2340
Ind. Eng. Chem. Res. 1997, 36, 2340-2345
Hot Gas Desulfurization with Zinc Titanate Sorbents in a Fluidized Bed. 2. Reactor Model J. T. Konttinen,*,† C. A. P. Zevenhoven,‡ and M. M. Hupa‡ Carbona Corporation, P.O. Box 610, FIN-33101, Tampere, Finland, and Department of Chemical Engineering, A° bo Akademi University, Lemminka¨ isenkatu 14-18 B, FIN-20520 Turku, Finland
High-temperature high-pressure sulfur removal is considered to be one of the key steps in the hot gas cleanup train of an integrated gasification combined cycle (IGCC) process. Regenerable mixed metal oxide sorbents such as zinc titanates are prime candidates for this purpose. As part of sulfur removal process development, Carbona Corp. is developing fluidized bed reactor models for scale-up. In the first part of this work, the parameters for the unreacted shrinking core and overlapping grain models were determined to describe the rate of zinc titanate in H2S capture. A method using these sorbent particle conversion models for fluidized bed application is presented in this paper. Both models show a reasonably good fit to the high-pressure fluidized bed sulfidation test data, with the same order of magnitude parameters as obtained in the previously reported solid conversion rate tests. Introduction H2S removal at high temperatures and pressures within the context of a simplified (IGCC) process has been investigated worldwide (Gupta and Gangwal, 1992, 1993; Harrison, 1995; Salo et al., 1995; Mojtahedi et al., 1994). With the use of regenerable sorbents, the amount of solid waste produced is minimized and the fuel-bound sulfur can be recovered as a commercial product, such as elemental sulfur (Portzer et al., 1995). Zinc titanate appears to be the leading sorbent for hightemperature high-pressure (HTHP) sulfur removal in fluidized bed reactors (Harrison, 1995; Salo et al., 1995). As part of sulfur removal process development, reactor models are needed for scale-up. It is essential for this work to apply a reliable and simple correlation for the conversion rate of solid sorbent such as zinc titanate in the sulfur capture reaction (Lew, 1990; Lew et al., 1992):
1/xZnxTiyOx+2y(s) + H2S(g) F ZnS(s) + y/xTiO2(s) + H2O(g) (1) In the previous part of this work (Konttinen et al., 1997) the parameters of two different models, namely the unreacted shrinking core model with variable effective diffusivity (USC) and the overlapping grain (OG) were determined on the basis of laboratory scale results for zinc titanate sulfidation. This paper concentrates on a method of applying the solid conversion rate model for bench-scale fluidized bed reactor modeling. The method will be shown to provide a simple way to apply laboratory-scale rate data to real-scale fluidized bed reactor modeling purposes. In addition to steady-state scaleup design, a dynamic sulfidation reactor model (based on the same principles) can be used for process control system design and operator training. Experimental Methods A unique HTHP reactor system has been used to evaluate candidate sorbents in cyclic sulfidation/regen† ‡
Carbona Corp. A° bo Akademi University. S0888-5885(96)00687-2 CCC: $14.00
eration tests (Mojtahedi and Abbasian, 1995a,b). The test unit includes simulated hot coal-gas-derived gas feed systems and a 7.5 cm diameter fluidized bed reactor with associated process instrumentation and control devices. The cyclic sulfidation/regeneration tests were conducted in this test unit in the temperature range 550-650 °C at 20 bar using a simulated coal gas mixture (Table 1) containing 1500 ppmv hydrogen sulfide. A detailed description of the reactor system and operating procedure in cyclic tests has been reported elsewhere (Abbasian et al., 1994; Mojtahedi and Abbasian, 1995a,b). Cyclic