United States Environmental Protection Agency Industrial Environmental Researc Laboratory Research Triangle Park NC 277 Research and Development EPA-600/S7-81-143 Dec. 1981 Project Summary Performance Evaluation of an Industrial Spray Dryer for SO2 Control James A. Kezerle, Steve W. Mulligan, Dave-Paul Dayton, and Patricia J. Perry TRW conducted a continuous moni- toring test program at the Amcelle Plant of the Celanese Fibers Company in Cumberland, MD, to evaluate the performance of a dry process flue gas desulfurization system. This system treated flue gas from a coal-fired stoker boiler. Tests involved methods specified by EPA for 30-day compli- ance testing, which requires a mini- mum of 22 days of data containing at least 18 hours of data per day and two data points per hour. Hourly and daily averages of results are presented as well as averages for the entire test period. Operating experience with the spray-dryer/ bag- house system is summarized for a 5- month period ending with the comple- tion of testing on September 3O, 1980. Brief descriptions of the test site, the flue gas cleaning system, and the continuous monitoring system are included. Manual sampling techniques for data verification are described and the systems for data acquisition, data analysis, and quality assurance, pre- pared specifically for this program, are presented. Raw process and emissions data are included in the appendices. Results based on 23 days of data showed the mean SO2 removal effi- ciency to be 70 percent over the compliance test when the sulfur content of the coal averaged 2 percent. In general, efficiency was 60-80 percent, except for periods of system upset. Particle removal efficiency was 99.7 percent. Participate emissions averaged 0.030 g/m3 (0.013gr/dscf) during the 2 days these data were taken. This Project Summary was devel- oped by EPA's Industrial Environmen- tal 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 TRW Inc., under contract to the U.S. EPA, tested the dry S02 control system serving the coal-fired (No. 5) boiler at the Amcelle Plant of the Celanese Fibers Company in Cumberland, MD. Celanese ordered the flue gas cleaning system in January 1979. Construction of the system by Rockwell International and Wheelabrator-Frye was completed in October 1979. Boiler installation was not completed, however, until mid- December 1979. Acceptance testing of the FGC system was completed on February 21, 1980. TRW began collecting data for the demonstration test phase in May 1980. Installation and certification of instru- mentation at the site for the performance testing were performed according to provisions for S02 compliance testing and began in late April 1980. The objective of the program was to collect 30 days of continuous monitoring data, representing proper operation of the flue gas cleaning system, using compli- ance test methods. Problems with the ------- boiler, the FGC system, and the con- tinuous monitoring system delayed completion of this test phase until September 30, 1980. System Description Celanese Fibers Company installed the coal-fired boiler in 1979 to supple- ment the existing oil- and gas-fired boilers at their Amcelle Plant. This installation was undertaken to improve the economics of supplying process steam for the production of synthetic fiber. A spray dryer and fabric filter combination was chosen to provide flue gas desulfurization (FGD) on the bases of cost, the lack of available space for ponding wastes from a wet FGD scrubber, and the need to provide good control of particulate emissions. The flue gas treatment system was purchased as a turnkey installation from Rockwell International and Wheelabrator- Frye, Inc. A flow diagram of the system is presented in Figure 1. Coal-Fired Boiler The coal-fired water-tube boiler at the Amcelle Plant is identified as the plant's No. 5 boiler. The boiler is an Erie City spreader-stoker with a traveling grate for continuous ash discharge. This boiler had previously been retired from service at a Celanese plant in Rome, GA. The boiler was retubed when it was reconstructed at the Cumberland, MD, plant. Table 1 specifies design data for this boiler. The coal-fired boiler is rated at 156 million kJ/hr (148 million Btu/hr) with secondary boiler fuels of gas or No. 6 fuel oil. At the boiler's maximum rating of 68,000 kg steam/hr (150,000 Ib/hr) when fired by a combination of coal and oil