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 1900°F 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 complementary—the SCS program will evaluate a large number of different catalysts at relatively fixed fuel composition and furnace design; in contrast the EPRI program will evaluate a limited number of similar catalysts over a wide range of fuel composition and furnace designs. PROGRAM STATUS The 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 710°F, and passes through an isolation damper, a venturi to monitor flow rate, and a 40 kW heater to adjust process temperature to desired values (680-700°F). Flue gas then enters an approximately 20 ft straight section in which ammonia reagent is injected and mixed. The flow is then equally split into two reactors, each containing catalyst from a different supplier. At the exit of each reactor are flow rate monitors and manual dampers which insure flow rates are equal in each section. An induced draft fan followed by a control damper is the last component prior to flue gas return. Post-FGD: NYSEG The post-side pilot plant is located at NYSEG's Somerset Station, approximately 40 miles northeast of Buffalo, New York. Figure 2 presents a simplified schematic of the pilot plant, which employs a recuperative heat exchanger and electric auxiliary heater to increase flue gas temperature to 625°F for acceptable NOX removal. The NYSEG/post-FGD process conditions are selected to provide 80% NOX removal (from boiler concentrations of 400 ppm) and control of residual 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 550°F by a recuperative (heat pipe) heat exchanger. Two electric heaters provide a total of 100 kW heating input to further increase flue gas temperature to 625°F. The gas then enters the reactor tower, which is identical to the 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 225°F. TEST PLAN A test strategy has been developed based on a two year operating period. The test plan will first establish baseline performance, then implement load-following operation. Documented changes in catalyst activity over two years will allow estimating the useful catalyst life. Additionally, a series of measurements will determine if SCR contributes to or reduces the concentration of trace species and particulates. For approximately 85% of the operating time, the pilot plant will operate in a simple load-following mode, and allow for monitoring NOx removal, residual NH3, and byproduct SC»3. 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 SC»3 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 400°F, based on ammonia and SO3 concentrations of approximately 10 ppm. However, the thermodynamics of these reactions for high sulfur coal conditions (up to 30 ppm 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 625°F. These unique conditions, not previously reflected in full-scale or pilot tests, could be responsible for persistent deposition at these relatively high temperatures. As of mid-February this problem at the pilot scale had been remedied with a special-purpose injection system. In this approach, two injectors are alternately used, allowing ammonium compound deposits on the injector not in 5A-7 ------- service to decompose to ammonia and 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 (250°F) Recuperative Heat Exchanger Gas out 550° F Electric Heater T in : 625°F 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 440°C. The space velocity (SV) ranged from 7,650 to 36,500 hr"1. The combustion gas was doped with nitric oxide (NO) and NH3, and the NH3/NO ratio ranged from about 0.6 to 2.2. Sulfur dioxide (S02) was added to the combustion gas in some runs to investigate its effect on NO conversion. The results obtained indicate that the SV has a significant effect on the conversion of NO for the in-house catalyst which was prepared primarily for start-up of this system before the commercial catalysts arrived. For the two commercial catalysts, the NO conversion was 90% and higher when the NH3/NO ratio was near or above unity. For the same catalysts, the NO conversion was approximately proportional to the NH3 concentration at the inlet of the reactor, when the NH3/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 440°C. The effects of temperature, space velocity, and NH3/NO ratio on the conversion of NO according to Equation 1 were examined. The amount of N02 detected in the 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 150°C to prevent moisture from condensing on the catalysts overnight. The natural gas burner was then fired up the next morning at a proper fuel/air ratio, and the preheater was turned off. Once the reactor temperature rose steadily, the fuel/air ratio was 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 5°C 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 475°C for 4 hours. A V205-containing solution (diluted H2S04) was then dip-coated on the calcined substrate, followed by calcination at 450°C for 4 hours. The catalyst blocks obtained were brownish yellow, but not uniform due to the dip- coating procedures used. Only very limited information on the commercial catalysts was released by the suppliers. Catalysts 2A and 2B were extruded V205/Ti02 based materials. Catalyst 2A, with a catalyst-flue gas contact area of 910 m2/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 406°C. 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 440°C, and in the presence of 95 ppm S02 . The results indicate that the catalyst is less active at 440°C than at 353°C, when 95 ppm of S02 is present in the flue gas. N20 Measurements The results of N20 measurements over Catalyst 2A are listed in Table 2. The NH3/NO ratio was varied from 0.586 to 2.17. The reaction temperature was 400°C. 