Introduction
At present there is considerable interest in advanced coal gasification processes such as integrated gasification combined cycle (IGCC) and pressurized fluidized-bed combustion (PFBC) systems. Advanced clean coal technology for electric power generation is currently under study because of high energy-efficiency as well as a cleaner environment. Porous silicon carbide (SiC) ceramic filters are commercially used for the combustion gas cleaning including IGCC and PFBC systems to clean combustion gas. The harmful submicron-sized particles cause corrosion of the turbine blade and ultimately reduce the energy efficiency in the PFBC system [1-3]. Therefore porous filters are used for the separation of the harmful submicron-sized particles in the combustion gas [1-5]. It can improve energy efficiency by producing clean gas and protect the environment by collecting the particles included in the gas.
In this filter system, cleaning process is indispensable for the continuous filtration because fouling is essentially caused by the agglomeration and deposition of dust particles included in the combustion gases. It is important to maintain a high cleaning efficiency during long operation of filters in order to reduce pressure drop as well as the operating costs [6-8].
In this hot gas filtration, high thermal and mechanical durability of porous filters are required to endure aggressive process environments containing steam, dust, pressure, and gaseous species [9-10]. For the PFBC, the conditions of environmental condition are usually at a temperature of 800~900oC and a pressure of 8~10 atm. In addition, these systems are under harmful gases such as SO2, CO, NO and NO2. Therefore a hot corrosion resistant material is desirable for the hot gas filter. To meet these requirements, SiC ceramic materials have been used as a hot gas filter.
In this study, porous SiC ceramic candle-type filters composed of highly porous support and filtration layer were fabricated by several processes such as uniaxial ramming, cold isostatic pressing (CIP) or extrusion forming. Filtering characteristics such as filtration efficiencies and dust cleaning efficiencies of porous SiC filters were characterized in continuous hot gas filtration. Also, the effects of the porosity of the filter support and the filter shape on the filtering characteristics were investigated. Additionally, the strength degradations of SiC filters exposed to PFBC conditions were measured to evaluate the durability of the filters.
Experimental Procedure
Silicon carbide (SiC) powder of mean particle size ~ 180 µm was mixed with inorganic binders, organic additives, and water. Inorganic binder such as clay (Gairome clay, Japan), mullite (Kyorith, Japan), or Al2O3 (Sumitomo, Japan) was added with fixed 1 wt% of calcium carbonate (CaCO3, Junsei, Japan). After the mixed batch was prepared, we made candle-type green bodies with dimensions of outer diameter 60 mm, inner diameter 40 mm, and length 500~1000 mm from the kneaded powder by uniaxial ramming with constant vibration forces, CIP or extrusion forming. The green bodies for the SiC supports were fabricated by control of the forming pressures during CIP, 20 MPa, 30 MPa, and 40 MPa to control the porosity (total pore volume). Different shapes were fabricated during the extrusion forming, tubular and cogwheel shape as shown in Figure 1. Some candle-typed filters were fabricated by uniaxial ramming technology with uniform vibration forces.
|
.jpg)
|
|
Figure 1. Sectional views of molds for SiC candle filters with different shapes: tube and cogwheel type.
|
The slurry that included finer powders of SiC, 14 µm, was spray coated onto the green body of the SiC support. Spray coating was performed uniformly while maintaining a constant rotation speed of the support. The coated green body was sintered at 1400oC for 3 h in air after removal of water and organic additives.
Filtration operating tests were performed at room temperature in air in the pilot-scale filtration unit. Figure 2 shows a schematic diagram of the test unit. The test unit consisted of a dust feed system, filter unit, pulse unit for back-pulse cleaning, and various measurement systems such as pressure loss and dust concentration.
|
.jpg)
|
|
Figure 2. Schematic diagram of filtration operational test unit.
|
Quantified dust with an average grain size of 5.2 µm, which was previously crushed from fly ash with size ≈ 17.2 µm, was fed into the filter unit by a screw.
