Ind. Eng. Chem. Res. 1988,27,434-439
434
Registry No. NiO, 1313-99-1; H2, 1333-74-0.
Pannell, R. B.; Chung, K. S.; Bartholomew, C. H. J. Catal. 1977,46, 340. Ross, J. R. H.; Steel, M. C. F.; Zeini-Isfahani, A. J . Catal. 1978,52, 280.
Literature Cited Bartholomew, C. H. J . Catal. 1976,45, 41. Bartholomew, C. H.; Pannell, R. B. J . Catal. 1980, 65, 390. Cimino, A.; Shiavello, M. J. Catal. 1971,20, 202. Farrauto, R. J. AZChE Symp. Ser. 1974, 70, 9. Gavalas, G. R.; Phichitkul, C.; Voecks, G. E. J . Catal. 1984,88, 54. Levenspiel, 0. Chemical Reaction Engineering; Wiley: New York, 1983; Chapter 12, pp 367-371. Martin, G . A.; Ceaphalan, N.; Moutgolfier, P. d. J. Chim. Phys. 1973, 70. 1422.
Shalvoy, R. B.; Reucroft, P. J.; Davis, B. H. J . Vac. Sci. Technol. 1980, 17, 209. Vedrine, J. C.; Hollinger, G.; Minh, 0. T. J . Phys. Chem. 1978, 82, 1515.
Wen, C. Y. Ind. Eng. Chem. 1968, 60, 34. Wu, M.; Hercules, D. M. J . Phys. Chem. 1979, 783, 2003.
Received for review March 20, 1987 Revised manuscript received October 26, 1987 Accepted November 14, 1987
Sorption of SO2 on Metal Oxides in a Fluidized Bed Bekir Zuhtu Uysal,*+Inci Aksahin,*and Hayrettin Yucelf Chemical Engineering Department, Jordan University of Science and Technology, Zrbid, Jordan, and Chemical Engineering Department, Middle East Technical University, Ankara, Turkey
A preliminary experimental study is described in which a low concentration of SO2 (0.2% v/v) is removed from air at temperatures between 358 and 640 K by reaction with bauxite and red mud, which consist of several metal oxides in a lab-size fluidized bed reactor of 5-cm diameter. CuO on activated alumina, a well-known sorbent for SOz,was also tested in the same system for comparison. A model based on the two-phase theory of fluidization was used to simulate the sorption process. It was shown that a kinetic rate expression for sorption of SO2 on a single metal oxide to yield a metal sulfate can also be used for mixtures of metal oxides by defining an overall apparent rate coefficient and an overall consumption coefficient for active sites. The work indicated that bauxite and red mud could also be considered as potential sorbents like CuO for SO2sorption in a typical range of stack gas temperatures. I t is generally accepted that emission of sulfur oxides from combustion of fossil fuels causes a serious air pollution problem. The control of sulfur dioxide emissions becomes more important due to the trend of using increasing amounts of high sulfur coal to meet energy requirements. A great number of processes have been proposed for removal of sulfur oxides from flue gases, and they are summarized in several references (Slack, 1971; Kohl and Riesenfield, 1979). All of the proposed processes for flue gas desulfurization can be classified into two basic categories: throwaway processes and recovery processes. In each category, the processes can be further classified into wet or dry processes. Wet throwaway processes such as alkali or limelimestone scrubbing have been preferentially developed on a commercial scale. However, dry recovery processes are thought to be more advantageous over wet systems since they may permit flue gas treatment a t elevated temperatures, avoiding flue gas reheating which is usually required for a wet process to maintain plume buoyancy. On the other hand, the dry recovery processes might have the usual problems of large-scale handling of regenerable solids, mainly high investment costs due to the complexities of the regeneration system and excessive losses of sorbent due to attrition. Recently, a significant amount of work has been done to develop dry recovery processes based on sorption of sulfur oxides on metal oxides (Bienstock et al., 1961; Thomas et al., 1969; Lowell et al., 1971). In a study by Tracor Corporation (Thomas et al., 1969; Lowell et al., 1971) the oxides of 48 metals were screened to determine which were best suited for the removal of sulfur oxides from flue gases by chemical reactions. The screening was t
