By Mike Patscheck and David Mitchell, Public Service of New Mexico, and Ajay Jayaprakash, Donald Fennesy and Raj Gaikwad, Sargent & Lundy
In 2013, Public Service of New Mexico (PNM) entered into an agreement with the New Mexico Environment Department (NMED) and the United States Environmental Protection Agency (EPA) to reduce the nitrogen oxide (NOx) emissions from the San Juan Generating Station. This agreement, achieved as part of a best available retrofit technology (BART) settlement, required the installation of selective non-catalytic reduction technology to be installed to reduce the NOx emissions from the current permitted levels to a maximum of 0.23 lb/MMBtu. This paper will discuss the salient points of the retrofit of this technology at the station, including discussion of the requirements of the agreement, the procurement strategy, balance of plant considerations, and the next steps.
The San Juan Generating Station is located in Waterflow, New Mexico and is comprised of four, pulverized coal boilers that are each firing New Mexico bituminous coal. San Juan Units 1 and 2 are sister units, as are San Juan Units 3 and 4.
San Juan Units 1 and 2 are 360 MWgross-rated Foster Wheeler, pressurized draft, front wall-fired boilers. Unit 1 was commissioned into service in 1976 and Unit 2 was commissioned into service in 1973. Each unit was designed to control particulate matter (PM) emissions through the use of a stacked hot-side electrostatic precipitator (HESP) upstream of the air preheater. Each unit is equipped with four, regenerative Ljungstrom-type air preheaters: two for primary air and two for secondary air. In 1998, both units were retrofit with wet flue gas desulfurization (WFGD) systems from Babcock & Wilcox (B&W) to control the SO2 emissions from the units. During 2007-2009, the units were each retrofit with B&W fabric filter technology to control the PM emissions from the units; at this time, the fields in the HESP were de-energized. In 2009, both units were retrofit with Low NOx Burners (LNB), overfire air (OFA) systems, Sub overfire air (SOFA) systems, and neural networks designed, furnished, and installed by B&W to control the NOx emissions from the units to the current permitted levels. In 2009, both units were retrofit with an activated carbon injection (ACI) system to control mercury emissions.
San Juan Units 3 and 4 are 550 MWgross-rated B&W, pressurized draft, opposed wall-fired boilers. Unit 3 was commissioned into service in 1979 and Unit 4 was commissioned into service in 1982. Each unit was designed to control PM emissions through the use of a stacked HESP upstream of the air preheater. Each unit is equipped with four, regenerative Ljungstrom-type air preheaters: two for primary air and two for secondary air. Similar to Units 1 and 2, both units were retrofitted with B&W WFGD systems in 1998 and B&W fabric filter technology during 2007-2009 to control the SO2 and PM emissions from the units, respectively. In 2008, both units were retrofit with B&W LNB, OFA, and neural network systems to control NOx emissions from the units to the current permitted levels. In 2008, both units were retrofit with an activated carbon injection (ACI) system to control mercury emissions.
The LNB, OFA and neural network systems were installed to control the NOx emissions from their uncontrolled values to a maximum emission of 0.30 lb/MMBtu as part of a consent decree agreement. As part of the agreement, the station entered into a 12-month test period for each unit where NOx data was collected to determine if the permitted emission limit could be further reduced from 0.30 lb/MMBtu based upon the capabilities of the installed technology. The data indicated that the NOx emission limit of 0.30 lb/MMBtu was appropriate for each of the units.
In August 2011, the EPA published the final Federal Implementation Plan (FIP) for the state of New Mexico to the Federal Register. In the FIP, the EPA stated that a portion of the New Mexico State Implementation Plan (SIP) was not approved. The argument made by the EPA was that the state had failed to include requirements for regional haze considerations for the San Juan Generating Station in the SIP. This subjected the station to BART requirements for NOx and, as a result, required the installation of selective catalytic reduction systems on all four units to reduce the NOx emissions.
PNM entered into negotiations with the NMED and the EPA, doing so with the backing of the governor of New Mexico. As part of the negotiated agreement, PNM agreed to install SNCR control technology on San Juan Units 1 and 4 as part of the overall NOx emission compliance program. Based upon the agreement between PNM, the NMED, and the EPA, the EPA was requested to hold in abeyance the consideration of the NOx BART determination for San Juan Generating Station set forth in the original SIP.
In response to the agreement between PNM, the NMED, and the EPA, PNM and Sargent & Lundy (S&L), the Owner’s Engineer, developed an “engineer and procure” specification for the SNCR systems to be installed on Units 1 and 4. The resultant specification was then released for competitive proposals from industry leaders in SNCR technology.
