Executive Summary Retrofit of Waste-To-energy Facilities Equipped with Electrostatic Precipitators

Transcription

1 Executive Summary Retrofit of Waste-To-energy Facilities Equipped with Electrostatic Precipitators INTRODUCTION The promulgation of municipal waste combustor [MWC] emissions guidelines in December, 1995 can seriously affect many communities. These communities are served by MWCs that were built before dry scrubbers and ultra-high efficient electrostatic precipitators [ESPs] or fabric filters became the de facto design standard in the mid- 1980s. In all. there are 135 ESP-equipped MWCs in 54 facilities. These units process more than 43,000 tons per day [TPD] of municipal solid waste [MSW] and have combined energy sales of 5,200,000 Ib/h of cogenerated and direct steam, and 640 MWe of electricity. While many of these units are large, a sizable number are small, e.g., capable of burning less than 250 TPD. Representing 5,300 TPD of disposal capacity, 860,000 Ib/h of steam sales and 70 MWe of electrical generating capacity, these small units employ over 700 people. It is evident that there is a need to find a way for these existing ESP-equipped facilities to economically meet whatever guidelines are ultimately promulgated for small units. Recognizing this need, the Department of Energy's [DOE] National Renewable Energy Laboratory [NRELI supported an American Society of Mechanical Engineers' [AS1\I1E] project designed to address the concerns. The ASME Center for Research and Development [CRTD] under the direction of the ESP Retrofit Subcommittee of the ASME Research Committee on Industrial and Municipal Waste [RCIMW]. Proof-ofconcept demonstration testing was conducted at two ESP-equipped facilities. This work demonstrated potentially economic retrofit technology which should help to keep the existing facilities operating, thereby contributing to our nation's energy future, and preserving the capital invested in the infrastructure. This program demonstrated the ability of a existing ESP-equipped MWCs using a combination flue gas temperature control, dry acid gas reagent injection and powdered activated carbon [P AC] addition to meet all the emissions guidelines for small «250 TPD) MWCs. The results show that large plant emissions limitations for acid gases probably were not met using this technique alone. Particulate, lead and cadmium emissions were unaffected and additional modifications may be needed depending on a plant's current performance. OBJECTIVE The objective of the proof-of concept demonstration testing was to determine the actual emissions level achieved by a combination of ESP inlet temperature control, acid gas reagent injection and P AC addition. During the two series of testing campaigns, ESP inlet temperature was controlled by latent heat reduction through evaporation of water injected into the gas stream and through sensible heat reduction using increased heat exchanger capacity.

2 The main purpose of the demonstration tests was to show achievement of the December 19, CFR 60 l Subpart Cb Emissions Guidelines (Table 1) for PCDD/F, mercury, HCl and SO2 for small <250 TPD) ESPequipped MWC facilities. A secondary objective was to determine if large ESP-equipped MWC facilities emissions guidelines could be met for particulates, PCDD/F, sulfur dioxide, hydrogen chloride and mercury using the same technique. Table emissions guidelines for pollutants potentially affected by dry sorbent injection. APPROACH To meet the objectives of the program, ESP-equipped MWCs smaller than 250 TPD were targeted. The available time and budget limited the field to those that already had extra heat exchanger capacity. To be generally useful, the duct runs between the last heat exchanger and the ESP had to be short. Facilities with longer duct runs should achieve better performance due to greater gas residence time between the incinerator outlet and ESP outlet. The program was divided into two parts: the first involved testing of latent heat reduction techniques; the second, the sensible heat reduction approach. In both cases, an acid gas sorbent and powdered activated carbon [P AC] were dry as injected into the gas stream before the ESP. The Facilities Phase I testing, the latent heat of evaporation temperature control procedure, was conducted at the Davis County Energy Recovery Facility [DCERF], Layton, Utah in November The DCERF is a nominal 420 TPD MWC with two furnaces. Steam generated by the facility is used in a back-pressure turbine to first generate electricity and

