- Email: [email protected]
Temperature effect on the pressure drop across the cake of coal gasification ash formed on a ceramic filter
Available online at www.sciencedirect.com
Powder Technology 181 (2008) 67 – 73 www.elsevier.com/locate/powtec
Temperature effect on the pressure drop across the cake of coal gasification ash formed on a ceramic filter Jin-Hyung Kim, Yu Liang, Kyoung-Min Sakong, Joo-Hong Choi ⁎, Young-Cheol Bak Department of Biological and Chemical Engineering/ ERI/BK21 for ECT, Gyeongsang National University, Jinju 660-701, South Korea Received 14 August 2006; received in revised form 14 June 2007; accepted 15 June 2007 Available online 26 June 2007
Abstract In order to predict the pressure drop across the cake of coal gasification (CG) ash formed on ceramic filter, an empirical equation was developed taking into account several factors, such as the face velocity, ash load, shape factor and size of particles, and especially the operating temperature. The hot air stream of well classified fine particles of CG ash was simulated as the syngas derived from the coal gasification process. The pressure drop behavior and cleaning efficiency of the filter were carefully investigated within the temperature range from room temperature to 673 K. The pressure drop across the ash cake was dominantly governed by the air viscosity, which increased with temperature. It was well expressed by the previously reported-empirical equation [J.H. Choi, Y.C. Bak, H.J. Jang, J.H. Kim, and J.H. Kim, Korean J. Chem. Eng., 21(3) (2004) 726.] with the modification of the viscosity term in the equation for different temperatures. The residual pressure drop rate across the ash cake also increased while the cleaning efficiency of the ceramic filter decreased as temperature increased. © 2007 Elsevier B.V. All rights reserved. Keywords: Temperature effect; Pressure drop; Cake; Filter; Gasification; Cleaning
1. Introduction The pressure drop is a primary factor in the design and operation of a ceramic filter unit. However, predicting the exact pressure drop can be difficult, as it depends on many codependent factors that originate from the design of the filter unit, the particle and gas properties, as well as the operating conditions. Moreover, the compression property of the ash cake accumulated on filter surface makes the situation even more complex. The compression phenomena of ash cake have been reported by several investigators [2–6]. Höflinger et al. [7] reported that reduction in the cake thickness occurred when particles became compacted by the exceeding shear force developed by the latterly formed-ash layers. Neiva et al. [8] proposed a cake build up model and reported that the compression of a given ash layer was related with the drag forces of its upper layers. Compression of the ash cake leads to compaction of the ash layer, which results in a reduction of the ⁎ Corresponding author. E-mail address: [email protected] (J.-H. Choi). 0032-5910/$ - see front matter © 2007 Elsevier B.V. All rights reserved. doi:10.1016/j.powtec.2007.06.006
cake porosity as well as an increase of pressure drop across the cake. Some computer programs [3,7,8] have been developed to calculate the pressure drop across ash cake (ΔPc), based on the theories mentioned above, which have shown good agreement with experimental results in limited cases or valid by specific parameters. Schmidt [3] considered local compressions of the cake for the progressive increase of pressure drop with filtration time. The model contains two empirical parameters that have to be determined by experimental results. Höflinger [7] applied the Mohr–Coulomb deformation criteria to spherical particles. The model proposed by Neiva et al. [8] based on cake build up to calculate the pressure vs. time, which requires the settling factor depending on the geometry of the filter vessel, process conditions such as velocity, concentration, density, particle size, agglomerates formation and gas dynamic viscosity. Additional works observed that the compression of ash cake depends on many co-dependent factors, including the particle properties (shape, size, and density) [9–11], the filter [12], the gas properties (density, viscosity and humidity) [13] and the operating conditions (the face velocity and the cleaning method) [14,15].
68
J.-H. Kim et al. / Powder Technology 181 (2008) 67–73
Endo et al. [5] derived an explicit equation for predicting the ΔPc of poly-dispersed particles taking into account the geometric mean diameter, the dynamic shape factor of particles and the particle size. However, his assumption of uniform porosity throughout the entire ash cake restricts its application. Choi et al. [1,2] developed a more general modified-Endo equation by adopting an extra two empirical equations, with respect to; (1) the ash cake porosity depending on the particle size, the face velocity and the ash load, and (2) the void function depending on the ash cake porosity and particle size. The predicted-values obtained using the modified-Endo equation showed good agreement with the experimental data found for CG fly ash at room temperature [1]. However, the effects of temperature to investigate the pressure drop across the ash cake have rarely been studied. The aim of this study is to develop an empirical equation for prediction of the pressure drop across the CG ash cake on a ceramic filter at moderate temperatures under 723 K in order to apply on the particle removal in a coal gasification process. Of the factors responsible for the pressure drop across the ash cake, the decrease in the porosity due to the particles sticking together should be dominant at high temperature, as a kind of liquid bonding due to the sintering effect. It has been pointed out that the sintering effect of fly ash is dominant above 1073 K, depending on the ash composition, especially in the case of particles originating from the PFBC (pressurized fluidized bed combustion) process, where the particles collection is carried out above 1023 K [19]. For gasification processes, such as the IGCC (integrated coal gasification combined cycle), the particles removal is preferably carried out under 823 K to meet the optimal conditions in combination with the drydesulfurization process which is carried out below 873 K [20]. Therefore, this study mostly focused on the removal of particles at the elevated temperature under 673 K in a coal gasification process. 2. Experimental Fig. 1 shows the schematic diagram of the experimental unit employed to measure the pressure drop across the ash cake at moderate temperature (RT −673 K) using the CG ash from a coal gasification unit. The constant volume of CG ash was fed into the air stream at room temperature using a screw feeder. Two cyclones were sequentially used to obtain the particles loaded-air streams of selected-particle sizes. The diameters of the first and the second cyclones were 64 and 48 mm, respectively. Consequently, the mean diameter of particles in the air stream decrease in the order of S2NS3NS4, where S2 and S3 are the flow streams originated from bottoms of the first and the second cyclone, respectively. S4 is the overflow stream from the second cyclone. The large particles in the raw ash were previously eliminated at the settling chamber (S1). One of the three streams was selected as the particle stream entering the filter unit. A ceramic filter was mounted between two flanges and tightened using long bolts as shown in Fig. 2. The particles loaded-air stream was introduced into the filter vessel, which