or gas, the flue gas to be treated by the dry FGD system is 41.4 mVs (87,000 acfm) at 216°C (420°F). At the boiler's nominal coal-fired rating of 49,900 kg steam/hr (110,000 Ib/hr), the flue gas to be treated is 30.7 mVs (65,000 acfm) at 193°C (380°F). Table 1. Boiler Data - Amcelle Plant Boiler No. 5 Erie City Spreader-Stoker Coal and Natural Gas Boiler Type Fuel Type Fuel Heating Value Sulfur Content Ash Content Coal Bituminous 29,056 kJ/kg (12,500 Btu/lb) 1.0 to 2.0 percent 8.0 to 20.0 percent Gas Natural Gas 37.2 MJ/m3 (1,000 Btu/ft3) 0.0 percent 0.0 percent Lime Fill (From Trucks)' Lime\ Storage Silo Be/t Feeder I \Slaker Lime '—|-1J SystemaaVatGi'it Screen Lime Slurry Opacity &S02 lnstruments\ Coal Hopper & Storage t!Z7 AAJ Coal Unloading Conveyor Silo Feedbucket Conveyor Silo Reclaim Conve yor/Ele vat or Ash Unloading (To Trucks') Figure 1. Celanese boiler and flue gas cleaning system. 2 ------- Analyses of randomly selected coal samples are presented in Table 2. The sulfur content of the coals received during the test period was 1.25-2.76 percent, with a mean of 2.02 percent (dry basis). Table 3 illustrates flue gas design conditions for various coal firings. Spray Dryer The gas cleaning system is designed to provide FGD removals of from 70 percent (for 1 percent sulfur coals) to 87 percent (for 2 percent sulfur coals) from half to full boiler load. Most of this SO2 removal takes place in the spray dryer where the S02-laden flue gas is passed through a finely dispersed fog of lime slurry and water. The spray dryer consists of a single, 6.1 -m (20-ft) diameter vessel containing a rotary atomizer (Figure 2). This rotary atomizer (Bowen wheel) is driven at approximately 16,000 rpm. The lime slurry is fed to the wheel at a liquid-to- gas ratio of 0.04 l/m3 (0.3 gal./1000 acf), where it is centrifugally dispersed into the gas stream. A swirling motion is imparted to the flue gas as it enters the top of the spray dryer through a fixed- vane rotary ring to increase turbulent mixing of the flue gas and the lime \ slurry. Approximately 20 percent of the flue gas bypasses the spray dryer, thus providing reheat to raise the gas temperature prior to its entry into the fabric filter. This is necessary for dry operation and compensates for the temperature drop in the fabric filter. The amount of water fed to the spray dryer is automatically adjusted to hold the gas temperature from the spray dryer at a set value. Lime System The lime system is depicted in Figure 3. The dry storage silo stores about a 10- day lime supply. High-calcium pebble quicklime is gravity fed into the lime slaker where it is mixed with water to provide a 20 to 30 percent (by weight) slurry. The lime system can provide 125 percent of required capacity when the boiler is fired at its maximum rate with 2 percent sulfur coal and overtired gas or oil. The lime system pumps and piping can be automatically flushed with water to prevent deposits. I I Spray } X Atomizer•£*' Figure 2. Spray dryer. Table 2. Selected Coal Analyses Vol* Ash Sample No. % % 1 (8-22-80) 2(8-29-80) 3(9-12-80) 4 (9-23-80) 19.9 31.6 17.2 33.14 14.95 18.96 13.97 16.81 Sulfur 1.58 1.92 1.36 2.24 HHV kJ/kg Btu/lb 29.860 27,998 30,153 29,411 12.846 12.045 12.972 12,653 * Volatile matter. Table 3. Flue Gas Characteristics - Amcelle Plant Boiler No. 5 Fuel Steam Production Flue Gas Temperature Flue Gas Flow Rate S02 Concentration SO2 Exhaust Rate Paniculate Loading Coal 49.900 kg/hr (110.000 Ib/hr) 193°C (380°F) 30.7 m3/s (65,000 acfm) 800 to 2,500 ppm 113 to 363 kg/hr (250 to 800 Ib/hr) 8.5 to 11.9 g/m3 (3.7 to 5.2 gr/dscf) Lime Slaker Water Slurry Tank Figure 3. Lime system. Fabric Filter The fabric filter, a four-compartment pulse-jet baghouse manufactured by Wheelabrator-Frye, Inc., is shown in Figure 4. Each compartment contains 225 bags. The baghouse can operate with three compartments on-line when the boiler is operating at its nominal coal-firing rate to produce 49,896 kg/hr (110,000 Ib/hr) of steam. The air-to-cloth ratio is 2.2 - 6.8 with a design pressure drop of 500 Pa (2.0 in. H20). The filter medium is a fiberglass- reinforced felt manufactured by Huyck. Description of Continuous Monitoring System Instrumentation The continuous monitoring system used in the performance evaluation consisted of four major groups: the filter probes and process sample lines, the gaseous analyzers, the data acquisition and