United States
Environmental Protection
Agency

_ ^

oEPA Research and

Development

PROCEEDINGS: 1991 JOINT SYMPOSUM ON
STATIONARY COMBUSTION NOx CONTROL
WASHINGTON, D, C., MARCH 25~28, 1991
Volume 2. Sessions .4 and 5

EPA- 600 / R- 92- 093b
July1992

Prepared for

Office of Air Quality Planning and Standards

Prepared by

Air and Energy Engineering Research
Laboratory

Research Triangle Park NC 27711


-------
EPA REVIEW NOTICE

This report has been reviewed by the U.S. Environmental Protection Agency, and
approved for publication. Approval does not signify that the contents necessarily
reflect the views and policy of the Agency, nor does mention of trade names or
commercial products constitute endorsement or recommendation for use.

This document is available to the public through the National Technical Informa-
tion Service, Springfield, Virginia 22161.

Copyright (c) 1991, EPRI GS-7447, Proceedings;
1991 Joint Symposium on Stationary Combustion
NOx Control, Volumes 1, 2, and 3. Since this
work was, in part, funded by the U. S. Government,
the Government is' vested with a royalty-free, non-
exclusive, and irrevocable license to publish, trans
late, reproduce, and deliver that information and
to authorize others to do so.


-------
TECHNICAL REPORT DATA /—
(Please read Instructions on the reverse before comply

l. REPORT NO.

EPA-600/R-92-093b

4. TITLE AND SUBTITLE

Proceedings; 1991 Joint Symposium on Stationary
Combustion NOx Control, Washington, D. C. , March
25-28, 1991, Volume 2. Sessions 4.and 5	

i II! IllII!IllIIII

PB93-212850

5. REPORT DATE

July 1992

6, PERFORMING ORGANIZATION CODE

7. AUTHOR(S)

Carolee DeWitt, Compiler

8. PERFORMING ORGANIZATION REPORT NO.

9. PERFORMING ORGANIZATION NAME ANO ADDRESS

William Nesbit and Associates

1221 Farmers Lane

Santa Rosa, California 95405

10. PROGRAM ELEMENT NO.

11. contract/grant no.

NA (EPRI Funded)

12, SPONSORING AGENCY NAME AND ADDRESS

EPA, Office of Research and Development
Air and Energy Engineering Research Laboratory
Research Triangle Park, North Carolina 27711

13. TYPE OF REPORT AND PERIOD COVERED

Proceedings; 3/89 ~ 3/91

14, SPONSORING AGENCY CODE

EPA/600/13

15. supplementary NOTES AEERL project officer is Robert E. Hall, Mail Drop 65, 919/541-
2477, Volume 1 includes Sessions 1-3, and Volume 3 includes Sessions 6-8.

is. abstract proceedings document the 1991 Joint Symposium on Stationary Combus-
tion NOx Control, held in Washington, DC, March 25-28, 1991. Jointly sponsored by
EPRI and EPA, the symposium was the sixth in a biennial series devoted to the
international exchange of information on recent technological and regulatory develop-
ments for stationary combustion nitrogen oxides (NOx) control. Topics covered inclu-
ded the significant increase in active full-scale retrofit demonstrations of low-NOx
combustion systems in the U.S. and abroad over the past 2 years; full-scale oper-
ating experience in Europe with selective catalytic reduction (SCR); pilot- and bench-
scale SCR investigations in the U. S.; increased attention to selective noncatalytic
reduction (SNCR) in the U. S. ; and NOx controls for oil- and gas-fired boilers. The
proceedings are published in three volumes.

17.
a.

KEY WORDS AND DOCUMENT ANALYSIS

DESCRIPTORS

b.IDENTIFIERS/OPEN ENDED TERMS

c. COSATI Field/Group

Pollution
Nitrogen Oxides
Combustion
Catalysis
Fossil Fuels

Pollution Control
Stationary Sources
Catalytic Reduction
Noncatalytic Reduction

13 B
07B
2 IB
07D
21D

18. DISTRIBUTION STATEMENT

Release to Public

19. SECURITY CLASS (This Report/

Unclassified

21. NO. OF PAGES

474

20. SECURITY CLASS (This page/

Unclassified

22. PRICE

EPA Form 2220-1 (3-73)

5B-105


-------
EPA-600/R-92-093b
July 1992

PROCEEDINGS:

1991 JOINT SYMPOSIUM ON STATIONARY COMBUSTION NOx CONTROL

Washington, D.C., March 25-28, 1991

Volume 2. Sessions 4 and 5

Compiled by

Carolee DeWitt
William Nesbit and Associates
1221 Farmers Lane
Santa Rosa, CA 95405

EPA Project Officer:
Robert E. Hall

EPRI Project Manager:
Angelos Kokkinos

U.S. Environmental Protection Agency
Air and Energy Engineering Research Laboratory
Research Triangle Park, NC 27711

Electric Power Research Institute
3412 Hillview Avenue
Palo Alto, CA 93404

Prepared for:

U.S. Environmental Protection Agency
Office of Research and Development
Washington, DC 20460

Electric Power Research Institute
3412 Hillview Avenue
Palo Alto, CA 93404


-------

-------
ABSTRACT

The 1991 Joint Symposium on Stationary Combustion N0„ Control was held in Washington, D.C.,
March 25-28,1991, Jointly sponsored by EPRI and EPAVthe symposium was the sixth in a biennial
series devoted to the international exchange of information on recent technological and regulatory
developments for stationary combustion NOx control. Topics covered included the significant
increase in active full-scale retrofit demonstrations of low-NOx combustion systems in the United
States and abroad over the past two years; full-scale operating experience in Europe with selective
catalytic reduction (SCR); pilot- and bench-scale SCR investigations In the United States; increased
attention on selective noncatalytic reduction in the United States; and NOx controls for oil- and gas-
fired boilers. The symposium proceedings are published in three volumes.

. • . r

ii


-------
PREFACE

The 1991 Joint Symposium on Stationary Combustion NOx Control was held March 25-28,1991, in
Washington, D.C, Jointly sponsored by EPRI and EPA, the symposium was the sixth in a biennial
series devoted to the international exchange of information regarding recent technological and
regulatory developments pertaining to stationary combustion NO„ control. Topics discussed
included the significant increase in active full-scale retrofit demonstrations of low-NO, combustion
systems in the United States and abroad over the past two years; full-scale operating experience in
Europe with selective catalytic reduction (SCR); pilot-and bench-scale SCR investigations in the
United States; increased attention on selective noncatalytic reduction in the United States; and NOx
controls for oil- and gas-fired boilers.

The four-day meeting was attended by approximately 500 individuals from 14 nations. Sixty-six
papers were presented by EPRI and EPA staff members, domestic and foreign utility companies,
federal and state government agencies, research and development organizations, equipment
vendors from the United States and abroad, and university representatives.

Angeios Kokkinos, project manager in EPRI's Generation & Storage Division, and Robert Hall,
branch chief, Air & Energy Engineering Research Laboratory, EPA, cochaired the symposium. Each
made brief introductory remarks. Michael R. Deland, Chairman of the President's Council on
Environmental Quality, was the keynote speaker. Written manuscripts were not prepared for the
introductory remarks or keynote address and are therefore not published herein.

The Proceedings of the 1991 Joint Symposium have been compiled in three volumes. Volume 1
contains papers from the following sessions:

¦	Session 1; Background

¦	Session 2: Large Scale Coal Combustion I

¦	Session 3: Large Scale Coal Combustion II

iii


-------
Papers from the following sessions are contained in Volume 2:

-Session 4A: v Combustion NOx Developments I .

Session 4B: Large Scale SCR Applications

Session 5A: Post Combustion Developments I

Session 5B: Industrial/Combustion Turbines on NO^.Control.

Papers from the following sessions are contained in Volume 3:

Session 6A
Session 6B
Session 7A
Session 7B
Session 8:

Post Combustion Developments II
Combustion NOx Developments II
New Developments I
New Developments II
Oil/Gas Combustion Applications

An appendix listing the symposium attendees is included at the end of Volume 3.

iv


-------
CONTENTS

Paper	VOLUME 1	page

SESSION 1:	BACKGROUND

Chair: I, Torrens, EPRI

"NOx Emissions Reduction in the former German Democratic Republic," B. Kassebohm

and S. Streng	1-1

"Top-Down* BACT Analysis and Recent Permit Determinations," J, Cochran and M. Fagan 1-15

"Analysis of Retrofit Costs and Performance of NOx Controls at 200 U.S. Coal-Fired

Power Plants," T. Emmel and M. Maibodi	1-27

"Nitrogen Oxides Emission Reduction Project," L Johnson	1-47

"The Global Atmospheric Budget of Nitrous Oxide," J. Levine	1-65

SESSION 2:	URGE SCALE COAL COMBUSTION I

Chair: B. Martin, EPA and G. Offen, EPRI

"Development and Evolution of the ABB Combustion Engineering Low NOx Concentric
Firing System," J. Grusha and M, McCartney	2-1

"Performance of a Large Cell-Burner Utility Boiler Retrofitted with Foster Wheeler

Low-NOx Burners," T. Lu, R. Lungren, and A. Kokkinos '	2-19

"Design and Application Results of a New European Low-NO_ Burner," J. Pedersen and
M. Berg	2-37

"Application of Gas Reburning-Sorbent injection Technology for Control of

NOx and SOj Emissions," W. Bartok, B. Folsom, T, Sommer, J. Opatrny, E. Mecchia,

R. Keen, T. May, and M. Krueger	2-55

"Retrofitting of the Italian Electricity Board's Thermal Power Boilers," R. Tarli, A. Benanti,

G. De Michele, A. Piantanida, and A. Zennaro	2-75

"Retrofit Experience Using LNCFS on 350MW and 165MW Coal Fired Tangential Boilers,"

T. Hunt, R, Hawley, R. Booth, and B. Breen	2-89

"Update 91 on Design and Application of Low NOx Combustion Technologies for Coal

Fired Utility Boilers," T. Uemura, S. Morita, T. Jimbo, K, Hodozuka, and H. Kuroda	2-109

v


-------
Paper

Page

SESSION 3:

LARGE SCALE COAL COMBUSTION II
Chair: D, Eskinazi, EPRI and R. Hall, EPA

3-1

3-23

3-51

3-74

3-99

3-123

"Demonstration of Low NOx Combustion Control Technologies on a 500 MWe Coal-Fired
Utility Boiler," S. Wilson, J. Sorge, L. Smith, and L. Larsen

"Reburn Technology for NOx Control on a Cyclone-Fired Boiler," R. Borio, R. Lewis, and
M. Keough

"Full Scale Retrofit of a Low NOx Axial Swirl Burner to a 660 MW Utility Boiler, and the
Effect of Coal Quality on Low NOx Burner Performance," J. King and J, Macphaii

"Update on Coal Reburning Technology for Reducing NOx in Cyclone Boilers," A. Yaglela,
G. Maringo, R. Newell, and H, Farzan

"Demonstration of Low NOx Combustion Techniques at the Coal/Gas-Fired Maas Power
Station Unit 5," J. van der Kooij. H. Kwee, A. Spaans, J. Puts, and J. Witkamp

'Three-Stage Combustion (Reburning) on a Full Scale Operating Boiler in the U.S.S.R.,"

R. LaFlesh, R. Lewis, D. Anderson, R. Hall, and V. Kotler

VOLUME 2

SESSION 4A:	COMBUSTION NOx DEVELOPMENTS I

Chair: W. Linak and D, Drehmel, EPA

"An Advanced Low-NOx Combustion System for Gas and Oil Firing," R. Lisauskas

and C. Penterson	4A-1

"NOx Reduction and Control Using an Expert System Advisor," G, Trivett	4A-13

"An R&D Evaluation of Low-NOx Oil/Gas Burners for Salem Harbor and Brayton Point

Units," R. Afonso, N. Molino, and J. Marshall	4A-31

"Development of an Ultra-Low NOx Pulverizer Coal Burner," J. Vatsky and T. Sweeney 4A-53

"Reduction of Nitrogen Oxides Emissions by Combustion Process Modification in

Natural Gas and Fuel Oil Flames: Fundamentals of Low NOx Burner Design," M. Toqan,

L, Berg, J. Befer, A. Marotta, A. Beretta, and A. Testa	4A-79

"Development of Low NOx Gas Burners," S. Yang, J. Pohl, S. Bortz, R, Yang, and W. Chang 4A-105

SESSION 4B:	LARGE SCALE SCR APPLICATIONS

Chair: E. Cichanowicz, EPRI

"Understanding the German and Japanese Coal-Fired SCR Experience," P. Lowe,

W. Ellison, and M. Perlsweig	4B-1

"Operating Experience with Tail-End-and High-Dust DENOX-Technics at the Power Plant
of Heilbronn," H. Maier and P. Dahl	4B-17

VI


-------
Paper

Page

"S03 Generation-Jeopardizing Catalyst Operation?," R. Jaerschky, A. Merz, and J. Mylonas 4B-39

"SCR Operating Experience on Coal-Fired Boilers and Recent Progress," E. Behrens,

S. Ikeda, T. Yamashita, G. Mittelbach, and M. Yanai	4B-57

'Technical Feasibility and Cost of SCR for U.S. Utility Application," C. Robie, P. Ireland,

and J. Cichanowicz	4B-79

"Application of Composite NOx SCR Catalysts in Commercial Systems," B. Speronelio,

J. Chen, M. Duriila, and R. Heck	4B-101

"SCR Catalyst Developments for the U.S. Market," T. Gouker and C. Brundrett	4B-117

"Poisoning Mechanisms in Existing SCR Catalytic Converters and Development of a New
Generation for Improvement of the Catalytic Properties," L. Balling, R. Sigiing, H. Schmelz,

E. Hums, G, Spitznagel	4B-133

SESSION 5A:	POST COMBUSTION DEVELOPMENTS 1

Chair: C. Sedman, EPA

"Status of 1 MW SCR Pilot Plant Tests at Tennessee Valley Authority and New York State
Electric & Gas," H. Flora, J. Barkley, G. Janik, B. Marker, and J. Cichanowicz	5A-1

"Pilot Plant Investigation of the Technology of Selective Catalytic Reduction of Nitrogen
Oxides," S. Tseng and C. Sedman	5A-17

"Poisoning of SCR Catalysts," J. Chen, R. Yang, and J. Cichanowicz	5A-35

"Evaluation of SCR Air Heater for NOx Control on a Full-Scale Gas- and Oil-Fired Boiler,"

J. Reese, M. Mansour, H. Mueller-Odenwald, L. Johnson, L. Radak, and D. Rundstrom 5A-51

"N20 Formation in Selective Non-Catalytic NOx Reduction Processes," L Muzio,

T. Montgomery, G. Quartucy, J. Cole, and J. Kramlich	5A-71

"Tailoring Ammonia-Based SNCR for Installation on Power Station Boilers," R. Irons,

H. Price, and R. Squires	5A-97

SESSION SB:	INDUSTRIAL/COMBUSTION TURBINES ON NOx CONTROL

Chair: S. Wilson, Southern Company Services

"Combustion Nox Controls for Combustion Turbines," H. Schreiber	5B-1

"Environmental and Economic Evaluation of Gas Turbine SCR NO„ Control," P. May,

L Campbell, and K. Johnson	5B-17

"NOx Reduction at the Argus Plant Using the NOxOUT* Process," J, Comparato, R. Buchs,
D. Arnold, and L. Bailey	5B-37

Vll


-------
Page

"Reburning Applied to Cogeneration NOx Control," C. Castaldini, C. Moyer, R. Brown,

J. Nicholson	5B-55

"Selective Non-Catalytic Reduction (SNCR) Performance on Three California Waste-to-
Energy Facilities," B. McDonald, G. Fields, and M. McDannel	5B-71

"Use of Natural Gas for NOx Control in Municipal Waste Combustion," H. Abbasl,

R. Biljetina, F. Zone, R. Usauskas, R. Dunnette, K. Nakazato, P. Duggan, and D. Linz 5B-89

VOLUME 3

SESSION 6A:	POST COMBUSTION DEVELOPMENTS II

Chair: D. Drehmel, EPA

"Performance of Urea NOx Reduction Systems on Utility Boilers," A. Abele, Y. Kwan,

M. Mansour, N. Kertamus, L. Radak, and J. Nylander	6A-1

"Widening the Urea Temperature Window," D. Teixeira, L. Muzio, T. Montgomery,

G. Quartucy, and T. Martz	6A-21

"Catalytic Fabric Filtration for Simultaneous NOx and Particulate Control," G. Weber,

D. Laudal, P. Aubourg, and M. Kalinowski	6A-43

SESSION 6B:	COMBUSTION NOx DEVELOPMENTS II

Chair: R, Hall, EPA

"Heterogeneous Decomposition of Nitrous Oxide in the Operating Temperature Range of
Circulating Fluidized Bed Combustors," T. Khan, Y.Lee, and L. Young	6B-1

"NOx Control in a Slagging Combustor for a Direct Coal-Fired Utility Gas Turbine,"

P. Loftus, R. Diehl, R. Bannister, and P. Pillsbury	6B-13

"Low NOx Coal Burner Development and Application," J. Allen	6B-31

SESSION 7A:	NEW DEVELOPMENTS I

Chair: G. Veerkamp, Pacific Gas & Electric

"Preliminary Test Results: High Energy Urea Injection DeNOx on k 215 MW Utility Boiler,"
D. Jones, S. Negrea, B. Dutton, L. Johnson, J. Sutherland, J. Tormey, and R. Smith	7A-1

"Evaluation of the ADA Continuous Ammonia Slip Monitor," M. Durham, R, Schlager,

M, Burkhardt, F. Sagan, arid G, Anderson	7A-15

"Ontario Hydro's SONOX Process for Controlling Acid Gas Emissions," R. Mangal,

M, Mozes, P, Feldman, and K. Kumar	7A-35

"Pilot Plant Test for the NOXSO Flue Gas Treatment System," L. Neal, W. Ma, and R, Bolli 7A-61

viii


-------
Paper

Page

"The Practical Application of Tunable Diode Laser Infrared Spectroscopy to the Monitoring

of Nitrous Oxide and Other Combustion Process Stream Gases," F. Briden, D. Natschke,

and R. Snoddy	7A-79

SESSION 7B:	NEW DEVELOPMENTS 11

Chair: C. Miller, EPA

"In-Fumace Low NOx Solutions for Wall Fired Boilers," R. LaFlesh, D. Hart, P. Jennings, and
M. Darroch	7B-1

"NOx Reduction on Natural Gas-Fired Boilers Using Fuel Injection Recirculation (FIR) -
Laboratory Demonstration," K. Hopkins, D, Czerniak, L. Radak, C. Youssef, and J. Nylander 7B-13

"Advanced Reburning for NOx Control in Coal Fired Boilers," S. Chen, W. Seeker, and
R.Payne	7B-33

"Large Scale Trials and Development of Fuel Staging in a 160 MW Coal Fired Boiler,"

H, Spliethoff and R. DoleZal	7B-43

"Computer Modeling of N20 Production by Combustion Systems," R. Lyon, J, Cole,

J, Kramlich, and Wm, Lanier	7B-63

SESSION 8:	OIL/GAS COMBUSTION APPLICATIONS

Chair: A. Kokkinos, EPRI

"Low NOx Levels Achieved by Improved Combustion Modification on Two 480 MW Gas-
Fired Boilers,* M, McDannel, S. Haythornthwaite, M, Escarcega, and B. Gilman	8-1

"NOx Reduction and Operational Performance of Two Full-Scale Utility Gas/Oil Burner
Retrofit installations,"' N. Bayard de Volo, L Larsen, L. Radak, R. Aichner, and A. Kokkinos 8-21

"Comparative Assessment of NOx Reduction Techniques for Gas- and Oil-Fired Utility
Boilers," G. iisonett and M. McElroy	8-43

"Analysis of Minimum Cost Control Approach to Achieve Varying Levels of NOK Emission
Reduction from the Consolidated Edison Co. of NY Power Generation Systems," D. Mormile,
J. Pirkey, N. Bayard de Volo, L. Larsen, B. Piper, and M. Hooper	8-63

"Reduced NOx, Particulate, and Opacity on the Kahe Unit 6 Low-NO,, Burner System,"

S, Kerho, D. Giovanni, J. Yee, and D. Esklnazi	8-85

"Demonstration of Advanced Low-NOx Combustion Techniques at the Gas/Oil-Fired Flevo
Power Station Unit 1J. Witkamp, J, van der Kooij, G. Koster, and J. Sijbring	8-107

APPENDIX A:	LIST OF ATTENDEES	A-1

IX


-------

-------
/

AN ADVANCED LOW-NOx COMBUSTION

SYSTEM FOR GAS AND OIL FIRING

R. A. Lisauskas
C. A. Penterson

Riley Stoker Corporation
Worcester, Massachusetts

ABSTRACT

A new 1ow-N0x combustion system for gas and oil-fired industrial and utility
boilers is discussed. The system consists of an advanced Riley low-NOx STS burner
used in conjunction with overfire air and recirculated flue gas. One of the
distinctive features of the low-NOx STS burner is the use of recirculated flue gas'
to form a separation layer between the primary and secondary flame zones,.

This advanced 1 ow-KGx combustion system has been implemented on several power
boilers in Western Europe. Combustion system modifications and emission test
results are summarized for two recent retrofit applications. Field emission data
are presented for both gas and oil-firing. N0„ emission levels of less than 50 ppm
have been demonstrated on a natural gas-fired 220 MWe utility boiler. The applica-
tion of this advanced system to U. S. gas and oil wall-fired boilers is also
discussed.

.INTRODUCTION

Environmental concern over power plant stack emissions has grown steadily over the
past decade. In spite of this concern, the 1980's saw little change in U. S. NOx
regulations. However, recent passage of new federal Clean Air amendments and
proposed new state regulations make it likely that U. S. industry will soon be
required to meet revised emission standards on both new and existing boilers.

Unlike the United States, Europe and Japan did impose new emission regulations
during the 1980's. In 1984, German legislators recommended stringent N0X emissions
standards for new and existing boilers(I). These standards .defined new emission
limits for all large combustion systems firing gas, oil and coal. As a result,
large numbers of industrial and utility boilers in Germany and other European
countries have been retrofitted with low-NOx systems. We believe this recent
European low-NOx retrofit experience is. of particular interest to the U. S. power
industry. This paper focuses on some of this experience applied to gas and oil
fired systems.

4A-1


-------
Deutsche Babcock, the parent company of Riley Stoker Corporation, has had
considerable experience in supplying combustion systems to meet the demands of
German and European air pollution codes. Low-NOx combustion systems have been .
implemented by Deutsche Babcock on a wide variety of industrial and utility
boilers. Since 1984, Deutsche Babcock has supplied low-NOx systems to over 110
liquid and gas-fired boilers. More than 520 l.ow-NOx burners have been retrofitted
to a variety of boiler configurations. In order to meet stringent emission limits
many of these retrofit applications incorporate combustion modification
techniques, such as flue gas recirculation and overfire air, in combination with
new low-NOx burners(2), NCL reductions of over 80% have been demonstrated with
these new systems. New fuel injectors have also been developed in response to the
changing quality of heavy fuel oils. This technology and experience is now
available to the U. S. power industry through Riley Stoker.

One new burner system - the Swirl Tertiary Separation (STS) burner - is
particularly well suited to U. S. wall-fired boiler retrofit applications. This
new burner system is the subject of this paper. In addition to presenting •
operating results from recent European retrofit installations, we will also discuss
the application of this new combustion system to U. S. boiler design
configurations.

DESCRIPTION OF L0W-N0x BURNER SYSTEM

New STS burner systems have been recently retrofitted on gas and oil wall-fired
boilers in both Germany and Sweden. In addition to reducing N0X, the burners were
designed to both minimize boiler pressure part changes and maintain acceptable
combustion conditions.

Figure 1 is an illustration of the STS burner equipped with swirl control. As is
typical in many European boiler designs, combustion air is controlled individually
to each burner. A spiral box, or scroll (shown in Figure 1) is used to supply the
combustion air to the burner. The scroll is divided between primary and secondary
air passages with control dampers and flow metering installed immediately upstream.
Total air flow to the burner is divided between the primary and secondary air
passages. The exact distribution of primary and secondary air can be adjusted
depending on the level of internal burner staging required for N0X control and
overall combustion performance.

The ability to independently control swirl imparted to the primary and secondary
air streams provides great flexibility in controlling flame length and shape. It
also ensures flame stability under low-NGx firing conditions. Adjustable air vanes
within the scroll are used to control the degree of swirl and subsequent fuel air
mixing. Between these two swirling air streams a separate recirculated flue gas
stream can be introduced forming a distinct separation layer between the primary
and secondary air.

The introduction of this separating layer of inert flue gas acts to delay the
combustion process and reduces N0X in the following manner:

• Peak flame temperatures, particularly on the surface of
the primary combustion zone, are reduced by a surrounding
blanket of inert flue gas.

» The rapid mixing of secondary air is prevented; thereby,

reducing the oxygen concentration in the primary combustion zone.

4A-2

t!


-------
Unlike flue gas mixed with the primary or secondary air streams, the flue gas
separation stream is unswirled and concentrated. This serves to delay secondary
air mixing until after first stage oxygen has been consumed and the flame has
cooled. The intent of the separation layer, therefore, is to control both thermal
N0X formation and N0X produced from nitrogen contained in the fuel.

Additional N0„ reduction is achieved through staged combustion, A portion of the
total combustion air can be introduced through overfire air ports above the burners
to provide external air staging. This overfire air is controlled and metered
independently of the combustion air to the burners, Low-NOx burners combined with
flue gas recirculation and overfire air offer an integrated.approach for maximizing
the reduction of NQX emissions on gas as well as oil firing.

As shown in Figure 1, oil is burned using a centrally located steam or mechanically
atomized oil gun. Natural gas is burned using spuds or canes located within the
primary core of the burner.

FIELD RESULTS

Arzberq Power Station

Low-NOx STS burners have been installed at Arzberg Power Station Unit NO. 6 in
Arzberg, West Germany. The boiler, shown in Figure 2, is a once-through Benson
boiler rated at 1.58 million lbs steam per hour and generates 220 MW of
electricity. The unit is currently equipped to fire natural gas or light NO. 2
oil. In 1988, the boiler was retrofitted with sixteen low-N0x burners, each rated
at 153 million Btu/hr heat input. Burners are arranged horizontally for opposed
firing on four levels. As stated earlier, the STS burner design was selected to
fit within existing burner openings.

NCL emission limits for this retrofit project were 50 ppm* for natural gas firing
ana 75 ppm for light oil. The retrofit combustion system was designed with the
flexibility of introducing recirculated flue gas through either the burner zone
separation annulus or having it mixed directly with the combustion air to the
burners. One tertiary air port was also installed in close proximity to each
burner but was later found to be ineffective for N0„ control. A level of overfire
air ports was added on both front and rear waterwalts above the burner array for
staged or off-stoichiometric firing. As shown schematically in Figure 3, all flows
including primary, secondary, tertiary and recirculated flue gas were independently
controlled and metered.

Prior to the retrofit, N0X emissions from natural gas firing averaged 300
ppm. Testing was conducted following the retrofit to optimize the operation and to
commission the boiler. Figure 4 illustrates the effect of mixing flue gas
recirculation into the combustion air on N0X emissions for natural gas firing. N0X
is reduced with increasing amounts of flue gas recirculation (FGR) flow. With 20%
FGR and 10% OFA flow, NQ„ emissions were reduced to 75 ppm. By increasing the
amount of recirculated ftue gas to 30%, N0X decreased to 50 ppm.

* All N0X and CO concentrations are dry and referenced to 3% Og.

4A-3

1


-------
Additional testing was then performed to evaluate the effect of Introducing FGR -
flow through the burner annul us for N0„ control. The total amount of FGR flow
remained at 30% with 10% OFA. Figure 5 illustrates the effect of introducing
increasing percentages of FGR flow through the burner annulus or separation layer.
When more than 50% of the total FGR flow was introduced through the separation
layer (the remaining amount being mixed in with the combustion air) NQX decreased
significantly. The lowest measured N0X emission approached 25 ppm when nearly all
of the FGR flow-was passing through the burner annul us. CO emissions remained less
than the 15 ppm throughout this testing and flame stability or scannabi1ity was not
a problem.

A limited amount of testing was performed on NO. 2 fuel oil. Data were collected
while operating at 15% FGR and 15% OFA flow rates. NO., emissions of 75 ppm were
achieved at full load and decreased to approximately 60 ppm at 50% boiler load. CO
emissions remained below 25 ppm for all test conditions.

Vartan Power Station

An advanced STS burner system has also been retrofitted at the Vartan Power Station
in Stockholm, Sweden. The Vartan unit, commissioned in 1976, is rated at 250 MW.
It is a once-through Benson style boiler designed for heavy oil firing. As shown
in Figure 6, the burners are mounted on a single wall in a 4 X 4 array. Each
burner is supplied individually with air and is equipped with a Deutsche Babccck
oil pressure/steam pressure atomizer. In addition to STS burners, the retrofit
combustion system includes both OFA and FGR. The existing FGR system was modified
to supply flue gas to each burner as well as the lower furnace.

The post-retrofit N0X guarantee limit for the Vartan unit is 0.27 lb/10® Btu or
approximately 210 ppm. N0X emissions measured during recent commissioning tests
are shown in Figure 7. Emission levels (at high load) for the new system are 30 to
40% lower than the guarantee value. The data spread is due to differences in
operating conditions and varying fuel oil nitrogen content. Average fuel oil
nitrogen content is 0.3%. During the recent tests, high load excess oxygen
measured 1.3-1.4% upstream of the air heater corresponding to an excess air level
of less than 7%. CO emissions were less than- 40 ppm. These results were achieved
with 10-11% OFA and 15% FGR. Approximately one third of the flue gas was
introduced through the burners. The remaining flue gas was introduced to the lower
furnace for steam temperature control.

APPLICATION TO U. S. BOILERS

The STS burner design has been adapted by Riley Stoker to U.S. wall-fired boiler
firing systems. Contrary to the European practice of individual burner air
supplies, U. S. wall-fired boilers are equipped with common windbox/multiple burner
arrangements. Because of this, the burner Inlet scroll, described in Figure 1, has
been replaced by primary and secondary air swirl vane registers surrounded by flow
control shrouds. All other burner components remain the same. As shown in Figure
8, the movable shrouds are operated by single actuators and can be automated with
boiler load. The shrouds control the primary/secondary air flow split
independently of swirl vane position. Flow measurement devices are positioned
between the burner barrels to provide a relative flow indication between the
burners.

4A-4


-------
A prototype 85 million Btu/hr STS burner designed for windbox applications (Figure
8) is currently being tested in Riley Stoker's large pilot combustion test facility
located at the Riley Research Center in Worcester, Massachusetts. This facility is
designed to simulate the near field combustion conditions of full scale
furnaces(3). Test variables include firing rate, flow biasing ratios, the amount
of flue gas recirculation and injection method, level of burner staging, swirl
setting, excess air and oil/gun positions. The test program has several
objectives: (1) to fully characterize the burner's low-NCL capability under U. S.
boiler operating conditions, and (2) to evaluate the sensitivity and trade-off of
various burner adjustments on N0X control and other combustion operating parameters
such as flame shape and particulate emissions. The prototype burner is being
tested on natural gas and NO. 6 fuel oil. The fuel oil selected for the test
program is a 1% sulfur oil with an asphaltene content of approximately 10%, Test
results will be available within the next several months.

SUMMARY

Advanced STS burners have been successfully retrofitted on several gas and oil
fired power boilers in Western Europe. These retrofits have been achieved within
existing burner openings. STS burners in combination with overfire air and flue
gas recirculation have exceeded their emission goals. N0X levels of less than 0.06
1b/10® Btu on natural gas arid less than 0.2 Ib/lO® Btu on heavy oil have been
demonstrated.

The introduction of a separate flue gas stream, or dividing layer through the
burner throat has been shown to be effective in reducing N0X on natural gas.
Additional testing is required to evaluate the effectiveness of this separation
layer during heavy oil combustion.

STS burner designs have been developed for U. S. wall-fired boiler burner/windbox
arrangements. Prototype burner tests are being carried out to ensure that European
experience is duplicated under U. S. boiler operating conditions.

REFERENCES

(1)	P.M. Dacey, "An Overview of International N0X Control Regulations,"
Proceedings; 1985 Symposium on Stationary Combustion N0„ Control, Vol. 1,
EPRI CS-4360, January 1986.

(2)	R. Oppenberg, "Primary Measures Reducing Nqx Levels on Oil- and Gas-Fired
Water Tube Boilers," Conference of the Association of German Engineers,
Duisberg, FGR, September 26, 1986,

(3)	R. Lisauskas, §t a].., "Experimental Investigation of Retrofit Low-N0x
Combustion Systems," Proceedings: 1985 Symposium on Stationary Combustion
NQX Control, Vol. 1, EPRI CS-4060, January 1986.

4A-5


-------
Primary air Secondary air



t

Flue gas

Figure 1. Low-NOx STS Burner Equipped for Gas and Oil Firing and Individual Air Supply


-------
Figure 2. Arzberg Power Plant Unit NO. 6

4 A-7


-------
a. Original System	b. Low-NOx Retrofit System

Figure 3. Low-NOx Combustion System at Arzberg Power Station


-------
Flue Gas Recirculation
Rate into Combustion Air, %

Figure 4.

N0X as a Function of FGR into
the Combustion Air -
Natural Gas Operation

Flue Gas Percentage
in the Separation Flow, %

Figure 5.

Nox as a Function of FGR into
the Annulus -
Natural Gas Operation


-------
r__i

Figure 6. Vartan Power Station

150

CM

o

f"~5

TO

50





























0,8 .	1.0	\2	1.4	1.6

Steam Flow, million Ib/hr

1.8

Figure 7. N0X Versus Boiler Load Post-Retrofit Heavy Oil Firing

4A-10


-------
Figure 8. Riley Low-NOx STS Burner (Model 90) for Windbox Burner Arrangements


-------

-------
N0X REDUCTION AND CONTROL
USING AN EXPERT SYSTEM ADVISOR

G. Michael Trlvett
Monenco Consultants Ltd.
Calgary, Alberta
T2P 3W3

4A-13


-------

-------
ABSTRACT

In a continuing effort to reduce emissions of N0X from their coal fired power
units, TransAlta Utilities has undertaken a program of combustion optimization for
low-NOx operation. In addition to testing and tuning these units, Monenco is
developing an on-line Expert System to enable operators to continuously maintain
1ow-N0x emissions.

Characterization and optimization tests for low-NOx operation were conducted at the
Sundance Generating Station on the 375 MWe tangentially fired Unit #6. The tests
produced an extensive database which will be incorporated into the Expert System.
The tests confirmed that a reduction of N0X emission of 5 to 15% could be achieved
by improved control procedures.

The Expert System advisor will incorporate real-time input from sensors such as
oxygen analysers, temperature indicators, N0X analysers, CO analysers,
burner/pulverizer status, etc. A graphical computer interface will show current
readings and a message board will provide recommended corrective actions to
minimize N0X emissions.

4 A-15

Preceding Page Blank


-------
INTRODUCTION

NOx emissions are one of the precursors of acid rain for which coal fired power
plants are a significant contributor. During tests conducted by Monenco at
TransAlta Utilities Sundance Plant Unit #6, modest N0X reductions of between 5 and
15 percent from baseline levels were achieved. Reduction of N0X emissions on a
continuous basis requires considerable diligence by the operator 1n balancing the
competing requirements of steam temperature control, ash slagging conditions and
low-NOx operations. An expert system can be developed to review the available
operating data and by use of a knowledge base make recommendations to the operator
to minimize N0X emissions while maintaining optimum unit performance.

For a moderate N0X reduction of 10%, the expected development costs for the expert
system represent a N0X avoided cost of $60/tonne based on a 35 year operating life.
By comparison a Selective Catalytic Reactor (SCR) with an 80% N0X reduction
efficiency represents a cost of over $2000/tonne. Therefore the cost effectiveness
of developing the N0X control expert system (N0XPERT) which could be applied to all
of TransAlta's coal fired units, appears very attractive.

The potential benefit of the application of an expert system to Sundance Unit #6
would be a reduction in NQX emissions from existing levels at a reasonable cost.
On an annual basis this amounts to a reduction of between 210 to 630 tonnes
NQx/year from an estimated baseline rate of 4200 tonnes NOx/year. In addition to
the total N0X reduction, a proportional decrease in NOg should result which may
reduce the contribution to the brown plume effect.

The general approach to be taken using an expert system would be to minimize N0X
formation by combustion modification or in-furnace techniques. Typical factors
that may be modified to reduce N0X emissions Include reduced excess air operation,
increased fuel fineness, increased furnace wall sootblowing frequency and fuel/air
staging in the combustion zone.

4A-16


-------
Each of the above factors can be optimized. However the Interrelationship between
factors requires some degree of compromise to achieve a reasonable low-NOx emission
with acceptable unit performance. These Interrelationships were developed as part
of the N0X testing program and the results of the tests form the majority of the
support data for the expert system development.

Expert systems have been developed as an extension of research In the field of
artificial intelligence (AI). An expert system 1s a computer program that combines
concepts, procedures and techniques derived from AI. These techniques allow the
design and development of computer systems that use knowledge and inference
techniques to analyze and solve problems in a way similar to human reasoning.

This specific application of an on-line or real-time expert system to control N0X
emissions is novel. The application of an expert system to real-time control 1s
becoming more advanced with chemical process companies applying systems to wider
and wider functions. However, the real-time aspect, particularly with future
closed loop control action, remains as leading edge technology. Expert system
"shells" or development tools are available from only a short list of potential
suppliers. Consequently the development of the N0X control expert system
represents leading edge development 1n both hardware/software and steam generator
control applications.

N0X EMISSION CHARACTERIZATION

TransAlta Utilities Sundance Plant has a total gross generating capacity of
2100 HWe. Units I and 2 are each 300 MWe and Units 3, 4, 5 and 6 are each 375 MWe.
All 6 units are Combustion Engineering design with forced circulation utilizing
tangential firing with a dual furnace arrangement. Units 3, 4, 5 and 6 are also
equipped with manual tilting overfire air nozzles. These units are supplied with a
low sulphur sub-bituminous *C" coal from the nearby Highvale Mine. Unit #6, 1n
particular, is equipped with 5 pulverizers, feeding coal to 5 elevations of
standard C-E tangentially fired coal nozzles on eight corners. The burner
arrangement is non-Low N0X except for the overfire air nozzles which are positioned
within the burner windbox setting.

The N0X tests were conducted on Unit #6 at Sundance and provided detailed
information on the relationships between N0X emissions and the control actions. A

4A-17


-------
test matrix was arranged to evaluate the response on N0X emissions by the
following;

•	Burner tilt position;

•	Fuel-air settings;

•	Auxiliary air settings;

§	Overfire air tilt position;

•	Increased coal fineness;

•	Reduced excess oxygen; and

•	Continuous operations.

The relationships between these control variables plus furnace cleanliness or
sootblowing frequency and unit load are depicted 1n Figure I. Fuel N0X, which 1s
dependent on fuel nitrogen content, Is shown with a dashed line since Its control
requires primarily fuel switching to a lower nitrogen content coal or combustion at
sub-stolchiometric conditions. Thermal N0X can be reduced by adjusting the control
parameters, essentially reducing the flame temperature,

A change in one of these control parameters may reduce NQX, however, 1t may also
result in an unwanted change, for example, to performance. The objective then is
to strike an acceptable balance between N0X reduction and overall unit performance.

Some guidelines were set regarding acceptable performance for the N0X tests. These
included the requirements not to: jeopardize safety of operations; increase the fly
ash carbon content above 0.7% by weight; and increase the average carbon monoxide
emissions above 50 ppm by volume. Also normal steam outlet temperatures from the
superheater and reheater would be maintained within their acceptable ranges.

One of the major objectives of the N0X test program was to determine the baseline
or existing emissions on a continuous basis. An initial test series was conducted
on a twenty-four hour basis over four days. All operating conditions were set
as-1s and in automatic mode where required. The data from these tests are
presented in Figure 2 and show the high, low and mean for each twenty-four hour
period. The descriptor on the x-axis represents both baseline and series 6 results
for days 1 to 4. The series G data, or optimized settings following the parametric
test matrix, are also shown for a direct before and after comparison.

The overall reduction in N0X emissions achieved was between 5 and 15 percent. In
addition the peak N0X emissions were reduced In the final test series as compared
to the baseline levels. The target settings for excess oxygen, burner tilts and

4A-18


-------
furnace cleanliness were achieved on Day 3 of the final test series. Consequently,
this test day produced the lowest mean N0X emissions and the lowest peak values.

Excess oxygen (Og) has an impact on N0X emissions as shown in Figure 2. Large
ranges of excess oxygen during the tests contributed to the large range in N0X
emissions, and the high oxygen peaks directly contributed to the high NQX peak
values. A reduction in excess oxygen during the optimized test series as seen in
Figure 2 aided in reducing the mean N0X emissions.

The effect of burner tilt position, either positive or negative degrees from
horizontal, has an effect on N0X emissions. Figure 3 shows the high and low range
for burner tilts for both the baseline data and series 6 data. The test days with
a large range in tilt position, had a corresponding large range in NQX emissions
and the highest peak levels.

As noted previously, changes to some control parameters may result in an
undesirable effect such as reduced performance. A decrease in performance is
typically indicated by two factors? the amount of unburned carbon in the fly ash
and the amount of carbon monoxide in the flue gas. Figure 4 shows the effect of
Increased fineness and excess oxygen on N0X emission and the resulting carbon 1n
ash content. The data represented by test number A was conducted essentially with
baseline coal fineness and fixed burner tilts at the horizontal position. The
remaining data points, B, C, D and E were tested with an increased coal fineness
and for tests C, D and E decreasing excess 02* As shown, an increase 1n fineness
results in a reduction in the carbon in ash content. By increasing fineness with
normal excess oxygen, an expected Increase 1n N0X resulted. However, the increased
fineness also allows the excess oxygen to be reduced, reducing N0X emissions. As
shown in Figure 4, as excess oxygen is reduced, carbon in ash increases. From this
information an optimum excess oxygen level, without degrading performance as
indicated by the carbon in ash level, was determined for these tests at a level of
2.5% 0?. Reduced excess oxygen also has the benefit of increased thermal
efficiency due to reduced heat loss, reduced fan power and draft losses.

The changes made to pulverizer/classifier settings to achieve a finer coal grind
were minor and consequently had little effect on the stack opacity. Greater
changes in fineness, however, may Impact the stack opacity and consequently opacity
data should be included for the expert system to review.

The overall effects on performance between the baseline data and the optimized
settings data as reflected in the carbon in fly ash levels and carbon monoxide

4A-19


-------
emissions are shown In Figure 5. Since the fineness 1n the baseline or as-1s
condition was considered to be out of specification, increased fineness was
expected to give an improvement 1n unburned carbon levels with lower CO emissions.
As noted the Increased fineness, as reflected in the reduced carbon content, allows
the excess oxygen to be reduced without degrading performance.

Overfire air (OFA), which is considered to be a primary N0X control technology,
allows separation of the fuel and combustion air resulting in lower flame
temperatures and lower thermal N0X formation. The angle that the overfire air
enters the furnace, relative to the horizontal position, was tested and as the
angle approaches the horizontal position, or closer relative to the flame, the N0X
emissions increase. Consequently the greatest separation angle, or 30 degrees
above horizontal, results 1n the lowest N0X levels. In addition this OFA tilt
position should remain fixed regardless of the automatic positioning of the burner
tilts.

The combination of both controllable and uncontrollable parameters 1n coal fired
steam generator operation results in a dynamic, complex system for N0X control.
Changes in burner tilt position in response to steam temperature control demands,
furnace wall slagging, burner to burner coal flow Imbalances and variations 1n
excess oxygen distribution from furnace to furnace result in fluctuations In N0X
emissions.

The amount of change or range in the control parameters noted gives a comparable
range In the resulting N0X emissions. Figures 2 and 3 show this effect for excess
oxygen and burner tilt position, respectively. In the optimized test series G the
movement of burner tilts was limited to within + or - 10 degrees with manual
adjustments and overall both the tilt range and excess oxygen ranges are smaller.
This resulted in a reduced peak for N0X emissions and an overall average reduction
between the baseline and series 6 emissions.

The N0X control expert system will review each of the noted parameters and based on
desired settings and expected relationships, should produce a reduction 1n
emissions similar to the difference between the baseline and optimized test data.

INPUT DATA FOR THE N0XPERT

The input data required for the N0X control expert system will be supplied from
three sources. The first source will be the flue gas analysis data. The second

4A-20


-------
will be panel board data with steam generator condition Information, and the third
source will be the operator/user.

The Input data for the flue gas analysis will Include the following:

t Nitrogen Oxides (N0X);

•	Carbon Monoxide (CO);

•	Excess Oxygen (02>I and

•	Stack opacity.

The oxygen and carbon monoxide monitors, two 1n each of the two ducts leaving the
economizer, should provide sufficient data for averaging and fault diagnosis.
Suspect data could be detected If outside the normal expected range and
recallbration might be suggested by the expert advisor. Also, the opacity monitors
will be used to detect high excursions possibly caused by too fine a coal grind.
The units currently are not equipped with N0X analysers and separate monitors would
be required for each of the dual furnaces.

The panel board will provide Input to determine what the steam generator current
operating conditions are corresponding to the flue gas analysis. These data will
Include the following;

•	Boiler/Turbine Load (HWe or steam flow);

•	Burner tilt position {+ or - from horizontal);

•	Superheater/Reheater outlet steam temperatures!

•	Windbox to furnace pressure differential;

t	Fuel A1r damper position (% open for all 5 levels);

§	Auxiliary Air damper position (% open for intermediate levels);

•	Bottom air damper position (% open);
t	Top air damper position (X open);

•	Overfire air damper position (OFA % open); and

•	Burner level status, mill coal feeder status (on/off).

The first two sources of data will be on-Hne or real-time at a specific recording
frequency. The third source of Information will be Input by the operator/user as
required by the expert system. This data will include information available only
at Infrequent periods such as the following:

•	overfire air tilt position (manual control, degrees from horizontal)!

•	carbon content of the fly ash5

4A-21


-------
•

t
t

In addition, the operator/user will respond to specific enquiries made by the
expert system for updates on the available data. For example time limits could be
set (time-stamping) for validity periods after which updates should be made.

The real-time data will also be recorded at a specified frequency so as to maintain
the latest Information and status available. This frequency, however, is expected
to be in the order of minutes for updates due to the large dead times as a result
of the boiler system thermal inertia. As a result, the rules for low-NOx firing
can be reviewed and acted upon over a reasonably long time period.

The input data mentioned above 1s depicted as a schematic diagram showing
communication with the expert system 1n Figure 6. The input data plus the
recommended control action to minimize N0X emissions would be echoed on-screen as
shown in Figure 7.

LOW N0X RULES

The N0X control expert system 1s currently being developed by Konenco for TransAlta
for application on Unit #6 at the Sundance Plant. The development plan includes
various stages from prototype to a fully fielded and implemented advisory system in
a one year time frame. Once fielded, a closed loop control scheme will be reviewed
for operation within a distributed control system (DCS).

The following details some of the specific concerns 1n developing this system and
outlines some generic features of an expert system.

An expert system comprises three parts:

•	facts;

•	rules: and

•	inference engine.

The facts describe aspects of the domain, for example, that the furnace excess
oxygen 1s at 3.2% by volume or that the top elevation of burners are out-of-

sootblower status and frequency;
coal fineness for all 5 mills: and
instrument calibration information.

4A-22


-------
service. Rules describe what an expert might do with the facts to reach the
objective, for example reduce the oxygen to reduce the N0X emission. In the same
way in which a human operator infers a solution to a problem based on the available
facts and previous experience with a similar problem, the inference engine combines
rules and facts in the knowledge base to reach a conclusion.

By definition rules are English-like sentences used for defining the knowledge of
an expert. Rules can be grouped as subsets which can be activated or deactivated
independently. The intent is to control which rules are used (fired) when a new
fact is introduced (asserted) into the knowledge base.

The facts that describe the domain are determined on-line from the various Input
sources such as the excess oxygen sensors. Once the facts are known by the expert
system, given constraints such as time validity, then the rules can be searched to
determine an appropriate decision or recommendation.

Rules are defined using an IF-THEN syntax that logically connects one or more
antecedent (or premise) clause with one or more consequent (or conclusion) clause.
A rule says that 1f the antecedents are true, then the consequents are also true.
The antecedents and consequents of rules refer to specific facts that describe the
state of the domain. Therefore each fact describes some particular aspect of the
domain's state. Together, the rules and facts make up the knowledge base.

The Inference engine analyzes the rules and facts for any rule antecedents that
match existing facts. The process of matching is finding a rule clause with the
same pattern of words (1n the same order) as a fact in the knowledge base. When
all of the antecedent clauses in a rule have a corresponding fact in the knowledge
base, the Inference engine can assert the consequent of the corresponding rule Into
the knowledge base as a new fact.

The Inference engine consists of a generalized computer program that knows about
reasoning strategies and various ways to combine rules and facts, but knows nothing
about any particular application. The knowledge base of rules and facts 1s
nonprocedural, while the inference engine 1s highly procedural. In other words,
rules and facts represent what the knowledge 1s, but the Inference engine
determines how that knowledge should be analyzed.

The following examples Illustrate rules which would comprise the knowledge base.
The order or sequence of the rules 1s not Important since the Inference engine will
use or fire the rule that satisfies the current facts.

4A-23


-------
•	If ppm N0X Is greater than N0X Limit

ppm CO Is less than CO Limit
Excess Og 1s greater than 2.5%

Load 1s 100% MCR
Then reduce Excess O2 to 2.5%

t If ppm N0X 1s greater than N0X Limit
ppm CO 1s greater than CO Limit
C 1n Ash 1s greater than C Limit
Any Fineness less than 65% thru 200 mesh
Any Fineness greater than 1.5% on 50 mesh
Then Recommend Increase in that Fineness to 65% thru 200 mesh and less
than 1.5% on 50 mesh

•	If ppm N0X is greater than N0X Limit

ppm CO Is less than CO Limit
Excess O2 Is 2.5%

C 1n Ash 1s less than C 1n Ash Limit
Burner tilt position less than 0 degree
Then Sequence wall sootblowers

Additional rules have been developed to set Initial conditions, such as equipment
availability (I.e. Overflre Air), optimal excess oxygen as a function of steaming
rate required and hierarchical rules to dictate priorities such as steam
temperature control or carbon monoxide limit override. The rules for the N0X
control aspect of the N0X PERT are Invoked only under steady load cases and during
transitional periods will be overridden in order to maintain safe operations.

The development of the rules, based on the example decision tree shown In Figure 8,
consists of logical If-Then clauses satisfying the various branches. Once the
complexity of the rules are fully defined, the expert system "Shell* will be
selected and a prototype N0X PERT will be assembled.

CONCLUSIONS AND RECOMMENDATIONS

There are three major techniques available for the reduction of nitrogen oxides
(N0X). They are classified as:

4A-24


-------
•	operational control,

•	combustion control, and

•	post-combustion control.

In each case there Is an associated economic cost with a corresponding Increase 1n
equipment complexity. The least expensive cost 1s to alter operational control of
existing burner equipment by means of the subject expert system. The next least
expensive Is a retrofit of the burner equipment to the low-NQx style of burners.
And finally, the most expensive control alternative is by post-combustion
techniques such as a selective catalytic reactor.

The costs associated with each technique and the level of N0X reduction offered
ranges from an estimated $60/tonne of NGX removed for an Expert System, $150/tonne
for low-NOx burners with overfire air and $2,000/tonne for an SCR. Utility
companies in Germany who have installed SCR's also recommend that N0X be reduced by
combustion first to minimize the cost of an SCR both 1n capital and annual
operating costs.

Based on the N0X test results from Sundance Unit #6, manipulating some key controls
can reduce N0X emissions by up to 15 percent from baseline emissions. By uniformly
applying rules for maintaining low-NOx operations, from operator shift to shift,
an overall reduction on both monthly and annual bases would be expected. Conse-
quently, an expert system that advises the operator using a consistent set of rules
could provide a N0X reduction up to and possibly surpassing the 15 percent from the
Sundance tests.

Based on these findings, an expert system to reduce N0X emissions from a utility
coal-fired steam generator represents a least cost approach with good probability
of success. Further, a successfully fielded expert system could be offered
commercially to other Utilities with similarly designed and equipped units.

ACKN0WLED6EMENTS

The author would like to acknowledge the support offered by TransAlta Utilities
Corporation for funding the work reported here. Also special thanks to
Richard Bane, Malcolm McDonald and Mike Blakely of TransAlta for technical guidance
through the various stages of this work.

4A-25


-------
Figure 1, NOx Emission Contributing Factors

BS-1 BS-2 BS-3 BS-4 - G-1 G-2 G-3

Baseline

Test Name

Optimized

Figure 2. NOx versus Excess Oxygen

4A-26


-------
BS-2

Baseline

BS-4 G-1
Test Name

G-2 G-3

Optimized

Figure 3. NOx versus Burner Tilt Position

Test Name

Figure 4. NOx versus Excess Oxygen and Fineness

4A-27


-------
Baseline	Test Name	Optimized

Figure 5. Carbon Monoxide and %C in Ash

I	OPERATING UNIT DATA

1-

Figure 6. Input Data for NOxPERT

4A-28


-------
>

I

ro
CD

NOx Control Expert System

Reheater T= Normal

Superheater T=Normal

OF A Tilt= + 30 Deg

Burner Tilt= -10 Deg

Coal Fineness

Most Recent Data:
Jan 17,1991

Current Load= 387 MW

Burner Status

Fuel Air Pos'n

Level A= ON

100«

Level B= ON

1009

Level C= ON

100%

Level D= ON

100%

Level E= OFF

50*

Carbon in Ash

*C= 0.6

NOX = 128 ng/j
= 176 ppm

Message Board

Date: Jan 31, 1991
Time: 09:45:33

Recommended Actions

NOX Level Higher Than 110 ng/j

*	Suggest Reduce 02 by 0.6 9S

*	Verify Carbon in Ash= 0.€>%

*	If NOX Doesn't Lower -
Then check coal fineness

Figure 7. Graphical Display for NOxPERT


-------
Figure 8.

NOx Control Decision Tree


-------
N0X PAPER FOR THE 1991 EPRI/EPA JOINT SYMPOSIUM ON A STATIONARY
COMBUSTION NOx CONTROL MARCH 25-28, 1991
THE CAPITAL HILTON
WASHINGTON D.C.

JANUARY 17, 1991 .

AN R&D EVALUATION OF LOW-NOx OIL/GAS BURNERS
FOR SALEM HARBOR AND BRAYTON POINT UNITS

Rui F. Afonso
Nino M. Molino
New England Power Service Company ,

Westboro, Massachusetts

John J. Marshal 1
Riley Stoker Corporation
Wcrchester. Massachusetts

4 A-31

I.


-------

-------
ABSTRACT

A thorough R&D program to evaluate low NQX, high efficiency oil/gas burners was
developed and conducted by New England Power Service Co. In anticipation of
retrofitting Brayton Point and Salem Harbor Units 4, the burner evaluation
project involved a series of evaluations designed to progressively identify
burner technologies most likely to meet New England Power performance
requirements. Detailed characterization of atomization quality, using an
Aerometrics Phase/Doppler particle analyzer was performed for five of the
initial eight candidate burners. Pilot scale combustion tests at Riley Research
test facility, including oil, gas and dual (oil/gas) firing conditions, were
conducted for the baseline and final candidate burners. Retrofit and cost
impact studies were conducted for the selected burners to ensure the most
efficient/economic application.. This paper focuses on combustion test results
at the Riley Research test facility.

4A-33


-------
INTRODUCTION

New England Power Service Company (NEPSCo) and Riley Stoker Corporation (Riley)
evaluated state of the art oil and gas burners for New England Power Company's
Salem Harbor, Unit 4 (Salem Harbor) and Brayton Point, Unit 4 (Brayton Point).

As originally designed by Riley for cycling/intermediate load operation, these
units typically operate with a capacity factor between 45% and 65%. Each of
these wall fired units produce 3,250,000 lb/hr of steam at S55°F superheat and
reheat, generating about 440 MWe net. Figure 1 shows a cross sectional view of
the units. Twenty four rear wall mounted burners fire into a pressurized
furnace. Figure 2 illustrates the general windbox arrangement. Two half depth,
vertical platen walls divide the furnace into three cells. Horizontal,
drainable superheater and reheater surfaces allow for timely start up and
shutdown in cycling duty.

The existing Rodenhuis and Verloop TTL7 burners are nominally rated at 200
million BTU/hr. Designed before 1970, they represented burner technology prior
to the 1971 New Source Performance Standards (pre-NSPS). Currently, N0X control
regulations do not apply at Salem Harbor. Brayton Point, designed in 1969,
adheres to a 0.3 lb/million BTU (-234ppm) N0X emission limit imposed by the
Massachusetts Department of Environmental Protection. Bias firing and Flue Gas
Recirculation (FGR) enable Brayton Point to achieve this level of control,
Particulate emission limits of 0.12 and 0.05 lb/million BTU apply to Salem
Harbor and Brayton Point, respectively, and are achieved with the installed
electrostatic precipitators. Salem Harbor operates near 10% excess air with a
boiler efficiency of about 87%. Brayton Point uses up to 15% FGR with 10%
excess air.

Performance goals for the new burners include 0.3 Ib/MBTU and 0.2 Ib/MBTU of
N0X, respectively, for oil and gas firing at 5% excess air, while maintaining
carbon in the flyash (oil firing) to less than 20%. This will reduce total
particulate emissions and improve precipitator performance. Typically, both
units average 30-50% carbon in the flyash.

A Rodenhuis & Verloop TTL5 burner served as the baseline burner for testing in
the Combustion Burner Test Facility (CBTF) at the Riley Stoker Research Center.
This 80 million BTU/hr burner is similar to the 200 million BTU/hr burners
installed at both Salem Harbor and Brayton Point. Field data from Salem Harbor

4A-34


-------
was compared to the baseline burner test results In the CBTF. This analysis
provides guidelines for predicting the potential field performance for the
candidate burners tested in the CBTF.

Comparative data illustrates improvements achieved with the latest design.
Additionally, results" are provided for natural gas firing and simultaneous
firing of both natural gas and residual fuel oil in various ratios.

THE R&D PROJECT

Previous papers (1) describe the background and technical approach to the
selection of the candidate burners for Brayton Point and Salem Harbor.

In summary, the program consisted of five primary tasks:

•	Evaluation of vendor proposals

•	Evaluation of atomization performance

•	Baseline burner combustion tests

•	Candidate burners combustion tests

•	Engineering and economic evaluation of burner retrofits

The evaluation process started with the submittal of proposals by eight
participating manufacturers, and evolved through detailed atomization
performance testing (five candidates), to pilot scale combustion tests for the
baseline and two final candidates.

Atomization testing was performed using surrogate fluids, air/water and
air/water-glycol to simulate steam/oil at various viscosities and operating
conditions. Spray droplet size distributions (spatial and temporal),
mass/volume fluxes and atomization efficiency were compared for each burner.
Review of these results combined with the initial proposal evaluation, led to
the selection of the two burners for combustion testing.

This paper focuses on a comparison of the results of the baseline burner
(RSlV TTL5) and a new generation Rodenhuis & Verloop TTL 22.5 burner capable of
residual fuel oil and natural gas firing.

4A-35


-------
PERFORMANCE CRITERIA

NEPSCo established the following full load performance goals and operational
characteristics for candidate burners:

Full Load Performance

•	NO), emissions less than 0.3 lb/10® Btu for 0.5% wt. nitrogen fuel;
0.Z Ib/Mbtu for natural gas

•	Particulate emissions, less than 0.05 lb/10® Btu after the
precipitator (precipitator efficiency = 60%)

•	CO emissions, less than 100 ppm

•	Flue gas oxygen content, less than 1% by volume (5% excess air)
Operational Characteristics

•	Turndown, 6:1 to 10:1

•	Oual fluid atomizer using air or steam

•	Fuel oil pressure, less than 300 psig at the gun inlet

•	Fuel oil viscosity at the burner 100-150 SSU

•	Fuel flow: 21 gpm #6 oil, 220 KCFH natural gas

In addition to the performance criteria, the following concerns were considered
for the evaluation:

•	Operability and reliability

•	Mechanical design

•	Materials of construction

•	Cost

•	Related experience of the various burner technologies

DESCRIPTION OF TEST BURNERS

Baseline Test Burner - TTL5: Over twenty years ago the Rodenhuis & Verloop B.V.
company of Holland developed the TTL5 burner. Compared to the typical utility
or industrial burners using pressure or steam assisted atomizers the TTL5 burner
was unique in that low pressure primary air, at a nominal 35"WS, atomizes tne

4A-36


-------
oil. No registers are available to control or shut off the secondary combustion
air flow to individual burners.

The TTL5, rated at 80 million BTU/hr, simulates the combustion aerodynamics of
the geometrically similar 200 million BTU/hr TTL7 installed at Salem Harbor and
Brayton Point. Figure 3 illustrates the TTL5 burner arrangement installed in
the CBTF test rig. Further details of the atomizing bullet assembly are shown
in Figure 4. The spiral atomizer head and the two stages (one adjustable) of
swirling primary air are indicated.

Primary air flow is about 1% of the total combustion air flow at full load.
During turndown the primary air flow stays essentially constant. In the case of
the Brayton Point and Salem Harbor installations primary air is taken from the
upstream side of the air heater. Booster fans distribute air to the burners.
The first stage of primary air supports and boosts the initial swirl of oil
issuing from the spiral atomizer to produce a conical shaped thin film sheet of
oil. The second stage primary air rotates in the opposite direction. The
interaction between these counter-rotating flow fields reduces the oil film to
fine droplets needed for combustion. An adjusting rod allows the second stage
primary air swirl to be changed. This is done by varying the second stage air
proportions through radial and tangential slots. Both extremes of 100% radial
to 100% tangential can be achieved. At full load oil pressures operate in a
range of 50 to 80 psig depending on oil type and viscosity.

The flame stabilizer is a conical annular diffuser ring mounted on the primary
air tube. Just upstream of the flame stabilizer an annular ring with a venturi
shape at the exit, helps to distribute the air uniformly to the diffuser. The
entire assembly penetrates just less than halfway into the venturi shaped
refractory throat.

Advanced Burner - TTL/HG22.5: Figure 5 depicts the new generation Rodenhuis &
Verloop TTL/MG22.5 selected for installation in the CBTF test rig. Rated at 80
million BTU/hr, the TTL/MG22.5 is similar in design and combustion aerodynamics
to the 200 million BTU/hr TTL/MG50 proposed for field retrofit.

The advanced burner incorporates many design changes over the baseline TTL5. A
sliding shroud register allows secondary combustion air flow control and

4A-37


-------
shutoff. No swirl of secondary air occurs at the register of this radial air
entry, axial flow burner.

The atomizer bullet assembly retains the operating parameters and design
philosophy of the TTL5 burner but new hardware designs enhance and improve
atomization. Two stages of primary air surround the spiral oil atomizer. The-
assembly shown in Figure 5 illustrates the design without an adjustable slide
for changing the second stage swirl. The. assembly is available with or without
this capability. In this program, testing included a temporary slide
arrangement to evaluate the impact of the additional hardware.

A multivane diffuser replaces the conical bluffbody flame stabilizer of the
TTL5. This changes the combustion aerodynamic patterns for mixing the primary
air/oil flows with secondary combustion air. The vaned diffuser provides a flow
that is counter in rotation to the second stage swirl of primary air. The
interaction of these flows increases the degree of mixing between streams.
Improved flame stability and turndown characteristics provide operational
advantages at low excess air operation.

The refractory throat area increases about 15% over the TTL5. Shape of the
throat changes to smaller entrance and exit angles with a longer axial section
(Figure 5). The combination of increased throat area and a 5% excess air
operating level reduces the secondary air velocity nearly 20%.

Natural gas enters the burner through an annul us created by a concentric tube
surrounding the primary air tube. A set of nozzles distributes the natural gas
at the inlet to the flame stabilizer. Pilot holes discharge a small amount of
fuel at the hub of the stabilizer directly to the primary flame zone.

TEST FUEL CHARACTERISTICS

To maintain a relatively consistent quality of residual fuel oil a storage tank
at Brayton Point was set aside to hold fuel throughout the test program. The
tank is rigid roof with heating but no mixing equipment, Though capable of
storing over two hundred thousand barrels, less than forty thousand barrels
(1,680,000 gallons) were in the tank at the start of the program. Over the
course of 18 months in the test program about 120,000 gallons were fired. Truck

4A-38


-------
shipments of 6000 to 8000 gallons were made from Brayton Point to the 10,000
gallon heated underground storage tank at Riley.

Each truck delivery was sampled as part of a quality plan. This assured that any
significant changes in fuel quality would not complicate the evaluation of test
results. Some minor variability was expected due to the design of the storage
tank, lack of mixing equipment, interrupted heating, the long peri.od of testing
and the relatively small size of a delivery. Table 1 lists the typical fuel
analysis obtained from truck deliveries. This fuel is consistent with the
residual fuel oil fired at Salem Harbor and Brayton Point.

RESIDUAL Oil COMBUSTION TESTS
Baseline Burner And Performance Targets

The Combustion Burner Test Facility (CBTF) at Riley accommodated all burner
testing described here. Design details of this nominally rated 100 million
BTU/hr coal, oil and gas fired facility appear in other publications (1,2). The
results of the baseline Rodenhuis & Verloop TTL5 burner testing in January and
February of 1989 have already been presented (1). Additional baseline burner
testing in April 1990 expanded and verified the earlier results. One of the
objectives of the baseline test was to establish correlations to field data from
Salem Harbor (3,4,5).

Due to various burner adjustments, two stable flame characteristics were found
for the TTL5. A long narrow flame and a shorter wide flaring flame were
obtained by changing the location of the spiral oil atomizer along with
adjustments to the primary first and second stage air ratio. The long narrow
flame is not acceptable as representative of field operations due to potential
flame impingement. To establish the baseline correlation only the TTL5 wide
flame test results were used.

Currently, at full load oil firing and less than 5% excess air, Brayton Point
N0X levels must not exceed 0.3 lb/million BTU (-234 ppm N0X at 3% Oxygen). This
is achieved in practice by utilizing flue gas recirculation (FGR) through the
TTL7 burners along with bias firing. Salem Harbor is not subject to N0X
regulatory limits.

4A-39


-------
The performance goal for carbon content in flyash is 20% or less. Full scale
field testing indicates a direct relationship between flyash carbon and
asphaltene content of the residual fuel oil (5). Currently at Salem Harbor, an
oil/water emulsification system helps control flyash carbon content.

Figures 6 and 7 respectively show the correlations for N0X and carbon content
reductions desired in the CBTF and field results. Burner area heat release
(BAHR) correlates the N0X data for the CBTF and the field. This technique has
been successfully applied in previous investigations for the Environmental
Protection Agency (6). Asphaltene levels in the residual fuel oil correlate the
carbon content results in the CBTF and the field.

Figure 6 illustrates the analysis used to develop potential NQX targets for new
burners applied in Salem Harbor or Brayton Point. The Salem Harbor field data
is scattered at about 350 ppm N0X at a BAHR of 350,000 to 363,000 BTU/hr-sqft.
Results from baseline burner tests in the CBTF range from 210 to 230 ppm NQX at
a nearly constant BAHR of 75,000 BTU/hr-sqft. Line LI connects the CBTF and
field data for similar burner operating and flame conditions and establishes the
methodology for projecting N0X emissions.

To determine the target NOx levels in the CBTF line L2 is drawn parallel to line
11 from the field requirement of 0.3 lb/million BTU (234 ppm N0X). The values
of line L2 represents a NOx reduction of about 35%. Accordingly, the NQX target
in the CBTF is 104 ppm N0X. Ideally the need for FGR, bias firing and
emulsification will be reduced or eliminated with the new burner, while meeting
the N0X emission goals.

Figure 7 shows the analysis to develop carbon loss reduction targets for new
burners applied in the field and firing neat oil. During the field test with
asphaltene levels in the oil at 6.1% and 14.7% the carbon content in the flyash
was 21.7% and 34% respectively (5). Considering carbon, ash and sulfate
contents in the flyash and the amount of carbon and ash in the as-fired fuel
oil, the carbon loss was 0.037% and 0.084% respectively. For the same amount of
ash and sulfates in the flyash, the flyash carbon goal of 20%, translates to
carbon loss values of 0.033% for the 6.1% asphaltene oil and 0,04% for the 14.1%
asphaltene oil. Figure 7 shows the actual field results and the equivalent
points needed to achieve 20% carbon content in the flyash.

4A-40


-------
This analysis indicates reduction requirements in carbon Toss from 11% to 52%
for the range of oils tested in the field. The following description
exemplifies this procedure: during TTL5 burner tests the asphaltene content of
the as-fired fuel was 11.2%; from Figure 7 the target reduction in carbon loss
at this level of asphaltenes is about 40%; at full load conditions in the CBTF
the TTL5 burner produced a carbon loss value of 1.84%; therefore, 40% reduction
from this level equates to 1.10% carbon loss. The actual percent reduction for
new burners will depend on the as fired content of asphaltenes at the time of
the test.

TTL5 And TTL/HG 22.5 Performance

A number of independent parameters were varied during testing: burner hardware
settings, fuel oil viscosity, excess air, overfire air and load turndown. The
burner hardware adjustments included variations in the relative location and
settings of burner components. Items such as atomizer and diffuser position and
primary air staging ratios were varied from their design points.

In comparing the NOx performance of the TTL5 with the TTI/MG22.5, excess air and
overfire air provided greater impact than viscosity. Viscosity varied from 21.0
to 31.5 cStokes (100 to 150 SSU). The TTL5 NOx remained relatively constant at
220 ppm over this range. The TTL/MG22.5 N0X decreased about 14% to 174 ppm
while the viscosity increased but the appearance of the flame at the root and
the end did not support operating at the higher viscosity levels.

The ability of the TTL/MG22.5 to operate efficiently at low excess air provides
its greatest advantage over the TTL5. At 10% excess "air (2.0% 02) the TTL5
produces about 220 ppm N0X. The TTL/HG22.5 produces slightly higher N0X (-230
ppm) at 10% excess air. For the 5% excess air design operating condition, the
TTL/MG22.5 produces 163 ppm N0X. This represents nearly a 25% reduction in the
level of N0X at design conditions. Figure 8 shows NOx versus exit oxygen.
Although the TTL5 burner operated stably at lower oxygen, the flame increasingly
contained dark areas and smoke. The TTL/MG22.5 produced good flame
characteristics down to 0.2% oxygen and perhaps lower but there was no practical
reason to pursue testing below this point. The level of CO was constant over
the range of 1.0% to 2.2% oxygen for each burner, with the TTL5 producing about
48 ppm while the TTL/MG22.5 yielded a lower 38 ppm. Down to 0.2% the TTL/MG22.5

4A-41


-------
CO level increased to about 43 ppm, still lower than the TTL5 at 1.0% oxygen.
CO is expected to be lower in the field due to multi burner interactions and
longer residence time.

After establishing the NOx/exit oxygen relation, most tests for the TTL5 were
conducted at about 2.2% oxygen which is typical of field operating conditions.
The TTL/MG22.5 tests stayed near 0.8% oxygen. Figure 9 shows the impact of OFA
on N0X emissions. The change in excess air accounts for most of the difference
between the two burners. At 25% OFA, the TTL5 N0X levels are lowered by 43%
while the TTL/MG22.5 levels dropped 37%. The TTL/MG22.5 achieves the target
level of NOx, 104 ppm, at 20% OFA. CO levels during OFA testing showed no
significant changes from the unstaged condition,

Figure 10 summarizes the N0X data in a comparison to the targets developed using
Figure 6. From this data the application of the new generation TTL/MG50 burner
is projected to decrease NOx about 17%, (see, 13 in Figure 10), Staging with
10% OFA would reduce N0X by 26% (L4). Using 20% OFA, (Figure 9) would achieve
the target N0X emission (L2).

The carbon loss reductions obtained with the TTL/MG22.5 reached the goal
established from the TTL5 burner and the guideline given in Figure 7. Three
sets of firing conditions were evaluated for the TTL/MG22.5 burner; low excess
air, low excess air with OFA and low excess air with a combination of OFA and
flue gas recirculation (FGR). The use of FGR had little impact on reducing N0X
during residual fuel oil firing. This is consistent with the findings of others
(8). As illustrated in Figure 11, the TTL/MG22.5 indicates a 47% to 57% carbon
loss reduction compared to the TTL5.

Natural Gas and Natural Gas/Residual Oil Dual-Firing

Brayton Point plans to add natural gas capability to in the future. Since the
unit has never fired natural gas, field data to develop a N0X correlation with
the CBTF are unavailable. The combustion testing of the TTL/MG22.5 with natural
gas in the CBTF provided information on the response of N0X emissions to various
hardware and operating parameter adjustments.

At design operating conditions of 1% Og, N0X emissions are about 60 ppm.
Variations in oxygen between 0.8% and 1.1% resulted in negligible changes in N0X

4A-42


-------
emission, CO did not vary for this range of excess air testing (12 ppm).
Staging from 0% to 15% reduced NQX from 61 ppm to 54 ppm indicating a very weak
response to staging. Again, the CO levels remained near 12 ppm."

Simultaneous firing of natural gas and residual oil in the TTL/MG22.5 burner was
tested in the CBTF. In the field Brayton Point wants the option to fire both
fuels simultaneously, either through the same burner, dual-firing, or through
separate burners, co-firing. As a single burner installation, the CBTF test rig
allows the evaluation of only dual-firing conditions. The state N0X emission
regulations for new sources are specific for gas or oil firing, but not stated
for simultaneous firing. For these conditions, a linear extrapolation may be
used as a first cut guideline by state environmental agencies.

Dual-firing tests were setup to evaluate N0X emissions at varying ratios of gas
and oil heat input. This was done with and without OFA. The results are
summarized in Figure 12. As natural gas is introduced into the burner a
significant increase in N0X emissions occurs, line 12. This trend stops near a
ratio of 80% oil/ 20% gas. Then the NQX decreases linearly to the level for 100%
gas. A similar situation occurs with 15% OFA but the effect is diminished at
these lower N0X levels. In this test program similar trends were found in a
more conventional steam atomized low-N0x burner except the initial rise in N0X
did not diminish with the use of 15% OFA.

SUMMARY

NEPSCo has supported a program that will assist in meeting commitments to
retrofit units at Salem Harbor and Brayton Point. The goals of the program aim
for improving unit efficiency, operation and fuel flexibility and simultaneously
reducing N0X emissions, carbon loss and particulate emissions. The program
evaluated several high efficiency, low-NOx burner technologies. Communications
with other users, burner proposal evaluations, atomizer tests, large scale
combustion tests, retrofit impact analysis and economic evaluations have all
played a roll in advancing the program. The selected technology, Rodenhuis and
Verloop TTL/MG50, is scheduled for installation at Brayton Point and Salem
Harbor in 1991, Full scale test results should be available in 1992.

4A-43


-------
REFERENCES

1.)	R. F. Afonso, N. H. Molino, D, C. Itse, 0. J. Marshall, J. F. Hurley and
S. Lindeman, Evaluation of Low-NO^, High Efficiency Oil and Gas Burners
for Retrofit to Utility Boilers. Presented at: American Flame Research
Committee 1989 International Symposium on Combustion in Industrial
Furnaces and Boilers, Short Hills, New Jersey September 25-27,1989 and the
EPRI Fuel Oil Utilization Workshop, Clearwater Beach, Florida November 1-
2, 1989, EPRI 6S-6919, July 1990.

2.)	R. A. Lisauskas, Experimental Investigation of Retrofit Low-No„ Combustion
Systems. Proceedings of the 1985 Symposium on Stationary Combustion NO^
Control, Vol. 1, EPRI CS-4360, January 1986.

3.)	D. V. Giovanni and T. Sonnichsen, Gaseous and Particulate Emissions Tests
at Salem Harbor Unit 4. Report submitted to New England Power Service
Company by KVB, Inc., NY, December 1972.

4.)	G. Dusatko, Salem Harbor Unit N0X Reduction Program. Report submitted to
New England Power Service Company by KVB, Inc., NY, September 1973 and
supplemental report in July 1974.

5.)	N. M. Molino, G. Dusatko, Field Test of a Processed and Emulsified
Residual Oil at Salem Harbor Station - Unit No. 4. Published by the
American Society of Mechanical Engineers, 345, East 47th ST., New York, NY
Ref. No. 87-JPGC-FACT-B.

6.)	C. C. Masser, R. A. Lisauskas, D. C. Itse. Extrapolation of Burner
Performance From Single Burner Tests to Field Operations. Presented at:
1985 Joint EPRI/EPA Symposium on Stationary Combustion N0y Control,
Boston, MA, May 6-9, 1985.

7.)	P. N. Garay. Add low-nitrogen, emulsified oils to list of emerging N0X
controls. Power Magazine, October 1990.

8.)	K. M. Bentley, S. F. Jel inek, N0X control technology for boilers fired
with natural gas or oil. TAPPI Journal, April 1989.