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 340°C 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 360°C is shown in Figure 7. The results indicate that Catalyst 1A is the least active. Although the exact reasons have not been investigated, it is possible that the difference in reactivity between the in-house and the two commercial catalysts is due to the difference in chemical composition which is reflected by the difference in color of the catalysts tested. It is also likely that the catalytic activity can be affected by the conditions under which the catalysts were made. CONCLUSIONS The U.S. Environmental Protection Agency has built a pilot plant to investigate the ammonia (NH3) based technology of selective catalytic reduction (SCR) of nitrogen oxides. One in-house catalyst and two commercially available catalysts were tested. The effects of temperature, space velocity (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 350°C. The NO conversion increased from an average value of 17 to 67% as the SV was decreased from 18,400 to 7,650 hr"1. For the two commercial catalysts, the NO conversion was 90% and higher when the NH3/NO ratio was near or above unity. For these two catalysts, the NO conversion was approximately proportional to the NH3 concentration at the inlet of the reactor when the NH3/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 406°C. For the same catalyst, flue gas S02 was found to be poisonous, and the poisonous effect of S02 was more severe at 440°C than at 353°C. At 400°C, NH3/NO ratios ranging from .0.586 to 2.17, and SV of 13,790 hr"1, the amount of N20 formed over the same catalysts was negligible. The difference of activity between the in-house and the commercial catalysts is attributed to the difference in chemical composition and how the catalysts were made. 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 process—Experiences 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 60°C for 24 hours and then at 120°C for 72 hours. After drying, the bulk titanium dioxide was crushed and sieved. The fraction between 20-32 mesh was collected and calcined at 600°C in air during the first hour, and then in He during the following six hours. The BET surface area of this support was 30.6 m2/g, which was measured by a Quantasorb surface area analyzer. The composition of the W03-V20s/Ti02 catalyst was the same as that described for a commercial SCR catalyst (3). The catalyst was prepared by co-impregnation of an aqueous solution of NHjVOs and (NH4)6 H2W12040 in oxalic acid. After impregnation, the catalyst was dried at 120°C for 15 hours and then calcined at 500°C in oxygen flow for 20 hours to decompose the ammonium salts into oxides. The elements identified as potential poisons for the SCR reaction in the earlier work are alkali (Li, Na, K, Rb, Cs, Ca), as well as arsenic (As), phosphorous (P), lead (Pb), and HC1. Accordingly, these elements were evaluated in this study by impregnation via incipient wetness with the precursor solutions of corresponding salts on the 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 120°C for 3-4 hours followed by calcination to decompose the precursor salts. The experimental setup and procedure were the same as reported earlier (1, 4). Briefly, the reactor was a quartz tubular reactor in which 1-2 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 f°r 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 250°C (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 300°C than at 350°C. 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 340°C blocked the active surface area, which was the reason that the deactivation was more pronounced at 300°C than 350°C. Effects of Sulfur Dioxide The results of SCR with 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, 300°C, 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, 300°C, = 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 200°C. 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 300°C. 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 540°F. The tests did provide the following results: 5A-55 ------- • With air heater gas inlet temperatures above 570°F, 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 1050°F. The boiler is equipped with primary and secondary superheater sections, a reheat superheater, economizer, and two air heaters. The boiler operates in a forced draft mode and is equipped with 24 combination gas- or oil- fired burners located on the front wall with four rows of six each. Selected burners in the upper rows are operated out of service for 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 740°F 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 740°F to 650°F is expected to reduce NOX removal from about 50 to 43 percent. Increasing the inlet temperature to 830°F would increase NOX removal to about 55 percent. The minimum temperature to obtain NOX removals is approximately 540°F. KAH recommends 900°F as a maximum gas inlet temperature. An extensive evaluation program to characterize the performance of the CAT-AH is planned following the initial start-up of the system. Key process parameters to be evaluated will include: • Air heater gas inlet temperature; • Space velocity; • Inlet 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 930°C. The peak NOX removal with urea was 82 percent at a temperature of 980°C. The calculations discussed previously yielded peak NO removals for NH3 and urea at nominally 927°C (1200K) and 1027°C (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 1100°C, 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 877°C, while injecting either urea or cyanuric acid resulted in peak emissions at 977°C. 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 1027°C (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 980°C. Similarly, urea injection resulted in a maximum NOX reduction of 57 percent at 930°C. 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 1100°C. 