The dust was fed from rotating powder disperser and ejected at Venturi type ejector. The dust was continuously moved by screw revolution and provided to the filter unit through inlet air diffuser. An atmospheric air stream was moved from the bottom to the upper part of the filter unit. The dust loadings were controlled by the revolution speed of the screw feeder. The test time is proportional to the dust loading and dust weight per area of filter, because the feeding rate was linearly proportional to the screw revolution speed. The relationship between ‘testing time,’ and ‘dust loading’ in our study is expressed by equation (1),
t = {DL/(Ci × Vf)} (1)
where t is testing time(sec), DL dust loading(g/m2), Ci initial inlet dust concentration (g/m3), Vf face velocity(m/s). For instance, 200g/m2 of dust loadings correspond to 1000 s of test time at 0.05 m/s of face velocity when the inlet dust concentration was fixed at 4 g/m3. The pressure drop (Dp) in the filter unit was measured with a pressure transmitter and manometer. The face velocity was varied from 0.02~0.05 m/s and inlet concentration was fixed as 4 g/m3 during the filtration operation.
The dust concentration was measured by an APS (aerodynamic particle sizer) at the inlet and outlet and we calculated separation efficiency (%), defined as,
η = {(Cin - Cout)/Cin} x 100% (2)
where Cin is inlet dust concentration and Cout outlet dust concentration. After the separation, back-pulse cleaning was operated automatically from the inside to the outside of the filter when the pressure drop reached a constant value, 5000~6000 Pa by injection of compressed air through pulse air header from the top of the filter unit. Here we define the cleaning efficiency (%) as,
ε = {(Pf - P)/(Pf - Po)} x 100% (3)
where Pf is cleaning pressure, Po is initial pressure drop, and P is a residual pressure drop after a back-pulse cleaning.
We made plate filter, 40 × 40 × 8.5 mm from the kneaded powder by uniaxial pressure of 300 kgf/cm2 to test thermal and mechanical durabilities. The green body was sintered at 1400oC for 3 h in air. The plate filters were cut to the size of 40 × 12.5 × 8.5 mm and was edge chamfered to evaluate the strength. The plate filters were exposed to similar PFBC conditions separately. First, a repetitive thermal cycling test was performed on the three types of filters with various additives (clay, mullite, or alumina with CaCO3). It was heated to 850oC at rate of 5oC/min and held at the maximum temperature for 1 h, and then cooled at the same rate. The thermal cycling was repeated in an air to a maximum of 800 cycles. Secondly, the filters were pressed under the static load, 10 kg/cm2, at room temperature for constant time. The pressurized period was varied from 0 to 106 s. Thirdly, we exposed SiC plate filters with different inorganic additives to hot corrosive environments for 175 h at 900oC. The detailed compositions of corrosive gases are shown in Table 1.
Table 1. Mixed gas composition used in the hot corrosion test.
|
|
|
Content(mol%)
|
25ppm
|
400ppm
|
200ppm
|
10ppm
|
7%
|
13%
|
balance
|
The center part of the filter was cut to characterize basic physical and mechanical properties. Sintered densities and porosities of the supports were calculated using the Archimedes method, and the average pore diameters were checked using a porosimeter (PoreSizer 9320, Micromeritics, USA). Microstructural characterization was carried out by Scanning Electron Microscope (SEM, XL-30, Philips, Netherlands) examination. Three-point flexural tests were run on the filter bars (span length 2.54 cm) before and after exposing to PFBC conditions. The bars were broken at a speed of 3 mm/min. O-ring strength tests were conducted for some filters.