4A-44


-------
¦ —1

Figure 1. Brayton Point Unit 4

4A-45


-------
x~~\

Figure 3. Rodenhuis and Verloop TTL5 Burner Arranqement

4A-46


-------
Burner assembly

M Combustion air
2/ Atomizing air
3/ Oil supply
4/ Gas supply
5/ Oil burner

6/ Igniter

7/ Flame detector
8/ First stage atomizing air
9/ Second stage atomizrg air
10/ Cylindrical air damper with drive

Oil and gas burner

Figure 5. Rodenhuis and Verloop TTL/MG Burner
and Atomizing Bullet Assembly

4A-47


-------
i u tl ii|in

0 EO- 100 150 200 250 300' 350 4C© 450 500

Burner Area Heat Release, 103 Btu/hr-ft-

Figure 6. Correlation of CBTF NOx
Emission to Full Scale

Q.100
0.095
0,090 %
0.085

o.oao \

0.075
0.070
^ 0.065
- 0.060
YX 0-55
O 0-°5°
—i 0.045
Z 0.040
0.035
0.030
0.025
0.020
0.015
0.010
0.005 -I
0.000

o

m

<
¦o

=































1































\































\































:































"































\































































\































































z































I































\





















——J—--

























1







:





















|





























|





























|







=





















|







:





















1







:





















1











\ ii %¦, * i i • i ; » i i

t . u





i I I i M i I I | I 1 k i ( > 4 I « | « i » i

Mil



0.0 1.0 2.0 3.0 4.0 5.0 6.0 7.0 8.0 9.0 10.0 11.0 12.0 13.0 14.0 15.0

•ASPHALTENES, %

¦ -ACTUAL	X - GOAL

Figure 7. Salem Harbor Unit Performance

{Asphaltenes Versus Carbon toss)

4A-48


-------
EXIT OXYGEN, I

Figure 8. Oil Firing - Effect of Exity Oxygen on NOx

or a, %

Figure 9. Oil Firing - Effect of QFA on NOx

4A-49


-------
400
350
300
250
200
150
100
50

Ll Baseline
12 Limit 0.3 1b/MMBTU
L3 TTIMG/22.5 - OSOFA
L4 TTLMG/22

C J VI Ml U it) I li It il 11 j U: '.mi I|1 II 111 II1111) 11 ll) I [ 111) Itl II] Ul I) Ik kt|l II i) 11 i l| 1IIII111 l|l IL111 III

0' 50 100 150 200 250' 300 350 400 450 500
BURNER AREA HEAT RELEASE, 1C3 BTU/HR FT2

go"
oo
o

z
o

CD

oc
•<

o

Figure 10. Projected NOx Emissions to Full Scale

CDlF-PAUriCULATEliAMii INGTEST GONHITIONS ANfl CAnilON LQD3

2.4

2.0 -

16

1.2

0.8

0.4

0.0

FACliiTY

ounwcn

DESIGN

TEST
NO.

CHL
ASPl IALT

OIL
VISGOSTTY
SOU

OfT
02

%%

OFA
Vs

Fen

CARBON
LOSS
to





MiVshi:'



iS.ir.if's"!



>:.Xou





CflTF

TfJ 5

B2

11 2

ICQ

2,2

0

0

1,01

CDTF

cutk

TARGET
TTUMG52-B-STl>. BUILT

M

11.2
tt.O

-

00

-

-

1.69

tos

0.7

c

0

9.13

CB7F

TTUMG22.5-STD. GUILT

9D

DO

103

0.0

u

0

Q 90

CQTF

TTUMGS2.5-STD CULLT

161

S.O

103

B.7

9

9

0.79

TTTT

TTT

TTTT

TTTT

TTTT

TTTT

98

94

lor

TTTT

TTTT

TTT!

TTTT

0 t 2 3 4 5 6 7 8 9 10 H 12 13 H 15

ASPHALTENES, %

O -TTL5;4-G0AL; +-R&V TTL/MG

Figure 11. Performance Comparison -

Asphaltenes vs. Carbon Loss

4A-50


-------
OIL/GAS LOW), %

Figure 12. Effect of Oil to Gas Load Ratio
@ 80MMBTU/hr with 01 and 15% OFA

Table 1

TYPICAL DELIVERED FUEL ANALYSIS

API GRAVITY, DEG. API	14.2

POUR POINT, DEG. F	30

FLASH POINT, DEG. F	240

BS&W, wt% .	0

CONRADSON CARBON, wt%	16.1

VISCOSITY @ 150 DEG. F, cStokes	168

I 180 DEG. F, cStokes	77

§210 DEG. F, cStokes	41

HIGH HEAT VALUE, BTU/lb	18,386

ASPHALTENES, wt%	9.5

VANADIUM, ppm	341

SODIUM, ppm	45.6

NICKEL, ppm	54

CALCIUM, ppm	8.7

IRON, ppm	7.7
ULTIMATE ANALYSIS, wt%

CARBON	86.00

HYDROGEN	11.00

SULFUR	2.00

NITROGEN	0.42

OXYGEN	0.49

ASH	0.09

4A-51


-------

-------
DEVELOPMENT OF AN ULTRA-LOW NOx PULVERIZED COAL BURNER

Joel Valsky, Director
Timothy W. Sweeney, Supervisor
Combustion and Environmental Systems
Foster Wheeler Energy Corporation
Ferryville Corporate Park
Clinton, New Jersey 08809-4000

I

1

4 A-53


-------

-------
ABSTRACT

Foster Wheeler has been utilizing the Controlled Flow/Split-Flame Low NO* burner for both new and
retrofit applications since 1979. This Internally Staged burner attains 50-60% NO* reduction, as
compared to pre-NSPS turbulent burners, without utilizing any staging ports. A new burner has been
developed which combines the internal staging concept with another patented Foster Wheeler
technology: fuel staging. This new design, which is defined as Internal Fuel Staging™, is consistently
achieving NOx levels of 0,25 lb/106 Btu with bituminous coals containing 22.-35% volatile matter and
fuel nitrogen of 1.8%. This represents at least 75% reduction from turbulent burner levels, This paper
discusses the results of comparative tests between the standard CF/SF Low NOx burner and the new
IFS design.

4A-55


-------
INTRODUCTION

An advanced low NOx burner has been developed which achieves NOx emission levels as low as 0,25

lb/10%tu; equivalent to reductions of up to 75% from turbulent burner levels. The Internal Fuel
TIM

Staged design is based upon concepts developed and patented by Foster Wheeler in the late 1970's.
Development of the IFS design was not undertaken at that time in favor of proceeding with the
Controlled Flow/Split-flame low NOx design. The latter is Foster Wheeler's standard low NOx burner
which as been in utility and industrial service, in both new unit and retrofit applications, since 1979.

The CF/SF burner was developed to achieve at least 50% NO* reduction in retrofit applications and
meet the 1979 New Source Performance Standard, of 0.5 lb/106 Btu for sub-bituminous and 0.6 lb/106
Btu for bituminous coals, without the simultaneous use of supplementary NO* controls such as overfire
air.

The IFS design, having a NO* control capability significantly greater than that of the CF/SF, was
delayed in development simply because there was no commercial market for a system with such low
NOx capabilities. By 1989 it became apparent that new source requirements were tending toward 0.3
lb/10® Btu and retrofit requirements toward 0.5 or lower.

Although the CF/SF design operates below the 0.5 level, in both new units and retrofits without
overfire air, and below 0,3 with overfire air, Foster Wheeler decided to advance the technological
capability by developing the IFS, Commenced in May 1990, the IFS development was completed in
September 1990 and has been offered commercially, with considerable success, since that time.

Development testing was done on Foster Wheeler's 80 million Btu/hr Combustion and Environment
Test Facility. Since substantial testing of the CF/SF burner had been performed on the CETF and
correlated with utility boiler data, and there are over 5,000 MW of retrofitted and over 7,000 MW of
new units with this design, all development testing of the IFS was comparative with CF/SF data.

Typical NOx emissions from the CF/SF burner, on theCETF, are about 0.4 lb/106 Btu, Typical IFS
emissions are 0.25-028 lb/106 Btu, at least one-third lower; both without overfire air.

It should be noted that the EFS design differs from the CF/SF in only a single component; the fuel injector's
nozzle. Consequently, it represents only a minor change in overall design since all other burner
components, and the operating method, are identical to the CF/SF design. The IFS coal nozzle is, therefore,
easily retrofittable to CF/SF burners currently in operation, thereby converting it to the IFS.

CONTROLLED FLOW/SPLIT-FLAME LOW NOx BURNER

TK/f

The CF/SF design shown in Fig. 1 is based upon the principle of Internal Staging of the flame. This
principle was developed and defined by Foster Wheeler in the 1970's.

4A-56


-------
INTERNAL STAGING™ is defined as:

A low NOx burner design which two-stages the secondary air flow and stages the primary air/fuel
flows within the burner's throat while maintaining classical turbulent burner flame patterns and low
pressure drop: < 4,0" H2O,

The Controlled Flow/Split-flame name is derived from the operating functions of the burner:

•	Controlled Flow for the dual register design which provides for the control of inner
and outer swirl zones along with a sleeve damper which allows independent control
of the quantity of secondary air flow to each burner.

« Split-Flame for the coal injection nozzle which develops a split-flame pattern for
obtaining low NOx emissions.

Key criteria within the overall design philosophy are summarized as follows:

Mechanical reliability to be such that after long term operation movable components
would still operate.

Combustion air flow and swirl to each burner to be independently controllable.

Adjustable primary air/coal velocity to ensure optimum relation between primary
and secondary air streams.

No increase in primary or secondary air pressure drop so that existing PA and FD
fans can be used.

Burner capacity to cover the complete range of industrial and utility use:
approximately 30 to 300 million Btu/hr.

•	Plug-in retrofitability, i.e., no pressure part changes, no burner piping
rearrangement and no major windbox modifications when installed on most
existing wall-fired boilers.

The CF/SF low NOx burner's components and their functions are described below:

•	Perforated Plate with Sleeve Damper: used to control secondary air flow on a per
burner basis. By measuring the pressure drop across the perforated plate an index of
air flow is obtained. The air distribution, vertically and horizontally, within the
windbox is thus optimized by adjusting the sleeve dampers to obtain equal burner
stoichiometrics. This is a one time optimization after which the "open" position is
fixed. The sleeve damper has "closed", "ignite" and "open" positions and is used,
instead of the main radial vane register, to shut off the air flow when the burner is
out of service. It is controlled by an electrically operated linear drive, but is not
modulated with load.

Dual Series Registers: provide improved flame shape control by two-staging the
secondary air. A key mechanical reliability feature of this configuration is that the
blades and drive mechanisms, set back from the furnace wall, are well "shaded"
from direct flame radiation. Consequently, the registers operate at windbox

4A-57


-------
temperature and do not overheat, warp or bind. Additionally, once the flame is
optimized for proper shape, the registers are fixed. They remain in their optimum
position and are not modulated with load or closed when the burner is taken out of
service since the sleeve damper performs the latter function, The burner essentially
becomes a fixed register type with the option for adjusting the register if a major
fuel change occurs; the drive mechanisms being manual-.

Split-Flame Nozzle: segregates coal into four concentrated steams. The result is that
the volatiles in the coal are driven out and are burned under more reducing
conditions than otherwise would occur without the split flame nozzles. Combustion
under these conditions converts the nitrogen species contained in the volatiles to N*2.
substantially reducing NOx formation.

~ Adjustable Coal Nozzle: allows primary air/coal velocity to be optimized without
changing primary air flow. The proper relationship between primary and secondary
air is important for both good combustion and flame shaping. Once optimized no
further adjustment is required.

Succinctly, only the sleeve damper, used to shut off the secondary air flow, is moved when the burners
are taken in or out of service. Thus, after optimization, the burners become fixed register types.

Mechanical reliability of the design concepts, materials and operational methodology has been fully
confirmed by commercial experience.

ADVANCED OVERFIRE AIR SYSTEM

NOx emissions from the burning of pulverized coal have three (2) sources: Thermal NOx generated
from thermal fixation of atmospheric nitrogen (Na> at high flame temperatures, conversion of bound
nitrogen in the coal's volatile fraction and conversion of bound nitrogen in the coal's char fraction. The
latter being the most difficult part of the emission to control.

Foster Wheeler's NOx control philosophy has been based upon using the CF/SF low NOx burner which
achieves a high degree of thermal and volatile-fraction fuel NOx reduction; with char-fraction fuel NOx
reduction to a lesser degree. This low NOx burner's effectiveness is uniformly consistent in reducing
NOx emissions by 50-60% from uncontrolled levels. When lower levels are required an advanced
overfire air system (AOFA) can be incorporated to increase NOx control to the 70-80% range. These
results are valid for both new steam generators and retrofittable existing units.

Figure 2 schematically illustrates a typical AOFA system. It is characterized by a set of overfire air ports
placed weU above the top burner level to provide relatively long residence time between the top burner
level and the overfire air port level. One port is located above each burner column with an additional
port near each sidewall. This is illustrated in Figure 2 where a four-burner wide arrangement uses six
overfire air ports on each firing wall.

Older overfire air arrangements, which used fewer ports and shorter residence time, can achieve only
about 20-30% NOx reduction from the low NOx burner's level. This AOFA system increases the NOx
reduction capability to' the 40-50% range. Foster Wheeler demonstrated this system's capability in the
early 1980's when two (2) new utility boilers, a 275 MW front-wall fired unit and a 550 MW
opposed-fired unit, were tested, The combination of the CF/SF low NOx burner and the above noted
AOFA principles enabled both units to operate at NOx levels of about 0.2 lb/106 Btu. During
subsequent commercial operation the overfire air ports were closed since the NOx regulations to be

4A-58


-------
attained were only those of the 1977 NSPS. Without overfire air both units operated at about 0.41b/10^
Btu.

The AOFA System was not commercialized by Foster Wheeler until the late 198CFS since there was no
regulatory need to achieve emission levels below 0.3 lb/106 Btu, This system is now being incorporated
on several new steam generators, ranging in size from 65MW to 550 MW. In early 1990 it was retrofitted
to a 500 MW unit, having the arrangement shown in Figure 2, where NOx emissions were reduced 78%
to about 0.26 lb/106 Btu.

FIELD EXPERIENCE: CF/SF AND AOFA

Since its introduction in 1979, when retrofitted to the 360 MW San Juan No. 1 unit of Public Service
New Mexico, the CF/SF low NOx burner has been successfully retrofitted to a total of ten (10) utility
boilers. Table 1 summarizes this experience, which totals 5,135 MW. The average NOx reduction
attained on these units is nearly 60% without overfire air; and nearly 80% with AOFA.

Table II is a listing of the projects underway for 1991: eight (8) units totalling 3,635 MW. Options in
these projects add another 2,380 MW for a total of over 11,000 MW and over 600 burners.

The NO* control capability and results are summarized on Figure 3, which contains data from four of
the ten utility units, two industrial units and Foster Wheeler's Combustion and Environmental Test
Facility. Table III lists the range of fuel properties in these retrofit applications. The NOx control
capability is consistent.

Figure 3 graphically illustrates the NOx control effectiveness of the low NOx Controlled
Flow/Split-Flame burner. The plot is of total NOx emission against Burner zone Liberation Rate (a
measure of heat input to the burner zone; the higher this number the hotter the lower furnace), for
turbulent burners, CF/SF low NOx burner and CF/SF in combination with AOFA. The data show both
industrial and utility units with burner capacities ranging from 30 to 300 million Btu/hr. The curves are
not load curves but, rather, represent the full load NOx emission for each value of Burner Zone
Liberation Rate.

Conclusions drawn from the summary information contained in this figure are:

1.	NO* control is independent of burner capacity: large burners arid small burners
achieve the same degree of NO* reduction.

2.	The low NOx burner (lower two curves) is much less sensitive in the thermal
environment than is the turbulent burner. There is a much smaller slope of the NOx
vs BZLR low NOx curves than for the turbulent burner curves indicating a very
small amount of thermal NOx is emitted by the low NOx burner (due to its lower
flame temperature).

3.	NO* reduction in the higher temperature units is somewhat greater than in the
lower temperature (lower BZLR). units due to the substantial decrease in thermal
NOx in the former.

4.	Uncontrolled NOx emissions from single wall-fired units is higher than from
opposed-fired units, yet this difference is eliminated by the CF/SF burner.

4A-59


-------
5. Foster Wheeler's Advanced Overfire Air System provides an additional NOx
reduction of 40-50% (to total reduction of 70-80%) below levels emitted by the low
NOx burner only.

Foster Wheeler has retrofitted wall-fired steam generators from all major domestic boiler
manufacturers, including an 800 MW opposed-fired unit originally equipped with three-nozzle "cell"
burners.' The effectiveness of the CF/SF low NOx burner is the same regardless of the boiler in which it
is installed.

Of utmost importance to steam generator operation are its performance and efficiency. In none of the
low NOx burner retrofits performed by Foster Wheeler have either boiler performance of efficiency
been deteriorated from pre-retrofit conditions. Typically efficiency is improved due to: .

Reduced excess air operation yielding lower stack losses and reduced forced draft
and induced draft fan power.

Lower burner pressure drop yielding further F.D. fan power savings.

Cleaner furnace walls (reduced slagging),

Unburned carbon levels equal to, or lower than, original equipment burners..

Succinctly, unit operations are equal to, or better than, pre-retrofit while NOx is reduced at least 50%
without overfire air.

Also shown on Figure 3 is the effectiveness of AOFA on three utility units and the CETF. All results,
without and with AOFA, are uniformly consistent in terms of NOx control.

Two projects are of particular interest: Units 1 and 7 on Figure 3. Unit 1 is an 800 MW boiler originally
equipped with 18-3 nozzle cell burners and was retrofitted with 48 CF/SF burners in early 1989. Unit 7
is a 500 MW boiler retrofitted with CF/SF burners and AOFA in early 1990. Since the results of these
two retrofits are typical of all others they will be briefly discussed below.

1. Four Corners Unit No. 4: Cell Burner Boiler - Arizona Public Service

This boiler is a Babcock & Wilcox opposed-fired, supercritical steam generator with
a maximum continuous rating (MCR) of 5,446,000 lb/hr main steam flow at
1000/1000F and 3590 psig. The unit was fired by 18, 3 nozzle cell burners (54
throats). Turbine rating is 820 MWG (780 MW net). Unit 4, along with its sister Unit
5, was built in the late 1960's; and went into commercial operation in 1969 and 1970,
respectively. In 1971 the State of New Mexico instituted a retroactive NOx emission
limit of 0.7 lb/l06Btu for coal-fired units constructed prior to 1971. Over the years
Arizona Public Service conducted several test and evaluation programs to arrive at
an acceptable means of achieving the NOx limit without degrading boiler
operability, performance or efficiency. In 1988 Foster Wheeler was contracted with
to provide a low NOx conversion with a guarantee lower than the State limit (0.65 vs
0.7). The unit was experiencing severe slagging, wide spa rial variation of Furnace
Exit Gas Temperature and resultant high superheater outlet metal temperatures.

Consequently, Arizona Public Service chose to proceed with a complete
revision of the lower furnace. This consisted of replacing both front and

4A-60


-------
rear firing walls with a new burner arrangement, six (6) burners wide and four (4)
burners high.

Foster Wheeler's scope of supply consisted of 48 CF/SF low NOx burners, twelve
burner wall panels (each containing four throats), new igniters and scanners and
miscellaneous equipment. Arizona Public Service had already contracted with B&W
to supply nine new MPS-89N pulverizers for each unit (Unit 5 mills were previously
installed and Unit 4 mills were installed during the low NOx conversion). Due to the
new firing configuration the total number of mills purchased for both units was
reduced from 18 to 16.

It should be emphasized that this conversion represents the most extensive
modification that can be performed on a cell-burner boiler's firing system. However,
the method chosen was designed not only to reduce NOx emissions but also to
improve boiler operability. In particular to produce a more uniform Furnace Exit
Gas Temperature (convert to one mill supplying a burner level) and reduce slagging
(vertically spreading the burners). Slag reduction also occurs because of the lower
flame temperature inherent in the CF/SF burner.

Other cell burner equipped boilers may not require modification as extensive as
Four Corners 4 & 5. In many units only a direct low NOx burner replacement may
be acceptable.

Also, of the total of 23 units either completed, underway or as contractual options,
the two Four Corners units are the only ones requiring panel replacement. The vast
majority of boilers, regardless of original equipment manufacturer, can be
retrofitted with Foster Wheeler low MOx burners without pressure part
modification.

The results of this retrofit are summarized on Table IV which compares pre and
post-conversion conditions.

Significantly, burner pressure drop has been reduced 2-2.5" H2O which will yield
substantial forced draft fan power savings.

During the past nearly two years since start-up the unit has operated according to
system load requirements, almost continually at full load. The following results
have been observed:

-	- There is no flame impingement on any heat transfer surface and no extension of

flames.

-	- Slagging and clinkering have been nearly eliminated on Unit 4. The furnace walls

are now the cleanest they have been since initial unit start-up 20 years ago. In
contrast the unconverted Unit 5 continues to have the same level of slag
accumulation that Unit 4 had prior to conversion.

-	- There has been no increase in FEGT, instead it appears to have decreased as a

result of increased furnace absorption caused by the clean walls.

4A-61


-------
- - Combustion efficiency has not been decreased, as indicated by CO and unburned
carbon levels.

In summary, NOx emissions, have been reduced about 60% with no degradation in
boiler performance or efficiency. Unit operability has been significantly improved
due to the near elimination of the slaggirtg/clinkering problems and elimination of
excessive superheat tube metal temperatures.

Unit 5 is currently undergoing retrofit.

Hsin-Ta #1: Taiwan Power Corp: This is a 500 MW Foster Wheeler natural'
circulation unit in commercial operation since 1978. Originally equipped with an
early non-split flame low NO* burner (FIV Controlled Flow design) and a basic
1970's vintage overfire air system, it was designed to meet a NO* regulation of 0,7
lb/106 Btu. More recently, the NOx regulations in Taiwan have been reduced. As a
result TPC purchased 24 CF/SF low NO* burners and the AOFA system for both
Foster Wheeler units on that site.

Unit 1 was converted in early 1990, and has been operating for approximately ten
(10) months. Because of the simultaneous use of CF/SF burners and AOFA this
unifs operating data will be presented in more detail than Four Corners #4, The
firing system and geometrical arrangement are as shown schematically in Figure 2.

After start-up and low NOx optimization a series of performance tests were done
over a period of several weeks. Since Taiwan imports most of its coal, fuels from the'
US, Australia and South Africa were fired as part of normal plant operations. Table
V lists the typical range of coals fired at this station. Fuels varied on a day-to-day
basis. As shown on the table volatile matter varied from 22.5% to 36.1 % with fuel
nitrogen contents covering the relatively high range of 1.85% to 2.28%.

However, the low NOx systems effectiveness is such that there is no significant
affect of these fuel properties on NOx emissions. Boiler performance and efficiency
variations were only those normally expected due to fuel properties affecting gas
weights and moisture content.

Figure 4 compares NO% emissions as a function of load for the CF/SF burner with
AOFA closed and open. Load was reduced in the manner normal for that station
with pulverizers taken out of service at the same loads they were prior to the low
NOx retrofit. The unit achieved full load with all mills in service or any single mill
out of service. Consequently full load testing was performed with all mills in or one
top or bottom mill out of service.

NOx emissions decrease monotonically from full load values average 0.464 and
0.266 lb/106 Btu with AOFA ports closed and open, respectively. For simplicity of
graphical illustration only seven (7) data points are shown, out of a much larger set
The temporal and fuel-property related variations in emissions cover a narrow
range: at not time did NO* emissions exceed 0.5 or 0.3 with AOFA closed or open,
respectively.

Figure 5 is a similar plot of windbox-to-furnace pressure drop as a function of load.
These results are also typical of the Foster Wheeler low NOx system. At full load

4A-62


-------
with AOFA closed and all mills in service AP does not exceed 3.0" H2O. When
AOFA ports are opened AP decreases to less than 1.5" H2O at full load. In both cases
AP decreases monotonically as load decreases.

Figure 6 illustrates the change in variability of the NOx emission as more
sophisticated NOx controls are used at full load. From a mean (X) emission of 1.23
lb/10® Btu and standard deviation (S) of 0.06 in the uncontrolled case, X and S are
decreased to 0.464 and 0.032 with CF/SF burner and no overfire air. When AOFA
ports are opened X = 0.266 and S = 0.012. The data includes all mill combinations
and the range of fuels listed in Table V.

The following summarizes the results of Figures 4,5 and 6.

-	- NOx does not increase as load is decreased.

-	- Full load pressure drop is no greater than for typical turbulent burner values, less

than 3.5" H20 at full load.

-	- High burner pressure drop is not needed to attain high NOx reduction and low

absolute emission level.

-	- NOx emission variability due to operating condition and fuel properties decreases

with the use of more sophisticated NOx controls.

Although the above results and conclusions are presented for two specific
retrofitted units, they are typical of data obtained on all other retrofits as well as
from thousands of additional MW's of new steam generators utilizing the same
equipment.

INTERNAL FUEL STAGED™ LOW NO* BURNER DEVELOPMENT

The development of the IFS low NOx burner has been performed on the Combustion and
Environmental Test Facility (CETF) located at Foster Wheeler's manufacturing facility in Dansville, NY.

Among the types of test work being performed at the CETF are burner development, low NOx furnace
design evaluations, sulfur dioxide control by dry sorbent injection, fuels evaluations, integrated
combustion/emissions control testing and customer support and problem analysis.

The CETF's furnace (Fig. 7) is arranged to produce conditions which closely match those of commercial
equipment, for example:

•	Furnace residence time is limited to about 2 sec. max. between the burner centerline
and furnace exit.

•	Furnace Exit Gas Temperature (FEGT) is about 2200 T and can be varied, by
adjusting total furnace absorption.

•	Burner/furnace aerodynamics are similar to those of commercial equipment as are furnace
mixing patterns, due to both overall geometry and the two-bumer-high arrangement.

4A-63


-------
The arch configuration, which is required to provide proper conditions for low volatile fuel
combustion, is also important to produce the desired velocities/residence times and FEGT with
horizontal firing. The reduced upper furnace cross-section is responsible for the increased
velocity and resultant decreased residence time. Decreased upper furnace heat transfer surface
reduces absorption above the burner zone, thereby increasing FEGT to levels normally achieved in
commercial equipment.

Although not currently equipped with a superheater, the unit has screen tubes at the furnace exit
followed by a water-cooled economizer which yield realistic gas quenching rates. In this way all flue
gas constituents, gaseous and particulate, are comparable to field equipment. Commercial practices,
used throughout the CETF system design, have been mated with research-oriented considerations
wherever practicable to maximize the usefulness and flexibility of the system.

The CETF utilizes a direct-fired system (i.e., hot-primary-air-swept ball mill) feeding either a single 75
million Btu per hour arch-fired twin-cycle burner assembly or two 40 million Btu per hour
horizontally-fired burners (shown) which fire into a refractory-lined water-jacketed furnace. The water
jacket operates under water-head pressure and utilizes natural circulation, producing steam which is
vented to the air through a steam drum above the furnace. Combustion gases leaving the furnace flow
over horizontal, convection tube surfaces (economizer) cooled by forced-water circulation. The gases
then pass through a two-stage air heater (a tubular air heater followed by a heat pipe air heater), a
baghouse dust collector,- an induced-draft fan, and then the stack. The level of sulfur dioxide is
controlled by injecting a sodium-based sorbent into the gas stream prior to the baghouse. In addition to
back-end sulfur dioxide cleanup, the CETF has the capability of furnace injection of calcium-based
sorbents for evaluation of in-situ sulfur dioxide control.

IFS DESCRIPTION

The IFS concept, as noted in the introduction, was conceived in the 1970's but not developed at that
time. Internal Fuel Staging is defined as;

A low NOx burner which two-staged the secondary air flow and internally stages
the fuel flow and primary air flow such that co-axial flames are developed within
the burner's throat while maintaining classical turbulent burner flame patterns and
low pressure drop; < 4.0" H2O.

Figure 8 schematically illustrates the IFS design. The philosophy of this development was to utilize as
much of the commercially-proven hardware of the Controlled Flow/Split-Flame low NOx burner as
possible. As can be seen in Fig. 8, the IFS differs from the CF/SF in only one respect; the coal injector's
nozzle.

All other mechanisms and functions are identical to those of the CF/SF, including operating method.
Externally, the two burners appear to be identical. However, internally the fuel injection's nozzle has
been redesigned to produce split-flames surrounding a co-axial internal flame. The result is NOx
emissions approximately one-third lower than those attained with the CF/SF design.

Table VI is a comparison of the key features of the CF/SF and the IFS. The similarity is so great that an operator
would note no difference between the two designs, other than the lower emission from the IFS.

The IFS burner is intended for use in both new and retrofit applications. Also, if units which are already equipped
with the CF/SF design require lower emissions all that is necessary is to replace the fuel injector.

4A-64


-------
IFS PERFORMANCE RESULTS

Due to extensive performance and emission data available for the CF/SF low burner, from the CETF
and new and retrofit field applications, IFS testing on the CETF is compared to CF/SF data-.

Boiler performance results demonstrate no adverse difference in either furnace absorption or Furnace
Exit Gas Temperature on the CETF between the two low NOx burner designs. Since extensive field data
already exits which shows no change in boiler performance when the CF/SF is retrofitted to boilers
initially equipped with, turbulent burners, it can be inferred the IFS-design will, similarly, not adversely
affect boiler performance.

A wide range of fuels, sub-bituminous and bituminous have been tested on the CETF using the CF/SF
burner. For the IFS development testing two baseline bituminous coals have been used: a low volatile
and several high volatile coals.

Table VII compares NOx and CO data at full load for the two burner designs with the high and low
volatile coals. NOx is reduced 33-35% below CF/SF levels using the IFS. For all data, overfire air ports
are closed and excess O2 is 3.4-3.6%. Note that for both fuels the variability in NOx emission decreases
significantly, to about the same standard deviation. Typical coal analyses are listed in Table VIII.

Figure 9 compares NOx and air-to-coal ratio (A/C) to load. Over a turndown ratio of nearly 3:1 NOx
decreases monotonically with load from about 0.27 to about 0.21 lb/106 Btu. A/C increases as load
decreases from about 2.1:1 to about 2.6:1, covering the range of typical vertical roller pulverizers. This is
a significant result in that reduced load operation does not deteriorate the IFS burner's performance:
NOx does not increase as load decreases.

It is also instructive to compare the CF/SF NO*, data from the 500 MW boiler, without and with
overfire, to the CF/SF. and IFS data from the CETF without overfire air. Figure 10 presents such a
comparison. The mean uncontrolled emissions for the 500 MW and the CETF are 1.23 and 1.04 lb/106
Btu for the fuel ranges listed in Tables V and VIII respectively. The 500 MW unit's emission in higher
than that of the CETF because it is a hotter unit and was firing higher nitrogen fuels than the CETF.

On each unit, NOx is reduced by over 60% with the CF/SF low NOx burner without overfire air, Note
that both the uncontrolled data and CF/SF data have nearly identical standard deviations
approximately 0.06 and 0.03, respectively. When overfire air'ports are open on the 500 MW unit NOx is
further decreased to a mean value of 0.266 lb/106 Btu: a total reduction of over 78%.

On the CETF the IFS low NOx burner yields a mean NOx emission of 0.269 lb/106 Btu without overfire
air: a total reduction of nearly 75%. Note that the standard deviations on the 500 MW unit with overfire
air ports open is about the same as that of the CETF when the IFS is used with overfire air ports closed.
Again illustrating that more sophisticated NOx controls become less sensitive to operating conditions
and fuel parameters.

A further comparison from this data is the NOx reduction from CF/SF levels due to AOFA or IK. On
the 500 MW unit NOx reduced 42.7% when the OFA ports are open. On the CETF, the IFS burner
reduces NOx 33.6% below CF/SF levels. The IFS low NOx burner is achieving a NOx reduction
capability that is 80% of the additional reduction AOFA attains beyond CF/SF levels.

The relationship of the IFS results to field data are graphically illustrated on Figure 11, which is a simplified version
of Figure 3. Clearly, many commercial boilers will be able to operate at levels below 0.4 lb/106 Btu without AOFA
Where lower levels are required the IRS low burner can be supplemented with the AOFA System.

4A-65


-------
SUMMARY

Foster Wheeler's standard Controlled Flow/Split-flame NOx burner has been successfully retrofitted
to ten (10) utility steam generators ranging in size from 225 MW to 800 MVV, with an additional' 6000
MW either underway or as options to existing contracts. In 1991 3635 MW will be retrofitted. Of these
three units, two B&W and one Foster Wheeler, will be receiving the new Internal Fuel Staged™ low
NOx burner design.

The EFS burner development has been successfully completed and it has been offered commercially for
new steam generator and retrofit use, The full range of emissions and boiler performance guarantees,
are offered. In addition to the above noted retrofits, the IPS will be installed on the following new steam
generators:

2 x 65 MW; NOx guarantee:	0.32

2 x 150 MW ; NQx guarantee: . 0.27
1 x 550 MW; NOx guarantee: . 0,32

*

These Units will also be equipped with Foster Wheeler's SCR system to achieve 0.1
lb/106 Btu at the Stack

Hie IFS design achieves NOx levels one-third lower than those attained by the standard CF/SF low
NOx burner, when tested on the Combustion and Environmental Test Facility. NOx levels below 0.27
lb/106 Btu without simultaneous use of overfire air, are routinely attained with a wide variety of
bituminous coals with NOx emissions showing little sensitivity to operating mode or fuel
characteristics. These performance results are obtained at low burner pressure drop and with short
flames which do not cause adverse changes to furnace absorption rates or Furnace Exit Gas
Temperature.

It is fully expected that these results will be duplicated in the new unit and retrofit applications
currently underway.

4A-66


-------
Table I
Low NOx Burner
Retrofit Experience

NOx (lb/10 Btu)

Reduction

Year

MW

Before

After

<%>

OEM

1979

1980
198,1

1986

1987

1989

1990
1990
1990
1990
Total

360
525
275'

650
500

800 (Cell)

800

500

500

225

5135MW

0,95
*

*

1.0

1.15

1.25

1.10

1.23

1.15

1.15

0,42

0.40

0.40

0.41

0.55

0.50

0.45

0,27+

0.50

0.50

60
59
78
56
56

FW
FW
FW
FW

FW-UK
B&W
FW
FW

FW-UK
FW

New Boiler Converted during erection

With Advanced Overfire Air System

NOTE: Boiler ages range from 10 to 25 years

Table II
Low NO* Burner
Retrofits Underway of 1991

Unit Size (MW)

No. of Burners

Boiler OEM

800 MW (Cell-Burner)
150 MW

200 MW (3 x 200)*
650 MW
500 MW
525 MW

650 MW (2 x 650)
160 MW (3x160)
Total 3635 MW

48
16

18(54)

24
24
24

24(48)
16(48)

194

B&W
B&W
B&W
FW
FW
FW
FW
FW

<

) = Contractual Options

4A-67


-------
Table III
Range of Fuel Properties

HHV 8000	- ' 13.500 Btu/lb

N2 1.0	- 2.3%

Si 0,5'	- 3.5 %

Ash 5.0	- 2,5 %

Average NOx Level	0.45 lb/106 Btti
NOx Reduction 55 - 60 % (No OFA))

Parameter

¦No. Mill in Service
Excess Air (%)

NOx (lb/106 Btu)

CO (ppm)

Burner A P ("H2O)

Carbon Loss (% Eff.).

Results:

Performance and efficiency not degraded
Eliminated slagging and clinkering
Significant power savings

Table IV
Four Corners # 4 Data
800 MW Cell-Burner Unit

Pre-Conversion

All (9)

18

1.27
<40
5-6
<0.1

Post-Conversion

All (8)
13-15
0.48 - 0.52
20-40
3-3.5
<0.1

Table V

500 MW Advanced Low NO* Retrofit
Ranee of Fuels Fired
(As Received)

VM (%)
FC (%)
Ash (%)
H2O (%)

22.51
54.05
15.55
7.49

28.07
52.63
12.45
6.85

36.15
41.47
12.01
10.37

C
H
O
N
S

HHV (Btu/lb)

64.49
3.79
5.48
2.07
0.73
10,909

67.39
4.11

6.27

2.28
0.65

11,605

61.99
4.79
8.70
1.85
0.59
11,018

4 A-68


-------
Table VI
Low NOx Burner Comparison

Desgn Parameter

CF/SF

IFS

Registers

Adjustments
Sec. Air Flow Control
Fuel Injector

Primary Air Velocity Control
NOx Reduction
NOx Level

Dual
Manual
Elec. Sleeve Damper
Split-Flame

Yes
55 - 60%
0.4 lb /106 Btu

Dual
Manual
Elec. Sleeve Damper
Split-Flame
Plus

Coaxial Internal Flame
Yes
70 - 75%
0.27 lb / 106 Btu

Table VII
CETF Comparative Data
' • CF/SF vs IFS

Fuel V.M. (%)
N2 {%)
Xs O2 (%)
Burner Type

23
1.87
3.5-3.6
CF/SF

IFS

33
1.59
3.5 - 3.6
CF/SF

IFS

NOx (lb/10 Btu); Avg.

Sid. Dev.

0.41
0.033

0.273
0.014

0.40
0.035

0.260
0.011

CO

Avg.
Std. Dev.

54
13

59
11

70
15

75
17

NOx Reduction {%)

33.4

35.0

4A-69


-------
Table VIII
CETF Fuels: IFS Testing

VM	32.70	23.49

FC	48.16	59.41

ASH	9.11 . 4.34

H2O	10.03	7.86

C	67.34	77.37

H	4.50	4.68

O	5.60	3.19

N ¦ 1	- 1.59 .	¦ 1.87

S	,1.85	0.79

HHV	11,935	13,877

NOx Reduction	79.5%	73%

NOx Level	0.266	0.273

No significant difference in NO* emissions due to coal characteristics

4A-7Q


-------
beciuc aevE

DJWER DW

tmm 01®

USSIER DHVE

Fig 1. Controllled Flow Split Flame Burner

town

WMIttt

«imoih

IfCOMPAIV
All DUCt

ithcw

OVEIFlit

»i»fon

CAMMti

Fig 2. Typical Advanced Overfire Airport System

4A-71


-------
(1)

(2)

o

(3)

800 MW Four Corners #4
626 MW Pleasants #2

275 MW Front Wall Fired
360 MW Front Wall Fired
525 MW Opposed Fired

1.2

NOx

Lb/M Btu

0.8

0.6

0.4

0.2

Single Wall Fired Units^

Pre-NSPS Burner

(4)	110,000 Lb/Hr. 4 Burner

(5)	125,000 Lb/Hr. 4 Burner

(6)	CETF

(7)	500 MW Opposed Fired

Opposed Fired Units

CF/SF Low NOx Burner

!-'R.J 	. . .

CF/SF + Advanced OFA

50 100 150 200 250 300 350 400 450

BURNER ZONE LIBERATION RATE

(10 3Btu/Hr-Ft 2

Foster Wheeler Energy Corp.
Combustion & Environmental Systems

Fig. 3. NO^ Reduction Summary


-------
N0X
lb/106 Btu

Ovariire A!r
Closed X

GT

Overfire Air'
Open

KEY

0 ALL MILLS IN SERVICE
®ED ONE MILL OUT OF SERVICE
4 TWO KILLS OUT OF SERVICE



300 350 400 450 500
LOAD (MW)

Fig, 4. Advanced Low NO( Retrofit 500 MW Boiler NOjvs Load

£L
<3
UJ
Q
<
!Z
£C

£
o

f-

s

03
Q
Z

4.0
3,5
3.0
2.5
2.0
1.5
1.0
0.5

KEY

© ALL MILLS IN SERVICE
®0 ONE MILL OUT OF SERVICE
A TWO MILLS OUT OF SERVICE

OVERFIRE AIR

CLOSED

OVERFIRE AIR
OPEN

300

450

350 400
LOAD (MW)

Fig, 5. Advanced Low NO_ Retrofit 500 MW Unit Pressure Drop vs Load

500

4A-73


-------
NOx 1-4
lb/106 Btu 1 3

1,2

1.1

1.0

0.9

0.8

0.7

0.6

0,5 ¦

0.4

0.3

0.2

0.1

UNCONTROLLED

S = 1.23
S = 0.06

REDUCTION
62.5% 78.6%



CF/SF	.

LNB mmmmm * - 0.464
ONLY *rn",r" s _ 03J2

LNB
&

AOFA



? =0.266
S ¦ 0.012

CF/SF = Controlled Flow/Split-Flame
LNB = Low N0X Burner
AOFA = Advanced Overflre Air
LOAD = 500 MW; 02 = 3.5%

Fig. 6.500 MW Boiler NO, Reduction

Tig. 7. Combustion and Enviornmental Test Facility

»4A-74


-------
Fig. 8. Internal Fuel Staged Low NO, Burner™

NO„

0.40 -

{lb/10 StU)

0.35-



0.30"



0.25"



0.20 -



0.15 "



0.10"



0.05"



o.oo-

A/C

LOAD (% MCR)

For all lest points; CO < 50 ppm ; 02 = 3.4 - 3.6%
4P = 1.S"HjO Maximum
No Overfire Air

Fig. 9. CETF Testing Internal Fuel Staged Burner

4A-75


-------
UNCONTROLLED

0.464
0.032

5 = 0.266
S = 0.012

500 MW
Steam Generator

UNCONTROLLED

t = 1.04
S = 0.057

74.6%

IFS
NO OFA

0.405
0.034

t = 0.269
S = 0.015

80 M Btu/hr
Test Facility

Fig. 10. Low NO< Burner Comparative Data Controlled Flow/Split Flame vs Internal Fuel Staged

4A-76


-------
KEY:

Uncontrolled Levels
CF/SF - with AO FA

O CF/SF - no OFA
j. IFS - no OFA

>

i

CF/SF Low NOx Burner

i : • :	

CF/SF + Advanced OFA

50 100 150 200 250 300 350 400

BURNER ZONE LIBERATION RATE

450

(103 Btu/Hr-Ft2)

Foster Wheeler Energy Corp.
Combustion & Environmental Systems

Fig. 11. NO^ Reduction Summary IFS vs. CF/SF


-------

-------
REDUCTION OF NITROGEN OXIDES EMISSIONS BY COMBUSTION PROCESS
MODIFICATION IN NATURAL GAS AND FUEL OIL FLAMES:
FUNDAMENTALS OF LOW NO>: BURNER DESIGN

M.A. Toqan, L. Berg, and J M. Beer

Massachusetts Institute of Technology
Cambridge, MA 02139

A. Marotta, A. Beretta and A. Testa
Eniricerche
Italy

4A-79


-------

-------
Abstract

Increasingly tight environmental regulations for NOx emission from coal-, oil- and gas-
fired utility boilers are forcing utility and industrial users of fossil fuels to pay greater
attention to control of NOx in oil, coal and even gas-fired units. Effective control of NOx
emissions requires the application of one or a combination of methods of combustion
process modification including staged air and staged fuel injection, the use of Iow-NO,
burners, and possibly even post-combustion clean-up such as NH3 injection into combustion
gases.

To date, the degree of NO, reduction achieved by these technologies has been
observed to vary widely and to depend on the combustion system in question. Primarily
this wide variation in performance of staged systems and of low NOx burners is due to lack
of understanding of the overlapping processes of the nitrogen-hydrocarbon chemistry and
the mixing/temperature histories of the fuel in the flame.

To address the variation of performance of low-NO, burners a theoretical and an
experimental investigation is being carried out at MIT which is focused on the
fundamentals of low-NOz burner design, applied to natural gas and fuel oil combustion.

In the experimental investigation a multi-annular burner developed by Massachusetts
Institute of Technology is used. Flame studies are carried out in the 1.2 m x 1.2 m x 4.5 m
test section of the MIT Combustion Research Facility (CRF). The CRF is a pilot plant
scale combustion tunneL, having a 3 MW^ multi-fuel firing capability and designed to
facilitate detailed investigation of industrial type turbulent diffusion flames. The burner is
equipped with a fuel gun surrounded by primary, secondary and tertiary air supplies. Mass
flow rates for each of the three air supplies external to the fuel gun can be independently
controlled, and for each supply the swirl can be adjusted over a wide range by means of an
independent moveable block swirler. A shroud-diffuser is used to maintain physical
separation of the secondary and tertiary air jets entering the combustion chamber.

The results from this investigation are pertinent to the desip principles of low-NO,
burners including scaling criteria.

4A-81


-------
INTRODUCTION

The most widely used design strategy for NOx reduction is staged combustion. The
creation of fuel-rich and -lean combustion zones in flames by means of staging the input
of either air or fuel is a successful method of NOx emission control. Nevertheless, the
degree of NOr reduction achieved by these technologies has been observed to vary widely
and to depend on the combustion system in question (1,2,3).

While the principles of staged combustion control of NO,, emission are well
established (4,5,6), their practical realization is hampered by lack of information on the
overlapping processes of the nitrogen-hydrocarbon chemistry and the mixing-temperature
history of the fuel in the flame. The problems are especially difficult in the case of the so
called "internal staging" process, in which the fuel-rich and -lean combustion zones must
be produced by appropriate fuel-air mixing in a single low-NO, burner, rather than by
producing fuel- rich and -lean combustion zones in the combustion chamber using overfire
air.

This problem is addressed in aa experimental research project at MIT focused on
the principles of low-NO„ burner design as applied to natural gas and oil combustion. This
paper reports results obtained with both natural gas and No. 2 and No. 6 oil as fuels. The
experimental RSFC (Radially Stratified Flame Core) burner was developed at MIT based
on the patented design of a multiannular burner (7). The RSFC burner is attached to the
flame tunnel (3 MWm, 1.2 x 1.2 x 4.5 m) of the MIT Combustion Research Facility (CRF).
Parallel with the experiments a mathematical modeling study is carried out; the progress
of combustion along the flame is computed for the effects of design and operating variables
of the burner, using the "Fluent" fluid dynamics code. In this paper, the relationships
between burner input parameters and emissions of CO and NO, are reported.

EXPERIMENTAL

The MIT Combustion Research Facility was designed to permit detailed in-flame
measurements of the flow field and of spatial distributions of temperature and chemical
species concentrations to be made. Variable heat extraction along the flame - by the use
of completely and partially water cooled furnace sections - is used to closely simulate large
scale flame systems. Access to the flame for optical or probe measurements is provided by
a 1.0 m long slot at the burner and by instrument ports at every 30 cm length further
downstream along the flaxne tunnel.

The experimental RSFC burner, the concept of radial flame stratification

The burner consists of three concentric annuli with each of the annular nozzles at
a larger radial position extending further in the axial direction (Fig. 1). Fuel is introduced
in the center through a fuel gun. The three sections of the burner can be axially adjusted
as may be required to maintain optimal geometry at turn down. Additional features of the
burner include independently variable swirl control in each annular air nozzle by means of
IFRF moveable block swirlers.

4A-82


-------
>

00
CO

TERTIARY AIR—£

SECONDARY AIR—I

BURNER QUARL

MOVABLE SHROUD

BURNER QUARL

PRIMARY AIR

SWIRL GENERATOR

Figure 1. Schematic of Low-NOx Radially Stratified Flame Core Burner


-------
The operation of the burner is based on the principle that a combination of a
positive radial density gradient and rotating flow field has a stratifying effect on the flame
by virtue of the damping of turbulence at the interface of the central flame zone and the
colder air flow radially surrounding the flame (8). This feature was chosen to give the
name; Radially Stratified Flame Core (RSFC) Burner. The flame so produced would have
mixing zones in the opposite sequence to the Flame type 1. used in the IFRF terminology,
as it would start with a narrow fuel jet flame of low air entrainment followed by an
internally recirculating flame region in which combustion is completed (Fig. 2).

The flexible design of the experimental burner permitted the variation of input
parameters over wide ranges. The burner parameters were:

Fuel jet velocity and angle

Fuel gun position (relative to the face of the burner)

Air distribution to Primary, Secondary and Tertiary air flows

Radial distribution of the swirl velocity in the air flow

Axial separation of the Primary, Secondary and Tertiary air flows

Experimental Matrix

Using natural gas and No. 2 and No. 6 oil and preheated combustion air (45Q"F)
parametric flame experiments were carried out with the RSFC burner. In the experiments,
burner input parameters were varied to determine their effect upon NOx and CO emissions.

The ranges of variables adopted were:

Fuel jet velocity:

Fuel jet angle:

Fuel gun position:

Primary air flow rate:

Secondary air flow rate:

Tertiary air flow rate:

Swirl number of primary air:

Swirl number of secondary air:

Swirl number of tertiary air:

Elemental analyses of the No. 2 and No. 6 fuel oils are listed in Table 1,

Measurements

Temperature, and gaseous concentrations of CO, C02» NO, and 02 were measured
at the exit of the combustion tunnel.

50 - 600 ft/sec.
0° - 25°
-45 - 0 cm
0 - 100%
0-100%
0 - 100%
0 - 2.79
0- 1.90
0 - 1.39

4A-84


-------
TERTIARY AIR

FUEL GUN

PRIMARY AIR

SECONDARY AIR

EXTERNAL RECIRCULATION
ZONE

FLAME ENVELOPE

INTERNAL

RECIRCULATION
ZONE

Figure 2. Schematic of a Radially Stratified Low NOx Flame

4A-85


-------
Table 1

Ultimate
Analysis

Weight %

No. 2

No. 6

C

86,80

86.46

H

12.44

9.67

N

0.14

.53

Ash

0.01

.08

h2o

_

.4

Asphaltene
Content

..

9.5

Heating value,
(Btu/lb)

19,640

18,236

EXPERIMENTAL RESULTS
Natural Gas Combustion

In the N.G. tests, 98 flames were investigated for the effects of burner input
variables upon NOx and CO emissions from the combustion tunnel. The input variables
found to have effect upon NOx and CO emissions are:

type of fuel nozzle

fuel gun position within burner

primary air fraction

radial displacement of swirl from flame axis
Type of Fuel Nozzle

Two parameters, the exit velocity of the fuel jet from the gun and the angle of the
jet relative to the flame axis, were considered in the design of the fuel nozzles. Several
nozzles were built to allow the velocity of the fuel to range from 50 ft/sec. to 600 ft/sec. and
the angle to vary from 0° to 25°. Results obtained from the combustion tests with these
nozzles are shown in F:gtires 3 and 4. It is noteworthy that while CO emission levels were
very low for all cases, they increased slightly with increasing fuel jet velocity. On the other
hand, NO, emission levels were more influenced by the fuel jet angle: i.e., increasing the
fuel jet angle from 18 to 25° increased NOx concentration at the exit by "* 25 %.

4A-86


-------
NATURAL GAS

NO-

CO

¦100
90
•80
¦70
60
50
40
30
20
10
0

20 30 40 50 60 70 80 90 100 110 120 130
FUEL JET VELOCITY (FT/SEC)

EFFECT OF FUEL JET VELOCITY ON NOx AND CO
EMISSIONS (02 at exit = 1.5 %)



100-



90

E

a.
a.

80-

z

70-

o



I—

60-

<



£C
H

50-

Z

40-

LU

o



Z

o

30-

o

20-

o

o

10-



0-

Prim, air
Sec. air
Tert. air

%	Swirl No.

low	high
zero

high	high

a.

¦S
z
g

I-
<
a-
H

z

UJ

u
z ¦

o
u

x

O

Figure 3.

4A-87


-------
NATURAL GAS



100^



90-

£



Q.

a

80-

z

70-

o



1—

60-

<



cc

f-

50-

z



UJ

40-

o



z

o

30-

o

20-

u

o

10-



0-*

%	Swirl No.

Prim, air low	high
Sec. air zero

Tert. air high	high

CO

0

15

FUEL ^\NC3LE

NO

25

200
180
H60
140
120

Moo

80

-60
-40
20
0

E

Q.

a.

a:

i—

z

Ui

U
z
o

Q
x
O
z

EFFECT OF FUEL JET ANGLE ON NOx AND CO
EMISSIONS (02 at exit = 1.5 %)

Figure 4.

4A-88


-------
Fuel Gun Position

The axial position at which the fuel is introduced within the burner is known to be
important in determining the flame structure. Fluid dynamically it affects the interaction
of the axial fuel jet and the swirling annular air flow. To investigate the effect of this
parameter upon NOx and CO emissions, several flames were investigated in which the
location of injection of fuel within the burner was varied. Figures 5 & 6 illustrate the
effect of this variable for the cases of strongly and weakly swilling primary air. The
negative values of the fuel gun positions shown in Figures 5 & 6 indicate the distance
between the end of the burner face and the fuel pn tip. A negative value implies that the
gun has been retracted into the burner throat It can be seen from Figures 5 and 6 that
the fuel gun position has little effect on NO, concentration. However, CO emissions were
observed to increase dramatically when the fuel gun was moved in the burner for certain
burner configurations.

Primary air fraction

The conditions represented in Fig, 5 with 51% of the air supplied as primary air give
high NO, values, ranging from 110 to 135 ppm, while CO concentrations are
understandably low because of the early aeration of the fuel. It is the case illustrated in
Fig. 6 that deserves further discussion. With the low primary fraction, N0X levels are in
the range of 75 to 85 ppm which shows that even at a low level of swirl in the primary air,
fuel/air mixing is damped in the near field. However, as the primary air fraction is raised
as illustrated in Figures 7 and 8, NOx emission levels increase due to the early mixing of
the fuel with the combustion air. It is noteworthy that for the cases which have low
primary air fraction, the lean stage mixing further downstream is inefficient without strong
swirl in the tertiary air. For the condition of high swirl degree of the primary air, NO
concentration is mainly dependent upon the primary air fraction. The CO emissions,
however, are more dependent upon the swirl degree of the secondary/tertiary air. For the
cases in Fig. 8 the CO concentration remains virtually constant over the full range of
primary air flow fraction as long as the tertiary air has a high degree of swirl. An optimum
flame was found in which the burner input conditions reflect the above trends: low primary
air fraction, with high swirl, high secondary mass flow fraction with over critical degree of
swirl, and low tertiary air flow with no swirl, ( NO emission at 3% 02: 70 ppm; CO; 56 ppm
and the 02 concentration in the exhaust 1.85%).

Similarly favorable conditions were obtained with low primary, low secondary and
high tertiary air flows as long as swirl was imparted both to the primary and the tertiary air
flows.

Radial displacement of swirl from flame axis

With the multi-annular burner it is possible to produce a wide range of different
types of swirling flows. The two extreme cases are the free and forced vortex flows.
Assuming a uniform axial velocity profile, free vortex flow is obtained by high swirl in the
primary air, low swirl in the secondary air and no swirl in the tertiary air. A forced vortex

4A-89


-------
NATURAL GAS

400

e

350-

Q.



Q.

300-

Z



o

c

250-

<



DC
H-

200-

Z



IU

o

150-

z



o
u

100-

o



o

50-

0

%	Swirl No.

Prim, air medium	high
Sec. air zero

Tert. air medium	high

NOx

200
180
160
140
(-120
100
80
60
40
20
0

E
a
a.

<
IT
I-
Z
HI
U

z

o
o

x

o
z

-27.5	-19	-13

FUEL GUN POSITION FROM QUARL OUTLET(cm)

EFFECT OF FUEL GUN POSITION ON NOx AND CO
EMISSIONS (02 at exit = 1.5 %)

Figure 5.

4A-90


-------
NATURAL GAS

CL

3

z
o

P

<
DC

z

Ui

o
z

Q
O

O
U

1000
900-
800-
700
600-
500
400
300-
200-
100-
0

NO

Swirl No.
Prim.air low low
Sec. air low low
TerL air high zero

-50

-45

^40

-35 -30~ -25 -20
FUEL GUN POSITION (CM)

-15

100
90
80
h70
60
i-50

h30
20
10

0
10

E
a

Ok

2H

o

<

a:

z

40 w
O

o

Q

O
z

EFFECT OF FUEL GUN POSITION ON NOx AND CO
EMISSIONS (02 at exit = 1.5 %)

Figure 6.

4A-91


-------
NATURAL GAS



100-1



90-

E



o.
a

80^

z

70-

o



I-

60-

<



DC

50-

Z



UJ

40-

O

z

o

30-

o
o

20-

u

10-



0-





NOx



CO



%

Swirl No.



Prim, air

high



Sec. air

high



Tert. air zero

*



0.0 0.2 0.4 0.6 0.8

FRACTION OF PRIMARY AIR

1.0

1.2

EFFECT OF (PRIM./SEC.) AIR RATIO
ON NOx AND CO EMISSIONS (02 at exit = 1.5 %)

Figure 7.

4A-92


-------
CL

Cl

Z

o

i—

<
DC

z

UJ

a
z

o
o

o
o

100

80'

60-

40

20

0-
0.0

NATURAL GAS













- NOx





"4"







1 1

o
°

%

PRIM AIR
SEC AIR 0
TCRT AIR .

Swirl No.
HIGH

HIGH

0.4

-140

0.8

200

180 i*
160 £

z

o

¦120 g
oc

100 H

80
60
40
20

1.2

FRACTION OF PRIMARY AIR

ui
O

2

O
a

X

O
z

EFFECT OF (PRIM./TERT.) AIR RATIO ON NOx AND CO
EMISSIONS (02 at exit = 1.5%)

Figure 8,

4A-93


-------
swirling flow is obtained by high swirl in the tertiary air, a lower swirl in the secondary air
and a no swirl in the primary air. In a Rankine type vortex a forced vortex in the core of
the rotating flow combines with a free vortex on the outside, A Rankine vortex can be also
produced by the appropriate adjustment of the radial distribution of the swirl velocity in
the RSFC burner. To investigate the effect of the radial displacement of the peak swirl
velocity (tangential velocity) component in the combustion air from the flame axis, several
flames were generated by imparting varying swirl degrees to the primary, secondary and
tertiary air jets. The effect of this parameter on NO, concentration is illustrated in
Figure 9. It can be seen that lowest NO, emission was obtained with a Rankine vortex type
swirl velocity distribution.

No. 2 and No. 6 Oil Combustion

Based on the experience gained from natural gas combustion, 41 flames were
investigated for the effect of burner input parameters upon NO, and CO emission levels.
The parameters varied included:

(1)	Type of fuel nozzle

(2)	Primary air fraction

(3)	Radial displacement of swirl from flame axis

Type of fuel nozzle

Six hole T jet twin fluid atomizers were used and the angle between the six fuel
jets and the axis was varied to range from axial (0°) to 25" half angle. NO, concentrations
obtained using these atomizers for several primary/secondary/tertiary air ratios and several
swirl numbers are shown in Figures 10 and 11. The experimental results show the
importance of the fuel jet angle upon NO, emission levels. For No. 2 oil, changing the fuel
jet angle from axial (0°) to 25° raises NO, emission levels from 54 to 90 ppm for the best
cases examined and from 90 to 250 for the worst cases. For No. 6 oil, using a 25° atomizer
instead of an axial (0°) nozzle has the effect of raising NO, concentration levels at the exit
of the combustion tunnel from 97 ppm to *" 170 ppm for the best cases and from ~ 240
to ~ 300 ppm for the worst cases.

Primary Air Fraction

The effect of the primary fraction ratio upon NO, emissions is shown in Figure 12.
It is noteworthy that for both fuel nozzle types when the primary air fraction constitutes
~ 0.5 of the total combustion air or higher, NO, emissions are high i.e., ranging from 86
to 140 ppm for No. 2 oil and ranging from 235 to 330 for No. 6 oil. As the combustion air
is diverted towards the tertiary air port, Jie concentration of NO, measured at the exit for
the axial (0°) nozzle drops to 59 ppm and 105 ppm for No. 2 and No. 6 oils, respectively.
A similar trend is observed with the 25° fuel jet atomizers. However, the lowest emission
level achieved with the latter could not match that obtained with the narrow angle fuel
nozzle. These results are similar to those obtained with natural gas. It is concluded that
only a small fraction of the combustion air has to be supplied around the fuel spray to

4A-94


-------
g

a.
Q.

130-

120

100

E 1,0

<
EC
)_
2

UJ

O

z
o
o

x

o

NATURAL GAS

RANK1NE VORTEX

FORCED VORTEX

FREE VORTEX

70-

0	1	2

OVERALL SWIRL NUMBER

EFFECT OF TYPE OF VORTEX AT BURNER EXIT ON
NOx EMISSIONS

Figure 9.

4A-95


-------
No.2 OIL

EFFECT OF FUEL JET ANGLE UPON NOx EMISSIONS

Figure 10.

4A-96


-------
No.6 OIL

EFFECT OF FUEL JET ANGLE UPON NOx emissions

Figure 11.

4A-97


-------
NOx

CO

Prim, air frac. Low High Low High Low High Low High

25

2g O

Fuel Jet Angle —	 -

EFFECT OF PRIMARY AIR FRACTION ON NOx AND CO EMISSIONS

FROM OIL COMBUSTION

Figure 12.

4A-98


-------
ignite the fuel and stratify the flame. The remaining fraction can then be introduced as
tertiary air.

Radial displacement of swirl from flame axis

Using narrow angle fuel sprays, combustion experiments were carried out with No.
2 fuel oil to investigate the effect of radial displacement of peak tangential velocity of tlie
air upon NO, emissions.

In Figures 13 and 14 the effects of the swirl degree, the radial distribution of the
swirl velocity of the combustion air and the angle of the fuel spray upon NO, emission are
illustrated. NO, emission is seen to decrease with increasing swirl number and "Ranktne1'
vortex is favored for the air flow and a narrow angle spray for the fuel oil. Minimum NO,
levels were 97 ppm for No. 6 fuel oil and 54 ppm for No. 2 fuel oil.

NO, emissions as a function of combustion air swirl vary differently in gas and oil
flames with the RSFC burner. In contrast to the gas flames in which NO, emissions were
increasing as the swirl degree was raised above its critical value for vortex breakdown
(S ~ 0.6), they continued to decrease in both No. 2 and No. 6 oil flames for much higher
values of the combustion air swirL This difference is thought to be due to the lower
entrainment rate and higher penetration depth of the narrow angle fuel spray compared
to the gas jet As a result of this, radial stratification of the flow can be maintained to a
higher value of the swirl number in the oil flames than in the gas flames.

The lowest levels of NO, and CO emissions obtained by purely combustion
aerodynamic means are shown in Fig. 15. In recent experiments in which gas recirculation
and steam injection were used, significant reductions were reached in NO, emission levels
while maintaining low CO concentration at the exit of the combustion tunnel. These
results will be reported in our paper prepared for the 1991 Fall Meeting of the American
Flame Research Committee.

CONCLUSIONS

An experimental investigation has been carried out with a flexible experimental
low-NOj burner of novel design. The burner is designed to achieve staged combustion by
a combination of radial flow stratification and axial air staging in the flame. Fuel/air mixing
is suppressed by radial flow stratification close to the burner but is then promoted by a
toroidal recirculating flow further downstream of the burner.

Parametric experimental studies carried out in the flame tunnel of the MIT
Combustion Research Facility (CRF) permitted optimization of the burner for low-NOx and
CO emissions by determining favorable conditions for the radial distributions of the air flow
and the swirl velocity at the exit from the burner and for the central fuel injection velocity
and angle. The results showed that for several operational modes of the burner input
variables, highly stable flames with low-NO, and CO emission levels were attainable.
Minimum values of NOx and CO emissions obtained by the optimization of the

4A-99


-------
No.2 OIL

0.6 0.8 1 i"!i
SWIRL NUMBER

0 0.2 0.4 0.6 0.8 1 1.2 1.4 1.5 1.8
SWIRL NUMBER

EFFECT OF TYPE OF VORTEX ON NOx EMISSION

Figure 13.

4A-1QQ


-------
Figure 14. Effect of Type of Vortex Upon NOx Emission

4A-101


-------
No. 6 Oil

I	1 NOx WM CO

N.G.	No.2	No.6

NOx AND CO EMISSIONS FOR AN OPTIMUM
BURNER CONFIGURATION

Figure 15.

4A-102


-------
aerodynamic variables (staging by flame stratification) were 70 ppm NOx and 56 ppm CO
for natural gas, 54 ppm NOx and 23 ppm CO for No. 2 fuel oil and 97 ppm NO, and
32 ppm CO for No. 6 fuel oil.

The data reported here are initial results only, recent experiments presently under
evaluation show that further significant reductions of NO, emission can be obtained with
the new radially stratified flame core (RSFC) burner by means of the controlled admixing
of small amounts of recirculated flue gas and steam.

ACKNOWLEDGEMENTS

Financial support from a consortium of utility companies; Empire State Electric
Energy Research Corporation (ESEERCO), Eniricerche, Electric Power Research Institute
(EPRI), Ente Nazionale Per L'Energia Elettrica (ENEL), Florida Power and Light
Company and Southern California Edison Company is gratefully acknowledged. "We thank
ABB-CE for their expression of readiness in producing a commercial burner should our
experiments be successful. The authors are indebted to Ms. Bonnie Caputo for the
production of the many drafts and the final form of the paper.

REFERENCES

1.	Morita, S., Kiyama, K., limbo, T., Hodozuka, K-, and Mine, IC, "Design Methods for
Low-NOx Retrofits of Pulverized Coal Fired Utility Boilers", EPRI/EPA Joint
Symposium on Stationary Combustion NO„ Control, Mar, 1989.

2.	Lisauskas, R.A., Reicker, E.L., and Davis, T., "Status of NOx Control Technology at
Riley Stoker", EPRI/EPA Joint Symposium on Stationary Combustion NOx Control,
Mar, 1989.

3.	Thompson, R.E., Shiomoto. G.H., Shore, D.E., McDannel, M.D., and Eslanaa, D.,"
NOj Emissions Results for a Low-NO, PM Burner Retrofit" EPRI/EPA Joint
Symposium on Stationary Combustion NO^ Control, Mar, 1989.

4.	Sarofim, A.F., Pohl, and Taylor, B.R., "Strategies for Controlling Nitrogen Oxide
Emissions During Combustion of Nitrogen Bearing Fuels, "AIChE Symposium Series
No. 175. 74, 67, (1978).

5.	Farmayan, W.F., M.Sc. Thesis, The Control of Nitrogen Oxides Emission by Staged
Combustion," Department of Chemical Engineering, Massachusetts Institute of
Technology, Cambridge, MA, April 1980.

6.	Beer, J.M., Jacques, M.T., Farmayan, W.F., Gupta, A.K., Hanson, S., Rovesti, W.C.,
"Reduction of NO, and Solid Emissions by Staged Combustion of Coal Liquid Fuels."
Nineteenth Symposium on Coal Liquid, Haifa, Israel, Aug. 1983,

4A-103


-------
7.	Beer, J.M. British Patent No. 1099 959. Jan. 17, 1968; U.S. Patent; "Low NO, Rich-
Lean Combustor Especially Useful in Gas Turbines", Jul 11, 1989,

8.	Beer, J.M., Chigier, N.A., Davies, T.N. and Bassindale, K.: "Lammarization of
Turbulent Flames in Rotating Flow Environments" Combustion and Flame, Vol. 16.
pp 39.45 (1971).

4 A-104


-------
DEVELOPMENT OF LOW NOx GAS BURNERS

Shyh-Ching Yang, John H. Pohl, Steven J, Bortz, .
Robert J. Yang, Wen-Chen Chang

Energy & Resources Laboratories
Industrial Technology Research Institute
Hsin Chu, Taiwan, R.O.C.

W. J. Schafer Associates, Irvine, CA 92718
R-C Enviornmental Service & Technologies, Irvine CA 92718

4A-105


-------

-------
ABSTRACT

The Energy Commission, Ministry of Economic AffairfEC, MOEA),

Republic of China, has a program to develop 2.5MW low NO* gas
burners. This paper reports the results on premixed and
nonpremixed swirl burners.

The versatile premixed and nonpremixed swirl burners were designed,
fabricated and tested in Energy and Resources Laboratories'(ERL) 10
x 10^ Btu/hr furnace. It is shows that the NOx can be controlled to
levels of less than 15ppm, The peak flame temperatures required to
maintain required NOx levels were achieved by mixing sufficient flue
gas, and/or other diluents and well control the mixing rate.

The ERL's test furnace is designed to simulate single burner industrial
boilers. The furnace is 1600mm square inside, and is lined with
500mm thick refractory for the first 2500mm. The last 3840mm of the
furnace is water cooled. The furnace is built of 250mm segments and
operated with residence time at 2-3 sec and 5-10 mm water positive.

The results showed that dilution of the premixed gas by air or flue gas
could reduce NOx emission to 12.5ppm(dry, 3% 02) and is relatively
temperature insensitive for premixed burner. The results of non-
premixed swirl burner showed that turndowns of 5/1, with CO
emissions less than 50ppm, 02 1 - 2.5%, NOx emissions about
10-12ppm were achieved with the conditions of flue gas recirculation
of 15%, and primary zone stoichiometry of 0.7.

4A-107


-------
INTRODUCTION

Recently, international attention has increased to the role of nitrogen
oxides in lake acidification, oxidant formation and forest damage.
Stringent regulations to reduce the allowable emissions of nitrogen
oxides are being promulgated in many industrial area of the wotU.
Under the clean air regislations, the combustion industry is being faced
with the necessary of having to reduce nitrogen oxides from its existent
units. Combustion modifications(i.e. low-NOx burners, reburning or
staged combustion) generally afford the least capital for achieving
reductions(l).

This paper summerizes the results of a program funded by the Energy
Commission Ministry of Economic Affairs(EC,MOEA) Republic of China to
develope and to determine the ability of burner designs and operating
conditions to achieve the low levels required and expected for NOx
emissions.

To achieve this objectives, chemical reaction and aerodynamics control
of the flame were used for burner design (2,3,5,6,7). In this program,
four burner designs, premixed (1), partially premixed (4), nonpremixed
fuel jet and swirl burner (3), were tested in Energy and Resources
Laboratories' 10 x 106 Btu/hr furnace.

EXPERIMENTAL

The Energy and Resources Laboratories' test furnace is nominally rated
for 10x106 Btu/hr and is designed to simulate single burner industrial
boilers. A schematic of the furnace is shown in Figure 1. The furnace is
1600mm square inside, and is lined with 500 mm thick refractory for
the first 2500 mm. The last 3480 mm of the furnace is water cooled.

4A-108


-------
The furnace is built of 250 segments. Each segment has a rectangular
port for veiwing and probing the flame. The furnace is operated with
force draft at 5 - 10 mm water positive,

The furnace is equipped with a rear port for flame photography and
measuring the exit concentration of the flue gas. The wall are
instrumented with thermocouples to monitor the change in wall
temperature and to measure the heat extraction in the wall-cooled
sections. The furnace is also equipped with a number of sheathed
thermocouples on axis to monitor the relative flame temperatures.

Flue gas is drawn by an induced draft fan from the stack. The flue gas
is cooled in a direct spray tower, monitored by an orifices plate and fed
to the burner. The direct spray tower cools the flue gas tol50°C
(300°F). Use of the direct spray will increase the water content of the
heat capacity, and to a limited extent heat which can be extracted from
the flame by the flue gas.

Gases are drawn from the furnace through a water-cooled probe for
continuous analysis. The gases are analyzed for CO and €02 with non-
dispersive infrared analyzer, for NO and N02 with chemiluminescence
analyzer, and for oxygen with an electrochemical analyzer. A suction
pyrometer was used to measure the temperatures in the flame and
sheathed thermocouples were used to continuously ttack changes in the
lower temperature regions of the furnace.

The versatile premixed, partially premixed, nonpremixed fuel jet and
swirl flow burners were designed, fabricated and tested in E&RL's
furnace. Those configurations are shown in Figures 2 and 3.

• Premixed Burner (Figures 2 and 3a)

In this configuration nature gas is introduced from the bottom, is mixed

4A-109


-------
in a chamber with air. Flue gas, if used, is drawn by a fan from the exit
of the furnace, cooled to 150°C by a water spray, measured with an
orifice meter, and fed radially into the chamber. The rear nature gas
chamber is isolated from the air chamber by two layers of 1mm
opening screen to help prevent flash backs. The face of the burner is
covered with two layers of 1mm opening screens to prevent flash back
from the furnace. The burner is set within a cast refractory quarl. The
quarl is equipped with a pilot flame and a flame detector.

•	Partially Premixed Burner (Figures 2 and 3c)

In this configuration, nature gas is introduced from the bottom, is mixed
in chamber I with air. The mixed air and flue gas is fed to the chamber
II through a perforated shroud.

•	Nonpremixed Fuel Jet Burner (Figures 2 and 3b)

In this configuration, nature gas is introduced from the bottom. Flue
gas, if used, is mixed in chamber I with nature gas. The air is fed to the
chamber II through a perforated shroud.

• Swirl Burner (Figure 3d)

In this configuration, nature gas is introduced in the central gas gun.
The air is drawn through a swirl generator to generate a low pressure
drop, low turbulence swirling flow to achieve the desired ignition, flame
geometry and burnout characteristics for a given fuel. The flue gas
recirculation and air staging were used in this testing.

ppct TT tc
tvJDyULfl d

These versatile burners were tested in the E&RL's 10x10^ Btu/hr

4A-110


-------
furnace. Results from the tests are reported in this paper.

Measurements were made of gas composition (NOx, CO, CO2, and O2) and
flame temperature. The temperature at 37.5mm from the quarl was

measured using a suction pyrometer; other temperatures were
measured using sheathed termocouples. Data was taken under the
follow conditions:

• Premixed and Nonpremixed Fuel Jet Burners

-	premixed flame and diffusion flame.

-	load 3.9-6.6xl06 Btu/hr

-	flue gas 02 2.0-9.0%(dry)

-	flue gas recirculation 0-10%

-	mixing factor, M=1 for premixed burner

-	mixing factor, M=0 for nonpremixed fuel jet burner

The results of NOx emissions and flame temperatures are shown in
Figure 4. The emissions of NOx decreased with increasing or decreasing

oxygen in case of without flue gas recirculation. Where £ is the
percentage of the flue gas recirculation, % is the total stoichiometric
and M is the mixing factor. The mixing factor is defined as the ratio of
air mixed with fuel per total air introduced at a given

M= air mixed with fuel/total air introduced at a given

0)

For premixed burner, M=1 whereas for partially premixed burner,
tkMcl,

The influence of a "flame temperature" is also shown in Figure 4. The
"flame temperature" used is the value measured by the suction
pyrometer at 375mm from the quarl. The NOx emissions decreased
with decreasing "flame temperature". For load of 3.9x10^ Btu/hr, the

4A-111


-------
NOx emissions of 0,02 lb/MM Btu corresponds the flame temperature
of 1245°C. The NOx emissions increases to 0.07 Ib/MMBtu at "flame
temperature" increases to 1500<>C. The track of the NOx emissions and
"flame temperature" are similar.

The relative "flame temperature", © = T/TBAS, and relative NOx

emissions, = NOx/NOxBAS of % =1.10-1.45 and $ =0-10%

for premixed and nonpremixed fuel jet burners are shown in Figure 5.
The emissions of NOx and "flame temperature" decreased with
increasing the amount of flue gas recirculation. The relations of ^ and
0 versus £ are

n = 1- 0.075 $	(2)

and

0 = 1- 0.006 S	(3)

Where £ is in percent. These results give the agreement of the ^ and
0 approach to 1 as £ equal to zero.

• Partially Premixed Burner

-	premixed flame

-	load 3.9-6.6xl06Btu/hr

-	flue gas 02 1.5-4%(dry)

-	flue gas recirculation 0-10%

-	mixing factor; 0
-------
concentrations of a given flue gas recirculation rate. These graphical
solutions can tell us the phenomena of NOx emissions of the partially
premixed burner and the design and operation methodology for this
type of burners. For example, the conditions of 02=2.1%, % =4%;
02=3.1%, ^ =6%; and 02=6%, S =6% will give the same N0X emissions of
0.03 lb/MM Btu for the load of 3.9x106 Btu/hr. Figures 8 and 9 give
the comparison between prediction and experiment results for those
various burners. The agreement of these values at various operation
conditions is very well.

It is shown that these burners have been developed to meet the strict
emissions regulations, NOX<0.03 lb/MM Btu, mandated by the South
Coast Air Quality Management District(SCAQMD).

• Nonpremixed Swirl Flow burner

-	diffusion flame

-	load 5.9x10^-10.Ox 1 ()6 Btu/hr

-	flue gas 02 1,0-5.0%(dry)

-	flue gas recirculation 0-20%

The NOx emissions of the swirl burner with the load of 10.0 x 10&

Btu/hr and flue gas recirculation rate of 0-10% are shown in Figure 10.
The total stoichiometric, 4*r is 1.05-1.2. The emissions of NOx decreased
with decreasing oxygen and/or increasing the amount of flue gas
recirculated. The NOx emission reduction can be defined as

Re(%) = (1 - ^°x ) x 100%	(4)

KUZBAS

After regression these data, which gives the reduction of NOx emissions
versus flue gas recirculation rate is

Rj:(%) = !(%)/( 1.33 
-------
The NOx emissions at the load of 5.9x10® Btu/hr, without flue gas
recirculation , are shown in Figures 11 and 12. With extrapolation of

equation (5) at each recirculation rate, ^ 5-15%, it shows that the
agreement of prediction and experiment of NOx emissions is well. The
NOx emissions at the load of 6.6x10^ Btu/hr and 8.7x10^ Btu/hr with
air staging of primary stoichiometric 0.6-0.64 and with 4-5% flue gas
recirculation are shown in Figure 13. The total stoichiometric is 1.05-
1.3. The NOx emissions for unstaged are 0.03-0.048 lb/MMBtu. The air
staging and flue gas recirculation can provide a significant further
reduction of the NOx emissions. The minimum level of NOx which could
be achieved was 10-12 ppm .

CONCLUSIONS

Four versatile premixed, partially premixed, nonpremixed fuel jet and
nonpremixed swirl burners were designed, fabricated and tested in
Energy and Resources Laboratories' lOxlO^Btu/hr furnace. Results
show that 12.5ppm(3% O2) can be achieved either by diluting the flame
with air or flue gas to lower the maximum flame temperature for the
premixed flame or delay the mixing and to lower the flame temperature
for the diffusion flame.

This paper has mentioned the minimum NOx levels for various
premixed and diffusion flame burners. Some results provide the
concept of NOx emissions reduction methodologies. The agreement
between the prediction and experiment of NOx emissions for various
burner and operation conditions are very well.

Use of excess air and/or flue gas recirculation to reduce NOx will depend
on individual circumstances of each boiler. Increased of excess air in a
premixed burner will reduce NOx emission, will not require installation
of ducts and high temperature fans but will reduce boiler efficiency.
Use of flue gas recirculation and air staging of a swirl burner will reduce
NOx will require ducts and high temperature fans, but will not
drastically influence the boiler efficiency.

4A-114


-------
REFERENCES

1.	Pohl, J.H., S.-C. Yang, C.-H. Chen and R.Yang, "The Performance of Low
NOx Gas Burner Configurations: I. Premixed," October, 1990, American
Flame Research Committee International Symposium, NOx Control,

Waste Incinerators and Oxygen Enriched Combustion, San Francisco, CA,
U.S.A.

2.	Pohl, J.H..A.W. Bell, S.-C. Yang, C.-H. Chen, and R.Yang," Development of
a Full Sized Ultra Low NOx Industrial Burner," 1990,April presented at
the American Flame Research Committee ^^e mbe-r s Meeting, Tu scon,
Arizona U.S.A.

