Use of ROFA and Rotamix to Reduce NOx in a Riley Turbo
Transcription
Use of ROFA and Rotamix to Reduce NOx in a Riley Turbo
Use of ROFA and Rotamix to Reduce NOx in a Riley Turbo-Fired Boiler Burning Landfill Gas and Natural Gas John Ralston, Jamie Fessenden, Geoffrey Green, and Brian Higgins Mobotec USA and Gil Espinosa Glendale Public Service Department ABSTRACT A MobotecSystem was installed on Glendale Water and Power’s Grayson Unit 5 during spring 2003 for NOx control. Grayson Unit 5 is a 44-MW Riley turbo-fired boiler with induced flue gas recirculation (IFGR) burning a blend of landfill gas (LFG) and natural gas (NG). The installed MobotecSystem consists of ROFA and Rotamix. The MobotecSystem reduced NOx emissions by 48% with a 30%-LFG and 70%-NG mixture. A 58% reduction in NOx occurred with 100% NG. While burning the LFG/NG blend at full load the NOx emission was 61 ppm without the MobotecSystem and 32 ppm with the MobotecSystem. The Grayson Power station is load limited by a fixed (lb/day) NOx emission rate. With the NOx reduction from the MobotecSystem, Unit 5 is able to nearly double its daily generation capacity. INTRODUCTION Due to tightening air quality regulations imposed by the South Coast Air Quality Management District (SCAQMD), the City of Glendale was required to reduce their yearly, daily, and hourly NOx emissions. In December 2002, a contract was signed between the City of Glendale and Plant Maintenance, Inc, for the installation of a MobotecSystem™ ROFA© and Rotamix© System at the Grayson Steam Plant Unit #5 to reduce NOx at 40 MW, while maintaining an ammonia slip of less than 10 PPM. Grayson Unit #5 is 44 MW Riley Turbo-Fired forced draft boiler. In 1996, Glendale Water and Power (GWP) modified the unit to burn a 30/70 blend of land fill gas (LFG) or 100% natural gas (NG). Unit 5 has six burners, 3 on the front wall and 3 on the back wall specifically designed for LFG and NG. LFG has approximately 30% of the heating value of NG and greater than 20 times more fuel-bound nitrogen. Table 1 presents the gas composition of LFG and NG. Burning landfill gas is desirable for economic and environmental incentives. Also in 1996, GWP installed an induced flue gas recirculation (IFGR) system. The IFGR provides a NOx reduction of 35-40% over and above Mobotec’s NOx reduction. 1 Table 1: Landfill gas and natural gas composition % by Volume Landfill Gas Natural Gas CH4 50 96.33 C2H6 0 1.57 C3H8 0 0.25 N2 16 0.71 CO2 31 1.14 O2 H2O 3 0 0 0 TOTAL % 100 100 HHV (Btu/SCF) 300 1009 BACKGROUND The MobotecSystem is a three tiered solution to multi-pollutant control. The MobotecSystem consists of ROFA, Rotamix, and an SCR. ROFA (Rotating Opposed Fired Air) directly affects combustion and NOx emissions by staging the furnace and turbulently mixing the flue gas with high velocity secondary air. Rotamix is an advanced selective non-catalytic reduction (SNCR) system that makes use of the well mixed flue gas and high velocity air to evenly distribute chemicals into the flue gas. Both SNCR reagent injection (NOx control) and sorbent injection (SOx and Hg control) can be utilized in a Rotamix system1 . An SCR (selective catalytic reduction) in-duct catalyst can be added with the MobotecSystem at reduced costs over a standard SCR due to the size reduction available by the NOx reduction from ROFA and Rotamix. With ROFA, the gas volume in the furnace is set in rotation via special asymmetrically placed air nozzles. The flue gases mix well with the added air, due to increased turbulence mixing and rotation in the entire furnace. This improves temperature and species distributions and improves particle burnout in the upper furnace. Mixing and rotation prevents the formation of stratified laminated flow, which enables the whole furnace volume to be used more effectively for the combustion process. The ROFA swirl reduces the maximum temperatures of the combustion zone and increases heat release, which may improve boiler efficiency. Mixing the combustion air more effectively with ROFA can also reduce surplus excess air. This results in less cooling of the furnace due to unused combustion air thereby increasing the heat absorption. Some the documented advantages of the ROFA technique are: • • Less temperature variation in the cross-section of the furnace. A more even distribution of combustion products in the cross-section of the furnace (e.g., O2 , CO, and NOx). 1 Haddad et al., “Full-Scale Evaluation of a Multi-Pollutant Reduction Technology: SO2 , Hg, and NOx”, MEGA Symposium #117 (2003). 