3 is then exported to the neighboring Hill Air Force Base district heating system. The units are refractory wall Seghers (combination rocking and sliding grate) furnaces coupled to Zurn waterwa1l, waste heat recovery boilers rated to raise 51,3441blh of 500 psig, 500 F steam. The DCERF was built with a powdered limestone furnace injection for acid gas control. This reagent system has been converted to inject Trona (a natural sodium reagent-sodium sesquicarbonate) between the boiler outlet and the economizer for acid gas control. A rapid dispersion lance system was developed that achieves virtually full duct coverage with Trona within a few feet of the injection point. Particulate emissions are controlled by a three field Environmental Elements ESP with a specific collector area of about 400 ft. 2 /1000 acfm. Energy Answers Corporation's resource recovery facility in Pittsfield, MA [EAC/Pittsfield] hosted the second phase testing conducted in December The facility has three refractory wall, excess air furnaces, each rated at 120 tons/day. The furnaces are manifolded to two heat recovery and gas cleaning trains consisting of a,vaste heat boiler followed by boiling and trim economizers, an electrostatic precipitator, a condensing economizer and. finally, a sodium based packed bed wet scrubber. The scrubbers exhaust to a common breaching which discharges to the atmosphere through a single flue stack. Each waste heat boiler is rated to generate 30,000 Ib/h of 250 psig, 515 F steam. The ESP is a four-field unit with a specific collector area of 490 tt2/1000 acfm designed to achieve particulate levels of % CO2. The flue gas temperature entering the ESP is between 300 and 350Of. Provisions were made at both sites to inject P AC into the flue gas stream. At DCERF, P AC,vas mechanically metered into the existing, pneumatic Trona injection system. Dry reagent is injected at the economizer outlet. At EAC/Pittsfield, two portable dry reagent dosing systems were used to drop powdered hydrated lime and P AC into a common eductor. The dry reagent was injected into the flue gas stream between the boiling and trim economizers on one APC train. Two rapid dispersion lances were used to achieve complete gas stream coverage in the rectangular ducts. At DCERF, the stack had separate flues for each unit exhaust. The flue exhausting the tested unit gas strean1 was sampled. At EAC/Pittsfield, flue gas sampling ports were located between the ESP outlet and condensing heat exchanger inlet. Limited sampling was also conducted in the stack to establish the relationship between ESP and scrubber outlet concentrations under normal operating (baseline) conditions. In both cases, facility compliance monitors were used to monitor stack emissions. Supplemental CEMS were used to monitor S~ and CO at the ESP outlet. Plant operating data, including furnace operating data, were recorded and monitored throughout the entire testing schedule at both facilities. Testing followed a randomized plan designed to illicit the effect of temperature and P AC and acid gas reagent injection rates on emissions. Sampling protocols that met EP A Level IV Quality Assurance requirements were followed.

4 Temperature Control There is considerable experimental evidence that once good combustion has been achieved. the predominant source of PCDD/F is formation downstream of the active combustion zone. Formation may take place in the gas phase, on surfaces, or within the bulk of particulate. Regardless of the actual mechanism, laboratory data indicates that temperature plays a significant role in the reaction. Below 250 C (482 F), the formation reaction is slow; above 400 C (752 F), the destruction reaction overwhelms formation; in between, the two reactions compete. Figure I shows the relationship exhibited.between PCDD/F concentration and air pollution control system [APCS] outlet temperature at 75 MWCs. Rather than plot each data point, the number of points in each sector of the graph are displayed as rays emanating from the data centroid. While the data are noisy, indicating that phenomena other than temperature are also important, the effect of temperature on PCDD/F concentration is clear. Emissions reductions are likely when temperature is reduced. While water injection has been used to reduce flue gas temperatures at ESPequipped facilities, it bas not always proven to be reliable. It appears that physical restrictions on the water injection location which limits the time available for the water to vaporize causes performance difficulties. Closely coupled ESPs limit the time for the water to evaporate.

