the duration of the
demonstration.
-------
Lab Performance Results
Field measurements were subject to more inaccuracies than measurements from a pilot
scale laboratory reactor. The scatter in the field data makes it difficult to detect changes in
catalyst performance. In order to more accurately measure the change in catalyst
performance over time, full-sized catalyst samples were tested fresh and at the end of the
demonstration in a pilot scale laboratory reactor. Such periodic testing is typical for SCR
systems in order to assure proper operation and manage catalyst life. For a description of
the pilot scale test, refer to "Quality Assurance of SCR Catalysts for the Southern
California Edison 480 MW Power Generating Plants Through Laboratory and Field
Performance Testing", Chris DiFrancesco, et. al., ICAC Forum '94.
The tests were performed at the design conditions. The catalyst pulled from the reactor
was tested "as is" without any cleaning. Only the first two layers of catalyst were
evaluated so that ammonia slip would be detectable (SV = 4164 hr-1). Ammonia slip
increased over time from 0.7 -1.4 ppmv, well below the 5 ppmv maximum, even with
only two catalyst layers. (Figure 3)
>
n.
a.
Ammonia Slip - Lab Measurements
E
10-
8-
6-
4-
2-
o'
(esign Conditions - 2 Layers SV=4164 hr-1
Max Allowable
, . -
..»••*"" Projected
3 5000 10000 15000 20000
Exposure Time (hrs)
. * *"
25000
Figure 3
Based on this rate of change in performance, we predict that the ammonia slip for two
layers of catalyst would remain below the 5 ppmv maximum for approximately an
additional 15,000 hours.
Pa
-------
Lab Chemical Analysis Results
Chemical analyses of field sample pulled at 10,600 hours was conducted by X-ray
Fluorescence. It was determined that the decrease in performance over the duration of the
demonstration was mainly due to a combination of arsenic (As) accumulation, surface
masking by fly ash components (Ca, Fe), and alkaline metal accumulation (K). The graph
below illustrates the observed increases in the X-ray intensities of these elements relative
to the total X-ray intensities of the titanium catalyst base. (Figure 4)
c
V
c
CB
u
F
o>
=
0>
a
Change in X-ray Intensities ratio of
Surface Contanrinants/Ti -vs- Fresh
0.10-
0.08-
0.06-
0.04 •
0.02-
^§ J
^^ r™i I
Irrj ^^ I
,11, ^ , 1 i
^ u u> + u- O <
^
Element
Figure 4
These performance deterioration factors are typical for coal fired applications and is
consistent with the coal analysis in the appendix and the experience of Cormetech and its
licensors. For a description of the deterioration mechanisms, refer to "Optimizing SCR
Catalyst Design and performance for Coal-Fired Boilers", by Scot Pritchard, et al.,
EPRI/EPA 1995 Joint Symposium on Stationary Combustion NOx Control.
PageS
-------
B. Low Dust Summary (5,800 hours in operation)
Field Results
Operation time was limited on this reactor. Pressure drop was somewhat erratic (4 - 8 in
H2O) caused by plugging of the small pitched catalyst due to unexpected carry-over of
large paniculate to the "low" dust reactor. More than 30% of the catalyst was plugged
with fly ash. This carry-over was due to the long duct runs of the test facility and a less
than optimum flue gas take-off scoop. This situation would not be expected in a full scale
unit
Despite these operational issues, ammonia slip remained below 1 ppmv for the duration of
the demonstration, well below the 5 ppmv maximum allowable slip. Similar to the high
dust case, no change in ammonia slip over time was detectable in the field.
SO2 oxidation rate remained well below the 0.75% maximum allowable rate for the
duration of the demonstration. Similar to the high dust case, average SC>2 oxidation rate
remained below 0.4% and no change in SO2 oxidation rate over time was detectable.
Lab Results
As with the high dust reactor, fresh samples and samples removed at the end of the
demonstration run were tested in the laboratory reactor. Although the reactor as a whole
was 30% plugged, the particular samples tested were only 3% plugged. Ammonia slip
increased over time from 0.6 -1.0 ppmv, well below the 5 ppmv max. Based on this rate
of change in performance, we predict that the ammonia slip would remain below the 5
ppmv maximum for more than 15,000 hours, excluding impact of the overall severe
plugging.
Through chemical and physical property analyses, it was determined that the very slight
decrease in performance over time was due mainly to a small amount of arsenic
accumulation.
Page 6
-------
VI. CONCLUSIONS
For the high dust application, deNOx performance (catalyst deactivation), SOo oxidation,
and pressure drop remained within design limits. Performance is expected to have lasted
much longer than the duration of the demonstration even with only two-thirds of the
reactor charge. Deterioration mechanisms and impact were consistent with expectations
based on the coal composition fired.
Also, note that the low ammonia slip values achieved are consistent with current design
practices (limit < 2-3 ppmv) to avoid fly ash contamination and excess air pre-heater
maintenance. Design ammonia slip limits are unit specific depending on ash disposal
method, sulfur content, and air pre-heater design.