Jordan University of Science and Technology. Middle East Technical University. 0888-5885/88/2627-0434$01.50/0
based on the thermodynamic requirement for efficient SO2 removal and product regeneration. Oxides of Ti, Zr, Hf, V, Cr, Fe, Co, Ni, Cu, Zn, Al, Sn, Bi, Ce, Th, and U were selected as a result of this screening process. Further screening of metal oxides by consideration of their reaction rates with SO2in a flue gas atmosphere showed that oxides of Cu, Cr, Fe, Ni, Co, and Ce would have economically feasible reaction rates with SO2. After further evaluation of factors such as sorption reaction stoichiometry, formation of product layers which affect the reaction rate, and SO, partial pressure over the sorption product, oxides of copper and iron were selected as the most promising. Of the potential metal oxide sorbents for SO2 removal, CuO received the most attention. Laboratory-scale work on copper oxide impregnated into porous alumina has been conducted by the US. Bureau of Mines (McCrea et al., 1970). The results showed that sulfate formation goes essentially to completion at temperatures of about 450 "C. The regeneration by the reducing gases such as hydrogen and methane could be accomplished at much lower temperatures. Work on a copper oxide process has also been conducted by Shell in the Netherlands (Dautzenberg and Nader, 1971). This process, which has been named the shell flue gas desulfurization (SFGD) process, also uses CuO on an alumina support. Two fixed bed units are used in the process. One is used for gas purification and the other undergoes regeneration by use of hydrogen, carbon monoxide, or methane. Both steps are accomplished at about 400 "C. The main object of this study was to test the possibility of using bauxite and red mud for SO2 sorption. Bauxite mineral and red mud-a byproduct of aluminum plants-contain oxides of metals such as iron, titanium, and aluminum which can be considered to be potential sorbents for SOz. Copper impregnated on activated alu1988 American Chemical Society
Ind. Eng. Chem. Res., Vol. 27, No. 3, 1988 435 mina, a well-known sorbent for SO2 and on which some fundamental studies have been carried out (Koballa and Dudukovic, 1970; Yates and Best, 1976; Best and Yates, 1977; Cho and Lee, 1983), was also used for SO2 sorption for comparison. As a contacting device, a fluidized bed reactor was used. The fluidized bed reactor, in addition to some obvious advantages as a gas-solid contactor, allowed use of fine powders of red mud, thus avoiding the formation of pellets which would be required in a fmed bed operation.
where K , = K1K2K3k,(02)1/2(MeO*)0, C A = (SO,), XR = fractional conversion of MeO*, and (MeO*)o = initial concentration of MeO*. The rate of consumption of the active site, MeO*, can be expressed as (Yates and Best, 1976) dX,/dt = Kd(1 - x,) (11)
Theory Reaction Mechanism for Sorption of SO2on Metal Oxides. When sulfur dioxide reacts with metal oxides in
Inserting eq 12 in eq 10 yields
the presence of oxygen at temperatures below the decomposition temperature of the sulfate, the principal product is sulfate and the overall reaction for a bivalent metal can be written as