recording system, and the remote temperature sensing system. Upon arrival at the test site, these four separate systems were assembled and aligned into one comprehensive system. Sampling System Sampling locations are shown in Figure 5. The inlet sample point was in the rectangular duct between the boiler and the spray dryer. The outlet sample point was in the circular cross-section of the stack. The intermediate sample point was not monitored. ------- To Stack From Spray Dryer Spent Sorbent and Fly Ash Figure 4. Fabric filter. The filter probe assemblies utilized stainless steel filters (20-/um mesh) attached to 76-cm (30-in.) long stainless steel tubes with an o.d. of 1.3 cm (0.5 in.) and an i.d. of 0.6 cm (0.2 in.). Connected to this tubing was about 30 m (100 ft) of electrically heat-traced sample line, constructed of 0.6-cm (0.2- in.) Teflon tubing. The sample lines were kept at 121°C (250°F) to prevent con- densation from the gas sample. The gas samples collected at the inlet to the spray dryer and at the outlet of the baghouse were dried with a gas condi- tioner. This gas conditioner contained two dual stainless steel condenser traps suspended in a medium of ethylene glycol cooled by a Hanke refrigeration unit with copper cooling coils to approx- imately 3°C (37°F). The sample was pulled through the condenser traps by two Teflon and stainless steel pumps and then delivered to the analyzers. The gas conditioner system was connected to a timer that allowed the condenser traps and heat-traced sample lines to go into the 700 kPa (100 psi) blowback Legend W v Contaminanted Air Flow Flue Gas - SOz and Fly Ash Scrubbing Solution Clean Air Flow Outlet Sample Point Inlet Sample Point Combustion Air '{Spent Dry Salts? Outside Air from Forced Draft Fan Air Preheater Absorbent A/kali Solution Scrubbing Tank Intermediate Sample Point Dry Product Disposal Induced Draft Fan Figure 5. Two-stage dry FGD system with TRW sampling positions indicated. 4 ------- mode for 3 minutes of every hour. A schematic of this sampling system (Figure 6) shows the path of sample gas from the probe to the analyzers and the path of output data from the analyzers to the data logger. Flue Gas Analyzers Flue gas was analysed using a Thermo-Electron Pulsed Fluorescent S02 Analyzer, Model 40, andaBeckman Paramagnetic 02 Analyzer, Model 755. These analyses were conducted con- tinuously for both the inlet of the spray dryer and the outlet of the baghouse. The inlet S02 analyzer operated on a 0- 5000 ppm full-scale range with a 1-V full-scale output, while the inlet O2 analyzer operated on a 0-25 percent of total gas volume full-scale range with a 10-mV full-scale output. The outlet S02 analyzer operated on a 0-500 or 0-1000 ppm full-scale range, depending on the concentration of S02 in the flue gas at the location of the outlet sample probe. The outlet Oz analyzer operated on the same range as the inlet O2 analyzer, but with a 1-V full-scale output. The SO2 and O2 analyzers were certified according to procedures out- lined in "Performance Specifications 2 and 3 for Continuous Monitors in Stationary Sources" as specified by EPA (44 Federal Register 58602, 1979). Relative accuracy was determined for the S02 analyzers using the average response times for the analyzers ob- tained during response time tests. These determinations ensured close agreement between the results from the S02 analyzers and EPA reference methods (specifically. Reference Method 6 for S02). Calibration errors were also deter- mined for each of the four analyzers. Directly after the daily calibration of the instruments, zero, mid-level, and high- level calibration gases were randomly introduced into the respective analyzer until a set of five points for each concentration (zero, mid-level, and high-level) was obtained. Both of the S02 analyzers and both of the O2 analyzers passed all of the certification requirements. In addition, all calibration gases used for instrument certification or instrument calibration were either traceable to National Bureau of Standards reference gases or underwent the calibration gas certifica- tion. The latter gases were obtained from the EPA repository and were certified by EPA personnel prior to use in the tests at Celanese. The S02 and 02 analyzers were calibrated daily between the hours of 0800 and 1200. Data Acquisition System The Data Acquisition System con- sisted of a Fluke Data