Pilot-Scale Fluidized-Bed Combustor Testing ... - ACS Publications

1828

Energy & Fuels 2006, 20, 1828-1835

Pilot-Scale Fluidized-Bed Combustor Testing Cofiring Animal-Tissue Biomass with Coal as a Carcass Disposal Option Bruce G. Miller,*,† Sharon Falcone Miller,† Elizabeth M. Fedorowicz,† David W. Harlan,‡ Linda A. Detwiler,§ and Michelle L. Rossman| The Energy Institute, The PennsylVania State UniVersity, UniVersity Park, PennsylVania, Cargill, Inc., Minneapolis, Minnesota, Virginia-Maryland Regional College of Veterinary Medicine, College Park, Maryland, and National Cattlemen’s Beef Association, Centennial, Colorado ReceiVed March 28, 2006. ReVised Manuscript ReceiVed July 21, 2006

This study was performed to demonstrate the technical viability of cofiring animal-tissue biomass (ATB) in a coal-fired fluidized-bed combustor (FBC) as an option for disposing of specified risk materials (SRMs) and carcasses. This is a potential option for beneficially utilizing SRMs and carcasses that must be removed from the food chain, allowing them to be used as an energy feedstock rather than disposing them via landfilling or incineration. The purpose of this study was to assess the technical issues of feeding/combusting ATB and not to investigate prion deactivation/pathogen destruction. Overall, the project successfully demonstrated that carcasses and SRMs can be cofired with coal in a bubbling FBC. Feeding ATB into the FBC did, however, present several challenges. Specifically, handling/feeding issues resulting from the small scale of the equipment and the extremely heterogeneous nature of the ATB were encountered during the testing. Feeder modifications and an overbed firing system were necessary. Through statistical analysis, it was shown that the ATB feed location had a greater effect on CO emissions, which were used as an indication of combustion performance, than the fuel type due to the feeding difficulties. Baseline coal tests and tests cofiring ATB into the bed were statistically indistinguishable. Fuel feeding issues would not be expected at the full scale since full-scale units routinely handle low-quality fuels. In a full-scale unit, the disproportionate ratio of feed line size to unit diameter would be eliminated thereby eliminating feed slugging. Also, the ATB would either be injected into the bed, thereby ensuring uniform mixing and complete combustion, or be injected directly above the bed with overfire air ports used to ensure complete combustion. Therefore, it is anticipated that a demonstration at the full scale, which is the next activity in demonstrating this concept, should be successful. As the statistical analysis shows, emissions cofiring ATB with coal would be expected to be similar to that when firing coal only.

1. Introduction Bovine spongiform encephalopathy (BSE) is a prion disease in cattle that leads to neuron damage in the brain resulting in brain degradation and death. BSE is caused by a conformation change of the PrP prion protein to the PrPsc prion protein. BSE infection occurs when a molecule of PrPsc enters a cow’s body and comes into contact with molecules of PrP. PrP is found throughout the cow’s body but occurs mainly in nerve cells and immune cell membranes.1 Therefore, infectivity is known to occur through ingestion or inhalation of PrPsc, peripheral exposure to a place such as damaged skin, or hereditary inheritance. Skin contact with an infected object is not believed to be a cause of infection.1 Catastrophic outbreaks of BSE have occurred in Europe fueled by the common practice of feeding meat and bone meal (MBM) made from BSE-infected cattle to other ruminants. These outbreaks have lead to new laws that restrict the utilization of material potentially infected with BSE. Since the enactment of such laws, European countries including Germany, the United * To whom correspondence should be addressed. Telephone: 814 865 3093. Fax: 814 863 7432. E-mail address: [email protected]. † The Pennsylvania State University. ‡ Cargill, Inc. § Virginia-Maryland Regional College of Veterinary Medicine. | National Cattlemen’s Beef Association. (1) Novakofski, J.; et al. Prion biology relevant to bovine spongiform encephalopathy. J. Anim. Sci. 2005, 83 (6), 1455-1776.

Kingdom, The Netherlands, France, and Switzerland are feeding MBM to incinerators, cement kilns, and power stations.2 Research shows that although PrPsc is resistant to common decontamination practices including autoclaving, chemical solvents, and biological proteases, incineration is successful in destroying the infective protein in MBM.3 No prior research has been conducted in studying the combustion of animal tissue that is potentially infected with BSE. If possible, it may be more efficient to burn tissue instead of first processing the carcass into MBM prior to combustion. Also, animal tissue combustion may be useful if a future catastrophic BSE occurrence calls for the need for large-scale carcass disposal.4 In the case of such an event, rapid decontamination is necessary to prevent spread of the infection. BSE transmission from carcasses has been documented. Also, it has been shown that prion diseases can remain infective in the environment for at least 3 years.5 (2) Nottrodt, -Ing. A. Technical requirements and general recommendations for the disposal of meat and bone meal and tallow; Report for German Federal Ministry for Environment, Nature Protection and Reactor Safety, February 23, 2001. (3) Brown, P.; et al. New studies on the heat resistance of hamster-adapted scrapie agent: Threshold survival after ashing at 600 °C suggests an inorganic template of replication. Proc. Natl. Acad. Sci. 2000, 97 (7), 34183421. (4) Detwiler, L. Overview of Animal Disposal Issues. Presented at the Workshop on the Utilization of Animal Tissue Biomass in Boilers and Other Industrial Processes, University Park, Pennsylvania, July 22, 2004.