2 • • Lower CO levels which means reduced excess air. Less excess air (O 2 ) which means less NOx and higher overall efficiency. Rotamix is a third-generation SNCR and sorbent injection system. The turbulent air-injection and mixing provided by ROFA allows for the effective mixing of chemical reagents with the combustion products in the furnace. The result is the efficient introduction of reducing chemicals directly into a well-distributed, rotating mixture. The Rotamix system is tuned to adapt to changes in load and temperature in the furnace, and only introduces reducing chemicals to the furnace where the temperature is most favorable for pollution reduction. This reduces chemical consumption and lowers chemical slippage by increasing the reaction efficiency. Relative to other SNCR systems, Rotamix can decreases recurring chemical costs. While NOx reduction with typical low-NOx burners (LNB) and OFA systems can sometimes exceed 50%, ROFA NOx reduction routinely exceeds 60% and Rotamix NOx reductions routinely exceed 35%. The minimum NOx reduction achievable with the MobotecSystem is 75% and often approaches 90%. The third tier to the MobotecSystem is the installation of an SCR. As shown by Haddad et al.2 , every dollar spent on the ROFA and Rotamix system SCR costs by $2.05. MOBOTECSYSTEM The MobotecSystem installed on Grayson Unit #5 consists of: • • • • • • • A ROFA Fan and inlet damper ROFA and Rotamix ductwork Individual flow-controlled ROFA & Rotamix boxes (each with several directed nozzles) An ammonia storage tank and pump skid A water storage tank and pump skid A Rotamix flow-control cabinet for ammonia and water delivery A Moboview control system ROFA Process: The ROFA Fan draws air from the ductwork between the FD fan and the air preheater inlet. The ROFA fan delivers pressurized air at 120 to 140°F to the ROFA and Rotamix boxes strategically positioned on the furnace side walls. The amount of secondary air diverted to the individual ROFA and Rotamix boxes and into the furnace varies with load and is controlled based on steam flow. The Rotamix system consists of a reagent tanks, water and reagent pump skids, and water and ammonia to the Rotamix cabinet, The water and ammonia are mixed inside delivery system including reagent and water storage delivery lines. The delivery lines supply pressurized located near the Rotamix boxes at the upper furnace. the Rotamix cabinet and are delivered to individual 2 Haddad et al., “The Viability and Economics of Adding a ROFA/Rotamix MobotecSystem to a Selective Catalytic Reduction (SCR) Installation,” NETL/DOE 2003 Conference on SCR and SNCR for NOx Reduction Pittsburgh, PA October 29-30, 2003. 3 injectors. The Rotamix injectors consist of delivery air, cooling air, and diluted reagent mixture. The locations of the ROFA and Rotamix boxes are determined from CFD modeling and field test data. CFD MODELING The performance of a MobotecSystem is strongly dependant on the design phase. A proven method to evaluate and validate the design from a fluid flow and combustion perspective is to use computational fluid dynamics (CFD). To perform a CFD analysis of a MobotecSystem, a baseline case is defined with full-load operating conditions including fuel flow rates, air flow rates, inlet temperatures, etc. Upon the completion of the baseline case a ROFA case is defined and incorporated into the baseline CFD model. Several key parameters such as O2 , CO, NOx, unburned fuel, and temperature are evaluated to validate the predicted performance. The CFD design methodology is an iterative process which includes multiple case studies. The finalized/optimized CFD ROFA model is used as the design basis for the installed system. The objective of applying CFD modeling to Grayson Unit 5 is to provide better understanding of how the unit behaves and to provide a determination of how the ROFA technology affects NOx production and unit operation. Model results can be helpful in providing improved understanding of furnace combustion. Because models inherently contain physical assumptions, results are best used to indicate trends in combustion behavior in comparison to modeled baseline levels. In addition, furnace measurements and observed furnace behavior are essential to understand model results. The computer model used for this study is a CFD-based reacting flow code, GLACIER, developed by Reaction Engineering International (REI). The code couples the effects of turbulent fluid mechanics, gas-phase combustion