5 Figure 1 Starburst diagram showing the relationship between dioxin concentration and inverse stack temperature. To overcome these known problems, an atomization system that economically and reliably delivers a very fine water spray is required. Based upon the gas retention time and the system geometry, a dual-fluid (air or steam assisted) spray with a 25 micron Sauter mean diameter droplet was selected for this effort. This size droplet theoretically evaporates within 0.4 seconds without wall or economizer tube impingement. To help achieve such fine atomization, the water's viscosity can be reduced by heating to 175 F which produces a finer droplet than cooler water for the same atomizer design. The system employed at DCERF included a spray lance built around a 3" Schedule 40 steel air supply pipe with a ½ Schedule 40 water supply pipe. The air supply pipe diameter was selected so that no branch connection to the atomizer nozzle would exceed 10 percent of the air supply pipe diameter. Three Bete Fog Nozzles, Inc. nozzles were used. The outer two nozzles produced a 60 spray angle (BETE SA308) and the center nozzle a 90 spray angle (BETE SA310). The three spray atomizers required 150 scfm of 100 psig air supplied through a spiral wrapped 150 psig Barracuda hose rated at 450 F. A digital vortex flow meter was installed in the y%" water supply line to monitor flow and provide an alarm signal if a nozzle head is lost and too much flow occurs.

6 An alternative approach to reducing the temperature is to use heat exchanger sensible heat cooling. Using extra heat exchangers to reduce flue gas temperatures produces an increase in salable energy. Should a facility be able to profitably sell more steam or electricity, the costs might be partially or totally defrayed by increased revenue. Heat exchangers also avoid the potential problems of wall wetting and particle agglomeration. Sorbent Addition Acid Gas Reagent The choice of a suitable acid gas reagent is normally governed by performance requirements and local cost. It is important to recognize that the amount of active sorbtive material is different in various reagents. For instance, if the stoichiometric amount of lime required. for the reaction is 76 Ib/h, either 110 Ib/h of sodium bicarbonate or 155 lb/h of Trona will be needed Assuming a sodium based reagent is desired, the delivered cost for sodium bicarbonate must be no more than 40 percent more than the cost of Trona for economic parity with equal technical performance. Another factor that might enter into the choice of a reagent is its effectiveness. Earlier tests at DCERF indicated the Trona removed more total acid gases than lime, but on individual species, the results were mixed. Trona removed more HCI than hydrated 1 lime did, but removed less S0 2. Simply adding more reagent may not be a workable solution at many facilities because the ESP may not be able to handle the additional load. At DCERF, a fixed rate of Trona (150 lb/h) was added to the flue gas stream. It was injected before the economizer so it could be calcined (drive off the C02) to increase its effectiveness. Nominal operation at DCERF calls for a stoichiometric ratio of 0.8:) with an anticipated reduction in excess of 50 percent with- out overloading the ESP. At EAC/Pittsfield, acid gases are normally controlled by the wet scrubber located after the ESP. During the demonstration test program, lime was introduced into the flue gas stream ahead of the ESP. A target stoichiometric ratio of 2: 1 was chosen as the starting point for lime addition with testing also conducted at 2.7:1 and 3:1. Test Procedures The sampling methods were the same at both sites: Method 23 for PCDD/F; Method 29 for Particulate and Trace Metals utilizing ICAP analysis for Cd and Pb and CY AA for Hg determinations; Method 26 for Hydrogen Chloride with determinations of both HCI and C1 2 ; and Continuous emissions monitoring using: * Method 6C for Sulfur dioxide; * Method 7E for Oxides of Nitrogen; 1 Laboratory testing reported in the Phase I suggests that sodium bicarbonate and Trona are equally effective at removing HCl.