Acquisition system, a dual-pen Fisher 5000 Record- all recorder, and a Leeds and Northrop six-channel multipoint recorder. Hi-Range Mid-Range Exhaust *Data Acquisition System. Figure 6. Flue gas sampling and analysis system. ------- Temperature Sensors Two 60-cm (24-in.) long chromel- alumel thermocouples were mounted parallel to the filter probes. These thermocouples measured flue gas temperatures at the inlet and outlet of the FGD system. They were hard wired into the Fluke Data Acquisition system, using about 30 m (100 ft) of chromel- alumel thermocouple wire. Results Data on S02 removal that were typical of the fully operating dry FGD system performance and whose relative accu- racy was fully documented were col- lected only during the final month of the tests. This period of "good" data collec- tion ran from August 28 through September 30, 1980. The boiler gen- erally ran at a steady load (about half of the rated value because of seasonal reduction in steam requirements) throughout most of this period, and the FGD system operated almost contin- uously. Hourly averages of S02 emissions in parts per million were calculated from a minimum of two data points per hour. These hourly averages were then corrected to zero percent oxygen dilution and converted to pounds per million Btu. Calculations of S02 removal efficiency were then based on these average hourly S02 emission values. S02 emissions in pounds per million Btu were determined using the F-factor technique. An F-factor for dry flue gas from coal of 9820 dscf/106 Btu (263.9 mVGJ) was used. Heat input to the boiler was calculated from available data. Hourly averages of steam flow were used to derive hourly values of coal feed rate from values of total daily coal consumption. Typical data for inlet and outlet S02 concentrations are shown in Figures 7 and 8. These data are representative of the 23 days when continuous monitoring methods met EPA's compliance criteria, including collection of data for over 18 hours per day with the FGD system treating boiler flue gas. Figure 7 shows that outlet S02 concentration, mea- sured in the stack, closely follows the SO2 concentration at the inlet to the spray dryer. This curve indicates no corrective action being taken to adjust slurry flow rate for varying inlet S02 concentration. With the FGD system in automatic control, the outlet SOa concentration (see Figure 8) was rela- tively constant, indicating that the slurry flow was adjusted to accommodate |2000-| s c | /eooH I 7200H O O oo 800- a 4oo-| Inlet* Inlet Figure 7. S' o -7600H i. c 1/200- o O 800- oo •O ~ 400- 0200 0400 0600 0800 1000 1200 1400 1600 1800. 2000 22002400 Time of Day Average hourly SO2 concentrations for September 3, 1980, with FGD system controlled manually. o u 5 Inlet Figure 8. 0200 0400 0600 0800 WOO 1200 1400 1600 1800 2000 2200 2400 Time of Day Average hourly SO2 concentrations for September 8, 1980, with dry FGD system controlled automatically. even rapid changes in inlet SO2concen- tration. Although the FGD system was de- signed to operate automatically, this was not always possible because of malfunctions in the stack S02 monitor which provided feedback to the spray dryer control system. Problems with this monitor necessitated extended periods of manual operation. Under manual operation, the slurry flow sometimes became so high that the outlet concen- trations were 50 ppm or less. On these occasions, SO? removal efficiencies exceeded 90 percent. The average daily S02 removal efficiencies for the continuous moni- toring period cited earlier are given in Figure 9. Except for periods of system upset, removal efficiency was 60-80 percent. The only prolonged period of low S02 removal occurred between September 3 and 6 and stemmed from inability to maintain steady boiler load and slurry pumping problems. The mean S02 removal efficiency for the 23 days of performance data was 70 per- cent, and the standard deviation from this mean was ±9 percent. However, over the last week of the tests, the ------- u .OJ S o 0) et M O oo Large Load Fluctuations Slurry Pumps Inoperable 8/28 Plugged Spray Dryer Repaired Baghouse Bags Changed I I I I I I I I I I I I I I I I I I I rl I I I I I I 8/31 9/5 9/10 9/15 9/17 9/24 9/30 Date 5/77 9/24 9/25 Figure 9. Average daily SO2 removal efficiency for dry FGD system. average daily SOa removal efficiency was 78.5 percent, based on 23 hours of hourly averaged