approach the filter from the outside and were then passed through the filter toward the inside, with clean gas finally passed to the outlet. The air flow was generated using a vacuum pump located downstream of the outlet. The face velocity across the filter was constantly controlled using a mass flow controller, even when the pressure drop changed during the run time. Using this method, the particles accumulated on the outer surface of the filter during filtration. The ash cake formed on the filter surface was dislodged by a pulse injection of air, via a pulse nozzle, at a pressure of 200 kPa. The pulse nozzle consisted of a 1/4 inch straight-tube located on the inside of a 1/ 2 inch co-centric tube. The pulsed amount of air was 0.65 g per pulse, with pulse duration of 0.6 s. The filter chamber was heated by an electrical heating unit. Temperature in the filter cavity measured with a 1/16 inch type K thermocouple. The temperature of the filter chamber was controlled using a PID controller connected to a thermocouple located in the filter chamber. The pressure drop across the filter was measured using a differential pressure transmitter connected to the filter chamber and the outlet of the filter unit, respectively. A pressure drop verse time (ΔP − t) curve was constructed at a constant face velocity. Part of particle stream entering the filter chamber was accepted for introduction into an aerodynamic particle size analyzer (API Aerosizer, manufactured by Amherst Process Instruments, Inc., Amherst, MA) to measure the particle size distribution under the experiment conditions. A cylindrical SiC filter (Dia-Shumalith 10–20, manufactured by Schumacher Umwelt-und Trenntechnik, Germany) [16], with outside and inside diameters of 60 mm and 40 mm, respectively, and a 50 mm long filter mounted inside the filter unit. The filter was comprised of a thin membrane layer on the outer surface, with a mean pore size of about 10 μm, which maintained collection efficiency of more than 99.0% for 0.5 μm sized particles [17]. The mass load (W) of accumulated-particulates on the filter surface was calculated by measuring the weight difference before and after the run using an electrical balance. The mass load was preliminarily confirmed to increase linearly with increasing run time in the short term, normally less then 30 min, but the feed rate then remained constant. The CG ash used in this study was obtained from the process development unit (PDU) of a coal gasification located at the Institute for Advance Engineering (IAE), Korea, and was collected using the filter unit with the SiC filters (DiaShumalith 10–20). The PDU was composed of an oxygen blown-entrainment type of gasification unit; with a coal feed capacity of 3 ton/day. Also, the dry coal was gasified at about 1573 K and 800 kPa. Fly ash from the gasification of Kideco coal at 1473–1623 K and 700–900 kPa was used in this study. The main gas composition during the gasification process was CO 57–65%, H2 20–30% and CO2 of about 20%. The ultimate analysis (moisture free basis) of the ash showed a composition of C 42.77%, H 0.12%, N 0.01%, S 0.19% and unburned-materials 56.91%. The physical properties of the classified-particle streams are shown in Table 1.
J.-H. Kim et al. / Powder Technology 181 (2008) 67–73
69
3. Results and discussion 3.1. Properties of the ash streams Well defined-particle streams with different particle sizes were produced using two cyclones. The probability density of the particle size (denoted by the geometric diameter) of the ash stream was measured using an API Aerosizer, and are presented in Fig. 3. The maximum peak of the probability density shifted to a smaller particle size in order S4NS3NS2, which supplies the opportunity to select the required particle stream of different particle size distribution. Stream S3 and S4 were quite suitable for simulating the particle stream of the field IGCC plant, where the mean particle size of the particles entering the ash collector was 2–3 μm when a cyclone was used as the pre-collector. The measured-data; geometric mean diameter (dg), volumetric mean diameter (dv) and geometric standard deviation (σg), are shown in Table 1. The adjusted-dynamic shape factor (κa) of the particles stream was calculated using Eq. (1), as defined in previously published work [2]. Cv and Cs are Cunningham correction factors of particles corresponding to diameters dv and ds, respectively, and were 1.16 and 1.0 for particles of 1 and 10 μm, respectively, under standard conditions (1 bar and room temperature), which decrease with increasing particle diameter. The Cunningham correction factor (C) for particles larger than 0.1 μm can be expressed by Eq. (2) [18], where λ is the mean free path of the fluid. The mean free path of air is 0.066 μm under standard conditions. The Stokes diameter was obtained using the sedimentation method [9]. In this study, the particles
Fig. 2. Detail schematic drawing for the filter assembly unit.
were suspended in ethanol solution contained within a glass tube with an inner diameter of 12 mm. The change in the concentration of the solids suspended over time was determined by measuring the change in the absorbance at 350 nm, which
Fig. 1. Schematic drawing of experimental set up to measure the filtration properties of a ceramic filter element at high temperature.
70
J.-H. Kim et al. / Powder Technology 181 (2008) 67–73
Table 1 Properties of particles in the CG ash loaded-air streams
3.2. Pressure drop across the ash cake
ρp dg dv ds σg κa (kg/m3) (μm) (μm) (μm) (μm)
Dusty stream
Carbon (wt.%)
The unburned carbon content (C%) was determined by measuring the weight change of the powder sample using thermal gravimetric analysis (TGA). The sample was initially heated from room temperature to 573 K, at a rate of 30 K/min, maintained at that temperature for 120 min, in order to eliminate the volatile compounds, and then heated to 1073 K, and maintained at this temperature for 60 min to burn out the carbon in the sample. The carbon was assumed to be naturally burnt out in the TGA cell at a temperature above 1073 K. The carbon contents were calculated using the equivalent weight loss corresponding to a temperature above 573 K, and was found to increase with decreasing particle size, as carbon particles are very fine particles [1]. The true density (ρp) of the particles was obtained using a Le Chatelier's density bottle by measuring the true volume of the particles corresponding to the determinedmass, and was found to decrease with decreasing particle size, because the carbon content increased with decreasing particle size, as mentioned above.
