EPRI
Electric Power
Research Institute
Keywords:
Nitrogen oxides
Combustion control
Denitrification
Flue gas treatment
Fossil fuel boilers
EPRI GS-7447
Volume 2
Project 2154
Proceedings
November 1991
                    Proceedings: 1991 Joint
                    Symposium on Stationary
                    Combustion NOX Control
                    Volume 2

-------
                              REPORT      SUMMARY
                              Proceedings:  1991 Joint Symposium on Stationary
                              Combustion NOX Control
                              Volumes 1 and 2
                              Proceedings of this 1991 symposium, sixth in a biennial series on NOX
                              control, provide an overview of current NOX control activities. The 66
                              presentations in these two volumes contribute significantly to the
                              development of cost-effective and reliable control systems for fossil-
                              fuel-fired power plants.
INTEREST CATEGORY

Fossil plant air quality
  control

KEYWORDS

Nitrogen oxides
Combustion control
Denitrification
Flue gas treatment
Fossil fuel boilers
OBJECTIVE  To foster an international exchange of information on developments
in NOX control technologies for stationary combustion processes.


APPROACH  EPA and EPRI cosponsored the sixth joint NOX control symposium,
held March 25-28, 1991, in Washington, D.C. Approximately 500 representatives of
electric utilities, equipment vendors, R&D groups, and government agencies heard
66 speakers report on control of NOX emissions from stationary combustion
processes. Reports focused on developments since the 1989 symposium that per-
tain to electric utility power plants and other stationary combustion sources. They
described progress in combustion technologies, selective catalytic reduction
(SCR), and selective  noncatalytic reduction (SNCR).


KEY POINTS
 R&D in the United States to reduce NOX emissions from conventional pulverized-
coal-fired boilers is oriented mainly toward retrofit combustion modifications. Low
NOX burners (LNBs) with or without the addition of overfire air (OFA) continue to
be the preferred approach, both economically and technically,  for  tangentially fired
and wall-fired units. Reburning remains the only widely discussed option for
cyclone boilers.
 Demonstrations of full-scale retrofit LNB and LNB/OFA systems have increased
considerably in the past two years. The trend in these demonstrations is toward
increasing  staging of air and fuel. With controls, emission levels (short-term mea-
surements) for tangentially fired boilers are commonly 0.30 to 0.50 Ib/MBtu, and
those for wall-fired  boilers range from 0.45 to 0.60 Ib/MBtu. Continuously achiev-
able levels would be higher.
 Many presentations suggested that the maximum NOX reduction achievable with-
out significantly affecting boiler operations depends on fuel characteristics, specifi-
cally on reactivity, nitrogen content, and fineness. A number of speakers reported
increases in unburned carbon (UBC) in fly ash when using combustion modifica-
tion techniques to control NOX. The increase depends on the above properties and
the amount of staging. Except for high-reactivity coals, UBC increases ranged
from 2 to 5%.
 SNCR technologies using NH3 or aqueous urea are receiving increased attention
in the United States and Europe. Full-scale tests indicate that NOX emission reduc-
tions up to  50% are possible with NH3  slip below 5 to 10 ppm.  Optimization of
EPRI GS-7447S Vols. 1 and 2
Electric Power Research Institute

-------
reagent mixing at 1700 to 1900F and accurate temperature measure-
ments are critical in obtaining these results.
 Experience with SCR reported by one utility in Germany indicates no
significant catalyst activity decrease, attainment of design NOX reduction
levels (75 to 80%), and control over NH3 slip, usually to less than 1 ppm.
 Retrofit capital costs for SCR on a conventional coal-fired boiler in the
United States are estimated at approximately $100/kW. Operating costs
are estimated at 5 to 7 mills/kWh and are dominated by catalyst replace-
ment costs.


PROJECT
RP2154
Project Manager: Angelos Kokkinos
Generation and Storage Division

For further information on EPRI research programs, call
EPRI Technical Information Specialists (415) 855-2411.

-------
DISCLAIMER OF WARRANTIES AND

LIMITATION OF LIABILITIES

THIS REPORT WAS PREPARED BY THE ORGANIZATION(S) NAMED BELOW AS AN
ACCOUNT OF WORK SPONSORED OR COSPONSORED BY THE ELECTRIC POWER
RESEARCH INSTITUTE, INC. (EPRI). NEITHER EPRI, ANY MEMBER OF EPRI, ANY
COSPONSOR, THE ORGANIZATION(S) NAMED BELOW, NOR ANY PERSON ACTING
ON BEHALF OF ANY OF THEM:
(A) MAKES ANY WARRANTY OR REPRESENTATION WHATSOEVER, EXPRESS OR
IMPLIED, (I) WITH RESPECT TO THE USE OF ANY INFORMATION, APPARATUS,
METHOD, PROCESS, OR SIMILAR ITEM DISCLOSED IN THIS REPORT,  INCLUDING
MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE, OR (II) THAT SUCH
USE DOES NOT INFRINGE  ON OR INTERFERE WITH PRIVATELY OWNED RIGHTS,
INCLUDING ANY PARTY'S INTELLECTUAL PROPERTY, OR (III) THAT THIS REPORT
IS SUITABLE TO ANY  PARTICULAR USER'S CIRCUMSTANCE; OR
(B) ASSUMES  RESPONSIBILITY FOR  ANY  DAMAGES OR OTHER  LIABILITY
WHATSOEVER (INCLUDING ANY CONSEQUENTIAL DAMAGES, EVEN IF EPRI OR ANY
EPRI REPRESENTATIVE HAS BEEN ADVISED OF THE POSSIBILITY OF SUCH DAMAGES)
RESULTING FROM YOUR SELECTION OR USE OF THIS REPORT OR ANY INFORMATION,
APPARATUS, METHOD, PROCESS, OR SIMILAR ITEM DISCLOSED IN THIS REPORT.

ORGANIZATION(S) THAT PREPARED THIS REPORT:
ELECTRIC POWER RESEARCH INSTITUTE
Printed on Recycled Paper
                                                          Proceedings:  1991  Joint
                                                          Symposium on Stationary
                                                          Combustion NOX Control
                                                          Volume 2
                                                          GS-7447, Volume 2
                                                          Proceedings, November 1991
                                                          March 25-28, 1991
                                                          Washington, D.C.
                                                         Symposium Cochairpersons
                                                         A. Kokkinos
                                                         ELECTRIC POWER RESEARCH INSTITUTE

                                                         R. Hall
                                                         U.S. ENVIRONMENTAL PROTECTION AGENCY
Prepared for
U.S. Environmental Protection Agency
Air and Energy Research Laboratory
Combustion Research Branch
Research Triangle Park,  North Carolina 27711

EPA Branch Chief
R. Hall

Electric Power Research Institute
3412 Hillview Avenue
Palo Alto, California 94304

EPRI Project Manager
A. Kokkinos

Air Quality Control Program
Generation and Storage  Division

-------
Electric Power Research Institute and EPRI are registered service marks of Electric Power Research Institute, Inc

Copyright -"  1991 Electric Power Research Institute, Inc All rights reserved

                              ORDERING  INFORMATION
Requests for copies of this report should be directed to Research Reports Center
(RRC), Box 50490, Palo Alto, CA 94303, (415) 965-4081. There is no charge for reports
requested by EPRI member utilities  and affiliates, U.S. utility associations, U.S. government
agencies (federal, state, and local),  media, and foreign organizations with which  EPRI has
an information exchange agreement On request, RRC will send a catalog of  EPRI reports.

-------
                                        ABSTRACT

The 1991 Joint Symposium on Stationary Combustion NOX Control was held in Washington, D.C.,
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
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 two volumes.

-------
                                        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 NOX control. Topics discussed
included the significant increase in active full-scale retrofit demonstrations of Iow-N0x 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.

Angelos 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 two  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
     Session 4A:   Combustion NOX Developments I
     Session 4B:   Large Scale SCR Applications

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

     Session 5A:   Post Combustion Developments I
     Session 5B:   Industrial/Combustion Turbines on NOX Control
     Session 6A:   Post Combustion Developments II
     Session 6B:   Combustion NOX Developments II
     Session 7A:   New Developments I
     Session 7B:   New Developments II
     Session 8:     Oil/Gas Combustion Applications

An appendix listing the symposium attendees is included in both volumes.
                                          VI

-------
                                       CONTENTS


Paper                                                                           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. Pagan  1-15

"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:         LARGE SCALE COAL COMBUSTION I
                      Chair: B. Martin, EPA and G. Often, 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-N0x Burners," T. Lu, R. Lungren, and A. Kokkinos                                 2-19

"Design and Application Results of a New European Low-N0x Burner," J. Pedersen and
M. Berg                                                                         2-37

"Application of Gas Reburning-Sorbent Injection Technology for Control of
NOX and SO2 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
                                           VII

-------
Paper                                                                           Page


   SESSION 3:        LARGE SCALE COAL COMBUSTION II
                     Chair:  D. Eskinazi, EPRI and R. Hall, EPA

"Demonstration of Low NOX Combustion Control Technologies on a 500 MWe Coal-Fired
Utility Boiler," S. Wilson, J. Sorge, L Smith, and  L. Larsen                               3-1

"Reburn Technology for NOX Control on a Cyclone-Fired Boiler," R. Borio, R. Lewis, and
M. Keough                                                                       3'23
"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. Macphai!           3-51
"Update on Coal Reburning Technology for Reducing NOX in Cyclone Boilers," A. Yagiela,
G. Maringo, R. Newell, and H. Farzan                                                3-74

"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             3-99

"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                                3-123
   SESSION 4A:        COMBUSTION NOX DEVELOPMENTS I
                      Chair: W. Linak and D. Drehmel, EPA

"An Advanced Low-N0x 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-N0x 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. Beer, 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
                                           VIII

-------
Paper                                                                            Page


"SO3 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. Speronello,
J. Chen, M. Durilla, 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. Sigling, H. Schmelz,
E. Hums, G. Spitznagel                                                            4B-133
   SESSION 5A:        POST COMBUSTION DEVELOPMENTS I
                      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 5B:        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 NOX 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
                                            IX

-------
Paper                                                                            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. Abbasi,
R. Biljetina,  F. Zone,  R. Lisauskas, R. Dunnette, K. Nakazato,  P Duggan, and D. Linz       5B-89
   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 a 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, and 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

-------
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 II
                      Chair: C. Miller, EPA

"In-Furnace 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 N2O 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. Bisonett and M. McElroy                                                 8-43

"Analysis of Minimum Cost Control Approach to Achieve Varying Levels of NOX 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, Paniculate, and Opacity on the Kahe Unit 6 Low-N0x Burner System,"
S. Kerho, D. Giovanni, J. Yee, and D.  Eskinazi                                         8-85

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

-------
           Session 5A



POST COMBUSTION DEVELOPMENTS I








      Chair:  C. Sedman, EPA

-------
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

-------
               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 I 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 complementarythe
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 TVA 1 MW pilot plant at the Shawnee Steam Station has been operating for
almost four months; baseline tests are 60% complete.  The TVA 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 TVA and NYSEG pilot plants are described as follows:

Hot-side SCR:  TVA 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 710F, and passes through an isolation
damper, a venturi to monitor flow rate, and a 40 kW heater to adjust process
temperature to desired values (680-700F). 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 625F 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 NHs 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 NHs to 2 ppm. Similar to the TVA 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 SC>2 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
550F 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 625F. The
gas then enters the reactor tower, which is identical  to the TVA design with the
exception that four catalyst layers are provided instead of five. After exiting the
reactor, flue gas is cooled by die recuperative heater, and exits the process at
approximately 225F.

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 SC3.

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
NH3/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 TVA 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 SC3 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 NH3/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 scfrrt (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., N20) 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 TVA 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 NOX 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 participate 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 SOs 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 acid  (from high SOs) 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 SC>3 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
SOs 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 TVA 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
star tup/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 400F, 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 SOs 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 625F.  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 803.  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 TVA 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 SOs 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/Utility
              Industry SCR Pilot Plant Program
        NYSEG*
        Niagara
        Mohawk

        CIPS
FUEL

3-4% S


2% S


1% S Oil


3-4% S

 *Post-FGD
FURNACE DESIGN

Pre-NSPS
(Wall-fired)

'79 NSPS
(Wall-fired)

Pre-NSPS
(Wall-fired)

Cyclone
             Table 2. Design Basis of Pilot Plants

        PILOT FEATURE           NYSEG
     Flowrate (scfm)
     Number of Catalyst Layers
     Dummy Layer
     Reactor Temperature (F)
     Inlet NOX (ppm)
     Inlet SO2 (ppm)
     Design Performance
     -  NOX(%)
     -  NH3 (ppm)

     Catalyst Manufacturer
     (all  honeycomb-type)
     Catalyst Pitch (mm)
          2000
            4
           no
           625
           400
           150

           80
            5
           WR Grace
           Englehard
            4
          TVA

          2000
             5
           yes
           700
           600
          2000

            80
             5

            Joy/KHI
             Norton
             6/7
                               5A-9

-------
Figure 1. Installation Arrangement of 1MW SCR Pilot
                        Plant At TVA
                                                        DAMPER
             OLD ELECTROSTATIC
             PRECIPITATOR
             (OEENERGIZED)
             SAFETY SHOWER
             
-------
     Figure 2. Schematic Of Post-FGD SCR
    Pilot Plant at NYSEG's Somerset Station
From
Scrubber
Outlet
(125T)

Return
To Plant
(250F)
 Recuperative Heat
   Exchanger
                     Gas out
                      550 F
                             Electric
                             Heater
   T in : 625F

  	  NH3
 SCR  I
Reactor[
U\J
                              F.D. Fan

-------
       Figure 3. Anticipated Relationship Between
         NOx Removal and Residual NH3 vs. Time
en
   ~ 90%


 NOx
Removal
                            3 months
                            (Baseline)
NH3,
ppm
                                           y

                                           X
                                -.90

                Ammonia/NOx Ratio (moles)

-------
                       Figure 4.  Test Schedule For
                        TVA High Sulfur Pilot Plant
     Activity    6/9 9/9  12/9  3/91 6/91   9/91   12/91  3/92  6/92  9/92  12/92  3/93
en
CO
     1. Start-up
2. Sampling/Analytical Trials
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 +
en
>
Removal on
        9
            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
01
en
           Activity
                 3/91  6/91   9/91  12/91 3/92  6/92  9/92  12/92  3/93  6/93
1. Start-up          ^^
2. Sampling/Analytical Trials
3. Baseline
4. Load-Following
5. Catalyst Activity
6. Trace Species/Particulate
7. Second Baseline

-------
PILOT PLANT INVESTIGATION OF THE TECHNOLOGY OF SELECTIVE
         CATALYTIC REDUCTION OF NITROGEN OXIDES

            Shiaw C.  Tseng,  Wojciech Jozewicz
                   Acurex Coporation
                     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

-------
    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
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 (NOX).   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 440C.
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/NO  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.
                             5 A-19

-------
INTRODUCTION

The  emission  of  nitrogen  oxides   (NOX)  into  the  atmosphere
contributes to the degradation  of air quality as well as to acid
rain and forest damage.'1'   NOX is formed during the combustion  of
fossil fuel.  Part of the oxides come from the thermal  oxidation  of
nitrogen  in  the combustion  air  (thermal  NOX) .     Thermal NOX
generally increases when the combustion temperature is increased.
The remaining NOX comes from the  oxidation of nitrogen-containing
species originally present in the fuel (fuel NOX) .   Compared with
thermal NOX, fuel NOX is not as sensitive  to  combustion temperature
and  depends highly on  the reactant  stoichiometry. (2)    The NOX
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 NOX 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  NOX  reduction  technologies  include
reduced-air preheat,   load reduction, low  excess  air,   flue-gas
recirculation,  overfire  air, deep-air staging,  fuel  staging  (or
reburning),  and various  low NOX  burner systems.'2'   In-furnace
reduction technologies could result in lower combustion efficiency
and  higher  CO emissions. (2)

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 NOX reduction reaction proceeds at the combustion  temperature,
and  the  combustion efficiency  is  usually reduced.

Removal  efficiencies  of  NOX 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.'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 (N02)  at the
catalyst surface to produce nitrogen and water :(5)


       4 NH3 + 4 NO  +  02  	>  4 N2    +6  H20       (1)

       4 NH3 + 2 N02 +  02  	>   3 N2    +6  H20       (2)


In the U.S., there  is little electric utility experience with SCR
NOX control  techniques.   A  demonstration of  this  technology is
being undertaken by Southern Company Services, Inc.(6' but no data
have been reported yet.    Several co-generators  (located mostly in
California)  are  testing  these  technologies; '4)  however,  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. (4/6)

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  combustion 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  440C.    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 combustion 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 N20,
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,
S02 cylinder, and 5% NH3 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
                              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  flame  sensor  is  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. x  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
made  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/min 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% NH3 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  are all thermally
insulated.  The flame temperature and the gas temperatures at the
outlet end  of  the burner and the  inlet  and outlet of 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 150C 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  then 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 5C could routinely be achieved.

As soon  as the targeted reactor temperature was  reached, NH3, 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, 3A, 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  iolite or
silicate of aluminum, magnesium, and iron)  substrate  with titanium
                             5A-23

-------
dioxide  (Ti02)  and subsequently  with vanadium  pentoxide  (V205) .
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  475C  for  4  hours.   A V205-containing  solution
 (diluted  H2S04)  was then dip-coated on  the calcined  substrate,
followed by calcination at  450C  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/m3, 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 N
-------
the reactor  temperature  to the geometric  volume  of the  reactor.
The results  show that when  SV was reduced  from  18,400 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 406C.   The NH3/NO  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 NH3/NO 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. (8)

Effect of S02

Figure  4   summarizes  the  effect   of  S02   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 440C,  and in the  presence of 95 ppm S02 . The
results indicate that the catalyst is less  active  at 440C than  at
353C, 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 0.586  to 2.17.  The
reaction temperature  was 400C.  The space  velocity was  calculated
to be 13,790  hr"1.    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 N20 is a  greenhouse  gas which  has been
blamed  for  both  increasing  the  alobal  temperature  <9'10>  and
destroying stratospheric ozone.'11'12'  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  340C and  SV of 36,500  hr'1.  The
NH3/NO  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 NH3 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  360C 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  (SV),  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  350C.   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/NO  ratio was  below  unity.

 The  NO conversion was found to be temperature insensitive  for  one
 commercial catalyst  tested  at three temperatures,  327,  360,  and
 406C.   For  the  same  catalyst,  flue  gas  S02  was  found to  be
 poisonous, and the poisonous effect of S02 was more severe at 440C
 than at 353C.  At 400C, 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.
 DISCLAIMER

 This  paper has  been  reviewed by  the  Air and  Energy Engineering
 Research Laboratory,   U.S.  Environmental  Protection Agency,  and
 approved for presentation.  The contents of this article should not
 be  construed  to  represent Agency policy nor does mention of trade
 names    or   commercial   products   constitute   endorsement   or
                              5A-26

-------
recommendation for use by the Agency.
REFERENCES

(1)   Mclnnes, R.G. and Van Wormer, M.B., Cleanup NOX emissions.
     Chem. Enqn. 130, 1990.

(2)   U.S. Environmental Protection Agency,  "Control  Techniques  for
     Nitrogen Oxides Emissions From Stationary Sources-Revised
     Second Edition," EPA-450/3-83-002  (NTIS PB84-118330),  1983.

(3)   U.S. Environmental Protection Agency,  "Municipal Waste
     Combustion-Background Information  for  Proposed Standards:
     Control of NOX Emissions, Vol. 4," EPA-450/3-89-27d  (NTIS
     PB90-154873), August 1989, p. 3-9.

(4)   Eskinazi, D., Cichanowicz, J.E.,  Linak, W.P., and Hall, R.E.,
     Stationary combustion NOX control. A summary of the  1989
     symposium. JAPCA. 39(8):  1131, 1989.

(5)   Schonbucher, B., Reduction of nitrogen oxides from  coal fired
     power plants by using the SCR processExperiences in the
     Federal Republic of  Germany  with pilot and commercial scale
     DeNOx plants.  In Proceedings:1989 Joint Symposium on
     Stationary Combustion NOX Control, Vol. 2, EPA-600/9-89-062b
     (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., Hall, 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 PB89-220529),  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.

(9)   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.,   Moe, T.M., and Hansen,
     J.E., Greenhouse effects  due to man-made perturbations of
     trace gases.  Science.  194:  685,  1976.
                               5A-27

-------
(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.  Geophvs.  Res.  86:  7185,  1981,
                             5A-28

-------
Ol
ro
CD
                              Burst

                              Disc
                        Vent
-Hll
                      Burner

                    Management
                                                     Flyash Feeder


                                                      Mechanism
Reactor Housing


with Backheating
                                                                                         Sampling Pump
                                                                                              V
                                                                                          Gas Analyzers
                                                                                                                       Filter
                                                                                                                        Cyclone
                                                                                                                       Collector
                    Figure 1.  Schematics of the bench scale pilot plant facility for testing  SCR DeNOx  catalysts.

-------
        100 -
    i   80 -
     C
     o
     u

     o
         60 -
40 -
         20 -
                          Catalyst  1A
B

T
T
o
= 360
O
= 350
C,
C,
SV =
SV =
7,
18,
650
400
hr.
hr.
J-
1
          0 . 0
                   0.5       1.0
                                      1 . 5
                                               2 0
                         NH  /NO  ratio
       Figure  2.   Performance of Catalyst  1A.
   C
   o
   H
   n
   M
   01
   >
   C
   O
   o


   o
   z


80 -
60 -
40 -

20 -

o -

9"'
_,-'
.-''

,-''
.-'




A
0
B
o
/ o. .


Catalyst 2A


T( C)
327
360
406

i '
- i
SV(hr )
12,500
13,360
14, 000






1 i  -
         0 . 0
                     0 . 5
                                  1 . 0
                                              1 . 5
                      NH3 /NO ratio
Figure 3.  Effect  of temperature on the  performance of
Catalyst 2A.
                         5A-30

-------
                 o
                 z
                    100
                     80-
                     60-
                     40 -
                     20-
                          Catalyst  2A
                       0.0


o
SO 2
(ppm)
None
95
T
(C)
360
353
SV
(hr "-1)
13,360
13,370
0.5          1.0
                                     NH  /NO  ratio
                                                             1 .5
Figure 4.   Effect of SC>2 on  the performance  of  Catalyst 2A.
                   100
               o
               z
                   80 -
                   60 -
                   40 -
                   20 -
                         Catalyst  2A.



                         SO=  95 ppm
         0  O o
           o

o

T
(C)
353
440
SV
(hr-1 )
13,370
14 , 680
                     0.0
                                               1.0
                                                           1.5
                                    NH3 /NO  ratio
   Figure 5.   Effect of temperature on the performance of

   Catalyst  2A in the presence of 95 ppm SO2.
                             5A-31

-------
    100
                                  -a	o-
     80 -
                            H/
     60 -
                               Catalyat  3A
o
z
     20 -
                             SV
                                  340 C
                                 36,500  hr
       0  0
                   0 . 5
                                1 . 0
                                             1 . 5
                      NH /NO ratio
 Figure 6.  Performance of Catalyst 3A.
      100
   O
   z
       80-
   o   60
   H
   n
   8   40
       20-
         0.0
Catalyst
 1A
a 2A
* 3A
T( C)
360
360
340
- 1
SV(hr )
7, 650
13,360
36,500
0.5
                            1 .0
                                      1 .5
2.0
                        NH  /NO  ratio
  Figure  7.   Comparison  of catalyst  performance,
                      5A-32

-------
     Table 1.  Nominal Gas Flowrates
total gas (liters/min)
combustion air (liters/min)
natural gas (liters/min)
NO (ml/min)
S02 (ml/min)
NH3 (ml/min)
120
105
5
120
120
120
Table 2.  Results Of N20 Measurements Over Catalyst 2A.
NH3/NO ratio
0.586
0.968
1.93
2.17
inlet N2O
(ppm)
3.19
2.74
2.86
2.13
outlet N20
(ppm)
3.37
2.75
2.86
1.79
                         5A-33

-------
 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

-------
                       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%
V20s/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% V20s/Ti02 and
8.2% W03 + 4.8% V20s/Ti02, the latter being similar to commercial SCR catalysts.
The addition of WOs 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 V20s/Ti02 catalysts, depending on other trace elements present
on the catalyst surface.
 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 (WO3), 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 60C for 24 hours
and then at 120C for 72 hours.  After drying, the bulk titanium dioxide was crushed
and sieved.  The fraction between 20-32  mesh was collected and calcined at 600C 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 NHjVOs and (NH4)6 H2W12040 in oxalic acid. After
impregnation, the catalyst was dried at 120C for 15 hours and then calcined at 500C
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
V20s/Ti02 or W03-V205/Ti02  catalysts.  The precursor solutions for the alkali
oxides, Li20, Na20, K20, Rb20 and Cs20 were, respectively: LiAc, NaNOs, KNOs, RbAc
and CsAc. For CaO, PbO, As203 and P205 doped catalysts, aqueous solutions of
Ca(Ac)2, Pb(Ac)2, As205 and P205, respectively, were used. The impregnated
catalysts were dried at 120C 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 cm3 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 02).  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 = 	     In(l-X)
                              [NO]0W

where Fo is the inlet molar flowrate of NO, [NO]o is the inlet molar concentration,
W is  the amount of catalyst, and X is the  fractional  conversion of NOX which is
defined as :

                        X= ([NOJin -[NO]out)/[NO]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 WOs in V2d5-based catalysts, two groups of catalyst were prepared to compare
the effects of WOs [i-e., 5% V205/Ti02, compared to 8.2% WOs + 4.8% V20s/Ti02].

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 As20s and
     are both weak poisons.
Figure 1 also shows the change of rate constants vs. M/V (M=metal atoms) over the
W0s-V20s/Ti02 catalyst. With the addition of WOs, the rate constant increased from
10.38 to 13.58 cm3/g/s for the catalysts with no poison doping. Moreover, catalysts
containing WOs always  exhibited higher activities than those without WOs fr the
same amounts of poisons. Figure 1 also reveals  that the addition of WOs to the
catalyst not only increased the catalytic activity,  but also improved the resistance to
alkali oxide poisons. Again, As20s is a weak poison compared to alkali for the WOs-
V20s/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 V20s-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 250C (with 1000 ppm each
of NO and NHs at 15000 hr1).
                                    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 300C than at 350C. The likely cause for the HC1
deactivation is as follows. First, the formation of NH4C1 by the reaction of HC1 with
NHs consumed the reactant NH3. Second, the reaction of HC1 with 205 forming
vanadium chlorides decreased the concentration of the active component of the
catalyst.  Third, the deposition of NFLjCl on the catalyst surface below 340C blocked
the active surface area, which was the reason that the deactivation was more
pronounced at 300C than 350C.

Effects of Sulfur Dioxide

The results of SCR with S02 and without SO2 are listed in Table 1 and Figure 2.
Similar to the case of chlorides, SO2 can be either  a promoter or poison.  Without
the presence of doped poisons, SO2 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 W03/V205
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 SC>2, 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 SC>2  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 V205/Ti02 catalyst.  In order to
obtain an understanding of the role of the W03 in the W03-V205/Ti02 catalyst,
ammonia chemisorption experiments were performed on a series of catalysts with
different K20 dopant amounts.  Figure 3 shows  the ammonia chemisorption
amounts at different doping amounts of K20 or Cs20 over V20s/Ti02 and WOs-
V20s/Ti02 catalysts. The ammonia chemisorption values of W03-V20s/Ti02 and
V20s/Ti02 catalysts were 2.31 and 1.93 cn\3 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-V205/Ti02 series) is always above curve B  (K20-
V205/Ti02 series), and curve C (Cs20-W03-V205/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-V205/Ti02 catalyst was caused by the
addition of W03-  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 W03,1M MAS
NMR experiments were performed.  The results showed that the addition of W03 to
the V205/Ti02  catalyst increased the Bronsted acidity.  The chemical shift increased
from 3.56 ppm for 5% V20s/Ti02 to 4.43 ppm for 8.2% WOs + 4.8% V205/Ti02-
However,  the doping of K20 in either 5%V205 or 8.2% WOs + 4.8% V20s/Ti02
catalyst resulted in  a reduction of Bronsted acidity. The chemical shift decreased
from 4.43 ppm for 8.25 WOs + 4.8% V20s/Ti02 to 3.14 ppm for 8.2% WOs + 4.8%
205 = 0.6% K20/Ti02-
                                    5A-43

-------
For a further understanding of the mechanism of the poisoning effect, the extended
Huckel 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 V205-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 WOs in the proportion of 8.2% in the catalyst  composition
     increases the rate constant for the SCR reaction.
                                     5A-44

-------
(2)   The inclusion of WOs 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 SC>2 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

-------
   16
                0.2     0.4     0.6     0.8
                   M/V   Atomic   Ratio
Figure 1.   SCR activity (expressed as first-order  rate constant) of 5%
V205/Ti02  (solid curves)  and 8.2% W03 +4.8% V205/Ti02 (dashed curves)
doped with different amounts of oxide poisons where M = Li, Na, K Rb,
Cs,  Pb,  As and P, 300C,  NO = NH3 = 1,000 ppm, 02  = 2%, N2 = balance,
GHSV = 15,000 hr.~l
                              5A-46

-------
18-
16-
   0.0     0.2    0.4     0.6     0.8
            M/V  Atomic   Ratio
                           1.0
Figure 2.   SCR activities (expressed as first-order rate constant)
of 8.2% WO. +4.8% V20 /Ti02 with doped metal  oxide poisons.
M = metal,  300C,
                    = 2%, NO =
HO = 8%,  N
                    = 1,000 ppm,
balance,  GHSV  =  15,000 hr"1
                                                 2 = 1,000 ppm,
                                            Solid curves are
without S0_  and H-0, and dashed curves are  with  SO,., and HO.
                            5A-47

-------
 co
X
 O
 E
 c
 O
 E
 4)
 .C
 O


  CO
 I
0.0      0.2     0.4      0.6      0.8


             M/V  Atomic  Ratio
                                                         1.0
      Figure 3.   NH  chemisorption amount over alkali oxides
                 doped  catalysts at 200C.
                 doped 8.2%


                 doped 5% V
                                           + 4.8%
                                         /TiO
                 C   ---   Cs20 doped 8.2% WO  + 4.8% V 0 /TiO

                 D   ---   Cs20 doped 5% V20 /Ti02

                 M   ---   K  or Cs
                               5A-48

-------
                                     Table 1
      Effects of  SO.  on NO Conversion and Rate Constant  (k)  for  SCR at  300C.
Catalyst
5% V205/Ti02 (A)
0.74% CaO/A
0.32% Li 0/A
0.68% As000/A
2 3
8.2% WO +4.8% V205/Ti
1.5% Na20/B
1.1% K20/B
5.1% As20.,/B
Without
Conv . , %
98.0
97.2
91.4
96.7

02 (B) 99.4
83
54
92
so2
k, cm /g/s
10.38
9.49
6.52
9.09

13.58
4.7
2.06
6.71
With
Conv. , %
99.2
99.2
99.1
99.2

99*
95.5*
98*
95.3*
so2
k, cm /g/s
12.82
12.82
12.63
12.80

12.23
8.23
10.39
8.12
Reaction Conditions:   NO = NH  = 1,000 ppm,   SO  = 1,000 ppm (when used),
                      0? = 2%,   HO = 8% (when used),   N2 =  balance,
                      GHSV = 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

-------
                      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 Chestnut 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, andD.A. Rundstrom
Southern California Edison Company
2244 Walnut Grove Avenue
Post Office Box 800
Rosemead, California 91770
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 NOX in the flue gas with NH3 in the presence
of catalyst-coated air heater elements to form N2 and H2O. 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, NOX  reductions, impacts on air heater  performance,  and process  economics.
Projected NOX reductions are presented as a function of  NH3 to NOX mole ratio and NH3
slip.   Air heater performance parameters considered include  heat transfer efficiency and
pressure loss characteristics.
                                     5A-53

-------
                      EVALUATION OF SCR AIR HEATER
                     FOR NOS CONTROL ON A FULL-SCALE
                          GAS-AND OIL-FIRED BOILER
INTRODUCTION
Rule  1135  currently under  consideration  by the South Coast Air  Quality  Management
District (SCAQMD) will require substantial NOX 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
NOX control technology  which can be integrated with other NOX controls, such as  low NOX
burners (LNBs), flue gas recirculation (FOR), and selective non-catalytic reduction (SNCR)
to achieve the required NOX reductions at a relatively low cost.

The  CAT-AH offers a low cost alternative for  installing SCR for NOX 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 NOX on the surface of a
CAT-AH. In addition to offering a NOX 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 NOX 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 coatings.   Development of the  catalyst  materials for NOX 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 NOX 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 NOX 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 NOX concentrations and  air heater gas inlet
temperatures were lower than expected.  During much  of the test, the temperature was
below 540F.  The tests did provide the following results:
                                        5A-55

-------
        With air heater gas inlet temperatures above 570F, NOX reductions of 30
         percent could be attained;

        Erosion of the catalyst from the abrasive fly ash (high S^ content) could
         be controlled.

A permanent retrofit of CAT-AH was not installed because RWE was able to meet the NO,
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:

        NOX 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 NOX 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;

        NH3 slip was limited to  3 ppm.

The NH3 injection did result  in unacceptably high NH3  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 SO2 scrubber.   The catalyst matrix was
supplied by KAH's Japanese Ljungstrom partner. The system was operated for 1,600 hours.
During the test, NOX reductions of 35 percent were achieved and NH3 slip was reduced from
300 ppm to zero.
APPLICATION OF CAT-AH TO MANDALAY GENERATING STATION

SCE 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 NH3 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 NH3 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 1050F.  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 NOX control. The unit is equipped with FOR 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 NOX
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 SO2 to SO3 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.
NH3 Injection System

The NH3  required  for  the  NOX  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 NH3 injection system.  The NH3 injection rate
for each test will be selected based on the inlet NOX concentration, flue gas flowrate, 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
NOX 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 H2O at full load.  Thus,  the increase in pressure loss across the CAT-
AH is expected to be less than 1.4 inches H2O.  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 NOX reductions as a function of NH3 slip.  NOX
reductions are shown for the baseline case of 180 ppm (@ 3% O2) NOX.  For comparison,
NOX reductions also are shown with reduced baseline NOX levels  of 120 ppm (@ 3% O2)
and 50 ppm (@ 3% O2) to reflect the emissions that could be attained with additional control
measures such as LNBs, FGR, and SNCR.

At the design condition  of 740F gas inlet temperature and a baseline NOX  of 180 ppm, a
NOX reduction of 50 percent is projected to maintain NH3 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 NOX reductions  can be obtained with  lower baseline
NOX values while maintaining NH3 slip less  than  10 ppm.  With an inlet NOX level of 120
ppm (@ 3% OJ, NOX removal is projected to be 60 percent.  With an inlet NOX of 50 ppm
(@ 3% O2), NOX removal increases to 76 percent.

NOX reductions are increased as the baseline  NOX is decreased because a smaller amount of
NH3 injection is  required.    The  ratio  of NH3 injected to NH3  slip remains  relatively
constant.  Thus,  to maintain a specified level of NH3  slip (10 ppm), higher mole ratios of
NH3 to NOX can be injected  at lower inlet NOX levels, resulting in a corresponding increase
in NOX removal.

The  figures also show the effect of flue gas inlet temperature on NOX removal.  With an
inlet NOX level of 180 ppm  (@ 3% O2) decreasing the inlet gas temperature  from 740F to
650F is expected to reduce  NOX removal from  about 50 to 43 percent. Increasing the inlet
temperature to 830F would  increase  NOX removal to about 55 percent.   The minimum
temperature to obtain NOX removals is approximately 540F.  KAH recommends 900F 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  NOX concentration;

       Allowable NH3 slip;

       NH3 distribution.
                                      5A-60

-------
Boiler and NH3 injection system operating conditions to be varied to evaluate the above
parameters include:

        Boiler load;

        NH3/NOX mole ratio;

        NH3 injection distribution;

        FGR rate to  hopper;

        Excess O2.

Boiler load affects several  operating variables simultaneously,  including inlet  flue gas
temperature,  inlet NOX concentration, and flue gas volume (space velocity). The NH3/NOX
mole ratio controls the availability of  NH3, but is limited by  the  allowable NH3 slip.
Typically,  an NH3/NOX 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 NOX removal.  Proper NH3 distribution will be verified by
traverse  measurements  of NH3 and NOX  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 O2 are variables that affect flue gas temperature, flue gas volume, and inlet NOX
concentration.  The impact of these variables on  air heater performance  also will be
determined.
INTEGRATED APPROACH TO NOX EMISSIONS CONTROL

The CAT-AH is well  suited for use with other NOX 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 NOX removal rate increases with decreasing air
         heater inlet NOX concentration.

Figure 5  shows  an example of how a CAT-AH can be combined in a system of multiple NOX
control technologies to achieve ultra low  NOX emissions levels.  Based on an uncontrolled
NO, emission level of 200 ppm (@ 3% O2) for a gas-fired utility boiler, it is conservatively
estimated that NOX emissions  can be reduced by 60 percent to 80 ppm (@  3% O2)  using
combustion modifications such as LNBs,  FGR and/or staged combustion.  The use of a
SNCR process could provide a NOX reduction of at least 25 percent, to 60 ppm (@ 3% O2).
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 NOX reduction of at least 67 percent, resulting in a NOX emission level of
20 ppm (@ 3%  O2), while limiting NH3  slip to 10 ppm.

Thus, the combined use of selected NOX emission control technologies can provide overall
NOX 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 $3.00/MWh for a
catalyst life ranging  from two years to five years.
CONCLUSIONS

Testing of the CAT-AH system at the Mandalay 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 NOX emissions;

        Requires no modifications to existing equipment and has minimal impact on
         performance;

        Provides substantial NOX reductions;

        Can be used to control NH3  slip from  SNCR or upstream catalyst processes;

        Can be integrated with other NOX control technologies to provide ultra low
         NOX emissions.

More quantitative conclusions will be developed following the evaluation at the Mandalay
station.
                                       5A-62

-------
Figure 1.  Mandalay Unit 2
          5A-63

-------
NH 3 slip, ppm
                        Flue Gas
                        Inlet Temperature
                              740 F
      10     20    30     40    50     60

         NOX  removal efficiency, %
     Figure 2.  NO x Curve for Inlet NO x  180 ppm
70
                   5A-64

-------
NH 3  slip, ppm
                          Flue Gas
                          Inlet Temperature
                               740 F
     10    20   30    40    50   60    70
         NOX  removal efficiency, %
     Figure 3. NO x Curve for Inlet NO x -  120 ppm
                   5A-65

-------
NH3  slip, ppm
                              Flue Gas
                              Inlet Temperature
                                   740 F
     10    20    30    40   50   60   70   80
         NOX  removal efficiency, %
      Figure 4.  NOX Curve for Inlet NO x  50 ppm
90
                   5A-66

-------
en
>

05
-vl
             250
             200
              150
100
               50
                0
                  NOX, ppm (at 3% O2 )
                       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
     RWE
     STEAG

Full-Scale Retrofits:

     RWE
     BEWAG
     BKG
Frimmersdorf "C1
Meppen
Voerde
Frimmersdorf "F"
Lichterfeld (Berlin)
Marl
Brown Coal
Gas
Bituminous Coal
Brown Coal
Oil
Bituminous Coal
                                     5A-68

-------
TABLE 2.  ORIGINAL AIR HEATER DESIGN OPERATING CONDITIONS
   Location
Flow,
Mlb/hr
Temperature,
     F
Pressure,
 In. H2O
   Air Inlet

  Air Outlet

   Gas Inlet

  Gas Outlet
 860.5

 790.0

 850.0

 920.5
     80

    646

    740

    267
(uncorrected)
    259
 (corrected)
  19.0

  15.1

  4.25

  1.00
                               5A-69

-------
                TABLE 3. NH, INJECTION DESIGN CONDITIONS
Location:




Flue Gas Flowrate:
Flue Gas O2 Content:
NOX Concentration:




NOX Flowrate:




NH3/NOX Mole Ratio:




NH3 Flowrate:




NH3 Carrier Air Flowrate:
Mandalay Unit 2 North Air Heater




850,000 Ib/hr




2.0% O2




180 ppm @ 3%  O2




212 Ib/hr




0.92




72 Ib/hr




4,000 Ib/hr
                                     5A-70

-------
                N2O FORMATION IN SELECTIVE NON-CATALYTIC
                        NOV REDUCTION PROCESSES"
                                 L. J. Muzio"
                              T. A. Montgomery
                               G. C. Quartucy
                       Fossil Energy Research Corporation
                            23342 C South Pointe
                            Laguna 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

-------
                     N2O FORMATION IN SELECTIVE NON-CATALYTIC
                              NO, REDUCTION PROCESSES
ABSTRACT
NOX  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 NOX reductions are not achieved at the expense of producing other undesirable species.
One  possible concern  is the production of N2O from NOX reduction processes.  The current work
addressed potential N2O production from selective non-catalytic NOX reduction (SNCR) processes using
ammonia,  urea  and cyanuric acid injection.  Previous work with SNCR  processes has shown  that
ammonia injection produces minimal N2O emissions, while cyanuric acid injection has, under certain
conditions, almost quantitatively converted NOX to N2O. 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 N2O production of these processes
over  a range of process parameters.  The data show that SNCR processes were all found to produce
some N2O as a byproduct.  Ammonia injection produced the lowest levels of N2O while cyanuric acid
produced the highest levels. N2O 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

-------
                     N20 FORMATION IN SELECTIVE NON-CATALYTIC
                              NO, REDUCTION PROCESSES
INTRODUCTION

The mean global concentration of N2O 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, N2O 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, N2O 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 N2O emissions from fossil fuel combustion have previously been reported to be equivalent to 25-
40% of the NOX levels (Hao, et al, 1987; Castaldini, 1983). However, recent tests have shown these
measurements to be  in error, most of the N20  having been formed as an artifact of the sampling
procedure (Muzio and Kramlich, 1988). Full-scale tests using an on-line N2O analyzer have confirmed
that direct emissions of N2O from fossil fuel-fired boilers are less than 15 ppm.  Further, N2O levels do
not generally correlate with the NOX 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 N2O emission levels.  This
paper describes experimental and  kinetic modeling studies of  selective non-catalytic NOX reduction
(SNCR) processes and the potential by-product N2O emissions  therefrom.

Selective non-catalytic NOX 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 NOX reduction technologies include:
        Ammonia (NH3) Injection, (Lyon, 1976)
        Urea (NH2CONH2) Injection, (Muzio  and Arand, 1976;  Mansour, et al, 1987)
        Cyanuric Acid ((HNCO)3)/Cyanic Acid Injection, (Perry,  1988)
                                         5A-74

-------
Figure 1 shows the possible major chemical paths leading to the reduction of NOX with these species
and possible paths leading to the formation and emission of N2O 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 N2O could be a product of the NOX 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 N2O 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 N2O as follows:

        OH + HNCO -> NCO + H2O
        NCO + NO - N2O + CO

Previously reported results with ammonia injection (Lyon,  1976; Muzio and Arand, 1976) indicate that
very little N2O is formed during the reduction of NOX. 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)3) or isocyanic acid injection (HNCO) indicate N2O
to be  a major intermediate species and product (Siebers and Caton, 1988).

The detailed reaction chemistry of urea (NH2CONH2) with NOX 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 N2O formation.  On the other hand, another decomposition path may be
2NH2  + CO, in which case little N20 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 N2O is
a  by-product of  SNCR processes,  and 2) determine  the  process parameters  and underlying
mechanisms leading to N2O 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
         NHs
 NHs+OH
NH2+NO
NH2CONH2
                                      I
NH3+HNCO
     (HNCO)3
     3HNCO
       L
                                                     f
                                             HNCOfH - NH24CO
                                                        HNCO+OH *- NCCM-H2O
                                                                 t
                                                         NCCMSD-  N2OCO
                                         N20+OH
                                          N2OfH
             Figure 1. Major Paths for Selective Non Catalytic NOX 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 N2O 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 (NH3, urea, cyanuric acid), and SO2 concentration.   Baseline conditions selected
for the modeling runs were an SR of 1.1 using a CH4/air flame, an initial NOX (NOJ concentration of
700 ppm and a molar nitrogen to NOX  (XN/NOX) ratio of 1.0.  There was no SO2 present during the
baseline runs.

The combustion products were produced by running  the model as an adiabatic 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 NOX level.

For all reducing agents, the injection temperature was varied from 900K to HOOK at 100K intervals.
A mixing time of 10 ms was used to model the addition of NOX reducing agents.

The decomposition routes of complex reducing agents such as urea and cyanuric acid are not currently
well understood.  This leaves some uncertainty as to how these materials should be treated during
modeling.  Cyanuric acid (HNCO)3 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:

                                 - 2NH2 + CO
             NH2CONH2          -  NH2 + H + HNCO
                                 -  NH3 + HNCO

Previous calculations (Chen, et al., 1988; Muzio, et al, 1989) have shown that only the latter product
set  (NH3 + HNCO) resulted in an acceptable prediction of NOX reduction with urea  injection.

Figure 2 (a,b) shows the calculated NOX reductions  and N2O 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-NOx molar ratio (N/NO)

                                          5A-77

-------
c
o
'
x
O
                                                            H Ammonia (NH3)
                                                            EJ Urea (NH3+HNCO)
                                                            D CyanuricAcid(HNCO)
           900    1000    1100    1200    1300    1400
                         Temperature, K
        a) Calculated NOX Reduction versus Temperature
80
P

-------
of 2.0.  The calculated NO reductions for NH3 injection are 97% with peak removals occurring at 1200K.
Calculated NO reductions for urea injection  (NH3 and HNCO) show peak removals at 1300K with the
peak removals somewhat less than ammonia.  For cyanuric 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 NH3 or urea.

Calculated N2O emissions corresponding to the NOX reductions in Figure 2a are shown in Figure 2b.
As seen in Figure 2b with NH3 injection, very little N2O 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 N2O levels of 90 and 68 ppm, respectively, at
1200K. For all chemical additives, essentially no N2O is found at temperatures above  1300K.

Figure 3 replots the results in Figure 2a,b showing the calculated N2O levels as a fraction of the NOX
reduced. For ammonia injection, the calculations indicate less than 1% of the NOX is converted  to N2O.
For urea injection, the calculations indicate a  peak  NOX to N2O conversion of 12% at 1200K. For
HNCO, the calculations indicate that over 50% of the NOX is converted to N2O 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 1200K. 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 SO2, 2)
combustion product stoichiometry, 3) initial NOX 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 al, 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

-------
U.fc)
0.5
0.4
X
0
z
0 0-3
C\J
0.2
0.1
n n

-
_


-

-
-








|








i



1













j








-
.


-

-
-
                                                   Ammonia (NH3)

                                                   Urea (NH3+HNCO)

                                                   Cyanuric Acid (HNCO)
 627     727    827    927   1027   1127

              Temperature, C
 Figure 3.  Chemical Kinetic Modeling Results: N2O Emissions
Normalized by the NO, Reduction. NO,=700 ppm, N/NO=2.0.
                         5A-80

-------
E
Q.
Q.
C
g
"w
(/>

E
LU

CO
I
    2000
1500
1000
     500

Ammonia (NH3)

Urea (NH3+HNCO)

Cyanuric Acid (HNCO)
           900
              1000   1100    1200   1300    1400


                    Temperature,  K


     a)  Calculated NH3 Emissions versus Temperature
kiUUU

|_ 1500
Q.
c
O
W)
w 1000
E
LU
O
O
^ 500
n


-




-




^
k\\\\\\V


















^
//


















V.
\\X\\\\X!

















L\\\\\\N











-




H Ammonia (NH3)
E3 Urea (NH3+HNCO)
D Cyanuric Acid (HNCO)









          900    1000   1100    1200   1300


                        Temperature,  K
                                            1400
        b) Calculated HNCO Emissions versus Temperature
                     Figure 4.  Chemical Kinetic Modeling Results
                           NO, = 700 ppm, N/NO = 2.0
                                      5A-81

-------
                                         BUHNER FLOW SYSTEM



/ \
7eom COMBUSTION
AND COOLING SECTION
EIOHT CONCENTRIC
COOLINQ PROBE |_
PORTS 	 ^
r

OAS AND SOLID
INJECTION PORT l_
; 	 + ^r
cm ADDITIVE INJECTION
SECTION

""""13

a
a

THERMOCOUPLE 1 1
PORT 	 ^-
a

a

a
Ocm TEST SECTION
a

a

3

u

a

[~[

i |

i| 	 	
/ TO
BAQHOUSE
	
	











!""

1
1

!

111





:;::,









,







BUHNE
r
r







































i
[
j

/



\
\
-^






































11 cm

1 I
7^
































-\
\<-

\


^

/
- a
k-

































T







cm
11cm


















































,,,]









|













:
I
	 ' 	  1 '
	 . 	 1 i
. :
1 	 1 i
{

1 ROTAUETERS
i i v
MAIN AIR ffl ffl ffl /METERING
0. 126-140 8LPU M [J Li/ V*LVE8
j BYPASS 	 f f t_DopEONH3
^^1 |
n ^^H NATURAL
^^^ O. B 6- 4 SLPU
28 ?"" / THEBUOCOUPtE PORT
P^JL/
-| ~/^ INJECTION PORT
.4 	 SUCTION PTHOMETEH


_j
~i

_i

C


30cm


r

L

c



_i

*

_i
4 	 SAMPLE PORT
r|
. DILUTION
* AIR
U
Figure 5.  Pilot-scale Combustor Facility
               5A-82

-------
Gas samples  taken  at  the  combustor exit were analyzed for  NO/NOX>  N2O,  CO,  CO2  and O2.
Continuous N2O measurements were made using an NDIR technique developed by Montgomery, et al
(1989).  NH3 measurements were made using wet chemical techniques.

The pilot-scale tests  investigated the effect of temperature, chemical  injection rate, and initial NOX
concentration on N2O production for selective non-catalytic  NOX reduction with NH3 (gaseous), Urea
(both a pulverized solid and an aqueous solution), and Cyanuric acid (pulverized solid).

Figure 6 shows both NOX reduction and N2O 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, NH3
exhibited the highest NOX reduction with a peak removal of 88 percent at a temperature of about 930C.
The peak NOX removal  with  urea was  82 percent  at a temperature of 980C.  The calculations
discussed previously yielded peak NO removals for NH3 and urea  at nominally 927C (1200K) and
1027C (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; NOX removals increased as the temperature  increased (at 1100C,  the NOX 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 N2O emissions data as a function of temperature. The data show
that ammonia injection resulted in the lowest N2O emissions, while urea injection provided the highest
in terms of absolute concentration.  With ammonia injection, N2O emissions peaked at 877C, while
injecting either urea or cyanuric acid resulted in peak emissions at 977C.

The N2O data from Figure 6 have been replotted in  Figure 7.  In Figure 7, the N2O  is shown  as a
fraction of the  NOX reduction  (e.g., the  fraction  of the  NOX reduced that is converted to N2O). These
results indicate that for NH3 injection, 2-5 percent of the NOX  reduced appears as N20 in the products.
For urea injection, 10-25 percent of the NOX reduced shows up as  N2O. Cyanuric acid exhibits the
highest conversion to N2O with up to 40 percent of the reduced NOX appearing as N2O in the products.

The calculated values shown in Figure 3b are in qualitative agreement with  the pilot scale results.
Experimentally, NH3 exhibits somewhat higher levels of N2O than the calculations.   Likewise, the
conversion of NO to N2O with urea is higher experimentally than calculated. Finally, the calculations
indicate  virtually no  N2O at temperatures of 1027C (1300K) and  above,  while experimentally the
window for N20 emissions is  broader.
                                          5A-83

-------
  >

 t>
 (D
 DC
 X
 o
100


 90


 80

 70


 60


 50


 40


 30


 20


 10


  0
            827    877    927    977   1027   1097


                         Temperature, C


         a) NOX Reduction versus Temperature
                                                            H  Ammonia (g)

                                                            E3  Urea(s)

                                                            D  Cyanuric Acid (s)
E
Q.
CL

C
O

"o
O

QL

O
CM
^uu
180
160
140

120
100
80
60
40
20
n
-
-
-

-
-
-







I







I













1
1
I
i
i

-





I
^
1
//
%
1
1
-
-

-



i


-

-
                                                          M  Ammonia (g)

                                                          0  Urea (s)

                                                          D  Cyanuric Acid (s)
           827    877    927    977    1027   1097


                         Temperature,  C

        b) N?O Production versus Temperature


        Figure 6.  Pilot-scale Test Results, NOX Reduction and N2O Production
                          NO, = 700 ppm, N/NO = 2.0
                                 5A-84

-------
X
o
o
c\j
     0.50
     0.40
0.30
     0.20
     0.10
     0.00
                                                    M  Ammonia (g)
                                                    0  Urea (s)
                                                    D  Cyanuric Acid (s)
            827    877    927   977    1027   1097
                         Temperature (C)
        Figure 7.  Pilot-scale Test Results, Conversion of NOX to N2O
                     NO, = 700 ppm, N/NO = 2.0
                              5A-85

-------
The  laboratory data show somewhat higher N2O  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 al, 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 tests. The
data show that regardless of the reagent injected, NH3 slip decreased as temperature increased. NH3
injection resulted  in the lowest measured NH3 slip. Cyanuric 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)3 was not measured as NH3,
thus  indicating that NH3  measurements reflected only NH3 emissions.

Pilot scale results at a lower initial NOX level of 300 ppm are shown in Figures 9 and 10.  Figures 9a
and 9b show NOX reduction and N20 emission, respectively, as a function of injection temperature. The
NOX  reduction data show that  NH3 injection provided peak removals of 88 percent at about 980C.
Similarly, urea injection  resulted in a maximum NOX reduction  of  57 percent at 930C. As with the
higher initial NOX  level, cyanuric acid injection did not exhibit any peak NOX removal over the range of
temperatures investigated. The maximum reduction of 52 percent was measured at 1100C. With the
exception of NH3  injection, maximum removals were lower for the tests performed with 300 ppm initial
NO than  those performed with  700 ppm initial NO.

N20  emissions (Figure 9b) show that at an initial NO level of 300 ppm, cyanuric acid injection yielded
the highest N20 emissions; 69  ppm at about 980C.  N2O emissions resulting from urea injection also
peaked at 980C, at 43 ppm. When injecting the NH3, peak N2O emissions of 21  ppm were measured
at 880C. 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  880C for NH3 and 980C for urea and cyanuric acid. Again, cyanuric acid was shown to
provide the highest conversion of NOX to  N20. 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 N20 conversion of 25 percent at this lower initial NOX level.
                                          5A-86

-------
E
Q.
Q.

W~
C
o
E
LU

CO
I
      1000
       800
600
400
       200
                                                         H  Ammonia (g)


                                                         0  Urea(s)


                                                         D  Cyanuric Acid (s)
                 877
                           977
1097
                           Temperature,  C
               Figure 8. Pilot-scale Test Results, NH3 Emissions
                         NO, = 700 ppm, N/NO = 2.0
                                5A-87

-------
c
g

o
13
T3
CD
cc
x
O
100


 90


 80


 70


 60


 50


 40


 30

 20


 10


  0
                          fl
              877        927        977


                         Temperature,  C


           a)  NO, Reduction versus Temperature
                                           1097
                                                              Ammonia (g)

                                                              Urea(s)

                                                              Cyanuric Acid (s)
E
Q.
CL

C
g

o
3
"O
o
O
CM
IUU
90

80
70
60
50
40
30
20
10
t w
0

-

:
-
-
-
-













1
SI











////
m











~].

877









|| |
||



1
i





927










1

H
p
i
I




























-

"
-
-
-
-




S Ammonia (g)

EJ Urea(s)
D Cyanuric Acid (s)







977 1097
                         Temperature, C


          b)  N.,0 Production versus Temperature
Figure 9. Pilot-scale Test Results, NO, Reduction and N2O Production
                   NO, =300 ppm, N/NO = 2.0
                           5A-88

-------
X
O
      0.50
      0.40
      0.30
      0.20
      0.10
      0.00
             II
              877
                            %
             I
927        977

Temperature (C)

                                  H  Ammonia (g)

                                  0  Urea(s)

                                  D  Cyanuric Acid (s)
1097
            Figure 10. Pilot-scale Test Results, Conversion of NOX to N2O
                          NO, = 300 ppm, N/NO = 2.0
                                 5A-89

-------
 The effect of the chemical injection rate on the NOX reduction and N2O emissions are shown in Figures
 11  and 12 respectively.  As expected, Figure 11 shows increased NOX removals with increasing
 chemicaladdition rate (N/NO ratio) for all three chemicals and at all temperatures tested.  As seen in
 Figure 12, the amount of  chemical injected, N/NO ratio, has little impact on the conversion of NOX to
 N2O. At the lower temperature conditions, less than 927C, there appears to be some increase in N2O
 production and conversion of NOX to N2O as the N/NO ratio increases.  The decrease in NOX to N2O
 conversion with increasing N/NO ratio for cyanuric acid injection at 827C is due to the increase in NOX
 reduction as the N/NOX ratio increases, rather than a decrease in N2O emission levels.

 DISCUSSION
 The results of this study showed that N2O can be a product of selective non catalytic NOX 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 N2O,  calculated as the  sum of the
 rate of  destruction of N20  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 1015 Btu (natural
 gas, oil and coal). An order of magnitude estimate of the annual N2O from SNCR processes can be
 made with the following assumptions (Eskinazi, 1991):
         Average utility NOX emissions are 0.7 Ib  NOX/106 Btu
         SNCR processes  result in 50% NOX reduction
         The various SNCR chemicals convert a fraction of the NOX to  N2O:
         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 NOX converted to N20 and the  fraction 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 N2O
 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 N2O (as N)
 is a conservative estimate of the contribution.

 CONCLUSIONS
 Both the chemical kinetic calculations and the pilot  scale test results show that N2O can be a product
of some of the SNCR processes.  NH3 injection yielded the lowest  N2O levels; typically less than 4%
of the NOX reduced.  With  cyanuric acid injection, conversion of NOX to N2O ranged from 12  to 40%.
The NO to N2O conversion with urea injection ranged from 7 - 25%.  The conversion of NO to N2O did

                                         5A-90

-------
 c
 g
 o
 a
"O
 0)
CC
 X
O
 c
 o
 o
CC
x
O
 c
 o
 '-'
 o
 ZJ
 ~D
 Q)
 CC
 x
 O
 100
  90
  80
  70
  60
  50
  40
  30
  20
  10
   0
                                                             N/NO
                                                           H 2.0
                                                           0 1.0
                                                           D 0.5
            827
              877
                            927
977
1027   1097
                         Temperature, C
          a)  NO, Reduction vs. Temperature, Ammonia Injection
100
 90
 80
 70
 60
 50
 40
 30
 20
 10
  0
                                                           N/NO
                                                          H  2.0
                                                          0  1.0
                                                          D  0.5
           827     877     927    977    1027
                       Temperature, C
                                            1097
         b)  NO, Reduction vs. Temperature, Cyanuric Acid Injection
           827    877    927    977    1027    1097
                         Temperature, C

         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

-------



X
0
fj
o
C\J
Z

0.5

0.4

0.3

0.2
0.1
n rt



-

-

'
"__ ^.m^.m^.m^.^
N/NO
M 2.0
E3 1.0
D 0.5




            827     877     927    977    1027   1097
                         Temperature,  C
          a) Conversion of NO, to N2O, Ammonia Injection
 X
O
Z

-------
w
c
Q  g
   0)
w
       1.0
       0.8
       0.4
       0.2
                       Urea
           NH   .'
              3 /
       o.o "?-'
                                             % of Fossil Fuel /+7.0
                                               Using SNCR /
                                                      100 %
        (HNCO)-'
             S 3
                                                              6.0
                                                               5.0
                                                             - -4.0
                                                      50 %
                                                      25 %
                                              		10 %"
                   1 0
20
30
40
50
                                                                      c
                                                                     .2
                                                                     B
                              Q- (/]

                              0|

                              z o>
                              -= <
                                                                     _ co
                                                             3.0    
                                                               2.0
                                - .1.0
                     Percent Conversion of NOx to N20
          Figure 13. Potential Annual N2O Emissions SNCR Processes
                 (Bars show annual N2O 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/NOX ratio) or the initial level
of NOX (over the range tested, 300 -700 ppm).

The experimental results are consistent with the chemical kinetic calculations suggesting that N2O
production with  SNCR processes occurs primarily due to the formation of NCO which subsequently
reacts with NO to form N2O,
                                        5A-94

-------
REFERENCES

Castaldini, C., et al., Environmental Assessment of Industrial Process Combustion Equipment Modified
for Low NOX 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 Section/The Combustion Institute Fall
Meeting, Dana Point, California, October 1988.

Chen, S.L., Cole, J.A.,Heap, M.P., Kramlich, J.C., McCarthy, J.M., and Pershing, D.W., Advanced NOX
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 Reburning 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. Geophys. 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 Section/The 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,O 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 NOX Control, Washington, D.C.,  March  1991, (this symposium).

Lyon, R. K., Longwell, J. P., "Selective Non-Catalytic Reduction of NOX 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 NOX Removal," Proceedings of the
1987 Joint Symposium on Stationary Combustion NOV Control, Vol. 2,  EPRI CS5361, 1987.

Montgomery, T.A., Muzio, L.J., Samuelson, G.S., "Continuous Infrared Analysis of N2O in Combustion
Products,"JAPCA, V. 39, No. 5, pp. 721-726, 1989.

Montgomery, T.A., et  al.  "N2O Formation from Advanced NOX Control Processes (Selective Non-
Catalytic Reduction and Coal Reburning), 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 N20 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 SectionAThe 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.

Teixeira, D.P., et al, "Widening the Urea Temperature Window", presented at the 1991 Joint Symposium
on Stationary Combustion NOX Control, Washington, D.C., 1991.

Tirpak, D. A., The Role of Nitrous Oxide (N2O) in Global Climate and Stratospheric Ozone Depletion,
Symposium on Stationary Combustion Nitrogen Oxide Control. V. 1, EPRI CS-5361, EPA Contract 68-
02-3994, WA93, 1987.

Weiss, R. F., The Temporal and Spatial  Distribution of Tropospheric  Nitrous Oxide, J. Geophys. 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.1.c.
    Ratcliffe  Technology  Centre
          Ratcliffe-on-Soar
             Nottingham
              NG11 OEE
           United Kingdom

-------
                   TAILORING AMMONIA-BASED  SNCR  FOR
                 INSTALLATION  ON  POWER  STATION BOILERS
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.
                                 5A-99

-------
   TAILORING  AMMONIA-BASED SNCR FOR INSTALLATION  ON POWER STATION
                               BOILERS

BACKGROUND
The  operators of  utility boilers  are  currently  seeking to develop
cost-effective  methods  for  controlling  NOX 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
NOX, 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 503  or  HC1  cause  both  fouling  and
low-temperature corrosion.

It is thus essential that, on industrial plant,  the point at which a
NOX  control  agent  is injected  is  matched to the  optimum temperature
for  the  de-NOx  process so that  acceptable NOX  reduction is obtained
without  significant ammonia slip.  It is  also  important  to match the
concentration of  agent  to  that  of NOX since  both NOX 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 NOX concentration.

In  order  to  engineer a  viable power  plant implementation  of  SNCR
technology,  it  is  necessary to  characterise the  behaviour 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 Rig (CADR)
(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 400  ppm,  which  is  typical  of the
emission level of a UK 500 MWe  wall-fired furnace fitted with first-
generation  combustion NOX  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
Marchwood Laboratories.  Total flue gas flow through the system was up
to 2000 Nm3/s.

Furnace Modelling Facility (FMF)

When used for this work,  the facility was configured as a l/5th scale
model  of  half   a  660   MWe  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%C>2)  .

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  adjusted by means of  control  valves. The  injectors
were mounted  downstream  of  a bank of vertically mounted, ceramically
shielded cooling tubes which had an array of Pt/Ptl3%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/NOX  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  1020C.  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/NOx 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  NOX  and ammonia emissions as the NH3-.NOX  ratio
is varied at  constant temperature (1074C)  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:NOX ratio was in use.

Effect of Initial NOX Concentration

A  series  of runs  was  carried out  (at 1093C, 1:1  NH3/NO ratio and
2.1%  02)  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-103

-------
decreased to 31% at an initial NOX level of 175 ppm.

Effect of Oxygen Concentration

The overall stoichiometry of ammonia-based SNCR,

4NH3 + 4NO + 02 = 4N2 + 6H2O

suggests  that  oxygen   concentration   might  affect  NOX  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 1000C, the reduction is effectively  independent of
oxygen  concentration  but that  at  908C,  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
NOX 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.' (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
NOX  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 attainable 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 NOX reduction of varying CH4:NH3 ratio  is summarised in
fig. 7- It  is apparent  that,  at the low  temperatures  (735 and  800C),
NOX reduction increases monotonically with  CH4:NH3  ratio.  At higher
temperatures  (865,915C)  ,  the curves begin to exhibit maxima beyond
which  conversion  falls as  methane  increases.  At  965C  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 1030C to 916C.  When  a
1:1 CH4iNH3 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 915C. 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

-------
150C in the absence of methane  but  around  200C  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-NOx process but that, at higher ratios, it
can  actually  cause NOX  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  NOX  level  of 341  ppm and  a temperature of 908C  at  the
injection point. At this low temperature, there was relatively little
interaction  between NOX  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 NOX reduction  (-60%)  but still gave
ammonia  slip of  over  60  ppm.  The  addition  of  methanol  at lower
temperature   (850C)   has  little   effect   on  NOX   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 870C. It
was  discovered  that,  at  low   injector  momentum,   helium  did  not
penetrate  to  the  centre of  the  duct.   As the  jet  momentum  was
increased,  a  peak of helium 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 NOX 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 NOX reduction is close to  1020C. 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 800C.
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 PowerGen'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  NOX  reductions at  conditions typical  of  the convective
sections of  industrial  p.f. furnaces.

The  effects of temperature, NH3:NOX 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-108

-------
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 NOX Reduction Processes
             Using -NH and  -CN  Compounds in Conjunction with Staged
          Addition', 22nd Symp.  (Int.)  on Combustion., The Combustion
         Institute,  (1989).

  2.        A.M. Dean,  J.E. Hardy,  R.K.  Lyon,  19th  Symp.  (Int.)  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  13-15th, 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

-------
7.       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

8.       D.  Wenli, K.  Dam-Johansen,  K. Ostergaard,  'Kinetics  of the
         Gas-Phase Reaction between NO, Ammonia and 2',  Preprint -
         40th Canadian  Chem.  Eng.  Conf., Dalhousie Univ.,  Halifax,
        Canada July,  1990.
                               5A-110

-------
cn
Propane
burner
N

T~
Coal burner A
secondary air
port
fi
A
h
ir
]


Mfllidiic ^"^
(nat. gas)

Combustion *
a a a a
Chamber
	 -
mmonia to
?rm NO*
> flue gas
0

CD
"<^7
p D a 9
React ant i_
injection
at port CSI
i
	

Flue gas Duct
o a
I
- 	  | Rig gas Am
| analysis ana
port pc
1 Secondary injection
For slip control
(at port CS4
Q5J Brooks Ho
control de
Liquid reactants
from syringe pump
nonia
ysis
>rl
w
vice
        Figure 1
                      COAL ASH DEPOSITION RIG AND REAGENT INJECTION SYSTEM

-------
                                                                      11,12 Fixed gas sampling positions
                                                                                              Sick
                                                                                              dust
                                                                                              monitor
                                                                                        Flue gas
                                                                                        recirculating
                                                                                        duct
                                                                                                         f D Ian
 
1

 31 Of> o o F=
"731 , i i cr
m 6' 
-------
              1.0
              0.8
              0.7
             0.6
         -   O.i
             o.i
             0.2
                                 900
                                                 1000               1100
                                            Reaction  temperature, *C
  Figure 4
          z
             100  -
Figure  5
                         OENO  ON CADR  AMMONIA SLIP  1071 CELSIUS   2.1% C>2
                                              5A-113

-------
            1)0
        o
        Z   JO
         u   20
         
                  +  IOOOT

                  D   -(08T
Figure 6
                                       2                       *
                                       Per cent onygen in Hue gas
                        DKNO  ON CADH   PHASE II LI'I CCT OF OXYCUN AT 908 AND 1000'C
                                     O.S         1.2        1.6
                                       Mrthane / ammonia mole rjlio
Figure 7
METHANE (NO  r;J=R:X;T  ON AMMONIA OENO
                                              5A-114

-------
         o
              700
Figure 8
             EFI-KC'.T-OF METHANE ON NH} DENO^ WINDOW-' NHj/ INITIAL HOf 1.0  OXYGEN 2.
-------
Ul
>
                      3:
                      2
                          350
                          300
                         250
                         200
O



!   150
                         100
                          50  -
                 NO^ with no ammonia or methane (3
-------
    1.0 r
u
u
                                  iDuct
 /^
-------
              Natural gas runs

              Runs without natural gas
          7JO
Figure 12
                        sx>
                                     950

                                     Temperature, "C
                                                   1050
                                                                 1130
          RACK END OVERALL NO^ REDUCTION CORRECTED TO I; I NHj /NO^.
  100
     750
                     Effect Of Scale on NOx Reduction
                                                              -B- REDN - FMF
                                                              -0- REDN - CADR
                                                                 REDN - MICRO
800
850      900      950     1000
         TEMPERATURE (DEQ. C)
1050
1100
1150
    Figure 13
                                    5A-118

-------
                 Session 5B



INDUSTRIAL7COMBUSTION TURBINES NOX CONTROL








   Chair: S. Wilson, Southern Company Services

-------
COMBUSTION NOx CONTROLS FOR COMBUSTION TURBINES
                    Henry Schreiber, P. E.
             Project Manager, Combustion Turbines
                Electric Power Research Institute
                     Palo Alto, California

-------
      COMBUSTION NOx CONTROLS FOR COMBUSTION TURBINES

                                    by

                            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 $300/KW for
simple cycles, and from about $400/KW to $700/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 and
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 NOX 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 Ib. water per
       Ib. 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.
B.      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 550F to 750F,
       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  550F  to
750F temperature window, it is necessary to cool the 950F to  1100F
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  970F depending  upon  the  specific
formulation.  The optimum operating temperature window for a
specific formulation  is 100F.  A synthetic  zeolite, ZSM-5,  has a
narrower operating  temperature range of 570 -  900F.  A new
synthetic zeolite, which is coated on a ceramic honeycomb structure is
claimed to be operational at temperatures between 675 and 1075F.
However, above  800F, NHs 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
       NOX 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) NQY Control.
                                     5B-8

-------
U.S. GAS TURBINES: YEAR ORDERED BY CUSTOMER TYPE
 FOR ELECRTIC POWER GENERATION, 1980 THROUGH 1989
  IOC We


  8GWe



  6GWe


  4GWe


  2GWe


  OGWe
 Electric utility
ED Non-utility generator
CHI Industrial generator
         1980 1981  1982 1983  1984 1985  1986 1987  1988  1989

       Source: UBS-Phillips & Drew Global Research Group
                   Figure  1.
          NOx  EMISSION REDUCTION
   Conventional coinbustor
   without steam injcclion
                         300 to
                        150 ppm

                          NOX
                         100 to
                         50 ppm

                         25 ppm
                         9 ppm
                                        Dry low NO*
                                      combustion system
                  Figure  2.
                        5B-9

-------
   NOX Production (ppmv/niscc)
   40
    10 
     0
0  0.2    0.5       1.0        1.5
        Fuel-Air Equivalence Ratio, <

              Fiqure  3.
                               Temperature (F)
                                         4000
                                             -3000
                                              2000
                                         1000
                                         1.8 2.0
NOx REDUCTION vs WATER-TO-FUEL RATIO

  NOX Emissions Reduction (%)
  60
  40
  20
         Water injection
                            Steam injection
                       Firing natural gas
                       	Firing distillate oil
          0.2    0.4     0.6    0.8     1.0
                 Water-to-Fuel Ratio (Ib/lb)

                Figure  4.
                                     1.2
1.4
                      5B-10

-------
        INCREASE IN HYDROCARBONS
          DUE TO WATER INJECTION
RHC
4
     R
     HC with water injection

'"c  HC without water injection

    Data from tests of
    four engines
               0.5            1.0
                 Water/Fuel  Ratio
             Figure 5.
     CARBON MONOXIDE INCREASE DUE
           TO WATER INJECTION
          CO with water injection
         CO without water injection
         Data from tests of
         four engines
              0.5            1.0
                Water/Fuel - Ratio
             Figure 6.
                  5B-11

-------
              NITROGEN OXIDES vs
         CARBON MONOXIDE EMISSIONS
 NOx Emissions (ppniv
 140
              100         200        300
                   CO Emissions (ppmv)
                                                400
                   Figure 7.
DYNAMIC ACTIVITY vs WATER INJECTION
 Overall rms Level Dynamic Activity, psi (10 chamber average)
 2.4
 2.0
 1.8
 1.6
 1.4
 1.2
 1.0
 0.8 K
 0.6
     [Baseline (SN)
     natural gas
   Qt. Comb. MN
   natural gas
                             Baseline (SN) no. 2 oil
                           Qt. comb. MN - no. 2 oil
0      10       20      30      40      50
           Water Injection Rate, (gal/min)
                                                60
                   Figure 8.
                    5B-12

-------
        DYNAMIC PRESSURE COMPARISON
nanuc Pressure, psi fpeak-lo-peak)
Dynamic Pressure, psi (pc;ik-to-pcak)

1.0
                    n.-isclinc (SN)       0.8
                    production liner

                                   0.6


                                   0.4


                                   0.2
                      Quid (MM)
                      cumbuslor liner
      Elect, noise
         246"
0    200    400    600    800    1000     0    200    400    600    800    1000
            Frequency, H/                           Frequency, Hz
                          Figure 9.
Pan


Ou
ter casing-i
n 1

rlow sleeve 1
m
_^^ 

Plane of
dilution holes

|j Conventional\ /
Jl ip.nn and
= =I
-S


Secondary zone

Dilution zone
V-'pre-mixins 1
Jl primary zone/ \



' '"-^^

J LU
           -End cover
                         Figure  10.
                           5B-13

-------
            DRY LOW NOX COMBUSTOR
              OPERATING MODES (49)
 l-irst Stage Burning            Two Stage Burning: Lean-Lean
 Second Stage Burning
               First Stage Premixed-
               Second Stage Burning
                   Figure 11.
 U.S. COMBUSTION TURBINE SCR INSTALLATIONS
        IN OPERATION (AND PROJECTED)
Generating Capacity (MWe)
   1986
1987
                    1988
                  1989
                                      1990
1991
                   Figure 12.
                     5B-14

-------
             SCHEMATIC DIAGRAM
         (V/Ti02or Zeolite SCR Catalyst)
 Steam-*-
Turbine
exhaust
   CO Oxidation]
        calalysl
Evaporator    SCR
           catalyst
                                              Clean
                                               gas
                                               Water
               Superheater
              Economizer
                   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

-------
 ENVIRONMENTAL AND ECONOMIC EVALUATION OF GAS TURBINE SCR NOX CONTROL
                                         by
                 Phillip A. May, Lisa M. Campbell, and Kevin L. Johnson
                                 Radian Corporation
                     Research Triangle Park, North Carolina 27709

                                     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

-------
 ENVIRONMENTAL AND ECONOMIC EVALUATION OF GAS TURBINE SCR NOX CONTROL
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.  All 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% O2) 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 750F temperature range.  Below 600F,  NO  conversion  slows
                                        5B-21

-------
dramatically; at temperatures above ~800-850F, 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 950F. 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 525F, with an optimum temperature of approximately 475F. 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 (>525F), this catalyst is an excellent NH3
oxidation catalyst, producing additional NOX.   Second,  it is limited to clean fuels because it is
also a good SO2 oxidation catalyst, forming SO3, 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 NOX 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 $30/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
$10/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                       5  21
               NOX Reduction, %                      60  95
               NO3/NOX Molar Ratio                    0.9 -  1.6
               Pressure Drop (across  catalyst),          1.9 - 6.1
               in We                                 170 - 3130
               Maintenance, man hours/year
                                       5B-23

-------
Outlet NOX concentrations range from 5 to 21 ppmvd (at 15% O2). The NH3/NOX 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  O2 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 (NH3), 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 reevaluation 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 bisulfate 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 NO2.  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 SO3.  In addition to the SO3 from combustion,
SO2 oxidation forms additional SO3 across the boiler tubes within the HRSG and across the
SCR catalyst. Available base metal catalysts oxidize between 1 and 5  percent of the SO2
present in the turbine exhaust gases to SO3. Some base metal catalysts offer an SO2
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%) SO2 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 SO2 in the gas turbine exhaust may be oxidized
to SO3. Therefore, CO catalysts can have a significant impact on  the SO3 content of the
exhaust gas stream.

There are two potential problems associated with increased SO3 in the exhaust gas  stream:
First, SO3 can be collected as a paniculate in the form of H2SO4 if the paniculate 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 SO3 and form either ammonium  sulfate and/or bisulfate  salts via the reactions:

     2NH3 + SO3 +  H2O- (NH4)2SO4 (ammonium sulfate)                             (1)
     NH3 + SO3 +  H2O ->  NH4HSO4 (ammonium bisulfate)                             (2)

Even at levels of a few ppm slip, NH3, SO3, and aerosol H2SO4 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 (<400F), only a few ppm of NH3 and SO3 are required
for reaction.  Therefore, at typical HRSG exit temperatures (300 to 350F), 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 SO3 concentrations
in Figure 10.  For a gas turbine firing 0.2 percent sulfur distillate the exhaust gas SO3
concentration is approximately 2 ppm.  As shown in Figure 10,  if the HRSG exit gas
temperature is 410F, 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 SO3 concentration in the
exhaust can be as high as 20 ppm (i.e., 50 percent SO2  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 all 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.


Utility-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:
          Lower 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 (SCR)  NOM Control.
2.    Saleem, A., M. Galagano, and  S. Inaba.  "Hitachi-Zosen  DeNox Process for Fossil
     Fuel-Fired Boilers." Proceedings:  Second NOX Control Technology Seminar.  Hosted
     by Electric Power Research Institute.  Denver, Colorado.  November 8-9, 1978.
     FP-1109-SR.  p 22-12.

                                       5B-31

-------
o
_c

I

-------
  150 
  140 
  130 
  120 
  110 
 100 
 I 9
 1 80
 <" 7
 1 60
 "5 en
 _c 50
   40
   30
   20
   10
               1       I       1       I       I       I       I
              10      20      30     40     50      60     70
                                Gas Turbine Size, MW
         Figure 3. Gas Turbine SCR Installed Capital Cost
 i
80
           18% Water Treatment
                                                      20% SCR
17%HRSG
                                                          7% CO Catalyst
  17% Water Injection
                                             21% Gas Turbine
             Figure 4. Facility Wide Failure Distribution
                               5B-33

-------
   25% Catalyst System
                                                       25% Ammonia System
                               50% CEM System
              Figure 5. SCR System Failure Distribution
                                                     45% NOx Analyzer
15% CO Analyzer
                                                                5% Q, Analyzer
                                                        15% Programming/Software
             20% Gas Conditioning
                   * Total frequency over six month period of 30 events.




  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*
                                            (A



                                           5
                                            o

                                           
                                           ^






-------
        Train A
                                             Train B
38.0
26
38.9
26.2
30.2
25.8
23.1
27
27

23.9
27.3
27.4
26.5
25.1
25.9
25.1
27

25


31.8
29.2
39.9
26.8
29.2
27
-
25










20.9
33.9
23.1
31
29.7
29
27.6
26.1
30.8
23.3
28.7
26.1
31.5
32.1
35.5
29.7
38.1
28.1
21.0
25.6
38.1
24.8
37.2
26.9
28.6
30
25.7
28.4
33.9
27.7
32.8
25.3
33.8
24.8
27.5
26.9
33.0
29.5
23.8
27.9
42.0
26.9
15.8
24.8
30.8
25.3
36.5
26.9
       Average 30.2
Average 26.5
 Figure 9.  SCR Inlet Velocity and NOx Concentration Maps
         500
                       5  10        50   100

                        SO3 Concentration, ppm
      500
              Figure 10. Ammonia Salt Formation
as a Function of Temperature and NH3 and S03 Concentration (2)
                         5B-36

-------
 NOx REDUCTION AT THE ARGUS PLANT
     USING THE NOxOUT* PROCESS

        Joseph R. Comparato
          Nalco Fuel Tech

          Roland A. Buchs
North American Chemical Corporation

          Dr .  D .  S .  Arnold.
          L.  Keith Bailey
       Kerr-McGee Corporation

-------
                        NOx Reduction At The Argus Plant
                            Using The NOxOUTR Process

                               Joseph R.  Comparato
                                 Nalco Fuel  Tech

                                 Roland A. Buchs
                       North American Chemical  Corporation

                                Dr. D. S. Arnold
                                 L.  Keith Bailey
                             Kerr-McGee Corporation
                                    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 bisulfate 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 test 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 (EPRI).   EPRI obtained the first patent on the fundamental urea
process in 1980.   The overall  chemical  reaction for reducing  NOx  with urea is:

                  NH2CONH2 +  2NO +1/2O2 > 2N2 + C02 + 2H2O

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 NOxOUT to new
applications.  Process performance is analyzed using Nalco Fuel Tech's chemical
kinetics computer model (CKM).   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 NOxOUT 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 Elektrizitatswerk 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 Ib/hr (322,580 kg/hr) at 950F (510C).

                                     TABLE  I

                                  COAL  ANALYSIS
            Type                          Utah Bituminous
            Ultimate Analysis             As Rec'd    Dry Basis
                  %Carbon                 70.52       73.27
                  %Hydrogen                4.91        5.10
                  %Nitrogen                1.37        1.42
                  %Chlorine               <0.1        <0.1
                  %Sulfur                  0.47        0.49
                  %Oxygen                 10.32       10.72
                  %Ash                     8.66        9.00
                  %Moisture                3.75        N/A
                  HHV, Btu/lb             12,592      13,083
                                      58-41

-------
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 SO2 scrubber.   The flue gas  is
exhausted without reheat at 120F from the stack.

In May  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 1800F (982C)
before modifications, to a maximum of 2200F (1204C) 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 base condition for applying the
NOxOUT process.   In  July  1989, the temperature profile in the upper furnace with
this SOFA configuration was measured.  An average temperature of 2020F (1104C)
and a peak of 2110F (1154C)  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 as a function of temperature (figure 4).3
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.
                                      5B-43

-------
Testing Results

Test series were identified in terms of eighteen test days.  The test objectives
were:
            Test days         Test Series Type
            # 1-4             Boiler SOFA Characterization
            # 5-9             Upper Level Injection
            #10-13            Lower Level Injection
            #14-16            48-Hour Demonstration Test
            #17-18            Miscellaneous Testing

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,
Inc.

The plant's continuous stack emissions monitor provided NOx and CO data,
corrected to a dry basis at 3% O2-   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 point 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
                                      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 ESP
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 entrainment of
brackish water from the  wet  scrubber.  Many of the decisions in subsequent tests
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 SOj 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 convective
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 as 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 shows 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 May, 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 Argus 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

-------
REFERENCES

1.    Muzio, L. J., and Arand, J. K. "Homogeneous Gas Phase Decomposition of
      Oxides of Nitrogen", EPRI Report No. FP-253, 1976.

2.    Arand, J. K., Muzio, L. J., Setter, J. G., U. S. Patent 4,208,386, June
      17, 1980.

3.    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.

4.    Michels, W.  F., Gnaedig, G., and Comparato, J. R. , "The Application of
      Computational Fluid Dynamics in the NOxOUT Process for reducing NOx
      Emissions from Stationary Combustion Sources", presented at the AFRC
      Committee Conference, San Francisco, CA, October 10-12, 1990.

5.    Hofmann, J.  E., von Bergmann, J., 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.

6.    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.

7.    Morgan, M. E., "Effect of Coal Quality on the Performance of Low-NOx
      Burners", presented at the  British Flame Days Conference, London,
      September 1988.
                                      5B-47

-------
       CONVECTIVE SUPERHEATER
           PLATEN SUPERHEATER   I
            EL 106'
     NOxOUT  
INJECTOR PORT LEVELS
                                                            ECONOMIZER
                                                            EMISSIONS SAMPLING
    COAL PULVERIZER
    400
    300
 Q_
 Q.