data for each day. The emission rate and removal efficiency of particulate matter for this FGD system were determined by three isokinetic sampling runs on June 2 and 13, 1980. Testing for particulate matter was conducted according to EPA Method 5 using a RAC Stacksampler sampling train. Results of these tests are sum- marized in Table 4. System Availability and Operating Experience Table 5 summarizes the availability of the boiler and FGD system during the tests. The boiler went down for refractory repairs on April 20. From then through the end of the program on September 30, the boiler was off-line approxi- mately 12 percent of the time. It was on- line but running abnormally an addi- tional 5 percent. Thus, boiler problems prevented representative characteriza- tion of the FGD system for about 17 percent of the time TRW was on site. This amounted to 672 of 3.912 hours in the period. The FGD system was off-line (not operating at all) about 23 percent of the time. The FGD system operated ab- normally an additional 12 percent of the time. During this time the slurry feed rates were so low or unsteady that monitoring of any significant S02 scrubbing was prevented. Thus, the Table 4. Particulate Emissions Results Date Run No. 6-2-80 BI5-1 6-2-80 BI5-2 6-3-80 BI5-3 Average Inlet Concentration g/m3 (gr/dscf) Run No. Outlet Concentration g/m3 (gr/dscf) Particle Removal Efficiency. % 9.43 (4.12) B05-1 0.0334 (0.0146) 99.65 8.44 (3.69) B05-2 0.0159 (0.00697) 99.81 11.78 (5.15) B05-3 0.0400 (0.0175) 99.66 Table 5. System Availability 9.88 (4.32) 0.0298 (0.0130) 99.70 Availability*, % Component Boiler FGD System Spray Dryer Lime Feed System Fabric Filter Apr-Sep 82.2 62.4 81.8 83.2 99.8 Aug-Sep (720 hr) 93.3 73.2 97.8 76.5 98.9 Sep 25-30 (144 hr) 100.0 96.2 100.0 96.2 100.0 *The percentage of time in the period that the component operated normally. FGD system was unavailable 35 percent of the time, or a total of 1,354 hours. Availability of the system was signifi- cantly improved in September when most of the continuous monitoring data were collected. During this period the FGD system was off-line less than 19 percent of the time and operated abnormally an additional 8 percent of the time, giving an availability of 73 percent. Operating problems and their effects on the program were broken down into four system components: the boiler, the lime feed system, the spray dryer, and the fabric filter. Problems with each ------- component impact the entire FGD system. Steam Boiler A problem which affected the per- formance of the FGD system was the variability of coal quality. Coal sulfur content varied widely throughout the early part of the program. The quality became less variable near the end of the program, but proximate analyses of daily coal deliveries showed sulfur contents of 1.25 - 2.76 percent. When operating in automatic control to keep the outlet S02 concentration at a set value, the FGD system responded to rapid changes in inlet SC>2 concentra- tion so that hourly averages of emissions remained constant. In manual control, the outlet S02 concentration followed the inlet concentration in the absence of operator adjustment. With uniform coal quality and automatic control of slurry flow, large fluctuations in inlet S02 concentrations were absent and a steady outlet S02 concentration was maintained. Another problem which relates to coal supply involves the amount of fines in the coal. Coal fines, when suddenly dumped into the furnace, cause rapid changes in boiler load and flue gas flow, changes in SC>2 emissions, and increased particulate matter and opacity levels in the stack. Fast changes in flue gas flow and SOz concentration made it difficult for the spray dryer to keep SOa emissions at a desired level. Such large and rapid load fluctuations occurred on September 5, 6, and 7. Data collected on these days were not included in overall averages. Lime Feed System The slurry is fed to the atomizer by progressing cavity pumps; under design conditions, one pump is in use and one is a spare. However, to cope with the higher sulfur coals encountered, the single pump had to be operated at high speeds and this led to rapid pump wear. To alleviate this, the system was modified to use both pumps in parallel. This was the normal mode of operation throughout the latter part of the test period. Most other problems with the lime feed system related to plugging