Furthermore, it is worth adding that for the particle sizes and the conditions involved the intraparticle resistance is negligible by the criteria of pore diffusion (Levenspiel, 1972; Smith, 1981). Modeling of Fluidized Bed. In recent years, a great number of models for fluidized beds have been developed, and extensive information about these models can be found elsewhere (Grace, 1971,1986; Rowe, 1972; Kunii and Levenspiel, 1977; Yates, 1983). The model employed in this research is based on the two-phase theory of fluidization. An approach similar to the Orcutt model (Yates, 1983) with slight modifications is used due to its simplicity and ability of satisfactory predictions of performance. Basic assumptions involved in the derivation of the model are as follows: 1. All the gas in excess of that required to incipiently fluidize the bed passes through the bed as bubbles, which are uniform in size and equally distributed throughout the bed. 2. Interfacial area per bubble volume, a; overall masstransfer coefficient, kbe; and fraction of fluidized bed occupied by bubbles, 6, are all independent of the height of the bed. 3. The gas in bubble phase is considered to be in plug flow. 4. The gas in emulsion phase is considered to be perfectly mixed. 5. There are no particles and, therefore, no adsorption in bubble phase. 6. Gas density is considered to be approximately constant. For a differential height along the bed, mole balance for the bubble phase, assuming pseudo steady state, can be written as -Puos, dCA, = k b e ( C A h - CAe)ClhS, dZ (14)
Me0
+ SO2 + 1/202 F! MeS04
(1)
A great number of mechanisms may be postulated for this reaction (Koballa and Dudukovic, 1970; Yates and Best, 1976; Best and Yates, 1977; Cho and Lee, 1983). One possible reaction scheme that explains the empirically determined rates (Cho and Lee, 1983) is as follows: S02(g)+ MeO* $ z l/202(g) MeO-S02*
+L
+ LO*
MeO-S03*
k,‘
MeO-S02* kd
(2) (3)
LO*
MeO-S03*
+L
A MeS04
(4)
(5)
It should be pointed out that actual reaction scheme may be more complex than that given above because chemistry of metal-oxygen-sulfur systems can involve multiple oxidation states, sulfite formation, disproportionation of sulfites to sulfates and sulfides to sulfate decomposition, and catalytic oxidation of SO2 to SOB. However, at the temperatures involved, reactions such as sulfite formation and catalytic oxidation of SO2 to SO3 are expected to be unimportant, and the above reaction scheme is assumed to be adequate to predict the system behavior. Considering the reaction mechanism expressed by eq 2-5 and assuming that the sulfate formation step is rate determining, then the rate expression can be written as rA = k,(MeO-S03*) Other steps are assumed to be in equilibrium (MeO-S02*) K1 = (MeO*)(S02)
(6)
where Kd is an overall consumption rate coefficient of MeO*. Integration of eq 11 gives 1 - XR = exp(-Kdt)
rA
= K r C A exp(-Kdt)
(12) (13)
Mole balance in the emulsion phase gives (7)
+
(1 - P)UOS,C, - (1 - P)UOS,C, LHfkb,U(CA,
-
c~.)6s, dZ = r A ( 1 - 6)s&f (15)
Integration of eq 14 yields (MeO-S03*) (L) K3 = (MeO-S02*) (LO*) Elimination of concentrations of sorbed species in eq 6-9 gives rA = K1K2K3k,(02)1/2(S02)(MeO*) The concentration of O2 can be taken as constant if it is in large excess; thus, the rate expression becomes (10) r A = K,cA(1 - x,)
CAb
=
CA,
+ (C,
- C4)e-yz
(16)
where * = -Izbea6
P UO
Substitution of eq 16 in eq 15 and then integration give
436
Ind. Eng. Chem. Res., Vol. 27, No. 3, 1988
Substitution of the rate eq 13 in eq 17 gives the emulsion-phase concentration
Combination of eq 16 and 18 gives the bubble-phase concentration as a function of height:
A mole balance a t the exit of the reactor gives CA,dt
= pcA,lZ=H, + (l - P)cAelZ=H,
(20)
Substituting eq 18 and 19 in eq 20 yields the exit gas SOz concentration -X
Table I. Chemical Analyses (Weight Percent) of Sorbents and Alumina Support on Dry Basis alumina bauxite red mud ignition loss 0.1 12.75 7.45 SiQz 0.03 7.23 16.85 A1203 98.5 56.49 16.96 Fe203 0.035 18.46 37.69 Ti02 22.12 4.73 v205 0.008 pzo5 0.005 CaO 1.17 3.76 Na20 0.05 10.27 COZ 1.4 H20 1.59 Table 11. Physical Properties of Sorbents and Alumina SUDDO~~ alumina bauxite red mud particle size, m m (Sauter mean) 0.0738 0.150 0.076 particle density, g/cm3 1.52 2.46 1.16 solid density, g/cms 3.55 2.95 3.12 porosity 0.57 0.17 0.63 surface area, m2 g 84.9 9.79 96.8 av pore radius, 94.4 68.2 79.4
