United States Environmental Protection Agency Industrial Environmental Research Laboratory Research Triangle Park NC 27711 * Research and Development EPA-600/S7-81-078 Mar. 1983 Project Summary Continuous Emission Monitoring at the Georgetown University Fluidized-Bed Boiler Charles W. Young, Edward F. Peduto, Peter H. Anderson, and Paul F. Fennelly The report gives results of a con- tinuous emission monitoring program for SO2. NO«, and paniculate matter at Georgetown University's 100,000 Ib steam/hr fluidized-bed boiler, to assess emissions control performance. Because the system was still in an extended shakedown phase, several key operating conditions (e.g., level of excess air, percent flyash recycle) were not operating in the intended design range. Consequently, an in- depth engineering analysis was neces- sary to interpret the emission data. On a daily average basis desulfurization was > 75% on all 24 days of record, > 85% on 12 days, and > 90% on 8 days. Although NO, emission rates were higher than 301 ng/J approximately half the time, they were shown to correlate with the flue gas 02 levels, typically in the off-spec range of 10- 12%. Average particulate emission rates for the 2 days of record were 36.5 and 24.3 ng/J. Implementation of recommendations resulting from the program are in most cases complete or in progress, and are leading to improved emission performance. This Project Summary was developed by EPA's Industrial Environmental Research Laboratory, Research Triangle Park, NC, to announce key findings of the research project that is fully documented in a separate report of the same title (see Project Report ordering information at back). Introduction The objective of this study is to describe the emissions control per- formance for SO2, NOx, and particulate matter for the Georgetown University fluidized-bed boiler (FBB), with emphasis on the relation of emission rates to boiler operating variables. This report is a preliminary assessment of the George- town FBB performance, and is based on measurements made during August and September 1980. A follow-up program was conducted in February and March 1982, results of which will also be published. Fluidized-bed combustion (FBC) of coal is a promising technology for low SO2and NO*coal-fired industrial steam generation. The 100,000 Ib (45,450 kg) steam/hr demonstration FBB at Georgetown University in the District of Columbia is the largest operational application of atmospheric FBC of coal using a limestone bed for SO2 capture. Because FBC technology has the potential to become an econom- ically and environmentally competitive technology for coal-fired industrial steam generation, the emissions control performance of the Georgetown FBB is of significant interest to potential users of FBC as well as to research and development officials of the U.S. EPA. Although much emission data exist for FBC technology, they result mostly from short-term testing on bench and pilot-scale units. Because the George- town FBB is a fully operational unit in the same capacity range as many industrial steam generators, it offers one of the first opportunities to obtain long-term emission control data for a coal-fired FBB. Additionally, long-term continuous emission monitoring data ------- on the order of 1 month for S02 and NOX are necessary to build a data base that adequately describes the generic emis- sion performance capabilities for FBC industrial steam generation. This type of data base is important to EPA in the development of New Source Performance Standards for industrial boilers. Due to the lack of full-scale operating units, the present data base for FBC technology is insufficient to support the current round of industrial boiler standards develop- ment by identifying FBC as an alternative coal-fired steam generation technology. The evaluation program at the George- town FBB is designed to provide the type of continuous emission monitoring data utilized in standards development. Furthermore, the results of this moni- toring program have been compared with three emission levels that have been proposed as being representative of stringent, intermediate, and moderate levels of control for SOa, NOX, and particulate emissions for coal-fired industrial boilers.1 The continuous emission monitoring (CEM) program at the Georgetown FBB was conducted from August 19,1980 to September 18, 1980. During this period, 23 days of continuous S02 emission data and 24 days of continuous NOX emission data were obtained. Additionally, particulate emissions were measured at the stack in triplicate sets on 2 days during the test program. The test program was designed to concurrently measure a number of boiler operating variables (e.g., percent flue gas Oz, calcium to sulfur molar feed ratio, bed temperature, and gas-phase residence time). The discussion of emission results focuses