10.1021/ef060128n CCC: $33.50 © 2006 American Chemical Society Published on Web 08/18/2006

Cofiring Biomass with Coal as a Disposal Option

Energy & Fuels, Vol. 20, No. 5, 2006 1829

Figure 1. Comparison of as-fired energy densities of fuels evaluated at Penn State.

In the United States, the estimated annual supply of fed-cattle (i.e., cattle less than 30 months in age) specified risk materials (SRMs, which are the age-dependent anatomical parts of an animal that contain the specific tissues that have been demonstrated to harbor the BSE infectivity in animals), cow carcasses from packinghouses, on-farm mortalities, and cull-cow (i.e., cattle greater than 30 months in age) SRMs is 850, 75, 2500, and 650 million lb, respectively.6,7 With heating values ranging from 2300 to 6200 Btu/lb as fired, approximately 15 trillion Btu of energy is available for use as a fluidized-bed boiler fuel.6,7 The heating value of carcasses and SRMs is illustrated in Figure 1 and compared to that of various boiler feedstocks, both fossil fuels and biofuels, which have been tested at Penn State.8 Feedstocks with energy densities as low as 4000 Btu/lb (e.g., poultry litter) are fired in boilers as sole fuels while fuels with even lower energy densities are successfully cofired with coal.8 To deal with the issue of carcasses and SRMs disposal, this testing explored disposal through combustion in a fluidizedbed combustor. The U.S. Environmental Agency (EPA) noted the following: there are several design options for industrial and utility boilers available offering significant capacity for utilizing carcasses and SRMs; boilers have good control over combustion processes; they provide particulate matter and acid gas control.9 Although the EPA identified disadvantages of this technique in that fluidized-bed combustors are not permitted to burn wastes, they lack suitable feeding equipment, and there (5) Brown, P.; Carleton Gajdusek, D. Survival of scrapie virus after 3 years’ interment. The Lancet, 1991, 337 (8736), 269-270. (6) Harlan, D. H. Review of Packer Issues. Presented at the Workshop on the Utilization Animal Tissue Biomass in Boilers and Other Industrial Processes, University Park, Pennsylvania, July 22, 2004. (7) Hamilton, R. Review of On-Farm Mortalities. Presented at the Workshop on the Utilization of Animal Tissue Biomass in Boilers and Other Industrial Processes, University Park, Pennsylvania, July 22, 2004. (8) Miller, B. G.; Falcone Miller, S. Utilizing Biomass in Industrial Boilers: The Role of Biomass and Industrial Boilers in Providing Energy/ National Security. Presented at The First CIBO Industrial Renewable Energy & Biomass Conference, Minneapolis, Minnesota, April 7-9, 2003.

will be permit issues dealing with biocontaminants, they also stated that no technical issues appear to be insurmountable. Handling the carcasses/SRMs is not considered problematic. Fluidized-bed combustors have historically utilized low-grade fuels such as paper mill sludge, coal-water mixtures/pastes, waste coal, and others.8,10 A variety of handling equipment is available for these fuels. Similarly, the rendering industry processes and handles carcasses and other animal-tissue biomass (ATB); consequently, the handling, grinding, and delivery of ATB to a boiler is not an issue. Also, fluidized-bed boilers have been used to combust hazardous wastes, plastics, and waste oils with complete combustion achieved. Fluidized-bed combustion is a proven technology for low-grade fuels and is deemed ideal for utilizing carcasses and SRMs. This testing explored the possibility of cofiring ATB in coalfired boilers as a disposal option. In July 2004, Penn State, Cargill Taylor Beef, and the McDonalds Corporation hosted a workshop at Penn State to discuss and develop strategies to utilize ATB as a fuel in industrial and utility boilers, brainstorm on the development of a national infrastructure that could utilize ATB as a fuel on both a routine and large-scale emergency basis, and stimulate public-private collaboration.11 A major highlight of the presentations was that the boiler vendors informed the audience that the concept of cofiring ATB with coal was technically sound. One of the key points developed during the (9) Kremer, F.; Lemieux, P. Treatment/Disposal of TSE Contaminated Wastes. Presented at the Workshop on the Utilization of Animal Tissue Biomass in Boilers and Other Industrial Processes, University Park, Pennsylvania, July 22, 2004. (10) Miller, B. G.; Falcone Miller, S.; Harlan, D. H. Utilizing AnimalTissue Biomass in Coal-Fired Boilers: An Option for Providing Biosecurity. Presented at the 29th International Technical Conference on Coal Utilization & Fuel Systems, Clearwater, Florida, April 18-22, 2004. (11) Miller, B. G.; Harlan, D. W.; Detwiler, L. A.; Falcone Miller, S. Utilizing Animal Tissue Biomass in Coal-Fired Boilers: A Step Closer Towards Implementation? Presented at the 30th International Technical Conference on Coal Utilization & Fuel Systems, Clearwater, Florida, April 17-21, 2005.