i
Experimental Section Apparatus. The experimental system which consists
where x = VHf
(22)
Hydrodynamic parameters and mass-transfer coefficients required in the models are estimated as follows: The minimum fluidization velocity, Umf,is calculated by Leva's equation (Yates, 1983):
- p,)0.94$.94
dp1.82(P,
umf= (1.1 x 10-3)
(23) p,0.@-Qe8
The terminal velocity of the particles, Ut, is found by using empirical correlation of Kunii and Levenspiel (1977). The average bubble diameter is calculated by using Kunii and Levenspiel's (1977) correlation. The height of the bed is calculated by
The fraction of fluidized bed occupied by bubbles is found from
The fraction of inlet gas entering the bubble phase is calculated as
The interfacial area per bubble volume is a = 6/db
(27)
The overall mass-transfer coefficient is calculated by using Sit and Grace's (1981) equation
which takes into account enhancement of mass transfer due to bubble interactions in freely bubbling beds. The effect of sorption on the mass-transfer coefficient is neglected since the SO2 concentration is low.
of mainly a fluidized bed reactor, gas handling and metering equipment, and a gas chromatograph is shown schematically in Figure 1. The fluidized bed reactor was a stainless steel tube of 5-cm inside diameter and 40-cm length. It was fitted with a perforated brass gas distributor. The distributor was designed according to the method outlined by Kunii and Levenspiel (1977) to give a uniform flow through openings. It had 69 holes of 1-mm diameter. The column was heated by nichrome heating wire, and the temperature in the bed was measured by a thermocouple. The reactor was insulated by glass wool lagging. The gas mixture was heated in a furnace preheater before being fed into the reactor. There was a cyclone at the top of the reactor to collect and return back the entrained particles. The exit gas was taken from the top of the cyclone and was given to the chromatograph (PerkinElmer Model F-20H) after passing through a double-pipe cooler. Materials. Three types of sorbents, CuO impregnated on alumina, bauxite, and red mud, were tested as sorbents. The chemical analyses of the sorbents and their important physical properties are given in Tables I and 11. CuO impregnated on alumina was prepared according to a procedure given by Yates and Best (1976) and Cho and Lee (1983). A concentrated solution of C U ( N O ~ ) ~ . ~ H ~ O in distilled water were prepared. The alumina particles previously activated at 750-850 K for 6 h were immersed in this solution for 48 h. The impregnated alumina was removed from the solution and washed with distilled water to remove any of the solution trapped between the particles. These particles were dried at 353 K for 48 h, and heated to 723 K, and vented to air for further 24 h to decompose Cu(NO& to CuO. The percentage of copper in the final product was determined by idometric methods (Kolthof and Elvin, 1961). This procedure gave about 6% (weight) Cu on the sorbent. Procedure. The system was operated semibatchwise with a single charge of particles. The bed was first loaded with an amount of sorbent to give a static bed height of 5 cm. The flow rate of air was set to the desired value, and after steady-state temperature was reached, SOz was introduced at such a rate as to give an inlet SOz concentration of 0.2% (v/v). The velocity was set equal to 11.1
Ind. Eng. Chem. Res., Vol. 27,No. 3, 1988 437 Temperature Controller TO Atmosphere
TO GOS
oln
P
0
n l
I
1
I
2
h
Rotameters
I
I
SO2
Ail
I
I
5
6
s
Figure 3. Breakthrough curves for bauxite. Symbols show experimental data, and lines show model equation.