chemistry, finite-rate NOx chemistry, and convective and radiative heat transfer. The REI combustion code assumes that the flow field is a continuum field that can be described locally by general conservation equations. The flow is assumed to be steady state and gas properties are determined through local mixing calculations. The fluid is assumed to be Newtonian and dilatation is neglected. The comprehensive model uses an Eulerian-Lagrangian framework and handles either Reynolds- or Favre-averaging. The code couples the turbulent fluid mechanics and the chemical reaction process, using progress variables to track the turbulent mixing process and equilibrium chemistry to describe the chemical reaction process. The CFD model predicted a slight increase in exit temperatures and CO emissions and a significant decrease in NOx. The CFD results as compared with actual results are presented in Table 2. Note from Table 2 that the predicted NOx concentrations are lower than the actual measured NOx concentrations. This is due to the fact that at these very low NOx concentrations, a high degree of accuracy in the computational modeling is expensive in terms of the number of nodes and the level of chemical kinetic complexity required to accurately model the furnace and the burners. While the NOx concentrations are not necessarily accurate, the trends between the two models are considered to be accurate. 4 Table 2: CFD model results versus actual results at 43.5 MW NOSE Temp (K) Baseline Model 1545 ROFA Model 1588 Baseline* Measured - ROFA** Measured - Temp (°F) 2321 2399 - - CO (dry, ppm) 5585 6057 - - O2 (dry, %) 2.89 2.65 - - 25 10 - - Baseline Model 1169 1645 ROFA Model 1195 1691 Baseline* Measured ROFA** Measured CO (dry, ppm) 388 555 O2 (dry, %) 2.38 2.37 NOx (dry, ppm) EXIT Temp (K) Temp (°F) NOx (dry, ppm) 24 9 61 32 lb-NOx/MMBtu 0.027 0.010 0.063 0.035 % NOx reduction *Baseline data at 40.4 MW **ROFA data at 42.9 MW 63% 48% On the following pages are NOx, temperature, and O2 profiles. These can be used to further understand the flow through the furnace and how ROFA affects this flow to reduce NOx. The ROFA condition as-shown defined the actual design case. That is, this case was used to specify the box locations, nozzle sizes, nozzle angles, and exit velocities as built and constructed. This case was arrived at after analyzing multiple cases and making design adjustments to affect NOx, temperature, and oxygen distributions. The first two figures show the NOx profiles for the baseline case without ROFA (Fig. 1) and the “as-designed” case with ROFA (Fig. 2). The difference in NOx concentrations between the baseline and ROFA case are considerable. The baseline case has more NOx throughout in the furnace. This is because NOx reduction with ROFA occurs throughout the furnace, not just at the ROFA ports. In the baseline case, it appears that there is a significant build up of NOx near the front left corner. This is due to high temperatures in that region (which is also evident in Fig. 3). For the ROFA NOx figure (Fig. 2), the overall NOx quantity is lower and it is isolated in the center of the furnace. The NOx in the exit plane is far more uniform for the ROFA case than for the base case. 5 Figure 1: Baseline NOx levels (0 to 50 ppm) NOx (ppm) Figure 2: ROFA NOx levels (0 to 50 ppm) NOx (ppm) 6 Figures 3 & 4 contain the predicted temperature for the baseline and ROFA cases. The hot spot in the front left corner of the base case is clearly discernable. The effect of ROFA is also seen as cold jets propagate across the furnace and mix with the combustion products coming up the center of the furnace from the burner level. Due to high rates of heat transfer in the convective backpass, the temperature profiles at the exit of the furnace are similar. Figure 3: Baseline temperature (0 to 2800ºF) Temp (°F) Figure 4: ROFA temperature (0 to 2800ºF) Temp (°F) 7 In Figures 5 & 6, the O2 levels are represented. Note the oxygen stratification in the baseline case as the products round the nose. The ROFA jets are clearly delineated in the bottom figures. Also, reduce O2 at the burner level is visible in the ROFA figure. Figure 5: Baseline O2 levels (0 to 10%) O2 (%) O2 (%) Figure 6: ROFA O2 levels (0 to 10%) 8 INSTALLATION Detailed (post-CFD) engineering of the installation began in December 2002. The technical engineering was performed by Mobotec USA and the Balance of Plant (BOP) Engineering was performed by Industrial Design Services. UCI, Inc. successfully completed the physical