As part of the specification, the SNCR Supplier would be required to perform combustion testing of the boilers, make recommendations for optimization of the combustion systems, and perform all required temperature and NOx mapping and all computation flow dynamics (CFD) modeling in addition to the design and supply of the urea solution storage, feed and injection equipment to reduce NOx emissions from the boilers.
As part of the evaluation, special consideration was given to balance of plant equipment that was not included in the scope of supply for the SNCR vendor, but is required for successful operation of the proposed design. This equipment includes water treatment systems required to provide minimum water quality for use as dilution water, closed cooling water systems, and compressed air requirements for pneumatic atomization of the injected urea solution. These specific requirements were included in the evaluation to determine the most cost-effective solution, not just on a specification basis, but on a project basis.
The agreement between PNM, the NMED, and the EPA required NOx emissions to be controlled to a value of 0.23 lb/MMBtu from the existing controlled permitted emission limit of 0.30 lb/MMBtu for each unit. In order to have margin below the permitted emission limit, the specification required a demonstrated NOx emission guarantee of 0.22 lb/MMBtu. This equates to a NOx reduction efficiency of greater than 26% for each boiler. At the time when the specification was issued, this NOx removal efficiency had not been demonstrated on pulverized coal-fired boilers of this type, size, and inlet NOx emissions.
In addition, similar to the requirements of the previous consent decree agreement to install the LNB, OFA and neural networks, PNM agreed to participate in Long-Term Performance Evaluation Period of nine (9) months subsequent to the installation and commercial operation of the SNCR systems. If, during the performance evaluation period, PNM is able to demonstrate NOx emission rates less than or equal to 0.20 lb/MMBtu on a daily 30-day rolling average, then the permitted NOx emission limit will be reduced commensurate to an agreed-upon percentage of the difference between the permitted emission limit and the demonstrated NOx emissions. During all periods, however, the SNCR systems shall be operated with an ammonia slip from the SNCR system below 10 ppm, which will limit how much urea can be fed into the boiler to reduce NOx emissions.
SNCR systems inject urea solution into an identified temperature range within the boiler that allows the urea to hydrolyze into ammonia and react with NOx to form nitrogen gas and water. In order to maximize the penetration of the urea solution across the boiler prior to hydrolyzing, the urea solution is injected in concentrations of approximately 5-10% by weight. However, it would not be economical for a station to take delivery of urea at this diluted concentration. Therefore, stations employing SNCR technology typically take delivery of either 52% (by weight) urea solution or dry urea pellets, which are then solutionized on site.
S&L performed a urea feed study evaluating the feasibility of installing a solutionizing system on site compared to taking delivery of urea solution. S&L originally estimated the consumables required to meet the outlet NOx emission using industry experience and engineering judgment as well as developed a differential order of magnitude capital cost estimate; these consumables included the urea consumption, the auxiliary power, steam and water requirements.
Unit costs were provided by PNM and their supply chain and S&L was able to develop annual fixed and variable O&M costs. The delivered reagent cost for urea was found by PNM to be $600/ton for dry pellets and $2/gal for urea solution. Using these costs, a net present value (NPV) comparison was developed to directly compare the two technology options. This study indicated that there was a potential that on site solutionizing of urea would be economically feasible for San Juan Generating Station. An addendum to the SNCR specification was prepared and issued to the SNCR technology suppliers to provide the equipment as an option. Once a SNCR vendor was contracted, the consumable values used in the study were revised based upon the guaranteed values provided by the successful technology supplier.
These updated results validate that solutionizing urea on-site at San Juan Generating Station is economically feasible. Provisions were allowed in the layout of the equipment to allow for the expansion of the urea tank farm to include the future installation of the solutionizing equipment. Detailed design of the dry urea unloading, storage, and solutionizing tanks has been completed by the SNCR vendor, but this option has not been fully exercised.
balance of plant considerations
Unrelated to the proposed SIP revision and the required reduction in NOx emissions, PNM is also undertaking other projects concurrently to improve the reliability and maintainability of the station. These include a bypass around the Unit 1 HESP equipment and conversion of both units from pressurized draft to balanced draft.
As stated above, the HESP systems on each unit have been de-energized subsequent to the installation of the fabric filter systems. However, the HESPs are still in the gas path and, due to the reduction in velocity in the equipment, a significant portion of the fly ash generated by the station is removed from the flue gas due to drop out. This requires the station to continually maintain the HESP systems and operate the ash collection system. S&L prepared a conceptual design to evaluate the removal of the HESP on Units 1 and 4 from the flue gas path. It was determined that it was cost effective to bypass the HESP on Unit 1, but not for Unit 4. Therefore, S&L began detailed design to remove the HESP from the flue gas path on Unit 1 only. As part of the detailed design, S&L performed additional studies to determine downstream impacts from removing the HESP from the gas path. These included evaluation of the fabric filters and the ash removal system from the fabric filter hoppers, evaluation of the air preheater baskets, and design of the ductwork to bypass the HESPs.