3.	Bortz, S.J. and S.-C. Yang,"Development of a Generalized Burner Design
Procedure" October, 1990, American Flame Research Committee
International Symposium, NOx control, Waste Incinerators and Oxygen
Enriched Combustion, San Francisco, CA, U.S.A.

4.	Pohl, J.H. S.-C. Yang, R. Chang and R. Yang, "The Influence of Burner
Geometry and Operation on NOx Emissions: II. Partially
Premixed",January, 1991 ,ASME Third Fossil Fuels Combustion
Symposium Houston, Texas, U.S.A.

5.	Yang S.-C., R.-S. Juang, W.-C. Chang, and J.-S. Chen, "Velocity
Measurements and Energy Distribution for Isothermal, suddenly-
Expanding, Swirling Flow in an Industrial Burner with Bluff-body",
Energy, H, NO.ll, pp.1015-1021,1990.

6.R.-S.	Juang, S.-C. Yang, W.-C. Chang, and J.-S.Chen, "Flow Characteristics
on Isothermal Sudden Expending Swirling Flow in an Industrial Burner
with Bluff Body,"(Accepted by J. of Chem. Eng. of Japan).

7.	Yang S.-C., et al., "Isothermal Swirling Flow in the Expanding Quarl of
Industrial Burner with a Bluff Body", November 1989,Proceeding of the
1989 International Gas Research Conference, Tokyo, Japan.

4A-115


-------
PREMIXED BURNER

>

CJ)

UJ
Cl

Figure 2 Versatile Natural Gas Burner

Figure 1 Energy and Resources Laboratories
lOxlO6 Btu/hr Test Furnace.


-------
PRE MIXED BURNER

FUEL

FLUE GAS

AIR

Figure 3a Versatile Premixed Gas Burner

NONPREMiXED - FUEL JET BURNER

JL

FUEL

FLUE GAS

AIR

Figure 3b Versatile Nonpremixed Fuel Jet Burner

4A-117


-------
PARTIALLY PREMIXED BURNER

FUEL

ltr

R ' AIR FLUE GAS

Figure 3c Versatile Partially Premixed Burner

N ON PREMIXED SWIRL BURNER

Figure 3d Versatile Nonpremixed Swirl Burner

4A-118


-------
008

0 07

006

C =0

£ •

CD 0.05

© 0.04

^ 0.03
m

S

2 0.02

•	O 3.9 XI06
' A A 6.6 X fO6

¦ O 3.9 XiO6

~	O6.6XJ06

1500

3.9Xi0%7U/HR'
6.6 X l(fiBTU/HR

1400

1300

0.01 -

1200

I 05

.15

1.25
0T

'.55

1,45

Figure 4 NOx Emissions and "Flame Temperatures'
of the Premixed and Fuel Jet Burners.

4A-119


-------
£ f %)

Figure 5 Relative N0X Emissions and Relative

"Flame Temperatures" Versus Flue Gas
Recirculation Rate

0.9S

096 ^
QQ
h~

0.94 \
0.92 11

090

4 A-120


-------
0 1 2 3 4 5 6 7 8 9 10

02 (%}

Figure 6 Prediction of NOx Emissions for Partially
Premixed Burner

4A-121


-------
0 I 234 567 89 10

02 (%)

Figure 7 Prediction of NOx Emissions for Partially
Premixed Burner

4A-122


-------

-------
\

0.09

0.08

0.07

0.06

g 0.05

I"—"

y

8 0.04

CC

0.

0.03

O
2

002

0.01



6.6X10® BTU/HR

02 1%) I ( %) M









_

0

, 2.5

0 I











¦

2.5

3.6 1











(1

2.4

5.1 . 1











A

3.4

6.6 1











~

4.4

0 1









-

O

7.4

0 1











•

9.0

0 1











<1

9.0

2.4 1















/Hk Jr



o2(%)

| (%)

M







d /

o

2,54

0

2

—





yo

•

1.80

0

2







y11

j

1.48

0

0.73

m



0/a

•

0

2.04

5.1

0.73







0

1.94



0.73

—

0

0 y

/ A



a

1.80

7.7

0.77



l l 1

I

I i



1

I



0 0.01 0.02 0,03 0.04 0.05 0.06 0.07

NOx (EXPERIMENT) ] ( LB/MMBTU)

0,08 0.09

Figure 9 Comparison Between the Prediction and
Experimental Results of NOx Emissions

4A-124


-------
I0.0 X106 BTU/HR , J3

Q03

\

5 0.02



X

o
2

0.0!

1.0

o

| = o %

•

5

~

10

R ^ (%) =(

N0X REDUCTION
, NOv I

NO

X.BAS

JXIOO%= ? /(I 330T-I I5J

J	L

1.05

.10

. J5

1.20

0T

Figure 10 NOx Emissions of Swirl Burner

4A-125


-------
. 0.04
0.03

3

h-

co
5

S 0.02

\

OD

-

x"

g 0,01

0

1.0	I.I	1.2	1.3	1.4

0T

Figure 11 N0X Emissions of Swirl Burner



5.9 XI06 BTU/HR

£-i



O i a o %

QO

-



/>///

-

w / / /
/ / /

_

o/ /



_



NOx , £ , PREDICTION
	 C = 5 %

	 j o

	 (5

_

/ NOX, £ ;
nox,bas

o.oi e /a.330j-us)



i I II

1 1 1 1

4A-126


-------
0.04

CD

\ 0.03
CD

O
h—
o

0,02

0
£

X

o 001

- 5.9X10®. BTU/HR . 02(%l

€ (%)





~

3.0

6.2





¦

1.9

62





A

4.0

11.7





_ A

. 4,0

1 1.8





o

3,0

ri.9





•

4.0
• /

i at y





_

/ °
/ "



0 (%)

i (%)

-

/*



O 2.6
• 2.5

16.0
16.!

/ 1

1 1 L

i

1 1

1

001

}x,i

0.0 2

0.03

0.04

N0V t (EXPERIMENT), (LB/MMBTUJ

Figure 12 Comparison Between Prediction and

Experimental Results of N0X Emissions

4A-127


-------
0.01

6.6XI0& BUT/HR
~ UNSTAGED

¦ PRIMARY ST0ICH=a60-0.64
HPR/MARYSTO(CH=0.60-0.64,4-5% PGR

8.7X10® BTU/HRI
O UNSTAGED

005

0.04

| 0.031

\
as

-j

-CO <=20 PPM





0

.05

1.15

1.2

1.25

1.3

0

T

Figure 13 N0X Emissions of Swirl Burner with Air
Staging and Flue Gas Recirculation

4A-128


-------
UNDERSTANDING THE GERMAN AND JAPANESE COAL-FIRED SCR EXPERIENCE

' /Phillip A, Lowe
INTECH Inc.
11316 Rouen Dr.
Potomac, MD 20854-3126

^ •, William Ellison
Ellison Consultants
4966 Tall Oaks Dr.
Monrovia, MD 21770

»	/ Michael Perlsweig

U.S. Department of Energy
Office of Fossil Energy
Washington, DC 20854

4B-1


-------

-------
ABSTRACT'

This paper examines the Japanese low sulfur and the German medium sulfur, coal-fired
selective catalytic reduction (SCR) costs and operating experiences.. Their
implications for high sulfur, U.S. coal-fired applications is also presented. It has
been observed that the costs are more strongly related to the application size than
they are to the location of the SCR catalyst section. However, operating costs and
issues are more strongly related to the location of the SCR system. The German Tail
Gas configuration technology, in which the SCR system is located downstream of the
flue gas desulfurization system, should be more easily transferred to high sulfur,
U.S., coal-fired applications, and if newer low temperature catalysts or less
expensive flue gas reheat designs are developed, it could become the configuration
of choice. Otherwise, site specific conditions, such as retrofit difficulty, will
probably dominate the selection process for applying High Dust or Tail Gas SCR
designs. A new issue not addressed in the German and Japanese SCR experience will
be how the spent catalysts are controlled, since they may be classified as a hazardous
waste in the U.S..

INTRODUCTION & BACKGROUND

In 1970, the Japanese initiated the use of SCR technology for NOx control on large,
electric utility boilers, including coal, oil, and gas service. The Japanese locate
their SCR reactor before the particulate collection equipment (High Dust
configuration), or when they have a high temperature particulate control system they
locate it after that equipment (Low Dust configuration). There were some initial
problems, principally with the formation of deposits on the catalysts and down stream
equipment. Studies showed that sulfur bisulfates were forming, and they were
depositing on the equipment or mixing with ash particles and the mixture was
depositing on the equipment. Considerable additional product development was
undertaken., and they eventually developed a long life, reliable SCR system for
reducing NOx emissions. Their approach included: reformulating the catalyst to reduce
its potential to form sulfur trioxide; during operations reducing the amount of
ammonia used by about 10-15% from the design specification and further controlling
the ammonia injection, if required, to assure that the ammonia leakage past the SCR
reactor (e.g., ammonia slip) is less than 10 ppm and preferably less than 5 ppm; only
treating about one-third of the boiler's uncontrolled NOx with the SCR system by using
combustion modifications to control the other two-thirds of the uncontrolled NOx; only
operating the SCR during steady plant operations (it can, however, be operated during
slow transients but it is not operated during start up or shut down); and operating
with low sulfur, low ash content coals (less than 1% sulfur and 10% ash). Because
of continuing problems with instrumentation, control of the process has been based
upon using calculated values of NOx and ammonia, and the measured values are
principally used as a trim signal for the control setting. Through 1990 they have
'installed 40 SCR systems on 10,852 MVe of coal-fired utility service.

In the early 1980s, the German utilities began an extensive pilot plant evaluation
of Japanese SCR designs for use at German electric utility power plants. Ultimately,

4B-3


-------
they conducted about 70 different pilot plant experimental studies. Since 1985, 129
SCR systems have been installed on 30,625 MWe of coal-fired utility service. Their
coal quality is quite different from the Japanese; they use low to medium sulfur and
ash coals (0.7 to 1.2% sulfur with low ash coals but on occasion up to 1.5% sulfur
and 25-30% ash, although such ballast coal is usually blended with the cleaner coal);
and their plant operations include intermediate duty (e.g., daily cycling and usually
shut down during the weekends). Also, the Germans retrofit their SCR systems to
existing plants, whereas the Japanese have both retrofit and new plant applications.
The Germans have both wet bottom firing (e.g., slagging boilers) and dry bottom
firing, whereas the Japanese use dry bottom firing. The principal German SCR design
is either the High Dust system or a Tail Gas system where the SCR is located after
the flue gas desulfurization (FGD! equipment. Initially, the Tail Gas configuration
was specified for wet bottom firing when the fly ash was recirculated to the furnace
section to be slagged (recycling produced high arsenic levels in the flue gas which
poisoned the catalyst). It is important to note that the Germans also do not operate
the SCR during start up or shut down periods (ammonia injection is typically keyed
to having the catalyst temperature at 554 'F or greater).

TECHNICAL IMPRCTS ON SYSTEM COSTS

Typically, the Japanese report SCR costs at $35-80/k¥tl1(, depending on the
application. The initial German experienceu,I,'! indicated that costs of $60-189/k¥
can be expected, depending on the boiler size and firing conditions, and the coal
used. U.S. cost estimates range from $80-100/kV for new plant installations!,,l,s|.
h key reason for the greater reported costs in the German literature is that they are
for retrofit applications at older, more congested plant sites, compared to the
Japanese and the new plant basis for the U.S. estimates. German studies prepared in
1985 and 1986 compared the expected costs for the High Dust and Tail Gas applications.
Figures 1 and 2 are typical of the German results reported'1,11. Often the capital
costs reported exclude the cost of the initial catalyst charge.

These figures and the results from the initial SCR installations (from 1985 through
1987) appear to be used by many authors to identify that the High Dust configuration
is less expensive than the Tail Gas configuration. However, if the entire data base
of applications through 1990 are considered, it appears that the Tail Gas
configuration is no more costly and may even be less expensive. High Dust
applications require long plant shutdown periods (during which the lost power must
be purchased from other sources! to allow for the connection of the SCR system to the
plant. On the other hand, the Tail Gas system requires a flue gas reheat system which
can consume as much as 3-4% of the fuel costs. Of course, costs are controlled, to
a very large degree, by the local site conditions, and thus either configuration could
be the least expensive at any - specific site. The High Dust system (in retrofit
applications) sometimes require penetrating the boiler primary pressure boundary to
remove or bypass part of the economizer so that the plant can continue to operate at
part power while maintaining the required SCR reactor inlet tempertures. This is an
expensive modification that is generally not identified in reported costs such as
provided in Figures 1 and 2. On the other hand, the Tail Gas systems requires a
second heat exchanger or auxiliary firing to bring the flue gas up to the SCR reactor
operating temperature (and to recover the excess heat before the flue gas is
discharged through the stack). The heat exchanger equipment can easily increase the
Tail Gas system's capital costs by 20%, which is significant.

In considering the capital costs, it should be recalled that Figures 1 and 2 are based
upon analysis assumptions and are not plots of data. The early capital cost estimates
w:ere important for establishing budgets, and thus cost assumptions that increased the
estimated cost were often used to conservatively predict the expected costs. In 1988
Jungm examined the reported costs for 19 German plants and reported that the total

4B-4


-------
capital expenditure had an arithmetic mean of about DM 66.3/kW,. More importantly,
his analysis showed that there was no statistical difference between wet and dry
bottom firing (and hence High Dust and Tail Gas SCR configuration)» but that the data
is represented by:

C = 0.358 x MW,"

where C is the total capital cost in millions of DM, expressed in January 1988

DM; and MW, is the power plant net thermal output in megawatts

The uncertainty factor needed to account for having a 90 percent confidence that all
of the actual cost data is represented by this expression can be expressed by applying
a multiplier on the coefficient 0.358:

C = 0.358 x (1.65 or 0.65) x HWte ,?1

where the multiplier 1.65 is used to define the upper bound of the cost data,

and the multiplier 0.65 is used to define the lower bound of the cost data.

Since the uncertainty can be accounted for in the coefficient, the 0.775 exponent can
be thought of as the factor that accounts for the effects of size or scale. The range
in the coefficient 0.358 (+65% to -35%) indicates that there is considerable
variability in the cost data base. This suggests that more detailed analyses of the
reasons for the cost differences would be useful.

However, the equation is important since it explains the large difference in the
reported cost/kV for High Dust and Tail Gas SCR applications as being primarily
related to the size of the system being retrofit. It is also important to note that
the inflation in Germany between 1985 and 1990 has been small enough that this cost
expression can be used to account for the DM in any year of convenience. Others11,11
have examined the reported capital costs and they report similar exponential
relationships between total costs and the size of the SCR application.

There is some distortion in the capital cost data that needs to be understood, and
it also needs to be recognized that the above expression does not account for those
distortions. First, the catalyst costs for the early SCR applications were 40,000-
60,000 DM/ir,'. By 1989 competition and the shrinking market for catalysts reduced the
costs to 20-25,000 BM/m1, and in one case the catalyst cost was reported at 17,000
DM/ir.'. Thus, as time went on the total capital costs were being reduced significantly
because of the reduced charge for the catalyst. To obtain a more realistic estimate
of the present cost for an SCR system, the data that was used to generate the above
equation should be normalized to assume the same unit costs for the catalyst.
However, that is not easy to accomplish because the German utilities do not always
identify the catalyst component of the total costs, nor do they use a standard chart,
of accounts. Thus, the reported capital costs can vary from plant to plant because
significantly different items are included in the total cost number. Also, regulatory
concern about the safety of anhydrous ammonia storage and preparation systems has
increased significantly during the application period. Now the regulators may require
remote location of the storage vessels, double wall vessels and piping, and increased
instrumentation and monitoring. These differences can increase the total plant costs
by an amount approximately equal to the changes accounted for in the cost decreases
that have occurred for the catalyst charge.

By 1987 the installed number of High Dust systems was about twice the number of Tail
Gas systems, suggesting that the market agreed with the interpretation that the High
Dust system was the least expensive. However, by the end of 1990 the number of Tail
Gas systems was slightly more than the number of High Dust systems'1". This suggests

4B-5


-------
that price (and technical) considerations have shown that the Tail Gas is at least
competitive if not less expensive. However, just as the cost data is not always
clear, there is other evidence that High Dust systems are preferred. For example,
the company Steag has reported that in 1989 the High Dust systems accounted lor 80%
of the dry bottom boiler applications and 16% of the wet bottom boiler applications
(for a total of 18,213 MW of capacity), while the Tail Gas accounted for the remaining
units {for a total 9,997 MW of capacity). Thus, in 1989 and 1990, more smaller units
applied the Tail Gas system. Since the cost/kv is greater for smaller units, on the
surface the raw cost data would imply that the Tail Gas is more expensive. Jung's
analysis, shows that this is not the case.

The earlier cost estimates assumed the same catalyst design for both the High Dust
and the Tail Gas configurations. Actual experience510 for honeycomb catalysts is that
the High Dust systems have used a catalyst pitch of 6-7.5 mm with a space velocity
of 2,000-3,000/hr. Tail Gas designs have used a 3.1-4.2 mm pitch with space
velocities of 4,000-6,500/hr., and they can also use a more reactive catalyst
formulation since the flue gas is cleaner than that which is treated by the High Dust
system. It has been observed in Germany and Japan that sticky, very small sized dust
particles can cause more serious catalyst deactivation problems than does arsenic"(see
below for a further discussion about arsenic). Also, the greater dust accumulation
on the equipment in High Dust systems has contributed to SCR fires in Japanese oil-
fired plants. The High Dust systems often employ a dummy leading edge to control
catalyst erosion. Theses physical conditions mean that a smaller catalyst and SCR
reactor can be specified for the Tail Gas system (it can be 50-60% smaller011),
reducing its system costs. Some of this impact is mitigated by the fact that the flue
gas saturation at the Tail Gas location means that the SCR reactor must process up
to 20% more flue gas volume than would a High Dust system on that plant, causing its
costs to increase (assuming that both designs operate at the same catalyst
temperature!.

Examination of the assumptions that were used to prepare Figure 2 also indicates that
some of them have biased the results to indicate that High Dust systems are less
expensive to operate. For example, the figure is based upon assuming the catalyst
life is 3 years for the High Dust system and 5 years for the Tail Gas system.
Experience indicates that 3-4 years (with occasional 5 years life) is consistent with
the High Dust catalyst design. However, some Tail Gas. system operators report that
they are not experiencing catalyst degradation, and that they expect the catalyst to
last for up to 80,000 operating hours (more than ten years). The Japanese on clean
flue gas, but not in coal service, have had catalysts last more than 10 years.
Catalyst life is an important item, others'" have reported that catalyst lifetime can
dominate the estimate of the levelized cost for SCR systems. Figure 2 was also
prepared with the Tail Gas system being charged for the cost of the flue gas reheat
system when a good part of that cost would have been incurred whether or not the SCS
system were present because some reheat is needed to add buoyancy to the flue gas
after FGD treatment. In reality, only the incremental costs beyond those required
for the FGD reheat system should be charged to the Tail Gas system. The figure also
assumes that all the reheat energy costs must be charged to the Tail Gas system costs.
The High Dust system requires that an economizer by-pass be installed to allow the
SCR system to operate when the plant is at part power. In the assumed 4,000
hours/year duty cycle that was used to generate Figure 2, the plant will be at part
power for a significant fraction of the operating time. Under those conditions bypass
of the economizer would lower the overall power plant efficiency by 2-4%, but that
cost impact was not included in the analysis used to generate the figure. The higher
assumed initial capital costs for the Tail Gas configuration require a larger
leveli2ed capital recovery factor for that design. All of these assumptions tend to
make the operating cost estimate for the High Dust system less expensive and the. Tail
Gas system more expensive.

4B-6


-------
Another factor that can be very important to account for is the fact that the retrofit
of a High Dust system' may be very difficult and more expensive due to the congested
available space, the need to reinforce the boiler building substructure and supports,,
the requirement to penetrate the boiler wall and to remove heating surfaces, and the
need to relocate soot blowers. Those costs and the plant down time (and the resulting
costs for replacement power while the plant is shut down) should be charged to the
High Dust system initial costs. On the other hand, the Tail Gas system can be sited
where it is convenient and the final hook up to the main plant can be performed more
quickly (typically in a few weeks). However, the more remote and accessible the Tail
Gas site, the potentially longer the flue duct runs and the higher would be their
corresponding costs. The greater dust loadings, the potential for fires involving
the SCR system (several fires were reported in Japan in 1989), the risk during plant
upset conditions for ammonium bisulfate deposits to form on downstream equipment, and
the need for washing the downstream equipment to remove deposits and then to control
and treat the wash water so it can be released to the environment all suggest a
greater down-side risk is associated with the High Dust system during the lifetime
of the plant. This risk should be reflected as a cost, but it is normally not
considered in the analyses used to prepare figures such as Figure 2.

The levelized cost assumptions do not include costs for control and disposal of the
spent catalysts. The shorter the catalyst lifetime the greater this problem could
be over the life of the plant, and the greater the expense associated with the
problem. In Germany and Japan it is assumed that the catalyst supplier will be
responsible for the spent catalyst, and in Germany the typical contract requires the
supplier to receive the spent catalyst. The experience in Japan (there has not yet
been enough experience in Germany with spent catalysts) is that it is not economical
to recover the catalyst materials, and they are simply disposed of by the catalyst
supplier. Thus, the assumption that the disposal cost are negligible is reasonable
for German and Japanese applications. The assumption needs to be checked for its
validity at local U.S. power plant sites.

U.S. cost estimates for new plants can also be used to help calibrate the reported
costs from German and Japanese plants. In 1984m the costs for a new 500 MV plant
were estimated at $70-80/kW. In 1989(5' similar applications were estimated at
SlOl/kW. The 1989 costs actually represent about a 40% reduction in the SCR system
costs, compared to the 1984 estimates, as is indicated when the 1984 costs are
escalated to the same basis as the reported 1989 costs'". However, the 1989 estimates
also include the 50% reduction in catalyst costs as reported in the German literature.
With retrofit cost factors added, the U.S. estimates compare favorably to the German
reported costs; they are, however, significantly greater than the reported Japanese
costs for new plants.

OPERATIONS AND MAINTENANCE

Another important lesson learned deals with the receipt, storage, use, and measurement
of ammonia. It has been established01,11,11® that if the SCR design is to provide a high
level of NOx control, extensive flue gas flow modeling and flow straightening are
needed to guide the design of the ammonia injection system and to assure that the flue
gas NOx-ammonia mixture is uniformly distributed across the catalyst cross sectional
area. The ammonia injection designs in Germany typically employ 30-40 injection
points per square meter of flow area. Each injection nozzle may have an individual
flow control (or each flow tube may have the controls if several nozzles are installed
on a single flow tube), so that the system can be optimized during the plant shakedown
period (optimization is the attempt to develop a uniform NH,/N0x ratio throughout the
entire inlet cross section of the SCR reactor). However, the use of many injection
points can lead to a false sense of security. It has been observed (see Figure 3)
in at least one design that used multiple injection' nozzles that dust and other

4B-7


-------
deposits collected inside the ammonia flow tube, indicating that conditions occurred
which caused the flue gas to enter some of the nozzles rather than for ammonia to exit
those nozzles. Since the resulting flow maldistribution built up slowly during
operations, it was not detected by the available monitoring instrumentation. But,
the hoped for NOx control that was supposed to be achieved by having a large number
of "tuned" ammonia injection nozzles was not achieved.

Much of the plant shake down and acceptance period is spent qualifying the valving,
motors, and switches in the ammonia distribution system (and in assuring that the
desired flow injection pattern is achieved). Typically, full flue gas velocity, NOx,
and temperature profiles are taken and a less extensive ammonia distribution profile
is taken. A 10% variation in the NH,/NOx ratio would be an excellent optimization,
a 25-30% variation would not be unusual. Often two measurement instruments are used
so that a average value can be determined.

The Germans initially used, in their High Dust applications, the Japanese criteria
of designing for an ammonia slip of less than 5 ppm, but during operations they have
found that they must limit the slip to less than 1 to 3 ppm'1,1 in order to provide fly
ash that can be used in other commercial processes. This also requires that the
ammonia injection be such that the ratio of NH,/N0x be less than 0.85 or the slip will
exceed the limit after a short operating period"1. The amount of acceptable ammonia
slip introduces a significant uncertainty in the life expectancy of the catalyst.
That is, later in its lifetime the catalyst may be able to provide for the design NOx
reduction with a 5 ppm ammonia slip, but not the needed 1 to 3 ppm slip. This item
remains to be evaluated, based upon actual, long term, operating experience. In the
Tail Gas configuration, because of the very clean nature of the gas, the ammonia slip
can be set by air emissions criteria, and slips as large as 20-30 ppm should not lead
to operational problems {ammonia odor and plumes are troublesome at 50 ppm or greater
ammonia concentrations).

As the German regulators dealt with additional SCR installations, they have become
more concerned about the safety implications of anhydrous ammonia, and increasing
design restrictions reflect those concerns. In Germany, anhydrous ammonia is
primarily delivered by rail, truck transport is prohibited except for volumes less
than 500 liters. The Germans generally store on site a 15 to 30 day supply of
'anhydrous ammonia. Often the ammonia is diluted to a mixture of 8% or less ammonia
before it is introduced into the flue gas. Normally, clean, cold, fresh air is the
diluent in order to help keep the ammonia nozzles clean. Two storage tanks are used
to allow for uninterrupted operation. They also increase the inherent safety margin
if a tank accident should occur. Many of the systems use double wall tanks and double
wall piping from the point of receipt to the exit from the ammonia vaporizer. Warm
water heating (compared to electrical heating used at some U.S. gas turbine
installations) is used to vaporize the ammonia. However, the safety requirements are
locally developed, and some installations use single wall systems, some facilities
bury their tanks, others were allowed to install above ground tanks. Increasingly
restrictive safety criteria have caused the ammonia system costs to increase almost
as much as the SCR catalyst costs have decreased1". Figure 4 is a presentation of the
purity specifications used for the purchase of anhydrous ammonia. The water and
oxygen are controlled such that the delivered ammonia is in region I of the figure.
Equipment corrosion is possible in region B, and corrosion will occur if the oxygen
and water content are such that the ammonia is in region C. However, corrosion by
water alone is unknown, and the storage of hydrous ammonia with 25% or greater water
content is common.

Actual plant operations in Germany have shown that some of the catalysts can store
ammonia. The storage has no effect during steady state operation, but this ammonia
inventory is released from or added to the catalyst during transients such as load

4B-8


-------
changes or shut downs. This phenomenon must be recognized in the design of the
ammonia injection control system, as the desorption process has been found, in at
least one plant, to take up to eight hours. For example, if the ammonia injection
control does not account for desorption during shut down, excess ammonia will be in
the flue gas during low temperature operations, causing too high an ammonia slip and
possibly forming and depositing bisulfate salts. Since ammonia desorption/absorption
can produce a 30 minute lag in the NOx control111, the selection of the averaging time
for meeting the NOx emission control requirement is very important for establishing
if the plant operation can remain within the permitting requirements. Since bisulfate
formation is not controlled by the ammonia slip as much as it is by the SO,
concentration, the lack of a continuous ammonia monitor and control by ammonia slip
levels is not serious to the overall satisfactory operation of the SCR system.

Other design areas that were not addressed during the initial German installations
but which have been indicated by operating experience are: 1) sulfur trioxide and acid
attack of the duct liners downstream of the SCR has occurred in some cases, indicating
that the oxidation potential of the catalyst (for converting sulfur dioxide to
trioxide) will be an important consideration when high sulfur coal is fired. 2) the
gas to gas heat exchanger for reheating the flue gas for the Tail Gas system has been
required to be redesigned to make it less complicated and to produce less leakage of
the untreated gas into the treated gas. Leakages as high as 7% are reported, but 1.5-
3% appears to be more common. It is possible to design for zero leakage, and some
plants have such heat exchangers. This suggests that the use of rotary heat
exchangers should be replaced with the use of non fluid mixing heat exchangers for
reheating the flue gas to the Tail Gas system operating conditions. 3) control
problems under load swing conditions have indicated that the ammonia instrumentation
remains an issue as well as the impact of ammonia absorption/desorption from the
catalysts. This could be even a more pronounced problem in high sulfur coal service.

The initial reason for developing the Tail Gas configuration was to overcome the
catalyst poisoning that was experienced in a few pilot plants that serviced wet bottom
boilers that also recirculated the fly ash back to the boiler to slag the ash. That
operation was found to increase the arsenic concentration in the flue gas by 10 to
100 times from the level found when no fly ash recirculation was used. That
information was initially used to conclude that the Tail Gas design was required when
wet bottom firing with full flue gas recirculation was used at the power plant.
Gutberlet"" has since examined the arsenic concentration in 14 wet bottom plants that
had varying amounts of fly ash recirculation. Figure 5U,) presents data from the 14
separate boilers. The upper plot presents the relative amount of arsenic in the flue
gas before the air preheater. The lower plot shows the relative amount of arsenic
in the coal being fired. The data was plotted so that the amount of arsenic in the
flue gas would increase as the data is viewed from left to right. The figure clearly
shows that arsenic content in the coal and operation with 100% fly ash recirculation
back to the boiler can cause both high or low arsenic concentrations in the flue gas
(Gutberlet did not identify the actual arsenic concentrations). Thus, fly ash
recirculation by itself is not a sufficient parameter to determine if arsenic
poisoning would be a problem for a High Dust SCR application. Gutberlet concluded
that the composition of the fly ash itself was a significant factor, and that the
greater the amount of calcium oxide in the fly ash the less arsenic would be found
in gaseous form in the flue gas. He recommended that the calcium oxide in the fly
ash be at least 3% and preferably greater than 5%; that fly ash recirculation be
restricted in general if possible and especially during boiler soot blowing; and that
coals with low arsenic content be used (most German coals have arsenic levels of 5-25
rag/kg, and the term "low arsenic content" was not defined or compared to those
levels). Separate Japanese51" studies (the coal arsenic levels were not reported)
have shown that calcium content in the fly ash is a primary factor for causing
catalyst deactivation in High Dust SCR configurations. This further supports the

4B-9


-------
German data that fine dust can be a greater catalyst "poison" than is arsenic. The
Japanese data is for dry bottom firing. The Japanese experience is that if the
calcium oxide content is less than 1% the catalyst will not have to be replaced even
after 38,000 hours of operation. Five to eight percent calcium oxide in the fly ash
can cause a comparable catalyst deactivation in 20,000 hours, and catalyst replacement
will be required before 25,000 hours. They also found that high sulfur coals that
produce a higher ratio of gypsum in the calcium compounds in the fly ash are less
likely to cause rapid catalyst deterioration. Thus, it appears that if there is a.
species such as arsenic in the flue gas, it can react with the calcium and is thus
removed as an active poison. However,- if there is no species available to-react with
the calcium, the calcium itself can blind the catalyst. The desirability of some
calcium content in the flue gas needs to be carefully studied for U.S. high sulfur
coal applications. Such a study will also have to evaluate the type of FGD design
used if Tail Gas SCR systems are being considered.

IMPLICATIONS FOR U.S. SCR OPERATIONS

Extensive overseas use of SCR has established design and cost criteria as well as
credibility for the use of this technology in low sulfur fuel applications, including
coal, oil, or gas. An important consideration in this commercial success for' High
Dust applications has been the recognition that the ammonia:sulfur trioxide reaction
to form ammonium bisulfite and bisulfate can be controlled by limiting the ammonia
slip.

Low Sulfur Coal

The large U.S. population of low sulfur, coal-fired, electric utility boilers,
primarily those firing western subbituminous coals, should be able to readily utilize
the commercial design and operating retrofit experience from Japan and Germany to
achieve NOx emission levels, when and if required, as low as 100 ppm or 0.17 lb
N0,/million Btu. The German operating experience is more important for such U.S.
applications because many of the U.S. installations are on peaking utility boilers,
whereas Japan coal-fired service has been essentially all in more simple, base load
boiler cases. An appropriate SCR system design strategy based on German practice on
low sulfur, coal-fired, dry bottom boilers would call for the retrofit of combustion
modifications including low NOx burners to reduce the gross emission to the range of
325 to 400 ppsi"". This greatly mitigates the overall cost and ammonia slip problems,
since the resulting SCR removal efficiency requirement is no more than 70 to 75% to
achieve a 100 ppm stack emission. Such an approach, if it achieved only half of the
four million annual ton of NOx emission inventory reduction in the U.S., which is
implicit in Title IV of the new Clean Air Act, would result in nearly 100,00 MW of
retrofit SCR system capacity. That would be more than the entire present worldwide
population of existing SCR facilities.

Key design premises for this coal-fired SCR service are tied to the use of a design
ammonia slip of 3-5 ppm. The suppliers and users in Japan"" have identified them as;

•	Vertical downward gas flow reactors to prevent ash accumulation.

•	' Linear gas velocities of 16-20 ft/sec (5-6 n/s) at maximum continuous

rating to prevent ash accumulation and erosion.

•	Use of a grid-shaped catalyst with a channel spacing (pitch) of 0.275-0.3
inches (7-7.5 mo) to prevent ash accumulation and erosion.

•	Catalyst layers formed without seams along the gas flow direction

4B-10


-------
(including optional use of a sacrificial initial stage) to prevent ash
accumulation and erosion.

•	Ash deposition removal by intermittent vacuuming or soot blowing.

Technical literature issued by Japanese SCR system suppliers emphasizes the following
key elements for a successful SCR installation on peaking utility boilers:

•	Reliable ammonia feed control.

•	Adequate ammonia feed distribution across the cross section gas flow
area.

•	Flue gas duct designs that ensure good nixing of flue gas and arononia
feed.

•	Provision for suitable control of ammonia feed at part load.

In Germany, important new SCR experience has been gained on slagging (wet bottom)
boilers, which produce different flue gas and fly ash characteristics that can impact
the SCR catalyst1"1. Full-scale experience since 1986 in the use of the High Dust SCR
designs for some wet bottom, boilers has confirmed excessive impacts on the catalyst
life and has lead to their reworking approximately 3,000 HW of such retrofits to
convert them to Tail Gas designs. It is clear that such boilers are best served by
the Tail Gas design.

High Sulfur Coal

The potential impacts of the much higher sulfur dioxide and especially the higher
sulfur trioxide concentrations in the flue gas fro® U.S. high sulfur, coal-fired
plants will be an item of special concern. It may cause a new optimum of catalyst
reactivity, linear velocity, ammonia slip, and operating temperature window to be
established. Since much of the impacts of ammonia absorption/desorption, sulfate and
bisulfate formation, poisoning of the catalyst, and blockage .or blinding of the
catalyst are surface effects, it is not possible to scale the existing German and
Japanese High Dust SCR results with confidence without the benefit of prior testing
at the expected operating conditions. Although the Tail Gas results can be scaled
with somewhat more confidence, questions about calcium poisoning or trace element
carry over from the FGD system indicates that these designs also need to be tested
prior to final design selection.

Patterning its work after the major pilot plant test program in Germany, EPJtl is
carrying forward a $15 million collaborative bench and pilot-scale research program
to include definition of costs and technical feasibility for the use of SCI in
domestic medium and high sulfur coal service"". Testing will be conducted at as many
as 14 separate facilities over a four year period to assess SCR process design,
catalyst life, instrumentation and controls, and plant integration.

Application of the Tail Gas system design is an important and possibly vital approach
for using SCI technology in high sulfur coal service. Such process applications would
avoid the potentially excessive rate of air preheater fouling and catalyst degradation
from high sulfur service, High Dust SCR design, even with an ammonia slip value as
low as 1 ppm.

Cyclone-fired wet bottom boilers, for high sulfur service and with their typical NOx
emissions of 0.8 to 1.8 lb N0S million Btu (500 to 1,100 ppm, represent a major market

4B-11


-------
for the Tail Gas SCI design. There are presently 105 operating cyclone units- They
represent about 14% of the pre-NSPS coal-fired generating capacity (over 26,000 MV).
However, these units contribute about 21% of the NOx emitted by pre-NSPS units because
their combustion design is conducive to NOx formation, and their firing design is not
conducive for low NOx burner technology. Furthermore, other conventional NOx
reduction techniques such as two-stage combustion cannot be applied to the full extent
due to associated operational concerns with corrosion. Although many of the units
are 20 to 30 years of age, many of the utility owners plan to operate them for an
additional 10 to 20 years. Since the majority of these units are located in the acid
rain emission control area (the Midwest!, cyclone boilers may be especially
appropriate for NOx control. "With the'well established use of Tail Gas SCI systems
on wet bottom boilers in Germany, this class of boilers appears to be a key target
market sector for Tail Gas SCI designs.

REFERENCES

1.	NOx Task Force, Economic Commission for Europe, Technologies for Controlling
NOx Emissions from Stationary Sources. July 1986.

2.	"Development ft Application of NOx-Flue Gas Treatment in the Federal Republic
of Germany." NATO/CCMS Meeting, Control of Air Pollution From Combustion
Systems, Duesseldorf, October 1988.

3.	B, Schaerer, N. Haug, and J~H. Oels. "Cost of Retrofitting Denitrification."
Proceedings of the Workshop on Emission Control Cost, Esslinger am Neckar, FGR,
1987,

4.	"Selective Catalytic Reduction for Coal Fired Power Plants." IPRI CS-3603,
October 1984.

5.	C. P. Robie, P. A. Ireland, and J. E. Cichanowicz. "Technical Feasibility and
Economics -of SCR NOx Control In Utility Applications." Proceedings 1989
Symposium on Stationary Combustion Nitrogen Oxide Control. EPRI GS-6423, July
1989.

6.	P, A. Lowe. Selective Catalytic Technology. Burns and Roe Service Corporation
report submitted to the U.S. Department of Energy, December 1989.

7.	J. Jung. "Capital Expenditures in S02- and NOx Reduction in the German
Electricity Industry." VGB Kraftwerkstechnik, Vol. 68, No. 2, February 1988.

8.	P. A. Lowe, and M. Perlsweig. "Recent Experiences for SCR Systems at Coal-Fired
Utility Boilers." Proceedings of the American Power Conference, March 1990.

9.	J. E. Cichanowicz, and G. R. Offen, "Applicability of European SCR Experience
to U.S. Utility Operation," 1987 Symposium on Stationary Combustion Nitrogen
Oxide Control, EPRI CS-5361, August 1987.

10.	P. Necker. "Operating Experience with the SCR DeNOx Plant in Unit 5 of
Altbach/Deizisau Power Station." 1987 Joint Symposium on Stationary Combustion
NOx Control, EPRI CS-5361, August 1987.

11.	M. Novak, and H. G. lych. "Design & Operation of the SCR-Type NOx-Reduction
Plants at the Duernohr Power Station in Austria." 1989 Symposium on Stationary
Combustion Nitrogen Oxide Control, EPRI GS-6423, July 1989.

4B-12


-------
12.	J. Arido., "NOx Abatement for Stationary Sources in Japan," EPA-600/7-83-027
(PB83-207639), May 1983.

13.	P. Necker. "Experience Gained by Neckarwerke From Operation of SCR DeNOx
Units." 1989 Symposium on Stationary Combustion NOx Control, EFRI GS-6423, July
1989.

14.	K. Staebler, et. al.. "Secondary Measures for NOx Reduction Experience with
Pilot- and Commercial-Scale Plants." VGB Kraftwerkstechnik, 68, No. 7, July
1988.

15.	H. Gutberlet, "Influence of Furnace Type on Poisoning of DENOX Catalysts by
Arsenic." VGB Kraftwerksteehnik, Vol 68,- No 3, March 1988.

16.	S. Nagayama, et. al.. "SCI Application for NOx Control in Coal Fired Utility
Boilers." Proceedings of the 7th Pittsburgh Coal Conference, September 1990.

17.	A. Kinoshita. "Trends in Environmental Control Costs for Coal-Fired Power
Plants in Japan." Proceedings of the 7th Pittsburgh Coal Conference, September
1990.

18.	M. Hildebrand. "Status of the Art of Flue-Gas Purification at E.S.C. Power
Stations S02/NOx-Reduction (in German)." Elektrizitaetswirtschaft, Jg. 89, Heft
9, 1990.

19.	K. Leikert, H. Reidick, and H. Schuster. "Low NOx-Emission Combustion of Fossil
Fuels in the Federal Republic of Germany." Proceedings of the Fourth Seminar
on the Control of Sulphur and Nitrogen Oxides From Stationary Sources, Graz,
Austria, 1986.

20.	P. A. Lowe. "Utility Operating Experience with Selective Catalytic Reduction
of Flue Gas NOx.", Proceedings of the Second International Conference on Acid
Rain, Washington, D.C., 1985.

21.	G. R. Offen, et. al.. "Stationary Combustion NOx Control." Journal of the Air
Pollution Control Association, July 1987.

22.	Electric Power Research Institute, "Selective Catalytic Reduction for NOx
Control, A Proposed Collaborative Bench- and Pilot-Scale Research Program,"
April 1988.

4B-13


-------
Too 200 300 400 500 IHo 700 *
CRPRCITY [MVt1 J

FIGURE 1. GERMAN SCR CAPITAL COSTS VERSUS CAPACITY

1.5-

N
X
Q

Q

Ci

*-«

tn
Ca
Ci

0. S-

4000 A/a OPERRTION TIME RT FULL LORD
3 - 5 TERRS CRTRLYST LIFE EXPECTRNCT
HO'HIGH-OUST TC=TRIL-CRS

CURL-FIRED

0 WET 8BTTSM HO
B MET BOTTOM TC
807. NO* REMOVAL

0 OPT BOTTOM HO
5 ORT BOTTOM TO

100 200 300 400 500

cflpficnr cm*, j j

600

700

FIGURE 2. GERMAN SCR OPERATING COSTS VERSUS CAPACITY

4B-14


-------
FIGURE 3. DUST DEPOSITS IN AN AMMONIA FLOW TUBE

4B-15


-------
arsenic in flue gas

I. 0

# 100% ash recirculation
© partial ash recirculation	•

©

° bo ash recirculation	0

o.i

o

o.ol

0.001

1,0
0.5

e

o • • 9	o

0

0

-L .	— I .

ARSENIC IN COAL (dry)

o

o

O	o	o

c

O	A 6	* ® O	*

1 2 3 4 5 6 7 8 9 10 n 12 13 14

TEST NUMBER

FIGURE 5. RELATIVE CONCENTRATION OF ARSENIC
THE FLUE GAS AND IN THE COAL

4B-16


-------
OPERATING EXPERIENCE WITH TAIL-END AND HIGH-DUST
DFNOX-TECHNICS AT THE POWER PLANT OF HE ILBRONN

Dr. H. Maier
Energie-Versorgung Schwaben AG
KriegsbergstraBe 32
7000 Stuttgart 10
Germany

P. Oahl

Energie-Versorgung Schwaben AG
Kriegsbergstrafle 32
7000 Stuttgart 10
Germany

4B-17


-------

-------
ABSTRACT

At the Heilbronn power station two different SCR-DENOX technics are installed.
The high dust SCR of the unit Heilbronn 7 has been started up in September 1986 and
operates till now appr. 25.000 hours.

It is a two line arrangement, each of it denitrificating 50 % of the total flue gas
amount.

The loss of catalysts activity and the ammonia slip have been measured in dependence
of operating hours and the results compared with expected or theoretical calcula-
tions.

Decrease of catalysts activity, fly ash plugging, irregular distributions in the SCR-
reactor (velocity, temperature, NH3/N0t-ratio) and side reactions {acid particles
formation and emission, ammonia sulfates) have been investigated in detail.

In the units Heilbronn 3-6 the tail-end configuration is installed.

The flue gases of 4 slag tap boilers are desulfurized downstream the electrostatic
precipitators in one FGD plant (wet limestone process yielding gypsum as byproduct)
with a capacity of approximately 1.800.000 m3/h. The tail-end SCR denitrificates the
flue gases, which are free of dust and desulfurized, with two lines, each of it
treating 50 % of the total flue gas amount.

The first line started up in the middle of 1988 and operates till now appr. 14.000
hours, the second one in October 1990. In contrast to high dust DENOX-plants, which
had been proofen in Japan for several years, there was no experience with tail-end
arrangements.	<

Therefore different start up times resulted from the demand, to research and optimize
such a configuration at the first line.

4B-19


-------
1.	INTRODUCTION

At Heilbronn the EVS operates following combined heat and power units;

•	Unit 3-6, with four bituminous coal fired boilers and a slag tap

firing process.

Technical parameters as shown in fig. 1. (1st slide)

•	Unit 7, with one bituminous coal fired boiler and dry firing

process.

Technical parameters can be seen in fig. 2 (2M slide).

In the Federal Republic of Germany the NQx-emissions from bituminous coal fired
boilers with more than 300 MW thermal output are limited to less than 200 mg/m3.

This usually needs post combustion flue gas cleaning technics. A well known and
widely used technology is the selective catalytic reduction or SCR-process, where
ami?,onia reduces with the aid of catalysts nitrogen oxides to nitrogen and water at
temperatures between 300 °C and 400 °C. (Fig. 3., 3ra slide).

The SCR-process can be installed between boiler outlet and air preheater (the so
called high dust arrangement) or downstream of the electrostatic precipitator and
the flue gas desulphurization plant (the so called tail end configuration).
At Heilbronn power station both possibilities are realized.

2.	THE HIGH DUST DEN0X OF HEILBRONN 7
2.1 General Informations

During start of erection (1982) and trial run (end of 1985) environmental protec-
tion laws have been actualised several times, of course each time they changed to
lower limit values for S0X- and N0x-envissions. Until 1982 the KO^-reduction by post

4B-20


-------
combustion flue gas technics has not been a severely discussed, feasible solution.
But in January 1984 the NOx-emission limit for new coal fired plants was reduced
from 800 mg/nr to 200 mg/m3. So, during the phase of erection of Hei1bronn 7 we were
forced to realize a SCR-piant as soon as possible. The erection of the DENOX-plant
started in the last quarter of 1985. One year later, in September 1986 the SCR was
put into operation.

Fig. 4 (4th slide) shows the high dust arrangement of the SCR-plant at Heilbrortn 7.
2.2 Operating Experience

Since beginning of operation the loss of catalysts activity and resulting ammonia
slip is measured periodically and compared with expected respectively theoretical
calculations.

Fig. 5 (5th slide) shows, that after 12.000 operating hours the status of the SCR-
plant was much better than expected (and guaranteed). But, as one can see in fig. 5
too, during the first months of 1990 (after appr. 18.000, operating hours) we
recognized increasing amounts of ammonia in the fly ash (up to 100 ppm) and
consequently we got problems with our waste water, where the ammonia concentration
is limited to 10 mg/1. Therefore we were forced to set measures immediately:

•	ordering an access layer of catalysts

•	manually cleaning of the DENQX plant
t measuring of catalysts activity

•	measuring the ammonia slip before and after cleaning.

The results, that partially can be recognized again in fig. 5, were:

•	The relative activity K./K, of the catalysts was 0.77, after 18.500
operating hours a really sufficient result

•	the cleaning, of the plant decreased the ammonia slip from appr.

2 vpm to 0.5 vpm

•	optically we estimated a loss of active surface of appr. 22 %,
caused by fly ash plugging (fig. 6, 6th slide).

These results caused (beside manual cleaning) further steps, to bring the plant
back to optimal operation conditions.

4B-21


-------
Fig. 7 (7th slide) shows, how the need of access catalyst volume or, at the other
hand, the loss of volume needed for good operation, is influenced by irregular
distributions of

•	NI^/NO,-ratio

•	temperature

•	velocity

It can be seen clearly, that optimizing the NH3/N0,-distribution is most effective.
Therefore, measures to be done are;

•	control of the NH3/N0x-distribution and, if necessary, optimize the
NHj-feed ing

•	we recommend generally and ordered for Heilbronn 7 sootblowers
between all catalyst layers, including the dummy, to keep the plant
free from surface losses caused by fly-ash plugging. The sootblowers
will be installed this summer

•	to keep further troubles far from us during that time, we replaced
the first layer by a new one.

In fig. 8 (8" slide) we tried to estimate the consequences of replacing one
catalyst layer in dependence of fly-ash plugging. You can see clearly, how impor-
tant the sootblowers are regarding the life time.

2.3 Side Reactions

The most important and well known side reaction in high-dust plants is the cataly-
tic oxidation of S02 to S03 and, depending on flue gas humidity and temperature, the
formation of sulfuric acid. This can cause troubles in two ways:

The Formation of Ammonia Sulfates like (NH,)? SO, or NH.HSQ,. These compounds can
cause pluggings on the gas-gas-reheater, therefore needs washing of it and creates
ammonia-loaded waste water.

According to the very low ammonia slip, which we try to run at Hei1bronn 7 because
of other reasons (saling of high-quality fly-ash), this is no problem till now. The
ammonia concentration behind SCR is not sufficient to form ammonia sulfates at
amounts, which could hurt.

4B-22


-------
Formation arid Emission of Acid Particles. Sulfuric acid condensates at the cold
raw-gas side of the gas-gas-reheater and is transported to the clean-gas side.
There an acid film forms at the walls of the flue gas ducts and the stack and is
adsorbed by fly ash or gypsum particles. That means, particles are created-with an
access of free sulfuric acid. Especially during starts after a few days stop the.
acid particles are solved from the walls by thermal expansion, transported to the •
stack and, after reaching full load and corresponding gas velocity, emitted within
a few minutes.

This phenomenon causes severe troubles if attacking the dope of cars nearby, the
power plant.

To fight these emissions, we have the possibility

•	to reduce the efficiency of the ESP and increase the amount of
neutralizing fly ash (this we are doing at the moment)

•	to set constructive measures in the clean gas duct like especially
precipitators (we are testing this)

•	in future, when catalysts have to be renewed," to use types with low
conversion efficiency.

2.4 Conclusions

After appr. 25.000 hours of operating the SCR-Plant at Heilbronn 7 we made follo-
wing experiences:

•	The decrease of catalysts activity is far from expected values and
much better than assumed.

•	Fly ash plugging can cause severe problems, so we recommend to
install sootblowers between each layer.

•	Irregular distributions of velocity, temperature and NH3/N0x-ratio
can not be avoided. The most effective measure at relatively low
costs is to optimize the distribution of NH3/NG„, i. e. optimizing
NHj-injection under respect of not avoidable irregular distributi-
ons.

•	Formation and emission of sulfuric acid and acid particles makes a
lot of troubles especially at plants with much starts and shutdowns.
Measures to solve this problem are possible and we try to realize
them at the moment.

4B-23


-------
• Formation of ammonia sulfates and consequences out of this is no
problem. This follows from the very good activity-slope and conse-
quently low NH3-slip.

3. THE TAIL-END DENOX OF HEILBRONN 3-6

3.1	General Informations

The start' up of these elder units at Heilbronn was between 1958 and 1966. As
mentioned before, also for these plants the NO,-emissions were limited to 200 ng/m3
in the middle of the 1980's.

That means again, that additional to primary measures post combustion technics had
to be installed.

But in contrast to the situation at unit 7, the well proofen high dust technic is
often not feasible for slag tap boilers.

Problems arise due to interference with the compact design of elder plants and to
relatively high content of catalyst poisening gaseous trace elements (like arsenic)
in the flue gas.

These were the reasons, why we decided to erect a so called tail-end SCR plant
behind electrostatic precipitators and flue gas desulphurization. This configura-
tion is shown in figure 9 (9th slide). ,

Because of no experience with such an arrangement, we realized this project within
three steps. The first was a pilot plant, which is not in discussion here. Just let
me say, that the principal feasabi1ity could be proofed.'

The second step was the commercial scale tail end plant for one half of the total
flue gas amount of the units 3, - 6, appr. 900.000 m3/h. This first line started up
in the middle of 1988 and operates till now more than 14.000 hours. An intensive
research project was running with this line from January until December 1990. The
first two steps have been partially sponsored by the EEC. The third step conse-
quently was DENOXing the second half of the flue gases of unit 3-6. This second
line started its operation in December 1990 with the trial run.

3.2	Operating Experience

Catalysts activity has been measured after 3.700 and 6.500 hours of operation with
the result of no activity loss. At the moment we expect the results after 13.000
hours. The ammonia slip measured at the DENOX-plant additionally after those times
was below 1 ppm.

4B-24


-------
Because of the very low dust- arid S02-concentration with the tail-end configuration,
the design of the catalysts is different to that of the high dust arrangement, as
one can see in figure 10 (10lf slide). Smaller pitches cause in spite of the low
dust concentrations some pluggings, so we are cleaning the plant from time to time
manually and we think about the installation of sootblowers.

In contrast to the high dust plant the flue gas temperatures in front of the tail-
end reactor is appr. 50 °C (behind FGD). Therefore we have to operate a reheating
system (9" slide), which is a combination of regenerative gas/gas preheater (GAVO)
and gas burner. The last one we need to compensate the hot side temperature
approach of the GAVO, which is below 30 °C.

Up to now the GAVO operates very well. The total leakage is about 3.7 % at full
load, that means an increase of the NO, clean gas concentration of appr. 25 mg/m3
caused by the GAVO. To prevent an increase of differential pressure according to
GAVO plugging, we clean it once per shift using sootblowers, but we don't wash the
GAVO.

3.3	Side Reactions

In principal we have to discuss the same side reactions as before, but obviously it
is expected to cause less problems because

•	the S02-concentration behind FGD is very low

t the conversion rate depends on the temperature (it decreases with
decreasing temperature rapidly) and we have significantly lower
temperatures with the tail-end configuration (290 °C - 320 °C)

•	also the ammonia slip is far from dangerous concentrations.

And indeed till now there are no problems with side reactions at the tail-end
plant.

3.4	Conclusions

After appr. 14.000 hours of operating the first line of the tail-end DEN0X plant at
Heilbronn 3-6, and after three months with the second line we made the experi-
ence, that this configuration is a suitable one for slag tap fired boilers:

4B-25


-------
The operating characteristics are shown in figure 11 (11" slide).

Again the decrease of catalysts activity is far from expected
values.

Fly ash plugging can be handled by manual cleaning, but sootblowers
should be more effective.

The danger of formation and emission of acid particles is less than
at the high dust plant and we don't expect troubles from this side
in the future.

Catalyst poisening caused by gaseous trace elements can be avoided,
if using a tail-end arrangement.

4B-26


-------
Flue gas flow rate Nm3/h	1,8 x 106

Output	MWnet	426

NOx-concentration

-	Inlet of SCR	mg/m3	< 1.000

-	outlet of SCR	mg/m3	< 200

NO-removal	%	> 80

SOx-concentratlon

-	inlet FGD	mg/m3	1.600 - 3.400

-	outlet FGD	mg/m3	< 160 - 340

SOx-removal	%	> 90

Dust concentration

-	Inlet electrostatic mg/m3	6.000
precipitator

-	outlet electrostatic mg/m3	< 200
precipitator

-	outlet FGD	mg/m3	< 50

Operation time

DENOX 1	h	14.000

DENOX 2	h	2.000

¦ Figure 1, Technical Parameters of H'eilbronn 3-6

4B-27


-------
Flue gas flow rate Nm3/h	2,3 x 106

Output	MW^,	700

NOx-concentration

-	inlet of SCR	mg/m3	<	800

-	outlet of SCR	mg/m3	<	200

NOx-remova!	%	75 - 80

SOx-concentration

-	inlet FGD	mg/m3	1.500 - 3.200

-	outlet FGD	mg/m3	< 200 - 400

Dust concentration

-	inlet electrostatic mg/m3	7.000 - 12.000
precipitator

-	outlet electrostatic mg/m3	<	50
precipitator

Start up of DENOX-system	September 1986

Operation time	h	25.000

Figure 2. Technical Parameters of He11 brorvn 7

4B-28


-------
4© + 4® + ©55«r-4© + 6®
2 @ + 4 @ + © 320-c!!m4 1 3 © + 6 ©

Figure 3. Principals of the SCR-Process


-------
4B-30


-------
-fc.

CD
¦

a3



1,0



0,9



0,8

rel.

activity

0,7

K/K0
r%i

0,6

L /oj

0,5



0,4



0,3



0,2



0,1



0







%





























: i ~i

. measured

























































•» ^















Ti-

- - -























































ex|

sect

ed ^

















































y

~

/

~































ex|

sect

ed -



' *
*

*

*













al

ter























*

*

*









/me

asui

red

cl

ean

ng

1















*























t
%

1











„ if



























V

<:
r —'

)

T—



—I—



—I—

Li	1



	1—1

NH 3-slip
[vpm]

5

4

3

2

1

2500 4600 6300 8500	12000

operating hours [h]

18500

Figure 5. Heilbronn power station unit 7; DENOX-plant. Loss of activity and ammonia slip


-------
4B-32


-------
nh3/nox

0 5 10 15 20 25 30

irregular distribution %

Figure 7. Heilbronn power station
unit 7; DENOX-plant. Need of access
volume vs irregular distributions of
temperature, velocity and NH3/N0x

4B-33


-------
00

1

CO

>»

¦>
4-»

o

(0

o
>
'¦4-*

ro

0)

1.1
1.0
0.9
0.8
0.7
0.6
0.5
0.4
0.3
0.2
0.1
0.0

200
180
160
140
120
100
80
60
40
20
0

5000 10000 15000 20000 25000 30000 35000 40000 45000 50000 55000 60000

t	operating hours









2 layers old + 1 layer new, not plugged

2 layers old + 1 layer new, 20% plugged



I











I







I*

y



		"I

' i



' i

' i

1

Figure 8. Heilbronn power station unit 7; DENOX-plant. Loss of activity and NH3
in fly-ash in dependence of plugging; replacing of the 1st layer


-------
f stack

gas preheater

Figure 9. Tail end arrangement of the DENOX plant at Heilbronn unit 3-6


-------
high-dust
system

tail end
system

a

DO

I

CO
O)

7.4 - 7.0
430 - 470
1.0

320 - 400
100

pitch
surface
activity
temperature
pressure loss

mm

m2/m3

°C

%

4.2
750
1.0-1.2
(270) 300 - 320
-180-250

Figure 10. SCR catalysts for slag-tap firing systems


-------
Temperature

N0X, behind SCR

NHg-slip

Pressure drop
total plant
reactor
GAVO

290 °C - 320 °C
150 mg/m3
< 1 ppm

37 mbar
7,5 mbar
14,0 mbar

Start up- I Shut down-times

heat up from 20 °C to 320 °C 5 1/2 hours
cool down from 320 °C to 20 °C 24 hours

Operational values

natural gas consumption
NH3 consumption
electrical power consumption

1.300 m3/h
170 kg/h
4 MW

Figure 11. Operating Characteristics of DENOX 3-6

4B-37


-------

-------
SO, GENERATION -
JEOPARDIZING CATALYST OPERATION?

R. Jaerschky
A. Merz
J. Mylonas
Isar-Amperwerke AG
Brienner StraBe 40
8000 MUnchen 2, Germany

4B-39


-------

-------
ABSTRACT

Isar-Amperwerke AG's modern, hard-coal-fired combined power and dis-
trict heating plant in Zolling went into operation at the end of 1985.
To ensure conformance with the applicable emission limits, the power
plant was initially equipped with primary N0X reduction measures, a
high-efficiency electrostatic precipitator and a desulfurization plant.
With the introduction of a more stringent limit for N0X emissions, it
became necessary to retrofit a DeNOx plant. This DeNOx plant, which
functions on the principle of selective catalytic reduction using
ammonia, went into operation at the beginning of 1988 and achieved the
required separation efficiency without difficulty.

After a brief period of operation, however, acidic particles started to
be emitted. Extensive investigations revealed that these emissions were
the result of the catalysts' having a very high S02/S03 conversion rate.

On the basis of the investigations results, steps were taken which
reduced the emission of acidic particles to an absolute minimum. It
became apparent, though, that a permanent solution to the problem would
require replacing the catalysts.

For this reason, the catalysts were replaced mid-1990 after approxi-
mately 12,000 hours of operation by a new type with a much lower con-
version rate.

This paper reports on the operating results obtained with the old
catalysts, the investigations carried out regarding S0s/S0s conversion,
and first experiences gained with the new fill.

4B-41


-------
ZOLLING POWER PLANT

Zolling Power Plant - formerly Known as Leiningerwerk Power Plant Unit
5 - benefits from a number of environmental protection features:

High efficiency thanks to supercritical steam conditions

Extraction of district heat

Highly efficient flue gas cleaning with DeNGK, dust removal
and desulfurization

A liquid waste processing system with ammonia stripper
A pleasant architectural design (Fig. 1)

Figure 1» Zolling Power Plant

4B-42


-------
The technical data of the Zolling Power Plant are listed in Table 1:

Table 1
TECHNICAL DATA

Max, thermal output of furnace	1,144 MW

Main steam mass flow	384 kg/s

Main steam temperature	54 0 °C

Main steam pressure	274 bar

Reheat temperature	540 °C

Reheat pressure	55 bar

Gross output.at terminals	450 MW

Net output at terminals	42 0 MW

Max. district heat extraction	270 MW

(at a gross output at terminals of:	392 MW)

The power plant burns a mixture of German hard coals with a low ash and
sulfur content (see Table 2). It operates primarily in the lower inter-
mediate load range and is frequently shut down at weekends and at night.

Table 2

COAL ANALYSIS (AVERAGE VALUES]

Water	8.51 %

Ash (free of water)	7.52 %

Volatile substances (free of water and ash)	28.44 %

C content	73.46 %
H content 4.32%
N content 1.54%
S content 0.96 %

Net calorific value	28.99 MJ/kg

FLUE GAS CLEANING FACILITIES

Table 3 correlates the applicalbe emission limits in Germany with the
flue gas values at the boiler outlet and at the stack of Zolling Power
Plant;

4B-43


-------
Table 3



Dimension

Raw gas

Limit

Clean gas

N0X

®9/»n3 (6% 02)

650

200

190

SO,

mg/mN3 (6% o2)

1,900

400*'

200

Dust

mg/mN3 (6% 08)

7,100

50

5

#> but at least 85% SOa separation

To ensure conformance with the applicable limits, the power plant is
equipped with a high-dust DeNOx plant, an electrostatic precipitator and
a flue gas desulfurization plant (FGD), the arrangement of which is
shown in Fig. 2.

Bom'

FGD

Stack

Figure 2. Flue Gas Flow Diagram

The following in-furnace NOx controls are also incorporated into the
boiler:

Dynamic classifier for finer pulverization
Low-NOx burners for quasi-stoichiometric combustion
Air staging with over fire air

4B«44


-------
These primary measures result in a flue gas N0X content of about 650 mg
N02/ms3 (6% 02) at the boiler outlet, together with a low nonburned
residue content of the fly-ash (less then 3%) [1, 2].

DENOx PLANT

Zolling power plant initially went into operation without a DeNOx plant,
but a retrofit was rendered necessary by increasingly stringent NOx
limits.

with a view to linking the boiler outlet and the air preheaters, we
selected the high-dust configuration (see Fig. 2), which is wore favor-
able from the energy point of view.

We decided on plate-type catalytic converters on account of their
superior corrosion resistance and lower pressure losses. The main design
data of the DeNOx plant are summarized in Table 4.

Table 4
DESIGN DATA OF DENOx PLANT

Flue gas volumetric flow (damp)	400 mN3/s

Flue gas temperature min./max.	300 °C/	400 °C

N0X reduction	70 %
Max. NH3 slip after 12,000 hours of operation 5 ppm

Volume of catalytic material	522 m3

Specific surface area of catalytic material	330 m8/m3

OPERATING RESULTS

The objective of 70% NQx reduction was reached without difficulty throu-
ghout the service life of the catalysts (approximately 12,000 hours)•
Activity checks carried out on the catalysts showed that deactivation
took place much more slowly than had been postulated in their design
(F xg * 3) . i^f ter 12 f 000 hours o f operat xon, the rema x n xng actx,vx^~y^ of the
catalysts was still 94% of the original value. This minimal drop in
catalyst activity resulted in only a small increase in the ammonia
content of the fly-ash from about 4 mg NH3/kg at the start to around 10
mg/kg prior to the catalyst replacement. Figure 4 shows the increase in
ammonia content of the fly-ash in the first quarter of 1990; these
values point to an NH, slip of approximately 0.5 ppm.

4B-45


-------
Hours of operation [h]

Figure 3. Activity Deterioration of Catalyst at Zolling Power Plant

Spot-check analyses 1990

Figure 4. Ammonia Concentration in Fly-Ash Prior to Catalyst Replacement

4B-46


-------
Throghout the entire duration of operation, no problems were encountered
with fly-ash deposits on the catalytic material; at the full-load duty
point the pressure loss across all the catalyst layers (htota, = 2m) re-
mained virtually constant at 3 mbar for the entire 12,000 hours of
operation.

After only a short period of operation all these positive aspects were,
however, overshadowed by the emission of acidic particles, which caused
damage to the paintwork of cars parked in the vicinity of the power
plant and jeopardized the good reputation of our power plant.

S03 GENERATION

After extensive investigations and measurements it became apparent that
the emission of acidic particles was the result of oxidation of S02 to
so3 in the DeNOx catalytic converters. This sulfur trioxid subsequently
reacts with the steam, of which there is plenty in the flue gas, to form
gaseous sulfuric acxd.

The S02 conversion rate K (SOs) is applied to quantify the oxidation of
SO,, to S03

C (S03) d0wn8tream catalyst	upstream catalyst

K (SOz) =						 x 100%

2'upstream catalyst

with the C (S02) and C (SO.,) concentrations entered in ppra,

The following parameters have a significant effect on the conversion
rate (cf. [3]):

The chemical composition of the catalytic substance -
particularly the vanadium pentoxide content

The ratio of the surface area of the catalytic material to
the flue gas volumetric flow

The flue gas conditions, particularly the temperature

Figure 5 shows, as a function of the flue gas temperature, the S03
concentration C (S03) measured upstream and downstream of the catalytic
converters in the course of the investigations.

4B-47


-------
30

30

25

20

E

CL

a,

pJ

15

(J

10

upstream of catalyst
downstream of catalyst

0'

300

&3&
® o 0



320

340-	360

Flue Gas Temperature [°C]

380

o

°„i
o%

O 0O



20

15

10

400

Figure 5. S03 Concentration in the Flue Gas

It became apparent that the S03 concentration in the flue gas was raised
considerably by the DeNOx catalytic converters, and that there was a
noticeable dependence on the flue gas temperature. At a flue gas S0s
content of 665 ppm, the SO, conversion rate calculated for 375 °C was
1.2% and for 405 "C 3.3%.

At low load (120 WW) and temperatures of around 315 °C, the S03 content
was lower downstream of the DeNOx plant than upstream, i.e. at this duty
point the catalyst stores $Oa.

S03 CONCENTRATION IN FLUE GAS PATH

In order to better understand the mechanisms involved in the formation
of acidic particles, several series of measurements were taken to deter-
mine the S03 concentrations at a given time at various points along the
flue gas path downstream of the DeNOx plant. Figure 6 plots the SOs
concentration downstream of the catalytic converter, downstream of the
air preheater, and in the stack (i.e. downstream of the regenerative
gas reheater); this series of measurements begins at low load (110 MW;
flue gas temperature 315 °C), with a rapid increase to full load between
6.00 and 7.00 a.m., causing the temperature to rise to 375 °c.

4B-48


-------
30

25

20

E
a

Q.

co
O
cn

O

15

10

downstream of catalyst
	^	

downstream of air preheater

in the stack

	B	;

110 MW i

450 MW

315 °C

i' 375 °C

£r '

Q- -

s

s

¦ 				

: -A- = = »... —O. — — — — a -

Time

jy- -

T

10

301

20

15

10

Figure 6. S03 Concentration in the Flue Gas as a Function of Time

30

25

SO

Q.

O.

«
O

JO,

o

15

10

downstream of catalyst

— — A- _ -	^

downstream of_ air preheater
in the stack

Yl.

250 MW |

450 MW

345 °C



390 °C

Activation of steam air preheater

30

25

20

15

10

			

Time

Figure 7. S03 Concentration in the Flue Gas as a Function of Tine

4B-49


-------
A similar experiment is shown in Figure 7, but in this instance there
was no soot blowing, which meant that the flue gas temperatures were
higher. Starting from 250 MW and tra = 345 °C, the plant was run up to
full load between 15.30 and 16.00, causing the flue gas temperature to
rise to 390 °C, At 18.00 the flue gas temperature downstream of the air
preheater was raised from 130 °C to 150 °C by activating the steam air
preheaters.

In both cases the S0a concentration downsream of the DeNOx plant rose
sharply on load increase and stabilized after about 3 to 4 hours at a
somewhat lower level; this phenomenon is the result of thermal desorp-
tion of stored S03 (see Fig. 5) .

The S0a concentrations downstream of the air preheater were considerably
lower , which can be accounted for by the fact that the temperatures in
the region of the air preheater are below the acid dew point. This
results in sulfuric acid precipitating on the fly-ash, of which there
are large quantities (sulfur content of the fly-ash 0.5%).

The amount of sulfuric acid which precipitates depends to a large
degree on the temperature in the air preheater, as is clearly illustra-
ted in Figure 7. Activating the steam air preheaters, thus causing the
flue gas temperature downstream of the air preheaters to rise from 130
°C to 150 °C, resulted in a decrease in the precipitation of S03 from
24 ppm at 130 °C to 15 ppm at 150 °C.

The measurements also revealed that almost all of the S03 still present
in the flue gas downstream of the air preheater precipitates in the
regenerative gas reheater; in all experiments the concentration of
gaseous S03 in the stack was close to the minimum detectable level
(approximately 0.5 ppm). As there is virtually no fly-ash in the region
of regenerative gas reheater to absorb the precipitated acid fraction,
further investigations were carried out to determine what was happening
to the sulfuric acid.

After one week's operation, a plate was removed from the regenerative
gas reheater. On the cold side of the heat exchanger plate there was an
oily deposit of concentrated sulfuric acid which, extrapolated for the
entire heat transfer surface of the regenerative gas reheater, amounted
to a stored volume of 600 kg of sulfuric acid [3].

Solids such as residual dust, iron oxides and gypsum are deposited in
this acid film. The rotation of the heat transfer plates allows these
acid-soaked particles to pass to the treated gas side of the regenera-
tive gas reheater, where they can break away and, given a sufficient gas
velocity, be carried out of the stack.

Acidic particles were emitted primarily on rapid load increase, particu-
larly on plant run-up after weekend shutdown.

4B-50


-------
MEASURES TO PREVENT ACIDIC EMISSIONS

On the basis ot the investigations, the following steps were taken
prevent the emission of acidic particles:

Soot blowing was stepped up to ensure that the flue gas
temperature at boiler outlet remains below 380 °C, even at
full load. As Figure 5 shows, this halves the S03 concentra-
tion downstream of the air preheaters, i.e. it reduces the
S02 conversion rate from 3.3 % at 400 "C to 1.2%. As Figures
6 and 7 show, most of this SOa fraction can be seperated
out in the air preheater.

The flue gas temperature downstream of the air preheater
was lowered, but at temperatures of less than 130 °C the
pressure losses began to increase considerably, which was
apparently the result of ammonia bisulfate deposits. The
temperature was therefore raised again to 135 °C.

Baffles were fitted to the floor of the flue gas ducts
upstream and downstream of the regenerative air reheater
for separation of the large particles (see Figure 8).

Figure 8. Baffles for Separation of Large Particles

4B-51


-------
During each weekend shutdown the flue gas ducts in the
region of the regenerative gas reheater were freed of
particles.

The blowing procedure of the regenerative gas reheater was
modified; the hot and cold untreated gas sides are blown
simultaneously so that the loosened particles remain on the
untreated gas side.

The plastic heat exchanger plates were removed from the
regenerative gas reheater in order to reduce cooling and
reheating and to achieve a smaller separator surface.

The facility for drying the treated gas from the FGD plant
using hot, S03-laden untreated gas was taken out of service.

During each weekend shutdown the regenerative gas reheater
was flushed out with large quantities of low-pressure water
(10 bar). After each flushing cycle (duration 3 hours,
volume of water around 150 m3) , the pH of the water dischar-
ged was measured; the flushing process was terminated as
soon as a virtually neutral pH (around 6) was reached.

As a result of applying these measures, no emmisions of acidic particles
have been detected since April 1989. However, the effort involved and
the damage done to the plant components by the sulfuric acid (corrosion,
shortening of service life, etc.) are tremendous. It was therefore
decided that, in the long term, the catalyst would have to be replaced
by a type with a lower conversion rate.

THE NEW CATALYST

Our Japanese suppliers, like all well-known manufacturers of catalytic
converters, invested a lot of effort in developing a low-conversion
catalyst.

Since the reduction of NO by NH3 takes place on the surface of the
catalyst, while the oxidation of S02 to S03 is a volumetric reaction,
i.e. it increases linearly with the volume of the catalytic material,
the plate thickness of the catalytic converters was,reduced, thus
decreasing the volume while maintaining the surface area for the DeNOx
process.

A further significant reduction in the conversion rate was achieved by
refraining from adding vanadium pentoxide to the catalytic material.
However, this heavy metal promotes DeNOx activity, particularly in the
300 - 370 °C temperature range, this measure resulted in a 10% increase
in the necessary catalyst volume to 574 m3.

4B-52


-------
We were assured that this new catalyst would, at the same N0X reduction
efficiency of 70%, have a maximum ammonia slip of 5 ppm after 16,000
hours of operation. An S0a conversion rate of 0.9% at 400 °C was antici-
pated; inservice measurements under normal power plant conditions
yielded values of around 0.7%. Right from the start, however, a notice-
ably higher NH3 slip of about 1.5 ppm was registered.

FIRST OPERATING RESULTS WITH THE NEK CATALYST

The catalyst was replaced during the unit outage in June/July 1990.
Acceptance testing was performed in September 1990 with the following
results:

The catalyst achieves the required N0X reduction without the increased
NH3 slip anticipated on the basis of the experimental measurements. Spot
measurements showed the maximum to be 0.5 ppm NHa and the average 0.3
ppm. These values were confirmed by the fact that the NHa content of the
fly-ash after catalyst replacement (Figure 9) is similar to that prior
to replacement (Figure 4).

Figure 10 presents the results of the SO, measurements upstream and
downstream of the catalytic converter.

Figure 9. NH3 Concentration in the Fly-Ash after catalyst Replacement

4B-53


-------
30

25 -

20

upstream of catalyst

downstream of catalyst

a.

CL

o
w

o

15

10

at

m

30

25

20

15

10

300

320

340	360

Flue Gas Temperature [°C]

380

400

Figure 10. SO, Concentration in the Flue Gas

At a constant S03 content upstream of the catalytic converter, the
values downstream of the DeNOx plant are considerably lower than those
shown in Figure 5. At 394 °C, only 7 ppm were measured (old catalyst >15
ppm) , thus confirming the S02 conversion rate of 0.7% at 400 °C in actual
power plant operation.

As was the case with the old catalyst, storage of S03 occured in the low
load range (ttG = 315 °C). Figure 11 reveals the surprising fact that no
detectable release of the stored S03 was observed on load increase from
200 to 450 MW.

Figure 11 also shows that the S03 concentration downstream of the air
preheater is close to the minimum detectable level of about 0.3 ppm. It
should be added that, under all load conditions, the gaseous S03 and
sulfuric acid aerosols detected in the flue gas ducts downstream of the
air preheater were always in the minimum detectable range.

As anticipated, the total S03 fraction can therefore be separated out
in the air preheater by reducing the temperatures to values below the
acid dew point. The fly-ash analyses confirm that the S03 content of the
flue gas upstream of the air preheater is now considerably lower; at
otherwise constant values, the sulfur content dropped from 0.5% to 0.3%.

4B-54


-------
30

25

20-

upstream ^of catalyst
downstream of catalyst
downstream of air preheater

E

CL
Q-

CO

o
w

o

15

200 MW i

310 CC I

450 MW

365 cC

10

30

25

20

15

10

Time

Figure 11. S0a Concentration in the Flue Gas Path as a Function of Time

Since catalyst replacement, there has apparently been no further preci-
pitation of sulfuric acid in the regenerative gas reheater. On flushing
the regenerative gas reheater with LP water, the first water discharged
was found to be approximately neutral (pH > 4.5), so it was decided to
extend the flushing intervals.

To summarize, the S02 conversion rate is one of the most significant
criteria to be considered when selecting a high-dust catalytic con-
verter.

Our experience with a low-conversion catalyst has shown that a trouble-
free operation of the DeNOx plant is possible, without risk of the
emission of acidic particles.

REFERENCES

G. Musset, U. Schroder, E. Swoboda and D. Kiefer.• "Operating Experi-
ence with the Low-No, Firing Concept in Unit 5 of the Leiningerwerk
Power Plant of Isar-Amp'erwerke AG".VGB-Kraftwerkstechnik. Vol. 69,
No. 4, April 1989.

0

R. Jaerschky and A. Merz. "NOx Reduction at Zolling Power Station
Pre-Combustion and In-Furnace Measures, SCR Catalyst Equipment".
ABME Paper 90-JPGC/FACT. October 1990.

4B-55


-------
H. Gutberlet, A. Dieckmann, A. Merz and L. Schreiber. "S02 Konver
sionsrate von DeNOx-Katalysatoren - Hessung und Auswirkung auf nach
gesehaltete Anlaaenteile". Cheroie im Kraftwerk 1989. pp. 86-96.

4B-56


-------
SCR OPERATING EXPERIENCE ON COAL-FIRED BOILERS
AND RECENT PROGRESS

Edward S. Behrens
Joy Environmental Equipment Company
Monrovia, California

Senichi Ikeda
Electric Power Development Company
Tokyo, Japan

Teruo Yamashita
Idemitsu Kosan KK,
Chita, Japan

Gunther Kittelbach, PhD
Deutsche Babcock Anlagen AG
Krefeld, Germany

Makoto Yanai, PhD
Kawasaki Heavy Industries, Ltd.
Kobe, Japan

:243-317-"!