installation of the MobotecSystem under the direction of Plant Maintenance, during spring 2003. Commissioning and start-up were finished the first week of June, 2003. MobotecSystem tuning started immediately after commissioning and was concluded in two months. DATA AND RESULTS The Mobotec System achieved significant reductions at Grayson Unit #5 with the NG, LFG, and the previously installed IFGR system. Tables 3 & 4 list the performance test results for the unit fired with natural gas (NG) only. Tables 5 & 6 list the performance test results for the unit fired with a blend of landfill gas (LFG) and NG. The NOx reduction achieved with 100% NG is 59%. The NOx reduction achieved with LFG and NG is 48%. The difference in NOx emission levels with and without LFG lies partly in the difference in fuel nitrogen content. With LFG, the total amount of fuel bound nitrogen increases from 0.7% to 5.0%. Also because LFG has a lower heating value than NG, the stoichiometric air requirement increases with LFG. Increased total air flow has two drawbacks: (1) thermal NOx increases and (2) total IFGR quantity decreases. An induced flue gas recirculation (IFGR) system is typically employed to allow a plant to run at low load and still achieve reheat temperatures. However, at this site, the IFGR was installed to achieve lower NOx through dilution of fuel and air during combustion. The MobotecSystem was tested with and without IFGR in service. The IFGR system improved the NOx reduction when coupled with the MobotecSystem. The NOx reduction achieved from the IFGR system was 35 to 40% when coupled with the MobotecSystem. Because NG has very low fuel nitrogen content, the majority of the NOx formed during combustion is thermal NOx. The IFGR dilutes the oxygen and thus reduces peak combustion temperature. Fundamentally, the NOx reduction from the IFGR and the MobotecSystem are additive. 9 Table 3: Natural Gas Only without Mobotec System Test Load, MW Stack Flow, Kdscfm NOx ppm, @ 3% O2 NOx lb/hr NOx lb/MW-hr Min Load 13.6 90 37.6 8 0.585 50% Load 23.1 52.8 44.7 14.4 0.625 75% Load 31.9 70.5 60.5 26.5 0.831 Table 4: Natural Gas Only with Mobotec System Test Load, MW Stack Flow, Kdscfm NOx ppm, @ 3% O2 NOx lb/hr NOx lb/MW-hr Reduction % Min Load 13.0 39 12.6 2.6 0.203 65.3% 50% Load 22.1 55.5 19.6 6.4 0.290 56.2% 75% Load 31.3 75.1 24.7 11.1 0.356 59.2% 100% Load 42.7 103.3 42.1 25.9 0.607 * *Baseline (Figure 1) 100% Load not measured due to hourly and daily NOx limit Figure 7: Baseline and ROFA NOx emission for 100% NG 30 25 NG without MobotecSystem NOx emission rate (lb/hr) NG with MobotecSystem Expon. (NG with MobotecSystem) 20 15 59% NOx Reduction 10 5 0 0 5 10 15 20 25 Load (MW gross) 10 30 35 40 45 Table 5: Landfill Gas with Natural, without the Mobotec System Test Load, MW Stack Flow, Kdscfm NOx ppm, @ 3% O2 NOx lb/hr NOx lb/MW-hr Min Load 17.7 51.0 24.3 7.0 0.397 50% Load 23.0 64.8 31.5 11.9 0.516 75% Load 32.2 80.6 40.9 20.2 0.625 100% Load 40.4 96.3 61.2 36.7 0.908 Table 6: Landfill Gas with Natural Gas, with the Mobotec System Test Load, MW Stack Flow, Kdscfm NOx ppm, @ 3% O2 NOx lb/hr NOx lb/MW-hr Reduction, % Min Load 18.0 52.7 13.2 4.1 0.228 45.8 50% Load 23.6 61.7 15.6 5.9 0.249 50.3 75% Load 33.0 83.9 23.9 12.1 0.365 41.6 100% Load 42.9 108.6 32.1 20.9 0.487 47.6 Figure 8: LFG and NG baseline and ROFA comparison 40 LFG and NG without Mobotec System LFG and NG with MobotecSystem 35 Expon. (LFG and NG with MobotecSystem) NOx emission rate (lb/hr) 30 48% NOx Reduction 25 20 15 10 5 0 0 5 10 15 20 25 Load (MW gross) 11 30 35 40 45 CONCLUSIONS The Mobotec ROFA/Rotamix System is an effective low-cost solution for NOx reduction in NG and LFG/NG fired units. In the case presented, NOx emissions were reduced nearly 48% at Grayson Unit #5 for the LFG/NG blend. NOx emissions were reduced substantially more when burning NG only. The quick turnaround time, from contract signature to operational unit, makes the Mobotec System even more attractive when coupled with total cost. GWP’s Grayson Unit 5 can be operated at twice the capacity factor while staying under the NOx emission limit with the installation of the MobotecSystem. The IFGR system worked very well with ROFA system. The NOx reduction achieved with the IFGR in addition to the Mobotec System was 35 to 40%. CONTACT INFORMATION Eric B. Fischer Mobotec USA Inc. [email protected] PO Box 2025, Orinda CA 94563 925-962-9040 ext 11 Edwin Haddad Mobotec USA Inc. [email protected] 8 McCarthy Circle, Framingham, MA 01702 508-872-1610 12