on relationships between these operating variables and emission rates. This analysis is an important part of the study because the boiler did not always operate according to design conditions and, therefore, performance was not necessarily typical of that expected in the future at Georgetown, or in general commercial applications of FBC. The Georgetown FBB burns coal of low to medium sulfur content and has a rated capacity of 100,000 Ib/hr steam. Limestone is added to the bed for SO2 control, and fabric filters are used for final particulate control. FBC provides inherent control of NO* emissions through low combustion temperature, approximately 843°C (1550°F), which practically eliminates thermal NOX formation and provides a favorable atmosphere for chemical reduction of fuel-derived NOX. Operation of the Georgetown FBB is one of the most successful and largest applications of FBC of coal in the U.S. The facility has demonstrated reliability in following steam demand for University heating and air conditioning. The boiler performed two continuous runs of about 15 days each during the preliminary monitoring program in August and September 1980. Soon after completing the moni- toring program, a continuous run of about 30 days was achieved. Monitoring Program The CEM system used for this program was the on-site system installed and operated by the Georgetown University Physics Department. This is an extractive system that continuously analyzes flue gas emissions to the atmosphere. Flue gas is extracted through a sintered probe filter and is transported to an instrument shelter, through heat-traced Teflon tubing. Inside the instrument shelter, the sample stream passes through a pump and is then split into two streams: one passes through a combined dilution/conditioner system; the other passes through a condensation system. Table 1 lists the instrument type and the mode of sample conditioning required for each gas species analyzed. SOz and NOXsample streams are diluted with air by a factor of 10 to 1. Dilution reduces the moisture level of the sample stream by a factor of 10 and provides a constant sample matrix (i.e., constant Oz and C02 concentration) that simplifies data interpretation in the S02 analyzer and eliminates possible fluo- rescence quenching. The CEM system monitors levels of CO, COz, and 02 directly without sample dilution. These sample streams are conditioned by filtering particulate through a coarse probe filter and backup fiber filters. Moisture is removed by a condenser coil. Calibration gases are injected at the probe interface connections through a motorized three-way ball valve. This injection point allows calibration gases to flow through the entire CEM system (except for the probe and stack filter). Calibration gas can also be injected directly into the analyzer; but this was not done during the test program. Individual strip chart recorders provided a permanent copy of continuous emis- sions of SOz, NO* CO, Oz, and CO2. The Oz measurement was used to compute diluent volume and emission rates, as opposed to COz, to avoid error introduced by COz production during limestone calcination. Before initiation of data collection, the continuous monitoring system was subjected to the performance tests in the October 10,1979, Federal Register.2 These tests, a proposed revision to Appendix B of 40 CFR Part 60, were conducted since the emission data are to be used in building a data base that would eventually support proposed new source performance standards. Future monitoring requirements specified under any new emission regulations would also probably incorporate use of these monitoring requirements. The performance specifications quantify short- and long-term drifts, system hysteresis (calibration error), total system response time, and accuracy of the system relative to the applicable reference method. EPA Reference Methods 3, 6, 7, and 10 are the applicable reference methods for Oz, SOz, NOx, and CO that were used during the relative accuracy tests. The relative accuracy portion of the performance tests was repeated during data collection to check the continued accuracy of the system and to provide additional data calibration. In general, the monitoring Table 1. Gas Analyzers for System Georgetown FBB Continuous Emission Monitoring Species analyzed SOz /VO, CO COZ Oz Monitor type Pulsed fluorescent Chemi- luminescent NDIR* NDIR* Electro- chemical Conditioning principle Filtration/ dilution Filtration/ dilution Filtration/ condensation Filtration/ condensation Filtration/ condensation Instrument range 0-100 ppm 0-100 ppm 0-1000 ppm 0-20% 0-25% Measuring range 0-1OOO ppm 0-1000 ppm 0-10OO ppm 0-20% O-25% "Nondispersive infrared. ------- system performed adequately. Because of potential problems with condensate in the sample line and subsequent intermittent reduction in flow to the instruments