-------

-------
1,0 - INTRODUCTION

Selective Catalytic Reduction (SCR) technology development traces
its roots back to the early 1970's. Its use for NOx reduction in
oil- and gas-fired energy-conversion plants has long been widely
accepted as Best Available Control Technology (BACT) by many U.S.
regulatory bodies. Its acceptance on coal-fired plants overseas
has demonstrated its ability to reduce MOx from these fuels. But
its acceptance in the U.S. for coal-fired power generation has
lagged somewhat due, in part, to perceived problems revolving
around the typically hightr sulfur coals found here. Nevertheless
over 50 commercial coal-fired plants overseas are proving that with
proper physical and chemical catalyst designs and operation these
perceived problems can be overcome. This paper will detail the
operating experience of three coal-fired commercial power plants
using this technology to successfully control their N0„ emissions
and present an up-to-date review of SCR technology.

2.0 - COAL-FIRED SCR OPERATING EXPERIENCE

Detailed below are the design specifications and operating results
of SCRs in three coal-fired power plants in Japan and Germany
including other items of operational interest with respect to
experience with the SCRs installed.

2.1 - Takehara Power Station

Electric Power Development Company's Takehara Power Station, Unit
1, in Hiroshima, Japan is a 250-KW, coal-fired boiler burning 2.3-
to 2.5-percent sulfur coal. It uses hot-side low-dust SCR
arrangements in two parallel SCR reactors A+B, each handling 50
percent of the flue gas. The reactor B SCR was placed in service
in 1981, and the current catalyst charge has been in service since
1985 using Type 470 catalyst elements having a 7 mm pitch. The
SCRs are located downstream of the hot-side electrostatic
precipitator (ESP) and upstream of the air preheater. Flue-gas
temperature is 658"F (343'C) with a N0X removal efficiency of 80
percent at full load.

Figure 1 is a photo of the downflow reactor at Takehara, and Figure
2 is a sectional elevation of the SCR reactor showing three layers
of catalyst and the vertical vanes used to assure proper gas-flow
distribution. Figure 3 is a computer generated diagram of the SCR
gas-flow showing even gas distribution across the catalyst modules.
Catalyst modules are loaded from the side of the reactor by means
of a fork—lift track assembly.

4B-59


-------
The Takehara SCR unit I-B

FIGURE #1

Sectional Elevation of Tbf
Taktharc SCR unit I B



t, /I i ¦¦¦



I i		'' »i~.

-•(>'*I xK\i. ¦ i

U>i

2mI «dm»i
On4i«t»0«i!

JU.oc

FIGURE #2

Gass Flow Analysis of

4B-60


-------
Figure 4 shows fine dust deposition from the ISP on the top
catalyst module after 34,000 hours of operation using the current
large openings, Type 470, 7 Dm catalyst type which is a thin-wall
catalyst. Flyash from the ESP is very fine and is highly adhesive,
and is one of the reasons a high opening type 470 catalyst was
selected. Despite the high S02 levels (1800 ppm) entering the SCR
no plugging of the air preheater by ammonium salts has been
observed. No additional preheater washings have been necessary.
This is accomplished by maintaining low KH3 slip levels.

Typical ash deposit with no soot blowing
after 34,000 hr. operallon al Takthara.

Table 1 details design data and Table 2 shows actual performance
results from annual performance tests of the SCR at the Takehara
Power Station.

TABLE i
r**afcara Pewar Static*
unit l-B
Elaetric fevar Davalapaan* cd.

a*atm

teafl '.mm pmt SOU



las

«*• Tim caefai



353,438

Taparaeurn (7)



SSI

no, in ippevdi1



304

SO, Ift (pp*v4j



1150

Mf eemraraion (I)



0.9

PIT (fr/aet) ity



0.04

DaN®, Iffieiancy l\:



ID

Slip Cppwd)1



4

wo, cue (ppwvd)1



SO

Catalyst daita P (in

HjOl

1.»

Cataiyat contact araa

(a*/*3?

470

1 c«rracc«d so n o,

* Ku. «llov»Ma bafor* Catalyst MpiacaaMnt

4B-61


-------
TABLE 2

TAKEHARA UNIT i-a SCR PERFORMANCE test result







DEC.

ms

_ ,NQV,
1986

OCT.
1587

FEB,
1988

OCT.
1988

OCT.
1989

JUNE
1990

Gaa Velvaa

x 1,

000 SCFH

246

256

235

24ft

229

246

235

Gaa tuNtatura

¦P



649

669

649

644

651

6S3

660

KO, 1 =

ppv



3 0*

383

251

249

270

315

279

NO, OW

pp®

<6% Oj}

61

60

43

52

57

64

51

mho, tillclancy

1

80

73

S3

79

79

30

82

siip sr

pp*

(61

0.1

C.l

0.2

o.x

O.X

0.2

0. 2

Slip KB.

ppm

fdeaiqn



















b&ais)

0.1

0.1

0.2

0.1

0.1

0.2

0.2

SO, It.

PP*



1,490

luao

I, 340

1,210

956

1,040

1. 170

SO, Is

ppm



5.2

4.9

3.2

.8

2.2

2.1

1.8

SO, OUT

ppm



7.6

7.4

3.2

4.6

2.9

3.3

2.8

SO, ccsrrarsion

1



0.18

3,21

0-08

0,15

O.QB

0.12

0.09

Catalyit vaa replaced in tiia pariodic inspection Ntwan Septambar and Novencar, lsiS.

2.2 - HKW Reuter West, Boiler E/D

The power station Reuter West of the Berliner Kraft- unci Licht AG
in Berlin, Germany does include 2 x 284 MW coal-fired power plants
burning various coals with sulfur contents up to 1,2 percent. The
boilers D and E of those two plants went on the line in 1988 and
has operated about 15,000 hours till November 1990. It is an
example of the hot-side, high-dust SCR arrangement- The SCR reactor
is located between the economizer and the air preheater, upstream
of any particulate-removal or flue gas desulfurization (FGD)
equipment. The flue gas temperature entering the SCR depends upon
coal burned, firing rate, and the condition of the boiler
(dirty/clean)? the normal average value is about 360" C. NO^
removal is more than 85% depending on removal requirements; the
average ammonia slip is about 1.5 ppmvd and the S02 conversion to
SOj about 0.5 percent.

Figure 5 is a photo of a model of the SCR unit at Reuter West, and
Figure 6 is a sectional elevation of one of the half-capacity SCR
reactors of one boiler. It shows space for three and one-half
layers of catalyst. Layer Ho. 1 uses only one module while layers
Ho. 2 and 3 use two nodules stacked on top of each other? the
fourth layer is a spare and is currently not used. It is held
vacant for use in future catalyst replacement programs or, if
needed, additional NO,, removal. Stacking modules facilitates

4B-62


-------
interchangeability of the modules and reflects the fact that
catalyst elements can only be extruded up to about 1 m xn length.
The typical level of potential catalyst poisons found in fly ash of
the Ruhr coal (K20 - 4 to 5%; CaO - 6.3%; MgO - 1.5%; P2o5 - °-6%)
have not appeared to significantly accelerate catalyst
deterioration. Table 3 shows examples of measured operation dates
of the SCR of one of the boilers at HKW Reuter West.

The SCR at Reuter IVeil

FIGURE #6

4B-63


-------
TABLE 3

KKK Reuter West
Boiler £
D«vag

OPERATION DftTA

Load (HWe)

230

1S5

Tesp»ratur« (F)

680 .

662

kvg, Oj%

5.3

6.3

HO, In (ppavd)"

350

350

SO, In (ppBvd)*'

865

¦ f!

SOj conversion (%!"

0,5



Ply Ash (gr/scf) dry"

5.11



DeNO, Efficiency (%!

85

86

Ifflj Slip (ppovdS'

1.1-2.1

.7-1.6

NO, Out (ppavd)1

52

50

Catalyst contact area (hVh3!

427



Corx«ct«d to «% Oj
11 Design Condition
15 Measured at 725 *F

The operation experiences are good. The activity losses of the
catalyst which were controlled after about 10,000 hours of
operation were much less than expected. Due to the relatively low
S02-conversion rate and low NHj-slippage no plugging at the air
preheater has occurred and no washing of it was necessary since the
start up of the plant. The soot blowing of the catalyst layers is
done only one time per week.

2.3 - Aichi Refinery

Idemitsu Kosan KK's Aichi Refinery Boiler No. 4 power unit in
Aichi, Japan is a 40-MW, coal-fired boiler burning 0.4-percent
sulfur coal. It was placed in service in 1986 and uses a hot-side,
high-dust SCR arrangement. The SCR was designed for an efficiency
of about 59 percejnt producing an outlet NO„ consistently below 90
ppm. Inlet NOx generally ranges between 200 and 250 ppm and the
SCR sometimes operates at >60% NO^ removal, maintaining outlet NO^
below 90 ppm.

The air preheater is a Lungstron type with corten elements. This
air preheater was designed to permit NH4S04 deposition at the cold
side of the hot-side element. Since start-up, there has been no
need for washing between the scheduled annual outages, nor has any
element been replaced or repacked.

4B-64


-------
Aichi No, 4 power unit is the earliest SCR system equipped with a
baghouse downstream. In general, it has been predicted that
baghouses downstream of an SCR would experience increased pressure
drop due to NHS leakage and S03 increase by the SCR. However, the
baghouse at Aichi continues to operate without increased pressure
loss, because of the low nh3 slip and low S02 to S03 conversion.

Figure 7 shows the Aichi boiler and SCR unit. Figure 8 is a
computer-generated gas-flow diagram for this SCR reactor which
utilizes a turning vane to improve distribution minimize gas eddies
and reduce pressure drop, across the catalyst module surface. Once
per day, soot blowers operate to clean the first catalyst layer
only. There has been no observation of catalyst plugging or
increased pressure drop. NHS in the ash was 28 ppm after 5000
hours of operation.

The SCR at the Aichi Refinery	Gas flow analysis of the Aichi Reactor

FIGURE #7

FIGURE #8

4B-65


-------
Activity of the catalyst has been tested annually and the activity
loss was very small compared with other high-dust systems (see
figure 10). The major cause of deterioration is deposition of Ca,
and Si on the surface of the catalyst.

As a test, KHI enclosed several catalyst elements of Type 555
geometry, among the usual Type 470 catalyst in the Aichi SCR. This
allows actual exposure of the test catalyst to the flue gas Cor
extended periods. Inspections of the Type 555 over a three-year
period showed very little plugging. Similar to Type 470 adjacent
to test elements.

Table 4 lists design conditions of the SCR installed at the Aichi
Refinery, Boiler 4,

ZkBlX *

kichi R»fin«ry
Unit 4
Iduitiu Koha&n Co.

Baaion Csndltlona

Load [Mfa)



40

Flow (seta)



IIS,965

Taaparatura if)



716

Avx}. 0,1



3.36

HO, In (ppwvd)'



230

SO, In (ppavtll



429

SOj conv«r»ion (»)



1.2'

Fly Uli (gr/sct! dry



3 .73 - 15.28

Dane, Efficiancy (II



59

Hit, Slip ippsvd),



3-5

NO, Out (ppmvd)'



90

Catalyst d«lta P (in X2S)

3.1

Catalyat cant-act araa



470

1 earraetad to 61 o,

* HaxlBU* allovibla batora Catalyst Raplaeaaant

4B-66


-------
2.4 - Present, and Possible Future Catalyst Geometries.

Present and possible future catalyst element geometries are shown
in Table 5. The thin-wall elements contemplated for future
installation exhibit improved N0X reduction and reduce S02 to S03
conversion (see 3.1). The thin-wall geometry is also less prone to
fly ash plugging. Since Type 555 has been successfully tested in
the Aichi reactor, and Type 572, which has a larger void/opening
ratio than Type 555, its future application looks promising.

TABLE 5

Examples of grid honeycomb geometry

CelU per Side

20*20

21 *21

25 x 25

35*35

Status

Conwoaraf

HighOper«n

Canwntfcmal

Migti 0b***)

Caimmnt



GtOfT... ./

IJentifieauon

Type 427

Type 470

Type 555

Type 572

Type 751

Type 816

Will

Thickness (mm;

1.35

LOO

1.00

0.80

0.80

0.55

Opening
rate (%)

64

71

68

73

64

76

Application













Gas Firing
Boiler / Turbine

. Oil Fifing Bailer

Diesel

Coll firing Boiler

Municipal Waste
Incinerator









































4B-67


-------
2 . S — NH, Injection,

Proper mixing of and NH3 is important to insure required N0X
reduction. Figure 9 shows the NH3 injection systems employed at
Takehara and Aichi which provides adjustability for gas
distribution through header regulation and individual orifice
replacement. During start-up of these NHj injection systems, it is
necessary to fine tune the AIG to ensure performance. However in
smaller plants, like Aichi, it has been demonstrated that the
standard deviation of NH3 distribution is usually within allowable
limits. However, if fine tuning is needed to optimize nh3
distribution, it is easily accomplished.

Typical SCR Ammonia Injection Grid

FIGURE #9

2.6 - Observed Catalyst Operation

At both Takehara and Aichi, catalyst physical and activity changes
over the life of the catalyst have been carefully monitored. This
is done as part of catalyst management program. Tests are made
annually. Since Takehara is a low-dust installation, erosion, of
course, has not been experienced. At Aichi, some slight erosion
has been noted in the top layer of catalyst, but it has not been
severe and will not compromise anticipated catalyst life.
Deterioration is relatively low for a high dust system. Primary
cause of deactivation is deposition of Ca,Si on catalyst surface.
Figure 10 shows the relative catalyst activity reduction for the
three coal-fired plants discussed versus cumulative operating
hours. All three plants exhibit better than expected catalyst

4B-68


-------
activity after 20,000+ hours of operation. Figure 11 shows the KHI
catalyst activity measurement facility, and Figure 12 shows their
erosion simulation facility.

, CATALTSt ACTIVITY OETEAlOAATiQN
LMDUn>er 0*r*

@ TMbtMl Ml 14	¦! • *<*B Kit*!?*

fi	«e • MU*. iMT¦ I Uiwii.

# flrw« •urn i

FIGURE #10

KHI Catalyst Activity
Test Facility

KHI Catalyst Erosion
Test Facility

FIGURE #11

FIGURE #12

4B-69


-------
3.0 - MODERN SCR TECHNOLOGY

The fundamental basis of SCR technology is based on the catalyzed
reduction of N0X (mixtures of NOz and NO) with ammonia (NH^) into
innocuous water (H20) and nitrogen Nz) in two general reactions:

NO + NHj + i/402 -* N2 + 3/2HzO (main reaction)

NO + NHj + 2NHj - 2Nj + 3 H20

NHj (in the form of liquid anhydrous or aqueous ammonia), which has
been vaporized it is diluted with air. The mixture is injected
into the flue-gas stream. The NH3 is injected upstream of the
catalyst appropriate to NO„ removal requirements through a
distribution grid.

3.1 - Catalyst Properties

Ceramic, homogeneous, honeycombed catalyst elements measuring
approximately 6-in. x 6-in. square are extruded up to 39-in. long
(150 mm x 150 mm x 1 m) . Titanium oxide (Ti03) as the base
material and is used to disperse and support the vanadium pentoside
(V2Os). Tungstein oxide (WOs) provides thermal and mechanical
stability. This titanium-based catalyst has been proven to provide
the highest durability and excellent reactivity. By changing the
mixing ratio of the active components, the catalyst can be tailored
to meet specific flue—gas requirements.

The vanadium content controls the reactivity of the catalyst. But
it also catalyzes the oxidation of S02 to S03. Therefore in high-
sulfur applications, it is necessary to minimize the vanadium
content. Through homogenous distribution of V205 throughout
catalyst elements, activity reduction of possibly low v2o5 catalyst
is minimized.

within the honeycombed catalyst elements, the incoming NO^/NHj
mixture enters micropores on the catalyst's surface and diffuses
back out after the chemical reactions have taken place within the
catalyst material itself. Therefore one of the goals in catalyst
development has been to attain a good mixture of macro-pores to
support gas diffusion and micro-pores to support the reaction
itself.

Honeycomb pitch (flue-gas passages) can also be varied to
accommodate a range of flue-gas dust loadings. The sectional
geometry of a few of the flue-gas-flow patterns used in these
catalyst elements are shown in Figure 13. Since the effective
depth of catalyst for N0X reduction occurs near the surface
(approximately 0.1 mm deep) it is possible to reduce the catalyst

4B-70


-------
volume if the catalyst surface area is increased by increasing the
number of cells. This produces a smaller cell pitch with thinner
walls. Alternatively, S02 oxidation occurs more slowly and takes
place deep inside the catalyst material. Therefore, the S0^ to S03
conversion can be decreased by decreasing inner-wall thickness
without reducing N0X-removal activity.

Honeycomb Catalyst Elements

FIGURE #13

All of these catalyst optimization features must, however, be
tempered in coal-fired applications by the fact that large-passage
honeycomb patterns permit freer flow of the dirtier flue gasses
typically encountered, but they also present less catalytic surface
area for reduction of N0X. In general, erosion-proof catalysts
demonstrate lower activity compared to catalysts used in low-dust
environments such as gas/oil-fired applications. Erosion-proof
catalysts inevitably have a smaller volume of micro-pores, the
major cause of their lower activity. Recent developments have
succeeded in increasing the activity of these lower-activity
catalysts to near optimum levels.

3.2 - Catalyst Modules and Reactor Design

The honeycombed catalyst elements are assembled into steel-cased
modules of the required size, Figure 14, for ease of handling and
installation. Modules can be inserted into the SCR reactor on
rollers or by an overhead crane. Modules are then stacked
horizontally or vertically within the SCR reactor on engineered
support structures. In coal-fired power plants, flue-gas flow is
vertically downward. Figure 15, to facilitate the passage of fly
ash through the catalyst elements with minimum drop out.
Horizontal dimensions of the SCR unit are set to optimize gas

4B-71


-------
velocity pressure drop and distribution through the SCR catalyst
elements; the number of vertical layers of modules (usually 2 to 4)
is determined by the desired N0K removal efficiency and the
temperature of the flue gas. NHj is introduced at the inlet of the
SCR reactor through an Ammonia Injection Grid (AIG) system which
nixes the NH3 thoroughly with the incoming flue gas before entering
catalyst-module array.

Catalyst Module

FIGURE #14

Typical Downflow SCR System



FIGURE #15

4B-72


-------
3.3 - SCR Catalyst Operating Conditions

Optimum flue-gas temperature for SCR reaction is between 500*F and
800* F (260*C to 427 °C). Below this temperature range, chemical
reactivity is impaired, and above it, physical catalyst damage can
occur through sintering. Catalyst degradation can also result from
poisoning through chemical reactions which tend to neutralize the
catalyst's reactivity and masking caused by ammonium bisulfate
(NH4HSOt) , ammonium sulfate ((HH4) 2S04) and fly-ash deposition. Any
of these adverse effects can cause NHS leakage to increase through
the SCR which is generally limited to <5 ppm for high-sulfur coals.
This occurs because the catalyst has become less reactive, thus
requiring more NH3 to achieve the same N0X removal. Figure 16 shows
NHj slip increases with time with constant DeNOx. While design
criteria calls for slightly over two year's operation before NH3
slip increases to the 5-ppm level, a predicted mean operating life
of over three years is expected. On the other hand, measured
operating unit experience shows that catalyst life of well over
four years can be anticipated. This is an example of the catalyst
maintenance program used at Aichi.

Aichi SCR Catalyst

'Oesignefl

' !

Predicted an

I

I	;	§ f!	la y; jj jj—

Cynw	Ope * «i < oh Meu'j Ml 328 n«g r >

111 • t KO.:2S0 Ppw, Qui l« i NOn ; 90 ppm, Typa*?@>, SViBiaP"1 i«H

D«i> e»»»a e« I tuor ito*> /	at ho tiiiiyn

FIGURE #16

Figure 17 shows that the mole relationship between NO^ removed and
NHj consumed is nearly linear except in the high efficiency range.
In order to achieve high removal efficiency with low ammonia slip,
the catalyst volume must be increased. This results in higher SOz

4B-73


-------
to S03 conversion. Increasing temperature also Increases S02 to S03
oxidation. This is# of course, undesirable and should be held to
<3 percent since SQ3 promotes the formation of ammonium sulfate and
ammonium bisulfate which can plug downstream heat transfer and
emission-control equipment. For a given catalyst volume a typical

interrelationship of tem^^eratiire7
reduction is shown in Figure 18.

SO, to SO, conversion and NO.

NO, nductkw »ad HI I, Ulp Nil,/NOs molt ratio

'0 4 0 6 0 8
NHj/NO* Mole Rairo

FIGURE #17

HO, rttlucHoft ami percent SCI, cntittnton »». SCR tovprvtan

£
>
£J
C

0

D
m
ir

90

85

80

75

70

High suilur,
catalyst

2 0
1 5
1 0
05

600 650 700 7SO 800
Temperature (*F)

u

<2

o

O

if)

FIGURE #18

4B-74


-------
Location of the SCR unit within a coal-fired power plant's flue-gas
stream is critical. Three options are available here, Figure .19;
(1) a hot-side, high-dust system is located after the economizer
and upstream of the air preheater and any emission-control
equipment such as an electrostatic precipitator (ESP), baghouse, or
FGD; and (2) A hot-side, low-dust ESP unit,* and (3) A cold-side,
1 ow—dust system i,s 1 ocated after the air preheater and ejnission—
control equipment.

aon.cn |—nm .

~1	1 Jto-Iio'c ESP		 rets

pJ^rfrnrrtP'

\i	a — * v ^ n—	 I |

VI	1	,	,	„			) ; STUCK

Hot Side High Dust System

FIGURE #19

In a hot-side, high-dust reactor, a large-pitch catalyst must be
used to accommodate the heavy dust loading. Furthermore,
reactivity of the catalyst is reduced to minimize oxidation of S02
to SOj. Conversely, in a cold-side, low-dust system, most of the
particulates and S02 have been removed from the flue gas.
The r e f o r e, a sssa 11 i^wCh^, ^gl*x^s\ir f ace a r ea cat a 1 y s^» can ^se lis ed •
Generally this requires less catalyst volume, and a more active
catalyst can be used since there is little concern for oxidation of
S02. But the primary advantage of this system as well as the hot-
side low dust system is the re^duced deterioration rate of .the
catalyst. However, reduced operating temperatures can require
reheating the flue gas. In most operating coal-fired power plants
the costs associated with reheating the flue gas outweigh the
savings from reduced catalyst volume and maintenance of the cold-
side SCR.

4B-75


-------
3.4 - Catalyst Module Replacement Schedules

Fifty to 60 percent of the cost of an SCR system lies in the
catalyst. Therefore catalyst life and associated replacement
schedules have a significant impact on the economics of the SCR.
The life limiting factor for these catalyst elements is increasing
NH3 slip with time which should be held to <5 ppia, for coal-fired
hot-side SCRs.

Figure 20 is a typical SCR configuration, and Figure 21 shows two
possible catalyst module replacement programs that can be used.

In the first example shown, three layers of catalyst modules are
used initially with provision for a fourth layer. When the NH3
slip has reached 5 ppm after approximately 24,000 hours of
operation, a new layer of modules is added in the vacant bottom
position. Then after an additional 16,000 hour of operation or so
when the KH3 slip has again increased, a new top layer of moduler
is installed, replacing the original top layer. When the NH3 slip
again increases after another 15,000 hours of operation, a new
intermediate layer is installed, again replacing one of the
original module layers. In subsequent replacements, a new layer is
added, and the oldest layer is removed.

In the second example shown in Figure 21, the full catalyst charge
is replaced every 24,000 hours. This results in a 70 percent

Catbaiyst Module Arrangement
in a typical SCR reactor

ms t(Uteris* control

FIGURE #20

4B-76


-------
increase in catalyst consumption over life of the plant. But, if
the additional spare layer module replacement program is chosen,
due consideration must be paid to the capability of the ID fan to
accommodate the resultant increased pressure loss caused by the
addition of another catalyst layer.

Proflra* C»f*eIHigh-Dust SCB

FIGURE #21

3.5 - Catalyst Regeneration

During operation, fine dust partical deposit on the catalyst's
surface causing the micropores to plug and reducing activity.
Generally, these particles can be removed, so that the catalyst can
be reused. However, regeneration methods do not fully restore the
original catalytic activity because of such factors as sintering
due to heat, but they do approach it. Currently, it appears that
the best method of catalyst regeneration involves sandblasting,
using a sand grain size of 0. lion which is blown through the
deteriorated catalyst's passages. However, further development of
this technique is required before full commercial practice is
available.

4.0 - CONCLUSION

This operating data on coal-fired commercial power plants burning
medium sulfur coal indicates that SCR is an effective technology
for reducing N0X emission and are not presenting abnormal operating
difficulties due to fly ash or S02/S03 in the flue gas, excessive
NH, slip, nor high S02 to S03 conversion.

4B-77


-------

-------
TECHNICAL FEASIBILITY AND COST
OF SCR FOR
U.S. UTILITY APPLICATION

C.P. ROBIE

P.A. IRELAND

UNITED ENGINEERS & CONSTRUCTORS INC.
WESTERN OPERATIONS

AND

J.E. CICHANOWICZ
ELECTRIC POWER RESEARCH INSTITUTE


-------

-------
ABSTRACT

The cost of utilizing Selective Catalytic Reduction (SCR) for N0„ reduction in
both new and retrofit applications is presented. Retrofit cases include hot-side
SCR technology applied to both FC and cyclone-fired units and post-FGD SCR
technology applied to a PC-fired unit. Technology status is assessed based
primarily on recent European experience. The impact of operational effects and
resultant modifications on downstream equipment are included in the analysis.
The hot-side capital costs (December 1989 dollars) range from $78 to 587/kW for
the new PC-fired case, $125 to J140/kW for the retrofit cyclone case, $96 to
$105/kW for the retrofit PC case. The single post-FGD SCR case evaluated is
estimated at $140/kW, The hot-side 1 eve!ized costs range from 5.3 to 5.9
mills/kWh for the new case, 8.2 to 9.1 mil 1s/kWh for the retrofit cyclone - fired
case, and 5.9 to 6.5 mil1/kWh for the retrofit PC-fired case. The levelized cost
for the single post-FGD SCR case presented is 6.8 mills/kWh.

INTRODUCTION

The feasibility and cost of applying ammonia-based selective catalytic reduction
(SCR) to control nitrogen oxide (NOJ emissions from power plants firing U.S.
coals is of considerable current interest. Although the N0X control requirements
of the 1990 Clean Air Act Amendments (CAAA) focus on low N0X burner technology
and other forms of combustion control, other factors - such as the CAAA NO,
emissions averaging provision, and strict N0X control requirements considered by
various state and local regulatory agencies provide the prospect of SCR
application in the U.S. In fact, applications for several low sulfur coal-fired
facilities developed by independent power producers in selected northeastern
states either require SCR, or a detailed, factual accounting of the feasibility
of SCR for the site. The considerable extent of SCR application in Japan and
Europe for low sulfur fuels has been a significant factor in promoting the
application of this technology in the U.S.

This paper completes the presentation of data from an EPRI-funded activity to
evaluate the feasibility and cost for various potential applications of SCR.

4B-81


-------
This study addresses the following six applications, proposed as representing the
range of potential SCR applications:

1.	New Plant - low sulfur coal

2.	New Plant - high sulfur coal

3.	Retrofit - cyclone boiler, high sulfur coal

4.	Retrofit - conventional (wall or T-fired) boiler, high sulfur coal

5.	Retrofit - post-FGD (e.g. reactor following S02 scrubber)

6.	Retrofit - oil-fired boiler

Results for cases 4 and 6 were reported at the 1989 Symposium (8). This paper
summarizes results for cases 2,, 3, and 5, with limited case 4 results repeated
for comparison.

DESIGN PREMISES

Key design Assumptions. SCR costs are significantly influenced by several key
design assumptions. The most important design variables used in this study are:

1.	Catalyst life - Several coal-fired European SCR installations have
operated for over two years without catalyst replacement and only
moderate measured loss in activity. A catalyst Tife of four years for
coal-fired hot-side SCR applications and four years for post-FGD SCR
applications has been used in this evaluation.

2.	Catalyst cost - Catalyst costs in Europe have decreased since 1985 by a
factor of approximately 2.5, primarily due to a very competitive supply
situation. Accordingly, this evaluation covers catalyst costs from

$330/ft3 to $660/ft3, covering the range seen in Europe. .

3.	Ammonia slip - Ammonia slip in European SCR installations is typically
specified at 5 ppm, while some utilities recommend even lower levels (2
ppm). For several coal cases in this study, both 5 ppm and 2 ppm slips
have been evaluated.

4.	Space Velocity - Advances have been made in catalyst formulation to
minimize S02 to S03 conversion, to develop smaller pitches and to
provide resistance to fouling by trace elements. These various advances
are reflected in the space velocities used for the cases evaluated.

Case Definition.- In order to develop representative costs for both the new and
retrofit SCR study cases, typical power plant layouts and design conditions were
selected. In the case of the retrofits, actual U.S. power plant layouts provided
the basis for design conditions selected. For the new plant application, design
conditions and layout were selected based on similar EPRI studies evaluating the
cost of flue gas desulfurization processes. Six study cases were evaluated in
this study, however, only Cases 2 to 5 are the subject of this paper and are
described in Table 1. General arrangement drawings for the study cases evaluated
in this paper are provided in Figures 1 to 4.

For the conventional hot-side SCR applications (reactor between economizer exit
and air heater inlet), the reactors were located above the particulate collection
device. In the post-FGD application, a wet FGD system precedes the reactors
which were placed above the heat recovery units (Gas-Gas-Heaters).

4B-82


-------
SCR Process Design. To obtain budgetary SCR system costs, a performance
specification for the catalyst and reactor was developed for each case. The
specifications were developed using fuel analyses, plant performance and
emissions data, and desired control of N0X, residual NH3, and byproduct S03.
Included in the specification was variation in certain process variables such as
N0X removal and ammonia slip for selected cases. Three SCR. system suppliers
provided quotations to these specifications.

The design basis and vendor supplied design data for each of the cases evaluated
in this paper are shown in Table 2. Sensitivity analyses are provided for the
new plant, hot-side design (Case 2) and the cyclone-fired, hot-side retrofit
design (Case 3), to show the cost and performance impacts of reducing the ammonia
slip from 5 to 2 ppmvd. For retrofit of hot-side SCR to a conventional
pulverized coal-fired boiler (Case 4), the effect of reducing the uncontrolled
N0X emission rate (by adding-combustion controls) while still meeting the same
N0X emission limit is evaluated; specifically, lowering uncontrolled NO, emission
rate from 0.60 to 0.40 lb NOx/MM Btu reduces the SCR NO, removal from 80% to 70%.

Consistent with typical practice, one reactor per air heater was used as the
design basis; the cyclone-fired retrofit (Case 3) uses a single reactor (1 x 100%
tubular air heater), while the other hot-side SCR cases utilize two reactors (2 x
50% trisector air heaters). The post-FGD case utilizes twin reactors because two
(2 x 50%) Ljungstrom heat recovery units were utilized.

The hot-side applications utilize downflow reactors, with additional capacity to
add a spare catalyst layer. Also, steam sootblowers are employed in the design
along with ash hoppers and ash transfer equipment. In the post-FGD application
the reactor is also designed as a downflow unit with capacity to add a spare
layer. The post-FGD reactor design does not require sootblowers and ash
collection hoppers.

The hot-side cases employ a catalyst with a 7.07 mm pitch (20 x 20 grid) while
the post-FGD case employs a catalyst with a 4.2 MM pitch (35 x 35 grid). The
lower pitch (higher specific area) and higher activity (per unit volume) of the
post-FGD catalyst allows a space velocity considerably higher than required for
the hot-side cases.

The ammonia storage and supply systems were designed using a truck unloading
station and a storage island providing seven days storage at an MCR rating.

Steam vaporizers are utilized for ammonia vaporization and dilution air is
provided from the discharge of the primary air fans in the hot-side cases, while
the post-FGD case utilizes separate dilution air fans.

SCR PROCESS IMPACT

The hot-side SCR process, because of its location directly downstream of the
boiler and upstream of the air heater, impacts every component of the flue gas
train and the boiler itself through its effect on the air heater (and in some
cases the economizer). The degree of impact varies with power plant
configuration, environmental control components, type of fuel, and emission
control requirements. The post-FGD SCR process impact is much less severe
because of its location at the end of the flue gas train.

Hot-Side SCR (coal) Impact

The impacts of hot-side SCR in coal-fired applications are summarized on Figure
5, The principal impacts are on the boiler, air heater and ID fan. Other

4B-83


-------
impacts are on the particulate collection device {ESP), wet limestone flue gas
desulfurization (FGD) process, FGD reheat system, waste disposal system and water
treatment system.

Boiler. The principal effects of hot-side SCR on the boiler will be the loss of
overall thermal efficiency, and additional operations and control complexity,
particularly for cycling units. Also, auxiliary power consumed by the SCR
process will reduce the net generating capacity.

Loss of thermal efficiency results from air heater modifications and an
economizer bypass which will result in higher air heater flue gas exit
temperatures. The result will be loss in the net generating capacity for the
same quantity of fuel consumption.

Air Heater. The potential for formation of ammonium sulfates and bisulfates
coupled with the presence of fly ash necessitates air heater modifications in the
hot-side SCR cases. Modifications to the air heaters in the PC boiler cases
include adding high pressure steam soot blowers at both the cold and hot ends,
adding high pressure water wash capability, replacing 24 gage heat transfer
surface material with 18 gage, replacing intermediate and cold end double
undulating (DU) heat transfer surface with notched flat (NF) surface, and adding
bypasses and dampers for on-line washing capability.

In the cyclone-fired boiler case, to reduce the rate of ammonium compound
deposition and build-up, all the existing 2" diameter tubes in the cold end, and
25% of the tubes in the hot end were replaced with 3" diameter tubes. Also, a
steam soot blowing system, utilizing medium pressure superheated steam at both the
hot and cold ends was added to reduce the rate of deposits.

In this case, it is expected that some residual ammonia may be captured by the
FGD system resulting in a build-up of ammonium species in the FGD liquor.

Although this may complicate scrubber sludge reuse or disposal, no cost impact
has been assigned.

Stack. The increase in the flue gas S03 concentration across the SCR could
result in increased opacity of the flue gas plume. Recent data from an EPRI
sponsored study with a member utility shows a direct correlation between stack
opacity and sulfuric acid concentration. To reduce opacity control measures may
be required to reduce the S03 concentration. A typical method of reducing S03 in
the flue gas would be to inject NH3 upstream of the ESP. The specific impacts or
costs associated with this effect have not been evaluated in this study, however.

ID Fan. To overcome additional pressure drop (up to 11" wc) associated with the
hot-side SCR, the existing ID fans were modified. For the retrofit cases it was
assumed that new, larger diameter wheels could be placed into the existing fan
housing to overcome the additional static pressure drop. The modifications
included replacing the fan wheel, shaft, bearings and motor.

ESP. SCR effects on the ESP include higher volumetric flowrate, higher negative
operating pressure, higher S03 concentration, higher flue gas temperature and
precipitation of ammonium compounds on fly ash.

Higher flue gas volume (an increase of up to 9.4% in the PC-fired cases) results
from higher flue gas temperature (20'F), lower flue gas static pressure, and
increased mass flow (the latter due to increases air heater leakage and dilution
air), and will have a significant impact on ESP operation. The increase in flue
gas volume will effectively reduce the Specific Collecting Area (SCA) and the

4B-84


-------
concentration of particulate in the flue gas. The result will be that the ESP
may require additional power to deliver the same particulate removal efficiency.

Greater negative operating pressure could require re-enforcement of the ESP.

This effect was not considered in the capital cost analysis.

In the high sulfur coal applications, the S03 concentration in the flue gas is
estimated to increase by 18 ppm across the SCR. Typically, an increase in S03
would be expected to reduce the fly ash resistivity significantly. However, the
increase in the flue gas temperature in the PC-fired cases (to keep the flue gas
above the acid dew point) may counteract the effect of the S03 increase, possibly
producing little net change.

Ammonium compound precipitation on the fly ash typically has a beneficial impact
on ESP performance by helping the fly ash agglomerate, preventing reentrainment.

The cumulative effects of all the above could be significant on an ESP', a pilot
test program would be required to.determine actual design and operations impacts.
In this study case it was assumed that the only net effect on the ESP operation
was an increase in power consumption by about 12%.

In the cyclone-fired boiler case the flue gas volume increase is expected to be
3.8%. This result is lower than the PC cases because of a negligible increase in
the leakage rate across the tubular air heater and only an 8°F flue gas
temperature increase at the air heater exit. Only a slight increase in the ESP
power consumption was assumed in this case.

Ash Disposal/Reuse. Ammonium compound content in the fly ash can have an impact
on waste disposal or marketing practices; for example, these compounds decompose
and release ammonia at elevated pH. While Eastern U.S. coals are not alkaline in
nature and ammonia would not be expected to gas off upon wetting, fixation with
alkaline species could result in an ammonia odor problem.

Similarly, reuse options for fly ash contaminated with ammonium compounds may be
limited. Direct use as an admixture in cement manufacturing may be jeopardized
if the ammonium compound content is too high.

FGD/Reheat. The chief effect on the F6D system is an increase in the water
evaporation rate and steam reheat requirement. The higher inlet temperature and
higher mass flow rate will result in an increase in water evaporation in the
absorber, as well as a significant increase in steam use by the FGD reheat system
(50°F reheat assumed).

A slight increase in power consumption could occur from having to increase the
FGD liquor recirculation rate in order to maintain the same S02 removal
efficiency. The higher liquor recirculation rate might be required as a result
of dilution of S02 in the flue gas, and higher flue gas volumetric flow rate
(saturated gas flowrate). This effect was not considered in this analysis.

FD Fan, In the PC-fired boiler cases (e.g. employing ljungstrom air heaters)' the
FO fan will consume slightly more power to account for a higher mass flow rate.
The mass flow increase results from an expected higher air heater leakage rate.

Water Treatment. Introduction of nitrogen species into the air heater wash water
requires additional water treatment equipment. Nitrogen species are introduced
into the wash water as ammonium bisulfate and sulfates. A biological treatment

4B-85


-------
process is utilized to convert the nitrogen species to free nitrogen. The
effluent is assumed to be discharged to the existing on-site water treatment
equipment,

Post-FGD SCR Process Impact

The impact of post-FGD SCR on power plant operations and equipment is less
significant than that expected with hot-side SCR, as the SCR reactor and
ancillary equipment follow all major process equipment. The impacts are shown by
Figure 6.

Boiler, The boiler is affected only insofar as auxiliary power consumption is
increased. The increase in the auxiliary power consumption (reduction in the net
generating capacity) will increase the Net Plant Heat Rate. Natural gas consumed
in elevating the SCR inlet gas temperature will also increase the NPHR.

ID Fan/Booster Fan, The increase in the flue gas pressure drop associated with
the post-FGD SCR process is estimated at 14.5 in w.c. The pressure losses are
principally across the inlet and outlet of the Gas-Gas-Heater (GGH) and the SCR
reactor. Addition of a booster fan into the flue gas train will increase the
complexity in flow and pressure control. In this case the booster fans are
located upstream of the stack; one booster fan is supplied for each SCR reactor
train.

Water Treatment. Nitrogen species will be introduced into the air heater wash
water as a result of ammonium bisulfate deposition on heat transfer surface.

With relatively little S03 capture expected within the FGD system, some
additional S03 generation across the catalyst, and the absence of fly ash, the
rate of chemical deposition on the GGH equipment is expected to be quite
significant. A biological treatment process was included to treat the
wastewater,

FGD. The SCR process affects the FGD system only indirectly. Because of the
location of the GGH, FGD system mist eliminator operation will be critical.
Excessive mist carryover could result in loss of heat recovery (resulting in
increased natural gas consumption) and an increase flue gas pressure drop,
possibly limiting generation capacity in addition to detracting from plant heat
rate.

Stack. Retrofit of the post-FGD SCR process will almost certainly have an impact
on the stack. If the original plant design included a wet stack, the 225"F GGH
exit gas temperature will require liner replacement. In this design case it was
assumed that the original design included steam reheat (50°F) and that the stack
was designed for approximately 180*F, The effect of the increase in the flue gas
temperature to 225"F was considered negligible.

Higher S03 concentration in the flue gas may result from oxidation of S02 across
the catalyst. While some of the S03 is likely to form ammonium/sulfur compounds
and deposit on the GGH surface, there may be a net increase in the S03
concentration which could increase plume opacity.

COST DEVELOPMENT

To develop total process capital costs, physical layouts of the ductwork and SCR
reactors were developed. From these drawings, lengths of ductwork and structural
requirements were estimated. All costs are presented in December 1989 dollars.

4B-86


-------
The operating and capital cost impact of SCR on other plant components was also
estimated. For major pieces of equipment, such as the air heaters, ammonia
storage system and ID fans, vendors were consulted in developing the cost of the
modifications. For smaller equipment items and piping runs, UE&C utilized in-
house data to arrive at equipment costs.

EPRI's Technical Assessment Guide (TAG) provided the basis to estimate fixed
operating and maintenance costs. Variable operating costs were determined by
calculating utility and raw material consumption rates. Considered in the
variable operating costs were the following:

•	SCR catalyst replacement

•	Ammonia consumption

•	Ammonia vaporization steam

•	Incremental Sootblowing steam

•	Incremental ID/Booster fan horsepower consumption

•	Incremental FD fan horsepower consumption
t	Incremental ESP power consumption

•	Water treatment chemicals

•	Air heater efficiency loss

•	Incremental FGD reheat steam consumption

•	SCR catalyst disposal

•	Incremental fly ash disposal cost

•	Natural gas consumption

RESULTS

Selected results from this study are summarized in Figures 7 to 10, while
sensitivity of results to catalyst cost and life are provided in Figures 11 to
14. Highlights are discussed as follows:

Capital Costs. Total capital requirement (TCR) for each of the cases is
presented, indicating the contribution of the reactor/catalyst, structural
modifications and/or support equipment, air heater, ductwork, NH3 injection, flue
gas handling, and contingencies. Figure 7 shows capital cost is least for new
units, due to the absence of retrofit considerations, and reduced catalyst
quantity from lower boiler exit N0„ emissions. These same factors, retrofit
considerations and boiler exit N0X emissions, are responsible for the cyclone
boiler having the highest cost for the hot-side application. Post-FGO capital
cost is high due to the GGH, which adds significantly more cost than is saved
through simplifying reactor design and reduced catalyst quantity.

Decreasing the ammonia slip from 5 to 2 ppm (shown for both cases 2 and 3) is
expected to increase the TCR by about 12% due to a larger catalyst volume
requirement.

The cost impact on the SCR of reducing the boiler NO, emission rate from 0.60 to
0.40 lb N0./MM Btu is shown by Case 4,0 and 4.1. Reduction of the boiler NO,
emission rate (through combustion modifications), while meeting the same emission
limit of 0.12 lb N0„ /MM Btu, reduces the SCR capital cost by 59.4/kW.

(Levelized costs reflecting both capital and operating costs must be compared to
judge the full benefit.)

The catalyst and reactor cost represents about 40-50% of the TCR in the hot-side
SCR cases. In the post-FGD SCR case, the catalyst cost represents only about 17%
of the TCR. The largest cost item in the post-FGD SCR case are the twin GGH's
used for heat recovery.

4B-87


-------
The contingency ranges from 14.4% to 18.2%. The highest contingency is assigned
to Case 3 due to uncertainties in high sulfur coal applications, coupled with
tubular air heaters and a very high boiler NGX emission rate.

levelized Cost. Figure 8 presents levelized costs for the same design cases,
depicting generally the same trends between costs for new units, cyclone boilers,
conventional PC boiler, and post-FGD application. The data shows that variable
operating costs and fixed charges represent about 50% of total levelized cost for'
the hot-side application. The most significant component of fixed charge is the
recovery of capital for the reactor and catalyst. Similarly, the most
significant component for variable Q&M is catalyst replacement cost. Comparison
of cases 4.0 and 4.1 shows the benefit of adding combustion controls to reduce
the N0X reduction requirement of the SCR; the results indicate that the SCR cost
can be reduced from 6.54 to 5.88 mills/kWh by reducing the boiler emission rate
from 0.60 to 0.40 lb N0X/MM Btu. In the case of the post-FGD SCR process, fixed
charges represent about 65% of the total levelized cost. ' Note that the results
consider a 0.93 mil 1s/kWh credit for a 50°F steam reheat system that is no longer
required upon retrofit of the post-FGD SCR process. This credit would, of
course, not apply for units that employ Wet stack operation.

The levelized costs for Case 3.0, the cyclone boiler, are significantly higher
than the costs expected with retrofit to a PC-fired boiler. This is due both to
higher capital requirement and catalyst replacement cost due to the large volume
of catalyst required in this application.

Figure 9 shows levelized costs in terms of $/ton NQX removed. Primarily, the
data shows the impact of the boiler N0X emission rate on the cost to remove a ton
of N0X. The cyclone-fired boiler (Case 3.0) shows the lowest levelized cost
(about $1,100/ton N0X ). Although the cost of SCR for application to cyclone
boilers is significant, the high uncontrolled boiler NQX emissions reduce costs
on a per ton basis.

The highest levelized cost is shown by Case 4.1 where combustion controls were
added to reduce the SCR N0X reduction requirement from 80% to 70%. Lowering the
boiler exit N0X emission rate correspondingly increased costs on a per ton basis.

Figure 10 provides a more detailed cost comparison between a post-FGD and hot-
side SCR process in terms of levelized costs (mi 11s/kWh), The power plant, fuel,
and N0X. reduction performance is identical for both cases. The levelized costs
for the two process options are comparable, however, as described earlier, the
reheat credit of 0,93 mills/kWh for the post-FGD process may not be applicable to
specific sites if a wet stack is used. Also, note that a 4-year catalyst life
was used in the post-FGD cost development, six years is closer to the currently
expected life. Catalyst replacement is the most significant O&M cost item for
the hot-side process, while natural gas cost (and heat rate penalty) is the most
significant O&M cost item for the cold-side process.

Effect of Catalyst life and Unit Costs. Sensitivities of the cost results to
both catalyst cost and life are provided by Figures 11 to 14. Base case
economics were developed assuming a four year catalyst life for both hot-side and
post-FGD SCR processes; a six year catalyst life for the post-FGD SCR is now
being predicted. Base case catalyst cost of $660/ft3 was utilized; this cost
reflected budgetary quotations from the primary U.S. SCR catalyst vendors with
coal-fired experience. It is possible that catalyst costs will approach those in
Europe ($40G-45G/ft3) due to world market competition.

4B-88


-------
The figures show that the SCR applications which require the largest quantity of.
catalyst are most sensitive to both catalyst life and cost. The post-FGD process
(Case 5) is the least sensitive due to its relatively small catalyst charge.

CONCLUSIONS

Conclusions developed from this study are;

•	The capital cost of SCR in 500 MW (nominal) size U.S. plants is expected
to be;

A.	$96 - S105/kW for hot-side retrofits to conventional (tangential or
wall) coal-fired power plants.

B.	$125 - $140/kW for hot-side retrofits to cyclone-fired boilers.

C.	$78-87/kW in new plant hot-side applications.

0. S140/kW for post-FGD retrofits.

t The levelized cost of SCR in U.S. coal-fired power plants (500 MW size
range) is expected to be:

A.	5.3-5.9 mills/kWh for new hot-side power plant applications.

B.	5.9 to 6.5 mills/kWh for hot-side retrofits to conventional-fired
units.

C.	8,2 to 9.1 mills/kWh for hot-side retrofits to cyclone-fired units.

D.	Approximately 6.8 mills/kWh for post-FGD retrofits to con-
ventional units assuming a credit for reheat (0.93 mills/kWh).

§ The levelized cost of removing a ton of NO, utilizing SCR is expected to
range as follows:

A.	$3,300 - $3,800/ton NQX for new coal-fired plant hot-side
applications.

B.	$1,100 - $l,250/ton N0X for coal-fired cyclone boiler hot-side
retrofits.

C.	$2,750 - $4,250/ton N0X for coal-fired conventional boiler hot-side
retrofits.

D.	52,850/ton N0X for post-FGD SCR retrofit to a conventional boiler.

•	The levelized cost of removing a ton of N0„ is lowest with high N0X
emission rates. The levelized cost of removing a ton of N0X for a
cyclone-fired boiler with a 1.80 lb N0„/MM Btu NO, emission rate is
estimated at $1,100/ton N0E,

•	The SCR capital cost in a new power plant application is substantially
less than in a retrofit application. The cost of a new plant SCR is
expected to be about 34% lower than a retrofit, the lower cost is due
largely to new boilers having lower N0„ emission rates and an attendant
reduced catalyst requirement, and the absence of costly existing
equipment modifications required in SCR retrofit applications.

•	SCR capital costs are higher for cyclone-fired boilers because of their
high NQ„ emission rate. The SCR capital cost for cyclone-fired units is
expected to be about 45% higher than that expected for conventionally-
fired power plants.

•	Catalyst life and catalyst unit cost significantly affect levelized
process costs. For most hot-side SCR applications, an increase in
catalyst life from 2 to 4 years reduces levelized cost by 30%. A
reduction in catalyst unit cost from $660/ft3 to $450/ft (for cases
assuming a four year catalyst life) reduces levelized costs by 15%.

4B-89


-------
• The levelized cost of N0„ removal for both hot-side and post-FGD SCR

processes is similar, but the components of the cost vary significantly.
Compared to hot-side SCR, post-FGD applications requires 30% more
capital, but feature lower catalyst replacement costs.

REFERENCES

1.	Bauer, T. K., Spendle, R. 6,, "Selective Catalytic Reduction for Coal-Fired
Power Plants: Feasibility and Economics," Steams-Roger Inc., EPRI CS-36G3,
October 1984.

2.	Cichanowicz, J. E., Offen, G. P., "Applicability of European SCR Experience
to U.S. Utility Operation," Proceedings: 1987 Joint Symposium on Stationary
NO. Control. EPA/EPRI, New Orleans, 1987.

3.	Cichanowicz, J. E. et. al., "Technical Feasibility and Economics of SCR NOx
Control in Utility Applications," Proceedings: 1989 Joint Symposium on
Stationary Combustion N0r Control, EPA/EPRI, March 1989.

4.	Electric Power Research Institute (EPRI), TAG - Technical Assessment Guide.
Volume I: Electricity Supply - 1986. EPRI P-4463-SR, December 1986.

5.	Ellison, W., "Assessment of S02 and N0X Emission Control Technology in
Europe," EPA-600/2-88-013, February 1988.

6.	Nakabayashi, Y., Abe, R., "Current Status of SCR in Japan," Proceedings: 1987
Joint Symposium on Stationary NO.. Control, EPA/EPRI, New Orleans, 1987.

7.	Necker, P., "Operating Experience with the SCR DeNOx Plant in Unit 5 of
Altbach/Deizisau Power Station," Proceedings: 1987 Joint Symposium on
Stationary NO.. Control, EPA/EPRI, New Orleans, 1987.

8.	Osborn, H. H., "The Effect of Ammonia SCR DeNO, Systems on Ljungstrom Air
Preheaters," C-E Air Preheater, EPRI RP 835-2, June 1979.

Table 1
Case Definition

PLANT DESCRIPTION

Case	2.0

Retrofit	No

Capacity, MW (gross}	546,600

Boiler Type	PC

Air Heaters	ljungstrom

Particulate Control	Baghouse

502 Control	Wet FGD

Reheat	yes

Gross Plant Heat Rate, Btu/kwh	9,13?

Capacity Factor, %	65

Remaining Life, years	30

3.0

4.0

5.0

Yes

Yes

Yes

536,000

536,000

536,000

Cyclone

PC

PC

Tubular

Ijunastrom

Ljungstrom

ESP

ESP

ESP

None

Wet FGD

yet FGO

NO

Yes

Yes

9,974

9,197

9,197

65

65

65

20

20

10

SITE CONDITIONS

Location	Kenosha, HI	Kenosha,	bl Kenosha, WI	Kenosha, wi

Seismic Zone	I	I	II

Urban site	No	No	No No

FUEL

Type	Coal	Coal	Coal	Coal

Area	Illinois No. 6	Illinois No. 6	Appalachian	Appalachian

Higher Heating Value, Btu/lb	10,533	" 10,533	13,100	13,100

Sulfur Content,' wt. %	3.74	3.74	2.60	2.60

Ash Content, ut. X	9.51	9.51	9.10	9.10

4B-90


-------
CASE NUMBER
DESCRIPTION

SCR DESIGN BASIS
Boiler Type

Economizer Outlet Temp. 3MCR, °F
Economizer Excess Air, %

Boiler NOx Emission Rate, Ib/MM Btu
NOx Concentration, ppmv (actual)
NOx Emission Limit, Ib/MM Btu
NOx Reduction Rate, X
NH3 Slip Rate, ppmvd (3 3% 02)
Guaranteed Catalyst Life, years
Reactor Configuration

Ammonia Storage, days

SCR DESIGN

Space Velocity, SCF*/ft3-hr
Linear Velocity, actual fps
Operating Temperature, °F
S02 Oxidation rate, %

Catalyst Geometry
Surface Area, m2/m3
Pitch, mn

Catalyst Layers (active + spare)
Soot Blowers

Amnonia Consuiption, Ib/hr
Gas-Gas-Heater (GGH)

-Number

-Untreated Gas In/Out, °F
-Treated Gas In/Out, °F

SCR COST DEVELOPMENT
Catalyst Cost, $/ft3
Expected Catalyst Life, years
Amnonia Cost, $/ton
Natural Gas Cost, $/MM Btu
Plant Life, years
Capacity Factor, %

* SCF 3 32°F

-P>.
CD

(b

2.0
New,
hot-side

PC
725
24
0.40
364
0.08
80
5
2

Twin,
Vertical
7

2,750
18.2
725
1.10
Grid
470
7.07
4 + 1
Yes
941.4

NA
NA
NA

660
4

145
NA
30
65

Table 2
SCR Process Design

2.1

3.0

3.1

4.0

4.1

5.0

New,

Retrofit,

Retrofit,

Retrofit,

Retrofit,

Retrofit,

hot-side,

hot-side

hot-side

hot-side

hot-side

cold-side

PC

CycI one

Cyclone

PC

PC

PC

725

682

682

725

725

NA

24

20

20

24

24

24

0.40

1.80

1.80

0.60

0.40

0.60

364

1700

1700

572

381

428

0.08

0.36

0.36

0.12

0.12

0.12

80

80

80

80

70

80

2

5

2

5

5

5

2

2

2

2

2

2

Twin,

Single,

Single,

Twin,

Twin,

Twin,

Vertical

Vert ical

Vertical

Vertical

Vertical

Vertical

7

7

7

7

7

7

2,300

1,800

1,500

2,530

2,960

6,000

18.2

18.2

18.2

18.2

18.2

22.0

725

682

682

725

725

625

1.20

1.10

1.10

1.20

1.20

0.39

Grid

Grid

Grid

Grid

Grid

Grid

470

470

470

470

470

795

7.07

7.07

7.07

7.07

7.07

4.2

4 + 1

6 + 1

6 + 1

4 + 1

4 + 1

2 + 1

Yes

Yes

Yes

Yes

Yes

No

932.3

4,478

4,468

1,383

807

1,383

NA

NA

NA

NA

NA

2 X 50%

NA

NA

NA

NA

NA

129/550

NA

NA

NA

NA

NA

625/226

660

660

660

660

660

660

4

4

4

4

4

4

145

145

145

145

145

145

NA

NA

NA

NA

NA

2.98

30

20

20

20

20

20

65

65

65

65

65

65


-------
PLAN

Figure 1. Case 2 Plan and Elevation General Arrangements.

4B-92


-------
N

Figure 2. Case 3 Plan and Elevation General Arrangements.

4B-93


-------
TO* El iSOf-O1

Figure 3. Case 4 Elevation General Arrangement.

4B-94


-------
AIR HEATER

•	Ammonium Blsulfate •
Fouling	a

•	Higher Exit Gas Temp. ,

•	Higher Leakage

•	Higher AP

•	Higher Steam Sootblow Rate

•	Higher Water Wash Rate

•	Higher Steam pressure &
Superheat

•	Additional Dampers For
On-Llne Wash

BOILER

NPHR Increase
Temp. Bypass
Reduced KW

FD FAN

•	Higher Mass Flow

•	Provide Dilution Air

•	Higher Hp Consumption

AMMONIA STORAGE

•	Operator Training
& safety

WATER TREATMENT

•	Treat AH Wash For
Nitrogen

ESP

Higher Inlet Gas Volume
Higher Gas Temp.
SOa NH3 Conditioning
Higher AP
Resistivity Affected

FLY ASH
Marketability Impact
Odor Problems
Additional Equipment
For SCR

BOILER

-fc.
03

CO
cn

1 AIR HEATER

-? nh3

ID FAN

Higher Mass Flow
Higher Volumetric Flow
Higher AP

REHEAT

Higher Mass Flow
Increased Steam Usage

FGD
Volume Increase
Higher Inlet Temp.

Increase In H20 Evap.
SO2 Concentration Dilution
FGD Wastewater Treatment
For NH3

STACK

Increased Opacity
Higher S03

J

j AH WASH

CbQ

WASTE TO
DEWATERING

TO EXISTING

FP FANS ||	||	W & WM SYSTEM

WATER TREATMENT

Figure 5. Hot-side SCR Design/Operations Impact.


-------
j PLANT

NPHR Increase

Reduced Kw

Natural Gas Supply
Required

Additional Plant
Complexity

WATER TREATMENT

» Treat GGH Water Wash
for Nitrogen
Compounds

FGD

Mist Eliminator
Operation Critical

AMMONIA STORAGE

• Operator Training & Storage

STACK

•	Higher S03

•	Increased Temperature

•	Increased Opacity

•	Increased Volume

BOILER



Alfl
HEATER

ESP

YW

FLY ASH

ID FAN

FGD

V

rt

STACK

GAS-GAS
heater

£H4

PUCT

BURNER

^NHi

| GGH WASH

DILUTION
AIR FAN

FD FAN

4m-

TO EXISTING
W 8l WM SYSTEM

WATER TREATMENT

Figure 6. Post-FGD SCR Design/Operations Impact.


-------
Total Capital Requirement, $/kW

COST ITEMS

irnno

CONTINGENCY

Emm

IWtHH

FACIUTIES,ENQ.,F EE

CD

ID FAN.WSWM,REHEAT

¦ cm

AIR HEATER/GQH

m

STRUCTURAL

o

DUCTWORK

»

NH3 STORAGE



REACTOR/CATALYST .

2,0 2.1 3,0 3.1 4.0 4.1 5.0
CASE

Figure 7, Total Capital Requirement.

Mills/kWh

COST ITEMS
EZ3 FIXED CHARGES
VARIABLE'O&M
FIXED O&M

Figure 8, Levelized Cost (mills/kWh).

4B-97


-------
$/ton NOx Removed iThouearids)

381 t

3430

I

n

2.0 2.1

1227

1103



4255

2766

If
I

3.0 3,1

CASE

4.0 4,1

26G7

MS®
5.0

COST ITEMS
LU FIXED CHARGES
IIII VARIABLE O&M
¦I FIXED O&M

Figure 9. Levelized Cost ($/ton NOx removed).

Mills/kWh

COST ITEMS

FIXED CHARGES
EFFICIENCY LOSS/GAIN

~	W4WM,STEAM .
[ZD NATURAL OAS
HB POWER

~	CATALYST
AMMONIA
FIXED O&M

CASE 4.0	CASE 5:0

Figure 10. Hoi-Side vs Post-FGD SCR Cost Comparison.

4B-98


-------
S/ton NOx Removed (thousands)

012345678

Catalyst Life, years

—— Case 2 —s- Case 3	Case 4	Case 5

Figure 11. Levelized $/ton NOx versus Catalyst Life.

Levelled Mills/kWh

2	3	4	5

Catalyst Life, years

Case 2

Case 3

Case 4 -e- Case 5

Figure 12. Levelized Mills/kWh versus Catalyst Life.

4B-99


-------
Levefized Mills/kWh

— Case 2 ~4— Case 3	Case 4 —B~ Case S

Figure 13. Levelized Mills/kWh versus Catalyst Cost.

Catalyst Cost, $/ft3

—¦— Case 2	Case 3	Case 4	Case 5

Figure 14, Total Capital Requirement versus Catalyst Cost.

4B-100


-------
APPLICATION OF COMPOSITE NOx SCR CATALYSTS IN
COMMERCIAL SYSTEMS

B.K. Speronello, J.M. Chen, M, Durilla, R.M. Heck

Engelhard Corporation
101 Wood Avenue South
Iselin, NJ 08830

4B-101


-------

-------
Abstract

Composite NOx SCR catalysts have been installed in a variety of
commercial NOx control systems, including coal fired power plants,
gas turbines, stationary engines, and chemical plants. This paper
reviews how such catalysts performed in these systems, and it relates
key features of the coinposxte catalyst d.esign to. catalyst
performance. These data illustrate how composite SCR catalysts can
cut catalyst volume and reactor size by over 50%- (relative to
conventional SCR catalysts), with no loss of NOx removal efficiency.

Background

The term composite honeycomb catalyst refers to a catalyst design
strategy where a layer of catalytic material is bonded to a strong,
thin-walled honeycomb support. This design has been used for years
to treat exhaust from a variety of sources, including; stationary
internal combustion engines1, gas turbines^, chemical
processes^, and, most notably, automobiles4'. 'In 1984, Engelhard
developed a composite catalyst for the selective catalytic reduction
(SCR) of NOx by ammonia5. This catalyst contains a catalytic layer
of V20s/Ti02 (V/Ti) supported on a cordierite ceramic honeycomb.
Recently, a composite zeolite SCR catalyst was developed to extend
the maximum operating temperature for the SCR reaction from nominally
450°C up to 600°C.

To date, composite catalysts have been demonstrated in 10 pilot tests
{see Table I) and 12 commercial installations (see Table II). In
addition, Table II lists another 12 commercial installations in
varying stages of design and construction. Overall, the treated
flows range from 3 lb/sec for a small chemical process to 650 lb/sec
for a 50 MW gas turbine. -This paper discusses some of the reasons
why the composite catalyst design was chosen for SCR, and summarizes
some of the experience- that has been developed with composite SCR
catalyst. In particular it focuses on factors pertinent to the
application of these catalyst to the exhaust from power generating
installations.

4B-103


-------
Discussion

Figure 1 illustrates the basic structure of a composite V/Ti SCR
catalyst, and compares it with a conventional SCR catalyst made by .
extruding catalytic material into the honeycomb shape. The composite
design is appropriate whenever mass transfer factors, such as
boundary layer diffusion or pore diffusion, limit the penetration of
reactant gases to a thin layer at the catalytic surface. This is the
case with NOx SCR.

The benefits of the composite design include:

1.	High strength due to the strength of the underlying
structural ceramic;

2.	, Thin walls made possible by the support's high

strength;

3.	High geometric surface area at constant pressure drop
due to the thin walls;

4.	Excellent abrasion resistance due to the hardness of
the ceramic support;

5.	High activity due to high geometric surface area and
qreater porosity within the catalytic layer;

JL	tt	»	-A	¦

6.	Wider temperature range of operation due to better NOx
mass transfer characteristics;

7.	Inherently low SO2 oxidation activity, and

8.	Contains 85% less heavy metals.

Composite SCR catalysts can be made with walls as thin as 12
thousandths of an inch (0.30 mm) compared to ca. 27 thousandths (0.65
mm) for conventional extruded SCR catalysts. Because they have
thinner walls, composite SCR catalysts can have much smaller openings
(and consequently higher geometric surface area) than conventional
catalysts. Figure 2 illustrates the benefits of thinner walls and
increased geometric surface area for increasing catalyst activity and
reducing catalyst volume and reactor size. Figure 2 compares the the
amount of SCR catalyst that would be needed to achieve either 80% or
90% NOx reduction using catalysts of different cell densities (ie.
channel opening sizes measured in cells per square inch, CPSI). For
example, in a design to achieve 80% NOx reduction at a pressure drop
below 3 inches v/ater column, a conventional 40 CPSI catalyst would
require a relative catalyst volume of 3.3. In comparison, a 200 CPSI
composite catalyst provides the same performance with 65% less
catalyst (a relative- catalyst volume of only 1.2).

Figure 3 illustrates another benefit of smaller channel openings,
wider range of operating temperature. It compares the curve of NOx
conversion vs. temperature for a 200 CPSI composite SCR catalyst
with a catalyst having a more conventional cell density, 25 CPSI.
The 200 CPSI composite catalyst was tested at much higher flow rate
than the 25 CPSI catalyst so that both provided similar NOx
conversion at temperatures below 350°C (660°F). As temperature was
increased, both catalysts exhibit the conversion peak and subsequent
drop off that is characteristic of the onset of excessive ammonia
oxidation in V/Ti SCR catalysts. However, the peak for the composite
catalyst is over 5 conversion points and 40°C higher than the 25 CPSI

4B-104


-------
catalyst. 'In addition, at 450°C (840°F), the normal maximum
continuous operating temperature for V/Ti, the 200 CPSI catalyst
still provided over 85% NOx conversion, while the 25 CPSI catalyst¦
had fallen to only 70%.

Figure 4 explains why composite catalysts demonstrate inherently low
SO2 oxidation activity. It is a graph- of SO2 conversion vs.
relative catalyst loading (in gram catalyst/in^ of catalyst, relative
to a base loading). A medium cell density composite SCR catalyst has
a relative catalyst loading of IX, while a conventional extruded SCR
catalyst has a typical loading of 10X on this scale. Because of the-
low SO2 oxidation activity of composite SCR catalysts, this test was
run under exceptionally severe conditions to xncrease oxi.dati.oTV
sufficiently to allow for precise measurement. SO2 oxidation in
commercial operation would be significantly lower than shown in
Figure 4.

At conversion levels of <. ca. 20%, the rate of SO2 conversion is
controlled by the number and potency of the active sites for SO2
oxidation. Under this rate limiting condition the extent of SO2
oxidation to SO3 increases in proportion to the amount of catalytic
material in the catalyst. As a result, composite catalysts have
inherently low SO2 oxidation activity, because they,contain only
about 10% as much V/Ti as extruded.SCR catalysts. To compensate for
this deficiency, the SO2 oxidation activity of extruded catalysts is
suppressed by incorporating SO2 oxidation demoters into the catalyst
formulation. These additives, however, are also reported to suppress
NOx removal activity6.

The relative hardness of the ceramic honeycomb support results in
excellent erosion resistance for composite catalysts in high dust
environments,. This characteristic, and catalytic performance in
several pilot and commercial installations are discussed in the
following sections.

Coal Fired Power Plants
High Dust, Hot Side:

The first experience with composite SCR catalyst for NOx control in a
high dust, hot side coal fired power plant environment began in 1988.
Small portions of the conventional SCR catalyst beds of two operating
coal fired power plants in Germany were replaced with composite
catalyst. One was on a dry bottom boiler, and the other a wet bottom
boiler. Pertinent operating characteristics are included in Table
III. Of particular interest were the dust loadings, which ranged up
to 15 g/Nm-3 {ca. 7 grain/DSCF) . Each SCR system contained 2 layers
of catalyst; each 1 meter deep. Several sleeves of composite'SCR
catalyst (25 CPSI, 5.1 mm cell pitch) were mounted in each catalyst
layer.

A 6 inch long block of catalyst was mounted at the inlet of each
layer to test for possible erosion. These inlet blocks were removed
after 18 months and returned to Engelhard for erosion analysis. The
remaining catalyst is continuing in operation.

4B-105


-------
Two types of erosion were evaluated; axial (ie, length) erosion of
the catalyst block and wall erosion (ie. thinning of the catalytic
layer). Neither type was detected.

Axially, there was no change in the length of the catalyst blocks,
nor was there evidence of honeycomb wear to the sharp edges at the
inlet face of the catalyst. The hard ceramic support completely
resisted abrasion in this environment.

Wall erosion was measured' by electron microscopy. Three samples were
cut from each aged catalyst block plus a fresh control. One sample
was taken 'from the center of the block, and the other two were taken
from spots located ca. 1 inch in from each end. These were mounted,
polished to reveal a cross section of each wall, and 5
photomicrographs were taken of each sample. These showed that there
was no evidence of erosion of the catalytic layer. There was no
difference in wall thickness between the aged and fresh catalysts,
and no correlation between depth,of the catalytic layer and sample
location within a block. Figure-5 shows representative
photomicrographs comparing aged catalyst with the fresh control. The
lighter layer near the center of each photo is the catalytic V/Ti,
and the darker material beneath it is the ceramic support. The depth
of the catalytic layer was -unaffected by aging.

These results are consistent with prior observations showing that
flow straighteners and hardening of the catalyst inlet face could,
control erosion of the SCR catalyst, and there' was no report of wall
thinning away from the inlet'face7.

Analysis at Engelhard indicates that catalyst erosion is limited to a
transition zone from turbulent to-' laminar flow at the inlet to the
honeycomb channel. The depth of this transition zone can be affected
by factors such as gas velocity and angle of incidence of the gas
with respect to the catalyst channel, but once the- flue gas develops
into full laminar flow {always within millimeters of the inlet face
of the catalyst) there is little interaction between abrasive
particles and the catalyst wall. Consequently there is negligible
wall thinning within the block. With composite catalysts, the hard
ceramic substrate resists erosion at the inlet, so, in addition to no
wall,thinning, there is also no axial erosion.

While the main purpose of this experiment was to study catalyst
erosion in the high dust environment, the SCR performance of these
aged inlet samples were also measured. The results are shown in
Figure 6. It is a graph of NOx conversion as a function of 'tem-
perature for aged catalyst taken from the SCR beds of both power
plants. Considering that the samples represent the most contaminated
and deactivated portion of the bed, the inlet 6 inches, it is notable
that both provide over 80% NOx reduction and little or no ammonia
slip at a typical commercial flow rate.

Medium Dust, Cold Side:

Composite SCR catalysts have been, and are continuing in pilot plant
tests on several coal fired power plants downstream of the flue gas
desulfurization systems. Figure 7 shows the results for a pilot

4B-1Q6


-------
reactor system that used a 20'0 CPSI V/Ti composite catalyst to
determine if catalyst with a pitch of less than 2 nun could operate in
that environment without plugging. Table IV summarizes pertinent
operating conditions. The test included a total of 2500 on-stream
hours (including 40'pilot plant start-ups and shutdowns)At
20, 000 ,1/hr VHSV and 350°C, NOx conversion averaged 92% with <5 ppm
ammonia slip. For this'application, the particulate concentration
(primarily gypsum, CaCC>3 and Si02) averaged about 50 mg/Km^ (0.02
grains/DSCF) with peaks of up to about 120 mg/Nm-* (0,05
grains/DSCF). p The inlet NOx concentration varied between 350 and 420
ppm. No apparent decline of catalyst performance-was seen throughout
the test period.

Since the channel size of the 200 CPSI catalyst was significantly
smaller than typically used for this dust level, provision was made
for soot blowers to prevent catalyst plugging. Initially the soot
blowing frequency was every 2 hours. Over the first 800 hours of
operation the soot blowing frequency was steadily reduced with no
increase in pressure drop across the catalyst. After 800 hours, soot
blowing was stopped completely. Pressure drop- rose slightly and
stabilized at a 15% increase from the original level. While every
installation is different, these results indicate that it is possible
to operate high cell density SCR. catalysts at moderate dust levels
without plugging.

Figure 8 shows on-line NOx reduction efficiency and ammonia slip
versus feed NH3/NOX ratio during this pilot test. It shows that by
using a 200 CPSI catalyst it -was possible to achieve >92% NOx
conversion with <5 ppm NH3 slip at with relatively little catalyst
(20,000 1/hr VHSV).

Aging of these catalyst modules plus several others containing lower
cell density catalyst was continued in the SCR reactor downstream of
the FGD unit of another coal fired boiler. Dust levels averaged 10
mg/Nm- (0.005 grains/DSCF), and inlet NOx ranged between 550 and 600
ppm. To date the catalysts have accumulated a total of 15,800 hours
of operation. Laboratory activity tests made at 60,000 1/hr VHSV on-
core samples taken at 0, 2500 and 7540 hours showed no decline of
catalyst activity from the fresh level (see Figure 9). Testing after
nearly 16,000 hours shows what may be a slight decline in conversion,
but even these results are within the range of normal test
variability.

These data, demonstrate that this composite SCR catalyst with a high
cell density can be applied successfully to boiler exhaust even where
the presence of particulates is significant. This can represent a
several fold reduction in catalyst volume for achieving the same high
NOx performance efficiency as conventional catalyst designs of lower
cell density.

Stationary Engines

Composite V/Ti based SCR catalyst technology has also been applied to
a wide range of stationary engine applications. Engine sizes have
ranged from 210 to 3900 horsepower. Fuels have been natural gas,
digester gas, and #2 diesel fuel. Ammonia control strategies have.

4B-1D7


-------
ranged from manual adjustment to continuous monitoring with control
by an in-plant host computer. In each application, the required NOx
conversion and NH3 slip performance limits were achieved.

Of particular interest are the results from one application which
utilized #2 diesel fuel. This application was operated on 300 CPSI
(1.5 mm pitch) composite catalyst as a demonstration unit by the EPA
at their Air and Energy Research Laboratory at Research Triangle Park
in North Carolina®. The demonstration was run for 4000 hours in
approximately 100' hour increments. Particulate levels in the exhaust
ranged from' 27 mg/Nx^ (0.013 grains/DSCF) during steady state
operation to up to 100 mg/Nm^ (0.05 grains/DSCF) during each start-
up. There was occasional evidence of increased pressure drop due to
accumulation of wet soot on the ultra-fine pitch catalyst during
periods of frequent repeated cold starts. As a result, the catalyst
was manually air la-nced four times during the demonstration. After
each cleaning, 85% NOx conversion was maintained. There was no
evidence of particulate buildup during periods-of continuous
operation.

This test demonstrated that ultra-fine pitch composite SCR catalyst
could provide continuous steady state NOx emission control on a
diesel engine operating on #2 fuel at a substantial reduction in
catalyst volume relative to conventional catalysts.

Conclusions

The composite catalyst design strategy using ceramic supports offers
several unique advantages, for NOx selective catalytic reduction.

They include greater mechanical strength, exceptionally high
activity, excellent erosion resistance, and inherently low SO2
oxidation activity. These benefits have been demonstrated in both
extensive pilot scale testing and commercial installations.

ACKNOWLEDGEMENT

The information included in this paper was generated in collaboration
with several Engelhard colleagues, including: Dr. J.W. Byrne, Mrs.
C.Hirt, Mr. J.Hansell, and Mr. M.Tiller.

4B-108


-------
1	K.Burns, M.Collins, R .K.Heck, Catalytic Control Of NOx
Emissions' From stationary Rich-Burning Natural Gas Engines,
ASME 83-DGP-12, 1983

2	J.M.Chen, R.M.Heck, K.R.Burns, M.F.Collins, Development Of
Oxidation Catalyst For Gas Turbine Cogeneration Applications,
82nd Annual Meeting And Exposition Of The Air Pollution
Control Association, June 1989, Anaheim, California

3	R.M.Heck, M.Durilla, A.G.Bouney, J.M.Chen, Ten Years 1
Operating Experience With Commercial Catalyst Regeneration,
81st Annual Meeting And Exposition Of The Air Pollution
Control Association, June 1988, Dallas, Texas

4	j.j.Mooney, C.E.Thompson, J.C.Dettling, Three Way Conversion
Catalyst - Part Of'A New'Emission Control System, SAE 77-
0365, 1977

5	R.M.Heck,*J.C.Bonacci, J.M.Chen, Catalytic Air Pollution
Controls - Commercial Development Of A Catalyst For The
Selective Catalytic Reduction Of NOx, 80^ Annual Meeting And
Exposition Of The Air Pollution Control Association, June
1987, New York, New York

6	J.Ando, "NOx Abatement For Stationary Sources In Japan", EPA
600/7-83-027(1983)

7	L.Balling, D. Hein, DeNOx Catalytic Converters For Various
Types Of Furnaces And Fuels - Development, Testing,

Operation, 1989 EPA/EPRI Joint Symposium On Stationary
Combustion NOx Control, March 6-9, 1989, San Francisco,
California

8	D.Dreschler, presented at the German VGB Conference, Feb.

1986

9	J.H.Wasser, R.B.Perry, Diesel Engine NOx Control With SCR,¦
EPA/EPRI 1985 Joint Symposium On Stationary NOx Control, May
1985, Boston, Mass.

4B-109


-------
Figure 1:

Two Different NOx SCR Catalyst Designs

Composite

Extruded Vanadia/Titania

Strong, _
Thiri Walled
Ceramic
Support

Catalytic
Layer Of
Vanadia/Titania



Wi

i »

J ll

k'„„	'< m

v-

Figure 2:

Catalyst Volume Can Be Cut 50 - 65% By
Increasing Geometric Surface Area

Pressure Drop, Inches Water Column

1.0

100 CPSI

['40 CPSI!