Results and Discussion
The Physical Characteristics of the Porous SiC Filter
Figure 3 shows the representative microstructure of porous SiC candle filters made in this study. It shows bilayer structure that consists of a coating layer for particle separation and substrate layer for maintaining strength and lower pressure drop during filtration. The coating thickness is similarly maintained around 200~250 μm by spray coating of SiC slurry, as shown in the sectional view of Figure 3.
|
.jpg)
|
|
Figure 3. SEM micrograph of porous SiC candle typed filter showing sectional view.
|
All grains in the substrate layer are bonded to each other because of the liquid phase sintering resulted from the inorganic additives such as clay and calcium carbonate. The pore diameters in the substrate layer are much larger, 47 μm in the support, than those of coating layer, 10 μm in the coating layer, to maintain lower pressure drop during filtration. In addition, the larger pore diameter also helps to remove the deposited cake by back pulsing for continuous operation while the diameter in the coating layer is smaller for filtering the harmful particles in PFBC or IGCC applications.
We controlled the porosity of the tube-typed SiC filter supports by controlling the size of starting powders and the pressure during forming. We used SiC powders, with particle size from 175 μm to 250 μm to give uniform pore size and distribution. Figure 4 shows the relationship between porosity and forming pressure. The porosity depended on the pressure during forming. It could be controlled from 28.7% to 35.8% by varying the forming pressure from 40 MPa to 20 MPa. The average pore size was measured from 45 μm to 65 μm. The strengths of the filter supports were included in Figure 4. The strength increase with the pressure relates to the improvement of green density by higher pressure. The inverse trend of the relation with porosity and strength is in good agreement with prior results [11]. Although the strength showed lower values at higher porosity, the strength was estimated at over 20 MPa. This high strength is closely related with grain-boundary necking as shown in Figure 3.
|
.jpg)
|
|
Figure 4. Porosity and strength as a function of forming pressure; circular data points correspond to porosity and rectangular ones indicate strength data.
|
The characteristics of porous SiC candle filters with different shape, tubular and cogwheel typed filter as shown in Figure 1, are summarized in Table 2. The sintered density and porosity of the cogwheel type filter show slightly smaller values than tube type, but the difference is negligibly small. Average pore diameters are almost the same for the two filter shapes. The values of O-ring strength are similar to each other within the error range. These results confirm that all properties are similar irrespective of the filter shape.
Table 2. Characteristics of SiC ceramic filter with different shape.
|
|
|
Tube type
|
1.84
± 0.01
|
40.3
± 0.02
|
10(coating)/
47(sub.)
|
26.1
± 5.2
|
|
Cogwheel type
|
1.83
± 0.01
|
39.1
± 0.02
|
10(coating)/
46(sub.)
|
22.7
± 2.0
|
The Filtering Characteristics of the Tube-typed Silicon Carbide Candle Filter with Different Porosities
High separation efficiency, >99.99%, was measured in the pilot-scale filtration unit for the fabricated tube-type filter as shown in Figure 5. The efficiencies were measured by the equation (2) after 1800 s had passed with a fixed inlet dust concentration of 4 g/m3. The efficiency was measured at various face velocities for filters with different porosities of the supports. As shown in Figure 5, all SiC showed high separation efficiency, > 99.99%, irrespective of filter porosity or face velocity. The result indicates that most of separation successfully occurrs in the fine pores of the coating layer, not in the support layer.
|
.jpg)
|
|
Figure 5. Separation efficiency of SiC ceramic filter as a funtion of porosity of filter supports at various face velocities.
|
The cleaning efficiencies measured by Eq. (3) were plotted in Figure 6 for filters with different porosities for the supports. Cleaning efficiency is an important parameter in evaluating filter lifetime because imperfect cleaning ultimately causes clogging of the filter. The result indicates that the packed residual dusts in the coating layer of a filter with a less porous support will prevent continuous separation during the operation.
|
.jpg)
|
|
Figure 6. Cleaning efficiency of SiC ceramic filter as a function of test times. The porosities of filter supports are indicated in the graph.