1
f
U
I
3 4 Time x
Preheoter
Figure 1. Schematic diagram of the experimental setup.
0-491
K
~ - - - - 5 0 8K
0.2i;
613 K
IO
01 0
I
I
I
I
I
1
2
3
4 s
5
Time x
Figure 4. Breakthrough curves for red mud. Symbols show experimental data, and lines show model equation.
01 0
I
I
1
2
I 3
Time x
I
I
I
4
5
6
7
s
Figure 2. Breakthrough curves for copper oxide on alumina. Symbols show experimental data, and lines show model equation.
cm/s in all runs. Throughout a run, gas samples were taken from the reactor inlet and outlet streams and analyzed by the gas chromatograph using a Porapak Q column. A run was continued until the outlet concentration of SO2 became almost equal to the inlet concentration.
Results and Discussion Experimental data in terms of sulfur dioxide breakthrough curves for the sorbents examined at various temperatures are given in Figures 2-4. Theoretical model eq 21 is also plotted in these figures for comparison. Experimental breakthrough curves are more or less sigmoid in shape, although there is some noticeable deviation from it particularly for bauxite and red mud. This is probably due to differences in the activities of various metal oxides present in these materials. The data show that as the temperature increases the breakthrough curves for all sorbents shift to the right, indicating a higher rate of sorption and sorption capacities
Table 111. Saturation Capacities of the Sorbents saturation capacity, g S02/kg adsorbent sorbent temp, K 6.41 CuO on alumina 384 430 6.71 466 6.43 14.2 529 392 bauxite 3.17 5.09 483 9.24 641 491 3.95 red mud 5.28 508 613 5.98
of sorbents for suA-drdiox ie. It is c-served that at temperatures lower than about 473 K the breakthrough curves converge to a single curve toward saturation. At temperatures higher than about 523 K, the saturation capacities of the sorbents increase appreciably. The saturation capacities of the sorbents were evaluated from experimental breakthrough curves by
where A is the area above the breakthrough curve. These values are shown in Table 111. The sorption capacities are in good agreement with those reported by Koballa and Dudukovic (1970) and Cho and Lee (1983). In the former study, the saturation capacity of a sorbent containing 20% Fe20, is reported to be 0.015 kg of S02/kg of adsorbent at 588 K and 0.008 kg of S02/kg of adsorbent at 533 K. These values are quite close to the values obtained in this
438 Ind. Eng. Chem. Res., Vol. 27, No. 3, 1988 10000
IO
IO
15
25
20 I/T x
30
I
I
I
J
10-
35
lo3, K - ’
Figure 5. Arrhenius plots of K, and Kd for CuO on alumina.
Table IV. Parameters in Arrheniue Equations for K , and KA Kr Kd ~ o - ~ A , , ER, i03A2, ED, s-l kJ/mol s-l J/mol 851.5 29.4 2.04 CuO on alumina 39.3 2.95 4142 38.1 bauxite 201.0 50.6 17.7 2.27 0.287 red mud
study for red mud containing 37.69% Fez03. In the latter study, it is reported that CuO impregnated on alumina has the saturation capacities ranging between 0.018 and 0.03 kg of SOz/ kg of adsorbent for the temperatures between 473 and 698 K. The model eq 21 was solved numerically by computer using a nonlinear regression analysis to calculate the kinetic parameters K, and Kd. The analysis involved the minimization of
1
IOlO
1
15
0
-
25
20 I / T XIO’,
4
30
K-’
Figure 6. Arrhenius plots of K, and Kd for bauxite.
t
5
+\
IO
IO
15
20 l/TxlO’,
2.5
30
K-’
Figure 7. Arrhenius plots of K, and Kd for red mud.