sulfidation/regeneration tests were carried out with sorbents “B” (UCI-2) and “C” (UCI-4) to determine the effectiveness of the sorbent in cycling. These sorbents would typically undergo hundreds of sulfidation/regeneration cycles in a commercial plant. The results of the cyclic tests indicate that the reactivity of both sorbents toward H2S gradually declines during successive cycles, as has been discussed in detail elsewhere (Mojtahedi and Abbasian, 1995a,b). The physical and chemical properties of the fresh sorbents used in the fluidized bed unit tests are shown in Table 2. The properties of sorbents B and C were also determined after the cyclic tests, particularly with regard to loss of porosity and surface area (Table 3) (Mojtahedi and Abbasian, 1995a,b). The data obtained with the test arrangements presented above will be presented in the modeling section of this text. Application of Time versus Solid Conversion Models for the Fluidized Bed Reactor The basic approach toward the modeling of the chemical performance of fluidized bed reactors is to describe the nonideal flow of gas and solids by considering separate phases. Thus, the reactive gas in its bulk concentration is one phase and the reactive solid has its own phase (Kunii and Levenspiel, 1991; Froment and Bischoff, 1990; Gupta and Gangwal, 1992). Kunii and Levenspiel (1991) have introduced a three-phase model, where the gas flow through the bed of solids is assumed to follow plug-flow behavior and the solid phase is perfectly stirred, which means that the concentration of solid reactant and the solid conversion are not © 1997 American Chemical Society
Ind. Eng. Chem. Res., Vol. 36, No. 6, 1997 2341
conversion of solid (zinc) in the bed, t is time (s), and N is the number of vertical slices in series. In eqs 2 and 3, the rate of solid conversion can be obtained from the USC and OG models (Konttinen et al., 1997) by discretizing, in other words, producing values of ∆X/∆t as a function of solid conversion. When the USC model is applied (Konttinen et al., 1997)
Table 1. Composition of Simulated Coal Gas Used in High-Pressure Fluidized Bed Cyclic Tests (Mojtahedi and Abbasian, 1995a) compd
vol %
compd
vol %
H2 CO H2O CO2
13 18 11 8
CH4 H2S N2
2.5 0.15 balance
where
property
B
C
Zn/Ti (molar ratio) zinc, w % bulk density, g/cm3 particle density, g/cm3 mercury pore volume, cm3/g porosity surface area, m2/g median pore diameter, Å average particle size, µm
1.46 46.6 1.32 1.51 0.273 0.412 3.3 4500 308
1.5 45.3 1.37 2.21 0.240 0.53 1.9 5000 267
F(X) )
f
mol of S in × gas conversion ) mol of Zn × time solid conversion (2) time
)
(
)
By using the symbols presented in Figure 1, this can be rewritten for the jth slice:
nS,0(1 - XS,1)(1 - XS,2)...(1 - XS,j-1)XS,j )
nZn dX N dt
RpFmol,s Rp2Fmol,s 1 + BX f (X) (4) fkin(X) + kUSC 6 Deff,0 1 + AX dif
the derivative can be obtained directly:
dependent on bed position. In the model, the effect of three phases in the bed can be lumped into one gas conversion rate parameter, which is a combination of reaction rate constant and mass transfer effects between the three phases for the purpose of calculating the concentration of gas at the bed exit. This model was shown to successfully predict the performance of fluidized bed catalytic reactors (Kunii and Levenspiel, 1991). The USC and OG models for the zinc titanate conversion rate in the sulfidation reaction (Konttinen et al., 1997) have a constant bulk-gas concentration as input. This cannot be directly applied into the fluidized bed reactor data, because the gas concentration at the inlet of the fluidized bed reactor is not the same as at the exit. In order to avoid this problem, Kunii and Levenspiel (1991) and Szekely (1976) suggest different forms of average concentration between the gas