from plugged capillaries, daily multiple-span calibration checks were made, as well as daily purging of condensate traps and replacement of secondary particulate filters. Passing the calibration gases through the entire system ensured quality control on data validity. Flue gas particulate concentrations were determined atthestackduringtwo test periods using EPA Reference Method 5. During each test period, three replicates were performed, each set constituting one test run. Each replicate test was conducted by travers- ing through the existing ports on the third stack sampling level, a location conforming to the sampling location criteria specified in EPA Method 5. Twelve traverse points were sampled for 5 minutes each, for a total sampling time of 1 hour replicate test. Coal samples were collected at the spreader stokers for subsequent deter- mination of ultimate and proximate analysis. Limestone samples were collected at the weigh belt feeders to each bed for subsequent analysis of Ca content. In addition to several daily composite samples of each material, hourly coal samples were taken for 7 days, and hourly limestone samples were collected for 2 days. Coal sulfur concentration and heating value were used to compute boiler inlet SOa loading in ng/J. This was used in conjunction with flue gas SOa to compute SOa reduction efficiency. Coal sulfur concentration was also used with limestone calcium concentration to compute Ca/S molar feed ratios for comparison with SO2 emission rates and desulfurization efficiency. A computerized data reduction system was used to process the continuous monitoring test data. Process operating data, other than coal and limestone feed rates, were taken from operator log books and keypunched onto computer cards. All emission rate and coal and limestone feed rate strip chart data were digitized to determine 15-minute increments. These 15-minute averages were used to calculate hourly and 24- hour average emission rates. Results Because the system is in an extended shakedown phase, several FBB operating conditions affecting emission control performance were outside intended boiler design ranges during the con- tinuous monitoring program. These conditions and the associated design values are: • High excess flue 02: 10-12% versus a 5% design. • High limestone feed rates: Ca/S ratios of 5-10 versus a 3 design. • Ineffective fly ash reinjection. • Torn and blinded fabric filter bags. Analyzed either singly or in combina- tion, these factors adversely impacted emission control performance SOa, NOx, and particulates, in terms of emission reduction capability and/or emission control cost effectiveness. SOa Emissions SOa removal efficiency was high, averaging more than 85% (more than 90% about a third of the time). Except for a few excursions on an hourly basis, outlet SOa was usually significantly less than 301 ng/J (0.7 lb/106 Btu), averaging 161 ng/J (0.37 lb/106 Btu},considering all hourly average emission rate results. Although SOa capture efficiency was high and outlet SOa emissions were low, limestone feed rates were high and calcium utilization was low. Limestone feed to the boiler was controlled manually by the operator during the program for two reasons: (1) the level of one of the two beds in the FFB could not be maintained without feeding limestone in excess of that required for emission control (it was later determined that high particle elutriation was occurring in the FBB because of injection of overfire air near the top of the bed); and (2) because of a problem with the feedback signal from the in-bed SOa monitor, the boiler operators adjusted the limestone feed rate manually and on the basis of the flue gas SOa level (this signal was intermittently disrupted by the calibration procedures and occasional monitor maintenance). Also contributing to higher-than-required Ca/S ratios was that the coal sulfur content was found to be 1.5-2.0%, which was lower than expected. This was not apparent until after completion of on-site activities since results of the analysis of coal samples taken during the program were not immediately available. Because limestone is fed on a mass basis relative to coal (about 1:3), Ca/S ratios would be abnormally high even in the absence of atypical conditions already noted. Overall, the Ca/S molar feed ratios were found to be about 5-10, in contrast to the design ratio of 3. Figure 1 shows SO2 removal efficiency as a function of Ca/S molar feed ratio for 48 hourly average values for 3 days during the continuous monitoring program. These data are shown because they represent days and hours for which there are analyses of coal sulfur content and heating value, and limestone 100 95 & £ 90 c i 85 <0 | 80 $ 70 65 • Day 240 hrs 10-20 ° Day 255 hrs 1-24 * Day 256 hr 15 — Day 257 hr 14 Stringent Intermediate Moderate 8 234567 Ca/S Molar Feed Ratio Figure 1. SOz emissions as a function of calcium to sulfur molar feed ratio. 