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 980°C. N2O emissions resulting from urea injection also peaked at 980°C, at 43 ppm. When injecting the NH3, peak N2O emissions of 21 ppm were measured at 880°C. Figure 10 shows the ratio of N20 emission to NOX reduction as a function of temperature for each of the three SNCR chemicals tested at this lower initial NO level. Test results showed maximum values at temperatures similar to those seen at higher initial NO levels (see Figure 7). The ratio peaked at about 880°C for NH3 and 980°C for urea and cyanuric acid. Again, cyanuric acid was shown to provide the highest conversion of 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 927°C, 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 827°C 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 1020°C. At higher temperatures, the ammonia reagent itself oxidises to form additional NOX and reduction efficiency decreases. At lower temperatures, the ammonia is not oxidised sufficiently rapidly to the amine radical, which is the species which actually interacts with NO (Dean et al. 1982), and unreacted ammonia passes through the SNCR reaction zone. Effect of Ammonia/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 (1074°C) and oxygen content. It is readily apparent that, at ratios not below 1.3, detectable ammonia slip occurs despite the fact that the temperature used for these experiments is considerably higher than the optimum observed when a 1:1 NH3:NOX ratio was in use. Effect of Initial NOX Concentration A series of runs was carried out (at 1093°C, 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 1000°C, the reduction is effectively independent of oxygen concentration but that at 908°C, where the effectiveness is lower, NOX reduction increases monotonically with O2 concentration. However, over the range of O2 contents which are likely to be encountered on industrial pulverised fuel 'p.f.' boilers (3-4%) the 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 800°C), NOX reduction increases monotonically with CH4:NH3 ratio. At higher temperatures (865,915°C) , the curves begin to exhibit maxima beyond which conversion falls as methane increases. At 965°C and above conversion falls as methane increases. The effect of methane addition on the temperature window is summarised in fig. 8. Introduction of 0.5 mol methane per mol of ammonia depresses the best reduction of the process from 68% to 60%, while the optimum temperature decreases from 1030°C to 916°C. When a 1:1 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 915°C. The 'window' of applicability of SNCR was generally wider when methane was in use. For instance, reductions of >40% were possible over a range of around 5A-105 ------- 150°C in the absence of methane but around 200°C when a 1:1 ratio of methane to ammonia was used. Effect of Methanol Addition on SNCR Studies were also carried out on the efficacy of methanol in modifying the behaviour of the process. Results are presented in fig.9. These show that, at low CH3OH/NOX ratios, methanol does yield some enhancement of the De-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 908°C 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 (850°C) 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 870°C. 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 1020°C. However, the maximum attainable reduction is considerably reduced (-40% compared to over 60% in the CADR) and the temperature window is considerably broadened. There is again an obvious movement of the temperature window to lower temperature when methane is injected with the ammonia and NOX reductions of 35-40% are attainable at a temperature of 800°C. EFFECT OF SCALE (MIXING) ON ATTAINABLE REDUCTIONS Fig. 13 presents a comparison of the NOx reduction vs. temperature plots obtained from the two rigs using ammonia injection alone. Also shown are the results of Wenli et al. (1990) which were obtained in an isothermal micro-scale quartz reactor. It is clear that the peak reduction efficiency decreases as scale increases. In fact, this variation is probably due to poorer mixing associated with increasing scale, rather than to scale itself. The apparent decrease in optimum temperature in the micro-scale results arises from the fact that they are obtained under isothermal conditions, whereas the other experiments are conducted in the 5A-107 ------- presence of significant cooling gradients (-200 K/s). The mean reaction temperature under these conditions will thus be lower than the injection temperature, which is the quantity plotted on the abscissa. FURTHER WORK The ultimate objective of 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 550°F to 750°F, and is capable of achieving NOX levels in the single digits. There is as yet no long term operating and maintenance experience with SCR on large utility gas turbines in the U.S. Also, (Ref. 3) there is no significant experience on the successful use of SCR on oil fired 5B-5 ------- gas turbines. This SCR technology was introduced in the U.S. fairly recently, and its use is expanding rapidly in response to stricter regulatory requirements (Fig. 12). A number of units are being operated by cogenerators and independent power producers. The technology and O&M costs of SCR are evolving. Since the majority of systems operating to date require that the combustion products undergoing the reaction be within a 550°F to 750°F temperature window, it is necessary to cool the 950°F to 1100°F exhaust from a gas turbine by means of either a heat recovery steam generator (HRSG) or dilution with ambient air. Because of the large gas mass flows involved, dilution with air to achieve uniform mixing is technically impractical and uneconomical. Therefore a HRSG is normally used for this purpose. The resulting steam can be used for steam injection into the gas turbine for NOX reduction and power augmentation, as input to a combined cycle (or repowered) steam turbine, or as process steam in a cogeneration plant. High temperature zeolite catalysts have been introduced more recently, but they are far more costly. Zeolite manufacturers claim that these catalysts are effective and stable over a wider temperature range (and especially at higher temperatures) than base metal oxide or precious metal catalysts. The wide operating temperature range of zeolite catalysts is the most important property for NOX control. However, the specific temperature range depends on the type