 8
200
    100
             COFA
                           ARGUS #26 COAL FIRED BOILER
                              NOx Baseline
                                                                     TO
                                                                     AIR PREHEATER
                                                                     FROM
                                                                     AIR PREHEATER
                                                                         FIGURE 1
                        0/0/100
                               0/50/100
                                              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

Q.
a. 40
CD
                                        A
                  A
                              A
 O
 c3
 O
   30
  20
   10
                         ...A
                    A
                                   A
                                      A
    160
                 180          200           220
                        NOx Emissions (ppm)
                                                     240

                                                       FIGURE 3
D_
D_
   500
   400
   300
   200
   100
                Critical NOx Concentration
     700      800
                         3% Excess Oxygen
                        NOxOUT Kinetic Model
                                      NOxi=500 PPM
                                            ,' NOxi=200PPM
                    900      1,000     1,100     1,200      1,300     1,400
                       Temperature (degrees C)
                                                        FIGURE 4
                            5B-49

-------
                                    CFD MODEL OF NOxOUT INJECTION
                                              PLAN VIEW AT ELEVATION 100'
cn
CD
cn
o
lonuentrat ion
     O.OOE+Lin
     1 .34E-04
     J1.69E-04
     4. 03 E- 04
     5.38E-IVI
                                                                                          9.41E-D4
                                                                                          1 .08E-03
                                                                                          1 .21E-03
                                                                                          1 .3AE-OJ
                                                                                          1 .A8E-OJ
                                                                                          1 .blE-OJ
                                                                                          1 .75E-03

                                                                                         Y
                                                                                     FIGURE 5

-------
                        NOxOUT INJECTION SYSTEM
        NOXOUT-A METERING PUMP
        WATER
        NOXOUT-34 METERING PUMI'
                                  PRESSURE REGULATOR
-N
                                 MOYNO PUMP
                        MIXING/METERING SKID
                  INJECTORS
                                                             FIGURE 6
   300
   250  -
Q.
6
   200
   150  -
   100
                         NOx Emissions
                     AT STAGING CONDITIONS
                  0.5           1           1.5           2

                NORMALIZED STOICHIOMETRIC RATIO (NSR)

                   0/0/100   0/100/100  40/100/100  loo/loo/loo
                     D      A       O        *
                             2.5
                                                               FIGURE 7
                               5B-51

-------
   240
   220  '-
Q. 20
Q.
   180

en
"F  160
LJJ
X
O  14
   120
   100
                    NOx Reduction vs NSR
                     0.5              1               1.5
                    Normalized Stoichiometric Ratio (NSR)
                        level 106   level 90    demo
                          D       A       O
                                                               FIGURE 8
                       EFFECT OF TOTAL FLOW
                              NSR RANGE 098 -1.19
ou 
40 -

O 30 -
|~"
O
Q
UJ 20.
X
O
10 -
0 -
3
D
D
o Q
DQ Q n
~*~ "*" _n ^ E3
+ + + D
+ + +
a
+

I I I I I I I I I I I | | | I 	 1
0 420 460 500 540 580 620 660 700
        D   LOWER LEVEL INJ
                               TOTAL FLOW (GPH)

                                       +   UPPER LEVEL INJ
                                                              FIGURE 9
                               5B-52

-------
NOx REDUCTION vs INJECTOR ARRANGEMENT
                   NSR RANGE 098-1 19
au 
40 -


z
O 30 -
b
D
Uj -
1
10 -
o -
B

0 D
B D
D O D Q
+ + + n o o B Q
+ + + Q
D
+

1 6 8 10 12 14 16
TEST DAY
     D   LOWER LEVEL INJ
                             4-   UPPER LEVEL INJ
                                                  FIGURE 10
              AMMONIA EMISSIONS
                 SAMPLED AT ECONOMIZED OUTLET
JU
f 	 s 28 -
^^
^ K
D_ x ~
5>
uJ -
3
-1 18 -
7 16 -
0
^ 12 -
^ 10 -
LU
O B ~
c 6 ~
LU 4 _
^ 2 -
0 -
C




+

D
+

+
D
O
 D
I I I I I I I I I I I I I I 1 	
2 4 6 8 10 12 14 1
                      TEST NUMBER
           + UPPER LEVEL INJ.   Q LOWER LEVEL INJ.
                                                   FIGURE 11
                    5B-53

-------
 LLJ
 Q

 g
 o
 CO
 DC

 O     10 -
                CO EMISSIONS WITH NOx REDUCTION
                               NSH RANGE 098-1.19
                               +

                               D Q
:~o~
             LOWER LEVEL INJ
                              NOx EMISSIONS (PPM)

                               +   UPPER LEVEL INJ
                                  180




                        O   48 HR DEMO


                             FIGURE 12
z
Q

O
^)
Q
LLJ
DC
X
O
         EFFECT OF FURNACE CLEANING ON REDUCTION


                        OPTIMIZED INJECTION DURING 48 HH DEMO
                         n	r
                           HOURS SINCE LAST SOOTBLOW
                                                     ~\	T
                                                             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

-------
               REBURNING APPLIED TO COGENERATION NO, 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
ABSTRACT

New cogeneration systems are increasingly regulated to stringent NOX levels based on control
technology precedents established in California.  NOX compliance costs can be a disincentive
to cogeneration markets. This project evaluated reburning to achieve low NOX 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 NOX 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 NOX reduction  results
corroborated the  subscale results.   Reburn burner stability  problems prevented going to
stoichiometric ratios  below 0.98,  however, so maximum  NOX 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 NOX reductions.  NOX 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, NOX  reductions of
70 percent were experienced. In all configurations, reburning 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 will 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 nonaftainment.  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  100,000 Btu/hr parametric configurational tests  for rich-burn,
          moderate O2, 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  NOX
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 NOX.

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
SR2 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-burn 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-burn 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  lean-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
not 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 NOX
reduction when metal oxide catalysts were introduced into the reburn chamber. The potential
benefits of the concept of catalytic enhancement of NOX 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 NOX 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 not possible
to test below stoichiometric ratios of about 0.99.  Since NOX 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 NO,-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 boiler staging. Stability tests showed that turbine exhaust oxygen levels of 14 percent or
greater were needed to maintain a stable boiler 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 1989.

    2.     Brown, R. W., Moyer, C., Nicholson, J., and Torbov, S., "Application of Reburn
          Techniques for NOX Reduction and Cogeneration Prime Movers:  Volume II,
          Lean-Burn Engine Applications," GRI 90/125, Gas Research Institute, Chicago,
          IL, March 1991.
                                        5B-61

-------
                          Air
                   Natural gas
                                                          H?0 In


1
T

Reburner
Rich A


i
     3
     9

     5

Waste heat
recovery boiler

<    Flue out
                                            Lean
                                                          Steam
                                                          out
                                                H,0 in
A1r 	 ^.
Air 	 .
Fuel 	 ^

_ turbine >
| I
IfJ

C
t

IAAAA/I
onventlor
>o 1 1 e r

i
[_

al
                                                                       Steam
                                                                       out
Figure 1.  Reburning Applied to Cogeneration or Repowering with Gas-Fired Prime Movers
                      Figure 2.  Reburn Subscale Facility Schematic
                                         5B-62

-------
  6o
  50 I
 40
   30 I
   10
                                               SR, = 0.8
                                    I
                                             _L
                                  _L
               10
   15         20        25         30
         Fuel  fraction (percent)
Figure 3.  Subscale Reburner
                                                                  35
       View port
                                                   2-1/2 in plunger
                                                       Gas
                    Figure 4. Effect of Fuel Fraction
                                   5B-63

-------

                                                                                      - 2,000
                                                           DIM gu eoodttooni
                                                                  - 07 pvrtwnl
                                                                T,  1.000'F
                                                                      100.000 Blu/ht
                 400
                         800
                                 1,200
                                           1,600    2,000     2,100
                                           Input NO  (ppm)
     2,800    3,200
                       Figure 5.  Effect of Input NO Concentration
   60

   55
'I  40
o-35

I  30
U
!2B
-  20
QJ
U
I  15

   10
    5
&  f  0.375 with preheat
A  I - 0.375 no preheat
  f - 0.20 with preheat
  I - 0.20 no preheat

     Flue gas conditions
     NO. - 1.500 ppm
     0, - 0.2 percent
     T, - 1,100'F
           -  100.000 Btu/hr
     0.6
             0.65
                      0.7
                               0.75
                                         O.B
                                                  0.85
                                                           0.9
                               SR,  rehurner  stoichiometry
                                                                    0.95
                         Figure 6.  Effect of Reburner Air Preheat
                                                                              1.0
                                            5B-64

-------
   60 _
~  50 -

IT  "5 _
   10

    5
SR, - OB

Flue gas conditions
    NO, = 1.500 ppm
    Oa = 0 2 percent
    T,  1,000'F
    FR^,  - 100,000 Btu/hr
                                           _L
                                       _L
                                                                              With burnout air
                                                                              Without burnout
                                                                              air
                        10       IB       20        25       30
                                     Fuel fraction (percent)

                            Figure 7.   Effect of Burnout Air
                                                                      35
                                                              ?-]/? 1n.
                                                              Gp  I/I In.
                                                              Hut gti centflttoni
                                                               MO,  1,500
                                                                  0.?
                                                               T3  I.100T
                                                                 -l. - 100.000 Btu/
                                     15       20       25       30
                                  Fuel  fraction (percent)
                      Figure 8.  CO  Level Versus Fuel Fraction
                                            5B-65

-------
                                                  !* |Mlfl
                                                  1-1/7 In.  Dlwff
                                                  ClD - J't 1".
                                                  fit* pi ttf*d< t(MI
                                                   wo,  too OP*

                                                   tj  I.IWT
                                                   n^,B . jf, ltu/h'
                     03     04     0.5    06    07
                                                            09    10
Figure 9.  Effect of Reburner Stoichiometric Ratio for Lean-Burn Conditions
                                                          SHIELD
                                                     AIR
                                                     QAS
           Figure  10.  Burner Configuration with Catalyst Insert
                                      5B-66

-------


2
g
^
<_>
n
D
LJ
c:
X
O
vp
c1;



100-
80-

60-


40-

20-
Q

e u %HF


O 30X FUEL FRAC
 20X FUEL FRAC

A 1 5X FUEL FRAC
A 1 0X FUEL FRAC
'
' 	 1 	
0.80 0.85 0.90
SR3

fcl* .
^k
*. *o
&. .
w
. A
* A 

A A *
^^ 4B
. AA
A A ^
A
1 1
0.95 1.00

Figure 11.  NOX Reduction with Catalyst Enhancement for a Space Velocity of 7,500 per hour



o
o

o
Ui
fy
IL.
X
O
K




90-

80-

70-


60-

50-
40-
30
*&& 6
W W
o
 .0 0
o c
0 ^ *>
o
o

0
Of\
0
o
o

5.0 15.0 25.0 35.0 45
                            SPACE VELOCITY (1/HR X 10~3)
    Figure 12.  Effect of Space Velocity on NOX Reduction with Catalyst Enhancement
                                     5B-67

-------
BYPASS ENGINE EXHAUST
                     LAYOUT FOR REBURN SYSTEM
       Figure 13.  Full-Scale System with 150-kW Caterpillar Engine
                      BASELINE -  NO CATALYST




z


=>
0
1, |
cc

0
2




"'O
60-
50-



40-

30-


20-

10-

0-
V V

V V 


V * J^ 
^7
V^ V
V
A 1 25 kw LOAD  V
 1 00 to. LOAD

 BO kw LOAD
V PILOT SCALE TESTS
	 1 	 1 	 1 	 i 	 i 	 i _ _ _ i
O.B9 0.91 0.93 0.95 0.97 0.99 1.01 1.03 1.C
                                 SR3
    Figure 14. Baseline Reburn NO, Reductions for Full-Scale System
                               5B-68

-------
           100
75
           50
           25
                   90 kw LOAD

                  A 100kwLOAD

                   90 kw LOAD

                  V PILOT SCALE TESTS
                             V    V
                           vv     v
                                           V
                                            I
                                                   v  v
             0.85
                           0.90
                                          0.95
                                          SR3
                                                          1.00
                                                                         1.05
                   Figure 15.  Full-Scale NOX Reduction with Catalyst
loll
                      Figure 16. Laboratory Repower Test Facility
                                         5B-69

-------
90 -
BO -
70 -
60 -
50 -
40 -
30 -
20 -
10 -
0 -
1
o EJ
J  * 
a
  .-?.*? ;.:< : 
+ t** *4











i i i i i i i i i ' i i i i 	
14 te 22 26 3 14 38
FUEL FRACTION,!
D STAGING + NONSTAGING
Figure 17. Effect of Fuel Fraction on NOK Reduction for Staging and Non-Stagin
CO REDUCTION ALL BURNERS
100 -
90 -
80 -
70 -
K
60-
g
^ 50 -
D
" 40-
o
u
30 -
20 -
10 -
n -
COLD & HOT WALL
B " V V V7 7
7 V y VV U
^ V

J
w
7
7
3
9
77

^

          0.2       04       06       08        1
                                (Thousonds)
                  INITIAL CO CONCENTRATION, ppm (0 7. 02)
                  V    WITH Si W/0 STAGING
                                                        1.2
                                                                 1.4
Figure 18.  Effect of Initial CO Concentration on CO Reduction
                            5B-70

-------
SELECTIVE NON-CATALYTIC REDUCTION (SNCR)
     PERFORMANCE ON THREE  CALIFORNIA
       WASTE-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

-------
                      SELECTIVE NON-CATALYTIC REDUCTION (SNCR)
                           PERFORMANCE ON THREE CALIFORNIA
                             WASTE-TO-ENERGY FACILITIES
ABSTRACT
Concern over NOX 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  NOX  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
NOX control  in  the form of  Exxon's Thermal DeNOx,  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 DeNOx  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  NOX data presented  in  this  paper  is given  on  a  dry  basis  corrected to 7%
oxygen.

                                      5B-73

-------
                      SELECTIVE NON-CATALYTIC REDUCTION (SNCR)
                          PERFORMANCE ON THREE CALIFORNIA
                             WASTE-TO-ENERGY FACILITIES
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
landfill ing 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 (NOX) 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 DeNOx 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-
Martin 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 Montenay 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 NOX
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 DeNOx on  MWC  furnaces.
Figure 1 provides general information  on  the  current  Thermal  DeNOx  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 DeNOx designs differ markedly.  Some of the unique designs
and operational features are:
Commerce
Stanislaus
            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.
            Ammonia feed rate  is controlled  automatically based on stack NOX 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.
                                       5B-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 NOX
            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 DeNOx process from adequately  controlling  NOX:   (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 NOX  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  NOX vs.  load  relationship
presented in Figure 5,  for four separate zone combinations.  Of particular  interest
is the ability of one zone (or  combination of zones) to provide low  NOX 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 LIMITS
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  NOX emissions but  also  of  particular  significance
is the averaging time designated for each limit.   Table 1 presents the NOX 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  NOX 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 NOX emissions  on a daily  basis  range from a low of  720
Ib/day at  SERRF to a high of  1130  Ib/day  at  Stanislaus; Commerce  has  a daily,  NOX
1imit of 825 pounds.
It is obvious that lower NOX 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  Ib/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 NOX 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 NOX LIMITS
The daily NOX mass emission limit at Commerce (825  Ib/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 NOX were to occur.

STANISLAUS NOX LIMITS
Stanislaus is unique in that NOX 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

-------
Ib/day which is roughly equivalent to  150 ppm.  Stanislaus is also unique in that the
plant has a stack ammonia limit of 50  ppm (raw).

NO^ 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 DeNOx 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 EPA study,  which reviewed NOX data from twenty-six  mass-burn/waterwall  facilities.
The study stated that  the average uncontrolled NOX 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 NOX
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 DENOX  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-NOx mole  ratio of about  1.5 through
the  upper  elevation  of nozzles.   Carrier  air  pressure  had  no  effect  on  DeNO
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 NOX 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  NOX  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 NOX 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 NOX 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 NOX control equipment  which
results  in some  ammonia  slip, the  ammonia  is  absorbed  by the  water creating  an
alkaline solution.  The solution acts  as  an  acid gas  scrubber  removing S02,  HC1  and
N02, 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  NOX  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 NO
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
NO  control logic has  stated that a  plume  no  longer is visible.
  X
                                      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 ppm2.   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 NOX emission permit limits.  Furthermore, all
three  plants  operate  below  the  NSPS NOX  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 NOX  compliance is  possible.
By reducing the time  intervals by which compliance is monitored, plants are forced  to
operate at lower NOX 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:
:Unlts:X X           X (1)330-400 TPD    XX:
                    Foster-Wheeler (115,000 Ib/hr)
                    Detroit.  ..'xx X. . .X /XXXXXX
                   .4 Levels oh Both      .  .XX
                    Side Walls   X  ':  XX. .X..
 Boiler Cross-Section:  iSis'fw) x 18'(d)
                          Stoker:
                         ;NHi Injection;
      Two Front Wall
      (10) NH3 Injection
      Nozzle Locations

STANISLAUS COUNTY RESOURCE RECOVERY
FACILITY:
Units:
. Boiler;
Stoker:
NH3 Injection:

Boiler Cross-Section:
(2) 400 TPD XX
Zurn {Not Available) .. ..
Martin 	 v.. ...Y
2 Levels on the
Front Wall
(Kot Available) : .
      Front Wall (15)
      and Side Wall
      (23) NH3 Injection
      Locations
     SOUTHEAST RESOURCE RECOVERY
              FACILITY (SERRH:	
Units:
Boiler.
Stoker:
NH3 Injection
                  (3) 460 TPD         .  ..  . ..:
                  L&C Stelnmueller (117,170 Ib/hr)
                  L&C Stelnmueller
                  2 Levels, Front Wall
                  and B.oth Side Walls
Boiler Cross-Section: 19'(w) x 18'(d)
Figure 1. Various Ammonia Injection Configurations at Three
 California MSW Incinerators Equipped with Thermal DeNOx
                          5B-82

-------
     CARRIER/PURGE AIR
OVERFIRE AIR

FAN - 30" H20
              AMMONIA STORAGE
F
*GE J
t^~A

VAPORIZE


"I

                                                               INJECTION

                                                                 ZONE
        Figure 2. Commerce Ammonia Receiving, Storage and Delivery System
         3
         u.
         o
             80 n
             60-
             40-
             20-
                       Limit Needed to  Meet

                          Daily NOx Limit
 0         50        100       150


            NOx - PPMc at 7% Oz




Figure 3. Commerce Refuse-To-Energy Facility

        Ammonia Feed Rate vs. NOx
                                                       200
                                  5B-83

-------
                    F1.F2
                     D
               G1.G2
               D
                              E1.E2
                              D
                    A1.A2  Bl.BI  C1.C2
                     ODD
                                       Furnace Penetrations for
                                     Ammonia Injection Nozzles
                                            Uppfmmofitifi)ctlon penetration
                                            ptarw ffflONT WALLHAS 1S NOZZLES)
                             Micfcte ammprita ipiection pkne
                             (23 NOZZLES 1CCATED ON BO1>|
                             SIDEWAU5J
                                   - (El, -S
                                            Lowera
                                            plan* (21 NOZZLES LOCATED ON BOTH
                                            SIDE WALLS;
Figure 4.     North side schematic of a typical SERRF Steinmuller-designed furnace.
            Observation ports through which temperature profiling was performed are
            shown.
 M
o
f-
4-1
CO


Q.
 I

X
O
    200 i
    150-
    100-
     50-
 ZONE 3
 ZONE 3,4
^ ZONE 2,3,4
n ZONE 2,3
                  20
             40
                                       1
                                      60
 1
80
                                                          100
                               %MCR
       Figure 5.   Commerce Refuse-To-Energy Facility NO, vs. Load
                 Utilizing Various Injection Zones.
                             5B-84

-------
   15
   10
Q.
Q.
IT
O
T.
X  ,
 n  O
T.
Z
X
D)
O
   (5)
                           (s)
                                        + HCI (g)
                         Explanation:
                            At any given temperature,
                            condensation will occur if
                            the log of the product of
                            mole-fractions XNH3-XHCI,
                            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

Air Quality District
 Commerce
Stanislaus
SERRF
South Coast   Stanislaus County  South Coast
   AQMD             APCD            AQMD

Pollutant
NOX ppm G> 7% 02
NOX ppm G> 7% 02
NOX ppm (? 7% 02
NO" Ib
NOX Ib
NH3 ppm (raw)
EPA-PSD
More stringent of
NOX ppm 0 7% 02
or
NO -Ib
and
More stringent of
NOX ppm @ 7% 02
or
NOX Ib
Averaging
Period
15 min. 175
1 hour
8 hour
1 hour 40
24 hours 825
--

3 hour


3 hour

24 hour


24 hour



--
200
--
1130
50

175


160.5

165


1200


175
116
--
34
720
--

--


--

--


~ 
 NOTE:   The EPA NSPS NOX limit  for MWC which are larger than 250 TPD  is  180  ppm NOX
 averaged  over  24  hours.
                                       5B-86

-------
                                      TABLE 2
                COMPARISON  OF  NOX EMISSIONS FROM THREE CALIFORNIA MSW
                      INCINERATORS EQUIPPED WITH THERMAL DENOX

Uncontrolled NOX
ppm @ 7% 0,
Ib/hr
Controlled NOX
ppm G> 7% 0,
Ib/hr
Ammonia Slip
ppm (raw)
Commerce

128-217
44-75

104
35.8

-2
Stanisl
Unit 1

298
90.4

93
28.1

3.7
aus
Unit 2

305
96.0

112
36.0

5.0
SERRF
Unit 1 Unit 2

210
74.8

49 72
16.5 22.7

--

Unit 3

259
93.1

54
17.9

35
                                      TABLE 3
                 PARTICULATE EMISSIONS AND THE IMPACT OF ADDING BACK
                          THE PSEUDO-PARTICULATE FRACTION

Permit Limit

Test Results
% of particulate
Commerce
5.5 Ib/hr

2.5
88%
Stani
0.0275
Unit 1
0.011
51%
si aus
gr/sdcf
Unit 2
0.011
79%
SERRF
5.0 Ib/hr
Unit 3
1.7
70%
caught in the back-half
of the sample train

Impact on particulate
level if neutral salts
were added back
60%
+ 34%
+38%
N/A
                                       5B-87

-------
USE OF NATURAL GAS FOR NOX 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 Takuma 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

-------
                  USE OF NATURAL GAS FOR NOX CONTROL
                     IN  MUNICIPAL WASTE  COMBUSTION
ABSTRACT
Natural gas injection  (NGI) technology for reducing NOX 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 106 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 NOX.   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-Energy
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.
                                 5B-91

-------
                  USE OF NATURAL GAS FOR NO  CONTROL
                     IN  MUNICIPAL WASTE  COMBUSTION
UTILIZATION OF NATURAL GAS IN MUNICIPAL WASTE COMBUSTORS (MWCs)
In 1986, following  GRI's  successful pilot-scale testing of natural gas
reburning  for NOX  reduction in coal-fired  applications,  GRI  and IGT
began  an investigation of the  potential  for utilizing natural gas in
MWCs  for the control of  NOX emissions.   At that time the control of
NOX 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 NOV  emissions from all MWCs.   The  limits  being
                       X
evaluated  were  based on the performance of the thermal de-NOx process,
which  uses ammonia  injection to reduce NOX emissions.  The thermal de-
NOX process has been installed on three MWCs operating in  California.

Figure 1 illustrates the NOX  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  NOX  precursors,  as  well  as  NOX.  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 NOX precursors (for example, NH3, 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  NOX 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
waterwall furnace technology.   Each unit was designed  to burn  MSW at
the   rate  of  100 tons/day   (90 metric  tons/day),   producing  about
24,000 Ib/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* NOX-  Without OFA
and at lower excess air, NOX emissions were reduced significantly, but
CO  and total  hydrocarbon  (THC)   emissions  increased  greatly.   The
baseline data  show that NO  can be reduced  by eliminating  OFA and
reducing  excess  air;   however,   incomplete  combustion  results    as
indicated  by  the high  CO  levels.  The  goal  of NGI  is  to reduce NOX
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  106  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  air;
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 NOX 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 NOV
                                                                     A.
precursors.   The first  3 seconds after NGI  reduced NOX  from  225  to
75 ppm.   Longer times   produce very  little additional NOX 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% NOV reduction.
* All of the NOX and CO emission values presented here are on a 12% O2
  and dry basis.  For a 3% 02 basis,  multiply values by 2 and for a 7%
  02 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 time  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,  NOX  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, NOX 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.  NO  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 NOX 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  NOX,  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
         and  the  variability  of  feed  quality  and  operating
         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 NOX 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
    t     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-furnace  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  NOX  emissions  on  a
         without any adverse effects
the  NO   emissions  on  a  full-scale  commercial  unit
    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,  CO2,  NO )  and temperature
         profiles  in the  reducing zone below  the OFA injectors
         and at the  furnace exit  above the OFA  injectors
         Flue   gas  composition   (O2,   CO,  CO2,   NOX)   at  the
         electrostatic  precipitator  (ESP)  inlet
         Flue  gas composition  (O2,  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  O2,  CO,  CO2,
and NO  analyzers  was installed near  the ESP  to  measure  the  gas
       ji.
composition at  the ESP  inlet;  and  another set of O2,  CO, THC, CO2, and
NO  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  NOX and  CO  emissions measured  at the ESP 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 NOX  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 Ib/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  Ib/h  or  6%  higher than  the  current
normal  baseline steam  flow of  26,700 Ib/h,  and 20% higher than the
1987  baseline level  of 23,500 Ib/h.   During  most  of the  tests with
NGI,  the steam  flow rate was maintained  at  29,000 Ib/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 require 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 60% and CO emissions
by  50%.   The NOX 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 NOV and
                                                                 A
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  follow  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  NOX  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 32 ppm  CO (expected regulatory
limit) was  about  137 ppm while the  average  NOX 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 NO  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-NOv  process for
                                                        A
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,  O.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,  C. ,  Wakamura, Y.
     and  Linz,  D.   "Development of Natural Gas  Reburning Technology
     for NO.. 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-90/0145   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

-------
    Undergrate Air
                                   Overflre Air
                                   Natural Gas/
                                   Reclrc. Flue
                                     Gases
   Figure 1.  The METHANE de-NOx Process
Figure  2.  Olmsted Waste-to-Energy Facility
                     5B-102

-------
  250
  200
 Q_
 Q.
 - 150
 X
O
  100
              n
                      D
   50
     012345
                Residence Time, seconds

    Figure 3.  Residence time  plays  a
    significant role in the effectiveness
    of natural gas
  180
  160
  140
p 120
Q_
D.
 -100
   80
  60
  40
  20
                D
  Baseline 87
     D
  Baseline 91
     A
   FGR Only
     O
FGR + Natural Gas
                                O
                                         D
    10    20    30    40    50    60    70    80    90
                      CO, ppm
      Figure 4.   Natural gas injection
      simultaneously decreases NO   and CO
      emissions
                     5B-103

-------
                                                   Table  1


                          AVERAGE OPERATING DATA - 1990/1991 FIELD EVALUATION TESTS
en
CD
MSW,* Ib/h


Natural Gas, %


Total Heat Input,* 106 Btu/h


FGR, %


Excess Air, %


Total Flue Gas,* Ib/h


Steam Flow, Ib/h


Economizer Exit Temperature, F
          Precipitator Inlet

             02,  %
             CO,  ppm at 12%
                             2




Baseline
1987
Test
6,450
0
33.5
0
73
44,800
23,500
417
9.3
30
135
1991
Test
7,760
0
40.3
0
76
54,100
28,250
425
10.5
46
117


FGR Only
(Average
Data)

0

9.5
54
47,100
27,670
423
7.6
47
70
FGR +
At Normal
1987
Baseline
MSW Input
(Average
Data)
6,500
14.0
39.9
9.5
37
45,400
29,000
422
6.5
22
48
NGas
At
Normal
1991
Baseline
MSW Input
Test
7,000
12.4
41.9
10.0
41
48,500
30,500
422
5.9
21
48
         *Estimated.

-------
           Session 6A



POST COMBUSTION DEVELOPMENTS II








       Chair: D. Drehmel, EPA

-------
PERFORMANCE OF UREA NOx REDUCTION SYSTEMS
                ON UTILITY BOILERS

       Andris R. Abele, Yul Kwan, and M.N. Mansour
               Applied Utility Systems, Inc.
                1140 East Chesnut Avenue
               Santa Ana, California 92701

              N.J. Kertamus and Les J. Radak
           Southern California Edison Company
               2244 Walnut Grove Avenue
               Rosemead, California 91770

                   James H. Nylander
           San Diego Gas and Electric Company
                4600 Calsbad Boulevard
                Carlsbad, California 92008

-------
             PERFORMANCE OF UREA NOX REDUCTION SYSTEMS
                              ON UTILITY BOILERS
                  Andris R. Abele*, Yul Kwan, and M.N. Mansour
                            Applied Utility Systems, Inc.
                             1140 East Chestnut Avenue
                            Santa Ana, California 92701
N.J. Kertamus and Les J. Radak
Southern California Edison Company
2244 Walnut Grove Avenue
Rosemead, California 91770
James H. Nylander
San Diego Gas and Electric Company
4600 Carlsbad Boulevard
Carlsbad, California 92008
ABSTRACT

Test results from the full-scale application of urea injection for NOX reduction on two utility
boilers demonstrate the sensitivity of urea NOX  reduction performance to boiler design,
operating conditions, and urea process variables.  The two utility boilers are both gas- and
oil-fired  boilers,  but  of different size and design.  The demonstration  sites include  a
Southern California Edison Company 320 MW tangentially-fired boiler and a San Diego Gas
and Electric Company (SDG&E) 110 MW front wall-fired boiler.

The  performance of the urea NOX reduction process at the  two sites  was dominated by
variables affecting the temperature at the injection location and the mixing of urea with the
combustion products.  Varying operating conditions, such as load and firing configuration,
changed the temperature distribution in the boilers as  well  as initial  NOX levels.   Such
changes affect the relative location of urea injectors within the urea reaction temperature
window and, thus, the level of NOX reduction achieved.   Available injection process
variables, including injector design, solution flow and pressure, injector location and spray
orientation,  were used to optimize the distribution  of urea within the reaction window at
varying loads to achieve maximum NOX reduction.

Minimum NOX  emissions  were  achieved  at both sites  by  coupling urea  injection  with
modified combustion  conditions.  Urea  NOX  reduction performance  at  these  modified
operating conditions was about 30 percent at NSR  = 2.0 over  the boilers' load ranges.
Resulting stack NOX emissions at both units were 20 to  45 ppm @ 3 % O2 depending on
load, while ammonia slip was less than 20 ppm.
* Currently with the South Coast Air Quality Management District.
                                        6A-1

-------
             PERFORMANCE OF UREA NOX REDUCTION SYSTEMS
                              ON UTILITY BOILERS
IMPLEMENTATION OF THE UREA NO, REDUCTION PROCESS

The urea NOX reduction process is a selective non-catalytic reduction (SNCR) process which
encompasses a sequence of steps.  Aqueous  urea  solution  is pumped to injection nozzles
which  spray the chemical into a boiler  or  furnace chamber.  The  droplets of injected
solution  evaporate and the urea thermally decomposes  into reactive  species.   The urea
droplets  and released reactive species mix with the NOx-laden combustion products.  Urea
species then react with the combustion products at the proper temperatures to  reduce nitric
oxide (NO) to elemental nitrogen (N2).

The NO-reducing reactions are temperature sensitive and occur within a narrow temperature
range. If the urea is released at too high a temperature, the chemical species can actually
be oxidized to NOX.  If the urea is released at low temperatures, the NO-reducing reaction
rates are limited and result in poor chemical utilization.

An additional complication in SNCR systems is that these temperature sensitive reactions
must occur not in a well controlled  reactor, but  in a load-following  utility boiler.  The
design of  these systems  must address the issues of  temperature  variations and  mixing
limitations to the extent possible.  Since a utility boiler presents a far from perfect reaction
chamber environment, efficient utilization of injected urea is not possible  for all boiler
operating conditions.   Since the  process is  imperfect,  excess  urea  must be injected to
maximize  the availability of NOx-reducing  species within the narrow reaction window
provided within utility boilers.  Unutilized ammonia (NH3)  will  be  a result if the injection
temperature is too low.  At high injection temperatures, excess NH3  is oxidized to NOX,
defeating the purpose of reducing NOX emissions.  Thus, tradeoffs will exist  between NOX
reduction and overall process performance.

To understand  the  effectiveness  of  the  urea  injection process,  the term Normalized
Stoichiometric Ratio (NSR)  was defined as  the ratio  between the actual amount  of urea
injected  and the theoretical amount required to react with all the NO present.  For example,
a urea flowrate of NSR =1.0 provides the exact amount of urea to react with 100 percent
of the NO present.  This Stoichiometric ratio  of NSR =  1.0 is equivalent to  a urea to NO
mole ratio  of  0.5,  since one  mole of urea  (NH2CONH2)  potentially has two moles of
nitrogen species (e.g., NH;)  available to react with NO.
                                        6A-2

-------
COMPARISON OF BOILER DESIGNS

The two boilers used for demonstrating the  urea NOX reduction process are  different  in
design and size.  Both boilers are located in Southern California.  The primary fuel for each
unit is natural gas, but each unit is also equipped to burn low sulfur fuel oil. Cross-sections
of the two boilers are shown in Figure 1, and their  design characteristics are compared  in
Table  1.

Encina Unit 2 is a 110 MW Babcock and Wilcox Company boiler.  The unit is fired from
the front wall with ten burners arranged in two elevations of five burners. The unit operates
with  balanced  draft  maintained by  forced  draft  and  induced draft fans.    Flue gas
recirculation (FOR) injected between water tubes on the back wall of the lower furnace is
a primary means  of  steam temperature control.   Final superheat steam  temperature  is
controlled by spray attemperation.   The final  reheat steam temperature is controlled by
distribution  dampers  in the backpass.   A  total of  sixteen existing observation ports are
available for urea injection in two elevations of the upper furnace. One elevation is located
adjacent to the furnace exit and entrance to the  convective pass, while the second elevation
is about 12 feet below,  near the arch of the furnace.

Etiwanda Unit 3 is a 320 MW Combustion Engineering boiler. This is a tangentially-fired
boiler with twin furnaces separated by a division wall.  Etiwanda Unit 3 operates  with a
pressurized furnace.  This unit is unique in its downward flow arrangement with the burner
assemblies located at  the top of the boiler.  The burner assemblies consist of three tiers  of
gas and oil burners located in the corners of each furnace. Tilt of the burner assemblies  is
a primary means of reheat steam temperature control.  FOR is injected into the windbox for
NOX control and for steam temperature control at low  loads. Spray attemperators maintain
final steam temperatures. Twelve existing observation ports arranged in two elevations near
the furnace exit were initially used for urea injection.  Additional ports were installed based
on initial test results and modeling efforts to improve NO, control performance over a wider
load range.

Etiwanda Unit 3 differs from Encina Unit 2 in a number of ways which can affect urea NOX
reduction performance.  These differences include:

       Boiler dimensions  and geometry  Etiwanda Unit 3 is physically larger than
         Encina Unit 2 with a larger furnace cross-section.  In addition,  Etiwanda
         Unit 3 has a divided furnace which limits access  to the furnace cross-section
         by urea injectors to two walls  rather than  three  walls as at Encina Unit 2;

       Firing  configuration   Etiwanda Unit 3  is  tangentially  down-fired while
         Encina  Unit 2  is a conventional front  wall-fired  boiler.   The  firing
         configuration, and the furnace geometry affect the furnace flow field and
         thus can be expected to  affect mixing  of injected urea with the furnace
         gases;
                                        6A-3

-------
        Thermal environment   At full load,  gas temperatures in the region of the
        furnace exit are  significantly higher  at Etiwanda Unit 3 (2400F) than at
        Encina Unit  2  (2250F).   Since  the  urea NOX reduction reactions  are
        temperature sensitive, differences in  injector configurations and  resulting
        performance can be expected;

        Combustion conditions   The  combustion  conditions at Etiwanda Unit  3
        result in significantly lower initial NOX levels than found at Encina Unit 2.
        At full load on gas fuel, for example,  NOX emissions at Etiwanda Unit 3 are
        as low as 90 ppm (@  3% O2) compared to  225 ppm (@ 3% O2)  at Encina
        Unit 2 with all-burners-in-service (ABIS). This is the result of NOX controls
        that have been in place since the 1970's, consisting of FOR and  two-stage
        combustion achieved with burners-out-of-service (BOOS).
SENSITIVITY OF UREA NOX REDUCTION PERFORMANCE

Results from urea injection trials of both Encina Unit 2 and Etiwanda Unit 3 are indicative
of the key factors affecting NOX reduction potential.  While  boiler operating  conditions
directly affected NOX reduction achieved with urea injection, the injection conditions and
configurations could be adjusted to  ultimately minimize stack NOX emissions over a range
of conditions on each unit.
Effect of Operating Conditions

Previous urea injection testing  at Encina Unit 2 was conducted  with the boiler operating
with ABIS(1).   Initially, urea  injection  was evaluated as a  cost-effective NOX control
alternative to the combustion modification techniques typically used in the SDG&E system
to meet current NOX regulations.  The combustion modification  techniques reduce overall
boiler efficiency compared to the higher,  NOx-producing ABIS  operating mode.  With urea
injection, however, NOX emissions could meet existing NOX regulations while operating with
the more efficient ABIS. Subsequent testing has been completed to evaluate urea injection
in conjunction with alternate, or modified,  combustion conditions.

The firing configurations evaluated included ABIS, air biasing, BOOS, and fuel biasing.
These  alternatives were evaluated to determine the overall NOX reductions possible by
coupling urea injection with modified combustion conditions. ABIS represents conventional
operation with balanced fuel and air for all the burners,  resulting in  high baseline NOX
emissions.  Air biasing was achieved with ABIS  by closing  the registers to the lower burner
elevation and thus diverting air to the upper level.  This results in staged combustion, with
the lower burners operating fuel-rich  and  the upper burners operating fuel-lean.  The effect
of staged combustion achieved  with air biasing not only reduced baseline NOX emissions,
but also affected the heat release distribution through the boiler by delaying the mixing of
fuel and air.  BOOS  operation was achieved by shutting  the  fuel off  to three of the  ten
                                        6A-4

-------
burners.  This redistributes the fuel to the remaining burners and results in those burners
operating fuel-rich.  BOOS operation thus also results in staged combustion and reduced
NOX emissions.  Since  the fuel  distribution  is changed with BOOS, the heat release
distribution also changes.

In gas fuel biasing, some of the fuel is diverted from the upper elevation of burners to the
lower elevations. This increases the heat release into the lower furnace and achieves staged
combustion.  Compared to air biasing and BOOS operation, which delay fuel and air mixing
by varying air distribution or by discrete changes in fuel distribution, fuel biasing provides
more uniform changes in fuel distribution such that slightly fuel-rich and slightly fuel-lean
zones are created. The result with fuel biasing is a more confined heat release zone due to
more balanced fuel and air mixing and, more importantly, the diversion of fuel to the lower
burner elevation.

The urea NOX reduction performance varied for the different combustion modes at Encina
Unit 2, as the data in Figure 2 illustrate.  Corresponding NOX emissions are shown in Figure
3.   The data presented  in Figures 2  and 3  represent urea NOX reduction performance
resulting from  the injection configuration optimized for  ABIS operation.  No attempt was
made in  these trials to optimize performance for each operating condition.  Thus, injection
nozzle characteristics and  injection configuration were  constant. The highest percentage
reductions  were achieved with ABIS operation and the lowest  with BOOS operation.
Differences  in  measured  performance  can be attributed directly to  changes  in boiler
conditions.  The data set  presented in the two figures indicates that differences in NOX
reduction performance can be attributed both to the different initial NOX levels produced by
the different combustion configurations  and to the effect on the temperature distribution
through  the boiler.

Analogous variations in urea NOX reduction performance with changing operating conditions
were documented at Etiwanda Unit 3(2). Figure 4 illustrates the effect of various combustion
conditions  on NOX reduction while Figure 5  presents  the corresponding  NOX  emissions
levels.  Included in the data presented  from Etiwanda Unit 3 are urea injection test results
with normal, as found fuel oil-fired conditions; normal, as found gas-fired conditions; and
modified gas-fired combustion conditions. The modified combustion conditions at Etiwanda
Unit 3 comprised adjustment of burner tilt to  horizontal for all loads with increased FOR
flowrate. As in the case for the Encina Unit 2 data set, the urea injection configuration was
not optimized for each operating condition.

The highest NOX reductions achieved at Etiwanda Unit 3 were with fuel oil. Fuel oil-firing
improves NOX reductions due to producing more favorable temperatures in the boiler (due
to differences in heat transfer characteristics between oil and gas fuels). Furnace exit gas
temperatures are about 200F lower for oil-firing than comparable gas-fired conditions.

NOX reductions over 30 percent were achieved with gas-firing over the load range of 80 to
320 MW.   Changes  in  combustion conditions, however, resulted  in  variations in  NOX
reduction performance.  Even with the variations in urea system performance, the lowest
                                         6A-5

-------
NOX emission levels, down to 21 to 45 ppm (@ 3%  O2) depending on load, were achieved
by coupling low NOX, modified combustion conditions with urea injection.
Effect of Urea Injection Parameters

Tests to optimize urea injection performance at each unit involved parametric evaluation of
urea  injection  process variables.  The variables considered  included:  atomizer design,
solution  flow and pressure, location  and injector orientation at each injection  location.
Conclusions from these parametric tests for both units include the following0'2':

         Atomizer design and the resulting  spray characteristics (spray distribution
          and angle, droplet size  distribution, and injection momentum) affect NOX
          reduction performance.  The effect of these atomizer specific characteristics
          are related to the penetration  of  urea  spray  into  the furnace flow,  the
          resulting mixing of urea with the furnace gases, and the rate of evaporation
          and the ultimate location of  release of urea into the furnace gases;

        The location of injectors and their orientation can improve NOX reduction
          performance by taking  advantage  of furnace flow  dynamics  to  enhance
          mixing of urea  with the furnace  gases and maximize residence time  at
          optimum reaction temperatures.

 Because of the fundamental differences in the thermal and mixing environments presented
 by the two different units, the  injector design and performance characteristics (i.e., droplet
 size  distribution, spray angle, injection momentum,  etc.) were significantly different.  In
 relative terms,  the  requirements  for  Encina Unit 2 compared to Etiwanda Unit 3 were
 injectors which produced small urea solution  droplets; lower injection momentum to cover
 the furnace  gas flow across the entire cross-section  of the boiler; and spray angle, shape,
 and location of ports to  inject across the cross-flowing stream.  These  requirements are
 consistent with the characteristic  differences between the two units, including:

        Favorable furnace gas temperatures in the region of injection at Encina Unit
          2  for urea  NOX reduction  reactions to  occur,  thus  requiring  the fast
          evaporation and release  of urea from small solution droplets;

        Small furnace cross-section dimensions in  the region of injection requiring
          only relatively low injection  momentum for adequate penetration and mixing
          of urea droplets with the furnace droplets;

         More uniform furnace gas flow with less cross-mixing due to the front wall
          firing  configuration  compared  to  the swirling flow field  resulting  from
          tangential firing, requiring use of ports physically spaced across the boiler.
                                         6A-6

-------
The requirements for Etiwanda Unit 3, on the other hand, were satisfied by urea solution
injection characteristics which included large droplets that would delay the evaporation and
release of urea from the high temperatures at the point of injection for reaction in lower
temperature regions. In addition, the injectors and locations were developed to optimize the
distribution and mixing of the urea  solution  by taking advantage of the furnace flow
dynamics of the tangentially, down-fired configuration.  In fact,  in a brief series of trials
to establish a direct comparison for urea injection between Encina Unit 2 and Etiwanda Unit
3, the injectors achieving optimum performance  at Encina  Unit  2 were  found to  achieve
essentially  no NOX reduction at Etiwanda Unit 3 at full load conditions.
OPTIMIZATION FOR VARYING CONDITIONS

The data from these two utility boilers demonstrate that unit design and operating conditions
can affect urea NOX reduction performance.  Since urea systems are designed by necessity
for  optimum performance  at  selected,  typical operating conditions,  NOX  reduction
performance will vary.  However, the design of urea injection and  control systems can
incorporate adjustable parameters to accommodate intermediate or varying conditions. This
potential to control over varying conditions has been demonstrated at both Encina Unit 2 and
Etiwanda Unit 3.
Multiple Level Injection

At Encina Unit 2, for example, simultaneous injection from multiple levels improved NOX
removal at both high and  low loads<2).  In a multiple injection configuration, a reduced
dosage  of urea (lower NSR) is injected at  each elevation.  This improves urea utilization
and, in turn, the overall NOX removal. This improved utilization also reduces byproduct
NH3 emissions.  Figure 6 compares NOX reduction performance at Encina Unit 2 achieved
with bi-level injection for natural gas and fuel oil-firing. The method of bi-level injection
reduced the  sensitivity  of NOX  removal to load.   In addition,  similar performance was
achieved for the two different fuels even though the resulting furnace temperature profiles
are distinctly different.
Injection Location and Orientation

Another technique used at both units to adjust for varying operating conditions was adjusting
injection location  by varying injector orientation.   In practical  applications of the  urea
injection process,  boiler penetrations to accommodate urea  injectors  will be selected  to
provide access into favorable temperature  regions for a limited number of conditions  or
loads.  To maintain urea NOX reduction performance for intermediate loads or changes  in
operating conditions, the orientation of the  injectors  can be used to adjust the  relative
location of urea injection.  Recent tests were conducted at Encina Unit 2  to evaluate the
optimization of urea injection with the combustion  modification technique of fuel biasing
                                        6A-7

-------
to achieve minimum stack NOX emissions. The test results illustrate how varying orientation
from available injection locations can improve performance and how orientation can be used
to maintain NOX reduction performance as operating conditions vary.

Urea NOX reduction performance was evaluated with and without fuel biasing by screening
injection location and orientation.  Tests were completed for loads of 80 MW and 50 MW.
At 80 MW with ABIS operation, the highest NOX reduction achieved was 44.3 percent using
the  lower level injectors only pointed up and urea injected at a rate of NSR = 2.0.  This
reduction resulted in NOX emissions of 50 ppm (@  3% O2) from a baseline of 91 ppm. With
fuel biasing at the same load, however, the highest  NOX reduction achieved was 29.1 percent
using simultaneous  bi-level injection with both the upper and lower elevations of nozzles
pointed up and urea  injected at NSR = 2.0.  The optimum urea injection configurations thus
shifted for the two different firing modes.

The reasons  for this shift appear to be a shift  in  furnace temperature.  Furnace exit
temperatures  increased about 40F with fuel  biasing.  As a result,  NOX  reduction was
improved by injection at a higher,  and therefore cooler, elevation for fuel biasing conditions
than for  normal ABIS operation.   Although relative urea NOX reduction performance was
decreased with  fuel  biasing compared to ABIS, stack NOX emissions were reduced from 50
ppm (@  3% O2) for ABIS and urea down to 38 ppm (@ 3%  O2) for fuel biasing and urea.

At  50  MW the data indicate that, for ABIS operation, injecting urea through  the lower
elevation with nozzles pointed upward achieved the highest NOX reduction. For fuel bias
operation, however, the best configuration  was bi-level injection with the upper elevation
injectors pointed down and the lower elevation injectors pointed up.  As for the 80  MW
case, the shift in optimum injection configuration  for the two operating conditions suggest
contributing affect  of a change in furnace gas temperature.  The  data  also indicate that
significant reductions can be achieved for low initial NOX levels, resulting in stack emissions
down to  23 ppm for an NSR =  1.7.
Dilution Water Flow and Injection Momentum

At Etiwanda Unit 3, three elevations of injection ports were determined to provide coverage
over the unit's  normal load range,  80 to 320 MW,  as shown  in  Figure 7.   However,
Etiwanda Unit 3 is also routinely operated down to 20 MW.  Test results demonstrated that
dilution water flow could be used in conjunction with injector elevation and orientation to
adjust the ultimate fate of urea droplets and achieve NOX reductions at loads less than 160
MW.  By varying dilution water flow, the solution concentration,  injection momentum, and
resulting droplet size distribution is changed.  The parameters directly affect the point at
which the urea is released from solution to react with the furnace gases.

The performance of the urea NOX reduction  system at Etiwanda Unit 3 is illustrated  in
Figure 8. The optimized system is used together with combustion modifications to achieve
NOX levels of 20 to 45 ppm  over the entire load range of 20 to 320  MW.  This represents
                                       6A-8

-------
significant reductions in  NOX compared to normal,  as  found conditions also shown for
reference.  In addition to the NOX reductions achieved, the available data indicate that
byproduct NH3  emissions  below 20  ppm could be  maintained  up to  urea  flowrates
corresponding to NSR = 2.0.   Figure 9  illustrates typical  NH3 emissions measured at
Etiwanda Unit 3.
CONCLUSIONS

The effectiveness of the urea NOX reduction process is sensitive to temperature and mixing
phenomena as well as chemical stoichiometry (NSR). Since the urea NOX reduction process
occurs within the  boiler  furnace, the ultimate performance  of the urea process is thus
dependent on boiler design and operating characteristics.  Although the design of SNCR
systems must attempt to address these factors,  realistic limitations must be imposed on the
range of expected boiler operating conditions (fuel type, load,  burner firing pattern, excess
air, FOR flowrate, etc.) over which the system performance can be optimized.

To accommodate differences in boiler design and variations in operating conditions, urea
injection process parameters can be adjusted and optimized.  Improvements in urea NOX
reduction performance and, ultimately stack NOX emissions, can be achieved by modifying
combustion conditions,  optimizing injection location and orientation, and adjusting injection
nozzle droplet size and injection momentum.   NOX reductions of about 30 percent at NSR
= 2.0 could be achieved over the load range of 20 to 320 MW at Etiwanda Unit 3, resulting
in stack NO, emissions in  the range of 20 to 45  ppm  (@ 3%  O2)  when  combined with
combustion modifications.  At Encina Unit 2,  similar reductions and stack NOX levels (23
to 38  ppm  @  3% O^ could be achieved when  urea injection was coupled with the
combustion modification technique of fuel biasing.   In general, the data trends suggest that
for  these  gas- and oil-fired boilers, more confined  heat release zones provide a more
favorable furnace environment than deeply staged,  delayed mixing conditions.
REFERENCES

1.   J.H. Nylander, M.N. Mansour, and D.R. Brown, "Demonstration of an Automated
     Urea Injection System at Encina Unit 2," in proceedings of the Joint Symposium on
     Stationary Combustion NO, Control, EPRI Report GS-6423, July 1989.

2.   A.R.  Abele,  D.R.  Brown,  Y.  Kwan,  M.N.  Mansour,   and J.H.  Nylander,
     "Demonstration of Urea Injection for NOX Control on Utility Boilers," in proceedings:
     GEN-UPGRADE 90,  EPRI Report GS-6986, September 1990.
                                        6A-9

-------
en
>
                      Encina Unit  2
                                                                     Etiwanda Unit 3
                                     Figure 1.  Demonstration Sites

-------
70
60
50
40
30
20
10
  NOx Removal (%)
 o
     NSR  2.0
              ABIS
Air Bias
BOOS

  0      20     40      60      80     100     120

                   Load (MW)
 Figure 2. Effect of Combustion Conditions on Urea
         NOx Removal at Encina Unit 2, Gas Fuel.
                     6 A-11

-------
90
80
70
60
50
40
30
20
10
   NOx (ppm @ 3% O0)
 0
     NSR = 2.0
  o
     ABIS
20
       Air Bias
BOOS
40     60      80

   Load (MW)
    100
120
    Figure 3. Effect of Combustion Conditions on
            Stack NOx Emission Levels with Urea
            at Encina Unit 2, Gas Fuel.
                     6A-12

-------
60
   NOx Removal (%)
50
40
30
20
10
     NSR = 2.0
Oil-As Found
Gas-Comb. Mod.
Gas-As Found
  0     50    100    150   200    250   300   350
                    Load (MW)
 Figure 4. Effect of Operating Conditions on Urea NOx
         Reduction Performance at Etiwanda Unit 3.
                     6 A-13

-------
90


80


70


60


50


40


30


20


10


  0
   NOx (ppm @ 3% O? )
NSR = 2.0
       Oil - As Found

       Gas - Comb. Mod.
Gas - As Found
   0     50    100    150    200   250    300   350

                    Load  (MW)
 Figure 5. Effect of Operating Conditions on Stack NOx
         Emission Levels with Urea at Etiwanda Unit 3.
                      6A-14

-------
en
                     80
                     70
                     60
                     50
                     40
                     30
                     20
                     10
                       NOx Removal, Percent
                      50
                               -- Gas Firing  ~V-Oil Firing
60
70     80
Load, MW
90
                                                         -o
100
                                      NOx Removal, Percent
                                    40
                                    30
                                    20
                                    10
                                                                  50
                                               --Gas Firing  -V- Oil Firing
                                            60
70     80
Load, MW
90
100
                             Figure 6. Comparison  of  NOx Reduction with Bi-Level Injection
                                       for Natural Gas and Fuel Oil Firing at Encina  Unit 2.

-------
O)
CO
                     Urea  Injection
                     Ports
                  O  Unused Ports
                                         Loop 3
                                                    El. 84'
                                                    El. 641
                                                    El. 61'


                                                    El. 54'
                                                                   O
  O     O


0*0
     Loop 2
                Division Wall
                     O
  o     o


0*0
   Loop 2
                                       Side View
         Front View
                            Figure 7.      Etiwanda Unit 3- Urea Injection  Port  Locations

-------
>
-vl
    NOx (ppm  @ 3% O2)
110
100
 90
 80
 70
 60
 50
 40
 30
 20
 10
  0
   0
                                                        As Found N Ox
                                                               x Combustion Modification N Ox
                                                       Urea * Combustion Modification NOx
                               50
100
150      200
Load (MW)
250
300
350
   Figure 8. Overall NOx Reduction Performance at Etiwanda Unit 3,
            Gas Fuel.

-------
en
00
                      75
                      60
                      45
                      30
                      15
                        NH3, ppm
                       o
                        0
O 320 MW
                                                               80 MW
                                                                       O
                                                                       o
                                                 NSR

                            Figure 9.    Typical NH 3 Emission from Optimized
                                       Urea System at Etiwanda Unit 3, Gas Fuel.

-------
TABLE 1. BOILER DESIGN CHARACTERISTICS

Design
Parameter
Capacity (MW)
Firing
Configuration
Burners
Dimensions
Height (ft)
Depth (ft)
Width (ft)
Steam Flow (Ib/hr)
SH Temperature (F)
RH Temperature (F)
Steam Press, (psig)
Encina
Unit 2
110
Front
Wall
2 Rows x
5 Burner
Peabody
77.0
20.0
34.0
700,000
1000
1000
1450
Etiwanda
Unit 3
320
Tangential
Down-Fired;
Divided Furnace
3 Elev/
Corner
x 8 Corner
CE
88.1
22.1
60.0 (30/30)
2,305,000
1050
1000
2450

                  6A-19

-------
WIDENING THE UREA TEMPERATURE WINDOW
               D. P. Teixeira
      Research & Development Department
       Pacific Gas and Electric Company
           San Ramon, CA  94583
                 L J. Muzio
              T. A. Montgomery
               G. C. Quartucy
                 T. D. Martz
      Fossil Energy Research Corporation
           Laguna Hills, CA  92653

-------
                     WIDENING THE UREA TEMPERATURE WINDOW
ABSTRACT
The results of laboratory tests to widen the effective temperature range while, at the same time,
minimizing byproduct emissions for the urea  injection  SNCR process are described.  Data are
presented showing the effect of a number of additives (methane, combination of hydrocarbons, carbon
monoxide, ethylene glycol,  HMTA,  and furfural) and initial NOX level (125  and 250 ppm) on NOX
removal efficiency and byproduct emissions (NH3, CO, N2O) as a function of temperature. Several new
phenomenon not previously observed are described.  Of particular interest is the strong effect of CO
on N2O emissions during urea injection.  In addition, many additives were found to improve NO
reduction  but not NOX reduction. In these cases, the presence of additives converted the NO initially
present to NO2 and/or N2O.
                                       6A-23

-------
                      WIDENING THE UREA TEMPERATURE WINDOW
INTRODUCTION
A variety of technologies is available to control NOX emissions from fossil power plants. One attractive
option is selective non-catalytic reduction (SNCR) with urea (1_).  However, the SNCR process, which
has many attractive features, does have several disadvantages. One drawback is the relatively narrow
temperature "window" over which  the process is effective.  Another potential  disadvantage is the
emission, at least under some operating conditions, of undesirable byproducts such as NH3 or CO.
These  issues become even more important for units  which are cycled frequently  or use multiple
fuels-which is the case for many fossil plants.

Results of a series of laboratory tests to address the issues noted above through the  use of additives
to the basic urea injection process are described in the  sections which follow.  The effects of additive
type, additive concentration and initial NOX level on NOX removal and byproduct emissions as a function
of temperature are presented.

PROCESS DESCRIPTION
Conceptually, the SNCR process with urea is quite simple. An aqueous solution of urea is injected
into, and mixed  with, the flue gas  at the correct temperature.  After the mixing has been completed,
the  urea then reacts selectively to remove the NOX.

In practical applications, however, the process (and the  equipment required)  can  be much more
complicated. Non-uniformities in velocity, temperature, and NOX concentration at the point of injection,
along with the variation in the physical location of the effective process temperature range within the
boiler, depend on various operating factors including load, type of fuel fired, and length of time on a
particular fuel.  These factors often lead to multiple levels  of injection  and/or use of additives to
accommodate the shifts in temperature.
                                         6A-24

-------
PILOT SCALE TEST FACILITY
A  schematic  of the pilot-scale facility used  for these tests is shown in  Figure 1.   The pilot scale
combustor fires natural gas, doped with  NH3 to control the initial NOX level.  The combustor and test
section are refractory lined with the test section being 15 cm in diameter and 240 cm long. At the firing
rates used for these tests, the residence time in the test section is nominally 0.5 seconds, while the
temperature drop along the test section is nominally 250C/sec (450F/sec). The SNCR solutions were
injected into the combustion products at the combustor throat through a small air assist atomizer,
above the test section. The atomizer was fabricated into a water cooled  holder. The atomizer was
located at the  center  of the throat with  the spray directed downward (i.e.,  co-flowing with the
combustion products). The solutions were pumped with variable speed peristaltic pumps and metered
with rotameters.  In order to maintain a constant thermal environment in the test section, the total
amount of liquid injection was held constant at nominally 1 liter/hr.  By diluting a concentrated urea (or
other SNCR chemical) solution with distilled water, the amount of chemical reagent was varied while
a total liquid flow rate of 1 liter/hr was maintained.

Gas samples were taken at the exit of the combustor with a water-cooled probe and transported to a
series of gas  analyzers (NO/NOX, N2O, CO, CO2, and O2). The continuous measurement of N2O was
made  using an NDIR based technique  (2).  NH3  was measured using a selective ion electrode
technique.

The pilot-scale tests investigated  the effect of temperature, additives, chemical injection rate, and initial
NOX concentration on NOX removal efficiency and byproduct emissions (specifically NH3, CO, and
N20).

RESULTS
During this study, experiments were carried out at initial NOX levels of 125 ppm and 250 ppm and
ISI/NO, molar  ratios of 1 and 2.  For brevity, most of the results shown in  this paper will  be from the
tests at an initial NOX level of 125 ppm.  Results at 250 ppm will be shown for situations where the
effect of the SNCR chemical, or additive,  exhibits different behavior from that observed at the 125 ppm
level.

Baseline Performance of Urea -  No Additive
To establish a reference for comparison of results from the various additives, a series of baseline tests
were performed using urea alone.  The baseline NOX removal and byproduct emission  results over
                                         6A-25

-------
                                        BURNER FLOW SYSTEM

7
It
34

lorn COMBUSTION
ANDCOOL1NOSECIION
EIOMT CONCENTRIC
COOLINQ PROBE |_
PORTS 	 ^
r~
0X3 AND SOLID
INJECTION PORT L_
' .- h.
\ r
cm ADDITIVE IHJECTION
SECTION
"-IT
r i
L I
THERMOCOUPLE (_|
PORT 	 ^
!~
I
LJ
Dem TEST SECTION
LJ
LI
LI
r
i
LI
n




BURNER
I 	 1
J

I1-
	 |


Horn

-------
the temperature range investigated for initial NOX levels of 125 and 250 ppm and a urea injection rate
corresponding to molar ratios of nitrogen to NOX (N/NOX) of 1 and 2 are shown in Figures 2 and 3.
Figure 2 shows the results for an initial NOX level of 125 ppm.  Figure 3 shows the same data but for
an initial NOX level of 250 ppm. The narrow effective process temperature range for NOX removal can
be clearly seen in both figures, as can the increasing levels of NH3 and CO byproducts as temperature
is decreased. Also shown are byproduct levels of N2O produced by the process at the test conditions.
Other investigators have also noted N2O byproducts associated with urea injection (3).

Carbon Monoxide Additive
A review of the general combustion chemistry literature showed that CO was a potential compound that
could alter the temperature dependence of the urea injection process.  This behavior was also
suggested by the data of reference 4 showing the effect of CO at high  concentrations (8000 ppm CO)
on NOX removal.  While the use of CO to modify the urea temperature window in power plant boilers
presents several difficult practical application issues, it was felt important to address the effect of CO
since all combustion devices emit some level of  CO.

For the data discussed below, the CO additive was introduced by injecting  it with the atomizing air.

NO. Removal Temperature Dependance.  Figure 4 shows the effect of CO on NOX removal as a
function of temperature at an initial NOX level of 125 ppm and N/NOX ratio of 2.  This figure shows
several interesting features:
     CO, even in relatively low amounts, has a significant impact on the NOX  removal efficiency at a
     given  temperature.  As CO levels  are increased, the NOX removal  versus temperature
     dependence shifts  to a lower temperature regime. Figure 4 shows that, increasing the CO levels
     from O ppm to 1000 ppm shifts the peak NOX removal temperature about 200F lower.
     As CO levels increase, the effective process temperature range is broadened.  For the conditions
     of Figure 4 when CO is in the 500-1000 ppm range, the window appears to be broadened by
     about 100F.
     Increasing CO also lowers the peak level of NOX removal possible. Figure 4 shows that peak NOX
     removal decreases from about 55% to 45-50% as CO increases from 0 ppm to 500 ppm; it further
     decreases to about 45% as CO is increased to 1000 ppm. Similar behavior is noted for the other
     conditions investigated.

CO Byproduct Emissions. The final CO levels resulting from addition of CO to the urea process are
shown in Figure 5. As can be seen, at the  lowest temperature evaluated,  1470F, CO emissions
increase as the initial amount of CO addition is increased.  However, for temperatures at or above
1600-1650F,  final CO levels are practically independent of the amount of CO added.
                                         6A-27

-------
E*.
Q- c
LU
  O
80




70



60




50




40




30




20




 10




  0




-10




-20
                      NH3
                                       ANOx
                                                                 (a)  N/NOX =
        1400  1500  1600  1700  1800  1900  2000  2100  2200  2300



                              Temperature, F
                                                               (b)  N/NO, = 2
     1400  1500  1600   1700  1800  1900  2000 2100  2200  2300
                           Temperature, F
        Figure 2.  NOX Reduction and Byproduct Emissions with Urea Injection

                            (Initial NOX = 125 ppm)
                                  6A-28

-------
E :
o. c
o. g
w "-

'E x

UJO
100





 90




 80





 70





 60




 50





 40





 30





 20





 10 -
                0.5XNII3
                                        ANOx
                                                                  (a) N/NO, = 1
       0

       1400  1500  1600  1700  1800  1900 2000  2100  2200  2300
                            Temperature, F
   .
Q- c
CL g
100




 90





 80




 70





 60





 50





 40





 30





 20





 10





 0

 1400  1500  1600   1700  1800  1900  2000  2100  2200  2300





                       Temperature, F
                                                                  (b) N/NO. = 2
         Figure 3.  NO, Reduction and Byproduct Emissions with Urea Injection

                              (Initial NOX = 250 ppm)
                                    6A-29

-------
"O

-------
NH, Byproduct Emissions.  Since minimum unreacted NH3 from the SNCR process is desirable both
from an environmental, as well as boiler impact standpoint, measurements of the byproduct NH3 were
made.  Figure 6 shows the results of these measurements for an initial NOX level of 125 ppm and
N/NOX ratio of 2.  As expected, NH3  emissions decrease as temperature increases.  However, NH3
levels at any given temperature, were found to decrease significantly as CO levels increased.

N;O Byproduct Emissions.  The most  interesting influence of CO on the urea injection process was on
the N2O byproduct characteristics (Figure 7). The effect of  CO on N2O is  strongly temperature
dependent. At higher temperatures (approximately 1900F and above),  N2O levels tend to merge to
a similar low level for all combinations of CO, initial NOX and N/NOX. At these high temperatures, N2O
tends to decrease rapidly to very low levels as temperature is increased.

However, at the lower temperatures investigated (1500-1600F), a very different behavior can be seen;
N2O levels increase with increasing CO levels. For example, at an initial NOX of 125 ppm and N/NOX
= 2, N2O increases from about  10 ppm to 35 ppm as CO is  increased from  0 ppm to  1000  ppm.
Although not shown, N2O emissions at these lower temperatures also increase as the amount of urea
(i.e. N/NOJ and initial level of NOX increase. At the highest initial  NOX (250 ppm),N/NOx (2), and CO
(2000 ppm) levels investigated, N2O concentrations approach 100 ppm.

At the intermediate temperatures (between 1500F and 1900F), there  is a transition from the low
temperature behavior to the  high temperature  behavior.   At  the  lower CO levels,  increasing
temperatures first produce an increase in  N2O then a decrease as temperature is increased,  with an
obvious maximum in the N2O as a function of temperature. At higher CO levels, N2O initially remains
relatively constant as temperature increases, then drops off abruptly.

Implications. There are several important practical implications regarding the influence of CO on the
urea injection process, in particular the N20 characteristics. First, to minimize N2O production in the
urea injection process it is important to maintain low CO levels.

Second, when using urea injection,  a "coupling" between the combustion process and the urea
injection process can occur, i.e. CO produced in the burner region influences the SNCR performance.
This may be especially true for low NOX burner systems where, as is well known, there are frequently
trade-offs between the NOX reduction  and CO  levels.

Lastly, the effect of CO on N2O formation may  explain some of the differences in N2O levels reported
by various researchers at a recent workshop on N2O (5).

                                        6A-31

-------
E
Q.
Q.
CO
     250
     200
     150
     100
      50
       0
 CO Addition
A   0 ppm
A  65 ppm
 *  125 ppm
      1400  1500  1600  1700  1800  1900  2000  2100  2200  2300

                            Temperature, F
e
a
0
CJ
         Figure 6.  Effect of CO Additive with Urea on Byproduct NH3 Emissions
                          (Initial NOX = 125 ppm; N/NOX = 2)
      so 1.	r
                                                                  CO addition
                                                                O    0 ppm
                                                                   500 ppm
                                                                n  1 ooo ppm
      1400  1500  1600  1700  1800   1900 2000  2100  2200  2300
                            Temperature, F
         Figure 7.  Effect of CO Additive with Urea on Byproduct N2O Emissions
                         (Initial NOX = 125 ppm; N/NOX = 2)
                                   6A-32

-------
Methane - Additive
Methane (CH4) was also investigated as a potential additive to alter the urea/NOx removal temperature
dependance. The results shown are for tests conducted at 1600F, N/NOX = 2, initial NOX levels of 125
ppm and 250 ppm, and CH4/NOX molar ratios of 0, 0.5 and 1. Figures 8 and 9 show the results.

For the initial NOX levels investigated, both NO and NOX (NO+NO2) levels decrease with the addition
of urea alone. However, when  methane is added, while the NO levels continue to decrease for both
initial NOX levels, the effect on NOX differs.  At an initial NOX level of 250 ppm, NOX levels continue to
decrease with the addition of CH4. However, at the lower initial NOX level of 125 ppm, while NO levels
decrease with CH4 addition, NOX levels remain constant. At this lower initial NOX level, the effect of
the CH4 is to oxidize NO to NO2, rather than to enhance the SNCR process.

The effect of methane additive with urea on N2O emissions is also included in Figures 8 and 9. At both
initial NOX levels, methane promotes the formation of N2O as a byproduct; the N2O levels increase with
increasing amounts of CH4.

Efforts to explain the significantly different behavior between the two initial NOX cases have to  date
been unsuccessful. The possibility of hydrocarbon interference with the N2O measurements, which is
known to occur for the instrument used, was considered but could not explain the results observed.

Multiple Additives

NOV Removal Efficiency.  A Japanese patent (6) identifies multiple hydrocarbon additives used with
urea  to broaden  the temperature window.  A  specific example was presented for  the  following
conditions: urea at N/NOX = 4; initial NOX = 990 ppm; temperature = HOOT; and additives consisting
of ethylene glycol, propane and carbon. Without the additives (i.e. urea only) the NOX reduction, as
expected, was low, under 10%.  With the additives, the NOX reduction was increased to almost 75%.

A series of tests were performed to verify the performance of the multiple additives under the following
conditions: initial NOX = 790 ppm; temperature of 1400F; N/NOX = 4; ethylene glycol/urea concentration
of 9.5%; propane  to urea of 57%; and carbon/urea of 33%.  All concentration  ratios are on a molar
basis. The tests were conducted sequentially to evaluate the individual, as well as combined, effect.
The results are shown in Figure 10.

As can be seen in Figure 10, the addition of glycol resulted in an increase of NOX removal from about
10% with urea only to about 20%.  Addition of propane increased the NOX removal to almost 55%.

                                         6A-33

-------
     150
 Q-
 Q.


O
C\J
CM
O
O
             Initial
                                                          NO + NO2 + N20
                    NO + NO2
                           Urea
   Urea +

0.5 ChM/Uiea
 Urea +

1.0 CH4/Urea
      Figure 8. Effect of Methane Additive with Urea Injection on NO, NO2, and N2O
               (Temperature = 1600F; Initial NOX = 125 ppm; N/NOX = 2)
      300
             Initial
                                                          NO + NO2 + N2O
                                                          NO + NO2

                                                          ND
                           Urea
    Urea+        Urea +

 0.5 CH4/Urea    1.0 CH4/Urea
      Figure 9.  Effect of Methane Additive with Urea Injection on NO, NO2 and N2O
               (Temperature = 1600F; Initial NOX = 250 ppm; N/NOX = 2)
                                   6A-34

-------
I
I
DC
o
100


 90

 80


 70


 60


 50


 40


 30


 20


 10


  0
                    Urea
                    + Glycol
                    + Propane
                    + Carbon
Urea
+Glycol
+ Propane
Urea
+Glycol
+Propane
+ Carbon -
                      D  %ANO

                      M  %ANOx
               (Reference 6)
                                    Present Tests
           Figure 10.  Effect of Multiple Additives on NOX Reduciton with Urea
 o
 TJ
 03
 cc.
 x
 O
      25
      20
      15
      10
                            Urea
                            + Glycol
                            + Propane
                   Urea
                            Urea
                            + Glycol
                                        Urea
                                        f Glycol
                                        + Methane
                       %ANOx
           Figure 11.  Effect of Multiple Additives on NOX Reduction with Urea
               (Temperature = 1400F; Initial NOX = 760 ppm; N/NOX = 1)
                                       6A-35

-------
 Further addition of the carbon actually resulted in a small deterioration in NOX removal.  While the 55%
 removal did not quite match the 75% value cited in the patent, the results were sufficiently encouraging
 that additional tests were conducted.

 The next series of tests were done under nominally the same conditions as above (initial NOX of 755
 ppm; temperature of 1400F; ethylene glycol/urea of 9.9%; propane/urea of 60%), but at a lower N/NOX
 ratio of 1.0.  Results of these tests are shown in Figure 11.  No improvements in NOX removal were
 noted for the case of glycol-only addition.  NOX removal increased to about 15% for the glycol plus
 propane  case.  No change was seen in NOX removal when methane was  substituted for propane.  It
 should also be noted that, in the urea plus glycol plus propane, or methane, cases, the NOX removal
 was significantly lower than the NO removal.

 The final series of tests considered the multiple additive concept at conditions of greatest practical
 interest:  temperature of 1600F; initial NOX levels of 125 ppm and 250 ppm; N/NOX of 2; CH4/NOX
 values of 0, 0.5, 1; ethylene glycol/urea of 10%. The results of these tests are shown in Figures 12
 and 13. At 125 ppm initial NOX  (Figure 12), little benefit of the multiple additives was observed. Some
 improvement in  the NO  removal  was noted  for glycol addition alone.   Very little NOX  removal
 improvement was noted. As discussed previously, no effect of methane on either NOX or NO removal
 was observed other than  to increase N20 emissions.   A case where methanol  was substituted for
 glycol at  a methanol/urea  of 10% was investigated and yielded virtually identical  results.

 Contrary  to the general lack of improvement in NOX performance at 125 ppm, meaningful improvement
 in NOX (and  NO) removal  was  observed  when glycol was added, and/or when CH4 was added at a
 higher  initial NOX level of 250 ppm (Figure 13). As will be seen in the next section, NO2 and N2O
 formation from the initial NO explains at least a portion of the difference  between  the NOX  and NO
 removal levels.

 NO0/N0O  Characteristics.  Data summarizing the NO2 and N2O characteristics of the multiple additive
concept are summarized in Table 1.
                                        6A-36

-------
             E
             o
             n_
             6
             (M
             C\J
             O
             z:
             o"
                 150
                 125 -
                 100 -
                                             Additive:  Elhylene Glycol
                                                                        NO + NO2 + N2O
                                                NO i NO2
    Utea N/NO (molar)
E.  Glycol/Urea  (molar)
     CH4/Urea (molar)
                        Initial
 0.0
 0.0
 0.0
Urea

2.0
0.0
0.0
                                            Urea/Add.    Urea/Add.    Urea/Add.
            2.0
            0.1
            0.0
           2.0
           0.1
           0.5
            2.0
            0.1
            1.0
           Figure 12.  Effect of Multiple Additives (Ethylene Glycol and Methane) on
                                  NO, Reduction with Urea
                 (Temperature = 1600F; Initial NOX = 125 ppm; N/NOX = 2)
                300
                                                                       NO + NO2 + N2O
                                            Additive:  Elhylene Glycol |
     Urea N/NO (molar)
  E.Glycol/Urea  (molar)
      CH4/Urea (molar)
                       Initial
0.0
0.0
0.0
                                  Uiea
2.0
0.0
0.0
                                           Urea/Add.   Urea/Add.    Urea/Add.
2.0
0.1
0.0
2.0
0.1
0.5
                                2.0
                                0.1
                                1.0
         Figure 13.  Effect of Multiple Additives (Ethylene Glycol and Methane) on
                                 NOX Reduction with Urea
                (Temperature = 1600F; Initial NOX = 250 ppm; N/NOX = 2)
                                     6A-37

-------
                                         Table 1
                           NO2 AND N2O  MULTIPLE ADDITIVES
T = 1600F N/NO = 2 Ethylene Glycol/Urea = 10%
Initial NO.
ppm
125
125
125
250
250
250
CHAlrea
molar
0
0.5
1
0
0.5
1
Final NO,
egm
19
24
25
33
37
31
N^O
ppm
13
21
28
26
42
63
N,O + NO,
ppm
32
45
53
59
79
94
Although not shown, virtually identical data were collected for methanol under the same test conditions.
As can be seen, a portion of the original NO appears in the products as NO2 and N2O.  NO2 levels
were roughly in proportion to the initial NOX levels and tended to increase as the CH4/urea increased.
Likewise, N20 increased approximately in proportion to the initial NOX and as CH4/urea was increased.

HMTA/Furfural Additives
A review of the patent literature also indicated that the addition of hexamethylenetetramine, C6H12N4
(HMTA),  and furfural (C5H4O2) to urea results in a broadening of the  effective process temperature
range for NOX reduction (7,8,9).

A series  of tests were conducted  to  evaluate the effectiveness of these compounds.   The tests
evaluated HMTA addition alone and in combination with furfural.  A temperature of 1650F was used
for these tests. The quantity of additives used in the tests was estimated based on the data contained
in References 7-9.  Test conditions were as follows: initial NOX level of 250 ppm; HMTA/urea of 0.2;
furfural/urea of 3.65 (all on a molar basis).  Results of these tests are  shown in Figure 14.

Examination of Figure 14 shows that the addition of HMTA alone, or the HMTA/furfural mixture, led to
a meaningful  improvement in  both  NO and  NOX  removal.  However,  the improvement  in NOX is
considerably lower than the improvement for NO removal.  Evaluation  of  the final NO2 levels (Table
                                        6A-38

-------
cc
X
O
     50
     40
     30
     20
     10
             Urea
             Alone
 HMTA    HMTA/Furfural
Addition       Addition
 Urea
Alone
D  %ANO

M  %ANOX
          N/NO = 1.0    N/NO = 1.0    N/NO = 1.0    N/NO = 2.0    Urea Alone
          N/NO = 1.0    N/NO = 1.8    N/NO =1.8    N/NO = 2.0    Urea and Additive
    Figure 14.  Effect of HMTA and Furfural on NO and NOX Reduction with Urea
                  (Temperature = 1650F; Initial NOX = 250 ppm)
                                    6A-39

-------
2) showed that a portion of the initial NO was being oxidized to NO2.  Unfortunately, data for N2O was
not collected during this test series, so a more complete assessment of the impact of HMTA/furfural
on byproducts could not be done.

                                         Table 2
                       NO2 LEVELS WITH HMT A/FURFURAL ADDITIVE

HMTA/Urea
(Molar)
0.2
0.2
Temperature =
Furfural/HMTA
(Molar)
0
3.65
1650F Urea/NOx = 1
Initial NO,
(ppm)
15
15

Final NO,
(ppm)
60
58
 Since the nitrogen in the HMTA increases the effective N/NOX ratio from 1 to 1.8, Figure 14 also shows
 the NOX removal expected for the urea only case at N/NOX = 2. This allows an alternative comparison
 of the behavior of HMTA since one alternative to the use of HMTA additives would be to increase the
 N/NOX by increasing the amount of urea injected in place of adding the HMTA.  As can be seen,
 increasing the amount of urea injected provided a comparable degree of NOX removal when compared
 to HMTA, or HMTA/furfural addition.

 Future Research

 Continuation of efforts to find additives or alternative reducing agents to improve the SNCR process
 will be pursued in the future. In  addition, a series of tests to evaluate the effect of CO additive with
 NH3 as a reducing agent will be conducted and compared to the urea plus CO additive results.

 CONCLUSIONS
 A number of unexpected results  were  observed when testing various additives to the urea injection
 process:
            CO shifts  and broadens  the temperature window even at low CO levels; in addition,
            significant changes in the byproduct emissions, especially for N2O, occur.
                                        6A-40

-------
            CH4 exhibits markedly different NOX and NO removal behavior depending on the initial
            NOX level. Reasons for this behavior are not understood. CH4 addition also leads to the
            conversion of NO to NO2 (oxidation) and the formation of N2O.

            As with CH4, the use of multiple hydrocarbon  additives leads to different NOX and NO
            removal behavior, depending on the initial NOX level.  The use of multiple additives also
            leads to the conversion of a portion of the initial NO to NO2 and N2O.

            The HMTA and furfural additives lead to the conversion of NO to N02.  As a result, NO
            removal improves to a greater extent than the NO,, removal.  Further, it appears that the
            improvement in NOX reduction can be attributed to the increased N/NOX injection  ratio
            that results from the addition of HMTA.
In addition to the specific conclusions reached above for the individual additives, overall examination

of the results indicates a more general conclusion:  The chemistry involved in urea NOX removal is

more complex than previously thought.  As a result, when considering employment of the process to

a specific application, careful consideration of the initial NOX level and the levels of trace combustion

product species,  including hydrocarbons and CO, is required.
                                        6A-41

-------
                                     REFERENCES


1.      Arand, J. K., Muzio, L. J., Setter, J.  G., U.S. Patent 4.208.386. June 17, 1980.

2.      Montgomery, T. A., et al, "Continuous Infrared Analysis of N2O in  Combustion Products",
       JAPCA Vol. 39, No. 5, May 1989.

3.      Jodal, et al, "Pilot Scale Experiments with Ammonia and Urea as Reductants in Selective
       Non-Catalytic Reduction of Nitric  Oxide", 23rd  International Symposium on Combustion,
       Orleans, France, July 1990.

4.      Siebers, D. L. and Caton, J. A., "Removal of Nitric Oxide from Exhaust Gas with Cyanuric
       Acid", Paper No. WSS/CI88-66,  1988 Fall Meeting of the  Western States Section of the
       Combustion Institute, Dana Point, California, October 1988.

5.      Second European Workshop on N2O Emissions, Lisbon, Portugal, June 1990.

6.      Kuze, T., et al, Japanese Patent 53128023, November 8, 1978.

7.      Bowers, E. B., U.S. Patent 4,751.065, June 14, 1988.

8.      Epperly, R. E. and Sullivan, J. C., U.S. Patent 4,770.863. September 1988.

9.      Epperly, W. R., O'Leary, J. H., Sullivan, J. C., U.S. Patent 4.780,289. October 25, 1988.
                                        6A-42

-------
    CATALYTIC  FABRIC  FILTRATION  FOR
SIMULTANEOUS NOx AND PARTICULATE CONTROL
   Greg F. Weber and Dennis L. Laudal
Energy and Environmental Research Center
       University of North Dakota
      Box 8213, University Station
          Grand Forks,  ND  58202

 Patrick  F. Aubourg  and Marie  Kalinowski
        Owens-Corning  Fiberglass
               P.O. Box 415
        Granville, OH  43023-0415

               Prepared for

    Electric  Power Research  Institute
          3412 Hillview Avenue
           Palo Alto,  CA 94303

-------
                          CATALYTIC FABRIC FILTRATION FOR
                      SIMULTANEOUS NO. AND PARTICULATE CONTROL
                         Greg  F.  Weber  and  Dennis  L.  Laudal
                      Energy and  Environmental  Research  Center
                            University of  North Dakota
                           Box 8213, University Station
                              Grand  Forks,  ND  58202
                      Patrick F. Aubourg and Marie Kalinowski
                              Owens-Corning Fiberglas
                                    P.O.  Box 415
                             Granville, OH  43023-0415
                                    Prepared  for

                         Electric Power Research Institute
                                3412  Hillview Avenue
                                Palo  Alto,  CA  94303
ABSTRACT

The Energy and Environmental Research Center (EERC) at the University of North
Dakota (UNO) has been working with Owens-Corning Fiberglas Corporation (OCF) for
several years evaluating Catalytic Fabric Filtration for simultaneous NOX and
particulate control.  Early work sponsored by OCF was presented at the 1989
EPRI/EPA NOX Symposium.   Since  April  1988,  the  U.S.  DOE  Pittsburgh Energy
Technology Center (PETC) has funded development activities at the EERC, with OCF
providing catalyst-coated fabric samples for testing.

The work has involved evaluating samples (1 ft2)  of catalyst-coated  fabric prepared
by OCF using actual  flue gas from the combustion of pulverized coal.  Dependent
variables included air-to-cloth ratio, ammonia/NO,,  molar ratio,  and  coal  type
(bituminous, subbituminous, and lignite).  Flue gas temperature was maintained at
65025F.   Resulting NOX removal  efficiency  and  ammonia slip  varied significantly
with air-to-cloth ratio.  As the air-to-cloth-ratio increased from 2 to 6 ft/min,
NOX reduction  decreased  from 85-95% to less  than  70% with corresponding ammonia
slip values ranging from 5 ppm to 360 ppm.  For the short-term (8-hour) tests
completed,  the four coals tested did not appear to have a significant effect on
catalyst-coated fabric performance.  Bench-scale tests have demonstrated that 90%
NOX reduction  can  be achieved with an  ammonia slip  of <5 ppm.
                                      6A-45

-------
                          CATALYTIC FABRIC FILTRATION FOR
                      SIMULTANEOUS NOX AND  PARTICULATE CONTROL
INTRODUCTION i BACKGROUND

In 1990, the first major reauthorization  of  the  Clean Air Act  since  1970 was
enacted by Congress and signed  into  law by the President of the United States.
Although S02 emissions are still the primary focus of acid rain control,  studies in
Europe and the United States  investigating the role  of  NOX in acid rain formation
and ozone chemistry have resulted  in NOX control  being an important component of
the new Clean Air Act (1,2).  Specifically,  the  Clean Air Act  Amendments of  1990
require a two million-ton  reduction  in NOX emissions by January 1, 1995.