90% Conversion
00% Conversion



10.0

Relative Catalyst Volume
380 dag. C. 20 ft/sec Velocity (• T)

100.0

4B-110


-------
Figure 3:

Increasing SCR Catalyst Geometric Area
Widens The Operating Temperature Window

NOx Conversion, %

~
+

~
¦ +

+

n 200CPSI. 60.000VHSV
+ 25CPSI, 12.000VHSV

40 '—
250

300 350 400 450
Temperature, deg C

+

500

550

1/1 NH3/NOX Ratio

Figure 4:

Composite SCR Catalysts Exhibit
Inherently Low S02 Oxidation Activity

S02 Conversion, %

0.0X 0.5X 1.0X	1.5X 2.0X 2.5X

Relative Catalyst Loading, (g/in3)/base

100 CPSI, 350 deg. C. 450 ppm S02


-------
Figure 5:

Composite SCR Catalyst Aged In High Dust, Coal Fired Boiler
Exhaust Shows No Evidence Of Wall Erosion

Fresh Catalyst


-------
Figure 6;

Inlet 6" Of Composite SCR Catalysts From
High Dust SCR Beds Of Coal Boilers

100

eo

60

40

20

NOx Conversion, %

Ammonia Slip, ppmv

Dry Bottom Boiler
Wet Bottom Boiler

0

300

100

ao

60

40

20

325	350	375	400

Temperature, degrees C

o

425

3000 hr-1 VHSV, 1/1 NH3/NOX

Figure 7:

200 CPS! Composite SCR Catalyst Did Not
Plug In A Moderate Dust Environment

NOx Conversion, %
100 	

80

60

40

20

Change In Pressure Drop, % ¦

100

80

8 24

No Soot Blowing

Soot Blowing
Frequency, hours

	g

60

40

20

0 400 800 1,200 1,600 2,000 2,400 2,800
On-Stream Time, hours

50-120n rr.g/Nm3 dust

4B-113


-------
Figure 8:

NOx Conversion And Ammonia Slip Over
Composite SCR Catalyst After Coal FGD

NOx Conversion. %	NH3 Slip, ppm

Inlet NH3/NOx Ratio

425 ppm NOx. 20,000 1/hr VHSV,
200 CPSI, 50-120 mg/Nm3 Dust

Figure 9:

Composite SCR Catalyst Unaffected By
2 Years In Coal Fired Utility Exhaust

NOx Conversion, %	NH3 Slip, ppm

NH3/NOx Ratio

400 ppm NOx, 320 dee C
60,000 1/hr VHSV, 200 CPSI


-------
Table

Engelhard Composite SCR Catalyst Pilot Tests

Location

United States
Germany
Germany
Germany
Germany
Germany
United States
United States
United States
United States

Type

Diesel Engine

Cold Side, Goal Fired Heating Plant
Cold Side, Coal Fired Power Plant
Cold Side, Coal Fired Power Plant
Hot Side, Coal, Dry Bottom Boiler
Hot Side, Coal, Wet Bottom Boiler
Gas Turbine
Natural Gas Boiler

Gas Turbine - Simple Cycle (980 F)
Gas Turbine - Simple Cycle (1085 F

Table II;

Engelhard's 25 Composite SCR Catalyst Systems





Approx.



Location

ADDlication-Catalvst-

3M£tuj?

Flow.#/sec

California

6 Engines-VNX

1984

200 to

4 Natural

Gas and 2 Digester Gas



4000 HP

Alabama

Chemical Process-ZNX

8/90

3

California

industrial Boiler -VNX

10/90

17

California

Refinery Heater-VNX

10/90

84



Refinery Heater-ZNX/VNX

10/90

46

California

50 MW Gas Turbine-VNX

11/90

650

California

1 MW Gas Turbine-ZNX

2/91

23

California

Annealing Furnace-VNX



29

California

S-Refinery Heaters-VNX

3/91

16-31

Texas

Chemical Plant-VNX

3/91

99

New Jersey

3-Dual Fuel Engine-ZNX

5/91

5

California

Refinery Heater-VNX

6/91

39

New Jersey

2-50MW Gas Turbines-VNX

10/92

844

*VNX(tm) & ZNX(tm) are V/Ti and Zeolite SCR Cat.

4B-1T5


-------
Table

Operating Conditions Of Hot Side SCR Systems

Coal Type
Ash Content, %
Sulfur Content, %

Wet Bottom

German/Foreign Ruhr/Saarland
6-12

< 1.5

Temperature, deg. C 360

Inlet NOx, mg/Nm3 800
Inlet S02, ppmv 1000
Particulates, g/Nm3 10 - 15

Soot Blowing	Monthly

4-7
< 1.5

360

1300 - 1500

1000

2

Every 10 Days

Table IV:

Moderate Dust SCR Pilot Plant Conditions

Location

Flow Rate, scfm
Temperature, deg. C
Space Velocity, hr-1

Inlet NOx, ppmv
Inlet S02, ppmv
Particulates, mg/Nm3

Coal Type
Sulfur Content, %
Ash Content, %

Slipstream Off FGD

2060

350

20,000

350 - 420
50

50 - 120

Bituminous
0.95 - 1.18
5.0 - 6.5

4B-116


-------
SCR CATALYST DEVELOPMENTS FOR THE U.S. MARKET

by T. R. Gouker

Research Division
W. R. Grace & Co.-Conn.

7379 Route 32
Columbia, MD 21044

C. P. Brundrett
Davison Chemical Division
W. R. Grace & Co.-Conn.

10 E. Baltimore Street
Baltimore, MD 21202

4B-117


-------

-------
ABSTRACT

This paper reviews SCR catalyst development from its invention in the U.S.
through power plant applications of the technology in Japan and West Germany.
Building on this experience, the requirements for adaption of the SCR process to
U.S. high-sulfur coal applications are discussed. Grace's SCR catalyst development,
SYNOX, is then reviewed for its application to U.S. boilers firing high sulfur coal.

JAPANESE SCR DEVELOPMENT

SCR was originally invented and patented by a U.S. company in 19591, but its
use was limited to a few industrial applications, such as pollution control from
nitric acid plants. It wasn't until the 1970's that SCR gained application to power
plant NQX emissions. The first utility applications took place in Japan, The
Japanese identified SCR as a suitable approach for controlling NO* and began a
stepwise application of the technology to all three types of fossil-fueled boilers:
gas, oil and coal. By 1985, there were more than 200 commercial installations
operating in Japan2.

Application of SCR to power plant exhaust was not a simple matter. A number
of problems were encountered during development. These included: (1) catalyst
poisoning by sulfur species in the fluegas; (2) ammonium bisulfate deposition in the
catalyst and on downstream equipment; and (3) equipment corrosion due to increased
S03 levels in the flue gas. The single most important contribution the Japanese
made to the development of SCR was to switch from noble metals to base metal oxides
for the catalyst3. The use of titanium dioxide supports with mixtures of vanadium
and tungsten oxides as catalysts, solved the major problems associated with oil and
gas-fired utility fluegas applications.

Additional developments were required, however, to address the issues of
flyash plugging and erosion for coal-fired service. In 1978, pellet catalysts were
given-up in favor of parallel-flow honeycomb or plate catalysts. The low conversion
targets (40-60%) coupled with the relatively low fly ash content of Japanese boilers
and the fact that most units were relatively new, allowed SCR application go
smoothly. Continued development led to ceramic honeycomb and plate-type catalyst
configurations which provided high geometric surface area with low tendency for
flyash plugging. In the early 1980's the focus of Japanese work on SCR was on
optimizing surface geometry and avoiding flyash plugging. This optimization reduced
the size (and cost) of SCR reactors by a factor of 2 and greatly improved the
economics of the SCR process. At this stage of development, ceramic honeycomb-
based catalysts had a pitch of 13 (wall thickness of 2 mm and channel opening of 11
mm) in 1978. By 1982, catalysts with a pitch of 7.5 (wall thickness of 1.4 mm and
channel opening of 6.1 mm) had been demonstrated. Since the mid-1980's, catalyst
research in Japan has principally focused on understanding the deactivation
mechanisms of SCR catalysts4. Mechanisms have been proposed for for the effects of
alkalis, alkaline earths, and heavy metals such as arsenic. Research on SCR
continues in Japan, but due to the more demanding commercial environment, the lead
role in SCR technology development shifted to West Germany in the mid-1980's.

WEST GERMAN SCR DEVELOPMENT

Oriven by tough regulatory standards, West German utilities made rapid

4B-119


-------
progress, lowering their coal-fired power plant NO* emissions to below 0.16
lbs/million Btu5, about half that of the Japanese national standard. To do this,
the SCR process could not run at the 40-60% conversion levels as in Japan, but
instead systems were designed for up to 90% NO* removal.

The first SCR facility to be retrofited in West Germany went into operation in
1985. By the end of 1990, more than 23 GWe of power plant capacity had SCR
installed to control NO* emissions6 (Figure 1). In support of this intensive
emission control program, about 60 pilot plants were constructed and operated. The
pilot plants were required to address the design issues which surfaced in the more
demanding environments of West German boilers. In addition to dry bottom units, as
in Japan, SCR was retrofitted on slagtap boilers. Since few German utilities has
low NOx burners, N0X concentrations were much higher than in Japan. Coal sulfur and
ash contents were also higher than in Japan, leading to higher SO* and flyash
concentrations in the fluegas. The SCR pilot program resolved several issues
involving SCR process technology. The effect of erosion, poisoning and fouling on
catalyst lifetime were quantified. The potential for ammonia siip was also
determined, along with the resulting effect of ammonium bisulfate deposition.

In general, the process experience with dry ash boilers in West Germany has
been similar to the results experienced by coal-fired utilities in Japan, The
majority of the loss in activity was due to interactions of the catalyst with the
flyash. The flyash had several types of effects on the monoliths: physical
fouling, poison transfer, and bulk plugging. Sub-micron ash particles accumulated
on the surfaces of the elements and blocked the pores of the catalyst. This
physical fouling prevents the N0X and ammonia from reaching the active sites of the
catalyst and leads to a reduction in catalytic performance. Figures 2 and 3 compare
the ESCA analysis of fresh and aged samples typical of dry ash boilers®. This data
demonstrates the buildup of the wide variety of flyash elements near the surface.
The reduction in the intensity of the titanium signal further demonstrates the
covering of the surface by flyash.

Flyash was also found to contribute poisons to the catalyst surface. The
transfer of alkali metals from the flyash lowered the activity of the catalyst.

This was found to occur primarily due to the leaching action of moisture on the
flyash during start-up and shutdown of the units7. Since these metal salts are
soluble in water, the presence of moisture can promote the redistribution of these
elements. In laboratory studies, alkali salts have been shown to be a catalyst
poison due to the formation of inactive complexes with the vanadium and tungsten.

In some cases, bulk plugging of the channels by accumulations of dust was
observed. Dust plugging occurred when flyash "flaked" off of upstream equipment.
Wire screens were installed on most units to break up these flakes as they
approached the catalyst. Soot blowers were also installed to blow the flakes back
and break them up during pilot plant start-up and shutdown.

In addition to occasional plugging problems, severe erosion problems due to
fluegas fldw maldistributions have occurred. The use of flow straightening vanes
and/or "dummy" catalyst layers has reduced gas flow distribution problems and
associated erosion.

While the experience in installing SCR on dry ash boilers was generally
similar in both Japan and West Germany, there has been a significant shift in
expectation of average catalyst life between SCR on the two continents. Perhaps
because of improvements in catalyst technology or improvements in process
installations, SCR catalysts in West Germany are experiencing better activity
maintenance than the Japanese high-dust, coal-fired SCR catalysts experience

4B-120


-------
suggested. Figure 4 shows a graph of relative catalyst life versus operating time5.
This figure illustrates a 5-10% higher level of activity per year for a number of
West German installations when compared to the curve based on Japanese experience.
This has turned out to be a surprising development for the industry in that most
organizations expected SCR to experience higher rates of activity loss in West
Germany. In planning for potential activity problems, a number of new catalyst
formulations were investigated for the West German market but nor commercialized.
These developments put the United States in a position to reap the benefits of
recent SCR catalyst advances as the technology is introduced on U.S. boilers.

U. S. SCR ISSUES

As mentioned earlier, SCR technology was originally invented in the United
States, While extensively installed abroad, U.S. SCR application has expanded from
its beginnings on nitric acid plants. SCR is in service on an increasing number of
gas and oil-fired boilers, turbines, and industrial applications such as refinery
process heaters. Grgce has been offering SCR systems into this growing market under
the trade name Camet for a number of years. Systems have been designed to provide
up to 90% N0X removal. Due to the low sulfur and low fly ash concentrations,
catalyst pitch for these applications can be greatly reduced. With conventional
ceramic plate type catalyst, pitch can only be reduced to 3.0 mm. Camet catalysts
are supported on a thin metal substrate (2.5 mils) and can be manufactured with
openings equivalent to a 0.2 pitch catalyst, greatly reducing the volume of an SCR
reactor. An example of a Camet SCR system is shown in Figure 5. The system is part
of the Santa Maria Cogeneration Project in California. It is owned and operated by
Bonneville Pacific Corporation of Salt Lake City, Utah.

Except for pilot testing by the EPA and EPRI8'9, SCR has not been implemented
on coal-fired boilers in the U.S. The boiler types in the U.S. are similar to those
in West Germany, therefore a great deal of information can be transferred from the
German experience. Several issues must be addressed, however, before SCR will gain
wide spread acceptance in coal service in the United States. An assessment must be
made of the impact of higher sulfur content of Eastern Bituminous coals on SCR life
and performance. The effects of differing fly ash constituents must be identified
and quantified. In addition, the possibility exists for some as yet unrecognized
boiler conditions or coal characteristics to impact SCR performance.

For SCR to be installed on U.S. facilities, especially those burning high-
sulfur U. S. coal, risks must be reduced to an acceptable level. To do this,
engineering companies and catalyst vendors will have to develop information on a
number of key areas including:

1.	The proper space velocity, linear velocity and reaction
temperature to minimize ammonia slip at required N0X removal
levels.

2.	The tolerable level of ammonia slip under high S02 and S03
conditions.

3.	The performance of catalyst and its deactivation rate in
flue gas and fly ash from U. S. coals.

4.	Performance of the air preheater when exposed to high S03
levels and subsequently high levels of NHtS0».

5.	Adhesion characteristics of fly ash from U. S. coals to SCR
catalyst. The effectiveness of soot blowing and the effects

4B-121


-------
of residual fly ash on catalyst activity.

The most efficient way for catalyst vendors to assist in the development and
demonstration of SCR technology on high-sulfur U. S. coals is through the design of
catalysts tailored to meet the above needs in a cost-efficient manner. High levels
of denox activity while minimizing ammonia slip can be met by catalysts designed to
have higher NO* conversion activity. The effects of NH*SQ4 deposition can be
minimized by designing catalysts to have a low S02 to S03 oxidation activity. The
detrimental effects of fly ash and deactivation can be counteracted by catalysts
designed to be more durable, and poison resistant.

U. S. SCR DEVELOPMENT

The design requirements and the recognition that a major portion of SCR
process costs are still associated with the catalyst has led researchers to
vigorously pursue improved catalyst designs. Grace's approach has been to focus on
increasing catalyst activity and life, thus reducing catalyst volume requirements
while extending the useful life and consequently reducing SCR process costs. To
determine the potential for major improvements, Grace undertook a fundamental study
of the limits of SCR catalyst performance, developing a mathematical model that
expressed catalyst performance (N0X conversion, ammonia conversion, S02 oxidation)
as a function of the properties of the catalyst.

The model accounts for the key design parameters of the catalyst, including
its composition, monolith channel shape and dimensions, monolith wall thickness,
pore structure, overall volume, and aspect ratio. The model has been found to
correlate well with Grace's commercial data base, adjusting only surface-kinetic
rate parameters. Modeling results indicated that a substantial reconfiguration of
the pore structure of the catalyst could increase NO* conversion by about 50%, while
simultaneously increasing resistance to poisoning and thereby extending catalyst
life. The model indicated that the NO* conversion improvement would be selective
with respect to S02 conversion. That is, the undesired S02 oxidation reaction would
not be enhanced by the pore structure reconfiguration.. This is due to the fact that
the reduction of NO is diffusion limited whereas the rate of oxidation of S02 is
kinetically controlled.

The model showed that an optimum balance between surface catalytic activity
and diffusivity could be provided by a bimodal pore structure with a substantial
percentage of macropores (Figure 6} . However, such pore structures could not be
attained using titania as the catalyst support.

To solve this problem, Grace succeeded in engineering silica to provide the
necessary macropores for a new catalyst pore structure while maintaining the
necessary intrinsic denox activity. Grace researchers developed a preparative route
to deposit titania within the silica to produce a novel catalyst support (Figure 7).
When extruded and impregnated with vanadia, the new catalyst, trade named SYNOX™,
resulted in the anticipated 50% improvement in activity11. We have demonstrated
excellent hydrothermal stability through near 3000 hours in simulated fluegas with
no noticeable decrease in activity (Figure 8) in testing of laboratory-scale pieces.
Proposals have been made to expose these new monoliths to power plant stack gas side
streams. Such pilot plant tests have been planned by the Electric Power Research
Institute and Southern Company Services (for the U.S. Department of Energy).
Independent of the pilot plant tests, Grace has initiated a slipstream test of its
own.

A slipstream reactor has been set up at the TVA Shawnee Steam Plant in
Paducah, KY, to study the SYN0X catalyst in fluegas from a 100 HW boiler burning

4B-122


-------
high-sulfur coal. The effects of flyash, alkali salts, and other constituents will
be studied and compared to their effects on a conventional titania catalyst. The
boiler effluent contains about 300 ppm NO*, 3000 ppm S02, and a flyash loading of
about 3 grains per standard cubic foot. The Grace slipstream reactor is located
downstream from the mechanical separators and air preheater, as shown in Figure 9.
The fluegas temperature entering the unit is about I50°C and must be reheated to
about 350"C before passing over the catalyst.

The reactor contains 3 catalyst baskets stacked one on top of the other and a
dummy basket at the top. The dummy basket contains inactive catalyst support and is
used as a flow straightener. Each catalyst basket contains 16 pieces of 2.5 cm x
2.5 cm x 23 cm-long monoliths.

A large bypass stream is maintained from the TVA supply duct to the Grace
slipstream reactor and back to the TVA return duct. This minimizes heat losses in
the 20 m-long pipes and reduces the chances of flyash plugging. A purge valve on
the inlet sample line helps keep the bypass lines clean. The primary purpose of
this slipstream reactor is to expose catalyst to fluegas under typical SCR operating
conditions.

The slipstream test is configured to expose the SYN0X and commercial catalyst
pieces to fluegas at typical linear velocities in order to evaluate poison and
erosion resistance. The slipstream reactor length of 27 inches versus the 9-12 feet
used commercially results in very low NO* conversions at these exposure conditions
which reproduce commercial linear gas velocities. Conversion, however, is measured
from time to time by reducing the linear velocity and operating at space velocities
typical of commercial installations. Figure 10 shows the results of testing upon
initial start-up of the slip stream reactor.

Quantitative catalyst analysis for denox efficiency is performed in the
laboratory under carefully controlled conditions. Accurate activity measurements
are possible only when the NO* and NH3 concentrations are constant and the catalyst
is operating at steady state. Due to a wide range of difficulties with Unit #9 at
the Shawnee Steam plant, data on aged catalyst has not been obtained yet.

SCR PROCESS COSTS

Since its introduction in Japan in the 1970's, the cost of SCR has dropped
continually, primarily because of technological advances. In Japan, the levelized
busbar cost of SCR decreased over a six-year period by more than a factor of 3
because of increases in catalyst life and reductions in catalyst volume requirements
as a result of Improved catalyst geometry and composition. In Germany the learning
curve continued, dropping costs by an additional factor of 2, again largely because
of technical developments: reduced catalyst installation cost; mechanized and
automated catalyst manufacture; and new catalyst replacement strategies that allowed
the extension of average catalyst lifetime guarantees to four years.

Additional progress on the learning curve toward reduced costs 1s expected
when SCR gains large-scale U.S. application. This can be best illustrated by
showing the cost sensitivity of the latest German SCR experience12, transposed to a
U.S.-equivalent basis, in 1988 dollars. Figure 11 shows, as mentioned previously,
that approximately half of the levelized busbar cost of SCR is still catalyst
related . This points to the possibility of further cost reductions as a result of
improvements in the catalyst.

Figure 12 quantifies this possibility by comparing costs for first- and
second-generation SCR control technologies. The first column in Figure 12, taken

4B-123


-------
from a 1989 United Engineers & Constructors study of first-generation SCR prepared
for EPRI'4, shows the costs associated with SCR for an 80% removal efficiency from
an uncontrolled emission level of 0.6 lbs/mm Btu.

Costs for second-generation technology, such as Grace's SYNOX catalyst, are
illustrated in the second column. With the 50% higher activity of SYNOX, an SCR
unit will need only two-thirds of the conventional reactor volume. Capital
requirements will be reduced to $62 per KM. With a reduced catalyst volume and
increased catalyst life, the levelized busbar cost will drop from 5.2 mi 11s/KWh to
2.1 mills/KWh and the cost expressed in $/ton of NO^ removed will also decrease. At
the level of 80% NO* removal the per-tonnage cost of a clean NSPS boiler will drop
from $2170 to $870 . Combining SYNOX SCR with low-NO* burners can reduce the cost
even further to $700/ton of NO* removed. On a cyclone unit, costs will be lowered
from the $600/ton range to $400/ton of NO* removed.

The work described in this paper was not funded by the U.S.

Environmental Protection Agency and therefore the contents
do not necessarily reflect the views of the Agency and no
official endorsement should be inferred.

1.	Anderson, H. C., W. J. Green and D. R. Steele, "Catalytic Treatment of Nitric *
Acid Plant Tail Gas," Ind. Eng. Chem., 53, 199 (1961).

2.	Ando, J., "Recent Status of Acid Rain and SO^NOx Abatement Technology in
Japan," 10th Symposium on Flue Gas Desulfurization, EPRI/EPA, Atlanta, GA,
November 18-21 (1986).

3.	Lowe, P. A., W. Ellison, L. Radak, "Assessment of Japanese SCR Technology
for Oil-Fired Boilers and its Applicability in the U.S.A.," Joint Symposium
on Stationary Combustion NO% Control," EPRI/EPA, San Francisco, CA March
6-9 (1989).

4.	Aoyagi, K., "Rapportuer's Report: Sessions on Environmental Control
Retrofit/Upgrade," GEN-UPGRADE 90, IEA/USDOE/EPRI, Washington, D.C., March
6-9 (1990).

5.	Haug, N., Material presented at the NATO Meeting on Coal Combustion Systems,
Copenhagen, Denmark, May 13-15 (1990).

6.	Gouker, T. R., J. P. Solar, J. C. Fu, C. P. Brundrett, "Evaluation of
Selective Catalytic Reduction Catalysts from West German Pilot Plant
Studies," 7th Annual International Pittsburgh Coal Conference, Univ. of
Pittsburgh, Pittsburgh, PA, September 10-14 (1990).

7.	Schallert, B., of VEBA Kraftwerke Ruhr AG, Private Communication (1987).

8.	Maxwell, J. D., T. W. Tarkington, T. A. Burnett, "Technical Assessment of
NO* Removal Processes for Utility Application," EPA Report 600/7-77-127,

March (1978).

9.	Shiomoto, G. H., L. J. Muzio, "Selectice Catalytic Reduction for Coal-fired
Power PI ants--Pi 1ot Plant Results," Final Report, EPRI Report CS-4386, April
(1986).

4B-124


-------
10.	Beeckman, J. W., L. L. Hegedus, "Design of Monolith Catalysts for Power Plant
N0X Emission Control, paper 72e presented at the AIChE Annual Meeting,
Washington, D.C. (1988); Ind. Eng. Chem. Res., in press, 1991.

11.	Solar, J. P., J. C. Fu., "Effect of Reaction Parameters on the Activity of
SYNOX Catalysts for the Selective Catalytic Reduction of NO*," 83rd Air and
Waste Management Annual Meeting, AWMA, Pittsburgh, PA, June 24-29 (1990).

12.	Schonbucher, B., "Costs of a DeNO* Plant on the Basis of the SCR Process,"
Proceedings of the Workshop on Emission Control Costs, Eds. 0. Rentz, et.
al., Inst, for Ind. Prod., Univ. of Karlruhe, Karlsruhe, West Germany,
September 28 - October 1 (1987).

13.	Boer, F. P., L. L. Hegedus, T. R. Gouker, K. P. Zak, "Controlling Power
Plant N0X Emissions," CHENTECH, 20, 312 (1990).

14.	Robie, C. P., P. A. Ireland, J. E. Cichanowicz, "Technical Feasibility and
Economics of SCR NO* Control in Utility Applications," Joint Symposium on
Stationary Combustion NO* Control, EPRI/EPA, San Francisco, CA, March 6-9
(1989).

4B-125


-------
35 -

103 MWeJ:

30 J
25
20 1

15 -¦

i
[

•10 t

1985 1986

1987

1988

1989
Year

1990

1991

1992

1993

Figure 1.

Flue-Gas Treatment NOx Controls

in West German Power Stations

p
e
a
k

I

n

t
e
n
s

i

t

y

A.J

1888 988 888 788 688 588 488 388 288 188

Binding Energy, eV

Figure 2.

ESCA of Fresh Catalyst Surface

8

4B-126


-------
12 T

I

11 -i
10 ••
9
8
7
.6
5
4
3

2 ->
1

A d

Na

f' As
08 N I Cs K G S

1884 983 883 783 682 582 482 381 281 181

Binding Energy, eV

Figure 3.

ESCA Showing Ash Constituents

on Aged Catalyst Surface

0.7
0.6

0.5
0.4 k
0.3
0.2

0.1 r

r

'01

Expected curve acc. to
Japanese experience

	1_

I	I	1

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15-10:'h 17

Figure 4.

SCR Catalyst (High-Dust) Activity
Loss in a Dry-Bottom Boiler

4B-127


-------
Figure 5.

Camet SCR System located at the
Santa Maria Cogeneratiori Installation

4B-128


-------
.6 vOv

Por® Dlam«ur|

Figure 6

Model Prediction of DeNOx Activity

Q.I

9?-

d

V
/
a

(

L &,«

0

S 0 J-.

D

j.



Figure 7
Pore Size Distributions

4B-129


-------
Comparison of the Activity of SYNOX, a second-
generation catalyst under development by Grace, with
Noxeram, a first-generation catalyst produced in West
Germany by a joint venture between Grace and Feldmuehle.

WATER

STEAM

COAL

AIR

MECHANICAL
SEPARATOR

AID

PREHEATER

UNIT \
/ #9
BOILER

vaAAMPi;

PREHEATED AIR

STACK

BAGHOUSE

150 C

SYNOX
TEST
REACTOR

Figure 9

Flue Gas System of TVA Boiler #9

Shawnee Steam Plant Paducah, KY

4B-130


-------
Conversion, %

100

80

60

40

/

/

Nominal Operating Conditions

,/*	Flue gas flow ¦ 1100 SCFH

'/	. Temperature « 350°C

NO» ¦ 300 ppm
S02 ¦ 3000 ppm
SOs * 25 ppm
20 r	H20 » 6%

02 ¦ 4%

Flyash « 3 grains/scf

0 12	1	1	

0	0.5	1	1.5

NH3 to NOx Ratio
Figure 10

SYNOX Catalyst Performance at Start-up
in Test Unit at Shawnee Steam Plant

Variable Operating Costs

Catalyst Replacement	34%

Ammonia	5%

Power	7%

Fixed Operating and Maintenance Costs

Maintenance Labor	11%

Administrative	8%

Capital Charges

Catalyst First fill	14%

Ancillary Equipment	21%

Figure 11

Levelized Busbar Breakdown of SCR in 1988

4B-131


-------
Second-generation SYNOX
First-generation catalyst catalyst. Grace estimates

Boiler Type	NSPS	NSPS

Uncontrolled Emissions, lb/mm Btu	0.6	0.6

Capital Cost, $/KW	100	62

Levelized Busbar Cost, mills/KWh	5.2	2.1

NQ Removal Cost, $/ton	2170	870

Figure 12

U.S. SCR Cost Projections in 1988 Dollars at 80% NOx Removal

4B-132


-------
POISONING MECHANISMS IN EXISTING SCR CATALYTIC CONVERTERS
AND DEVELOPMENT OF A NEW GENERATION FOR
IMPROVEMENT OF THE CATALYTIC PROPERTIES

L. Balling
R, Sigling
H. Schmeltz

E. Hums
G. Spitznagel

Siemens AG Power Generation Group (KWU)
Hammerbacherstrasse 12 + 14
8520 Erlangen, Germany

4B-133


-------

-------
ABSTRACT

For better understanding of the processes involved in catalytic NOx reduction using
titanium, tungsten, molybdenum and vanadium oxide catalysts, extensive investigations
have been can-led out by Slemens/KWU In recent years, also focussing on explaining the
deactivation phenomena in greater detail.

On the basis of research carried out on wet-bottom furnaces, heavy oil combustion and
glass melting furnaces, this paper discusses the mechanisms of poisoning and the factors
which cause changes in the catalytic properties. The catalysts were analyzed by
appropriate methods such as XRD, XPS, XANES, EXAFS, etc.

Taking arsenic oxide as a well known catalyst poison, this paper explains Its formation and
accumulation In the wet-bottom boiler type. Deactivation, mechanisms and poisoning
models are considered.

Finally, the paper points out the way In which this knowledge has been incorporated Into
further developments, resulting In a new generation of catalyst, which Is currently being
prepared for introduction onto the market.

4B-135


-------
1. INTRODUCTION

Catalyst aging means "normal" loss of activity during operation as a result of irreversible
processes in the catalyst e.g. sintering. Catalyst poisoning by contrast is a much more
rapid deactivation caused by components of the flue gas.

At the beginning of SCR catalyst development in Germany, unusually high deactivation
rates were, however, measured in the high-dust region downstream of wet-bottom
furnaces, see Fig.1.

Figure 1. Relative activity of SCR catalysts in dry and wet-bottom boilers versus the service
time.

Extensive measurements In power plants showed that the rapid deactivation Is caused by

gaseous arsenic oxide or very fine arsenic oxide covered dust particles. The

concentrations of arsenic oxide in wet-bottom boiler flue gases upstream of the air

preheater is about 100 times higher than in dry-bottom boiler flue gas.

One of the first basic questions was therefore, to find the reason for the much higher

arsenic oxide concentration and to find measures to reduce this influence of catalyst

poisoning.

4B-136


-------
2. WET-BOTTOM BOILERS

The ash recirculation and the very high combustion chamber temperature in wet-bottom
boilers are the main differences to the dry-bottom boiler type. In dry-bottom furnaces the
arsenic oxide concentration is mainly a function of the concentration in the fuel. In wet-
bottom boilers however, ash recirculation increases the concentration from a second
source, the melting of arsenic laden flyash in the combustion chamber see Flg.2.

f02; 
-------
POSSIBILITIES OF PREVENTING ARSENIC OXIDE ENRICHMENT

Preventing enrichment by complete or partial dust removal

Measurements of arsenic oxide concentration In the Individual sections of an electrostatic
precipitator have shown that about 30% of the total of arsenic oxide on the flyash was
bound to the fines fraction of flyash. By partial extraction of this fines fraction which
accounts for only 18% of the total amount of dust, the gaseous arsenic oxide
concentration can be reduced by a factor of 2 - 3, as shown In Fig.3. This potential solution
was not, however, pursued further, since the question of what to do with these fines is still
open for some power plants, rendering this concept financially unattractive for them.

soQtblower operation

HlthOUt dUBt fBDOVftl
with IBS dust ranoval

10 - 20 30 40 SO 60 70 SO 30 100
Operating period (hours)

Figure 3. Effect of partial dust removal on the gaseous arsenic concentration

Improvement of Adsorption of Gaseous Arsenic on Flyash or on Additives

Binding gaseous arsenic to a suitable fuel additive proved useful for two reasons. It is a
well-known fact that when the crystalline fraction contained in the ash Is high, only very
low concentrations of gaseous arsenic will be present In the flue gas, even if the arsenic
content of the coal was high. In a wet-bottom furnace plant where limestone was added to
the coal to enhance slag flowability, it was also observed that the catalyst deactivation was
much lower. For this reason, systematic measurements have been performed to

4B-13B


-------
investigate the effects of limestone dosing to bind arsenic and other volatile metals as
well. Results are given in Fig.4. It can clearly be seen that In this special case, admixture of
limestone to coal reduces the concentration of gaseous arsenic oxide In the flue gas from
700 pg/m3 to less then 100 fig/m3.

Metering of limestone {*!)

Figure 4. Effect of limestone metering on the gaseous arsenic concentration

The reduction of the arsenic deposited on catalyst specimens is particularly striking proof
of the effectiveness of this measure. In Fig.5 the uptake of arsenic by a catalyst as a
function of gaseous arsenic concentration is shown.

Figure 5. Arsenic concentration In the catalyst versus gaseous arsenic concentration

4B-139


-------
All results shown above where gathered (measured) by Siemens from a wet-bottom boiler
with the utilities kind assistance. In spite ot these positive results, we do not consider
these measures alone sufficient, since the catalyst could suffer serious damage in matter
of hours in the event that limestone dosing is not available. For this reason It was
imperative to develop catalysts which:

•	adsorb less poisoning matter, I.e. have lower
affinity for arsenic than conventional dry-
bottom furnace catalysts

•	are less susceptible to arsenic poisoning

3. CATALYTIC MECHANISMS

After having explained the reasons for the high concentration of arsenic oxide in the flue
gas and measures to reduce it, we will now elucidate why this compound can act as a
poison for a DeNOx catalyst.

MODEL OF DENOX CATALYSIS

To thoroughly understand the poisoning mechanisms it is necessary to know about the
undisturbed catalytic mechanisms on the surface of a DeNOx catalyst. Fig.6 shows a more
general model of the SCR process, illustrating the pore system of a DeNOx catalyst
containing pebbles of titanium dioxide joined together by sintered bridges and covered
with an active catalytic layer comprising a mixture of oxides ot vanadium and either
molybdenum or tungsten.

4NHj + 4NO + Q2 4N2 + 6H20

Pore system «-
cavity system of
a pebble bed

Catalytic converter
Figure 6. Model! of DeNOx Catalysis

4B-140


-------
We thus consider the catalyst to have the form of a pebble bed In which the surface
structures of the pebbles at the bottom of the bed are accessible via the cavity system of
the bed. Using this model as a basis, the SCR process can now be broken down into 5
steps (see also Fig.7).

1.	The reactants first have to be drawn off from the flue gas, which pass the catalyst at a
rate of about 10 m/s, and conducted via the system of pores to the inlerior surfaces. Large
pores enable gases to be transported rapidly, but at the same time the pebbles should be
as small as possible in order to provide a large specific surface area. A compromise has to
be reached In this respect, because pore size and pebble diameter in any pebble bed are
normally correlated functions. For practical purposes, pebbles with a primary grain size of
about 20 nm, resp. 75 m2/g of specific surface area have proven to be useful.

2.	The reactants NH3 and NO are adsorbed by the active sites.

3.	The reaction occurs between NH3 and NO at the active sites.

4.	The reaction products, N2 and HgO are desorbed.

5.	The reaction products leave the pore system via the same route as the reactants were
admitted and become entrained in the flow of flue gases.

Figure 7. DeNOx-reaction-path

413-141


-------
DEACTIVATION CAUSED BY ARSENIC OXIDE

During exposure to flue gases the arsenic oxide is deposited on the surface of the
catalysts. In the event of condensation of the arsenic oxide, this occurs preferentially in
the smallest pores with the highest curvature, so that there may be a narrowing of small
pores, Inhibiting the gas transport in steps 1 and S.

Apart from such physisorptlve coverage, a chemical attack of the surface structures by
arsenic has been observed. Both effects reduce the number of active sites necessary for
step 2.

Investigations by X-ray absorption spectroscopy show that the Initially deposited As3 + In
arsenic Is oxidized to Ass + , forming a structure of isolated orthoareenate on the catalyst
surface. This Implies a reduction of the catalyst and might lead, for Instance, to a change
in its reoxidation ability necessary for step 3.

An Irreversible selective blocking of NH3 adsorption sites by arsenic oxide is suggested by
infrared spectroscopy studies. These show, in the case of tungsten type catalyst,
pertubations of W=0 (terminal oxo group) oscillations by orthoarsenate, analogous to the
reversible adsorption of Lewis bases, such as H2O, NH3 or CO.

As a consequence of all these considerations, the pores of a wet-bottom boiler catalyst
should be somewhat larger than in a dry-bottom boiler in order to limit the blocking of
active sites. Furthermore, the chemical composition on the surface should be optimized to,
account for arsenic loading. The Investigations on this subject are in progress and focuss
especially on differences between tungsten and molybdenum, because catalyst types
based on these active components have shown different behaviour in service.

4B-142


-------
4. NEWLY-DEVELOPED CATALYTIC MATERIAL

Based on our experience with 45 DeNOx-plants in operation with Siemens catalytic
converters (wet and dry-bottom boiler, oil fired, up and downstream of FGD) we are in a
good position to further improve our catalysts. The long-term experience in pilot plants
and in large scale reactors downstream of wet-bottom boilers with arsenic oxide
concentrations greater than 1000 jjg/Nm3 indicate how the catalytic material based on
Ti02, WO3, M0O3 and V205 was chemically and physically changed. Investigations in
arsenic-deactivated catalytic material show a chemical reduction of its active components
such as vanadium.

For the explanation of the above-mentioned phenomenon, arsenic poisoning tests with
V2O5/M0O3 as well with Tl02/V20&/Mo03 systems were performed and compared. Ti02-
free as well Ti02-based material containing M0O3 and a composite oxide of V and Mo. It
was found that this composite oxide is reduced by As203, forming a certain ratio of V5 +
to V4+ by phase transformation without arsenic incorporation. The special feature of this
reduced phase is a portion of stable Vs + which can hot be further reduced by arsenic.
Using this stabilized V/Mo precursor for the catalyst preparation, chemical and physical
properties can be achieved which differ from those of the Japanese-licensed catalytic
material.

5. COMPARISON TESTS IN A WET-BOTTOM BOILER

To demonstrate the behavior of different catalytic materials in a long-term test, we exposed
samples prepared as specimen plates to the flue gas of a wet-bottom boiler with a very
high gaseous arsenic oxide concentration of about 700-1200 jjg/Nm3. To accelerate the
deactivation, we exposed the material to a very high gas (resp- arsenic) mass flow in
relation to the surface area.

Catalyst A is a tungsten type material prepared with a coprecipitation method, catalyst B is
a molybdenum type Siemens innovation for wet-bottom boilers and type C is a new
catalyst also based on Ti02, prepared with this special arsenic oxide resistant V/Mo-
precursor.

4B-143


-------
Figure 8 compares the physical and chemical properties of the fresh material.





Type A

Type B

Type C

Properties









Activity Iresh 350*0

Nm3/mzh

34,5

38,1

46,5

SC>2 • conversion const

Nm3/m3h

83

112

221

BET surface area

m2/g

79

76

73

Porevolume

mm3/g ,

272

229

292

Porema*imum

•

A

70-100/40-50

100-150/50-60

900-1500/150

Figure i. Comparison of different catalytic materials

Figure 9, 10 and 11 compare the properties of catalytic activity, inner surface and pore
volume before and after 1800 h operation In the above-mentioned flue gas.

Figure 9 clearly shows, that the newty-developed catalyst has the highest activity
combined with lowest deactivation which can be traced to the different pore size
distribution (see Fig. 12) and the highly stabilized precursor as regards poisoning.

measuring conditions:
T = 350* C
Oi = 4 Vol%
850 = 10 Vol*
NO = 100 vpm
NH, * 400vpm
LV mi m/s (356X)

| before exposure
»fler exposure

Type	A	B	C

Figure 9. Comparison of the catalytic activity before and after exposure

4B-144


-------
Fig, 10 end 11 demonstrate the possibility of reducing the Influence ot the arsenic oxide
condensation to the inner structure.

90

bo -

E 60

S 50

f

40

30 -

S 20

m

10

Type

['.!



rv\;:3

s

before exposure
Bfter exposure

ail



Figure 10. Comparison of the inner
surface before and after exposure

350

300

E

£

E

200

too

J. before exposure

after exposure

IS

Type	ft	B	C

Figure 11. Comparison of the pore volume
before and after exposure

Figure 12 shows the pore size distribution of the catalyst type A and type C in comparison.

pore volume

"rei/g

€.20

0.02

TtTTTTTT I 11111 I T

1C';

to*

pore radius A

0,26

0,20

0,08

0,02

-



/		 	

i



type C

/

-



r

;



/

i

-



f

i

i

j



	

)

W	10*	10*

pore radius A

to2

Figure 12. Comparison of the pore size distribution ot type A and C

4B-145


-------
To improve a catalytic materia! for applications in S02-laden gases, especially for fuels
with a very high sulfur content such as some American coals, it is necessary to reduce the
SOg oxidation rate. The feature of the composite oxide and the pore size distribution (see
Fig. 12) leads to an increase in DeNOx catalytic activity without increasing the S02
conversion rate. In other words, the SO2 conversion and the SO3 production rate can be
limited to very low levels.

The activity and SOg conversion rate of the newly-designed catalyst can be adjusted by
the amount and composition of the precursor. In Fig.13 the catalytic activity and the
reaction constant for SOg conversion are compared for different compositions of type A
(tungsten type) and type C (newly-developed type).

it can be clearly seen, that at fixed S02 reaction constant, the catalyst volume can be
reduced by about 15 %. The major advantage tor high-dust applications however Is the
reduction of the SO3 production by a factor of about 2.6 (350°C), consequently minimizing
the corrosion problems in downstream facilities.

Figure 13. Comparison of the k^Ox VB,ue versus ksOx value 'or two different catalyst
types

4B-146


-------
6. CONCLUSIONS

When retrofitting power plants in the US with SCR systems, the very different flue gas
compositions have to be considered on a case-to-case basis, tailoring the catalytic
converter to specify plant requirements. Improved catalysts feature durability against
poison laden gases and reduced SO2 conversion rates.

Faced with this spectrum of requirements, Siemens AG Power Generation Group (KWU) Is
In a position to offer an improved catalytic material with low SO2 oxidation and low
deactivation rates combined with the advantage of a plate type shape for high-dust
applications.

To demonstrate these combined advantages, we are ready to supply this type of catalytic
converter to pilot or demonstration plants in the US.

4B-147


-------

-------
STATUS OF 1 MW SCR PILOT PLANT TESTS AT
TENNESSEE VALLEY AUTHORITY AND
NEW YORK STATE ELECTRIC & GAS

H, Flora and J. Barkley
Tennessee Valley Authority

G. Janik and B. Marker
New York State Electric & Gas

J. E. Cichanowicz
Electric Power Research Institute

ABSTRACT

EPRI and member utilities are sponsoring a pilot plant test program to evaluate
SCR NOx control for potential application by the U.S. utility industry. This
program will employ up to six SCR pilot plants of nominally 1 MW capacity, and
focus on evaluating catalyst life and process performance for medium and high
sulfur coal application. The first pilot plant in operation is located at TVA's
Shawnee Test Facility, operating on high sulfur content (3-4%) coal. Initial results
from baseline tests show catalyst performance for NOx removal and control of
residual NH3 after 4 months operation meets the design values estimated by the
catalyst suppliers. A two year test program including periodic extraction and
analysis of catalyst samples is planned for all pilot plants to track any changes in
catalyst performance and activity. The results will provide a basis for estimating
catalyst life and process feasibility for U.S. conditions.

INTRODUCTION

In recent decades, environmental agencies in Japan and Europe have implemented
regulations to significantly reduce NOx emissions. Generally, these reductions
necessitate control of NOx to limits beyond the capabilities of combustion controls.
For example, since the 1970s, allowable NOx emissions for coal-fired power stations
in Japan have been as low as 150 ppm. Several western European nations in the
1980s implemented NOx regulations for coal-firing to approximately 100 ppm.

This international trend in NOx regulations raises the prospects for increasingly
stringent requirements in the U.S. Without major improvements in the NOx
control performance of combustion technology, postcombustion control may be
required to meet the most strict NOx regulations.

The most widely commercialized postcombustion technology to date is selective
catalytic reduction (SCR). Considerable experience with SCR exists in Europe with
low sulfur coal; and in Japan with low sulfur coal, oil, and natural gas. In contrast,
there is no meaningful experience with SCR for medium/high sulfur U.S. fuels in
combination with furnaces of heat release characteristics that typify U.S.
applications. Recent results from a fundamental investigation of SCR catalyst
poisoning (1) suggests that sulfur, in combination with certain trace elements in

5A-1


-------
coal (such as alkali) can contribute to catalyst poisoning. Accordingly, meaningful
pilot plant experience is desirable prior to full-scale SCR application.

To provide this experience, EPRI and member utilities plan to operate up to six
SCR pilot plants on medium and high sulfur fuels on U.S. power plants. The
proposed pilot plants will provide the basis for realistic estimates of catalyst life and
SCR process impacts. A companion paper at this Symposium (2) has identified the
significant impacts of SCR on balance of plant equipment, and documented the
influence of catalyst life on SCR levelized costs. Data from this pilot plant program
will be used by EPRI to refine engineering study results estimating the feasibility
and cost of SCR for the U.S. utility industry. This paper describes the pilot plant
design and test plans for the first two units scheduled for operation, at the TVA
Shawnee Steam Station, and the New York State Electric & Gas (NYSEG) Somerset
Station. Initial results from the TVA pilot plant are summarized,

PROGRAM SCOPE AND OBJECTIVE

This empirical test program will address both the conventional "hot-side" SCR
process (reactor located between the boiler economizer and air heater) and the
alternative "post-FGD'" SCR application. The test objective is to provide realistic
information for key SCR design variables such as space velocity (e.g. catalyst
quantity), the level of residual ammonia that can be tolerated, byproduct SO3
formation, catalyst lifetime, and the formation of byproduct ammonium/sulfur
compounds. This information will reflect authentic U.S. utility operating
conditions, as defined by fuel properties and furnace design characteristics. A
generic pilot plant design was defined for all six planned sites, thus the only
changes between sites will be fuel properties, furnace design, and operating modes.
For the "hot-side" application, tests will focus on the quantity and lifetime of
catalyst necessary to maintain control of residual NH3 while delivering required
NOx removal, and generation of byproduct SO3. For the post-FGD process, tests
will similarly evaluate the catalyst quantity and lifetime necessary for control of
NOx and residual NH3, and generation of acidic compounds; but also evaluate the
thermal performance of the heat exchanger necessary to elevate flue gas
temperatures to reaction levels.

A fundamental premise of this program is that fuel composition and furnace
design uniquely determine catalyst life, by defining the conditions for transport of
trace species to the catalyst surface. Transport conditions are defined by both the
composition and concentration of trace species in flue gas, particularly the amount
of trace elements volatilized; thus both fuel composition and furnace
temperature/time history are important. A total of six pilot plants will be
employed to simulate the-wide range of transport conditions typifying the U.S.
utility industry. Table 1 summarizes the fuel characteristics and furnace types at
four pilot plant sites that are either operating in a test mode, are in startup, or are
in a design/planning stage. High sulfur coal SCR testing on a pre-NSPS
conventional boiler (e.g. tangential- or wall-fired) is underway at TVA's Shawnee
Steam Station. The post-FGD SCR application on a medium sulfur coal is being
evaluated at the Somerset Station of NYSEG. SCR application to high sulfur
content (-1% sulfur) fuel oil will be conducted at Niagara Mohawk's Oswego
Station. Also planned is an SCR pilot reactor followed by an air heater on a high

5A-2


-------
sulfur coal-fired, cyclone type boiler, presently designated for the Coffeen Station of
Central Illinois Public Service. Two additional pilot plants are planned, although
specific utilities and fuel types have not yet been identified.

A unique feature of this program is a cooperative venture with catalyst suppliers to
assess deactivation mechanisms and estimate catalyst life based on the pilot plant
results. Each pilot plant will be capable of evaluating two catalysts, at identical
process conditions. Catalyst suppliers will extract samples at approximately 3 or 4
month intervals for analysis in their laboratories. Measurements will both
document catalyst activity (as inferred from NO removal) and the accumulation on
the catalyst surface of trace species suspected to be poisons. Results over a two year
period will provide a factual basis for estimating catalyst lifetime.

Data from these pilot plants will be supplemented by results from the evaluation of
SCR conducted by Southern Company Services (SCS) under the Department of
Energy's Clean Coal Technology program. The SCS program, which EPRI is
cofunding, will also be conducted for a nominal 3% sulfur coal, on a pre-NSPS
conventional boiler, similar to the fuel/furnace conditions reflected by the TV A
Shawnee Station. The objectives of these two activities are complementary-the
SCS program will evaluate a large number of different catalysts at relatively fixed
fuel composition and furnace design; in contrast the EPRI program will evaluate a
limited number of similar catalysts over a wide range of fuel composition and
furnace designs.

PROGRAM STATUS

The TV A 1 MW pilot plant at the Shawnee Steam Station has been operating for
almost four months; baseline tests are 60% complete. The TV A pilot plant is
evaluating catalysts supplied by Joy Environmental Equipment Company and
Norton Company. The NYSEG pilot plant, evaluating the post-FGD SCR
application, is initiating startup/shakedown tests at this writing. Catalysts will be
supplied by W.R. Grace Co, and Englehard Industries. The pilot plant at Niagara
Mohawk's Oswego Steam Station has been fabricated and is presently being
installed; a mid-1991 startup is planned. The SCR reactor/air heater pilot plant
planned for the Coffeen Station of Central Illinois Public Service is still in the
formative stages of planning and funding; no significant activities are anticipated
until late 1991 /early 1992.

PILOT PLANT DESIGN

A generic 1 MW pilot plant was designed based on experience gathered from
numerous SCR pilot plants tested in Europe in the mid-1980's, and from the 3 MW
SCR pilot plant operated by EPRI on low sulfur coal from 1980 through 1982 at the
Arapahoe Test Facility. The key design premises based on this experience are:

•	pilot plant flue gas should promote process conditions replicating a full-
scale reactor in terms of flue gas residence time, temperature, gas species
and trace element composition, etc.

•	full-scale catalysts representative of commercial systems should be tested.

5A-3


-------
•	pilot cross section should ensure at least one full-scale catalyst element is
not adjacent to a wall, and thus experiences erosion, mass transfer, and
heat transfer conditions typifying full-scale conditions.

•	two catalysts should be evaluated at identical process conditions, with
samples capable of being extracted at nominally 3 or 4 month intervals.

The TV A and NYSEG pilot plants are described as follows:

Hot-side SCR: TV A Shawnee

The hot-side SCR high sulfur coal pilot plant is shown in Figure 1. Pilot process
conditions are selected to provide 80% NOx removal (from boiler exit
concentrations of 600 ppm) and maintain residual NH3 at the exit at 5 ppm. Four
catalyst layers are employed to meet the design conditions; a fifth layer exists to
evaluate the required catalyst quantity and pressure drop to reduce residual NH3 to
2 ppm or less. Pilot design and operating conditions are summarized in Table 2.
Flue gas composition measurements can be obtained at the exit of any of the five
layers.

Flue gas is extracted from the economizer exit of Unit #9 at the Shawnee Steam
Station (Paducah, KY) at approximately 710°F, and passes through an isolation
damper, a venturi to monitor flow rate, and a 40 kW heater to adjust process
temperature to desired values (680-70Q°F). Flue gas then enters an approximately 20
ft straight section in which ammonia reagent is injected and mixed. The flow is
then equally split into two reactors, each containing catalyst from a different
supplier. At the exit of each reactor are flow rate monitors and manual dampers
which insure flow rates are equal in each section. An induced draft fan followed by
a control damper is the last component prior to flue gas return.

Post-FGD: NYSEG

The post-side pilot plant is located at NYSEG's Somerset Station, approximately 40
miles northeast of Buffalo, New York. Figure 2 presents a simplified schematic of
the pilot plant, which employs a recuperative heat exchanger and electric auxiliary
heater to increase flue gas temperature to 625°F for acceptable NOx removal.

The NYSEG/post-FGD process conditions are selected to provide 80% NOx removal
(from boiler concentrations of 400 ppm) and control of residual NH3 to 10 ppm and
5 ppm (at the exit of the second and third catalyst layer, respectively). A fourth
catalyst layer is included to evaluate the additional catalyst and pressure drop
required to reduce residual NH3 to 2 ppm. Similar to the TV A pilot, two different
catalysts can be evaluated at identical process conditions. Pilot design and operating
conditions are presented in
Table 2.

Flue gas is extracted following the exit of the host station's wet limestone flue gas
desulfurization process at approximately 125 °F. The flue gas concentration typifies
that of FGD exit conditions, with low SO2 and particulates (150 ppm and 0.006 gr/scf,
respectively). Design values for the concentration of NOx and O2 at this location are
400 ppm and 6%, respectively. After extraction with the isokinetic scoop flue gas

5A-4


-------
passes through an isolation damper, a venturi to monitor flow rate, and is heated to
550°F by a recuperative (heat pipe) heat exchanger. Two electric heaters provide a
total of 100 kW heating input to further increase flue gas temperature to 625°F. The
gas then enters the reactor tower,, which is identical to the TV A design with the
exception that four catalyst layers are provided instead of five. After exiting the
reactor, flue gas is cooled by the recuperative heater, and exits the process at
approximately 225°F.

TEST PLAN

A test strategy has been developed based on a two year operating period. The test
plan will first establish baseline performance, then implement load-following
operation. Documented changes in catalyst activity over two years will allow
estimating the useful catalyst life. Additionally, a series of measurements will
determine if SCR contributes to or reduces the concentration of trace species and
particulates. For approximately 85% of the operating time, the pilot plant will
operate in a simple load-following mode, and allow for monitoring NOx removal,
residual NH3, and byproduct SO3.

Figure 3 presents the anticipated form of one specific result that will be used to
characterize catalyst performance and lifetime. Figure 3 describes the relationship
exhibited between NOx removal and residual NH3 concentration, as a function of
NH3/NOx ratio. Residual NH3 concentration is relatively constant until an
NHs/NOx ratio of approximately 0.90; further increases in NH3/NOx ratio
significantly elevate residual NH3. Experience with SCR pilot plants and full-scale
applications in Europe, as well as the SCR pilot plant operated by EPRI at the
Arapahoe Test Facility, shows that residual NH3 is one of the most sensitive
indicators of catalyst activity. Accordingly, residual NH3 as a function of ammonia
injected will be periodically documented during the two year tests to characterize
any changes with time. This data, in addition to NOx removal and residual NH3
measured between catalyst layers at selected test conditions, will supplement the
analysis of catalyst samples for use in projecting catalyst life.

Figure 4 depicts the test schedule for the TV A pilot plant. The major components of
the test plan are described as follows:

Baseline. Selected baseline tests completed to date document NOx removal, residual
NH3, and byproduct SO3 as a function of key design variables. Additional tests
scheduled for completion by late April will document the effect of flue gas
temperature, space velocity, and NHs/NOx ratio, among others. A second baseline
test period of 4 weeks is planned after two years.

Load-following. This activity will be fully implemented by June 1991, and will
employ a process control system to simulate actual load-following. The pilot will
operate at a fixed reactor design flow rate of 2000 scfm (1000 scfm per catalyst), but the
ammonia injection will be tailored to maintain a fixed NH3/NOx removal over the
daily variable conditions of inlet NOx, O2, temperature, etc.

Trace Species/Particulate. Over the two year period, two measurement campaigns
will be conducted to determine the fate of trace metals across the reactor, and if trace

5A-5


-------
byproducts (e.g., N2O) are created or removed by the reactor or the NOx reduction
reactions.

Catalyst Activity. At three month intervals, the reactor will be removed from load-
following operation, and selected test conditions from the baseline series repeated.
The reactor will be removed from service and inspected, and catalyst samples
extracted for bench-scale testing by the supplier. The first samples were removed in
late March 1991.

Each catalyst supplier has modified the center catalyst so that samples can be
extracted for further testing and analysis in a bench-scale laboratory rig. Samples
will be tested under well-controlled operating conditions of gas composition and
temperature to define NO removal, allowing catalyst activity to be assessed. In
addition, catalyst suppliers will employ special-purpose diagnostic techniques to
monitor the surface composition. It is anticipated that changes in catalyst activity
will correlate with the surface concentration of trace species suspected to be poisons
for SCR catalysts. Samples will be extracted at approximately 3 or 4 month intervals,
allowing trends in activity and surface composition to be established that can be
used to estimate catalyst life.

RESULTS

As of late March 1991 testing with the TV A pilot plant had progressed
approximately 60% through baseline operation, accumulating almost 2000 hrs (one
fourth year) operation. The NYSEG unit had not yet started operation but was in
the final stages of construction and check-out. Selected results from the TVA unit
are summarized as follows.

TVA.

Two categories of results have been obtained to date with the TVA pilot plant: (a)
process performance data, and (b) operating experience that could minimize
operating problems and maintenance costs at full-scale.

Process Performance. Preliminary measurements defining NO* removal and
residual NH3 as a function of ammonia injection rate are shown in Figure 5. Data
analysis is not yet complete, thus data for each specific catalyst is not identified;
rather the general range of results is shown along with several points for illustrative
purposes. Figure 5 indicates that the catalyst in a new state (e.g. 3 months duty or
less) meets the design performance specifications. The measured residual NH3
concentration is two ppm or less for NH3/NOX ratios less than 0.85. We are
conducting additional diagnostic tests to insure all residual ammonia both in the
flue gas and adsorbed by particulate is accounted for.

Initial measurements of SO3 show flue gas concentration entering the pilot plant is
generally 20-30 ppm, depending on boiler operating factors such as load, excess air,
etc. Measurements also show that depending on the specific catalyst and process
conditions up to 40 ppm SO3 can be added to the flue gas, producing concentrations
exiting the reactor in excess of 70 ppm. The high SO3 content (from both inherent
levels associated with high sulfur coals and SO2 oxidation) compared to Japanese

5A-6


-------
and European applications could be responsible for the two operating experiences
described below.

Ash Deposition. Significant deposits of fly ash adhered to the wall of the SCR
reactor in the initial stages of operation. In general, most of the adhered fly ash was
hardened with a cementitious surface, or glazing. Analysis of surface deposits by
scanning electron micrograph show a high content of sulfate compounds -
specifically calcium sulfates - well above the content usually observed in fly ash. It
is theorized that sulfuric add (from high SO3) condensed on the fly ash, leached out
calcium, and subsequently formed the sulfates. The condensation of sulfuric acid
was likely due to frequent startup/shutdown operation in the early phases of pilot
plant testing, exposing the catalyst to SO3 and moisture at temperatures below the
condensation threshold. These hardened deposits blocked up to 10% of the catalyst
surface, and if allowed to further accumulate, would remove a significant portion of
the catalyst from operating duty.

As a result of this experience, a procedure for proper startup/shutdown was
developed that in principle could be adopted to full-scale. To avoid condensation
during startup the catalyst was preheated with ambient air to above both the flue gas
SO3 and moisture dewpoints (-300 °F and 100 °F, respectively. During shutdown,
the reactor is purged with air as the catalyst cools from operating temperatures (~700
°F) to below the SO3 and moisture dewpoint. This is accomplished at the TV A pilot
plant by installing an inlet valve in the flue gas ductwork to allow ambient air to be
inducted. The ambient air was heated to above 350 °F by either an electric heater
(during startup operation) or the relatively hot duct walls (during shutdown
operation). This experience has been documented and will be used to develop
startup/shutdown guidelines for full-scale.

Deposit Formation On NH^ Injectors. Additional operating experience addressed
ammonia injection equipment. To date, no full-scale installations in Japan or
Europe have reported in the open literature problems with ammonia
sulfate/bisulfate formation on the injector nozzles. However, operation during the
first three months of startup documented the formation of ammonium
sulfates/bisulfates on the injectors in quantities sufficient to block ammonia
injection and/or cause maldistribution of ammonia and reduced NOx removal.
These injectors were of a special design to provide rapid mixing and a uniform
distribution of NH3 and NOx; however the solids deposition is believed possible on
conventional injectors.

The usually reported temperature for deposition of such compounds is
approximately 400°F, based on ammonia and SO3 concentrations of approximately
10 ppm. However, the thermodynamics of these reactions for high sulfur coal
conditions (up to 30 ppm SO3 in flue gas, and ammonia concentration up to 50,000
ppm in the transport air) suggests that such compounds can form at temperatures
up to 625°F. These unique conditions, not previously reflected in full-scale or pilot
tests, could be responsible for persistent deposition at these relatively high
temperatures. As of mid-February this problem at the pilot scale had been remedied
with a special-purpose injection system. In this approach, two injectors are
alternately used, allowing ammonium compound deposits on the injector not in

5A-7


-------
service to decompose to ammonia and SO3. We are presently evaluating concepts
that could be applied at full-scale.

NYSEG Post-FGD.

Installation of this pilot plant was completed in mid March 1991, with check out
activities and startup tests scheduled to begin in late March. The test plan for the
NYSEG unit is similar to that for the TV A pilot plant, and is presented in Figure 6.

SUMMARY

Selective catalytic reduction has been applied extensively in Japan and more recently
in Europe to control NOx emissions to extremely low levels. Although no serious
problems have been reported to date for these low sulfur coal applications, several
critical concerns remain for high sulfur coal application in the U.S. For the
conventional hot-side application, these concerns address primarily catalyst life and
quantity to control residual ammonia, and the quantity and fate of residual SO3
generated by the catalyst. For post-FGD applications, the cost and materials of
construction required for a recuperative heat exchanger that can survive the
potentially corrosive, low temperature environment following conventional wet
FGD processes is critical.

EPRI and member utilities plan tests employing up to six pilot plants to empirically
evaluate these issues for U.S. application. The first pilot plant is addressing hot-side
SCR on high sulfur coal at the TVA/Shawnee Test Facility, with early results
confirming catalyst suppliers predictions for catalyst performance, but identifying
two operating issues that potentially relate to the high SO3 content of flue gas.

A second pilot plant to evaluate post-FGD SCR at NYSEG's Somerset Station will be
operational in April 1991. Results from these pilots and two additional units
planned (at Niagara Mohawk Power Corp. and Central Illinois Public Service) will
be used with EPRI engineering studies to predict with confidence the feasibility and
cost of SCR for U.S. application.

REFERENCES

(1)	"Poisoning of SCR Catalysts," presented at the 1991 Joint Symposium on
Stationary Combustion NOx Control, March 1991, Washington, D.C.

(2)	'Technical Feasibility and Cost of SCR for U.S. Utility Applications", presented
at the 1991 Joint Symposium On Stationary Combustion NOx Control, March
1991, Washington, DC

(3)	'Technical Feasibility and Cost of SCR NOx Control In Utility Applications,"
Draft Report for EPRI Project 1256-7, August 1990.

5A-8


-------
Table 1. Fuels. Furnace Designs Evaluated In The EPRI/Utilitv
Industry SCR Filpt Plant Program .

HOST
TVA

NYSEG"

Niagara
Mohawk

OPS

FUEL
3-4% S

2% S

1% S Oil

3-4% S
*Po$t-FGD

FURNACE DESIGN

Pre-NSPS
(Wall-fired)

'79 NSPS
(Wall-fired)

Pre-NSPS
(Wall-fired)

Cyclone

Table 2. Desien Basis of Pilot Plants



PILOT FEATURE

NYSEG

TVA

Flowrate (scfm)

2000

2000

Number of Catalyst Layers

4

5

Dummy Layer

no

yes

Reactor Temperature (°F)

625

700

Inlet NOx (ppm)

400

600

Inlet SO2 (ppm)

150

2000

Design Performance





- NOx (%)

80

80

- NH3 (ppm)

5

5

Catalyst Manufacturer

• WR Grace

• Joy/KH3

(all honeycomb-type)

* Englehard

• Norton

Catalyst Pitch (mm)

4

6/7

5A-9


-------
Figure 1. Installation Arrangement of 1MW SCR Pilot

Plant At TVA

• 1 I Cm 111 I IV

t EL 430'—5*"

COWER ** OF MANLIFT
FENCING {BY FIELD)

¦ PORTABLE
SUPS

5A-10


-------
Figure 2. Schematic Of Post-FGD SCR
Pilot Plant at NYSEG's Somerset Station

From
Scrubber
Outlet
(125°F)

Return
To Plant
(250°F)

Recuperative Heat
Exchanger

Gas out

Electric
Heater

F.D. Fan


-------
Figure 3. Anticipated Relationship Between
NOx Removal and Residual NH3 vs. Time

-90%

NOx
Removal

NH3,
ppm

y

X

3 months
(Baseline)

.90

Ammonia/NOx Ratio (moles)


-------
Figure 4. Test Schedule For
TVA High Sulfur Pilot Plant

Activity 6/90 9/90 12/90 3/91 6/91 9/91 12/91 3/92 6/92 9/92 12/92 3/93



1.	Start-up

2.	Sampling/Analytical Trials EK3

3.	Baseline

4.	Load-Following

5.	Catalyst Activity

6.	Trace Species/Particulate

7.	Second Baseline






-------
Figure 5. Relationship of NOx Removal,
Residual NH3 - Preliminary TVA Baseline Results

NOx

100

95 +

Removal gn

85
80

75 ~

.80 .85 .90 .95 1.0 1.05
Ammonia/NOx Ratio (moles)


-------
Figure 6. Test Schedule For
NYSEG Post-FGD Pilot Plant

3/91 6/91 9/91 12/91 3/92 6/92 9/92 12/92 3/93 6/93

Activity

1.	Start-up

2.	Sampling/Analytical Trials ESJ

3.	Baseline	ISMS!

4.	Load-Following	¦ i

5.	Catalyst Activity	^

6.	Trace Species/Particulate

7.	Second Baseline

® ® ^ ®

mm


-------

-------
PILOT PLANT INVESTIGATION OF THE TECHNOLOGY OF SELECTIVE
CATALYTIC REDUCTION OF NITROGEN OXIDES

Shiaw C. Tseng, Wojciech Jozewicz
Acurex Corporation
P.O. Box 13109
Research Triangle Park, NC 27709

Charles B. Sedman
Gas Cleaning Technology Branch, MD-04
Air and Energy Engineering Research Laboratory
U.S. Environmental Protection Agency
Research Triangle Park, NC 27711

This paper has been reviewed in accordance with the U.S. Environmental
Protection Agency s peer and administrative review policies and approved for
presentation and publication.

5 A-11


-------

-------
ABSTRACT

The U.S. Environmental Protection Agency has built a bench scale
pilot plant to investigate the ammonia (NH3) based technology for
selective catalytic reduction (SCR) of nitrogen oxides (N0X) . A
key objective of this task is to establish the performance of
commercially available SCR catalysts on U.S. fuels and combustion
sources.

One rudimentary catalyst produced in-house and two commercial
catalysts were tested over the temperature window of 327 to 440°C.
The space velocity (SV) ranged from 7, 650 to 36,500 hr_1, The
combustion gas was doped with nitric oxide (NO) and NH3, and the
NH3/NO'ratio ranged from about 0.6 to 2.2. Sulfur dioxide (S02)
was added to the combustion gas in some runs to investigate its
effect on NO conversion. The results obtained indicate that the SV
has a significant effect on the conversion of NO for the in-house
catalyst which was prepared primarily"for start-up of this system
before the commercial catalysts arrived. For the two commercial
catalysts, the NO conversion was 90% and higher when the NH3/NO
ratio was near or above unity. For the same catalysts, the NO
conversion was approximately proportional to the NH3 concentration
at the inlet of the reactor, when the NH3/N0 ratio was below unity.
For one commercial catalyst,, the NO conversion was .lower when 95
ppm of S02 was present in the flue gas. Over the same catalyst,
the amount of nitrous oxide (N20) formed was practically
negligible. The difference of activity between the in-house and
the commercial catalysts is attributed to the difference, in
chemical composition.

5A-19


-------
INTRODUCTION

The emission of nitrogen oxides (N0X) into the atmosphere
contributes to the degradation of air quality as well as to acid
rain and forest damage.N0X is formed during the combustion of
fossil fuel. Part of the oxides come from the thermal oxidation of
nitrogen in the combustion air (thermal N0X) . Thermal NO*
generally increases when the combustion temperature is increased.
The remaining N0X comes from the oxidation, of nitrogen-containing
species originally present in the fuel (fuel N0X) . Compared with
thermal N0X, fuel NOx is not as sensitive to combustion temperature
and depends highly on the reactant stoichiometry.The KQX
emissions can be reduced by several approaches such as in-furnace
NOx reduction, selective non-catalytic reduction (SNCR), and
selective catalytic reduction (SCR).

Some in-furnace N0X reduction technologies involve modification of
the combustion process to reduce peak flame temperature and create
fuel-rich conditions by reducing the ratio of fuel to combustion
air. . Other in-furnace N0X reduction technologies include
3^ e du ce d— a x r p r e h e at ,, 1 o ad r e duct ion, 1 o^^ e x ce s s ar r, fl ue—g a s
recirculation, overfire air, deep-air .staging, fuel staging {or
reburning), and various low N0X burner systems. :2) In-furnace
reduction technologies could result in lower combustion efficiency
and higher CO emissions. 125

In SNCR processes, ammonia (NH3) or aqueous urea solution is
injected into the combustion chamber. <3! The vaporization of water
reduces the flame temperature and the reducing agent reacts with
NOx to form nitrogen and water. Since no catalysts are employed,
the N0X reduction reaction proceeds at the combustion temperature,
and the combustion efficiency is usually reduced.

Removal efficiencies of N0X ranging from 20 to- 80% have been
reported by in-furnace NOx reduction'1* and SNCR^3' , technologies.
However, simultaneous deployment of several of these technologies
is often required to achieve the targeted emission level.
Furthermore, complicated mechanical modifications are involved, and
the application of these technologies has to be reviewed on a
case-by-case basis. ,

SCR is an established technology capable of removing 80 to 90% of
the NOx present in the flue gas.<1,4' This technology was first
commercialized in Japan and is widely utilized in Europe to control
NOx emissions from fossil fuel fired power plants.11,1 The SCR
processes have the advantage of being applicable to all types of
conventional boilers and even municipal solid waste incinerators.
The SCR unit can be incorporated into the present process in three
configurations. It can be placed upstream of the air preheater
(the high-dust system), between the electrostatic precipitator (the
low-dust system) and the flue gas desulfurization (FGD) unit, or
'downstream of the FGD unit (the tail-end system).

In SCR processes, anhydrous or aqueous NH3 is injected into the

5A-20


-------
flue gas upstream of a catalyst bed. 'In the 'presence of oxygen,
NH3 reacts with nitric oxide (NO) and nitrogen dioxide (N025 at the
catalyst surface to produce nitrogen and water:<5)

.4 NH3 + '4 NO +
4 NH3 + 2 N02 +

In the U.S., there is little electric utility experience with SCR
NGX control techniques. A demonstration of this technology is
being undertaken by Southern Company Services, Inc.but no'data
have been reported yet. Several co-generators (located mostly in
California) are testing these technologies; howev,er, the
operating data of these facilities are not readily shared, and the
performance of the units is not easily verified. .

The Japanese and European experience with the SCR technologies
cannot be blindly applied to the U.S. There remain two significant
uncertainties about design, performance, operating parameters, and
cost of the SCR-technologies. First, U.S. electric power plants
operate under more variable loads. Second, the' amounts and types
of trace elements in U.S. coals are different from those in the
fuel consumed in Japan and Europe.

Acurex Corporation operates U.S. Environmental Protection Agency's
pilot plant which is designed to evaluate commercially available
catalysts used in the NH3 based SCR technologies. A key objective
of this task is to establish the performance' of commercially
available SCR catalysts on U.S. fuels and combustxon sources.

Reported in this paper are the preliminary results obtained by
testing catalysts from three sources over the typical SCR
temperature window ranging from 327 to 440°C. The effects of
temperature, space velocity, and NH3/NO ratio on the conversion of
NO according to Equation 1 were examined. The amount of N02
detected in the combustxon gas was very small, about 5 ppm, and the
reaction according to Equation 2 was therefore neglected. The
possible poisoning effect of flue' gas sulfur dioxide (S02) on the
NO conversion was investigated. The issue of the formation of \'2C,
a greenhouse gas,' over the catalysts was also examined. To the
best of our knowledge, such data have not been reported.

EXPERIMENTAL

Pilot Plant Test Facility

Figure 1 shows a schematic diagram of the pilot plant facility used
in this work. The facility includes (1) a simulated flue gas
generating station consisting of a natural gas burner, NO cylinder,
SO, cylinder, and 5% NHo in air cylinder, (2) a section of about 3
m of' heated combustion gas transport duct, (3) a reactor which is
also externally heated, (4) a dust collecting system consisting of

-> 4 N-

+ 6 KoO

0,

-> 3 N2 + 6 H20

(1)

(2)

5A-21


-------
a cyclone separator, a dust collector, and a ceramic particulate
filter, (5) a section of -3.6 m long exhaust duct, and (6) a flyash
feeding mechanism. The above components, except for the flyash
feeder, were made of stainless steel (SS); the flyash feeder was
made of Pyrex glass. Not shown in this diagram were the air
preheater, the control panels for the natural gas burner and the
flyash feeder, and mass flow controllers for natural gas and
combustion air including burner and dilution air.

The natural gas burner is rated at 2,110 W. The burner is equipped
with a Fenwal Series 05-14 ignition/proof-of-flame mechanism which
provides positive ignition of the burner when heat is required and
therefore eliminates the need for pilot burners. The ignition
spark operates until the flame is established and then is
immediately shut off. A positive rlame sensor rs installed to
detect the ionized species present in the combustion chamber during
normal burning of the natural gas. If the flame is not present or
the ionized species are below the detection limit of the flame
sensor, the burner management system will shut off the natural gas
supply valve automatically until the flame is re-established. For
each re-ignition, air will purge the combustion chamber for
15 seconds (approximately 9 combustion volume changes) to ensure
that no residual natural gas remains in the combustion chamber. A
rupture disc made of aluminum foil is teed to the outlet of the
burner for additional safeguard.

The reactor was originally made of Pyrex glass with the dimensions
of 5.1 cm O.D. k 60 cm long. The breakage and frequent replacement
of the Pyrex tube were alleviated by employing a SS tube of the
same dimensions in later runs. . Blank runs with no catalyst were
made1 and both confirmed no reduction of NO by' the SS tube. A 5.1
cm square SS tubing was also used to test a square catalyst. The
reactor section is also externally wrapped with a beaded heater to
aid maintaining the temperature.

A cyclone separator is installed downstream from the reactor to
remove all particulate matter from the flue gas. The dust is
separated and accumulated in the collector at the bottom of the
cyclone. The gas' leaving the cyclone passes through the
particulate filter and is then vented into the tubing connected to
the main exhaust pipe.

The body of the flyash feeder is made of Pyrex glass columns' in
two sections. A maximum of 1,500 g (1 week supply) of flyash can
be charged into the bottom section of the feeder. The flyash
particles are then air-fluidized and fed into the flue gas stream
at a nominal rate of' 1.10 g/tr.in through two 0.16 cm O.D. SS lines
alternatively. To avoid pressure buildup in this feeder due to
possible clogging of the tubing, two solenoid valves are employed
so that when one tubing is feeding flyash into the system, the
other tubing is back-flushed with air to sweep any particles back
into the fluidizing chamber. The fluidizing air, "passing through
a glass filter mounted on the upper section of the feeder, is then
vented via SS tubing tapped into the main exhaust duct.

5A-22


-------
The composition of the combustion gas can be adjusted • by
introducing NO, S02, and NH3 from ' cylinders to bring the
concentrations of these gases to the desired level. The combustion
gas leaving the burner was first mixed with NO and S02 at a common-
port midway of the combustion gas transport duct. Anhydrous
ammonia (5% KH2 in air) was introduced into the reactor through a
port located at the 180° connecting elbow between the reactor and
the transport duct. The distance between this port and the
reactor is about 20 cm.

All hot spots of the unit, the burner, combustion gas transport
duct, reactor, and the cyclone/duct collector ar.e all thermally
insulated. The flame temperature and the gas temperatures at the
outlet end of the burner and the inlet and outlet or the reactor
are constantly monitored.

The test facility is operated at ambient pressure. The- nominal
gas flowrates are given in Table 1. All the flowrates are measured
at ambient temperature.

Operating Procedures

Catalyst blocks were first loaded into the reactor, usually 1 day
ahead of the scheduled test date. The air preheater and the beaded
heater were then turned on to.keep the reactor at-a temperature of
at least 150°C to prevent moisture from condensing on the catalysts
overnight. The natural gas burner was then fired up the next
morning at a proper fuel/air ratio, and the preheater was turned
off. Once the reactor temperature rose steadily, the fuel/air
ratio was theiv adjusted to keep the reactor temperature at the
desired value. The flue gas temperatures at the inlet and outlet
of the reactor were constantly monitored. A temperature difference
of less than 5°C could routinely be achieved.

As soon as the targeted ire act c r temperature was reached,	NO,

and S02 were then introduced into the combustion gas. The NO
concentrations at the inlet and outlet of the reactor were then
measured and NO conversion was then calculated.

Catalysts

More than 10 catalyst vendors were invited to participate in this
program by providing their SCR catalysts. So far only three of them
have provided catalysts for this work. Since these commercial
catalysts arrived rather late, EPA had to make its own catalyst for
system start-up. The catalysts were labelled 1A, 2A, 2B, 3k, and
4A. Catalyst 1A was made in-house. The others were commercial
catalysts. Catalysts 1A, 2A, and 3A were tested. Testing of
Catalysts 2B and 4A is in progress. Described below is the
information on Catalysts 1A, 2A, 2B, and 3A.	information

regarding Catalyst 4A will be reported after testing is completed.

Catalyst 1A was made by coating a cordierite (a form of ioliteor
silicate of aluminum, magnesium, and iron) substrate with titanium

5A-23


-------
dioxide (Ti02) and subsequently with vanadium pentoxide {V2O5)',
A 15.2 x 15.2 x 7.6 cm, cordierite,block was cut into six pieces of
~4.5 cm O.D. x 7,6 cm long substrate. Each piece was first coated
with Ti02-containing solution (concentrated H2S04) followed by
calcination at 4750C for 4' hours. A V205-containing solution
(diluted H2S04) was then dip-coated on the calcined substrate,
followed by calcination at 450°C for 4 hours. The catalyst blocks
obtained were brownish yellow, but not uniform due to the dip-
coating procedures used.

Only very limited, information on the commercial catalysts was
released by the suppliers. Catalysts 2A and 2B were extruded
V205/Ti02 based materials. Catalyst 2A, with a catalyst-flue gas
contact area of 910 m2/m^, is marketed for clean-gas applications.
Catalyst 2B has a contact area of 470 m2/m3 and is for high-dust
applications. Both catalysts are square with dimensions of 4.4 x
4.4 x 50 cm. Catalysts 2A and 2B are green and light yellow,
respectively. (Note: Catalyst 2A contained some tungsten.) For the
reactivity test, a single piece of Catalyst 2A was used.

Catalyst 3A was a greyish brown,- extruded precious-metal-based
ceramic material, 3.5 cm O.D. x 7.6 cm long. For the reactivity
test, six blocks of this catalyst were used.

Measurement of NO!f, SOg, and N^O

Model 10A NQX and Model 40 S02 analyzers, both made by Thermo
Electron Corporation, were employed to perform on-line measurements
of NOx and S02 concentrations, respectively. A Hewlett-Packard
Model 5980 gas chromatograph (GO equipped with an electron capture
detector (ECD) was employed to measure the concentration of N20.
Such an instrument had been deployed to measure flue gas N20
concentration from various combustion sources. <7) A back-flush
feature has been incorporated into this GC to keep the column from
being contaminated; however, the previous method did not have this
feature.