7 * Method 10 for Carbon Monoxide; and * Method 3A for Oxygen. Methods 23 and 29 were run simultaneously on opposite traverses at DCERF and in parallel using a quad- (4-sampler) train at EAC/Pittsfield Test run sampling time at DCERF was two hours. This duration was adequate to measure total PCDD/F concentrations well below the 60 ng/dsm 7% O 2 guideline limit for large ESP equipped MWCs and to detect Section 129 metals. Two hour samples also allowed two replicate measurements for a test condition to be completed in a single day. At EAC/Pittsfield, a 2.4 hour sample time was used and an average sample volume of 3.7 dsm3 was collected. This volume is more than double the minimum sample volume required by Method 29. In both cases, separate l-hour Method 26 (midget impinger) samples were obtained during the main sampling campaign. At EAC/Pittsfield, four 1.7 dsm3 quad-train HCl Method 26 samples were taken on the last day using Greenburg Smith impingers. Plant operating conditions were monitored and recorded continuously using the process instrumentation and the facility's data historians. Plant operating data were collected prior to testing, throughout testing, and following testing. These data were analyzed to verify that normal operations were achieved during each test run. Data Handling and Analysis The plant operating data, including the APCS parameters such as acid gas reagent flow and P AC dosing rate, were entered into a master data spreadsheet along with the test run information and the laboratroy analysis results. Summary statistics for each test were generated within the spreadsheet and commercial statistical analysis packages (SPSS for Windows.) were used for detailed analysis and exploration of the data. Robust regression based analysis of variance [ANOVA] techniques and theoretical performance models were used to determine if there were changes in emission characteristics as a result of designed differences t in temperature or reagent feed rates. FINDINGS AND CONCLUSIONS DCERF-Evaporative Temperature Control Field Observations. The water spray lance performed as expected during testing. Water flow rates and flue gas temperature drops corresponded to the anticipated values. Some fouling had occurred on the lance by the end of testing and this was suspected as being the case of reduced performance towards the end of the program. The water sprays are also thought to have contributed to particle agglomeration and settling. This lead to a partial blockage of the duct and ways need to be found to minimize the potential for such action. ESP performance improved with water addition. This improvement was reflected in a reduced spark rate in the ESP during water injection coupled with an increased potential difference (applied voltage). Opacity was reduced.

8 Performance Results. The effect of temperature reductions on PCDD/F concentrations was not very significant between the 420 F normal operating temperature and the 300 F lowest tested temperature. The HCl and S0 2 removal benefited from reduced temperature. Overall, the results indicate that the facility can meet the 1995 small MWC performance requirements in the 40 CFR 60 Subpart Cb for all pollutant categories and would likely meet the large facility guidelines for all but the acid gases. The PAC data all indicate that lb/h (300 mg/dsm 7% O 2 ) provides an ample margin of safety on the large plant emissions limits - 30 ng/dsm 7% O 2 for PCDD/Fs 1 and 80 µg/dsm 7% O 2 for mercury. Interestingly, PAC addition at the highest tested temperatures (450 F) was sufficient to achieve regulatory compliance. To make sure that there is plenty of room, however, some temperature reduction is indicated. There is little apparent value in dropping the APCS temperature below 350 F. The acid gas results for Trona injection only incorporate the variability induced by changes in stack gas I flow and nominal Trona feed ratio fluctuations because the experiment did not call for adjusting this feed. rate. The results indicate that feed rates of around 1,700 mg/dsm 7% O 2 (125 lb/h) should keep the HCl and S0 2 removal rates safely above the 50% removal level required for small MWC facilities. Injecting reagents increased particulate emissions slightly compared to zero addition TSP measurements. With temperature control, outlet particulate concentrations returned to no-injection levels. Cadmium and lead concentrations remained unchanged and below both the 1995 small and large plant guidelines regard- less of the planned interventions. EAC/Pittsfield Sensible Heat Reduction Field Observations. The addition of lime and PAC reduced the current to the first field of the test ESP to avoid excessive sparking, but the voltage was unchanged as were the characteristics of the other fields. P AC addition alone reduced particulate emissions, likely because of changes in the conductivity of the particulate cake. No noticeable changes in opacity were recorded. The ESP residue handling system was overloaded by the high lime injection tests. Blockage of the double dump values occurred. The increased volume of residue requires upgrading of the residue handling system at this installation. Dry powdered hydrated lime h3nd1ing required replacement of the commercial venturi that worked fine with P AC, but failed after two hours at a 1: I stoichiometric flow rate of powdered hydrated lime. After installation of a purpose-built entrainment device, the reagent addition system worked satisfactorily. Performance Results. The testing at EAC/Pittsfield confirmed the findings at DCERF. Dry sorbent injection and P AC addition allows the MWCs to meet the 1995 emissions 1 60 ng/dsm 7% O 2 for ESP equipped plants