The overall pressure drop across the filter indicates the pressure difference between the inside and outside of the filter. However, the pressure drop across the ash cake was identified by detecting the extra value over that found across the filter only, which should be preliminarily measured under ash freeconditions; therefore, it denotes the difference between P and Pi (that is P − Pi) in Fig. 5. Fig. 5(A) represents a typical pressure drop pattern measured involving the conditioning period of a fresh filter. It was represented in a schematic diagram in Fig. 5 (b) to describe the notations. At a certain time i, particles start to be accumulated on the regenerated surface of ceramic filter and pressure drop reaches in its maximum value P during time Ti+1 − Ti. The pressure drop across the ash cake can be divided into two regions: the temporary pressure drop (ΔPc = P − Pi+1) and the residual pressure drop (ΔPr = Pi+1 − Pi) across the ash cake. ΔPr originates from several factors such as the residual ash cake remaining on the filter surface after the pulse cleaning, the pores plugging with the penetrated particles, and the reduction in the pore size of the ash cake due to compression under the high drag force of the gas flow or from being sintered at high temperature. In general, the pattern of ΔPc shows an upward curvature according to the mass load [1], which means the rate increase is low during the initial stage of cake layer formation, but considerably increases with run time (denoted in the mass load in the figure), as shown in Fig. 6. This phenomenon indicates that the compression of the ash cake occurs after the formation of a certain thickness of the ash cake and develops high compaction effect with multilayer structure of the ash cake. The experimental date in Figs. 6–8 were obtained during the first 30 min run time using the clean filter for the each case. So the pattern of pressure drop for each case is similar with that shown in Fig. 5(A). Considering the particle shape, particle size deviation and compressible property of the ash cake, ΔPc is well expressed by Eq. (3) for the operation at room temperature [1]. Where, the overall porosity (εo) is expressed by Eq. (4) with factors of particle size, face velocity, and ash load on the filter surface.
Fig. 3. Probability density of particles in the dust stream classified by cyclones.
Fig. 4. Plot of the accumulated mass fraction of particles to measure the Stoke's mean diameter.
S2 2,250 (1st cyclone bottom) S3 2,230 (2nd cyclone bottom) S4 2,200 (2nd cyclone top)
4.2
4.5
3.8
1.6
1.48 34.4
3.5
3.7
3.2
1.5
1.32 46.5
2.2
2.4
2.1
1.4
1.30 41.1
was converted into the accumulated mass fraction corresponding to Stoke's particle size (ds), as shown at Fig. 4 (for the case of S2). Stoke's mean particle size was obtained at the point where the cumulative mass fraction reached 0.5. ja ¼
C v dv 2 C s ds
C ¼ 1 þ2:52k =d:
ð1Þ ð2Þ
J.-H. Kim et al. / Powder Technology 181 (2008) 67–73
Fig. 5. (A). Typical pattern of the pressure drop change during the continuous filtration with pulse cleaning. (B). Schematic representation of pressure drop change.
This is a more advanced approach than that of Endo et al. [5] as considering the porosity change with the cake thickness in this study. DPc ¼
18 1010 lja ð1 eo Þ2 d ½qP dg1:08 expð4ln2 rg Þ1 vW e3o ð3Þ
eo ¼ 1 0:27dg0:15 v0:27 W 0:18 l ¼ l0
T 293
71
Fig. 8 also shows that Eq. (3) corresponded with the experimental data for the particles streams of S3 (dg is 3.5 μm) at different temperatures. These results indicate that the modified-Endo equation (Eq. (3)) can be valid for expressing the pressure drop across the ash cake formed on a ceramic filter, with respect to the complex effects originating from several factors such as the particle properties (the particle size, the particle size distribution, and the dynamic shape factor), operating conditions (face velocity, the mass load, and temperature) and compression effect. The main contribution of this equation is in the confirmation that the effect of temperature on the pressure drop across the ash cake is mainly related with the gas viscosity (which can be adjusted by Eq. (5)) within the elevated temperature range (RT −673 K). The experimental results imply that the sticking effect of the ash cake due to the liquefaction of GC ash is insignificant in the given temperature range. Meanwhile, several papers have reported [18–21] that fly ash from the coal combustion had a tendency to melt at high temperatures above 1027 K (which is recommended to achieve the high thermal efficiency of PFBC plant) and maintained a strong melting aspect when the fly ash contained high volatile materials, such as CaO and MgO originating from the desulfurization of compounds for PFBC. For the IGCC, in conventional cold gas cleaning systems, the CG ash removal is preferably performed at temperatures below 573 K, with some loss of the cycle efficiency, in order to fit the next purification unit for the removal of other impurities such as sulphur and/or nitrogen-containing compound using a wet cleaning system. An even higher elevated temperature (523– 923 K) is recommended for providing a beneficial hot gas cleaning system to increase the thermal efficiency, with the additional benefits of dry ash handling and the elimination of water treatment units. However, the successful long-term operation for gasification cycles in the field plant has been proven only within the limited temperature of 523–673 K [21]. In this temperature range, the sticking effect of the ash cake due to melting of GC ash, according to a previously published report, would be negligible [18]. Ninomiya et al. [18] reported
ð4Þ
0:75 :
ð5Þ
In Eq. (3), the most sensitive term relating to the temperature is the viscosity (μ). From the viscosity data for air, given in a reference [12], the viscosity correlated well with the power law according to the temperature change, as shown in Eq. (5). Where, μ0 is the viscosity of air at 298 K. Although Eq. (5) has no theoretical background, it is representative, with a good correlation confidence (Rsqr = 0.9993), of the viscosity of air at different temperatures. Figs. 6 and 7 show the experimental data for ΔPc are a good match with the values calculated by Eqs. (3)–(5) for the particle streams with different particle sizes at 473 and 673 K, respectively, at a face velocity of 0.03 m/s.
Fig. 6. Pressure drop change across the dust cake with the variation of the dust load and the particle size at 473 K and v = 0.03 m/s. Solid lines denote the values calculated by Eq. (3).
72
J.-H. Kim et al. / Powder Technology 181 (2008) 67–73
Fig. 7. Pressure drop change across the dust cake with the variation of the dust load and the particle size at 673 K and v = 0.03 m/s. Solid lines denote the values calculated by Eq. (3).
that CG ashes from an entrained bed coal gasification process did not show any considerable melting behavior at temperatures below 1300 K even when it contained high volatile compounds such as CaO. Fig. 9 shows the pressure drop rate (rc) across the temporary ash cake increased with both temperature and face velocity as well as with reduction in particle size. rc denotes the time rate of the increase in ΔPc per mass load as shown in Eq. (6). The reason for the increase in rc with temperature can be explained by the increasing effect of ΔPc with the increase in the air viscosity at higher temperatures. rc ¼
ðP Pi Þ : ðtiþ1 ti ÞW
ð6Þ
3.3. The residual pressure drop and cleaning efficiency The residual pressure drop rate (rr) indicates the time rate of the pressure drop across the residual ash cake remaining on the
Fig. 8. Pressure drop change across the dust cake with the variation of the mass load and temperature at the face velocity 0.03 m/s for 3.5 μm particle. Solid lines denote the values calculated by Eq. (3).