RESULTS AND DISCUSSION

The resources and time available limited the experimental matrix to
those runs without flyash addition to the flue gas system. The
results reported below are preliminary. Experimental work is in
progress to derive more kinetic information such as flowrate (SV)
effect - and values of activation energy and pre-exponential
constant.

Catalyst 1A

Shown in Figure 2 is the performance of Catalyst 1A at two space
velocities, 7, 650 and '18', 400 hr"1. The temperatures differed
slightly, 360 and 35Q°C, respectively. The NH3/N0 ratio was
changed from about 0.6 to 1.75, at the inlet of the reactor. Space
velocity (SV) is defined as the volumetric flue gas flow rate at

5A-24


-------
the reactor temperature to the geometric volume of the reactor.
The results show that when SV was reduced from 18,40-0 to 7,650
hr-1, the averaged NO conversion increased from about 17 to 67% in
the range of NH3/NO ratios tested.

Catalyst 2A

Effect of Temperature

Shown in Figure 3 are the NO conversions at three temperatures:
327, 360, and 406°C. The NH3/NG ratio was from'0.7 to 1.45. The
results indicate that" the NO conversion is not sensitive to the
reaction temperature. In such a temperature range.more than 90% of
NO is reduced when the NH3/NO ratio is near and above unity. When
the NHj/NQ ratio is below unity, the NO conversion is approximately
proportional to- the amount of NH3 entering the reactor, as
indicated by the dotted line. This result is in agreement with
that observed by others employing vanadia/titania-silica
catalyst.1 '

Effect of S02

Figure 4 summarizes the effect of SO2 on the performance of
Catalyst 2A. The concentration of S02 is 95 ppm. The NH3/NO ratio
was varied from 0.65 to 1.25. The result indicates that this
catalyst is more active without the presence of S02. Shown in
Figure 5 is the performance of the same catalyst at two
temperatures, 353 and 440°C, and in the presence of 95 ppm SO2. The
results indicate that the catalyst is less active at 4400C than at
353°C, when 95 ppm of S02 is present in the flue gas.

N20 Measurements

The results of N20 measurements over Catalyst 2A are listed in
Table 2. The NH3/NO ratio was varied from 0.5B6 to 2.17. The
reaction temperature was 400°C. The space velocity was calculated
to be 13,790 hr-4. No S02 was added to the combustion gas. The
results shown in Table 2 indicate that practically no N20 formed
over Catalyst 2A at the testing conditions chosen. This fact is
very significant because N2Q is a greenhouse gas which has been
blamed for both increasing the global temperature	and

n 1 l 7\

destroying stratospheric ozone. ' ' There was N20 in both the- NH3
and NO tanks.

Catalyst 3A

Shown in Figure 6 is the performance of Catalyst 3A. The test was
conducted at a temperature of 34 0 °C and SV of 36, 500 hr-1. The
NH3/NOs ratio was varied from 0.75 to 1.25. More than 90%
reduction of NO is achieved when the NH3/NO ratio is above unity.
The amount of NO removed is approximately proportional to the
amount of NR3 present when the NH3/NO ratio is below unity. This
result is in agreement with that observed by others.(8)

5A-25


-------
Performance Comparison

The performance of the three catalysts tested in the temperature
range of 340 to 360°C is shown in Figure 7. The results indicate
that Catalyst 1A is the least active. Although the exact reasons
have not been investigated, it is possible that the difference in
reactivity between the in-house and the two commercial catalysts is
due to the difference in chemical composition which is reflected by
the difference in color of the catalysts tested. It is also likely
that the catalytic activity can be affected by the conditions under
which the catalysts were made.

CONCLUSIONS

The U.S. Environmental Protection Agency has built a pilot plant to
investigate the ammonia (NH3) based technology of selective
catalytic reduction (SCR) of nitrogen oxides. One in-house
catalyst and two commercially available catalysts were tested. The
effects of temperature, space velocity CSV), and NH3/NO ratio on
the conversion of NO were investigated. In some runs, sulfur
dioxide (S02) was added to the combustion gas to investigate its
effect on the performance of a commercial catalyst. The formation
of nitrous oxide (N20)' was also examined.

For the in-house catalyst, the SV has a significant effect on NO
conversion at about 350°C. The NO conversion increased from an
average value of 17 to 67% as the SV was decreased from 18,400 to
7, 650 hr"1. For the two commercial catalysts, the NO conversion
was 90% and higher when the NH3/NO ratio was near or above unity.
For these two catalysts, the NO conversion was approximately
proportional to the NH3 concentration at the inlet of the reactor
when the NH3/N0 ratio was below unity.

The NO conversion was found to be temperature insensitive for one
commercial catalyst tested at three temperatures, 327, 360, and
406°C. For the same catalyst, flue gas S02 was found" to be
poisonous, and the poisonous effect of S02 was more severe at 440°C
than at 353°C. At 400°C, NH3/NO ratios ranging from 0,586 to 2.17,
and SV of 13, 790 hr"1, the amount of N20 formed over the same
catalysts was negligible.

The difference of activity between the in-house and the commercial
catalysts is attributed to the difference in chemical composition
and how the catalysts were made.

5A-26


-------
REFERENCES

(1) Mclnnes, R.G. and Van Wormer, M.B., Cleanup ICO... emissions.
Chem. 5r.gr.. 130, 1990.

(25 U.S. Environmental Protection Agency, "Control Techniques for
Nitrogen Oxides Emissions From Stationary Sources-Revised
Second Edition," EPA-450/3-83-0Q2 (NTIS P384-118330), 1983.

(3)	U.S. Environmental Protection Agency, "Municipal Waste
Combustion-Background Information for Proposed Standards:
Control of NOx Emissions, Vol. 4," EPA-45Q/3-89-27d (NTIS
PB90-154873), August 1989, p. 3-9.

(4)	Eskinazi, D., Cichanowicz, J.E., Linak, W.P., ana Hall, R.E.,
Stationary combustion N0X control. A summary of the 1909
symposium. JAPCA, 39(8): 1131, 19B9.

(5)	Schonbucher, B., Reduction of nitrogen oxides from coal fired
power plants by using the SCR process—Experiences in the
Federal Republic of Germany with pilot and commercial scale
DeNOx plants. In, Proceedings:1989 Joint Symposium on
Stationary Combustion N0X Control, Vol. 2, EPA-6C0/9-89-062b
(NTIS PB 89-220537), June 1989, p. 6A-1.

(6)	U.S. Department of Energy, "Comprehensive Report to Congress
Clean Coal Technology Program. Demonstration of Selective
Catalytic Reduction (SCR) Technology for the Control of
Nitrogen Oxide (NOx) Emissions from High-sulfur-coal-fired
Boilers. A Project Proposed by Southern Company Services,
Inc.," DOE/FE-0161P, April 1990.

(7)	Linak, W.P., McSorley, J.A., Kali, R.E., Ryan, J.V.,
Srivastava, R.K., Wendt, J.O.L., and Mereb, J.B. N20
emissions from fossil fuel combustion. _In Proceedings:
1989 Joint Symposium on Stationary Combustion NOx Control,
Vol. 1, EPA-600/9-89-062a (NTIS P389-220529), June 1989,

p. 1-37.

(8)	Odenbrand, I.C.U., Lundin, S.T., and Andersson, L.A.H.,
Catalytic reduction of nitrogen oxides. 1. The reduction of
NO. Appl. Catal., 18: 335, 1985.

(95 Donner, L. and Ramanathan, V., Methane and nitrous oxide:
Their effects on the terrestrial climate. J. Atmos. Sci. 37:
119, 1980.

(10) Wang, W.C., Yung, Y.L., Lacis, A.A., Hoe, T.M., and Hansen,
J.E., Greenhouse effects due to man-made perturbations of
trace gases. Science. 194: 685, 1976.

5A-27

I


-------
(11)	Crutzen P.J., Ozone production rates in an oxygen-hydrogen
nitrogen oxide atmosphere. J. Geophvs. Res. 76: 7311, 1971.

(12)	Weiss,- R.F., The'temporal and spatial distribution of
tropospheric nitrous oxide. J.1 Gecphvs. Res. 86; 7185, 1981

5A-28


-------
Vent

Figure 1. Schematics of the bench scale pilot plant facility for testing SCR DeNOx catalysts.


-------
•100 -

f. 80

c

o

-A

m

u

-------
100

. 0 o o °

O

o

o

u
a>



SO 2
{ppm)

T

<°C)

SV

(ht *x)

¦

None

3S0

13,36Q

0

95

353

13,370

0 i—:	>	1	•	r—			1

0.0	0,5	1.0	1.5

NK 3 / NO ratio

Figure 4. Effect of S02 on the performance of Catalyst 2A.

CataXyat 2A

80

60"

100

80

c

.2 so

m
M
SI

40

20

0

OX)	05	1.0	1.5

H»3 /NO ratio

Figure 5. Effect of temperature an the performance of
Catalyst 2A in the presence of 95 ppm SO2.

Catalyst 2k.
SO 2 — 95 ppm



0 O 0
0

0





. *

O •

•



•





•







T

SV



<°c>

(hr " 1 )

0

353

¦13,370

•

440

11,680

5A-31


-------
o

2

100 •

80

SO

40 '

20

0 . 0

A
' ~

-a	 	o-

0

/o

Catalyst 3A

T

0

340 C

SV

-1

36,500 hr

0.5	1.0

NH /NO ratio

3

Figure 6. Performance of Catalyst 3A.

l. s

H

0)
>

e

a
u

o

35

100

80

60

40

20

0.0

\ Da'

4

* ~

4)
*

Catalyst

T( °C)

-1

SV(hr )

* 1A

360

7, 650

D 2A

360 .

13,360

* 3A

340

36.500

0.5	1.0

1 ,5

NH 3 /NO ratio

2.0

Figure 7. Comparison of catalyst performance.

5A-32


-------
Table 1. Nominal Gas Flowrates

total gas (liters/min)

12C |

combustion air (liters/min)

105

natural gas (liters/min)

5 |

NO (ml/min)

120

S02 
-------

-------
POISONING OF SCR CATALYSTS

Jianping Chen, Ralph T. Yang*
Department of Chemical Engineering
State University of New York
Buffalo, New York 14260

J. Edward Cichanowicz
Generation & Storage Division
Electric Power Research Institute
Palo Alto, California 94303


-------

-------
ABSTRACT

Results are summarized from a comprehensive study of the activity of 5%
V2O5/Ti02 catalysts for SCR, addressing the influence of all major possible poisons
encountered in combustion gases. The strongest poisons are the alkali metal oxides.
The effects of the strong poisons are compared for two catalysts: 5% V2Os/Ti02 and
8.2% WO3 + 4.8% V205/Ti02, the latter being similar to commercial SCR catalysts.
The addition of WO3 increases both the catalyst activity and the resistance to
poisoning. A general observation from this study is that the strength of the poison
is directly related to its basicity. Concerted experimental and theoretical results
indicate that the Bronsted acid sites are the active sites for SCR. Deactivation is
caused by reducing the strength and the number of these sites. Results also show for
this case of pure compounds (e.g., without real effects of pore plugging and
blocking), SO2 in the gas phase can either decrease or increase SCR activity for ..
tungsten-containing V205/Ti02 catalysts, depending on other trace elements present
on the catalyst surface.	1

* Corresponding Author.

5A-37


-------
INTRODUCTION

This paper updates the status of a fundamental investigation into the role of trace
elements in coals on catalyst poisoning for the selective catalytic reduction (SCR)
reaction. The objective of this research is to identify and assess the role of potential
poisons for SCR catalysts, particularly for application to high sulfur coal. Although
considerable research has been conducted in this area, no systematic analysis of the
effect of potential catalyst poisons on catalyst activity in high sulfur coal is available
in the "open literature. Results from this effort will support analysis of data from the
1 MW pilot plant tests sponsored by EPRI to evaluate catalyst performance and
activity with authentic fuels.

The results of the initial phase of this activity were reported at the 1989 Symposium
on Stationery Combustion NOx Control in San Francisco, and summarized in
reference (1). Results from the initial investigation identified the alkali metal
oxides as the most potent poisons for vanadium-based catalysts (without tungsten
oxide), with relative poisoning strength increasing with basicity. Other elements
such as lead and arsenic were identified as exhibiting a poisoning effect on SCR
activity.

This phase of the research addresses the poisoning influence of these and other
elements on SCR catalysts that include tungsten oxide (W03), thereby more closely
simulating the composition of catalysts in commercial applications. In addition, the
effects of SO2 are included in this study. To aid in understanding the nature of
active sites and the mechanism of poisoning, several special-purpose diagnostic
techniques were included in this phase of the study. These are Proton NMR,
Extended Huckel Molecular Orbital Calculations (EHMO), and NH3 chemisorption
results.

SCOPE

The scope of this research is to identify changes in catalyst activity due to strictly
chemical effects of pure compounds that are potential poisons. It is important to
note that this investigation is not intended to simulate the actual mechanism of
poisoning of SCR catalysts with real fuels. In actual commercial application,
additional factors such as blockage or plugging of pores, or the physical obstruction
of active sites to access by the reactants is important. Also, this study at present does
not address the details of the surface conditions with real fuels, such as the
distribution and concentration of multiple poisons. Rather, purely chemical

5A-38


-------
influences of single compounds are addressed. The role of sulfur in the context of
this fundamental evaluation is confined to the chemical influence of SO? as a gas,
in conjunction with other species that form on the catalyst surface. Insight into the
real deactivation mechanisms in authentic fuels will be addressed with analysis of
catalyst samples from the 1 MW SCR pilot plants operated by EPRI-member utilities,
described in a companion paper at this Symposium (2).

EXPERIMENTAL

Details of the preparation of the Ti02 support were described in our previous paper
(1). Titanium dioxide powder (P-25, Degussa) was mixed with distilled water at a
ratio of 1:1.75 by weight. The resulting paste was first dried in air at 60°C for 24 hours
and then at 120°C for 72 hours. After drying, the bulk, titanium dioxide was crushed
and sieved. The fraction between 20-32 mesh was collected and calcined at 600°C in
air during the first hour, and then in He during the following six hours. The BET
surface area of this support was 30.6 m2/g, which was measured by a Quantasorb
surface area analyzer.

The composition of the W03-V20s/Ti02 catalyst was the same as that described for a
commercial SCR catalyst (3). The catalyst was prepared by co-impregnation of an
aqueous solution of NH4VO3 and (NH4>6 H2W12O4O in oxalic acid. After
impregnation, the catalyst was dried at 120°C for 15 hours and then calcined at 500°C
in oxygen flow for 20 hours to decompose the ammonium salts into oxides.

The elements identified as potential poisons for the SCR reaction in the earlier work
are alkali (Li, Na, K, Rb, Cs, Ca), as well as arsenic (As), phosphorous (P), lead (Pb),
and HC1. Accordingly, these elements were evaluated in this study by impregnation
via incipient wetness with the precursor solutions of corresponding salts on the
V2O5/T1C2 or W03-V20s/Ti02 catalysts. The precursor solutions for the alkali
oxides, Li20, Na20, K2O, Rb20 and Cs20 were, respectively: Li Ac, NaN03, KNO3, RbAc
and CsAc. For CaO, PbO, AS2O3 and P2O5 doped catalysts, aqueous solutions of
Ca(Ac)2, Pb(Ac)2, AS2O5 and P2O5, respectively, were used. The impregnated'
catalysts were dried at 120°C for 3-4 hours followed by calcination to decompose the
precursor salts.

The experimental setup and procedure were the same as reported earlier (1, 4).
Briefly, the reactor was a quartz tubular reactor in which 1-2 cm^ of catalyst particles
were supported on a fritted glass. The temperature was controlled by a
thermocouple in a quartz well inserted in the catalyst bed. The NO conversion was

5A-39


-------
measured by the effluent NO concentration. The reactant flowrate and the catalyst
particle size were chosen in a manner that the rates were free of mass transfer
effects. The reactant gas supply was controlled by using rotameters for higher flow
rate gases and mixtures (i.e., N2, NH3 + N2, and NO +N2) and by using mass flow
controllers (FM 4575, Linde Division) for lower flow rate gases (SO2 and O2). The
premixed gases (0.8% NO in N2, and 0,8% NH3 in N2) were supplied by Linde
Division.

The walls of the gas mixing system were heated with heating tapes to maintain their
temperatures above that for formation and deposition of ammonium sulfates.

Also, to avoid possible analytical errors caused by the oxidation of ammonia in the
converter of the chemiluminescent NO/NOx analyzer, an ammonia trap was
installed prior to the sample inlet of the analyzer (1, 4). The NO concentration was
continuously monitored by a chemiluminescent NO/NOx analyzer (Thermo
Electron Corporation, Model 10).

The first order rate constants were calculated by the following formula:

Fo

k = .	ln(l-X)

[NO]0W

where Fo is the inlet molar flowrate of NO, [NO]0 is the inlet molar concentration,
W is the amount of catalyst, and X is the fractional conversion of NO* which is
defined as :

X= ([NO]in -[NO]out)/lNO]in

5A-40


-------
RESULTS AND DISCUSSION

Effects of Alkali Oxides. Arsenic Oxide and Chlorides

The potential poisons which were studied in this work included alkali oxides,
alkali-earth oxides, phosphorous, arsenic oxides, lead oxide, and chlorides.

The poisoning effect is expressed in terms of the decrease of the first order rate
constant vs. poison doping amount. To identify the poisoning mechanism and the
role of WO3 in V'205-based catalysts, two groups of catalyst were prepared to compare
the effects of WO3 [i.e., 5% V205/Ti02, compared to 8.2% WO3 + 4.8% V2Cs/Ti02l-

The poisoning effects on the 5% V205/Ti02 catalyst are shown in Figure 1. Of the
various poisons, alkali oxides are the strongest. Comparing alkali and alkali-earth
metal oxides, the poisoning effect is directly related to their basicity. An oxide with a
higher basicity gives a stronger poisoning effect.

Compared to alkali, lead oxide is a medium-strength poison for SCR, and AS2O3 and
P2O5 are both weak poisons.

Figure 1 also shows the change of rate constants vs. M/V {M=metal atoms) over the
W03-V20s/Ti02 catalyst. With the addition of WO3, the rate constant increased from
10.38 to 13.58 cm3/g/s for the catalysts with no poison doping. Moreover, catalysts
containing WO3 always exhibited higher activities than those without WO3 for the
same amounts of poisons. Figure 1 also reveals that the addition of WO3 to the
catalyst not only increased the catalytic activity, but also improved the resistance to
alkali oxide poisons. Again, AS2O3 is a weak poison compared to alkali for the WO3-
V205/Ti02 catalyst.

The effects of chlorides were more complex. Both promoting and poisoning effects
were observed, depending on the overall basicity of the chlorides. Experiments with
NaCl and KC1 doped catalysts showed a weak poisoning effects. The overall effect of
these compounds was a net result of poisoning by alkali and promoting by chlorine.
In fact, a small amount of NaCl acted as net promoter for the \r205-based catalysts in
SCR. Some transition metal chlorides are actually active catalysts for SCR. For
example, 2% CuCl/Ti02 gave a 99./3% NO conversion at 250°C (with 1000 ppm each
of NO and NH3 at 15000 hr"l).

5A-41


-------
The effects of hydrogen chloride on SCR activity depended strongly on the reaction
temperature. HC1 was found to significantly deactivate the SCR catalyst (1). The
deactivation was stronger at 300°C than at 350°C. The likely cause for the HC1
deactivation is as follows. First, the formation of NH4CI by the reaction of HC1 with
NH3 consumed the reactant NH3. Second, the reaction of HC1 with V2O5 forming
vanadium chlorides decreased the concentration of the active component of the
catalyst. Third, the deposition of NH4CI on the catalyst surface below 340°C blocked
the active surface area, which was the reason that the deactivation was more
pronounced at 300°C than 350°C.

Effects of Sulfur Dioxide

The results of SCR with SO2 and without SO2 are listed in Table 1 and Figure 2.
Similar to the case of chlorides, S02 can be either a promoter or poison. Without
the presence of doped poisons, S02 reduces catalyst activity. Alternatively, a strong
promotion effect is noted for catalysts doped by poisons.

Figure 2 shows the effects of SO2 on SCR activity over the W03-V205/Ti02 catalysts
doped with various amounts of alkali oxides. SO2 significantly decreases the
activities of the undoped catalysts, but increases the activity of the doped WO3/V2O5
catalysts for low concentrations of poisons. For example, Figure 2 shows that when
Na/V < 0.5, the activity of the doped catalyst actually exceeded that of the undoped
catalyst due to the presence of S02 in the gas phase. In the presence of SO2, the
minimum NO conversion reached 98% even at an atomic ratio of M/V (M=Na, K)
of 0.5. Alternatively, for K/V > 0.5, the net effect is a decrease in catalytic activity.
This result indicates that although the addition of SO2 initially recovered the
catalytic activities of these doped catalysts, catalyst activity eventually decreases.

The ability of SO2 to resist the poisoning effect of alkali oxides was probably caused
by the gas-solid reaction of SO2 (or SO3) with the alkali oxides. The gas-solid
reaction reduced the surface basicity of the catalyst by forming surface sulfates.
Sulfates are known to possess Bronsted acidity when water is chemisorbed on the
surface. Our recent study on transition metal sulfates (iron, cobalt and nickel
sulfates) indicated that these sulfates are highly active SCR catalysts even at near
ambient temperatures (4).

5A-42


-------
ANALYSIS OF DATA

The above results show that the W03-V20s/Ti02 catalyst yields a higher SCR activity
and a stronger alkali poison resistance than the V20s/Ti02 catalyst. In order to
obtain an understanding of the role of the WO3 in the W03-V20s/Ti02 catalyst,
ammonia chemisorption experiments were performed on a series of catalysts with
different K2O dopant amounts. Figure 3 shows the ammonia chemisorption
amounts at different doping amounts of K2O or Cs20 over V20s/Ti02 and WO3-
V20s/Ti02 catalysts. The ammonia chemisorption values of W03-V205/Ti02 and
V205/Ti02 catalysts were 2.31 and 1.93 cm3 STP/g catalyst, respectively. This result
indicates that the acid site density of W03-V205/Ti02 was higher than that of
V20s/Ti02.

In Figure 3, curve A (K20-W03-V20s/Ti02 series) is always above curve B (K2O-
V205/Ti02 series), and curve C (Cs20-W03-V20s/Ti02 series) is always above curve D
(Cs20-V205/Ti02 series). This result, again, indicates that the strength of the poison
coincides with its basicity.

The higher acid site density on the W03-V20s/Ti02 catalyst was caused by the
addition of WO3. This was supported by results from Proton Magic Angle Spin
Nuclear Magnetic Resonance (1H MAS-NMR) experiments (5). Bronsted acidity is
caused by the donation of proton from the surface hydroxyl group. The proton
nuclear magnetic resonance shift is a direct measure of the Bronsted acidity, and
such shifts are measured relative to a standard, [commonly used is tetramethyl
silane (TMS)], in terms of ppm. A positive value means a shift of the resonance
toward a lower magnetic field, corresponding to a smaller shielding by the electron
shell. This, in turn, means a weaker bond between the proton and the oxygen atom,
hence a stronger Bronsted acidity. The "ideal" Bronstead acid, i.e., proton without
electron shell, gives a shift of 30.994 ppm relative to TMS. To understand the nature
of the sites (Bronsted or Lewis) which were created by the addition of WO3,1M MAS
NMR experiments were performed. The results showed that the addition of W03 to
the V2O5/T1O2 catalyst increased the Bronsted acidity. The chemical shift increased
from 3.56 ppm for 5% V2O5/T1O2 to 4.43 ppm for 8.2% WO3 + 4.8% V20s/Ti02.
However, the doping of K2O in either 5%V20s or 8.2% WO3 + 4.8% V20s/Ti02
catalyst resulted in a reduction of Bronsted acidity. The chemical shift decreased
from 4.43 ppm for 8.25 VVO3 + 4.8% V2O5/T1O2 to 3.14 ppm for 8.2% WO3 + 4.8%
V2O5 = 0.6% K20/Ti02-

5A-43

1


-------
For a further understanding of the mechanism of the poisoning effect, the extended
Huckei Molecular Orbital (EHMO) calculations were performed to examine the
nature of the surface hydroxyl groups. The results were expressed in terms of the
extraction energy (EH+) of the hydrogen atom and the net charge of the hydrogen
atom (H+). The calculation results showed that the hydrogen of terminal group was
easier to be abstracted (i.e., stronger proton donicity hence stronger Bronsted acidity)
than that of the bridge hydroxyl group. The doping of alkali oxides lead to a decrease
in Bronsted acidity on the catalyst surface, whereas the addition of SO2 on the
surface lead to an increase in Bronsted acidity.

The above results, summed together indicate that Bronsted acid sites are the active
centers for the SCR reaction on the V20s-based catalysts. Therefore, we may
conclude that a V20s-based catalyst with a higher Bronsted acid site density results in
a higher activity for the SCR reaction. The poisons reduce the Bronsted acidity
hence the SCR activity.

POISONING CONSIDERATIONS IN REAL FUELS

Catalysts operating in authentic flue gas will experience different surface conditions
(defined by the number of trace compounds on the surface and their distribution)
than observed of this experiment. Specifically, the role of sulfur - to be a poison or
promoter - is unclear. If sulfur combines with strong alkali - and thus acts as a
means to add net basicity to the surface - catalyst activity will decrease.

Alternatively, if sulfur combines with alkali in a manner to increase the net acidity
of the surface - catalyst activity could increase.

A possibly more important role of sulfur could be to combine with various trace
elements (including alkali) and deposit on the catalyst surface, thereby restricting
access of the site to reactants and decreasing catalyst activity. The specific surface
conditions - defined by the specific types of compounds and their concentration -
will play an important role in the ultimate effect of trace elements on catalyst
activity. Further investigations into such surface conditions are being considered to
resolve the role of sulfur on catalyst activity,

CONCLUSIONS

(1) The inclusion of WO3 in the proportion of 8.2% in the catalyst composition
increases the rate constant for the SCR reaction.

5A-44


-------
(2) The inclusion of WO3 improves the resistance of the SCR catalyst to poisoning.
However, strong poisons such as alkali compounds still have a pronounced
effect on catalyst activity, with poisoning strength in proportion to basicity.
Lead, arsenic, and phosphorous are also poisons, but exhibit less poisoning
strength compared to strong alkali.

(3 The addition of gaseous SO2 decreases the activity of tunsten-bearing
V205/Ti02 catalysts without poisons deposited on the catalyst surface,

However, SO2 increases the activity of the catalysts doped with alkali oxides.

(4)	Chlorides can act to either promote or poison the catalyst, depending on the
form of compound deposited. If vanadium chlorides ultimately form, catalyst
activity will decrease significantly.

(5)	Ammonia chemisorption analysis of K20-doped catalyst samples suggest that
W03-containing catalyst offer higher acid site density.

(6)	The actual role of sulfur on catalyst activity will depend on the nature and
concentration of sulfur-bearing compounds deposited on the surface. This
study identified that sulfur could either decrease or increase catalyst activity. If
sulfur acts as a means to add basicity (or acidity) to the surface, catalyst activity
will decrease (or increase).

(7)	These results, supported by NMR analysis (Proton Magic Angle Spin) and
calculations (Huckel Molecular Orbital) further support the suggestion that
SCR activity can be interpreted in terms of the density of Bronsted acid sites.
The presence of elements that decrease Bronsted acidity on the surface (such as
alkali compounds) causes a corresponding decrease in activity.

REFERENCES

1.	J. P. Chen, M.A. Buzanowski, R. T. Yang and J. E. Cichanowicz. Air Waste
Manage. Assoc., 40,1403 (1990).

2.	H. B. Flora, J. Barkley, G. Janik, B. Marker, and J, E. Cichanowicz. Proceeding of
the 1991 Joint Symposium on Stationary Combustion NOx Control, March 1991.

3.	G. Tuenster, W.F.V. Leeuwen and L. J. M. Sheprangers. Ind, Eng. Chem. Res., 25,
633 (1986).

4.	J. P. Chen, R. T.Yang, M.A. Buzanowski and J. E. Cichanowicz. Ind. Eng. Chem.
Res., 29,1431 (1990).

5.	B. M. Reddy, K. Narsimha, D. K. Rao and V. M. Mastikhin. J. Catal., 118, 22
(1989).

5A-45


-------
M/V Atomic Ratio

Figure 1. SCR activity (expressed as first-order rate constant) of 51
V205/Ti02 (solid curves) and 8.2% WO3 + 4,8% V2O5/T1O2 (dashed curves)
doped with different amounts of oxide poisons where M = Li, Sa, K Rb,
Cs, Pb, As and F, 300'C; NO - N% - 1,000 ppm, 02 = 2%, N2 - balance,
GHSV - 15,000 hr."1

5A-46


-------
M/V Atomic Ratio

Figure 2. SCR activities (expressed as first-order rate'constant)
of 8.2% WO^ + 4,81 VjO^/TiOj with doped metal oxide poisons1.
M = metal, 300°C, C>2 = 2%, NO' = NH3 = 1,000 ppm, S02 - 1,000 ppm,
H^O = B%, = balance, GHSV = 15,000 hr *. Solid curves are
without -SO^ and H20, and dashed curves are with SC»2 and H^O.

5A-47


-------
M/V Atomic Ratio

Figure 3. NH^ cheuiisorption amount over alkali oxides
doped catalyses at 200°€.

A		,	K20 doped 8,22 WC>3 + 4.8% 'V^/TiC^

B			K20 doped 5% V205/Ti02

C			Cs20 doped 8.2% Wt>3 + 4.8% V,0'5/Ti02

D	—-	Cs20 doped 5% V^/TiOj

M			K or Cs

5 A-48


-------
Table 1

Effects of SO2 on NO Conversion and Rate Constant (k) for SCR at 300°C,

Catalyst

Without SC>2	With SC^

3 '	3

Conv,, % k, cm /g/s Conv., % k, cm /g/s

5%. V„Oc/TiO_ (A)	98.0	10.38	99.2	12.82

I j 1

0.741 CaO/A	97.2	9.49	99.2	12.82

0.32% Li„Q/A	91.4	6.52	99.1	12.63

2

0.68% As-0»/A	96.7	9.09	99.2	12.80

L 3

8.2% W0.+4.8% V.O./TiO. (B)	99.4	13.58	99*	12.23
3 2 5 2

1.5% Na20/B ¦	83	4.7	95.5*	8.23

1.1% K20/B	54	2.06	98*	10.39

5.1% As203/B	92	6.71	95.3*'	8.12

Reaction Conditions; TO = NH^ = 1,000 ppn, SO, = 1,000 ppm (when used):

0, = 2%, H,0 = 8% (when used), SL = balance,
i.	2.	i.

C-HSV = 15,000 ppm.

~¦reaction with 8% water vapor.

5A-49


-------

-------
EVALUATION OF SCR AIR HEATER
FOR NOx CONTROL ON A FULL-SCALE
GAS- AND OIL-FIRED BOILER

J, L. Reese and M. N, Mansour
Applied Utility Systems, Inc.

1140 East Chesnut Avenue
Santa Ana, California 92701

H. Mueller-Odenwald,

Kraftanlagen AG, Heidelberg
Im Brietspiel 7
Postfach 103420
D-6900 Heidelberg 1, Germany

L. W, Johnson, L. J. Radak, and D. A. Rundstrom
Southern California Edison Company
2244 Walnut Grove Avenue

Post Office Box 800
Rosemead. California 91770

5 A-51


-------

-------
ABSTRACT

A selective catalytic reduction air heater (CAT-AH) system is being demonstrated by
Southern California Edison Company (SCE) on Mandalay Generating Station Unit 2, a gas-
and oil-fired unit. The CAT-AH is installed on one of two air heaters and treats flue gas
from an equivalent of 107.5 MW of electrical generation.

The CAT-AH process involves the reaction of NO, in the flue gas with NH3 in the presence
of catalyst-coated air heater elements to form N2 and H20. The elements are designed to
provide optimum NOx reductions while maintaining air heater performance. This technology
was developed and is supplied by Kraftanlagen AG Heidelberg (KAH).

Projected results of the demonstration are presented, showing the design of the CAT-AH
system, NO, reductions, impacts on air heater performance, and process economics.
Projected NO, reductions are presented as a function of NH, to NO, mole ratio and NH3
slip. Air heater performance parameters considered include heat transfer efficiency and
pressure loss characteristics.

5A-53


-------
INTRODUCTION

Rule 1135 currently under consideration by the South Coast Air Quality Management
District (SCAQMD) will require substantial NOK reductions on all utility boilers located in
the South Coast Air Basin. Conventional selective catalytic reduction (SCR), the technology
identified by the SCAQMD to achieve compliance with Rule 1135, can be very expensive
to install on existing boilers. Catalyst coated air heater elements represent an alternative
NO, control technology which can be integrated with other NO, controls, such as low NOx
burners (LNBs), flue gas recirculation (FGR), and selective non-catalytic reduction (SNCR)
to achieve the required NO, reductions at a relatively low cost.

The CAT-AH offers a low cost alternative for installing SCR for NO, control on a utility
boiler. An air heater is a device with a large surface area compacted in a small volume.
The heat transfer surface of an air heater is designed to ensure intimate contact with the
boiler flue gas. Placing a catalyst on the surface of an air heater satisfies most of the design
and operational requirements of a conventional SCR system.

The CAT-AH also is complementary with SNCR processes. NH3 breakthrough, which is
a byproduct of SNCR, can be used to provide further reduction of NO, on the surface of a
CAT-AH. In addition to offering a NO, reduction, a CAT-AH can eliminate or reduce NH3
discharge to the atmosphere.

DEVELOPMENT OF THE CAT-AH

The CAT-AH technology has been developed by KAH in response to strict environmental
regulations in Germany. KAH is a licensee of the Ljungstrom air heater technology and has
been supplying industrial air heaters since the 1920's. KAH began development of a
catalyst-coated air heater for NO, control in 1984. Since then, extensive development work
has been carried out to develop a catalyst and, more importantly, a process to apply the
catalyst to the varied profile geometries used by KAH in their Ljungstrom air heater
designs. This is particularly important because the catalyst-coated element must reduce NOx
emissions while maintaining high heat transfer and low pressure loss characteristics.

Initially, ceramic monolithic catalysts available from catalyst manufacturers were evaluated.
These proved inadequate due to erosion and structural problems associated with thermal
shock. An alternative approach was developed based on KAH's expertise in the manufacture

5A-54


-------
of ceramic coalings. Development of the catalyst materials for NO, removal involved
optimizing the catalyst composition to obtain the desired temperature and porosity to
maximize reactivity while minimizing plugging potential. An additional key factor involved
obtaining good adhesion to the heating element while not compromising the element profile.

The manufacturing process includes the following steps;

•	Form profile;

•	Cut to length;

•	Apply catalyst coating,

A similar process is used in the manufacture of conventional enamel-coated elements. A
key feature of this process is that the profile is formed prior to the application of the
catalyst. This approach allows the elements to be formed into the desired profiles to
optimize air heater performance characteristics.

A large number of laboratory-scale tests have been carried out using the flue gas from
different fuels to evaluate NOx removal as well as correlations of thermal performance and
pressure loss. These tests have been carried out for KAH at the Karlsruhe University, at
the Heat and Mass Transfer Laboratory of Svenska Rotor Maskiner AB in Sweden, as well
as at KAH's laboratories. These tests have shown that satisfactory NO, removal can be
obtained while not sacrificing air heater thermal or pressure loss performance. The
laboratory tests have been followed by full-scale field trials and retrofits.

LARGE-SCALE EXPERIENCE

Table 1 lists KAH's full-scale experience with CAT-AHs, This experience includes
application of the technology to different types of boilers and a full range of fuels. Specific
details of this experience are discussed below.

Frimmersdorf Station

In 1988, KAH CAT-AHs were evaluated on 150 MW units firing brown coal at the
Frimmersdorf Generating Station operated by RWE, The objective of this application was
to evaluate the use of air heater catalysts in supplementing combustion modifications in
achieving the NO, emission limit of 200 mg/m3.

The test period lasted approximately 3,500 hours and included eight start-ups and
shutdowns. During the test period, the actual NO, concentrations and air heater gas inlet
temperatures were lower than expected. During much of the test, the temperature was
below 540°F. The tests did provide the following results:

5A-55


-------
•	With air heater gas inlet temperatures above 570"F, NO, reductions of 30
percent could be attained;

•	Erosion of the catalyst from the abrasive fly ash (high Sj02 content) could
be controlled.

A permanent retrofit of CAT-AH was not installed because RWE was able to meet the NOx
limits only with combustion modifications.

Lichterfeld Station

KAH CAT-AH was evaluated on an oil-fired boiler in 1988 at the Lichterfeld station in
Berlin operated by BEWAG. Air heater catalyst was tested in conjunction with a furnace
urea injection system installed by Fuel Tech, Inc. The primary objective of installing the
CAT-AH was to control NH3 slip. Results of these tests were as follows:

•	NO, emissions were reduced by 75 percent with the combined urea injection
and CAT-AH;

•	NH3 emissions were reduced from 20 ppm to 1 to 2 ppm with the CAT-AH
despite stratification of NH3 at the air heater inlet;

•	Catalyst activity and air heater performance were not impaired by water
washings.

Although favorable test results were obtained with the KAH air heater catalyst, the utility
ultimately installed a conventional SCR system due to a lack of time to meet the NO, limits.

Marl Station

CAT-AH coated with catalysts supplied by BASF have been installed at the slag-tap (wet
bottom) boilers operated by BKG at the Marl Generating Station. Additional elements were
installed upstream of the catalyst elements to reduce gas temperatures to optimum levels.
Results of the installation are:

•	NO„ emissions reductions goal of 30 percent was achieved;

•	NHj slip was limited to 3 ppm.

The NHj injection did result in unacceptably high NH, content in the fly ash, which is
recycled for the production of cement. KAH and the catalyst supplier are currently
assessing potential solutions to the fly ash problem.

5A-56


-------
Voerde Station

A pilot test was conducted in 1987 at the Voerde station operated by STEAG on a gas to gas
heat exchanger used to reheat flue gas from a S02 scrubber. The catalyst matrix was
supplied by KAH's Japanese Ljungstrom partner. The system was operated for 1,600 hours.
During the test, NO„ reductions of 35 percent were achieved and NH3 slip was reduced from
300 ppm to zero.

APPLICATION OF CAT-AH TO MANDALAY GENERATING STATION

SCB recently installed a CAT-AH system on the north air heater at Mandalay Station Unit 2.
This is the first CAT-AH installation in the United States. Catalyst hot end elements were
supplied by KAH under subcontract to Applied Utility Systems, Inc. (AUS).

Intermediate and cold end elements were supplied by ABB Air Preheater, Inc., the United
States licensee of Ljungstrom air heater technology. ABB Air Preheater, Inc. also supplied
new conventional elements for the south air heater at Mandalay Unit 2. An NH, injection
system using anhydrous NH3 was provided by SCE's Engineering and Construction
Department. Design criteria for the NH3 injection grid was provided by KAH and AUS.
Structural design of the injection grid was provided by Charles T. Main, Inc.

Installation of the CAT-AH was completed on December 1, 1990. The unit was returned
to service and has been operating normally since then. Evaluation of the performance of
the CAT-AH has been delayed pending the completion of the installation of the NH3
injection system. This installation is now virtually complete and NH3 injection is expected
to be initiated during the week of April 1, 1991. Evaluation of the CAT-AH will be under
the direction of SCE's Engineering, Planning and Research Department.

EQUIPMENT DESCRIPTION

The Mandalay Generating Station is located in Oxnard, California. The station consists of
two identical steam generating units, Units 1 and 2. Each unit is equipped with two air
heaters. The CAT-AH is being evaluated on the north side air heater of Unit 2. The CAT-
AH and the NH} injection system are described briefly below.

5A-57


-------
Mandalav Unit 2

Figure 1 shows a diagram of Mandalay Unit 2. This is a Babcock and Wilcox unit that was
installed in 1959. The electrical generation is 215 MW at full load and 20 MW at minimum
load. Superheated steam conditions at full load are 2400 psig and 1050°F. The boiler is
equipped with primary and secondary superheater sections, a reheat superheater,
economizer, and two air heaters.

The boiler operates in a forced draft mode and is equipped with 24 combination gas- or oil-
fired burners located on the front wall with four rows of six each. Selected burners in the
upper rows are operated out of service for NO, control. The unit is equipped with PGR to
the hopper of the furnace for steam temperature control at reduced loads.

Air Heaters

Mandalay Unit 2 is equipped with two vertical shaft Ljungstrom regenerative air heaters.
The air heaters are Model 1588, Type 26 VIX, originally supplied by ABB Air Preheater,
Inc. The air heater has three layers of elements, consisting of hot end , intermediate, and
cold end layers. The heights of the original elements are as follows:

•	Hot end layer: 42 inches;

•	Intermediate layer: 16 inches;

•	Cold end layer: 12 inches.

The hot end and intermediate layers are constructed of open hearth steel. The cold end
layer is 409 stainless steel. Table 2 shows the design operating conditions for the original
air heaters. Flue gas outlet temperatures are shown both uncorrected and corrected for air
leakage.

The CAT-AH has been installed on the north side air heater. The catalyst has been applied
to the hot end elements. The height of the hot end has been increased to maximize NO,
reductions, The new intermediate layer is constructed from Corten steel. The height has
been reduced to compensate for the increased hot end height. The cold end elements have
been replaced with enameled steel elements with the same dimensions as the original
elements. Enameled elements have been used due to the potential for increased corrosion
caused by the increased conversion of S02 to S03 when firing oil fuel.

The height of the new elements are as follows:

•	Catalyst hot end layer: 1150 mm (45.25 inches);

•	Intermediate layer; 355 mm (14.0 inches);

5A-58


-------
• Cold end layer: 305 mm (12.0 inches).

The south side air heater has been replaced with new elements conforming to the original
design.

NH, Injection System ,

The NHj required for the NO„ reaction is provided by an air injection system using
anhydrous NH3. The NH3 will be mixed with carrier air and injected into the flue gas using
a grid of nozzles in the horizontal duct located downstream of the economizer. Table 3
shows the nominal operating conditions of the NH, injection system. The NHj injection rate
for each test will be selected based on the inlet NOx concentration, flue gas flow rate, and
desired NH3/NOx mole ratio.

The NH3 is supplied to seven horizontal header pipes located in the duct. Nozzles from
adjacent header pipes are offset to provide uniform dispersion of NH3. There are a total of
165 injection nozzles.

The NH3 supply line to each header pipe is equipped with a manual valve and pressure
indicator to allow control of NH3 to each header pipe. In this way, the NH3 injection can
be biased to compensate for stratification of the flue gas flow in the vertical direction;

PROJECTED PERFORMANCE OF THE CAT-AH

Projections of the performance of the CAT-AH in terms of heat transfer, pressure loss, and
NO, removal were provided by KAH using correlations they have developed based on full-
scale and. laboratory-scale experience.

The projected thermal and pressure loss performance are as follows:

•	Thermal performance: no change;

•	Pressure loss: increased by less than 20 percent.

Air heater outlet gas temperature, and thus boiler efficiency, are not expected to be changed
by the CAT-AH. The design total pressure loss for the gas and air side of the original air
heater is 7.15 inches H20 at full load. Thus, the increase in pressure loss across the CAT-
AH is expected to be less than 1.4 inches H20. This is not expected to have any significant
adverse impacts on boiler operation. Preliminary indications at the Mandalay station are
that the performance of the CAT-AH with respect to gas and air side temperatures and
pressures are in line with projections.

5A-59


-------
Figures 2 through 4 show the projected NO, reductions as a function of NH-, slip. NO,
reductions are shown for the baseline case of 180 ppm (@3% 02) NO,. For comparison,
NO, reductions also are shown with reduced baseline NO, levels of 120 ppm (@3% 02)
and 50 ppm (@3% 02) to reflect the emissions that could be attained with additional control
measures such as LNBs, FGR, and SNCR,

At the design condition of 740°F gas inlet temperature and a baseline NO, of ISO ppm, a
NO, reduction of 50 percent is projected to maintain NH} slip below the specified 10 ppm.
Higher NOx reductions can be obtained, but with a corresponding increase in NH3 slip.

As shown in Figures 2 and 4, higher NO, reductions can be obtained with lower baseline
NO, values while maintaining NH3 slip less than 10 ppm. With an inlet NO, level of 120
ppm (@ 3% 02), NOx removal is projected to be 60 percent. With an inlet NOx of 50 ppm
(@3% 07), NO, removal increases to 76 percent.

NO, reductions are increased as the baseline NO, is decreased because a smaller amount of
NHj injection is required, The ratio of NH3 injected to NH3 slip remains relatively
constant. Thus, to maintain a specified level of NHa slip (10 ppm), higher mole ratios of
NH3 to NO, can be injected at lower inlet NO, levels, resulting in a corresponding increase
in NO, removal.

The figures also show the effect of flue gas inlet temperature on NO, removal. With an
inlet NO, level of 180 ppm (@3% 02) decreasing the inlet gas temperature from 740°F to
650°F is expected to reduce NO, removal from about 50 to 43 percent. Increasing the inlet
temperature to 830°F would increase NO, removal to about 55 percent. The minimum
temperature to obtain NO, removals is approximately 540°F. KAH recommends 900°F as
a maximum gas inlet temperature.

An extensive evaluation program to characterize the performance of the CAT-AH is planned
following the initial start-up of the system. Key process parameters to be evaluated will
include:

•	Air heater gas inlet temperature;

•	Space velocity;

•	Inlet NO, concentration;

•	Allowable NH3 slip;

•	NHj distribution.

5A-60


-------
Boiler and NH3 injection system operating conditions to be varied to evaluate the above
parameters include:

•

Boiler load;

•

NHj/NOx mole ratio;

•

NH3 injection distribution;

•

FOR rate to hopper;

•

Excess 02.

Boiler load affects several operating variables simultaneously, including inlet flue gas
temperature, inlet NO, concentration, and flue gas volume (space velocity). The NH3/NO,
mole ratio controls the availability of NH3, but is limited by the allowable NH3 slip.
Typically, an NH3/NO, mole ratio of less than one is used to avoid excessive NH3 slip.
Because the available NH3 is limited, adequate NH3 distribution across the air heater is
important to obtain maximum NO, removal. Proper NH3 distribution will be verified by
traverse measurements of NHS and NO, stratification at the air heater outlet. The
distribution of ammonia can be controlled by the manual control valves installed on each
horizontal supply pipe or, if necessary, by using injection nozzles of different sizes. FGR
and excess O, are variables that affect flue gas temperature, flue gas volume, and inlet NO,
concentration. The impact of these variables on air heater performance also will be
determined.

INTEGRATED APPROACH TO NO, EMISSIONS CONTROL

The CAT-AH is well suited for use with other NO, control techniques because of the
following:

•	CAT-AH can be used to control NH3 slip from upstream SNCR or SCR
processes;

•	For a constant NH3 slip, the NO, removal rate increases with decreasing air
heater inlet NO, concentration.

Figure 5 shows an example of how a CAT-AH can be combined in a system of multiple NO,
control technologies to achieve ultra low NO, emissions levels. Based on an uncontrolled
NO, emission level of 200 ppm (@3% 02) for a gas-fired utility boiler, it is conservatively
estimated that NO, emissions can be reduced by 60 percent to 80 ppm (@3% 02) using
combustion modifications such as LNBs, FGR and/or staged combustion. The use of a
SNCR process could provide a NO, reduction of at least 25 percent, to 60 ppm (@3% 02).
A CAT-AH could then be applied to provide further NOx reductions while controlling NH3

5A-61


-------
slip from the SNCR process. Based on the projections for the Mandalay unit, the CAT-AH
could provide a NO, reduction of at least 67 percent, resulting in a NO, emission level of
20 ppm (@ 3% 02), while limiting NH3 slip to 10 ppm.

Thus, the combined use of selected NO, emission control technologies can provide overall
NO, reductions comparable to conventional SCR, in the range of 90 percent. This approach
can be much less costly than a conventional SCR system in a retrofit application.

COSTS

Costs of the CAT-AH control technology are dependent on a number of site-specific factors
including the unit size, fuel characteristics, number of air heaters, air heater design and
operating conditions, and catalyst life. Rough approximations of capital costs are in the
range of $20/kW. Operating and maintenance costs are controlled by catalyst life, which
is yet to be determined. Approximate costs range from $1.25/MWh to $3.0O/MWh for a
catalyst life ranging from two years to five years.

CONCLUSIONS

Testing of the CAT-AH system at the Mandalav station to be conducted in the upcoming
months will provide a detailed assessment of CAT-AH performance. At the present time,
the following qualitative conclusions can be reached:

•	CAT-AH provides an additional technique to reduce NO, emissions;

•	Requires no modifications to existing equipment and has minimal impact on
performance;

•	Provides substantial NO, reductions;

•	Can be used to control NH3 slip from SNCR or upstream catalyst processes;

•	Can be integrated with other NO, control technologies to provide ultra low
NO, emissions.

More quantitative conclusions will be developed following the evaluation at the Mandalay

station.

5A-62


-------
Figure 1. Mandalay Unit 2

5A-63


-------
N0X removal efficiency, %

Figure 2. NO x Curve for Inlet NO x ¦ 180 ppm

5A-64


-------
NH 3 slip, ppm

10 20 30 40 50 60 70 80

NOx removal efficiency, %

Figure 3. NO x Curve for Inlet NO x ¦ 120 ppm

5A-65


-------
NH 3 slip, ppm

N0X removal efficiency, %

Figure 4. NOxCurve for Inlet NO x - 50 ppm

5A-66


-------
Ol

>
I

CT)
-vj

250

200

150

100

50

0

NOx> ppm (at 3% 02 )

200

80

60

20

Uncontrolled

Combustion
Modifications

SNCR

Catalyst
Air Heater

Figure 5. Integrated Approach for Ultra-Low NOx Emissions


-------
TABLE 1. KAH FULL-SCALE EXPERIENCE



Utility

Power Station

Fuel

Pilot Trials:





RWE

Frimmersdorf "C"

Brown Coal

RWE

Meppen

Gas

STEAG

Voerde

Bituminous Coal

Full-Scale Retrofits:





RWE

Frimmersdorf "F"

Brown Coal

BEWAG

Lichterfeld (Berlin)

Oil

BKG	Marl	Bituminous Coal

5A-68


-------
TABLE 2. ORIGINAL AIR HEATER DESIGN OPERATING CONDITIONS



Location

Flow,
Mlb/hr

Temperature,

op

Pressure,
In. H20

Air Inlet

860.5

80

19.0

Air Outlet

790.0

646

15.1

Gas Inlet

850.0

740

4.25

Gas Outlet

920.5

267
(uncorrected)

259
(corrected)

1.00

I

5A-69


-------
TABLE 3. NHj INJECTION DESIGN CONDITIONS

Location:

Flue Gas Flowrate:

Flue Gas 02 Content:
NOx Concentration:
NO, Flowrate:

NH3/NO, Mole Ratio:
NH3 Flowrate:

NH3 Carrier Air Flowrate*.

Mandalay Unit 2 North Air Heater
850,000 lb/hr
2.0% 02

180 ppm @ 3% 02

212 lb/hr

0,92

72 lb/hr

4,000 lb/hr

5A-70


-------
n2o formation in selective non-catalytic

NO, REDUCTION PROCESSES'

L. J. Muzio"
T. A. Montgomery
G. C. Quartucy
Fossil Energy Research Corporation
23342 C South Pointe
Lag una Hills, CA 92653

J. A. Cole
J. C. Kramlich

Energy and Environmental Research Corporation
18 Mason
Irvine, California 92718

' Work sponsored by U.S. DOE AR&TD (DE-AC22-88PC88943)
" Corresponding Author

5A-71


-------

-------
ABSTRACT

NO, control techniques currently under development include combustion modification and post-
combustion techniques. As these technologies are developed and implemented, it is Important to
ensure that NO, reductions are not achieved at the expense of producing other undesirable species.
One possible concern is the production of N?0 from NO, reduction processes. The current work
addressed potential N20 production from selective non-catalytic NO, reduction (SNCR) processes using
ammonia, urea and cyanuric acid injection. Previous work with SNCR processes has shown-that
ammonia injection produces minimal NsO emissions, while cyanuric acid injection has, under certain
conditions,.almost quantitatively converted NO, to N20. While little data exists for urea injection, It has
been suggested that it might behave as a hybrid between ammonia and cyanuric acid. Pilot-scale
testing and chemical kinetic modeling was used to characterize the N20 production of these processes
over a range of process parameters. The data show that SNCR processes were all found to produce
some N20 as a byproduct. Ammonia injection produced the lowest levels of Na0 while cyanuric acid
produced the highest levels. N20 formation resulting from these processes was shown to be dependent
upon the reagent used, the amount of reagent injected, and the injection temperature.

5A-73


-------
INTRODUCTION

The mean global concentration of N20 is approximately 300 ppm and has been increasing at a rate of
0.2-0.4% per year {Tirpak, 1987, Weiss, 1981). In the troposphere, N20 is a relatively strong absorber
of infrared radiation and, therefore, has been implicated as a contributor to the "Greenhouse Effect".
Being stable in the troposphere, NaO is transported to the stratosphere where it is the largest source
of stratospheric NO. NO in turn is the primary species responsible for establishing the equilibrium
stratospheric 03 concentration (Kramlich, et al, 1988).

Direct N20 emissions from fossil fuel combustion have previously been reported to be equivalent to 25-
40% of the NO, levels (Hao, et al, 1987; Castaldini, 1983). However, recent tests have shown these
measurements to be in error, most of the NsO having been formed as an artifact of the sampling
procedure (Muzio and Kramlich, 1988). Full-scale tests using an on-line N20 analyzer have confirmed
that direct emissions of N20 from fossil fuel-fired boilers are less than 15 ppm. Further, N20 levels do
not generally correlate with the NO, emissions (Muzio, et al, 1990).

While N20 emissions from conventional combustion equipment are low, a number of advanced
combustion and emission control systems could be responsible for significant N20 emission levels. This
paper describes experimental and kinetic modeling studies of selective non-catalytic NO, reduction
(SNCR) processes and the potential by-product N20 emissions therefrom.

Selective non-catalytic NO, reduction (SNCR) processes involve the reaction of NO with a nitrogen-
based chemical in a temperature region of nominally 1000K to 1350K. Representative processes in
this category of NO, reduction technologies include:

Ammonia (NH3) Injection, (Lyon, 1976)

Urea (NH2CONH2) Injection, (Muzio and Arand, 1976; Mansour, et al, 1987)

Cyanuric Acid ((HNCQ)3)/Cyanic Acid Injection, (Perry, 1988)

5A-74


-------
Figure 1 shows the possible major chemical paths leading to the reduction of NO, with these species
and possible paths leading to the formation and emission of N20 as a by-product.

Since all of these processes involve reactions between NO and nitrogen species in the temperature
window between 1000-1350K, there is some concern that N20 could be a product of the NO„ reduction
process. Kramlich, et al, (1987,1989) showed that there is a temperature window in the region from
1200 - 1500K for the formation and emission of N20 by the reaction of cyano species and NO,
essentially the right hand path in Figure 1. This involves the formation of NCO which subsequently
reacts with NO to form N20 as follows:

OH + HNCO -» NCO + HjO
NCO + NO -> N20 + CO

Previously reported results with ammonia injection (Lyon, 1976; Muzio and Arand, 1976) Indicate that
very little N20 is formed during the reduction of NO,. This is consistent with the path shown in Figure 1;
the NH3 decomposes to NH, species, which in turn react with NO forming N2 as the primary product.
Reported results with cyanuric acid injection ((HNCO)a) or isocyanic acid injection (HNCO) indicate N20
to be a major intermediate species and product (Siebers and Caton, 1988).

The detailed reaction chemistry of urea (NH.CONH2) with NO, Is not presently known. The actual
reaction path is dependent on the urea decomposition products upon injection into high temperature
combustion products, of which a number can be postulated. It has been suggested that the urea might
decompose into NH3 and HNCO (Caton and Siebers, 1988); this path is shown In Figure 1. If the urea
decomposes to NH3 and HNCO, as suggested by the results of Caton and Siebers (1988), then the
HNCO may ultimately lead to NaO formation. On the other hand, another decomposition path may be
2NHZ + CO, in which case little NsO would be expected as a product.

OBJECTIVES AND APPROACH

The specific objectives of work reported in this paper were to 1) determine the extent to which N20 is
a by-product of SNCR processes, and 2) determine the process parameters and underlying
mechanisms leading to N20 emissions.

The formation and emission of N20 from SNCR processes has been addressed through a combination
of theoretical and experimental efforts. Chemical kinetic calculations were performed using a
mechanism and model developed by Energy and Environmental Research Corp,, (Cole and Kramlich,
1990). Pilot scale tests were conducted in a research combustor at Fossil Energy Research Corp.

5A-75


-------
AMMONIA

UREA

CYANURIC ACID

NH3

NHs +OH —NH2+H2O

NH2+NO	^"N2+H20

NH2CONH2

Nhb+HNCO

(HNCO)3

3H

f

MOO

HNCO+H —~ NH24CO

HNCOKW —NCOfH20
ndo+no-1^ NzO+CO

N2O1M—^N2+Q+M
N20+OH —N2+H02
N2CM-H—~N240H

Figure 1. Major Paths for Selective Non Catalytic NO^ Reduction

5A-76


-------
CHEMICAL KINETIC CALCULATIONS

A series of chemical kinetic calculations have been performed to predict the conditions under which
SNCR processes may result In N20 formation. These calculations were performed using the gas-phase
one-dimensional model and kinetic data set referred to above.

The calculations investigated parameters including temperature, combustion product stoichiometry (SR),
reducing agent type (NH9, urea, cyanuric acid), and S02 concentration. Baseline conditions selected
for the modeling runs were an SR of 1.1 using a CH4/air flame, an initial NO, (NO*,) concentration of
700 ppm and a molar nitrogen to NO, (XN/NO,) ratio of 1.0. There was no S02 present during the
baseline runs.

The combustion products were produced by running the model as an adiabatlc well stirred reactor
followed by a plug-flow reactor. This approach has been previously shown to successfully simulate
effluents from premixed and diffusion burners. The gases were then "quenched" to the desired starting
temperature and the NO concentration adjusted to provide the baseline NO, level.

For all reducing agents, the injection temperature was varied from 900K to 14QQK at 100K intervals.
A mixing time of 10 ms was used to model the addition of NO, reducing agents.

The decomposition routes of complex reducing agents such as urea and cyanuric add are not currently
well understood. This leaves some uncertainty as to how these materials should be treated during •
modeling. Cyanuric acid (HNCO), was assumed to decompose into either HNCO or HOCN. For urea,
more complex chemicals such as biuret may result from pyrolysis, thus leading to a more complex set
of final decomposition products. Since kinetic data are available for only a few rather simple species,
it is necessary to assume that urea is essentially a combination of simpler species such as:

—» 2NHj. + CO
NHjCONH2	—> NH2 + H + HNCO

-» NH3 + HNCO

Previous calculations (Chen, et al., 1988; Muzio, et al, 1989) have shown that only the latter product
set (NHj + HNCO) resulted in an acceptable prediction of NO, reduction with urea injection.

Figure 2 (a,b) shows the calculated NO, reductions and N20 production, respectively, as a function of '
temperature for ammonia (NH3), cyanuric acid (as HNCO), and urea (as NH3 + HNCO) addition. These
calculations are for the baseline condition described previously at an additive-to-NO, molar ratio (N/NO)

5A-77


-------
900

a)

1000 1100 1200 1300
Temperature, K

Calculated NO, Reduction versus Temperature

H Ammonia (NH3)
® Urea (NH3+HNC0)
Q Cyanuric Add (HNCO)

B Ammonia (NH3)
0 Urea (NH3+HNC0)
~ Cyenuric Acid (HNCO)

900 1000 1100 1200 1300 1400
Temperature, K

b) Calculated N20 Production versus Temperature

Figure 2. Chemical Kinetic Modeling Results, NO,=700 ppm, N/NO - 2.0

5A-78


-------
of 2,0. The calculated NO reductions for NH, Injection are 97% with peak removals occurring at 1200K.
Calculated NO reductions tor urea injection (NH3 and HNCO) show peak removals at 13QQK with the
peak removals somewhat less than ammonia. For cyan uric acid injection (HNCO), peak removals of
90% occur at 1300K. Also, as seen in Figure 2a, the calculated temperature window with HNCO is
narrower than with NHa or urea.

Calculated N20 emissions corresponding to the NO, reductions in Figure 2a are shown in Figure 2b.
As seen in Figure 2b with NH3 injection, very little N20 is calculated as a product, with a peak level of
1 ppm at a temperature of 1200K. This is consistent with experimental results reported by Lyon (1976)
and Muzio and Arand (1976), For the assumed scenario for cyanuric acid decomposition (HNCO), and
for urea (NH3 + HNCO), the calculations show peak N20 levels of 90 and 68 ppm, respectively, at
1200K. For all chemical additives, essentially no N20 is found at temperatures above 1300K.

Figure 3 replots the results in Figure 2a,b showing the calculated NzO levels as a fraction of the NO,
reduced. For ammonia injection, the calculations indicate less than 1% of the NO., is converted to NzO.
For urea injection, the calculations indicate a peak NO, to N20 conversion of 12% at 1200K. For
HNCO, the calculations indicate that over 50% of the NO, is converted to N20 at 1200K.

Calculated byproduct emissions of NH3 and NHCO are presented in Figure 4 (a,b). In both instances,
no byproduct emissions were found at temperatures in excess of 1200K. When considering NH3
emissions (Figure 4a), NH3 injection gave peak emissions of 1397 ppm at 900K, while they were 700
ppm for urea injection at the same temperature. Cyanuric acid injection resulted in maximum NH3
levels of 4 ppm at 120QK. The HNCO emissions, plotted versus temperature in Figure 4b, showed that
cyanuric acid injection resulted in maximum HNCO emissions. Urea injection showed peak HNCO
levels of about one-half of those seen when injecting cyanuric acid, while no HNCO emissions were
seen with ammonia injection. These data show that the byproduct emissions consisted primarily of the
initial reactants, and that at lower temperatures they passed through unreacted.

Additional calculations have been performed Investigating the effect of 1) the presence of S02, 2)
combustion product stoichiometry, 3) initial NO, level, and 4) amount of SNCR chemical added. These
results show similar trends and while they have not been included In this paper, they are discussed in
the project report (Montgomery, et at, 1990).

PILOT-SCALE TEST RESULTS

A series of tests were also conducted in a small pilot-scale combustor. A schematic of the combustor
is shown in Figure 5. This combustor is the same one described by Teixeira (Teixeira, et al, 1991).

5A-79


-------
X

o

z

<

b

C\J

Z

<3

IS Ammonia (NH3)
B Urea (NH3+HNC0)
O Cyanuric Acid (HNCO)

627 727 827 927 1027 1127
Temperature, °C

Figure 3. Chemical Kinetic Modeling Results; NsO Emissions
Normalized by the NO, Reduction. NO,=700 ppm, N/NO=2.0.

5A-80


-------
2000

1500

1000

500

H Ammonia (NH3)
0 Urea (NH3+HNCO)
~ Cyanurlc Acid (HNCO)

900 1000 1 100 1200 1300 1400
Temperature, K
a) Calculated NH3 Emissions versus Temperature

2000

1500

1000

500

a Ammonia (NH3)
Q Urea (NH3+HNCO)

~ Cyanurlc Acid (HNCO)

900 1000 1 100 1200 1300 1400
Temperature, K

b) Calculated UNCO Emissions versus Temperature

Figure 4. Chemical Kinetic Modeling Results
NO, = 700 ppm, N/NO = 2.0

5A-81


-------
BUHNER FLOW SYSTEM

OflUTON
AW

Figure 5. Pilot-scale Combustor Facility

5A-82


-------
Gas samples taken at the combustor exit were analyzed for NO/NOx, NaO, CO, CO.i and 02.
Continuous NaO measurements were made using an NDIR technique developed by Montgomery, et al
(1989). NH- measurements were made using wet chemical techniques.

The pilot-scale tests investigated the effect of temperature, chemical injection rate, and initial NO,
concentration on N20 production for selective non-catalytic NO, reduction with NHa, (gaseous), Urea
(both a pulverized solid and an aqueous solution), and Cyanuric acid (pulverized solid).

Figure 6 shows both NO, reduction and N?0 emissions as a function of temperature for NH3 (gaseous),
urea (solid), and cyanuric acid (solid) at an injection rate corresponding to an N/NO molar ratio of 2.0.
For the test results shown in Figure 6, the Initial NO level was 700 ppm. At these conditions, NHS
exhibited the highest NO, reduction with a peak removal of 88 percent at a temperature of about 930°C.
The peak NO, removal with urea was 82 percent at a temperature of 980°C. The calculations
discussed previously yielded peak NO removals for NH3 and urea at nominally 927°C (1200K) and
10276C (1300K) respectively. These differences are most likely due to 1) the finite mixing time in the
combustor, 2) the non-isothermal nature of the combustor, and 3) the finite time for urea evaporation

and decomposition. Cyanuric acid did not exhibit a peak in removal over the range of temperatures
studied; NO, removals increased as the temperature increased (at 1100°C, the NO, removal was 73
percent). Again, this difference in high temperature behavior of cyanuric acid relative to the calculations
is due to the relatively slow decomposition rate of cyanuric acid in the combustor.

Figure 6b shows the corresponding N£0 emissions data as a function of temperature. The data show
that ammonia injection resulted in the lowest N20 emissions, while urea injection provided the highest
in terms of absolute concentration. With ammonia injection, NaO emissions peaked at 877°C, while
injecting either urea or cyanuric acid resulted in peak emissions at 977°C.

The N20 data from Figure 6 have been replotted in Figure 7. In Figure 7, the Ns0 is shown as a
fraction of the NO, reduction (e.g., the fraction of the NO, reduced that is converted to N20). These
results indicate that for NH3 injection, 2-5 percent of the NO, reduced appears as N?0 in the products.
For urea injection, 10-25 percent of the NO, reduced shows up as N20. Cyanuric acid exhibits the
highest conversion to N?0 with up to 40 percent of the reduced NO, appearing as N20 in the products.

The calculated values shown In Figure 3b are in qualitative agreement with the pilot scale results.
Experimentally, NH, exhibits somewhat higher levels of N,0 than the calculations. Likewise, the
conversion of NO to N?0 with urea is higher experimentally than calculated. Finally, the calculations
indicate virtually no N20 at temperatures of 1027°C (1300K) and above, while experimentally the
window for N.O emissions is broader.

5A-83


-------
827 877 927 977 1027 1097
Temperature, °C

¦ Ammonia (g)
0 Urea (s)
~ Cyanuric Acid (5)

a) NO, Reducfion versus Temperature

¦ Ammonia (g)
E3 Urea(s)
Q Cyanuric Add is)

827 877 927 977 1027 1097
Temperature, °C
b) NjO Production versus Temperature

Figure 6. Pilot-scale Test Results, NO, Reduction and N?0 Production
NO, = 700 ppm, N/NO = 2.0

5A-84


-------
0.50

0.40

0.30

0.20

0.10

0.00

B Ammonia (g)
0 Urea (s)
~ Cyanuric Acid (s)

827 877 927 977 1027 1097
Temperature (C)

Figure 7. Pilot-scale Test Results, Conversion of NO, to NzO
NO,, = 700 ppm, N/NO = 2.0

5A-85


-------
The laboratory data show somewhat higher N20 levels relative to the calculated data at lower
temperatures. This may be a result of the CO present in the combustor at lower temperatures. ThisCO
is an artifact of the way that the laboratory combustor is operated; temperatures are varied by adjusting
the combustor gas fuel flow rate. Low temperature conditions are obtained by operating the combustor
at very lean conditions that also produce CO, These CO levels of nominally 100 ppm have been shown
(Teixeira, et a), 1991} to result in increased N20 emissions with SNCR processes.

Figure 8 shows NH3 emissions as a function of temperature, measured during the pilot-scale teste. The
data show that regardless of the reagent injected, NH3 slip decreased as temperature increased. NH3
injection resulted in the lowest measured NH3 slip. Cyan uric acid injection was found to give relatively
high NH3 emissions. This is in contrast to the modeling data, which predicted that NH3 slip would be
minimal. The data suggest that the HNCO is converted to NH3 before it exits the combustor unreacted.
To determine if HNCO slip was being measured, three samples were prepared by dissolving cyanuric
acid in an aqueous solution. The resulting solutions were analyzed using the same specific Ion
technique used to detect NH3. Test results showed that dissolved (NHCO)s was not measured as NHa,
thus indicating that NH3 measurements reflected only NH3 emissions.

Pilot scale results at a lower initial NO, level of 300 ppm are Shown in Figures 9 and 10. Figures 9a
and 9b show NO, reduction and NsO emission, respectively, as a function of injection temperature. The
NO, reduction data show that NH3 injection provided peak removals of 88 percent at about 980°C.
Similarly, urea injection resulted in a maximum NO, reduction of 57 percent at 930cC. As with the
higher initial NO, level, cyanuric acid injection did not exhibit any peak NO, removal over the range of
temperatures investigated. The maximum reduction of 52 percent was measured at 1100°C. With the
exception of NH, injection, maximum removals were lower for the tests,performed with 300 ppm initial

NO than those performed with 700 ppm initial NO.

NjO emissions (Figure 9b) show that at an initial NO level of 300 ppm, cyanuric acid injection yielded
the highest NzO emissions; 69 ppm at about 980°C. N20 emissions resulting from urea injection also
peaked at 980°C, at 43 ppm. When injecting the NH3, peak N.O emissions of 21 ppm were measured
at 880°C. Figure 10 shows the ratio of N20 emission to NOx reduction as a function of temperature for
each of the three SNCR chemicals tested at this lower initial NO level. Test results showed maximum
values at temperatures similar to those seen at higher initial NO levels (see Figure 7). The ratio peaked
at about 880°C for NH3 and 980°C for urea and cyanuric acid. Again, cyanuric acid was shown to
provide the highest conversion of NO, to NaO. For NH3, the peak value was about 9 percent, while a
peak value of 42 percent was observed with cyanuric acid injection. Urea exhibited a maximum NOx
to NjO conversion of 25 percent at this lower initial NO, level.

5A-86


-------
Temperature, °C

Figure 8. Pilot-scale Test Results, NH3 Emissions
NO, = 700 ppm, N/NO = 2.0

H Ammonia (g)
E3 Uraa (s)
D Cyanuric Acid (s)

5A-87


-------
c
o

u

3
T>
05

cc

x

O
2

877	927	977

Temperature, °C
a) NO, Reduction versus Temperature

1097

¦ Ammonia 
-------
X

o _
z e

< a

O %

CM

Q-

0.50

0.40

0.30

0.20

0.10

0.00

I

¦ Ammonia (g)
E2 Urea (s)
~ Cysnuric Acid (sj

877

927

977

1097

Temperature (C)

Figure 10. Pilot-scaie Test Results, Conversion of NO, to N20
NO, = 300 ppm, WHO = 2.0

5A-89


-------
The effect of the chemical injection rate on the NO, reduction and N20 emissions are shown in Figures
11 and 12 respectively. As expected, Figure 11 shows increased NO, removals with increasing
chemicaladdition rate (N/NO ratio) for all three chemicals and at all temperatures tasted. As seen in
Figure 12, the amount of chemical injected, N/NO ratio, has little impact on the conversion of. NO, to
N20. At the lower temperature conditions, less than 927°C, there appears to be some increase In N20
production and conversion of NO, to N.O as the N/NO ratio increases, The decrease in NO, to N20
conversion with increasing N/NO ratio for cyanuric acid injection at 827°C Is due to the increase in NO,
reduction as the N/NOx ratio increases, rather than a decrease in N20 emission levels.

DISCUSSION

The results of this study showed that NzO can be a product ot selective non catalytic NO, reduction

processes. The question is whether implementation of SNCR processes will have a significant impact

on the global N20 budget. The annual atmospheric production of NaO, calculated as the sum of the

rate of destruction of NaO and the rate of increase in the atmosphere, is estimated to be 13-14

megatons (metric) of N20 (as N) (Levine 1991). The potential contribution of SNCR processes can be

estimated using the results of this study and the amount of fuel burned in industry. For instance,

consider the U.S. utility industry which has an annual fuel consumption of about 20 x 10'5 Btu (natural

gas, oil and coal). An order of magnitude estimate of the annual NsO from SNCR processes can be

made with the following assumptions (Eskinazi, 1991):

Average utility NO, emissions are 0.7 lb NOS/106 Btu

SNCR processes result in 50% NO, reduction

¦ ¦ The various SNCR chemicals convert a fraction of the NO, to N?0:

NH3 - 3%, Urea - 15%, cyanuric acid - 30%.

The results of these calculations are plotted parametrically in Figure 13 as a function of the percent of
the NO, converted to N?0 and the Taction of the utility fuel burned that use SNCR processes. As seen
in Figure 13, even if all of the fuel burned in the utility industry used SNCR technology, annual Nz0
production would be 0.06 megatons N for NH3; 0.3 megatons N for urea; and 0.6 megatons N for
cyanuric acid. While the use of SNCR technology may become wide spread, it is not likely that all of
the fuel burned would utilize SNCR. In this context, the calculated 0.06 - 0.6 megatons of NsO (as N)
is a conservative estimate of the contribution.

CONCLUSIONS

Both the chemical kinetic calculations and the pilot scale test results show that N20 can be a product
of some of the SNCR processes, NHa injection yielded the lowest N.,0 levels; typically less than 4%
of the NO, reduced. With cyanuric acid injection, conversion of NO, to N20 ranged from 12 to 40%.
The NO to NsO conversion with urea injection ranged from 7 - 25%. The conversion of NO to N20 did

5A-90


-------
c
.9
o

3
"O

a>
£E

X

o

C

o

,

o
~
X!
05

DC
x

O
Z

c
q
o

3
T3
ffi

CC

X

O

827 877 927 977 1027 1097
Temperature, °C

a) NO, Reduction vs. Temperature, Ammonia Injection

827

877

927

977

1027 1097

Temperature, °C

b) NO, Reduction vs. Temperature, Cyanurlc Acid Injection

027 877 927 977 1027
Temperature, °C

1 097

c) NO, Reduction vs. Temperature, Urea Injection

Figure 11. Pilot-scale Test Results, NO, Reduction as a Function of N/NO Ratio

NO, = 700 ppm

5A-91


-------
0.5
0.4
0.3

N/M2
H 2.0

m i.o

~ 0.5

0.2

0.1

0.0

t

,fc-i,Pfcn,

immi i ii i

827 B 7 7 927 977 1027 1 09 7
Temperature, °C

a) Conversion of NO, to N20, Ammonia Injection

N/NQ

b)

827 877 927 977 1027 1097

Temperature, °C
Conversion of NO, to N2Op Cyanurlc Acid Injection

N/NO
H 2.0

~	1.0

~	0.5

827 877 927 977 1027 1097

Temperature, °C
c) Conversion of NO, to N,0, Urea Injection

Figure 12. Pilot-scale Test Results, Conversion of NO, to NzO as a Function of
Temperature. NO, = 700 ppm, Varying N/NO Ratios

5A-92


-------
1.0

0.8

vy
c
©
ifi

~ 2

UJ in
O o

CM ¦£

z 55

0.6

ra

3
C

c
<

o>
-"

		10 %m~"

¦ /I;;..--;;.!.

t

1

I

1
1

1

I
1
1

1
1

1



iVrfr't	T"	

1



7.0

6.0

5.0

-¦4.0

3.0

2.0

1.0

1 0

20

30

40

50

Percent Conversion of NOx lo N20

Figure 13. Potential Annual N20 Emissions SNCR Processes
(Bars show annual NaO production from SNCR
processes if all of the utility fuel burned used SNCR.)

5A-93


-------
not appear to be a strong function of the amount of chemical injected (N/NO, ratio) or the initial level
of NO, (over the range tested, 300 -700 ppm).

The experimental results are consistent with the chemical kinetic calculations suggesting that NzO
production with SNCR processes occurs primarily due to the formation of NCO which subsequently
reacts with NO to form N20.

5A-94


-------
REFERENCES

Castaldlni, C„ et al., Environmental Assessment of industrial Process Combustion Equipment Modified
for Low NO„ Operation, Proceedings of the 1982 Joint Symposium on Stationary Combustion NO.
Control. V. II, 46,1-46,24, EPA-600/9-85-0266, U.S. Environmental Protection Agency, 1983.

Caton, J. A. and Siebers, D. L, "Comparison of Nitric Oxide Removal by Cyanuric Acid and by
Ammonia," Paper 88-67, presented at the Western States SeetionHTie" Combustion Institute Fall
Meeting, Dana Point, California, October 1988.

Chen, S.L., Cole, JA.Heap, M.P., Kramlich, J.C., McCarthy, J.M., and Pershing, D.W., Advanced NO,
Reduction Processes Using -NH and -CN Compounds in Conjunction with Staged Air Addition. In
Proceedings: Twenty-second Symposium (International) on Combustion. 1988, The Combustion
Institute. Pittsburg, PA. pp. 1135-1145.

Cole, J.A., Kramlich, J.C., Chemical Kinetic Study of Fuel-Rich Rebuming Chemistry. Combustion and
Flame (1990), (submitted).

Eskinazi, D., EPRI, Personal Communication, 1991. .

Hao, W. M., Wofsy, S.C., McElroy, M. W., Beer, J.M., and Toquan, M. A., Sources of Atmospheric
Nitrous Oxide from Combustion, J. Geophvs. Res., V, 15, 1369, 1987.

Kramlich, J.D., Cole, J.A., McCarthy, J.M., Lanier, W.S., and McSorley, J.A., "Mechanisms of Nitrous
Oxide formation in Coal Flames". Paper 1A-006. Presented at: Fall Meeting, Western States
Section/Japanese SectionfThe Combustion Institute. Honolulu. HI. November, 1987.

Kramlich, J.C., Cole, J.A., McCarthy, J.M., Lanier, W.S., and McSorley, J.A., Mechanisms of Nitrous
Oxide Formation in Coal Flames, Combustion and Flame. (1989) 77 (3,4), pp. 375-384.