where f j is the value of C,/C, calculated from eq 21 for time t j and Fj is the corresponding experimental point. The method employed in the nonlinear regression analysis was a combination of Gauss and steepest-descent method (Marguardt, 1959; Kuester and Mize, 1973). Arrhenius plots of the rate coefficients K , and Kd are given in Figures 5-7. From these plots, the values of activation energies and preexponential factors in the Arrhenius equations K, = Ale-ER/RTand Kd = A2e-ED/RT were calculated and presented in Table IV. Quite satisfactory fit of the theoretical breakthrough curves with the experimental data shows that the model equation and the kinetic expression give a realistic description of the process. However, it should be pointed out that the rate constants determined for bauxite and red mud are overall values including the contributions of all the active metal oxides present in these materials. The slight deviation of the theoretical breakthrough curves from the data in the early part of the curves is probably due to overestimation of the mass-transfer resistance which considers bubble interactions. This is expected because the value of the mass-transfer coefficient decreases due to bubble interactions, resulting in a higher mass-transfer resistance. As time proceeds, the active sites sorb SO,; hence, the rate of reaction decreases and mass-transfer resistance becomes relatively smaller. This
leads to a better agreement between the experimental and theoretical breakthrough curves for the longer times.
Conclusion Sorption of SO2 on CuO impregnated on alumina, bauxite, and red mud in a fluidized bed showed that these materials can be considered as potential sorbents for SO, in the typical range of stack gas temperatures. The study of the effect of temperature indicated that the efficiency of the reactor increases as temperature increases. The kinetic rate expression for sorption of SO2on metal oxides (Kobda and Dudukovic, 1970; Yates and Best, 1976; Best and Yates, 1977; Cho and Lee, 1983) was verified to be satisfactory for sorption of SO2 on CuO as well as mixtures of metal oxides such as bauxite and red mud by the defining overall apparent rate constant, K,, and overall consumption coefficient of active sites, K,+ The fluidized bed model developed by using the two-phase theory of fluidization is found to represent fluidized bed sorption process adequately. The results of the semibatch experiments can establish the basis for the design of continuous fluidized bed processes. The required residence time of adsorbent particles and thus the flow rate of fresh adsorbent particles in continuous fluidized bed processes can be determined by using the results of the fluidized sorption
Ind. Eng. Chem. Res., Vol. 27, No. 3, 1988 439 processes carried out in this study. Furthermore, model suggested in this work can be helpful in adjusting the design parameters of fluidized beds for the required conversion or retention of SO2.
Nomenclature A = area, m2 AI, A2 = preexpbnential factors, l/s a = interfacial area per bubble volume, CA = concentration of SO2 in air, moi/m3 CAb= concentration of SO2 in bubble phase, mol/m3 C , = concentration of SO2 in emulsion phase, mol/m3 C , = concentration of SO2 at the inlet of the fluidized bed, moi/m3 ,,C = concentration of SO2 at the exit of the fluidized bed, mol/m3 d b = average bubble diameter, m ER,ED = activation energies, J/mol g = gravitational acceleration, m/s2 Hf = height of fluidized bed, m Hmf= height of bed at minimum fluidization, m K,, K2 = equilibrium constant, m3/kg K3 = equilibrium constant, dimensionless Kd = overall consumption coefficientof active sites to absorb so29
l/s
K , = overall reaction rate constant, l/s kbe = overall mass-transfer coefficient between bubble and emulsion phases, m/s k,, kd, k:, kd', k,', k2/, k, = reaction rate constants L = active sites on sorbent for oxygen MeO* = active site of metal oxide for SO2 sorption (MeO*) = concentration of active sites (MeO*)o = initial concentration of active sites MW = molecular weight Q = volumetric flow rate, m3/s rA = reaction rate of SO2, mol/(m3-s) S = saturation capacity, kg of S02/kg of sorbent S, = cross-sectional area of the bed, m2 T = temperature, K t = time, s U,, = minimum fluidization velocity, m/s U, = inlet gas velocity to the bed, m/s Ut = terminal velocity of bubbles W = weight of sorbent, kg x = vHf, dimensionless XR = fractional conversion of active sites to sorb SO2, dimensionless z = distance from the distributor, m Greek Symbols a = ( Ubtyf)/ U,!, dimensionless p = fraction of incoming gas entering bubble phase
v = (kbaQ)/(SUo),m': 6 = fraction of fluidized bed occupied by bubbles, dimensionless emf = minimum fluidization voidage, dimensionless pg = density of gas, kg/m3 pp = particle density, kg/m3 Registry NO.SO2,7446-09-5; CuO, 1317-38-0; AlzO3,1344-28-1; 1314Si02, 7631-86-9; Fe203,1309-37-1; TiOz, 13463-67-7; V205, 62-1; P206,1314-56-3; CaO, 1305-78-8 Na20, 1313-59-3; bauxite, 1318-16-7.