Fig. 9. The plot of pressure drop rate across the temporary dust cake with the variation of the face velocity for different temperature and particle size.
filter surface after the pulse cleaning. Where, rr means ΔPr change over the time per mass load, as shown in Eq. (7). Therefore, the rr value indicates the permanent filtration resistance, which is mainly due to the residual ash cake originating from insufficient filter cleaning. Fig. 10 shows the increases in rr as a function of the temperature rise for all ash streams in relation to the particle size and face velocity. A high value of rr corresponds to a low cleaning efficiency. The cleaning efficiency of the filter was defined by Eq. (8). rr ¼
ðPiþ1 Pi Þ ðtiþ1 ti ÞW
ð7Þ
g¼
ðP Piþ1 Þ ðP Pi Þ
ð8Þ
Fig. 11 shows the decreases in the cleaning efficiency with increasing temperature for all the cases with variations in the particle size and face velocity. The simultaneous increase of rr with decreasing η with increasing temperature is related with the increase in the pressure drop across the ash cake. Thus, a high pressure drop should increase the compression effect of the
Fig. 10. The change of pressure drop rate across the residual dust cake with the variation of the face velocity for different temperature and particle size.
J.-H. Kim et al. / Powder Technology 181 (2008) 67–73
rr T v W ΔPc ΔPr εo η κa λ μ ρp σg
73
Pressure drop rate across the residual ash cake (Pa·s−3 or ms−3) Temperature (K) Face velocity (m·s− 1) Ash load based on the filter surface area (kg·m− 2) Pressure drop across the ash cake (Pa) Pressure drop across the residual ash cake (Pa) Overall porosity of the entire cake layer (−) Cleaning efficiency (−) Adjusted-dynamic shape factor (−) Mean free path of air (m) Air viscosity (kg·s− 1·m− 1) Particle density (kg·m− 3) Geometric standard deviation (−)
Fig. 11. The change of cleaning efficiency with the variation of the face velocity for the different temperature and particle size.
Acknowledgement ash cake [17], and consequently induce a highly bound-ash cake, which forms a relatively strong attachment force to the filter surface. However, it seems that this effect would originate from the fluid property of air, such as an increase in the viscosity, but not from an increase in the chemical binding force due to particulate liquefaction at moderated temperatures below 673 K.
The authors gratefully acknowledge the financial support by the Center for New & Renewable Energy Development and Dissemination (CNEDD), of the Korea Energy Management Corporation (KEMCO), Korea. References
4. Conclusions The effect of temperature on the pressure drop across ash cake of coal gasification ash on a ceramic filter was carefully investigated within an elevated temperature range (RT −673 K). The pressure drop across the ash cake increased with increases in the filtration temperature. This temperature effect was found to be deeply related with the fluid viscosity, which also increased with temperature. The temperature dependency of the air viscosity can be presented as the equation: μ = μo (T / 293)0.75. Using this adjusted-viscosity, the pressure drop across the ash cake of GS ash was well expressed by the previously developed modified-Endo Eq. (1). The modified-equation correlated the experimental results more closely than the original one proposed by Endo et al. [5] by applying the flexible cake porosity according to particle size, face velocity, and especially cake thickness represented by mass load. Therefore, the increase in the pressure drop across the ash cake at high temperature was mainly due to the increase in the viscosity at higher temperatures below 673 K. The increased pressure drop at high temperature results in an increase in the residual pressure drop rate, as well as the decrease in the cleaning efficiency of the filter, as a high pressure drop leads to a high compaction of the ash cake.
List of symbols C dg ds dv P Pi rc
Cunningham correction factor, Cv for dv, and Cs for ds Geometric mean diameter of particles (m) Stoke's mean diameter of particles (m) Volume average mean diameter of particles (m) Total pressure drop (Pa) described in Fig. 4 Total pressure drop (Pa) at a certain pulse cycle, as described in Fig. 4 Pressure drop rate across the ash cake (Pa·s−1 or ms−3)
[1] J.H. Choi, Y.C. Bak, H.J. Jang, J.H. Kim, J.H. Kim, Korean Journal of Chemical Engineering 21 (3) (2004) 726. [2] J.-H. Choi, S.-J. Ha, H.-J. Jang, Powder Technology 140 (2004) 106. [3] E. Schmidt, Filtration & Separation (May 1998) 365. [4] M.L. Aguar, J.R. Coury, Industrial & Engineering Chemistry Research 35 (1996) 3673. [5] Y. Endo, D-R. Chen, D.Y.H. Pui, Powder Technology 98 (1998) 241. [6] J.H. Choi, S.J. Ha, Y.C. Bak, Y.O. Park, Korean Journal of Chemical Engineering 19 (6) (2002). [7] W. Höflinger, Ch. Stöcklmyer, A. Hackl, Filtration & Separation (December 1994) 807. [8] A.C.B. Neiva, L. Goldstein Jr., P. Calvo, High Temperature Gas Cleaning 2 (1999) 83. [9] K. Schulz, M. Durst, Filtration & Separation (January/February 1994) 25. [10] J.H. Choi, S.J. Ha, Y.O. Park, Korean Journal of Chemical Engineering 19 (4) (2002) 711. [11] J.H. Choi, S.J. Ha, Y.C. Bak, Y.O. Park, Korean Journal of Chemical Engineering 19 (6) (2002) 1085. [12] R.H. Perry, D.W. Green, Perry's Chemical Engineers' H/B, 6th ed., McGraw-Hill, 1973, pp. 20–98. [13] A. Gupta, V.J. Novick, P. Bisawas, P.R. Monson, Aerosol Science and Technology 19 (1993) 94. [14] R. Dennis, J.A. Dirgo, Filtration and Separation 18 (1981) 394. [15] C.R.N. Silva, V.S. Negrini, J.R. Aguiar, M.L. Coury, Powder Technology 101 (1999) 165. [16] K. Schulz, M. Durst, Filtration and Separation (January/February 1994) 25. [17] G. Hemmer, S. Berbner, H. Umhauer, G. Kasper, in: A. Diller, et al., (Eds.), High Temperature Gas Cleaning, vol. 2, 1999, p. 220. [18] Y. Ninomiya, A. Sato, R. Backman, International Conference on Ash Behavior Control in Energy Conversion Systems, March 18–19, Pacifico Yokohama, Japan, 1998, p. 170. [19] J.P. Hurley, B.A. Dockter, S.E. Allan, V. Kühnel, High Temperature Gas Cleaning 2 (1999) 25. [20] C.R. Black, 12 th EPRI Conference on Gasification Power Plants, San Francisco, 1993, p. 4. [21] S.C. Mitchell, Hot Gas Particulate Filtration, IEA Coal Research — The Clean Coal Centre, 1997.