1830 Energy & Fuels, Vol. 20, No. 5, 2006

workshop included performing pilot-scale testing to demonstrate to regulatory agencies, the U.S. Department of Agriculture (USDA) and the Food and Drug Administration (FDA), and industry the technical viability of this option.11 This paper presents the results from the pilot-scale testing, showing the feasibility of fueling a bubbling fluidized-bed combustor with ATB. The results obtained from the tests were statistically analyzed to determine if the introduction of ATB into the fuel stream affects combustion efficiency in a statistically significant manner. In the analysis, CO emissions were used as an indicator of combustion efficiency. 2. Experimental Section 2.1. ATB Preparation and Analysis. Cargill Taylor Beef provided the ATB for the project. Cargill Taylor Beef attempted to prepare mixtures that approximated cull-cow carcasses and both cull-cattle SRMs and SRMs from fed cattle. The samples, referred to as ATB1, ATB2, and ATB3, respectively, were prepared as follows: ATB1sCull-Cow Carcasses. This dead-cow mix was approximately 47% skeletal muscle (inedible beef), 30% cow slaughter plant offal (heads, legs, intestinal tract, etc.), 20% bones and fat trim from a cow carcass deboning operation, and 3% hide trimming. This material was coarse ground and mixed using commercial rendering equipment. ATB2sCull-Cattle SRMs. This high-bone mixture was approximately 55% bones and fat trim from a cow carcass deboning operation and 45% cow slaughter plant offal. This material was coarse ground using an inedible meat grinder with a final 1/8th inch plate. ATB3sFed-Cattle SRMs. This low-ash, high-moisture mixture consisting of 100% small intestine with contents harvested during slaughter, predominately from fed cattle. This material was ground using an inedible meat grinder with a final 1/8th plate. All materials were immediately blast frozen in the form of 50-lb blocks (≈22 in. × 18 in. × 5 in.) and stored in such form until delivery to Penn State. The ATB received from Cargill Taylor Beef was extremely heterogeneous and could not be used as received. Standard ASTM, ISO, and EPA representative sampling preparation techniques are not available for this type of fuel. Traditional coal and wood sample preparation equipment were employed in an attempt to both homogenize the ATB and deliver a size consistently amenable to the auger feed system used with the fluidized-bed combustor (FBC). ATB preparation using jaw crushing, plate milling, roll crushing, and hammer milling techniques were attempted. ATB was introduced to the equipment frozen, thawed, and freeze-dried with liquid nitrogen. All attempts were unsuccessful in size reduction. Elasticity and compressibility were the characteristics of the biomass that interfered with the traditional sample preparation techniques. The preparation problems associated with the elasticity of the biomass could be somewhat overcome by preparing the fuel while wholly or partially frozen. After much trial-and-error, it was determined that ATB1 (i.e., cull-cow carcasses) that was mostly thawed and fed into a chipper/ shredder in small amounts, with the cutting blades on, produced a product that was somewhat amenable to the biomass feed system. Similarly, it was determined that ATB2 (i.e., high-bone mixture) that was partially thawed and fed into the chipper in larger amounts, with the cutting blades removed, produced a product that was somewhat amenable to the biomass feed system. ATB3 (i.e., small intestines) that was fully thawed and fed into the chipper in any quantity, with the cutting blades removed, produced a product that was quite amenable to the biomass feed system. In a full-scale system, it is envisioned that the ATB will not be fed frozen into an FBC. Rather, a two-stage grinding system, common to the rendering industry, would be utilized to prepare a feed material with the consistency of ground beef. This option was not available for the test program because Cargill Taylor Beef prepared the various ATB

Miller et al. Table 1. Characteristic Proximate Analyses and As-Fired Heating Values of ATB Samples and Coal moisture (as vol matter fixed C ash HV (asdeterm wt %) (dry wt %) (dry wt %) (dry wt %) fired Btu/lb) ATB1 ATB2 ATB3 coal