Literature Cited Best, R. J.; Yates, J. G. Ind. Eng. Chem. Process Des. Dev. 1977, 16(3), 347. Bienstock, D.; Field, J. H.; Meyers, J. G. Report 5735, 1961; U.S. Bureau of Mines, Washington, D.C., p 25. Cho, M. H.; Lee, V. K. J. Chem. Eng. Jpn. 1983, 16(2), 127. Dautzenberg, F. M.; Nader, J. E. Chem. Eng. Prog. 1971, 67, 86. Grace, J. R. AIChE Symp. Ser. 1971, 67(116), 159. Grace, J. R. In Chemical Reactor Design and Technology; de Lasa, H. I., Ed.; Martinus Nijhoffi Dordrecht, 1986; p 245. Koballa, T. E.; Dudukovic, M. P. AIChE Symp. Ser. 1970, 73(165), 199. Kohl, A.; Riesenfield, F. Gas Purification, 3rd ed.; Gulf: Houston, TX, 1979; Chapter 7. Kolthof, I. M.; Elvin, P. I. Treatise on Analytical Chemistry; Interscience: New York, 1961; Vol. 3. Kuester, J. L.; Mize, J. H. Optimization Techniques with Fortran; McGraw-Hill: New York, 1973. Kunii, D.; Levenspiel, 0. Fluidization Engineering; Robert E. Krieger: Huntington, NY, 1977. Levenspiel, 0. Chemical Reaction Engineering, 2nd ed.; Wiley: New York, 1972. Lowell, P. S.; Schwitzgebel, K.; Parsons, T. B.; Sladek, K. J. Ind. Eng. Chem. Process Des. Dev. 1971, 10(3), 384. McCrea, D. H.; Fourney, A. J.; Meyers, J. G. J. Air Pollution Control Assoc. 1970, 20,819. Marguardt, D. W. Chem. Eng. Prog. 1959,55(6), 65. Rowe, P. N. "Fluidized Bed Modelling". Proceedings of the 5th Europe/2nd International Symposium on Chemical Reaction Engineering, Amsterdam, 1972. Sit, S. P.; Grace, J. R. Chem. Eng. Sci. 1981, 36, 327. Slack, A. V. Pollution Control Review 4, 1971; Noyes Data Corp., Park Ridge, NJ. Smith, J. M. Chemical Engineering Kinetics, 3rd ed.; McGraw-Hill: New York, 1981. Thomas, A. D.; Davis, D. L.; Parsons, T.; Shroeder, G. D.; De Berry, D. NAPCA PB. 185-562, 1969. Yates, J. G. Fundamentals of Fluidized Bed Chemical Processes; Butterworths: London, 1983. Yates, J. G.; Best, R. J. Ind. Eng. Chem. Process Des. Dev. 1976, 15(2), 239.
Received for review April 23, 1987 Accepted October 28, 1987