|
The Filtering Characteristics of the Silicon Carbide Candle Filter with Different Geometries
Figure 7 plots the separation efficiencies for the two filter shapes, circular data points for the tube type filters and rectangular ones for cogwheel type filter. The separation efficiency (η) from Eq. (2) was plotted as a function of particle size from very fine particles, 0.5 μm, to coarse particles, 15.0 μm. Inlet dust concentration and the flow rate was fixed as 4 g/m3 and 0.94m3/min, respectively. All filters tested in this study showed similar separation efficiency, >99.99%, irrespective of filter shape. It is noteworthy that the two filters showed similar separation efficiency although the cogwheel shape has a larger surface area.
|
.jpg)
|
|
Figure 7. Separation efficiency of porous SiC candle filters as a function of particle size.
|
However, the cogwheel shape crucially affected the cleaning characteristics. Figure 8 shows cleaning efficiencies for the SiC filters with different shapes. As the particles are introduced into the filters, the dust particles are successfully collected in the finer pores of the coating layer, which caused the increase of pressure drop. The pressure drop of the two filters increased as time passed and the filter surfaces were packed with dust. However, the increased rate of the pressure drop was affected by different geometries because of the difference of face velocity. At a certain stage, back-pulse cleaning was required in order to conduct continuous filtering. Cleaning by back-pulse compressed air was performed at 5000 Pa of pressure drop in this study. Therefore the pressure drop diminished by the cleaning. Although complete cleaning was not obtained after the back-pulse cleaning by compressed air, the result of Figure 8 suggests better cleaning performance by the cogwheel type filter. As seen in Figure 8, there is not a large difference between the efficiency of the tube type filter and that of the cogwheel type filter at the first stage of particle mass loading, up to around 2.5kg. However, they showed an evident difference when a large amount of dust particles were introduced.
|
.jpg)
|
|
Figure 8. Cleaning efficiency of porous SiC candle filters as a function of particle mass loading.
|
The Results of Thermal and Mechanical Durability Tests for the Silicon Carbide Plate
Filter in the simulated PFBC condition
Figure 9 plots strength as a function of number of thermal cycles on SiC filters at pressureless air. The maximum temperature is maintained at 850oC for 1 h at each cycle. Therefore the number of thermal cycle exactly means holding time at 850oC. Data points are means and standard deviations of a minimum of five specimens. The graph indicates that SiC hot gas filters fabricated in this study have superior thermal stability in the PFBC condition up to 800 h. Thus, it is thought that only temperature of the PFBC condition does not cause the strength degradation much in air.
|
.jpg)
|
|
Figure 9. Flexural strengths of the SiC filters with various additives as a function of number of thermal cycles.
|
Figure 10 plots the strength of SiC filters as a function of holding time at fixed static load, in air. A uniaxial pressure was applied for a constant time to simulate the pressure of the PFBC condition, 10 kg/cm2, and then post strength measurements were performed. The test of uniaxial pressing maybe an extreme case because the pressure that closes to hydrostatic pressure applies on the filter in real PFBC condition. Nevertheless, the strength degradations do not occur under 10 kg/cm2 for 106 sec at room temperature in air as shown in the figure.
|
.jpg)
|
|
Figure 10. Flexural strengths of the SiC filters with various additives as a function of static times.
|
Figure 11 plots the strength data of SiC plate filter as a function of hot corrosion time at 900oC in mixed gas. The exposed time is varied to the maximum of 255 h.
|
.jpg)
|
|
Figure 11. Strength of hot gas filter bonded with various inorganic additives as a function of hot corrosion exposure time at 900oC.
|
We measured the porosity and strength of SiC filters bonded with various additives such as clay, mullite, and alumina before a hot corrosion test. The results are summarized in Table 3. The porosities of three filters were similar to each other irrespective of the types of additives. However, the strength degradation occurred in the corrosive environments for the clay and mullite-bonded filters as shown in Figure 11. Pastila [10] reports that the strength of SiC filter is reduced in the PFBC corrosive environment due to crystallization (SiO2 formation) and microcracks of grain boundary. The other study reported that clay-bonded SiC filters are subject to creep during long time exposure in PFBC or IGCC operations, because the filter element can fail during the operation due to the deformation of inorganic binder at high temperature [9]. However, the SiC filter bonded with alumina additive shows little strength degradation in this study. While the laboratory strength of alumina-bonded filter shows relatively lower than mullite-bonded filter, the strength after a long time-exposure in hot corrosive environment exhibits the highest value among the three filters. The results indicate that the type of additive should be determined by considering the strength degradation in a real environment.