When the fabric filters were originally designed, the basis was the full dust loading of the flue gas, which is roughly 18.5 lb fly ash/MMBtu for the coal fired at this station. It was then concluded that no modifications would be required for the fabric filters to operate successfully while treating the full dust loading. The ash removal system was designed to remove all of the fly ash that would be collected from the flue gas should the Unit 1 HESP be removed from the gas path, but all of the required equipment was not installed. The original equipment manufacturer (OEM) design included a bifurcated hopper design and installed feeders, valves, and piping on one connection. As part of this project, feeders, valves, piping and all required cross-ties were procured for the second connection on the bifurcated hopper.
Because the HESP had always served to remove a significant portion of the fly ash in the gas path, even in a de-energized state, there was also concern that the removal of the HESP from the gas path would result in an overload of fly ash to the air preheaters. This, in turn, could lead to erosion of the baskets and/or decreased performance of the boiler. In order to mitigate this issue, PNM contracted with the air preheater OEM to evaluate the air preheaters and determine if the installed baskets were suitable for the increased ash loading. As part of this evaluation, the OEM was also tasked with determining if there would be any adverse impacts on the air preheaters due to the introduction of ammonia into the gas path from slip from the SNCR system. The OEM concluded that the baskets that were currently installed were suitable for the full dust loading that was expected. In addition, the OEM concluded that due to the relative ratios of ammonia and SO3 present in the flue gas, the ammonia present in the flue gas would likely exist as ammonium sulfate, which would have no adverse impact on the air preheaters performance.
Unit 1 has a split economizer resulting in two parallel gas paths that exit the boiler. When the flue gas exits the Unit 1 boiler, each path of the ductwork penetrates the boiler building wall and immediately splits into two smaller ducts to feed the upper and lower HESPs which are in a stacked arrangement. As part of the conceptual design, S&L proposed to remove the ash hoppers from the upper HESP casing and use the resulting space as the location of the HESP bypass duct. This would minimize the modification required to the existing steel and allow tie-in to the existing primary and secondary air connections more readily. As part of the detailed design, S&L performed CFD modeling of the proposed ductwork arrangement to ensure that the new duct design would be optimized, all required turning vanes designed, and to minimize locations of extremely high and low velocities. By minimizing the locations of high and low velocity, it could be ensured that there was low risk for erosion of the ductwork as well as low risk for drop out and deposition of fly ash in the ductwork. The results of the CFD analysis are presented in detail below.
S&L developed a CFD model using the preliminary ductwork design in a commercially available CFD modeling program. The preliminary ductwork design was developed in such a manner to minimize the construction costs and impacts on existing structures, but without consideration for internal members. As stated above, the goal of the modeling was to optimize the duct design to the greatest extent possible, design turning vanes, and minimize locations of extreme velocity. With physical limitations due to existing structures and constructability concerns in mind, the proposed ductwork was modeled at both full and low load conditions. Low load had been defined by PNM as 40% of MCR conditions. Due to this large operating load range, it was expected to be very difficult to completely avoid extreme velocities. In these cases where extreme velocities were unavoidable, recommendations for operation and maintenance were to be provided.
Because the ductwork is mirrored across the economizer, only one side was modeled to determine the results. The baseline duct geometry for the South train is shown Figure 1 on page 30. The economizer outlet is shown on the left side of the figure, with the tie-ins to the existing air preheater inlets on the right side of the figure.
Boundary conditions for the model were developed as part of the design basis calculation prepared in advance of the SNCR specification. The expected flow split between the primary and secondary air preaheaters was determined by the geometry of these connections.
Using these boundary conditions, the flow model was developed to determine the baseline results. These results show that for the large area of the common duct, the flow is relatively homogenous with no areas of concern. When the ductwork splits to direct flow to the two air preaheater inlets, the flow becomes extremely stratified and areas of extreme velocity are readily identifiable. The CFD results from baseline, full load conditions are shown for the primary air preheater inlet and secondary air preheater inlet in Figures 2 and 3, respectively.
These results show that in the primary air heater inlet duct, the flow is generally homogenous entering the first downward turn. At this point the bulk of the flow is along the back wall leading to recirculation zones at the top of the ductwork and along the near wall. Just prior to entering the primary air inlet, the ductwork takes a horizontal turn, causing an area of low flow along the top of the ductwork. In each of these areas, the velocity is low enough that drop out of fly ash is expected. In the case of the low flow zone at the primary air heater inlet, there is sufficient flue gas velocity immediately below the low flow zone to re-entrain any fly ash that may drop out of the flue gas. However, it was determined that the flue gas was too stratified in the down-comer duct and the velocity should also be increased as part of the optimization of this section of ductwork to prevent drop out of fly ash.