some- where in the system because of grit in the lime. Although grit was supposed to have been removed by screens inside the slaker, damaging quantities of it passed through or bypassed the screens into the rest of the system. Failure to remove grit caused excessive wear in the pumps and plugging in the slaker, in the flow lines, in the slurry pump, and in valves. Dual-element screen filters were eventually installed in the feed system, but not enough time elapsed before the end of the program to assess whether they solved the problem. Spray Dryer Maldistribution of lime slurry in the atomizer resulted in the wetting of the dryer wall and discharge of damp material from the dryer. It was corrected by redesign. Other problems encountered with the spray dryer also related to the atomizer. The rotary atomizer was subject to clogging with grit particles if they were not screened sufficiently from the slurry. The spray dryer was shut down for cleaning when this clogging occurred. Another problem was failure of the bearings supporting the shaft of the atomizer wheel caused by an imbalance due to grit plugging the atomizer wheel. Fabric Filter The most serious problem with the baghouse was the unexpectedly high pressure drop through the fabric filter. This was apparently caused by moisture on the bags which occurred during an upset and combined with ash and lime to form a coating that increased the resistance to flow. To lower the pressure drop through the baghouse, design and process changes were made, including increasing the pulse-jet air volume by approximately 15 percent. Tests since this modification indicate that this has solved the problem. Conclusions and Recommendations The two-stage dry FGD system in- stalled at the Celanese Fibers Company's Amcelle Plant required 28 - 46 hours of maintenance each week and close operating supervision for continuous operation. Some of this maintenance was performed while the system was operating so there was no interruption in SOa removal. The average S02 removal efficiency demonstrated over a 30-day period, based on 23 days of acceptable data, was 70 percent. This level of performance was achieved while burning coal with an average sulfur content of about 2.0 percent on a dry basis. System guarantees called for 70 percent SOa removal for 1 percent sulfur coal and 87 percent SO2 removal for 2 percent sulfur coal. Compared with these goals, the demonstrated S02 removal was low. Over the last 6 days of A the tests, after several operational \ difficulties had been resolved, SO2 removal efficiency averaged 78.5 per- cent, a marked improvement over earlier results but still below the stated goal with 2 percent sulfur coal. Boiler operators operate the FGD system along with their other duties. Modifications made to the system after operating experience had been gained have the potential to make this a more reliable system. As described above, most of the'operating problems relate to plugging caused by grit in the slurry and water vapor condensing in the flue gas due to low operating temperatures. Both of these problems can be solved by changes in operation and design. Main- tenance needs will also be reduced by these modifications. Because of problems experienced thus far, redundancy of critical compo- nents is recommended. Specifically, three slurry pumps are needed with two on-line at all times and one as a spare. A spare atomizer will limit spray dryer shutdowns due to atomizer failure. Filters should be set up to provide uninterrupted slurry flow to the spray dryer while one filter element is being replaced or cleaned. A means of keeping the outlet SOa monitor operating con- tinuously is needed. Since feedbac from the outlet S02 monitor is used i controlling lime slurry flow to the spray dryer, this will permit a steadier outlet SOa level and more consistent FGD system performance via operation in automatic control. 8 ------- J Kezerle. S. Mulligan, D. Dayton, and P. Perry are with TRW, Inc., P.O. Box 13000, Research Triangle Park, NC 27709. Theodore G. Brna is the EPA Project Officer (see below). The complete report, entitled "Performance Evaluation of an Industrial Spray Dryer for SCb Control," (Order No. PB 82-110 701; Cost: $21.00, 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 •ft U.S. GOVERNMENT PRINTING OFFI CE :1 981 --559-092/3356 ------- (D f q = Is D : 3 fi 0) - -'« O: X O> 00 C! m ------- |