Kramlich, J.C., Lyon, R.K., and Lanier, W.S., EPA/NOAA/NASA/USDA N„Q Workshop, Volume I:
Measurement Studies and Combustion Sources, EPA-600/8-88-079, 1988.

Levine, J., The Global Atmospheric Budget of Nitrous Oxide, presented at the 1991 Joint Symposium
on Stationary Combustion NO, Control, Washington, D.C., March 1991, (this symposium).

Lyon, R. K., Longwell, J. P., "Selective Non-Catalytic Reduction of NO, by NH3," Proceedings of the NO,
Control Technology Seminar. EPRI SR39, February 1976.

Mansour, M. N., et al. "Full-Scale Evaluation of Urea Injection for NO, Removal," Proceedings of the
1987 Joint Symposium on Stationary Combustion NO, Control, Vol. 2, EPRI CS5361,1987.

Montgomery, T.A., Muzio, L.J., Samuelson, G.S., "Continuous Infrared Analysis of N20 in Combustion
Products,"JAPCA, V. 39, No. 5, pp. 721-726, 1989.

Montgomery, T.A., et al. "N20 Formation from Advanced N0X Control Processes (Selective Non-
Catalytic Reduction and Coal Rebuming), Report prepared for DOE project DE-AC22-88PC8894,1990.
(Draft)

Muzio, L. J. and Arand, J. K., "Homogeneous Gas Phase Decomposition.of Oxides of Nitrogen, EPRI
FP253, August 1976.

Muzio, L.J., and Kramlich, J.C., An Artifact in the Measurement of N20 from Combustion Sources,
Geophvs. Res. Lett- V. 15, 1369, 1988.

5A-95


-------
Muzio, L.J., Montgomery, T.A., Samuelsen, G.S., Kramlich, J.C., Lyon, R.K., Kokkinos, A., "Formation
and Measurement of N?0 in Combustion Systems" presented at the 23rd' Symposium (International)
on Combustion, Orlean, France, July 1990.

Perry, R. A,, "NO Reduction Using Cyanuric Acid: Pilot-Scale Testing," Paper 88-68, presented at the
Western States Section,H"he Combustion Institute Fall Meeting, Dana Point, California, October 1988.

Siebers, D. L. and Caton, J. A., "Removal of Nitric Oxide from Exhaust Gas with Cyanuric Acid,"
presented at the Fall Meeting of the Western States Section of the Combustion Institute, Dana Point,
CA, 1988.

Teixelra, D.P., et at, "Widening the Urea Temperature Window", presented at the 1991 Joint Symposium
on Stationary Combustion NO, Control, Washington, D.C., 1991.

Tirpak, D. A., The Role of Nitrous Oxide (N20) in Global Climate and Stratospheric Ozone Depletion,
Symposium on Stationary Combustion Nitrogen Oxide Control, V. 1, EPRICS-5361, EPA Contract 68-
02-3994, WA93, 1987.

Weiss, R. F., The Temporal and Spatial Distribution of Tropospheric Nitrous Oxide, J. Geophvs. Res.
Lett.. V. 86 (C

5A-96


-------
TAILORING AMMONIA-BASED SNCR FOR
INSTALLATION ON POWER STATION BOILERS

Robin M.A. Irons
Helen J, Price
Richard T. Squires

PowerGen p.I.e.
Ratcliffe Technology Centre
Ratcliffe-on-Soar
Nottingham
NGll OEE
United Kingdom

5A-97


-------

-------
ABSTRACT

An ammonia-based SNCR installation on a power station boiler must be
capable of giving acceptable NOx reductions over a range of furnace
conditions without excessive ammonia slip.

Experimental characterisation of SNCR has been carried out on two
combustion test facilities - a 0.15 MW linear furnace and a 6 MW
scale model of a power station furnace - with the ultimate aim of
determining suitable conditions for a power station installation.

The smaller facility has been used to characterise variables
affecting SNCR performance and, particularly, to identify the
efficacy of additives in both altering the temperature window of SNCR
and in controlling ammonia slip. It has been demonstrated that a
combination of methane injection (to follow temperature changes at a
given injection point) and lower temperature methanol injection (to
limit ammonia slip) is potentially suitable for power station
installation.

The 6MW facility has been used to develop a practical ammonia
injection system and to determine the NOx reduction achievable on an
installation with finite mixing rates.

Preceding Page Blank i

I	5A-99


-------
BACKGROUND

The operators of utility boilers are currently seeking to develop
cost-effective methods for controlling N0X emissions- One of the
possibilities receiving consideration is Selective Non-Catalytic
Reduction (SNCR) in which a nitrogenous compound is injected into a
flue gas stream and reacts with NOx (primarily NO) to form molecular
nitrogen.

A number of nitrogenous compounds have been proposed as agents for
use in SNCR. These include ammonia (Lyon(1976)), ammonium sulphate
(Chen et al. (1989)) and urea (EPRI(1985)). The urea-based process is
protected by patent and is available under licence from EPRI.

Although the different agents do not have the same effectiveness, "all
have similar traits since each exhibits a relatively narrow
temperature window over which it is useful.

At high temperatures, the nitrogenous compound is itself oxidised to
N0X, while, at low temperatures, reaction is too slow and unreacted
or partially reacted nitrogenous species pass downstream (often in
the form of ammonia). This so-called 'ammonia slip' is potentially a
major problem on power plant since the ammonium salts which form from
reaction between ammonia and SO3 or HC1 cause both fouling and
low-temperature corrosion.

It is thus essential that, on industrial plant, the point at which a
N0X control agent is injected is matched to the optimum temperature
for the de-HOx process so that acceptable N0X reduction is obtained
without significant ammonia slip. It is also important to match the
concentration of agent to that of NOx since both N0X reduction
efficiency and ammonia slip are functions of this ratio.

These aims are complicated by a number of factors:-

1. Variation of gas temperature with boiler load.

5A-100


-------
2.	Variation of gas temperature due to changes in
patterns and extent of slagging and fouling.

3.	Non-uniform cross-duct temperature profiles.

4.	cross-duct distribution of N0X concentration.

In order to engineer a viable power plant implementation of SNCR
technology| xt xs necessary to characterise the behavxour of the
process over a wide range of process conditions. In addition, it is
extremely desirable to be able to alter the temperature window of
operation of the process to allow SNCR to be effective over a range
of furnace operating conditions.

This paper describes work carried out on two combustion rigs (of 0.15
and 6 MW thermal input) to characterise the behaviour of SNCR using
ammonia as the de-NOx agent. It describes tests aimed at illustrating
the effects of temperature, NH3/NOx ratio, NOx inlet concentration
and flue gas oxygen concentration. The use of additives to alter the
temperature window of the process and to control ammonia slip is also
described.

COMBUSTION FACILITIES USED FOR SNCR STUDIES

Two rigs were used to carry out the SNCR studies. Both are located at
PowerGen's Marchwood Engineering Laboratories near Southampton in
southern England.

Coal Ash Deposition Ria (CAPRI
(see figure 1)

This is a horizontally-fired 0.15 MW facility comprising a
refractory-lined combustion chamber contracting to a U-shaped length
of 100mm square section exhaust ducting. It has been described by
Jones (1987).

As its name suggests, the CADR was designed primarily to study coal
fouling mechanisms but, for the current SNCR work, it was used solely
as a source of hot flue gas and was fired with propane. The propane
fuel was doped with ammonia to give independent control of the NOx

5A-101


-------
concentration in the gas entering the test section. In most runs, NOx
concentration was set at around 4 00 ppm, which is typical of the
emission level of a UK 500 MWe wall-fired furnace fitted with first-
generation combustion N0X control. Ammonia (injected with an air
carrier gas to increase its momentum) was introduced via a single
jet. Flue gas analysis (O2, CO, NOx, N2O) was carried out 3.3m
downstream. Ammonia slip was measured a further 1.6m downstream using
a continuous wet chemical ammonia probe developed at PowerGen's
Karchwood Laboratories. Total flue gas flow through the system was up
to 2000 Nm3/s.

Furnace Modelling Facility fFMFl

When used for this work, the facility was configured as a l/5th scale
model of half a 660 MWa oil-fired power station furnace. The
refractory-lined combustion chamber was opposed-fired with 6 residual
fuel oil (RFO) burners on both the front and rear walls. A side
elevation of the FMF is presented in figure 2.

For the duration of the SNCR tests, temperature at the furnace exit
plane was controlled by changing the number and configuration of
burners in service. In contrast to the CADR, it was not, therefore,
possible to obtain independent control of NOx levels and temperature
on the FMF. NOx levels at the furnace exit were typically in the
range 200-300 ppm (3%02).

Ammonia (injected with an air carrier gas) was introduced via two
arrays of five 13mm injectors mounted on each of the side-walls of
the convective section of the rig (see fig.3). Ammonia and air flows
were monitored with turbine flow meters. Maximum flows of 5 kg/h and
400 kg/h of ammonia and air respectively could be maintained. The
flow of the gas mixture through each injector could be monitored
individually and ad}usted by means of control valves. The xnjectors
were mounted downstream of a bank of vertically mounted, ceramically
shielded cooling tubes which had an array of Pt/Ftl3%Rh thermocouples
attached to their downstream side. These thermocouples were used to
determine the 'injection' temperature'of ammonia.

5A-102


-------
CHARACTERISATION OF SNCR PERFORMANCE AT 0.15 MW SCALE.

A series of tests were carried out on the CADR to quantify the
influences governing SNCR performance. These included temperature,
NH3/N0x ratio, oxygen concentration and the effect of various
additives to the gas stream.

Effect of Temperature

The results (see fig.4) demonstrated the characteristic temperature
'window' of SNCR performance. The optimum NOx reduction is obtained
at a temperature of around 1020°C. At higher temperatures, the
ammonia reagent itself oxidises to form additional NOx and reduction
efficiency decreases. At lower temperatures, the ammonia is not
oxidised sufficiently rapidly to the amine radical, which is the
species which actually interacts with NO (Dean et al. 1982), and
unreacted ammonia passes through the SNCR reaction zone.

Effect of Ammonia/NOy Ratio

The onset of ammonia slip is also affected by changes in the NH3:NOx
ratio used in the process. This is illustrated in figure 5, which
shows the variation of N0X and ammonia emissions as the NH3:N0X ratio
is varied at constant temperature (1074°C) and oxygen content. It is
readily apparent that, at ratios not below 1.3, detectable ammonia
slip occurs despite the fact that the temperature used for these
experiments is considerably higher than the optimum observed when a
1:1 NH3:N0X ratio was in use.

Effect of Initial NQv Concentration

A series of runs was carried out (at 1093°C, 1:1 NH3/NO ratio and
2.1% O2) to examine the effect of initial NOx concentration on
conversion. It was found that NOx conversion remained constant at
38% as inlet NOx was decreased from 385 to 250 ppm but that it

5A-1Q3


-------
decreased to 31% at an initial N0X level of 175 ppm.

Effect of Oxvaen Concentration

The overall stoichiometry of ammonia-based SNCR,

4NH3 + 4N0 + O2 = 4N2 + 6H2O

suggests that oxygen concentration might affect N0X reduction
efficiency. This was tested experimentally by varying the oxygen
content of the flue gas in the CADR from 1 to 5 per cent at two
different temperatures. The results obtained are shown in fig.6. It
is clear that, at 1000°C, the reduction is effectively independent of
oxygen concentration but that at 908°C, where the effectiveness is
lower, NOx reduction increases monotonically with O2 concentration.
However, over the range of O2 contents which are likely to be
encountered- on industrial pulverised fuel 'p.f.' boilers (3-4%) the
N0X removal efficiency is not a strong function of O2 concentration.

Effect of Addition of Pulverised Fuel Ash

As mentioned above, the work carried out on the CADR used propane
flames. A limited number of runs was carried out to determine whether
the addition of pulverised fuel ash *p.f-a.9 (collected from earlier
runs of combustion rigs, stored and refired) to the air supply to the
rig would have any significant effect on the process. These runs
proved to be closely similar to those carried out without p.f.a.
addition. In addition, no ammonia was found to be adsorbed on the
recollected ash. It is, however, possible that some heterogeneous
effects might occur in the presence of 'fresh' p.f.a..

USE OF ADDITIVES TO MODIFY SNCR PERFORMANCE

Various compounds have been suggested in the literature (e.g. Lodder
and Lefers (1985)) as additives capable of altering the temperature
window of the SNCR process. All the additives act in a similar way.
Their main function is to increase the concentrations of free

5A-104


-------
radicals in the flue gas stream in order to allow the destruction of
N0X to take place at lower temperatures. The compounds used to
achieve this effect are generally fuels. Hydrogen, carbon monoxide,
light alkanes and alcohols have all been suggested as possibilities.
All these additives exhibit similar behaviour in that they :-

1.	Decrease the optimum temperature of SNCR performance.

2.	Broaden the effective temperature window.

3.	Decrease the best attaxnable reduction.

The choice of the best additive for an industrial installation will
depend on its cost., availability, toxicity and ease of storage.
Natural gas is a strong candidate for use in UK power stations since
it is readily available, non-toxic and easy to store. Thus, a series
of runs were carried out on the CADR to determine the efficacy of
natural gas as a means of controlling the ammonia SNCR temperature
window. Natural gas in the UK is typically over 90% methane.

Effect of Methane (Natural Gas) Addition

Natural gas was pre-mixed with the ammonia before injection into the
flue gas stream.

The effect on N0X reduction of varying CH4:NH3 ratio is summarised in

fig. 7. It is apparent that, at the low temperatures (735 and 800°C),
NOx reduction increases monotonically with CH4:NH3 ratio. At higher
temperatures (865,915°C) , the curves begin to exhibit maxima beyond
which conversion falls as methane increases. At 965°C and above
conversion falls as methane increases.

The effect of methane addition on the temperature window is
summarised in fig.8. Introduction of 0.5 mol methane per mol of
ammonia depresses the best reduction of the process from 68% to 60%,
while the optimum temperature decreases from 1030°C to 916°C, When a
1:1 CH4:NH3 ratio is employed, effectiveness again decreases slightly
but there" is no longer a clearly defined optimum temperature since
the conversion remains constant between 800 and 915°C. The 'window'
of applicability of SNCR was generally wider when methane was in use.
For instance, reductions of >40% were possible over a range of around

5A-105


-------
I50ac in the absence of methane but around 200°c when a 1:1 ratio of
methane to ammonia was used.

Effect of Methanol Addition on"SNCR

Studies were also carried out on the efficacy of methanol in
modifying the behaviour of the process. Results are presented in
fig.9. These show that, at low CH3OH/NOX ratios, methanol does yield
some enhancement of the De—NO^ process but that, at higher ratios, it
can actually cause N0X formation. The most significant finding of
these runs, however, was the ability of methanol to control ammonia
slip at lower temperatures. A demonstration of this effect is shown
in fig. 10. This summarises the results obtained in a test with a
baseline N0X level of 341 ppm and a temperature of 908°C at the
injection point. At this low temperature, there was relatively little
interaction between NQX and ammonia so that 90% of each passes
through the reaction zone. The addition of 0.49 mol of methane per
mol ammonia led to significant N0X reduction (-60%) but still gave
ammonia slip of over 60 ppm. The addition of methanol at lower
temperature (850°C) has little effect on N0X emissions but
significantly decreases ammonia slip. By using a methanol to ammonia
ratio of 2.4:1 it was possible to reduce ammonia slip to around
lOppm.	.

Although the conditions used in this test were not typical of those
which are desirable in a utility installation, the results do
establish the principle of using methanol to limit ammonia slip.

RESULTS FROM 6MW FURNACE MODELLING FACILITY

Trials were carried out on the FMF to determine the effectiveness of
SNCR in a system where mixing is imperfect.

Before any SNCR runs were carried out, the mixing of injected gas
with the bulk gas flow was assessed using helium tracer tests. In
these tests, helium was injected through the five injection ports on
the north side of the FMF and its concentration measured via probes
inserted through the corresponding positions on the south wall (refer

5A-106


-------
to fig.3)

Some of the results obtained are presented in figure 11 which shows
data from tests carried out at an injection temperature of 870°C. It
was discovered that f at low inj sctor momentum , hel iuiq did not
penetrate to the centre of the duct. As the jet momentum was
increased, a peak of hel 1 uin concentration formed towards the centre
of the duct and this peak became sharper as the momentum increased
further. For all the SNCR tests described below, a momentum ratio of
1.5 was used. Therefore, the injection system will have produced a
higher concentration of ammonia towards the centre of the duct.

The N0X reduction results obtained on the FMF using ammonia injection
(both with and without methane addition) are summarised in fig.12
There is considerably more scatter in these data than in those from
the CADR, but the general behaviour is similar. Again, the most
effective temperature for N0X reduction is close to 1020°C. However,
the maximum attainable reduction is considerably reduced (-40%
compared to over 60% in the CADR) and the temperature window is
considerably broadened.

There is again an obvious movement of the temperature window to lower
temperature when methane is injected with the ammonia and NOx
reductions of 35-40% are attainable at a temperature of 800°C.

EFFECT OF SCALE (MIXING) ON ATTAINABLE REDUCTIONS

Fig. 13 presents a comparison of the NOx reduction vs. temperature
plots obtained from the two rigs using ammonia injection alone. Also
shown are the results of Wenli et al. (1990) which were obtained in
an isothermal micro-scale quartz reactor. It is clear that the peak
reduction efficiency decreases as scale increases. In fact, this
variation is probably due to poorer mixing associated with increasing-
scale, rather than to scale itself.

The apparent decrease in optimum temperature in the micro-scale
results arises from the fact that they are obtained under isothermal
conditions, whereas the other experiments are conducted in the

5A-107


-------
presence of significant cooling gradients (-200 K/s). The mean
reaction temperature under these conditions will thus be lower than
the injection temperature, which is the quantity plotted on the
abscissa.

FURTHER WORK

The ultimate objective of PbwerGen's work on SNCR is to develop the
technology to the point where a large-scale power station
installation is viable.

More experimentation is planned on the CADR to characterise further
the interaction of ammonia, methane and methanol and to elucidate the
role of the latter in controlling ammonia slip. This experimental
work will be supported by kinetic modelling of the free radical
chemistry of the SNCR process.

A further project is under way to predict variations in furnace flue
gas temperature with position, load, fouling and firing pattern. This
work is using steady-state power plant modelling system - Ready
(1988) - to examine these effects. The results of these simulations
will be verified by on-site temperature measurements.

CONCLUSIONS

Experiments carried out on 0.15 and 6 MW scale combustion rigs have
demonstrated that ammonia-based SNCR is potentially capable of giving
significant N0X reductions at conditions typical of the convective
sections of industrial p.f. furnaces.

The effects of temperature, NH3:N0X ratio, oxygen content and inlet
NOx level on reduction efficiency have been determined.

The use of methane (natural gas) as an enhancer to alter the
effective temperature range of SNCR has also been demonstrated at
both 0.15 and 6 MW scale.

5A-1Q8


-------
The use of methanol addition to the process has been shown to have
potential as a means of ammonia slip control.

Attainable NOx reductions have been shown to decrease with increasing
rig scale, possibly due to poorer mixing at larger scale.

ACKNOWLEDGEMENTS

The content of this paper draws on work which was undertaken by staff
of the Central Electricity Generating Board who are now employed by
PowerGen p.I.e. and National Power p.I.e.

The authors are particularly grateful for the contributions to this
work of Brian Billinge, John Pye and David Hoadley.

This paper is published by permission of PowerGen p.I.e.

REFERENCES

1.	S.L. Chen, J.A. Cole, M.P. Heap, J.C. Kramlich, J.M.
McCarthy, D.W, Pershing, 'Advanced N0X Reduction Processes

Using -NH and -CN Compounds in Conjunction with Staged
Addition' , 22nd Svitro, flnt.l on Combustion. r The Combustion
Institute, (1989).

2.	A.M. Dean, J.E. Hardy, R.K. Lyon, 19th Symp. flnt.^ on
Combustion. p97, The Combustion Institute, (1982).

3.	EPRI - Report KVB 802200-2029, EPRI RD102A, 1985

4.	A.R. Jones , EPRI Conference on Effects of Coal Quality on
Power Plants. Atlanta, Georgia October 13-I5th, 1987

5.	P. Lodder, J.B. Lefers, Chem. Eng. Journal, 30 161-7
(1985).

6.	R.K. Lyon, (1976) Int. J. Chem. Kinetics. 8, p315

5A-109


-------
A.B. Ready, 'The Use of Steady-State Plant Models in the
Analysis of Fouling-Related Problems Found in Power Station
Boilers', Second UK National Conference on Heat Transfer,
14th-16th September, 1988 Mechanical Engineering
Publications, London

D. Wenli, K. Dam-Johansen, K. Ostergaard, 'Kinetics of the
Gas-Phase Reaction between NO, Ammonia and O2', Preprint -
40th Canadian Chem. Eng. Conf.. Dalhousie Univ., Halifax,
Canada July, 1990.

5A-110


-------
Propane
burner

Figure 1

COAL ASH DEPOSITION RIG AND REAGENT INJECTION SYSTEM


-------
11,12 Fi«ed gas sampling positions

TO lan

Recirculated Hue gas

Figure 2

	 Sampling planea

*	Fiied thermocouple lor Hue gas temp,

•	•	•	¦ metal temps.

Sir® VIEW OF FURNACE MODELLING FACILITY (PMF).

5A-112


-------
Reaction temperature, "<

AMMONIA UUNO UN THIi CAOR; INITIAL NM./NO RATIO 1.0

Figure 4	1

Figure 5

DtMO> ON CADR AMMONIA SLIP 1076 CFLSIUS 2.1* 0?

5A-113


-------
100

fQ

SO

70

I 60

m

%0

o

2 W

10

"I" lOOO'C

~ wrc

Figure 6

Per ceni o«*g«i m ll<* g«

OfiNO ON CADR CHASE II til CCT OT OXYCCN AT AND IWC

0.1	J.2	I*

Methane / ammonia rnoir rai»o

Figure 7

METHANE (NCI FJTiaiT ON AMMONIA DtW^

5A-114


-------
1.0

^ Ammonia only
+ Methane / ammonia i.O
O Methane / ammonia 0,5

700

900

Tcinperafure *C

1100

Figure 8

0.4	0,1

Oil / NO talk)

Figure 9

DENOt ON CAOR EF.-eCT OF MCTHANOt INH, I NOi - II

5A-115


-------
cn
>

cn

X

z

350

300

230

200

O
z

I. 130
>

100

30

NO^ with no ammonia or melhanc (34lvpm)

jj NO^ with ammonia only (308vpm)

Nllj slip with ammonia only (304vpm)

02 2.25%
CH^/NHj 0.49

0.6	1.0	1.4

Methanol (OH) / NHj mole ratio

NO

METHANE ENHANCED DENO USING

Figure 10

AMMONIA'.CH^AT 908T

FOLLOWED BY CH}OH AT 850°C


-------
Distance from duct wall, m
(b). Traverse at top port

Distance from duct wall, m
(a). Traverse at mid port

Figure 1 1 HELIUM CONCENTRATION PROFILES.

5A-117


-------
60

»

A Natwai ga* rtn
¦ Rims without natural

7»

Figure 12

IH

Tconptratiac, *C

toso

If JO

BACK ENf> OVERALL Nt^ REDUCTION CORRECTED TO l;l Nt^/NO .

Effect Of Scale on NOx Reduction

900 950 1000
TEMPERATURE (DEG, C)

1150

Figure 13

5A-118


-------
COMBUSTION NOx CONTROLS FOR COMBUSTION TURBINES

*

Henry Schreiber, P.E.

Project Manager, Combustion Turbines
Electric Power Research Institute
Palo Alto, California

ABSTRACT

The three major currently available nitric oxide (NOx) abatement techniques and
their effect on carbon monoxide (CO) emissions, i.e., water or steam injection, dry
low NOx combustors and selective catalytic reduction are discussed. The advantages
and adverse factors for each method or methods that must be considered in making
a site specific selection of NOx reduction technology are described. A way of
approaching an economically advantageous selection of a site specific NOx reduction
concept is outlined.

5B-1


-------
INTRODUCTION

Gas turbine generators are inexpensive compared to other generation equipment,
are easily installed, highly reliable and achieve very high thermal efficiency as
combined cycles. In the simple cycle configuration they provide a fast start capability
ideal for peaking service. Installed costs range from about $200/KW to S300/KW for
simple cycles, and from about $400/KW to S700/KW for combined cycles. Simple
cycle efficiencies are generally in the high thirties, and combined cycle efficiencies
are close to 50%. There has been a large surge of gas turbine procurement in the
1980s by utilities, cogenerators and independent power producers (Fig. 1). Over
30,000 MW of additional gas turbine capacity is predicted to come on line in the
1990s. Advanced gas turbine technology, benefiting from large government outlays
for improvement of military jet engines, has resulted in much higher reliability and
efficiency than was characteristic of the gas turbines sold in the 1960s and early 1970s.
Concurrently, increased emphasis on NOx and CO emissions abatement by
regulatory agencies, has resulted in the need to devise new approaches to meet
compliance levels (Fig. 2), Gas turbine manufacturers have made considerable
progress in this direction. Post combustion treatment of exhaust gas by chemically
reacting ammonia with NOx on the surface of a catalyst (selective catalytic reduction,
or SCR for short) is also becoming a viable technology.

GAS TURBINE EMISSIONS

Since gas turbines normally fire natural gas, light distillate oil or syn-gas made from
coal, and since combustion efficiency at normal base load operating conditions is
high (very close to 100%), particulate and unburned hydrocarbon emissions due to
incomplete combustion are not of major concern.	-

The NOx emissions from a gas turbine can result from the oxidation of atmospheric
nitrogen in the intense high temperature flame in the combustor, (called thermal
NOx), or from the conversion of fuel bound nitrogen that may be present in some
liquid fuels (called fuel NOx). Some in-engine NOx abatement techniques, such as
water or steam injection into the combustor to cool and dilute the flame, can result
in some loss of combustion efficiency and produce increased CO and unburned
hydrocarbon content in the exhaust. Because of the short residence time of the
working fluid in the combustor of engines with can-annular or annular combustion
systems, full CO burnout may not occur under these conditions.

5B-2


-------
IN-ENGINE NOx ABATEMENT TECHNIQUES

The rate of formation of thermal NOx is directly related to flame temperature arid
residence time at flame temperature (Fig. 3), Consequently, reducing the peak flame
temperature, or reducing the amount of fuel burning at the highest temperature in
the combustor will reduce thermal NOx formation. Fuel NO* cannot be materially
reduced by these means.

A. Water or Steam Injection

When liquid water is injected (usually as a fine spray) into a gas
turbine combustor, heat from the burning fuel vaporizes the water
and brings the resulting mixture of fuel, air, water vapor and
combustion products to a lower working fluid temperature level than
uncontrolled combustion would achieve. Since the residence time of
air in the combustor is unchanged by water injection, the lower rate
of thermal NOx formation resulting from the lower flame
temperature causes a decrease in NOx emission. At the upper limit of
water injection rate for single fuel nozzle can annular combustors,
such as on the G.E. MS7001 series engine, NOx levels can be reduced
by about 70% from uncontrolled conditions (Fig. 4). The upper limits
of water injection flow rate are set by the onset of flame instability,
high CO emissions, increased unburned hydrocarbon emissions (Figs.
5, 6, 7), severely increased wear rates of combustion hardware, and
possibly by surge margin. This accelerated wear is the result of
mechanical vibrations of the combustor liner assembly and the
transition piece induced by high amplitude pressure fluctuations at
acoustic frequencies in the combustor (Figs. 8, 9).

In an EPRI cofunded project (Ref. 1) completed in 1985, General
Electric Company developed a modified combustion chamber design
for its MS7001 series engines. It has six fuel nozzles per combustor
instead of one. This multi-nozzle "Quiet" combustor generates less
(lower amplitude) acoustic noise and suffers less mechanical damage
when heavily water injected. It can operate for up to 12,000 hours
between combustion inspections compared to 3,000 hours for the
single nozzle design.

Water injection mass flow rates in the range of 0.75 to 1.2 lb. water per
lb. fuel have been used. The additional mass flow rate through the
turbine results in a relatively large power output increase because the
parasitic mechanical energy to bring water to combustor injection
pressure is far less than the mechanical energy that would have had
to be expended by the turbine to compress an equivalent mass flow of
air. The water must be demineralized, adding parasitic load. There is
also a resulting heat rate penalty, because the latent heat of
vaporization of the water which was provided by burning the fuel is

5B-3


-------
not fully recovered, due to the atmospheric exhaust from the gas
turbine.

Steam injection also cools and dilutes the flame. Since the heat of
vaporization to make the steam was provided by a heat source
external to the gas turbine combustor, there is a lesser flame cooling
effect per pound of steam injected. Steam injection has a lesser causal
relationship to flame instability-induced dynamic pressure pulsation
and attendant combustion hardware wear rate. The additional mass
flow of the steam increases power output and improves the engine
heat rate (since its energy as a working fluid was only partially
provided by the engine combustion system). Care must be taken to
ensure that adequate compressor surge margin is maintained at the
higher steam flow rates, which could be as high as 2:1 steam/fuel
ratio. The engine manufacturer must define the maximum
allowable steam and water injection rates at all possible engine
operating modes (load, transients, limiting ambient temperatures).

Silo type combustors such as are found on the Siemens and current
models of ABB engines have a much larger volume than can-
annular combustors: and therefore the working fluid has a higher
residence time in them. This allows more time for CO to burn to CO2
and reduces CO emissions at high water or steam injection rates.

Dry Low NOx Combustion

All of the dry low NOx combustors currently being offered by the
major utility gas turbine manufacturers operate on the lean pre-mix
principle. Siemens and ABB are offering dry low NOx silo type
combustors. These combustors are capable of dry low NOx operation
on gas fuel only, but are also capable of firing oil in the diffusion
flame mode while using steam or water injection for NOx reduction.
General Electric Company is offering a can-annular combustor with
the same fuel constraints (Fig. 10). Westinghouse has a can-annular
dry low NOx system in development.

The principle of operation of the lean pre-mix type of dry low NOx
combustor is to create as uniform as possible a fuel lean mixture of
fuel and air prior to combustion. This mixture is then introduced to
the combustion zone in the combustion chamber at a controlled
velocity sufficiently higher than the local speed of flame propagation
so as to prevent the flame from flashing upstream into the pre-mix
zone (Fig. 11). The velocity of the pre-mixture must also be low
enough so as not to blow the whole flame downstream.

When burning in this mode, there is no diffusion flame front where

a high temperature stoichiometric flame exists, because all parts of
the mixture are at below stoichiometric fuel/air ratio. The resulting
flame volume is therefore lower in temperature, since the chemical

5B-4


-------
fuel energy released by combustion must heat a greater mass of air in
intimate contact with it at the moment of combustion. Since burning
rate is also a function of local air and fuel temperature, the cooler
lean pre-mix flames require more time to achieve full burnout of fuel
than the hotter diffusion flame. These combustors are more complex
than diffusion type combustors as they require precise control of local
velocities and sequencing of fuel/air ratios during transients, starts,
and stops. Pilot diffusion flames generating NOx at a high rate (but at
a low total mass flow) may be employed to prevent lean blowout of
the main flame. In a water injected mode firing liquid fuel, NOx
levels of about 42 ppm have been offered. Where regulatory
requirements dictate lower emission levels with liquid fuel,
operating hour limits or post combustion treatment may be needed
for compliance.

The technology of dry low NOx combustion using a lean pre-mix
flame is relatively new and is still being actively developed and
refined. It has not been proven in long-term problem-free service in
the United States. Reliability characteristics have not been established
by user experience. Manufacturer's claims should be carefully
evaluated by the potential buyer against available experience. It must
be emphasized that the present designs being offered are the product
of as much as ten years of research and development work by major
engine manufacturers, attesting to the difficulty of achieving these
objectives.

Other low NOx combustion techniques that avoid long residence
time at high temperature, as well as catalytic combustion designs
have been explored, but none of these techniques have achieved
commercial viability in large utility gas turbines. Due to the fact that
gas turbines are usually purchased on a lump sum competitive bid
basis, the incremental price of the dry low NOx system is not known
to the buyer, unless specified as a separate option. At this time, the
technology is too new to allow normal, commercial pricing.

C Post Combustion Treatment

A third method of NOx abatement involves treatment of the
combustion products after they leave the engine. CO abatement can
also be achieved this way. Selective catalytic reduction (SCR) is a
process whereby ammonia is reacted with NOx in the gas stream to
form N2 and H2O on the surface of a catalyst interposed across the gas
stream (Ref. 2). The reaction proceeds in the desired direction when
the gas and catalyst are in a temperature window of 550°F to 750°F,
and is capable of achieving NOx levels in the single digits.

There is as yet no long term operating and maintenance experience
with SCR on large utility gas turbines in the U.S. Also, (Ref. 3) there
is no significant experience on the successful use of SCR on oil fired

5B-5


-------
gas turbines. This SCR technology was introduced in the U.S. fairly
recently, and its use is expanding rapidly in response to stricter
regulatory requirements (Fig. 12). A number of units are being
operated by cogenerators and independent power producers. The
technology and O&M costs of SCR are evolving.

Since the majority of systems operating to date require that the
combustion products undergoing the reaction be within a 550°F to
750°F temperature window, it is necessary to cool the 950°F to 1100°F
exhaust from a gas turbine by means of either a heat recovery steam
generator (HRSG) or dilution with ambient air. Because of the large
gas mass flows involved, dilution with air to achieve uniform
mixing is technically impractical and uneconomical. Therefore a
HRSG is normally used for this purpose. The resulting steam can be
used for steam injection into the gas turbine for NOx reduction and
power augmentation, as input to a combined cycle (or repowered)
steam turbine, or as process steam in a cogeneration plant-

High temperature zeolite catalysts have been introduced more
recently, but they are far more costly. Zeolite manufacturers claim
that these catalysts are effective and stable over a wider temperature
range (and especially at higher temperatures) than base metal oxide or
precious metal catalysts. The wide operating temperature range of
zeolite catalysts is the most important property for NOx control.
However, the specific temperature range depends on the type of
zeolite. For example, a naturally occurring mordenite zeolite can
operate between 430° to 970°F depending upon the specific
formulation. The optimum operating temperature window for a
specific formulation is ±100°F. A synthetic zeolite, ZSM-5, has a
narrower operating temperature range of 570s - 900°F. A new
synthetic zeolite, which is coated on a ceramic honeycomb structure is
claimed to be operational at temperatures between 675° and 1075°F.
However, above 800°F, NH3 begins to be oxidized to NOx, which is
counter productive. Because zeolites contain no heavy metals,
manufacturers claim that spent catalyst disposal presents less
problems than for conventional catalysts.

Due to the large cross sectional area of the duct necessary to
accommodate the large gas mass flow with acceptable pressure drop, it
is usually not practical to achieve completely uniform mixing of
ammonia and combustion gas. Thus, at a given NOx level, there will
be unreacted ammonia (ammonia slip) emitted from the stack. Other
issues affecting the use of SCR are the catalyst initial cost and
replacement cost, catalyst disposal (hazardous waste), and the fouling
of catalyst and heat transfer surfaces in heat recovery boilers that
results from the formation of ammonium bisulfate and ammonium
sulfate when sulfur bearing liquid fuels are burned, or when the
ambient atmosphere contains sulfur.

5B-6


-------
Since SCR works on the basis of a percentage reduction of exhaust gas
NO* content, it is desirable to reduce NOx levels to a minimum by
less expensive means such as water or steam injection prior to SCR so
as to reduce the amount of catalyst and ammonia required. Latest
indications are that SCR, on this basis, adds $30 to $50/KW of capacity
to a gas turbine installation. Since SCR requires ammonia storage,
there is also a safety issue involved.

Ultimately, an engineering evaluation (Ref. 2) is required on a site
specific basis to determine the most economical combination of
methods to achieve the required emissions level. For example, it
may be desirable to bring NOx from 150 ppm to 50 ppm (67%
reduction) with water injection and from 50 ppm to 9 ppm (80%
reduction) by SCR. The volume of catalyst required increases at a
greater than linear rate with the percent NOx reduction needed.
Obviously, when catalyst beds and ammonia distribution grids are
introduced into the flow stream, additional combustion gas pressure
drop results, causing a heat rate penalty.

When water injection is used in conjunction with SCR, the CO levels
leaving the engine may be excessive. A CO oxidation catalyst may be
required ahead of the SCR system (upstream of the ammonia
injection grid) at a region in the HRSG where the appropriate
temperature level for CO oxidation exists.

OVERALL APPROACH TO NOx REDUCTION IN GAS TURBINES

A utility faced with siting a new plant, repowering an existing plant or retrofitting
an existing plant may need to provide for NOx abatement in response to increased
stringency of emissions control regulation. A very broad view of the problem is
required to achieve the best site specific solution. It is important to take advantage
of as much lead time as possible to plan the strategy to be employed in achieving a
cost effective response to regulatory requirements. Over the past few years,
mandated NOx levels have been ratcheted downwards, and have been somewhat of
a moving target. An early understanding of the local regulatory process and the
posture of the regulatory body or bodies is advantageous. Early public education
campaigns have been helpful in some cases to alleviate public concerns about new
plant sitings or modifications. A thorough knowledge of the various technical
aspects of NOx reduction in gas turbines is essential. Unless a utility has a sizeable
technical staff and has kept abreast of rapidly evolving technology and regulatory
developments and decisions, it would appear highly desirable to engage outside
consulting organizations that have expertise in these areas when an undertaking of
this kind is contemplated.

CONCLUSIONS

The increasing popularity of gas turbine generating systems coupled with the greater
regulatory stringency of emissions levels, makes it important to have a thorough

5B-7


-------
understanding of the technical as well as administrative aspects of gas turbine
operations compatible with environmental requirements. Water injection, steam
injection, dry low NOx combustion and post combustion treatment for NOx and CO
by using SCR and CO catalysts are all currently available means of NOx and CO
emissions abatement. All of these technologies have adverse economic affects,
necessitating careful study of the best combination of alternatives to meet regulatory
requirements.

REFERENCES

1.	EPRI Report AP-3885, Project RP1801-1, May 1985; High Reliability Gas
Turbine Combustor Project, prepared by the General Electric Company.

2.	EPRI Project RP2936-1; Gas Turbine Best Available Control Technology
Guidebook: to be published third quarter 1991.

3.	EPRI Report GS-7056, Project RP2936-1, December 1990; Evaluation of Oil
Fired Gas Turbine Selective Catalytic Reduction (SCR) NOv Control.

5B-8


-------
U.S. GAS TURBINES: YEAR ORDERED BY CUSTOMER TYPE
FOR ELECRTIC POWER GENERATION, 1980 THROUGH 1989

IOC We
SGWe
6GWe
4GWe

2GWe
OGWe

1980 1981 1982 1983 1984 1985 1986 1987 1988 1989

© 1990 UBS-Phillips and Drew Global Research Group
(Reproduced with Permission)

Figure 1.

Electric utility

NOx EMISSION REDUCTION

Figure 2.

5B-9


-------
N0S Production (ppmv/mscc)

40 r

Tempcraturc (F)

4000

3000

2000

1000

0 0.2 0,5	1.0	1,5 1.8 2.0

Fuel-Air Equivalence Ratio,O

Figure 3.

NOx REDUCTION vs WATHR-TO-FUF.L RATIO

NO, Emissions Reduction (%)
80

60

40

20

Water

—— Firing natural gas
—— Firing distillate oil

0 0.2 0.4 0.6 0.8 ].0
Water-to-Fuel Ratio (lb/lb)

1.2 1.4

Figure 4.

5B-10


-------
.INCREASE IN HYDROCARBONS
, DUB TO WATER INJECTION

Water/Fuel - Ratio
Figure 5.

CARBON MONOXIDE INCREASE DUE
TO WATER INJECTION

Water/Fuel - Ratio
Figure 6.

5B-11


-------
NITROGEN OXIDES vs
CARBON MONOXIDE EMISSIONS

NO * Emissions (ppmv)

0	_i			1—		—		1

o	100	200	300	400

DYNAMIC ACTIVITY vs WATER INJECTION

Overall rms Level Dynamic Activity, psi (10 chamber average)

CO Emissions (ppmv)'

Figure 7

1.6

0.8
0.6

1.0

1,2

Qt. comb. MN - no. 2 oil

0

10	20 30 40	50 60

Water Injection Rate, (gal/min)

Figure 8

5B-12


-------
DYNAMIC PRESSURE COMPARISON

namsc Pressure, psi jpcak-fo-peak}



nrlXJ

Dyn> mic Pressure, psi (pealMo-peak)

i.o r

Bnscline(SN)	O.B

proihikliun I iter

0,6
Q.4
0,2

Quwl (MN)
cciiibtiior liner

0 ' 200 400 60Q BOO 1000 0 200 400 fiOO 800 1000
frequency, Hz	Frequency, 1U

Figure 9.

Ouier casing-i rlTIow sleeve

Cap



Convent ion al\ ;/
lean and

Secondary zone
rVentuii

pre-mmng
primary zone

IE

-End cover

Plane of
dilution holes

Dilution zone

Figure 10.

5B-13


-------
DRY LOW NOx COMBUSTOR
OPERATING MODES (49)

i'irsi Stage Burning.

jxf

Second Stage Burning

jxf

Two Stage Burning, Lean-Lean

nf

First Stage Premised-
Second Stage Burning

nf

Figure 11,

U.S. COMBUSTION TURBINE SCR INSTALLATIONS
IN OPERATION (AND PROJECTED)

Generating Capacity (M We)

1,600



.5,200

~

1,800

r

2,400

•

2,000



1,600

•

1,200



800

•

400



0



1986

1987

1991

Figure 12.

5B-14


-------
SCHEMATIC DIAGRAM
(V/T1O2 or Zeolite SCR Catalyst)

Figure 13.

5B-15


-------

-------
ENVIRONMENTAL AND ECONOMIC EVALUATION OF GAS TURBINE SCR NOx CONTROL

Phillip A. May, Lisa M, Campbell, and Kevin L. Johnson
Radian Corporation
Research Triangle Park, North Carolina 27709

5B-17


-------

-------
ABSTRACT

Approximately 3600 MW of gas turbine SCR capacity is in operation or start-up in the U.S.
Total gas turbine SCR operating time is approximately 600,000 hours with a mean operation
per unit of 10,000 hours. Many additional sites are either under construction, permitted, or in
the process of obtaining a permit.

Experience obtained from operating SCR sites will assist in defining both actual control costs
and key procurement/technical feasibility issues pertinent to future U.S. applications. This
paper characterizes the state of the art with respect to the application of SCR. Operating and
cost data collected in a SCR site field program are presented along with a discussion of key
technical and procurement issues identified in conjunction with designers and manufacturers.
Key design and procurement issues include correct catalyst placement within the operating
temperature window, ammonia distribution, and the potential formation and deposition of
ammonium salts associated with combustion turbine SCR systems firing sulfur-bearing fuels.

5B-19


-------
INTRODUCTION

The use of gas turbines in cogeneration and utility applications has risen sharply over the
past decade. In conjunction with this rise, SCR as a NOx control technology has been
introduced over the past five years in regions of the U.S. with acute air quality problems. In
addition, NOx emissions are receiving increased attention at both the federal and state level
because of new Clear Air Act requirements and growing regional, state, and local regulatory
requirements. No database from which to evaluate the reliability, cost, and performance of
SCR systems under anticipated' operating conditions has been available.

This paper summarizes the results of a study performed to characterize the current status of
SCR applications to gas turbines, determine the true cost of applying SCR controls, and
identify key design and procurement issues. Included is a summary of the state of the art in
the U.S., a characterization of the SCR systems included in this study, an example of the
capital and operating costs associated with the application of SCR, and a discussion of the
key design and procurement issues identified,

STATE-OF-THE-ART

At the end of 1990, the total installed SCR capacity for gas turbines operating in the U.S. was
approximately as follows:

80 sites
110 units
3600 MW

Almost all of the operating units are in California, with units also operating in New Jersey,
Massachusetts, and Rhode Island. AH but one of these units is in a cogeneration application.
Most of the units (-85%) are in the 20 to 80 MW size range, with some units in the 3 to 10

5B-20


-------
MW range. No gas turbine SCR systems in the 3 to 10 MW size range operates outside of
California.

Figure 1 shows the total cumulative SCR system capacity for gas turbines in operation over
the last five years (1985-1990). The figure indicates that most of the U.S. capacity came on-
line in the last 3 years. Relative to current catalyst guarantees of 2 to 3 years this indicates
that the industry is still young.

The NOx permit limits for gas turbine SCR sites in operation, under construction, or with
active permits are presented in Figure 2. A few of the earlier California sites and a New
Jersey site have NOx permit limits on the order of 15 to 25 ppmvd (@15% 02) but, most of
the recently permitted sites have NOx limits of 9 ppmvd. Currently, 9 ppmvd is the most
common level outside of California. The levels shown in Figure 2 for units below 9 ppmvd
are all in California. A number of California sites have NOx emissions levels of less than 9
ppmvd in order to minimize offset requirements. Essentially all of the gas turbine SCR
installations operating achieve these low NOx levels by applying SCR in combination with wet
injection, either steam or water. This is typically done by reducing NOx emitted from the
turbine with wet injection down to 25-42 ppmvd and then applying SCR. The distribution of
NOx reduction performance at a number of gas turbine SCR sites operating in the U.S. is as
follows:

NOx Reduction (%)	Percent of Sites

80	70

75-80	5

70-75	5

65-70	10

60-65	10

Most of the gas turbines operating or planned to operate with SCR use natural gas as their
primary fuel, with a few of these units firing refinery gas. There is very limited experience in
the U.S. firing distillate oil at gas turbine SCR facilities.

Most SCR catalysts in use are composed of base metal oxides, primarily vanadia and titania,
on titania, silica, or tungsten supports. Optimum NOx reduction for these conventional SCR
catalysts occurs in the 600° to 750°F temperature range. Below 600"F, NOx conversion slows

5B-21


-------
dramatically; at temperatures above ~8Q0-850DF, these catalyst materials can lose surface
area and reactivity. This requires location of the SCR catalyst within a heat recovery steam
generator (HRSG) to obtain the proper operating temperature window. In the last few years,
molecular sieve zeolites have been commercially marketed in the U.S. Zeolites have
reportedly extended the SCR operating range up to approximately 95CTF. Currently, there
are four gas turbine zeolite applications.

The other major SCR catalyst type recently applied commercially in the U.S. is a precious
metal-based (e.g., platinum) catalyst. This catalyst has an operating temperature window of
about 425 to 525°F, with an optimum temperature of approximately 475°F. This low-
temperature operation allows placement of the catalyst outside the high pressure section of
the HRSG, upstream of the economizer and stack. However, there are two limitations to this
catalyst type. First, at higher temperatures (>525°F), this catalyst is an excellent NH3
oxidation catalyst, producing additional NOx. Second, it is limited to clean fuels because it is
also a good S02 oxidation catalyst, forming S03, with potential for forming ammonium
sulfates and increasing downstream corrosion,

GAS TURBINE SCR OPERATING EXPERIENCE

To characterize the cost and operating experience of gas turbine SCR applications in the
U.S., information was collected from approximately 20 sites, including capital cost, operating
and maintenance, and reliability/availability data.

Study Group Characterization. A total of 37 operating SCR units applied to gas turbines
ranging in size from 3.5 to 80 MW were included in the study. Gas turbine
manufacturers/models included in the study are: General Electric (GE)/LM 2500, LM5000,
and Frame 5,6, and 7EA; ASEA Brown Bovari (ABB)/Type 8; Solar/Centaur and Mars; and
Allison/501 -KB.

Permit levels for NO„ emissions range from 5 to 21 ppmvd. Some of the permits also include
ammonia emissions limits ranging from 15 to 20 ppmvd, All of the gas turbine SCR systems
included in the study use natural gas as their primary fuel. Back-up fuels include distillate oil
and Jet A. Operation with the back-up is almost nonexistent. The following SCR system
suppliers are included in the study group: Babcock/Hitachi, Engelhard, Hitachi Zosen,

5B-22


-------
Ishikawajima-Harima Heavy Industries (IHI)/Foster Wheeler, Johnson Matthey, Mitsubishi
Heavy Industries, Norton, and Steuler. Nine of the sites also include carbon monoxide (CO)
catalyst systems. All but two of the systems included in the study use an anhydrous
ammonia system. Both of the aqueous ammonia systems started up in the latter half of
1990.

Study Group Results. The results from the study group of 37 operating SCR units are
divided into three categories: SCR capital costs; SCR operating parameters; and SCR
maintenance history.

SCR Capital Costs. Installed capital cost data for the SCR reactor and subsystems was
collected from 11 sites in the study group, representing 16 total SCR units. The SCR units
from which capital costs were collected range in size from 3.5 to 80 MW. Figure 3 presents
the installed capital cost data ($/kW) versus gas turbines size (MW) for the 11 sites. The
installed capital cost ranges from S30/kW to $100/kW. This represents 5-25% of the total
installed capital cost of a combined cycle combustion turbine system.

The sites which were installed earliest, or first generation U.S. SCR sites, were found to have
a higher installed capital cost than the equivalent size units which are newer. This is shown
in Figure 3 by the two upper data points at 22 MW and 37 MW. This trend indicates that
catalyst costs have declined. The two data points at the 80 MW size differ in cost by about
S10/KW. This difference in cost is attributed to one site procuring and purchasing the SCR
unit as a change order, after the initial design and equipment specifications were made.

SCR Operating Parameters. The key SCR operating/cost parameters for the 37 operating
SCR units in the study group are summarized below.

SCR OPERATING PARAMETERS

Operating Parameter

Actual Operating Range

Outlet NOx, ppmvd
NOx Reduction, %

N03/NOx Molar Ratio
Pressure Drop (across catalyst),
in Wc

Maintenance, man hours/year

5 - 21
60 ¦ 95
0.9 - 1.6
1.9 - 6.1
170 - 3130

5B-23


-------
Outlet NOx concentrations range from 5 to 21 ppmvd (at 15% 02). The NH3/NOK molar ratio
ranges from 0.9 to 1.6, with a corresponding NOx reduction range of 60-85 percent with one
site achieving 95 percent. This site is unique in that five turbines exhaust to a single SCR
system and only two turbines are currently fired.

The pressure drop across the catalyst systems ranges from 1.9 to 6.1 inches water. The
pressure drop across the SCR catalyst bed increases the back-pressure on the turbine. This
reduces the power generating capacity and increases the heat rate of the turbine.
Maintenance labor required for the SCR systems at the study group sites are reported to
range from 170 to 3130 man hours per year. Most of this time is devoted to the CEM
system.

The history of SCR catalyst replacement or additions within the study group is limited based
on the low total operating hours of the units. The total operating hour range of the study
group is about 1200 to 40,000 hours. Only three sites out of the 20 sites in the study group
have replaced or added catalyst. The experience of these three is as follows;

Site	Total Operating Hours	Catalyst Replacement/Addition

Site 1	-40,000	6 catalyst replacements or additions

Site 2	6,000	1 catalyst addition

Site 3	24,000	1 catalyst addition

With the limited operating hours represented in the study group, no conclusions can be
made on the frequency of catalyst replacement.

The SCR operating parameters presented can be used to determine the annual operating
cost range for a specific SCR unit. For an 80 MW combustion turbine application, the
resulting range in annual operating cost is 1.30 to 3.2 mil/kWh. The cost components and
their contribution to the total annual operating cost are as follows:

5B-24


-------
Cost Component

Percent of Annual Operating Cost

ammonia usage	2-10

heat rate penalty	6 - 8

replacement catalyst	36-46

maintenance cost	1-6

overhead cost	0.2 - 2

capital charges	22-44

G&A, taxes, insurance	6-12

SCR Maintenance History. A maintenance history was collected from each of the SCR study
group members. The number of events for each of the plant sections including the gas
turbine, HRSG, water treatment, water injection, SCR, and CO systems was then totaled.
Figure 4 shows the percentage of events reported for each part of the facility. As shown, the
SCR system, including all SCR subsystems, represents about 20 percent of all events
reported and is on balance with the other major plant systems. As a check on the results
presented in Figure 4 plant operators in the study group were polled to assess the order of
priorities when starting a new shift. The order of priority determined was as follows: 1) water
treatment; 2) SCR: 3) HRSG; 4) gas turbine.

SCR system events were divided among three subsystems: catalyst, ammonium, and
continuous emissions monitoring (CEM). The percentage of failures attributed to each of the
SCR subsystems, is presented in Figure 5. The ammonia subsystem includes the ammonia
storage, vaporization, mixing, injection, and ammonia control system. The catalyst
subsystem includes only the SCR reactor housing and catalyst itself. The CEM subsystem
includes the NOx, CO, and 02 sample probes and analyzers, and the gas conditioning
systems. As shown, 25% of the events reported are attributed to the ammonia system and
catalyst system, respectively, while the majority of the failures (50%) are attributed to the
CEM system.

The failure distribution for the CEM and ammonia subsystems are shown in Figures 6 and 7,
respectively. The CEM subsystem failure distribution indicates that the component with the
highest failure rate (45% of the total) is the NOx analyzer. The gas conditioning system has
the second highest (20%) malfunction rate. No root cause identification has been performed,
so it possible that the high rate of NO^ monitor failures is linked to gas conditioning system

5B-25


-------
failures. The CEM subsystem failure distribution is based on 28 reported events within a six
month period of operation.

The ammonia subsystem failure distribution of Figure 7 indicates that two components, the
ammonia vaporizer and the ammonia flow control valve, have had the highest failure rate
(each component represents 40% of the total ammonia system failures). This failure
distribution for the ammonia subsystem is based on 10 reported events occurring within a six
month operating period.

GAS TURBINE SCR DESIGN ISSUES

Key areas identified included: placement of the SCR catalyst in the optimum temperature
window, flexible distribution and adjustment of ammonia (NHg), and HRSG impacts that result
from firing sulfur- bearing fuels. Other areas identified included: communication channels for
procurement and continuous emissions monitoring interfaces with regulatory reporting
requirements. The following section is an overview of the information obtained. Where
possible the experience of actual sites is used to illustrate the potential impact.

Optimum Catalyst Placement. Proper placement of the catalyst within the HRSG is essential
to achieving consistent NOx reduction. Incorrect placement or variations in the boiler
temperature outside of the SCR system's design range can result in deviations in the
operating temperature for the catalyst which in turn can lead to unnecessary increases in
ammonia usage, reduced catalyst performance, and unnecessary or premature catalyst
replacement.

To include the assurance of correct catalyst placement in the HRSG and SCR procurement
process, anticipated operating conditions are included in the HRSG and SCR system
specification packages. Of particular Importance are anticipated gas turbine load swings,
shifts in HRSG steam demand, duct firing impacts, and changes in HRSG performance with
time. Open communication of anticipated operating conditions between the HRSG and SCR
system suppliers is key to a well-designed system.

The importance of this concept is best illustrated through the experiences of one of the study
group members. The operators of this site were considering catalyst replacement because

5B-26


-------
of unanticipated rises in the NH3 injection rate. Figure 8 shows the relative outlet NOx level
and NH3 usage as a function of time for this site. Also shown are the related plant events.
As the figure indicates, NH3 usage increased following each of the facility's outages over a
two-year period. Although site personnel were aware of the rise in NH3 usage and the
potential implications, no cause for this increase was identified until repeated boiler tube leaks
resulted in adjustments to the steam flow within the system and subsequent revaluation, of
the temperature path within the HRSG. Following each of the outages the performance of
the HRSG had been altered resulting in a temperature shift at the SCR catalyst position within
the boiler. The temperature shift was undetected because of inadequate and poor
thermocouple placement. In response to this problem, the site installed additional
thermocouples and now monitors the temperature at several locations in the HRSG on a daily
basis.

Ammonia Injection and Distribution. Ammonia injection and distribution is key to achieving
required NOx emissions limits, meeting any NH3 slip permit requirement and preventing
ammonium sulfate and bisutfate formation in SCR applications where sulfur-bearing fuel is
fired.

The reaction of NH3 and NO is equimolar, but approximately 2 moles of NH3 are required to
react with each mole of N02. Because gas turbine exhaust is primarily NO, a slight molar
excess of NH3 is required to react with NOx. For an optimally designed and perfectly mixed
SCR system, an approximate 1.0 NH3/NOx mole ratio is required to achieve 80 to 90% NOx
reduction when the catalyst is new. Because a perfectly mixed SCR system is not possible,
care should be taken in the design of the HRSG to ensure even flow at the catalyst surface
and flexibility in the NH3 distribution system.

The need for a flexible NH3 distribution system is best illustrated by data collected at one of
the study group sites. The HRSG at this site is a split boiler. Figure 9 presents the results of
a velocity and NOx traverse performed at the inlet to the catalyst. The average velocities of
30.2 and 26.5 ft/sec (with ranges of 21 to 38 ft/sec) for trains A and B, respectively, indicate
uneven flow between the two halves of the unit, and widely variable flow within each train.
There was relatively less variation in the inlet NOx, with Train A (25 - 29 ppm) and Train B
(23 - 34 ppm). Therefore, the ability to adjust NH3 flow distribution is critical to meet NOx
reduction performance requirements and minimizing NH3 slip.

5B-27


-------
Sulfur-Bearing Fuel-firing Issues, Combustion of sulfur-bearing fuels creates SOx emissions;
a portion of these emissions is in the form of S03. In addition to the S03 from combustion,
S02 oxidation forms additional S03 across the boiler tubes within the HRSG and across the
SCR catalyst. Available base metal catalysts oxidize between 1 and 5 percent of the S02
present in the turbine exhaust gases to S03. Some base metal catalysts offer an S02
oxidation potential of less than one. However, these low- oxidation formula catalysts also
have a decreased NOx reduction activity per unit volume. Thus, greater catalyst volumes are
require to achieve an equivalent reduction. Zeolite catalysts are claimed to offer the
advantage of significantly lower (<1%) S02 oxidation rates.

One of the unique features of U.S. gas turbine SCR applications is that they may also be
combined with a CO catalyst upstream at the entrance to the HRSG. When a CO catalyst is
present in the system, as much as half of the S02 in the gas turbine exhaust may be oxidized
to S03. Therefore, CO catalysts can have a significant impact on the S03 content of the
exhaust gas stream.

There are two potential problems associated with increased S03 in the exhaust gas stream:
First, S03 can be collected as a particulate in the form of H2S04 if the particulate collection
train used for compliance measurements is operated at temperatures below the acid gas dew
point. This is the case in certain states including California and New Jersey where the
sample is collected at ambient conditions. Second, unreacted NH3 slip from the SCR system
can react with S03 and form either ammonium sulfate and/or bisulfate salts via the reactions:

2NH3 + S03 + H20- (NH4)2S04 (ammonium sulfate)	(1)

NH3 + S03 + H20 - NH4HS04 (ammonium bisulfate)	(2)

Even at levels of a few ppm slip, NH3, S03, and aerosol H2S04 can react to form ammonium
sulfate and bisulfate deposits.1

Ammonium bisulfate is a sticky substance which deposits on downstream equipment,
particularly HRSG tubes at lower tube metal temperatures. These deposits can cause
corrosion and plugging, eventually resulting in loss of heat exchange efficiency, increased
pressure drop, and shortened equipment life. Ammonium sulfate is a white, crystalline (flaky)

5B-28


-------
compound which deposits on lower temperature surfaces. Corrosion and plugging problems
can also occur with the sulfate. The potential for salt formation increases as temperature
decreases. At very low temperatures (<400°F), only a few ppm of NH3 and S03 are required
for reaction. Therefore, at typical HRSG exit temperatures (300 to 350°F), ammonium salt
deposits are expected to form in the HRSG when firing sulfur-bearing fuels.

Ammonium salt formation temperature is shown as a function of NH3 and S03 concentrations
in Figure 10. For a gas turbine firing 0.2 percent sulfur distillate the exhaust gas S03
concentration is approximately 2 ppm. As shown in Figure 10, if the HRSG exit gas
temperature is 410°F, then to avoid salt formation the NH3 slip must be controlled to less than
5 ppm. However, if a CO catalyst is present in the system, the S03 concentration in the
exhaust can be as high as 20 ppm (i.e., 50 percent S02 oxidation across the CO catalyst).
For this case there is no NH3 slip level which will guarantee against salt deposition.

Although many gas turbine SCR systems have been designed to fire a sulfur-bearing
secondary fuel, few have operated with such a fuel. As a result, there is little gas turbine
SCR operating experience in the U.S. with sulfur-bearing fuels. Two sites were identified with
operating experience firing refinery gas as a secondary fuel.

SCR PROCUREMENT PROCESS

Several approaches to SCR procurement have and can be used and the degree of
involvement for each party differs among them. The utility has the option to develop
contracts with: (1) the architectural/ engineering (A/E) firm which, in turn, has a contract
with the HRSG vendor to procure the SCR system; (2) the SCR vendor directly; (3) the A/E
firm which, in turn, has a contract directly with the SCR vendor; or (4) the A/E firm acting as
the owner's agent. The advantages and disadvantages associated with each method are
discussed below.

Utility - A/E - HRSG - SCR Vendor. The most common procurement method involves the
A/E firm procuring the SCR as a part of the HRSG system. In this arrangement, the HRSG
vendor includes a performance guarantee for the SCR system as part of the HRSG package.
This performance guarantee is identical to the guarantee provided by the SCR vendor;
however, the HRSG vendor is legally liable to the A/E and utility for SCR performance.

5B-29


-------
Inclusion of the SCR system within the scope of the HRSG package is preferred for the
following reasons;

Integration of the SCR system into the HRSG is enhanced;

Design changes that may affect the interface of the two systems are
more readily implemented;

Optimization of the SCR reactor operating temperature and catalyst
placement in the HRSG are easier to achieve; and
* A single vendor provides performance guarantees and is responsible
for both the HRSG and SCR systems.

Another variation of this procurement method is for the buyer or A/E firm to procure the
HRSG and SCR from the gas turbine manufacturer. This has the added advantage of
obtaining a single point responsibility for ail emissions and velocity distributions, and it more
closely integrates the HRSG and SCR systems with the gas turbine cycle performance.
However the SCR procurement experience of some gas turbine manufacturers may be
limited.

Utilitv-SCR Vendor. Some buyers have the engineering staff and expertise required to
design, procure, and, in some cases, construct a power generation facility in-house without
the assistance of an independent A/E or engineering/construction (E/C) firm. In this
scenario, the utility may work directly with the SCR vendor to secure a contractual
agreement. Some of the advantages of this direct working relationship between the buyer
and SCR vendor include:

Procurement of the SCR separately from the HRSG allows the lowest

cost (i.e., initial capital cost) system to be selected for each, rather

than the low cost bid package including both systems.

HRSG and A/E fees are not included in the SCR cost, but the SCR vendor and

the HRSG vendor incur coordination labor costs.

Closer contact between the SCR vendor and the end user helps

ensure that the needs of the end-user are met satisfactorily. This close

contact also ensures that the utility is aware of system features, such

as unique design or technology, which may impact cost.

It should be noted that the greatest potential disadvantage is missing direct coordination
between the HRSG and SCR manufacturers. In any scenario, the HRSG vendor must be

5B-30


-------
involved in determining the location of the optimum temperature range in the HRSG. Another
disadvantage is that most buyers have less experience in procuring an SCR system than
A/Es, C/Es, or HRSG vendors.

Utility - A/E and/or E/C. There are three potential working relationships between a buyer
and an A/E and/or E/C firm with respect to procurement of an SCR system. The first
involves the utility implementing all stages of SCR procurement with the exception of
construction, which is contracted out to an E/C or construction management A/E firm. In
this arrangement, the utility and not the E/C has a contract directly with the SCR vendor.
The arrangement between the utility and A/E firm involves the A/E working as the owner's
agent to develop a detailed specification for the gas turbine/HRSG/SCR system. The A/E
firm acting as the owner's agent also reviews the bids to verify that the proposals meet the
bid specification. The third arrangement between the utility and the E/C firm involves the
E/C acting as the turnkey contractor responsible for detailed system design and
construction. In this case, the E/C firm provides the SCR system specification along with the
HRSG specification to the HRSG vendor. The E/C firm also provides the balance of the plant
design and procurement, and manages overall plant construction.

Utility - A/E - SCR. In some arrangements between the buyer and the A/E firm, the A/E
procures the SCR system directly from the vendor. Some A/E firms prefer to procure the
SCR system directly for the following reasons:

tower cost of the SCR system;

Direct accountability of the SCR vendor to the A/E firm; and
Direct communication between A/E and SCR vendor.

The same potential disadvantage of missing coordination between the HRSG and SCR
manufacturers also exists.

References

1.	EPRI Report GS7056, Project 2936-1, December 1990; Evaluation of Oil Fired Gas
Turbine Selective Catalytic Reduction fSCR) NO:i Control.

2.	Saleem, A., M. Galagano, and S. lnaba. "Hitachi-Zosen DeNox Process for Fossil
Fuel-Fired Boilers." Proceedings: Second NO, Control Technology Seminar. Hosted
by Electric Power Research Institute. Denver, Colorado. November 8-9, 1978.

FP-1109-SR. p 22-12.

5B-31


-------
4,000
3,750
3,500
3,250 -
3,000
2,750
2,500
2,250 H
2,000
1,750
1,500
1,250 -
1,000 -
750
500
'250

86

07

90,

88	89

Year

Figure 1. Total U.S. GT SCR Capacity in Operation

91

w
cr

a
B

1

E

3

z
¦a

6.5 7 8 9 10 12 13 14 15 16 22 25 26
NOx Permit Levels, ppm (@ 15% Cy

Figure 2. U.S. GT SCR NOx Emission Permit Levels

5B-32


-------
30
20
10

i : i 1 i i i i i r-
10 20 30 40 50 60 70 80
Gas Turbine Size, MW

Figure 3. Gas Turbine SCR Installed Capital Cost

18% Water Treatment

17%HRSG

17% Water Injection

20% SCR

7% CO Catalyst

21% Gas Turbine

Figure 4, Facility Wide Failure Distribution

5B-33


-------
50% OEM System

Figure 5. SCR System Failure Distribution

Figure 6. Continuous Emissions Monitoring Failure Distribution*

5B-34


-------
40% Ammonia Flow Control Valve

40% Ammonia Vaporizer

10% Ammonia
Injection Nozzles

10% Ammonia Mixer to Injection

* Total fequency over six month period of 10 events over 15 sites.

Figure 7. Ammonia Injection System Failure Distribution*

«
e

9
.2
O
Q_

©

m
c

0
£

1
1

ft

S

c
£
£
o

1t87

1988

1089

NH,

NOx

1990

Figure 8. Study Site Relative Outlet NOx and NH3 History

5B-35


-------
Train A

38.0
26

23.9
27.3

•

•

38.9
26.2

27.4

26.5

31.8
29.2

:

30.2
25.8

25.1
25.9

39.9
26,8

-

23,1
27

25.1
27

29.2
27

-

2?

•

•

*

-

25

25

*

Train B

20,9

31,5

28.6

33.0

33,9

32.1

30

29.5

23.1

35,5

25.7

23.8

31

29.7

28,4

27.9

29,7

38.1

33.9

42.0

29

28.1

27.7

26.9

27.6

21.0

32,8

15.8

26.1

25,6

25,3

24.8

30.8

38,1

33,8

30.8

23.3

24,8

24,8

25.3

28.7

37.2

27.5

36.5

26,1

26.9

26,9

26.9'

Average 30.2

Average 26.5

ft/sec

ppm

Figure 9. SCR Inlet Velocity and NOx Concentration Maps

500

100

¦50

C

0

1

110

i

z

1	2 5 10	50 100	500

SO, Concentration, ppm

Figure 10. Ammonia Salt Formation
as a Function of Temperature and NH3and S03 Concentration (2)

5B-36


-------
NOx REDUCTION AT THE ARGUS PLANT
USING THE KOiOUTB PROCESS

Joseph R. Comparato
Nalco Fuel Tech

Roland A. Buchs
North American Chemical Corporation

Dr. D. S, Arnold
L. Keith Bailey
Karr-McGee Corporation

5B-37


-------

-------
ABSTRACT

Urea injection using the NOxOUT Process was. demonstrated at the Kerr-McGee Argus
No. 26 unit. The earlier installation of burner modifications had reduced NOx
emissions from 330 ppm to about' 225 ppm. The NOxOUT Process further reduced NOx
emissions to below a target level of 165 ppm.

Testing of the hybrid NOx control system included furnace characterization,
injection optimization, and a 48-hour demonstration test. Process performance
was analyzed from extensive data logged with a computer data acquisition system.
A computer model of the furnace flow dynamics provided information for selecting
injector locations and performance settings. Optimization reduced the ammonia
slip to 2 ppm. CO slip was limited to 6 ppm.

Subsequent long-term evaluation examined the impact on plant operation. The air
heater was inspected for possible accumulation of ammonium .bieulfate and was.
found free of such deposit build-up. The storage, pumping, and injection
equipment operated reliably. Chemical consumption has been consistently within
expected projections. The successful NOxOUT demonstration is being upgraded to
a permanent installation.

5B-39


-------
The NOxOUT Process for controlling oxides of nitrogen (NOx) emissions was
installed on the Kerr-McGee Argus No. 26 coal-fired boiler in June 1989.
Parametric testing was conducted in August 1989 to characterize and optimize the
process application. The matrix testing concluded with a 48-hour continuous
demonstration run. The achievement of 30% reduction in NOx emissions below the
level of reduction previously accomplished with low NOx combustion system'
modifications was demonstrated. The combined result of NOxOUT and combustion
system modifications was an overall NOx reduction of more than 50%.

The process optimization during start-up of the NOxOUT system concentrated on
achieving the required NOx reduction while controlling ammonia slip to below 5
ppm. The purpose of this objective was to prevent potential fouling of the
regenerative air preheater surfaces. The limit was chosen to avoid any
significant formation of ammonium bisulfate from the combination of ammonia with
fuel sulfur products. The demonstration teat showed that ammonia slip was held
to 2 ppm. It was also important to prevent any significant increase in carbon
monoxide emissions. A target of less than 10 ppm, CO increase was achieved with
a CO slip of 6 ppm.

Following the formal testing, the program continued with Phase II, a four-month
period, that was extended to seven months, to observe the long-term effects of
operating the NOxOUT system. The process equipment performed reliably.
Inspections of the unit conducted during and after the Phase'II operation
verified successful control of potential air preheater deposits.

NOxOUT Process Technology

In the NOxOUT process, the products of combustion are treated with an aqueous
solution of chemicals. .NOxOUT A, sometimes enhanced with other chemicals,
combines with NOx in reduction reactions to yield molecular nitrogen, water, and
carbon dioxide. The technology initially emerged from research on the use of
urea1 to reduce nitrogen oxides conducted in 1976 by the Electric Power Research
Institute (EPHI). EPRI obtained the first patent on the fundamental urea
process in 1980.^ The overall chemical reaction for reducing NOx with urea is:

NH2CONH2 + 2NO +l/202 	> 2N2 + COj + 2HgO

Nalco Fuel Tech is the exclusive licensing agent for the EPRI technology. Nalco
Fuel Tech has developed' the technology with added know-how and patented
advancements. NOxOUT is the tradename for this post-combustion technology for
NOx reduction.

The NOxOUT technology comprises methods and experience for effectively treating
a wide range of applications. Combustion laboratory testing provides data for
proprietary chemical formulations that extend effectiveness beyond the
conditions limiting the performance of the basic urea process. The NOxOUT A

5B-40


-------
formulation insures consistent product quality control and includes additives
which prevent problems such as injector fouling.

Performance design tools increase confidence in applying NQxOUT to new
applications. Process performance ia analyzed using Naleo Fuel Tech's chemical
kinetics computer model (CKM),3 Process conditions are evaluated using
computational fluid dynamics (CFD) modeling techniques.4 The CFD modeling also
enables the simulation of injector design configurations to evaluate chemical
dispersion effectiveness. Used "together, the CKM and CFD'models provide a sound
basis for predicting expected performance.

Research in injector development, including laboratory analysis using laser
equipment for measuring droplet size and velocity, provides a database for
selecting injection equipment for a specific application. Process equipment
designs incorporate experience from both demonstration and commercial projects.

The NQxOUT technology was fully applied in treating the Kerr-McGee Argus No. 26
unit. Successful experience with a similar unit in Germany, a 75-MW brown coal
fired power plant operated by Rheinisch-Westfalisches Elektri2itatswerk A. G.
(RWE), provided a basis for confidence.5 However, there are'often significant
differences between similar coal fired units. Thus, extensive modeling and data
analysis were conducted in support of the testing.

Argus No. 26 Boiler Description

The Kerr-McGee Argus No. 26 unit (figure 1) is a tangentially fired, pulverized
coal, VU 40 type, ABB Combustion Engineering boiler. Western bituminous coal is
burned in the furnace with three coal elevations, each supplied ,by a bowl mill
pulverizer. Table I is a typical fuel analysis. The unit has a normal
operating steam output of 710,000 lb/hr (322,560 kg/hr) at 950°F (510°C).

TABLE I

COAL ANALYSIS

Type

Ultimate Analysis

•Utah Bituminous

As Rec'd

Dry Basis

%Carbon

^Hydrogen

%Nitrogen

%Chlorine

%Sulfur

%Oxygen

%Ash

%Moisture
HHV, Btu/lb

70.52
4.91
1.37
<0.1

0.47
10.32
8.66
3.75
12,592

73.27
5.10
1.42
<0.1

0.49
10. 72
9.00
N/A
13,083

5B-41


-------
Flue gas heat recovery is accomplished with an economizer followed by a ,
horizontal shaft regenerative air preheater. After the air preheater, the
combustion products pass through an electrostatic precipitator (ESP) for dust
control, and then through a sodium-based wet S02 scrubber. The flue gas is
exhausted without reheat at 120°F from the stack.

In Kay 1989, the firing system was modified to reduce NOx emissions. As
originally built, the unit had close coupled over-fire air (COFA) for NOx
control. In this configuration, baseline NOx levels were about 360 ppm {dry,
corrected to 3% O2) when firing 60% coal and 40% petroleum coke (330 ppm when
firing 100% coal}. The modifications included LNCFS (Low NOx Concentric Firing
System) nozzles, flame attachment nozzles, and the addition of SOFA (Separated
Overfire Air) ports,6 NOx emissions were reduced to a typical value of under
225 ppm under normal operating conditions.

Operation with varied overfire air configurations had a strong effect on the
baseline conditions for NOxOUT treatment. Figure 2 shows the NOx emissions with
different SOFA damper positions. The numbers identifying the SOFA conditions
correspond to the upper/middle/lower damper percent opening.

As overfire-air dampers were opened, the combustion air was redirected from the
burner zone to higher elevations. While the total oxygen available for
combustion in the furnace was relatively constant, less oxygen was available in
the burner zone as overfire dampers were opened. Fuel burning was effectively
staged. Fuel-rich conditions were created in the burner zone to promote
reduction reactions that destroy some of the NOx formed from fuel nitrogen,7
Combustion was distributed over a longer portion of the furnace. ' Peak
temperatures were lowered to avoid the thermal formation of NOx from nitrogen in
the combustion air.

Temperatures in the regions suitable for NOxOUT injection were affected by the
degree of staging. A reduction in peak furnace temperatures to control NOx
also reduced the radiant heat transfer to the furnace walls. Consequently, the
flue gas temperature in the upper portion of the furnace increased as NOx is
reduced with deeper degrees of staging. Some data indicated an increase in
temperatures in the upper furnace (elevation 106') from about 1800°F (982°C)
before modifications, to a maximum of 2200°F (1204°C) with the SOFA dampers
fully open.

The 40/100/100' SOFA configuration was considered- the typical operating mode for
the Argus #26 unit. As evident in figure 2, the benefits of additional NOx
reduction began to diminish with deeper staging. Figure 3 is a plot of CO
emissions as a function of NOx level. CO emissions tended to increase as NOx
level decreased. This resulted in part from- increasing difficulty in tuning the
burner air flows as more air was redirected to the SOFA ports.

5B-42


-------
The 40/100/100 SOFA staging was chosen as the baee condition for applying the
NOxOUT process. In July 1909, the temperature profile in the upper furnace with
this SOFA configuration was measured. An average temperature of 2020°F (1104°C)
and a peak o'f 2110°F (1154°C) in the center of the plane were observed. The
temperature was of concern since the critical level of NOx increases with
increasing temperature.

Chemical kinetics modeling and data from laboratory and field tests have shown
that a "critical NOx" level exists aa a function of temperature (figure 4).^
Critical NOx is also strongly affected by the oxygen concentration and the
presence of reducing species such as carbon monoxide. CO concentrations were
also sampled during the temperature traverse and found to be less than 200 ppm
at the furnace outlet plane. A benefit of the high temperatures is that the
reactions are rapid, requiring less residence time than at lower temperatures.
The tendency for residual formation of ammonia and CO byproducts is also
decreased.

A CFD model (figure 5) of the Argus #26 unit was prepared to provide guidance
for the testing. The upward spiraling flow typical of a T-fired furnace was
predicted. The model provided simulations of the .injection trajectories and
chemical dispersion. In applying the results, care was taken to identify
guidelines for preventing droplet impingement on tube surfaces.

The NOxOUT Installation

Injection ports were installed at two levels. The upper level, at elevation
106", provided a region where fine droplets could be promptly evaporated in the
lowest available gas temperature conditions. The lower level, at elevation 90',
allowed the injection of larger droplets to enable greater penetration into the
gas stream, but into higher temperatures. The injectors were designed with
interchangeable atomizing tips to facilitate testing different spray pattern
options.

Skid-mounted pumping equipment was installed on site. Chemical injection pumps
metered the reagents into a mixing header. Dilution water also entered the
mixing header. A rotary positive displacement pump mixed the reagents and water
by recirculation through the header and pressurized the mixture for supply to
the injectors. Air was used as the atomizing medium for the injectors. A
pressure-settable air regulator controlled the atomizing medium conditions.

Figure 6 is a simplified schematic of the process system. An analog controller
provided output to the electronic stroke controlled chemical injection pumps.
It also provided PID loop control of the pressure control valve to maintain a
settable constant mixture discharge pressure.