of zeolite. For example, a naturally occurring mordenite zeolite can operate between 430° to 970°F depending upon the specific formulation. The optimum operating temperature window for a specific formulation is ±100°F. A synthetic zeolite, ZSM-5, has a narrower operating temperature range of 570° - 900°F. A new synthetic zeolite, which is coated on a ceramic honeycomb structure is claimed to be operational at temperatures between 675° and 1075°F. However, above 800°F, 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 750°F temperature range. Below 600°F, NO conversion slows 5B-21 ------- dramatically; at temperatures above ~800-850°F, 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 950°F. Currently, there are four gas turbine zeolite applications. The other major SCR catalyst type recently applied commercially in the U.S. is a precious metal-based (e.g., platinum) catalyst. This catalyst has an operating temperature window of about 425 to 525°F, with an optimum temperature of approximately 475°F. This low- temperature operation allows placement of the catalyst outside the high pressure section of the HRSG, upstream of the economizer and stack. However, there are two limitations to this catalyst type. First, at higher temperatures (>525°F), this catalyst is an excellent NH3 oxidation catalyst, producing additional NOX. Second, it is limited to clean fuels because it is also a good 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 (<400°F), only a few ppm of NH3 and SO3 are required for reaction. Therefore, at typical HRSG exit temperatures (300 to 350°F), ammonium salt deposits are expected to form in the HRSG when firing sulfur-bearing fuels. Ammonium salt formation temperature is shown as a function of NH3 and 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 410°F, then to avoid salt formation the NH3 slip must be controlled to less than 5 ppm. However, if a CO catalyst is present in the system, the 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 950°F (510°C). 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 120°F 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 1800°F (982°C) before modifications, to a maximum of 2200°F (1204°C) with the SOFA dampers fully open. The 40/100/100 SOFA configuration was considered the typical operating mode for the Argus #26 unit. As evident in figure 2, the benefits of additional NOx reduction began to diminish with deeper staging. Figure 3 is a plot of CO emissions as a function of NOx level. CO emissions tended to increase as NOx level decreased. This resulted in part from increasing difficulty in tuning the burner air flows as more air was redirected to the SOFA ports. 5B-42 ------- The 40/100/100 SOFA staging was chosen as the 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 2020°F (1104°C) and a peak of 2110°F (1154°C) in the center of the plane were observed. The temperature was of concern since the critical level of NOx increases with increasing temperature. Chemical kinetics modeling and data from laboratory and field tests have shown that a "critical NOx" level exists 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. G«p • 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 ------- !»*» |M»lfl 1-1/7 In. • Dlw«ff 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 Upp»f«mmofit«ifi)«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 (2400°F) than at Encina Unit 2 (2250°F). 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 200°F 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 40°F 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 250°C/sec (450°F/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 200°F 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 100°F. • 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, 1470°F, CO emissions increase as the initial amount of CO addition is increased. However, for temperatures at or above 1600-1650°F, 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 1900°F 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 (1500-1600°F), 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 1500°F and 1900°F), 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 1600°F, 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 1400°F; 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 = 1600°F; 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 = 1600°F; 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 = 1400°F; 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 1400°F; 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 1600°F; 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 = 1600°F; 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 = 1600°F; Initial NOX = 250 ppm; N/NOX = 2) 6A-37 ------- Table 1 NO2 AND N2O MULTIPLE ADDITIVES T = 1600°F 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 1650°F 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 = 1650°F; 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 1650°F 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 650°±25°F. 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 650°F. 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 970°C," 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 900°C. 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 850°C. 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 700°C. 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 850°C. 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 870°C, 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 10°C, 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 fraction—essentially 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 (3000°F) 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 (650°F) pre-heated air 1.3 to 1.4 3.2 kg/s (7 Ib/s) 1300 K (1850°F) 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 3140°F) for PC firing and some 150 to 200 K (270 to 360°F) 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 1850°F, 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 2000°F. 