inlet and exit concentration for steady-state calculations and then calculate the performance of the reactor by iteration between gas and solid molar balances. The predictions will lead to incorrect results if the solid in the bed is so reactive that only the lower part of the bed is reacting, which means that the upper part of the bed is practically free of reactive gas. Another way to describe the nonideal flow of gas is to model it as several perfectly mixed tanks in series (Levenspiel, 1972, 1989; Froment and Bischoff, 1990). The division into N identical vertical tanks or slices is shown graphically in Figure 1. For each individual slice the following mass balance should be satisfied (1 mol of sulfur gas (H2S) reacts with 1 mol of solid zinc (in zinc titanate particle)):
(
1 F(X) CS,j
t)
Table 2. Chemical and Physical Properties of the Fresh Sorbents Used in the Tests (Mojtahedi and Abbasian, 1995a)
(3)
where nS,0 is the sulfur inlet flow (mol/s), XS,j is the conversion of sulfur in the jth slice, nZn is the total amount of zinc in the fluidized bed (mol), X is the overall
CS,j dX ) dt F′(X)
(5)
where CS,j is the concentration of sulfur gas in slice j (mol/m3). Equations 3 and 5 can be combined for slice j to give
nS,0(1 - XS,1)(1 - XS,2)...(1 - XS,j-1)XS,j ) nZn CS,0(1 - XS,1)(1 - XS,2)...(1 - XS,j) (6) N F′(X) This reduces to
nS,0XS,j )
nZn(1 - XS,j)rateZn N
where
rateZn )
CS,0 F′(X)
(7)
In eq 7, the symbol rateZn describes the rate of solid reaction according to the USC model (1/s) when exposed to the gas inlet concentration. This rate can be obtained similarly from the OG model by discretizing the model into small solid conversion steps. Rewriting eq 7, we get
R)
XS,j 1 - XS,j
where
R)
nZnrateZn NnS,0
(8)
This form is analogous to the performance equation of the first-order irreversible reaction in a perfectly mixed reactor (Levenspiel, 1972, 1989; Froment and Bischoff, 1990). It should be noted, however, that for parameter fitting using fluidized bed data at sufficiently high temperatures the equilibrium limitation of the sulfidation reaction should be taken into account. Figure 2 shows the equilibrium saturation concentration of H2S at the conditions of the TGA and fluidized bed tests. For 550 °C with a steam content of 10 vol % the equilibrium H2S content is only 2-4 ppmv, whereas for 650 °C it is 7-15 ppmv (depending on the crystal
2342 Ind. Eng. Chem. Res., Vol. 36, No. 6, 1997 Table 3. Chemical and Physical Properties of the Sorbent Samples during Life-Cycle Sulfidation/Regeneration Tests (Mojtahedi and Abbasian, 1995a) sorbent B zinc, w % Zn/Ti (molar ratio) bulk density, g/cm3 particle density, g/cm3 mercury pore volume, cm3/g porosity surface area, m2/g median pore diameter, Å average size, µm a
sorbent C
cycle 7
cycle 17
cycle 25
cycle 30
cycle 22
cycle 30
cycle 40
cycle 50
NAa
NA 1.51 1.61 1.7 0.16 0.272 2.2 2900 NA
NA 1.48 1.67 1.9 0.16 0.304 2.6 2500 NA
NA 1.47 1.69 1.96 0.14 0.274 2.7 2600 300
NA 1.49 1.54 2.48 0.167 0.414 1.4 4500 255
NA 1.45 1.57 2.58 0.16 0.413 1.4 4500 252
NA 1.45 1.6 2.56 0.16 0.41 1.2 4500 252
NA 1.442 1.625 2.51 0.153 0.384 1.3 4500 250
1.5 1.45 1.62 0.205 0.332 2.6 2700 NA
NA ) not available.
(
Mj + XS,eqj R)
Mj + 1
where
Mj )
yH2Oj-1 ) yH2Sj-1
(
)
XS,j XS,eqj - XS,j
yH2O,0 + yH2S,0 j-1
yH2S,0
)
-1
(1 - XS,j) ∏ 1
(9)
where XS,ej is the conversion gas when at equilibrium concentration and yH2Oj-1 and yH2Sj-1 are the fractional H2S and steam contents entering slice j. yH2S,0 and yH2O,0 are the fractional steam and H2S contents at the bed inlet. The conversion of gas in slice j in the reversible sulfidation reaction can be found by rearranging eq 9:
XS,j )
R XS,eqj β+R
where
Figure 1. Model equations when dividing the gas-phase flow into N identical tanks (or slices) in series.