31.6 44.8 57.3 1.2

81.6 80.3 96.9 32.5

2.2 1.6 1.3 62.5

16.2 18.1 1.8 5.0

3 841 2 425 3 893 13 897

Table 2. Characteristic Ultimate Analyses of ATB Samples and Coal

ATB1 ATB2 ATB3 coal

carbon (dry wt %)

hydrogen (dry wt %)

nitrogen (dry wt %)

sulfur (dry wt %)

oxygen (dry wt %)

40.8 33.3 30.3 81.3

12.8 13.1 7.9 5.9

0.3 1.4 1.8 1.4

0.1 0.1 0.1 0.7

26.4 32.9 58.1 5.8

samples by mixing the components on a concrete slab to generate the required samples. To do this, the material could only be ground in one stage prior to mixing. Proximate, ultimate, and higher heating value (HHV) analyses were performed on the ATB samples according to ASTM methods D5142, D5373, D4239, and D5865. Characteristic results of the proximate and as-fired heating value (HV) analyses are listed in Table 1. Characteristic results of the ultimate analyses are listed in Table 2. 2.2. Coal Preparation and Analysis. A Middle Kittanning seam coal from Pennsylvania crushed to 1/4 in. × 0 was used as the coal feedstock for the tests. To ensure consistent coal samples for the multiple tests, the roll-crushed material was homogenized by long piling according to ASTM method D346. Characteristic results of the proximate and as-fired HV analyses are listed in Table 1. Characteristic results of the ultimate analyses are listed in Table 2. The coal particle size was ≈100% < 1000 µm. 2.3. Description of the Fluidized-Bed Combustor System. The FBC used during the test program, shown in Figure 2, is designed to operate as a circulating fluidized-bed combustor. It was modified to operate in a bubbling-bed mode for this testing. The combustor includes a 1-ft diameter main bed and freeboard. The freeboard height is approximately 115 in. At a fluidization velocity of 6 ft/s, the bulk gas residence time is approximately 1.5-2 s in the freeboard. A wall separating the sluice from the main bed was extended to prevent the loss of bed material into the sluice during testing. Fluidizing air was supplied by bubble caps located on four standpipes below the bed. A slide gate valve allows for the removal of bottom ash as necessary. During this test program, makeup sand and coal were fed from separate feeders. The feeders rested on load cells from which the feed rates could be monitored. The sand and coal combine in a transition hopper before being introduced to the feed tube through a rotary valve. The fuel is then gravity fed into the combustor approximately 1-2 ft above the bed. The products of combustion are monitored at the FBC outlet with a complete analytical package consisting of on-line analyzers. An additional gas sample was collected in a 1 L Tedlar bag approximately every 30 min and analyzed for C1-C4 hydrocarbons. 2.4. ATB Feeding Issues. Feeding ATB into the fluidized-bed combustor presented several new challenges not encountered with other types of feedstocks previously tested in The Energy Institute’s FBC. Some of these challenges are attributed to the small scale of the equipment used, while others resulted from the composition of the ATB. The original intent was to feed the ATB into the dense bed to ensure maximum mixing, high heat transfer to the ATB, and high combustion efficiency. However, feeding ATB into the dense bed with an air-cooled auger was met with moderate success. Difficulties were encountered with sand, used as the startup bed material, which eventually prevented the auger from rotating. After several attempts to feed ATB into the dense bed, it was determined that overbed feeding of the ATB was the best possible option at



Figure 2. Generalized schematic diagram of the fluidized-bed combustor.

this scale. Therefore, an access port located within the freeboard was enlarged to accept the injector and the feed system was relocated. (Note that the final feed location is indicated by the red horizontal access port.) While the use of overbed feeding of the ATB eliminated plugging of the injector with sand, the slugging of the ATB into the bed produced considerable fluctuations in the emissions. Plus, the ATB was introduced higher in the freeboard than optimum and overfire air could not be used; hence, CO emissions spikes were observed. 2.5. Operating Procedures. Prior to each test, 200 lb of medium grade industrial sand was added to the FBC. The sand served as the initial bed material. The induced draft fan was then turned on, and the controller was set to maintain a draft pressure of a -0.10in. water column within the boiler. The FBC typically operates with a slightly positive internal pressure during testing. Next, the combustion air blower was started and adjusted to achieve a bed velocity of approximately 6-7 ft/s at operating temperatures. This velocity was maintained throughout the test. The FBC natural gas burner was then ignited and the combustor allowed to preheat until temperature T2 (upper bed temperature)

was approximately 1300-1400 °F. Coal was then fed into the main bed, and the flow rate of natural gas decreased until a transition to burning 100% coal was completed. The total air flow through the bed was then adjusted to maintain the bed velocity. Additional sand was added to the combustor to achieve a bed depth of approximately 15-20 in. Each test began with at least a 1-h baseline period of burning coal. During this period, the feed rate of coal was adjusted to maintain a bed temperature of approximately 1600-1700 °F. If flue gas recirculation (FGR) was to be applied, the throttling valves were adjusted to achieve the desired percent recirculation. Following the baseline period, tests performed cofiring ATB with the coal were initiated by feeding ATB while reducing the feed rate of the coal. The thermal input contributed by the coal and ATB was adjusted to achieve the desired percent cofire and to maintain the bed temperature. Recirculation of flue gas (0-20%) into the main bed of the FBC was also performed. After establishing each baseline or cofiring condition, the various temperature, pressure, flow rate, and emissions data were collected during each test at 30-s intervals by the data acquisition system.