9 guidelines for small plants. Indeed, minimal quantities of PAC were required to bring both mercury and PCDD/F emissions into line with these requirements. Dry sorbent injection and P AC addition should allow large ESP equipped facilities to meet all the 1995 performance targets except acid gas removal providing, of course, the installed ESP must meet particulate emission standards. ESP improvements, such as particle conditioning and enhanced enerization control may be needed. The testing extended the lower end of the PAC addition data from 300 mg/dsm3 (DCERF) to 90 mg/dsm3 (EAC/Pittsfield). Even at this low level, emitted concentrations met the large plant guidelines for PCDD/F and mercury. Particulate emissions, while generally meeting the small facility performance standard under all conditions, doubled from the normal operating level when the loading to the ESP was doubled. The addition of 120 Ib/h of lime is estimated to be equivalent to the normal dust loading from the furnaces. This suggests the marginal ESPs will require turning or retrofitting to continue to meet the particulate emissions standards. Metals generally associated with particulate emissions, namely lead and cadmium, display no change with respect to lime or P AC addition. This is consistent with observed DCERF performance. Lead emissions at EAC/Pittsfield were higher than those measured at DCERF.But, the lead concentration displayed the same proportion to TSP concentration. Improved ESP performance should reduce lead emissions. Emissions Control Within normal limitations, the results presented and discussed in this report show that acid gas reagent injection and PAC addition are capable of meeting the December 19, 1995 emissions guidelines for small plants (<250 TPD MWC). With respect to the large plant guidelines, the test results indicate that dry sorbent injection can meet the PCDD/F and mercury limitation. However, sulfur dioxide and hydrogen chloride emissions limitations were not achieved. An interesting finding is that P AC addition alone can bring PCDD/F and mercury into line with the emissions guidelines without additional temperature control or acid gas reagent injection. Addition of PAC showed a pronounced effect on PCDD/F concentrations. With the low PAC injection rate of 4 Ib/h (100 mg/dsm 7% O 2 ) removal efficiencies were in the eighty percent range. PCDD/F removal was enhanced by the addition of lime. Addition ofpac also produced pronounced reductions in mercury emissions. With PAC injection. mercury concentrations well below the 80 µg/dsm 7% O2 emissions guideline were routinely achieved The test results indicate that lime is capable of achieving better than 50 percent reductions in S0 2 and HCl emissions. The 75% removal efficiencies for SO 2 and 95% removal efficiencies for HCl for large plants were not achieved at the test sampling

10 location - the electrostatic precipitator outlet. However, the flue gases are scrubbed downstream in the wet scrubber further reducing the S02 and HCl concentrations. The data indicates that EAC/Pittsfield was operating with very low NOx concentrations throughout testing. This is probably a consequence of recycling flue gas to the primary and tertiary chambers for temperature control. Overall, there was no discernible change in emitted TSP concentrations over the range of conditions tested. Adding lime and PAC did cause a reduction in the current to the first precipitator field, but the voltage remained essentially the same. As a result, there should be less spark-over and back-corona which can enhance removal. But, this effect can be offset by a lower charging rate which can produce lower removal. Additional gas conditioning, ESP tuning or ESP modifications might be needed to meet large plant emissions guidelines or improve the emission characteristics of existing units that have marginal performance. The particulate associated metals, cadmium and lead, generally exhibited no change in emitted concentration with P AC or dry lime injection. So, facilities that meet these requirements before retrofit should continue to do so. Other Findings The reagent injection system worked satisfactorily. Permanent installations for reagent systems should include loss-of-weight feeders and pneumatic line flow monitors to detect feed interruptions and blockages before an extensive site clean-up is needed. Dry powdered hydrated lime handling systems require easy access for preventative maintenance access, especially for the reagent entrainment systems. A standard commercial venturi worked well for P AC entrainment, but the same venturi quickly plugged (on the order of a few hours) when used to handle dry powdered hydrated lime. While Method 29 appears satisfactory for Section 129 metals, proof rinses and blank trains (instead of reagent blanks) are indicated to verify recovery and background contamination. With ever more restrictive emissions limitations. this becomes critical. Poor recovery can set' unrealistically low regulatory limits. Limits set within I method precision (less than the limit of quantitation) can only be achieved by chance.. To meet particulate limitations. the acid gas reagent additional rate is limited by the particulate removal capability of the ESP. The ESP satisfactorily separated the particulates from the gas stream; however the residue discharge system could not effectively convey the combined residue to the designated discharge area. Modifications to handle the increased residue flow may be needed.