Expectations are that NOX emissions will  be regulated more strictly at the local
level (state and local regulatory  agencies)  than as  currently  addressed under the
reauthorized Clean Air Act.   Therefore, technology capable of  achieving higher
levels of NOX control  than those demonstrated by low NOX burners  must be developed.

For the past six years, the  Energy and Environmental  Research  Center (EERC), using
fabrics developed by  Owens-Corning Fiberglas  (OCF),  has  pursued  the  development  of
the catalytic fabric  filtration concept as an  advanced  NOX control technology.   The
overall objective of  the project is  to evaluate  the  potential  of  a catalytic
fabric filter for simultaneous  NO, and particulate control.  Specific goals include
the following:

        90% NOX removal  efficiency with <25 ppm ammonia slip.

        A  particulate removal  efficiency of  >99.5%.

        A  bag/catalyst life  of >1 year.

        A  20% cost savings  over conventional  baghouse  and SCR control
         technology.

        Compatibility with  S02  removal  systems.

        A  nonhazardous waste material.

 Even  though  promising results were obtained  in the early bench-scale work  funded
 by OCF,  a continued effort was  needed to  further develop the  product that  would
 give  the best combination  of high  NO, removal capability, low ammonia slip,  high
 particulate removal efficiency,  and  long  catalyst/bag life.

 Specific activities have progressed  from  bench-scale  experiments  using simulated
 flue  gas  (Task A)  and flue gas  from  a pc-fired source (Task B) to pilot-scale
 experiments with  catalyst-coated bags.  Specific parametric and  fabric-screening
 tests  using simulated flue gas  (Task A) were  conducted  in which  the  fabric weave,
 coating  composition,  and coating process  were  adjusted  to develop acceptable
 fabrics  for further testing.  Task B, which  is the focus of this  paper, involved
 the testing of ten  catalyst-coated fabric samples developed by OCF using  a


                                       6A-46

-------
slipstream of flue gas from EERC's Participate Test Combustor (PTC).  Based on the
results of these bench-scale experiments, tests with catalyst-coated filter bags
are scheduled to begin in the summer of 1991.

RESULTS & DISCUSSION

The purpose of Task B was to further evaluate catalyst-coated fabric samples in
the presence of flue gas generated during pulverized coal combustion.  This was
considered necessary to begin evaluating the potential effects of fly ash on
catalytic performance:  specifically, the effects of submicron particles, volatile
species, and trace elements that could not be addressed using synthetic flue gas.
Ten catalyst-coated fabric samples (Fabrics #2, #3, #4, #5, #7,  #13, #14, #15,
#17, and #18) developed by OCF were selected for testing.  The criteria for
selecting these fabric samples for further evaluation were high NOX  removal
efficiency and/or low ammonia slip, based on Task A results.  Detailed
descriptions of eight catalyst-coated fabric samples were presented in a previous
report (3).  Fabrics #17 and #18 were catalyst-coated fabric samples recently
developed by OCF.  Fabric #17 was similar to previously tested Fabric #2, except
that a different vanadium source was used to prepare the coating, and
modifications were made to increase the surface area.

The catalyst coated on Fabric #18 was a new iron-based catalyst.  Iron compounds
have been shown to be effective catalysts for reducing NOX (4).   In  addition,  it
may broaden the temperature window for the NOX reduction reactions.

Four coals were selected for Task B testing, a medium-sulfur washed Illinois #6
bituminous (the baseline coal), a high-sulfur Pyro Kentucky bituminous, a Jacobs
Ranch subbituminous, and a South Hallsville, Texas, lignite.  Each of the ten
fabrics was tested with the washed Illinois #6 bituminous coal at air-to-cloth
ratios of 2, 3, 4, and 6 ft/min.  Ammonia slip and S03 measurements  were  made  at
each air-to-cloth ratio.  The ammonia/NOx molar ratio was to be  held constant  at
0.9; however, due to an error in calculating an orifice coefficient, several tests
were conducted at an ammonia/NOx molar ratio of 1.1.   Cloth weight in all  instances
was 14 ounces per square yard.

Based on the results of the first eight fabric-screening tests,  two fabric
samples, #2 and #13, were selected to be tested using the remaining three coals.
For the first 6 hours of the test, the air-to-cloth ratio was held constant at 3
ft/min.  However, near the end of each test, the air-to-cloth ratio was adjusted
to 2 ft/min for 1 hour and then 4 ft/min for 1 hour.  The ammonia/NOx molar ratio
was held constant at 0.9.  The slipstream sample system used to perform the tests
is shown in Figure 1.

The results of the Task B fabric-screening tests are presented in Table 1.  These
results are consistent with the values reported for Task A.  As expected, there
was a marked decrease in NOX  removal  efficiency with  increased air-to-cloth ratio.

An example of this is shown in Figure 2.  Although there was some variability in
the operation of the combustion system, NOX  removal  efficiency was relatively
constant with time.  Fabric #2 appeared to have demonstrated the best overall
performance of the first eight fabric samples tested, with respect to high NOX
removal and low ammonia slip.
                                       6A-47

-------
The results for Fabric #17, with the new vanadium source, compared favorably to
Fabric #2, which is similar in all other respects.  The two fabrics are compared
directly in Figure 3.  As can be seen, with the exception of the ammonia slip at
an air-to-cloth ratio of 2 ft/min, the results are very similar.  Figure 4 shows
the actual ammonia/NOx molar  ratio as  a  function  of  time  for Fabric  #17.   As  is
shown in the figure, the ammonia/NOx molar  ratio  averaged about  0.95 for  the  test
at an air-to-cloth ratio of 2.2 ft/min.   This may have been the reason for the
higher ammonia slip at the lowest air-to-cloth ratio.  Figure 4 data are typical
of the variability in ammonia/NOx molar  ratio for all  the tests.

For Fabric #18, the results did not seem to be very impressive (an NOX removal
efficiency of 64% at an air-to-cloth ratio of 2 ft/min);  however, this is
promising, as the coating process for iron has not been optimized.  As stated
earlier, iron presents several potential advantages over vanadium; however,
further development by OCF will be necessary to improve its performance.

From the fabric-screening data, the maximum air-to-cloth ratio that can be used
and still obtain >85% NOX removal  efficiency is 3 ft/min,  which  is consistent with
the bench-scale results using simulated flue gas  (Task A).  For all  the catalyst-
coated fabric samples, there was a marked decrease in catalytic performance at
air-to-cloth ratios of 4 and 6 ft/min.

Following completion of the first eight fabric-screening tests,  fabric samples #2
and #13 were chosen to test the effects of coal  type on fabric performance.  Both
fabrics were tested using the three remaining coals:  South Hallsville, Texas,
lignite; Jacobs Ranch subbituminous; and a Pyro Kentucky bituminous at an air-to-
cloth ratio of 3 ft/min, ammonia/NOx molar  ratio  of  0.9,  and temperature  of 650F.
Table 2 summarizes the results from these tests as well as data from the previous
screening tests using the washed Illinois #6 bituminous coal.  The data are also
represented graphically in Figures 5 and 6.

From the data, it appears that NOX removal  efficiency with Fabric #2 was  similar
(85% to 90%) for three of the four coals fired in the pilot-scale combustor.   The
exception was observed when firing the South Hallsville,  Texas,  lignite.   Although
an obvious explanation of this result (80% NOX removal  efficiency and 121  ppm
ammonia slip) is not apparent, EERC believes that the filtration characteristics
of the South Hallsville fly ash may have contributed to the observed result.
Specifically South Hallsville, Texas,  lignite is  known to produce an ash that is
difficult to collect in a fabric filter (5).  A large number of pinholes were
present in the dust cake at the conclusion of the test.  Pinholes may result in
localized areas of very high air-to-cloth ratios  which, depending on the number
and size of the pinholes, can limit contact between the flue gas and the catalyst,
resulting in decreased NOX removal  efficiency and increased ammonia  slip.

For Fabric #13, the results using South Hallsville,  Texas, lignite were more
successful, as excessive pinholing did not occur.  Although the NO^  removal
efficiency was somewhat lower, about 83% compared to 86% and 90% for the Jacobs
Ranch and Illinois #6 coals,  respectively,  the data is not conclusive.  Therefore,
the effect of coal type, if any, on catalyst-coated fabric performance has not yet
been determined.  The results using the Pyro Kentucky bituminous coal with
Fabric #13 are suspect due to an upset in the pilot-scale combustion system.
                                        6A-48

-------
Excessive slagging resulted in an unstable flame in the burner, causing an early
shutdown of the test.

Table 3 presents surface area and catalyst data for each of the catalyst-coated
fabric samples tested.  Both were measured prior to exposure to the flue gas and
after completion of the reactivity tests.  In all cases, there was a substantial
decrease in surface area after exposure to flue gas.  But, for most of the fabric
samples tested, the catalyst concentration decreased only slightly or remained
constant with exposure to flue gas.  However, this indicates that the decrease in
surface area is not due to sluffing of the catalyst from the fabric surface.  The
decrease in surface area may be due to a slight sintering effect, possible
plugging of the surface pores by submicron aerosols or fly ash particles, or due
to residual carbon burnout in the coatings.

The initial BET surface area for both Fabrics #17 and #18 was higher than previous
fabrics.  However, the surface area for Fabric #17 after exposure to flue gas
(which gave results very similar to Fabric #2) decreased to a level that was
essentially the same as that observed for Fabric #2.  For Fabric #18 (iron
catalyst), there seems to have been almost a complete collapse of surface area.
The reason for this is not known at this time; however, it was speculated by OCF
that there may be some temperature effects.  Figure 7 shows the NOX removal
efficiency as a function of the surface area after exposure to flue gas.  One
surface area point does not fit the curve.  This data point represents Fabric #7,
and a final determination concerning its validity has not been made.  Fabric #7
may be tested again during upcoming pilot-scale activities.  Although other
factors such as weave texturization may also be important, the figure shows that
NOX removal  efficiency is  directly proportional  to  the  surface  area.   Based  on  this
data, the minimum BET surface area needed to achieve 85% NO, removal  efficiency at
an air-to-cloth ratio of 3 ft/min is about 4-5 m2/g.

For Fabrics #17 and #18, N20 was measured at  the inlet  and outlet of the catalyst-
coated fabric.  The measurements are shown in Table 4.   Within the limits of the
instrument, the table shows that there is no apparent conversion of NOX  to N20
across the catalyst-coated fabric.  Downstream N20  values  ranged  from 4  to 6 ppm.
This is consistent with results presented by other researchers (6,7) for a
pulverized coal-fired boiler.  Additional measurements will be made when pilot-
scale bag tests begin.
                                        6A-49

-------
CONCLUSIONS

Based on the results of Task B testing,  several  conclusions can be made.

   1.    There was a substantial  decrease in NOX  removal  efficiency  with increased
         air-to-cloth ratio for all  the  catalyst-coated fabric samples tested.  It
         appears that for the 14 ounce per square yard fabric samples tested, in
         the bench-scale system,  the maximum air-to-cloth ratio at which 85%-90%
         NOX removal  can be achieved  is  3  ft/min.

   2.    Although there was some variability in  the data, the NO,  removal
         efficiency appeared to be constant with time over the short (eight hours)
         duration of these tests.

   3.    Of the fabric samples tested, Fabrics #2 and #17 appear to provide the
         best performance with respect to NOX  removal  efficiency and ammonia slip.

   4.    Although three of the coals, the two bituminous coals and the
         subbituminous coal, resulted in similar catalyst-coated fabric
         performance, there appeared to  be a reduction in NOX  removal  efficiency
         for the South Hallsville, Texas,  lignite.   This may have  been a result of
         pinhole formation.

   5.    When the catalyst-coated fabric is exposed to flue gas,  there is a
         decrease in the total surface area.  A  minimum BET surface area after
         exposure to flue gas of 4 to 5  m2/g is  necessary to  provide good  catalyst-
         coated fabric performance.   Therefore,  in  order to improve performance,
         it would be beneficial to increase the  surface area of the catalyst or
         the catalyst-coated fabric.

   6.    There does not seem to be any decrease  in  catalyst-coated fabric
         performance using the new vanadium source.  Although the  NOX removal
         efficiency using the iron catalyst is relatively low, it  does show
         promise, as the coating process for the iron catalyst has not been
         optimized.

   7.    For these initial tests, there  is no apparent conversion  of NOX to N20
         across the catalyst-coated  fabric.


REFERENCES

   1.    Hjalmarsson, A.K.; Vernon,  J.  "Policies for NO, Control  in  Europe,"
         Presented at: 1989 EPRI/EPA Joint Symposium on Stationary Combustion NOX
         Control, San Francisco,  CA,  March 1989.

   2.    Bruck, R.I.  "Boreal Montane Ecosystem  Decline in Central Europe and the
         Eastern United States:  Potential Role  of Anthropogenic Pollution with
         Emphasis on Nitrogen Compounds,"   Presented at 1985 EPRI/EPA Joint
         Symposium on Stationary Combustion NOX  Control,  Boston, MA,  May 1985.
                                       6A-50

-------
3.    Weber, G.F.; Laudal,  D.L.  "Final Technical Project Report for April 1988
      through June 1989 for Flue Gas Cleanup,"  Work performed under DOE
      Contract No. DE-FC21-86MC10637, Grand Forks, ND, November 1989.

4.    Kato, A.; Matsuda, S.; Nakajima, M.I.; Watanabe, Y.  "Reduction of Nitric
      Oxide on Iron Oxide-Titanium Oxide Catalyst,"  Journal of Physical
      Chemistry 1981, 85, (12), 1710-1713.

5.    Miller, S.J.; Laudal, D.L.  "Flue Gas Conditioning for Improved Fine
      Particle Capture in Fabric Filters:  Comparative Technical and Economic
      Assessment,"  Vol II. Advanced Research and Technology Development, Low-
      Rank Coal Research Final Report, Work performed under DOE Contract No.
      DE-FC21-86MC10637, Grand Forks, ND, 1987, Vol. III.

6.    Aho, M.J.; Rantanen,  J.T.; Linna, V.L.  "Formation and Destruction of
      Nitrous Oxide in Pulverized Fuel Combustion Environments between 750 and
      970C,"  Fuel 1990,  29,  957-1005.

7.    Kokkinos, A.  "Measurement of Nitrous Oxide Emissions," EPRI Journal
      1990, April/May, 36-39.
                                     6A-51

-------
                   Thermocouples
                               To Baghouse
                                           To Gas Pump and
                                           Dry Gas Meter
                      To Sample Conditioner
                      for Flue Gas Analysis
      Figure 1.  Slipstream Sample System
   100
^ 90-


 >^80-

 c
 CD  70-
]
-------
                                       Fabric *2 Fabric #17
              A/C = 2 ft/min A/C = 3 tt/min  A/C = 4 ft/min  A/C = 6 ft/min
                         NH3/NOx Molar Ratio = 0.9
    Figure 3.   Comparison  of the NOX  Removal  Efficiency as
    a  Function of Air-to-Cloth Ratio  for  Fabrics #2  and #17
CO
DC
_00
O
o
o
                                             .A/C..= .3.ft/m.in	A/C.=_2.,2.ft/m|n
                                                    700
                                                           800
                                                                  900
                                Time (min)
        Figure 4.   Ammonia/N0x Molar Ratio as a  Function
        of Time  for Fabric #17
                               6A-53

-------
                 Fabric #2
  Air-to-Cloth Ratio (ft/min)
                Illinois #6    Jacobs Ranch  Pyro Kentucky South Hallsville
                           NH3/NOx Molar  Ratio = 0.9

       Figure 5.   Comparison  of the Catalytic  Performance
       Using Four Different Coals  for  Fabric  #2
              Fabric #13
Air-to-Cloth Ratio (ft/min)  2 ^ 3 |gg 4
 o
 c
 CD
'o
it=
LJJ
"ro
 O
 E
 
-------
  10
   2 -
    50
             60
                      Air-to-Cloth Ratio = 3 ift/min

                      NH3/NOx Molar Ratio = 0.9
                      70
                               80
                                         90
                                                 100
                NO  Removal  Efficiency (%)
Figure  7.   NOX  Removal Efficiency as a  Function
of Catalyst-Coated  Fabric Surface Area  after Exposure
to Flue  Gas
                       6A-55

-------
                                             Table  1

             RESULTS FROM  TASK  B   BENCH-SCALE  FABRIC-SCREENING  TESTS "b


Fabric
No.
2
2
2
2
2
2
2
2
3
3
3
4
4
4
4
5
5
5
5
7
7
7
7
13
13
13
13
14
14
14
14
15
15
15
15
17
17
17
17
18
18
18
18

A/C
Ratio
fft/min)
2
3
4
4.5
2
3
4
6
2
4
6
2
3
4
6
2
3
4
6
2
3
4
6
2
3
4
6
2
3
4
6
2
3
4
6
2.2
3
4
5.5
2
3
4
6

NH3/NO,
Molar
Ratio
1.1
1.1
1.1
1.1
0.9
0.9
0.9
0.9
0.9
0.9
0.9
0.9
0.9
0.9
0.9
0.9
0.9
0.9
0.9
0.9
0.9
0.9
0.9
1.1
1.1
1.1
1.1
1.1
1.1
1.1
1.1
1.1
1.1
1.1
1.1
0.9
0.9
0.9
0.9
0.9
0.9
0.9
0.9

NO,
Inlet
(pom)
765
716
740
735
540
550
590
630
760
710
720
715
695
675
645
730
700
760
730
700
675
650
660
673
686
688
671
703
729
772
838
847
789
761
656
306
292
268
287
372
401
413
381

NO,
Outlet
(ppm)
20
38
83
64
58
83
112
175
226
390
490
171
235
310
436
90
125
190
305
75
95
175
200
34
64
126
209
89
151
228
433
40
68
98
193
27
31
66
101
134
166
212
224
NO,
Removal
Efficiency
(%)
97.4
94.7
88.8
91.3
89.3
84.9
81.0
72.2
70.3
45.1
31.9
76.1
66.2
54.1
32.4
87.7
82.1
75.0
58.2
89.3
85.9
73.1
69.7
94.9
90.7
81.7
68.9
87.3
79.3
70.5
48.3
95.3
91.4
87.1
70.6
91.2
89.4
75.4
64.8
64.0
58.6
48.7
41.2

Ammonia
Slip
(ppm)
187
63
129
121
5
7
22
76
ND
NO
357
87
127
179
288
28
54
76
163
4
13
33
50
64
58
88
108
107
153
256
179
57
58
104
122
45
17
28
73
102
88
122
172
Particulate
Removal
Efficiency
(%)


99.8



99.8


90.4



99.5



99.9



99.8



99.4



99.8



99.9

99.8



99.9



Each catalyst-coated fabric sample was evaluated using a slipstream of  flue gas  from a pc-fired pilot-scale
combustor firing a washed Illinois #6 bituminous coal.
"NO" denotes  data that  are not available due to problems encountered with the sampling system.
                                             6A-56

-------
                                  Table 2
                 RESULTS  FROM TASK B  EFFECTS OF COAL TYPE


Fabric
No.

2
2
2
13
13
13

A/C
Ratio
(ft/min)

2
3
4
2
3
4

NH3/NOX
Molar
ratio

0.9
0.9
0.9
1.1
1.1
1.1

NO,
Inlet
(ppm)
Washed 11
540
535
590
673
686
688
Jacobs Ranch
2
2
2
13
13
13
2
3
4
2
3
4
0.9
0.9
0.9
0.9
0.9
0.9
785
760
800
645
680
675
South Hall
2
13
13
13
3
2
3
4
0.9
0.9
0.9
0.9
900
820
810
825

NOX
Outlet
(ppm)
1 inois #6
58
81
112
34
64
126
, Wyoming,
59
75
90
80
105
195
NOX
Removal
Efficiency
m
Bituminous
89.3
84.9
81.0
94.9
90.7
81.7
Subbituminous
92.5
90.1
88.8
87.6
84.6
71.1

Ammonia
Slip
(ppm)


7


58



86


99

Particulate
Removal
Efficiency
(%)


99.8


99.4



99.9


99.9

sville, Texas, Lignite
175
110
140
195
80.6
86.6
82.7
76.4
121

75

	

99.8

Pyro Kentucky Bituminous
2
2
2
2
3
4
0.9
0.9
0.9
970
930
925
93
130
178
90.4
86.0
80.8

10


99.7

13
0.9
810
170
79.0
30
99.6
                                  6A-57

-------
                                     Table 3

                   CATALYST  CONCENTRATION  AND BET SURFACE AREA
                 FOR EACH OF THE CATALYST-COATED FABRICS TESTED"
                  Catalyst Concentration"
                              Surface Areac
Unexposed
fmq/q)
0.03
9.1
8.4
4.7
4.7
5.5
7.6
6.8
8.4
3.4
7.7
13.2
7.1
exposed
(mq/q)

9.0
8.3
3.7
4.2
5.4
6.3
6.1
8.0
3.6
5.7
13.4
7.4
Change
(%)

1.1
1.2
21.3
10.6
1.8
17.1
10.3
4.8
(5.9)d
26.0
(1.5)'
(4.2)d
Fabric No.


   Blank
     2
     2
     3
     4
     5
     7
    13
    13
    14
    15
    17
    18

Unexposed and exposed refer to exposure to flue gas.
Catalyst concentration is mg catalyst per g of coated fabric.
Fabric surface area is m2 per  g  of  coated  fabric  (BET surface  area).
(  ) Indicates there was a measured  increase in catalyst concentration.
Unexposed
(m2/q)
0.56
9.50
10.68
3.31
4.28
5.79
6.62
5.76
6.52
3.09
6.24
14.61
16.60
Exposed
(mz/q)

6.19
5.11
1.54
2.02
3.74
2.74
4.04
4.00
1.90
3.79
5.05
2.19
Change

34.8
52.2
53.5
52.8
35.4
58.6
29.9
38.7
38.5
39.3
65.4
86.8
                                     Table  4
                       N,0 CONCENTRATION IN THE FLUE GAS
     Air-to-cloth
         Ratio
       (ft/min)
          2.2
           3
           4
          5.5
           2
           3
           4
           6
  Inlet N20
Concentration
    (ppm)

 Fabric #17

     4.0
     3.5
     4.0
     4.0

 Fabric #18

     5.5
     4.5
     4.0
     3.5
  Outlet N20
Concentration
	ppm)
     5.0
     4.5
     4.5
     4.5
     6.0
     5.0
     4.5
     4.0
                                    6A-58

-------
           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.Y. Lee
                             L. Young
                      Ahlstrom Pyropower Inc.
                        8970 Crestmar Point
                     San Diego, California 92121

-------
         HETEROGENEOUS DECOMPOSITION OF NITROUS OXIDE IN THE OPERATING
           TEMPERATURE RANGE OF CIRCULATING FLUIDIZED BED COMBUSTORS
                                          T. Khan
                                          Y.Y. Lee
                                         L. Young
                                  Ahlstrom Pyropower Inc.
                                    8970 Crestmar Point
                                San Diego, California 92121
ABSTRACT
There is growing concern over the increasing atmospheric nitrous oxide concentration. This concern
stems from the realization that nitrous oxide contributes to the depletion of the ozone layer and to
the greenhouse effect. A research program has been developed at Ahlstrom Pyropower Inc. to study
the emission of nitrous oxide  from circulating fluidized  bed combustors (CFBCs).   The program
involves, in part, an investigation into the mechanism of nitrous oxide formation and destruction in
the operating temperature range of CFBCs. This paper describes a study directed at understanding
the decomposition of nitrous oxide on solid materials known to be present in the combustor.

An electrically heated tubular quartz reactor (2.3  cm I.D.) was used to study the decomposition of
nitrous oxide on  six different  solid materials;  alumina, silica, ceramic beads,  sulfated limestone,
calcined amorphous limestone  and calcined crystalline limestone.  Approximately 10 cm3 of each
solid material was placed in turn in the reactor and a mixture of nitrous oxide (200 ppm) in helium
was passed  through the reactor.   The concentration of nitrous oxide at the  reactor outlet was
measured  to determine  the  extent  of  N2O  decomposition.   As a basis  for comparison,  the
homogeneous phase decomposition of nitrous oxide in the reactor was also  studied.

Results showed that a  significant amount of N2O decomposed even in  the  absence  of any solid
material in the reactor.  It was observed that the presence  of solid materials in the reactor enhanced
the decomposition of nitrous oxide and that the degree of enhancement was dependent on the solid
material  being tested;  calcined limestone, for  example,  was  seen to be highly effective  in
decomposing nitrous oxide while ceramic beads showed little or no effect.
                                           6B-1

-------
INTRODUCTION
There is growing concern over the increasing concentration of atmospheric nitrous oxide.  This
concern stems from the realization that nitrous oxide contributes to the greenhouse effect and to
the depletion of the ozone layer.  The mean concentration of N2O in the atmosphere is 330 ppbv
and it is estimated that it is  increasing at a rate of 0.2% per year JJJ.

It has been suggested that fossil fuel combustion is a major contributor  to the atmospheric nitrous
oxide inventory.  Measurements [21  show that nitrous oxide emissions from  circulating fluidized
bed combustors (CFBCs)  range from 20 to 120 ppm. Based on these emission values, it is doubtful
that nitrous oxide emissions from fluidized beds contribute more than a minor fraction to the global
inventory.  Nonetheless,  in accordance  with its dedication to developing an environmentally safe
product, Ahlstrom Pyropower Inc. has instituted a project directed at the reduction of nitrous oxide
emissions from AHLSTROM PYROFLOW* boilers.  The project involves, in part, an investigation into
the formation and destruction of nitrous oxide in circulating fluidized bed combustors.

Knowledge of the principal reactions involved in the formation and destruction of nitrous oxide in
fluidized bed environments is limited at best.  In order to minimize  nitrous oxide emissions it is
necessary that:

    1.     reactions that play a dominant role in the formation and destruction of nitrous oxide be
          identified and  that
    2.     the effect of process parameters on the kinetics of these reactions be studied in detail.

Studies [3.41 indicate that hydrogen cyanide (HCN), released during the devolatilization of coal, is
a major precursor of nitrous oxide.  It is believed that HCN undergoes oxidation to NCO which in
turn reacts with nitric oxide  (NO) to form nitrous oxide (N2O).  There is relatively little debate
about the importance  of this reaction path as a means of formation of nitrous oxide.  Doubts about
it being the only major nitrous oxide formation path have however been expressed.   De Soete [51
and Arnand and Andersen [61 have reported the formation of nitrous oxide by the reduction of NO
on char surfaces. De Soete [51 has also reported that nitrous oxide may be formed by the oxidation
of char nitrogen (1-5%) during  combustion.

Nitrous oxide destruction in the fluidized bed environment may  occur through both homogeneous
and heterogeneous phase reactions.   Kramlich et al. 4J and Emola et al. [31 have suggested that
the principal nitrous oxide destruction reaction is its homogeneous phase reduction to nitrogen by
hydrogen radicals.  Relatively very little is  known  about  the heterogeneous phase destruction of

                                           6B-2

-------
nitrous  oxide.   It  is believed [51 that nitrous oxide reduction on  char is  one of  the  major
heterogeneous N2O destruction pathways.  Little or no information currently exists on the interaction
of nitrous oxide with solids, other than char, present in a fluidized bed environment.

This paper describes a study directed at investigating the heterogeneous decomposition of nitrous
oxide in the operating temperature range of a CFBC.  An electrically heated tubular quartz reactor
(2.3 cm I.D.) was used to study the decomposition of nitrous oxide on six different solid materials;
alumina, silica,  ceramic beads,  sulfated limestone,  calcined  amorphous limestone and  calcined
crystalline limestone.  Approximately 10 cc of each solid material was placed in rum in the reactor
and a mixture  of  nitrous oxide  (200 ppm) in helium  was passed through the reactor.   The
concentration of nitrous oxide at the reactor outlet was measured to determine the extent of N2O
decomposition.   As  a basis for comparison, the homogeneous phase decomposition of nitrous oxide
in the reactor was also  studied.

EXPERIMENTAL SET-UP
The experimental set-up used in the course of this study (Fig. 1) consists of three major components:

   1.     An electrically heated quartz tube that serves as a reactor.
   2.     Mass  flow  controllers  used to  deliver a  measured amount of a nitrous-oxide/helium
          mixture to the reactor.
   3.     A HORIBA Non-Dispersive Infrared nitrous oxide analyzer.

Reactor
The reactor, for the major part, is a  91.5 cm long, 2.3 cm I.D. quartz glass tube.  Caps at the end
of the tube house ports  for the inlet and the outlet of reactant and product gas mixtures. The end
caps  also house inlet ports for  thermocouples used  in measuring and controlling the reactor
temperature. A sintered quartz glass filter is provided 50.8 cm from one end of the tube and  serves
to support a bed of the solid material being tested.  The reactor is heated by a three zone, 61 cm
long electric furnace.  The two outermost zones of the furnace are 15.25 cm long and the central
zone is 30.5 cm in length. Each furnace zone is independent of the others in its temperature control.
Thermocouples inside the reactor serve as sensors for controllers that control the temperature of each
furnace zone.

Mass Flow Controllers
Two mass  flow controller modules, one for  a nitrous-oxide/helium  mixture (0.4% N2O)  and the
                                            6B-3

-------
other for pure helium were used in the course of this study.  Using these controllers it was possible
to feed mixtures of nitrous oxide in helium at predetermined concentrations and flow rates to the
reactor.  It may be mentioned here that helium was chosen as a 'balance gas' due to its chemical
inertness and its high thermal conductivity.  The high thermal conductivity was necessary to minimize
radial temperature gradients and the heat up  zone within the reactor.

Nitrous Oxide Analyzer
A HORIBA Non Dispersive Infrared N20 analyzer was used to measure the concentration of nitrous
oxide in the inlet and outlet gas streams of the reactor. The analyzer was equipped with a 7.8 ^m
wavelength filter.

EXPERIMENTAL PROCEDURE
Homogeneous Phase Decomposition Study
The reactor was heated to the desired temperature and a 200 ppm mixture of N2O in helium was
fed to the reactor at three different flow rates (500, 1000 and 1500 cmVmin).  At each condition,
the concentration of nitrous oxide at the outlet of the reactor was measured to determine the extent
of nitrous oxide decomposition. This procedure was repeated for six reactor temperatures; 650, 700,
750, 800, 850 and 900C. The results obtained are presented in the following section.

Heterogeneous  Phase Decomposition
Approximately  10 cm3 of the material being tested (250jim>mean particle diameter>125/tm) was
placed in the reactor and the reactor was heated to 850C.  A 200 ppm mixture of N2O in helium
was fed  to the reactor  at a flow rate of 500, 1000 and 1500  cmVrnin.  At each condition, the
concentration of nitrous oxide at  the outlet of the reactor was  measured to  determine  the extent
of nitrous oxide decomposition. A comparison between the results obtained for each  solid material
tested is presented in the following section.

RESULTS
Results of the homogeneous phase nitrous  oxide decomposition study are shown  in Table 1. As
may be seen from the data, nitrous oxide does not decompose to any significant extent below 700C.
It is also evident that the rate of nitrous oxide decomposition increases with reactor temperature and
residence time.  It is most likely that the products of the nitrous oxide decomposition were nitrogen
and oxygen; no nitric oxide  (NO)  was detected in the  outlet stream from the reactor.

Reaction rate constants for the homogeneous phase decomposition of nitrous oxide  were calculated

                                           6B-4

-------
from the obtained data.  It was assumed, in the calculation, that the decomposition of N2O occurs
via a first order reaction.  The reaction rate constant, k, is presented as a function of temperature
in Table 2.  Fig. 2 is a plot of -ln(k) versus 1/T.  As may be seen, the plot is a
straight line.   This  indicates that the assumption  that nitrous oxide undergoes  a first order
decomposition reaction was correct.  The rate of homogeneous phase nitrous oxide decomposition
may thus be written  as:

                       d[N20]/dt = -k  [N20]

where, [N2O] is the nitrous oxide concentration at time  t. The reaction rate constant, k, a function
of temperature, may  be expressed as:

                       k  =  koexp[-E/RT]

The value of the activation energy, E, derived from the slope of the plot  (E/R)  is 246.6 kJ/mol.
The frequency factor, ko, may be derived from the y-intercept of the plot,  -ln(ko), and is equal to
2.813 x 1010 sec".

The results of the heterogeneous phase N2O decomposition studies are shown in Table 3.  Also
included in the table, for the  purpose of comparison, are the results from the corresponding empty
tube (homogeneous  phase)  experiments.   The results  show the fraction of  nitrous oxide that
decomposes on passage through the reactor. The  residence times entered at the top of the table
are the residence times of the gas mixture in the entire  heated length of the reactor.  The numbers
(1) and (2) are used to distinguish between the two  types of limestones used;  respectively, the
calcined crystalline limestone  and the calcined amorphous limestone. The variation of nitrous oxide
decomposition with total reactor residence time, is shown, for each solid material and the empty tube
experiment, in Fig. 3.

As may be seen from the results, the presence of ceramic beads or sulfated limestone in the reactor
does not significantly affect the decomposition of nitrous oxide.  The  presence  of silica sand or
alumina enhances the decomposition of nitrous oxide to a small extent.  The most dramatic results
are those obtained in the presence of calcined limestone.   It  may be seen  that nitrous oxide
decomposes completely in the presence of the calcined  crystalline limestone at  850C.  As may be
seen from the graphical results, the conversion in the presence of calcined limestone is dependent
on the kind of limestone used.  There is an almost 50% difference in the conversions  for the two
types of limestones used at  a reactor residence times of 3.2 sec.  As in the case of the homogeneous
                                           6B-5

-------
phase decomposition studies, no NO was detected in the outlet stream from the reactor.

DISCUSSION AND CONCLUSIONS
Based on  the observations and results described in this paper, the following conclusions may be
drawn.

1. The homogeneous phase thermal decomposition of nitrous oxide is a very important pathway for
   nitrous oxide destruction in a fluidized bed combustor.  A simple calculation shows  that under
   typical operating conditions in a circulating  fluidized bed, that is, a gas residence time of 6
   seconds at an average furnace temperature of 870C, over 60 percent of the nitrous oxide present
   at the  bottom  of the combustor would be  destroyed  before it reached the  combustor exit.
   Furthermore, if the average  operating temperature of the unit were to be increased by 10C, the
   extent  of N2O decomposition would be increased to  70%. It has been seen in measurements on
   commercial  scale CFBCs  [21 that the  nitrous oxide  emission  level does in  fact  decrease
   significantly with increasing bed temperature.  It must be realized, of course,  that the rate of
   nitrous oxide formation is also temperature dependent.

2. Of the solid materials tested, calcined limestone was seen to  decompose nitrous oxide most
   efficiently.  Alumina and silica sand were seen to slightly enhance the  decomposition of nitrous
   oxide and ceramic beads  and sulfated Limestone were seen to have virtually no effect on the
   extent  of nitrous oxide decomposition.  One would expect, in the light of these observations, to
   see a dramatic  decrease in nitrous oxide emissions with increasing feed Ca/S ratio in a CFBC.
   This, however, has not been the case.  Studies on a 0.8 MW^, Ahlstrom Pyroflow pilot plant [21
   show only a slight reduction in nitrous oxide emissions with increasing feed Ca/S ratio;  no
   definite relationship between nitrous oxide emissions and feed Ca/S ratio could be detected for
   a similar  study  [21 carried out on commercial scale  units.

3. The efficacy of  calcined Limestone in decomposing nitrous oxide was  dependent on the type of
   Limestone  used.   Calcined  crystalline Limestone was seen to  decompose nitrous oxide  more
   effectively than  was calcined amorphous Limestone.  At a reactor residence time of 3.2 seconds,
   the calcined crystalline Limestone was seen to completely decompose the nitrous oxide, where, the
   calcined amorphous Limestone was seen to decompose only 50% of the inlet nitrous oxide.

ACKNOWLEDGEMENT
The authors  gratefully acknowledge partial funding of  the described study by the Finnish Ministry
of Trade and Industry through  the LIEKKI program.

                                          6B-6

-------
REFERENCES

1.  R.F. Weiss, Journal of Geophysical Research, vol. 86, 1981, p. 7185.

2.  M. Hiltunen, P. Kilpinen, M. Hupa and Y.Y. Lee, "N2O Emissions from CFB Boilers: Experimental
   Results and Chemical Interpretation."  To be presented at the 11th Int. Conf. on Fluidized Bed
   Combustion. Montreal, 21-24 April, 1991.

3.  P. Ernola &  M.  Hupa,  "Kinetic Modelling of Homogeneous  N2O Formation and Destruction in
   Fluidized Bed Conditions."  Proceedings of the Joint Meeting of the British and French Sections
   of the Combustion Institute. Rouen,  1989, p. 21.

4.  J.C. Kramlich, J.A. Cole, J.M. McCarthy, W.S.  Larder & J.A. McSorley, "Mechanisms of Nitrous
   Oxide Formation in Coal Flames." Combustion and Flame. 1989, vol. 77, p. 375.

5.  G.G. De  Soete,  "Heterogeneous  NO  and N2O Formation from Bound Nitrogen  during Char
   Combustion." Proceedings  of the Joint Meeting of the British and  French  Sections of  the
   Combustion Institute. The  Combustion Institute, 18-21 April, Rouen,  1989, p. 9.

6.  L.E. Amand  & S.  Andersen, "Emissions  of Nitrous Oxide (N2O) from Fluidized Bed Boilers."
   Proceedings of the 1989 International Conference on Fluidized Bed Combustion, vol. 1, pp. 49-
   56.
                                          6B-7

-------
MASS FLOW CONTROLLERS
              GAS SUPPLY CYLINDERS
REACTANT GAS INLET _
^n-
r


A

N20/He

eJ&i

i
61
I
i
cm



s


QUA

I

[*
                                                                 QUARTZ GLASS TUBULAR
                                                                      REACTOR
                                                                        THREE ZONE
                                                                     ELECTRIC FURNACE
                                                                      QUARTZ GLASS FRIT
                                                                        PRODUCT GAS TO
                                                                         ANALYZERS
                                                       THERMOCOUPLES
                                                                  -oki-
                                            REACTOR BYPASS LINE
        Figure  1.    Experimental Setup  for Quartz  Tube  Reactor  Studies
                                                                         1.0E-3
                                  Figure 2. -in(k) Vs. 1/T
                                        6B-8

-------
o
 eg
'in
 o
 CL

 E
 o
 o
 v
Q

o
  CM
       0.9  -
       0.8  
       0.7  -
  c

o'
 CM
  o    0.6  
O
 CM
       0.5  -
0.4  -
       0.3  -
       0.2  -
       0.1  -
       0.0
                                                    v	v Empty Tube



                                                    O	o Alumina

                                                    o  o Ceramic Beads

                                                    	 Silica Sand

                                                    A	A Sulfated Limestone

                                                    	 Calcined Limestone (1)

                                                    	 Calcined Limestone (2)
                                                                                 T
                                                                                            10
                                       Reactor Residence Time (sec)
                      Figure 3.  Fractional N 0 Decomposition vs. Reactor Residence Time
                                            6B-9

-------
                              Table 1

         HOMOGENEOUS PHASE DECOMPOSITION OF NITROUS  OXIDE

Reactor Pressure  :  3 psig        Inlet N2O Concentration :  200 ppm
Temperature
<*C)
650
700
750
800
850
900
Residence Time
(sec)
11.7
5.8
3.9
11.1
5.5
3.7
10.5
5.3
3.5
10.0
5.0
3.4
9.6
4.8
3.2
9.2
4.6
3.1
NHO Outlet Concentration
(ppm)
200
200
200
197
200
200
185
200
200
150
174
182
78
125
148
12
52
81
                              Table 2

        HOMOGENEOUS PHASE N2O DECOMPOSITION REACTION RATE
                    CONSTANT  VS.  TEMPERATURE
Temperature
(*C)
700
750
800
850
900
k
(sec'1)
0.001350
0.007399
0.028640
0.097444
0.301750
                              6B-10

-------
                           Table 3




FRACTIONAL N2O DECOMPOSITION VS. TOTAL REACTOR RESIDENCE TIME




  Reactor Temperature :  850 C   Reactor Pressure  :  3  psig
Material
Alumina
Ceramic Beads
Silica Sand
Sulfated Limestone
Calcined Limestone (1)
Calcined Limestone (2)
t=9.6s
0.65
0.61
0.63
0.61
1.00
0.92
t=4.8s
0.41
0.38
0.40
0.38
1.00
0.65
t=3.2s
0.29
0.26
0.28
0.26
0.98
0.50
Empty Tube
0.61
0.38
0.26
                           6B-11

-------
   NOx  CONTROL IN A SLAGGING COMBUSTOR FOR A
      DIRECT  COAL-FIRED UTILITY GAS  TURBINE

          P.  J. Loftus and R. C. Diehl

Energy Technology Office/Textron Defense Systems
       (Formerly AVCO Research Laboratory)
           2385 Revere Beach Parkway
                Everett,  MA 02149

                       and

      R.  L.  Bannister and P.  W. Pillsbury

          Westinghouse Electric Corp.
       The Quadrangle, 4400 Alafaya Trail
             Orlando, FL 32826-2399

-------
              NOX CONTROL IN A SLAGGING COMBUSTOR FOR A
                 DIRECT  COAL-FIRED  UTILITY  GAS  TURBINE

                     P.  J. Loftus and R. C. Diehl

           Energy Technology Office/Textron Defense Systems
                  (Formerly  AVCO  Research Laboratory)
                      2385  Revere  Beach Parkway
                          Everett,  MA 02149

                                 and

                  R.  L.  Bannister and  P. W. Pillsbury

                     Westinghouse Electric  Corp.,
                  The Quadrangle, 4400 Alafaya Trail
                        Orlando,  FL 32826-2399
    Joint EPA/EPRI Symposium on Stationary Combustion NOX Control
                  Washington, D.C.,  March  25-28,  1991
ABSTRACT

     A modular combustion concept, which emphasizes controlled coal
thermochemistry, has been developed for application in direct coal
firing of utility gas turbines.  The approach under investigation is
based on a multi-stage, slagging combustor, which incorporates NOX,
SOX and particulate emissions control.   This  approach  allows raw
utility grade coal to be burned, thereby maintaining a low fuel cost.
The cost of electricity from combined cycle plants incorporating a
direct coal-fired gas turbine is expected to be significantly lower
than that from conventional pulverized coal steam plants.

     The first stage, the primary combustion zone, is operated fuel-
rich to inhibit NOX formation from fuel-bound nitrogen and has a jet-
driven, toroidal vortex flow field, which provides for efficient,
stable and rapid combustion at high heat release rates.  Impact
separation of molten mineral matter is accomplished in the second
stage, which is closely integrated with the primary zone.  The second
stage may also include a slagging cyclone separator for additional
slag removal.  This is a novel application for a cyclone separator.
Final oxidation of the fuel-rich gases and dilution to achieve the
desired turbine inlet conditions are accomplished in the third stage.


                                6B-15

-------
Rapid quenching and good mixing with the secondary air are employed to
avoid thermal NOX  formation  by  minimizing  peak  flame  temperatures and
residence times in the third stage.

     The combustor concept has been extensively tested at a thermal
input of 3.5 MWt  (12  MM Btu/hr)  and a  pressure  of  6 atmospheres.   Both
pulverized coal and coal-water mixtures have been successfully fired.
The combustor has demonstrated stable and intense combustion, with
excellent carbon conversion, efficient slag capture,  retention of most
of the coal alkali in the slag and low pressure and heat losses.  The
staged combustion NOX control  strategy has  proved  very  effective: NOX
emissions are approximately one fifth of the New Source Performance
Standards requirements.

BACKGROUND

     The authors'  companies are working under Department of Energy
sponsorship to develop the technology base for direct coal-firing of
utility gas turbines.  The approach under investigation is based on a
multi-stage, slagging combustor, which incorporates NOX,  SOX and
particulate emissions control.  This approach allows  raw utility grade
coal to be burned, thereby maintaining a low fuel cost.  The cost of
electricity from combined cycle plants incorporating a direct coal-
fired gas turbine is expected to be significantly lower than that from
conventional pulverized coal steam plants with flue gas
desulfurization (Pillsbury et al., 1989).

     The program objective is to develop an efficient combustor
capable of meeting the New Source Performance Standards (NSPS) for
NOX,  S02 and particulates upstream  of the turbine.  The program is
divided into three key tasks.  The first of these is  the design,
fabrication and testing of a subscale slagging combustor (6 atm, 3.5
MWC).   This task  is  in progress:  combustor  testing commenced in  late
1988 at the Textron Defense Systems/Energy Technology Office  (ETO)
Haverhill test facility.  The second task involves testing the final
subscale combustor configuration with a stationary cascade to study
the effect of deposition, erosion and corrosion on air-cooled turbine
vanes.  Based upon the data generated, the final task is to design,
fabricate and test a full size combustor (14 atm,  20  MWt) .   This  paper
discusses the design and performance of the subscale slagging
combustor from the point of view of NOX emissions  control.

COMBUSTOR CONCEPT

     The three stage combustor is illustrated schematically in Figure
1.  The design of the first stage, the primary combustion zone,   is
based on Avco Research Laboratory's toroidal vortex combustor concept,
and provides for efficient, stable and rapid combustion at high  heat
release rates  (Mattsson and Stankevics, 1985, Stankevics et al.,
1983).  Coal and preheated air are fed coaxially into the primary zone
through four separate injectors which are inclined upward at
approximately 60  to  the horizontal.   The  coaxial  injection promotes
intense coal/air mixing, leading to rapid coal particle heating  and
devolatilization,  which minimizes carbon burn-out time.  The four
inlet coal/air jets converge at the combustor centerline and form a
single vertically directed jet.  This jet impacts the center uf  the
primary zone dome, where it is turned and a toroidal vortex is formed.
This arrangement forces a high degree of controlled combustion product


                                 6B-16

-------
re-circulation which leads to extremely intense and very stable
combustion of a wide variety of fuels.  The toroidal vortex design
gives very high volumetric heat release rates for coal combustion  (up
to 40 MWt/m3) .  These heat release rates are some three to four times
that for cyclone-type combustors, leading to smaller combustor sizes
and lower wall heat losses.   Fuel-rich conditions in the primary zone
inhibit NOX formation from fuel-bound nitrogen.   Extensive  use was
made of 3-D combustion modelling techniques in the preliminary design
of the combustor  (Chatwani and Turan, 1988, Loftus et al.,  1988).

     The toroidal vortex provides the mechanism for flame
stabilization and also for inertial separation of larger ash/slag
particles.  Partial separation of mineral matter and char at the top
of the toroidal vortex zone results in initiation of wall slagging
there, with continued deposition and flow over all exposed wall
surfaces.  In order for successful slag deposition in the dome region,
enough coal particle residence time and combustion product re-
entrainment must be provided to ensure rapid coal particle burnout,
resulting  in molten, free mineral matter.  Larger, partially
devolatilized coal particles will continue to burn, either in
suspension or in the wall slag layer.  The primary zone was designed
for approximately 90 percent suspension burning and 10 percent wall
burning.   The primary zone particle residence time is of order 100
msec for a 75 micron diameter particle.  The primary zone slag layer
provides thermal and erosion protection for the combustor walls,  in
addition to a mechanism for oxidation of deposited char.  The slag
layer formed from this portion of the mineral matter eventually
reaches the impact separator, where it is collected in the slag
bucket.

     The major fraction of mineral removal from the gas is obtained in
the second stage impact separator which is at the exit from the
primary zone.  The separation of combustion and slag removal duties
between the two stages has two substantial benefits.  First, it
results in maximum removal of carbon free slag: at the primary zone
exit there is a very high carbon conversion fractionessentially all
the coal char has been oxidized, leaving free mineral matter behind.
Second, due to the low density of the combustion products,  a simple
impactor allows high efficiency separation of fine mineral matter
particles  at low cost in pressure drop.  Overall, the air pressure
drop is optimally distributed, first for combustion stabilization and
second for mineral matter removal.

     Pulverized limestone sorbent is used for control of sulfur
emissions.  The sulfur control technique used is based on related ETO
work on super-equilibrium sulfur capture (Abichandani et al., 1989).
The sorbent is injected into the primary zone combustion products,
which generally contain a mixture of S02/  H2S and COS, at the  exit  of
the primary zone, just upstream of the exit nozzle.  Reacted sorbent
is removed along with the coal ash in the second stage impact
separator.

     Final oxidation of the fuel-rich gases and dilution to achieve
the desired turbine inlet conditions are accomplished in the third
stage.  Rapid quenching and good mixing with the secondary air are
employed to avoid thermal NOX formation by  minimizing peak  flame
temperatures and residence times in the third stage.
                                 6B-17

-------
NOX  CONTROL  APPROACH

     Emissions of nitrogen oxides in the products of combustion are
controlled by adopting the following approach:

         Sub-stoichiometric (fuel-rich) combustion conditions in the
          first combustor stage.

         Effective control of the gas temperature and stoichiometry
          histories during final oxidation/dilution in the third
          combustor stage.

     The main source for formation of nitric compounds in the
combustion of coal is fuel-bound nitrogen.  Part of the fuel nitrogen
is released with the volatiles in the early stages of combustion and
the remainder is retained by the char residue and released during
subsequent char oxidation.  Nitric oxide can be produced by the
oxidation of the nitrogen in the volatiles or in the char.  NOX
formation from fuel bound nitrogen is very sensitive to the combustion
stoichiometry.  It is known that HCN and NH3  are formed in the  gas
upon evolution of coal nitrogen.  These can subsequently be oxidized
to NO or can react with NO to form harmless molecular nitrogen,
depending upon the availability of oxidants in the gas.  Fuel-rich
operation promotes the formation of molecular N2 as  the end product  of
the fuel nitrogen, whereas fuel lean operation, with the availability
of oxidants, results in NO formation.

     Volatile nitrogen is defined as that which is produced from the
volatile coal fractions and reacts in the gas to form N2,  NO,  HCN  or
NH3.   Char nitrogen is  that  which is  associated with a  solid,  either
as a pyrolysis product of tars or as the original coal char.  The
distribution of nitrogenous species between volatile nitrogen and char
nitrogen is critically dependent on the coal particle heating rate,
the peak temperature, the residence time at high temperature and the
nitrogen distribution within the coal  (Smart and Weber, 1989) .  Fuel-
bound nitrogen is the major source of NOX in  conventional  PC
combustion, typically accounting for more than 80 percent of total NOX
emissions  (Pershing and Wendt,  1979).

     For staged combustion to be effective, it is important to avoid
the carry over of either volatile or char nitrogen to the final
oxidation zone, where these can be converted to NO.   The intense and
rapid mixing produced by the toroidal vortex design leads to rapid de-
volatilization of .the coal, homogeneous combustion conditions and
efficient oxidation of the char to a fuel-rich gas in the first stage.
These conditions favor conversion of fuel bound nitrogen to molecular
nitrogen and minimize the possible carry-over of volatile or char
nitrogen to the third combustor stage.

     For the case of PC combustion, the calculated equilibrium
concentrations of nitrogen oxides in the combustion gas are shown in
Figure 2 for various primary zone fuel air equivalence ratios and
temperatures.  This plot includes NH3 and HCN,  which have  been
converted to total NOX  and included in the concentrations  shown.   (The
contributions from these species are typically small.)   NOX
concentrations at the adiabatic flame temperature and at 100 K and 200
K below the adiabatic flame temperature are shown.  The NOX
concentrations in the gas corresponding to the NSPS limitsX for
                                 6B-18

-------
bituminous coal  (0.6 Ibs per MM Btu) and sub-bituminous coal  (0.5 Ibs
per MM Btu) are also shown as a function of fuel-air equivalence
ratio.  The strong temperature dependence of NOX is clearly seen:  a
temperature drop of 200 K typically reduces the equilibrium NOX by a
factor of three or more.  The equilibrium NOX concentration in the gas
becomes less than the NSPS standard at fuel-air equivalence ratios
greater than about 1.2 and at the primary zone nominal operating point
(equivalence ratio in the range 1.3 to 1.4) the equilibrium NOX in the
primary zone is more than a factor of ten less than the NSPS
requirement.

     The control of stoichiometry and temperature  in the third
combustion stage is key to minimizing the formation of thermal NOX.
The formation of thermal NOX is  governed by the highly temperature
dependent reactions between nitrogen and oxygen, the Zeldovich chain
reactions.  The rate of formation is significant only at temperatures
above approximately 1900 K  (3000F)  and increases  with increasing
oxygen concentration.  Thus temperatures in the final oxidation zone
should be maintained below this value to avoid thermal NOX formation.
The secondary air for final combustion in the last combustor stage is
added in such a manner that the gas is rapidly quenched and maintained
at a temperature below which thermal NOX can form,  while  final
oxidation of the unburned species in the gas is completed.  As soon as
the final oxidation is complete, the dilution air  is introduced, again
with rapid and complete mixing, in order to quench all further NOX
generation.

     Kinetic calculations were performed to determine the desired
temperature and operating conditions during final  oxidation and the
appropriate split between quench/final oxidation air and dilution air.
These calculations showed that an adiabatic flame  temperature of about
1800 K is reached for a fuel air equivalence ratio of 0.6 in the
intermediate zone and that the final oxidation of  the gas is completed
within a few milliseconds, see Figure 3.  At these conditions thermal
NOX formation is insignificant  and the predicted final NOX
concentration in the gas will be only a small fraction of the NSPS
limit.  It is important to obtain effective mixing of the secondary
combustion air with the hot fuel-rich primary gases.

     Three-dimensional aero-thermal calculations and analysis of the
mixing process in both the intermediate and dilution zones of the
third combustor stage were conducted.  The number, size and
orientation of the intermediate and dilution zone  jets were varied to
arrive at the optimum mixing performance, expressed as a minimum exit
temperature pattern factor.  The final design analysis involved
extending the three-dimensional aero-thermal flow  modelling of the
third stage to the full reacting flow field.  However, no attempt was
made to optimize the lean-burn combustor from the  point of view of NOX
control.  The principal purpose of the experimental work was to tackle
the major technical issues in this development effort, which are
related to obtaining efficient primary zone and slag separator
performance.

TEST ARRANGEMENT AND COMBUSTOR OPERATION

     The combustor concept has been extensively tested at a thermal
input of 12 MM Btu/hr (3.5 MWt)  and a pressure of 6 atmospheres.
Tests have been conducted with both pulverized coal  (PC)  and coal-
                                 6B-19

-------
water mixture (CWM)  fuels.  A photograph of the subscale slagging
combustor test arrangement as currently installed at ETO's Haverhill
test site is shown in Figure 4.   The nominal test conditions for the
subscale combustor are as listed in Table 1.  An oil fired air pre-
heater is used to heat the combustion air in order to correctly
simulate the gas turbine compressor discharge conditions.  A
downstream sonic orifice is used to control the combustor chamber
pressure.  After pressure let-down, the combustor exit gases are water
quenched before being led to an exhaust stack.  The subscale combustor
is water cooled, the cooling water being re-circulated via a cooling
tower.  All fuel-rich zone components are lined with a high alumina
refractory.  This is both to reduce heat losses in this small scale
experimental combustor and to promote slagging during the relatively
short tests.

     Start-up and operation of the system proved to be simple and
reliable.  After establishing the correct air flow rates through the
system and allowing the air pre-heater to come up to design
temperature, a methane/oxygen torch in the primary zone is ignited.
The torch is used to ignite a fuel oil flame and is then extinguished.
Fuel oil is then burned for approximately 15 minutes, in order to pre-
heat the refractory liner.  The oil is injected via two spray nozzles
in the primary zone.  After the refractory liner has reached operating
temperature, the coal  (PC or CWM) flow is started, and the fuel oil
flow is  stopped.  In PC testing, a pneumatic conveying system is used
to feed  coal to the primary zone.  For CWM testing, a Moyno
progressing cavity pump is used to supply CWM to the combustor.  The
CWM atomizers are Parker-Hannifin air-assist atomizing nozzles.
Atomizing air for CWM tests is supplied from a high pressure tube
trailer  via a heat exchanger.  The heat exchanger warms the expanded
high pressure air back up to approximately ambient temperature.

     A detailed fuel specification for the proposed application was
prepared by AMAX Extractive Research and Development.  Choice of coal
 (and consequently of mineral matter composition), coal particle size
and CWM  composition affects certain primary design constraints for the
slagging combustor.  These include liquid slag formation, combustion
efficiency, downstream corrosion, erosion and deposition and pollutant
generation.  The primary  zone of the combustor was designed to operate
at highly fuel-rich  (i.e. low flame temperature) conditions.  The
flame temperature is obviously even lower for CWM fuels.  Consequently
a low ash fusion temperature coal was desirable.  The ratio of ash to
sulfur content is of interest: the higher the coal sulfur content, the
higher the ratio of limestone sorbent to ash in the slag to be
separated and the greater the effect of sorbent injection on slag
properties.  The preferred coal fuels were determined to be high
volatile eastern bituminous coals.  These coals have the advantages of
a high heating value, leading to favorable combustion characteristics
with high flame temperatures and rapid combustion.  They also tend to
have low to medium sulfur contents and soluble alkali contents below
0.05 percent.  From this  general specification, several specific seams
were identified for use in the subscale combustor testing.  These
included a low and a high sulfur eastern bituminous coal and a low
sulfur western sub-bituminous coal.  Detailed coal specifications are
given in Table 2.  The CWM fuels tested were prepared from close to
standard grind  (95 percent through 200 mesh) coals.

     A full test program  was conducted with PC feed before switching


                                 6B-20

-------
TABLE 1
SUBSCALE SLAGGING COMBUSTOR NOMINAL TEST CONDITIONS
Coal Thermal Input
Coal Feed
Atomizing Air/CWM Mass Flux Ratio
Oxidizer
Primary Zone Equivalence Ratio
Total Mass Flow Rate
Exit temperature
Pressure
Sorbent
Sorbent Molar Ratio
3.5 MWt (12 MM Btu/hr)
95% < 200 mesh PC
95% < 200 mesh, 60% solids CWM
1.0
620 K (650F) pre-heated air
1.3 to 1.4
3.2 kg/s (7 Ib/s)
1300 K (1850F)
6 atm
-325 mesh limestone
Ca/S = 2
over to CWM feed.  From the outset of combustor testing,  a stable,
flowing slag layer was formed on the primary zone dome and walls.
Some dissolution of the refractory layer was observed in the early
runs, but after a few hours of operation an equilibrium insulation
layer of slag and refractory was formed.  Equilibrium slag layer
thicknesses in the primary zone, where heat fluxes are high, are on
the order of 1 mm.  The corresponding thickness in the slag separator
is on the order of 3 mm.  No obstruction or fouling of any of the
primary zone coal/air injectors or of the relatively small diameter
primary zone exit nozzle with slag was observed.  The impact separator
worked as planned, and a flowing slag layer was observed on the top
and sides of the impactor centerbody and on the slag bucket walls.

TEST RESULTS

     A full series of tests with PC fuels demonstrated that the
combustor primary zone produces excellent carbon conversion
performance, see Figure 5.  At the nominal primary zone operating
point (fuel/air equivalence ratio of 1.3 to 1.4) the carbon conversion
for PC firing is better than 99 percent.  For PC firing,  better than
98 percent carbon burnout in the primary zone was obtained for
fuel/air equivalence ratios as high as 1.6.  In order to obtain good
carbon conversion performance on CWM fuels, it was necessary to
increase the primary zone aspect ratio  (length/diameter).   For PC
firing the aspect ratio of the primary zone was 1.25  (L/D = 1.25) .
The optimum configuration for CWM firing was a primary zone aspect
ratio of 1.50.   In this configuration, better than 98 percent carbon
conversion was obtained for equivalence ratios up to approximately
1.4.   The increase in aspect ratio increases the particle residence
time, thus allowing more time for evaporation of the water in the CWM
droplet.   The performance on CWM is slightly worse than that obtained
                                 6B-21

-------
TABLE 2
ANALYSES FOR COALS TESTED TO DATE
Coal Analysis As
Received
% Moisture
% Ash
% Volatile Matter
% Fixed Carbon
% Sulfur
% Chlorine
% Carbon
% Hydrogen
% Nitrogen
% Oxygen
MJ/kg (Btu/lb)
Dorchester,
VA
1.00
6.24
33.30
59.46
0.96
0.04
80.43
4.79
1.72
4.82
32.95 (14,234)
Pittsburgh #8
1.49
7.59
38.28
52.64
2.35
0.14
76.73
5.21
1.34
5.15
32.00 (13,822)
Hanna Seam,
WY
9.09
5.37
38.33
47.21
0.57
0.05
67.09
5.06
1.44
11.33
27.28 (11,784)
for PC, but this is to be expected, given the lower heating value and
flame temperatures of CWM fuels.  Measured flame temperatures in the
dome region of the primary zone for PC firing are shown in Figure 6.
The primary zone temperatures at the nominal primary zone operating
point are 2100 to 2000 K  (3320 to 3140F)  for PC firing and some 150
to 200 K  (270 to 360F)  lower than  this  for  CWM  firing.

     Figure 7 shows measured primary zone NOX concentrations for
pulverized coal firing.  These measurements were made at the exit from
the primary zone, just upstream of the main exit nozzle.  The
measurements are compared both with calculated equilibrium NOX levels
for PC firing and also the NSPS limits,  as described above.  The limit
of resolution of the chemiluminescent analyzer used in making these
measurements is of order 10 ppm.  The measured NOX  concentrations are
well below the NSPS limits and are in general agreement with the
calculated equilibrium concentrations at 100 to 200 K below the
adiabatic flame temperature.  The measured flame temperatures, shown
in Figure 6, are typically 150 to 200 K below the adiabatic flame
temperature.

     Corresponding primary zone results and equilibrium calculations
for the case of 60 percent solids CWM firing are shown in Figure 8.
The results for CWM firing are substantially different from those for
PC firing.  While the calculated equilibrium NOX concentrations  for
CWM firing are lower than those for PC firing, because of the lower
flame temperatures,  the measured NOX  concentrations  at  the primary
zone exit for CWM firing are considerably higher than those for PC
firing.  The CWM measurements are also considerably higher than the
                                 6B-22

-------
calculated equilibrium concentrations for CWM firing.

     This increase in NOX concentrations  for  CWM firing is  also
reflected in the lean zone exit, or exhaust emissions, measurements.
Measured NOX exhaust  emissions,  corrected to  15  percent oxygen,  for
three PC fuels and for 60 percent solids Dorchester CWM are plotted as
a function of primary zone fuel-air equivalence ratio in Figure 9.
These measurements were made at the combustor exit, downstream of the
lean-burn zone.  The overall combustor fuel-air equivalence ratio at
the lean zone exit was fixed as the primary zone equivalence ratio was
varied.  The dramatic reduction of NOX levels with  increased staging
of the combustion is clearly illustrated.  The NSPS limit  (0.6 Ib/MM
Btu for bituminous coals), scaled for the combustor exit conditions,
is shown for reference.  At the nominal design operating point, the
combustor NOX emissions for both PC and CWM firing  are well below the
NSPS limit. Not enough information is available to partition the
exhaust NOX emissions between contributions from (1)  primary zone NO
generation;  (2) lean-burn zone oxidation of volatile or char nitrogen
carried over from the fuel-rich zone; and  (3) generation of thermal
NOX in the lean-burn  zone.   It is obvious,  however,  that NOX is
generated in the lean-burn zone.  For example, NOX  levels at the  rich
zone combustor exit  at equivalence ratios in the range 1.3 to 1.4  (the
nominal primary zone operating point) for PC firing have been measured
at 20 to 40 ppm.  The primary zone typically has one third of the
total gas mass flow  rate.  If no NOX was  generated  in the lean-burn
zone, the primary zone NOX would therefore be diluted by a  factor of
approximately three,  giving emissions on the order of 10 to 15 ppm.
The actual emissions at this condition are of order 30 to 50 ppm.
Thus some 20 to 40 ppm NOX are generated in the  lean-burn zone.   These
20 to 40 ppm are either from thermal NOX in the  lean-burn zone or from
lean zone oxidation  of char of volatile nitrogen carried over from the
primary zone.

     The exhaust NOX  emissions for CWM firing are slightly  higher than
those for PC firing.   At the nominal primary zone operating point, the
CWM emissions are in the range 60 to 80 ppm, compared with 30 to 50
ppm for PC firing.   While the precise mechanisms leading to the higher
levels of NOX with CWM firing are not clear at present,  several
contributing factors may be identified.  As discussed above, the
measured NOX levels at the primary zone exit  for CWM firing are much
higher than those measured at the same location for PC firing.  In
fact, for CWM firing the measured NOX is  in excess  of the
thermodynamic equilibrium level.  Thus NO destruction would be
expected downstream  of the primary zone exit.  This indeed appears to
be the case: if the  assumption of no NOX generation in the  lean-burn
zone is again made,  and the primary zone NOX concentration  is diluted
by a factor of three, the NOX concentration so obtained is  of order
160 ppm, considerably in excess of the measured NOX emission for CWM
firing of 60 to 80 ppm.  This suggests that NO is destroyed between
the primary zone exit and the lean zone inlet.

     The large differences in primary zone NOX between PC and CWM
firing are indicative of significant differences in temperature,
heating rate and stoichiometry histories in the fuel-rich primary  zone
for the two fuels.  As discussed above, the partition of the  fuel-
bound nitrogen between volatile and char nitrogen and the  subsequent
conversion of NOX  precursors to molecular nitrogen  are strongly
dependent on such parameters.  Because of its high moisture content
                                 6B-23

-------
and large size, a CWM droplet will experience both a lower heating
rate and a lower final temperature than a pulverized coal particle.
This may lead to both less complete evolution of fuel-bound nitrogen
and also less efficient conversion of released fuel-bound nitrogen to
molecular nitrogen and consequently to higher NOX emissions.

     The post-run appearance of the slag layer in the primary zone
would also indicate that more wall burning occurs for CWM firing than
for PC firing, possibly because of the production of relatively large
coal particle agglomerates on evaporation of the moisture in the CWM
droplet.  These larger coal particles will be inertially separated
from the toroidal vortex onto the slagged wall before burning out
completely.  Thus the gas phase stoichiometry for CWM burning is
leaner than the global stoichiometry based on air and fuel inputs.
NOX levels at the exit of the primary zone  may therefore reflect the
equilibrium levels at leaner conditions, and given enough residence
time, would eventually be reduced to levels indicative of the global
stoichiometry.

     Figure 10 shows the exhaust NOX emissions plotted as a  function
of the combustor outlet temperature.  The nominal design outlet
temperature is 1850F,  at which temperature the NOX emission is  some
40 ppm.  There is only a moderate increase in NOX emissions  as  the
outlet temperature is increased to 2000F.

     NOX generation and destruction in staged combustion are
controlled by an extremely complex series of phenomena.  Given the
limited amount of experimental information available from a practical
staged slagging combustion system such as the one currently being
tested, it is difficult to completely identify the precise mechanisms
responsible for the results obtained.  However, the general concept of
staged combustion for NOX reduction has  worked extremely well  in this
application,  leading to NOX emissions on the order of one fifth of the
NSPS requirements.

CONCLUSION

     A three-stage combustion concept has been developed for
application to direct coal-firing of utility gas turbines.  A key
aspect of combustor performance is the effective control of NOX
emissions.  A subscale combustor  (3.5 MWt,  6 atm)  is  currently  being
tested.  Results for various coal fuels fired as either PC or CWM have
shown extremely good coal particle burnout leading to effective
slagging in the primary zone.  The combustor employs staged combustion
 (fuel-rich conditions in the primary zone to inhibit NOX production
from fuel-bound nitrogen; rapid quench/good mixing in lean-burn zone
to reduce peak flame temperature and minimize thermal NOX production)
for NOX emissions control.   For primary  zone fuel-air equivalence
ratios greater than approximately 1.1 for PC firing and  1.15 for CWM
firing, the subscale slagging combustor NOX emissions are well  below
the NSPS limit.  Given the high levels of fuel-bound nitrogen in the
coals burned  (typically 1.3%), the staged combustion has worked
extremely well to control NOX emissions.

ACKNOWLEDGEMENTS

     The work described in this paper is sponsored by the U. S.
Department of Energy through the Morgantown Energy Technology Center
                                 6B-24

-------
under Contract No. DE-AC21-86MC23167.  Mr. Donald W. Gelling  is  the
METC Program Manager.

REFERENCES

Abichandani, J. S., Loftus, P. J., Diehl, R. C., Woodroffe, J. A., and
Holcombe, N. T.  (1989) "Nonequilibrium Sulfur Removal  from High
Temperature Gases," Proceedings:  Sixth Annual Pittsburgh  Coal
Conference, Pittsburgh, PA, September, 1989.

Chatwani, A. U., Turan, A., and Stickler, D. B.  (1988)  "Design and
Sizing of the Primary  Stage of a  Toroidal Vortex Gas Turbine  Combustor
Using a 3-D Flow Field Modelling  Code," Western States  Section Meeting
of the Combustion  Institute, Salt Lake City, UT, March, 1988.

Loftus, P. J., Chatwani, A. U., Turan, A., and Stickler,  D. B.  (1988)
"The Use of 3-D Numerical Modelling  in the Design of a  Gas Turbine
Coal Combustor," Heat  Transfer in Gas Turbine Engines  and Three-
Dimensional Flows, ASME HTD-Vol.  103, pp. 95-105, edited  by E. Elovic,
J. E. O'Brien, and D.  W. Pepper,  New York.  Also presented at ASME
Winter Annual Meeting, Chicago, IL,  December, 1988.

Mattsson, A. C. J., and Stankevics,  J. 0. A.  (1985) "Development of  a
Retrofit External  Slagging Coal Combustor Concept," Proceedings:
Second Annual Pittsburgh Coal Conference, Pittsburgh,  Pennsylvania.

Pershing, D. W. and Wendt, J. 0.  L.  (1979) "Relative Contributions of
Volatile Nitrogen  and  Char Nitrogen  to NOX Emissions from Pulverized
Coal Flames," Industrial and Engineering Chemistry: Process Design and
Development, 18  (1);  60-66, 1979.

Pillsbury, P. W.,  Bannister, R. L.,  Diehl, R. C., and  Loftus, P-  J.
(1989)  "Direct Coal Firing for Large Combustion Turbines: What  Do
Economic Projections and Subscale Combustor Tests Show?"  ASME Paper
89-JPGC/GT-4, Joint ASME/IEEE Power  Generation Conference, Dallas,
Texas, October, 1989.

Smart, J. P- and Weber, R.  (1989) "Reduction of NOX and Optimization
of Burnout with an Aerodynamically Air-Staged Burner and  an Air-Staged
Precombustor Burner,"  Journal of  the Institute of Energy, December
1989, pp 237-245.

Stankevics, J. 0.  A.,  Mattsson, A. C. J., and Stickler, D. B.  (1983)
"Toroidal Flow Pulverized Coal-Fired MHD Combustor," Third Coal
Technology Europe  Conference, Amsterdam, The  Netherlands.
                                 6B-25

-------
        STAGE
PRIMARY
ZONE
                                                  STAGE I I I
                                                  LEAN BURN/
                                                  DILUTION ZONE
      CENTERBODY
          STAGE I I

          IMPACT SEPARATOR

          CYCLONE SEPARATOR
                                                    ORIFICE PLATE


                                                        >  TO STACK
  Figure  1   Schematic diagram  of three  stage slagging combustor
             concept  including  slagging  cyclone separator
       10000
            NOx (ppm)
                                             NSPS Bituminous Limit

                                             NSPS Subbituminous Limit
        1000 r
         100
                                              Equilibrium NOx at AFT

                                                      AFT  100 K

                                                      AFT  200 K
          10
                    1.1       1.2       1.3       1.4
                            Fuel-Air Equivalence Ratio
                                               1.5
1.6
Figure  2  Calculated thermochemical equilibrium NOX
           concentrations  in the  fuel-rich  zone as  a  function  of
           fuel-air equivalence ratio for three gas temperatures:
           adiabatic flame temperature  (AFT),  100 K below AFT,  and
           200  K below AFT
                                 6B-26

-------
           Species Mass Fraction
          NO Concentration (ppm)
                 0.5
   1.5
Time (msec)
Figure 3  Variation of species concentrations showing final  CO
          burnout and NO generation  in  lean burn zone at  a  fuel-
          air equivalence ratio of 0.6
    Figure  4   Photograph  of  subscale slagging  combustor test
               arrangement
                              6B-27

-------
        Carbon Conversion (%)
               PC L/D - 1.25
               CWM L/D - 1.50
               CWM L/D  1.25
      80
       0.9
1.1    1.2     1.3    1.4     1.5    1.6
Primary Zone Fuel-Air Equivalence Ratio
1.7
1.8
 Figure 5   Measured primary zone  carbon conversion for  PC and CWM
            firing as  a function of fuel-air  equivalence ratio
         Measured Flame Temperature (K)
zouu
2500
2400

.. ^Tn^

D Pittsburgh #8
^ Wyoming Rosebud
0 Dorchester


    1600
        0.6   0.7   0.8  0.9    1    1.1   1.2   1.3    1.4    1.5
                     Primary Zone Fuel-Air Equivalence Ratio
                                        1.6
      1.7
Figure  6  Measured primary  zone flame  temperatures for PC  firing
           as a  function of  fuel-air equivalence  ratio
                                 6B-28

-------
          NOx (ppm)
     10000=	
                                           NSPS Bituminous Limit

                                           NSPS Subbituminous Limit
       100k
        10 c
                                            Equilibrium NOx at AFT

                                                    AFT  100 K

                                                    AFT  200 K
             -X-  Measured NOx (PC)
                   1.1       1.2       1.3       1.4
                          Fuel-Air Equivalence Ratio
                          1.5
1.6
Figure  7  Measurements of  NOX  concentrations at exit  of primary
           zone  for PC firing and calculated equilibrium NOX
           concentrations for PC combustion as  a  function  of fuel-
           air equivalence  ratio
      10000
            NOx (ppm)
       1000 =
        100
                                            NSPS Bituminous Limit

                                            NSPS Subbituminous Limit
                                         Equilibrium NOx at AFT


                                                 AFT  100 K

                                                 AFT  200 K
                    1.1
 1.2      1.3       1.4
Fuel-Air Equivalence Ratio
                                                       1.5
 1.6
Figure  8  Measurements of  NOX  concentrations at  exit of primary
           zone  for CWM firing and  calculated  equilibrium NOX
           concentrations  for CWM combustion as a function of
           fuel-air equivalence ratio
                                 6B-29

-------
      600
      500
      400
      300
      200^
          NOx (ppmv, dry, corrected to 15% O2)
       100
                                             0  PC L/D - 1.25

                                             *-  CWM L/D  1.50
                                                      NSPS Limit
        o  	
         0.8   0.9
   1     1.1    1.2    1.3    1.4   1.5    1.6
   Primary Zone Fuel-Air Equivalence Ratio
1.7
1.8
Figure  9   Measured  lean-zone exit  NOX  concentrations (dry,
           corrected to 15 percent  oxygen) for  PC and CWM  burning
           as a function of primary zone fuel-air equivalence
           ratio
       100
        80
        60
        40
          NOx (ppmv, dry, corrected to 15% O2)
                                     A
        20
         0
        1500
1600      1700      1800      1900      2000
    Combustor Outlet Temperature (deg F)
                                                              2100
Figure 10 Measured lean-zone  exit NOX concentrations  (dry,
           corrected to 15 percent oxygen)  as a function of lean-
           zone  outlet temperature
                                 6B-30

-------
       LOW NOX COAL BURNER

   DEVELOPMENT AND APPLICATION
           J.  W.  Allen
NEI-International  Combustion Ltd
           Sinfin  Lane,
     Derby, England  DE2 9GJ

-------
                              LOW NOX COAL BURNER
                          DEVELOPMENT AND APPLICATION

ABSTRACT

The paper describes the development and  application  of a front  wall  low  NO   coal
                                                                         A
burner in the U.K.  power industry.

Target NO  emission  levels  set  by European Community  Directives,  for the  U.K.
         X
industry, were met both  in  full  scale single burner  thermal trials  and in  the
multi burner boiler operation.

The  paper  highlights  the  basic  differences   between  test   rig   and  boiler
installations,  not  only  in  combustion  performance   but  also  in  the boiler
operational  effects  which  influence  the  selection  of materials of  construction
for the critical  burner parts.

In order to optimise the boiler  performance,  the characteristics of the low  NO
                                                                              A
burner  must be recognised in the boiler  operating  procedures.
                                     6B-33

-------
INTRODUCTION



Current UK NO  emission targets for large combustion plant  (i.e.  plant  with  heat
             X

input greater than  50  MW thermal.,  are based on  a  European Economic  Community


(EEC)  Directive  (88/609/EEC)  issued  in  December,   1988   (1).    The  Directive


stipulates  limits  for  new  large  plant  and  also  NO   reduction  targets  to  be
                                                    X

achieved by the various EEC countries over the  decade  to  1998.   NO  limits  for
                                                                   X

the various fossil  fuels are given in  Table 1.





                                    Table 1



                  EMISSION LIMIT VALUES  FOR NO   FOR  NEW PLANTS
                                             X



                      Type of Fuel                Limit Values  (mg/NmJ)




           Sol id in General                            650

           Solid with less than 10% volatiles         1300

           Liquid                                      450

           Gaseous                                      350
Although these NO  levels refer to  new  plant  they have become target norms  for
                 X

the retrofitting  of  power  generation  boilers  in the UK for  low NO   operation.


Furthermore the  UK  is  required to reduce NO   emission levels  by  15% prior to
                                            X

1993 and 301 prior to 1998,  based  on NO   emission  levels in  1980.
                                       X




European units for NO  concentrations  are frequently quoted in mg/Nm3, although
                     X

most concentration measurements are made in terms of  parts  per million  (ppm).


For comparison purposes the ppm concentration is  referred  to  either  a 3% or 6%


dry waste gas oxygen  concentration.   Table  2 gives the interconversion  factors


for terms commonly used for the expression of  NO  concentrations.
                                                X
                                      6B-34

-------
                                    Table  2
                   INTERCONVERSION OF NOX CONCENTRATION TERMS
         To convert              To        Multiply by
            From                    >
                                                          D6

         mg/Nm3                   -        0.487     8.14 x 10
                                mg/Nm3       ppm      lbs/10D Btu
                                                              -4
                                                              -3
         ppm                     2.05        -       1.67 x 10
         lbs/106 Btu             1230        598
Table 2 is based  on  coal  combustion with a dry  flue  gas 02  content of  6%.   To
correct
used:-
correct NO  concentrations,  at differing 0?  levels,  the  following  formula  can  be
          A
            N0y (ppm at 02 n J  .    21	p_2 m      NOY (ppm at 02 m )
              A           \  I f               \ L- )        A           \ L. I
                                    21   02(1)

Prior  to  the  privatisation of  the  electricity  industry  in  the  UK  the  CEGB
announced a E170M programme in  order  to achieve the reductions  in  NO   emission
                                                                     X
levels  as required  by the EEC  Directive.    The   two major  privatised  power
generators, National Power and PowerGen, are  continuing with this programme.

Progress in the conversion of corner fired units in the UK has  proceeded quickly
following  the  successful  demonstration  of the  'Low   NO   Corner Firing  System
                                                        A
(LNCFS)1 installed in a single 500 MW boiler  in the CEGB,  North  Western Region,
in 1985(2),(3).  The 500  MW+  corner  firing capacity of both National  Power  and
PowerGen is committed to this low NO  system.
                                    X

Conversion of the wall  fired  units has  proceeded  more slowly,  at  the  time  of
writing around 25-30% of  the UK wall  fired coal capacity  has  been  converted  or
committed  to  low  NO   burner  retrofit.    This slower  progress  has   enabled  the
                    X
power generating and manufacturing  organisations  to proceed via a  well  defined
programme  based on  isothermal  and  mathematical modelling,  single   burner  full
scale rig testing  and the testing of individual burners within  an  actual boiler
environment,  before  commencing  a  full  boiler commercial retrofit.    All  the  low
NO  burner developments, including corner firing,  have  been  based  on combustion
  X
staging techniques,  which have been demonstrated as capable of  achieving the NO
                                                                               X
                                      6B-35

-------
reduction  requirements  of  the  EEC  Directive.    The  burner  development  and
operations described  in  this paper relate to  a front  wall  low NO  coal  burner
                                                                  X
incorporating both fuel  and  air  staging into the basic design.   Although these
burners  are  capable  of meeting the  NO   emission requirements up to 1998  it is
                                       X
anticipated  that a  tightening of  the  regulations  within the  EEC will  occur
before that  date.  Improved  internal  staging,  furnace  staging  and,  perhaps,  post
combustion  No  reduction  techniques  will  have to  be  introduced  to  meet these
              X
more  strict  emission  limits.

If  post  combustion  reduction techniques  are  eventually  required,  an  accepted
basic low NO   burner  system will  enable any  future  emission  regulations to be
             X
met  effectively both  in terms of  speed of  implementation  and minimum  capital
cost.

PRINCIPLES  OF BURNER DESIGN AND DEVELOPMENT
The  current  NEI-ICL  low  NO  wall  burner  design  is  shown  in  Figure  1.   Air
                             X
staging  is  achieved by  splitting  the combustion air into independently swirled
secondary and tertiary streams.   Fuel  staging  is achieved  by  means  of  fuel flow
redistributors  (FFR)  located in the  pulverised  coal/primary air  stream  close to
the  burner  exit.  Situated  in this  location the FFR  produce a  fuel  lean/fuel
rich  profile at the burner  mouth.   Ignition of the  main pulverised coal  fuel
(PF)  is  achieved via a centrally located oil burner with its  integral combustion
air  supply  fan.  PF is  supplied,  from  the  PF  supply  piping,  via  a tangential
inlet and scroll distribution system to the   annular  burner  fuel  duct.    The
design concepts were developed using isothermal modelling  techniques, to examine
both  the  flow of fuel  and  the  air  distribution within the burner system.   Fuel
flow  work addressed  the problem of  roping  within  the  burner fuel  annulus  and
produced  an evenly distributed flow into the FFR system which  then  produced  the
required  fuel staging effect at the  burner  exit.  Various forms of FFR  devices
were  tested using flow visualisation techniques.  Air distribution  and  air swirl
were  studied in relation to  the  recirculation and  general  mixing   patterns
produced  both   in the  near  burner  region  and  further  downstream.   Figure  2
illustrates a typical recirculation pattern  from an  early  burner design.

Following the isothermal  model  work a series of potential  low  NO  burner  design
                                                                X
configurations  were  selected for  thermal  testing,  at full scale, in  the  88  MW
NEI-ICL burner test  facility.  The initial  full  scale tests related  to  a  37 MW.,
burner design which  would  be required for several   48  burner  500 MW front wall
                                      6B-36

-------
coal fired  units  in  the UK.   During  this work,  operating  parameters,  such  as
those  relating  combustion  air preheat  and  heat  input  to  NO   levels,  were
                                                                A
established  (Figures  3a and  3b).   In  this  work  the  principle  of  good  flame
retention at the burner mouth, as a pre-requisite of low NO   operation, was also
                                                           A
established.  This and the effectiveness of the FFR, in controlling  overall  NO
                                                                               X
emission levels, is illustrated in Figures 4a and 4b.

In order to  relate the  test rig burner performance to  site  boiler  performance,
particularly with respect to NO  emissions, the test rig was refractory lined in
                               X
a pattern  determined  by computer  calculations,  such  that  the  rig  centre  line
temperature was similar to the boiler  centre  line flame  temperature,  as shown in
Figure  5  (4).    To  demonstrate the effectiveness of this  approach  a  standard
burner  from a  500 MW boiler  was  rig  tested  under  these   conditions and  did
reproduce site NO  levels of around 700 ppm at 3% Qz .   Thus  a 1:1 rig  factor in
                 X
respect of  NO  emission levels was established.
             X

Further work was  carried  out on flame retention,  which  resulted in  successful
patent  applications for the burner design(5) and also  up-rating  of the  design
from 37 MW., to 58 MW.,  without an increase in NO  emissions.   The 58  MW  burner
was also  required  to operate  with a  primary  air to pulverised  coal  ratio  of
1.2:1 compared to the more usual 1.5/2:1  range.  Furthermore the primary  air was
vitiated by the  use  of recycled flue gas  into  the ball  mills  for  coal  drying
purposes.    This  primary   air  vitiation  and  low  pa:pf  could  aid   low  NO
                                                                               X
performance  of the  burner   but  also  adversely  affect flame  stability and  burn
out.