Miller et al.

Table 3. Summary of Average Operating Conditions for Selected Tests fuel

100% coal 100% coal 100% coal 87% coal 87% coal 83% coal 85% coal 90% coal 78% coal 82% coal 18% FGRa 30% FGR 13% ATB1 13% ATB1 17% ATB1 15% ATB2 10% ATB2 22% ATB3 18% ATB3 27% FGR 23% FGR 25% FGR 24% FGR

firing rate Btu/h coal Btu/h ATB ATB feed location main bed velocity (ft/s) temperatures (°F) T2: upper bed T4: middle freeboard T5: upper freeboard emissions O2 (%) CO (ppm) @ 3% O2 CO2 (%) @ 3% O2 SO2 (ppm) @ 3% O2 NOx (ppm) @ 3% O2 HCc (ppm) @ 3% O2 a

417 000 N/Ab N/A 6.3

333 600 N/A N/A 6.3

250 290 N/A N/A 6.1

229 396 31 147 inbed 6.5

258 535 38 152 overbed 6.6

246 779 51 882 overbed 6.6

222 563 37 990 overbed 6.6

230 154 24 710 overbed 6.5

196 849 56 744 overbed 6.7

242 789 54 899 overbed 6.6

1 597 1 269 1 201

1 681 1 366 1 290

1 633 1 367 1 295

1 690 1 314 1 233

1 663 1 600 1 532

1 716 1 598 1 503

1 678 1 439 1 350

1 702 1 551 1 472

1 688 1 521 1 415

1 696 1 627 1 532

10.3 1 689 15.9 537 439 50

8.4 1 344 16.6 507 411 35

7.8 1 415 16.3 492 362 49

10.1 1 364 15.8 471 493 74

3.5 1 656 15.3 482 311 160

4.0 2 570 14.9 420 342 207

9.7 1 952 15.4 475 508 184

4.6 2 098 15.3 327 353 221

8.7 2 912 15.1 545 466 281

2.8 2 607 14.9 465 307 338

Flue gas recirculation. b N/Asnot applicable. c HCstotal hydrocarbons.

The operators recorded the ATB feed rate, along with other manual readings and observations, at 15-min intervals. 2.6. Experimental Procedures for Continuous Emissions Monitoring for Flue Gas Composition Measurement. EPA stack sampling was performed using EPA methods 3A, 6C, 7E, 10, and 25A for O2, CO2, SO2, NOx, CO, and total hydrocarbons, respectively. In addition, gas samples were analyzed using a Perkin-Elmer Autosystem gas chromatograph fitted with a flame ionization detector and an Alltech Chemipak 018 80/100 column for C1-C4 hydrocarbons.

Table 4 variable fuel type

recirculated flue gas (RFG) feed location

3. Results and Discussion 3.1. Summary of Testing and Operating Conditions. Table 3 is a summary of selected tests representing coal baseline and ATB/coal cofiring tests performed at various levels of flue gas recirculation, coal/ATB ratios, and feed locations. In all, 35 tests were performed and used in the statistical analysis. A summary of the average operating conditions for the selected tests is provided in Table 3. Selected temperatures listed in Table 3 correspond to the location of the thermocouples shown in the FBC schematic in Figure 2. The emissions listed in Table 3 have been normalized to 3% oxygen in the flue gas in order to compare emissions results from tests with varying O2 concentrations. The CO emissions during the coal-fired tests were higher than normally observed in the FBC. The coal contained a significant quantity of fines (the coal size was ≈100% < 1000 µm), which did not get mixed into the bed but rather were entrained and not completely consumed. In a typical FBC system, the coal is coarser with an average size of ≈1000 µm and a top size of ≈5000 µm.12 Typically, coal with an average size of ≈4500 µm is used in the Penn State FBC with good results.13 Overfire air was not used in this test series, which would have resulted in lowering the CO emissions. The increased CO and total hydrocarbon emissions were also due to the difficulties encountered in feeding the ATB into the small system. For example, pieces of hide often bound the screw, resulting in ATB slugging into the FBC and hence variability in emissions. 3.2. Statistical Analysis of the Test Data Using ANOVA. A general linear model was used to conduct an analysis of variance (ANOVA) and analysis of covariance using MINITAB (12) Miller, B. G. Coal Energy Systems; Elsevier: Oxford, United Kingdom, 2005. (13) Miller, B. G.; Falcone Miller, S.; Wincek, R. T.; Wasco, R. S. Fuel Flexibility in Boilers for Fuel Cost Reduction and Enhanced Food Supply Security; Prepared for the Pennsylvania Energy Development Authority, June 30, 2006.