Powder Technology 181 (2008) 67 – 73 www.elsevier.com/locate/powtec
Temperature effect on the pressure drop across the cake of coal gasification ash formed on a ceramic filter Jin-Hyung Kim, Yu Liang, Kyoung-Min Sakong, Joo-Hong Choi ⁎, Young-Cheol Bak Department of Biological and Chemical Engineering/ ERI/BK21 for ECT, Gyeongsang National University, Jinju 660-701, South Korea Received 14 August 2006; received in revised form 14 June 2007; accepted 15 June 2007 Available online 26 June 2007
Abstract In order to predict the pressure drop across the cake of coal gasification (CG) ash formed on ceramic filter, an empirical equation was developed taking into account several factors, such as the face velocity, ash load, shape factor and size of particles, and especially the operating temperature. The hot air stream of well classified fine particles of CG ash was simulated as the syngas derived from the coal gasification process. The pressure drop behavior and cleaning efficiency of the filter were carefully investigated within the temperature range from room temperature to 673 K. The pressure drop across the ash cake was dominantly governed by the air viscosity, which increased with temperature. It was well expressed by the previously reported-empirical equation [J.H. Choi, Y.C. Bak, H.J. Jang, J.H. Kim, and J.H. Kim, Korean J. Chem. Eng., 21(3) (2004) 726.] with the modification of the viscosity term in the equation for different temperatures. The residual pressure drop rate across the ash cake also increased while the cleaning efficiency of the ceramic filter decreased as temperature increased. © 2007 Elsevier B.V. All rights reserved. Keywords: Temperature effect; Pressure drop; Cake; Filter; Gasification; Cleaning
1. Introduction The pressure drop is a primary factor in the design and operation of a ceramic filter unit. However, predicting the exact pressure drop can be difficult, as it depends on many codependent factors that originate from the design of the filter unit, the particle and gas properties, as well as the operating conditions. Moreover, the compression property of the ash cake accumulated on filter surface makes the situation even more complex. The compression phenomena of ash cake have been reported by several investigators [2–6]. Höflinger et al. [7] reported that reduction in the cake thickness occurred when particles became compacted by the exceeding shear force developed by the latterly formed-ash layers. Neiva et al. [8] proposed a cake build up model and reported that the compression of a given ash layer was related with the drag forces of its upper layers. Compression of the ash cake leads to compaction of the ash layer, which results in a reduction of the ⁎ Corresponding author. E-mail address: [email protected] (J.-H. Choi). 0032-5910/$ - see front matter © 2007 Elsevier B.V. All rights reserved. doi:10.1016/j.powtec.2007.06.006
cake porosity as well as an increase of pressure drop across the cake. Some computer programs [3,7,8] have been developed to calculate the pressure drop across ash cake (ΔPc), based on the theories mentioned above, which have shown good agreement with experimental results in limited cases or valid by specific parameters. Schmidt [3] considered local compressions of the cake for the progressive increase of pressure drop with filtration time. The model contains two empirical parameters that have to be determined by experimental results. Höflinger [7] applied the Mohr–Coulomb deformation criteria to spherical particles. The model proposed by Neiva et al. [8] based on cake build up to calculate the pressure vs. time, which requires the settling factor depending on the geometry of the filter vessel, process conditions such as velocity, concentration, density, particle size, agglomerates formation and gas dynamic viscosity. Additional works observed that the compression of ash cake depends on many co-dependent factors, including the particle properties (shape, size, and density) [9–11], the filter [12], the gas properties (density, viscosity and humidity) [13] and the operating conditions (the face velocity and the cleaning method) [14,15].
68
J.-H. Kim et al. / Powder Technology 181 (2008) 67–73
Endo et al. [5] derived an explicit equation for predicting the ΔPc of poly-dispersed particles taking into account the geometric mean diameter, the dynamic shape factor of particles and the particle size. However, his assumption of uniform porosity throughout the entire ash cake restricts its application. Choi et al. [1,2] developed a more general modified-Endo equation by adopting an extra two empirical equations, with respect to; (1) the ash cake porosity depending on the particle size, the face velocity and the ash load, and (2) the void function depending on the ash cake porosity and particle size. The predicted-values obtained using the modified-Endo equation showed good agreement with the experimental data found for CG fly ash at room temperature [1]. However, the effects of temperature to investigate the pressure drop across the ash cake have rarely been studied. The aim of this study is to develop an empirical equation for prediction of the pressure drop across the CG ash cake on a ceramic filter at moderate temperatures under 723 K in order to apply on the particle removal in a coal gasification process. Of the factors responsible for the pressure drop across the ash cake, the decrease in the porosity due to the particles sticking together should be dominant at high temperature, as a kind of liquid bonding due to the sintering effect. It has been pointed out that the sintering effect of fly ash is dominant above 1073 K, depending on the ash composition, especially in the case of particles originating from the PFBC (pressurized fluidized bed combustion) process, where the particles collection is carried out above 1023 K [19]. For gasification processes, such as the IGCC (integrated coal gasification combined cycle), the particles removal is preferably carried out under 823 K to meet the optimal conditions in combination with the drydesulfurization process which is carried out below 873 K [20]. Therefore, this study mostly focused on the removal of particles at the elevated temperature under 673 K in a coal gasification process. 2. Experimental Fig. 1 shows the schematic diagram of the experimental unit employed to measure the pressure drop across the ash cake at moderate temperature (RT −673 K) using the CG ash from a coal gasification unit. The constant volume of CG ash was fed into the air stream at room temperature using a screw feeder. Two cyclones were sequentially used to obtain the particles loaded-air streams of selected-particle sizes. The diameters of the first and the second cyclones were 64 and 48 mm, respectively. Consequently, the mean diameter of particles in the air stream decrease in the order of S2NS3NS4, where S2 and S3 are the flow streams originated from bottoms of the first and the second cyclone, respectively. S4 is the overflow stream from the second cyclone. The large particles in the raw ash were previously eliminated at the settling chamber (S1). One of the three streams was selected as the particle stream entering the filter unit. A ceramic filter was mounted between two flanges and tightened using long bolts as shown in Fig. 2. The particles loaded-air stream was introduced into the filter vessel, which