XS,eqj ) 1 -
Mj K
and
β)
Mj + XS,eqj Mj + 1
≈ 1 (10)
In eq 10 K is the equilibrium constant of the sulfidation reaction. Because at the test conditions of interest the volume fraction of steam is more than 60 times higher than the volume fraction of H2S, the term β can reasonably be assumed to be 1. The volume fraction of H2S at the bed exit is
yH2S,e ) yH2S,0 (1 - XS,1)(1 - XS,2)...(1 - XS,N) ) N
yH2S,0
Figure 2. Fractional equilibrium content of H2S at the temperature range of interest in fluidized bed sulfidation (H2O, 10 vol %).
structure of the zinc reactant), which certainly has an influence on fluidized bed modeling parameters, especially on the reaction rate constant. For the conversion of gas in reversible reaction in slice j, an equation presented by Levenspiel (1972, 1989) can be applied:
(1 - XS,j) ∏ 1
(11)
The model described above for gas conversion in nonideal flow contains the previously introduced parameters reaction rate constant (k′), product layer diffusion coefficient (Dpl), and the number of tanks in series (N). By varying N, the test data of sorbent C at 20 bar and 650 °C, cycle 1 were modeled by adjusting the other two parameters to fit the data best. The results of this are shown in Figure 3. Figure 3 shows that best results are obtained when N is larger than 3. Thus, the plug-flow model can be used instead (Levenspiel, 1972, 1989; Froment and Bischoff, 1990). An equation that describes the firstorder reversible reaction in the plug-flow reactor (Levenspiel, 1972, 1989) can be directly applied:
Ind. Eng. Chem. Res., Vol. 36, No. 6, 1997 2343
Figure 4. USC and OG model fits into life-cycle fluidized bed sulfidation data of sorbent B at high pressure during cycles 1, 17, and 30 (H2O, 10 vol %). Figure 3. OG model fit into high-pressure fluidized bed sulfidation data of fresh sorbent C when alternating the number (N) of slices in series (H2O, 10 vol %).
(
)
nZnrateZn M + XS,eq XS,eq ln ) nS,0 M+1 XS,eq - XS,e where
M)
yH2O,0 yH2S,0
and
XS,eq ) 1 -
M K
(12)
XS,e is the total fractional conversion of H2S at the bed exit, and XS,eq is the conversion of H2S when reaching equilibrium concentration. Rearranging eq 12 gives the volume fraction of H2S at the bed exit
(
(
(
yH2S,e ) yH2S,0 1 - XS,eq 1 - exp
)))
-nZnrateZn nS,0β
where
β)
M + XS,eq ≈1 M+1
(13)
Fit of the Solid Conversion Models into Solid Conversion versus Gas Conversion Data. Results and Discussion In the modeling of the batch fluidized bed solid versus gas conversion data, the solid phase is assumed to be perfectly mixed, which means that the concentration of reactive solid is not dependent on bed position. The content of H2S at the fluidized bed exit can be found by using eqs 10 and 11 (N tanks in series) or eq 13 (plugflow). In order to minimize the number of adjustable parameters, plug-flow of gas through the bed is assumed in the following, which is supported by the results in Figure 3. Figure 4 shows the results of the fit of the USC model and the OG model into the results of sorbent B at 550 °C with the help of eq 13. Figure 4 shows the fit into the sulfidation results from consecutive cycles number 1, 17, and 30. A similar fit was obtained from cycles 7 and 25. The fits of the models into the test data with C at 650 °C at cycles 1, 30, and 50 are shown in Figure 5. The list of parameters for the two models obtained is shown in Table 4. In Table 4, the reaction rate constants reported by Lew (1990) and Lew et al. (1992) were used in the OG model. The intrinsic reaction rate constants were used in the USC model as reported in the previous paper
Figure 5. USC and OG model fits into life-cycle fluidized bed sulfidation data of sorbent C at high pressure during cycles 1, 30, and 50 (H2O, 10 vol %).