Figure 6 demonstrates the NO  performance of  this larger burner  showing not only
                            X
the usual trend of increasing NO  with waste  gas Oz  content  (with a NO   level of
                                X                                     X
375 ppm  at  3%  02),  but also  that  the burner  can operate  at  lower excess  air
rates than  normally   used  for  coal firing  without the  generation  of high  CO
levels.   Corresponding  with CO levels below 100  ppm the  carbon in  dust  levels
measured on the rig tests were a maximum  of  2%.   During the thermal trials  the
opportunity  was  taken  to  collect   in-flame  gas  samples   and   temperature
measurements.   Contour  plots  of  gas  and  temperature  variations  are  shown  in
Figures  7a-7d.    These emphasise  the  importance  of   the  near  burner  region
aerodynamics in establishing  a centrally  located  reducing  atmosphere with  the
flame envelope which  encourages  the  formation  of Ha  rather than NO  from  the
                                                                     X
nitrogen contained in the  fuel.   High  NO  levels were  produced in  the outer
                                           X
regions of  the  flame,  close to the  burner, corresponding with  the  mixing  of

                                      6B-37

-------
secondary air and the outer layers of the fuel  stream.   This NO  mixed  later  in
                                                               A
the flame with the reductants produced in  the  flame  core,  thus producing  a  low
overall NO  emission from the flame.
          X

Depending on  the  particular  conditions  rig NO  levels were  in  the 300-400  ppm
                                              A
range (related to 3% Oz,  dry) which  represents an approximate 50%  reduction  in
NO .
PERFORMANCE OF BURNERS IN SITE INSTALLATIONS
Prior to the  possible  retrofitting  of a full boiler  set  of  low NO  burners  it
                                                                   A
was considered prudent to replace just one  or two  standard burners,  with  the  low
NO  designs,  in an operating boiler.   This  preliminary installation  would  enable
the  compatibility  of  the  low  NO  burners,  within  a  hot  multi-burner  furnace
                                 X
environment,  to be assessed from an  operational  and  durability standpoint.   Two
37 MW. .   low  NO   burners were installed,  on  a  48 burner  500 MW boiler,  in  the
     L n        x
wing and centre top row locations and  a single  58  MW.,  burner installed  in  the
centre  top  row position  of  a 32 burner  500 MW  boiler.    The centre  top  row
location was  considered  to  give  the most  hostile  conditions regarding  burner
component temperatures, particularly in the non-firing mode.   The wing  position
enabled a qualitative  assessment of  the burner, in  operation,  to  be made.   The
centre top row burners were  inspected,  in-situ, using a water cooled  periscope
inserted into the burner  via de-ashing ports, critical components  of the  burner
were  instrumented  with   thermocouples to  provide  burner   metal   temperature
variations in both the firing and non-firing operational modes.

Temperatures recorded  from the single low  NO  burner, installed in  the  standard
                                            X
burnered furnaces,  gave  some  cause  for  concern,   as  in   the  non-firing  mode,
temperatures  approaching  recommended  limits  for  the material   used   in  the
critical burner  areas were  recorded,  with the normal  10-15% MCR  cooling  air
equivalent passing through the burner.

Computer calculations  of  heat  flux  based  on  test rig data,   of  low NO   burner
operation,  showed that with a full boiler  set of low NO burners the temperature
                                                       X
of the  critical  burner components would be  satisfactory.   The  main reason  for
this was  the lower peak  flame temperature  of  the  combustion staged  low  NO
burner  system  which also  occurred  further  down  stream  from  the  burner  exit.
There was  also a  change  in the gas  recircul ation pattern  at the  furnace  front
wall as a result  of the low NO  burner design.
x
                                      6B-38

-------
Periscope observations  indicated  the possibility of some ash  deposition  in the
low NO  burner installed in a conventional boiler burner system.   From both test
      X
rig experience  and  computer predictions  it  was  postulated  that the  change  in
front  wall  flow  patterns  from  a  full  boiler  set of  low  NO   burners  would
                                                               X
eliminate this possibility.
Although  both  rig  operating experience and computer predictions  indicated  that
neither  high material  temperature or ash deposition would  be  a  problem with  a
full  set  of low NO   burners,  material  specifications  for  the critical  burner
                   X
components  were  selected and  a  minor modification  made to the  secondary  air
stream  aerodynamics  to provide further assurance.   In  practice,  with  the  full
boiler  set  of low NO   burners, the computer and test rig  predictions,  regarding
                    X
critical  burner metal  temperatures  and  ash  deposition,   were  verified.    By
carrying  out  these  investigations  a  considerable  data  bank  was  compiled  on
potential materials for burner construction covering fabricated,  cast  materials,
coated  materials and  ceramics.   Data on erosion  resistance of these  materials
exposed to  flowing pulverised coal streams were also obtained.   Table  3 compares
the temperatures measured in the single low NO  burner  and  the multiple  low  NO
                                              X                                X
burners after the boiler modification.

                                     Table 3
                      BURNER METAL  TEMPERATURES COMPARISONS
                 BEFORE AND AFTER  LOW NO  BOILER  MODIFICATIONS
                                        A

      Burner Component                        Temperature C
                               Before Modification       After  Modification
                                   mean   peak              mean     peak

    Tertiary Air Duct               880    980               868     1011
    Secondary Air Duct              870    950               838      915
    Oil burner core tube            730    810               707     792

Temperatures  in  Table 3  relate  to the  non-firing mode  with  10-15% of  normal
firing air supply passing through the burner.
                                      6B-39

-------
Also, prior to the installation of a full  boiler  set  of  low  NO   burners,  the  NO
                                                             X                A
and CO levels were measured on  an  unmodified  boiler  (6).   The results are  shown
in Figure 8.   In general the unmodified boiler  NO   emissions were in the  range
                                                 X
550-730 ppm (related to 3% Qz ,  dry),  depending upon the  excess air  level, with a
mean level  of 633 ppm at 3% 02.  Thus a 30%  reduction  in NO  would require  the
                                                            X
boiler to operate  at  a mean figure of 443 ppm  well  within  the  capacity of  the
burner,  from  the  rig test  data   (see  Figure  6).   Carbon  in  dust  from  the
unmodified  boiler was in the range 0-6%    3.3%  (mean  1.93%)  depending upon mill
groups in operation  and  excess air levels,  under similar  conditions  CO levels
were recorded in the 60-200 ppm range.

Figure 9 shows  the  results from the  initial  commissioning  trials  of  the full
boiler set  of low  NO   burners,  covering  the whole  range  of  mill  groups  and
                      X
excess air  levels,  equivalent  to the 2-5% waste  gas  02  range and compares them
wit  te  test  rig burer  performace.    Summarising  these  early  results  from  te
boiler, the low  NO   burners,  in  combination,  can operate  under  the conditions
                   X
outlined in Table 4.

                                    Table  4
                        INITIAL COMPLIANT OPERATING RANGE
                               OF  LOW NO   BURNERS
                                       A
                                                Oz level
                                           3%             4%
                 NO  ppm                  330            430
                 CO  ppm                   25             10
                 C in Dust %                 52

NO   levels  in Table 4 refer to  ppm at 3% 02 dry.
  X

The  results  confirm  the  1:1  rig  factor to  boiler  factor relating   to  NO
                                                                              X
emissions,  in the 3-4% waste gas 02  range.  The CO emission  results in Figure 9
indicate that the 100 ppm CO level  would not  be  exceeded  until excess air levels
equivalent  to 1.8% 02 were obtained,  this  compares to 2.6%  02  in the unmodified
boiler.  Over the  3-4%  waste  gas  02  range  the CO  levels  in  the  boiler were
similar to  those in the rig tests,  however there  is a tendency for  a more  rapid
increase in CO generation,  below 3% Qz , in the boiler compared to the test  rig.
                                      6B-40

-------
The  average of  all  the  boiler  NO   level  results  gave  399  ppm  NO   which
                                    X                                   X
corresponds to a 37% reduction in NO  compared to the mean level  of NO  from the
                                    X                                 X
unmodified  boiler.    This  reduction  should  be  even   greater   when   burner
optimisation is complete  to  enable  the  burners  to operate  at  lower Oz  levels
without excessive CO generation in the boiler.  Carbon in  dust  levels increased
in the low NO  burnered boiler to an average of around 5% (at 3% 02) compared to
             X
2% in the unmodified boiler  (see  Figures  8&9).   The general  practice with  this
boiler is to over-fire on the bottom rows of burners in  the  unmodified  boilers,
as a  means  of  controlling superheater temperatures and  this practice has  been
continued on the modified boiler.   Some burners are therefore operating  at lower
overall air to fuel ratios, however, the increased swirl  and hence  shorter flame
length  of  the unmodified  burners  produces  sufficient  in  furnace  time  and
turbulence to produce a low C in dust loss overall.

As a  result of staged  combustion effects  low  NO  burners  have  a low  overall
                                                  X
swirl producing increased  flame  lengths  and low furnace turbulence levels.   We
now know that higher carbon in dust  levels are generated from the  burners  which
are operating  at  lower  overall  air  levels.   The  time,  temperature  and  mixing
history (Oz  availability), which  controls the  combustion  reactions within  the
boiler,  including   NO   emissions  is  influenced  by furnace  geometry  and  air
                     X
quality.    The  10  m depth (with  an  approximate 3:1  width:depth  ratio)  of  the
boiler coupled with the use of  vitiated  air for coal  conveying have  an  adverse
effect on the  final  burn-out  characteristics.  Optimisation  of  the  boiler  and
burner performance, fully  recognising  the low swirl characteristics  of  the  low
NOx burners, should improve this situation.

CONCLUSION
Single full  scale burner test facilities can be used to  indicate multi-burnered
boiler NO   emission  levels.    Combustion  staged  low  NO   burner  designs  are
          X                                                X
capable of meeting current legislation relating to NO   emission  levels.
                                                     X

Front wall  environments  are  less  hostile to  burner  components  in  a  low  NO
                                                                               X
system compared to a conventional  front wall coal  burner system.

Low NO  burner characteristics  must  be fully recognised in  the optimisation  of
      X
low NO  front wall  burner boiler operations.
      X
                                      6B-41

-------
ACKNOWLEDGEMENTS

Thanks are due to the Directors of NEI-ICL for permission to  publish  this paper
and to many  colleagues  within  NEI-ICL responsible for  providing both  test  rig
and boiler commissioning data.

Thanks  are  also  due  to  PowerGen  Technical   and  Station  personnel  for  the
provision of  boiler  operating  data and continued  enthusiastic  interest  in  the
project.

REFERENCES

1.    Official  Journal  of  the  European   Communities  L336  "Council   Directive
      88/609/EEC  of  24th  November,   1988  on  the  Limitation  of  Emissions  of
      Certain Pollutants into the Air from Large Combustion  Plants"
      7th December,  1988.

2.    J.   W.  Allen,  W.  J.  D.  Brooks,  N.  A.  Burdett,  F.  Clarke  and  G.  Foley.
      "Reductions in NO   Emissions  from  a 500 MW Corner  Fired  Boiler."   Joint
      Symposium on Stationary NO  Combustion Control.   New Orleans,  1987.
      	X	

3.    J.   W.   Allen    "NO   Reductions  in  Coal  fired  Boilers."    Modern  Power
                        X                                          	
      Systems.  June, 1987.

4.    Private  Communications.    M.   J.  Sargeant,  S.  Cooper  -  CEGB,  Marchwood
      Engineering Laboratories, 1984.

5.    UK Patent 8805208
      USA Patent 317743
      European Patent 89302101.4

6.    Private Communication.   CEGB
                                      6B-42

-------
                  Secondary air
                  control damper
          Secondary air
          swirl vanes

         Tertiary
         air vanes
         Outer  back  plate



 Sight tube
Conical liner

     Core air  tube

           PA/PF inlet
Secondary
air tube
    Tertiary
    air tube
                                                                               Entry chamber
                                                                         Rodding tube
                                       Support tubes
                         Secondary/tertiary
                         air  shut  off damper
Fuel  flow
redistributors
                   Figure  1.    Low  NOx  Front  Wall Coal  Burner.
                  Axial distance (m)
                         1.5-1
                                      Burner centre  line
                         1.0-
                                                                Flame
                                                                boundary
                        0.5-
                                                            Central
                                                            recirculation
                                                            zone
                     Figure 2.    Low NOx  Coal  Burner  Model.
                     Typical  Recirculation  Pattern.
                                          6B-43

-------
NOx  (ppm)

500 -i
400 -


300


200 -


100
  0
     0         100       200

          Air  preheat temp. (C)

      3a Effect  of Air  Preheat on
      NOx  (Excess Air  = 3% O2 )
                                  300
                      NOx  (ppm)

                      500 -i
                      400 -


                      300 -


                      200


                      100
                                             0
                                       I
                                      50%

                                   Burner  load
                                                                       100%
                           3b Effect of Burner  Load on
                           NOx (Excess Air =  3% 02 )
                           100% Load  = 58MW.
     Figure  3.    Effect of  Air  Preheat  and  Burner Load  on  NOx.
NOx  (ppm)

700 -i
600 -


500 -


400 -


300 -


200 -


100 -
  0
Fully lifted
flame
Well anchored
flame
     01     2345

           % 02 in  waste  gas

 4a  Effect of Flame Retention on  NOx
                      NOx  (ppm)

                      700 -i
600 -


500 -


400 -


300 -


200 -


100 -
                                             0
Burner
without
FFR  ^-
Burner
with  FFR
                           0
           Figure  4.
                                12345
                                 % 02 in  waste  gas

                        4b Effect of Fuel  Staging on NOx

   Effect  of  Burner  Parameters  on  NOx
                                   6B-44

-------
 Centre  cell
 gas temp.  (K)
  2000 -i
   1750-

   1500 -

   1250-

   1000	

   750 -
   500
                                          	 Test  rig
                                          	Boiler
 I
10
 I
12
                                        l
                                        14
                         16   18
   I
  20
                    Axial distance  (m)
 Figure  5.   Comparison of Refractory  Lined Rig  and
 500MW Boiler Centre  Line Temperatures.
NOx  (ppm)
500 -i
400 -

300 -

200 -

100 -
                             CO (ppm)
NOx
         CO
-100
-80
-60
-40
 20
                                                     0
                    % O2 in waste gas
Figure 6.   Test Rig  Performance  of 58MW  (Thermal)
Front Wall  Coal Burner.
                         6B-45

-------
CD
DO
 I
-P-
O)
                 O-i


                 1 -


                 2-


                 3-
                   0     2
O-i
                                         Burner centre line
                                     280
                    250
                                                   I
                                                   12
                              4    6    8    10    12   14

                                Distance along axis (m)

                                  7a NOx  Contours (ppm)
                  Burner centre line
                                             I
                                            16
                   0    2     4    6    8    1012

                                Distance along axis  (m)

                                  7c CO Contours (%)
 \
14
                                            16   1!
                                                           O-i
                                       Burner  centre  line
                                       1
                                                                10
                                                                                                     2.5
                                                           O-i
                                                                             1 -
                                                                        4    6    8    10    12    14

                                                                          Distance along axis (m)

                                                                            7b  02 Contours  (%)
                                                                             Burner centre line
                                                                 ! 6   1!
                                                                                              900
                                                                                     900
                                                                                                  800
                                                                                                    I
                                                                                                    8
                                                                                                   I
                                                                                                  1 A.
                                                                        4    6    8    1012

                                                                          Distance along axis (m)

                                                                        7d  Temperature Contours  (C)
                                                                 16   18
                                         Figure 7.    In Flame  Gas and  Temperature  Contours.

-------
NOx (ppm)

700 -i
600 -
500 -
NOx
400
                       I
                       2
CO (ppm)

100 -i
 80 -


 60 -


 40 -


 20 -


  0
CO
    o
  % c

  8 -


  6 -


  4 -


  2 -
                                I
                                3
                   % 02 at economiser
                   Unburnt Carbon
                      i
                      4
                       I
                       2
                   % 02 at economiser

     Figure  8.   Unmodified Boiler Performance
                       6B-47

-------
NOx (ppm)

600 -i
500 -


400 -


300 -


200 -


100 -


  0
                                            KEY
 NOx
                               o  Boiler
                               x  Test rig
    0
                                                 I
                                                 5
 CO (oom)

 80 -i
 60 -


 40 -


 20 -
CO
  o
  %  c
 10-,
  6 -
  4 -
  2 -
Unburnt Carbon
             '         '         1	1	i
    01         2345
                  % 02 at economiser

 Rgure  9.   Modified  Burner  Performance  on  Boiler
 During  Commissioning,  Compared  to  Single  Burner
 Test  Rig  Performance
                      6B-48

-------
             Session 7A



        NEW DEVELOPMENTS I








Chair: G. Veerkamp, Pacific Gas & Electric

-------
                  Preliminary Test Results
High Energy Urea Injection DeNOx on a 215 Mw Utility Boiler

             Dale G. Jones, Ph.D., P.E., Noell, Inc.
                Stefan Negrea, P.E., Noell, Inc.
                   Ben Dutton, Noell, Inc.
       Larry W. Johnson, P.E., Southern Calif. Edison Co.
      J. Paul Sutherland, P.E., Southern Calif. Edison Co.
            Jeff Tormey, Southern Calif. Edison Co.
     Randall A. Smith, Fossil Energy Research Corporation

-------
                        Preliminary Test Results
    High Energy Urea Injection DeNOx on a 215 MW UlUlly Boiler

                                    by

                    Dale G. Jones, Ph.D., P.E., Noell, Inc.
                       Stefan Negrea, P.E., Noell, Inc.
                           Ben Dutton, Noell, Inc.
              Larry W. Johnson, P.E., Southern Calif. Edison Co.
             J. Paul Sutherland, P.E., Southern Calif. Edison Co.
                   Jeff Tormey, Southern Calif.  Edison Co.
            Randall A. Smith, Fossil Energy Research Corporation

                               ABSTRACT

Initial tests of a high energy urea injection SNCR  DeNOx system have been
completed  at Southern  California Edison's Huntington Beach Unit 2 gas- and
oil-fired boiler.  The SNCR  DeNOx temperature window in  this 215 MW utility
boiler occurs in narrow cavities and between boiler convection sections.  The
Huntington Beach SNCR DeNOx project  Is a demonstration  of high energy
urea injection  in narrow cavities  to evaluate various DeNOx alternatives and
to bring  such installations in  compliance  with South  Coast  Air  Quality
Management District regulations for  the metropolitan area.

Following contract award in June, 1990, Noell proceeded with injection system
design, installation and start up.  Initial tests of high energy injection into the
2nd cavity and other boiler  zones  were conducted between  Jan. 15 and March
5, 1991. Pressurized urea-water mixtures were  Injected into cross-flowing flue
gas using  high velocity air-driven nozzles.  Initial  2nd cavity injection tests
showed that 25% to 40%  DeNOx Is achieved at full load despite adverse
conditions  of short cavity residence  times (i.e.  40 milliseconds) and floor-to-
roof adverse temperature gradients (l.e. about 200  F). Such adverse conditions
in the 2nd cavity also caused unacceptably high levels of NH3 slip.

Additional  tests were therefore performed to investigate  urea injection into the
1st cavity where the full load temperature is about 2050 F.   Using only four (4)
sldewall  Injection  nozzles, 20% to 25% full load DeNOx was obtained at urea
feedrates from NSR = 2 to  NSR = 4 (NSR is moles of NHi injected vs. moles of
Initial NOx).  Under these conditions, NHs slip  measured upstream from the air
preheater averaged less than 3 ppm, or less than about 1.5%  of NHi feedrate,
Noell Is proceeding with further development of advanced injection systems to
be considered for installation and additional testing at Huntington Beach.
                                  7A-1

-------
1.0   Introduction and  Background

Initial  tests of a  high energy urea injection SNCR DeNOx system have been
completed at Southern  California Edison's Huntington Beach Unit 2 gas- and
oil-fired boiler.  The SNCR DeNOx temperature window in this 215 MW utility
boiler occurs in narrow cavities  and between boiler convection sections.  The
Huntington Beach SNCR DeNOx project  is a demonstration of  high energy
urea injection in  narrow cavities to evaluate various DeNOx  alternatives to
comply with South Coast Air Quality Management District regulations.

Urea (NH2.CO.NH2) reacts at high  temperatures with  NOx in combustion flue
gases,  approximately as follows:

      2 NO  + NH2.CO.NH2 +  0.5 O2  =  2 N2 + 2 H2O  +  CO2

Amine radical (NH2) resulting  from thermal decomposition of the urea reacts
with NO.  The chemical feedrate  vs. quantity of NOx is called the normalized
stoichlometric ratio (NSR),  defined  as  the molar ratio of NHi  being injected
divided by initial NOx.  At Isothermal conditions, the SNCR DeNOx process
operates best over a narrow 'temperature window' between  1600  F  and 1900 F.
If  the flue gas temperature Is too  hot, some of the NH2  radicals form additional
NOx and  DeNOx performance decreases.  If the flue gas temperature Is too
cold, some of the  NH2 radicals form byproduct NH3, called 'ammonia  slip* and
DeNOx performance goes down.  Thus, a 'temperature  window* exists.

This narrow temperature window is the primary drawback  of boiler Injection
SNCR  DeNOx technology.  When boiler  operations change, temperatures at an
injection  location also  change.   Therefore, multiple levels of Injection are
usually required  to provide  good DeNOx performance over  a range of boiler
conditions.   At  low load,  the  temperature may be  too  cold,  and  Injection
should occur at a location closer to the furnace.  At high load, the  temperature
may be too hot, and Injection should be at a location further  from the furnace.

Noell's  boiler injection DeNOx  system uses high  velocity  Injection  Jets to
provide Intense flue  gas mixing.  These Jets can overcome distribution  problems
typically observed, such as non-uniformities In temperature, flowrate,  and/or
composition of the flue gas.  As In any chemical process, intimate and complete
mixing is  Important.  By proper design and  operation of the injection system,
close  approximation  to a well-mixed reactor can be  achieved.  Noell's boiler
Injection Jets are used  for  flue gas  mixing and operate Independently from
chemical feeding, accomplished using feed  pumps for higher or  lower levels of
DeNOx. Chemical distribution occurs first In the Injection Jet, and then as the
injection jet(s) mix with cross-flowing flue gas.  Noell's boiler injection concept
is  Illustrated in Figure 1, which provides results of Jan, 1988 Injection  system
flow model testing for the KVA/Basel MSW incineration plant. The left picture
shows 'channelling1, where a smoke  stream passes through  the  flow path
without much mixing.  The  right picture  is similar except  that scaled-down
injection Jets were installed Into the sidewall(s) of the  flow model to determine
effects  on  mixing.  As can be seen, such high energy injection Jets have a major
Impact on flue  gas mixing.   Similar full-size Injection Jets were subsequently
installed in the 330  TPD Basel MSW Incinerator.  At maximum boiler output at
330 TPD incinerator feed rate, NOx removal  of 70% was obtained  at urea NSR
= 1.3, along with  relatively low  levels of NH3 slip.  (Reference  1).
                                   7A-2

-------
        Figure  1: Photographs of Flow Model Test Results

         KVA Basel 330 TPD MSW Incineration Furnace
                        January, 1988
"Channelling" Effect
  (left-hand picture)
Injection Jet Effect
(right-hand  picture)
                              7A-3

-------
Noell  has also installed its high energy boiler  injection SNCR DeNOx process
at the 325 MW coal-fired power plant of BKB/Offleben in  Germany, which was
started  up for commercial operation  in Sept. 1989.  In this coal-fired  boiler.
Noell's steam-driven nozzles are used  for urea injection to  achieve 95 ppm NOx
at full load. At full load,  the urea NSR is about 0.64, corresponding to about
32% DeNOx with NH3 slip of less than 1.0 ppm. Due to the SO2 content of the
flue  gas,  the Offleben requirement  is less than  5.0 ppm NH3 slip to  avoid
forming ammonium bisulfate deposits in the air  preheater.   (Reference 2)

In more recent developments, Noell has been awarded a contract by the  Public
Service  Company of  Colorado (PSCC) to  design and procure boiler injection
SNCR DeNOx equipment for a Clean  Coal III project at PSCC's Arapahoe coal-
fired  station.  This boiler  injection SNCR DeNOx project is being co-sponsored
by the U.S. DOE  and by EPRI.  Noell  has  also been awarded a contract by the
Tennessee Valley Authority (TVA)  to  conduct perform field testing of flue gas
temperatures, and conduct boiler  flow  model testing  of injection system
options  for a project  being considered by TVA  to  demonstrate boiler injection
SNCR DeNOx at a large coal-fired power plant

2.0   Description of Huntlngton  Beach Unit 2 Boiler

This gas- and oil-fired 215 MW boiler incorporates a pressurized furnace with
front  wall-fired burners arranged  6  wide  by 4  high.  The  drum-type natural
circulation steam generator includes pendant secondary superheater and
reheat  superheater   convection sections.  It is In the area of these pendant
sections that  flue gas  temperatures at  full load on gas fuel reach levels of
interest  for SNCR DeNOx.   Full load  superheater outlet  conditions are
1,560,000 Ib/hr  at 2450 psig and 1050 F.  Flue gas from  the furnace  passes
horizontally through  the secondary superheater, a water screen formed  by the
rear wall tubes of the furnace, the reheater, and  the pendant loop portion of
the primary superheater.  Following  the  rear cavity, the  flue gas then  passes
vertically  downward  through  the  balance of the convection sections, air
preheater and stack.  Flue gas recirculatlon fans are provided for  accurate
control of superheated steam temperatures.  At  full load on gas fuel, about 8%
of the flue gas is recirculated to the furnace bottom hopper.  A side sectional
elevation  of the boiler is shown  in Figure 2.  The furnace cross section  In the
vertical  upflow direction is 24 ft  wide oy 50 ft. deep.

Detailed  description of the boiler convection sections goes beyond the scope of
this report. It is sufficient to say that the flue gas velocities at full load on
gas fuel are such that the residence times in the 1st and  2nd cavities between
convection sections are on the order of 40 milliseconds (msec), and  that flue
gas temperatures initially decrease  at a  rate  of  about 4 F/msec in the first
pendant section, and  then at a rate of about 2 F/msec In  the second and third
sections.  These narrow cavities  and very short residence times are typical for
many gas- and  oil-fired boilers,  and offer perhaps the most  difficult type of
challenge  for  application  of  boiler  injection SNCR DeNOx.  An  earlier
publication by  Mittelbach, et.al. indicates that at 1800  F or  above, flue gas
residence times of about 100 msec would  be sufficient to  complete most of the
SNCR DeNOx reactions (Reference 3).  In the  case of the Huntlngton  Beach
Unit 2 boiler, this expectation was overly  optimistic.
                                   7A-4

-------
Figure 2:  Side Sectional Elevation, Huntlngton Beach Unit 2 Boiler
               Southern California Edison Company
                          StCONOARYl UREHEAT
                          SUPERHEATER SUPERHEATER
                                 7A-5

-------
3.0   Flue Gas Temperatures

Prior  to design of the injection system, flue gas temperature data was obtained
using HVT probes at the upper furnace front  and side-wall observation doors,
and oy acoustic pyrometer to obtain average flue gas temperature at the Inlet
of the first pendant  tube section.  The various field  measurements of flue gas
temperatures were compared with boiler manufacturer design  data  as  follows:

         Table 3.1   COMPARISON OF FLUE  GAS TEMPERATURES
              Huntlngton  Beach Unit 2 at Full Load   (Gas Fuel)

      Source of Data              SSH Inlet     1st Cavity    2nd Cavity

HVT Probe @  Observation  Doors    2230 F        n/a           n/a
Acoustic Pyrometer @ Obs. Doors   2280 F        n/a           n/a
HVT Probe @ Manway Doors       n/a           1910 F (?)     1760 F
Manufacturer  Design Sheets       2340 F        n/a           1775 F

The field data seemed to be in reasonable agreement with boiler manufacturer
data.   Computer-generated  prediction of 2nd cavity temperature contours (full
load  on gas fuel) were also  provided by  the boiler manufacturer,  which
indicated cooler zones averaging 1700-1800 F  near the 2nd cavity  floor, hotter
zones of about 1850-1950  F in the middle, and then 1800 F or above nearly all
the way to the 2nd cavity roof.  Based on the foregoing, there was no reason
to doubt that  the 2nd cavity was the preferred Injection zone.  The 2nd cavity
measures approximately 16 ft. high by 50 ft. wide  In cross-section.

Following Installation of the  2nd cavity Injection nozzles,  further data was
obtained.  Temperature profiles from HVT measurements In the 2nd cavity are
provided In Figures  3 and  4, where  the strong Influence of burner  patterns
under otherwise  Identical operating  conditions Is easily seen.  Burner pattern
adjustment caused average  flue  gas  temperatures to  Increase (or decrease)  up
to 100-150 F.   The entire SNCR DeNOx temperature window Is only 300 F, and
changes of 100-150  F are quite significant  As seen In Figures 3 and 4,  flue
gas temperatures also decreased up  to 200 F from the floor  to the roof. This
adverse temperature gradient substantially shortened the 2nd  cavity Injection
residence times within the 1600-1900 F SNCR DeNOx temperature  window.


4.0   Description of  2nd Cavity  Injection System

The Initial full load NOx concentration was generally about 120 ppm (corr. 3%
O2, dry).  Except as  noted, this Initial NOx was used  for NSR calculations.

Tube  shields  were designed and Installed by Noell  on the first row of boiler
tube at the downstream edge of the 2nd cavity.  Discussion between Southern
California Edison and Noell confirmed that tube shields  would provide a way
to evaluate effects of high velocity Injection Jets on metal thicknesses, without
any metal loss on the boiler  tubes themselves.  In  coal-fired applications  of
high energy boiler Injection  for SNCR DeNOx,  Noell generally recommends the
use of tube  shields so that  the potential for Increased erosion In specific high
velocity zones  can be determined without risk to the boiler tubes themselves.
                                   7A-6

-------
Figure 3:  2nd Cavity Flue Gas Temperatures Near Boiler Centerline
       Huntlngton Beach Unit 2 Boiler. Full Load, Gas Fuel
               Southern California Edison Company
HVT Temperature (F)
























































r"
\
A-









S**

\









>

\\
\


Ni






\
\
^
fv
\
rma







\
\.
oo
V

BO
V,







r^~
- 

DS
X






AI

,-

/







^


J*
S







^\

\
r~s









\
A
\\
\\
\










\
\
\\










i
i

                              4     t     a    10     12    14    16

                               Height Above Floor of 2nd Cavity, ft
HVT Temperature (Fl





























































*-
tf




N





^

N



>TE




/0(
Q
-
PII




S-E
,,oJ
n
mo




jt-
U
M
BOOS-We
i -v

B0(

. i
S-Taat

N
Tali




^v
'et
^
7

,

en a





X
/
/
/
/

,r





^
'S




, fro





~N
\
*v^



n Si





k
^
\
\


ie







^
\\
\
Jli










1
                    0    J     4     6         10     IS

                               Height Above Floor of 2nd Cavity, ft
         Figure 4:  2nd Cavity Flue Gas Temperatures Near Boiler Walls
               Huntlngton Beach Unit 2 Boiler. Full Load, Gas Fuel
                      Southern California Edison Company
                               7A-7

-------
5.0   Results of 2nd Cavity Injection Tests

System tests Involved selection of pump settings for controlling the urea-water
mixture ratio.  The liquid mixture was  then pumped  to the boiler  level and
injected Into the cross-flowing flue gas using air-driven nozzles  operating at
sonic Jet velocities.  A number of higli velocity  Injection nozzles were installed
in the floor zone of the 2nd cavity.  By means of aspirated ports, these nozzles
could be extended or retracted up to 8 ft. into the pressurized flue  gas  zone,
without influencing boiler operations.  Two (2) air orifice sizes were tested, the
larger orifice(s)  requiring  an  injection air flowrate of  about 2.1% of the full
load flue gas flowrate, and the smaller orifice(s) requiring about 1.2%.

Figure 5  shows  the effect of boiler  load and burner  pattern on percentage
DeNOx for 2nd cavity  injection at  NSR  = 2 for the two (2) sizes of Injection
nozzles.   As can be seen, the effect of  increasing boiler  load with  ABIS (all
burners  In  service) was to increase the  DeNOx performance.   With normal
BOOS  (burners out of service), increasing boiler load at a constant  urea
feedrate  for NSR = 2 at full load caused a decrease in DeNOx  performance.
With  the  smaller nozzles,  reduced DeNOx performance especially at full load
was partially caused by reduced flue gas mixing at  higher flue gas velocities.

Figure 5  Illustrates the effect of adjusting  the burner pattern  from normal
BOOS to  ABIS,  which causes increased  flue gas temperatures (Figure 3  & 4).
The increased flue gas  temperatures,  in  turn, caused  a full  load DeNOx
performance Increase from 27% to 40%. Since the change In burner pattern
caused 2nd cavity  flue gas temperature changes of 100-150 F, and  since the
resulting DeNOx Increase (at otherwise identical conditions) was  relatively
large, it was concluded that SNCR DeNOx in the 2nd  cavity at full  load was
operating at the  colder edge of the  1600-1900 F temperature window.   The
injected  urea behaved  as if the isothermal temperature was about 1600 F,
regardless that full load HVT temperatures  in the 2nd cavity itself  averaged
1720-1780 F.  These Initial full load  results  up to 40% DeNOx were  achieved
despite adverse conditions of short cavity residence time (i.e. 40  milliseconds)
and 2nd  cavity floor-to-roof adverse temperature gradient  (i.e. about 200  F).
Despite moderate DeNOx  levels which were achieved, such  adverse conditions
in the 2nd cavity caused unacceptably high levels of NHs slip.

Further analysis of these  initial test program results showed  that the hotter
1st cavity or upper furnace zones offered better locations at full load for high
energy SNCR DeNOx Injection than the 2nd cavity.

6.0   Tests of 2nd Cavity Injection Nozzle Supply Pressure

Additional tests were conducted using the larger 2nd cavity nozzles.  In  these
tests, the  boiler was held at full load, and urea NSR feedrate was  increased to
determine DeNOx vs.  NSR.  The results are presented  In Figure 6, where it is
seen that with a lower nozzle pressure, the DeNOx cannot be Increased beyond
about 20% regardless how much the chemical feedrate  Is Increased.  This type
of response curve Is  Indicative  of relatively  poor flue  gas  mixing, where the
SNCR DeNOx process become mixing limited.  At the higher nozzle  pressure,
there  Is a continuing Increase in DeNOx  performance up to  about  37% as NSR
is  increased up to about 5.  This second type of response curve is  Indicative of
relatively  good flue gas  mixing.
                                   7A-8

-------
Figure 5:  Effect of Boiler Operations on 2nd Cavity Injection DeNOx
             Huntington Beach Unit 2 Boiler, Gas Fuel
               Southern California Edison Company
      Percent NOx Removal
lb-
10 
15
JO
J5


5

0


NC

















TE:








	 ..
















 -,
















;  -~
















1  ~-
















 	
Nc
Full
Si












^


~~ 
rmaJ
Load
nail











j
 _


 ~-.
sod
NSR
^ozzl






Fixed Urea Feedrates for Normal
NSR = 2 at Full Load (215 Mtt





>-"-"
  _


~~ 
)S
= 2
es






BOO
r)


BIS (
Largi
^

~.