Table 3. Porosity and strength of SiC hot gas filters with various additives.
|
|
|
Clay
|
36.13±2.02
|
33.14±0.90
|
|
Mullite
|
33.87±1.45
|
36.36±1.68
|
|
Alumina
|
35.90±0.16
|
32.06±1.81
|
Conclusions
The filtering characteristics and durability of porous SiC filters have been characterized in this study. The following results were obtained.
• Most dusts were collected in the SiC coating layer, >99.99% at room temperature and atmospheric conditions irrespective of the porosities and the shape of filter.
• Higher porosity of the support layer was more effective in the back-pulse cleaning.
• The filters with high surface area such as cogwheel type filter exhibited better performance with better cleaning efficiency.
• The filters fabricated in this study showed high thermal and mechanical resistance with some strength degradation by hot corrosive gas. The porous filter with alumina additive showed better damage tolerance against hot corrosion.
References
1. J. Stringer. and A. J. Leitch, “Ceramic Candle Filter Performance at the Gremethopre (UK) Pressurized Fluidized Bed Combustor”, J. Eng. Gas Turbines Power-T ASME, 114 (1992) 259-276.
2. R. A. Newby, T. E. Lippert, M. A. Alvin, G. J. Burck and Z. N. Sanjana, “Status of Westinghouse Hot Gas Filters for Coal and Biomass Power Systems”, J. Eng. Gas Turbines Power-T ASME, 121 (1999) 401-408.
3. J.H. Choi, S.M. Keum and J.D. Chung, “Operation of Ceramic Candle Filter at High Temperature for PFBC Application”, Korean J. Chem. Eng., 16 (1999) 823-828.
4. S.K. Woo, K.S. Lee, I.S. Han, D.W. Seo and Y.O. Park, “Role of Porosity in Dust Cleaning of Silicon Carbide Ceramic Filters”, J. Ceram. Soc. Japan, 109 (2001) 742-747.
5. K.S. Lee, S.K. Woo, I.S. Han, D.W. Seo, S.J. Park and Y.O. Park, “Filtering Characteristics of Porous SiC Filter with High Surface Area”, J. Ceram. Soc. Japan, 110 (2002) 656-661.
6. T. Lücke and H. Fissan, “The Prediction of Filtration Performance of High Efficiency Gas Filter Elements,” Chem. Eng. Sci., 51 (1996) 1199-1208.
7. H.S. Yu, “Optimization of HEPA Filter Design”, Proceedings of Institute of Environmental Sciences, 37 (1993) 35-43.
8. D.R. Chen, David Y. H. Pui and Benjamin Y. H. Liu, “Numerical Study and Optimization of Pleated Gas Filters”, Proceedings of Institute of Environmental Sciences, 37 (1993) 414-422.
9. R. Westerheide, J. Adler, A. Walch, W. Volker, H. Buhl and D. Fister, “High Temperature Gas Cleaning”, Volume II, Dittler A., Hemmer G. and Kasper G., (Ed.), Institut für Mechanische Verfahrenstechnik und Mechanik der Universität Karlsruhe, Germany, (1990), pp. 255-287.
10. P. Pastila, V. Helanti, A.-P. Nikkila and T. Mantyla, “Environmental Effects on Microstructure and Strength of SiC-based Hot Gas Filters”, J. Eur. Ceram. Soc., 21 (2001) 1261-1268.
11. R. W. Rice, “Comparison of Stress Concentration Versus Minimum Solid Area Based Mechanical Property-Porosity Relation”, J. Mater. Sci., 28 (1993) 2187-2192.
Contact Details
|