(Konttinen et al., 1997). Thus, the product layer diffusion coefficient was used as a fitting parameter. The fits of the models produced this way in Figures 4 and 5 seem to fit the data best in cycle 17 with sorbent B and cycle 30 with sorbent C. A similarly good fit was obtained with cycles 7 and 25 of sorbent B data and cycles 22 and 40 of sorbent C. The differences in model fit between different cycles could be attributed to some changes in fluidization properties of sorbents B and C in cyclic testing, thus causing gas bypass through the bed in bubbles. The slight increase of the free ZnO phase in the sorbent, as reported by Mojtahedi and Abbasian (1995b) during cyclic operation, can increase sulfur capture efficiency, because ZnO shows a 1.5-2 times faster sulfidation rate than zinc titanates (Lew, 1990). The values of the product layer coefficient in Table 4 are at the same level as in previously reported kinetic tests (Konttinen et al., 1997) with the same sorbents. The reasonably good model fitting results in Figures 4 and 5 indicate that the contact of zinc titanate particles with H2S in the bench-scale batch fluidized bed reactor was efficient; thus the possible mass transfer limitation (due to gas bypass in bubbles) did not have a big influence on the results. Figure 6 shows the results of a sensitivity analysis, where the OG model fits with reaction rate parameters by Lew et al. (1992) are compared with those obtained by using the ambient-pressure test rate parameters reported in the first part of our study (Konttinen et al., 1997). The data used for product layer diffusion coefficient fitting in Figure 6 are the same as shown in Figure 5 (sorbent C, 650 °C, 20 bar). At 650 °C, the reaction rate constant by Lew et al. (1992) is about 3 times higher than that determined by us. In Figure 6, the product layer diffusion coefficient values to fit the
2344 Ind. Eng. Chem. Res., Vol. 36, No. 6, 1997 Table 4. Parameter Values of the USC Model and OG Model Obtained by Using the Life-Cycle Fluidized Bed Data for Sorbents B and C sorbent B cycle 1a
cycle 7
cycle 17a
sorbent C cycle 25
cycle 30a
cycle 1a
cycle 22
cycle 30a
cycle 40
cycle 50a
Dpl (m2/s) (OG) 2.3 × 10-14 3 × 10-14 2.8 × 10-14 1.3 × 10-14 8.7 × 10-15 5 × 10-14 1.1 × 10-13 9.2 × 10-14 1.1 × 10-13 7.1 × 10-14 k′ (OG) 1/sb 33.85 28.61 25.41 33.56 35.95 105.6 87.32 90.84 77.26 82.06 3.5 × 10-9 2.2 × 10-9 Dpl (m2/s) (USC) 2.5 × 10-9 2.8 × 10-9 2.1 × 10-9 1.3 × 10-9 9.5 × 10-10 3.8 × 10-9 4.5 × 10-9 4 × 10-9 a
Shown in Figures 4 and 5. b Calculation method described in Konttinen et al. (1997).
tions, by taking all the phenomena mentioned in the previous paragraph into account, the computational complexity can become a concern. Because in such cases the pressure and temperature and sorbent physical properties are very close to the values of Figures 4 and 5,the USC model can be preferred due to the possibility of obtaining analytical results without numerical integration. Conclusions
Figure 6. OG model fit into high-pressure fluidized bed sulfidation data at 650 °C using reaction rate constant parameters obtained from the ambient-pressure data (Konttinen et al., 1997) and from the literature (Lew et al., 1992).