approach the filter from the outside and were then passed through the filter toward the inside, with clean gas finally passed to the outlet. The air flow was generated using a vacuum pump located downstream of the outlet. The face velocity across the filter was constantly controlled using a mass flow controller, even when the pressure drop changed during the run time. Using this method, the particles accumulated on the outer surface of the filter during filtration. The ash cake formed on the filter surface was dislodged by a pulse injection of air, via a pulse nozzle, at a pressure of 200 kPa. The pulse nozzle consisted of a 1/4 inch straight-tube located on the inside of a 1/ 2 inch co-centric tube. The pulsed amount of air was 0.65 g per pulse, with pulse duration of 0.6 s. The filter chamber was heated by an electrical heating unit. Temperature in the filter cavity measured with a 1/16 inch type K thermocouple. The temperature of the filter chamber was controlled using a PID controller connected to a thermocouple located in the filter chamber. The pressure drop across the filter was measured using a differential pressure transmitter connected to the filter chamber and the outlet of the filter unit, respectively. A pressure drop verse time (ΔP − t) curve was constructed at a constant face velocity. Part of particle stream entering the filter chamber was accepted for introduction into an aerodynamic particle size analyzer (API Aerosizer, manufactured by Amherst Process Instruments, Inc., Amherst, MA) to measure the particle size distribution under the experiment conditions. A cylindrical SiC filter (Dia-Shumalith 10–20, manufactured by Schumacher Umwelt-und Trenntechnik, Germany) [16], with outside and inside diameters of 60 mm and 40 mm, respectively, and a 50 mm long filter mounted inside the filter unit. The filter was comprised of a thin membrane layer on the outer surface, with a mean pore size of about 10 μm, which maintained collection efficiency of more than 99.0% for 0.5 μm sized particles [17]. The mass load (W) of accumulated-particulates on the filter surface was calculated by measuring the weight difference before and after the run using an electrical balance. The mass load was preliminarily confirmed to increase linearly with increasing run time in the short term, normally less then 30 min, but the feed rate then remained constant. The CG ash used in this study was obtained from the process development unit (PDU) of a coal gasification located at the Institute for Advance Engineering (IAE), Korea, and was collected using the filter unit with the SiC filters (DiaShumalith 10–20). The PDU was composed of an oxygen blown-entrainment type of gasification unit; with a coal feed capacity of 3 ton/day. Also, the dry coal was gasified at about 1573 K and 800 kPa. Fly ash from the gasification of Kideco coal at 1473–1623 K and 700–900 kPa was used in this study. The main gas composition during the gasification process was CO 57–65%, H2 20–30% and CO2 of about 20%. The ultimate analysis (moisture free basis) of the ash showed a composition of C 42.77%, H 0.12%, N 0.01%, S 0.19% and unburned-materials 56.91%. The physical properties of the classified-particle streams are shown in Table 1.
J.-H. Kim et al. / Powder Technology 181 (2008) 67–73
69
3. Results and discussion 3.1. Properties of the ash streams Well defined-particle streams with different particle sizes were produced using two cyclones. The probability density of the particle size (denoted by the geometric diameter) of the ash stream was measured using an API Aerosizer, and are presented in Fig. 3. The maximum peak of the probability density shifted to a smaller particle size in order S4NS3NS2, which supplies the opportunity to select the required particle stream of different particle size distribution. Stream S3 and S4 were quite suitable for simulating the particle stream of the field IGCC plant, where the mean particle size of the particles entering the ash collector was 2–3 μm when a cyclone was used as the pre-collector. The measured-data; geometric mean diameter (dg), volumetric mean diameter (dv) and geometric standard deviation (σg), are shown in Table 1. The adjusted-dynamic shape factor (κa) of the particles stream was calculated using Eq. (1), as defined in previously published work [2]. Cv and Cs are Cunningham correction factors of particles corresponding to diameters dv and ds, respectively, and were 1.16 and 1.0 for particles of 1 and 10 μm, respectively, under standard conditions (1 bar and room temperature), which decrease with increasing particle diameter. The Cunningham correction factor (C) for particles larger than 0.1 μm can be expressed by Eq. (2) [18], where λ is the mean free path of the fluid. The mean free path of air is 0.066 μm under standard conditions. The Stokes diameter was obtained using the sedimentation method [9]. In this study, the particles
Fig. 2. Detail schematic drawing for the filter assembly unit.
were suspended in ethanol solution contained within a glass tube with an inner diameter of 12 mm. The change in the concentration of the solids suspended over time was determined by measuring the change in the absorbance at 350 nm, which
Fig. 1. Schematic drawing of experimental set up to measure the filtration properties of a ceramic filter element at high temperature.
70
J.-H. Kim et al. / Powder Technology 181 (2008) 67–73
Table 1 Properties of particles in the CG ash loaded-air streams
3.2. Pressure drop across the ash cake
ρp dg dv ds σg κa (kg/m3) (μm) (μm) (μm) (μm)
Dusty stream
Carbon (wt.%)
The unburned carbon content (C%) was determined by measuring the weight change of the powder sample using thermal gravimetric analysis (TGA). The sample was initially heated from room temperature to 573 K, at a rate of 30 K/min, maintained at that temperature for 120 min, in order to eliminate the volatile compounds, and then heated to 1073 K, and maintained at this temperature for 60 min to burn out the carbon in the sample. The carbon was assumed to be naturally burnt out in the TGA cell at a temperature above 1073 K. The carbon contents were calculated using the equivalent weight loss corresponding to a temperature above 573 K, and was found to increase with decreasing particle size, as carbon particles are very fine particles [1]. The true density (ρp) of the particles was obtained using a Le Chatelier's density bottle by measuring the true volume of the particles corresponding to the determinedmass, and was found to decrease with decreasing particle size, because the carbon content increased with decreasing particle size, as mentioned above.