3 10 ------- calcium concentration. This information was used to calculate inlet and outlet SOa, S02 control efficiency, and Ca/S molar feed ratios. The S02 removal efficiencies shown illustrate the expected trend of increased removal efficiencies at higher Ca/S ratios. Note: this effect levels off at Ca/S ratios of about 5-6 and S02 removal efficiencies of about 90- 95%. Comparing the S02 removal efficien- cies to the various emission control levels described in reference 1 shows that the moderate control level of 75% removal was attained for all but one hourly average shown. Also, the inter- mediate and stringent levels were supported consistently for Ca/S ratios greater than 5. Note: even under the adverse reinjection condition and decreased gas phase residence time, a moderate SC<2 emission standard of 75% removal could have been achieved with a much lower limestone consump- tion rate. NOx Emissions NOx emission rates were higher than expected primarily because of high levels of excess 02. Ineffective fly ash reinjection may have also decreased NOx control. During the 30-day CEM program, flue gas 02 was 7-12%, as opposed to a 5% design value. Because of problems with the in-bed 02 monitor, automatic control of combustion airwas not used. Combustion air, controlled manually by the operator, was held high to avoid forming reducing zones in the bed with attendant water-tube corrosion problems. Two other sources of air introduced to the FBB further complicated the control of combustion air. Overfire air could be injected in only two discrete volumetric rates, and air introduced through the eductors used to recirculate flyash was not easily controlled or measured. The lower operating temperatures of 815-870°C used in FBC suppress formation of thermal N0>. It is generally considered that nearly all NOX from FBC is derived from oxidation of fuel nitrogen. It is also postulated that carbon char, volatile fuel nitrogen species, CO, and other reducing species play a role in the chemical reduction of NO to molecular N2. It is therefore not surprising that, at very high excess O2 levels, these reduction mechanisms would be sup- pressed and NO emissions would increase.3 The relationship between NOX emis- sions and excess air, indicated by percent 02 in the flue gas, is shown in Figure 2. Examination of over 400 hours of record when flue gas 02 was less than 12% indicates that a moderate control NOx level of 301 ng/J (0.7 lb/106 Btu) would be supported about half the time. Hourly average emissions less than the stringent level of 215 ng/J (0.5 lb/106 Btu) were recorded for 27 hours when flue gas 02 was 5-9%. However, higher NOx emissions were also recorded at these 02 levels; the maximum was 335 ng/J (0.78 lb/106 Btu) at 9% 02. Although flue gas O2 correlated strongly with N0« emissions, the variation in the data in Figure 2 shows that other variables were also affecting NOx emissions. To examine the relation of NOx emissions to other process variables, a multiple linear regression analysis was conducted. The resultant relationship is expressed by: NOx (ng/J) = 31 O2-202 TR-66 F + 138 where Oa is % flue gas oxygen, TR is the average gas residence time (in seconds) in the bed, and F represents the fly ash reinjection system being on or off; F = 1 is "on" and F=0 is "off." The multiple correlation coefficient "r" for the equation was 0.83. The regression line in Figure 2 shows the relation between NOX and flue gas O2 for an average gas residence time of 0.38 seconds and F=1 (i.e., fly ash reinjection "on"). The standard error of the estimate for this equation is 35.7 ng/J, indicating that a 95% confidence interval for the regression line at average conditions observed during the monitoring program would result in a range for NOK of ± 70 ng/J. As actual conditions (specifically 02 level) varied from the average level observed, this range increased. It should be pointed out that the O2 levels observed indicate an amount of excess air significantly above what is considered good practice for efficient operation of industrial boilers. As some of the problems with in-bed 02 monitor- ing and flyash reinjection are resolved, the Georgetown FBB should run at 1.0 0.9 0.8 0.6 "§ 0.5 0.4 0.3 0.2 o.; '.• « .£» * • Moderate •»•••»• Intermediate .. *»*4* * •• 7* •••*• • Stringent . I *• ., 450 400 350 300 250 200 * 750 7OO 50 Figure 2. 6 7 8 9 10 Flue Gas Oxygen, percent A/0, emissions as a function of flue gas oxygen. 