covariants temperature stack O2

no. of levels/values 4 coal (1) coal + ATB1 (2) coal + ATB2 (3) coal + ATB3 (4) 2 no (0) yes (1) 2 inbed (0) overbed (1) continuous T1-T6

release 14 software.14 The general linear model ANOVA tests the hypothesis that the means and variances of several sample sets are equal; therefore, the samples are from the same general population. The variance is a measure of how far the data are spread about the mean. In the case of the tests conducted, ANOVA tests the hypothesis that the factors listed in Table 4 do not affect the combustion efficiency in a statistically significant manner. CO levels were used as an indicator (response) of combustion efficiency because incomplete combustion produces some unreacted carbon in the form of CO. ANOVA was run at a 95% confidence interval or R ) 0.05. An R level of 0.05 means that there is a 5% probability of making a type 1 error, i.e., concluding that there is a significant effect by the factors (i.e., fuel, feed location, and FGR) on the response (CO) when there is not. The ANOVA analysis develops a model, which best describes the CO level as a function of the model factors and covariants. A covariant is a variable consisting of continuous data that are not controlled but may have an effect on the response. Covariants used in the model are temperatures 1-6 and the oxygen level. The most important statistic in the ANOVA table is the P-value (P-value in bold in Table 5). There is a P-value for each term in the model (except for the error term). The P-value for a term indicates whether the effect for that factor is significant in determining the response, i.e., CO level: • If P is less than or equal to the R level selected, then the effect for the term is significant. The R level is 0.05 at a 95% confidence interval. (14) MINITAB Statistical Software, release 14; Windows, 2003.



Table 5. Summary of ANOVA Model Resultsa ANOVA model conditions

1

2

3

4

fuels 1,2 feed locat inbed and overbed

fuels 1,2,3,4 feed locat overbed

fuels 1,2,3,4 feed locat overbed

fuels 1,2,3,4 feed locat overbed outliers omitted fuel/FGR factor

P-Values and Fit test factors fuel type feed locat FGR O2 T1 T2 T3 T4 T5 T6 fuel×locat fuel×FGR Sc R2 d

0.000 0.000 0.000 0.000 0.002 0.001 0.000 0.000 0.000 0.000 0.020 N/A 1015.6 20.05%

0.000 N/Ab 0.061 0.889 0.000 0.000 0.000 0.000 0.000 0.000 N/A N/A 1387.1 26.46%

0.000 N/A N/A N/A 0.000 0.000 0.000 0.000 0.000 0.000 N/A N/A 1387.6 26.39%

N/A N/A N/A N/A 0.000 0.000 0.000 0.000 0.000 0.000 N/A 0.000 333.4 23.35%

Figure 3. Plot of actual data means for each fuel FGR factor (fuel types 1-4/FGR 0,1).

a P-values are for response variable CO. b Not applicable, factor not used in the ANOVA model. c S is the square root of S2: the estimate of variance. d Coefficient of determination: proportion of data explained by the model factors.

• If P is larger than the R level (0.05) selected, the effect is not significant. • P-values given as 0.000 can be any value less than 0.0005. If the effect of a fixed factor is significant, then the level means for the factor are significantly different from each other. If the effect of an interaction term is significant, then the effects of each factor are different at different levels of the other factor(s). For this reason, it does not make sense to try and interpret the individual effects of terms, which are involved in significant higher-order interactions. The results of the various ANOVA tests are given in Table 5. An initial ANOVA was conducted on fuel 1 (baseline coal) and fuel 2 (coal + ATB1) fired inbed (0) and overbed (1) to determine the effect of feed location on CO emissions (Table 5, ANOVA 1). Fuel, location, and T1-T6 were shown to be statistically significant in predicting CO levels (i.e., P < 0.05 in bold); however, the ANOVA model was only able to explain 19.9% (R2) of the variation in the data. The CO concentrations measured during the overbed feeding of fuel 2 (2, 1) are higher (mean CO increased by 315 ppm or 22.7%) and more variable than those measured during inbed feeding of fuel 2 (2, 0). The difference in the mean CO level measured between the baseline tests (fuel 1) and fuel 2 (coal + ATB1 overbed feeding) is 398 ppm, whereas the difference between fuel 1 and fuel 2 (coal + ATB1 inbed feeding) is only 83 ppm. In addition, there appears to be an interaction between fuel and location (fuel × location). Unfortunately, no further analysis of the effect of feed location could be conducted as fuels 3 and 4 were only fired above the bed due to feeding problems (see Table 3). The analysis of the variance of only the overbed fired tests (location 1) for all the fuel types using fuel type and FGR as factors is given in Table 5, ANOVA 2. All of the P-values (P < 0.05 in bold) indicate that T1-T6 and fuel type are significant. The FGR factor and O2 are not significant in determining the CO levels. The model fitted the data with an adjusted R2 of 26.3%. This means that 26.3% of the data can be explained by the model factors (T1-T6, O2, fuel type, and FGR). A third ANOVA model was run eliminating the FGR factor and O2 covariant. The results are given in Table 5,

Figure 4. Plot of fitted means for each fuel FGR factor (fuel types 1-4/FGR 0,1).