~~-~







5 Tes


SNS
; Noz
^

 


~~~~.







ts


R = :
iles


 



i^-












Nc
Full
Li
-i


  .










^

rma

-------
7.0   Results of 1st Cavity Injection Tests

Additional  tests were performed to investigate  1st cavity injection at higher
full load temperatures,  which averaged about 2050 F in the  1st cavity.  This
was  several hundred degrees Fahrenheit hotter  than the average  full load
temperature  in the 2nd cavity.   The existing 1st cavity sootblowers were
removed and air-driven  nozzles were  installed Into these existing membrane
wall  aspirated ports.  Using  four (4) sidewall nozzles with known limitations
in flue gas  cross-sectional  coverage, 20%  to 25%  full load  DeNOx was
obtained with urea feedrates from NSR = 2 to NSR = 4  (Figure 7).  For these
same urea NSR  feedrates and  operating conditions. NH3 slip as measured
upstream from  the air preheater was well below  1.5% of the NHi injection rate,
and  averaged less than 3 ppm.  Despite the very high 2050 F temperature, the
SNCR DeNOx process operated beyond expectations, especially considering the
relatively poor flue  gas cross-sectional coverage  and mixing afforded  when
using only  four (4) sidewall nozzles.

8.0   Results of Upper Furnace Injection Tests

Further tests were also performed to determine upper  furnace  injection DeNOx
as a function of boiler load.  Again, only four (4)  sidewall nozzles were used
where  existing observation doors  (aspirated) were available.  The chemical
feedrate during  these  tests  was  maintained  at a constant value  which
provided NSR =  2 at  full load  conditions.  As  shown in Figure 8,  the
percentage  DeNOx decreased  from a maximum of about 40% at a reduced load
of 120 Mw. At full load on gas fuel, the flue  gas temperatures are about 2300
F at  the inlet of the first boiler tube bank. This is too hot for SNCR DeNOx,
and  as  shown  in  Figure 8, the DeNOx decreased down to about 5%  at  full
load.  NH3  slip characteristics are also shown  in Figure 8, where it is seen that
at about 145 MW or 150 MW boiler load,  upper  furnace  flue gas temperatures
are most favorable for optimum SNCR DeNOx  performance.

9.0   Further Work In Progress

Noell is proceeding with  further  development  of advanced injection systems to
be considered for installation and additional  testing at Huntington Beach.
                                  7 A-10

-------
Figure 6: Effect of Nozzle Pressure on 2nd Cavity Injection DeNOx
       Huntington Beach Unit 2 Boiler, Full Load. Gas Fuel
              Southern California Edison Company
           Percent DeNOx from InlUaJ NOx .110 ppm
               215 MW, Gas Fuel
           40 
                                          13
                                           P'g

           33
   AK)ii (Off*.
           30
           55
           20
           10
                              2        3

                       Normalized Stolchlometrlc Ratio (NSR)
                                                        to
         Figure 7:  1st Cavity Sidewall Injection DeNOx vs. NSR
           Huntington Beach Unit 2 Boiler, Full Load, Gas Fuel
                  Southern California Edison Company
                              7A-11

-------
                                                                    ro

                                                                    CO
  FVNAc  Uflf*
4 435. ZxfcKS,  6/fS FUEL
     PI
  :r3
  ^r?
r n C o
o 0 r*
                                                                 o
                                                               p (T)  f1
                                                               Cr P  o
                                                               K.O  p
                                                               o crp.
                                                               p
                                                               -^   Tl
                                                                 DO rt
                                                               o -   "i

                                                               Bog
                                                              TJ P  K

                                                               EJ K  
                                                               P >T1 _
                                                                    D
                                                                    fD
                                                                    2

                                                                    O
                                                                    X

-------
Results and Conclusions

1.     Narrow  cavities and very short residence  times In many  gas- and oil-
      fired boilers offer perhaps the most difficult challenges for application of
      boiler Injection SNCR DeNOx.

2.     Flue gas temperature variations caused by  normal  boiler operations can
      and  will have  significant effects on  boiler injection  SNCR DeNOx,
      even when there are no changes in boiler steam production or load.
      Successful load-following SNCR  DeNOx systems  must  have multiple
      injection zones and relatively sophisticated controls.

3.     Detailed field temperature measurements and flow model  optimization
      tests of  injection Jets are considered prerequisites for the design of high
      performance (boiler-specific) SNCR DeNOx injection systems

4.     Despite  adverse time/temperature conditions in narrow cavities between
      adjacent convection sections in  the Huntington  Beach gas-fired boiler,
      full load DeNOx performance was obtained  as follows:

      Injection Zone     Nozzle Posltlon(s).        DeNOx      NH3 Slip

      2nd Cavity        Multiple  Floor Nozzles    25%-40%    high
      1st Cavity         Sldewall Nozzles (4)       20%-25%    low <  3 ppm
      Upper Furnace     Sldewall Nozzles (4)       0%-5%      zero

5.     This initial Huntington Beach  test  program has shown that SNCR
      DeNOx  is a function  of available DeNOx reaction time plus injection
      system  cross-sectional coverage and mixing. In this application at full
      load with short residence times,  injection  into the 1st cavity at  a flue
      gas temperature  of about 2050  F  appears to provide  the best SNCR
      DeNOx results.

6.     Noell  is  proceeding with further development  of  advanced injection
      systems  to  be  considered for installation and additional testing at
      Huntington Beach.


References

1.    Jones,  D.G.,  et.  al.,  'Two-Stage  DeNOx Process  Test Data  from
            Switzerland's Largest Incineration Plant', EPA/EPRI Symposium
            on Stationary Combustion NOx Control, San  Francisco, California,
            March 6-9. 1989.

2.    Negrea,  S., et. al., 'Urea Injection NOx  Removal on a  325 MW Brown
            Coal-Fired Electric Utility  Boiler in West Germany', 52nd Annual
            Meeting,  American Power Conference, Hyatt Regency Chicago,
            April 23-15, 1990.

3.    Mlttelbach,  G., et.  al.,  'Application of the SNCR Process to  Cyclone
            Firing', Special Meeting on NOx Emissions Reduction of the VGB,
            German Power Industry Association, June 11-12, 1986.
                                   7A-13

-------
EVALUATION OF THE ADA CONTINUOUS AMMONIA SLIP MONITOR

     Michael D. Durham, Richard J. Schlager, Mark R. Burkhardt,
             Francis J. Sagan and Gary L Anderson

                    ADA Technologies, Inc.
               304 Inverness Way South, Suite 110
                    Englewood, CO 80112

-------
          EVALUATION OF THE ADA CONTINUOUS AMMONIA SLIP MONITOR

              Michael D. Durham, Richard J. Schlager, Mark R. Burkhardt,
                       Francis J. Sagan and Gary L. Anderson

                              ADA Technologies, Inc.
                         304 Inverness Way South, Suite 110
                              Englewood, CO 80112


ADA Technologies,  Inc.  has  developed a continuous emissions  monitor  for use  with
advanced NOX control technologies that is capable of simultaneously monitoring ppm levels
of NH3 and NO in flue gas. The instrument can also measure SO2 when it is present in the
flue gas.  The  instrument is based on ultraviolet light absorption using a photodiode array
spectrometer.  It has  unique  advantages over other  ammonia instruments as it directly
measures  ammonia  as opposed to the indirect chemiluminescent techniques which must
infer the NH3 concentration from the difference  between two large numbers. The monitor
has undergone extensive laboratory  and field  evaluation  and data  are presented which
demonstrate sensitivity, accuracy and drift of the instrument. The analyzer has been field
tested at a gas  turbine with SCR, a coal-fired circulating fluidized bed with ammonia injection,
a refinery  boiler with SNR, and  a  utility boiler  with urea injection.   The accuracy of the
instrument was determined by comparison with extractive wet chemical measurements.
                                     7A-17

-------
                                I. INTRODUCTION
ADA Technologies, Inc. has developed a continuous, real-time analyzer for measuring part-
per-million levels of ammonia  (NH3) and nitric  oxide  (NO) in  flue gas  associated with
advanced NOX reduction systems. A two-year long development program sponsored by the
U.S. Department of Energy resulted in an analyzer that is specific to ammonia, reliable, and
accurate.  Other common flue gas  components do not interfere with the measurement of
NH3.


This instrument fills the need created by advanced NOX control technologies for an ammonia
slip monitor which can be used as part of the process control system. Ammonia is a primary
ingredient in virtually all of the advanced NOX control  processes such as selective catalytic
reduction (SCR) and  selective non-catalytic reduction (SNR)  technologies.   However,
because of severe problems  related to the penetration of unreacted NH3 through the flue
gas treatment system,  it is extremely important to measure and control the downstream
concentrations of NH3.


The instrument is an effective diagnostic tool for optimizing De-NOx systems, and will be a
valuable component of NOX control equipment in many applications including: coal-, oil- and
gas-fired utility boilers,  co-generation plants, refineries, municipal solid waste incinerators,
and research programs.


The monitor has been operated as both an in-situ and extractive instrument. The extractive
mode of operation allows a testing team to evaluate the stratification of NH3 gas across the
diameter of a duct.  This capability is particularly important in evaluating  whether ammonia is
dispersed uniformly within the flue gas of a SCR or SNR De-NOx system.
                           II. MEASUREMENT PRINCIPLE
A. MEASUREMENT PRINCIPLE


Ammonia and NO absorb light in the ultra violet (UV) range at specific wavelengths, and the
shape of the absorption spectra are characteristic of the identity of the particular gas. Figure
1 shows absorption spectra for NH3 and NO in a selected UV wavelength region.  In this
spectral range,  NO absorbs at two characteristic wavelengths, and  NH3 absorbs at four
characteristic wavelengths. The two large doublet peaks identify the absorption due to NO,
and the four smaller peaks, which include two characteristic doublets, are due to NH3. The
quantity of light absorbed by a  gas is proportional  to  its concentration, as  defined by
principles of Beer's Law. Since the NO doublet located near diode 400 overlaps with one of
the ammonia peaks, this region cannot be used for  analysis.  However, the NO peak at
diode 850 and the ammonia peaks at either diode 200 or  diode 600 do not interfere  and
therefore can be selected and analyzed to determine the concentrations of these two gases.
The data available from the multichannel  spectrometer allow measurement of these  two
gases directly and accurately.
                                      7A-18

-------
       3.20
                        AMMONIA AND  NITRIC  0X11113 SIM3CTHA
       3.10 -
       3.00 -\
        2.90 -1
      CJ^

      s- 2.80 -

      (si
       2.70 H
      H

      ,_ 2.60 -
      re
      o
        2.50 -


        2.40 :


        2.30 -


        2.20 -
                    10  ppm NH3
                10  ppm Mi-la  and  200 ppm  NO
               ' ' i|iii  | i i  i |  iii| i  > i|JTi |  i i  i |  i ii|iirniiiir
             0   100   200   300  400   500   600   700  BOO   900   1000
                                  PHOTODIODE ELEMENT  #

Figure 1.   Absorbance spectra for ammonia and ammonia/nitric oxide mixture.


B. DETECTION SYSTEM


Photodiode  array  detectors  provide  a  technology to  improve  upon  the design  of
conventional scanning monochromator-based spectrometers.  The improvement involves
the placement of a series of detectors across the focal plane of a polychromator, each with
its associated readout  electronics.  The  most advanced  of  these systems use a  linear
photodiode array (LPDA) detector.  The LPDA is a large-scale integrated circuit fabricated on
a single monolithic silicon crystal. It consists of an array of diodes, or pixels, each acting as
a light-to-charge transducer and a storage device. These detectors are ideally suited for use
in UV spectrometers because they have a large quantum efficiency (40-80%) throughout the
range as well  as geometric, radiometric, and electronic stability.  The array itself can be
mounted and operated so as to be tolerant  of high  temperature, humidity, vibration, and
electrical and magnetic fields.

An LPDA  spectrometer  system, shown schematically in Figure 2,  operates by passing  a
continuous light source through the sample and into the polychromator.  The polychromator
disperses  the  light across the LPDA,  which has replaced the exit slit of a  conventional
spectrometer.  The array is located in the focal plane of the polychromator so that each
diode corresponds to a particular wavelength  resolution of the  UV-VIS spectrum. The diode
array provides an almost ideal sensor for the digital  acquisition of spectra, as the array itself,
by its presence in the focal plane  of the spectrometer, digitizes the spectrum  into discrete
intervals.   Unlike the scanning spectrometers, whose wavelength accuracy is mechanically
limited, the LPDA spectrometer is limited only by geometric constraints  of the detector itself,
                                       7A-19

-------
                                      Deuterlun
                                      Lanp
                                         Gas Cell
                                    Gas
                                        Outlet [~ |
                                                    Gas Inlet
^Quartz Pr,sn^ p^
w> Entrance Silt
Colllnatlng Mirror
Peltier Cooler
T
^L-^

I


Xlrror

ife^l
Thernocouple
ontrollor

Figure 2. Schematic diagram of Linear Photodiode Array spectrometer system.


by vibration and thermal expansion of the optical components,  and by the stability of the
source.  Wavelength accuracy is equivalent to the diode spacing multiplied by the linear
dispersion  of the spectrograph.  Its geometric registration  and, therefore,  its wavelength
accuracy and precision, are greater than any mechanically scanned spectrometer


With the PDA detector it is possible to develop algorithms which use the unique structure of
the absorbance spectrum to quantify the concentration of the gas. This approach eliminates
the need  to maintain  the initial  intensity (IJ reference  and simplifies and  speeds  the
calculation.   Since the analysis  procedure searches  for  characteristic features  of  the
absorption  spectrum rather than  a fixed wavelength,  it is less sensitive to drift or lamp
intensity fluctuations.


The  photodiode  array  detector  has unique  advantages  over all the  other  ammonia
instruments. It provides a direct measurement of ammonia and is, therefore, inherently more
sensitive than the indirect chemiluminescent measurement techniques which must infer the
NH3 concentration from the  difference between two  large numbers.   In addition,  the
photodiode array spectrometer has the following unique features.

       The instrument can  be  built with no  moving  parts  which  will  reduce
        maintenance and increase reliability in an industrial environment.

       The software is written to provide built-in checks for alignment of the optics.
       Changes in light intensity to do create a drift problem.

       Finally, the interferences  are well known  and can be accurately handled by
        the PDA  detector.
                                       7A-20

-------
                          III. LABORATORY EVALUATIONS
A. TEST SET-UP


Performance parameters of the  analyzer were determined in a series of laboratory tests.
Gases used in the evaluation were supplied in cylinders containing the individual gases in a
background of nitrogen gas.   The concentrations of the gases  were certified by the
manufacturer  through  analysis.   Gases were  mixed  in  various  combinations  and
concentrations using mass flow  controllers and manifold system.  The gas flow was then
metered into the analyzer for evaluating performance.  Tests were conducted using a gas
cell with a path length of 90-cm. The cell was heated to maintain an internal gas temperature
of 300 F.  Results of the evaluation follow.


B. LINEARITY OF NH3, NO, AND S02


The linearity of the response of the analyzer was evaluated by initially calibrating the analyzer
using nitrogen and a span gas for each component of interest. Gas concentrations were
then decreased in steps and resulting analyzer measurements noted.  Results of the linearity
evaluation for NH3, NO, and SO2  are shown in Figures 3 through 6.


Ammonia results are shown for two ranges of measurement, 0 to 70 ppm and 0 to 10 ppm.
Figure 3 shows that when calibrated at 70 ppm, measured concentrations are within  1 ppm
of the input concentration.  For the low range, Figure 4 shows that measured concentrations
are within 0.5 ppm of the input concentration.


Prior  to  measuring the  linearity of the NO,  the  instrument was  calibrated using  two
concentrations of NO because the absorbance of NO requires a second order equation to fit
the calibration curve. Using this technique, the linearity of the instrument is within 2% of the
actual concentration over a concentration range of 0 to 200 ppm as shown in Figure 5.  If
only a single gas is used for calibration, there is a maximum 10% deviation from linearity in
the middle of the range.


Figure 6 shows the linearity of  the analyzer for S02 calibrated at 80 ppm.  For all  gas
concentrations, the  measured values  are within 1  ppm of the input concentrations.  The
capability to  accurately  measure sulfur dioxide  provides the basis for  eliminating its
absorbance as an interference to the measurement of NO and NH3.


C. LONG-TERM NOISE AND DRIFT


Analyzer noise and drift were estimated by observing instrument readings over a 36 hour
period of time  as a mixed  gas stream of fixed  composition was  passed  through the
measurement cell.  Analyzer measurements for NH3, NO, and SO2 are shown in Figures 7.
The composition of the gas stream was 10 ppm NH3, 55 ppm NO, and 80 ppm SO2.
                                      7A-21

-------
             100-


           p. 90-
           o.
              80-


              70-
           g  60H
           o
           g  50H
           u

            "40-
            OJ
            u
            3  20-
            n
            m
                     10
                                   i
                                    40
              I
               50
20   30    40    50    60    70    80
 Input NH3 Concentration (ppm)
i
 90
                                                                 100
Figures.  Linearity of NH3 measurements when analyzer is calibrated using 70 ppm
          standard gas.
           d
           o
           d
           u
              8-
              4-
           55  2H
           cd
           QJ
           a
                          Input  NHS Concentration (ppm)
Figure 4.  Linearity  of NH3 measurement  when  analyzer is spanned using  10 ppm
         calibration gas.
                                      7A-22

-------
             250
                         40    60    80   100   120   140   160

                             Input NO Concentration (ppm)
                                                           180   200
Figure 5.  Linearity of analyzer to NO input concentrations when calibrated using two span

          gas concentrations.
            100
           a  8CH
           o


          "d


           d  60
           0)
           a
           d
           o
          U

           N  40-
          a
          tu
          in
          id
          0)
          a
             20-
                                   I
                                    40
i
 60
20        40         60         80

 Input  SOZ  Concentration  (ppm)
                                                                  100
Figure 6.  Linearity of SO2 measurements when analyzer calibrated using 80 ppm span gas.
                                        7A-23

-------
au-i 	
'"IT
a
0,70-4
>s
fl6o4
.2 * 
A * 	 ft - -- ft 	 	 	 6- 	 	 ft 	
	 ^ 	 	 $ - v ^ v w
00000 BO ppm S0t Input




_ e 	 (5 	 a 	 e  - 	 Q 	 e 	
          a
          40-
          a
          o
          <-> 30-
          tn
          cd
          O 20-
            10-.'
                               OOOOO 55 ppm NO Input
                                     10 ppm NH3 Input
                    iiiiiiiiiiiiiiiiiiiii
                     4      8     12    16     20    24
                              Measurement Period  (hr)
 T"1
28
32
      36
  Figure 7.   Noise and drift characteristics  of NH3,  NO, and SO2 measurements over 36-
            hours.

 As can be seen from Figure 7, with unattended operation, the output is extremely stable for
 all three oases  Analyzer noise is defined as the short-term peak to peak signal variation,
 and is equal to'jt 0.3 ppm for NH3, 0.15 ppm for NO, and _ 0.1 ppm for SO2.  Analyzer
 drift is defined as the long-term  variation in  analyzer signal around an average value.
 Analysis of the measurements shows that the drift is  0.3 ppm for NH3, and jf 0.3 ppm for
 NO drift, and is _ 0.4 ppm for SO2.  These noise and drift measurements are well within the
 accuracy capabilities of the gas flow delivery system using the mass flow controllers.


 D.  RESPONSE TIME


The response time of the analyzer is a function of how quickly a sample of gas is delivered to
the light path and the time it takes to process the spectral information into gas concentration
units. Since the data processing time is very short, on the order of a few seconds, the rate of
response becomes directly related to the volume of the gas cell and the flow rate of the gas
through that cell.  For example, 90% of full scale response is achieved  to a  known NO
calibration gas input within five equivalent volume changes of the cell. This rate of gas flow
through the sample system is typically done within 1-minute. The response to ammonia gas
is slightly slower than observed for NO, due to the nature of ammonia gas which requires
conditioning of tubing surfaces during its travel to the measuring cell.
                                        7A-24

-------
E.    MINIMUM DETECTION LEVELS


The minimum detection level for a particular gas is defined as twice the noise value. Based
on data shown in Figure 7, the minimum detectable level using a 0.9 meter log gas cell is 0.6
ppm for NH3 and 0.3 ppm for NO.


The  minimum detectable level and maximum concentration measurable using absorption
spectroscopy are a function of the path length that the light travels through a gas sample.
Higher gas concentrations can be measured using a shorter path length, but minimum
detection levels increase in proportion.  In  actual practice, gas measuring cells lengths are
specified based on the particular application and accuracy requirements.


F. INTERFERENCES


Several gases that  are typically found in flue gas absorb light in the lower UV region and
present a potential for  interfering with  the  measurement of  NH3  and  NO.   However,
experiments were conducted which demonstrated that at typical flue gas concentrations,
NO2, CO,  CO2, O2> and H2O did not  interfere with the measurement of NO and NH3.  The
most predominant  interference is SO2 which, depending upon the concentration, can  be
accounted for using spectral  subtraction which has  been described previously (Durham et
al., 1989).  The  maximum  SO2 concentration that can be accurately subtracted from the
absorbance spectrum depends upon  the length of the gas cell.  For example in a 0.9 meter
cell,  the maximum concentration  of SO2 is  80 ppm.  If the cell is reduced to 4 cm, then the
maximum  concentration increases to 1800 ppm SO2.  However, with the smaller cell the
minimal detection limit for NH3 is increased to 13 ppm.  Therefore, a gas cell needs to  be
selected for the specific application.
                             IV.  FIELD EVALUATIONS
A.    GAS TURBINE WITH SCR


The ADA Analyzer was used to evaluate the De-NOx system of a gas-fired co-generation
facility.  At this site, the Analyzer was evaluated  as both an in-situ  and an extractive
instrument. The in-situ instrument avoids sample biasing and minimizes the operating and
maintenance  requirements.   The extractive  version  is  designed  for  traversing  ducts
downstream of the NOX control system to optimize the ammonia injection configuration.


At this site, ammonia is injected upstream of a selective catalytic reduction (SCR) bed to
control the NOX emissions.  The plant did not have an ammonia detector but did monitor the
concentration of NOX at the inlet and outlet of the SCR and measured the quantity of
that was injected.  The target NOx emission from the facility was 18 ppm.
                                       7A-25

-------
Verification of the Accuracy of the Instrument


The  measurement accuracy of the analyzer was  determined by comparing instrument
emission measurements against a standard wet chemical technique. This manual technique
involves extracting a sample of the flue gas from the  stack and bubbling it through an acidic
solution which collects the ammonia.  The solution is then analyzed in a laboratory using a
selective ion electrode to determine the quantity of NH3 collected. Although this technique is
very manpower intensive, accurate measurements can be obtained if the procedures are
followed carefully. An experienced third party testing firm was contracted to perform the wet
chemical measurements.


Several wet chemical tests were conducted while the analyzer continuously measured NH3
concentrations.  The analyzer was used in-situ,  while wet chemical tests were conducted
from a different,  neighboring port on the duct.  In spite of the fact that the measurements
were made at different points in the stack, there is excellent agreement between the  two
methods. Figure 8 shows a comparison of the ammonia concentrations measured  by the
continuous analyzer and the manual method. The straight line represents a 1:1 correlation.
The  numbers inside the data points are the ports where the extractive measurements were
made.   The ADA instrument was operated at a  port midway between the two orthogonal
ports 1 and 4.  The  different ammonia levels in  the stack  were achieved when the facility
operator manually adjusted the ammonia injection rate. This data demonstrates that the
instrument is capable of accurately measuring the concentration of ammonia in a flue gas
stream.
              25-
             in
             I
              20-
               15-
             E
             a.
             a.
               10-
             5 5H
             o
             o
             o
                             Numbers Represent Extractive Sampling Ports
                                                  
Sample Port Configuration

    3 /   ^~\ 2
        ADA
                         5        10       15       20       25      30
                      NH3 CONCENTRATION (ppm)  BY WET CHEMICAL ANALYSIS
Figure 8.  Comparison of NH3 measurements using the ADA In-Situ monitor and extractive
          wet chemical analysis at a co-generation facility.
                                      7A-26

-------
Continuous Operation


The instrument was operated on a 24-hour per day basis during the test week. Algorithms
were written to eliminate any detrimental effects due to fouling of the lenses or mirror. During
the operation of the instrument some fouling of the mirror did occur due to the deterioration
of the purge blower.  This resulted in a reduced  magnitude of light detected by the
photodiode array. However, the algorithms operated as designed to account for loss of light
level, and the fouling had no effects on the measurements of NHg and NO concentrations.


Figure 9a shows a plot of the data obtained during a 24-hour period. The trends in the NH3
and NO measurements show a gradual decline in the NO concentration while the ammonia
slip is increasing. Whenever a sharp change in NO level occurs, there is a corresponding
change in the opposite direction for NH3.  The ammonia injection rate is plotted in Figure 9b.
As can be  seen there is a strong correlation between the ammonia injection rate and the
ammonia slip.  This data indicates the variability that occurs in even a stable combustion
system such as the gas turbine combustor.


Evaluation of the SCR System


The data obtained during the  continuous in-situ measurements were reduced to determine
the relationship between the NO  level and the NH3 slip. These data, which are plotted  in
Figure 10, provide very valuable information relative to the performance of an SCR system.  It
can be seen that for higher concentrations of NO there is very little slip and the amount of slip
increases as  the NO is reduced.  However, at some  point any further decrease in NO can
only be achieved with a significant increase in ammonia slip.


This data is extremely important relative to the cost-effective  operation of an SCR and the
resulting emissions. If the facility is operating  under a permit that specifies only a maximum
NO concentration, without considering the ammonia slip, the minimum level of emissions will
not be obtained.  In this example, in  order to obtain a 2 ppm reduction in NO from 19 to 17
ppm, the NH3 slip will increase by 20  ppm. It would be more desirable to operate at the knee
of this curve to minimize the total release of pollutants.


Operating at this point would  also make economic sense. At an ammonia slip level of 25
ppm, half the injected ammonia is  going up the stack unreacted. This means that the cost  of
the ammonia is double what  it would be if the system were  controlled with the slip as a
parameter.  This data also demonstrates the importance of a continuous ammonia slip
monitor.  Since  the performance of the catalyst in the SCR is going to change over time, the
continuous monitoring of  the flue gas can  be used to identify the  optimum  operating
conditions at all times.


Evaluation of the Extractive Analyzer


The analyzer was also used in an extractive mode in order to measure gas concentration
gradients in the system.  A probe was used to draw samples of flue gas from discreet points
across the diameter of the stack  and into  the analyzer.   Since there was  no access
immediately downstream of the catalyst, a traverse was made at the stack. The traverse was
                                       7A-27

-------
           25-
          -20-
                Concentrations of NHj and NO During Continuous Operation




                     r     NO Concentration
                             -NHj  Concentration
                                                                    -50
                                                                    -40;
                Tiiiiiiiiiiiiiiiiiii  r
             0  1  2  3  4  5  6  7  8  9  10111213141516171819 20 21
                                                                    60
                                                                       a.
                                                                       a.
                                                                       QL
                                                                       O


                                                                       6


                                                                   1-20

                                                                       O
                                                                   -10
                               OPERATING TIME  (hrs)
Figure 9a. Continuous NH3 and NO measurements from a co-generation facility.
                  MH3 Injection Rote During Continuous Operation
                              II  I   I  iIIi11I||Ir
             01  23456789   101112131415161718192021

                              OPERATING  TIME (hrs)
Figure 9b. Ammonia injection rates during emissions measurements.
                                         7A-28

-------
            25-
           E
           Q.
           CL
          v,

          Q_
           ,20-
15-
          5 10-
             5-
             0	1  i r  i i i iir|i r~r -T~;iirn|ii1-1 ] ~I~T~T i| "T i1 i i | i i i -i|- T -i i i | i i i
               0     5     10    15   20   25   30    35   40    45   50
                           OUTLET  NO  CONCENTRATION  (ppm)

 Figure 10. Nitric oxide emissions as a function of ammonia slip.


 made parallel to the ammonia injection grid. The results presented in Figure  11 show the
 presence of strong gradients in both NO  and  NH,, concentrations across the stack.  The
 higher levels of NO correspond with lower levels of NH3. Both the gradients and the inverse
 relationship  between  NO and  NH3 are due to  an improper balancing of the ammonia
 injection valves. This shows the usefulness of the extractive instrument in providing a means
 to optimize the ammonia injection system.


 B. COAL-FIRED FLUIDIZED BED WITH SNR


 The ADA Continuous Ammonia Analyzer was field tested at a 49.5-MW coal-fired circulating
 fluidized bed co-generation facility.  The plant injects ammonia into the primary cyclone for
 control of NO . The on-site CEM  system  incorporates a chemiluminescent instrument to
 beasure both NH3 and NOX levels using a thermal  converter for ammonia. Flue  gas samples
 are withdrawn from the center of the stack (approximately 100 feet above ground level) via a
 heated in-situ probe. The flue gas is pulled down approximately 100 feet of heated sample
 line to  an instrument enclosure.   Moisture is removed from the flue  gas sample before it
 entered the NO^ analyzer.  In the NH3 measurement mode,  a solenoid valve is activated
 periodically, forcing the flue gas through a thermal converter which converts the NHL to NO.
The signal generated from the flue gas that by-passes the thermal converter is subtracted
from the signal generated when the flue gas passes through the thermal converter to obtain
the NH3 concentration present in the sample.
                                      7A-29

-------
                                      N
            14 4

            136

            1L4
               31.5

               36.5

               37.3

               385










s

c

R

o







 From
Turbine
                                                   NH3

                                                Injection
                                                  Valves
Figure 11. Measured concentration gradients for NH3 and NO.


The field test program was performed to determine the accuracy of the ADA Continuous
Ammonia Analyzer for measuring NH3, SO2 and NO in a flue gas environment containing low
levels (5-40 ppm) of SO2.  As was done in the previous field study, the NH3 concentrations
measured by the ADA monitor were compared with those obtained using the standard
ammonia wet chemical technique performed by  a third party.  In addition, a comparison
between the ADA ammonia monitor and the chemiluminescent ammonia monitor determined
how well the two techniques agreed with each other  and with the standard wet chemical
method.


Simultaneous NH3 measurements were taken using  the wet chemical method, the ADA
ammonia monitor, and the chemiluminescent  ammonia  monitor.  The chemiluminescent
ammonia monitor took samples from the center of the stack through a heated sample probe.
The ADA ammonia monitor measured NH3 directly in the stack through a port positioned at a
90  angle from the  chemiluminescent  monitor  sample  probe.   The  wet  ammonia
measurements were performed  by positioning the wet ammonia sample probe adjacent to
the ADA in-situ probe.  This was done by placing the sample probe through the sample port
90 from the ADA monitor (180 from the chemiluminescent ammonia monitor) and then
bending the sample line to physically contact the ADA in-situ probe.


Figure 12 shows the comparison of the NH3 concentrations measured by the ADA ammonia
monitor, the chemiluminescent ammonia monitor, and the wet chemical ammonia method.
All data were corrected for 7.8%  moisture and 5% oxygen.  These conditions were measured
in the stack at the time of sampling.  The sample points  are averages taken over the wet
ammonia method sampling time.   Measurements of  different NH3 levels were attempted
                                      7A-30

-------
when the facility operators manually adjusted the ammonia injection rate.  However, the
vaporizers were not functioning properly at the time of the test, and the ammonia control
valves were opened fully.
As shown in Figure 12, the wet chemical and the ADA methods agree well.  This test also
shows the effectiveness of the ADA processing package in eliminating the interfering effects
of SO2 on the NH3 measurements. The chemiluminescent ammonia monitor response was
approximately 3-5 times higher than the standard wet chemical method.  This inaccurate
measurement of the ammonia slip could result in the injection of an insufficient quantity of
ammonia to react with NOX.
           a
            '5-
           d
           o
           S4-
           -U
           fl
           QJ
           O
           fl ,
           o 3"
           O
Chemiluminescent Indirect
                                 2         3
                                     Time  (Hours)
                        4
Figure 12. Ammonia slip measurements on a  coal-fired fluidized  bed boiler using three
          methods.
C. REFINERY BOILER WITH SNR
The ADA analyzer was used to measure NH3 and NO emissions from a thermal De-NOx
system used on a refinery boiler gas stream.  Ammonia gas was injected into the hot exhaust
gas from a furnace in order to effect the NOX reduction reaction. The gas stream contained
several hundred parts per million SO,.  Therefore, a gas measuring path length was chosen
to most effectively accommodate the 1lue gas SO2 content, while still providing the necessary
degree of accuracy for NH3 and NO measurements.
                                      7A-31

-------
Accuracy Determination


The analyzer was again used in both an  in-situ and extractive mode to gather data.  The
facility performed several wet chemical NH3 evaluations while the analyzer operated in-situ.
These results compared as follows:

                        Wet Chemistry            ADA Analyzer

                           51 ppm                 60 ppm
                           171 ppm                225 ppm
These results indicate good agreement between the methods, especially given the rapid
short-term changes in NH3 emission levels observed in the flue gas stream using the real-
time analyzer.


De-NOx System Evaluation


Ammonia slip and NO emissions data were collected as De-NOx system variables were
adjusted.  Figure  13 shows the relationship between NO emissions and NH3 slip measured
over a range of operating conditions.  Because of the proprietary nature of the information,
the data are plotted in relative concentration terms. This figure has a very similar shape as
the plot obtained from the SCR tests in that there is a point of diminishing returns relative to
the amount of ammonia injected.  This is  represented by the point where  only minimal
reduction in the concentration of NO  is obtained at the expense of significant increases in
ammonia slip .  Figure 14 shows the relationship between NH3 slip and NH3 injection rates.
Data such as these,  when collected  in combination with otner process information, can
produce  a significant data base for  use  in characterizing  a De-NOx system,  and  for
troubleshooting purposes.


The data  presented on the  De-NOx system evaluation were collected in only a few days of
testing.   These  results demonstrate the  ability  of  a  real-time analyzer for effectively
characterizing emissions from a full-size control system.


D.  UTILITY BOILER WITH UREA SNR


The final field test program  was conducted during  a demonstration of urea injection into a
utility boiler.  This program was  conducted during  October to December,  1990  and is
described in  the  paper by Abele  (1991) which  is presented  at the 1991  NOX Control
Symposium.  During this program, the instrument was successfully operated during the test
program.  The  calibration of the instrument  was checked at the beginning and end of  the
program.  After nearly two  months of operation, the  calibration constants had drifted less
than 2%.
                                      7A-32

-------
           co  6
           c
           u
           o
           O
          U
          O
                    1234567
                                 Ammonia Slip, Relative
                                                                10    11
Figure 13. Nitric oxide emissions as a function of ammonia slip at a refinery boiler.
              12
              10
          55  6
          .2
          o
          E  ,.
                        2        4        6        8        10
                             Ammonia Injection Rate, Relative
12
Figure 14. Relationship between ammonia slip and ammonia injection rate for refinery SNR
          system.
                                        7A-33

-------
                                   V. STATUS
ADA continues to provide testing services and analyzers for evaluations of De-NO^ systems.
ADA has been working toward commercialization of the analyzer technology with instrument
manufacturers.  ADA will be  participating in a round-robin performance  evaluation of
commercially available analyzers with regulatory agency involvement beginning in March.
ADA highly endorses such programs and will report results at upcoming meetings.
                                 VI. REFERENCES
Durham, M.D., T.G. Ebner, M.R.  Burkhardt,  and F.J.  Sagan (1989).  "Development of an
      Ammonia Slip Monitor for Process Control of NH~ Based NOX Control Technologies",
      presented at the AWMA International Specialty Conference on Continuous Emission
      Monitoring-Present and Future Applications, Chicago, IL November 12-15.


Abele, A. (1991).  "Performance of Urea NOx Reduction System on Utility Boilers", EPRI-EPA
      1991 Joint Symposium on Stationary  Combustion NOX Control, Washington D.C.,
      March 25-28
                                      7A-34

-------
ONTARIO HYDRO'S SONOX PROCESS FOR
  CONTROLLING ACID GAS EMISSIONS

        R. Mangal and M.S. Mozes
      Ontario Hydro Research Division
           800 Kipling Avenue
            Toronto, Ontario
            M8Z 5S4 Canada

                 and

       P.L. Feldman and K.S. Kumar
 R-C Environmental Services and Technologies
          US Highway 22 West
         Branchburg, New Jersey
              USA 08876

-------
                               ONTARIO HYDRO'S SONOX PROCESS FOR
                                 CONTROLLING ACID GAS EMISSIONS

                                         R. Manga! and M.S. Mozes
                                      Ontario Hydro Research Division
                                            800 Kipling Avenue
                                             Toronto, Ontario
                                             M8Z 5S4 Canada

                                                   and

                                       P.L. Feldman and  K.S. Kumar
                                R-C Environmental Services and Technologies
                                           US Highway 22 West
                                          Branchburg, New Jersey
                                              USA  08876
                                               ABSTRACT

 An in-furnace slurry injection process for the simultaneous control of SO, and NO, from power plant flue gases has been
 developed at Ontario Hydro's 640 MJ/h (0.6 x 10* BTU/h) Combustion Research Facility. The process known as SONOX
 involves the injection of an aqueous slurry of a calcium-based sorbent such as limestone, dolomite, hydrated lime, etc and
 a nitrogen-based additive into the furnace at temperatures ranging between 900 to 1350C. Coals varying  in sulphur
 content from 0.54 to 2.8% with NO, emission levels of 450-620 ppm were studied.  Operating parameters  have been
 optimized for maximum SO, and NO, capture. Under optimized operating conditions the  technique removes  up to 85%
 of the SO2 and effective NO, removal is 63-80%. The specific removal levels obtained depend upon the type  of sorbent
 and nitrogen-based additive, temperature, stoichiometry and coal. The effluent gas stream has been characterized for NH,,
 HCN and N2O.  The solid waste produced is composed of fly ash, CaSO4 and CaO which can be collected by the ESP.
 Due to the high dust loading that results from  the process, the ESP performance deteriorates somewhat.  A levelized cost
 estimate indicates that a SONOX system is about half the cost of a wet FGD system to own and operate. Negotiations
 are in progress to demonstrate this process on full scale boilers.


                                            INTRODUCTION

 In December 1985,  the Ontario  government announced a more stringent acid gas emission  policy limiting  Ontario
 industries in atmospheric emission of SO, and NO..  Ontario Hydro's limits were reduced to 430,000 tonnes/year starting
 in 1986 and to 215,000 tonnes/year starting in  1994.  This regulation is challenging in that Ontario Hydro must  stay below
 the regulated tonnage limit regardless of changes in the demand for energy or in other forms of generation. Although the
 regulation limits the amount of SO2 emissions, the level of NO, emissions is not specifically regulated and Ontario Hydro
 is  free  to trade between SO2 and NO, as long as the aggregate emissions of the two (SO2 and NOJ does not exceed
 215,000 tonnes/year and no more than  175,000 tonnes/year may be SO,(1,2).  Specific  NO, legislation is  now  being
 negotiated between the Federal and Provincial Ministers of the Environment

 Consequently, Ontario Hydro embarked on a program to curtail acid gas emissions from its coal burning plants. This
 program was initiated to meet the above mentioned regulations.

 Several options are being considered to reduce both SO, and  NO.. In the case of SO,, some options include:  sorbent
 injection processes, burning low sulphur coals with flue gas conditioning, wet flue gas desulphurization and the limestone
dual  alkali process.  Ontario Hydro is committed to two scrubbers being in operation at the beginning of 1994. For NO,
control, the options can  be classified as non-retrofit  and retrofit technologies.  Non-retrofit options would be to reduce
NO,  emissions by installing fossil replacement generation that has lower NO, emission rates than are currently generated
by existing stations and to reduce coal generation. Burning natural gas is an example. Retrofit options include: low NO,
burners, selective catalytic reduction and selective non-catalytic NO, reduction processes-(additive injection).
                                                 7A-37

-------
Of the options considered to meet the above regulations in-fumace sorbent injection and selective non-catalytic NO,
reduction processes were investigated extensively at Ontario Hydro's 640 MJ/h Combustion Research Facility.  As a result
Ontario Hydro's SONOX process which injects a calcium-based sorbent slurry and an additive to simultaneously abate
S02 and NO, was developed and is the subject of this paper.

The SONOX process is an in-fumace injection technique of an aqueous slurry of a calcium-based sorbent and a soluble
additive injected at temperatures ranging  between 900 and  1350C. The calcium-based sorbent reacts with SO2 and the
additive reacts with NO,.  The furnace  which serves as the chemical reactor provides  sufficient residence time and
favourable temperature for the reactions.  The following reactions represent globally, the SOj/NO, (SONOX) removal
paths:
                CaCO3  ->  CaO + CO2

                CaO  + SO2  +  1/2 O2  - >  CaSO4

                NO  +  Reagent (Additive) - >  N2 + H2O

The technique provides excellent distribution and mixing with the  flue gas for the above reactions to be efficiently
completed(3).  A schematic of the process is  shown in Figure la.  The process steps can be visualized as follows:

          Atomizauon of Ca sorbent and additive;
          Water  droplet evaporation;
           Particle disintegration for the Ca  sorbent and thermal cracking of the additive;
           Calcination of the Ca sorbent;
           Development of reactive sorbent  and  additive (CaO and
           SO2 and NO, capture.

The above steps are Illustrated in Figure Ib for limestone.
                                             EXPERIMENTAL

Combustion Research Facility

The study was conducted at Ontario Hydro's Combustion Research Facility (CRF) designed for a maximum coal feed rate
of about 20 kg/h (44 Ib/h)  U.S. bituminous coal at a firing rate of 640 MJ/h (0.6 x 10* BTU/h) (Figure 2).  The furnace
is  a refractory-lined cylindrical chamber, fully equipped for monitoring gas and wall temperatures.  There are multiple
ports for flame observation and for insertion of solid sampling probes.  There are also probes to determine  slagging and
fouling rates. The pulverized coal is delivered down-draft to the burner with the combustion air which can be electrically
preheated  to temperatures up to 350C (662F). Gas burners on each side of the coal burner are used to heat the furnace
to operating temperatures before  beginning to  feed the coal.

The coal burner, designed  and constructed by  Research Division staff, is equipped with a vortex generator and four air
vanes to assure good mixing and adequate residence time of the fuel-air mixture in the combustion zone. The combustion
gases in the  furnace are cooled by water and/or air circulating in the cylindrical  Inconel jacket around  the furnace. This
cooling  system is equipped with  temperature sensors and flow meters to control furnace quenching rates.

The combustion gases leaving the furnace are  further cooled by a series of air-cooled heat exchangers prior to entering
the resistivity probe housing and ESP. The ESP consists of a cubic stainless steel chamber, and is equipped  with two sets
of interchangeable cells. One set has  an 11-plate electrode with 2.5 cm (1 in) spacing, the other a 5-plate electrode wiih
5 cm (2 in) spacing.  The design specific collection areas (SCA, m2/m3/s) for the two sets of cells are 39 (0.2 ftVcfm) and
17 (0.09 ftVcfm) respectively for baseline firing conditions using a high volatile U.S. bituminous coal.
                                                  7A-38

-------
The CRF instrumentation permits systems temperatures, and flue gas composition (O,, COj, CO, SO, and NO.) to be
monitored continuously. Gas temperatures in the furnace are measured with a suction pyrometer and flame temperatures
with an optical pyrometer. Flow rates and pressures are measured by flow meters and manometers. All measuring and
monitoring systems are linked to a computerized data acquisition system.  Paniculate mass loading in the flue gas before
and after the ESP is measured with an  isokinetic sampling system.   In-situ resistivity is measured with a point-plane
resistivity probe situated in the resistivity probe housing and particle size distribution of the ash is measured with a Pilot
Mark 3 Cascade Impactor.

A more complete description of the facility is given elsewhere/4/.

SONOX Hardware

A general overview of the hardware used is shown in Figure 3.  A positive displacement pump pumps the slurry/additive
mixture from a continuously stirred mixing tank under a pressure of 650 to  720 kPa.  Recirculation and a static mixer
upstream of the furnace kept the panicles in  suspension and prevented settling.  A small metering pump delivered the
slurry/additive mixture to the atomizer through which  fine droplets were  injected into the flue gas stream.

Injection was in the  middle of the furnace through a twin-fluid high pressure nozzle (5 or 3 mm) with an internal mixing
chamber, shown in Figure 4.  Operating  pressures range between 40 to 60  psig.  The stainless steel nozzle was purchased
from Turbotak Inc.  The MMD  of the droplets was about  12 |im for the  5 mm nozzle and approximately 6 ^im for the
3 mm nozzle.  The nozzle was equipped with a cooling jacket which was necessary to avoid evaporation of the water and
hence drying of the  slurry causing deposition of particles.

Fuels and  Sorbents

Several coals ranging in sulphur content from 0.54% to  2.8% were used with the SONOX technology.  These coals
include a 0.54% beneficiated western Canadian coal, supplied by Unocal  Canada,  a 1.1% S coal resulting from a  blend
of western  Canadian and eastern U.S., a  1.7% S eastern U.S. bituminous  and a 2.8% S coal from Nova Scotia, Canada.
The proximate and ultimate analyses of the coals are shown in Table 1.

Sorbenis used include two local calciuc limestones from Ontario, namely Beachville and PL Anne.  A Beachville hydrated
lime was also studied. Also from Ontario, a dolomitic stone was used supplied by E.C.  King. A Mosher limestone from
Nova Scotia was used with the Nova Scotia coal. The chemical and physical properties of the raw sorbents are shown
in Table 2.  These analyses were performed  by ORTECH International - a  research foundation in the province of Ontario.
Of the additives used to remove NO., the three best are described in this  paper and are labelled A, B and C.

Procedures

After steady state was  achieved  with the baseline coal, injection of the  sorbent  slurry/additive  into  the middle of the
furnace was initiated.  Temperature-lime and radial  profiles simulating Lakeview and  Lambton TGS  were  studied.
Changing the quenching rate allowed the effect of residence time to be studied.  Data collected during each test include
system temperatures, and pressures, slurry/additive-feed rates and stoichiometry, flue gas constituents concentrations (CO2,
O2, CO, SO2 and NOJ, in-situ ash resistivities and particle size distribution. Coal, sorbents feed and fly ash samples were
collected during the  tests. Analysis of samples include chemical composition and panicle size distribution. In selected
runs, NH3,  N2O and  HCN were monitored.  Calcines and sulphated calcines were  analyzed for CaO, CaCO3 and CaSO.
content. Porosity, mass median  diameter and BET surface area of some  samples were also determined. The analytical
methods used are described in reference(4).
                                                   7A-39

-------
                                       RESULTS AND DISCUSSION

The most important parameters that were found to affect process performance (SO2 and NO  capture) are classified under
the following categories:

Sorbent/Additive
           Chemical and physical characteristics;
            Concentration; and
            Addition rate (stoichiometric ratio).

 Injection Parameters
            Mode of injection;
            Droplet size, distribution and mixing with the  flue gas;
            Temperature; and
            Residence time.

 Coal
            SO, and NO, concentration.

These parameters were optimized for maximum SOj/NO, capture on the pilot furnace. It is important, however, to address
some of the advantages of the SONOX process and to mention that negotiations are in progress to demonstrate SONOX
on the full scale. Some of the advantages are:

            SONOX provides a low cost solution to the removal of acid gas from flue gases;.
            SONOX is suitable for retrofit application;
            SONOX is applicable to coals with  various SO2 and NO, levels; and
            SONOX requires short lead time for installation.

Sorbents Comparison

For SOj control using alkaline-based sorbents,  sorbent composition and physical properties are important  factors in
determining overall capture performance(5,6,7,8,9). Significant variability in the reactivity of the various sorbents has been
observed and it was recognized that surface area and porosity play a vital role in sorbent reactivity. Figure 5 illustrates
the effect of porosity on sulphur capture for various sorbents.  Pt Anne limestone with an initial porosity of 55% gave
significantly higher removal than Beachville  limestone with  an initial porosity of 17%  (70%  removal for Pt. Anne
compared to 55% for Beachville) at a Ca/S ratio of 3.0. The  Nova Scotia limestone slurry was used with the Nova Scotia
coal.  Thus a direct comparison of process performance between this sorbent and the local calcitic stones  was not possible.
Data indicate, however, that similar sulphur capture can be obtained with Nova Scotia limestone (porosity 57%) and the
Pi. Anne limestone (porosity 55%) even if they are used for two different coals  (2.8% and 1.7% sulphur content).

Since the additives for NO.-removal are water soluble, only ihe effect of concentration and chemical composition were
evaluated.

Effect of Injection Parameters

Injection parameters that influence SO^NO, capture include:  atomizer type, injector location, atomizing air pressure, and
particle  size distribution or mass median diameter  (MMD) of the atomized droplets.  High atomization air pressure
improves the quality of atomization and promotes an early release of the sorbent/additi ve to engage in the sulphation/NO,
reduction reactions.  It also increases the discharge momentum of the droplets leading to enhanced penetration and mixing
with the flue gas stream. These experiments were conducted with the Turbotak nozzle.

The effect of atomizing air pressure on droplet size is illustrated for limestone slurry in Figure 6. SO2 capture was found
to be a function of droplet size distribution, and  quality of atomization and mixing with the flue gas.  The best mixing
was observed while spraying a 40% aqueous Pt. Anne limestone slurry into the furnace cocurrently at an  injection location
which was close to the flame zone where increased turbulence  exists. Increasing the atomizing pressure from 40 psig 10
70 psig reduced droplet MMD from 12 nm to 6 |im and improved SOj  capture from about 62% to 70% at Ca/S ratio of
3.0.
                                                   7A-40

-------
Effect or Temperature and Injection Mode

(a)   Slurry Injection for SO, Control

     The effect of temperature on SO2 capture was evaluated for the different sorbents (Pt. Anne limestone, Beachville
     limestone, Beachville hydrated lime, Nova Scotia limestone and E.G. King dolomite) while  burning the 1.7% S
     eastern U.S. coal, the 1.1% S eastern U.S./westem  Canadian coal blend and the 2.8% S Nova Scotia coal.  The
     results are shown in Figure 7a. Cocurrent injection gave higher SO2 capture than the countercurrent mode and the
     opumum injection temperature for the recurrent mode was found to be 1200C.  The comparative performance for
     the different coal/sorbent pairs was done with the Turbotak 3 mm nozzle as is illustrated in Figure 7a.

     The highest capture, 85% was observed with hydrated lime to be followed by 83% with the E.G. King dolomite, 65-
     70% with the porous Pt. Anne limestone and 55% with the Beachville limestone at a Ca/S ratio  of 3.0 while burning
     the 1.7% S U.S. coal.  Under the same operating conditions, using the same limestone, SO2 capture from the western
     Canadian/U.S. coal blend was slightly less than from the U.S. coal as is shown in Figure 7a.  Injecting the  Pt. Anne
     limestone with the high sulphur Nova Scotia coal (2.8%)  resulted in 76% SO2 removal at a Ca/S ratio of 3.0.

     Sulphur removal efficiency was 58 to 63% using a 2.8% S  Nova Scotia coal with the porous Nova Scotia limestone,
     at a Ca/S ratio of 2.2. (Because of the presence of grits with this limestone and limited pump capacity, this was the
     highest rate at which  this sorbent could be fed to the furnace.)  However, this compares favourably well with the
     60% capture obtained at a Ca to S ratio of 2.5,  using the porous PL Anne limestone with  the 1.7% S U.S. coal.
     Replacing 5% of the calcium from the PL Anne limestone by an equivalent  amount of dolomite (dolomite doping)
     resulted in 80% SO2 capture, up by  10% from what was achieved with pure Pt.  Anne limestone.

(b)  Additive Injection for NO. Control

     The effect of temperature on NO, removal is shown for the three additives. A, B and C, in Figure 7b while they were
     being injected cocurrently only.  The data indicate that additives A and B show a common optimum  at around
     1100C, while additive C shows a "flat" profile between  975 to 1100C.  At 1100C, additives A and B  removed
     90 and 84% NO. respectively, while between 975 to  1100C additive C removed 77 to 80% NO.. This can be quite
     a desirable feature for full scale boilers where load is constantly changing resulting in changing temperatures.  The
     reason for additive C behaving differently from the  others is not fully understood and further studies may be able
     to provide an explanation.

     Slip  Gases

     The concentration of nitrogen containing species such as ammonia (NHj), hydrogen cyanide (HCN) and nitrous oxide
     (N2O) in the slip gases during additive injection  has been investigated.

     Results indicate that NH3 slippage for additive A ranged between 7 - 26 ppm  and for additive  C up to 49 ppm.  HCN
     was found to be between 3 - 9 ppm. With no NO. removal additive present the N2O produced ranged from 10-25
     ppm at an initial NO. concentration of - 550 ppm.  Decomposition of additive A has a side reaction which could
     lead  to the formation of N2O.  The amount of N2O  produced when additive A was injected ranged from 59 - 150
     ppm at 1100C and an additive/NO stoichiometric ratio of 2.0.  These data demonstrate that 11 to 27% of the NO.
     is converted to N2O thus the effective NO, removal  for additive A is 63 to 80%  instead of 90%.  It was found that
     NjO formation is affected by injection temperature, additive stoichiometry and NO. level  in the flue gas.  More
     studies are required to optimize operating conditions for minimum conversion of NO. to N2O. Additives B and C
     showed  an increase in N2O levels of 5 - 15 ppm from the baseline.

(c)   SONOX Process for SO/NO. Control

     Simultaneous  capture of S02 and NO, was undertaken  by  adding additive A  to an aqueous slurry of PL  Anne
     limestone and dolomite doped PL Anne limestone while burning the 1.7% S eastern U.S. bituminous coal with an
     initial SO2 concentration of 1350 -  1400 ppm and  NO, concentration of 550 ppm. The results  are illustrated in
     Figure 7c for the following optimized conditions:
                                                  7A-41

-------
           40% aqueous calcium-based slurry (Pt. Anne and dolomite doped)
           Ca/S ratio = 3.0
           Additive A concentration of 13.5% (w/w) in slurry
           Addinve/NO mole ratio = 2.0
           Injection mode: cocurrent
           Nozzle:  Turbotak 3 mm, droplet size = 6 ^m MMD

     The graph of Figure 7c shows the effect of temperature on SOj/NO, capture for additive A combined with PL Anne
     and dolomite doped PL Anne slurries.  SO2  capture for  the PL Anne slurry and additive A  at the optimun
     temperature of 1200C is 70% and nominal NO, capture is 90%.  With the 5% dolomite doped PL Anne slurry and
     additive A,  SO2 capture  is 80% and nominal NO, capture is still  90%.

Effect of Stoichiometry

(a)  Ca/S Ratio for SO. Control

     The effect of Ca/S ratios on sulphur capture and sorbent utilization was studied while using the Pt. Anne (porous)
     limestone with the U.S. coal, ihe U.S.-western Canadian blend and the Nova Scotia coal.  The Beachvillc (non-
     porous) limestone, dolomite and hydrated ume  were studied only with the U.S. coal.   Injecting the Pi. Anne
     limestone with the U.S. coal was done at 1200C and 1300C while all other coal-sorbent combinations were done
     at  1200C.  In all  cases  injection took place cocurrently using a 40% aqueous slurry. Ca/S ratios varied from 1.5
     to  3.0 and the furnace quenching rate was held constant at 500C/s.  The results are shown in Figure 8a.  Sulphur
     capture and sorbent utilization are plotted vs Ca/S ratios for the various sorbem-coal pairs.

     Sulphur capture  decreases, but sorbent utilization increases  with decreasing Ca/S ratios for all coal-sorbent pairs
     tested.  At the optimum temperature of 1200C, dolomite and hydrated lime showed the highest capture. Dolomite
     removed 78% of the SC^ at a Ca/S ratio of 1.5 for a calcium utilization of 52%,  while hydrated lime removed 75%
     and 83% SOj at Ca/S  ratios of 1.5 and 2.5 respectively. Sorbent utilization was 50 and  33%.

     At all Ca/S  ratios, the more porous PL Anne limestone outperformed the less porous BeachviUe limestone  both in
     terms of sulphur capture and sorbent  utilization.  At 1200C using the U.S. coal with the PL Anne limestone  at a
     Ca/S ratio of 3.0, sulphur capture and sorbent utilization were 65 to 70% and 22 to 23% respectively as compared
     to  55% and 18% with the  Beachville  limestone.  Using the PL Anne limestone  with the high sulphur Nova Scotia
     coal, sulphur capture at a ratio of 2.0  is 72% and at a ratio of 3.0 is 76%. Under the same operating conditions at
     a Ca/S  ratio of 1.5 sulphur capture for the Pt. Anne and Beachville limestones dropped to 50 and 31 respectively,
     but utilization increased to 33 and 21%.  With the Nova Scotia coal and Pt. Anne limestone at a Ca/S = 1.5,  sulphur
     capture is 64% with a sorbent utilization of 43%.

(b)  Additive/NO Ratio for  NO. Control

     The effect of additive jjormalized stoichiometric ratio, NSR (NSR = moles of  additive injected to the theoretical
     moles required to remove 100% NOJ  for the three additives.  A, B and C, was studied while burning the  eastern U.S.
     bituminous coal. In all cases injection of each additive took place  cocurrently at 1100C while NSR was varied from
     1.2 to 3.0.  The concentrations of the additive solutions were  as follows: A -13.5% by weight, B - 5.6% by weight,
     and C - 16.1% by weighL  The baseline NO, from the U.S. coal was 500-550 ppm.

     NO, capture is illustrated  in Figure 8b.  NO, capture by A and C increases with increasing NSR  up to  1.7  to a
     maximum of 90% (nominal) and 80% respectively, and by  B up to NSR = 2.0 to a maximum of 84%.  Reagent
     utilization drops with increased Stoichiometry for all three additives. The best utilization with A was 55-56% at an
     NSR of 1.2 to 1.5, with B, 56% at a NSR of 1.0 and with C, 41 to 42% at a NSR of 1.5 to 1.7.
                                                 7A-42

-------
(c)   Ca/S - Add/NO for SONOX

     The effect of Ca/S  mole ratio and additive/NO normalized stoichiometric ratio was studied by injecting the 5%
     dolomite doped  Pt. Anne limestone combined with additive A.  The coal burned was the 1.7% S eastern  U.S.
     bituminous and injection was carried out cocurrently at the optimum temperature of 1200C.  The results in Figure 8c
     show thai at a Ca/S ratio of 3.0, 80% SO2 capture is achieved and at an additive to NO stoichiometric ratio of 1.7
     to 2.0, a nominal NO, capture of 90% is achieved.

Low Sulphur Coal Application

The development of the SONOX technology has been carried out mainly on a medium S (1.7%) eastern U.S. bituminous
coal and a high S (2.8%) coal from Nova Scotia with SO: emissions of 1350-1400 and 1700-1725 ppm  and NO, emissions
of 550 and 450-520 ppm respectively.

The effectiveness of the SONOX  process was also demonstrated on a western Canadian Obed coal sample, prepared by
UnocaJ Canada. The  sulphur content of the coal is 0.54% with initial SO2 concentration of 349 ppm.  NO, level initially
measured  620 ppm.  A 40% aqueous  dolomite doped Pt. Anne limestone slurry (10%  dolomite) with additive A  was
injected cocurrently in the pilot furnace and the effects of injection temperature and stoichiometry observed.  The results
are illustrated in Figure 9.

In Figure 9a, SO^NO, capture as  a function of injection temperature is plotted for constant stoichiometries, Ca/S = 3.0
and additive/NO normalized stoichiometric ratio of 3.0.  The results indicate that the optimum temperature was around
1100C for both pollutants with SO2 removal  being 81% and nominal NO. removal being 89%.

The effects of  Ca/S  ratio  and additive/NO stoichiometric ratio is shown in Figure 9b.   Removal of both acid gas
components increases with increasing Ca/S and add/NO ratios. Optimum Ca/S ratio for SO2 is 2.0 to 2.5 and for NO,,
optimum add/NO stoichiometry is 2.0.  Utilization of both sorbents improves with decreasing addition ratios as is shown
in Figure 9b.  Under optimized operating conditions (injection temperature = 1100C, Ca/S = 2.0-2.5  and add/NO = 2.0)
80% SO2 and 85% NO, was removed from the flue gas stream.  Sorbents utilization and 32-40% and 43% respectively.

These results indicate that the  SONOX technology is applicable to coals with various levels of sulphur content and NO,
levels.

Impact on Ash Characteristics,  Collectibility and Deposition

The SONOX process  produces increased amounts of waste composed mainly of CaSO<, unreacted CaO and fly ash. Any
impact on ESP  performance and deposition on the radiant section and convective passes will depend on the type and
chemical composition, the particle size distribution and amount of Ca-based sorbent injected and waste produced.

Waste Characteristics

Particle size distribution of isolrinetically collected waste samples from the injection of various limestone sorbent slurries
while burning a 1.7% S U.S. bituminous coal are compared to that of an ash sample from  the same coal in Figure 10.
The mass  median diameter of the baseline ash  is about 8 (am compared to 6 \im for  the  Pt. Anne and 9 \in\ for the
Beachville limestone slurry. The  slightly finer waste resulting  from the injection of the very fine Pt. Anne limestone is
not expected to  affect panicle migration velocity and ESP collection efficiency.

In Table 3 a typical waste from slurry injection is compared to the baseline ash and to a waste from dry sorbent injection.
High levels of calcium compounds and the quantity produced must be considered for handling and disposal. CaO content
of a typical slurry waste is 302 g/kg and CaSO4 content is about 220 g/kg.

Dust Electrical Resistivity and ESP Performance

The resistivity of the baseline  fly  ash measured in-situ with about 10 ppm SO, naturally occurring in the flue gas from
the  medium sulphur eastern U.S. bituminous coal is about 10* ohm.cm. During injection of  all slurries, resistivities
consistently increased by one to two orders of magnitude to 109 to 1010 ohm.cm yet the electrical operating conditions of
                                                7A-43

-------
the ESP were not seriously affected and collection efficiencies were not seriously degraded (see Table 4) from a baseline
level of 89%  during slurry injection.  Dry injection on the other hand results in a resistivity of 10" ohm.cm and an 8%
drop in collection efficiency.  It is possible that due to the increased moisture level in the flue gas (up to 23%  relative
humidity) a thin acidic film forms around the panicles and acts as a conditioning agent aiding the ESP in its performance.
Inlet mass loading to the ESP has increased 2 fold from a baseline level of 1.4 g/m3 with a resulting increase in paniculate
emissions by  a factor of about 2 times during slurry injection.  Thus the main problem with the SONOX process is the
high dust loading  to the ESP which depends on the Ca/S ratio.

Slagging and Fouling  Properties of the Waste

Soft deposits, which form at low temperatures and are generally characteristic of deposits found on  air heaters and
economizers were observed on the furnace walls and heat exchanger surfaces. These deposits could be easily blown away
by compressed  air suggesting that, conventional soot  blowing  equipment may suffice for full scale application of the
SONOX process.


                                 SONOX COMMERCIALIZATION ISSUES

Electrostatic Precipitator Performance Following SONOX Application

The application of the  SONOX technology in  the upper furnace region affects the nature  of paniculate  mauer entering
the existing electrostatic precipitator.  While the additives for NO, control do not add to the paniculate content entering
the ESP, the calcium sorbents for SO2 control in the furnace result in higher paniculate loading depending on coal sulphur
content and Ca/S  ratios. The precipitator inlet loading can double for most applications.  In addition  to the increase in
inlet paniculate loading, an increase in paniculate resistivity is to be expected because of the uptake of SO, from the flue
gas by free lime in entrained solids.  While dry sorbent injection technologies increase paniculate resistivity from about
109 ohm.cm to  the 10" levels, paniculates from the slurry injection  process show resistivity levels of about 10*  10'
ohm.cm  due  to the higher moisture content in the flue gas.  Hence, the  electrical operation of the ESP is expected to
remain unaffected and  only the solids loading  will have  to be dealt with.

Precipitator upgrades will be needed in most cases following  sorbent injection in order to handle both high loadings and
increased resisuvity. Research-Cottrell has conducted a detailed study on  behalf of the Electric Power Research Institute
and  proposed solutions for the precipitator degradation problems following furnace sorbent uijection(lO).  The  most
economical solution is humidification and subsequent evaporative cooling  of flue gas to restore resisuvity to pre-injection
levels.  At the  lower temperature, due to increased gas density,  the precipitator can be operated at increased power
compared to the  pre-injection level operation at 150C.  The precipitator can thus be operated at higher collection
efficiency to  overcome the increased loading effect. The humidification concept for restoring precipitator operation has
been successfully carried out at two full-scale plants by EPRI and DOE(ll). The humidification concept has also been
demonstrated earlier by Research-Cottrell at the pilot scale in a CONOCO supported program.

The requirements for cooling to restore ESP performance are significantly reduced for the SONOX process because of
reduced paniculate resistivity. We expect the stack paniculate emissions to be restored to pre-injection levels by operating
at ESP inlet gas temperature between 110 to 120C.

Economics

Economic analysis of the SONOX technology indicates that capital costs can vary between 30 to 60 $/KWe, including
moderate precipitator upgrade costs, for combined SO2 and NO, removal rates at SO to 70% each. This can be compared
to the wet FGD capital costs of ISO to 400 S/KWe, the higher cost numbers being applicable to smaller plants in  the 150
MW size range. The operating costs  of SONOX  will  be higher because of higher sorbent consumption  when compared
to wet FGD.  A levelized  cost estimate, however, indicates that a SONOX system is about half the cost of a wet FGD
system to own and operate.

SONOX technology has been demonstrated at the pilot plant level.  Application of the  SONOX concept on a full-scale
coal-fired boiler does impact the overall system and the following questions need to be addressed to assure a successful
commercialization path:
                                                   7A-44

-------
           what is the optimum nozzle  array configuration and slurry size distribution to assure proper gas-slurry
           contact?

           what is the optimum sulphation and NO, removal temperature window in the upper furnace region?

           vhat is the effect of increased solids loading on boiler tube erosion?

           what is the effect of increased loading and resistivity on ESP performance, and what is the best precipitator
           upgrade approach?

           what is the best approach to increased solids handling of the calcium-rich ash?

Many of the answers to  the above questions can be obtained from the experience with full-scale dry furnace sorbeni
injection systems already operating in Germany and other parts of Europe.  Ontario Hydro/Research Cottrell are currently
seeking to demonstrate the SONOX technology on a full-scale coal-fired utility boiler.


                                    SUMMARY AND CONCLUSIONS

The SONOX process, an in-furnace injection of a calcium-based sorbent and a nitrogen-based additive is a very efficient
way  of removing SO2 and NO, from  flue gases.  This technique facilitates unproved distribution and mixing of the
sorbent/additive with the gas flow, reduces deactivation of the sorbent/additive and allows sufficient residence time at
favourable temperatures for the reaction between CaO and SO2, and NH2 and NO to be efficiently completed. The process
was developed at Ontario Hydro's 640 MJ/h (0.6 x Iff  BTU/h) Combustion Research Facility. Coals studied ranged in
sulphur content from 0.54 to 2.8% and calcium sorbents used  include two local calcitic limestones and a hydrated lime
from Ontario, a local dolomitic stone and a limestone from Nova Scotia. NO, levels in the flue gas ranged between 450-
620 ppm and several nitrogen-based additives were investigated.  The following is a summary of the findings:

           Sorbents chemical and physical properties are very important in determining the degree of SO^NO, removals.
           Dolomite with a high magnesium content was very effective in removing SO2 as was the case for hydrated
           lime.  PL Anne limestone with an initial porosity of 55% was superior to Beachville limestone with an  initial
           porosity of 17%. Five percent dolomite doped Pt. Anne limestone increased SO2 capture from 70%  to 80%.
           The nitrogen-based additives did not vary substantially in their ability to remove NO,.

           Injection parameters were  found to be also very important in removing SO2 and NO,.  High atomizing air
           pressure which improves the quality of atomization, promotes and early release of the sorbeni/additive mixture
           and increases the discharge momentum of the droplets, increased SOj/NO, capture significantly.  In  the case
           of  SO2 removal, increasing the atomizing air pressure from 40 to 70 psig increased SO, capture from 62 to
           70%  for the PL Anne limestone.

           The optimum injection temperature for SO2 control was 1200C while NO. was 1100C. However,  with the
           SONOX technology (simultaneous control of both SO2 and NOJ the optimum  temperature was found to be
           1200C. Injecting 5% dolomite doped PL Anne limestone slurry and additive A at the optimum temperature
           of  1200C resulted in 80% SO, capture and nominal NO. capture is 90%.  However, the effective NO.
           removal is 63 to 80% because  11 to 27% of the NO, is converted to N2O.  Hydrated lime removed up to 85%
           SO2 from the flue gas.

           Both SO2 and NO,  improves  with increasing Ca/S and Add/NO stoichiometric ratios.  Optimum Ca/S and
           Add/NO stoichiometric ratios  were found to be 2.5 to 3.0 and 1.5 to  1.7 respectively.  Burning the 1.7% S
           eastern U.S. bituminous coal and injecting 5% dolomite doped PL Anne limestone at a Ca/S ratio of 3.0 and
           additive A at a normalized stoichiometric  ratio of 1.7 removed 80% SO2 and nominally  90% NO,  at the
           optimum temperature of 1200C.
                                                 7A-45

-------
           SONOX was also found lo be very effective for low sulphur coal application.  Firing a low sulphur western
           Canadian Obed coal supplied by Unocal Canada with a sulphur content of 0.54% and injecting 10% dolomite
           doped Pt. Anne limestone slurry and additive A (Ca/S = 2.0-2.5 and add A/NO = '..7-2.0), removed 80% SO2
           and nominally 85% NO, from the flue gas.

           Particle size distribution of the waste from the Pt. Anne slurry was slightly finer than the baseline ash.  The
           waste contains fly ash and calcium compounds (CaO,  CaSO4, etc) and the quantity produced must be
           considered  for handling and disposal systems.

           Ash resistivities increased by one to two orders of magnitude from 10* ohm.cm to 10' to 10' ohrn.cm but
           ESP collection efficiencies  were  not seriously affected.  The increased level of the flue gas  moisture is
           believed to act as a conditioning agent.

           Slagging does not appear to  be  a  problem and the soft deposit formed on  the  furnace walls and heat
           exchanger surfaces was easily removable.

           A levelized cost estimate indicates that a SONOX system is about half the cost of a wet FGD system to own
           and operate and negotiations are in progress to demonstrate this process on the full scale.
                                             FUTURE WORK

Studies are planned whereby other nozzles will be investigated. Other additives that have the potential for high NO,
removal while at the same time ensuring cost effectiveness of the SONOX technology will be studied  Fundamental
studies  to better understand the  SOj/NO. removal paths will be undertaken.  Activating and recycling waste from the
process is being investigated and utilization studies arc being conducted at the University of Calgary.

More importantly, negotiations are in progress to demonstrate this process on full scale boilers.

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.
                                         ACKNOWLEDGEMENTS

The authors wish to express a special thanks to Ontario Hydro's New Business Ventures Division for their dedicated
efforts in conducting negotiations to commercialize the SONOX technology. In particular, we recognize the efforts of
Mr. F. Schneider and Mr. R. Kozopas.
                                              REFERENCES

1.    Taborek, R., Dawson, C.W., and Stuart-Sheppard, IJL, "Acid Gas Emission Control - The Requirements, Technology
     and Hardware"  Ontario Hydro, Design and Development Division, Special Report, March  1986, 3799H.
2.    Bumham, C.. "Ontario Hydro's Acid Gas Control Programs". Paper presented to the Standing Committee on General
     Government, June 15, 1989.
3.    Mangal, R., Mozes, M.S., Thampi, R., and MacDonald, D., "In-Fumace Sorbent Slurry Injection for SO2 Control"
     Presented at the Sixth Annual International Pittsburgh Coal Conference, September 25-29, 1989, Pittsburgh, Penn.
4.    Mozes, M.S., Mangal, R., Thampi, R., and Michasiw, D.L., "Pilot Studies of Limestone Injection Process Phase I:
     Simulating Lakeview TGS Quenching Rate". Ontario Hydro Research Division Report No 86-62-K, May 30, 1986.
                                                  7A-46

-------
5.    Kirchgessner, D.A., Gullett, B.K., and Lorrain, J.M., "Physical Parameters Governing the Reactivity of Ca(OH), with
     SO2".  Presented at the 1986 Joint Symposium on Dry SO2 and Simultaneous SOyNO, Control Technologies, June
     2-6, 1986, Raleigh, North Carolina.
6.    Dismuk;-.-;, E.G., Berttel, R., Gooch, JP., and Rakes, S.L., "Sorbent Development and Production Studies". Presented
     at the 1986 Joint Symposium on Dry SO2 and Simultaneous SOj/NO, Control Technologies, June 2-6,1986, Raleigh,
     North Carolina.
7.    Szekely, J., Evans, J.W., and John, H.Y., "Gas Solid Reactions".  New  York, Academic Press, 1976.
8.    Simmons, G.A., "Rate Controlling Mechanism of Sulphation". Proceedings 1986 Joint Symposium on Dry SO2 and
     Simultaneous SO^NO, Control Technologies, Vol 2, EPRI  CS^966, December 1986.
9.    Mozes, M.S., Mangal, R., and Thampi, R., "Sorbent Injection for SO2 Control:  (A) Sulphur Capture by Various
     Sorbents and (B) Humidification. Ontario Hydro Research Report No 88-63-K, July 1988.
10.  Helfritch. D.J., et al., "Electrostatic Precipitator  Upgrades for Furnace  Sorbent Injection",  EPRI Final  Report
     GS 6282, April 1989.
11.  Altman, R.F., "Precipitation of Particles Produced by Furnace Sorbent Injection  at Edgewater", 8th Symposium on
     the Transfer and Utilization of Paniculate Control Technology, March 1990, San Diego, California.
                                                 7A-47

-------
                                                                Stack
 Sorbent.
  Slurry
    +
 Additives
      Ln
              Esp
                                                           Disposal
                     a) Schematic of SONOX Process
              Heat
              	*
             Water Drop
             Evaporation
               Heat>

           0 Calcination
Limestone Slurry
Atomization
Dry Limestone
   Particles

Particle
Disintegration
-Calcination
-High Pore Structure
 Development
-Sintering Process Avoided
Sulphation
                       b)  Chemical and Physical Steps
                              FIGURE 1

                          SONOX PROCESS
                              7A-48

-------
CD
                                                                                                  Furnace
                                                                                                2) Burner Assembty
                                                                                                3) Air Supply
                                                                                                7) Heat Exchangers
                                                                                                ?) Fttter Unit & Coal Bin
                                                                                                6) Door To Control Room
                                                                                                7) Resisttvity Housing
(a) Electrostatic Predpltator
(9) To Exhaust
(to) Propane Gas Control
(ft) Sortwnt Injection System
(fg) Isoklnedc Sampling System
(g) Water Injection System
(u) Furnace Quenching Pipe*
(1%) HumidifcaBon Chamoer
                                                                                                      FIGURE 2
                                                                                     COMBUSTION RESEARCH FACILITY

-------
                        Air In
                          ft
Cooling
Water In
                |	J   Positive Displacement  ^	|
                      Recirculating Pump
Stanc Mixer
                             FIGURE 3

                       SONOX HARDWARE
  Slurry In
              Air In
                 Water In
                 Water Out
                Internal
             Mixing Chamber
                             FIGURE 4

               TURBOTAK "EXTENDED" NOZZLE
                              7A-50

-------
    70
Ca/S
3.0




2.2-2.5
o
o
A
a
Coal
US
US-W.Can.
N.S.
US
Sorbent
PA
PA
N.S.
B.
I

Q.
re
O
C/3
    60
    50
    40
             10     20     30     40     50


                         Porosity





                      FIGURE 5



            CAPTURE VS LIMESTONE POROSITY
60
                        7A-51

-------
en
IV)
                      12
                                Turbotak 3 mm Nozzle,
                                40 % Apueou* Slurry ol Pt. Anne Limestone
                      10
                    E

                    o>
                    y   6
                    Q.
                    O
                        30         40         50          60

                                   Atomizing Air Pressure, psig

                        a)  Droplet Size vs Atomizing Air Pressure
70
                  Ca/S - 3.0
                  Slurry Rowrata 70 ml/min
                                                                                70
       O
                                                                                65
                                                                                60
2468
             Slurry Droplet Size, \im


 b)  SO2  Capture vs Droplet Size
10
                                                           12
                                                                      FICURE  6

                                        SO2CAPTURE VS SLURRY DROPLET SIZE (ATOMIZING MEDIA -  AIR )

-------
Ul
                   80
                    70
                   60
                 o

                 8"
                 5?
                    40
                    30

 
Coal
U.S.
U.S.
U.S.
U.S-WC
Nova Scotia
Nova S
-------
-J
>
 I
01
                    75
  100

   90

   80

   70

  60

O* 50

88  40

   30

   20

   10
                               US Coal (1.7% S)
                               Initial NO, Cone. - 500 - 550 ppm
A, SR . 2.0
B, SR . 2.0
C, SR - 2.0
                        900    1000
                   1100    1200   1300
                     Temperalure,C
                           1400
                                                                      o
                                                                       x
100

 90

 80

 70

 60

 50

 40

 30

 20
                                                               US Coal (1.7% S)
                                                               Initial NOX Cone. . 500 - 550 ppm
                                                               Ca/S Ratio - 3.0
                                                               AddvNO Stoichiometry . 1.7
                                                                                             Dolomite Doped P.A. Limestone
                                                                                P.A Limestone
                                                                                               SO 2 Removal
                                                                                               NO x Removal "
                                         100

                                         90

                                         80
                                            0)
                                         70 3
                                            CL
                                            co
                                         60 o
                                            ox
                                         50 Z
                                            5?
                                         40

                                         30

                                         20
                    " Effective NOX Removal 63-80 % due to N2O formation
  900    1000   1100    1200   1300    1400
             Injection Temperature, C

Effective NOX Removal 63-80 % due to N2O formation
                                     FIGURE 7b

                        NO- CAPTURE - EFFECT OF INJECTION
                                   TEMPERATURE
                                                                       FIGURE7c
                                                                   SONOX PROCESS
                                                       SO>/NO. CAPTURE - EFFECT OF INJECTION
                                                                    TEMPERATURE

-------
Ul
en
                                    40 % Aqueous Slurry
                                    Co Current Injection
                                    Droplet MMO - 6fim

                                    -o-  1200C
                                    -  1300Q
2              3
    Ca/S Ratio
                  BHL -Beachville Hydrated Lime
                  B    Beachville Limestone
                  PA   Pt. Anne Limestone
                  US  -U.S. CoaJ
                                     65

                                     60

                                     55

                                 c  50
                                 g
                                 To  45
                                 N
                                 ^  40

                                 en  -3C
                                 O  J0

                                 5s  30

                                     25

                                     20

                                     15
U.S. - WC - U.S. Western Canadian Coal Blend
D       * Dolomite
NSC     - Nova Scotia Coal
                                                                                           .-US
                                                                                                                D-US
                                                                                                                PA-NSC
                                                                                                                PA-US
                                                                                                              0 PA - US - WC

                                                                                                              OB-US
                                                                                  1
                                                                                                     Ca/S Ratio
                                                                       FIGURE 8a

                                                        SO2 CAPTURE - EFFECT OF Ca/S RATIO

-------
Ul
05
                       U.S. Coal
                       Initial NOX Cone.  500 - 550 ppm
                       Injection Temperature - 1100 C
                                23
                               Additive Stoichiometry
                                                                 
-------
en
                           100
                            80
                        (0
                        

                        DC

                        ox
                         
-------
  100
  40% Slurry
_ 1.7% S Eastern US Coal
  Ca/S-3:1
2
a.
a

           Ft. Anna Slurry
                                            I
         2.5   5.0  7.5  10.0  12.5  15.0  17.5  20.0

                   Sieve Opening, (am
                   FIGURE 10

  PARTICLE SIZE DISTRIBUTION OF BASELINE
         FLYASH AND SLURRY WASTES
                     7A-58

-------
en
CD
                                     TABLE I
                          CHARACTERISTICS OF COALS
                 Proximate Analysis, g/kg
Ultimate Analysis, g/kg

Moisture
Ash
Volatile Mattel
Fixed Carbon
Healing Value
MJ/kg
US
Coal
14
80
357
548
32

Nova Sea 13
Coal
12
96
314
577
31

W Can US
Coal Blend
32
99
321
548
29

UNOCAL
Coal
33
135
367
465
28









Carbon
Hydrogen
Nitrogen
Sulphur
Ash
Oxygen
US
Coal
756
57
16
17
80
74
NovaScoiia
Coal
756
50
12
28
96
58
W Can US
Coal Blend
751
47
14
11
71
101
UNOCAL
Coal
673
50
15
5
135
122
                                     TABLE 2
              CHEMICAL AND PHYSICAL PROPERTIES OF SORBENTS

LJ2O. g/kg
NajO
K2O
MgO
CaO
F23
AI203
Si02
Ca(OH)2
LOI
BET area, nf/g
MMD. |im
p. g/cm
POROSITY. %
Beach villa
Limestone
<003
003
06
80
5240
01
220
<12.0

4340
1.3
86
26
170
Pi. Anne
Limestone
.
0.1
05
48
5354
2.1
43
21.0


29
39
2.3
550
Mosher
Limestone
(Nova Scotia)

0.1
l.t

538.0
64
65
25.0


1 86
110
25
570
E.G. King
Dolomite
.
04
<\0
212.1
300.9
2.3
10
7.9

4610
06
33.0
25
420
Beachville
Hydrated
LJme
0002
03
<005
7.6
1410
1.5
22
5.1
7880

126
82
2.1
264
        TABLE 3
  WASTE COMPOSITION
1.7% S US Coal with Limestone Sluuy
Ca/S = 2.5
Temperature = 1200 C

CaO
CaSO<
CaCOj
MgO
LCH
kg Waste/I 00 kg Coal
Baseline
9*9
34
39
30
9
47
88
Sorbent Slurry Injection Waste
g/kg
302
257
44

50
17
Dry Injection Waste
g/kg
316
220
36

44
17
                                                                                                   TABLE 4
                                                                                 WASTE RESISTIVITIES AND ESP PERFORMANCE
                                                                                        Injection Temperature 1200 C (Dry = 1100 C)
                                                                                        40% Limestone Slurry
Coal
U.S.
US.
US.
us
Sorbent

Beachville Limestone
(Slurry)
Pt. Anne Limestone
(Slurry)
BeachvHIo (Dry)
SO2
Removal
%

55
62
43
Flue Gas
Relative
Humidity
%
8-10
23
-19
8-10
Ash
Resistivity
ohm-cm
83 x 107
47 x 109
l.lxtO10
1 1 x 10n
ESP
Performance
Efficiency
%
89
88
87
at

-------
PILOT PLANT TEST FOR THE NOXSO FLUE GAS TREATMENT SYSTEM

                           L.G. Neal
                          Warren T. Ma
                       NOXSO Corporation
                          P.O. Box 469
                    Library, Pennsylvania 15129

                          Rita E. Bolli
                          Ohio Edison
                       76 South Main Street
                        Akron, Ohio 44308

-------
        PILOT PLANT TEST FOR THE NOXSO FLUE GAS TREATMENT SYSTEM
                                       L.G. Neal
                                     Warren T. Ma
                                   NOXSO Corporation
                                      P.O. Box 469
                               Library, Pennsylvania  15129
                                      Rita E. Bolli
                                      Ohio Edison
                                   76 South Main Street
                                   Akron, Ohio 44308
ABSTRACT

The NOXSO process is a FGT system that employs a reusable sorbent.  A fluidized bed of sorbent
simultaneously removes SO2 and NOX from flue gas.  The spent sorbent is regenerated by treatment
at high  temperature with  a reducing  gas.  Adsorbed NOX is  evolved  on  heating the sorbent to
regeneration temperature. The concentrated stream of NOX produced is returned to the boiler with the
combustion air.

NOXSO Corporation, MK-Ferguson, W.R. Grace & Co., and Ohio Edison will conduct a pilot test
of the NOXSO system  at Ohio Edison's Toronto station.  The plant treats 12,000 SCFM of flue gas
containing 2300 ppm SO2 and 350 ppm NOX, which is roughly 1/20 the size of a commercial module.
The paper summarizes  the system design.

An additional test of the NOX recycle concept will be conducted at the Babcock & Wilcox Research
Center in Alliance, Ohio.   The  test apparatus is a 6 million Btu/hr small boiler simulator.   It is a
scaled-down version of B&W's single cyclone front wall fired boiler design.  The proposed test plan
and the data from previously reported NOX reduction tests using a pc-fired system at the Pittsburgh
Energy Technology Center are included.
                                         7A-63

-------
INTRODUCTION

The NOXSO Process  simultaneously  removes SO2 and NOX from the flue gas of coal-fired boilers
using a dry, regenerable sorbent. Three previous tests of the NOXSO Process have been conducted.
The first was a bench-scale program conducted at TVA's Shawnee Steam Plant for the purposes of
establishing process chemistry and  kinetics, quantifying sorbent attrition rates, and establishing the
corrosion properties of different metals for use in specific applications within the NOXSO Process.
The kinetic tests were all performed in a fixed bed reactor (1.2).  Funding was provided by NOXSO
and by the U.S. Department of Energy's (DOE) Pittsburgh Energy Technology Center (PETC). The
second and third test programs were funded and conducted by DOE at PETC with technical guidance
provided by NOXSO Corporation.  The second test program was designed to test laboratory data in
a scaled-up process, 3/4 MW in size (3). The third test program was a life-cycle test to  determine
sorbent physical and chemical performance over repeated cycles of adsorption and regeneration (4).
The current test program is a 5 MW  pilot plant that will provide the data necessary to scale up to a
full size (100 MW) module (5).  The pilot plant is currently under construction at  Ohio Edison's
Toronto Station and is scheduled to begin operation in May 1991.  NOXSO Corporation is responsible
for operation of the pilot plant while  funding comes from DOE, the Ohio Coal Development Office,
NOXSO Corporation, W.R. Grace & Co., and MK-Ferguson Co.  A brief comparison of these four
test programs is given in Table 1.  Detailed information on  test facility design, test results, and data
analysis can be obtained from the previously referenced reports.

PROCESS  DESCRIPTION

The NOXSO Process is a post-combustion flue gas treatment technology that simultaneously removes
both SO2 and NOX from the flue gas generated by coal-fired utility boilers.  The process utilizes a high
surface area 7-alumina substrate impregnated with sodium to achieve removal efficiencies of 90% for
SO2 and 70%-90% for NOX.  A process flow diagram is shown in Figure 1, and a description of the
process is given below.

Flue gas leaving the boiler passes through the combustion air preheater,  the electrostatic precipitator,
and into the NOXSO flue gas treatment system.  The flue gas is first cooled to 120C by vaporizing
a water  stream  sprayed directly in the  ductwork.   The cooled  flue gas  is  then passed  through  a
fluidized bed of sorbent where the SO2 and NOX are simultaneously adsorbed.  The  clean flue gas
                                           7A-64

-------
flows through a cyclone where attrited sorbent is separated and returned to the adsorber bed. Finally,
the flue gas is returned to the power plant duct and exhausted through the stack.

After the sorbent is loaded with SO2 and NOX, it is removed from the adsorbers and pneumatically
conveyed to a sorbent heater. The sorbent heater is a three-stage fluidized bed where a hot air stream
is  used to heat the sorbent to 660C.  During the heating process, NOX and loosely bound SO2 are
desorbed and transported away in the heating gas stream. The hot air stream exiting the sorbent heater
is  recycled back to the boiler replacing a portion of the combustion air while providing an energy
credit to the NOXSO Process. At normal boiler operating conditions, the recycled NOX will either be
reduced by hydrocarbon fuel or suppressed by the formation of additional NOX so that a steady-state
equilibrium concentration of NOX is attained.

Once the  sorbent reaches a regeneration temperature of 660C, it is transported from the  sorbent
heater  to a moving bed regenerator.  In the regenerator, sorbent is contacted with natural gas in a
countercurrent fashion.  The natural gas reduces sulfur compounds on the sorbent (mainly  sodium
sulfate) to primarily SO2 and H2S with some COS also formed (less than 2%  of total inlet  sulfur).
Approximately 48% of the sodium sulfate is reduced to sodium sulfide which must subsequently be
hydrolyzed in the  steam treatment vessel.   The moving bed steam treatment is obtained from the
reaction of steam with Na2S.  The regenerator off-gasses are sent to a Glaus plant where SO2 and H2S
are reacted to  form elemental sulfur.  The sulfur is sold as a by-product of the NOXSO Process.

From the steam  treatment vessel, the sorbent is fed to a sorbent cooler.  The cooler is a three-stage
fluidized bed where the sorbent is cooled to 120C using an ambient air stream.  The warm air exiting
the cooler is further heated in a natural gas  fired  heater before being used to heat the sorbent in the
fluidized bed  heater. The cooled sorbent is returned to the adsorber completing one full cycle.

PROCESS CHEMISTRY

The NOXSO sorbent is prepared by spraying Na2CO3 solution on the surface of 7-alumina sphere (1.6
nominal diameter).  Both sodium and alumina contribute to the NOXSO sorbent's capacity to adsorb
SO2 and NOX from  flue gas.  Our laboratory tests show that the presence of steam in the flue gas helps
the SO2 sorption. The reaction of the sodium can be described as follows:
                                          7A-65

-------
                            Na2C03  + A1203      2NaAl02  +  CO2
                            2NaAlO2 + H2O <> 2NaOH + Al2O3                        (2)

                           2NaOH +  S02 +  -O2  >  Na2SOt  + H2O                     (3)
                          2NaOH + 2NO +  O2 <> 2NaNO3  +  H2O                     (4)
                          2NaOH + 2NO2  +  O2  > 2NaNO3  + H2O                     (5)
                                            Zj
 Adsorbed nitrogen oxides are decomposed and evolved on heating the spent sorbent to regeneration
 temperature.  The concentrated stream of NOX produced on heat-up is returned to the boiler with the
 combustion air.  This results in no significant increase of NOX concentration in the boiler flue gas
 because of the reversibility of NOX formation in the boiler (1.2).

 The spent sorbent can be regenerated at high temperature with many kinds of reducing gases, such as
 H2S, CO, H2, natural gas, etc.  The regeneration reaction, for example, using natural gas at 610C
 is described below:
                                                  O2 +  CO2 + 2H2O                    (6)

                         4Na2SO3 + 3CHi  > 4Na2S + 3CO2 + 6H2O                    (7)
                            A1203  + Na2S03 <>  2NaAlO2 +  SO2                       (8)
                         A12O3 + Na2S  +  H2O <>  2NaAlO2  + H2S                     (9)

Although sulfite has not been identified in our studies, it is a likely intermediate in sulfate reduction.
A detailed discussion on the existence of sulfide during regeneration had been given by Gavalas it.al.
(6)  who used  CO  to study the regeneration of alkali-alumina.  The  SO2 and H2S produced from
regeneration are then converted to elemental sulfur in a Claus-type reactor.
                                           7A-66

-------
                              S02  + 2H2S < > XS3/X + 2H20                         (10)
The sulfur produced is a marketable by-product of the process.

PROOF-OF-CONCEPT PILOT TEST

Background
On May 10, 1989, a consortium assembled by NOXSO Corporation signed a cost-shared contract with
the DOE/PETC to conduct a POC test of the NOXSO process. The consortium consists of NOXSO,
MK-Ferguson, W.R. Grace & Co.,  Ohio Edison and the Ohio Coal Development Office.  The POC
project will take approximately three years to complete, and the test will be conducted at a coal-fired
Ohio Edison plant in Toronto, Ohio.

POC Test Site
The POC unit will treat flue  gas from  either Boiler #10 or Boiler #11 at Ohio Edison's Toronto
Station.  Two sources of flue gas will be tapped so that  the POC test can continue as long as one of
the boilers is operating.  A slipstream of flue gas will be taken downstream of the Toronto plant's
electrostatic precipitators.  The  Toronto boilers are pc-fired and burn Ohio coal containing 3.7%
sulfur. The flue gas typically contains 2300 ppm SO2 and 350 ppm NOX.

POC Test Schedule
Detailed design engineering has been completed and the major pieces of equipment have been ordered.
Construction began in April 1990 and will be completed in May 1991. The test will run through
February 1992.

POC Process
The process flow diagram for the POC has shown in Figure 1. The system differs from a commercial
application of the NOXSO technology in  two important areas.  First, the POC facility does not include
a Claus plant, which in the commercial design would be used to produce a sulfur by-product from the
concentrated stream of SO2 and H2S produced in the regenerator.  This is because Claus technology
is commercially available and therefore does not require  testing at pilot scale.  Second, the POC does
not include NOX recycle to  the coal combustor.  In the commercial design, NOX in the air leaving the
sorbent heater is recycled to the combustor as part of the combustion air.  Since NOX formation in the

                                          7A-67

-------
coal combustor is a reversible reaction, addition of NOX to the combustion air suppresses the formation
of NOX in the combustor.  However, NOX recycle is impractical in the POC test since the POC treats
less than 10% of the flue gas produced by Toronto Unit 10 or 11.

POC Test Unit Design
Data from three previous tests of the NOXSO process were used to design the POC test facility.  A
comparison  of the three previous  test programs was given in Table 1.  The design specifications for
the major equipment in the POC test facility are listed in Table 2.

Materials of Construction
During development of the NOXSO process, some corrosion problems were encountered, particularly
in the regenerator.  Different materials of construction were tested to withstand the high temperature
environment of SO2, H2S, elemental sulfur, and sulfated sorbent.  Corrosion results were documented
in an earlier report  (2), the practical results of the test program are discussed here.

In tests performed at the Shawnee Steam Plant, sorbent was heated with electrical resistance heaters
made of Inconel 600, Monel 400, type 316 and type 316L stainless  steel (SS).  All these materials
exhibited severe corrosion in areas of sorbent contact attributed to hot sulfation of nickel.  It should
be noted that the temperature of the heating elements themselves were substantially higher than the bed
temperature of 600C.  The reactor, made out of either type 316 or type 316L SS, showed scale on
the inside surfaces after use.  When the reactor was made of type 446 SS or alonized type 316L SS,
there was no scale and only a slight discoloration of the metal surfaces observed.

In the LCTU, the regenerator was made of alonized type 304 SS and showed no visible evidence of
corrosion at the end of 330 regeneration cycles. Based on these results, it was felt that either 446 SS
or alonized 304 or 316L SS  would be satisfactory for the POC regenerator.

The sorbent heater also encounters hot sulfated sorbent and will therefore be made of type 304 SS.
The bottom  bed  of  the sorbent heater where the temperature is 660C will be alonized.  All other
vessels will be made of standard A-285 or A-283 grade C carbon steel, as no  corrosion problems are
anticipated.
                                           7A-68

-------
The other area in the NOXSO process that requires special consideration for materials is between the
flue gas cooler and the adsorber. In this area, sub-acid dewpoint corrosion can occur. All previous
NOXSO tests have cooled the flue gas indirectly while at the POC the flue gas will be cooled by a
direct water spray in the ductwork. The flue gas temperature in this portion of the  system will be
112F so that an acid-resistant epoxy coating will be used to line the ductwork from the cooler to and
including the bottom of the adsorber. This epoxy has not been tested previously by NOXSO, but there