For the low dust application, deNOx performance (catalyst deactivation) and SO2
oxidation remained within design limits even with 30% of the catalyst plugged. Pressure
drop increased significantly due to the plugging but was a result not realizing a truly low
dust situation. If a truly low dust situation was realized, the catalyst performance, as in
the high dust case, is expected to have lasted much longer than the duration of the
demonstration.
Page 7
-------
APPENDIX
Boiler Type:
Particulate Control:
Design Fuel Analysis:
Tangentially-Fired, Dry-bottom
Hot and Cold-side Electrostatic Precipitator
C, wt %
H, wt %
S, wt %
N, wt%
Cl, wt %
Ash, wt %
Moisture, wt %
Oxygen, wt% (by diff.)
67.80
4.60
2.90
1.40
0.25
9.50
7.90
5.65
Actual Fuel Analysis from March 1993 to July 1995 Based on Monthly As-Burned
Composites. Alabama Power Company Results, Dry Basis
Test
Moisture, Total
Ash
Gross Caloric Value
Sulfur, Total
Sulfur, Ib/MMBtu
Carbon
Hydrogen
Nitrogen
Oxygen
Carbon, Fixed
Volatile Matter
Continued on Next Page
Method
ASTM D 3302
ASTMD3180
ASTM D 3180
ASTM D 3180
ASTM D 3180
ASTM D 3180
ASTM D 3180
ASTM D 3180
ASTM D 3180
ASTM D 3180
ASTM D 3180
Units
% by Wt
% by Wt
Btu/lb
% by Wt
Ib/MMBtu
% by Wt
% by Wt
% by Wt
% by Wt
%byWt
% by Wt
Ave.
10.87
9.30
13268
2.58
1.95
74.82
5.00
1.25
6.73
52.83
37.88
Std.
Dev.
0.97
0.63
130
0.04
0.31
0.81
0.07
0.03
0.66
1.31
1.17
PageS
-------
Continued
Test
Aluminum
Antimony
Arsenic
Barium
Beryllium
Cadmium
Calcium
Chlorine
Chromium
Cobalt
Copper
Flourine
Iron
Lead
Lithium
Magnesium
Manganese
Mercury
Molybdenum
Nickel
Phosphorous
Potassium
Selenium
Silica
Sodium
Titanium
Vanadium
Zinc
Method
ASTM D 3682
ASTM D 3683
Sec. Chem. Acta. 44B
ASTM D 3683
ASTM D 3683
ASTM D 3683
ASTM D 3682
ASTM D 4208
ASTM D 3683
ASTM D 3683
ASTM D 3683
ASTM D 3761
ASTM D 3682
Sec. Chem. Acta. 44B
ASTM D 3683
ASTM D 3682
ASTM D 3683
ASTM D 3684
ASTM D 3683
ASTM D 3683
ASTM D 3682
ASTM D 3682
Sec. Chem. Acta. 44B
ASTM D 3682
ASTM D 3682
ASTM D 3682
ASTM D 3683
ASTM D 3683
Units
% by Wt
mg/kg
mg/kg
mg/kg
mg/kg
mg/kg
% by Wt
mg/kg
mg/kg
mg/kg
mg/kg
mg/kg
% by Wt
mg/kg
mg/kg
% by Wt
mg/kg
mg/kg
mg/kg
mg/kg
% by Wt
% by Wt
mg/kg
% by Wt
%byWt
% by Wt
mg/kg
mg/kg
Ave.
1.09
<1.0
L_ 3-2
40
3
<1.0
0.24
1767
19
7
9
56
1.08
11.7
9
0.06
24
0.07
7.78
15
0.02
0.20
<2
2.27
0.06
0.06
41
39
Std.
Dev.
0.11
1.9
18
1
-
0.03
812
4
2
2
27
0.17
4.5
5
0.02
4
0.04
4.33
2
0.02
0.06
-
0.19
0.02
0.01
10
29
Page 9
-------
Design Flue Gas Composition:
Design Flyash Composition:
N2, vol %
O2, vol %
CO2, vol %
H2O, vol %
SO2, ppmv
SOs, ppmv
NOx, ppmv
Hcl, ppmv
73.29
3.01
13.82
9.61
2210
20
400
104
SiO2, wt %
A12O3, wt %
Fe2O3, wt %
TiO2, wt %
CaO, wt %
MgO, wt %
K2O, wt %
Na2O, wt %
SOs, wt %
P2O5, wt%
LOI, % (typ represents UC)
50.4
19.9
18.1
1.0
4.2
1.0
2.6
0.7
1.4
0.3
6.5
Page 10
-------
Design Particulate Loading:
Average: High Dust - 8000 mg/Nm
Range: 6000-11,000
Average: Low Dust 30 mg/Nm Avg.
Design Particle Size Distribution
High Dust
20 40 60 80
Diameter, Microns
100
Page 11
-------
NOTES
-------
NOTES
-------
NOTES
-------
NOTES
-------
NOTES
------- |