ANOVA 3. The R2 was increased only slightly to 26.3% so that the model did not do a “good” job at being able to predict the data. The model indicated that the FGR factor was not significant in affecting CO levels given the large variance of the CO data. The addition of FGR resulted in a net increase of 298 ppm in mean CO in the tests where the fuel was fed overbed, whereas a change in fuel factor (Fuels 1-4) resulted in a net increase in mean CO of 1458 ppm. The “poor” fit of the model is attributed to the large fluctuations in CO levels due to feeding issues. The test data set was truncated to eliminate the high number of outliers to address the large fluctuations in CO. Outliers were defined as CO > 2000 ppm. ANOVA was run using the fuel, location, and FGR factors and the temperatures as covariants. The data show that FGR is significant in affecting the CO level; therefore, a combined fuel × FGR factor was used (Table 5, ANOVA 4). The model shows that the fuels were statistically indistinguishable from one another when fired under the same FGR conditions and the average CO emissions were reduced by 59.3% when fuels were fired with FGR. The one exception is fuel 4 (coal + ATB3) with FGR, which had a higher average CO emission level. A plot of the measured mean CO and the fitted mean CO as predicted by ANOVA for each fuel FGR combination is given in Figures 3 and 4, respectively. The overall measured and fitted CO for all tests is indicated by the horizontal line on each graph. The ANOVA 4 results indicate that the tests with FGR (excluding fuel 4) which plot above the horizontal line are statistically distinct from the tests where no FGR was present which plot below the line. In Figure 3, fuel 3 without FGR (30) lies on the line matching the mean of all tests. The predicted


Miller et al.

Figure 5. Plot of fitted mean versus data mean for fuel FGR factors. Table 6. Summary of Statistically Indistinct Fuel Firing Conditions Using Truncated CO Emissions Data group

RFG

fuel FGR factor

mean CO

I II

no yes

10 ≈ 20 ≈ 30 ≈ 40 (exception 41 w/FGR) 11 ≈ 21 ≈ 31

3426 1394

means from the ANOVA 4 model shows fuel 3 with FGR being grouped with the other tests in which FGR was present. An effect is present when changes in a factor (e.g., fuel or FGR) affect the response differently. The magnitude of an effect is related to its distance from the overall mean of the population (horizontal line) and the slope of the line from each fuel FGR factor. Different levels of the factor affect the response differently. The greater the difference in the vertical position of the plotted points (the more the line is not parallel to the x-axis), the greater the magnitude of the main effect. The plot shows that the introduction of FGR had the greatest effect in reducing CO emissions for fuel 1 (coal) and fuel 2 (coal + ATB1). Fuel 4 (coal + ATB3) did not show this trend. Figure 5 is a plot of the fitted mean versus the data mean with a regression line (R2 ) 0.62). The data fall into two clusters consistent with the measured and fitted means located above and below the total test mean plotted in Figures 3 and 4 and are identified in Table 6. The fuel × FGR combinations within each group are indistinguishable from one another based on the AVOVA results. 4. Conclusions and Summary The test data was analyzed to determine if the introduction of ATB into the fuel stream affected CO emissions in a statistically significant manner. In general, the presence of ATB did not statistically affect CO levels. The data suggests that the feed location had a greater effect on CO emission than the fuel type due to feeding difficulties. The presence of FGR was also important in affecting CO levels. In summary, the following represent our findings: • ANOVA could only predict 19.9, 26.3, and 23.2% of the variability of the emission data from the inbed feed tests (fuel 2), overbed feed tests, and combined tests (CO data truncated), respectively. The poor fit is attributed to the extreme variability of the data due to feeding limitations. • The feed location had the following effect. The feed location had a greater effect on measured CO levels than fuel type, and the location increased the variability in the CO emissions. Fuels 1 (baseline coal) and 2 (coal + ATB1) were statistically indistinguishable when fuel 2 was fed directly into the bed. The average CO emissions differed only by 83 ppm. Mean CO emissions of fuel 2 (coal + ATB1) increased by 315 ppm