exists ample literature that supports its use as an acid resistant material in other similar applications.
NOX RECYCLE
NOX recycle will be implemented at the full-scale commercial demonstration plant.  The concept of
NOX recycle has been tested previously using the 500 Ib/hr coal combustor used for the 3/4 MW tests
and also using a tunnel furnace capable of being fired with a variety of fuels including gas, fuel oil,
coal, and coal-water mixtures.

Previous NOX Recycle Results
NOX recycle was tested by spiking the combustion air with varying concentrations of bottled NOX and
measuring  the outlet NOX concentration  from the  combustor.   The net NOX production rate was
determined by a material balance on the combustor as shown schematically in Figure 2. The NOX flow
rate at the exit of the combustor minus the NOX feedrate into the combustor equals the rate that NOX
is produced in the combustor, which is defined as the net NOX production rate (R). For data reduction
purposes,  the NOX production rate (R) and the NOX feedrate (F) were normalized with  respect to
conditions at zero NOX feed according to R*=R/R0 and F*=F/R0 where R0 is the NOX production rate
at F =  O.  Results from the 500 Ib/hr combustor are compiled in Table 3.  The measured data are
NOX concentration at the exit of the combustion  system and the  flow rate  of NOX fed into the
combustor with the combustion air.  Data provided  in the other columns were calculated.

A plot of R* versus F* is shown in Figures 3  and 4 for both the 500 Ib/hr combustor and the tunnel
furnace, respectively.  In each case, the data fall in a straight line, but with different slopes. The two
lines are described by the equation R* = 1 - aF*. The parameter "a" is the slope of the line and also
represents the fraction of NOX fed to the combustor that is destroyed,  The value of "a" is 0.65 for
the 500 Ib/hr combustor and 0.75 for the tunnel furnace. The data for the tunnel furnace include both
natural gas combustion and coal-water slurry combustion.
                                           7A-69

-------
These  results  demonstrate that the nature  of the  fuel has no affect on  the  effectiveness  of the
combustion system to reduce NOX fed through the combustion air.  Also, the NOX reduction capability
of a combustion system is independent of the  amount of NOX fed with the combustion air. Finally,
the most important variables are those associated with the combustor design.  NOX recycle will be
extensively studied at the Babcock & Wilcox Research Center in Alliance,  Ohio.

Pilot-Scale NO. Recycle Test
The power plant selected for the NOXSO full-scale demonstration (Ohio Edison's Niles Station, Niles,
Ohio)  uses cyclone burners.  Since the destruction efficiency of NOX recycle has not previously been
tested  with cyclone type burners, a demonstration of NOX recycle with this type of coal combustor is
necessary for the proper design of the NOXSO full-scale plant.

Pilot-scale NOX recycle  tests will be done using Babcock & Wilcox's 6 million Btu/hr Small Boiler
 Simulator (SBS) shown  in Figure 5.  The water-cooled furnace is a scaled-down  version  of B&W's
 single-cyclone, front-wall fired boiler design. The cyclone has been in operation since February 1985.
 The SBS cyclone furnace simulates a large cyclone unit very well.  A comparison between the SBS
 cyclone furnace and commercial units is given in Table 4.

 The NOX recycle tests will begin with three loads and three excess air levels to establish the baseline
 of the NOX emission from the SBS furnace.  NO will then be injected in multiples of the baseline NOX
 production  levels.  The NO concentration at  the air inlet  duct to the cyclone will be measured to
 document the inlet level. Stack NOX will be measured to determine NOX destruction occurring in the
 flame.  The series of tests with different NO  injection rates will also be performed at three furnace
 loads  and three excess  air levels.  This test result will assist the determination of a second injection
 location for the next  series of tests.

 In the second  series of tests, NO and NO2 will be injected separately for two furnace loads and two
 excess air levels.  Volumetric  flowrate of the injected NO and NO2 will be based on the  proportion
 of these gasses that are present in the NOXSO sorbent heater off-gas.  The addition of methane to the
 air stream to assist the NOX destruction (7) will also be tested.  The NOX recycle test will be finalized
by  burning  the coal from  the Niles plant  in the SBS  furnace.   Since the  coal-ash slagging
characteristics are important to the power plant operation, the use of Niles plant coal will assess the
change of the coal ash's "flowability" in the Niles plant when the NOX recycle  stream is installed.
                                            7A-70

-------
FUTURE WORK


On December 21,  1989, NOXSO Corporation, in association with MK-Ferguson Company, W.R.

Grace & Co., and Ohio Edison, received an award from DOE's Clean Coal Technology Program to

conduct a $66 million, full-scale commercial demonstration of the NOXSO technology.  The U.S.

DOE  will provide $33 million and  the  remaining funds will  be provided  by the Ohio Coal

Development Office, the Electric Power Research Institute, the Gas Research Institute, the East Ohio

Gas Company, and the aforementioned NOXSO development team.  The 115 MW demonstration plant

will be installed at Ohio Edison's Niles Power Plant in northeastern Ohio.  Construction is scheduled

to begin in early 1993 with plant startup scheduled in May 1994.  This project is the final step in the

program to commercialize the NOXSO technology.


REFERENCES
 1.   J.L. Haslbeck, CJ. Wang, L.G. Neal, H.P. Tseng, and J.D. Tucker. Evaluation of the NOXSO
     Combined  NOX/SO2 Flue Gas Treatment Process.   NOXSO  Corporation Contract Report
     submitted to U.S. DOE Report No. DOE/FE/60148-T5. November 1984.

 2.   J.L. Haslbeck, L.G. Neal, CJ. Wang, and C.P. Perng. Evaluation of the NOXSO Combined
     NOX/SO2 Flue Gas Treatment Process.  NOXSO Corporation Contract Report submitted to U.S.
     DOE Report No. DOE/PC/73225-T2.  April 1985.

 3.   J.L. Haslbeck, W.T. Ma, and L.G. Neal.  A Pilot-Scale Test of the NOXSO Flue Gas Treatment
     Process. NOXSO Corporation  Contract Report submitted to U.S. DOE Contract No. DE-FC22-
     85PC81503.  June 1988.

 4.   J.L. Haslbeck, J.T.  Yeh, W.T. Ma, J.P. Solar, and  H.W. Pennline.  Life-Cycle Test of the
     NOXSO Process:  Simultaneous Removal  of NOX and SO2 from Flue Gas. Presented at the 1989
     AWMA Annual Meeting, Anaheim, California.  June  1989.

 5.   J.L. Haslbeck, M.C. Woods, R.E. Bolli, R.L. Gilbert, and C.P. Brundrett. Proof-of-Concept
     Test of the  NOXSO Flue Gas Treatment System. Presented at the EPA/EPRI 1990 SO2 Control
     Symposium.  New Orleans, Louisiana. May 8-11, 1990.

 6.   G.R. Gavalas, S. Edelstan, M. Flytzani-Stephanopoulous, and T.A. Weston.  Alkali-Alumina
     Sorbents for High-Temperature Removal of SO2. AIChE Journal  Vol. 33, No. 2, p. 258.  1987.

 7.   J.T. Yeh, J.M. Ekmann, H.W. Pennline, and CJ. Drummond.  New Strategy to Decompose
     Nitrogen Oxides from Regenerable Flue Gas Cleanup Processes. Presented at the 194th ACS
     National Meeting. New Orleans, Louisiana.  Aug. 30 Sept. 4, 1987.
                                         7A-71

-------
   NOx RECYCLE
                         TO CLAUS PLANT
                             REGENERATOR
                             AIR
NOXSO PROCESS FLOW DIAGRAM
            FIGURE 1

              7A-72

-------
        	-t-
                    I          E,
         Adsorber
             R+F
         Combustor
FIGURE 2. SCHEMATIC DIAGRAM OF NITROGEN
         OXIDE RECYCLE.
                     7A-73

-------
CC

bJ

*
CC


o
f-
o

o
o
cr
QL
o
LJ
cc
o
-2.0
               1.0       2.0      3.0

            NORMALIZED NOx FEED RATE, F*
                                      4.0
       FIGURE 3. NORMALIZED NOx REDUCTION

                DATA-PC COMBUSTOR.
      *
      OC

      UJ


      CC.


      o
      h-
      o

      o
      o
      oc.
      o.
       x
      O
      z

      o

      N
      QL
      O
     + 1



      0



     -I



     -2



     -3


     -4



     -5



     -6



     -7



     -8
           0        5        10        15

            NORMALIZED NOX FEED RATE, F*


            FIGURE 4.  NORMALIZED NOx

                     REDUCTION  DATA-

                     TUNNEL FURNACE.
                    7A-74

-------
    STACK
                  STEAM
 REHEATER
 DEPOSITION 
 PROBE
                                     SUPERHEATER
                                     FOULING TUBE
                                     DEPOSITION PROBE
FLUE GAS
RECIRCULATION
                                                 FURNACE ARCH

                                                PRIMARY AIR
                                                AND COAL
                                                    TERTIARY AIR
                                                     SECONDARY
                                                     AIR
                            SLAG TAP
                                             MOLTEN SLAG
                                               SLAG COLLECTOR
                                               AND FURNACE
                                               WATER SEAL
  FIGURE 5. SMALL BOILER SIMULATOR (SBS) SCHEMATIC
                               7A-75

-------
           Table 1.  Comparison of NOXSO Test Programs

Operating Parameter
Coal Burned, Ibs/hr
Flue Gas Volume, SCFM
Adsorber Type
SO2 Inlet Concentration, ppm
NOX Inlet Concentration, ppm
SO2 Removal Efficiency, %
NOX Removal Efficiency, %
Reducing Gas for Regeneration
Operating Mode
Test
TVA
NA
0.35
Fixed Bed
2300
600
90
90
H2S, H2, CO
Batch
Program
3/4 MW
500
1200
Fluid Bed
1465-5000
470-720
90-99*
80-92*
H2, H2+CO, CH4
Batch
Test Program
Operating Parameter
Coal Burned, Ibs/hr
Flue Gas Volume, SCFM
Adsorber Type
SO2 Inlet Concentration, ppm
NOX Inlet Concentration, ppm
SO2 Removal Efficiency, %
NOX Removal Efficiency, %
Reducing Gas for Regeneration
Operating Mode
LCTU
40
120
Fluid Bed
1450-3000
240-800
60-90*
60-90*
H2, CH4
Continuous
POC
NA
12000
Fluid Bed
2300
350
**
**
Natural Gas
Continuous
NA = Not applicable, i.e., small slipstream was drawn from the power plant ductwork.

  * = In the 3/4 MW and LCTU tests, removal efficiencies cover a wide range since
      operating conditions were intentionally varied to study their effect on process
      performance.

 **  = Pilot plant is under construction.
                                  7A-76

-------
         Table 2.  POC Major Equipment Specifications*
Fluidized Bed Adsorber
   Diameter                          10.5 ft
   Temperature                       120C
   Settled Bed Height                 2 ft
   Sorbent Residence Time            45 min
   Superficial Gas Velocity            3 ft/s
   Transport Disengaging Height       7.7 ft
   Material of Construction            Carbon Steel

Fluidized Bed Sorbent Heater
   Number of Stages                 3
   Diameter                          7.7 ft
   Settled Bed Height                 0.9 ft
   Sorbent Residence Time            30 min
   Superficial  Gas Velocity            3 ft/s
   Transport Disengaging Height       2.8 ft
   Material of Construction            Type 304 SS

Fluidized Bed Sorbent Cooler
   Number of Stages                 3
   Diameter                          5.7 ft
   Settled Bed Height                 1.2ft
   Sorbent Residence Time            20 min
   Superficial  Gas Velocity            3 ft/s
   Transport Disengaging Height       4.3 ft
   Material of Construction            Carbon Steel

Moving Bed Regenerator/Steam Treater
   Diameter                          4 ft
   Bed Height                        10.3 ft/6.8 ft
   Sorbent Residence Time            30 min/20 min
   Material of Construction            Alonized Type 304H SS
Air Heater
   Design Flow (Air)
   Temperature Rise
   Type

Pneumatic Conveyor
   Sorbent Circulation Rate
   Lift Distance

Adsorber Cyclone
   D-50
   D-100
   Gas Flowrate
* At base case operating conditions.
      14,300 Ibs/hr
      330C
Natural gas fired in duct burners
      9,994 Ibs/hr
      80ft
      20/xm
      100 MHI
      16,257 ACFM @ 120C
                           7A-77

-------
Table 3.  NOX Reduction Data; 500 Ib/hr Combustor (3)

Test >
No.#
1
2
3
4
5
6
7
8
9
10
Tests 1
Tests 4
Tests 7

F

JOxExit NOxExit NOX Fed R
ppm Ih/hr Ib/hr Ib/hr R* F*
550 3.59
1370 8.94
875 5.71
650 4.24
850 5.55
930 6.07
700 4.56
1100 7.17
1200 7.82
820 5.34
0 +3.59 1.0
14.09 -5.15 -1.43
8.29 -2.58 -0.72
0 +4.24 1.0
4.66 +0.89 0.21
5.49 +0.58 0.14
0 +4.56 1.0
6.64 +0.53 0.12
7.98 -0.16 -0.04
1.60 +3.74 0.82
3. Coal feedrate = 223 Ibs/hr, Flue gas flowrate =
moles/hr (dry), and Temperature = 2500F.
6. Coal feedrate = 352 Ibs/hr, Flue gas flowrate =
moles/hr (dry), and Temperature = 2500F.
10. Coal feedrate = 431 Ibs/hr, Flue gas flowrate =
moles/hr (dry), and Temperature = 2500F.
0
3.92
2.31
0
1.10
1.29
0
1.46
1.75
0.35
122.1
160.0
180.4
 Table 4.  Comparison of Baseline Conditions Between
          the SBS Facility and Commercial Units


Cyclone Temperature
Residence Time at full load
Furnace Exit Gas Temperature
NOx Level
Ash Retention
Unburned Carbon
Ash Particle Size (MMD; Bahco)

SBS
>3000F
1.4 sec
2265 F
900-1200 ppm
80% -85%
< 1 % in Ash
6-8 microns
Typical
Cyclone-Fired Boilers
>3000F
0.7-2.0 sec
2200-2350F
600-1400 ppm
60% -80%
1%-20%
6-11 microns
                      7A-78

-------
 THE PRACTICAL APPLICATION OF TUNABLE DIODE LASER INFRARED
SPECTROSCOPY TO THE MONITORING OF NITROUS OXIDE AND OTHER
             COMBUSTION PROCESS STREAM GASES

                         Frank E. Briden
            Air and Energy Engineering Research Laboratory
                U.S. Enviornmental Protection Agency
             Research Triangle Park, North Carolina 27711

                        David F. Natschke
                        Richard B. Snoddy
                       Acurex Corporation
                      4915 Prospectus Drive
                   Durham, North Carolina 27713

-------
       THE PRACTICAL APPLICATION OF TUNABLE DIODE LASER INFRARED
      SPECTROSCOPY TO THE MONITORING OF NITROUS OXIDE AND OTHER
                    COMBUSTION PROCESS STREAM GASES
                                 Frank E. Briden
                   Air and Energy Engineering Research Laboratory
                        U.S. Environmental Protection Agency
                    Research Triangle Park, North Carolina 27711
                                David F. Natschke
                                Richard B. Snoddy
                                Acurex Corporation
                              4915 Prospectus Drive
                           Durham, North Carolina 27713
                                   ABSTRACT

There are a number of gases associated with combustion process streams which are difficult
to monitor because of their physical properties and interferences from other gases.  Tunable
diode laser infrared (TDIR) spectroscopy offers a reliable, specific means for the continuous
monitoring of many of these gases. Some of the gases that can be efficiently monitored by
this technique are N2O, NO, NO2, H2O, H2O2, O3, NH3, HCN, SO2, SO3, OCS, CO2, CO,
HCI, HBr, HF, CH3CI, CH4, CH3OH, and C2H5OH, to name a few.

This technique requires the use of sophisticated electronic components, but provides an
extremely rugged, simple to operate, stable, sensitive, and reliable instrument.  This paper
describes how the Air and Energy Engineering Research Laboratory of the Environmental
Protection Agency at Research Triangle Park, NC, designed, built, and tested, with  a coal
burning furnace, a TDIR monitor for N2O. The present diode mount is limited to the
simultaneous use of only two 2 diodes and therefore only two analyte gases per optical cell.
Newer mounts allow the simultaneous use of four diodes.  The conversion  of the system for
other gases will be described. TDIR in-stack monitoring and long-range atmospheric
monitoring will also be discussed.
                                     7A-81

-------
                                   INTRODUCTION

The measurement of atmospheric N2O and its sources is of great interest since it is a
potential contributor to global warming and its atmospheric concentration is increasing.  The
principal sampling method uses an evacuated container to collect a grab sample of the gas
stream of interest, so the containers could be taken back to a laboratory and analyzed later.
The original data indicated a linear relationship between the concentrations of N2O and NOx
in the stack gases.  The validity of this data began to be questioned in the mid-1980s when
studies showed the  detection of N2O when none was expected.  Muzio et al. reported on the
formation of N2O as a sampling artifact while studying natural gas flames injected with SO2
and NH3. (1)  Another report showed that the artifact could be reduced by drying the gas
before sampling, and the artifact could be effectively eliminated by removing the SO2. (2)  It
was evident that a gas-phase aqueous reaction between SO2, NOX, and H2O was generating
N2O in the  sample container.  These reactions have been known since the 18th century and
reported as early as 1924. ^ Discovery of this sampling artifact led to research on the
development  of sampling and analysis techniques which would provide accurate results.
One project in this area, by the Air and Energy Engineering Research Laboratory,  used a
heated sample line and then filtered and desiccated the gas before it was analyzed by an on-
line GC/ECD  (for N2O) and continuous emission monitors (for O2, CO2, CO, and NO).  This
research indicated that the N2O concentrations were less than 5 ppm and were not a
function of  the NOX  concentration. ^

This project was undertaken to demonstrate the ability of a laser diode system to accurately
and correctly  measure the concentration of N20 in stack gas in real time, and to verify the
lower N2O  concentrations reported with modified sampling methods.
                                       7A-82

-------
                                     EXPERIMENT

The detection of trace gases using second-derivative spectroscopy was first used in  1978 by
Reid et al. at McMaster University. ^4^ Second-derivative or modulation spectroscopy
consists of using a modulated source to scan the absorption line of interest.  The detector
output is amplified using a phase-sensitive amplifier referenced to twice the modulation rate.
In addition to significantly reducing the background noise by rejecting all signals which are
not in phase with the reference  signal, operating the amplifier at twice the modulation rate
produces a pseudo-second-derivative signal as the output.  This signal is proportional to the
absorption of the line being scanned but the signal must be calibrated for each line of
interest.  A beamsplitter, lock cell, and a second detector are used to provide a feedback
signal to correct for any drift in the source.  For the feedback circuit, a phase-sensitive
amplifier referenced to the modulation frequency reduces the noise level and provides the
stabilization signal. ^4'  In this system, an infrared diode laser modulated at 2000 Hz  was
used as the source. The output frequency of a diode laser can be broadly tuned by adjusting
its operating temperature and finely tuned by varying the applied current.  This particular
diode is tunable over the range  2200 to 2215 cm"1.  Figure 1 diagrams the optical system.
The cold head contains part of the cooling system for the diode and also provides an
insulating vacuum for the diode since it is operated at 28 K.  A monochrometer is used to
isolate the laser line of interest. The beamsplitter deflects  a portion of the laser light through
a lock-cell containing a high concentration of N2O, and into a detector to generate the
stabilization signal.  The rest of the laser energy passes through the beamsplitter, into the
analytical cell, and then into the analytical detector.  The analytical cell is a two-pass, 0.5 m
cell with an external retroreflector.  Both detectors are single element mercury cadmium
telluride photoconductive detectors with low noise preamplifiers. The first and second
derivative signals are generated by setting the reference channel of the phase-sensitive
amplifier to either the "f" (first derivative) or "2f" (second derivative) mode.  In the "2f" mode,
the reference channel of the phase-sensitive amplifier is driven internally at twice the input
frequency, eliminating the need for an external, stabilized 4000 Hz reference signal.  The
output signal from the amplifier  (for either mode) is a pure  DC signal  reflecting the magnitude

                                        7A-83

-------
of the input signal which is in phase with the reference signal. Any AC component is the
result of noise in the system and is reduced by the AC filter at the output.  This AC filter has
a variable time constant which can be adjusted from 1 ms to  100 sec.  A higher setting of the
time constant will reduce the noise level, but will  also eliminate the corresponding time
variations in the analytical detector signal.  The output signal is then displayed on the chart
recorder.  The change in the magnitude of the signal, as measured from the baseline
(determined using dry nitrogen gas), is directly proportional to the concentration of N2O in the
analytical  cell.

Before beginning the tests,  the N2O line with the least interference from the other gases
commonly found in stack gas (H2O, CO2,  CO) at various pressures, temperatures, and
concentrations was determined. Theoretical spectra were calculated using the FASCODE
algorithm  which was developed by the Air Force Geophysics Laboratory. (5) Examples of
these spectra are shown in Figures 2, 3, and 4.  During this work, the gas pressure in the
analytical cell was maintained at 5000 Pa by continuously pumping on the outlet side of the
cell with a vacuum pump and limiting the flow at the cell inlet port. This kept the pressure-
broadening of the lines to a minimum and, during sampling of furnace gases, cooled the
furnace gases to reduce thermal-line-broadening. The line at 2208.75 cm"1 was chosen for
this work.  Initial tests using mixed gases from cylinders verified the detection of N2O and no
response to the CO2, CO,  SO2, and H2O vapor.

The equipment was moved from the laboratory and connected to the Innovative Furnace
Reactor (IFR), a furnace designed to evaluate various methods of scrubbing stack gases.  It
is a down-fired, tunnel-fired furnace burning powdered coal.  Figure 5 diagrams the system .
During these tests, the IFR was being  used to evaluate the  efficiency of powdered lime to
reduce SO2 emissions.  The stack gases were sampled at two different positions (see Tables
 1 and 2), one at the end of the furnace before the gas  is filtered in the bag house, and the
other near the roof just before the gases were vented to the  atmosphere.  These are
indicated in Figure 5 as #1  and #2, respectively.  The gases at the two sampling positions
are significantly different. At position # 1, the gases reflect the actual combustion products of
the furnace.  After leaving  the furnace, the gases are diluted and cooled to protect the bag
house filter elements and the roof-mounted blower from damage  due to excessive heat.
Although  the gases sampled at position # 2 reflect what is discharged to the atmosphere, the
gases have been diluted, cooled, and filtered.

                                        7A-84

-------
The operating parameters of the TDIR system are listed in Table  1.

These operating parameters are typical for each sampling position, but the actual values
were adjusted slightly to optimize the system  each day. The system was calibrated each day
using N2O in dry nitrogen at concentrations of 0.108, 0.514, 0.970, 1.99, and 4.82 ppm. A
sample of the data collected from sample position #2 is in Figure  6.  This section of the chart
paper shows the time variations in the N2O concentration  which is attributed to fluctuations in
the coal feed rate.  Also visible are the areas where dry nitrogen is used to verify the
baseline. The addition  of powdered lime to the stack had no effect on the measured N2O
concentrations.  The  average concentration is 0.9 ppm with a maximum excursion of 1.0 ppm
and a minimum of 0.8 ppm. Figure 7 shows data collected at sample position #1. There are
several differences evident in this chart. First, the level of N2O is much lower, about 0.3
ppm.  Second, as the system is switched from sampling dry nitrogen to stack gas, there is a
spike in the  N2O concentration which  is not seen in the data  from position # 2. Third, the
two spikes at the end of the trace are observed each time the coal feed is stopped and only
air is blown  into the burner section of the furnace.

                                  INTERPRETATION

The calibration data were fitted using a linear function to correlate a given deviation from the
baseline to concentration.  The results are summarized in Table 2. These concentrations are
much lower  than those  in the original  N2O database and are  also lower than the more  recent
data indicated.  The higher concentrations in the  stack at position #2 are caused by the
formation of  N2O in the baghouse. The concentration is reduced, by dilution of the gas
stream in the baghouse and after the  baghouse, to cool the gas before it  is vented. The
data from position #1  is a more accurate measure of N2O produced by the furnace since it is
sampled before there is any chance of dilution and the gas temperature (300 C) is high
enough  to keep the water in vapor form.  It is assumed that both the higher temperature and
the reduced  time between sampling and analysis work to reduce the amount of N2O
generated as an artifact.

The spikes in the data from position #1 are the result of N2O generation in the filter unit and
the short section pipe connecting the heated sample line to the furnace. The filter and
connector pipe were  not heated and would cool off when the furnace gases were not flowing

                                       7A-85

-------
through them.  This permits water condensation and the formation of N2O in these unheated
parts.  When gas was subsequently drawn from the furnace, the small volume of gas in the
pipe and  filter would precede the hot furnace gases into the analytical cell and cause a spike
in the output. The  fact that this effect was not observed in the data from position #2
indicates that the gas components had already interacted, producing N2O, and could not
generate more N2O in the filter set.  It is assumed that this reaction most likely took place in
the baghouse where the ash and lime reaction products were collected and the temperature
fell below 100 C causing the water vapor to condense and initiate the reaction.

The fluctuations  in  the N2O concentration both during furnace operation and at the end,
when the coal feed unit was turned off, were well correlated to similar fluctuations in the
concentration of  CO which was continuously measured as part of the SO2 scrubbing tests.
This may indicate that the  N2O is a result of a lower concentration of oxygen in the furnace
which also generates more CO.
                                   CONCLUSIONS

 In this study, it was found that the N2O concentration, immediately after the combustor
 (position #1, Figure 5) varied above  and below ambient which was measured at 280 ppb.
 However, conditions in the baghouse caused an increase of N2O  up to about 3 times
 ambient  (position #2, Figure 5).  The major source of N2O in the stack gas appears to  be its
 formation when  the water vapor condenses and reacts with other components of the stack
 gases.

 This study also shows great promise for the use of laser diode modulation spectroscopy for
 other applications  where continuous monitoring of one or more trace gases is required. The
 system is easily modified to monitor other gases by  replacing the  diode with one that will
 operate  in the spectral region of interest. By operating both diodes simultaneously and
 adding more optical components, the current system can be configured to simultaneously
 monitor two gases in the sample stream. There are also cold head systems available  which
 will allow the use of four diodes simultaneously and  therefore the  monitoring of four distinct
 trace gases.
 This method may also be used to directly measure species concentrations in the stack by
 using optical windows mounted in the stack access ports (e.g., the sulfun'c acid
 measurements of Pearson and Mantz. (6)).  Measurements of atmospheric contaminants over

                                        7A-86

-------
 long path lengths are feasible and could provide significant information on the generation,
distribution, and dissipation of pollutants which are not generated from single sources.  It is
proposed to use this technique to monitor methane emissions from landfills or pasture land.
                                        7A-87

-------
REFERENCES



(1)    L.J. Muzio, et al. "Errors in Grab Sample Measurements of N2O from Combustion



      Sources." JAPCA, Vol. 39 No. 3, 1989, pages 287-293.



(2)    L.J. Muzio and J.C. Kramlich. "An Artifact in the Measurement of N2O from



      Combustion Sources." Geophysical Research Letters, Vol.  15 No. 12, 1988, pages



      1369-1372



(3)    W.P.  Linak,  et al. "N2O Emissions from Fossil Fuel Combustion." In Proceedings:



      1989 Symposium on Stationary Combustion NOX Control, San Francisco, CA, March



      6-9, 1989, Volume  1, EPA-600/9-89-062a (NTIS) PB89-220529), June 1989.



(4)    R.S. Eng, et al.  "Tunable Diode Laser Spectroscopy: An Invited Review."  Optical




      Engineering, Vol. 19 No. 6, pages 952-953



(5)    FASCODE  Fast Atmospheric Signature Code (Spectral Transmittance and



      Radiance), H.J.P. Smith et al., AFGL-TR-78-0081 Air Force Geophysics Laboratory,



      Air Force Systems Command, United States Air Force, Hanscom AFB,  MA 01731.



(6)    E.F. Pearson, A.W. Mantz. "A Tunable Diode Laser Stack Monitor for Sulfuric -Acid



      Vapor." EPA-600/Z-80-174 (NTIS PB80-202 690), 1979.
                                    7A-88

-------
Q.



if




O

>
                                    Chart Recorder
                                   Lock-in Amplifier

                                   for Signal Analysis
                                   Lock-in  Amplifier

                                 for Reference Analysis
                                      Laser
                                      Control
                                      Module
                                     Oscilloscope
Figure 1. Laser diode setup
        7A-89

-------
-vl
CD
O
                   2200
2201
                                         N20  1 PPM,  50 MB,  25  C
'9 8.0
7.0
6.0
I 5.0-
Q
^ 4.0-

O 3.0-
2.0-
1.0-
0.0-















n






11






I .11 .














I , 1
i  i  i  i < i 
2202       2203       2204

       WAVENUMBER (CM'1)
    Figure 2. N2O Spectrum at 25 C
2205
2206

-------
                                        N20 PPM, 50MB, 100 C
CO
? -j
6.0-
5.0-
~T~
& *o-
Q
S 
E 3--
0 :
2.0-
1.0 -i
0.0-
22







A ^^^^












111 . ll ll















I
L . ill Jv. i A Jv i
DO 2201 2202 2203 2204 22






D5
                                              WAVFNUMBER (CM'1)
                                           Figure 3. N2O Spectrum at 100 C

-------
CD
                        o
                           10.0
                                          N2Q. OQZ M20, CO AT 25 C, 50 MB
                            2208      2208.2
22084     2208.6      22088
    WAVENUMBER (CM'1)
                                                                                                Legend
                                                                                                10X002
                                                                                                10  too
                                                                                                1000 PPM CO
                                                                               2209      2209.2
                                                 Figure 4. Combined Spectrum

-------
                             Coal Feeder
CD
CO
                  Sorbent Feeder
                 Sorbent/Slurry

                 Injection Probe
                  Sampling Ports
                                                                                     Rool


                                                                 N2O Sampling Port #2 
                                                          SO? Sampling Port
                                                                                        o
                                                                                        Q.


                                                                                        I
                                                                                        Q.
                                                                                        o
                                                                                         OJ
                                                                                   O
                                                                             CEM Sampling Port  I
                                                                                                           Fan
                                                                                                   Stack
                                                                                                      1
                                                                                                                                        Baghouse
                                                              Figure 5.  Innovative Furnace Reactor

-------
CD
                                                   o
                                                  At
                                                   Baseline'
                                              Figure 6. Position #2 N2O Data

-------
CD
cn
                                                                                                                                    I I
                                                                                                                       N2 Baseline
                                                                     Figure 7. Position #1 N2O Data

-------
                      TABLE 1.  OPERATING PARAMETERS
Parameter
Current, mA
Temperature, K
Frequency, Hz
Scan Width", mA
Sensitivity, mV
Time Constant, sec
Sample Position #1
217
28
2000
5
0.01
3
Sample Position #2
217
28
2000
5
0.025
3
* A current scan width of 5 mA equates to a frequency shift of 0.75 cm"1
                  TABLE 2. OBSERVED N2O CONCENTRATION
Data                     Sample Position #1           Sample Position #2
                               ppm                        ppm

Average                         0.30                        0.74
Maximum                       0.46                        1.27
Minimum                        0.14                        0.75
                             ( 0.053)                   ( 0.025)
                                   7A-96

-------
     Session 7B



NEW DEVELOPMENTS








 Chair: C. Miller, EPA

-------
     IN-FURNACE LOW NOX SOLUTIONS
        FOR WALL FIRED BOILERS
                  By
R.C. LaFlesh, D. Hart, and P. Jennings
      ABB Combustion Engineering
            Michael  Darroch
      City  of  Jamestown,  New York

-------
                                  ABSTRACT

Since the early 1940's, several thousand Type R pulverized coal burners have
been installed and are operating in wall fired boilers ranging up to 160 MWe
in size.  In response to the low NOX Emission requirements,  ABB Combustion
Engineering Services, Inc. has undertaken development of the RO-II coal
burner based on proven Type R wall firing technology with additional NOX
control capability.

Extensive laboratory tests were conducted at a large pilot scale (50 x 10
Btu/hr) in order to optimize the RO-II coal burner configuration.
Specifically, a number of coal nozzle/air register configurations were
evaluated in terms of their combined ability to meet specific emissions and
operational performance criteria.  The RO-II burner reduced NOX from a
baseline uncontrolled level of 0.9 #/106 Btu to 0.5 #/106 Btu during the
laboratory trials.

This paper will review laboratory development activities and report on RO-II
field demonstrations currently in progress.

Background

As a result of the recent Clean Air Act and specific local regulations,
boiler operators are addressing the need to reduce stack gas emissions.
Current attention is focused upon controlling acid rain precursors, oxides of
nitrogen (NOX) and sulfur dioxide (SO,,).  Under Phase I of the Clean Air Act,
a number of pre NSPS   coal burning wall fired boilers will be required to
reduce their NOX emissions by the mid 1990's.  The proposed Federal upper
limit for NOX emissions from wall  fired units is 0.50 Ib/MBtu fired.

ABB Combustion Engineering Services, Inc. (ABB-CE) has been actively
developing and commercially demonstrating low NOX technologies for coal fired
tangential and cyclone boiler arrangements.  In order to meet the NOX
reduction needs of coal wall fired boilers, ABB-CE has embarked on  an
extensive low NOX coal  burner development and commercial  demonstration
program building on its substantial wall fired experience base with the
ABB-CE Type R burner.

The Type R horizontal burner was developed by Combustion Engineering  Inc.  in
the early 1940's.  This burner was designed to burn pulverized coal,  oil,  or

                                    7B-1

-------
gas, is applicable to single wall  or opposed wall  firing in either single or
multiple burner arrangements.  In  terms of experience,  several  thousand Type
R burners have been installed and  operated in a wide variety of boiler
configurations ranging up to 160 MWe in capacity.   Individual  burner
capacities have ranged from 20 MBtu/hr to 120 MBtu/hr.   As a result of this
extensive field experience, ABB-CE has established Type R design standards
which delineate proven materials of construction and fabrication techniques,
Type R operating procedures are also firmly established.

The Type R coal burner, illustrated in Figure 1, has several key hardware
features.  The centrally located coal nozzle has spiral rifling along the
inner walls to promote swirl of the pulverized coal/primary air stream which
is  initially established by a tangential inlet nozzle.   A convergent nozzle
tip is located at the end of the coal nozzle.  Five (5) deflector vanes,
located near the tangential inlet nozzle, can be adjusted  in terms of
incident angle to vary coal/primary air stream swirl which in turn,
influences final luminous flame shape.  On the combustion  air side, the total
combustion air flow passes through an adjustable angle  flat blade swirler
assembly. Combustion air angular momentum can be varied to optimize the
burner's flame stabilizing aerodynamic recirculation zone, directly
influencing both flame stability and flame shape.

Laboratory Development Program

 In  order to respond to low NOX requirements for wall fired-coal  boiler
retrofit market, ABB-CE embarked on a laboratory development program with  the
objective  of  developing a  new low NOX wall fired burner product.
The new  burner, named  the  RO-II burner, would be capable of meeting the
 following  performance  targets:

          NOX  less than 0.5#/106 Btu Fired
          Zero or  nominal  increase  in carbon loss and/or CO emissions  under
           low NOX conditions.
          Acceptable flame envelope  (length).
      t    Zero or  nominal  increase  in fuel system or combustion air windbox
           static pressure(s).

At  the onset  of the development program, ABB-CE assessed the NOX reduction
potential  of  the Type  R burner design; upon  review  it was  decided to
                                    7B-2

-------
incorporate certain key design features of the Type R design into the new
RO-II burner design.   These features specifically included the tangential
inlet,  spirally rifled coal nozzle and an adjustable coal  stream deflector
vane assembly.   The Type R combustion air register assembly was determined to
not offer any advantages in terms of reducing total NOX  so alternative  air
register assemblies were reviewed for incorporation into the new low NOX  RO-
II burner design.

ABB-CE selected a  patented, commercially available, air  register for
incorporation into the RO-II burner.  Key features of the register are
highlighted in Figure 2.  These features include:
     1.   Two separate plenums which permit staged introduction of combustion
          air.
                 pilot air which is introduced concentrically adjacent to the
                 centrally located coal nozzle
                 main air which surrounds the pilot air stream
     2.   Involute (spirally shaped) air inlets for each plenum which swirl
          total combustion air flow.
     3.   Separate flow control dampers for both the pilot and main air
          streams.
     4.   Integral instrumentation which permits burner operators to balance
          combustion air flow to multiple burner arrays located within a
                    common windbox.
     5.   Unique helical flow vane assembly which enhances combustion air
          swirl and improves air distribution within the register.
     6.   A shadow vane assembly which enhances combustion air swirl but more
          importantly protects the flow vane assembly and fuel nozzle from
          damage due to flame radiation in multiple burner installations.

Photo 1, an end-on view of the RO-II register assembly,  highlights the
involute (spirally shaped) air plenum, for both the pilot and main combustion
air  streams, and the shadow vane assembly.  Photo 2 highlights the flow  vane
assembly utilized in the RO-II register.  The helical vane arrangement is
shown separate from the air register.  Note that the pilot combustion air
stream passes through six  (6) vanes at the rear of the burner  (i.e. the
widest part of the vane assembly), the main combustion air stream passes
through eight (8)  vanes near the burner front (i.e. the narrowest part of  the
vane assembly).  It should also be noted that the register design requires
minimal maintenance since the only moving parts are the pilot  and main air

                                    7B-3

-------
dampers.   These same dampers also provide the register with the ability to
compensate for burner to burner combustion air flow imbalances in multiple
burner/common windbox arrangements.

The RO-II development program was largely comprised of extensive combustion
trials of potential RO-II firing system hardware.  These trials were
conducted in one of ABB-CE's front wall fired large scale laboratory test
furnaces.  ABB-CE's development philosophy was to conduct tests with hardware
designed to operate at a heat input rate of 50 x 106 BTU/HR.   This rate is
identical to the design heat input rate of the burners to be installed in two
units in Jamestown, NY.  By adopting this development philosophy, ABB-CE
could confidently  accelerate the process of transitioning laboratory hardware
developments into  commercial application.

Prior to conducting the laboratory combustion trials, ABB-CE evaluated the
air register's near-field aerodynamics.   The objective of these tests was to
define key aerodynamic characteristics of the register in order to support
the design of compatible coal nozzle configurations.  Recirculation zone size
and strength as well as the air register's potential to control stoichiometry
in the burner near field (through internal air staging) were assessed.  These
aerodynamic properties were consistent with the low NOX objectives of the RO-
II development program.

Laboratory combustion trials began following the register aerodynamic study.
The focus of these trials was to evaluate the combustion performance of a
variety of air register/coal nozzle configurations.  The performance of each
configuration was  evaluated in comparison with the overall performance
targets for the RO-II burner.  It should be noted that the air register
configuration remained fixed throughout the trials.  Development activities
concentrated on combining advanced low NOX Type R coal  nozzle arrangements
with the existing  air register design.

The combustion trials generated the data necessary to assess RO-II burner
performance.  Flue gas 02,  NOX,  and  CO  concentrations were measured  at  each
test condition, along with coal/primary air static pressure at the coal
nozzle inlet, windbox and furnace static pressures, and total combustion air
and primary air mass flows.  Qualitative assessments of flame shape, length,
and stability were also made throughout the development program.  In
addition, flyash samples were taken for subsequent carbon in ash analysis.
                                    7B-4

-------
Furnace horizontal exit gas temperatures were also quantified using suction
pyrometry.

The combustion test program parametrically evaluated a number of key RO-II
design and  operating variables.  Some of the variables investigated included
coal nozzle/tip configurations, firing rate (MCR and reduced load), excess
air level,  coal/primary air velocity at the coal nozzle tip exit, pilot and
main air damper position (pilot/main air flow split) and coal stream swirl.

All laboratory trials were conducted with a Pennsylvania bituminous coal
having 10% ash, with a fixed carbon to volatile ratio of 1.65 and a fuel
nitrogen content of 1.5% by weight.  Coal preparation for the laboratory
tests was consistent with typical utility practice; the pulverized coal grind
averaged 70.3% through 200 mesh (75 microns), with 0.6% remaining on a 50
mesh (300 microns) screen.

The laboratory test furnace utilizes a dilute phase (1.5   2.0 # primary
air/# coal) indirect coal feed system.  A schematic of the feed system is
shown in Figure 3.  Figure 3 highlights the fact that a gravimetric feeder is
employed to accurately quantify coal feed rate.  The figure also illustrates
the location of static pressure taps in the coal feed system.  These
pressures were documented throughout the test program for comparison to
performance targets.

Photo 3 shows the installed RO-II Burner register as viewed from outside the
furnace.  Note the use of the tangential entry fuel nozzle inlet,
characteristic of both the Type R and RO-II burner designs.  Photo 4 shows
the installed RO-II from the furnace side and highlights the shadow vanes and
divergent refractory quarl similar to typical field installations.

Note also in Photo 4 that there is refractory material on the furnace walls.
The laboratory test furnace has atmospheric pressure water cooled walls.  The
furnace gas temperatures and heat release profile are adjusted by altering
the refractory configuration depending upon test objectives.  The refractory
configuration selected for these trials was chosen to create a furnace
thermal environment where relatively high levels of thermal NOX would be
generated.   In addition to refractory modifications, the test furnace was
intentionally operated at a volumetric cubic heat release rate of 39,800
Btu/hr/ft3.   This  volumetric heat release rate in effect far exceeded a
                                   7B-5

-------
typical  coal-designed boiler's volumetric heat release range of 9,000-16,000
Btu/hr/ft3.   As  a result  of this  (and  combined with  the  refractory insulation
thickness and pattern in  the furnace), measured furnace  gas outlet
temperatures  (horizontal  furnace outlet plane) were in the 2500   2700F
range, far exceeding typical boiler horizontal furnace gas outlet
temperatures  of 1900   2000F.   The implication of high  temperature furnace
operation during the RO-II laboratory trials is that NOX generated thermally
via the Zeldovich mechanism (1) was projected to be conservatively higher
than would be expected in subsequent field RO-II installations.

Eleven different coal nozzle configurations were evaluated during the
combustion trials.  Baseline tests were conducted with a conventional Type R
nozzle; ten advanced Type R nozzle configurations were also evaluated.  The
baseline nozzle  (Figure 4) was comprised of the tangential fuel inlet, coal
stream deflector vanes, and a spirally rifled nozzle with a convergent tip.
A  furnace side view of the baseline Type R coal nozzle is shown in Photo 5.

Combustion test data from the "Baseline" RO-II configuration is shown in
Figure 5 which depicts NOX (ppm corrected to 3% 02)  as a  function  of  flue  gas
02 concentration.  As is  characteristic of a diffusion flame burner,  NOX
increases with increasing excess air level.  The primary point of the figure
is that at a nominal excess air level  of 20% (approx. 3.5% 02), measured NOX
was approximately 650ppm  (approx. 0.9 #/MBtu).  Under all excess air
conditions, NOX exceeded  the target value of 0.5 #/MBtu.

The most optimum coal nozzle arrangement of the ten tested is  shown in
Figure 6.  As shown in the schematic,  the optimum RO-II  coal nozzle retains
the tangential fuel/primary air inlet, deflector vane assembly, and spirally
rifled nozzle of the Type R design.  The optimum RO-II arrangement includes
the addition of a venturi diffuser assembly, which is a channeled flow
control device, and a convergent nozzle tip with axial rifling vs. spiral
rifling as in the baseline case.

Photo 6 is a "furnace side" view of the optimum nozzle arrangement.

Figure 7 graphically depicts the NOX emission performance of a number of the
tested RO-II  coal nozzle  concepts.  Data in this figure highlights the  fact
that the coal nozzle design employed had a dominant influence  on NOX levels
observed.  One can summarize the data contained in Figure 7 by directing
                                    7B-6

-------
attention to the solid line plotted in the center of the graph.  All data
below the solid line represents the NOX performance of the venturi  diffuser
concept, all data above the line represents alternative tested concepts.
Clearly, the venturi diffuser concept generated lower total NOX at  any given
operating excess air level, as compared with all other tested coal  nozzle
concepts.  Most importantly, at a nominal flue gas 02 concentration of 3.5%
(20% excess air), total measured NOX was 360 ppm (corrected to 3% 02),
meeting the overall project goal of 0.5 #/MBtu NOX.  The venturi  diffuser
coal nozzle assembly, as a result of its success in meeting the NOX reduction
target established for the project, has been chosen as the coal nozzle design
to be utilized in the RO-II burner.

Beyond its NOX reduction capability, the RO-II burner met all  other
established performance targets.  These targets were set to ensure that the
firing system hardware developed in the laboratory would be retrofitable to
most existing wall fired boiler arrangements.  Most units, for example, have
fan limitations in terms of achievable windbox to furnace delta static
pressure.  The RO-II coal burner is capable of operation at less than 3.0"
W.C. static windbox to furnace delta pressure at MCR.  Most existing boiler
F.D. fan systems are capable of achieving at least that pressure differential
at MCR.  In a similar vein, primary (coal transport) air static pressure at
the coal nozzle inlet is a critical factor from a retrofit standpoint.  Any
low NOX burner installation should  operate within existing coal feed system
pressure limitations.  The RO-II burner operated at MCR with a primary air
static pressure at the nozzle inlet of less than 4.5 inches W.C., an
acceptable operating primary air static pressure for most existing wall fired
installations.

Many low NOX coal  firing system laboratory tests and field demonstrations to
date have reported that, under low NOX conditions,  carbon in fly ash levels
tend to increase (2, 3, 4).  In some cases, CO emissions also  increase under
low NOX conditions.  These results  are, of course,  very dependent on coal
type, coal  particle size distribution, and furnace configuration.  In
practical terms, most low NOX coal  firing systems must strike an acceptable
balance between NOX reductions and  carbon in fly ash/CO increases.   In the
case of the RO-II coal burner, operated at 0.5 #/106 Btu, both carbon in fly
ash and CO emissions did increase,  however, the final emission levels
documented were within acceptable operating ranges.  For example, under
baseline, high NOX conditions,  carbon  in fly ash and CO were 1-2% and 30-
                                    7B-7

-------
BOppm, respectively.  Under low (0.5 # MBtu) NOX conditions,  carbon in fly
ash and CO increased to 3% and 40-70 ppm, respectively.  These laboratory
results indicate that nominal increases in carbon in flyash may be expected
in RO-II field applications, again dependent on coal type and furnace
configuration.

Several low NOX coal firing systems evaluated to date for wall  fired boiler
applications have experienced increased flame lengths as compared to pre-
retrofit cases (5,6). As in the case of the relationship between NOX,  carbon
loss, and CO, one must in most situations strike a balance between NOX
reductions and increasing flame length.  Operating experience with the RO-II
coal  burner to date is good in this regard.  Baseline (high NOX)  conditions
produced a luminous, stable flame about 12' long.  Under low NOX  (0.5 #/MBtu)
conditions, flame length increased to approximately 16'-18' long.  The
increase in flame length was deemed acceptable because since the field units
targeted for the first RO-II coal demonstrations can accommodate a similar
increase in flame length without direct flame impingement on rear wall tube
surfaces.  Future boiler retrofits will be assessed on an individual basis
not only to ensure  compatibility between furnace depth and the luminous flame
volume of the RO-II low NOX coal  burner,  but also to ascertain  potential  for
changes in post-retrofit boiler thermal performance.

Field Experience

Following successful laboratory development trials, the RO-II coal burner has
presently been retrofitted to three (3) field installations.  Figure 8 is a
schematic of the as-installed RO-II coal  burner.  The tangential  inlet,
spirally rifled coal nozzle with venturi  diffuser assembly and convergent
nozzle tip can be seen in the figure.  The pilot and main air plenums,
helical flow vanes, and shadow vanes are also depicted.


The current RO-II field installations are listed in Figure 9 with other
pertinent information.  City of Jamestown Unit 10 and BPU Kansas City are
currently undergoing start-up and demonstration testing.
Conclusions
ABB-CE's RO-II coal burner, specifically designed for retrofit wall fired
                                    7B-8

-------
boiler applications, has undergone extensive laboratory testing and is now
commercially available.  In laboratory trials, the burner was shown to meet
the NOX target  of 0.5  #/MBtu  firing  Eastern  U.S.  bituminous  coal  while
limiting increases in  carbon  loss, CO, and flame length to commercially
acceptable levels.  The burner also demonstrated the ability to operate
within the capacity of most existing boiler combustion air fan and coal
delivery systems in terms of  static pressure requirements.  The RO-II burner
offers advantages in terms of its simplified construction and operation.  In
addition, the RO-II burner has the ability (via adjusting the main/pilot air
damper system)  to equalize burner to burner combustion air flow imbalances in
multiple burner/common windbox plenum arrangements.

                                  References

1.   Zeldovich, Y. et  al. (1947), Oxidization of Nitrogen in Combustion,
     Academy of Sciences of the USSR, Institute of Chemical  Physics,
     Moscow-Leningrad, Translated by M.  Shelf, Scientific Research
     Staff, Ford Motor Co.

2.   Beard, P.  et al "Reduction of NOX Emissions  form a 500  MW Front
     Wall Fired Boiler" 1989  Joint EPA/EPRI Symposium on Stationary
     Combustion NOX Control.

3.   Grusha, J. and McCartney M., "Development and Evolution of the ABB
     Combustion Engineering Low NOX Concentric Firing System   1991
     Joint EPA/EPRI Symposium on Stationary Combustion NOX Control.

4.   Kinoshita, et al  "New Approach to NOX Control  Optimization and
     Unburnt Carbon Losses"   1989 Joint EPA/EPRI Symposium on
     Stationary Combustion NOX Control.

5.   Clark, M.J. et al "Large Scale Testing and Development of the B&W
     Low NOX Cell  Burner"   1987 EPA/EPRI  Symposium on Stationary
     Combustion Nitrogen Oxide Control.

6.   LaRue, A.  et al "Development Status of B&W's Second Generation Low
     NOX Burners   The XCL Burner"   1987  EPA/EPRI Symposium on
     Stationary Combustion Nitrogen Oxide Control.
                                    7B-9

-------
          Figure 1: Type R Coal Burner
Photo 1: End-On View of the RO-II Register Assembly
                                                   Figure 2: Exploded View of RO-II Burner Assembly
                                                        Photo 2: Helical Flow Vane Assembly
                                                        Figure 3: Coal Feed System Schematic
                                            7B-10

-------
 Photo 3: Installed RO-II Burner Register as Viewed
             from Outside the Furnace
                                      TANGENTWL
                                     FUELPFB MARY AIR
                                        INLET
SPIRALLY-BIFLED TIP
                                        MI
                                          v
                                           DEFLECTOR
                                             VANES
Figure 4: "Baseline" Type R Coal Nozzle Schematic
                          2D%EA

                         02, %
 Figure 5  "Baseline" Nozzle Assembly, NOx vs. O
                                                        Photo 4: Installed RO-II Burner from the Furnace Side
                                            I
                                                         Photo 5: Furnace Side View of the "Baseline" Type R
                                                                            Coal Nozzle
                                                                                          INLET
                                                                                          'I
                                                                         ^- L-
                                                                                                 VEWTURI DtfFUSER
                                                                                                 ADJUSTMCWT BOOS
  STUFFWG BOX Fi
AXIAL NOZ2LE ADJUSTMENT
                                                                                    X    \
                                                                                   SPTRlALLY-fOOZD
                                                                                      NOZZLE
                               DEFLECTOR VANES
                                                                    NOZZLE TP WITH
                                                                    AXIAL RJFL**S
                                                           Figure 6: Venturi Diffuser Nozzle Assembly, Test
                                                                        Equipment Schematic
                                                  7B-11

-------
                                                                              02. %

                                                            At and Below the Line - Venturl Dlfluser Concept!
                                                            Above the Line - Other Tested Concepts
                                                       Figure 7: RO-II "Advanced Coal Nozzle Concepts"
                                                                         NOx vs. O,
Photo 6: "Furnace Side" View of the Optimum Nozzle
         Arrangement for the RO-II Burner
Customer Unit
Bd ol Public Utll 9
City of Jamestown
Bd of Public UtIL 10
City of Jameetown
Bd ot Public Utll. Oulndaro
Kansas City Unit 2

Unit
Type
CE-VU40

CE-VU40

Rlley


Steam Flow
Ib/hr
165.000

165.000

1.126.000


No. ol
Burners
4

4

9


Fuels
E. Bit

E. Bit

Sub-Bit.
Natural Gas
Propane
                                                                Figure 9: RO-II Experience List
             Figure 8: RO-II Burner
                                               7B-12

-------
 NOx  REDUCTION  ON  NATURAL  GAS-FIRED  BOILERS
USING FUEL INJECTION RECIRCULATION (FIR) -
          LABORATORY DEMONSTRATION
    Kevin C. Hopkins, David 0. Czerniak
                   Carnot
       15991  Red  Hill  Ave.,  Suite  110
           Tustin, CA 92680-7388

                 Les Radak
     Southern California Edison  Company
          2244 Walnut  Grove  Avenue
               P.O.  Box 800
             Rosemead,  CA  91770

               Cherif  Youssef
      Southern California Gas Company
         3216 North Rosemead Blvd.
            El Monte, CA  91731

               James  Nylander
      San  Diego Gas &  Electric Company
               P.O. Box 1831
            San Diego, CA 92112

-------
                     NOx REDUCTION ON NATURAL GAS-FIRED BOILERS
                    USING FUEL INJECTION RECIRCULATION  (FIR) -
                              LABORATORY DEMONSTRATION
ABSTRACT
Increasingly stringent NOx regulations on industrial  and  utility boilers  may  require
implementation of expensive post-combustion NOx control  techniques.   Fuel  Injection
Recirculation (FIR) is a relatively low cost NOx reduction strategy  for  natural-gas
fired boilers in which the  fuel  is  diluted prior  to combustion with  air,  steam,  or
flue gas.  This technique  is different from conventional flue  gas recirculation  (FGR)
because it is conceptually believed  to impact  prompt  as well  as  thermal NO  formation
mechanisms and is therefore capable  of greater NOx  reductions.   Furthermore,  the  two
technologies when applied in conjunction are additive is terms of NOx reduction.
As  a preliminary  step  towards  full  scale  implementation  of FIR,  a  laboratory
demonstration was performed to determine the feasibility  of the  technology.   FIR  was
demonstrated on a 2.0 MMBtu/hr test  facility designed  to simulate burners used  on full
scale utility boilers.    The  test   facility employed combustion  air preheat,   FGR,
staged-air firing,  and was modified  to inject  flue  gas, air,  or  saturated steam into
the  fuel  stream  prior to combustion.   The effectiveness of  FIR was determined  at
varying injection rates,  firing rates, air preheat levels, FGR  rates,  and  excess 02
conditions.
Results show  that FIR  is  more  effective  that  FGR  in  reducing  NOx,   and  that  a
additional 50% NOx reduction was achieved when FIR is  used in conjunction  with FGR.
The test program  demonstrated  that in a full-scale application,  FIR may be capable of
reducing NOx to low levels, at an attractive cost relative to  post-combustion  control
retrofits.
                                      7B-15

-------
INTRODUCTION
Carnot was contracted by the Southern California Gas Company, the Southern California
Edison Company  (SCE), and the San Diego Gas and Electric  Company (SDG&E)  to perform
a laboratory demonstration of a potential  new NOX  reduction  technology for gas-fired
boilers  which  has  been  designated  Fuel   Injection   Recirculation   (FIR).    As  a
preliminary  step  towards  full-scale  implementation,  this demonstration  program  was
performed to determine the feasibility of the technology.
Fuel  Injection  Recirculation involves recirculation of a  portion  of  the  boiler flue
gas  and  mixing  it  with  the   gas  fuel  at  some point  upstream  of  the  burner.
Additionally, the  FIR concept  can  be  expanded to  include the fuel injection  of  any
inert  diluent  such as  steam or air.    This  method conceptually  is  believed  to  be
capable  of  greater NOX  reductions  than can  be  achieved by  conventional  Flue  Gas
Recirculation  (FGR),  which  is  mixed with  the combustion air.   Furthermore,  it  is
anticipated that when implemented on a  utility boiler,  the two technologies would be
to some extent,  additive in  terms of NOX reductions, ultimately resulting in very low
NOX emissions.   The principal  motivation for  pursuing this  concept is  the potential
cost  benefit in comparison post-combustion NOX control technologies  such  as SCR  and
urea  injection, which are presently  being  considered  to  meet the  stringent new NOX
limits specified  in  the  South  Coast Air Quality  Management District  Rules  1135  and
1146.  The FIR concept is also attractive because full-scale application of FIR would
require relatively few modifications to existing equipment.
The approach taken  for  this laboratory demonstration  program was to  apply the  FIR
technology on  a test facility  which  incorporates many  key design  and  operational
attributes of  burners  in  use   on  utility  boilers,  and  which  employs  NOX control
techniques commonly used in  these large scale boilers.  The primary  emphasis of the
this feasibility  study  was  a  practical  evaluation of  FIR over ranges of important
operating conditions such as firing rate,  air preheat, overfire air,  and  FGR.
                                      7B-16

-------
TECHNICAL OBJECTIVES
Throughout this study,  FIR was evaluated primarily in terms of flue gas concentrations
of NOX,  02,  C02, and CO,  and  in  terms  of  burner  stability  and  flame  characteristics.
The specific technical objectives of the investigation were as follows:
          1.   Evaluate  the  NOX reduction  effectiveness  of  FIR using  a
               laboratory-scale  burner  similar  in   design  and thermal
               environment to burners used on electric utility boilers.
          2.   Evaluate the  NO  reduction efficiency of FIR alone,  and  in
               combination with FGR.
          3.   Evaluate the effect  of FIR on minimum operable 02 level,  and
               on burner stability.
          4.   Evaluate  the  effect   of  reduced  firing  rate  on   the
               effectiveness of FIR.
          5.   Evaluate the effect of air staging on the effectiveness  of
               FIR.
          6.   Compare the effect  of  air relative to  flue gas as the  FIR
               diluent.
          7.   Compare the effect of steam relative to  flue  gas as the  FIR
               diluent.

BACKGROUND
Fuel Injection Recirculation  (FIR)  is a potential new NOX control strategy for natural
gas-fired boilers  which is defined as the injection of any inert diluent into the  fuel
gas  at  some point  upstream  of  the  burner.   The concept  originally  involved  the
extraction  of  flue  gas from the exit  of the boiler,   cooling  it if necessary,  and
finally compressing  it for  injection  at  gas  header  pressures  into the fuel  line.
Operating expenses and  equipment costs may be reduced by injecting other diluents  such
as air or steam, or by lowering gas header pressures  through burner modifications.
FIR and  Prompt  NO  Formation:  NOX formation in natural  gas-fired boilers is associated
with two mechanisms known as  thermal NO and prompt NO.   Thermal NO refers to the  high
temperature reaction of nitrogen and oxygen from the combustion air.   This mechanism,
which is commonly termed the "Zeldovich" mechanism after its discoverer, is thought
to occur  in the post-flame  or  burned gas zone.   Low excess  air  firing,   flue  gas
recirculation,  burners-out-of-service  (BOOS), and  air  staging are commonly  used  on
utility  boilers to control  thermal  NO formation.
The existence of another  NO formation mechanism was first suggested by Fenimore whose
measurements showed that  reactions other than  the Zeldovich  mechanism  were  taking
place,  and that some NO was being formed in the flame  region.  Because  of  the early
                                      7B-17

-------
formation of NO, Fenimore coined the name "prompt"  NO.   Fenimore  proposed  that  C2 and
CH radicals present  in hydrocarbon  flames  contribute  to the formation of prompt NO.
The  formation  of prompt  NO is  greater  in  fuel-rich flames, and  decreases  with the
increase  in local 02 concentrations.  Similar experiments  have shown that prompt NO
formation  is   a  function  of  flame  temperature  as well  as  stoichiometry.   Other
measurements made in flat flame  burners demonstrate that prompt NO can account for 10-
40 ppm of the total  NO formed.   In  utility  boiler systems, prompt NO  is assumed to be
less  than  50  ppm while  the thermal  NO contribution can be  as  high  as 125-200  ppm.
Thermal NO control techniques such as FGR and BOOS can decrease NO to concentrations
approaching prompt NO concentrations.  The South Coast Air Quality Management District
Rule  1135  for  utility boilers will  require NOX emission limits translating  to  about
25 ppm, and therefore the control of prompt NO formation  is important  if new emissions
limits  are to  be met  without   installation  of  expensive  post-combustion  control
techniques.
FIR appears to be a  effective and relatively inexpensive technique for the control  of
prompt NO formation.  It  is believed  that FIR reduces prompt NO formation by diluting
the  fuel   prior  to   combustion  thereby  reducing  the  concentration   of  hydrocarbon
radicals  which  produce prompt NO.   In addition,  FIR also acts like  FGR  in  reducing
thermal NO production.   It is  anticipated  that  FIR in combination  with  FGR,  could
reduce  NOX emissions  to  levels  below  25 ppm  by controlling   both NO  formation
mechanisms.

TEST  DESCRIPTION
Test  Facility:  The  laboratory facility  selected for this evaluation  of FIR was  an  80
hp Scotch fire-tube  boiler. This boiler was modified to incorporate many  significant
components of  a  full-scale  utility boiler  furnace.  The test  facility comprised the
fire-tube boiler, which  is  capable of firing  up to  3.0  x  106 Btu/hr  on natural  gas,
a forced draft fan, a separately fired air preheater (APH), a 5  1/2" diameter gas fuel
ring, a ceramic quarl, and  a windbox with a  sixteen blade variable air register.   Off-
stoichiometric firing was achieved by diverting a portion of the pre-heated combustion
air to the overfire  air  (OFA)  ring  placed downstream of the  burner face.   A  separate
fan was used to  recirculate a portion  of the flue gas  to the  combustion  air  (FGR).
The FGR flowrate was determined  by  measuring the  windbox 02  concentration  along with
the flue gas 02  concentration.  The  mass  flowrate of  the flue gas recirculated was
subsequently determined  from stoichiometric calculations.
Natural gas was  supplied to the boiler  via a 10 psig  supply, and  metered using a
rotameter.  The maximum  firing  used  in  this study was  2.0  x 106  Btu/hr.   The burner
consisted  of 3/8  inch ring with 11  equally  spaced  holes  drilled radially, each  of
                                      7B-18

-------
0.189 inches diameter.   The  ceramic burner quarl,  six inches long with  a  nine-inch
exit diameter,  was geometrically similar to those used on small  Peabody ring burners
in utility boilers.  The  air register vanes were set  initially to target  a baseline NOX
level characteristic of full-scale units.   The air register vanes were set at 20  off
radial and were not varied throughout the remainder of the tests.
The FIR concept was tested using  three fuel diluents:  flue  gas,  air,  and  saturated
steam.  Most of the  testing was performed using flue gas as the diluent.  The flue  gas
injection system consisted of a 5 hp rotary lobe  type compressor capable of a delivery
pressure of up to 8 psig at  a flow rate of 30  scfm  of  flue gas.   Flue gas,  extracted
at the stack plenum, was  compressed  and  injected into a  2 inch fuel line  through  a
sparger.  FIR tests  with air  injection were performed using the same configuration as
above with the inlet to the blower disconnected from the stack plenum.
Steam injection  was accomplished  using  a separately  fired  2-1/2 hp  Parker  Boiler
providing saturated steam at approximately 180  psig.   The flow rate was controlled
using a gate-valve and was metered using  an Annubar flow  sensor.   Steam was injected
through the sparger into a heat-traced fuel line.
Test Conditions: The principal  objective  of this laboratory demonstration program  was
to determine the effectiveness  of FIR  in reducing NOX at conditions characteristic of
large industrial or utility boilers.  Conditions and  parameters which  significantly
impact NOX on full-scale units include combustion air temperatures,  off-stoichiometric
firing,  excess air levels, load variations, flue gas recirculation to the combustion
air, burner  configuration, and  air register orientation.   It was not practical  to
systematically  investigate the  influence  of  each  of  these  characteristics  in  the
laboratory  facility.    Once  baseline  configurations  were  established,  the  burner
hardware and the air register orientation were not  changed throughout the testing.
Excess Air  Levels:    Tests  were  performed at  a  "minimum"   or  "nominal"   excess  02
condition.   The minimum 02 condition was  defined by the following  criteria:
          1.   the excess air level producing  200   400 ppm CO,  or
          2.   an excess 02 concentration  of   0.3 %
The second criteria  was  necessary because at some test conditions, CO did not  exceed
100 ppm even  at extremely low  02  concentrations.   The  0.3  % 02 concentration  was
necessary as a  lower safety  limit for those tests where CO remained below  100 ppm.
The nominal  02 condition was defined as  the amount of excess air necessary to increase
the minimum 02  concentration  by  0.5 %.
Flame Characteristics:   Since  an  important objective of  this test  program is  to
determine  the limits of applicability of  FIR with respect to  flame characteristics,
                                      7B-19

-------
the test series involving fuel  dilution with flue gas, steam, or air, the diluent was
added  to the  point  of  flame instability.    Flame stability  and  general  flame
characteristics were determined primarily form observations.  The flame was considered
to be unstable if any of the following was observed:
          1.   Any tendency for the flame to lift-off from the burner face
               and re-attach downstream on the OFA ring.
          2.   Excessive fluctuations in furnace draft
          3.   Excessive fluctuations of NOX,  CO,  or 02  concentrations.
                                          'X'
RESULTS AND DISCUSSION
The results of the Fuel Injection Recirculation  (FIR)  test  program are presented in
this section.  The NOX results  presented  below  are  expressed in ppm corrected to 3%
02 on a dry basis.   The NOX  reductions  achievable,  and the  limitations  in  terms of
flame stability are considered for FIR used in  conjunction with varying firings rates,
flue gas recirculation rates, air preheat  levels, and both with, and without overfire
air.  For  each  test  series,  the injection rate of flue gas  was  increased  until  the
limit off  flame stability was reached.  The flame stability  limit  is defined as  the
maximum  injection rate at  which  the  flame  remains  attached to  the burner  face.
(Higher injection rates would cause the flame  to detach from the burner face and  re-
attach to  the overfire air ring).
For  the  purposes  of  later  comparison,  the "baseline"  condition  is defined  by  the
following  parameters:

               firing rate:         2,000,000 Btu/hr  2 %
               Op  condition:         minimum (defined by CO ~ 200-400 ppm)
               OFA condition:       nominal (defined by  10% of total air)
               APH temperature:      480   495 F
               Windbox FGR:         0 %
The baseline NOX concentration  for  this  test facility was  87.6 ppm @ 3% 02.   Without
OFA, the NOX  concentration  was  167.6  ppm @  3%  02.  The use of OFA reduced NOX by 48%.
This is consistent with full-scale NOX reductions attainable using NOX  ports and/or
burners-out-of-service  (BOOS).   The effects  of other parametric  variations  are
presented below.
Summary of Baseline Characteristics
              The baseline  NO   concentration  is 87.6 ppm  0  3%  02  with
               approximately  10%   overfire  air with   a  combustion  air
               temperature  of approximately 490 F.
              NOX is  very sensitive  both to  excess  air   level  and  to
               combustion  air temperatures, especially  at  lower  FGR rates.
                                      7B-20

-------
              The measured  NOX  vs  FGR relationship  is  typical of  full-
               scale units.
              The NO  vs firing rate relationship  is characteristic  only
               of smaller industrial boilers.
Flue Gas as FIR Diluent
The effect of Fuel  Injection  Recirculation using flue gas as the diluent is  presented
in this section.  The amount of FIR injection is expressed  in  two ways.  First  as  a
percent fuel  dilution defined as  the percentage of the volume of flue gas injected to
the  total  volume flow  through the  burners.   Alternatively,  for  the purposes of
comparison to conventional flue gas  recirculation,  it is  expressed  as  the percent of
the weight of the flue gas injected to the total  weight of the air and fuel.
FIR vs Windbox FGR:    The  effect of FIR without windbox flue gas recirculation  (FGR),
and at an  optimum and maximum FGR rates are presented in this section. The firing  rate
is 2.0 x  106  Btu/hr  with  nominal  OFA  at  the minimum 02 condition.   The  results are
shown in Figure 1-A and 1-B.
Figure 1-A shows NOX  concentration  vs  FIR  injection rate expressed as  percent  fuel
dilution.   NOX decreases  uniformly  with  increasing  FIR injection.   With no windbox
FGR, the rate of decrease  is  approximately  1.7  ppm per % fuel dilution.   At  higher
windbox FGR rates,  the rate of  decrease is approximately 0.6 ppm per % fuel  dilution.
The decreasing effectiveness  at higher windbox FGR rates indicates that FIR reductions
are partially thermal NOX reductions and that the two techniques are  to  some  extent
redundant.  However,  since further decreases are measured even at the maximum windbox
FGR rate,  the two techniques  also appear to be additive.
This additive effect can be more clearly seen in Figure  1-B where the effect  of FIR
on NOX is  plotted as  a function of the total  flue  gas recirculated (to  windbox  and to
fuel).  For each of the three data sets shown on the graph, the  windbox  FGR is  held
constant while the FIR flowrate is increased.  The  dotted line  on the graph defines
the relationship between  NOX  and the windbox FGR  alone.   At  both  the 15% and 23%
windbox FGR rates,  FIR injection  is capable of additional  reductions of approximately
50%.  Table  1 summarizes the  maximum  reductions  achievable with  FIR when used in
conjunction with FGR.  Furthermore, it  is  evident  that FIR alone is  more  effective
than FGR:  5%  of the flue gas injected into the fuel results  in lower NOX than 23%  flue
gas injected  into the  combustion air.  This is shown  graphically in Figure 2 where NOX
reduction  is  plotted  vs  the  total flue gas  recirculated.   The  NOX reduction curve
rises more steeply  with FIR than  without.   It should be re-stated here that the  flue
gas  recirculation  to the  fuel  requires  significantly  higher  compression  that
recirculation to the  combustion air.
                                       7B-21

-------
As postulated earlier,  FIR is  believed control prompt, as well as thermal NO, both by
reducing  peak flame  temperatures  and  by  lowering  the  concentration  of  certain
hydrocarbon radicals which are  thought  to  produce prompt  NO.  The  concentration of
prompt NO formed in utility combustion systems is thought  to be 25 ppm or less.   The
tests performed in the present  study are not  intended to  distinguish  between prompt
NO reductions and thermal NO  reductions, or even  to  confirm the  existence  of prompt
NO.   It  is  not  possible  to conclude whether the  additive NOX  reductions are due to
more  efficient  mixing  of flue  gas  with  the fuel and air,  or whether  FIR actually
suppresses  prompt NO formation.   What  can  be  concluded  however  is  that  FIR is  more
effective than windbox  FGR,  and  that together there is a  measurable additive benefit.
The  use  of  FIR  does  not  significantly  affect  flame stability  up  to a fuel dilution
ratio of approximately 35%  Higher injection rates create  a  tendency to lift off the
burner face creating a  "boiler rumble" and large fluctuations in NOX and 02 and furnace
draft.   At  lower  injection  rates,  the  appearance of the  flame  is not significantly
different  from  the  flame  appearance  with  no  FIR  injection,  other  than  decreased
brightness  which is  indicative  of lower peak flame temperatures.
The  Effect  of Overfire Air on FIR:  The effect FIR when  used without overfire air is
shown  in Figure 3-A  and Figure  3-B.   FIR  is  equally effective with, or without
overfire  air.   Without OFA,  FIR reduces  NOX concentrations  by  60% at  0% FGR and 15%
FGR.
It was also expected that overfire air would affect flame stability by decreasing the
burner throat velocities.  The  tests demonstrated that  overfire  air does not affect
flame stability.  Figure  3-A shows that the limit of flame  stability is approximately
at 35% fuel dilution regardless  of the OFA rate.   Figure 3-B shows that the effect of
overfire air has a  decreasing  effect at higher FGR rates.   For example,  at 15% windbox
FGR  with the maximum FIR  injection rate,  10% air staging results  in  less than a 5 ppm
NOX reduction.
The  Effect  of  Firing  Rate On  FIR:   The  effect of  FIR  at  three firing  rates is
presented in Figure 4.  FIR injection using  flue gas results  in approximately the same
NOX reductions at  1.0,  1.5,  and  2.0  x 106 Btu/hr.  The slopes of the curves on Figure
4  are  not  a function of the  firing rate.   With  no windbox  FGR,  FIR  reduces NOX at
approximately 5 ppm/%fuel dilution up to 35% fuel  dilution.  At  an  optimum windbox FGR
rate, the slope decreases to  .6 ppm/%fuel dilution up to  35% fuel  dilution.
It  is important to  note that  reduced  firing does  not  significantly  affect flame
stability.  The limit of  flame stability occurs at approximately 35% fuel dilution at
each firing rate tested.    It  is difficult to  extrapolate  this  characteristic to the
full-scale  application primarily due to the  non-characteristic  NOX  vs  firing  rate
                                       7B-22

-------
relationship, i.e. the  relative  increase  in  NOX  at  the  mid-firing  rate.   It is also
important to remember that the minimum 02 condition  at the lower firing rates results
in  significantly higher 02 concentrations.  The  air register  vane  setting is likely
to  affect  flame stability and the  minimum  02  condition,  however the effect  of air
register adjustments was not  examined during this test  program.
Air As FIR Diluent
The original concept of Fuel  Injection Recirculation involved  injecting flue gas into
the fuel.  In principle  any diluent could have the same affect  on prompt NO formation.
The advantage of using  air as  a  fuel diluent  is  that compressing dry air up to fuel
pressures is less expensive than  compressing hot flue gas.  In  addition, problems with
moisture  condensation  in the  fuel  delivery  system are eliminated  if  air  is  used
instead  of  flue gas.   The effectiveness of air  injection was  explored  in a limited
test matrix  intended to compare  air to  flue gas  as  FIR  diluents.
Air was  injected as an FIR diluent at the following  conditions:  high combustion air
temperatures, at a nominal overfire air rate, and at two FGR rates.   The results are
shown  in Figure  5,  where the  results  for  flue gas injection  are re-plotted  for
comparison.  These results demonstrate  that  air injection is not as  effective as flue
gas injection in overall NOX  reductions.  For the 0% FGR case, NOX actually increases
at  low air  injection  rate.    The  characteristic  is not  measured  at  the 15%  FGR
condition.  Table 2 shows that the overall  NOX reductions achieved using air injection
are less than half of the reduction measured using  flue gas injection.
Steam as FIR Diluent
Steam is another fuel diluent which in principle should reduce NOX  much  the same way
as  flue  gas.  The use  of  steam as an FIR diluent for full-scale application  may be
attractive on a  cost basis since it would require  no  additional compressors.  Provided
that steam could be extracted at relatively low pressures,  the impact on boiler heat
rate should not  be prohibitive.   The use of  steam injection as a means of NOX control
on  large boilers  is  not a  new technique.   However,   it  is usually  injected into the
combustion air upstream of the burner rather than into  the fuel.
Particular experimental  difficulties precluded a  more expanded test  matrix with steam
injection.   The primary difficulty  was the  high  fluctuation in  steam flow:   the
flowrate fluctuated by approximately 25 %.   This  made measurement of steam flow rate
difficult and caused high fluctuations  of  NOX  and  especially CO.   Figure  6 shows a
example  time  trace  taken  from data logger  records.   Note  that  the NOX  has  been
corrected to 3%  02.  NOX, CO,  and  02 fluctuated regularly at the same frequency of the
steam generator  fluctuation.    The  period  of  the  fluctuation  was  approximately 4
minutes.   As the steam  flow  cycled  to  a maximum, about 62  Ib/hr,  the NOX reached a
                                       7B-23

-------
minimum, and CO was  in excess of 1000 ppm.   At the minimum steam flow, about 48 Ib/hr,
the NO  reached a relative maximum,  and CO reached a minimum.  Since the  fuel  flow
could not be adjusted for  changes  in  back pressure, the fuel flow also cycled causing
small fluctuations  in 02.  Despite the fact that the  steam  generator flow rate could
not be held constant, the results generated are  still  valuable.  The steam flow cycled
in a very regular, repeatable manner, and accurate  data were obtained by averaging the
continuous emissions data over many  cycles.
The results of the  steam injection test are presented in Table 4-11 and in Figure 7.
The steam  injection tests were performed without overfire air.  When steam injection
was used in conjunction with overfire air,  excessively high CO emissions resulted as
well  as poor  flame  stability.   Overall  NOX reductions are 54% without  FGR,  and 36%
with  15%  FGR.   Figure 7  presents  a  comparison   of steam  injection  and  flue gas
injection.  Also shown on this figure are the minimum and maximum NOX concentrations
corresponding  to the maximum and minimum steam flow rate. The results show that with
no overfire  air, steam injection  is  nearly as effective  as flue gas  injection.

CONCLUSIONS
Fuel  Injection  Recirculation  (FIR) was  demonstrated  on  a  laboratory  scale  test
facility designed to simulate the significant combustion  characteristics of full-scale
utility natural gas burners.  FIR was evaluated  in  terms  of NOX reductions and burner
stability.   While,  the absolute  values of N0x  emissions  results  presented in this
report  should not extrapolated  directly to full-scale boilers, relative NOX reductions
and general  trends  measured on the  sub-scale facility,  should  be representative of
results expected on full-scale  units. The major conclusions drawn  from the laboratory
evaluation are presented  below:
Baseline Characteristics
                At test conditions typical of utility boilers, the baseline
                NOX   concentrations   on    the   sub-scale    facility   are
                representative  of  full scale  units.
                The  measured N0x  dependencies   on  FGR,  air   staging,  air
                preheat    temperatures,   and   excess   air   levels   are
                representative  of  trends seen in full  scale units.
                The  measured relationship  between  NOX and firing rate  is
                typical of smaller package  boilers.
Flue  Gas  as  FIR Diluent
                FIR  is an effective NOX  reduction technique  to be applied  to
                natural gas-fired  boilers,  and NOX reductions  achieved using
                this technique are additive to those achieved  by windbox FGR
                and  air staging.
                                       7B-24

-------
              FIR is more effective than windbox  FGR,  per pound of flue
               gas recirculated,  in  reducing  NOX emissions.

              FIR in combination with air staging  and windbox FGR results
               in   additional  NO  reduction  of approximately  50%.   NO
               concentrations  below  25 ppm were achieved at full load with
               nominal  air staging,  15%  FGR  and 35% fuel dilution.

              FIR has  no  adverse effects on  maintaining  minimum 02 levels.

              FIR is equally effective at reduced firing  rates  and when
               used without overfire air.

              FIR operates with good flame  stability at  high  combustion
               air temperatures and  nominal air staging at FIR levels up to
               35%  fuel  dilution.    However,  the  maximum  level  of  FIR
               consistent  with acceptable  burner  stability  decreases with
               decreasing  combustion air temperature.

              With no air staging,  FIR  operates with good flame stability
               at  low combustion air temperature up  to a  35% fuel dilution.
Air as FIR Diluent
              Air as an FIR diluent is less  effective  than  flue gas and
               leads  to  flame  instabilities  at lower  injection  rates.
Steam as FIR Diluent
               Steam as an FIR diluent when applied  in combination with air
               staging  results   in   poor   flame  stability  and  high  CO
               concentrations.

               Steam when  applied  with  no  air  staging  is  nearly  as
               effective  as  flue  gas  as  an FIR diluent.

               CO  concentrations  are  generally  higher with steam than with
               air,  or  flue  gas as the  FIR diluent.
                                      7B-25

-------
                        STEAM INJECTION, O OFA, O% FGR
HI
CO
LL
o


 E
 Q.
 a

 x"
O
      1.1



      1.0



      0.9



      0.8



      0.7



      0.6



      0.5



      0.4



      0.3



      0.2



      0.1



      0.0
           TEST 186, 187
                        CO 6-1000 ppm
                                            19 November 1990

                                              <   n

                                              \   !' i   | :
                                              I   i    (I
NOx@3%O2


0-250 ppm
              O2 0-10%
                          10:40
                                        10:55


                                    TIME
             Figure 6.  Example Emissions Time Traces with Steam Injection
    180





    160




    140




O   120



CO
                                             FIRING RATE: 2.0 x 1C6 Btu/hr

                                             APH:       490"F

                                             NOOVERFIREAIR


                                              STEAM INJECTION

                                                  FLUE GAS INJECTION  -
    100





     80





     60





     40





     20





      0
             15% FGR
            0     5     10    15    20   25    30    35   40    45    50




                      PERCENT FUEL DILUTION
                  Figure 7.  Effect of Steam Injection
                                    7B-26

-------
cf1
CO


Q.
Q.
x"
O
110

100

 90

 80

 70

 60

 50

 40

 30

 20

 10

  0
                                  INJECTION
FIRING RATE:  2.0 x 106 Btu/hr
APH:         490 F
NOMINAL OFA
MINIMUM O2
A-  AIR INJECTION
     FLUE GAS INJECTION
                         10    15    20    25    30    35    40    45
                         PERCENT FUEL DILUTION
                         Figure 5. Effect of Air as FIR Diluent
                                                                   50
                                   7B-27

-------
O
<
CO
Q.
0.
x"
O
110

100

 90

 80

 70

 60

 50

 40

 30

 20

 10

  0
                                                        NOMINAL OFA
                                                        MINIMUM O2
                                                        APH - 490 F
                                                        A 2.0 x106 Btu/hr
                                                        - 1.5 x106 Btu/hr
                                                         - O  1.0 x106 Btu/hr
                          10    15    20    25    30    35    40   45    50
                         PERCENT FUEL DILUTION
               Rgure 4. Effect of FIR at Three Firing Rates; NOx vs. Dilution
                                  7B-28

-------
    180



    160



    140


O  120


CO

    100
0.   80
0.


O   w
z

     40


     20


      0
            I | I r I 7 | T T  I 1  [ I I I


            O

            O-  -, ^ 0% WB FGR
                    \
                      \
                        \
                         \
FIRING RATE:  2.0 MMBtu/hr

APH:        488 "F

MINIMUM O2

-ANOMINAL OFA

-O- NO OFA
                                  NO
                                   N
                0% WB FGR

                \

          15% WBFGR
                                               ~~O
          15% WB FGR
                       10    15    20   25    30    35   40    45    50
                       PERCENT FUEL DILUTION
                   Figure 3A. FIR With and Without Overfire Air






CM
O
a?
n

E
Q.
a.
x~
O
z





IOU


160

140

120

100
80

60


40

20
n

! 0 FIRING RATE: 2.0 MMBtj/hr
~ APH: 488DF
"\ MINIMUM Og
\\ A NOMINAL OVERFIRE AIR
\ '\  O- NOOVERFIREAIR
\O : ^^
\
\
\ .
\
- \. <$ "...
\ ' . \ 0% FIR 	 >'""-...
V ""-^ I o'vQ.
\ ' ... | \

'. \A 	 \O 15% WBFGR
\ 	 \ NO
0% WB FGR A V. \
^^-""-i. VN

'..!,..,! . . . I . , . , I . . , I . , , , I . , , , '
                              10
                                       15
                                                20
                                                         25
                                                                  30
                PERCENT FLUE GAS RECIRCULATION

                              (FGR + FIR)


                Rgure 3B.  Effect of FIR with and Without Overfire Air
                                  7B-29

-------
o
D
O
UJ
DC
X
O
LU
O
DC
UJ
O.
100


 90


 80


 70


 60


 50


 40


 30


 20


 10


  0
         FIRING RATE: 2.0 x 106Btu/hr

         APH:       4S8F

         NOMINAL OFA

         MINIMUM O2
                          10
                               15
20
                                                       25
                  PERCENT FLUE GAS RECIRCULATION
                    Figure 2. Maximum NOx Reduction with FIR
30
                               7B-30

-------
cf
vP
5-*


a.
a.
x"
O
110

100

 90

 80

 70

 60

 50

 40

 30

 20

 10

  0
                   0% WB FGR
               23% WB FGR
                                               FIRING RATE: 2.0 MMBtu/hr
                                               APH:        4886F
                                               NOMINAL OFA
                                               MINIMUM O2
             0     5     10    15    20    25    30    35    40   45    50

                       PERCENT FUEL DILUTION
        Rgure 1A. Effect of FIR at Three FGR Rates; NOx vs. Percent Fuel Dilution
O
5?
CO

a.
a.
x"
O
z
    110

    100

     90

     80

     70

     60

     50

     40

     30

     20

     10

      0
       0% WB FGR
                          0% FIR
                                           FIRING RATE:  2.0 MMBtu/hr
                                           APH:       4886F
                                           NOMINAL OFA
                                           MINIMUM 02
                              7
                                 15%WBFGR
                                                    23% WB FGR
                               10
                                        15
                                                 20
                                                      25
                                                                    30
                  PERCENT FLUE GAS RECIRCULATION
                                 (FGR + FIR)
             Rgure 1B. Effect of FIR at Three FGR Rates; NOx vs. Total FGR
                                 7B-31

-------
                                   TABLE 1
                    MAXIMUM NO. REDUCTIONS WITH FIR
Windbox
FGR,%
0
15
23
NO
0% FIR
89.2
40.6
35.3
. & 3% O,
Max FIR
37.8
21.9
17.0
% Reduction
57.6
46.1
51.8
                                   TABLE 2
                       COMPARATIVE NOS REDUCTIONS;
                    AIR INJECTION VS FLUE GAS INJECTION
                       Air Injection
               0        MAX                    0
              FIR        FIR    %Reduction      FIR
                                        Flue Gas Injection
                                            MAX
                                             FIR       %Reduction
0% FOR
94.1
73.
21.6
89.2
                                                           33.1
                                                          62.9
15% FOR      41.0
          31.2
          23.9
            40.6
                                                          21.9
                         46.1
NOTES    1. Firing Rate = 2.0 x 10" Btu/hr
          2. Nominal OFA
          3. Minimum O2
          4. APH = 490 F
                                    7B-32

-------
     ADVANCED  REBURNING  FOR NOX CONTROL
            IN COAL FIRED BOILERS
                  S.  L.  Chen
                W.  R. Seeker
                  R.  Payne
Energy and Environmental Research Corporation
                  18 Mason
          Irvine,  California  92718
                (714)859-8851

-------
                         ADVANCED REBURNING FOR NOX CONTROL
                                IN  COAL FIRED  BOILERS
ABSTRACT
This paper summarizes  an  experimental  study which was  conducted  to investigate the
chemical  constraints  of   the  reburning  process  and  identify  advanced  reburning
configurations for optimal NOX  reduction  in  coal-fired boilers.  Tests were performed
initially on a bench scale tunnel furnace to characterize and optimize the fuel-rich
reburning zone and  fuel-lean  burnout zone  independently.   Based  on the results, an
advanced  reburning  process was  designed  which integrated  reburning  with selective
reducing  agent  injection  to enhance the  burnout  zone efficiency.   The concept was
subsequently tested in a pilot  scale  facility and yielded over 80 percent reduction in
NOX emissions.
                                       7B-35

-------
INTRODUCTION
Reburning is an NOX control  technology which  uses fuel to reduce N0.1"A  The main  heat
release zone can be operated normally to achieve optimum  combustion conditions  without
regard for NOX control.   With  reburning,  a  fraction of the  fuel is injected above the
main heat release zone.   Hydrocarbon  radicals from combustion  of reburning  fuel react
with nitric oxide to form molecular nitrogen.  This process occurs  best in the  absence
of oxygen.  Thus sufficient reburning fuel,  between 15 and 20 percent of the  total  heat
input, must be added to produce an oxygen deficient reburning zone.  Subsequently, air
is provided to  combust fuel fragments which  remain  at  the exit of this  zone.  Since
reduced nitrogen species  NH3 and HCN are  also  present, air addition may allow a  further
NO,, reduction.
  X
Previous studies showed that 60 percent reduction in NOX emissions could be  achieved
with natural  gas  reburning.5   Recently research  has  been conducted  to examine  and
enhance the  NOX reduction  chemistry  in the burnout zone.6   The  burnout zone can be
considered as  an  excess-air "flame" burning  the  remaining  fuel  fragments  from  the
reburning zone.  Oxidation  of  the  fuel  fragments,  particularly CO, could  generate a
significant amount of radicals via chain branching:
               CO + OH  =  C02 + H
                H + 02  =  OH  + 0
               0 + H20  =  OH  + OH

These radicals  play  an  important role  in  the conversion of XN  species to  N2 or NO
during burnout.

Figure 1 is an experimental  examination of  the burnout zone chemistry,  in particular,
the conversion efficiency of NH3 to N2.  The rich  zone was assumed to  supply 600  ppm
each of NO and NH3, or an N  to  NO ratio  of  1.0.  Under excess air  conditions, ammonia
gas was  mixed with various amounts of CO and injected at  temperatures  between 1300  and
2200F.  The  solid symbols  represent  the injection  of  NH3 alone, which  is basically a
                                      7B-36

-------
simulation of Thermal De-N0x.   For the open  symbols,  0.2  percent  CO was  included with
NH3,  thereby yielding  a burnout  like  environment.   The presence of CO lowered the
optimum temperature  for NOX reduction  from  1800F to 1500F.  It is  readily  apparent
that a reduction in the burnout temperature from the 2200-2400F  normally employed  in
the reburning  process would  increase the  conversion  efficiency of  NH3 +  NO to  N2
because of the presence of CO.

This  paper  summarizes the results  of a  pilot  scale study which was undertaken  to
investigate the possibility of positive synergism between the injection of selective
reducing  agents,  such as ammonium  sulfate, to provide  the reducing specie  NH3,and
combustion modifications, such as reburning,to  serve as  the source of CO.

EXPERIMENTAL
The 3.0  MWt,  down-fired tower  furnace5  used  in the pilot-scale investigations was
refractory-lined and water jacketed with inside dimensions of 1.2 x  1.2  x 8.0 m.  The
four  main diffusion  burners  each consisted of  an  inner  pipe  for axial  primary fuel
injection and an outer pipe, equipped  with  swirl vanes,  for the  main  combustion air.
This  four burner array produced  relatively  uniform velocity and  composition  profiles
at the primary zone exit.  The furnace contained  seven rows of ports for reburning fuel
and burnout air  injection.   The temperature profile was manipulated  by insertion  of
cooling panels, positioned against the furnace walls.   The reburning  fuel and  burnout
air injectors were designed to maintain jet  mixing  similarity  between the pilot-scale
furnace and a full scale boiler based on  empirical  correlations  for  entrainment rate
and jet penetration.

Exhaust gas samples were withdrawn through a stainless steel, water-jacketed probe and
analyzed for NOX  (chemiluminescence), 02 (paramagnetic),  C0/C02 (NDIR), and S02 (NDUV).
A water jacketed probe with an  internal water quench spray near the front end  was used
for extracting in-flame samples.   Gas phase HCN and  NH3 species were collected in a gas
washing unit and  subsequently analyzed  for CN"  and dissolved ammonia using specific ion
electrodes.  Gas temperatures were characterized with a  suction  pyrometer.

RESULTS
Recent studies6 have  suggested  that the key parameters for the  enhancements of burnout
zone  chemistry in  staged combustion or reburning are:
      f    Reaction temperature  (850C)
      0    CO levels  (0.5% or less), and
          NH3 species.
                                       7B-37

-------
Advanced Reburninq
Apparently  the  conventional  reburning  process  does   not   provide   the  required
environment.  An advanced reburning process, which combines reburning  with selective
NOX reduction (SNR)  via ammonium  sulfate injection, was designed.   Figure 2 shows two
hybrid schemes with  20 percent and 10 percent gas reburning,   respectively.   With 20
percent reburning (SR2  =  0.9), the burnout air was divided into two streams to yield
an  SR3 of  1.03  and  an  SRt of  1.15.   With 10  percent  reburning,  the  reburning  zone
stoichiometry (SR2)  was 1.03  and the burnout air stoichiometry  (SRt) was 1.15.  In both
cases, an aqueous solution of ammonium sulfate was atomized with the final burnout air
an  injected at 850C at an N to NO molar ratio of 1.5.

Verification Tests
Figure 3 shows the advanced reburning  results obtained with natural gas  as the primary
fuel.  The natural gas fired at 4.5 x 106 Btu/hr was doped with NH3 to yield primary
NOX levels  of 600 and 400 ppm (dry, 0 percent 02).  Twenty and ten percent  advanced gas
reburning were applied, respectively.  Similar final emissions,  approximately 125 ppm
NOX, were achieved with both  concepts.   Experiments were subsequently carried out  with
an  Indiana coal  as  the primary fuel.   The Indiana coal produced  an  uncontrolled NOX
emission of 800 ppm (dry,  0  percent 02)  at 15  percent excess air.   The  primary NOX at
SR,  =  1.13  was  680 ppm.   Figure  4 presents the results and indicate that  as seen in
the  bench  scale  studies6,   both  advanced concepts  were  equally effective in  NOX
reductions.  It is  apparent that  there exists  a  tradeoff between natural  gas  premiums
and the cost of ammonium sulfate.

Ammonia Slip and $0x Emissions
The  injection  of ammonium  sulfate into  the  furnace  has a  potential  of  producing
unwanted emissions  such as NH3 and S02/S03.  A series of  exhaust  measurements were  made
to evaluate the slip of ammonia using  selective  ion electrode and the emissions of S02
and  S03  via  controlled condensation  during  the  Indiana  coal  tests.    Exhaust  NH3
concentrations were  negligible in all  cases including those obtained with Utah coal and
natural gas as the primary fuel.    Higher S02 emissions  were obtained with  10 percent
gas reburning.  However, the  uncontrolled S02 level  was  maintained  with  20 percent gas
reburning due to  dilution.  No increase  in S03 emissions was observed for both cases,
suggesting  favorable conversion of the sulfate to S02.

Thus, there exists  a control  strategy  to prevent the  increase  in S02 emissions due to
injections  of ammonium sulfate.   For  the application  of  advanced reburning  to  high
                                      7B-38

-------
sulfur coals,  10  percent  gas reburning is recommended, whereas  for low sulfur coal
applications, the 20 percent gas reburning concept is preferred.

CONCLUSIONS
In summary, these results  suggest that selective  reducing  agents  can be combined with
combustion modification techniques to provide NOX  reductions that are larger  than those
that are  possible  by applying the technologies  simultaneously but separately.  By using
the stoichiometry control  associated  with  reburning  to  produce a slightly fuel rich
region for selective reducing agent injection,  reductions can be achieved at  relatively
low temperatures without the use of stainless steel or other catalysts.

ACKNOWLEDGEMENTS
This work was primarily supported by the U.S.  Department of Energy, Pittsburgh Energy
Technology Center  (Contract  No.  DE-AC22-86PC91025) with  Dr.  Richard  Tischer as the
Project Manager.  We also would like to acknowledge the contributions of our colleague
Mr. Loc Ho in conducting the experiments.

DISCLAIMER
Reference herein to any specific commercial  product, process, or service by trade name,
trademark, manufacturer, or  otherwise, does not  necessarily  constitute or imply its
endorsement, recommendation,  or favoring by the United States  Government or  any agency
thereof.   The views and options of authors expressed herein do not necessarily state
or reflect those of the United States Government or any agency thereof.

REFERENCES
  1.  Myerson,  A.  L.,   et  al.,  Sixth  Symposium  (International)  on  Combustion,  The
     Combustion Institute, 1957, p. 154.
  2.  Reed, R. D., "Process for  the  Disposal of  Nitrogen  Oxide."  John Zink  Company,
     U.S. Patent 1274637,  1969.
  3.  Wendt, J. 0.  L.,  et al.,  Fourteenth Symposium  (International) on Combustion, the
     Combustion Institute, 1973, p. 897.
  4.  Takahashi, Y., et al.,  "Development of Mitsubishi 'MACT'  In-Furnace NOX Removal
     Process."  Presented at the U.S.-Japan  NOX Information Exchange, Tokyo,  Japan, May
     25-30,  1981.  Published  in  Mitsubishi  Heavy Industries, Ltd. Technical Review,
     Vol. 18, No. 2.
  5.  Chen, S. L., et al.,  21st Intl. Symp., Combustion Institute, 1986,  p.  1159.
  6.  Chen, S. L., et al, JAPCA. Vol. 39,  No.  10  (1989).
                                      7B-39

-------
Z
UJ
O
5  80
ox 60
 X
O
2
   40
   20
                   SR - 1.1


                   (NOx)p - 600 PPM (DRY, O% O2)


                             - 1.0
                  O NH3 + 0.2% CO



                   NH3 ONLY
          1400  1600  1800  2000  2200


       PEAK INJECTION TEMPERATURE 
-------
                                   Reburning
                             V77\ Advanced Reburning
i
600
~ 500
CM
0
o 400
o
^ 300
Q.
Q.
0* 200
z
4 f\f\
100
n
,
Primary NOX

^H


^

Ml


_










^/,






.g Primary NOX
= 4
1 I
^ t>
O ^
00









Q)
cc
^
2
%,
         20% Gas
10% Gas
Figure 3.  Results obtained with  natural gas as  primary  fuel
                       7B-41

-------
    800
7,  600
oc
Q   400
Q.
Q.
    200
             UNCONTROLLED NO
                INDIANA COAL
CO


O
               0
               CM
                               CO

                               O
                                         O
                                         UJ
                                         CC
                                         CO
                                         CO

      Figure 4.  Pilot scale results with Indiana coal
                       7B-42

-------
LARGE SCALE TRIALS AND DEVELOPMENT OF FUEL STAGING IN A 160 MW COAL
                            FIRED  BOILER
                           H.  Spliethoff
                       Universitat Stuttgart
        Institut fUr Verfahrenstechnik und Dampfkesselwesen
                    Prof.  Dr.  techn.  R.  Dolezal
                        Pfaffenwaldring 23
                     7000  Stuttgart 80, Germany

-------
                   LARGE SCALE TRIALS AND DEVELOPMENT OF FUEL
                      STAGING IN A 160 MW COAL FIRED BOILER
                                  H.  Spliethoff
                              Universitat Stuttgart
               Institut fur Verfahrenstechnik und Dampfkesselwesen
                           Prof.  Dr.  techn.  R.  DoleZal
                               Pfaffenwaldring 23
                           7000 Stuttgart 80, Germany
ABSTRACT

In a study under the contract of the Saarbergwerke AG it is planned to  achieve NOX
emissions near 200 mg N02/m3, i.e. 98 ppm NO without expensive DENOX technology.
By application of retrofit primary methods  (air staging,  flue gas  recirculation)
the NOX emissions  from  the coal  fired boiler  Fenne  3  (slag tap furnace, 160 MW
electric power) could be reduced  from  900 to 520 ppm NO at 5% 02-  In the year 1988