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 'e«t ^ 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)»i«i (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 FV£NAc£ 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—|—i—i—i | i i i | i—i—i—| i > i—|—J—T—i | i i i | i i—i—|—i—i—rn—i—i—i—i—r 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 —i—i—i—i—i—i—i—i—i—i—i—i—i—i—i—i—i—i—i—i—i— 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; T—i—i—i—i—i—i—i—i—i—i—i—i—i—i—i—i—i—i—i 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 i—I—I—i—1—1—I—|—|—I—r 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 i—i—r—|—i r~r -T~;—i—i—rn—|—i—i—1-1 ] ~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 1350°C. 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 1350°C. 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 350°C (662°F). 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 1200°C. 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 1100°C, while additive C shows a "flat" profile between 975 to 1100°C. At 1100°C, additives A and B removed 90 and 84% NO. respectively, while between 975 to 1100°C 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 1100°C 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 1200°C 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 1200°C and 1300°C while all other coal-sorbent combinations were done at 1200°C. 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 500°C/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 1200°C, 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 1200°C 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 1100°C 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 1200°C. 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 1100°C 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 = 1100°C, 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 150°C. 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 120°C. 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 1200°C while NO. was 1100°C. However, with the SONOX technology (simultaneous control of both SO2 and NOJ the optimum temperature was found to be 1200°C. Injecting 5% dolomite doped PL Anne limestone slurry and additive A at the optimum temperature of 1200°C 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 1200°C. 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- 1200°C —•- 1300°Q 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 F«2°3 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 120°C 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 660°C. 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 660°C, 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 120°C 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 610°C 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 600°C. 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 660°C 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 112°F 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 120°C 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 330°C Natural gas fired in duct burners 9,994 Ibs/hr 80ft 20/xm 100 MHI 16,257 ACFM @ 120°C 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 = 2500°F. 6. Coal feedrate = 352 Ibs/hr, Flue gas flowrate = moles/hr (dry), and Temperature = 2500°F. 10. Coal feedrate = 431 Ibs/hr, Flue gas flowrate = moles/hr (dry), and Temperature = 2500°F. 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 >3000°F 1.4 sec 2265 °F 900-1200 ppm 80% -85% < 1 % in Ash 6-8 microns Typical Cyclone-Fired Boilers >3000°F 0.7-2.0 sec 2200°-2350°F 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 2700°F range, far exceeding typical boiler horizontal furnace gas outlet temperatures of 1900 2000°F. 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 -A—NOMINAL 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: 4S8°F 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: 488±6°F 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: 488±6°F 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 2200°F. 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 1800°F to 1500°F. It is readily apparent that a reduction in the burnout temperature from the 2200-2400°F 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 (850°C) 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 850°C 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 <°F) Figure 1. NH3 conversion in the "burnout zone." FUEL + AIR FUEL + AIR 20% NAT. GAS AIR AIR AND 110 0.90 1.03 1.15 10% NAT. GAS AIR AND 1.13 1.03 1.15 \ Figure 2. Advanced reburning concepts. 7B-40 ------- 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 4£v" *• 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£> o°o 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 25°C (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: 1,000°F 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 - B€ST, 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 ------- I£U ^ 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 H«H nUtUII UMIAT tnAU TlktftUTUII. I LOCI 8-55 ------- uMACt tut 4- co en 05 HOI po«r«-j- •U«Nf t(_ . •U*M<*1 - . NOPPCII —r- PO«TS -•UINCIS NOI PORTS EXISTING ARRANGEMENT NEW TWO-STAGE COMBUSTION SYSTEM •URHttf •UONCMC •UDHttC cn m oo o m r> m 2—1 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 A°™LFU 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 ------- <£OU CO 0 200 O A O CO" § 150 CO ^ LJJ co Q 10° CO ^ LJJ Q_ o: I 50 C\J 0 c JANUARY k B A° ^ A OU(3) .0F A°U(5) ALF — ^ LU(5) A OUS(2) . ou A LEGEND A LHJ TECHNOLOGIES APPLIED TO ALL APPLICABLE UNITS A LFUS<5) TECHNOLOGIES APPLIED TO SELECTED UNITS LFU TO ALL APPLICABLE UNITS, SCR TO (5) UNITS B BASELINE, NO CONTROL O OFF-STOICHIOMETRIC FIRING L LOW-NOx BURNERS F FLUE GAS RECIRCULATION g GAS FIRED UNITS ONLY U UREA INJECTION S SELECTIVE CATALYTIC REDUCTION . s A ALFU A LFgU m LFUS A ii ii 1 23 45 6 78 9 10 11 1! LEVELIZED COST OF CONTROL, Mill/kwh FIG. 7 Optimum System NOx Control Strategy to Achieve Varying Levels of Emission Reduction for Peak Generating Day in January ------- 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