(≈22%) when fed overbed. The increased variability in CO measurements in all the tests (as reflected in the standard deviation) in which fuel was fed overbed is attributed to pulsing in the feed, combustion of fats taking place in the freeboard, and poor mixing in the bed. • The fuel had the following effect. Mean CO levels were statistically increased for fuels 3 and 4 when fed overbed compared to the total population sample. The increase in mean CO from fuels 1-4 was 1306, 1705, 2027, and 2765 ppm, respectively. The population mean was 1720 ppm. • FGR had the following effect. FGR did not have a significant effect on CO emissions when evaluating the total data population; however, FGR did have a significant effect on CO emissions when the test data was truncated (i.e., CO > 2000 ppm was eliminated). The model shows that the fuels were statistically indistinguishable from one another when fired under the same FGR conditions. This suggests that coal+ ATB blends fired equally as well as 100% coal based on CO emissions. CO emissions were lower when the fuels were fired with FGR. • The oxygen level was not statistically significant in determining CO levels. • Temperature was significant in determining CO levels. The objective of the project was to demonstrate the technical viability of cofiring ATB in a coal-fired FBC as an option for disposing of SRMs and carcasses. This testing is necessary to demonstrate to the regulatory agencies, USDA and FDA, and the industry the technical viability of this disposal option prior to securing funding for a full-scale demonstration. The purpose of this testing was to assess the technical issues of feeding/ combusting the animal-tissue biomass and not to investigate prion deactivation/pathogen destruction. Overall, the project successfully demonstrated that carcasses and SRMs can be cofired with coal in a bubbling FBC. Feeding ATB into the FBC did, however, present several new challenges not encountered with other types of feedstocks previously tested in Penn State’s FBC. Specifically, handling/feeding issues were encountered during the testing; however, they were primarily artifacts of the small scale of the equipment and the specific feeders available for use with such a heterogeneous material. These issues would not be expected at the full scale since fullscale units routinely handle low-quality fuels. After several feeder modifications and trials at different injection locations in the pilot-scale FBC, an overbed feed system was selected for the testing. In addition, a 4-in. diameter screw feeder was used, and this large feed size, relative to the combustor diameter of 1 ft, resulted in fuel slugging into the combustor. This slugging, in turn, resulted in fluctuating carbon monoxide and total hydrocarbon emissions, which are indicators of incomplete combustion. However, the average emissions from ATB cofiring tests in many cases were similar to the average emissions from coal baseline testing, indicating that under the conditions tested in the small-scale unit, the ATB performed similarly to coal. In a full-scale unit, the disproportionate ratio of feed line size to unit diameter would be eliminated, thereby eliminating feed slugging. Also, the ATB would either be injected into the bed, thereby ensuring uniform mixing and complete combustion, or would be injected directly above the bed with overfire air ports used to ensure complete combustion. Emissions of SO2 were reduced when firing ATB since the overall sulfur content in the fuel decreased with the addition of ATB. However, in a full-scale system, limestone is used for SO2 control, whereas it was not for the pilot-scale testing; hence, SO2 emissions would be similar in both scenarios. Emissions of NOx increased with the addition of ATB; however, this


increase was less significant as the level of flue gas recirculation was increased. In summary, the objectives of the project were met in that cofiring carcasses and SRMs with coal was successfully demonstrated. While the test conditions were not optimum, due to the equipment limitations and atypical coal particle size distribution, the performance of the ATB cofire tests was comparable to coal baseline testing. Statistically, it was shown that the ATB feed location had a greater effect on CO emissions, which were used as an indication of combustion performance, than the fuel type due to feeding difficulties. Baseline coal (fed about 2 ft above the bed) tests and tests cofiring ATB1 into the bed were statistically indistinguishable. This indicates that a demonstration at the full scale, which is the next major activity in demonstrating this concept, should be successful since equipment limitations would not be a factor. Hence, emissions cofiring ATB with coal would be expected to be similar to that when firing coal only. In testing currently underway in which additional ATB/coal testing is being performed, the fines have been removed from


the coal and overfire air is being used, thereby CO emissions have been reduced significantly. In addition, the FBC has been modified and the ATB feed location has been lowered to just above the bed (i.e., same feed location as the coal). The purpose of this testing is to further demonstrate the ATB/coal cofiring concept with improved combustion performance even in the small-scale unit. Acknowledgment. Funding for the testing was provided in part by the Beef Checkoff through the National Cattleman’s Beef Association and by Cargill Taylor Beef. Cargill Taylor Beef is acknowledged for preparing the various ATB mixtures. Mr. David Kraft from The Babcock & Wilcox Company and Mr. Neil Raskin from Foster Wheeler North America Corporation are acknowledged for their technical support prior to the start of the test program. The staff and undergraduate students of The Energy Institute are acknowledged for performing system/feeder modifications, conducting pilot-scale testing, providing analytical support, reducing data and preparing graphs, and providing photographic support. EF060128N

Pilot-Scale Fluidized-Bed Combustor Testing ... - ACS Publications

Recommend Documents