the boiler was  equipped  with  an  arrangement for fuel staging. Reburning fuel is
coal gas with 50 % H2 and 25 % Cffy.
Experiments from September 1988 to July 1990 showed  that  reburning can reduce NOX
emissions from 520  ppm to 180 ppm NO (5% 02). The influence of different parameters
(primary zone stoichiometry, reducing zone stoichiometry  etc.) was  investigated.
The reduction zone  stoichiometry  and the reburn fuel  mixing were pointed  out to be
the most important parameters for low NOX emissions by reburning /!/.
In order to optimize reburning the following work has been done:
    t   distribution  of  flue  gas  concentrations  was  measured  (primary  zone,
        reducing zone, burnout  zone),
       reburning  fuel  mixing  was optimized  by  three-dimensional  fluid  flow
        computations,
       fuel  staging with synthetic gases was examined in a  0.5 MW  test  facility
        and
       the influence of ammonia addition into the reduction zone was investigated.

By optimizing  the   reburning  gas injection  and by  addition of  ammonia to the
reduction zone the NOX emissions could be reduced to a minimum of 130 ppm NO  (5%
02) up to now.  Reburning has only a slight  impact  on  the  burnout  of the  coal. The
carbon content in the fly ash is less than five percent.
                                     7B-45

-------
INTRODUCTION

In the  last years  there  had  been  large  efforts  to  lower  NOX  emissions  from
stationary combustion sources.  For combustion systems with a thermal  load  of more
than 300 MW NOX emissions of 200 mg/m3 NO calculated as N02  (98 ppm  NO) at  5%  02
(molten ash furnace) or 6% 02 (try ash furnace) are demanded  in Germany.

Applied and commonly used techniques for NOX abatement can be devided in
       combustion modifications,
       selective  non catalytic  reduction  (SNCR)  by ammonia or urea and
       selective  catalytic reduction (SCR)  by ammonia.

Due to the short period for retrofitting existing  power plants and equipping new
power plants with NOX abatement  techniques,  most German  hard  coal  fired  power
stations are or will be soon equipped with the SCR DENOX technology.

Measures  to  influence  the  NOX  emissions  of   coal  furnaces  by  combustion
modifications are:
       optimized  boiler  operation (low oxygen operation),
       flue gas recirculation,
       air staging (single burner or in the furnace)  and
       fuel  staging, reburning  (single burner or in the furnace).

In  the  past years  air staging  has  proved  to  be  an  effective method  for NOX
reduction.  For  German lignite  it seems  possible  to  achieve the  required NOX
emissions without  expensive DENOX-technology by improved  air staging in the  furnace
/2/. A  further  technique  of  minimizing  NOX  emissions  is a  method  called  fuel
staging, reburning  or  In-Furnace  NOX Reduction.  Results of fuel  staging  in  test
facilities are very promising.
A  published  application  of  reburning  to  coal  combustion furnaces  is  the  MACT
process. By fuel staging  at a coal dust furnace NOX emissions of less  than  150 ppm
could be achieved /3/.

Figure 1 shows  the principle of  fuel  staging. In the first zone, which is the main
heat release zone,  the fuel  can  be  burnt under  fuel lean conditions to ensure
complete burnout.  The addition of reburning fuel creates a fuel rich,  NOX reduction
zone. The reduction  of nitrogen oxides is initiated by hydrocarbon radicals. In the
final zone the combustion is completed by  addition of air.


DESCRIPTION OF THE PROJECT "BRENNSTOFFTRENNSTUFUNG (BTS)H

To lower the NOx emissions  in coal  dust furnaces  the  project  "Combined minimizing
of NOX production and reduction  of formed NOX - Brennstofftrennstufung  (translated:
Fuel Splitting and Staging)" has been  initialized.

Coal is  divided by  a devolatilization process in a reduction gas with volatile
nitrogen and the remaining  coal  (char). Both fractions are burned in a fuel staged
combustion with char as primary fuel  and pyrolysis gas as  reburning fuel.
                                     7B-46

-------
The project consists of several steps:

       Investigation of reburning  at  a  0.5  MW  gas  fired  combustion  facility with
        synthetic fuel,
       Large  scale tests of reburning with coal gas as reburning fuel in a slag
        tap furnace,
       Investigation of the process "fuel splitting and staging* in  a  small scale
        test facility.

The investigations of reburning in the 0.5  MW  combustion facility with synthetic
fuels and the  trials  at  the 160 MWe] slag tap furnace are  subjects  of this report.
Results of performance and emissions of the process  "Fuel Splitting and Staging"
in a small scale test facility are soon expected.
MECHANISMS GOVERNING NOX PRODUCTION AND REDUCTION AT FUEL STAGING

Figure 2 shows the NOX production and NOX  reduction  mechanism for  the  three  zones
of a fuel  staged  combustion  with coal  dust as primary fuel  and gas as  reburning
fuel.

In the  main  heat release  zone  the formation of  NOX is mainly due  to the  fuel
nitrogen. During devolatilization of coal  a part of fuel  nitrogen is released with
the  pyrolysis  gases, the  other part  remains in the  coal  char. The amount  of
nitrogen released with the  pyrolysis products depends on coal  properties  (volatile
matter  content)  and  temperature.  The volatile  nitrogen  and  char  nitrogen  are
converted to NOX in a different way and in different amounts.
The volatile nitrogen  quickly  forms  the  intermediate species HCN, which  is  then
converted  in  a slow  reaction  to  NH3-  Depending on  the  fuel/air ratio  and  on
temperature,  NH3 is  either  reduced to molecular nitrogen  or it forms NO. The degree
of nitrogen oxide  formation  from the volatile fuel  nitrogen can  be  affected  by
primary combustion modifications, such as air staging or flue gas recirculation.
The production of NOX from Char-N is generally low with conversion rates  between
10 and  20 percent. The heterogeneous production of nitrogen oxide is less  sensitive
to process parameters as  the  formation  from volatile  sources.  Therefore  it  is
assumed, that Char-N  is  responsible  for minimum  NOX emissions, which can not  be
lowered.

In the  reduction zone the nitrogen oxides  formed in the main  heat release zone are
reduced by  homogeneous reactions.  If the reburning fuel  contains hydrocarbons, the
gas phase reduction of NO  is initiated by CH-j in a fast reaction
           NO      +   CHi        >   HCN    +   products.       (1)
This  fast step is followed by the relatively  slow conversion  of HCN to  NH-j. This
reaction is significant for the overall reduction.
NH-j then either forms NO by reaction with 0 or OH radicals
           NHi     +  0 / OH     >  NO      +   products       (2)
or is reduced  by NO to N2
                                     7B-47

-------
           NHi     +  NO         >  N2      +   products      (3).
Because  of the  fuel   rich  atmosphere  in  the  reduction  zone  reaction  (3)   is
predominant.

Investigations of Bose /4/ confirm, that the gas phase reactions are dominant  in
fuel   rich  combustion  zones  and that  the  heterogeneous  reduction is  of minor
importance for coal dust combustion.
The gas phase  nitrogen  reactions in the first and second stage are quite the same,
as to be seen in figure 2.

By addition of air the N-containing species NO, HCN and NH^ are converted to NOX
in the burnout zone.  NO and HCN are almost completely transformed to NOX, NH-j only
in a  very  small  amount  /5/.   If  the  burnout  air  is  added  to  the flue  gas   at
temperatures of about 900 C,  a further NOX reduction is possible.
REBURNING WITH SYNTHETIC COAL GASES IN A TEST FACILITY

In order to study the reduction  efficiency with  a pyrolysis gas as reburning fuel
experimental  investigations  were  carried   out   under   the  contract  of  the
Saarbergwerke in a gas fired combustion facility at the University of Karlsruhe.
The  synthetic  pyrolysis gas  consists of 60% H2  and  30% CH4-  The  watercooled
combustion chamber is described  elsewhere /6/. The  residence  time in the reducing
atmosphere is about  one  second,  the flue gas  temperature  at  the location of gas
injection  is about  1300  C,  at  the location of air  injection  about  900 C. The
stoichiometric ratio of the first fuel lean  zone is \\ = 1.1 with a measured NOX
level after the first stage of 600 ppm.  The overall stoichiometric ratio was kept
constant at ^3 = 1.2.
The keypoint of the  tests was to evaluate the influence of  ammonia addition to the
reburning  fuel,  as  pyrolysis  gases  contain  nitrogen   species  such  as  NH3-
Furthermore the pilot scale results are compared  to  the results of reburning  in the
slag tap furnace in order to demonstrate  optimization potential for the large  scale
application. The experiments at  a  pilot scale test facility allow the variation of
parameters which cannot be changed at  a utility power plant.

Earlier investigations showed, that the  addition of a nitrogen species such  as NH3
to a reburn fuel makes no difference at  the optimum stoichiometry X2> but outside
this optimum the N containing reburn fuel  resulted in higher NOX emissions /7/.

Figure  3  shows  the  final NOX emissions and  the corresponding measured nitrogen
species after the reduction zone  for using a reburn fuel containing no NH3, 1.5%
and 3 % NH3- For pure pyrolysis  gas  (0%  NH3),  NOX  is  reduced  from 600 ppm (5% 02)
after the primary zone to 115 ppm after the burnout  zone at X2 = 0.85. The addition
of 3 Vol% ammonia results  in a shift of the optimum stoichiometry to A2)0pt  = -89
and  a  further  reduction of the total NOX emissions  to 60 ppm  NOX  (5% 02). The
corresponding N-species of the reduction zone show  an increased reduction of NOX,
the concentration of NH3 rises drastically for X2 < A2)0pt,  while the HCN emission
                                     7B-48

-------
is not affected by the increased NH3 input. At the  optimum  stoichiometry  without
ammonia, N-species of 130  ppm NO and 20 ppm NH3 are converted to 115 ppm NOX in  the
burnout zone. For the maximum NH3 addition (3%) 60  ppm NO and  100  ppm  NH3  form 60
ppm final NOX emissions.
Further experiments at the University of  Karlsruhe /8/ outside this project with
natural  gas  as  reburn  fuel  showed  similar  trends  as  in  the case  of  ammonia
addition.

In contrast to other investigations the addition of ammonia to the reburning  gas
enhances the reduction efficiency of reburning significantly. The discrepancy of
the presented results to  those of other authors are believed  to be caused  by  the
high temperature of about  1300 C in the reduction zone, optimized mixing injection
and a residence time of one second. These  conditions  favour the formation  of  NH3
rather than HCN in the reduction  zone for  all three cases studied.  While the NO of
the  reduction  zone  is  completely  converted  to  NOX   in  the burnout  zone,   the
conversion of  NH3 to NOX is small. The  high conversion  of  HCN to  NOX  can  be
avoided. This is in agreement to Tagaki,  who reports a low conversion  rate  of  NH3
to NOX and a high rate of HCN to NOX /5/.
 INVESTIGATION OF REBURNING IN A 160 MW SLAG TAP FURNACE

 In order to show the effectiveness of NOX  reduction with pyrolysis gas as reburning
 fuel and to find out the main parameters, the fuel staged combustion was applied
 to a 160 MWe] power plant.
Furnace design and performance of the trials

Figure 4 shows the furnace  of the steam generator and the zones of the  fuel staged
combustion. The furnace consists of two molten  ash  chambers.  The  two burner rows,
consisting of four air staged burners, are arranged in two stages  at each chamber.
To lower the  NOX emissions of the molten  ash chambers, the old unstaged  burners had
been retrofitted by air staged burners. As a second method to reduce NOX by primary
measures, flue gas recirculation to the pulverizer mills had  been installed. The
achievable NOX emissions  by primary  NOX reduction had to be evaluated as the basic
emission level before starting reburning.

After the fuel lean combustion of coal dust in the molten  ash  chambers reduction
gas can  be  injected  to  the flue  gas by twelve  nozzles  for  each  chamber.  The
arrangement of reburning  fuel  injection is shown  in figure  5.  The flue gas at the
end of the first  zone  has a temperature of about 1400 - 1500 C. The injected fuel
is coke oven  gas,  which mainly consists of H2 (50%)  and CH4 (25%). The  addition  of
reburning fuel  causes  the formation of fuel radicals, which start the NOX reduction
process. The  residence time of the  flue  gas  under fuel  rich conditions in the
reduction zone is about one second at maximum thermal  load.
By addition of burnout air at the end of the separated flue gas channels behind the
chambers the  combustion is completed.
                                     7B-49

-------
The entire experimental program from September 1988 till September  1990  included
trials without reburning to determine the  initial  emissions,  trials  with  coal gas
as reburning fuel and experiments with ammonia addition into the reduction  zone and
to  the  burnout zone.  During  the experiments  about 100 process  variables were
measured for On-Line monitoring  and  stored  for later data analysis. Besides the
operational  flue  gas   analysis  in  the  furnace   and  at  the  stack,  flue gas
concentrations  and  temperatures  were  measured   in  cross  sections  behind the
chambers, in the reduction  zone and in the burnout  zone  for  a  better  understanding
of NOX formation and  destruction and to point out possibilities for optimization.

As the results of NOX emissions by reburning are a function  of the  stoichiometry
of  the  main heat  release  zone,  the  reburning zone and  the  burnout  zone, the
stoichiometries of the zones had  to  be calculated  accurately.  While  the air  flows
and the  reburning  gas  flows were measured,  a measurement  of the pulverized coal
flow was not available.

The  air  stream,  necessary  for the  stoichiometric  combustion   of  coal,  is
proportional to the ratio of thermal power and the  efficiency  of steam generation.

    Vair,stoich. = A * Pth  / ^F
The thermal power P^h can be calculated  by  the  superheater and reheater Jetstream
and the  temperatures and pressures necessary  for  determining  the corresponding
enthalpies. The efficiency of steam generation r?p is dominated by the  heat loss of
the flue gas. The variable A gives the necessary air for combustion of coal with
a thermal  input of 1 MW.  A is constant for  a  large  range of coals and not varying
with changing water or ash contents of the coal.

The stoichiometries computed by this method were verified by comparison with the
stoichiometries calculated from flue gas composition.
Results
Primary methods.  The results of the  primary  NOX reduction  (air  staging at the
burner, flue gas recirculation) are summarized in figure 6. The NO emissions are
plotted as a function of the recirculated  flue  gas stream.  Each point in  figure 6
relates to a value, measured every ten seconds.
The application of air  staging is  for this slag tap  furnace the more effective
method for reducing NOX emissions  than the application  of flue gas recirculation.
By air staging  at the burner without flue gas recirculation the NOX  emissions could
be lowered from 644 ppm to 500 ppm  NO  (5%  02).  When  10% of  the whole flue gas was
recirculated to the mills,  air staging  caused a reduction from 560 to 490 ppm NO.
By  application of  different methods  for NOX  reduction  at  the  same  time the
effectiveness of the single measure decreases.
                                     7B-50

-------
The initial  emissions for the reburning trials were  500  -  550  ppm, which could  be
obtained by air staging at the burner  and by  flue gas recirculation. The  initial
emissions refer  to an  unstaged  operation in  the  furnace,  what  means  that the
stoichiometry of the chambers and the overall  stoichiometry were  kept constant  at
1.2.

Reburning results.  Figure  7 shows the result of reburning with varying gas streams.
Each value corresponds  to a  trial of  at  least two  hours.  At a steam generation
power near full  load  the  NOX emissions without reduction gas are 520  ppm for  a
stoichiometry of 1.2.  By air  staging  in the furnace and at  a constant thermal load
the NOX emissions could be lowered to 460  ppm (\\  =  1.1, ^3 =  1.2).  The reduction
of the thermal power caused  in  this test no  significant change of  NO emissions.
Other tests  showed  a maximum  influence of reduced thermal load  of 20 ppm NO  for the
staged case. The reduction of the thermal power corresponds to the  heat input  of
the maximum gas stream.

By  increasing  the gas  stream at  a  constant  first  zone  stoichiometry,  the  NO
emissions decrease sharply. By supplying twenty percent of the total  heat input  by
the reburning fuel, NOX  emissions of 180 ppm (5% 03) could be  achieved. The  unburnt
carbon in the fly ash was 4%.

The dominating parameter for  reburning  is the stoichiometry of the  reduction zone.
Figure 8 shows NO  emissions  for  trials in 1989 and  1990 without measures for  an
improved reburning  as  described  later.  The  trials  were  performed  at  different
primary zone stoichiometries, burnout  zone stoichiometries and different  thermal
loads. If sufficient  air  is  provided  for the  coal  combustion in the molten ash
chambers, reburning  caused   no  increase  of  unburnt carbon  in the  fly  ash.  The
operation of the  first  zone  with  a  stoichiometry greater  than  1.09  for  the
existing,  non  optimized  coal   dust  distribution  to  the   burners  secured   a
satisfactory burnout of the  coal below the 5%  threshold value.

Figure 9 compares  the  measured NO concentrations in  the  reduction zone  without
reburning gas and with  a reburning fuel of 20% of the total thermal  input.  Without
reburning gas an uniform  distribution  of NO concentrations of 550 ppm (at 0% 02)
was  measured in  the cross  section  before  burnout  injection.   By  addition  of
reburning fuel of  20 % the  cross section measurements  showed NO concentrations
between  100  and  300  ppm  NO.  The  concentrations of NO  are  corresponding  to the
measured concentrations of CO, H2 and  CmHn (Figure 10). Near  the  furnace  wall  on
the side of the gas injection (left  side in figure 9 and 10)  and  in  the middle  of
the furnace  the  concentrations  of the combustible species are maximum. The  non-
uniform distribution is mainly caused  by an incomplete mixing of the  reburning gas
with the flue gas from the molten ash chambers. Further cross  section measurements
of flue gas concentrations behind the chambers show that the coal  dust distribution
to the burners  also  contributes to  an unbalanced distribution in  the reduction
zone. In the scope of  the investigations  the coal dust/air distribution  was not
optimized,  but it is  assumed  that a  control of coal  dust supply  to the individual
burners can  contribute to obtain lower NOX emissions.
                                     7B-51

-------
Mixing calculations. Experimental  investigations of Kolb /6/ with  natural  gas  as
reburning fuel pointed out the influence  of mixing on the NOX emissions for a fuel
staged combustion. By an optimized mixing of reburning gas  he  could achieve a 50%
reduction compared to the case without optimization. The effect of mixing phenomena
on the results at  the test  facility  of the University of Karlsruhe was minimized
by an optimized mixing.  The reduction zone in the slag tap furnace "Fenne 3" at  an
optimum mean  stoichiometry  consists  of areas with stoichiometries, which  differ
from the optimum stoichiometry, so resulting in higher NOX emissions.

In order to improve the mixing of the reburning gas and to  optimize NO reduction,
mixing  of the  reburning  fuel  was  calculated  by three-dimensional  fluid  flow
computations.
The grid used for the computations is shown  in  figure  11. Because of the  symmetry
of the  furnace  the fluid  flow was calculated  for a  half  of one chamber.  As the
combustion  of coal  dust  is mainly  completed  in  the chambers,  the  computation
disregards heat transfer processes by reaction and radiation.

The choice of the computation domain  considers  the asymmetric distribution  of the
velocities (Figure 12) at the location of reduction gas injection.  This is  caused
by the  return of  flue gas  from the chambers  to the  upstreaming gas in the  first
furnace duct. In the cross section above  gas  addition  an non-uniform distribution
of velocities can be seen with maximum velocities  near the  side wall and  the wall
opposite to the gas nozzles. At the wall  near the gas  nozzles recirculation  takes
place. In the following  cross sections the velocities  are more balanced, but still
showing basically the same tendencies.

The calculated stoichiometries in figure  13a  confirm the measured distribution  at
a cross section at the end of the reduction  zone. As  it was evaluated  in  the test
facility with a reburning fuel containing ammonia, NOX reduction  is optimum at  \2
= 0.9 and satisfactory for a reduction 0.82 < \2 < 0.92.
The computations  indicate  that  the area  with a stoichiometry  for  a satisfactory
reduction covers only 15% of the cross section. In 35 % of the cross  section the
flue gas atmosphere  is fuel  lean.

In order to improve  gas injection  the cooling air duct of the gas  nozzles  should
be connected  to  the existing flue  gas  recirculation.  Before  installation the
influence of  an  increased  mixing  momentum on the  stoichiometry distribution was
computated, as shown in figure 13b.
With flue gas as  additional mixing  momentum the area with a  satisfactory reduction
covers 60 % of the cross section at  the end of the reduction zone.  These results
of calculation were the  reason  to install  a provisional connection of the  flue gas
recirculation  to  the  gas  nozzles.   A  comparison  of measured and   calculated
stoichiometries showed a good agreement /9/.
Trials of improved  reburning. The  impact  of an increased mixing momentum on the
final NO emissions is shown in figure 14. The decrease in NO emissions in this test
was about 25 ppm.  The  effect  of  the more uniform distribution of reduction gas on
                                     7B-52

-------
the NO concentrations measured  at the end of the  reduction  zone is depicted  in
figure  15.  With  flue  gas  as  additional  mixing  momentum  the  average   NO
concentrations  are  reduced  by  40  ppm.  An  increased  reduction  of  local   NO
concentrations seems to be equalized by an increased conversion of the N-species
of the reduction to NO in the burnout zone.

The recirculation of  flue gas provided the possibility of ammonia addition  into the
reduction zone.  In order  to  quench the flue gas,  water  or  ammonia water can  be
injected into the flue gas. For these tests  a 15% NH3 concentration was used.
The  results  confirmed  the positive  effect of  ammonia  on  NOX  reduction.  The
experiment shown in figure 16 was carried out at a reduced thermal load  in order
to examine a  wider range of reducing zone  stoichiometries.  In the case without
ammonia addition (with flue gas)  no NOX minimum could be determined, with ammonia
injection the NOX  emissions  were minimum at \2 =  0.89.  Only for very fuel rich
conditions in the reduction zone \$ < 0.85 (reduction gas  fraction  > 25%) ammonia
addition leads to higher NO emissions. Figure 16 also demonstrates the effect  of
burnout stoichiometry ^3.  A decrease of X3 from 1.2  to 1.1 causes a decrease  in the
NO emissions  for the  case  with and without ammonia addition. The unburnt carbon  in
the fly ash was less than 4%.
Laser measurements /10/ of NH3  concentrations in  the  flue gas at the end of the
furnace detected in no case a measurable ammonia slip.

The addition of ammonia to the burnout air had only a positive effect for higher
NO emissions  or stoichiometries  \2  > 0-92 (Figure 17).  The temperature of  the flue
gas  after  burnout  air injection  is  between  1000 and 1150  C,  measured  at full
thermal load over the complete cross section of the furnace.

The reported results refer to a  two  chamber operation.  In one chamber operation
lower emissions could be determined, as shown in figure 18.  Each value in figure
18 corresponds to one test over several  hours.
The difference between one chamber and two chamber operation is the possible use
of an air stream to the chamber out of operation  as a further  burnout air, so that
in one chamber operation the burnout air can  be added in two stages.  In one chamber
operation  minimum  emissions of  130  ppm  at  5  %  02  could  be obtained   at
stoichiometries of the burnout zone beetween  >3 =  1.05 - 1.1 (without regarding the
air from the chamber out of operation).
In figure  19  the  unburnt  carbon  in the fly ash  is  plotted  as a function of the
reduction zone stoichiometry for the one chamber tests.
CONCLUSIONS

By application of reburning to a slag tap furnace a NO reduction from  520  ppm  to
minimum emissions of  130  ppm were obtained. The  investigations pointed out the
strong influence of reduction zone  stoichiometry  on the NO emissions. Mixing  of
reburn fuel  has  to be optimized and burnout zone stoichiometry should be as  low  as
possible to achieve  low NOX emissions.
                                     7B-53

-------
For the slag tap furnace "Ferine 3"  there  exists  a  further  NOX  reduction  potential
by
       optimizing the reburn fuel  mixing into the reduction zone,
       optimizing of the coal  dust distribution to the burners,
       arranging the burnout air injection in at  least two stages  and by
       addition of ammonia above the reburning  gas injection.
Measures to  increase  the fineness  of the  coal  dust  would allow to minimize  the
reburning fuel fraction.
ACKNOWLEDGEMENTS


This work was conducted under the contract of the  Saarbergwerke AG with  financial
support of the federal Ministry of Research and Technology (BMFT), Germany.
REFERENCES

1.  H.  Spliethoff.  "NOx-Minderung  durch  Brennstoffstufung  mit  kohlesta'mmigen
    Reduktionsgasen." VDI-Bericht 765, 1989, pp. 217-230

2.  K.R.G. Hein, D. Kallmeyer. "Stand der NOx-Minderung bei braunkohlebefeuerten
    GroBkesselanlagen." VGB Kraftwerkstechnik, June 1989, pp 591-596

3.  M. Araoka, A.  Iwanaga, M. Sakai.  "Application of Mitsubishi "Advanced MACT  "
    In-Furnace Removal  Process." 1987 Joint Symposium on Stationary Combustion NOx-
    Control, New Orleans 1987

4.  A.C. Bose, J.O.L.  Wendt.  "Pulverized  Coal  Combustion:  Fuel Nitrogen Mechanics
    in  the rich Post-Flame." 22ndt  Symp.  (Int.) on  Combustion,  The Combustion
    Institute, 1988, pp 1127-1134

5.  T. Tagaki, T. Tatsumi, M. Ogasawara. "Nitric Oxide Formation from Fuel Nitrogen
    in Staged Combustion:  Roles of HCN and NHi." Combustion and Flame 35, 1979, pp
    17-25

6.  T. Kolb,  W.  Leuckel. "Reduction of NOx Emission in Turbulent Combustion  by Fuel
    Staging / Effects of Mixing and Stoichiometry in the Reduction Zone."
    22nd Symp. (Int.) on Combustion,  The  Combustion  Institute,  1988,  pp 1193-1203

7.  S.L. Chen, J.M. McCarthy, W.D. Clark, M.P. Heap, W.R. Seeker, D.W. Pershing.
    "Bench and  Pilot  Scale Process Evaluation of  Reburning for In-Furnace NOx-
    Reduction"
    21st Symp.(Int) on Combustion, The Combustion Institute, 1986, pp. 1159-1169

8.  J. Ritz, T. Kolb, P. Jahnson, W.  Leuckel. "Reduction of NOx Emission  by Fuel
    Staging  Effect of Ammonia Addition to the Reburn  Fuel." Joint Meeting of the
    British and French Section of the Combustion Institute  (1989), Rouen, France

9.  H. Spliethoff,  B.  Epple, D. Renner. "Einmischung von Reduktionsbrennstoff oder
    Reduktionsmitteln in technische  Feuerungen" 6. TECFLAM Seminar,  Stuttgart 1990

10. H.  Hemberger,   H.  Neckel, J.  Wolfrum.  "LasermeBtechnik  und mathematische
    Simulation von  SekundarmaBnahmen zur  NOx-Minderung in Kraftwerken." 3.  TECFLAM
    Seminar, Karlsruhe 1987
                                     7B-54

-------
                       Main Fuel / Air
                        Reduction Fuel
                        Burnout Air
                                        Primary Zone

                                            X >  1
                    Reduction Zone

                        X < 1


                     Burnout Zone

                        X >  1
               Figure  1.  Principle of fuel staging (reburning)
  MAIN HEAT RELEASE ZONE      REDUCTION ZONE
                                       BURNOUT ZONE
 COAL DUST
AIR
REDUCTION GAS
BURNOUT AIR
            CharN   I
  FuelN
                      \
          Volatile N j OH, (
Figure  2. NOX production  and  reduction for a fuel  staged combustion with  coal  as
primary fuel and gas as reburning fuel
                                   7B-55

-------
CD

cn
CO
                   300
                E
                OH
                   200
 d
_o
'to

    100
                X
               O
                                    1.5
                                      NH3 (VolX)
                                         o* NO/NO,
                          0.80       0,85      0,90       0,95

                          Stoichiometry Reduction Zone \2
                                                                     O
                                                                    N
                                                                    "8
-



I

-------
   Burnout

   Air
  Coaldust

  burner \
            Reduction Gas
Figure 4.  Furnace of the 160 MWe]  steam generator Fenne 3
      Molten Ash Chamber 1
                                A
o
o
oo
o
o
(M
I   f  t f  f  I  I  M  t  I  t


I   t  t M  I  M  t  M  1
Molten Ash Chamber 2
                                       Coal Gas
Cooling Air /

Flue Gas
               9068
           Figure 5.  Reduction gas nozzles
                     7B-57

-------

1 1

ex
ex


CM
O
5?
e
_o
S
'6

I

.? Unstaged Burner Operation
i
v
>'>_ _

**. m-
";;; 1

i Staged Burner Operation
i t 4v"
* m j3S5*
'-'^: ! '^^&'
1 ?f
                  0       10000     20000    30000     40000    50000     60000

                 Flue Gas Recirculation (to the mills )     f m3/h 1


          Figure 6.  Results of air staging (burner) and flue gas
          recirculation (to the mills)
r-,
0-






OJ
o
.V
in
0
z
DCJU
550
500
450
400
350
300
250
200
150
100
50
n
LOAD
92% ^ Unstaged
927 MJ?^ D S^S6*1 Operation
73% mFumaCe

Reburn Fuel Fraction
81% ^& 10%
' 86% ^ 14%
C*
'91% o^ 19%


   Stoichiometry  Primary  Zone


Figure 7.  Reduction by reburning -
influence  of reburn fuel  fraction
btau
'E 550
D.
Q.
1-1 500

450


400
350
300

^250
O
.v 200
in
o 150
inn
Two Chamber Operation
Non optimized Reburning o <^
<3>
oo o
%^O 
&* 0 0<>

^^^^
&&%
^^r
*
   .8    .9     1      1.1    1.2
  Stoichiometry  Reduction  Zone


Figure 8.   Trials  in  1988 and 1989
                                   7B-58

-------
                               Cmm
                without
1200
  ppm
  NO (05$ 02)reburning
'0
   I
 280   . .
  ppm With
  NO  reburning
                                                       4200
                                                                  [mm]
Figure 9.  Cross section measurement of NO in the reduction zone with and
without reburning gas (half cross section behind one chamber, gas injection
is located 12 meters below the depicted cross section on the left side)
   Figure 10.  Cross section measurement of unburnt gas in the reduction
   zone (half cross section behind one chamber, gas injection is
   located 12 meters below the depicted cross section on the left side)
                                   7B-59

-------
   Figure  11. Grid for fluid
   flow computations
Figure 12. Calculated distribution
of velocities
0,93 < X < 1,0


0,87 < X < 0,93

0,82 < X < 0,87 '

0,77 < X < 0,82 I
                                X > 1,0
           t    MM     !
                 0,93 < X < 1,0

                 -N-	^
              0,87 < X < 0,93
     HIM     I
        a)  Without Flue Gas Momentum     b)  With Flue Gas Momentum

       Figure 13.  Calculated distribution of stoichiometries without and
       with flue gas  as mixing momentum
                                     7B-60

-------
                 350
               |I
               E 325
               a
               D.
               "-1 300

                 275

                 250

                 225

                 200

               ~ 175
               C\J
               O
               .V 150
               in
               ^^
               ~ 125
                  100
Boiler Load    92%
Reburn Fuel
Fraction       19%
X Without Flue Gas Momentum
V With Flue Gas Momentum
                    .8     .85     .9     .95      1
                   Stoichiometry  Reduction  Zone
              Figure  14.  Effect  of flue gas  momentum
              on final  NOX emissions
280
 ppm Without Flue
 NO  Gas Momentum
0
                   middle of the furnace /
                      4300 /	
                  3800 /	
         Momentum
              3300 /	
           2800 /	
       2300 /	
         Figure 15. Effect of  flue gas momentum on  local
         NOX  emissions  in the  reduction  zone
                                7B-61

-------
JDU
'e 325
a.
Q_
1-1 300


275
250
225
200
~ 175
OJ
O
.\- 150
in
i 125
i on
Boiler Load 78-86%

V With Flue Gas Momentum
A With NH3 Addition to Flue Gas

V A \3 = 1,2
T A \3 = 1,1

/
/
//
/ /
/* /^ '
^^>
'"vT^^
--*-
25% 20% 15% Reburn Fuel
Fraction
JDK)
1  1
E 325
Q.
D.
1-1 300


275
250
225
200
~ 175
OJ
o
.V 150
in
o 125
i not
Boiler Load 78 - 86%

X Without Flue Gas Momentum
O With NH3 Addition to Burnout Air












25% 20% 15% Reburn Fuel
Fraction
100.6 .85 .9 .95 1 11JIJ.8 .85 .9 .95 1
Sto i ch i omet ry Reduction Zone St o i ch i ome t r y Reduction Zone
Figure 16. Effect of ammonia Figure 17. Effect of ammonia
addition to the flue gas addition to the burnout air
5001 	 in
'E 450
CL
CL
^ 400

350


300


250


200


~ 150
OJ
o
.v 100
in
o 50
~z.
Pi
One Chamber Operation
NH3 Addition to the Flue Gas
\3 = 1,05 - 1,2

A



.

A

&

A
A &
A

A 4^A




i V
9
n
.v 8
^
x 7
M
H
x B

L_
5
c J
.^
c 4
0 ^
J3
(0 g
0 J

ID 2
c
-D 1
D
n
One Chamber Operation
NH3 Addition to the Flue Gas
A
A X3 = 1,05 - 1,2





A
A
A


A A A
& A A
A A

"
^ A
A

AA



i .b .9 i 1.1 ^7 .8 -;g 	 i 	 ,;, 	
   Stoichiometry  Reduction Zone

Figure  18.  NOx emissions  for one
chamber operation  with  ammonia
addition to the  flue  gas
    Stoichiometry  Reduction Zone

Figure 19. Unburnt  carbon  in  the
fly ash  for  one chamber operation
(Corresponding to Figure 18)
                                7B-62

-------
COMPUTER MODELING OF N20 PRODUCTION BY COMBUSTION SYSTEMS

   Richard  K.  Lyon,  Jerald A.  Cole,  John C.  Kramlich,
                      and  Wm.  Lanier

      Energy and Environmental Research Corporation

                        18 Mason

                  Irvine,  CA 92718-2798

-------
            COMPUTER MODELING OF NaO PRODUCTION BY COMBUSTION SYSTEMS

    Richard K.  Lyon, Jerald A. Cole, John C. Kramlich, and Wm. Stephen Lanier

                  Energy and Environmental Research Corporation
                                    IB Mason
                              Irvine CA,  92718-2798


                                    ABSTRACT

       The observed rate of increase of NaO (0.181/. to 0.26'/. annually) is a
matter of environmental concern.  While it is generally agreed that this
increase is a result of human activity, there is no consensus as to the
relative importance of different sources.  Several studies have suggested that
pulverized coal fired combustion systems  might be responsible, but the high
levels of NeO found in these studies were later found to be an artifact, the
results of chemical reactions which occur during sample aging.  Measurements in
which precautions are taken against this  problem show very low NeO levels for
flue gas from pulverized coal firing but  do show substantial NeO concentrations
for fluid bed combustion.
       In this paper computer modeling calculations are done for two mechanisms
of NeO production,  the selective reduction of NO by HCN and sample aging.  The
former plausibly accounts  for the production of NS0 in fluid bed combustion and
may also be responsible for the small but apparently real amounts of NS0 found
in flue gas from pulverized coal firing.   Calculations for sample aging,
however, show that  preventing this mechanism from producing small amounts of
NE>O may be substantially more difficult than was initially believed.  Thus
sample aging may also account for the small amounts of NS0 presently found in
flue gas from pulverized coal firing.
       There have been speculations in the literature that the flue gas from
pulverized coal firing may be an important indirect source of N^O, i.e.  it was
speculated that chemical reactions which  occur during sample aging may also
occur in the flue gas after it is released to the atmosphere.  Our calculations
indicated that  this does occur but only to a very minor extent.
                                     7B-65

-------
INTRODUCTION

       The observed rate of increase of N^O (0.181/. to 0.267, annually) is a
matter of concern both because NP0 is a greenhouse gas and because it has a
major and unfavorable influence on the ozone layer (1,2,3).  While it is
generally agreed that this increase is a result of human activity, there is no
consensus as to the relative importance of different sources.  While McElroy's
calculations ( 3 , *t )  suggest that denitrification of chemical fertilizers could
account for the observed increase, others have criticized his calculations as
an order of magnitude too high (5,6). Weiss and Craig (7), Pierotti and
Rasmussen (8),  Hae et al (9),  and C. Castaldinin et al (10), have all reported
measurements of N^O emissions  by large stationary combustion systems, i.e.
pulverized coal fired utility  boilers and the like (11).  For combustion systems
fired with fuels containing chemically bound nitrogen (i.e. coal and heavy oil)
NF0 levels of approximately 25'/. of the NO emissions were found and there was a
strong suggestion that emissions at this level would be sufficient to explain
the observed increase.
       Recent experimental and computer modeling studies (12,13), however, cast
doubt on this conclusion.  In  all the studies mentioned above, grab samples of
flue gas which  contained both  NO and SOs were analyzed by GC several hours or
days after being taken.   Table 1 shows literature values for the rate constants
and/or equilibrium constants of a number of chemical reactions.   These
reactions are all well established processes.   Figure 1 from reference 13 shows
the results of  modeling  calculations done with this set of reactions.  The
prediction of these calculations is, that as the sample ages,  the NO in the
sample is converted to NO^, which undergoes solution phase reduction by sulfite
ion, first to nitrite ion and  then to the N0~ ion, with the N0~  ions then
reacting with each other to form NeO.  The amount of NE0 which this completely
a prior model predicts is in reasonable agreement with the amount observed
during the aging of a sample.
       Thus it  is entirely possible that the NeO concentrations  reported in
references 7 -11 are largely or entirely artifacts.  As discussed in references
I'*, 15 and 16,  recent measurements have been done in which precautions to
prevent this artifact were taken.  For conventional utility combustion systems
N.-.?0 levels of only Ippm  were typically found,  but considerably higher levels
have been found for fluid bed  combustion systems.  While NeO emissions of Ippm
would not appear to be of environmental concern, the mechanism by which they
are formed is still of scientific interest and the higher levels found for
fluid bed combustors are potentially an environmental concern.
       One of the issues to be addressed in this paper is the mechanism by
which this N,?0  formation occurs.  The other issue to be addressed herewith
relates to the  environmental importance of the NO/NOs/sulfite reaction
mechanism.  As  is pointed out  in reference 16 the absence of NeO in the flue
gases which combustion systems discharge to the atmosphere does  not necessarily
mean that these systems  are not important sources of NaO.   If the
NO/NOe/sulfite  mechanism is important in nature, the NO and 502  emissions of
combustion systems may cause substantial NE0 production after the flue gases
enter the environment.
                                      7B-66

-------
COMPUTER MODELING METHODS

       Calculations were done with the reaction rate model shown in Table 1
using an Acuchem program (17).  Additional calculations were also done with the
model shown in Table 2 using the PC version of ChemKin developed by Albert
Chang of Stanford University (18).
RESULTS AND DISCUSSION

Mechanism of Direct N20 Production during Pulverized Coal Firing

       As discussed above in recent measurements of N50 in flue gases of
pulverized coal fired systems precautions were taken against NS0 formation
during sample aging.  Since these measurements show greatly reduced but still
apparently real amounts of N^O one might conclude that some small production of
NpO does in fact occur during pulverized coal firing.  Since it is well proven
that fluid bed combustion produces large amounts of NeO one might plausible
concluded that whatever mechanism is involved there, is also operative to a
small degree during pulverized coal firing.  Alternatively one might conclude
that the precautions taken against the production of NaO during sample aging
were largely but not completely effective.
       The production of N^O by sample aging shown in Figure 1 is
oversimplified in one important respect: in Figure 1 it was assumed that all
the NOx in the sample is initially present as NO.  Figure 2 shows calculations
for the removal of NOx from the vapor phase by reaction with sulfite ion
solution for two cases, a gas mixture containing 600ppm NO and one containing
5^*0 ppm NO plus 60ppm N0e.  While the former shows a slow steady decay of the
NOx, in the latter case there is an initial drop which consumes much of the
N0;=.  Figure 3 shows the corresponding calculations for the production of hlNDs.
in the liguid phase.  As one might expect, when N0a is not initially present,
HNOa is formed slowly and only after an induction period, while when N0e is
initially present, there is a burst of HNOe formation at the start of the
reaction.  As shown in Figure ^ when N0e is initially absent, NeO is produced
only after a significant induction, but when it is present, the formation of
Nfc.0 begins immediately.  Indeed when NOK is initially present the sample need
only age for 12 seconds to produce 2ppm NeO.
       Thus for samples which initially contain NOS it is considerably more
difficult to avoid the production of NE0 by sample aging.  Consequently, if one
tests one's experimental procedures using synthetic gas mixtures which contain
NO but no IMOH, these procedures may appear adequate to prevent NaO production
during the sampling process, but still fail for real flue gases which do
contain NeO.
       In this regard, it is interesting to note, that in reference 15,
measured NsO/NOx ratios of 0.01 or less were found for 10 different coal fired
installations, but for a gas turbine a value of 0.21 was found.  If the NaO
found in these measurements is a result of inadequate precautions against
sample aging, one would expect the highest N;=0/N0x ratio to be found for the
installation in which the NOx contained the largest fraction NO^.. It is well
known that the NOx emitted by gas turbines can contain a much larger fraction
of N0a than found in other combustion systems.
                                      7B-67

-------
Indirect NgQ Production during Pulverized Coal Firing

       As mentioned above there is a question of whether or not the NOx and S02
in flue gas may not represent an indirect source of NaO.  When flue gas exits
the top of a stack, it both mixes with the atmosphere and cools to a
temperature that allows some of the water vapor it contains to condense.  Thus
two competing processes occur, i.e. formation of an aqueous phase allows the
processes which produced N^O in aging laboratory samples to occur in the flue
gas, but mixing with the ambient atmosphere will rapidly quench those
processes.  Thus one can imagine two ways in which flue gas can act as an
indirect source of NeO; some NS0 production can occur immediately after release
to the atmosphere and a much slower N,0 production might occur after the mixing
with the atmosphere via NOx and S0e  reacting in clouds.
       The former is a complex process and would be difficult to model
accurately but from the calculations shown in Figure ^ it seems likely that it
is a real but minor source of NeO.  In order to do calculations for the
production of NeO by reaction of NOx and S0e once they have been diluted to
ambient atmospheric concentrations a set of typical conditions was assumed.
Thus ambient concentrations of 6ppb and 10 ppb were assumed for NOS and  S0e
respectively.  L, the ratio of liquid phase to gaseous phase, was taken at  ^.8
x lO"7, a typical value for a cloud.    It was also assumed that the reaction
of NOp with SOE to form Nf?0 was in competition with other reactions and that
the most important of these was the reaction of N0e with OH to form HNOa.  The
ambient concentration of OH free radicals in the cloud was assumed to be 1.7 x
lO'6 molecules/cc and a rate constant of 1.1 x 10-! was used for the reaction
NOs. + OH = HN03.  (IB)
       Figure 5 shows the results of these calculations for a case in which the
aqueous phase was assumed to have an initial pH of 7.  The NOE + OH = HN03
reaction is found to be faster than NE0 formation by a factor of more than
101*.  Assuming an initial pH of less than 7 made NaO formation even less
important.  Thus production of N^O from NOx and SOe after they have mixed  in
the ambient atmosphere is trivial and combustion systems are indirect sources
of N^O only to the minor extent that NaO forms prior to the mixing of the  flue
gas with the atmosphere.


N50 Production during Fluid Bed Combustion

       While the very small concentrations of NeO currently being found in  the
flue gas of pulverized coal fired systems may or may not be real, the fact  that
fluid bed combustion can produce large concentrations of NaO seems to be well
proven.  Reference 19 reports an interesting set of experiments which may
provide an explanation for this high NeO production.  In reference 19 it is
reported that substantial NO reductions can occur in the free board of a fluid
bed combustion system and that these reductions can occur at temperatures  as
low as 1050C.K and reaction times as short as O.S sec. Since these NO reductions
occurred in the presence of V/, Os, some form of selective noncatalytic
reduction is clearly involved, but the observed NO reduction does not appear to
be due to reaction with NH3.  Thus the mechanism by which the NO was reduced is
unclear.
                                      7B-68

-------
       Figure 6,  quoted from reference 20 shows the result of flame modeling
calculations done with a reaction mechanism very similar to that shown in Table
2.  The model's prediction is that there exists a narrow range of temperatures
in which HCN selectively reduces NO,  the product of this reduction being N20.
Reference 20 also reports experimental results which confirm this prediction.
       Based on these results reference 20 suggested that NeO in the flue gases
from pulverized coal firing was produced by the following mechanism.  Nitrogen
containing char is produced in the primary combustion.   Some of this char
escapes the primary combustion zone and reacts to form HCN down stream at lower
temperature where the reduction of NO by HCN to form NaO is favorable.  This
reaction only produces NsO in a narrow range of temperatures because at
temperatures above this range N^O decomposes and at temperatures below the
range the HCN/NO reaction does not occur.
       Looking at Figure 6 one might  suppose that this mechanism for NeO
production is not applicable to fluid bed combustion systems because they
operate below the temperature window.  Figure 7, however, shows that the
temperature window for N^O production is a sensitive function of the reaction
time.  Selective reduction of NO to N^O by HCN can occur in the free board of a
fluid bed combustion system and thus  may be the explanation of the NO removal
reported by reference 19.
Practical Implications

       Fluid bed combustion is generally regarded as a developing technology
and hence the fact that fluid bed combustors may emit N^O might seem to be a
potential rather than an actual problem.  There is, however,  one application in
which fluid bed combustion is a major industrial process, fluid bed catalytic
cracking.  Within the cat cracking process the catalyst used  to "crack" higher
molecular weight hydrocarbons to smaller molecules becomes coated with  coke
and catalytic activity is restored by fluid bed combustion of the spent
catalyst.  The temperature of this combustion is low to protect the catalyst
and consequently any NE0  produced would survive.  Further,  the amount of
nitrogen in the coke which is available for NaO is substantial, since
chemically bound nitrogen in the hydrocarbon feed goes preferentially into the
coke.  Thus, since a major fraction of the world's total oil  production goes
through the fluid bed cat cracking process, it is quite possible that this
process contributes significantly to anthroprogenic NeO emissions.
                                      7B-69

-------
CONCLUSIONS

       Recent measurements of the N&0 levels in flue gas from pulverized coal
firing typically show'concentrations of a few ppm.   These NaO levels may be
real and the result of  the reduction of NO by traces of HCN,  or they may be an
artifact, a result of the fact that it is more difficult to prevent NE0
production by sample aging than was initially believed.
       While there has  been speculation that the emissions of S0e and NOx by
pulverized coal firing  may indirectly be a substantial  source of NS0, our
modeling calculations indicate that indirect NeO production is a minor process.
       Fluid bed combustion,  however, can produce substantial emissions of NaO
and our modeling calculations suggest that these emissions can plausibly be
explained in terms of the reduction of NO by HCN.  It is regrettable that no
data are presently available for the production of  NeO  by fluid bed catalytic
cracker regenerators, since these installations may be  a substantial source of
N,=,Q.
                                      7B-70

-------
                                    REFERENCES


 1  Weiss., R.F., J. Beophy. Res., 86,7185-7195 (1981).

 2  Khalil, M.A. and Rasmussen, R.A.,  Tellus, 35B, 161-169 (1983).

 3  Mat-land, G., and Rotty, R.M. J.A.P.C.A., 35, 1033-1038 (1985).

 4  McElroy, M.B., as reported by J. E. Bishop, The Wall Street Journal, p.9,
Nov. 13, 1975.

 5  Crutzen, P.J., Geophys. Res. Lett., 3, 169-172 (1976).

 6  Liu, S.C., Cicerone, R. J., Donahue, T.M.,  and Chameides, W.L., Geophys.
Res. Lett., 3, 157-160 (1976).

 7  Weiss, R.F. and Craig, H., Geophys. Res. Lett., 3, 751-753, (1976).

 8  Pierotti, D. and Rasmussen, R.A.,  Beophys.  Res. Lett., 3, 265-267  (1976).

 9  Hao, W.M., Wofsy, S.C., McElroy, N.B., Beer, J.M., Toqan, M.A., J. Geophy.
Res., 92, 3098-3194 (1987).

10  Castaldini, C., Water land, L.R., and Lips,  H.I.,  EPA-600-7-86~003a, 1986.

11  Ryan, J. V., and R. K. Srivastava, EPA/IFP workshop on the emission of
nitrous oxide from fossil fuel combustion (Ruei1-Malmaison, France, June 1-2,
19B8), Rep. EPA-600/9-89-089, Environ. Prot. Agency,  Research Triangle Park,
N.C., 1989. (Available as NTIS PB90-126038 from Natl. Technol. Inf. Serv.,
Springfield, Va.)

12  Muzio, L. J., and Kramlich, J. C., Geophysical Research Letters, 15, 1369-
1372, (1988)

13  Lyon, R. K., and Cole, J. A., Combustion and Flame, 77, 139 (1989)

14  Muzio, L. J., Montgomery, T. A., Samuelsen, G. S., Kramlich, J. C., Lyon,
R. K., and Kokkinos, A., 23rd Symposium (International) on Combustion, in
press.

15  Kokkinos, A, ECS UPDATE, Spring-Summer 1989, No 15 pp 8-10

16  Linak, W. P., et. al., Journal of Geophysical Research, 95, 7533-7541
(1990)

17  Braun, W., Herron, J. T. and Kahaner, D. K., Int. J. Chem. Kin. 20 51-62
(1988)

18  Baulch, D. L., Drysdale, D. D., Home, D. S. and Llyod, A. C., Evaluated
Rate Constants, Butterworth, 1976

19  Walsh, P. M., Chaung, T. Z., Dutta, A.,  Beer, J.  M., and Sarofin,  A. F..
19th Symposium (International) on Combustion, 1281-1289 (1982)

20  Kramlich,J. C., Cole, J. A., McCarthy, J. M., Lanier, W. S., and McSorley,
J.  A., Combustion and Flame, 77, 375-384, (1989)
                                      7B-71

-------
                   1000
DO
                    800 -
                    600
                 E
                 Q.
                 0.
                    400
                    200
EXPERIMENTAL

RESULTS
                                                     150       200       250       300       350
                                                    TIME, MINUTES
                Figure 1.     Experimental  and Kinetic Calculations of N?0 Formation in  Sampling Containers

-------
                           ppm
DO
I
-vl
CO
                                                                     [NO2]0 = 60ppm
                                  Figure 2.   Modeling of NOX  Removal from the Gas Phase by
                                              Reaction with  Sulfite  ion

-------
-xl
DO
                       40
                       30
                       20
                        10
                          ppm
                                                                           A
                                                                                        A
                              A
                                                                     J	L
                          0     10     20    30    40    50    60    70    80    90    100   110
                                                         t,  sec

                                          +  [NO2]o = 0     A  [NO2]o = 60ppm
                    1000ppm SO2, 600ppm NOx, 0 or 60ppm NO2,
                    40C, 6.52rnole% liquid water
Concentration of HNO2 expressed
   as ppm based on gas phase
                                Figure 3.   Modeling of HN02  formation  with N02 initially
                                           present and absent

-------
-si
00
Al
01
                     40
                     30
                     20
                     10
                        ppm
                      0
                       0
                    Time to form 2ppm N2O  12 seconds
100
200
  300
t,  sec
400
500
                                          [NO2]o = 0    -*- [NO2]o = 60ppm
                 1000ppm SO2, GOOppm NOx, 0 or 60ppm NO2
                 40C, 6.52mole%  liquid water
600
                              Figure 4.   Modeling of N20 formation with N02 initially
                                        present and absent

-------
-J
CD

-^J
CO
                                 10
20
           30     40    50    60     70
             TIME, seconds X1000

[NO2]/[NO2]i   -3-[N2O]/[NO2}\ X -\0000   &~
80
90    100
                                                                             [HNO3]/[N02]i
                     10ppb SO2, 6ppb NO2, Initial pH = 7
                     40C, L = 4.8E-7 ccL/ccG
                                 pH at 10E+5 sec - 3.33
                              Figure  5.   Competition between N20 formation and HN03
                                         formation after the flue gas mixes with the
                                         atmosphere

-------
  800-
 
-------
-vl
en
I
~sl
CO
                       200
                       150 -
                       100 -
                           ppm
                                  1.05
                                t = 0.02 sec
11      115.     1.2      1.25      1.3
       T, K  (Thousands)
                  1.35
           1.4
      t = 0.04 sec
t = 0.10 sec
t = 0.20 sec
                    200ppm HCN, 600ppm NO, 10% O2, 5% H2O,
                    balance inert
                               Figure 7.  Calculation of the Effect of Reaction Time on N20
                                        Formation

-------
                                    TABLE  I
    Chemical mechanism, rate  constants and equilibrium constants at  25C
             (rate constants  are  in  units  of L/mol/s or L /mol /s)
                                                                           12
         Gas Phase Reaction
1 .  NO  + NO  + 0
                  N02 + N02
                                                       Rate  Constant
                                                         6.73 E + 3
       Liquid Phase Reactions
   N02 + HS03-
                 N02- + HS03
   HS0
          HS0
                  H0)
       H2S03
   2N02 + H20 = HN02
                       HNO,
   HN02 + HS03-
NOS03-
                           H20
            H
   NOS03-
   HNO + HNO
                +' H20)  = HNO
               N2O
   NOS03-
           I- HS03-
9 .  HNO (SO3) 2~  + H*
10. HNO(S03)2~  + H2
   H20
f HNO(S03)
 (+  H20)  =
i  =  HONHSO,
                            2-
3.00 E+5
5.00 E+5
7.00 E+7
2.40 E+0
5.00 E+l
3.00 E+4
8.50 E+l
1.90 E-2
1.50 E-6
        Equilibrium  Processes
11.N02(gas)   N02(aq)
12 . S02(gas)   S02(aq)
                                                  Henry's  Law  Constants
                                                      H = 0.01 M/atm
                                                      H = 1.30 M/atm
13. S02(aq)
14.HNO,
             H  + HS03-
      ,   H  f NO -
15. HS04- = H'  S04
                                Equilibrium  Constants
                                    K = 1.54 E-2 M
                                    K = 5.10 E-4 M
                                    K = 1.20 E-2 M
                                    7B-79

-------
                              TABLE 2
                 ELEMENTARY REACTIONS USED IN MODELLING
REACTION

1   NH3+M=NH2+H+M
2   NH3+H=NH2+H2
3   NH3+0=NH2+OH
4   NH3+OH=NH2+H20
5   NH2+H=NH+H2
6   NH2+0=NH+OH
7   NH2+OH=NH+H20
8   NH2+02=HNO+OH
9   NH2+NONNH+OH
10  NH2+NO=N2+H20
11  NH2+HNONH3 + NO
12  NH2+NNH=N2+NH3
13  NNH+M=N2+H+M
14  NNH+NO=N2+HNO
15  NNH+OH=N2+H20
16  HNO+M=H+NO+M
17  HNO-t-OH=NO+H20
18  NH+02=HNO+0
19  OH+H2=H2O+H
20  H+02=OH+0
21  0+H2=OH+H
22  20H=0+H20
23  H+02+M=H02+M
    H20/21./
24  H+H02=20H
25  0+HO2=02+OH
26  OH+H02=H20+02
27  H02+NO=N02+OH
28  N02+H=NO+OH
29  N02 + ONO+02
30  N02+M=NO+0+M
31  0+0+M=02+M
32  N20+H=N2+OH
32  N20+M=N2+O+M
33  N20+0=N2+02
34  N20+0=NO+NO
35  CO+OH=C02+H
36  CO+H02=C02+OH
37  CO+02=C02+0
38  CO+0+M=C02+M
39  NCO+0=NO+CO
40  NCO+NO=N20+CO
41  NCO+H=NH+CO
42  NCO+NH2=NH+HNCO
43  0+H2=HNCO+H
44  NCO+OH=NO+CO+H
45  HNCO+OH=NCO+H20
4t>  HNCO+H=NH2+CO
47  HCN+OH=HNCO+H

-------
            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
            Mark D. McDannel,  P.E.
          Sheila M. Haythornthwaite
                    CARNOT
       15991 Red Hill  Avenue,  Suite 110
            Tustin  California  92680

          Michael  D. Escarcega, P.E.
            Barry L. Gil man, P. E.
      Southern California Edison Company
           2244 Walnut Grove Avenue
          Rosemead, California 91770

-------
                   LOW NOX LEVELS  ACHIEVED  BY IMPROVED COMBUSTION
                    MODIFICATION ON TWO 480 MW GAS-FIRED BOILERS
                               Mark 0.  McDannel,  P.E.
                             Sheila M. Haythornthwaite
                                       CARNOT
                          15991 Red Hill  Avenue,  Suite 110
                             Tustin, California  92680
                             Michael  D.  Escarcega,  P.E.
                               Barry L. Gil man, P.E.
                         Southern California  Edison Company
                              2244 Walnut  Grove Avenue
                            Rosemead, California   91770
ABSTRACT
While most  applications  to meet  new  and emerging  NOX  regulations  have focused  on
retrofit technologies (low-NOx burners, urea, SCR), there are still opportunities for
additional  NOX reduction  via improved  combustion optimization.
Southern California Edison, as part of their compliance  efforts  for  a new  NOX  rule,
which ultimately requires  NOX  limits  of approximately  20  ppmc,  retained Carnot  to
assist them in designing  and conducting  a combustion optimization program on two 480
MW gas-fired boilers.  As a result of  detailed  combustion optimization test  programs
on the two boilers,  NOX was reduced by 24  to  56% over the  load  range  at an average
cost-effectiveness  of  $.59/lb  NOX.   Through  increased  windbox  FGR,  improved  BOOS
patterns and overfire air,  NOX  levels  at  full  load were reduced from 91 to  62 ppmc.
These reductions will  help  SCE  meet  current  and  near-term  NOX limits,   and  will
substantially reduce  construction and  operating costs of any future  SCR systems.
                                        8-1

-------
INTRODUCTION
While  most  applications to meet  new and emerging  NOX  regulations  have  focused  on
retrofit technologies (low-NOx burners, urea, SCR), there are still  opportunities for
additional NOX reduction via improved combustion optimization.
Southern California Edison, as part of their overall  compliance plan  for South  Coast
Air Quality Management District (SCAQMD)  Rule 1135, retained Carnot to assist them in
designing and  conducting combustion optimization programs  on  two 480 MW  gas-fired
boilers (Alamitos 5 and Redondo 8).  This paper presents the results  of the two test
programs, which provided immediately implementable NOX reductions of 24  to  56% over
the unit load range at an average cost-effectiveness of $.59/lb NOX.
Included in the paper is a  description  of the technical  and  regulatory  background  on
NOX emissions from the two boilers, a description of the two boilers,  a  description
of  the approach  taken in  designing  and  executing the  program,  the  results of  the
program, and a discussion of the results.

BACKGROUND
All of SCE's  boilers  in  the  South  Coast  Air Basin are subject to SCAQMD Rule  1135,
which  includes system-wide  24-hour  average NOX limits that start at 1.10 Ib NOx/MW-hr
(approximately 100 ppmc*) in  1990 and steps  down to  0.25 Ib  NOx/MW-hr (approximately
23 ppmc) in  1999.   Additionally, Alamitos 5 and Redondo 8 are subject to  rule 475,
which  was  passed in  1970  and  limits  NOX  on  gas fuel  to  125 ppmc (approximately
1.38 Ib/MW-hr) for a 15-minute averaging  period.  Figure 1  presents  a summary of NO
limits on these two boilers.
When the 125  ppmc limit  was  imposed,  SCE implemented off-stoichiometric combustion
(overfire air ports and/or burners out of service) on 24 boilers in the  South  Coast
Air Basin,  and additionally implemented  flue gas  recirculation (FGR) to  the windbox
on four of these boilers, including  Alamitos  5 and  Redondo 8.  Implementation of  these
       ppmc = parts per million by volume,  corrected  to 3% 02, on  a dry  basis

                                        8-2

-------
techniques  reduced NOX  levels  from  approximately  900  ppmc  to  100  ppmc  on  both
Alamitos 5 and Redondo 8.
In  SCE's  overall   Rule 1135  compliance plan,  there  are a  number of  NOX  reduction
efforts either planned or already evaluated on these two units.   On Alamitos 5,  urea
injection and  installation  of one row of low-NOx burners have been tested,  and  the
installation  of  a Selective  Catalytic  Reduction   (SCR)  system  is   planned.    On
Redondo 8, an SCR  system consisting of  blocks  of  (honeycomb)  catalyst  placed  in  the
duct between the  economizer  and  air  preheater is scheduled for  1991.   It  is  within
this context that combustion optimization was evaluated and implemented.

UNIT DESCRIPTION
Alamitos 5 and Redondo 8 are two  of four  identical 480 MW  Babcock & Wilcox  opposed-
fired units operated by SCE  in the South  Coast Air basin.   The units are  capable of
firing either natural  gas or fuel  oil.  This program addresses gas firing only,  since
Rule 1135 has limited application to oil  firing and since oil is rarely burned.
Relevant details on the boilers are listed below:
              Manufacturer:  Babcock & Wilcox
              Rated Capacity:   480 MW (net)
              Steam temperature:  1,000F superheat and reheat
              Steam pressure:   3500 psig (supercritical)
              Burner arrangement (see Figure 2):
               --  Opposed  fired
               --  32  burners,  16 per wall
               --  4 rows of 4  burners each on each  wall
               --  furnace  split  by division wall
              NOX control:
               --  third  elevation of burners out of service
               --  FGR to windbox
               --  OFA ports
              Newly installed  Rosemount digital  control system
              02  trim system in  service
              CO trim system installed but not yet in service
PROGRAM DESCRIPTION
The objective of the program was to determine what  level  of NOX  reductions  could be
achieved by modifying and optimizing combustion and boiler operating conditions prior
                                        8-3

-------
to the installation  of  SCR or other back-end NOX reduction technologies.   Specific
benefits expected were:
          1.   Help meet Rule 1135 limits immediately.
          2.   By reducing  inlet  NO   levels,  reduce the size  and  cost of
               future SCR installations.
A comprehensive program  involving  five discrete phases was designed.  The five phases
are listed below, followed by a brief description of each phase:
              Records search
              Interview operating staff
              Physical  inspection and repair
              Optimization test program
              Load  following tests

Records Search
The first step of the program was  to  review  available  test  and  operating data on the
units to help plan the test program.

Interview Operating  Staff
Interviews  were  held  with  station  engineers,  maintenance   and  instrumentation
supervisors, shift supervisors, and  boiler  operators  to familiarize test  personnel
with unit operation  and to familiarize station personnel with  the  objectives  of the
program.  Unit  operation was observed with at least two different shifts of operators.

Physical Inspection  and Repair
Prior to performance of  the combustion  optimization test programs,  thorough  boiler
inspections  were conducted  during  maintenance  outages.   The objectives  of  the
inspections and outages were to:
          1.   Evaluate the condition of  all  fireside operating equipment
               including fans, dampers,  and burners.
          2.   Identify any equipment requiring repairs or adjustments,  and
               verify that repairs were made.
          3.   Allow  the test  crew to  become familiar with boiler design
               and equipment.
          4.   Wash  boiler to provide a known cleanliness lever.
                                        8-4

-------
Performance of the inspections and repairs ensured that equipment problems would not
adversely impact unit operation during the test program.

Optimization Test Program
The optimization test programs consisted  of  105 tests  on  Alamitos  5  and 51  tests on
Redondo 8.  The test matrices were  designed  to  evaluate the impact on NOX emissions
from the following variables:
              Unit load
              Excess 02
              Flue gas recirculation (FGR) to the windbox
              Overfire air ports
              Alternate BOOS patterns
          t    Air register throttling to selected burners
              Superheat/reheat proportioning dampers  (Alamitos 5 only)
              Fan balancing

Each test included collection of gaseous emission data  at the economizer exit,  a full
set of unit operating data  from  the control  room,  and external unit  data as  needed
(damper  positions,  air  register  settings,  windbox  02,  etc.).   For  most  tests,
North/South composite data was collected.  This involved  collecting  average gaseous
data from the North side, average gaseous data from the South  side,  and a composite
sample.  For selected tests,  full  32-point gaseous traverses were performed.   When
test conditions were established  and unit data were  collected,  the  impact of test
variables on unit  heat  rate was watched  carefully.    The need  to  isolate one test
variable at a  time to  determine  its  impact  on combustion  did  result  in some test
conditions where operation was not optimum;  this was  considered in  evaluation  of the
results.

Load Following Tests
The test  programs  on both units were concluded with two sets of load following  tests.
These  tests  involved  establishing  recommended  low-NOx  operating  conditions  and
monitoring NOX,  02, and CO while ramping boiler  load  between 160  MW and  480  MW.  The
purpose of these  tests was  to  determine  if the   low-NOx operating  modes could  be
maintained,  and  expected  NOX  values  seen,   over  the  entire  load   range  with  no
operational  problems.
                                        8-5

-------
RESULTS
The results are presented separately for the two units,  as  follows.   For the sake of
brevity,  detailed  impacts  of  individual  test  variables   are  presented  only  for
Redondo 8; similar results were obtained for Alamitos 5.

Redondo 8
The tests identified  two  modifications to baseline operation (as  described under Unit
Operation) that resulted in significant  NOX  reductions over the  full  load range,  and
two further  modifications  that resulted  in  small  additional NOX  reductions.   The
modifications which reduced NOX significantly are:
             t      increasing  flue  gas recirculation to the  windbox  to
                   the  maximum  achievable; and
             t      minimizing excess  02 until CO formation  is seen.
Modifications which produced smaller NOX  reductions  are:
                   taking burner pair 6 out of service (while leaving the
                   air  registers open); and
                   opening  of the  OFA ports  at  480  MW and  during  load
                   following  tests.

The results of  combining  these techniques are  summarized in Table 1,  detailed in Table
2, and  illustrated in  Figure 3.  Note that Figure 3 does not include the opening of
the OFA ports,  which were only evaluated during  the  load following tests.

Increased  Flue  Gas Recirculation  effects are  shown in  Figure  4.    Test points  on
Figure 4 are scattered  somewhat due to the inclusion  of all  test  variables.   However,
the trend of NOX reduction with increased GR is  clear.   This is most notable  at  480
MW.  Higher GR was limited at this  load  because  of a  fan amp  limit.   If fan  capacity
could be increased to enable 25% GR,  the  projected NOX would  be  approximately  40  ppm
@ 3% 02 (see dotted extension line  on graph).

Minimizing excess  02 was performed  at 250,   360,  and 480 MW.   The 02  setpoint  for
minimum 02 was  determined by gradually lowering excessive air until  100  to 200  ppm of
CO was seen consistently at that condition.
Table 2  shows  the percent  reduction  attributable  to minimizing  02  at  the various
loads.  This reduction  increases with lower  load, and more  reduction  may be  possible
at 160 MW, where significant CO formation had not begun.
                                        8-6

-------
Taking burner pair 6 out of service reduces the NOX fairly uniformly across the load
range, as shown  in Table 2.   Figure  5 shows graphically the impact on NOX of taking
6 OOS.   The  reduction  caused by this modification is small,  but  the  improvement in
boiler operation  is significant.   Figure  6  shows  the  CO level  both with and without
6 OOS.  At 480 MW extremely high CO was created with all burners in service; this was
removed  by taking 6 pair OOS.
Another  impact of this modification was to  improve the excess 02 balance between the
north and south  sides  of the boiler.   A series of tests  led  to the conclusion that
Burner 6 south is starved for air.  This results in lower 02 and higher CO levels on
the  south  side.   Taking  Burner 6 out  of  service improved  both  02 and  CO  balance
between  the two  sides.

Opening the overfire air ports at 480  MW reduced NOX by 7 ppm, or 11%.  This condition
was established while at full  load. Opening the OFA ports was not evaluated at other
loads due to difficulty in determining the  positions  of the  ports  early in the test
program.  Once the open  position was established by observing NOX reduction at 480 MW,
the ports were  kept  open  for one  set of  load  following tests.   Figure  7  shows  the
reduction  achieved  across the  load  range  by opening  the  NOX ports.   While  this
reduction  is  small,  the modification does  not impact  boiler operation,  and  could
easily be made a permanent operating  condition.

Load following tests showed that optimum low-NOx conditions  could  be maintained over
the full unit  load range,  without any operating  problems.  The results  of the load
following tests are shown in Figure 7.  NOX  levels are shown with NOX ports both open
and closed.   A slight  reduction  with NOX ports  open is  seen  over the  entire load
range.

Other variables that were investigated during the program were air register throttling
on  inboard  burners to  provide  increased  air flow  to  starved  outer burners,  and
alternate BOOS  patterns.    These tests provided  insight into  unit operation,  but
implementation caused undesirable effects  such as increased NOX,  difficult operation,
or a large 02 or CO  imbalance between  the  north and south sides of the boiler.

Alamitos 5
The tests  on  Alamitos 5 identified  three  modifications  to baseline   operation  (as
described under  Unit  Operation)  that  resulted   in  significant  reductions  in  NOX
emissions over the full  load range:  increased flue gas recirculation to the windbox,
opening  of the OFA ports,  and taking  burner  pair 6 out of service  (while leaving the
                                        8-7

-------
air registers open).  The results of combining these three techniques  are  summarized
in Table 3, and illustrated  in  Figure 8.
The results show that substantial NOX reductions were achieved across the load range,
with the percentage reductions decreasing as unit  load  increases  (from 56% at minimum
load to 27% at maximum  load).
Table 4 shows  the incremental reductions achieved  by  each  of  the three techniques.
The reductions achieved by each technique were cumulative  across  the full load range.
The largest reductions  (11  to 36%) were achieved by increasing  FGR to  the  windbox.
Reductions of 10 to 18% were achieved by taking Burner Pair 6 OOS, and reductions of
1 to 9% were achieved by opening the NOX ports.
Load following tests showed  that these  conditions could be  maintained over  the  full
unit load range, without any  operating  problems.  The  results of the  load following
tests are shown in Figure 9.
Other  variables  that were  investigated during the program were  excess  02  level,
superheat/reheat  proportioning  damper  position,  air register  throttling  on lower
burners to provide increased combustion  staging, air register throttling on  selected
burners in  an  effort  to overcome an air/fuel  imbalance,  FD and GR fan biasing  and
balancing,  and  alternate BOOS  patterns.    Those  tests provided  insight  into  unit
operation, but did not provide substantial NOX reductions.
Reductions  in excess 02 did  provide  some  NOX  reductions,  but the existing boiler 02
curve is so low (1% 02 over most  of the load range)  the 02 levels could only  be reduced
approximately 0.2%  before  the onset of CO.  Placing the  CO trim control  system  in
service will allow maintenance of minimum 02  levels over  the load range, and  should
result in additional NOX reductions of  2 to 5% (based  on  minimum 02 tests conducted
during this program).
The tests also identified a  significant north/south 02 imbalance  in the furnace.   A
series of tests led to  the  conclusion that  the imbalance  is mostly due to  burner 6
North  (an  upper,  rear,  corner  burner)  being starved  for  air.   The  problem  was
partially alleviated by taking the burner pair out of service for NOX  control.

DISCUSSION

This section presents discussions  on  the  potential  impact of the three recommended
combustion modification techniques (increased windbox  FGR,  Burner 6 out of  service,
minimum excess 02) on unit operation, including heat rate.   This  discussion applied
to both units.
                                        8-8

-------
Heat Rate

Any change in operation should be evaluated  in terms of its impact on unit heat rate.
Operating costs for a NOX  technique can become significant if they have a significant
impact on boiler efficiency.  Emissions data, unit operating data,  heat rate factors
and fuel cost factors were  combined to determine  an operating  cost  in  terms of $/lb
NOX reduced for increased  FGR,  taking Burner 6  out of service,  and  opening the NOX
ports.  The cost benefit of reduced excess oxygen levels was also considered.
Tables 5 and 6 summarize  the  heat rate penalties, and present  the operating cost of
the techniques combined in dollars per pound of NOX reduced.  On Alamitos 5, heat rate
penalties of $0.34 to $0.83/lb  NOX were seen.  On Redondo 8, the only  load  at which
a cost is seen  is 250  MW.  Here NOX costs $0.31/lb  reduction.  At all  other loads, the
heat rate is improved by reducing excess 02.
The results showed  two areas in  which  unit heat rate penalties were  incurred, and one
in which heat rate was  improved:  increasing FGR  to the windbox  increased auxiliary
power  consumption,  and  taking Burner 6 out  of  service  increased average excess 02
levels as measured  by the  test van.  Minimizing 02 reduced the NOX level  and  improved
heat rate by lowering the excess air  used.
It should be noted  that these cost-effectiveness  values  are so low  in  part because
these techniques involve an incremental extension  of NOX  reduction techniques already
implemented on the  boilers.  Costs for boilers which do not  already  have windbox FGR
or some form of off-stoichiometric firing would be higher.

Other  Impacts on Unit Operation
None of the four low NOX techniques used in this  study  had  any deleterious effects on
unit operation that  were detected during the test programs.  When  the techniques were
implemented unit load was  stable, flame appearance and  stability were acceptable, and
there were no significant  changes in tube metal  temperatures.
There are some areas in which the techniques might impact  unit  operation in  the long
run.   The most  important may be a loss in load capacity safety margin while operating
with Burner 6  out of  service.  In the baseline condition  there are 24 firing  burners,
and with Burner 6 out  of service  there are 22 firing burners.  If  a  burner pair trips
at full  load,  there would  be either  two or four fewer  firing  burners  in service
(depending upon whether  it is  an  upper or lower burner pair that trips).  With Burner
6 out  of service, it would be more likely that available  unit load would be curtailed
if a  burner pair tripped.   Prior to implementing  Burner 6 DOS, the  magnitude of the
possible curtailments  would need to  be  determined and  an  evaluation  made  of the
relative value of reduced  NOX  emissions vs.  the risk of increased load  curtailments.

                                       8-9

-------
Another area of concern with taking Burner 6 out of service is that the increased  heat
release rate per firing burner (an increase of 9% would occur) might cause overheating
in the burner throat area.   This would  have to be evaluated  prior  to  implementation.
Implementation  of  increased  flue  gas  recirculation  to  the  windbox   should  be
coordinated with appropriate safeguards,  since the  booster  fans have a high  enough
capacity that they can blow out the flames at lower loads.  New digital controllers
have been installed  on the booster fans,  the hopper control dampers, the FGR fans,  and
superheat/reheat proportioning dampers.  With the new booster fan controllers,  curves
of damper position vs. unit load can be programmed in.  However, to  protect against
injecting too much FGR there should be a windbox 02 monitoring  system. Such a  system
could be  either used for operator  information or  tied into  the  control  system  to
provide an alarm and/or  feedback signal.
Operating with the OFA ports open should not provide any operation problems. As noted
before,  it is currently difficult to access the OFA ports to  open  or close  them.   The
ports should  be welded  open.   The  chains currently  installed  do  not  allow  easy
operation.
An important aspect  to consider  in applying combustion  optimization techniques  is  the
boiler control  system.  These two boilers have newly installed digital  control systems
that  allow  effective and  safe  control  of  the  fuel  and  air  systems  within  close
tolerances.   On boilers with older control  systems it  may not  be possible  to achieve
such tight control.

CONCLUSIONS
The major conclusions of  the  program are:
             1.     Improved   combustion    optimization   can    provide
                   significant NOX reductions  (23  to 56%) beyond  those
                   achieved to meet compliance with the first generation
                   of SCAQMD NO, rules.
                              X
             2.    The incremental operating cost of these NOX reductions
                  is  negligible (average  of  $.59/lb NOJ  compared  to
                  retrofit  technologies.    In  some  cases  operating
                  savings are achieved due to excess 02 reductions.
             3.    These  techniques  can be  implemented  safely with  no
                  adverse impact on unit operation.
                                       8-10

-------
         IS
         
         a.
         Q.
1SO
165
150
135
120
105
 90
 75
 60
 45
 30
 15
                  NOTE:
                  ASSUMES UNIT HEAT RATE --
                  9,000 Btu/kW-hr
1.5
                                                         5
                   1970        1980        1990
                                  YEAR
                                                    2000
   Figure 1.  Gas Fuel NOx Limits on Alamltos Unit 5 and Redondo Unit 8
WEST FIRING WALL (VIEW FROM INSIDE)
D
0
75 0
0
3S0
D

8S
0
04S
D
0
5N0
0
1N0
D i
 ^
06N \
 ;
_2N
0 ;
 	 M< 	 W 	 < 	 M< -.W ..v
                                                      PLAN VIEW
                                              FIRING WALL
                                              FIRING WALL     FIRING WALL
                                                          DIVISION
                                                           WALL
                33'
                                                             FIRING WALL
                                                         48'
       Figure 2.  Burner and NOx Port Locations on Alamitos Unit 6
       (Redondo 8 Is a Mirror Image)
                                   8-11

-------
    I
          Q BASELINE
          O BEST. BUHNER B 003
              100    200    300    400

                UNIT LOAD, MW NET
 Figure 3.  NOx versus Load for Baseline and Best
 Conditions for Redondo Unit 8
         100


         90


         80


         70


         60


         50


         40


         3D


         20


         10


          0
            A 160 MW
            O 250 MW
            O 360 MW
            D 080 UW
INCREASING FGR
                    WINDBOX 02, %

Figure 4.  NOx vs. WindboxO  for Redondo Unit 8
                     8-12

-------
     o
     *
     n
     
     I
100


 M


 SO


 70


 80


 50


 40


 JO


 20


 10
             O BA8ELME, BURNER S M SERVICE
             A BEST. BURNER B M SERVICE
             O BEST, BURNER 6 009
             NOTE 3RD ELEVATION OF
                BURNERS DOS FOB ALL TE3T3
                100    200    JOO    400

                  UNIT LOAD, MW NET
    Figure 5.  Impact on NOx of Taking Burner 6
    Out of Service for Redondo Unit 8
      I
      o"
      u
             D BASELINE, BURNER i IN SERVICE
             - - A -  BEST. BURNER  M SERVICE
             - O - BEST, E DOS
            NOTE: 3RD ELEVATION OF
                BUHNER3 OO3 FOR ALL TE3T3
           100  150  200  250  300  ISO  400  450  500
                   UNIT LOAD, MW NET


Figure 6.  Impact on CO of Taking Burner 6 Out of
Service for Redondo Unit 8
                       8-13

-------
               -O BEST WITH OFA PORTS CUOSiD
               -A- - BEST WITH OFA PORTS OPEN
                   100    200    300    400

                     UNIT LOAD, MW NET
Figure 7.  NOx versus Load for Load Following Tests
at Best Conditions for Redondo Unit 8
        I
               	O	BASELINE, CLEAN
               	A	BEST, CLEAN
               - - -O- - - BASELINE. DIHTY
               ---O - BST, DIRTY
                              NOTE:
                              CLEAN AND DIRTY REFER
                              TO FURNACE CLEANLINESS
                   100    200    300    400

                      UNIT LOAD, MW NET
  Figure 8.  NOx versus Load for Baseline and Best
  Conditions for Alamltos Unit 5
                          8-14

-------
 C\J
CO
a
a

x"
O
100





 90





 80





 70





 60





 50





 40





 30





 20





 10
                100      200      300      400



                   UNIT LOAD, MW NET
                                               500
      Figure 9. NOx versus Load for Load Following Tests at Best

              Conditions for Alamltos Unit 5
                            8-15

-------
                                      TABLE 1

                              SUMMARY OF NOX REDUCTIONS
                               ACHIEVED IN REDONDO 8
                          COMBUSTION OPTIMIZATION PROGRAM
Unit Load
MW Net
Basel ine,
ppm NO 0 3% 02
Ib/MW-hr1
lb/MW-hr2
Best Case,
ppm NO @ 3% 02
lb/MW-hr1
lb/MW-hr2
% Reduction
(ppm 0 3% 02)
160

26
0.38
0.32

20
0.22
0.22
23%

250

39
0.55
0.47

24
0.28
0.29
38%

360

63
0.84
0.72

30
0.39
0.35
52%

480

88
1.19
1.02

55
0.73
0.64
38%

1   First lb/MW-hr number is calculated from plant CEM data divided by plant MW data
2   Second lb/MW-hr number is calculated from trailer N0x  ppm  and  Rosemount  heat  rate
by I/O method
                                      TABLE 2

                        PERCENT REDUCTION ACHIEVED BY THREE
                     NOX REDUCTION  TECHNIQUES  AT  REDONDO  UNIT  8
Unit Load
MW Net
Increased GR
to windbox
Minimize 02
Take Burner
6 OOS
Combining all 3
techniques
160

5%

21%*
5%

23%

250

26%

13%
7%

38%

360

43%

12%
5%

52%

480

37%

7%
6%

34%

* At 160 MW, 02 could be reduced further before  significant CO formation
                                       8-16

-------
                                      TABLE 3

                             SUMMARY OF NOX REDUCTIONS
                               ACHIEVED  IN  ALAMITOS  5
                          COMBUSTION OPTIMIZATION PROGRAM
Unit Load
MW Net
Clean Furnace
Baseline NOX:
ppm @ 3% 02
Ib/MW-hr
Best Case NOX:
ppm @ 3% 02
Ib/MW-hr
% Reduction
Dirty Furnace
Baseline NOX:
ppm @ 3% 02
Ib/MW-hr
Best Case NO :
ppm @ 3% 02
Ib/MW-hr
% Reduction
150 250 360


32 51 59
0.42 0.61 0.67

---- * 29 35
0.35 0.40
4*3
-------
               TABLE 4

     PERCENT REDUCTIONS ACHIEVED
BY THE THREE NOX REDUCTION  TECHNIQUES
            ON ALAMITOS 5
Unit Load
MW Net
Increase GR to
windbox
Take Burner 6 DOS
Open NOX ports
Combined techniques
150
36%
18%
9%
56%
250
36%
10%
9%
43%
360
29%
13%
9%
41%
480
11%
13%
1%
27%
                8-18

-------
                                      TABLE 5

            HEAT RATE PENALTIES ASSOCIATED WITH NO,  REDUCTION  TECHNIQUES
                                   REDONDO UNIT 8
Load
Increase FGR
(higher aux. power)
Burner 6 DOS
(higher 02)
Minimum 02
Net heat rate
penalty (gain)
Avg. heat rate,
Btu/kW-hr*
Base hourly fuel
cost, $/hr**
Efficiency penalty (gain),
$/hr
Ib/hr NOX Reduced
$/lb NOX Reduced
160 MW
0.06%

0.12%

-0.48%
(0.30%)
10,209
5,717
($17)
12
(1.42)
250 MW
0.16%

0.04%

-0.04%
0.16%
9,645
8,439
$14
45
0.31
360 MW
0.33%

-0.04%

-0.32%
(0.03%)
9,327
11,752
($4)
135
(0.03)
480 MW
0.21%

0 . 08%

-0.32%
(0.03%)
9,415
15,817
($5)
167
(0.03)
* Average of data collected during test program

** Assumes $3.50/MMBtu fuel cost
                                       8-19

-------
                                      TABLE 6

           HEAT  RATE  PENALTIES ASSOCIATED  WITH  NOX REDUCTION TECHNIQUES
                                   ON ALAMITOS 5
Load
Increase FGR
(higher aux. power)
Burner 6 OOS
(higher 02)
NO Ports Open
(higher 02)
Net heat rate
penalty
Avg. heat rate,
Btu/kW-hr*
Base hourly fuel
cost, $/hr*
Efficiency penalty,
$/hr
Ib/hr N0x Reduced
$/lb NOX Reduced
150 MW
0.33%

0 . 28%

-0.10%
0.51%

10,880
5,710
$29
35
0.83
250 MW
0.20%

0.12%

0.14%
0 . 56%

9,820
8,590
$48
66
0.73
360 MW
0.21%

0.04%

0.12%
0.37%

9,430
11,880
$44
99
0.44
480 MW
0.06%

0.12%

0.12%
0.30%

9,320
15,660
$47
137
0.34
* Average of data collected during test program

** Assumes $3.50/MMBtu fuel cost
                                       8-20

-------
NOx REDUCTION AND OPERATIONAL PERFORMANCE OF TWO FULL-SCALE
       UTILITY GAS/OIL BURNER RETROFIT INSTALLATIONS
                     N. Bayard de Volo
                         L. Larsen
            Energy Technology Consultants, Inc.
                     Irvine,  California

                          L.  Radak
                         R. Aichner
               Southern California Edison  Co.
                    Rosemead,  California

                        A. Kokkinos
             Electric Power Research Institute
                   Palo Alto, California

-------
             NOx REDUCTION AND OPERATIONAL PERFORMANCE OF TWO FULL-SCALE
                    UTILITY GAS/OIL BURNER RETROFIT INSTALLATIONS

                                  N.  Bayard de Volo
                                      L.  Larsen
                         Energy Technology Consultants,  Inc.
                                  Irvine,  California

                                      L.  Radak
                                     R. Aichner
                           Southern California Edison Co.
                                Rosemead,  California

                                     A.  Kokkinos
                          Electric Power Research Institute
                              Palo Alto, California
ABSTRACT

      In 1989-90 Southern California Edison Company replaced the original burners
firing natural gas and residual oil fuels in two large, opposed-fired boilers of
different capacities and design.  The replacement burners were manufactured by Todd
Combustion,  Inc of Stamford, Connecticut.  The principal objectives of the retrofit
were: 1) to improve flame shape and stability, and 2) to achieve NOx emission levels
with all burners in service at full load, in combination with Flue Gas Recirculation
(FGR), equal  to or less than the levels previously achieved by Off-Stoichiometric
firing with FGR.


      Tests were conducted on both boilers, firing gas and oil fuels separately, to
define the flame shape and stability and the NOx emissions over a wide range of
load, excess  air and FGR rate for both pre- and post-retrofit configurations.
Further reduction in NOx emissions achievable with the new burners firing in an Off-
Stoichiometric mode, with FGR, was also determined over the same range of
operational  variables.

      This paper is an interim status report presenting preliminary results of the
pre- and post-retrofit testing program funded by SCE and EPRI.
                                        8-23

-------
 INTRODUCTION

       In  1987 Southern California Edison Company (SCE) initiated projects to replace
 existing  gas/oil burners on two large boilers, Alamitos Generating Station, unit o,
 and Ormond Beach  Generating Station, Unit 2.  The principal  motivation in eacn case
 was to  improve  flame quality (stability, attachment, etc.) over the load range, out
 especially at low firing rates.  Additional motivations included improving boiler
 efficiency and  reducing NOx emissions.

       In  order  to define the actual improvements achieved by each retrofit, SCE
 instituted a program to perform comprehensive testing of both boilers before and
 after the burner retrofits.  EPRI provided additional funds to expand the parametric
 testing and to  promote the dissemination of the NOx technology results to the
 electric  utility industry.  Energy Technology Consultants, Inc. (ETEC) was retained
 to provide consulting services to plan and conduct the testing program, to analyze
 the test  results and to report on the program findings.  This paper is written to
 present some preliminary results comparing pre- and post-retrofit NOx emissions for
 natural gas and oil fuels.  The program is still in progress and a considerable
 portion of the  post-retrofit testing remains to be completed for both gas and oil
 fuels.  Nevertheless, because there is currently so little public information
 available on full-scale, Low-NOx gas/oil burner performance, it was thought to be
 useful to present these preliminary results at this time.

      Considerable success has been achieved by utilities having to comply with
restrictive NOx regulations applying to existing gas/oil fired units by implementing
Off-Stoichiometric (O.S.) firing.  In this mode of operation, selected burners are
taken out-of-service (BOOS) while fuel flow is compensatingly increased to the
remaining burners to maintain boiler load requirements.  As a result, the active
burner combustion process is made fuel rich and consequently NOx formation is
reduced.  Although NOx emissions can be significantly reduced in this manner for
both gas and oil fuels, operational performance can also be degraded somewhat as a
consequence of  having to raise excess air levels to maintain acceptable CO
concentrations  on gas fuel and plume opacity/particulates on oil fuel.  In addition,
a degradation in flame holding and stability can also result.  SCE has employed O.S.
firing on all of its units for many years achieving significant reductions in NOx
emissions but has also experienced the deterioration of boiler performance and
combustion on selected units.

      The basic concept of low NOx burners is to achieve fuel rich combustion, and
 hence reduced NOx formation, by controlling local  mixing of fuel and air.  This
 approach offers the promise of equaling or exceeding the NOx reduction capability of
O.S. firing while avoiding the possible performance and operational  deficiencies
 associated with the latter approach.   The potential gains however must be balanced
 against the capital cost of the burner retrofit in comparison to O.S.  firing which
 is implemented  operationally without equipment expenditure.

      This paper should be of interest to utilities who anticipate having a future
 need to reduce  NOx emissions from their gas/oil  fired boilers.   The subject program
 represents one  of the few instances in which data  are to be  developed for a low NOx
 burner utility  boiler installation and for which a comparison of the relative NOx
 reduction capabilities and overall performance of  the two NOx control  approaches can
 be established.  It is for this reason that SCE and EPRI have jointly funded the
 program reported herein.
                                         8-24

-------
PRE-RETROFIT OPERATION

      Unit 6 at the Alamitos Generating Station (AGS-6) is a B&W, opposed-fired,
gas/oil fuel boiler/turbine/generator set rated to produce 480 MWe.  The boiler was
designed with 16 two-burner cells, arranged in two rows of four cells on the front
and rear furnace walls.  Ring type gas spud burners and constant-differential,
pressure-atomized, swirl-tip oil burners were provided.  Dampered overfire air
ports, fed from the windbox, were provided above each top elevation burner cell.
Two gas recirculation fans were originally provided to extract flue gases from the
economizer exit and direct those gases to the furnace hopper area as an aid to
controlling steam temperatures at low firing rates (FGR).

      The boiler began operation in 1966 and subsequently became subject to a Los
Angeles County APCD regulation limiting NOx emissions to 225 ppm (dry, 3% 0?)  for
natural gas fuel and 325 ppm for fuel oil.  The uncontrolled NOx emission with gas
fuel at full load was approximately 700 ppm.  NOx emissions were reduced to within
the regulatory limit on both fuels by implementing O.S. firing.  The optimum firing
configuration was determined to be with the bottom burners of the upper cells (i.e.
3rd elevation) out of service for both gas and oil fuels and with the OFA ports
closed.

      Subsequently, the APCD NOx emission limit for natural gas was reduced to 125
ppm and 225 ppm for oil.  Two booster fans were installed to extract flue gas from
the main gas recirculation fan outlets and to inject the flue gas into the
combustion air through orifices in the flow-metering air-foils within the air ducts
between the air preheaters and the windbox as depicted schematically in Figure 1.
The combination of windbox FGR (WFGR) and O.S. operation achieved compliance with
the reduced emissions limits for both fuels and the boiler has been operated in this
mode ever since.

      Unit 2 at the Ormond Beach Generating Station (OBGS-2) is a Foster Wheeler,
opposed-fired, gas/oil fuel boiler/turbine/generator set capable of producing 800
gross MWe.  The boiler was constructed with two sets of 2-burner cells at each of
four elevations on the front and rear furnace walls.  Each two-burner cell is fed by
one gas and one oil supply pipe/valve, however, each individual burner had its own
air register control.  Each burner had a constant-differential pressure-atomized,
swirl-tip oil gun and a cane-type gas burner with (8) eight canes fed from an
external ring manifold.

      The boiler was originally designed to produce NOx emissions below 500 ppm
(dry, 3% OJ  for both gas  and oil  fuels.   This was to be accomplished by including
overfire air (OFA) ports fed by the windbox.  In 1969 it appeared that the Ventura
County APCD intended to establish a NOx emission limit of 250 ppm (dry, 3%02)  for
both fuels.   During construction of the OBGS units (1 & 2) WFGR was added to both
units.  For each unit one dual-inlet fan extracted flue gas from the economizer
outlet ducts and injected the gas into the two combustion air ducts leading to the
windbox.  The general configuration is depicted schematically in Figure 2.  The WFGR
injection is accomplished through an array of perforated pipes located within each
air supply duct a few feet upstream of the rear windbox.

      Upon commercial operation of OBGS-2 in 1973, compliance with the 250 ppm NOx
limit was achieved with either fuel at full  load by a combination of FGR, OFA and
limited O.S.  firing.   In 1975 the Ventura County  APCD reduced the allowable NOx
emissions with gas fuel  to 125 ppm (dry,  3% 0?).   Because oil  fuel  was used
exclusively  for several  years, compliance with the 125 ppm limit for gas fuel was
not demonstrated until 1977.  Compliance was achieved by operation with 8 BOOS,
                                        8-25

-------
maximum FGR (around 18%)) and load restriction to about 720 gross MWe.  The use of
the OFA ports was discontinued.

      Both units at OBGS have experienced severe boiler vibration under a variety of
"normal" operating conditions, possibly aggravated by the use of low-NOx firing
procedures.  The optimum operating modes were determined on the basis of compliance
with NOx emission limits and acceptable vibration control, and consisted of maximum
FGR at full load (throttled back at reduced load) and with 8 out of 32 burners out
of service (3rd elevation-gas fuel, 2nd  elevation-oil fuel).

      Several substantial efforts were made to alleviate the incidence of boiler
vibrations, including installation of burner air register shrouds and readjustment
of boiler back-pass dampers.  These efforts were partially successful in reducing
vibration.

      As with the ACS units, operation at OBGS increasingly emphasized reduced load
operation at times of off-peak-demand.  SCE determined that the flame conditions at
lower loads (ca 250 MWe) were not as secure as they desired.  In addition, the OBGS-
2 steam system was modified in 1985 to permit continuous generation as low as 50
MWe.  This increased the concern with flame stability (lift-off, etc.) at the
extremely low firing rates.

LOW NOx BURNER RETROFIT

      In  1986, the Steam Generation Division at SCE, in conjunction with the System
Planning  and Research Department,  contracted with Todd Combustion (formerly a
Division  of Fuel Tech,  Inc.) to provide 32 gas/oil burners to replace the existing
burners at AGS-6, principally to improve low-firing-rate flame conditions but also
to  provide reduced NOx  emissions.  Shortly thereafter, the Steam Generation Division
solicited competitive bids to provide 32 gas/oil burners for installation on OBGS-2.
The contract was also awarded to Todd Combustion.  Again, the emphasis was on stable
combustion at all firing rates, with low-NOx and increased efficiency as additional
objectives.

      Prior to installation of the Todd burners at AGS-6, SCE obtained a Permit to
Construct  from the South Coast Air Quality Management District (SCAQMD), which
stipulated that the NOx emissions post-retrofit must not exceed 113 ppm on gas fuel
and 203 ppm on oil fuel.  An additional requirement was that NOx emissions over the
load  range must be at least 10% below comparable emissions pre-retrofit, and that CO
emissions  could not increase.

      The Todd Dynaswirl burner relies upon control of the combustion air in
several component streams, as well as the controlled injection of fuel into the air
streams at selected points, for maintaining stable, attached flames with low NOx
generation.  Figure 3 schematically illustrates the internal configuration of the
burner.

      For  gas firing, fuel is introduced through six pipes, or pokers, fed from an
external  manifold.  The pokers have skewed, flat tips, perforated with numerous
holes and directed inward toward the burner centerline.  Gas is also injected
through a  central gas pipe with multiple orifices at the furnace end.  A single oil
gun is  located along the burner centerline, inside the gas pipe.

      Primary and secondary air streams flow from the surrounding windbox plenum
through a  spun cone inlet to the burner.  A shut off damper is provided at the
burner  inlet.  The primary air stream flows into the burner and down the center of
the venturi around the  center fired gas gun where it mixes with the center gas
                                        8-26

-------
forming a stable flame in front of the swirlers.  The secondary air flows  into the
burner flows near the outer walls of the venturi where it mixes with fuel  from the
gas pokers and is ignited by the stable center flame.  The testing air stream is
controlled by a separate slide damper and flows between the venturi evase  and the
burner throat quarl.  A piezometer ring is provided at the venturi vena contracta
for comparison to pressure at the burner inlet; the pressure signal of about 2.5
times the windbox to furnace pressure loss provides an accurate measurement of
combustion air flow rate.

      The oil gun is a conventional constant-differential, pressure-atomized burner.
The original single orifice swirl tip was replaced with a multi-orifice proprietary
design to reduce boiler vibration, however the turndown ratio is still of  some
concern, and efforts continue to improve the turndown while maintaining good flame
quality and low NOx emissions.  A swirl impeller is attached to the oil gun support
pipe just at the end of the primary sleeve section.

      In performance of the retrofit contract, Todd Combustion performed flow model
analyses of the windbox air flow distribution.  Based upon those analyses, baffles
and turning vanes were installed at selected points in the windbox to improve the
uniformity of air flow to all burners.

      Following selection of the Todd Dynaswirl burner for retrofit to OBGS-2, SCE
obtained a "Permit to Construct" from the Ventura County APCD.  The permit
conditions specified that the new burners would produce no increase in the emissions
of NOx, CO, total particulate and Volatile Organic Compounds (VOC), over the
operating load range, as compared to pre-retrofit emissions.  Windbox modifications
to improve air flow uniformity were also made on this unit.
TEST METHODOLOGY

      Comprehensive measurements of gaseous emission species (NOx, CO, 02) were made
for the pre- and post-retrofit testing phases of both boiler retrofits.  The scope
and conduct of both boiler test programs were essentially identical.

      Gaseous emissions were measured by an extractive sampling/conditioning/
measurement system contained within a mobile van.  Gaseous analyses included
chemiluminescent (NOx), non-dispersive infrared  (CO, C02)  and fuel cell (oxygen)
types.  All measurements were made after drying the sample gases.

      The sample flue gas was extracted through stainless steel probes located  in a
matrix across the economizer exit ducts.  Measurements could be made of any single
probe sample or a composite of any combination of probes.  Composite samples ensured
an equal portion from each probe by passing each individual sample through a
valve/bubbler prior to mixing within a common manifold.

      At AGS-6, a similar matrix of probes was located in the air supply ducts
between the air foils (FGR injection) and the windbox.  At OBGS-2 the  FGR/Air
mixture was measured by sampling from pressure-tap tubing located adjacent to each
burner air register.

      The FGR rate was calculated as the volumetric percentage of the  flue gas
extracted from the exit ducts and injected into the combustion air.  The calculation
was made based upon the dilution of gas species caused by the mixing process, i.e.
the comparative concentrations of 02,  C02  and  NOx within  the  flue  gas  alone and  the
flue gas/air mixture supplied to the burners.
                                        8-27

-------
      Since the OBGS-2 permit required a demonstration that particulate and
hydrocarbon emissions did not increase following the burner retrofit,  tests were
conducted to measure TSP (oil fuel only) and VOC (both fuels) pre- and post-
retrofit.  TSP was measured using a modification of EPA Method 5,  in which the back
end catch was analyzed in addition to the front end catch (filter plus probe
washing).  VOC was measured by capturing flue gas in Tedlar bags and analyzing for
C2 to  C8  by  GC/MS.   Triplicate measurements  of  Total Suspended Particulate  (TSP)  and
Volatile Organic Carbon (VOC) were made for each of four load levels from 250 to 700
MWe.  Analyses were made to determine the carbon content of the TSP filter catch and
the organic hydrocarbon content of the back-end catch.

      Each test was conducted with operation as close to steady state as possible,
with the load blocked on manual  control.  The boiler fuel,  air and steam controls
were generally on "automatic" except that excess air trim and FGR settings were
manually controlled.  In general, each test lasted from 30  minutes to 2 hours,
depending upon the complexity of gas measurement desired.  In addition to the
emissions measurements, considerable data were recorded regarding operating
conditions  (e.g., fuel and air flows, pressures and temperatures,  control/damper
settings, steam conditions, motor amps, boiler excess 02 and  stack opacity).


TEST RESULTS

      This  section of the paper presents a brief discussion of selected test results
acquired to date.  As pointed out previously, although pre-retrofit testing has been
completed,  only limited test data have been acquired for the post retrofit, low NOx
burner configuration for the two units.  Due to the limited extent of this latter
data and some present uncertainty in calculated WFGR rates  (discussed below), it is
premature to draw definitive conclusions as to the demonstrated NOx control
capabilities of the two Todd burner installations and comparison with the pre-
retrofit NOx control configurations.  This paper should be  viewed therefore as an
interim  status report which will be superseded by a future  publication documenting
the completed program.

      The testing of both units was constrained by the necessity to continue to
comply with the regulatory NOx limits of 125 PPM and 225 PPM respectively on gas and
oil fuels.  This constraint prevented testing to determine  the NOx reduction
capability  of the Todd burner by itself in the absence of the utilization of WFGR at
higher loads, since emissions compliance could not have been maintained.  This same
constraint  applied to the pre-retrofit testing relative to  demonstrating the
individual  control capabilities of WFGR and O.S firing on the two units.  Some
estimate of these individual influences for both NOx control  configurations for
Alamitos #6 have been made using historical data and FGR effectiveness trends as
discussed later in the paper.


ALAMITOS UNIT #6

      Figure 4 shows representative test results acquired for the Todd burner
installation on AGS-6 over the load range.  The calculated  WFGR rate and measured
average exhaust gas 0, concentration associated with each test  data point is
indicated.  In general, the data reflect the maximum NOx reduction capability of the
installation.  The indicated 0,  levels  at  the higher loads   ( >260 MW)  are  the
minimum achievable within the SCE constraint of maintaining exhaust gas CO
concentration below 300 PPM.  The lower load minimum 02 levels  are constrained by
the necessity to maintain a minimum level  of air flow as dictated  by safe operating
procedures.  The indicated WFGR rate at the highest loads is near  the maximum
                                         8-28

-------
capability of the WFGR system for the test conditions.  At the lower loads, the
indicated maximum WFGR rate is constrained by flame stability concerns although no
flame degradation in this regard was noted for the indicated levels.

      The upper data point shown in Figure 4 at 480 MW applies to the all- burner-
in-service (ABIS) mode of operation which was the intended employment by SCE for the
for Todd burner installation.  The level of NOx emissions achieved represents a
reduction of 87% for the combination of burner and 16% WFGR from the uncontrolled
level of approximately 700 PPM (best estimate based on historical data, could
possibly be higher).  At 19% WFGR, the maximum capability of the FGR system, NOx
emissions would have been in the range of 75 PPM (extrapolated from Figure 5 data)
representing an 89% reduction from uncontrolled baseline.

      The curve in Figure 4 is for O.S. operation with 8 BOOS.  Although the O.S.