Due to the existing physical constraints in this area and expected ductwork support locations, the new ductwork to the secondary air preheater inlet makes multiple transitions prior to the connection with the existing secondary air preheater inlet ductwork. In order to minimize construction impacts, the connection was determined to be at the expansion joint upstream of the dampers located at the secondary air preheater inlet. These transitions occur in quick succession and result in the bulk of the flow taking the path of least resistance to the secondary air preheater inlet.
This concentrates the flow through the one path, experiencing high localized velocities that are greater than 100 fps along the inside corner of the turns, which is expected to cause erosion of the duct plate. Conversely, opposite these points of high velocity, there exist large areas of low velocity.
These zones with flue gas velocity below 20 fps pose a potential risk for fly ash deposition due to drop out. Turning vanes and other flow correction devices were recommended to be added to these sections to reduce the stratification of the flow and distribute the gas more evenly across the ductwork.
The optimized ductwork arrangement was developed by S&L via collaboration of the process, mechanical, and structural disciplines working on the project. In order to increase the velocity entering the primary air heater inlet duct, the size of the ductwork was decreased and turning vanes were located along the downward turn to distribute the flow more homogenously across the duct cross section. In the secondary air preheater inlet duct, numerous sets of flow correction devices were required to distribute the flow across the cross section of the duct. These included turning vanes, diverter vanes and chamfered corners of the ductwork.
The optimized geometry resulted in favorable results for the flue gas flow into the primary and secondary air preheater inlets.
The areas of recirculation and low flow are greatly reduced and the overall flow profile entering the primary air preheater inlet is more homogenous.
The optimized results are a significant improvement in flow distribution when compared with the baseline results in Figure 3. The areas of both extremely high and low velocity have been largely mitigated with exception to the final expansion in the new ductwork segment. This expansion in the ductwork is required to match up with the existing expansion joint frame at the secondary air preheater inlet. No iteration of turning vanes was able to distribute flow adequately to this corner to prevent potential of fly ash dropout.
Overall, the results of the CFD modeling were determined to be very favorable and duct design was able to continue to the next phase. The flow distribution between the primary and secondary air preheater inlets, as shown in Table 3, was expected to be 25%/75%.
Balanced Draft Conversion
To further improve the operability, and maintainability of the station, as well as reduce fugitive dust emissions to meet the requirement of the Station’s Title V permit, PNM determined that it would be beneficial to convert Units 1 and 4 from pressurized draft operation to boiler balanced draft operation.
In order to accomplish this, modifications to the furnace, boiler, electrical systems, forced draft (FD) and induced draft (ID) fans and motors, and ductwork and equipment design may be required. As part of this evaluation, S&L performed an evaluation of the FD and ID fans to determine their viability for future use at the site.
It is expected that the conversion to balanced draft will change the required pressure rise from current pressurized draft operation. The total pressure rise from the FD fans will decrease, while the total pressure rise from the ID fans will increase commensurately. In order to validate the fan curves and accurately measure the existing system resistance, performance tests were conducted at the station on the ID fans.
The Unit 1 ID fans are currently damper controlled with a full pressure rise of 27″ at a flow of 580,000 cfm (per fan). The conversion to balanced draft is expected to increase the required full load pressure rise by 10″. However, because it was determined that the HESP would be removed from the flue gas path for Unit 1, the total pressure rise could be reduced by 2.5″ and any increase in flue gas volume attributed to infiltration from the HESP casing could be removed. Based upon this adjustment in total pressure rise and flow, the existing ID fans were determined to be acceptable for future use as long as variable frequency drives (VFD) were added to allow the full operational capacity of the fans to be used and controlled. One added benefit of this conclusion is that the inlet dampers to the fan could be left at full open, thereby further reducing the pressure rise required of the ID fans.
The Unit 1 FD fans were found to be acceptable for future use without modification.
The Unit 4 ID fans are currently inlet vane controlled with a full pressure rise of 26″ at a flow of 920,000 cfm (per fan). The conversion to balanced draft is expected to increase the required full load pressure rise by 10″.
Currently, the project is progressing in accordance with the timeline set forth in the agreement between PNM, the NMED, and the EPA to ensure reduction in NOx emissions by the SNCR system by January 31, 2016. The SNCR vendor is nearing completion of engineering and will be starting fabrication and delivery of equipment in the first quarter of 2015.
Upon completion of the installation of the SNCR system in 2015, the system shall be tested by a third-party testing contractor to verify achievement of the required performance guarantees. Subsequently, the systems will begin the Long-Term Performance Evaluation Period as discussed above.
Power Engineerng Issue Archives
View Power Generation Articles on PennEnergy.com