data best with lower k values are only 1.2-1.3 times higher than those reported in Table 4. The same is true with all the high-pressure data (including sorbent B at 550 °C). It can be concluded that at the test conditions in Figures 4 and 5, the sulfur capture performance of zinc titanate sorbents is not much dependent on the initial sulfidation rate. Product layer diffusion resistance starts control the rate of sulfur capture at sorbent conversions higher than 10 mol %. Because of these reasons, the possible error caused by using the reaction rate constant values obtained in our ambient-pressure test conditions (Konttinen et al., 1997) with possible mass transfer effects seems not to be significant. The reasons for the change of physical and chemical properties of regenerable mixed metal oxide sorbents (such as zinc titanates) in consecutive sulfidation/ regeneration cycles are not discussed here, since they are beyond the scope of this study. These issues have been discussed closely by Gupta and Gangwal (1993), Mojtahedi et al. (1995b), and Harrison (1995). A simple way to predict the performance of a a realscale fluidized bed reactor can be obtained by using the method and its results presented above, with the addition of mass and energy balances of solids and gases entering and exiting the reactor. In a large-scale fluidized bed some reactant gas (H2S) can bypass the solid fluidized bed in bubbles (Froment and Bischoff, 1990; Kunii and Levenspiel, 1991). For large-scale applications the sulfidation reactor model will have to include equations for predicting the possible mass transfer limitation effects, which can be obtained from a model such as given by Kunii and Levenspiel (1991). A reactor model based on these principles was used in the sizing of the high-temperature high-pressure pilotscale sulfidation reactor with successful results (Salo et al., 1995). The advantages and disadvantages of the USC and OG models as part of the reactor model were discussed in the first part of this study (Konttinen et al., 1997). As the results show, both models can be used for modeling of a batch fluidized bed reactor. However, in industrial-scale sulfidation reactor modeling applica-
Two different models, the unreacted shrinking core and the overlapping grain models were applied for modeling of zinc titanate sulfidation in a fluidized bed reactor. The results presented show that the models used generally for fitting time versus solid conversion data can be applied directly to model fluidized bed results with gas conversion versus solid conversion data. All the experimental data can be fitted with two parameters, the reaction rate constant and the product layer diffusion coefficient. In order to minimize the number of adjustable parameters, the reaction rate constant was obtained from the literature data on zinc titanates, thus leaving the product layer diffusion coefficient as a fitting parameter. The assumptions one has to make for batch fluidized bed modeling are firstorder reversible sulfidation reaction, plug-flow of reactant gas through the fluidized bed, and perfectly mixed flow of solid reactant in the bed. The method forms the most essential part of a large-scale sulfidation reactor model which can be applied directly for scale-up and dynamic modeling purposes. Acknowledgment The work on the development and testing of regenerable sulfur removal sorbents and processes has been carried out as part of the “LIEKKI 2” Combustion Research Program and “SIHTI 2” Energy and Environmental Technology Program, which are partly financed by the Technology Development Centre of Finland (TEKES). The development of the sulfidation models was funded partly by the Imatran Voima Foundation. The authors greatly appreciated the comments and suggestions given by Prof. Dr. Maria Flytzani-Stephanopoulos of Tufts University, Medford, MA, during the evaluation of the modeling results. The authors acknowledge the valuable contribution of Dr. Javad Abbasian of the Institute of Gas Technology in producing the test data used for the model parameter fit in this article. Notation A, B ) parameters of the USC model (Konttinen et al., 1997) b ) stoichiometric factor C ) concentration of the gaseous reactant, mol/m3 gas D ) diffusion coefficient, m2/s F(X) ) parameter defined by eq 4 F′(X) ) first derivative of F(X) with respect to X
Ind. Eng. Chem. Res., Vol. 36, No. 6, 1997 2345 fkin(X), fdif(X) ) parameters of the USC model (Konttinen et al., 1997) K ) equilibrium constant k ) reaction constant, m/s k′ ) intrinsic reaction rate constant, 1/s Mj, M ) parameters defined by eqs 9 and 12 nS,0 ) flow of reactant sulfur gas, mol/s nZn ) total amount of zinc in fluidized bed, mol N ) total number of vertical slices in series Rp ) sorbent particle radius, m rateZn ) parameter defined by eq 7 t ) time, s X ) solid fractional conversion XS,j ) fractional conversion of reactant sulfur gas in vertical slice j y ) volume fraction of gas Z ) stoichiometric volume ratio of solid product to solid reactant R ) parameter defined by eq 8 β ) parameter defined by eqs 10 and 13 Fmol,s ) molar concentration of solid reactant, mol/m3 particle Subscripts e ) at fluidized bed exit eff ) effective eq ) equilibrium j ) order number of the vertical slice mol ) molecular pl ) product layer S ) sulfur gas (H2S) 0 ) initial
Literature Cited Abbasian, J.; Bachta, R. P.; Wangerow, J. R.; Mojtahedi, W.; Salo, K. Advanced High-Pressure Bench-Scale Reactor for Testing with Hot Corrosive Gases. Ind. Eng. Chem. Res. 1994, 33, 1. Froment, G. F.; Bischoff, K. B. Chemical Reactor Analysis and Design, 2nd Ed.; John Wiley & Sons: New York, 1990. Gupta, R. P.; Gangwal, S. K. Enhanced Durability of Desulfurization Sorbents for Fluidized Bed Applications. Topical Report, 96U-4274, Contract DE-AC21-88MC25006; Research Triangle Institute: Research Triangle Park, NC, November 1992. Gupta, R. P.; Gangwal, S. K. High-Temperature, High-Pressure Testing of Zinc Titanate in a Bench-Scale Fluidized Bed Reactor for 100 Cycles. Topical Report, Contract DE-AC21-88MC25006; Research Triangle Institute: Research Triangle Park, NC, June 1993.