The overall pressure drop across the filter indicates the pressure difference between the inside and outside of the filter. However, the pressure drop across the ash cake was identified by detecting the extra value over that found across the filter only, which should be preliminarily measured under ash freeconditions; therefore, it denotes the difference between P and Pi (that is P − Pi) in Fig. 5. Fig. 5(A) represents a typical pressure drop pattern measured involving the conditioning period of a fresh filter. It was represented in a schematic diagram in Fig. 5 (b) to describe the notations. At a certain time i, particles start to be accumulated on the regenerated surface of ceramic filter and pressure drop reaches in its maximum value P during time Ti+1 − Ti. The pressure drop across the ash cake can be divided into two regions: the temporary pressure drop (ΔPc = P − Pi+1) and the residual pressure drop (ΔPr = Pi+1 − Pi) across the ash cake. ΔPr originates from several factors such as the residual ash cake remaining on the filter surface after the pulse cleaning, the pores plugging with the penetrated particles, and the reduction in the pore size of the ash cake due to compression under the high drag force of the gas flow or from being sintered at high temperature. In general, the pattern of ΔPc shows an upward curvature according to the mass load [1], which means the rate increase is low during the initial stage of cake layer formation, but considerably increases with run time (denoted in the mass load in the figure), as shown in Fig. 6. This phenomenon indicates that the compression of the ash cake occurs after the formation of a certain thickness of the ash cake and develops high compaction effect with multilayer structure of the ash cake. The experimental date in Figs. 6–8 were obtained during the first 30 min run time using the clean filter for the each case. So the pattern of pressure drop for each case is similar with that shown in Fig. 5(A). Considering the particle shape, particle size deviation and compressible property of the ash cake, ΔPc is well expressed by Eq. (3) for the operation at room temperature [1]. Where, the overall porosity (εo) is expressed by Eq. (4) with factors of particle size, face velocity, and ash load on the filter surface.
Fig. 3. Probability density of particles in the dust stream classified by cyclones.
Fig. 4. Plot of the accumulated mass fraction of particles to measure the Stoke's mean diameter.
S2 2,250 (1st cyclone bottom) S3 2,230 (2nd cyclone bottom) S4 2,200 (2nd cyclone top)
4.2
4.5
3.8
1.6
1.48 34.4
3.5
3.7
3.2
1.5
1.32 46.5
2.2
2.4
2.1
1.4
1.30 41.1
was converted into the accumulated mass fraction corresponding to Stoke's particle size (ds), as shown at Fig. 4 (for the case of S2). Stoke's mean particle size was obtained at the point where the cumulative mass fraction reached 0.5. ja ¼
C v dv 2 C s ds
C ¼ 1 þ2:52k =d:
ð1Þ ð2Þ
J.-H. Kim et al. / Powder Technology 181 (2008) 67–73
Fig. 5. (A). Typical pattern of the pressure drop change during the continuous filtration with pulse cleaning. (B). Schematic representation of pressure drop change.
This is a more advanced approach than that of Endo et al. [5] as considering the porosity change with the cake thickness in this study. DPc ¼
18 1010 lja ð1 eo Þ2 d ½qP dg1:08 expð4ln2 rg Þ1 vW e3o ð3Þ
eo ¼ 1 0:27dg0:15 v0:27 W 0:18 l ¼ l0
T 293
71
Fig. 8 also shows that Eq. (3) corresponded with the experimental data for the particles streams of S3 (dg is 3.5 μm) at different temperatures. These results indicate that the modified-Endo equation (Eq. (3)) can be valid for expressing the pressure drop across the ash cake formed on a ceramic filter, with respect to the complex effects originating from several factors such as the particle properties (the particle size, the particle size distribution, and the dynamic shape factor), operating conditions (face velocity, the mass load, and temperature) and compression effect. The main contribution of this equation is in the confirmation that the effect of temperature on the pressure drop across the ash cake is mainly related with the gas viscosity (which can be adjusted by Eq. (5)) within the elevated temperature range (RT −673 K). The experimental results imply that the sticking effect of the ash cake due to the liquefaction of GC ash is insignificant in the given temperature range. Meanwhile, several papers have reported [18–21] that fly ash from the coal combustion had a tendency to melt at high temperatures above 1027 K (which is recommended to achieve the high thermal efficiency of PFBC plant) and maintained a strong melting aspect when the fly ash contained high volatile materials, such as CaO and MgO originating from the desulfurization of compounds for PFBC. For the IGCC, in conventional cold gas cleaning systems, the CG ash removal is preferably performed at temperatures below 573 K, with some loss of the cycle efficiency, in order to fit the next purification unit for the removal of other impurities such as sulphur and/or nitrogen-containing compound using a wet cleaning system. An even higher elevated temperature (523– 923 K) is recommended for providing a beneficial hot gas cleaning system to increase the thermal efficiency, with the additional benefits of dry ash handling and the elimination of water treatment units. However, the successful long-term operation for gasification cycles in the field plant has been proven only within the limited temperature of 523–673 K [21]. In this temperature range, the sticking effect of the ash cake due to melting of GC ash, according to a previously published report, would be negligible [18]. Ninomiya et al. [18] reported
ð4Þ
0:75 :
ð5Þ
In Eq. (3), the most sensitive term relating to the temperature is the viscosity (μ). From the viscosity data for air, given in a reference [12], the viscosity correlated well with the power law according to the temperature change, as shown in Eq. (5). Where, μ0 is the viscosity of air at 298 K. Although Eq. (5) has no theoretical background, it is representative, with a good correlation confidence (Rsqr = 0.9993), of the viscosity of air at different temperatures. Figs. 6 and 7 show the experimental data for ΔPc are a good match with the values calculated by Eqs. (3)–(5) for the particle streams with different particle sizes at 473 and 673 K, respectively, at a face velocity of 0.03 m/s.
Fig. 6. Pressure drop change across the dust cake with the variation of the dust load and the particle size at 473 K and v = 0.03 m/s. Solid lines denote the values calculated by Eq. (3).
72
J.-H. Kim et al. / Powder Technology 181 (2008) 67–73
Fig. 7. Pressure drop change across the dust cake with the variation of the dust load and the particle size at 673 K and v = 0.03 m/s. Solid lines denote the values calculated by Eq. (3).
that CG ashes from an entrained bed coal gasification process did not show any considerable melting behavior at temperatures below 1300 K even when it contained high volatile compounds such as CaO. Fig. 9 shows the pressure drop rate (rc) across the temporary ash cake increased with both temperature and face velocity as well as with reduction in particle size. rc denotes the time rate of the increase in ΔPc per mass load as shown in Eq. (6). The reason for the increase in rc with temperature can be explained by the increasing effect of ΔPc with the increase in the air viscosity at higher temperatures. rc ¼
ðP Pi Þ : ðtiþ1 ti ÞW
ð6Þ
3.3. The residual pressure drop and cleaning efficiency The residual pressure drop rate (rr) indicates the time rate of the pressure drop across the residual ash cake remaining on the
Fig. 8. Pressure drop change across the dust cake with the variation of the mass load and temperature at the face velocity 0.03 m/s for 3.5 μm particle. Solid lines denote the values calculated by Eq. (3).