56-43


-------
Testing Results

Tes.t aeries were identified in terms of eighteen teat days. The- teat objectives
weres

The demonstration utilized an on-line data logging system to provide continuous
monitoring of the boiler and NOxOUT system operation. Display screens of the
current operating conditions facilitated assessing test progress and decision
making for proceeding with steps in the test program. Analog signals from the
boiler control room and instrumentation from the chemical injection equipment
were transmitted to an analog-to-digital.converter. The digital values were
read by an 80286 based micro-computer using THE FIX software by Intellution,

The plant's continuous stack emissions monitor provided NO* and CO data,
corrected to a dry basis at 3% 02, Signals from the control room provided data
on the boiler operating conditions. Calculations were performed with THE FIX
software to compute NOx on a mass flow basis, values for NOxOUT chemical flow
rates were taken from analog outputs from the pumping skid controller.

The main parameter for determining the NOxOUT treatment rate is normalized
stoichiometric ratio (NSR). .As can be seen from the basic chemical reaction,
one mole of urea combines with two moles of NOx under perfect conditions. NSR
is the ratio of the actual molar flow of urea to the molar flow required for
stoichiometry, or perfect reaction. NSR values were computed from the chemical
flow rates and NOx massflows identified as baseline'conditions for the various
test. runs.

Ammonia analysis utilized a manual batch extractive method. The very low levels
of ammonia measurements required a technique with high sensitivity. Filtered
flue gas samples were drawn through heated probes from ports in the economizer
outlet. During the 48-hour demonstration run, 12 point samples, on a 4 port by
3 paint insertion grid, were collected. Ammonia was captured in an impinger
train containing dilute sulfuric acid. The impinger samples were cooled to a
controlled temperature, then made alkaline to release the ammonia for
measurement with an ion specific electrochemical cell.

Figure 7 is a plot of the NOx emissions as a function of NSR for various SOFA
settings observed during the boiler characterization tests, series 1-4.

External mix injectors producing 100 micron volume mean diameter droplets were
used in the seven ports available at the 106* level. Over 50% NOx reduction was

#	1-4

#	5-9
#10-13
#14-16
#17-18

Test davs

Test Series Type

Boiler' SOFA Characterization

Upper Level Injection
Lower Level Injection

48-Hour Demonstration Test
Miscellaneous Testing

Inc.

5B-44


-------
achieved with an NSR of 2.2 in the 0/0/100 SOFA condition and a high NOx
baseline. However, lower NOx emissions were obtained using less chemical with
deeper staging.

The data at 0/0/100 SOFA suggested, as was expected, that the chemical was not
fully dispersed in the flue gas. It should be noted that the chemical flow for
an NSR of 2,2 at a baseline of 288 ppm is 3.8 times the flow for a. NSR of 1.0 at
a baseline of 166 ppm. The curve for the 0/0/100 condition suggests that the
performance was limited by the ability to treat all of the gas. The CFD model
indicated that with injection at the- 106' level, a large portion of the gas
would pass below the injection plane.

It was noticed that the stack opacity visibly increased during injection and
persisted for more than an hour after injection was discontinued. A "plume",
appeared that was attached to the stack outlet as opposed to the detached water
vapor plume normal during the cooler times of day. Opacity readings at the ISP
outlet did not increase. It was assumed that the plume was caused by ammonia
slip combining with trace amounts of chloride and/or sulfate in the stack gas.
Traces of chloride and sulfate were present in the stack gas from entrainmant of
brackish water from the wet scrubber. Many of the decisions in subsequent teste
were aimed at minimizing ammonia emissions.

The plume was minimized as ammonia slip was reduced in the later injection
optimization series but at the expense of some NOx reduction. An S03 injection
system was installed after the demonstration test series was completed. This -
was previously planned to reduce particulate emissions. After installation of
the ESP injection system, the plume was eliminated.

Series 5-9 and 10-13 tested injection at the upper (106*) and lower (90*)
levels. It was found that roughly equal NOx reduction performance could be
achieved at either level. Large droplet sprays (1000 micron) with high total
liquid flows were effective at the lower, hotter level. The large droplets had
longer lifetimes and evaporated in the cooler upper furnace.

The NOx reduction results are shown in figure 8. Somewhat better performance
was achieved with injection at the lower level. This is in part the result of -
improved dispersion of the chemical in the flue gases and a slight quenching
effect from the increased liquid flow. High liquid flows were not desirable at
the upper level since complete evaporation could not be assured prior.to
reaching tube surfaces, A trend of increased NOx reduction with increased
liquid flow can be seen in figure 9. Injection was optimized by adjusting
atomizing and liquid pressures and using angled internal mix tips with varied
orientation. Figure 10 shows the progress of NOx reduction as different
injection arrangements were tested.

Ammonia slip control was the principal guide in selecting injector arrangements.

5B-45


-------
The results, are seen in 'figure 11. In general, injectors were operated to avoid
the release of chemical in regions too close to the inlet to the connective
pass. Chemical released where gas temperatures are rapidly quenched would form
ammonia. Thus, the optimization achieved a balance between excessively high and
low temperature zones. Ammonia slip values of 2 ppm were measured in the two
12-point traverses conducted during the 48-hour demonstration run.

CO slip was controlled to 6 ppm during the demonstration run. Figure 12 is a
plot of CO emissions versus NOx emissions for all tests. CO emissions increased
from the 11 ppm baseline shown in figure 3 to 17 ppm. As with the baseline
data, CO emissions tended to increase ae NOx emissions were decreased.

The scatter in the NOx reduction-data reflect the influence of a number of
factors in operating a coal-fired furnace. Routine adjustments in the burner
dampers would result in changes in baseline NOx. Furnace cleanliness influenced
flue gas temperatures. Figure 13 showB a trend of slightly decreasing NOx
reduction with time after cleaning with furnace wall blowers during the 48 hour
demonstration run.

Phase II operation showed that consistent performance can be achieved. The air
preheater was inspected in January, 1990, and Hay, 1990, and found to be free of
deposits that could be caused by the NOxOUT system. In June, 1990, changes were
made to the boiler aimed at reducing carbon loss. However, the NOxOUT
application was not adjusted for the new conditions. Ammonium bisulfate
deposits accumulated apparently as the result of an undetected increase in
ammonia slip resulting from changes in the furnace conditions. In October,
1990, the injector operating conditions were adjusted to reduce droplet size and
in November, 1990, changes were made in the operation of the air heater
sootblowers. Subsequent operations have been-too short to determine whether the
problem has been fully resolved.

Demonstration Results

NOx emissions during the 48-hour demonstration, using an NSR of 1.1, were
reduced 31% below the test baseline. Ammonia and CO slip were controlled to 2
and 6 ppm, respectively. The equipment operated reliably with minimal need for
operator attention. Phase II extended operation confirmed that the system is an
effective means for reducing NOx emissions from the large coal-fired boiler.

As an outcome of the demonstration, the NOxOUT system for Argua unit #26 is
being upgraded to a permanent installation and integrated with the plant control
system. The process will also be installed on the identical unit #25.

5B-46


-------
Muzio, L. J., and Arand, J. K. "Homogeneous Gas Phase Decomposition of
Oxides Of Nitrogen", EPRI Report No. FP-253, 1976. -

Arand, J. K. , Muzio, L. J., Sotter, J. G., U. S. Patent 4,206,386, June •
17, 1980.	'

Sun, W. H., and Hofmann, J. E., "Post Combustion NOx Reduction with Urea;
Theory and Practice", presented at the Seventh Annual International
Pittsburgh Coal Conference, Pittsburgh, PA, September 10-14, 1990.

Michels, W. F., Gnaedig, G., and Comparato, J. R., "The Application of
Computational Fluid Dynamics in the NOxOOT Process for reducing NOx
Emissions from Stationary Combustion Sources", presented at the'AFRC
Committee Conference, San Francisco, Cft, October 10-12, 1990.

Hofmann, J. E., von Bergrnann, 3., Bokenbrink, D., Hein, K., "NOx Control
in a Brown Coal-Fired Utility Boiler", presented at the EPRI/EPA Symposium
on Stationary Combustion NOx Control, March, 1989.

Buchs, R. A., Bailey, L. K., Dallen, J. V., Hellewell, T. D., Smith, C.
W., "Results from a Commercial Installation of Low NOx Concentric Firing
System (LNCFS)", presented at the 1990 International Joint Power
Generation Conference and Exhibition, October 21-25, 1990, Boston, MA.

Morgan, K. 1., "Effect of coal Quality on the Performance of Low-NOx
Burners", presented at the British Flame Days Conference, London,

September 1988.

5B-47


-------
CCNVECTWE SUPERHEATER
PLATEN SUPERHEATER

¦ EL 106?

NOxQU
INJECTOR PORT LEVELS

¦ EL 9C?

COAL PULVERIZER

ECONOMIZER ,

EMISSIONS SAMPLING

TO

mPHEHEATER

FROM '

AIR PHEHEATER

ARGUS #26 COAL FIRED BOILER

FIGURE 1

400

300

a
&

200

100

COFA

NOx Baseline

0/0/100

0c'50/100 0/100/100 40/100/100 100/100/100

Staging Condition (SOFA Damper positions)

FIGURE 2

5B-48


-------
CO EMISSIONS AT BASELINE NOx LEVELS

60

50

40

30

20

10

A

d

3

CD

"Q
'x
o
c
o

c
o

¦Q
CG

O

160

160	200	220

NOx Emissions (ppm)

500

400

Critical NOx Concentration

3% Excsss Oxygen
NOxOUT Kinetic Model

5' 300
0_

z

9 200

100

TOO

N0xi=500 PPM

N0xi=200 PPM

NOxi= 100 PPM

800	900	1,000	1,100	1.200

Temperature (degrees C)

1,300	1,400

FIGURE 4

5B-49


-------
CFD MODEL OF NOxOUT INJECTION

PLAN VIEW AT ELEVATION 100'

U1

03

I

CJl

o

(.'oncentratinn

0.00E+00
1.34E-04
2.69E-04
4.03E-04
5.38E-04
6.72E-04
8.07E-04
9.41E-04
1.08E-03
1.21E-03
1.34E-03
1.48E-03
1.61E-Q3
1.75E-03

2.5000E+01 m/s.

FIGURE 5


-------
NOxOUT INJECTION SYSTEM

FIGURE 6

NOx Emissions

AT STAGING CONDITIONS

NORMALIZED STOICHIOMETRIC RATIO (NSR)

0W100 ft'100/100 4a'100/100 100/100/100
~	A	O	*	FIGURE 7

5B-51


-------
NOx Reduction vs NSR

240

220

JZ

Q.

s

w
c

o

"(0

"E

LU
X

O

200

180

160

140

120

100

0,5	1	1.5

Normalized Stoichiometric Ratio (NSR)

level 106 level 90 demo
o	a	o

FIGURE 8

EFFECT OF TOTAL FLOW

O

I—

o

D
O
LU

DC

*
o

z

NSR RANGE 098-1.19

C LOWER LEVEL INJ,

TOTAL FLOW (QPH)

4 UPPER LEVEL INJ.

FIGURE 9

5B-52


-------
NOx REDUCTION vs INJECTOR ARRANGEMENT

NSR RANGE 0,98-1.19

4-

+

0	p

O
~

Q

D °	n

~	Q	„

	1 	1 	1 .... ..j	j	|	|	|	¦¦ r ¦		j 	]	 ^	i ¦

6	8	1C	12	14	16

TEST PAY

D LOWER LEVEL iNJ	4- UPPER LEVEL INJ.

FIGURE 10

AMMONIA EMISSIONS

SAMPLED AT ECONOMIZED OUTLET

+

~

C



~

~

Q D

_1	_!	f	!	!	(	!	J	,	,	,	,	^	

t	6	B	10	12	14	16

TEST NUMBER

+ UPPER LEVEL INJ. ~ LOWER LEVEL INJ.	FIGURE 11

5B-53


-------
CO EMISSIONS WITH NOx REDUCTION

NSR RANGE 0.98-1.16

£3 LOWER LEVEL !NJ.

NOx EMISSIONS (PPM)
+ UPPER LEVEL INJ

O 48 HR DEMO1

FIGURE 12

EFFECT OF FURNACE CLEANING ON REDUCTION

optimized injection duhing 4tm demo

HOURS SINCE LAST SOOTBLOW

FIGURE 13

5B-54


-------
REBURNING APPLIED TO COGENERATION NOx CONTROL

C. Castaldini
C. B. Moyer
Acurex Corporation
Mountain View, California

R. A. Brown
Electric Power Research Institute
Palo Alto, California

J. A. Nicholson
ABB Combustion Engineering
Windsor, Connecticut

5B-55


-------

-------
ABSTRACT

New cogeneration systems are increasingly regulated to stringent NO, levels based on control
technology precedents established in California. NO, compliance costs can be a disincentive
to cogeneration markets. This project evaluated rebuming to achieve low NO, levels at lower
costs than postcombustion catalytic reduction. Subscale tests were run at the 100,000 Btu/hr
scale to simulate combustion conditions with both rich-burn and lean-burn reciprocating-engine-
based cogeneration and lean-burn turbine-based cogeneration. Results showed NO, reductions
in the range of 50 to 70 percent for rich-burn conditions with a reburn-to-engine fuel ratio of 0.2
to 0.3. Reductions with lean-burn engine conditions were nominal unless the reburn zone was
operated at a locally substoichiometric condition. For rich-burn conditions, introduction of a
metal catalyst into the reburn zone increased the NO, reductions to greater than 90 percent
by presumably accelerating the NO, reduction reactions under fuel-rich conditions.

Full-scale rich-burn reburn tests were run with a 150-kW Caterpillar engine feeding flue gas to
a new design reburn section. Over the range tested, the full-scale NO, reduction results
corroborated the subscale results. Reburn burner stability problems prevented going to
stoichiometric ratios below 0.98, however, so maximum NO, reductions were 50 percent
without the catalyst and 75 percent with the catalyst.

Pilot-scale lean-burn repower tests were run with the boiler fired at a high fuel fraction to
produce a locally substoichiometric condition. Air staging in the boiler was also used to further
improve NO, reductions. NO, reductions of 50 percent were achieved with no air staging at
boiler-to-engine fuel ratios of 1.5 and above. With air staging In the boiler, NO, reductions of
70 percent were experienced. In all configurations, rebuming was very effective in destroying
90 percent or more of the CO emitted by the prime mover.

5B-57


-------
INTRODUCTION

The cogeneration of electricity and process steam has grown at a steady rate, stimulated by
favorable economics of on-site generation and by the Public Utilities Regulatory Policy Act
(PURPA), New cogeneration is expected to increase annual gas consumption by 815 billion
cubic feet per year over the next 10 years (1,2). Turbine-powered cogeneration or repower
configurations wiil contribute about 585 BCF of this growth, or 70 percent. Rich-burn or lean-
burn reciprocating engine-powered systems will contribute about 230 BCF, or 30 percent.

The cost of NO, controls for new cogeneration systems is increasingly taking on a larger
fraction of the total system cost. With increasingly stringent control technologies required
during permitting,, the incremental costs of NO, compliance may be decisive in' making
cogeneration noncompetitive. This trend is accelerating as a result of two recent regulatory
developments: the top down BACT policy, and Title I of the 1990 Clean Air Act Amendments.
The top down BACT procedure causes permit applicants to consider implementing the most
stringent NO, control technology adopted elsewhere for similar equipment. This is causing
considerable downward pressure nationwide on BACT levels set during permitting because of
the California cogeneration precedent. In several districts in California, selective catalytic
reduction-is required as BACT for turbines and nonselective catalytic reduction is required for
rich-burn reciprocating engines. Title I of the 1990 Clean Air Act Amendments promotes NO,
controls for attainment of ozone air quality in areas designated as in extreme, severe, or
serious nonattainment. This is increasing both the number of sources under control as well
as the severity of new or retrofit control levels.

In many cases in California and elsewhere, consideration of catalytic postcombustion controls
has diminished the return on investment for the cogeneration project to the point where other
energy options are preferred. The present project was initiated by the Gas Research Institute
to evaluate reburning as a means to achieve improved NO, reductions at lower costs than
postcombustion controls.

A market applications study at the outset of the project indicated that two types of
engine/boiler configurations, shown in Figure 1, could gain a significant market share with
reburning. The conventional cogeneration system, shown at the bottom normally feeds the
prime mover exhaust directly to an unfired heat recovery steam generator. For reburn NO,
control, the fuel staging is most easily done with installation of a reburner section in the engine
exhaust gas ducting to the HRSG. This configuration, shown at the top is most readily
packaged for new units. Repowering is a cogeneration alternative for existing boilers that can
be retrofitted with a reciprocating engine or turbine.

For both reburn configurations, developmental testing is needed to identify the preferred reburn
stoichiometry, temperatures, engine-to-reburn fuel ratio, and primary/reburn mixing geometry.
In the present program, testing was done in three stages to address these issues:

•	Subscale 100,OCX) Btu/hr parametric configurational tests for rich-burn,
moderate 0?, and lean-burn cogeneration conditions.

•	Full-scale 150-MW rich-burn reciprocating engine cogeneration configuration
tests

5B-58


-------
• Pilot-scale one million Btu/hr repowered boiler burner configuration testing

A detailed discussion of these tests, as well as associated market applications studies and
economic comparisons, is contained in References 1 and 2.

SUBSCALE TESTS

The subscale facility used for parametric reburn configurational testing is shown in Figure 2.
The test combustor was assembled in two main sections: a 100,000 Btu/hr down-fired engine
exhaust simulator; and a reburner and burnout section. Doping with nitric oxide and CO was
done between the two sections to achieve NO levels representative of engines or turbines.
Independent regulation of natural gas and combustion air to the reburner and burnout air
downstream of the reburner allowed parametric variation of the reburner stoichiometry, SR2,
and the postreburn stoichiometry, SR3. Combustion air preheat capability was added to study
temperature effects on the reduction reactions.

Initially, a hardware screening series of tests was done to identify the sensitivity of NO,
reduction to burner geometry, and to iterate to the preferred burner design. These tests
showed that NO, reduction was sensitive to the method of reburn mixing with the engine
exhaust. For cases where the mixing was enhanced to promote NO, reduction, the percent
reduction was sensitive to the inlet level of NO,.

Based on the initial screening tests, the reburner design shown in Figure 3 was selected. Early
tests showed the benefit of the bluff body over the flame with a tight spacing to promote
mixing. The forced mixing of the reburn flame with the primary flue gas stream promoted NO,
reduction by exposing the carryover NO, from the engine simulator to the fuel-rich reactants.
With this burner, optimum performance was experienced at a reburner stoichiometric ratio of
SR? of about 0.8. Figure 4 shows the improvement in NO, reduction with increasing fuel
fraction as the quantity of flue gas generated in the burner becomes a larger fraction of the
engine exhaust volume.

The rich-burn tests showed a significant effect of inlet NO, concentration on NO, reduction
efficiency. Figure 5 shows that for the rich-bum engine with a reburn stoichiometric ratio of 0.8
and a fuel fraction of 20 percent, the reburn efficiency decreases as carryover NO, increases.
This may indicate an increasing depletion of radical species in the fuel-rich region.

Increasing temperatures in the reheat zone is apparently effective in accelerating the reburn
reactions within the available residence time. Figure 6 shows that addition of preheated air to
the reburner improves the reduction efficiency significantly for fuel fractions of 20 and
37,5 percent. There is also a beneficial reburn effect in the downstream zone where burnout
air is injected when the reburn region is operated at an overall substoichiometric condition.
Figure 7 shows an improvement in NO, reduction of over 10 percent with a rich-bum exhaust
when reburn air is added to complete combustion.

As would be expected, the reburner acts as an afterburner for CO destruction. Figure 8 shows
that with sufficient heat addition to the reburn section, the carryover CO can be effectively
destroyed.

5B-59


-------
With Jean-burn engine simulation, the NO, reductions were less effective because a
substoichiometric condition was not achieved for the fuel fractions tested. The lean-burn tests
showed that a much richer reburn stoichiometry was most effective compared to the SR2 = 0,8
optimum observed with rich-burn conditions. Figure 9 shows the improvement with richer
reburn conditions. The best reductions achieved were around 35 percent. These moderate
reductions would probably not justify use of the reburn hardware. The effect of fuel ratio was
riot significant over the range of 30 to 37.5 percent tested. For the lean-burn conditions of
Figure 9, a fuel ratio of about 100 percent would be required to achieve an overall fuel-rich
reburn zone.

Exploratory tests made during,the initial parametric study showed a dramatic increase in NO,
reduction when metal oxide catalysts were introduced into the reburn chamber. The potential
benefits of the concept of catalytic enhancement of NO, reduction was sufficiently strong that
the burner was modified for catalyst inserts, as shown in Figure 10. Figure 11 shows the
reduction resulting from use of a nickel oxide ceramic catalyst added at the end of the reburn
mixing zone. For an overall reburn zone stoichiometry of 0.95 or lower, the reduction of the
carryover NO, from the rich-burn engine was essentially complete. Figure 12 shows the effects
of several catalyst configurations that give variations in effective surface area. Although there
is considerable scatter, the data show that higher effective surface area strongly improves
reduction.

FULL-SCALE RICH-BURN TESTS

Based on the favorable subscale test results, a full-scale rich-burn cogeneration configuration
was tested at the Air and Energy Engineering Research Laboratory of the Environmental
Protection Agency in Research Triangle Park, North Carolina, Figure 13 shows the reburn
reaction chamber fabricated for the testing and the overall laboratory configuration. The
noncatalytic baseline and the catalytic testing agreed fairly well with the subscale tests.
Figures 14 and 15 show the NOx reduction without and with the catalyst section. Due to flame
stability problems experienced with the reburner under fuel-rich conditions, it was riot possible
to lest below stoichiometric ratios of about 0.99. Since NO, reduction is very sensitive to
stoichiometric ratio at these conditions, this was a constraining factor. The trends indicate that
if the stability problem was resolved, considerably higher reductions would be experienced.
Apart from the burner issue, the reburn reactor section performed well and showed promise
for sustained commercial usage.

LEAN-BURN REPOWER TESTS

The cogeneration tests discussed above centered on reburn-to-primary-fuel ratios of around
0.2 to 0.375, which would be characteristic of a duct reburn section upstream of a HRSG. For
repowering of existing boilers, the fuel ratios are much higher since the prime mover exhaust
is used as combustion air for the boiler and sufficient fuel is added to nearly deplete oxygen.
To simulate these repower conditions, the test facility shown in Figure 16 was tested. The
prime mover simulator had a firing capacity of one million Btu/hr. The exhaust from tne
simulator was directed to the primary boiler test burner. The firing rate of the prime mover
simulator together with heat exchangers and NO or CO doping were adjusted to obtain a
reasonable simulation of lean-burn turbine repowering temperatures and flue gas composition.
The boiler had additional provision for stage air above the test burner.

5B-60


-------
Three different boiler repower burners were tested to study effects of mixing NQ„-bearing
combustion air with the primary boiler flame. Despite significant differences in mixing patterns,
the three burners produced comparable NO, emissions reductions. Figure 17 shows NO,
reduction results with and without stage air. The NO, reduction improved with boiler-to-engine
fuel ratio, and reductions in excess of 70 percent were experienced at representative fuel ratios
with boiier staging. Stability tests showed that turbine exhaust oxygen levels of 14 percent or
greater were needed to maintain a stable boiier flame. Repowering is effective in destroying
any carryover CO, as shown in Figure 18. The lower efficiency at low CO levels is due to
residual boiler CO concentrations.

CONCLUSIONS

The following conclusions were reached in this study;

•	Reburning, without catalyst assist, reduced NO, by 50 percent at a fuel fraction
of about 30 percent. With this performance the process presents little
economic attractiveness.

•	Catalyst, assist reburn was shown to achieve 70 to 99 percent'NO, destruction.

This performance is required for reburn to become a viable and competitive
technology for gas-fired engine NO, control.

•	Continued research is needed to evaluate catalyst and improved mixing on
NO, reduction potential and applications.

ACKNOWLEDGEMENTS

This project was sponsored by the Gas Research Institute. Dr. F. R. Kurzynske was the Gas
Research institute Project Manager. The Coen Company assisted in selecting model burner
designs for testing. The Todd Burner Division of Fuel Tech, Inc., contributed the reburner
reactor used in the full-scale testing. The U.S. Environmental Protection Agency made available
the host site for the full-scale testing,

REFERENCES

1.	Brown, R. A., Lips, N., and Kuby, W. C., 'Application of Reburn Techniques for
NO, Reduction to Cogeneration Prime Movers; Volume I, Rich-Burn
Applications," GRI 88/0341, Gas Research Institute, Chicago, IL, March 19B9.

2.	Brown, R. W., Moyer, C., Nicholson, J., and Torbov, S., "Application of Reburn
Techniques for NO, Reduction and Cogeneration Prime Movers: Volume II,
Lean-Burn Engine Applications," GRI 90/125, Gas Research Institute, Chicago,
IL, March 1991,

5B-61


-------
Figure 1. Rebuming Applied to Cogeneration or Repowering with Gas-Fired Prime Movers

56-62


-------
fuel fraction (percent)

Figure 3. Subscale Reburrter

Figure 4. Effect of Fuel Fraction

5B-63


-------
Figure 5. Effect of Input NO Concentration

-





&

1 • 0.375 with preheat

	



\ -

A

1 ¦ 0 375 no preheat





~~ " %

V

©

1 • 0.20 wMi preheat







•

1 ¦ 0 20 no preheat

-



o



Put gai condSlons

	





NO. - 1J00 ppm



A





O, - 05 percent

—







T, . 1.100'F

-







- 100,000 Btu/hi

- ,



















-

	1	

i 1 i i

1

( (

0.6 0.65 0.? 0.75 0.8 0.85

SRj returner slelcMopetry

0.9

0.95

1.0

Figure 6. Effect of Reburner Air Preheat

5B-64


-------


SR, - 0.8





Rue sat conditions



_

NO, - 1,500 pptn





0, « 0 J peretm

With bumoui •»



T, - 1.000'F





FR^„ . 100.000 Btu/hr

-

0/"







0 Wllhoul burnout





y' «lr



0 ar









-

/ *

t i I I i i

1 1

0

4 10 15 20 25 30

35 40



fuel fraction (percent)





Figure 7. Effect of Burnout Air



s«| • 0.*

Sluff bumf
.? fit. l
-------


€0



S5



50

Z-

ih





i

*C

|

tt

w



s

3C

O



i



t

?0





c



m

V

55

*



o.

10



S



c









flu" tmif tw-*.

-







t m, l«K|









• |M»»(

-







t-iti i*. • tiwp*









bio ¦ l(l 1»,

-







»I«r ftt









ior • loe m-

-







C| * II W>t«M









*| « I.IBC'f

-







fi*,,, »tac.eee its"-



1ft. fl

D, VA ptrctM fyf"

frlCtiO*











		 6 f - 0.3D

_







"""" "—} 1 • a. 375

-

	I	 s	

I I

I

I I 1

0 ? O.J

0.4

0.5 0 6 M 0.8. O.t IP

Figure 9. Effect of Reburner Stoichiometric Ratio for Lean-Burn Conditions

MFTAl
CATALYST

Figure 10. Burner Configuration with Catalyst Insert

5B-66


-------
z
o
pr
o
=>
a
u
ct

«

100

80

60

40

20

-tr

O 20* fuel rmc

• 2DX FUEL FRAC
£15* FUEL FHAC ¦
A 1 OX FUEL FRAC





*°o
•

A A

A

0.80

0.85

-+-

0.90
SR3

0.95

1.00

Figure 11. NO, Reduction with Catalyst Enhancement for a Space Velocity of 7,500 per hour

z
o
p

o

£i

i

K

15.0	25.0	35.0

SPACE VELOCITY (l/HR X ID''}

45.0

Figure 12. Effect of Space Velocity on NO„ Reduction with Catalyst Enhancement

5B-67


-------
W*TtH

BYPASS ENQIHE E»h»UST

LAYOUT FOR REBURN SYSTEM

Figure 13. Full-Scale System with 150-kW Caterpillar Engine

BASELINE - N0; CATALYST

z
o
p=
u

D
O
Lul

tr

K

o
2

70
60
50
40
30
20
' 10'



V

V

A 125 I™LOAD
• 100 tar LOAD
¦ 96 tar LOAD

V pilot scale tests

V*

\

V



¦4-

H-

—H	1	1	1	

0.89 0,91 0.93 0.95 0.97 0.99 1.01 1.03 1.05

SR3

Figure 14, Baseline Reburn NO, Reductions lor Full-Scale System

5B-68


-------
too

v	y	v	

• 90 kw LOAD
75 h	V

—	A 1 DC- kw LOAD	_

g.	V

¦ 90 kw LOAD

2 50

25

V PILOT SCALE TESTS	^ y

V

V V

OSS	0 SO	0.95	100	1.05

SR3

Figure 15. Full-Scale NO, Reduction with Catalyst

l»PP*t Dtf*.

Figure 16. Laboratory Repower Test Facility

5B-69


-------
too

z

o

p
u
~
~
y
cc

o

2

90

SO
7Q
&0
50
*0
30 ¦
20 ¦
10 •

b° 
-------
SELECTIVE NON-CATALYTIC REDUCTION (SNCR)
PERFORMANCE ON THREE CALIFORNIA
HASTE-TO-ENERGY FACILITIES

Barry L. McDonald, P.E.

Gary R. Fields
Mark D. McDannel, P.E.
CARNOT

15991 Red Hill Ave., Suite 110
Tustin, CA 92680-7388

5H-71


-------

-------
ABSTRACT

Concern over N0X emissions from municipal waste combustors (MWC) has increased to the
point where recently the EPA determined DeNOx to be BACT on several MWC facilities.
In addition, in February of this year, the EPA issued new source performance standards
(NSPS) which establish N0X limits for facilities larger than 250 tons/day, at 180 ppm,
corrected to 7% oxygen.*

Three MWC located in California were the first incinerators to install post-combustion
N0X control in the form of Exxon's Thermal DeN0x, a selective non-catalytic reduction
(SNCR) technology. Other examples of SNCR technologies which have been applied or
proposed for NOx control on MWC units include: (1) urea injection (NOxOUT), (2)
cyanuric acid (RAPENOJ, and (3), ammonium sulfate. This paper discusses the practical
(rather than the theoretical) aspects of the DeNO, technology such as: 1) installa-
tion, 2) control strategies, 3) regulatory limits, 4) system performance, 5)
startup/shutdown considerations and 6) secondary effects (i.e., plumes and increased
particulate emissions).

All N0X data presented in this paper is given on a dry basis corrected to 7%

oxygen.

Preceding Page Blank	5B-73


-------
INTRODUCTION

In nearly three decades, waste generation in this country has doubled, from 88 million
tons in 1960 to nearly 180 million tons in 1988, This is the equivalent of each
person in the U.S. generating four pounds of waste every day. The EPA now projects
that by 2000, we will produce 216 million tons per year, or close to 4-1/2 pounds per
person per day.

Of the 180 million tons being produced annually in 1988 roughly 76 percent was
landfilled; 11 percent was recycled; and 13 percent was incinerated. With more
stringent regulations involving the siting and operation of landfills the cost of
landfilling has increased and the available capacity decreased. By 1992 the EPA
projects that the fraction of the nation's waste that is incinerated will have
increased to roughly 19 percent.

Recognizing the growth of incineration, currently there are approximately 130 MWC.
facilities operating in the U.S., the EPA has moved to establish controls on the
emissions from these facilities. On February 11 of this year the EPA promulgated
final standards for new and existing MWC. Relative to air emissions, the New Source
Performance Standards (NSPS) established limits for new facilities for: particulate
matter, dioxins/furans, sulfur dioxide, hydrogen chloride, nitrogen oxides and carbon
monoxide. The EPA also promulgated guidelines with the intended effect to initiate
state action to develop state regulations controlling emissions from existing MWC.
The guidelines covered the same air contaminants as those covered under NSPS, with the
exception that there was no guideline given for nitrogen oxides.

The NSPS set for nitrogen oxide (N0X) emissions for new large MWC (those constructed
or modified after December 20, 1989 with a greater throughput than 250 TPD) is 180
ppm, averaged over a 24-hour period.

Currently, the Exxon Thermal DeN0x process had been operational from two to three and
one-half years on three state-of-the-art facilities built in California. It is
understandable that DeNOx was first demonstrated in California since the state and the

5B-74


-------
area regulated by the South Coast Air Quality Management District (SCAQMD), in
particular, are recognized as regions in which emission controls are especially
strict, due to regional air quality.

The first MWC in California to install Thermal DeNOx was the Commerce Refuse-to-Energy
Facility which is operated by the Los Angeles County Sanitation District (LACSD). The
Stanislaus County Resource Recovery Facility which is owned and operated by Ogden-
Hartin also employs DeNOx. Finally, the third MWC to have installed Thermal DeNOx was
the Southeast Resource Recovery Facility (SERRF), which is owned by the City of Long
Beach and operated by Hontenay Pacific Power Corporation.

THERMAL DeNOx INSTALLATION AND CONTROL

Mass-burn waterwall MSW incinerators are ideally suited, with respect to Thermal DeNOx
performance, as compared to utility boilers. Incinerators generally have an ideal
temperature region (1600-1800 F) in which to inject the ammonia and obtain good N0X
destruction. Furthermore, flue gas velocities are lower giving longer residence times
and there is good mixing due to overfire air ports. These factors all enhance the
performance of DeN0x on MWC furnaces.

Figure 1 provides general information on the current Thermal DeN0x installations at
the three incineration plants. The plants are remarkably similar relative to design
steam flow (each unit is large by EPA NSPS standards, throughput >250 TPD), but it is
easy to observe that the DeN0x designs differ markedly. Some of the unique designs
and operational features are:

Commerce

Four injection zones are provided. The lower two injection zones
were added to assist in meeting permit conditions at reduced load
and during startup and shutdown.

Although originally equipped with an air compressor to provide 30
psi carrier air, overfire air at 1 psig is presently utilized.
This provides substantial power savings with no loss in perfor-
mance. The system configuration (Figure 2) includes purge air for
unused nozzles and remote zone selection.

Stanislaus

Ammonia feed rate is controlled automatically based on stack N0X as
shown in Figure 3. The control logic minimizes ammonia flow and
hence ammonia slip when the emissions are within permit limits.
Reagent flow increases substantially during off nominal periods.

Two injection zones are provided, however, only the upper level is
utilized during normal operation. The lower level is utilized
during startup and shutdown transients.

50-75


-------
•	Ammonia feed rate is controlled automatically by a proprietary
control system.

SERRF

•	Two injection zones are provided, however, only the upper level is
utilized during normal operation. Ammonia flow is proportioned
between the upper and lower zones using an algorithm which uses
upper furnace temperature as the only input.

•	Ammonia feed rate is controlled automatically based on stack N0X
concentration.

Having worked closely on the SERRF plant it would be helpful to other facilities
considering the Thermal DeNOx technology to report some of the early work conducted
shortly after startup. Initially, NOx control was inadequate and several measurements
were taken to assess why NOx could not be maintained continually below permit limits.
Temperature profiling was performed using suction pyrometry. Sample locations are
shown on Figure 4. Temperature profiling identified three problems which prevented
the DeNGx process from adequately controlling N0X: (1) rapid flue gas temperature
swings, (2) an increasing temperature gradient from the front towards the rear wall
of the furnace, and (3) excess temperatures. Working with Dravo and Steinmuller the
combustion logic and overfire air operation were significantly modified. While these
modifications stabilized temperatures in the furnace the injection location was
determined to be too low in the furnace. Ammonia was being injected into a region
where the flue gas temperature was above the optimum for DeNOx performance and some
of the ammonia was being oxidized. The optimum temperature was located near the next
higher level of boiler nozzle penetrations. Since the upper front wall nozzle
penetrations were already in place, it was relatively simple to connect an ammonia/air
header and insert the proper nozzles. The combinations of these modifications allow
the SERRF boilers to operate in compliance with their N0X limits.

Recent operational data for Commerce has demonstrated that some flexibility in
injection location is possible for operation under steady controlled firing
conditions. Four months of operational data provided the N0X vs. load relationship
presented in Figure 5, for four separate zone combinations. Of particular Interest
1s the ability of one zone (or combination of zones) to provide low N0X over a wide
operating range. Although DeNOx system performance is regarded to be highly dependent
on the temperature at the point of injection, the actual window can be rather wide
when a removal efficiency of 50% is acceptable.

5B-76


-------
REGULATORY EMISSION UNITS

Before reviewing the performance of the Thermal DeNOx systems at these three
facilities it is important to understand the regulatory limits or targets that each
facility was designed to achieve. It is interesting to note that although all three
facilities are located in California (two are even located in the SCAQMD) the
regulatory limits for each facility is uniquely different. The difference is not
solely the magnitude of allowable N0X emissions but also of particular significance
is the averaging time designated for each limit. Table 1 presents the N0X regulatory
limits for Commerce, Stanislaus and SERRF.

Each individual unit, (units are similarly sized from a steam throughput standpoint),
at the three facilities have a broad range of N0X limits to comply with. Considering
mass versus concentration limits and the five different averaging periods it is
interesting to note that there is only one common emission limit for all three
facilities. The allowable N0X emissions on a daily basis range from a low of 720
lb/day at SERRF to a high of 1130 lb/day at Stanislaus; Commerce has a daily, N0X
limit of 825 pounds.

It is obvious that lower NQX emission limits-are more difficult to achieve. However,
the averaging period and concentration versus mass limits have an important effect.
For example, even though Commerce, in order to avoid an emission exceedence, cannot
exceed 175 ppm for a fifteen minute period, the plant must operate below roughly 120
ppm so as not to exceed the 40 lb/hour limit. (Note: The 175 ppm limit for Commerce
and SERRF is not in either plant's authority to construct permit but is a prohibitory
limit in SCAQMD Rule 476, Rule 476 limits the N0X concentration from liquid or solid
fuel fired units in the Basin to 225 ppm corrected to 3% 02. This value is equivalent
to 175 ppm corrected to 7% 02.)

COMMERCE N0X LIMITS

The daily N0X mass emission limit at Commerce (825 lb/day) is equivalent to roughly
34 Ib/hr which translates to about 100 ppm. Consequently, the plant needs to operate
consistently below 100 ppm in order to comply with the daily mass limit. A safety
margin below 100 ppm would be required if frequent upsets resulting in large spikes
of NQX were to occur.

STANISLAUS NO, LIMITS

Stanislaus is unique in that N0X emissions are regulated by both the Stanislaus County
Air Pollution Control District (SCAPCD) and the EPA, due to EPA's PSD permit. The
most stringent limit from a continuous basis is the SCAPCD daily mass limit of 1130

5B-77


-------
lb/day which is roughly equivalent to 150 ppm. Stanislaus is also unique in that the
plant has a stack ammonia limit of SO ppm (raw).

nqk compliance test results

Emissions data taken from initial compliance tests and some more recent results are
presented in Table 2, Uncontrolled NOx data is not as plentiful as an analyst might
desire since all three plants are required to operate the DeN0x system when the plants
are on-line and/or burning refuse. To obtain uncontrolled emissions data, therefore,
a variance is required. Uncontrolled emissions are in-line with levels reported in
an EFA study, which reviewed NOx data from twenty-six mass-burn/waterwall facilities.
The study stated that the average uncontrolled N0X concentration was 242 ppm. This
is in the range of the data from Commerce, SERRF and Stanislaus. It should be noted
that the 68 ppm listed for SERRF in the EPA study was incorrect. The study stated
that the low NOx value was due to flue gas recirculation, which as previously stated,
is incorrect.

A limited amount of work was initially performed to evaluate FGR injected in the first
three undergrate zones on the SERRF units. Preliminary indications were that some NQX
reduction was achievable at a recirculation rate of roughly ten percent. Since those
early tests there have been numerous modifications to the SERRF units. In order to
establish a more definitive answer as to the effectiveness of FGR a research plan was
submitted to the SCAQMD. The goals of the research plan are:

1.	to quantify the effect of FGRs contribution to NOx reduction during
simultaneous FGR/Thermal DeNOx use,

2.	to quantify FGR's contribution to reduced ammonia usage and slip
during simultaneous FGR/Thermal DeNOx use, and

3.	to assess the impact of FGR on primary combustion zone location and
on boiler/grate operation.

Work, under a SCAQMD research permit, is currently on-going. Along with the FGR
study, an extensive DEN0x optimization program is being conducted.

Carnot conducted a DeNOx optimization program at Commerce. At Commerce the study
evaluated injection level (there were only two injection levels at the time), carrier
air injection pressure and ammonia injection rate. The study concluded that optimum
performance was achieved by injection of an NH3-to-N0x mole ratio of about 1.5 through
the upper elevation of nozzles. Carrier air pressure had no effect on DeN0x
performance. Further, it was observed that even when there was substantial ammonia
slip levels at the economizer exit the level at the stack due to the spray dryer
baghouse was held to less than 5 ppm.

5B-78


-------
The controlled N0X data given In Table 2 was taken at nominal full load. The lower
levels achieved by SERRF are due to a higher rate of ammonia being injected as
compared to Commerce and Stanislaus. The higher ammonia injection rate also explains
the higher ammonia slip numbers experienced at SERRF.

STARTUP AND SHUTDOWN TRANSIENTS

With the advent of continuous emissions monitors (CEMS) plant operators are able to
observe emission levels during all operational phases. CEMS have proven to be
invaluable tools, however, some problems, which were not originally anticipated have
developed with the data they provide. Before CEM data were available, emissions were
measured using integrated, sampling techniques. Normally emissions tests were
conducted at full load.

CEM data now permits plant operators to monitor emission levels during transient
conditions such as startup and shutdown. Because these periods are transients, the
emission rates are not characteristic of normal steady-state operation. Regulations
in establishing permit limits have only had to deal with what emissions are expected
to be at steady load. Once it was determined that steady state emission levels could
be exceeded during startup/shutdown transients, regulators were forced to modify
emission requirements. As an example, the SCAQMD adopted Rule 429 which recognizing
this problem provided startup/shutdown NQX relief for refinery boilers, refinery
process heaters, gas turbines, utility boilers, industrial boilers, industrial process
heaters and nitric acid plants.

Emission transients can occur for both N0X and CO during startup and shutdown. Since
Thermal DeNOx is a temperature dependent process it is critical that special
procedures be developed to control emissions during these transients. In addition,
regulators need to develop acceptable permit language which provides plant operators
sufficient margin to transition these periods safely.

IMPACT OF AMMONIA SLIP ON PARTICULATE EMISSIONS

As a result of the way particulates are defined by California regulators ammonia use
for N0X control has resulted in higher particulate values being reported. This has
caused concern among plant operators as well as particulate control suppliers who are
being asked to guarantee particulate emission levels but have no way of collecting the
gaseous components that make-up this excess particulate, which we refer to as pseudo-
particulate.

Pseudo-particulate is an artifact of the standard EPA Method 5 sampling procedure.
In the back-half of the sampling train are two impingers containing water. Normally

5B-79


-------
gaseous species pass through the water and when the impinger solution is evaporated,
there is little material found. On plants equipped with N0„ control equipment which
results in some ammonia slip, the amnionia is absorbed by the water creating an
alkaline solution. The solution acts as an acid gas scrubber removing S02, HC1 and
N0Z, forming the associated ammonium salts. When the impinger solution is evaporated
these salts remain leaving the particulate residue referred to as pseudo-particulate.

When test protocols were being developed for Commerce, the SCAQMD accepted a procedure
which excluded the neutral salts caught in the back-half fraction. All of the
particulate tests conducted at Commerce were adjusted to exclude these neutral salts.
Similarly, the Stanislaus County APCD accepted the premise behind the particulate
adjustment and the initial particulate compliance tests at Stanislaus were corrected
for neutral salts.

Recently, however, the SCAQMD in evaluating the test protocol for SERRF concluded that
the neutral salt adjustment was unwarranted. Their logic was that since the gaseous
species combined in the atmosphere forming particulate that it was incorrect to back
them out from the particulate determination simply because the components were gaseous
when they passed through the sampling train. Consequently, particulate tests at SERRF
include this pseudo-particulate fraction. It is interesting to note that the SCAQMD
draws a distinction between plants using ammonia for N0„ control and those using
ammonia for ESP performance improvement. When measuring particulates from facilities
using ammonia as an ESP performance enhancement SCAQMD allows the neutral salts
collected in the impinger solution to be backed-out of the particulate determination.

The impact of including pseudo-particulate in the particulate emission determination
is shown in Table 3.

As might be expected, the higher the ammonia slip, the more prevalent this problem
becomes. Individuals considering projects that employ ammonia or other SNCR
technologies, as well as regulators need to understand the impact ammonia can have on
particulates when setting particulate emissions levels.

IMPACT OF AMMONIA SLIP ON PLUME FORMATION

With the wide application of ammonia injection and other SNCR technologies for N0X
control, there have been frequent occurrences of plumes from sources which have
chlorine in the fuel. Typically these plumes are detached but once formed continue
for long distances. SERRF has a detached plume and frequently a plume can be observed
at Commerce. Stanislaus was reported as having a plume in the past but due to the new
N0X control logic has stated that a plume no longer is visible.

5B-80


-------
Analysis of the situation at SERRF in terms of chemical equilibrium calculations
indicates that the plume problem is explainable in terms of ammonium chloride (NH4C1)
condensation in the atmosphere above the stack. These calculations also show that
ammonium sulfate or bisulfate should not be contributing factors.

Principles of chemical thermodynamics show that NH4C1 condensation is governed by the
product of NH3 and HC1 concentrations in the stack ([NH3] x [HC1], the "concentration
product") and the stack and ambient temperatures. The thermodynamic relationship
showing the critical value of [NH3] x [HC1] above which condensation will occur versus
temperature is shown in Figure 6. For any combination of stack temperature, ambient
temperature and concentration product in the stack, there is a dilution vector on
Figure 6 along which the stack conditions will decay as ambient air mixes with the
flue gas leaving the stack.

Once NH4C1 forms, its visibility is dependent upon plume diameter. This is known to
be a logarithmic dependence for simple opacity but becomes more complicated when back
scattering is included, which must be the case for a white plume. The plume diameter
is, of course, related to stack diameter and air infiltration.

Based on a study conducted at SERRF, to avoid NH4C1 formation requires extremely low
values of NH3 and/or HC1 concentrations, such that NH3 x HC1 does not exceed approxi-
mately 10"4 ppi2. This criterion is impractical for SERRF to achieve and total
avoidance of NH4C1 formation therefore does not appear to be an option. Further, the
plume visibility is essentially proportional to the lesser concentration of NH3 and/or
HC1.

SUMMARY

Thermal DeNOx is successfully providing adequate NOx control such that Commerce, SERRF
and Stanislaus can meet their individual N0X emission permit limits. Furthermore, all
three plants operate below the NSPS N0X limits recently promulgated by the EPA.
Critical to the success of this technology is stable combustion and the ability to
inject and properly mix the ammonia at the proper optimum flue gas temperature. When
done correctly, continuous N0X compliance is possible.

By reducing the time intervals by which compliance is monitored, plants are forced to
operate at Tower N0X levels to avoid emission upsets associated with variations in
feed quality or equipment upsets. Furthermore, the use of ammonia injection is not
without secondary problems, specifically potentially higher particulate emissions,
depending on what regulatory agencies define particulate to be, and visible plume
formation.

5B-81


-------
Four Side Wall
(8) NH3 Injection
Nozzle Locations

COMMERCE REFUSE-TO-ENERGY FACILITY:

.';8oij#nX;"XX;^

xstokenx'x^	x'x'x'xvxxx': xx

vNHsvJnj?pilOn;X;; '..' 4 Leyels/bn Botiv XX

xxxxxxxxxxxx	side waiis'-xx-xxxx xxxxxxxxxxxxxxx

vBoIlis^'Cross-Sectlbn-XiSi'S^w^x f8<(d)vX-XXXXX.XvXv

Two Front Wall
(10) NH,Injection
Nozzle Locations

STANISLAUS COUNTY RESOURCE RECOVERY
	FACILITY:	__

XUnlfsiXXXXX-': -X-Xv" .X(2J 400 TPD ¦¦X-X'--"
X Boiler;-X'X' XXXX iZurri (Not Available)
~;Stoker:7v;./.'vvv . .Martin
XN.tt,'_Injection: ' x-X!;! 2 Levels on the •

Boiler Cross-Section:

Front Wall X ;!
(Not Available)/

awwBi

Front Wall (15)
and Side Wall
(23) NH3 Injection
Locations

SOUTHEAST RESOURCE RECOVERY
FACILITY (SERRF};

'.Units;.v . X (3) 460 TP.D XXXX	'/XXXXX

••Boiler.' X..	XX- L&C Stelnmuelter (117,170 ib/hr)

Stoker;; ;X . ¦ ' L&C Stelnmueller ; . Xy
NHj Injection: .... 2 Levels, Front Wall
and B.oth Side Walls
Boiler. Cross-Section: 19*(wl	X"



Figure 1. Various Ammonia Injection Configurations at Three
California MSW Incinerators Equipped with Thermal DeNO,

5B-82


-------
NOx - PPMc at 7% O2

Figure 3. Commerce Refuse-To-Energy Facility
Ammonia Feed Rate vs. NOx •

5B-83


-------
Temperature Profiling







Ffont OFK
14 nozzles of-

3-5/8" DiAMETEH

FI.R
~

01,02
~

C*,E2

o

A1.A2 61,S3 CIC3

~ 0 ~

Furnace Penetrations

for

Ammonia Injection Noz



B;';

€11



zles



^t<0771F.3VDO.TE30N
SIDE. WALLS};



uwh

^"^pi"Nb2ZLEStb^tEDDN.BOlfK"'

FEED I GBATE DIRECTION
END

Figure 4. North side schematic of a typical SERRF Steinmuiler-designed furnace.
Observation ports through which temperature profiling was performed are
shown.

200

«
o

150

f>-

4^

<9

f 100

0.

CL

X

O
z

50

¦ ZONE 3
• ZONE 3,4
a ZONE 2,3,4
Q ZONE 2,3

—j—

20

,—|—
60

—,—
80

40

%MCR

100

Figure 5. Commerce Refuse-To-Energy Facility NO, vs. Load
Utilizing Various Injection Zones.

5B-84


-------
NHjCI (s)

NKj(g) + HCI (g)

Explanation:

At any given temperature,
condensation will occur Ii
the log of the product ot
mole-fractions XNH3*HCI,
expressed as ppm2, lies
above the curve.



100

200

300

400

500

600

700

Temperature F

Figure 6. NH4C1 Equilibrium Curve

5B-85


-------
TABLE 1

REGULATORY LIMITS FOR COMMERCE, STANISLAUS, AND SERRF

Plant



Commerce

Stanislaus

SERRF .

Air Quality District



South Coast
AQMD

Stanislaus County South Coast
APCD AQMD

Pollutant

Averaging
Period







NGX ppm @ 7% 0,
NO. ppm § 7% 0,
NO, ppm @ 7% 02
NO lb
N0X lb

IS min.
1 hour
8 hour
1 hour
24 hours

175

40

825

200
H30

175
116

34
720

NH3 ppm (raw)



..

50

--

EPA-PSD









More stringent of
NOj, ppm @ 7% 02

or
HO--lb
anS

More stringent of

N0X ppm * 7% 02
or

N0X lb

3 hour
3 hour



175
160.5



24 hour
24 hour

--

165
1200



NOTE; The EPA NSPS NO, limit for
averaged over 24 hours.

MWC which are

larger than 250 TPD

is 180 ppm N0X

5B-86


-------
TABLE 2

COMPARISON OF NO_ EMISSIONS FROM THREE CALIFORNIA MSW
INCINERATORS EQUIPPED WITH THERMAL DENOx

Uncontrolled NO

*

Commerce	Stanislaus		SERRF	

Unit 1 Unit 2 Unit 1 Unit 2 Unit 3

ppm @ 7% 0,	128-217	298	305	--	210	259

Ib/hr	44-75	90.4	96.0	--	74.8	93.1

Controlled NO,

ppm 0 7% 0,	104 93	112	49	72 54

Ib/hr	35.8	28.1	36.0	16,5	22.7	17.9

Ammonia Slip

ppm (raw)	-2	3.7	5.0	--	--	35

TABLE 3

PARTICULATE EMISSIONS AND THE IMPACT OF ADDING BACK
THE PSEUDO-PARTICULATE FRACTION

Commerce

Stanislaus

SERRF

Permit Limit	5.5 1 b/hr

Test Results	2.5

% of particulate	88%

caught in the back-half
of the sample train

Impact on particulate	+ 60%

level if neutral salts
were added back

0.0275 gr/sdcf

Unit I
0.011

• 51%

+ 34%

Unit 2
0.011
79%

+38%

5.0 Ib/hr
Unit 3
1.7
70%

N/A

5B-87


-------

-------
USE OF NATURAL GAS FOR NQX CONTROL
IN MUNICIPAL WASTE COMBUSTION

H. Abbasi and R. Biljetina
Institute of Gas Technology
3424 South State Street
Chicago, Illinois 60616

F. Zone and R. Lisauskas
Riley Stoker Corporation
Riley Research Center
45 McKeon Road
Worcester, Massachusetts 01610

R. Dunnette
Olmsted Waste-to-Energy
2128 Campus Drive, S.E.
Rochester, Minnesota 55904

K. Nakazato
Itoh Takuaa Resource Systems Inc.
335 Madison Avenue
New York, New York 10017

P. Duggan and D. Linz
Gas Research Institute
8600 West Bryn Mawr Avenue
Chicago, Illinois 60631

f					5B-89

J Preceding Page Blank


-------

-------
ABSTRACT

Natural gas injection (NGI) technology for reducing N0X emissions from
municipal waste combustors (MWCs) is being developed in a joint
program between the Gas Research Institute (GRI), the Institute of Gas
Technology (IGT), Riley Stoker Corporation (Riley), Olmsted Waste-to-
Energy (Olmsted), and Takuma Company, Ltd. (Takuma). The approach
developed by IGT and Riley (termed METHANE de~NOx) is based on
extensive, full-scale, MWC in-furnace characterization followed by
pilot-scale testing using simulated combustion products that would
result from the firing of 1.7 X 10® Btu/h (0.5 MWth) municipal solid
wastes (MSW). The approach involves the injection of natural gas,
together with recirculated flue gases (for mixing), above the grate to
provide reducing combustion conditions that promote the destruction of
NOx precursors, as well as N0X. Extensive development testing was
subsequently carried out in a 2.5 X 106 Btu/h (0.7 MWth) pilot-scale
MWC firing actual MSW. Both tests, using simulated combustion
products and actual MSW, showed that 50% to 70% NOx reduction could be
achieved. These results were used to define the key operating
parameters.

A full-scale system has been designed and retrofitted to a 100-ton/day
Riley/Takuma mass burn system at the Olmsted County Waste-to-Inergy
facility. The system was designed to provide variation in the key
parameters to not only optimize the process for the Olmsted unit, but
also to acquire design data for MWCs of other sizes and designs.
Extensive testing was conducted in December 1990 and January 1991 to
evaluate the effectiveness of NGI. This paper concentrates on the
METHANE de-NOx system retrofit and testing. The results show
simultaneous reductions of 60% in NOx, 50% in CO, and 40% in excess
air requirement with natural gas injection.

Preceding Page Blank

5B-91


-------
UTILIZATION OF NATURAL GAS IN MUNICIPAL WASTE COMBUSTORS (MWCs)

In 1986, following GRI's successful pilot-scale testing of natural gas
reburning for N0X reduction in coal-fired applications, GRI and IGT
began an investigation of the potential for utilizing natural gas in
MWCs for the control of NQX emissions. At that time the control of
N0X was required in the State of California; however, it was not yet
being seriously discussed elsewhere in the United States. By 1989,
the U.S. Environmental Protection Agency had announced its intention
to set limits for N0X emissions from all MWCs. The limits being
evaluated were based on the performance of the thermal de-NOx process,
which uses ammonia injection to reduce N0X emissions. The thermal de-
N0X process has been installed on three MWCs operating in California.

Figure 1 illustrates the N0X reduction ¦ approach proposed for MWCs.
This approach, termed METHANE de-NOx, involves the injection of
natural gas, together with recirculated flue gases (for mixing), above
the grate to provide reducing combustion conditions that promote the
destruction of N0X precursors, as well as NGX. Secondary overfire air
(OFA) is then injected at a higher elevation in the furnace, after
sufficient residence time at these reducing conditions, to burn out
the combustibles. Applying this approach to MWCs is challenging
because of the low heat content of the waste being fired, the presence
of significant amounts of N0X precursors (for example, NH-j, HCN) above
the grate, and the high excess air levels that are typically used in
these types of combustors. These conditions result in relatively low
temperatures and high oxygen and N0X precursor levels in the primary
combustion zone compared with conditions in the same location in a
coal-fired boiler. Further complexities include the distribution of
air, which includes a relatively large amount through the burnout
grate at the discharge end of the combustor, and a large amount of air
infiltration due to the negative operating pressure of the combustor.
Also, because of the variability of the waste being burned, conditions
in the furnace are typically variable. The initial concern,
therefore, was that if NGI could be made to work at all in MWCs, it

5B-92


-------
might require either large amounts of natural gas, or extended furnace
zones to increase the residence time, or both.

The objectives of the development program were to 1) characterize the
in-furnace conditions of a commercial MWC to define the variability of
operation, the gas compositions within the furnace, and the flow
distribution patterns for oxygen, CO, NOx, and other flue gas species,
2) evaluate the gas-phase chemistry in laboratory furnace simulation
experiments (0.5 MWth) and define regions of operation in which NGI
could be effective using simulated MWC flue gases, 3) design and build
a pilot combustor (0.7 MWth) firing actual MSW, in which the NGI
process could be developed and tested, and 4) design and conduct a
full-scale evaluation of the NGI process on a commercial MWC.

The experimental program was conducted from 1987 to 1989. The
installation of the full-scale field evaluation was completed in late
1990, and NGI testing was completed in January 1991. The remainder of
this paper summarizes the research conducted over the last 3 years
that led to the design of a full-scale system and the results of NGI
testing on the full-scale commercially operating MWC.

RESULTS OF COMMERCIAL COMBUSTOR CHARACTERIZATION

The baseline data were acquired on one of the two units at the Olmsted
County Waste-to-Energy Facility (Figure 2) located in Rochester,
Minnesota. The design of the combustor is an integration of the
Takuma MWC stoker and combustion control technology with the Riley
water-wall furnace technology. Each unit was designed to burn MSW at
the rate of 100 tons/day (90 metric tons/day), producing about
24,000 lb/h (11,000 kg/h) of 615-psig (42-bar) superheated steam.

The unit was tested while varying load, total stoichiometric ratio
(TSR), allocation of undergrate air (UGA) flow, and OFA location. Two
general types of tests were conducted; in-furnace measurements by IGT
and overall system performance data acquisition by Riley. Test
details have been presented earlier (1) and the results are briefly
described below.

5B-93


-------
In normal operation, with 60% to 80% excess air to ensure complete
combustion, this unit produced about 125 to 175 ppm* N0X. Without GFA
and at lower excess air, N0X emissions were reduced significantly, but
CO and total hydrocarbon (THC) emissions increased greatly. The
baseline data show that NOx can be reduced by eliminating OFA and
reducing e x c 6 s s a i r j how e v e r, incomplete c omJDu s t i on results **™ as
indicated by the high CO levels. The goal of NGI is to reduce N0X
emissions without the corresponding increase in CO emissions. The
furnace characterization data that were acquired also show that it
would be possible to create the substoichiometric NOx reducing
conditions within the furnace with NGI.

Furnace Simulator

A pilot furnace at IGT was fired with No. 2 fuel oil using preheated
air and adding appropriate amounts of oxygen, moisture, and ammonia
(to simulate fuel-bound nitrogen). Thus, the pilot furnace closely
simulated the baseline combustion products from the stoker firing
1.7 X 10® Btu/h (0,5 MWth) of MSW. Tests investigated the impacts of
reducing zone residence time, stoichiometry, and gas temperature;
amounts of natural gas and fuel bound nitrogen* overall excess airj
and the amount of flue gas recirculation (FGR) for mixing the natural
gas with the combustion products. These test details have also been
presented earlier (2^3).

In typical excess air operation (without NGI), the furnace simulator
produced relatively steady N0X levels of 200 to 225 ppm - independent
of residence time. As illustrated in Figure 3, however, residence
time plays an important role when natural gas is injected, because
sufficient time must be available for the natural gas to decompose N0X
precursors. The first 3 seconds after NGI reduced N0X from 225 to
75 ppm. Longer times produce very little additional N0X reduction.
The results showed that if NGI is to be effective, it must be injected
into the MWC such that sufficient residence time at high temperatures
is provided before OFA is injected for combustible burnout. An NGI
level of 15% was found to be sufficient for 50% to 70% NOx reduction.

* All of the NOx and CO emission values presented here are on a 12% O2
and dry basis. For a 3% O2 basis, multiply values by 2 and for a 7%
O2 basis, multiply by 1.56.

5B-94'


-------
Pilot MWC Combustor •

Because of the encouraging furnace simulator test results, it was
decided to make follow-up tests in the pilot combustor at Riley's
Research Center. A pulverized coal combustor at Riley was modified to
simulate the commercial unit at Olmsted, and several different batches
of MSW were tested to investigate the impacts of reducing zone
residence tine and stoichiometry, natural gas injection location and
amount, and overfire air injection location. The results have been
presented earlier (3,4) and show that without NGI, N0X emissions
ranged from 110 to 165 ppm - a fairly good simulation of the baseline
results obtained in the commercial combustors. With 10% to 15%
(percent of total heat input) NGI, N0X emissions were reduced by as
much as 70%, depending on the natural gas and OFA injection points and
the residence time in the reducing zone. NOx emissions decreased from
100 to 130 ppm at 0.6 seconds residence time and 40 to 80 ppm at
1.2 seconds residence time. These results verify the beneficial
effects of residence time as observed in the furnace simulator tests.
A reducing zone stoichiometric ratio of between 0.8 to 1.0 was found
to be sufficient for effective N0X reduction. With NGI, it was also
possible to operate the unit with significantly lower excess air.

FIELD EVALUATION OF NATURAL GAS.INJECTION

In light of the favorable test results obtained from both the IGT and
Riley pilot-scale investigations of NGI, a field evaluation was
undertaken. The NGI technology was retrofitted to one of the Olmsted
units. This facility was also used to acquire all the baseline data
reported here.

The pilot-scale work had demonstrated the potential of NGI for
reducing the emissions of N0X, CO, and THC. A number of issues
remained, however, before it could be commercialized as a viable
emissions reduction technology. The major issues were as follows: -

•	Can NGI be as effective on a commercial unit,
considering the actual conditions of high excess oxygen

ir 2.21I3 i 1 i f f cl ua 11	^sjipisrcitpincj

temperature?

•	Can the already low CO and THC levels (<50 ppm) be
further lowered and stabilized on the full—scale unit,
as evidenced in the pilot unit?

5B-95


-------
•	Can proper furnace aerodynamics be maintained or
improved? In other words, can adequate distribution of
natural gas in the reducing zone and OFA in the burnout
zone be accomplished in full-scale systems?

•	What would be the impact on thermal efficiency,
slagging, corrosion, steam superheat, and other boiler
performance parameters?

•	What are the costs and advantages over thermal de-NOx
and/or other alternative N0X control measures?

The results of the field evaluation would help resolve many of these
issues. As with the experimental program, this 15-month effort was
conducted jointly by IGT and Riley in consultation with Olmsted and
Takuma. The work effort was divided into three major activities. The
first involved finalization of site selection and engineering and
design of a flexible NGI retrofit system. The second was the
procurement and installation of the retrofit system. The third was
the field evaluation testing of NGI for emissions reduction, as well
as other impacts, which began in early December 1990 and was completed
in late January 1991.

The primary goal was to reduce NOx to below 70 ppm from the current
uncontrolled level of over 140 ppm without adversely affecting other
emissions such as CO and THC. Additional goals were to maintain or
improve the steam capacity while increasing the boiler thermal
efficiency.

The retrofit METHANE de-NOx system was designed by IGT and Riley based
on the pilot-scale testing results. The primary variables (determined
during the pilot testing) for design of the NGI system are -

•	15% natural gas above grates to create substoichiometric
conditions

•	15% FGR above grates for mixing the natural gas with the
furnace gases

•	Variability in reducing zone stoichiometry; reducing
zone residence time; and natural gas, FGR and OFA flows,
injection locations, and velocities.

The retrofit included installation of an FGR system and modification
of the furnace walls to accommodate several nozzles and
sampling/observation ports at multiple levels. The design also

5B-96


-------
provides for acquisition of the necessary in-fumace and flue gas
composition and temperature data, as well as other relevant data.
Recirculated flue gas, taken from the economizer outlet, is used to
introduce natural gas above the stoker.	,

ofa injectors are installed in two locations in the upper half of the
furnace for combustible burnout. The two elevations enabled testing
of different residence times for the reducing zone. Residence time
has a significant effect on NOx reduction and combustible burnout.
Inserts were employed during the testing to evaluate higher injection
velocities for the OFA, natural gas, and FGR,

FIELD EVALUATION TESTS

Extensive testing was carried	out on the 100-ton/day commercially

operating MWC during December	1990 and January 1991. These tests

investigated the impacts of the	following variables.

•	OFA location — to change the residence time in the
reducing zone

•	OFA amount, injector size, and number of injectors — to
optimize combustible burnout

•	Natural gas and FGR amounts, distribution, injector
sizes, and injector locations - to modify" reducing zone
mixing

•	UGA amount and distribution - to modify MSW combustion
profiles.

As indicated earlier, the objective of the testing was twofold:

1.	To prove the effectiveness of natural gas in reducing
the N0X emissions on a full-scale commercial unit
without any adverse effects

2.	To acquire design data for the application of the NGI
technology to MWCs of other sizes and designs.

As a result, the system was instrumented to provide an extensive data
base for the impacts of NGI on both the furnace side, as well as the
steam side parameters. The following is a list of measurements made
during the tests.

5B-97


-------
•	Full spectrum of furnace and steam side operating data
including temperatures, flows, pressures, etc, through a
specially installed computer data acquisition system and
manually

•	Gas composition (O2, CO, THC, C02, N0„) and temperature
profiles in the reducing zone below the OFA injectors
and at the furnace exit above the OFA injectors

•	Flue gas composition (02, CO, C02, N0X) at the
electrostatic precipitator (ESP) inlet

•	Flue gas composition (©2# CO, NOx) in the recirculated
flue gases

•	Oxygen concentration in the reducing zone (continuously)

•	Ash samples

•	MSW samples.

The in-furnace gas composition and temperature measurements were made
using water-cooled gas sampling and suction pyrometer probes that were
installed at various elevations to traverse the furnace. Two sets of
continuous emission monitors were employed. One set of 02, CO, C02,
and N0X analyzers was installed near the ESP to measure the gas
composition at the ESP inlet; and another set of 02, CO, THC, C02, and
N0X analyzers was installed in the control room to measure the gas
compositions inside the furnace and in the recirculated flue gases.
The gas composition at the ESP inlet was measured continuously for the
duration of each test, while the gas composition in the recirculated
flue gases was measured periodically between the in-furnace traverses.
The moisture contents of the flue gases and the flue gas flow rates
were also measured during some of the tests.

The extensive data that were acquired during the field evaluation
tests have not been fully reduced and analyzed at this writing. The
composition of the actual MSW burned during the tests is also not yet
available. Consequently, the data presented here are limited. The
results will focus on N0X and CO emissions measured at the ISP inlet
and their preliminary relationships with some of the significant
operating parameters. In general, these relationships were consistent
with the pilot-scale results. The data presented here are further
limited to the configurations that provided the optimum results with
NGI. Data are presented for three types of tests. First, these data

5B-98


-------
are presented with the baseline configuration as the unit is normally
operated,* second, in the NGI configuration with FGR injected into the
lower furnace and OFA moved up to a higher elevation; and third, also
in the NGI configuration with both FGR and natural gas injected into
the lower furnace and OFA injected at the higher elevation.

Table 1 summarizes the average values of selected operating data, as
well as CO and N0X emissions for these three test configurations.
Data are also presented from the 1987 baseline testing and for one
test with NGI that was carried out at a higher steam flow to maintain
the MSW rate at the current normal baseline value of 7000 lb/h. The
MSW feed rate and the total flue gas flow rate shown have been
estimated assuming typical MSW composition and heat content. The
actual values might be somewhat different, but the trends are expected
to be unaltered. It must be noted that the steam flow during the 1991
baseline test was about 28,250 lb/h or 6% higher than the current
normal baseline steam flow of 26,700 lb/h, and 20% higher than the
1987 baseline level of 23,500 lb/h. During most of the tests with
NGI, the steam flow rate was maintained at 29,000 lb/h or 9% higher
than the current normal baseline level (as there was no need for the
additional steam) which automatically decreased the MSW feed rate to
the 1987 baseline value. However, as shown, one test was carried out
with the MSW rate maintained very close to the current normal baseline
level by increasing the steam flow by about 14%. This was to prove
that NGI retrofit may not necessarily recjuire a decrease in MSW feed
rate. Table 1 shows that 12.5% to 14% [total heat input) NGI allowed
a reduction in excess air from over 70% to about 40% which may
increase the boiler thermal efficiency.

The data presented in the table also show that, compared to the 1991
baseline test, NGI decreased the NOx emissions by 601 and CO emissions
by 50%. The N0X emissions were decreased by 40% with FGR alone,
however, the CO emissions were more than double compared with the
average CO with NGI. The CO level with FGR was comparable to the 1991
baseline test value, but higher than the average value for the 1987
baseline tests. Figure 4 illustrates the relationship between NOx and
CO emissions for the Olmsted combustor that was found in 1987 for the
baseline operation. The relationship represents baseline operation at
different UGA and OFA flow distributions and excess air levels. The

5B-99


-------
current (1990-1991) data at baseline configuration, as well as with
FGR, show scatter but appear to fallow the 1987 trend. The average
NOx/CO values with FGR fall close to the average baseline curve. This
suggests that the effectiveness of FGR in reducing N0X may not be
significantly better than some of the other simpler combustion
modifications that were tested in 1987. The figure also illustrates
the effectiveness of NGI in controlling both NOx and CO emissions
simultaneously. Both NOx and CO emissions were significantly lower
with NGI. The average baseline NOx at 3 2 ppm CO (expected regulatory
limit) was about 137 ppm while the average N0X with natural gas was
about 50 ppm at an average CO level of about 22 ppm.

SUMMARY OF RESULTS

As discussed, the data acquired during the field evaluation tests have
not yet been fully reduced and analyzed. Based on the current
analysis, however, the following can be stated;

•	In general, the relationships between the significant
operating parameters and the emissions were consistent
with those found on the pilot-scale units.

•	Proper injection of 12% to 15% (heat input basis)
natural gas simultaneously decreased the NOx emissions
to below 50 ppm and the CO emissions to below 25 ppm,
which represents a 60% reduction in NOx and a 50%
reduction in CO compared to the 1991 baseline test
values.

•	NGI also allowed a reduction in excess air to 40% (from
the baseline levels of 70% to 80%), which may provide an
increase in boiler thermal efficiency.

•	An FGR level of 6% to 8% was sufficient to inject and
effectively mix the natural gas with the furnace gases.

•	Because of the reduced excess air requirement, it was
possible (as demonstrated in one test) to maintain the
MSW feed rate at the baseline level by increasing the
steam output to accommodate the additional heat input
with natural gas.

In conclusion, the effectiveness of the METHANE de-NOx process for
controlling NOx and CO emissions from MWCs has now been demonstrated
on a commercially operating MWC. Further analysis of the data should
provide additional information for application of this process to MWCs
of other sizes and designs, including refuse derived fuel (RDF).

5B-100


-------
ACKNOWLEDGMENT

Many sponsors played important roles in the development of the METHANE
de-NOx process. Considerable funding and guidance were provided by
the Gas Research Institute, Brooklyn Union Gas Co., Minnegasco,
Northern Illinois Gas Co., Northern Natural Gas Co., Peoples Gas Light
and Coke Co., Southern California Gas Co., and IGT's Sustaining
Membership Program member companies

The Olmsted County Waste-to-Energy officials and plant personnel
warrant special thanks for interrupting commercial operations to not
only accommodate but also vigorously assist the researchers in the
birth of a new process that can serve both the waste-to-energy and
natural gas industries.

REFERENCES CITED

1.	Fleming, D.K., Khinkis, M.J., Abbasi, H.A. , Linz, D.G. and
Penterson, C.A. "Emissions Reduction From MSW Combustion Systems
Using Natural Gas." Paper presented at the Conference on Energy
From Biomass and Wastes, XII, New Orleans, Louisiana, February
15-19, 1988.

2.	Abbasi, H., Khinkis, M.J., Itse, D., Penterson, G,, Wakamura, Y,
and Linz, D. "Development of Natural Gas Reburning Technology
for N0X Reduction From MSW Combustion Systems." Paper presented
at the 1989 International Gas Research Conference, Tokyo, Japan,
November 6-9, 1989.

3.	Emissions Reduction From MSW Combustion Systems Using Natural
Gas. Task 2. Pilot-Scale Assessment of Emissions Reduction
Strategies.	GRI-90/014 5 Final Report, Institute of Gas
Technology and Riley Stoker Corp., July 1990.

4.	Penterson, C.A., Itse, D.C., Abbasi, H.A., Khinkis, M.J.,
Wakamura, Y. and Linz, D.G. "Natural Gas Reburning Technology
for NOx Reduction From MSW Combustion Systems." Paper presented
at the ASME 1990 National Waste Processing Conference, Long
Beach, California, June 3-6, 1990.

5B-101


-------
]

^ Overlire All

Natural Gas/
Reclrc, Flue
Gases

Undergrate Air *

5B-102


-------
0	1	2	3,4	5	6

Residence Time, seconds

Figure 3. Residence time plays a
significant role in the effectiveness
of natural gas





~



Baseline 87

-





D









Baseline 91





~



A



o\









~

\ ~



FGR Only
0





\ O





-

D

~\

A

FGR » Natural Gas
•







O











0

\*



0

o

\ °

\

•







• 1





o



•

V

		1



[

I i 1

10 20 30 40 50 60

CO, ppm

70

80

90

Figure 4. Natural gas injection
simultaneously decreases N0X and CO
emissions

5B-103


-------
Table 1

AVERAGE OPERATING DATA -1990/1991 FIELD EVALUATION TESTS









FGR +

NGas

. ¦ I-







At Normal

At









1987

Normal









Baseline

1991



Baseline

FGR OnlV

MSW Input

Baseline



1987

1991

(Average

(Average

MSW Inpu



Test

Test

Data)

Data)

Test

MSW,* lb/h

6,450

7,760

—

6, 500

7,000

Natural Gas, %

0

0

0

14.0

12.4

Total Heat Input,* 10® Btu/h

33.5

40.3

—

39.9

41.9

FGR, %

0

0

9.5

9.5

O
O

Excess Air, %

73

76

54

37

41

Total Flue Gas,* lb/h

44,800

54,100

47,100

45,400

48,500

Steam Flow, lb/h

23,500

28,250

27,670

29,000

30,500

Economizer Exit Temperature, °F

417

425

423

422

422

Precipitator Inlet











o2, %

9.3

10.5

7.6

6.5

5.9

CO, ppm at 12% O2

30

46

47

22

21

NOx, ppm at 12% O2

135

117

70

48

48

~Estimated.


-------
;—— 	TECHNICAL REPORT DATA

¦ > , 1 -_ 1 _ • (Please read Ixsimctions on the reverse before completing)

1. REPORT NO. 2.

EPA-600 /R-92~093b

3. RECIPIENT'S ACCESSION NO.

A. TITLE AND SUBTITLE

Proceedings; 1991 Joint Symposium on Stationary
Combustion NOx Control, Washington, D. C. , March
25-28, 1991. Volume 2. Sessions 4 and 5

5. REPORT DATE _ .

''July 1992	,

6. PERFORMING ORGANIZATION CODE

7. AUTHOR®

Carolee DeWitt, Compiler

8. PERFORMING ORGANIZATION REPORT NO.

9, PERFORMING ORGANIZATION NAME AND ADDRESS

William Nesbit and Associates

1221 Farmers Lane

Santa Rosa, California 95405

10. PROGRAM ELEMENT NO.

11. CONTRACT/GRANT NO.

NA (EPRI Funded)

12. SPONSORING AGENCY NAME AND ADDRESS

EPA, Office of Research and Development
Air and Energy Engineering Research Laboratory
Research Triangle Park, North Carolina 27711

13. TYPE OF REPORT AND PERIOD COVERED

Proceedings; 3/89 - 3/91

14. SPONSORING AGENCY COOE

EPA/600/13

is. supplementary notes AEERL project officer is Robert E. Hall, Mail Drop 65, 919/541-
2477. Volume 1 includes Sessions 1-3, and'Volume 3 includes Sessions 6-8.

is. abstract proceedings document the 1991 Joint Symposium on Stationary Combus-
tion NOx Control, held in Washington, DC, March 25~28, 1991. Jointly sponsored by
EPRI and EPA, the symposium was the sixth in a biennial series devoted to the
international exchange of information on recent technological and regulatory develop-
ments for stationary combustion nitrogen oxides (NOx) control. Topics covered inclu-
ded the significant increase in active full-scale retrofit demonstrations of low-NOx
combustion systems in the U. S. and abroad over the past 2 years; full-scale oper-
ating experience in Europe with selective catalytic reduction (SCR); pilot- and bench-
scale SCR investigations in the U.S. ; increased attention to selective noncatalytic
reduction (SNCR) in the U. S. ; and NOx controls for oil- and gas-fired boilers. The
proceedings are published in three volumes.

17. KEY WORDS AND DOCUMENT ANALYSIS

a. descriptors

b.IDENTIFIERS/OPEN ENDEO TERMS

c. coSati Field/Group

Pollution
Nitrogen Oxides
Combustion
Catalysis
Fossil Fuels

Pollution Control
Stationary Sources
Catalytic Reduction
Noncatalytic Reduction

13 B
07B
21B
07D
21D

18. DISTRIBUTION STATEMENT

Release to Public ¦

19. SECURITY CLASS (This Report)

Unclassified

21. NO, OF PAGES

474

20. SECURITY CLASS (This page)

Unclassified

22. PRICE

EPA Form 2220-1 (9-73)	5B~ 105


-------

-------