mode of operation was not intended by SCE at the time for normal employment, SCE
wanted to demonstrate the maximum NOx reduction achievable since it now must comply
with a significantly reduced emission limit.  As Figure 4 indicates, the O.S. mode
of operation combined with 19% WFGR resulted in a further full load NOx reduction of
35%  (from 75 PPM to 49 PPM) which represents a 93% NOx reduction from the
uncontrolled baseline level.  This NOx control mode has been implemented by SCE for
normal operation.

      A comparison of pre and post retrofit test results for a range of WFGR rates
is shown in Figures 5-7.  The measured average exhaust gas 0? concentration
associated with each data point is indicated.  The single data point shown in Figure
5 for the Todd burner operating in an ABIS mode indicates that less NOx reduction
was achievable than for the pre-retrofit O.S. mode.

      With respect to the O.S. mode of operation, most of the post retrofit data
acquired thus far have been for higher WFGR rates than for the pre-retrofit data and
the minimal overlap for the two sets of data prevent a direct comparison over a
range of WFGR rates.  However, the data do seem to demonstrate consistent trends
indicating that the Todd burner is capable of achieving lower NOx levels in an O.S.
mode than was possible pre-retrofit.  This result appears to be due primarily to the
burner's capability to operate  at lower 0, levels  (discussed later)  since both  sets
of data show a clear trend of decreasing NOx with decreasing excess 02.   This may be
only a partial explanation and the Todd burner may in fact produce lower NOx
emissions than pre-retrofit operation at identical  excess 02 and WFGR levels.   A
regression analysis will be performed on the expanded future data base to more fully
assess this question.

      The WFGR rates were determined according to the procedure previously outlined.
There is a degree of uncertainty associated with the indicated values, however,
since a comparison between the calculated rates determined by the different methods
(02 or NOx  dilution)  showed  random differences in the range  of 10-15%.   Since FGR
exerts a strong influence on NOx level, this degree of uncertainty could result in
appreciable error in the data as plotted and misleading apparent trends.  This
potential  deficiency will be more fully assessed in the continuing program and it is
believed that the relative level  of uncertainty in calculated WFGR rates can be
reduced.

      Figure 8 shows a comparison between pre and post retrofit NOx control
performance capability for the various control configurations.  The NOx levels for
uncontrolled baseline and BOOS configurations are estimated based on 20 year old
test data.   The indicated NOx levels for the other configurations are either current
measurements or extrapolations from these measurements.  The comparison is tentative
since it  is based on current limited data but is presented to provide the reader
                                         8-29

-------
with a present estimate of the Todd burner NOx control  capability for the Alamitos
unit as well as a comparison with the pre-retrofit control  capability.  The
comparison indicates that for like configurations, there is little difference in pre
and post retrofit NOx control capability in absolute terms, the maximum being either
91% or 93%.  However, in relative terms, the difference of 16 PPM is significant to
SCE's NOx emission reduction objectives.

      The demonstrated percent reductions are measured  from an uncontrolled NOx
baseline level of 700 PPM.  Experience with implementing O.S. firing has shown that
the percent reduction  achievable on a particular unit  is dependent on the magnitude
of the initial, uncontrolled NOx emission rate and decreases as this rate is
reduced.  Therefore, it is likely that lower NOx control capability could generally
be expected for Todd burner installations on boilers exhibiting lower uncontrolled
NOx emission rates.

      Figure 9 compares pre and post retrofit C0/02 trends.   As shown,  the Todd
burner demonstrated significantly improved performance  over that achievable for the
pre-retrofit NOx control configuration.  This gain in minimum achievable excess 02
level is partly responsible for the lower NOx emission  rate obtainable with the Todd
burner retrofit and also offers a benefit in terms of boiler thermal efficiency.

      The  improved C0/02 performance of the Todd burner installation can be
attributed in  part to improved air/fuel flow uniformity to the burner arrays on the
two firing walls.  This was achieved by a combination of windbox modifications made
in conjunction with the burner installation and balancing of the burner fuel and air
flows during shakedown testing.  Therefore, part of the  NOx and heat rate gain can
be credited against the windbox modifications independently of the burner
installation and the remaining part to the burner itself.  The relative contribution
of these two factors has not yet been assessed but answering this question is useful
in terms of comparing the NOx control capability of O.S. firing (whose
implementation could be accomplished in conjunction with windbox modification) with
the installation of a Todd LNB.

      Figure 10 is a plot of recorded CEM data (note scale is in LB/HR) acquired
post retrofit  during the month of August, 1990 for unit operation over the normal
load range in  both AGC and operator control modes.  The significant data scatter can
be attributed  to the normal variability of key parameter settings such as excess 02
and FGR rate and instrumentation variability.  A similar plot has been prepared for
the pre-retrofit NOx configuration for the same period  in 1987.  Figure 11 shows the
best curve fits for each of the mentioned data sets and also a replot of the lowest
obtainable post retrofit NOx emission demonstrated as shown previously in Figure 4
(all in LB/HR).

      The plots illustrate that single point data  acquired in controlled testing of
the maximum NOx control  capability configuration can significantly underestimate
achievable operational emissions as monitored by a CEM  for demonstration of
regulatory compliance purposes.  A comparison of the upper two curves also confirms
that the Todd  burner installation was successful in reducing NOx emissions during
normal AGC operation.

      Figure 12 shows pre-retrofit NOx emissions at selected loads on oil firing for
the ABIS and BOOS modes  of operation.  Post retrofit oil firing data have not yet
been acquired  and the data are shown for general interest.

      In terms of operational performance, the Todd burner installation has
satisfied all  of SCE's original objectives with the exception of turndown on oil
firing which has not yet been demonstrated.  Flames are stable over the load range
                                        8-30

-------
including minimum load and do not exhibit any tendency to lift off under normal
operating conditions.  In addition, operating excess 02 level  has been significantly
reduced for gas firing thereby yielding a meaningful improvement in boiler thermal
efficiency.


ORMOND BEACH UNIT #2

      Figure 13 shows representative pre and post retrofit test results over the
load range for OBGS-2 firing gas fuel.  The data points apply to minimum excess 0,
levels and approximately to the same near maximum WFGR rate at each load level.  The
data indicate that the Todd burner installation reduced NOx emissions to below
obtainable pre-retrofit levels for the ABIS mode of operation and a further
increment in NOx reduction was achievable for O.S. operation (third row BOOS).

      Uncontrolled full load NOx emissions are believed to have been in the range of
1200-1500 PPM and therefore the controlled full load emissions for any of the
configurations (LNB or original burner with O.S. and with WFGR) represent a
reduction of at least 92%.  This magnitude of percent NOx reduction is nearly
identical to that achieved on AGS-6.  Unlike that unit however, post retrofit ABIS
NOx emissions at OBGS-2 are lower than the best obtainable pre-retrofit NOx
emissions by approximately 10% at full load.  The test results in the O.S. mode
shows an incremental reduction of 20% from the pre-retrofit level at full load as
indicated in Figure 13.

      The general range of pre and post retrofit CO concentrations measured verses
excess 02 is shown in Figure 14 for gas fuel  at loads of 550 MW and above.   The C0/02
trends are approximately the same for the pre-retrofit O.S. and post retrofit ABIS
modes of operation while post retrofit operation in an O.S. mode exhibited higher CO
concentrations at comparable 0, levels.   These results  are at   variance with those
demonstrated for AGS-6 which showed an improvement in the C0/02 post  retrofit trend
for the O.S. operating mode in comparison to pre-retrofit results.   CO
concentrations for this latter unit operating in an ABIS mode have not yet been
measured.  The results are surprising since the windbox modifications made to
improve air flow uniformity were expected to result in an improvement in the C0/02
trend as compared to pre-retrofit conditions.

      A comparison of pre- and post-retrofit NOx emissions for oil  firing is shown
in Figure 15.  The data indicate that the Todd burner achieved lower NOx emissions
at full load operating in an O.S. configuration than was obtainable for pre-
retrofit.  Since the data are limited and there is some uncertainty in the indicated
WFGR rates, further analysis is required to confirm this result.

      For gas fuel there was no increase in measured VOC emissions for operating
conditions consistent with lowest-NOx emissions, (O.S.  operation, low excess 02 and
high FGR rate).   Similarly for oil  fuel there was no measured increase in either
solid carbon or condensible hydrocarbons, again under lowest-NOx operating
conditions.

      The post-retrofit condition of the flames was substantially better than pre-
retrofit under all operating conditions, even at 50 MWe with all air registers open,
high FGR rates (up to 40%) and high excess air (25% of rated flow).  Under all
conditions the flames were closely attached to the burner tip/throat area and were
steady and symmetrical.   Prior to retrofit the flames were frequently detached from
the burner throat by as much as three to four feet, pulsated irregularly and were
occasionally irregular in shape.
                                        8-31

-------
      Prior to the burner retrofit, severe boiler vibration (rumbles and furnace
wall pulsations) were experienced under certain "normal"  conditions of load, excess
air, FGR rate and burner firing pattern.   Although the severe vibration could
usually be avoided, or corrected by an experienced operator,  the condition was of
concern to the operating and engineering  staff.   Following  the burner retrofit, the
unit generally operates more smoothly and the most severe vibrations no longer
occur.   It should be noted that simultaneously with the burner retrofit, the FD fans
were modified from constant-speed with inlet  vane flow control  to variable speed
with no inlet vanes.  Although it is uncertain whether the  fan modifications
contributed to the reduced vibration, the change has definitely reduced the
operating noise level  and has significantly improved the  control  and steadiness of
the air flow.


CONCLUSIONS

      It is premature in view of the limited  post-retrofit  test data acquired thus
far to  draw definitive conclusions relative to the pre and  post retrofit NOx
emission control performance comparison.   It  is  possible, however,  to make some
observations on the basis of the data that have  been acquired which are expected to
be valid at the conclusion of the program.

      1)  Full load gas fired NOx emissions for  both units  with the Todd burner
          installation combined with approximately 20% WFGR have been reduced by 93%
          from the uncontrolled baseline  NOx  level.   This reduction was achieved by
          operating in an O.S.  mode with  25%  BOOS.

     2)  The pre-retrofit NOx control  configuration of O.S.operation (25% BOOS)
          combined with 20% WFGR demonstrated nearly the  same NOx reduction as post-
          retrofit from the uncontrolled  baseline level for full  load gas fired
          operation.   The difference in demonstrated relative NOx control capability
          amounting to a further reduction of about  20% from  the pre- retrofit level
          could be meaningful  for utilities facing very stringent NOx emission
          control  regulations such as SCE.

     3)  Achievable NOx emissions employing  either control  configuration during
          normal  AGC operation  will  be significantly higher than that demonstrated
          in  the controlled testing conducted in this  program.

     4)  The C0/0?  performance  demonstrated  by  the  Todd  burner  installations  owed
          conflicting  trends in comparison to the pre-retrofit test results.   VOC
          emissions  on gas fuel  and particulates on  oil fuel  did not increase with
          the installation.

     5)   The burner retrofit demonstrated significantly  improved operational
          performance  relative  to pre-retrofit in terms of  flame holding, stability
          and boiler vibration.
                                        8-32

-------
00
I
CO
CO
        FUEL OIL ATOUIZER
        AIR SLIDE ACTUATOR
                               Figure 1: Todd Dynaswirr Low NOx Burner
                                                                                             Todd
                                                                                             Combustion
 GENERAL ARRANGEMENT
 DYNASWRL - LN BURNER
C/W CENTEfl FIRED CAS GUN
  AND POKERS

-------
CXI

CO
         Front

       windbox
     4

     3
Burner

level  2

     1
                                                                         Air from FD fans (2)
                                                              Damper


                                                          FGR fan (2)
                 Figure 2: Alamitos - Unit 6 Air/FGR Configuration

-------
00

CO
01
             Front
           Windbox
                                               FGR
                                             Injection
                                             Array (2)
                                                               Flow
Measurement
 Venturi (2)
                   Figure 3: Ormond Beach Unit 2 Air/FGR Configuration

-------
o
CO
Q.
QL
8
   100
    80 -
    60
    40
    20
            ALAMITOS #6
            GAS FUEL
             ABIS
            A 3rd Row BOOS
           Uncontrolled NOx Emission 700 ppm
               44
                100
                                                        16%WFGR
                                                        0.8 % Excess O.
                                                        500
                                                         600
                      200        300        400
                    GROSS GENERATION (MW)
Figure 4: Minimum Achievable Post Retrofit NOx Emission Over the Load Range
 O
 CO
 Q.
 Q.
 x"
 O
 z
I^U

110


100
90

80



60

50

40
2.0
A A 2.2

/
A2.2

-

-

ALAMITOS
GAS FUEL,
" Pre Retrofit



^2.3

A 2.8
/ .\ / \ / \ ^k ' 2
1.8 A2.2 *
A1.9
A2.0 Kl'7
Ai.e
#6 A
480 MW 1-3 A145
- BOOS
-  Post Retrofit - ABIS Aa9
- ^ Post Retrofit - BOOS M A

i
1.0
i i i i
               8
                        10
                                                 16
                                                          18
                        12       14
                        WFGR, %
Figure 5: Comparison of Pre and Post Retrofit NOx Emission at 480 MW
                                                               20
                                 8-36

-------
IU


^ 100
0"
^
CO
(8) 80
TJ,
C
Q. 60


o
""^
Z 40


20
ALAMITOS
GAS FUEL
A Pre Retrc
A2.6 A2.7 ApostRett
! 1 i i /\fc. 1
A2.5 A2.6
A1.8 A1.8
A2.3
~
A2.6
2.1AA1.9
A1p8
^i 4
A 2.5 % Excess 0 2
A. 95
-
A1.4 1' A
^K ^& M
A1.4
I I I I I

0 5 10 15 20 25
5 #6
., 360 MW
>fit - BOOS
ofit - BOOS












1.3
AL6
J1.0 A1-3
I '85

30 3
                          WFGR, %
Figure 6: Comparison of Pre and Post Retrofit NOx Emission at 360 MW
au

80

^
Cf 70
^
*? 60

S 50

a 40
x"
Q 30
Z
20
in
ALAMITOi
A GAS FUEl

2'8 A Pre Retrc
^2.3 A Post Retr


A.8
A2.3% Excess 02

A2.5
1.7 A A>.2
^1.7 1.8 ^21 A3.0
' 2'3AA?27A,.o A2.1
IQ A . Z_i2.6
* A 'a ^ A c
0.7 A 1.3 A A ^.6
0.9 *'
I I I
5 #6
_, 260 MW

>fit - BOOS
ofit - BOOS










*f
i
          10
40
                       20           30
                          WFGR, %
Figure 7: Comparison of Pre and Post Retrofit NOx Emission at 260 MW
50
                         8-37

-------
1,000
800
E 600
0.
0.
x"
O
Z 400
200
0
- 700
ppm
-

-
-
<*
:';X
v^'

"t

Pre Retrofit * * Post Retrofit



79% Reduction from ^^_
150 ^Ml Baseline 174(^^1
Uncontrolled BOOS BOOS ABIS ABIS BOOS
Baseline 19%FGR 19%FGR 19%FGR
   Figure 8: Comparison of Alamitos #6 Pre and Post Retrofit NOx Emission at Full
           Load on Gas Fuel
   700
   600
   500
Q.
   400
O
   300
   200
   100
  ALAMITOS #6
  GAS FUEL

  250-480 MW
Post Retrofit
 (O.S.)
     0.5
                              1.5           2

                          EXCESS 02, PERCENT
                                               2.5
            Figure 9: Comparison of Pre and Post Retrofit CO Emission
                                8-38

-------
   500


   450


   400


   350


   300
 . 250
X
O
Z  200
   150


   100


    50


     0
ALAMITOS #6
GAS FUEL
            50    100    150    200   250    300    350   400    450    500
                                    LOAD, MW
                 Figure 10: CEM Data for the Month of August, 1990
                                                                   550
500


450


400


350


300


250


200


150


100


 50
         ALAMITOS #6
         GAS FUEL
                                   Beat Fit of August, 1987.
                                   CEM Data, (Pre-Retroflt)
                                                        Post Retrofit
                                                        Minimum Achievable
                                     Best Fit of August, 1990.
                                     CEM Data, (Post-Retrofit)
            50
          100
                                                      400    450    500    550
                   150    200   250    300    350
                              LOAD, MW
Figure 11: Comparison of Pre and Post Retrofit "Best Fit" Curve of CEM Data
          and Post Retrofit Minimum Achievable NOx Emission
                                     8-39

-------
   170
   160
   150
CO


   140
   130
Q.
Q.
   120
   110
   100
   ALAMITOS #6
   OIL FUEL
   o ABIS
   A BOOS
                                                           O12.6%FGR
                                                             3.5 %Q,
                                 Ol7.9
                                   3.5
                                                             2.8 % O2
          Q24.2
           3.5
           20.7
           3.2
                                       \18.4
                                         3.6
     250
                 300          350           400           450
                                  LOAD, MW
                       Figure 12: Pre Retrofit Oil Fuel NOx Emission
                                                                     500
         ORMOND BEACH #2

         GAS FUEL
         A Pre Retrofit, O/S Firing
           Post Retrofit, ABIS
         A Post Retrofit, O/S Firing
         Uncontrolled NOx Emission
         1200-1500 ppm
     200
                                                                      800
                     400        500         600
                 GROSS GENERATION,  MWe
Figure 13: Comparison of Minimum Achievable Pre and Post Retrofit NOx Emission
         over the Load Range for Gas Fuel
                               8-40

-------
   1,200




   1,000




.^   800




 I.   600
 Q.

O
   400




     200
        Pre Retrofit (O.S.)

        Post Retrofit (ABIS)
                                                  ORMOND BEACH #2
                                                  GAS FUEL
                                                  550 MW - 750 MW
                 0.5                1                1.5
                    BOILER EXCESS 02, % (dry)
          Figure 14: Comparison of Pre and Post Retrofit CO Emission
 200


 180


 160
j

 140
ORMOND BEACH #2
OIL FUEL
O  Pre-Retrofit (ABIS)
A  Pre-Retrofit (O/S Firing)
  Post-Retrofit (ABIS)
  Post-Retrofit (O/S Firing)
() 120


Q 100


I.  80
Q.

X"  60
O

Z  40


    20


     0
                                       38, 2.05
                                          O
                                  A
                                   31,3.78
                                                     33, 2.55
                                                 26, .77'
                                           27, 2.0 '
                                                                 24, .73
                                                             25,2.950
                                                         7, 2.53
                                  438,2.13
                              41,1.0^
                                                                        . 20, 2.22
               200               400                600
                    GROSS GENERATION, MWe

Figure 15: Comparison of Minimum Achievable Pre and Post Retrofit NOx Emission
         Over the Load Range for Oil Fuel
                                                                               800
                              8-41

-------
COMPARATIVE ASSESSMENT OF NOx REDUCTION TECHNIQUES
         FOR GAS- AND OIL-FIRED UTILITY BOILERS

                       Gary L. Bisonett
                 Steam Generation Department
                Pacific Gas and Electric Company
                 San Francisco, California 94106

                        Mike McElroy
                Electric Power Technologies, Inc.
                   Berkeley, California 94705

-------
         COMPARATIVE ASSESSMENT OF NOx REDUCTION TECHNIQUES
                    FOR GAS- AND OIL-FIRED UTILITY BOILERS
                                 Gary L. Bisonett
                              Steam Generation Department
                            Pacific Gas and Electric Company
                            San Francisco, California 94106

                                  Mike McElroy
                            Electric Power Technologies, Inc.
                              Berkeley, California 94705
ABSTRACT
Pacific Gas and Electric Company conducted a comparative assessment of commercially available
NOx control technologies that might be applicable to our gas- and oil-fired boilers.  One phase
of the assessment, cofunded by EPRI, was a comparative cost and feasibility analysis of various
commercially available technologies, including combustion modifications, low NOx burners, and
selective catalytic reduction.  The results of this study are being incorporated into efforts
to identify a cost-effective system wide NOx control strategy for our system.  The comparative
assessment was conducted based on a typical boiler in our system to allow technology comparisons
to be made on a consistent basis.  Once the information for each technology was developed, the
site specific factors that affected each technology were identified so that the results could be
generalized and modified for other boilers in our system. One aspect of the project was to
develop a computer program, also cofunded by EPRI, to help PG&E compare various NOx control
strategies for possible application in our system. The computer program provides a first-cut
analysis of NOx reduction costs given different projected NOx limits and compliance strategies.
                                          8-45

-------
         COMPARATIVE ASSESSMENT OF NOx REDUCTION TECHNIQUES
                     FOR GAS- AND OIL-FIRED UTILITY BOILERS
INTRODUCTION

Pacific Gas and Electric Company (PG&E) performed a multi-faceted engineering program to
identify and evaluate options for reducing NOx emissions from its gas- and oil-fired electric
generating units. The program,  involving  the 39 boilers in the PG&E system, had two primary
goals:  (1) Evaluate  and compare the technical and economic merits of commercially available
retrofit NOx control technologies and their applicability to PG&E's boilers; and (2) Develop a
computer model to assist PG&E in developing an optimum system-wide NOx control strategy.
The program was prompted by concerns for lower NOx emission requirements  for California
utility boilers.

The program was performed with cofunding and technical participation from the Electric Power
Research Institute (EPRI).  The involvement of EPPJ was in recognition that the PG&E program
would be a valuable case study for the utility industry, and  the results could assist other
utility companies planning  or engaged in similar NOx control assessments.

PG&E is one of the  largest investor owned gas and electric utilities in the United States.
PG&E's fossil fuel fired electric generating capacity is centered in seven stations located
throughout the Company's service territory which encompasses much of northern and central
California. PG&E's gas- and/or oil-fired boilers total over 7,600 megawatts of electrical
capacity, and represent a wide cross-section of manufacturers, furnace designs,  combustion
systems, equipment  sizes, and vintages.  PG&E's 345 MW opposed-fired boilers (manufactured
by Babcock and Wilcox) comprise one-third of the capacity, and were the focus of the program.
NOx control measures have been previously implemented on these  and several other PG&E boilers,
including overfire air, flue gas recirculation, low excess air operation, and biased firing.

The California Clean Air Act which was passed in 1988 requires local air pollution control
districts to develop plans to attain ambient air quality standards in California. The
California ozone ambient air quality standard is 25 percent  more stringent than  the Federal
ozone standard. This requires a very aggressive program on the part of regulators to develop
plans to attain the California ozone standard. PG&E's goal is to work closely with regulators
to identify emission  reduction  plans that are both cost effective and responsive to the air
quality needs of the  communities we serve.

Since the completion of this study, PG&E has continued to develop site-specific information to
identify cost effective strategies for reducing NOx emissions. This program is  ongoing and will
continue as information from other installations, R&D, and the regulatory process becomes
available.
                                          8-46

-------
PG&E NOx CONTROL ASSESSMENT PROGRAM

Program Scope

The PG&E program consisted of the following work elements:

     1.  Verify existing boiler NOx emissions as a function of load for each boiler,
        using existing field test data, supplemented as necessary with NOx
        emission predictions based on furnace heat release rate correlations.

     2.  Compile detailed listings of boiler-specific operating and physical data
        that are related to NOx formation.

     3.  Evaluate the applicability and NOx reduction potential of operational
        modifications (e.g., bumers-out-of-service and biased firing) for the
        entire PG&E boiler population.  This work was based upon previous
        experience with such  controls within PG&E and elsewhere in the utility
        industry.

     4.  Assess the technical feasibility of retrofitting state-of-the-art low-NOx
        combustion systems for three selected boilers, and develop NOx reduction
        and cost factors for the technically feasible options.

     5.  Perform limited field  tests on one unit (Contra Costa Unit 6) to validate
        predictions of NOx reduction achievable by combustion modifications.

     6.  Conduct comprehensive technical and economic assessments for low-NOx
        combustion and Selective Catalytic Reduction (SCR) for a selected boiler
        (Contra Costa Unit 6).

     7.  Rank each potential NOx control option evaluated during the study by cost,
        NOx reduction potential, and technical risk.  Also, identify the site
        specific factors that influenced the rankings.

     8.  Construct a NOx emission forecast model which utilizes the above results to
        identify the NOx controls required to meet specified system-wide or
        regional emission limits at minimum cost.

     9.  Develop hypothetical NOx compliance strategies for different levels of
        system-wide NOx reduction utilizing the NOx emission forecast model.

Contra Costa Unit 6 was selected for the retrofit feasibility study  (Item 4 above), and for
detailed engineering and cost evaluations (Item 6), because it is representative of a boiler
design that constitutes one-third of the PG&E fossil system capacity.  Less detailed feasibility
studies where also performed on two other PG&E boiler designs which posed distinctly different
retrofit situations (Moss Landing Units 6 & 7,  and Pittsburg Units 5 & 6).  Each of the three
selected boilers were already operating in a reduced-NOx mode (with flue gas recirculation to
the windbox and combustion staging) which was the baseline condition for the feasibility and
engineering studies.
                                         8-47

-------
For the three plant sites, operation with natural gas and residual oil was considered.  Fuel oil
with a 0.5 percent sulfur content,  based on the maximum allowed by regulatory requirements, was
assumed.  Since fuel nitrogen content is not constant and variations would affect the NOx
reduction attainable by a given combustion NOx control, values of 0.3 percent and 0.6 percent
(by weight) nitrogen in the oil were considered for purposes of NOx predictions.

Description of Study Boilers

Contra  Costa Unit 6 - The unit is a forced draft,  opposed-fired, drum-type boiler manufactured
by Babcock &. Wilcox with a rated generating capacity of 345 MW (gross).  The unit was built in
1964. The unit fires oil and natural gas through 24 circular register burners arranged in two
rows of six burners on each firing wall.  The furnace contains two division walls separated from
the furnace end walls and each other by two columns of burners.  An elevation drawing of the
boiler is provided in Figure 1.  In 1973-1974, overfire air ports were installed to reduce NOx
emissions in order to meet new  NOx emission limits. Overfire air ports were installed  in the
windbox,  one above each burner column, for a total of twelve ports. In addition, the existing
hopper gas recirculation system was upgraded to mix up to 18 percent flue gas into the secondary
air duct feeding the windbox.

Moss Landing Units 6 and 7  These two identical units, rated at 750 MW  (gross), began operation
in 1967-68. These  units, manufactured by Babcock & Wilcox, are  forced-draft,  supercritical
boilers.  The units are opposed  wall fired and were originally equipped to fire oil or natural
gas with 3-nozzle cell burners arranged in a two-high by four-wide  array on each firing wall (a
total of 24 burner throats on each wall).  In the early  1970's, the existing hopper gas
recirculation  system was modified to permit operation with up to 18 percent flue gas
recirculation  with provisions to  direct recirculated flue gas to the windbox for NOx control.
Also, the top nozzles of the upper four cell burners on each wall were modified to pass air
only, acting as localized overfire air ports to provide  an additional NOx reduction.

Pittsbure Units 5 and 6 - The  two  identical units, designed by Babcock and Wilcox, began
operation  in 1960-61.  The units are forced draft,  natural circulation boilers, with a rated
generating capacity of 330 MW (gross when fired with either natural gas or oil fuel.  The units
were designed for future coal firing with a conversion to balanced draft.  The boilers are
opposed fired with 24 burners arranged in two-high by six-wide array on each wall.  In the early
1970's,  the units were modified to reduce NOx emissions by adding flue gas recirculation to the
windbox and installation of overfire air ports above the top burner row.

Program Participants

A majority of the work was performed by outside contractors selected on a competitive basis.
The major participants and their areas of prime responsibility are as follows:

      EPRI - Cofunding and  participation in project technical direction.

      Babcock  & Wilcox Company - Retrofit evaluation of low-NOx combustion
       equipment options and Selective Catalytic Reduction.

      Fossil Energy Research Corporation - Development of NOx Emission Forecast
       Model

       KVB. Inc.  - Compilation of current (baseline) boiler NOx emission factors,
       and evaluation of NOx  reduction via operational modification.
                                            8-48

-------
      Electric Power Technologies. Inc. - Provide technical and administrative
       support to PG&E, including assistance in program planning,  selection of
       subcontractors, and analysis of results.

      PG&E - Overall project management and NOx reduction field tests at Contra
       Costa Unit 6.
NOx CONTROL TECHNOLOGIES EVALUATED

The NOx control technologies that were considered in the NOx control evaluation include:

     1.  Operational Modifications to Existing Equipment

     2.  Combustion Equipment Modifications

          Two Stage Combustion (TSC)

          Reburning

     3.  Postcombustion NOx Control

          Selective Catalytic Reduction (SCR)

Operational Modifications.  The operational modifications evaluated were:  (1) low excess air;
(2) bumers-out-of-service (BOOS), including selected gas spuds out of service for natural gas
firing, (3) fuel biasing, (4) optimization of existing overfire air ports (where installed); and
(5) optimization of existing windbox flue gas recirculation (where installed).  Other
modifications considered, but not found to be cost-effective, were reduced combustion air
preheat and water injection.

Combustion Equipment Modifications.  The combustion equipment modifications were commercial
combustion systems, offered by B&W.  Each involved retrofit of low-NOx PG-DRB burners,
installation of dual register overfire air ports, and installation of a compartmentalized
windbox. Conceptually, the systems differed primarily in the arrangement and number of burners
on the firing walls, location of overfire air ports, requirements for additional furnace height,
and the control and distribution of air and fuel among the overfire air ports and burner
elevations.  Each system was evaluated for a range of flue gas recirculation rates, both within
the existing FOR capacity and under conditions of increased FOR capacity. The scope of
modifications and retrofit equipment associated with each combustion technology is summarized in
Table 1.

Four versions of rebuming were evaluated:

     (a) In-Fumace NOx Reduction (IFNR)
     (b) Pseudo-In-Fumace NOx Reduction (Pseudo-IFNR)
     (c) Derate In-Furnace NOx Reduction (Derate-IFNR)
     (d) Dual-Mode In-Furnace NOx Reduction (DM-IFNR)
                                           8-49

-------
Versions (b), (c) and (d) were essentially compromise designs which attempt to minimize boiler
modifications [e.g., minimize or eliminate need for additional furnace height] compared to a
non-compromise, full rebuming system [version (a)]. Pseudo-IFNR utilized minimum furnace
residence time criteria for rebuming reactions, Derate-IFNR involved a load reduction on  the
unit to satisfy rebuming residence time requirements, and DM-IFNR involved operation in an IFNR
mode below a certain load and TSC operation at higher loads.

A limited evaluation of B&W's XCL burners was also performed, as this technology became
commercial during the course of the study.

Selective Catalytic Reduction.  The postcombustion SCR technology was a commercial system
offered by B&W through a licensing agreement with Babcock-Hitachi in Japan.  The scope of
modifications and retrofit equipment are summarized in Table 1.
RESULTS

Operational Modifications

Maximum NOx reductions achievable from implementation of operational modifications to existing
combustion equipment were predicted to range from approximately 10 percent to as high as 60
percent from boiler to boiler (at full load).

The range reflects the varying degrees of NOx control already in place, and the  site-specific
factors that influence the applicability and performance of these controls. The NOx reductions
typically associated with each control technique are as follows:

                Operational Modification              NOx Reduction

                 Low Excess Air                       5-10 percent
                 Bumers-Out-Of-Service                15-60 percent
                 Fuel Biasing                          20-50 percent
                 Overfire Air Optimization               10-15 percent
                 FOR Optimization                     5-20 percent

In general, due to the low cost of implementing operational changes, these options should be
considered as a first NOx control alternative.

Combustion Equipment Modifications

State-of-the-art low-NOx combustion controls, aimed at achieving minimum NOx emissions via
modifications to combustion equipment  specifically, TSC and rebuming  were not universally
applicable to all boilers in the PG&E system.  Moreover, the predicted NOx reductions with these
technologies,  where technically feasible, varies considerably from unit to unit.  Predicted NOx
reductions range from 20 percent to as high as 70 percent from existing  levels, reflecting the
impact of site-specific factors, associated compromises in NOx control system design, and
specific NOx  control design and operating conditions.  These NOx reductions were calculated from
existing "baseline" boiler operating conditions in which the current use of flue gas
recirculation and various degrees of conventional combustion staging already result in reduced
NOx emissions.  Larger percentage NOx reductions would be expected if the study boilers had  not
been previously equipped with these NOx control measures.
                                          8-50

-------
The boiler-specific results concerning technical feasibility are summarized in the following
paragraphs.  The predicted NOx emissions are summarized in a separate subsection below.

Contra Costa Unit 6:  TSC could be applied, with burner rearrangement and significant ductwork
and windbox modifications. The relatively tight furnace, originally designed with minimum
residence times, would not accommodate any version of reburning without major extensions in
furnace height.  The change in furnace height required for implementation of IFNR is illustrated
schematically in Figure 2.

Moss Landing Units 6 & 7:  Application of low-NOx combustion systems is difficult due to
the 3-nozzle cell burner design, and the physical interferences from steam headers and mixing
equipment located halfway up the furnace walls in the windbox.  A TSC system could be installed
but would require major modifications to the firing walls, including complete rearrangements of
the burner array and windbox to accommodate new burners and overfire air ports. Pseudo-IFNR is
the only rebuming option determined to be feasible, but would require a substantial increase in
furnace height as well as firing  wall modifications similar to TSC. For both control options,
use of XCL burners instead of PG-DRB burners could reduce the retrofit complexity and cost.

Pittsburp Units 5 & 6:  The relatively high residence time in the furnace (originally designed
for future coal conversion) greatly enhances  retrofit feasibility.  TSC can be retrofitted with
only minor modifications to the overfire air ports (the PG-DRB burner would fit into existing
burner openings). IFNR can also be applied without  major furnace modifications~an additional
row of burners and new overfire air ports would be required.

Selective Catalytic Reduction

It is feasible to retrofit Selective Catalytic Reduction (SCR) to the Contra Costa Unit 6 to
achieve postcombustion NOx removals of approximately 80 percent. The design conditions and
operating parameters were concluded to be similar to SCR units operating in Japan.

Two possible SCR arrangement were evaluated for Contra Costa Unit 6:  (1) Base Case -single SCR
reactor located in the existing air heater location, requiring relocation of air heaters and FD
fans towards the stack; and (2)  Alternate Case - two  SCR reactors located above the existing
air heater locations,  with air heaters and fans undisturbed.  Schematics of both configurations
are shown in Figures 3 and 4.
                                           8-51

-------
NOx Reduction Summary for Control Options

The predicted NOx reductions for the combustion modification options are summarized in Table 2
for the three boilers evaluated.

Figure 5 compares the NOx reductions predicted for combustion modifications and SCR applied to
Contra Costa Unit 6.

Plant Impacts

For SCR, and the advanced combustion systems that were technically feasible, there appear to be
no adverse impacts on power plant performance, operation, or reliability that would preclude
their implementation.  However, potential impacts were identified and incorporated into the
overall evaluation of control options. The potential impacts considered include:

     Combustion Modifications

            Increased auxiliary power for higher FOR rates, where required.

            Potential increase in furnace  tube wastage due to reducing
            conditions.

         -  Boiler control system complexity.

            Changes in furnace excess air and resulting effects on plant heat
            rate.

         -  Boiler startup and shutdown procedures.
         -  Potential for flame impingement.
         -  Burner turndown.
         -  Restrictions on rate of load change.
            Potential localized connective pass tube overheating.

     Selective Catalytic Reduction

         -  Potential air heater plugging  when burning oil fuel.

         -  Increased minimum load or economizer bypass to maintain minimum SCR
            temperature.

         -  FD fan upgrading to overcome increased system pressure drop.

            Boiler startup and shutdown  procedures.

         -  Increased maintenance for SCR catalyst replacement and air heater
            cleaning.

         -  Air heater wash water treatment.

         -  Ammonia emissions.
                                           8-52

-------
Cost of NOx Control

The cost of retrofitting combustion modifications and SCR (1989 dollars) were evaluated
according the standard EPRI Economic Premises.  Capital costs ($/kW) included all materials,
engineering, installation, contingencies, and home office fees for a turn-key retrofit project.
Levelized costs (mills/kWh) included all operating and maintenance labor and materials,
administrative costs, and carrying charges. Levelized costs reported herein are for a base case
30-year levelization period and 30 percent capacity factor (other assumptions were evaluated in
the study to examine cost sensitivity to these parameters).

Low-NOx combustion system  costs estimated for Contra Cost Unit 6 ranged from approximately
$40/kW to $50/Kw, with total levelized costs ranging from approximately 3 to 4 mills/kWh.  These
cost  estimates are higher than generic cost estimates in the open literature.

The  capital cost of SCR ranged from approximately $72/kW to $82/kW, and total levelized costs
range from approximately 3 to 8 mills/kWh, depending on specific design and operating
assumptions.

A comparison of the costs of technically feasible NOx control options (TSC and SCR) for Contra
Costa Unit 6 are compared in Table 3.  Approximately 30 percent of the Engineering & Material
costs for TSC-are for low-NOx burners, burner accessories, and overfire air ports.  For SCR,
approximately 40 percent of the Materials & Engineering cost is for the SCR reactor vessel,
including the casing, framework,  and initial catalyst charge.

General Observations.  The  results of the study reinforce the following considerations
regarding the evaluation of utility boiler retrofit NOx controls:

     1. The selection of an optimum NOx control approach for a specific boiler is
        rarely obvious, without first performing detailed engineering and cost
        analysis of the available technology options.

     2. To provide a meaningful comparison of NOx control options, it is imperative
        that a systematic approach be used which analyzes each potential control
        technology under the  same technical and economic premises.

     3. Relying on generic technical and cost data is not advisable for evaluating
        retrofit feasibility, NOx control cost, and potential NOx reductions for a
        specific boiler or a utility generating system.  Such an approach could
       easily lead to substantial  errors relative to a systematic, detailed
       engineering and cost analysis of the same boilers.

     4. Depending on site-specific constraints  and NOx reduction requirements, it
       is likely that a combination of NOx reduction techniques will provide the
       overall least cost means of achieving those requirements.

Applicability and Value to Industry

The PG&E retrofit analyses involved a single boiler manufacturer's NOx control technology
applied to  a few specific boilers.  Although the technologies are representative of generic
classes of NOx controls that are offered by other vendors, it is likely that conclusions
regarding technical feasibility and cost would differ if performed by another manufacture
applying its versions of these technologies.
                                          8-53

-------
There are other boiler design types within the U.S. utility industry that are not represented by
the units selected for evaluation in this study. Such boilers, including tangentially-fired
units and cyclone-fired boilers burning gas and oil fuels, can be anticipated to pose
substantially different retrofit constraints.  Thus, a comparable feasibility analysis performed
on these units could have different results than those in this study.

Although the technical and cost evaluations may be pertinent to some retrofit situations
encountered elsewhere in the industry, for the reasons enumerated  above, feasibility and
engineering/cost analyses specific to each utility company are required.  However, the
methodology used in this study is generally applicable across the industry, and can be applied
by other utility companies performing NOx assessments of their generating systems.

The value of this methodology will be further demonstrated as PG&E proceeds towards final
selection and application of NOx controls for their generating system.


PG&E NOx EMISSION FORECAST MODEL

The PG&E NOx Emission Forecast Model determines the NOx emission controls required to meet
specified emission limits and their related cost to PG&E.  The costs are calculated both in
terms of capital costs and levelized costs.  The model also determines changes  in the system
heat rate due to the application of NOx controls. The model will allow  PG&E to evaluate various
load and fuel use scenarios with different emission  limits imposed. The model calculates annual
NOx emissions using boiler-specific information on operating hours and the loading, combined
with information on boiler specific NOx-versus-load and heat rate-versus-load curves.  The model
has the capability to take PG&E's "adjusted  load data" (a slightly modified version of the Total
Daily Production, or TDP,  files) and produce seasonal, monthly, and annual load profiles and
capacity factors for each boiler. Therefore,  although the model calculations are designed
around a system annual operating basis, year-to-year variations in load demand and fuel  use may
be accommodated.

A generic version of the model will be made available to EPRI member utilities as part of a
software system now being  assembled by EPRI.


CURRENT PG&E ACTIVITIES

PG&E is continuing  to develop information  on NOx control technologies that might be applicable
to our power plants.  We are conducting studies to evaluate NOx control cost and feasibility for
more of the boilers in our system. This information will be used as input to the NOx emission
forecast model to help us develop a cost effective system-wide NOx reduction  strategy.  Our goal
is to identify a range of NOx reduction strategies that are both  cost effective and responsive
to the  needs of the communities we serve.

We are also planning to  conduct a "proof of concept"  test using urea injection  on a 345 MW
boiler.  Urea will be injected into one-third of the flue gas in the convective pass of the
boiler.  The test boiler has two division walls that divide the furnace and flue gas paths into
three  flow streams.  The results of this test will be used to determine if urea injection has
the potential to provide cost effective NOx reductions on our 345 MW boilers.
                                           8-54

-------
                                                                             FIGURE  1
                                                                                       ot/nrr
                           PACIWC 0*5 t HKTtK COMPANY
                    CONTtA COnA TOWII flANT  UNrtJ NO.  AND 7

                                AKDOCM, CALIFOINLA
                           MiTPCf ft WHCO1 ftAAtAMT MHtAT MUtl
CATACTT. u HUM rv MOU*	      riueeo     KJHIHL'I cxjrxn nmuti
MAIMUU AllOWAtU WO*UN HUIML FV . .     l.UO     HH nUtUII
                          UMIAT tnAU TlktftUTUII. I       LOCI
                                       8-55

-------
            uMACt tut   4-
co

en
05
HOI por-j-






  UNf t(_ .







  U*M<*1 - .




  NOPPCII
                                             r-
                                                     POTS
                                                  -UINCIS
                                                                                  NOI PORTS
                      EXISTING   ARRANGEMENT
                                         NEW TWO-STAGE

                                      COMBUSTION  SYSTEM
                                                                                    URHttf
                                                                                    UONCMC
                                                                                    UDHttC
                                                                                                                          cn
                                                                                                                          m
                                                                                                                          oo
  o
  m


r> m
  21

  5  -n

S^
  i

  t/>
                                                                                      IN-FUHNACE NOX REDUCTION

-------
00
 I

Cxi
                                                                 EXISTIM5JNEW (EXCEPT

                                                                             NOTED)
                                                                                                                                                              CO
                                                                                                                                                              O
                                                                                                                                                              JO
   m
   3
   m


53
m 

5S
                                                                                                                                                              o
                                                                                                                                                              O
                                                                                                                                                              CO
                                                                                                                       EXISTING t LOCATION
                                                                                                                          OF F.D. FANS

-------
CD
 i
cn
CD






PLATFORM
EL. 83' -0-





1
PLATFORM
EL. 63--0-


R_ATFOW
EL. 49*-3*













'
L
SAMPLING CONN..

AMHO4IA
DISTRIBUTION 	 ,


\
\
WID \ \


1
,
k


^ 1
\ !

-

"7^
Y ;; i j!|
X

f
rr
i
i
\/..MUJlLUJ
,1
1









1












1 J t=;
-|.
 1



i




EXISTING










^
s 	


P













	 ~* '





-








-SELECTIVE
CATALYTIC
REACTOR







/-SAMPLING CONN.
j
I



/

I
1


l
EL. 8'-IO
NEW (EXCEPT
 ~ AS
NOTED)
?


A
3
-3/V J






--





t A.M.









T
 t 	 1
,
'
1








^
1



NEW LOCATION






	 .





OF EXISTING"

AIR HEATER

                                                                     !
                                                               	I
!   s
   i-
   A
                                                           EL. 9'-6-
                                                                ?6'-0-
                                                                                  18' -0'              32' -6"
                                                                                                         t  PROPOSED
                                                                                                         M COLUMN
                                                                                                         LOCATION
                                                     EXISTING  LOCATION
                                                         OF  F.D.  FANS
                                                                                                                                                                           l/>
                                                                                                                                                                           yo

                                                                                                                                                                           70


                                                                                                                                                                           tn
                                                                                            li;
                                                                                                                                                                           r>
                                                                                                                                                                           a
                                                                                                                                                                           


-------
                                                     FIGURE 5
                                             PREDICTED NOx EMISSIONS FOR
                                           CONTRA COSTA UNIT 6 - FULL LOAD
00
cln
CO
                        NOx, ppm @3% O2 (dry)
                  500
                  400 -
                          Original
                           Design
 Existing
(FGR+OFA)
TSC
IFNR
SCR
                                        Fuel Oil (0.3 N2)
                          Natural Gas

-------
                                                Table 1


                  MAJOR MODIFICATIONS AND EQUIPMENT ITEMS FOR
                     NOx CONTROL OPTIONS - CONTRA COSTA UNIT 6
      Two Stage Combustion
 In-Fumace NOx Reduction
         SCR (Base Case)
Fans and Ductwork:
-  Replace FGR fan rotor.
-  New FGR outlet ducts and
   dampers.
-  OFA ducts and dampen.
-  PC ductwork/piping and dampers.
-  Replace air heater outlet ducts.
Generally, same items as for TSC.
(Detailed design not performed)
   New FD fans, drives, and
   foundations.
   Increased stiffening on flues and
   ducts.
   Structural supports, platework,
   expansion joints, dampers, turning
   vanes, etc. for installation of SCR,
   relocated air heater, and new FD
   fans.
Boiler Modifications:
-  Partial replacement of sec.
   superheater (SSH) tubes.
-  Replace SSH attemperator to
   increase capacity.
-  Compartmentalized windbox.
Major extension of furnace height
(furnace bottom extended
downward) requires
modifications/replacement of
furnace wall panels, structural
supports, and water circuitry.
Compartmentalized windbox.
   Reposition air heater toward stack
   (install SCR reactor in existing air
   heater location).
   Modify furnace convection pass
   buckstay/support systems.
Combustion Equipment:
-  24 PG-DRB burners with
   accessories (installed in existing
   furnace openings).
-  12 Dual Register OFA ports
   (installed in existing furnace
   openings).
-  Modified fuel supply valving.
Generally similar equipment
items as for TSC except for
additional row of burners (i.e., 36
PG-DRB burners required).
-  None
Other
-  Boiler control system modifications
   (minimal)
Boiler control system
modifications and
instrumentation expected to be
more extensive than for TSC.
   SCR reactor vessel, incl. catalyst.
   Ammonia storage, vaporization, and
   injection systems.
   SCR controls and instrumentation.
   Modified underground utilities (due
   to interferences).
                                                 8-60

-------
                                                                    Table 2

                                             LOW-NOx COMBUSTION FEASIBILITY STUDY RESULTS

Test Case Description:
PG-DRB Burners
Burner Arrangement
Overfire Air Ports
FGR Rate
Predicted NOx Reduction
at Full Load:
Fuel Oil (0.3%N)
Natural Gas
Increased Furnace Height
Other Considerations
Preliminary Feasibility
Contra Costa Unit 6
TSC

24
2Hx6W
Opposed
12
20%

31%
61%
No

Yes
IFNR

36
3Hx6W
Opposed
12
20%

52%
73%
Yes

No
P-IFNR

36
3Hx6W
Opposed
12
20%

45%
70%
Yes

No
D-IFNR

36
3Hx6W
Opposed
12
20%

58%
75%
No
(1)
No
DM-IFNR

36
3Hx6W
Opposed
12
20%

30%
62%
Yes

No
Moss Landing 6 & 7
TSC

36
3Hx6W
Opposed
12
18%

21%
50%
No
(2)
Yes
P-IFNR

36
3Hx6W
Opposed
12
18%

54%
69%
Yes
(2)
No
Pittsburgh 5 & 6
TSC

24
2Hx6W
Opposed
12
18%

40%
58%
No
(3)
Yes
IFNR

36
3Hx6W
Opposed
12
18%

47%
66%
No
(3)
Yes
CO
 I
O)
            (1) Load restricted to 55-60% of MCR.
            (2) Existing 3-nozzle cell burners require extensive changes in burner arrangement and windbox to accommodate PG-DRB
               retrofit. Physical interferences from steam piping and mixing devices along furnace wall complicate retrofit.
            (3) Coal-design furnace provides sufficient residence time for combustion staging within existing furnace cavity.

-------
                               Table 3

            COSTS OF TSC AND SCR FOR APPLICATION TO
                       CONTRA COSTA UNIT 6
                                  Two Stage      SCR        SCR
                                  Combustion  (Base Case)  (Alternate)
   Capital Cost ($/kW)

   Material & Engineering               17.5        30.8        33.7
   Installation                         12.7        15.6        19.3
   Other (1)
   TOTAL CAPITAL REQUIREMENT    45.7        72.3        82.5
Levelized Cost (mills/kWh)

   Fixed and Variable O&M              0.8         1.1          1.2
   Consumables (2)                      0.0         1.3          1.3
   Carrying Charges (Capital)             23         4.5          5.2

   TOTAL LEVELIZED COST             3.7         6.9          7.7
Notes:
(1)  Includes contingencies, general facilities, taxes, and pre-production costs.
(2)  Includes replacement catalyst and ammonia for SCR.
                                8-62

-------
          ANALYSIS OF MINIMUM COST CONTROL APPROACH
     TO ACHIEVE VARYING  LEVELS OF NOx EMISSION REDUCTION
FROM THE CONSOLIDATED EDISON CO.  OF  NY POWER GENERATION SYSTEM
                          D.  Mormile
                          J. Pirkey
             Consolidated Edison Co. of New York
                         New York,  NY
                      N. Bayard de Volo
                          L. Larsen
                           B. Piper
                          M. Hooper
              Energy Technology Consultants,  Inc.
                          Irvine,  CA

-------
                      Analysis of Minimum Cost Control  Approach
                 to Achieve Varying Levels of NOx Emission  Reduction
           from the Consolidated  Edison Co. of NY Power Generation  System
                                     D. Mormile
                                      J.  Pirkey
                         Consolidated Edison Co.  of New york
                                     New York,  NY
                                  N.  Bayard de Volo
                                      L.  Larsen
                                      B. Piper
                                      M.  Hooper
                         Energy Technology  Consultants,  Inc.
                                      Irvine, CA
ABSTRACT

      Con Edison of New York operates a system of gas and oil  fired boilers for
power generation and district heating which is located in New York City.  Although
current NOx emissions from these boilers are in the range of NSPS limits, a further
reduction could be mandated as a consequence of a future NOx regulatory strategy to
achieve compliance with ambient ozone standards.   In recognition of this
possibility,  Con Edison initiated a program in 1989 to determine how NOx emissions
might be best controlled and at what cost.

      Tests have been conducted on each unit type/fuel combination to determine
current NOx emission levels and the reduction potential  achievable by employing
operationally implemented off-stoichiometric firing.  A PC based model of the system
has been formulated which can predict system NOx emissions integrated over any
potential  compliance period for the application of any unit specific combination of
NOx control technologies.  The model considers capital and operating costs on a unit
specific,  control  concept design basis and calculates system cost levelized over a
specified period for each case considered.

      This paper presents a review of the program status and a preliminary summary
of results obtained to date.  The program is not yet completed.
                                       8-65

-------
INTRODUCTION

      In 1989, The Consolidated Edison Company of New York,  Office of Environmental
Affairs, initiated a program to define cost-effective strategies to contend with
possible future NOx emission regulations.  The purpose of the program was threefold:

      1)  To assess the cost and effectiveness of all viable NOx control
          technologies as applied to the Con Edison fossil  fuel  boilers and to
          define the optimum means of achieving any specified level of NOx
          emissions.

      2)  To provide information to assess the economic and  emissions impacts of
          proposed regulation levels and forms so that Con  Edison might formulate
          a corporate position relative to rulemaking activities of regulatory
          agencies.

      3)  To identify areas to which Con Edison might best  direct internal R&D
          funding to nurture the development of NOx control  technologies to serve
          its future needs.

The program, still in progress, comprises four major tasks:  1)  testing of
representative boilers to characterize both the baseline NOx emissions throughout
the Con Edison system and the emissions reductions possible  with O.S. firing
techniques; 2) compilation and assessment of information on  the  control
effectiveness and application costs of all pertinent NOx control technologies;
3) formulation of a PC-based computer model of the Con Edison fossil  fuel boiler
system to permit assessment of baseline NOx emissions and the cost and NOx emissions
resulting from application of selected control technologies; and 4) analysis of
optimum NOx control strategies to achieve compliance with a  variety of potential
emission requirements, using the results from the previous  three tasks.

      The testing portion of the program consists of the measurement of NOx
emissions from a selected set of boilers representing the total  Con Edison
population of boilers.  Each boiler was tested with normal  firing procedures over
its firing range (load) and for each fuel (natural gas or residual oil) commonly
burned.  The baseline NOx emissions were characterized vs excess 0? level  at each
load level tested.  Measurement of 02,  CO and  NOx was made at multiple locations in
the boiler exit ducts using a mobile flue gas analysis laboratory.  On some boilers
tests were also performed to define the potential NOx reduction  achievable by firing
in an off-stoichiometric (O.S.) mode, consisting of shutting off fuel to selected
burners while leaving their air registers open, thus stratifying the air/fuel mix in
the combustion zone.  In all, 21 boilers have been tested,  out  of a total population
of 31 electric generation and 33 steam sendout boilers.

      The compilation and assessment of NOx control technology  effectiveness and
costs was accomplished with a combination of public and proprietary NOx emissions
test data for a wide range of control technologies.  To the  extent possible, the
available data were adjusted to reflect the most likely control  effectiveness and
cost of implementation which would occur upon application to specific Con Edison
boilers.

       A PC-based, spreadsheet model was composed to calculate  the NOx emissions,
electric and steam production, and fuel consumption of each   Con Edison boiler for
any specified time period, load schedule, fuel mix and NOx  control technology
implementation.  A discussion of some features of the program is contained below.
                                        8-66

-------
      Some preliminary analyses of optimum NOx control strategies have been
completed using the computer model.  The initial results are discussed in the paper.
The purpose of this paper is to present these preliminary results, which may be of
some interest to other utility and regulatory investigators.  The authors emphasize
that the analysis is incomplete at present.  Additional boiler testing is planned,
refinements are being incorporated into the computer model and the assessment of NOx
control technologies continues to be updated.

CURRENT OPERATION

      Con Edison operates a system of 64 fossil-fuel-fired steam boilers located
within the city of New York, ranging in size from 150,000 Ib/hr to over 8 million
Ib/hr steam capacity.  Eleven large boilers generate only electricity (173 to 972
MWe each) with condensing turbines.  An additional twenty boilers produce
electricity and also send out live, extraction or exhaust steam for commercial
heating use.  Thirty-three smaller boilers produce steam only for send-out.  The 64
boilers are distributed among thirteen separate plants in the boroughs of Staten
Island, Brooklyn, Queens and Manhattan.  Table 1 presents a summary description of
the boilers operated by Con Edison and included in the current analysis.  Additional
electric generating plants, partially owned by Con Edison but operated by others,
are not included in this study.  Similarly, combustion gas turbines are  excluded at
present.

      As shown in Table 1, some units burn either gas or oil fuel (or a combination
of both) while the remainder burn exclusively natural gas (60th St) or residual oil
(all of the rest).  Boilers with dual-fuel capability are generally restricted to
oil fuel in the months of December through February due to curtailment of gas
supplies.  When both fuels are available, current fuel prices generally favor gas
firing.  In recent years the relative system-wide fuel mix has been from around 50
to 75% oil on an annual basis (BTU value).

      The electric generating boilers represent a spectrum of tangential, face and
opposed fired boilers manufactured by CE, B&W and FW.  Most of these were originally
designed for coal firing and therefore represent relatively large furnace volumes
(and consequently, low NOx emissions)  for the unit firing capacity.  This
characteristic is discussed further below.

      The total capacity of Con Edison-operated fossil-fuel electric generation is
approximately 6,700 MWe of which about 5,100 is steam-electric located in New York
City.  The remainder comprises gas turbines and shares of steam-electric units
located elsewhere.  Figure 1 depicts representative monthly generation and fuel
usage projected for the early 1990's.  From the figure it is clear that two annual
peak generation periods occur, one in December/January and the other in July/August.
In 1990 the peak generation days were on January 8 and July 5.  As can be seen in
Figure 1 the total actual  generation by fossil-fuel steam units is around 40% of the
maximum possible over the year.

      From Figure 1 the seasonal shift in fuel mix is clearly seen, with oil
predominating from October through April  and gas fuel sharing the load throughout
the summer.   This seasonal fuel-mix characteristic has significant implications on
NOx emissions and control  strategies.

      As mentioned above,  the Con Edison boilers were, for the most part, designed
for coal  firing and therefore exhibit low NOx emission characteristics.  Table 2
shows a comparison between similar classes of boilers (size, design) at Con Edison
and at other utilities with typical gas/oil-design boilers.  All data shown are from
test  data acquired within the past several years.   The Con Edison baseline emissions
                                        8-67

-------
measurements have not been completed.  It is clear that the  Con Edison boilers have
considerably lower baseline NOx emissions with gas fuel than comparable boilers
elsewhere.  With  oil fuel the difference is not as clear, although the Con Edison
emissions are among the lower emission levels.  The principal implication of the low
initial (baseline) NOx emission levels at Con Edison is that the percentage
reduction in NOx emissions achievable with most NOx control technologies depends to
some degree on the initial NOx level prior to application of the technology.

      The baseline NOx emissions shown in Table 2 and used for analysis of potential
NOx reduction capability are derived from short-term, carefully controlled
engineering tests performed with steady-state boiler operation.   While these data
are useful for defining the effects of various controllable operating parameters on
NOx emissions, it should be understood that continuous, day-to-day operation of a
unit does not necessarily produce, on average, the same NOx emissions as a short-
term engineering test, even at nominally the same firing conditions.  Thus, there is
a degree of uncertainty as to the actual  NOx emission to be expected over a longer
time span.

      Under Automatic Generation Control  (AGC) the load on a unit (firing rate) is
controlled by a central dispatch computer and can cycle continuously over its normal
load range.  In this transient mode of operation it is not always possible to
maintain the "optimum" specified firing conditions (excess 02,  burner pattern,  etc)
vs. load.  In order to avoid unsafe conditions as the unit is automatically
controlled over the load range, operators will tend to set a safety margin of excess
02 above the ideal,  steady state point at a  given load  level,  and thus the NOx
emission will be increased somewhat.  Also,  over a longer period of time, boiler
furnace walls may become dirty between soot-blowing periods, burners may deteriorate
slightly and other uncontrollable factors may tend to increase NOx emissions over
the values defined in short-term testing.  Figure 3 illustrates the considerable
variability of baseline NOx emissions with AGC control  in comparison to the baseline
NOx emissions derived from short-term testing.  Thus, in order to maintain NOx
emissions consistently below a specified regulatory limit, the operator would have
to either reduce the average NOx emission well below the limit (so that the peak NOx
emission was still below the limit) or reduce the variability of the NOx emissions
about the average value by maintaining tighter control  of excess 02,  boiler wall
cleanliness, etc.

NOx CONTROL TECHNOLOGIES

      The technologies selected for inclusion in the study are those which have been
historically employed on an operational  basis for NOx control on gas/oil fired
utility boilers and certain other developing technologies close to
commercialization.  Descriptions of these technologies have been well documented in
the published literature and the discussion presented here is confined to pertinent
information relating to NOx control capabilities.  Considerable uncertainty exists
as to the control capabilities of most of the candidate control  options.  The NOx
reduction algorithms employed in the preliminary analysis are current best
estimates.  An effort is being conducted as part of the program to refine these
estimates for final  analysis.

OFF-STOICHIOMETRIC FIRING fO.S.)

      This control option has been effectively employed by a number of utilities to
achieve significant NOx reductions on gas/oil fired boilers.  Figure 2 (abstracted
from Ref.  1) shows the results achieved by one utility (Southern California Edison
Co.) employing O.S.  firing on a range of boilers firing natural  gas fuel.  These
results are representative of those demonstrated in other utility systems which
                                        8-68

-------
generally indicate a NOx reduction dependency on initial, uncontrolled NOx level.
The shaded area in the figure depicts the range of NOx reductions demonstrated in
the current Con Edison test program and confirms the dependency of control
capability on initial NOx level.  Similar trends have been demonstrated for oil fuel
firing.  The Con Edison O.S. test data generally fall in the range of 30% NOx
reduction, which is substantially less than the control  capability normally
associated with this technology but is explained by the low baseline NOx levels.

      The steady state, short term data acquired in the test program for O.S. firing
have been used in the analysis for the performance of this control option.  This
data may substantially overstate the magnitude of NOx reduction that could actually
be achieved during normal AGC operation.  Figure 3 shows a comparison between steady
state and AGC test data for one of Con Edison's units in uncontrolled and O.S.
operating modes.  The AGC data shows considerable scatter and does not reflect any
NOx reduction benefit for O.S. firing in comparison in the steady state data.
Similar data scatter has been observed for baseline operation.  The data scatter is
due primarily to variations in operating excess air and to boiler cleanliness
effects resulting from switching back and forth between natural gas and fuel oil
firing.  It may be possible to narrow the data scatter band by improving operating
procedures and air flow control, but differences between steady state and AGC NOx
emissions cannot be eliminated.  The implication of these results is that both
baseline and O.S. operating mode NOx emissions should be predicted on the basis of
AGC operation, which is the intent for the final analysis.

LOW NOx BURNERS (LNB)

      There are very few installations of LNB's on gas/oil fired utility boilers and
there is little published data reporting NOx control performance.  Ref. 2 provides
preliminary data for installation of one such burner design on two gas/oil fired
utility boilers.  The test results demonstrated an improvement over that which had
been achieved for O.S. firing in the range of 10-20%.  On the basis of these
results, the analysis assumes an NOx control performance for the LNB control
technology of 10% greater NOx reduction than that achieved in the O.S. testing of
the Con Edison units.

UREA INJECTION (UREA)

      UREA injection is a developing technology which is likely to have widespread
future application in utility systems for NOx control  Versions of this technology
are currently being demonstrated on several boilers in the Southern California
Edison system.  NOx reduction data acquired in these programs have been employed for
the present study to formulate a NOx control algorithm.   The data have been
extrapolated to lower initial NOx levels than tested by kinetic analysis.  The model
thus formulated was used in the analysis and is shown in Fig. 4.

      The EXXON Thermal DeNOx technology which is similar to UREA injection except
that the reagent is ammonia, could be employed as an alternative to UREA  injection.
For the purposes of this initial study, the UREA technology has been assumed to be
representative of this general category of NOx control approach.

WINDBOX FLUE GAS RECIRCULATION (WFGR)

      WFGR has been employed on both new and existing gas/oil fired boilers for NOx
control.  The technology has been demonstrated to be a very effective NOx control
option but little data exists in the published literature pertaining to  it's control
performance.  Reference 2 reports some data for two retrofit installations  in  the
Southern California Edison system.  This data has been utilized to formulate a  NOx
                                        8-69

-------
control model for natural gas and fuel oil firing which is shown in Fig. 5.  The
nitrogen content of the fuel oil applying to the test data is 0.3% which is
essentially the same as for the Con Edison fuel.

REBURNING

      The Con Edison boilers are particularly suitable for the application of the
Reburning technology because of their uncharacteristically large furnaces  for
gas/oil fired units.  This technology was not considered in the analysis,  however,
due to the lack of sufficient data to estimate NOx control performance, particularly
at low initial NOx levels.

SELECTIVE CATALYTIC REDUCTION (SCR)

      SCR was assumed to  have a NOx reduction capability of 80% for all initial NOx
levels.

SYSTEM NOx MODEL

      A PC-based spreadsheet model was written to calculate the NOx emissions and
cost of control for any combination of control technologies for the Con Edison
system, and for each  boiler unit individually.  The model  comprises three
functional areas: data input, calculations and summary.

       In the data input area the user enters the conditions defining the specific
case to be evaluated.  After the first run, only those data which change from case
to case need to be entered each run.  The input data fall  into three categories:
general description of the case, NOx control selection, and unit loading schedules.
The general description data include case number and narrative description of the
case conditions.  The NOx selection input consists of completing a matrix table of
NOx control technologies  for each unit in the system.  The final  data input consists
of loading schedules for  each unit for both short term (1  hour to many days) and
annual periods.  The short-term period is intended to provide the total and average
NOx emissions from each unit over a specified duration (e.g.  8 hours, 1 day, 1 week,
etc).  The annual period  is used to calculated the NOx emissions, generation, fuel
consumption and variable  control costs over a year's time.   For each time period the
user inputs the hours of  operation of each unit, at each of five (5) load levels and
for each fuel used.  The  specification of hours of operation at each load level is
important since NOx emissions are variable (usually non-linear) with load, and
therefore the load history must be known in order to calculate integrated NOx
emi ssions.

      Also located in the data input area, but usually not changed by the user, are
tables of NOx reduction effectiveness and generic costs (capital  and O&M) for each
control technology.  Capital costs are specified in $/KW and variable O&M costs in
terms of $ per unit of generation or of tons of NOx removed.

      The calculation area of the model  begins with tables of baseline NOx
emissions, (Ib/mmBtu) vs  load for each unit and each fuel  fired.   Similar tables of
NOx emissions vs load are provided for O.S. firing conditions.

      Controlled NOx emissions (in Ib/mmBtu) are calculated sequentially for each
technology specified in the data input area.  Thus, each technology effectiveness
(and resulting NOx output) is dependent upon the output NOx level of the preceding
technology.  For example, if both LNB's and FGR are selected for a unit, then the
FGR effectiveness at each load level of the unit will depend upon the LNB output NOx
level at the corresponding load.  Of course, each technology not selected has no
effect on the NOx level.


                                       8-70

-------
      Following the last application of NOx technology to each unit, the final
outlet NOx level is determined at each load level for each fuel.  Based upon the
hours of operation at each load level for each fuel specified in the input tables,
the total short-term and annual NOx emissions (Ib/NOx) are integrated for each unit,
along with the total generation (kwh) and thermal input (Btu).

      The cost of NOx control  is calculated for each unit by summing each cost
element (capital, fixed O&M, variable O&M) for each technology used.  The capital
cost for each selected technology is the generic cost ($/kw) times the unit rating
(kw) times a unit-specific multiplier which represents the degree of difficulty of
applying each technology to that unit.  Similarly, the variable O&M cost of each
unit is calculated as the sum  of each applied technology's variable O&M cost, which
is the product of the generic  cost ($/kw or $/ton NOx) times the annual usage (kwh
or tons NOx) times a unit-specific cost multiplier.  Fixed annual O&M costs are the
specified generic fixed O&M costs ($/yr) times a cost multiplier for each unit.
Finally, capital costs are level ized by multiplying the total capital cost for each
unit by a  recovery factor representing a specified time period (e.g. 20 years) and
a rate of return (e.g. 10%).   Similarly, the total annual  O&M costs are levelized
according to standard procedures to account for rising O&M costs over the economic
life of the project, essentially in accordance with the EPRI TAG procedures.  The
capital and O&M levelizing factors are entered by the user.

      The final function of the spreadsheet model is to compile the emission and
cost results for each unit into a total for the system (including appropriate system
averages, such as Ib/mmBTU NOx emission) and to present the results in a concise
tabular format.

      By calculating the-unit  specific emissions and costs (and therefore the system
emissions and costs) for a successive series of varied NOx control  applications,  the
user can determine the lowest-total-cost combination of controls which will  result
in total system emissions meeting any specified level  for any specified time-
averaging period.

ANALYSIS RESULTS

      The Con Edison System NOx model has been constructed and is fully operational,
but preparation of input information has only been partially completed.  Selected
analyses have been performed,  however, by utilizing that information which has been
developed and by otherwise employing prior information in ETEC's possession and best
estimates.   The results of these analyses are reported herein and although they are
subject to some level of uncertainty in terms of magnitude,  derived trends and
observations based on these trends are believed to be generally valid.

      Figures 6 and 7 show calculated system NOx emissions for 24 hour periods
coinciding with peak generating days in July and December for baseline operation and
for various NOx control  strategies.   Each plotted data point corresponds to a
specific control strategy consisting of the application of various combinations of
NOx reduction technologies to each unit in the system.  Solid symbols denote that
the indicated control combination has been uniformly applied to all units in the
system while open symbols indicate selective utilization.   In this latter case, the
letters "Fg" indicate WFGR applications on only gas/oil  fired boilers (excluding oil
only units)  and a numeral denotes the limited number of unit applications of the
technology identified by the end letter in the sequence (ie OU(5) denotes O.S. on
all  units and UREA on 5 units).
                                        8-71

-------
      The results apply to actual unit load duration curves for 1990 but the fuel
mix has been altered to reflect maximum gas burning in July and maximum oil burning
in January (ie. dual fuel units burn either all gas or all oil depending on the
month).  This allocation of fuels burned approximates that shown in Fig. 1 which  is
based on a PROMOD projection.  The indicated NOx emissions for each strategy have
been determined by summing the respective integrations over each unit's load
duration curve of the emission rate applying to the fuel  burned and the combination
of control technologies installed on the unit.

      The baseline (uncontrolled) NOx emissions indicated in the figures have been
determined on the basis of the steady state test data acquired to date and estimates
for as yet untested units.  The levels shown understate actual NOx emissions since
they do not reflect the effects of AGC operation,  dual fuel  firing and boiler
cleanliness in switching between fuels.  Each of these factors would tend to
increase unit baseline, and hence system, NOx emissions.   The reduced emission
levels shown to be achievable by the application of the various strategies are also
overstated in this regard since they are based on  the baseline emissions.   Aside
from this factor, the achievable reductions have been determined employing
potentially overly optimistic estimates of the NOx control capabilities of the
individual control options, as pointed out previously.  As a consequence of the
above factors, the results as shown are probably too low and the rate of decline in
achievable emissions with increasing control cost  is too steep.

      The analysis results shown in Figure 6 and 7 are primarily of interest to Con
Edison.  It is possible, however, to draw certain  observations based on the
indicated trends that may be of more general interest to other gas/oil  utilities and
these are discussed below.

OPTIMUM NOx CONTROL STRATEGIES

      The purpose of the analysis was to determine the minimum control  cost to
achieve varying levels of NOx emission reduction.   This cost would be represented by
a curve defining the locus of minimum control  cost strategies for achieving
successively reduced levels of NOx emission.  Defining such  a curve by employing the
model is an iterative procedure in which various strategies  are analyzed and the
calculated NOx emission levels and costs are compared.  This procedure was followed
in the present case and the optimum strategies determined are those shown  in Figures
6 and 7 as being the lowest points at any cost level.

      The strategies that were analyzed only broadly define  the optimum curve since
intermediate steps have not yet been evaluated.  For instance, the locus of
strategies between O.S. on all units and O.S.  plus UREA on all units would be
defined by the intermediate steps of sequentially  adding  UREA combined with O.S. to
successive units.  Two such intermediate steps are shown  in  Figures  6 and 7 for
OU(3) and OU(5).

      The analysis results indicate that the optimum strategy to achieve a specific
level of NOx emissions would consist of maximizing the system wide utilization of
the lowest cost technologies first before employing on any unit the next most costly
technology.  For instance, it would always be more cost effective to employ UREA on
additional units compared to employing the next most costly  technology, which in
this case would be WFGR, on any additional unit.  This analysis result is  summarized
below:
                                       8-72

-------
      Strategy for Increasing                   Order of Control Option Application
      Levels of NOx Control                     All units         Successive units

            I                                                           O.S.
            II                                  O.S.              +     UREA
            III                                 O.S. + UREA       +     WFGR
            IV                                  O.S.+ UREA + WFGR +     SCR

      LNB could be employed as a substitute technology for O.S., providing an added
10% increment in NOx reduction.  However, the combination of O.S. plus UREA would
always be more cost effective than the utilization of LNB's.  WFGR would be employed
in an optimum strategy only on gas/oil fired boilers since it's control capability
for reduced initial NOx levels is too low for cost effective utilization on oil-only
boilers.

      The above ranking order for utilization of control technologies would apply
only to situations in which an emission regulation were expressed as a LB/day
emission limit averaged over a system.  Alternative forms of emission limits would
likely result in a different ordering of technologies for optimum employment.

DIMINISHING RETURN

      Figures 6 and 7 graphically illustrate the diminishing return of increasing
expenditure to reduce NOx emission from the Con Edison system.  This observation is
quantified in the table below which applies to the optimum locus of strategies in
Figure 6.

            System NOx Emission                       Cost of control
               Reduction, %                              Mill/KWH

                  50                                         .4
                  70                                        1.4
                  75                                        1.8
                  80                                        4.5


      The table values show for instance that an 80% emission reduction would
require a factor of three greater expenditure than a 70% reduction.  This trend is
actually understated since the achievable emission reductions shown in the figures
are optimistic as explained previously.

SEASONAL INFLUENCE ON COST OF CONTROL

      Figure 8 replots the optimum strategies defined in Figures 6 and 7 in terms of
daily emissions averaged on a LB/MMBTU basis.  Peak day NOx emissions are shown to
be higher in January than in July.  The reason for this is attributable to higher
baseline NOx emissions in January due to substantially increased oil firing, to
differences in unit loading schedules and to generally reduced NOx control
effectiveness for some of the technologies for oil firing.

      The difference in emission rates for the two seasons is particularly
significant if a regulation were passed of a form limiting emissions on a LB/MMBTU
basis.   The inset table in the figure shows that the cost of compliance in this
instance would be at least a factor of two greater for January in comparison to
July.   The purpose of such a regulation, however,  would be to reduce ambient ozone
concentrations,  which tend to be most pronounced during the summer months because of
                                        8-73

-------
meteorological conditions favoring their formation.   Therefore,  a regulation of this
form would result in an additional expenditure that  would serve  no environmental
purpose.  In such a situation, the emission limit should be formulated to cost
effectively achieve it's intended purpose.

CONCLUSIONS

1)  A system NOx emissions model of the type described can be a  useful tool in
    assessing the implications of a potential  regulation in advance of it's
    promulgation for preparing a utility for the regulatory process.

2)  The Con Edison boilers have low uncontrolled baseline NOx emissions because of
    their design and low capacity factors.   In such  instances,  it is  more difficult
    to reduce NOx emissions because of the  reduced effectiveness of NOx control
    technologies for low initial NOx levels.

3)  The process of establishing NOx emission regulations should  recognize that
    relatively small differences in control limits can have a dramatic effect on the
    required cost of control.

4)  The form of an emission regulation can  inadvertently result  in the expenditure
    of unnecessary control costs if it does not specifically address  it's intended
    purpose.


REFERENCES

      1)  Bagwell, F.A., et.al., "Utility Boiler Operating Modes for  Reduced Nitric
          Oxide Emissions", JAPCA, November, 1971

      2)  Bayard de Volo, N., et.al.,  "NOx  Reduction and Operational  Performance of
          Two Full-Scale Utility Gas/Oil  Burner Retrofit Installations",  1991 Joint
          Symposium of Stationary Combustion NOx Control,  Washington,  D.C.,
          March 25-28, 1991
                                       8-74

-------
          TABLE I
CON EDISON GENERATING UNITS
Plant
Function

POWER









POWER PLUS
STEAM
SENDOUT




















STEAM
SENDOUT




Plant


ARTHUR KILL

ASTORIA




RAVENSWOOD


EAST RIVER

(Pwr Only)
59TH St.





WATERSIDE







74TH ST.


HUDSON AVE.



RAVENSWOOD
E.RIVER SO.
59TH ST.
74TH ST.
60TH ST.
Unit


20
30
10
20
30
40
50
10
20
30
50
60
70
110
111
112
113
114
115
41
42
51
52
61
62
80
90
120
121
122
71,72
81,82
100

4 units
10 units
3 units
6 units
6 units
Capacity
MW

345
440
187
173
365
375
375
95
395
900
148
148
180
72
43
43
43
79
79
71
71
71
71
97
97
160
160
64
64
64


187
MLB/HR
275 EA
150 EA
150 EA
150 EA
150 EA
Mfg


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

B&W
FW
FW
FW
FW
Firing
Config.

Face
Corner
Face
Face
Face
Corner
Corner
Corner
Corner
Corner
Opposing
Opposing
Face
Face
Face
Face
Face
Corner
Corner
Corner
Corner
Corner
Corner
Corner
Corner
Corner
Corner
Corner
Corner
Corner
Face
Face
Face

Face
Package
Package
Package
Package
No of Burn.


32
40
22
22
32
32
32
32
32
64
12
12
18
8
5
5
5
8
8
8
8
8
8
8
8
16
16
8
8
8
8
8
16

6
2
2
2
2
Fuel
O-Oil
G-Gas
O
O
G.O
G,0
G,O
G,O
G,O
G,O
G,0
0
G,0
G,O
G,O
0
0
O
O
0
0
G,0
G,0
G,O
G,0
G,0
G,0
G,0
G,O
0
O
0
0
0
0

0
0
O
0
G
        8-75

-------
                 TABLE II

 CON EDISON BOILER CURRENT NOx EMISSIONS
AND COMPARISON WITH OTHER UTILITY BOILERS


FIRING
CONFIGURATION


Single Face Fired







Opposed Fired





T Fired





SIZE
MW


175
175
180
187
215
230
345
365
148
225
230
350
480
750
320
395
440
900/2
FULL LOAD, UNCONTROLLED NOx EMISSIONS,
PPM
GAS

OTHER
UTILITY
405
750


520
337



550
360
890
700
1200
335



CON ED



300
175



225
275






150


OIL

OTHER
UTILITY

450


250
370



...
250
425
320
750
225



CON ED



250
300


250
325
250






175
200
275
                   8-76

-------
               2,500
00
           LU

           CO
               1,500
g   2,000

O


O


I
LJJ
Z
LJJ
O
=!   1,000
CO
CO
O
500
                   0
                                       OIL
                                                                    Maximum Fossil Fuel

                                                               Generating Capability for one month


                                                                       4880 GWH
                                                                           50 million
                                                                                           40
                                                                           30
                                                                                           20
                                                                           10
                     JAN   FEB   MAR   APR   MAY   JUN   JUL   AUG   SEP   OCT   NOV   DEC

                                                    MONTH


                             FIG. 1  Projected Con Edison Fossil Fuel Generation

                                      and Fuel Consumption for the early 1990's
CO
E
E

z"
O
CO
z
O
O
_i
LU
=)

-------
                   Normal
                  Operation
             High
            Excess
             Air
           Furnace
          Operation
Fuel-Rich Burner Operation
                                                 1.4 Burner Equivalence Ratio

                                                 71  Burner % Air

                                                     Furnace % Excess 0 
FIG. 2   Off-stoichiometric Combustion for Natural Gas Firing
                                8-78

-------
a>
-^i
C>
            OJ
            O
E
Q.
Q.
V)
z
O
V)
uy

LU
x
O
                300 -
                20
                100
                  0
0
                  AGC OPERATION


                /\ UNCONTROLLED


                O OS




                STEADY STATE TEST DATA
                                UNCONTROLLED

                                O.S.
                                   50
100

MW
                                                        150
200
                        FIG. 3 NOx EMISSIONS BAND DURING AGC OPERATION
                                 ON GAS FUEL FOR ASTORIA UNIT # 10

-------
40
80
                                                    240
280
                            120      160       200
                            Initial NOx, ppm
Figure 4.  NOx Control Effectiveness of UREA Injection versus Initial NOx Level
                                           240
                                            280
                          120       160      200
                          Initial NOx, ppm
FIG. 5   NOx Control Effectiveness of 20% WFGR versus Initial NOx Level for
        Gas and Oil Fuels
                    8-80

-------
C\l\i
CD
_l A
O
0
g 150
CO"
0
CO
CO
5
gj 100
?~
^c
Q
oo *:
<
LJJ
n
c *
x

4
 TECHNOLOGIES APPLIED TO SELECTED UNFTS
LFU TO ALL APPLICABLE UNFTS. SCR TO (5) UNITS
B BASELINE, NO CONTROL
O OFF-STOICHIOMETRIC FIRING
L LOW-NOx BURNERS
F FLUE GAS RECIRCULATION
n GAS FIRED UNITS ONLY
9
U UREA INJECTION

S SELECTIVE CATALYTIC REDUCTION














AOU . LF
^ Lu A OUS(2)
^
OFgU A LUS(2)
~ A ALFU A
U=gU A

LFUS(5)
A

. LFUS
A
II 1 1 i I I

01 23 45 6789 10 11 1-
          LEVELIZED COST OF CONTROL,  Mill/kwh

FIG. 6 Optimum System NOx Control Strategy to Achieve Varying Levels
       of Emission Reduction for Peak Generating Day in July

-------

-------
             0.35
00
do
CO
        CD
         E
        .E
        m
        _j
        CO
        CO
        CO
        LU
        LU
        Q_
        DC
        CM
             0.30
0.25
POSSIBLE
EMISSION
LIMIT
LB/MMBTU
A 0.13
B 0.08
% ADDmONAL COST OF CONTROL
FOR JANUARY COMPLIANCE
IN COMPARISON TO JULY

100
120
                                                   JULY, 1990


                                                   JANUARY, 1990
                                                                       10
        2468

          LEVELIZED COST OF CONTROL,  Mill/kwh

FIG. 8  Added Cost of NOx Control to Comply With LB/MMBTU

       Emission Limit in January in comparison to July
                                                                      12

-------
REDUCED NOx, PARTICULATE, AND OPACITY ON THE
      KAHE UNIT 6 LOW-NOx BURNER SYSTEM

                 Stephen E. Kerho
                 Dan V. Giovanni
      ELECTRIC POWER TECHNOLOGIES, INC.
               Menlo Park, California

                    J. L. B. Yee
       HAWAIIAN ELECTRIC COMPANY, INC.
                Honolulu, Hawaii

                  David Eskinazi
      ELECTRIC POWER RESEARCH INSTITUTE
                Palo Alto, California

-------
             REDUCED NOx, PARTICULATE, AND OPACITY ON THE
                   KAHE UNIT 6 LOW-NOx BURNER SYSTEM

                               Stephen E. Kerho
                               Dan V. Giovanni
                    ELECTRIC POWER TECHNOLOGIES, INC
                             Menlo Park, California

                                  J. L. B. Yee
                    HAWAIIAN ELECTRIC COMPANY, INC
                              Honolulu, Hawaii

                                David Eskinazi
                   ELECTRIC POWER RESEARCH INSTITUTE
                              Palo Alto, California
 ABSTRACT

 Hawaiian Electric Company (HECO) completed major combustion system
 modifications in mid-1988 on Kahe Unit 6, a Babcock & Wilcox (B&W) oil-fired unit
 rated at 146 MW. The modifications were undertaken to reduce emissions of NOx and
 particulate matter, and to restore operational flexibility that had been restricted with
 burner-out-of-service operation  previously used for NOx control. Modifications
 included installation of the B&W PG-DRB burners, front and rear wall overfire air
 (OFA) ports, extensive ductwork for the OFA and flue gas retirculation (FGR) flows,
 and upgrading of the automatic burner control system. This installation represented
 the first application of this type  of low-NOx firing system to a utility boiler in the
 United States.

 As reported in 1989, the NOx reduction goal of emissions below 0.23 Ib/MBtu was
 achieved and particulate emissions  were controlled to below 0.1 Ib/MBtu.  However
 opacity levels increased from pre-retrofit levels of approximately 6% to between 15-
 20%. In an attempt to reduce opacity levels and still comply with NOx emission
limits, HECO and the Electric Power Research  Institute jointly sponsored a follow-on
Phase 2 performance improvement  program conducted by  Electric Power
Technologies, Inc to evaluate the potential of new atomizer designs to reduce NOx,
particulate, and opacity. The program demonstrated significantly reduced opacity and
particulate levels while maintaining NOx emissions below 0.23 Ib/MBtu even though
the levels of OFA and FGR were reduced.
                                     8-87

-------
INTRODUCTION

In July 1987, the Hawaiian Electric Company (HECO) contracted with the Babcock &
Wilcox Company (B&W) to retrofit a low NOx combustion system on their 146 MW
(grossT oil-fired Kahe Unit 6. The unit is front wall-fired and burns oil with a
maximum sulfur content of 0.5%. Up to this time, the unit had been operating with
flue gas recirculation (FGR) to  the combustion air and burners-out-of-service (BOOS)
in an attempt to satisfy the operating permit requirement for maximum NOx
emissions of 0.23 Ib/MBtu (180 ppm, dry, 3% O2). Typical emissions using these
controls were 0.28 Ib/MBtu NOx (219 ppm) and 0.06 - 0.08 Ib/MBtu particulate matter
(PM). Normal opacity levels were in the 4-6% range, which is below the visible
threshold.

The principal objective of the retrofit was to reduce NOx emissions to below the
regulatory requirement while minimizing particulate matter (PM) emissions.
Additionally it was intended that the retrofit technology would allow a return to all-
burners-in-service operation, thereby improving the operating flexibility of the unit
which had been impaired with BOOS operation. Specifically, a higher turndown was
expected from improved flame  stability at low loads (the lowest load for dispatch was
95 MW with BOOS), and  a higher reliability in achieving full load was expected with
the ability to accommodate burner maintenance outages without load reduction.  The
project was the first installation in the United States of the integrated application of
low-NOx burners, FGR to both the combustion air and directly to the burners, and a
state-of-the-art front and rear wall overfire air (OFA) design to a heavy oil-fired utility
boiler. The combustion system, designated "PG-DRB",  is licensed by B&W from
Babcock-Hitachi (BHK) who commercialized  the technology in Japan.

The retrofit was successful in meeting the NOx requirement of the operating permit
and in providing the desired improved operating flexibility.  However, operating
problems such as undesirable opacity levels led to a follow-on Phase 2 program of
combustion optimization work and equipment modifidation.  This paper presents the
results of the follow-on program which was conducted in 1990.

OVERVIEW OF 1987 NOx SYSTEM RETROFIT

Kahe 6 is a radiant reheat type steam-electric unit manufactured by B&W.  An
elevation view is presented in Figure 1. For NOx control, the boiler was originally
equipped with nine B&W dual  register burners arranged in a 3 X 3 array on one wall,
and flue gas recirculation to the windbox which permitted up to 20% of the flue gas to
                                      8-88

-------
be mixed with combustion air prior to the burners. The retrofit PG-DRB system
consisted of the following elements:

         1.  PG-DRB burners
         2.  Dual fluid (steam/oil) atomizers
         3.  Utilization of existing FGR to the windbox combustion air
         4.  Primary gas (PG) system which supplies FGR directly to the burners
            unmixed with the combustion air
         5.  Overfire air system
         6.  Upgraded control system

The PG-DRB burner, shown in Figure 2, consists of an oil atomizer/impeller located
axially in the primary (core) air zone of the burner. The core air is introduced into the
center zone through slots located at the back of the burner. Core air flow is limited to a
maximum of approximately 10% of the total air flow. The flow to this region can be
controlled with a small sliding disk.  The core zone is surrounded by the PG zone,
which is enclosed by the inner and outer air zones. Pure gas recirculation is fed
through a perforated plate located at the entrance to the PG zone annulus which helps
to distribute the flow around the periphery of the zone.  A butterfly-type valve
provides controllability of the PG flow to individual burners.  Air to the inner and
outer air zones is controlled by a single sliding disk. An impact-suction pilot tube grid
is installed prior to the inner and outer air zones to allow measurement of the airflow
in these zones.  The pilot grid consists of a manifold which encompasses Ihe burner
with six finger-like extensions into Ihe total air flow zones.  These measurements,
togelher wilh air slide position, provide Ihe capability of controlling  air flow to  the
individual burners.  The inner air zone contains gear driven spin vanes, while  the
outer zone has fixed spin vanes followed by gear driven spin vanes. The manually
operated gear driven vanes provide the ability lo vary swirl characteristics  and  Ihus
Ihe resulting flame shape of the burner.

The OFA system was designed to divert up to 30% of the tolal combustion air to six
OFA ports located on the front and rear boiler walls (three ports per wall),
approximately 10 feet above the top burner elevation. Each OFA port is equipped with
damper assemblies and air spin vanes to allow independent control of air quantity,
velocity, and furnace penetration.  A schematic showing the port design is  provided in
Figure 3. Like the burners, the OFA ports were equipped with flow monitors, allowing
on-line measurement of separate flows through the spin annulus and central core of
each overfire air port.  Flow modeling tests using a scale model of the windbox  and
furnace were used by B&W to obtain air flow distribution information for the windbox
                                     8-89

-------
 and OFA system.  The model results were used to establish placement and sizing of the
 OFA ports for optimum mixing. The modeling results were the basis for the decision
 to use six ports (instead of three) and the recommendation for a nominal 70:30 rear-to-
 front wall distribution of overfire air.

 Summary of Retrofit Low NOx System Performance Evaluation

 The results of the program were presented in detail at the 1989 Symposium (Reference
 1) and are summarized below. The retrofit realized its principal goal to reduce NOx
 emissions to below 0.23 Ib/MBtu with all burners in service. At 145 MWg, NOx and
 PM emissions levels of 0.21 and 0.07 Ib/MBtu respectively were achieved with a stack
 opacity of 15%. The fuel nitrogen content was approximately 0.3% (wt). The  test was
 performed using 10%  FGR (defined as the amount of recirculated flue gas divided by
 the sum of the total air and fuel flows) to the windbox and 27% of the total air to the
 OFA system (split 70% to the rear ports  and 30% to the front ports). These acceptance
 test results typified the best overall emissions performance achieved  and required an
 extensive test effort during the commissioning of the equipment to control PM
 emissions and opacity. Although the opacity levels noted above are  within the
 regulatory  requirement of <20% for a 6  minute average, they are considered
 undesirable because a visible plume results.  These results represented an over 75%
 reduction in NOx from pre-retrofit levels with all burners in service  (ABIS) and
 without FGR.

 During commissioning, a strong inverse relationship between NOx and PM/opacity
 was encountered.  Initially, when the combustion equipment was tuned to achieve
 NOx levels below 0.23 Ib/MBtu, the corresponding PM emissions were typically 0.13-
 0.15 Ib/MBtu and opacity exceeded 20%. The magnitude of this trade-off was
 unexpected from previous experience reported by BHK in Japan, where over 10,000
MW of PG-DRB is operational. It appears that this trade-off is  a fundamental  feature
of the PG-DRB system when fired with  heavy oils. Further assessment of the  Japanese
experience  in the light of these results led to the conclusion that a similar trade-off
exists at Japanese installations, however it is not an issue there because the boilers are
equipped with electrostatic precipitators for participate and opacity control.

Initial Oil Atomizer Selection

In order to reduce PM emissions, a comprehensive program was implemented by
B&W during commissioning to optimize oil atomization with the PG-DRB burner
system. Improved atomization would result in smaller oil droplets which burn out
                                      8-90

-------
more completely, resulting in reduced PM emissions.  During the course of the
program, a number of B&W dual fluid (steam/oil) atomizer designs and atomizer
spray cone angles were evaluated. These included the Y-Jet, Racer, modified Racer
(Racer with increased steam rates), T-Jet, and a developmental I-Jet design. These
atomizer types are characterized by their geometry, steam-to-oil mass flow  rates, and
the size of the oil droplets produced.  The Racer, Y-Jet, and T-Jet designs were flow
characterized using water and air as the working fluids. Drop size distribution
information was obtained using an Aerometrics Phase Doppler Particle Analyzer.  The
conversion of water/air data to oil/steam was done using viscosity, surface tension,
and mass ratio corrections which were obtained from the literature. For oil properties
and operating conditions at Kahe, the nominal Sauter Mean Diameter (SMD) of the oil
droplet size distributions were 400, 320, and 235 microns for the Racer, Y-Jet, and T-Jet
respectively. The T-Jet was judged to provide the best performance and was selected by
B&W for continuous operation.  The importance of reducing  drop size was
demonstrated by the reductions in PM and opacity achieved from the initial levels:  PM
emissions were reduced from 0.13 - 0.15 Ib/MBtu to 0.07 Ib/MBtu and opacity levels
from over 20% to 15%.

LONG-TERM OPERATING EXPERIENCE

Operation at Kahe 6 after approximately two years was characterized by a number of
combustion related problems. Although NOx level