77 12 ------- excess air levels closer to design, and the N0« emissions should drop to a level that is generally considered more representative of FBC. For example, this analysis indicates that, even at an 02 concentration of 9.8% (still high), a moderate control level of 301 ng/J would be supported on the average. Similarly, the intermediate NOX level of 258 ng/J could be achieved at a level of 8.4% Oz, and the stringent NO* level of 215 ng/J at 7.1 %O2. Particulate Emissions During the first fewdays on site, there was noticeable puffing from the stack at regular intervals, indicating leakage in a compartment of the baghouse. To locate leaking bags, a hot slump procedure was undertaken on August 19, 21, and 22 to gain access to the baghouse through the compartment lids. The FBB was shut down for short periods (0.5-2 hours) by shutting off the induced- and forced-draft fans and the coal and limestone feed. Each period, several (5- 10) bags were replaced. This reduced, but never completely eliminated, the puffing. Three EPA Method 5 particulate measurements were conducted on August 23, as part of the performance specification tests. The resulting emis- sions rates were 24.6, 37.8, and 47.0 ng/J (0.0572, 0.0879, and 0.109 lb/106 Btu), or an average of 36.5 ng/J (0.0848 lb/106 Btu). This performance is consistent with the optional inter- mediate particulate control level of 43 ng/J (0.1 lb/106 Btu). The FBB was shut down over the Labor Day weekend to clear clogging of the bed spent solids drain standpipe, remove some refractory lining, and reinspect the baghouse. Three days of downtime allowed the baghouse to cool so that operators could enter and inspect bags and seals at the bottom of the unit. As a result of this inspection, some additional bags were replaced. Overall baghouse performance im- proved after this inspection, shown by the results of Method 5 testing on September 13. The three measurement results were 19.6, 20.9, and 32.3 ng/J (0.0456, 0.0487, and 0.0751 lb/106 Btu), or an average of 24.3 ng/J (0.0565 lb/106 Btu). As with S02 and NOX emissions, atypical operating conditions may have adversely affected particulate control performance; e.g., high excess air, high limestone feed rates, ineffective flyash reinjection, and torn or blinded fabric filters could all be expected to decrease the control efficiency for particulate emissions. Better identification and resolution of these problems should improve particulate emissions control. Summary Continuous emission monitoring at the Georgetown FBB shows that this coal-fired steam generation system can meet stringent S02 and NOX emission control levels, although several atypical operating conditions during the test program hindered continuous achieve- ment of optimal control. SC>2 control performance was adversely affected by bed height maintenance problems, lack of a reliable in-bed S02feedback signal for automatic control of limestone feed, and ineffective fly ash reinjection. NQ2 emissions were generally higher than expected for typical applications in the future due to high excess air operation, shorter-than-design gas residence time, and ineffective fly ash reinjection. Baghouse performance suffered due to torn bags, bag blinding, and inefficient multicyclone performance. Although uncertain, high baghouse inlet loadings may have caused the tearing bags and/or bag blinding. Modification to the FBB (e.g., improving automatic control by using feedback signals from 02 and SOz monitors, rebuilding the reinjection system, and improving the baghouse) is planned and, in some cases, already in progress. These changes should further improve control. A further monitoring program was conducted in February and March 1982. A report of the results of the follow-up program will also be published. References 1. Young, C.W., et al.. Technology Assessment Report for Industrial Boiler Applications: Fluidized-bed Combustion, EPA-600/7-79-178e (NTIS PB 80-178288), November 1979. 2. U.S. Environmental Protection Agen- cy, Federal Register, Vol. 44, No. 197, pp. 58602-58636, Wednesday, October 10, 1979. 3. Beer, J. M., et al., /VOX Emissions from Fluid/zed Coal Combustion. Draft Final Report for EPA Grant No. R804978, Massachusetts Institute of Technology, 1980. Charles W. Young. Edward F. Peduto, Peter H. Anderson, and Paul F. Fennellyare with GCA/Technology Division. Bedford, MA 01730. John O. Milliken is the EPA Project Officer (see below). The complete report, entitled "Continuous Emission Monitoring at the George- town UniversityFluidized-BedBoiler,"(OrderNo. PB83-151837;Cost:$16.00, subject to change) will be available only from: National Technical Information Service 5285 Port Royal Road Springfield. VA 22161 Telephone: 703-487-4650 The EPA Project Officer can be contacted at: Industrial Environmental Research Laboratory U.S. Environmental Protection Agency Research Triangle Park. NC 27711 U. S. GOVERNMENT PRINTING OFFICE: 1983/659-095/1921 ------- Post3ge and United States Center for Environmental Research pees pg^ Environmental Protection Information Environmental Agency Cincinnati OH 45268 Protection Agency EPA 335 Official Business Penalty for Private Use $300 5 S DtAtfBGRix IL 60604 ------- |