Fig. 9. The plot of pressure drop rate across the temporary dust cake with the variation of the face velocity for different temperature and particle size.
filter surface after the pulse cleaning. Where, rr means ΔPr change over the time per mass load, as shown in Eq. (7). Therefore, the rr value indicates the permanent filtration resistance, which is mainly due to the residual ash cake originating from insufficient filter cleaning. Fig. 10 shows the increases in rr as a function of the temperature rise for all ash streams in relation to the particle size and face velocity. A high value of rr corresponds to a low cleaning efficiency. The cleaning efficiency of the filter was defined by Eq. (8). rr ¼
ðPiþ1 Pi Þ ðtiþ1 ti ÞW
ð7Þ
g¼
ðP Piþ1 Þ ðP Pi Þ
ð8Þ
Fig. 11 shows the decreases in the cleaning efficiency with increasing temperature for all the cases with variations in the particle size and face velocity. The simultaneous increase of rr with decreasing η with increasing temperature is related with the increase in the pressure drop across the ash cake. Thus, a high pressure drop should increase the compression effect of the
Fig. 10. The change of pressure drop rate across the residual dust cake with the variation of the face velocity for different temperature and particle size.
J.-H. Kim et al. / Powder Technology 181 (2008) 67–73
rr T v W ΔPc ΔPr εo η κa λ μ ρp σg
73
Pressure drop rate across the residual ash cake (Pa·s−3 or ms−3) Temperature (K) Face velocity (m·s− 1) Ash load based on the filter surface area (kg·m− 2) Pressure drop across the ash cake (Pa) Pressure drop across the residual ash cake (Pa) Overall porosity of the entire cake layer (−) Cleaning efficiency (−) Adjusted-dynamic shape factor (−) Mean free path of air (m) Air viscosity (kg·s− 1·m− 1) Particle density (kg·m− 3) Geometric standard deviation (−)
Fig. 11. The change of cleaning efficiency with the variation of the face velocity for the different temperature and particle size.
Acknowledgement ash cake [17], and consequently induce a highly bound-ash cake, which forms a relatively strong attachment force to the filter surface. However, it seems that this effect would originate from the fluid property of air, such as an increase in the viscosity, but not from an increase in the chemical binding force due to particulate liquefaction at moderated temperatures below 673 K.
The authors gratefully acknowledge the financial support by the Center for New & Renewable Energy Development and Dissemination (CNEDD), of the Korea Energy Management Corporation (KEMCO), Korea. References
4. Conclusions The effect of temperature on the pressure drop across ash cake of coal gasification ash on a ceramic filter was carefully investigated within an elevated temperature range (RT −673 K). The pressure drop across the ash cake increased with increases in the filtration temperature. This temperature effect was found to be deeply related with the fluid viscosity, which also increased with temperature. The temperature dependency of the air viscosity can be presented as the equation: μ = μo (T / 293)0.75. Using this adjusted-viscosity, the pressure drop across the ash cake of GS ash was well expressed by the previously developed modified-Endo Eq. (1). The modified-equation correlated the experimental results more closely than the original one proposed by Endo et al. [5] by applying the flexible cake porosity according to particle size, face velocity, and especially cake thickness represented by mass load. Therefore, the increase in the pressure drop across the ash cake at high temperature was mainly due to the increase in the viscosity at higher temperatures below 673 K. The increased pressure drop at high temperature results in an increase in the residual pressure drop rate, as well as the decrease in the cleaning efficiency of the filter, as a high pressure drop leads to a high compaction of the ash cake.
List of symbols C dg ds dv P Pi rc
Cunningham correction factor, Cv for dv, and Cs for ds Geometric mean diameter of particles (m) Stoke's mean diameter of particles (m) Volume average mean diameter of particles (m) Total pressure drop (Pa) described in Fig. 4 Total pressure drop (Pa) at a certain pulse cycle, as described in Fig. 4 Pressure drop rate across the ash cake (Pa·s−1 or ms−3)
[1] J.H. Choi, Y.C. Bak, H.J. Jang, J.H. Kim, J.H. Kim, Korean Journal of Chemical Engineering 21 (3) (2004) 726. [2] J.-H. Choi, S.-J. Ha, H.-J. Jang, Powder Technology 140 (2004) 106. [3] E. Schmidt, Filtration & Separation (May 1998) 365. [4] M.L. Aguar, J.R. Coury, Industrial & Engineering Chemistry Research 35 (1996) 3673. [5] Y. Endo, D-R. Chen, D.Y.H. Pui, Powder Technology 98 (1998) 241. [6] J.H. Choi, S.J. Ha, Y.C. Bak, Y.O. Park, Korean Journal of Chemical Engineering 19 (6) (2002). [7] W. Höflinger, Ch. Stöcklmyer, A. Hackl, Filtration & Separation (December 1994) 807. [8] A.C.B. Neiva, L. Goldstein Jr., P. Calvo, High Temperature Gas Cleaning 2 (1999) 83. [9] K. Schulz, M. Durst, Filtration & Separation (January/February 1994) 25. [10] J.H. Choi, S.J. Ha, Y.O. Park, Korean Journal of Chemical Engineering 19 (4) (2002) 711. [11] J.H. Choi, S.J. Ha, Y.C. Bak, Y.O. Park, Korean Journal of Chemical Engineering 19 (6) (2002) 1085. [12] R.H. Perry, D.W. Green, Perry's Chemical Engineers' H/B, 6th ed., McGraw-Hill, 1973, pp. 20–98. [13] A. Gupta, V.J. Novick, P. Bisawas, P.R. Monson, Aerosol Science and Technology 19 (1993) 94. [14] R. Dennis, J.A. Dirgo, Filtration and Separation 18 (1981) 394. [15] C.R.N. Silva, V.S. Negrini, J.R. Aguiar, M.L. Coury, Powder Technology 101 (1999) 165. [16] K. Schulz, M. Durst, Filtration and Separation (January/February 1994) 25. [17] G. Hemmer, S. Berbner, H. Umhauer, G. Kasper, in: A. Diller, et al., (Eds.), High Temperature Gas Cleaning, vol. 2, 1999, p. 220. [18] Y. Ninomiya, A. Sato, R. Backman, International Conference on Ash Behavior Control in Energy Conversion Systems, March 18–19, Pacifico Yokohama, Japan, 1998, p. 170. [19] J.P. Hurley, B.A. Dockter, S.E. Allan, V. Kühnel, High Temperature Gas Cleaning 2 (1999) 25. [20] C.R. Black, 12 th EPRI Conference on Gasification Power Plants, San Francisco, 1993, p. 4. [21] S.C. Mitchell, Hot Gas Particulate Filtration, IEA Coal Research — The Clean Coal Centre, 1997.