Harrison, D. Control of Gaseous Contaminants in IGCC Processes, An Overview. In Twelfth Annual International Pittsburgh Coal Conference, Pittsburgh, PA, September 11-15, 1995; 1995; pp 1047-1052. Konttinen, J. T.; Zevenhoven, C. A. P.; Hupa, M. M. Hot Gas Desulfurization with Zinc Titanate Sorbents in a Fluidized Bed. 1. Determination of Sorbent Particle Conversion Rate Model Parameters. Ind. Eng. Chem. Res. 1997, 36, 2332-2339. Kunii, D.; Levenspiel, O. Fluidization Engineering, 2nd Ed.; Butterworth-Heinemann: Boston, 1991. Levenspiel, O. Chemical Reaction Engineering; John Wiley & Sons: New York, 1972. Levenspiel, O. The Chemical Reactor Omnibook; OSU Book Stores: Columbus, OH, 1989. Lew, S. High-Temperature Sulfidation and Reduction of Zinc Titanate and Zinc Oxide Sorbents. Ph.D. Dissertation, Massachusetts Institute of Technology, October 1990. Lew, S.; Sarofilm, A. F.; Flytzani-Stephanopoulos, M. Modeling of the Sulfidation of Zinc-Titanium Oxide Sorbents with Hydrogen Sulfide. AIChE J. 1992, 38 (8), 1161-1169. Mojtahedi, W.; Abbasian, J. H2S Removal from Coal Gas at Elevated Temperature and Pressure in Fluidized Bed with Zinc Titanate Sorbents. 1. Cyclic Tests. Energy Fuels 1995a, 9 (3), 429-434. Mojtahedi, W.; Abbasian, J. H2S Removal from Coal Gas at Elevated Temperature and Pressure in Fluidized Bed with Zinc Titanate Sorbents. 2. Sorbent Durability. Energy Fuels 1995b, 9 (5), 782-787. Mojtahedi, W.; Salo, K.; Abbasian, J. Desulfurization of hot coal gas in fluidized bed with regenerable zinc titanate sorbents. Fuel Process. Technol. 1994, 37, 53-65. Portzer, J.; Gangwal, S. K.; Dorchak, T. Slipstream testing of HotGas Desulfurization with Sulfur Recovery. In Twelfth Annual International Pittsburgh Coal Conference, Pittsburgh, PA, September 11-15, 1995; 1995; pp 1073. Salo, K.; Konttinen, J.; Ghazanfari, R.; Feher, G.; Lehtovaara, A.; Mojtahedi, W. Pilot Scale Experience on IGCC Hot Gas Cleanup. In Proceedings of the Advanced Coal-Fired Power Systems ’95 Review Meeting, Volume 1, June 1995; U.S. Department of Energy, Morgantown Energy Technology Center: Morgantown, WV, 1995. Szekely, J.; Evans, J. W.; Sohn, H. Y. Gas-Solid Reactions; Academic Press: New York, 1976.
Received for review October 28, 1996 Revised manuscript received January 30, 1997 Accepted March 15, 1997X IE960687P
X Abstract published in Advance ACS Abstracts, May 1, 1997.