Ind. Eng. Chem. Process Des. Dev. 1980, 79, Edmister, W. C., Okamoto, K. K., Pet. Refiner, 38(8), 117 (1959a). Edmister, W. C., Okamoto, K. K., Pet. Refiner, 38(9), 271 (1959b). Edmister, W. C., Pollock, D. H., Chem. Eng. Prog., 44, 905 (1948). Graboski. M. S.,Daubert, T. E., Ind. Eng. Chem. Process Des. D e v . , 17, 443 (1978a); 17, 448 (1978b); 18, 300 (1979). Grayson, H. G., Streed, C. W., Sixth World Petroleum Congress, Frankfurt am Main, 111, Paper 20-PD7, p 233, 1963. Hoffman, E. J., Chem. Eng. Sci., 23, 957 (1968). House, G. G., Braun, W. G., Thompson, W. H., Fenske, M. R., "Documentation of the Basis for the Selection of the Contents of Chapter 3-ASTM, TBP, and EFV Relationships for Petroleum Fractions", American Petroleum Institute, Dept. of Refining, Washington, D.C., 1966. Jeter, L. T., Thompson, W. H., Braun, W. G., Fenske, M. R., "Documentation of the Basis for Selection of the Contents of Chapter 2-Characterization of Hydrocarbons", American Petroleum Institute, Dept. of Refining, Washington, D.C., 1965. Lee, B. I., Erbar, J. H., Edmister, W. C., AIChE J., 18, 349 (1973). Lee, B. I., Kesier, M. G., AIChEJ., 21, 510 (1975). Lion, A. R., Edmister, W. C., Hydrocarbon Process., 54(8), 119 (1975). Maxwell, J. B., Bonneli, L. S., "Vapor Pressure Charts for Petroleum Engineers", Esso Res. and Eng. Co., Linden, N.J., 1955.
393
393-396
Maxwell, J. R., "Data Book of Hydrocarbons", Van Nostrand. New York, N.Y., 1950. Nelson, W. L., Harvey, R. J., Oil Gas J., 47(7), 71 (1948). Nelson, W. L., Souders, M., Jr., Pet. Eng., 3(1), 131 (1931). Okamoto, K. K., Van Winkle, M., Pet. Refiner, 28(8), 113 (1949). Okamoto, K. K., Van Winkle, M., Ind. Eng. Chem., 45, 429 (1953). Packie, J. W., Trans. AIChE, 37, 51 (1941). Peng, D. Y., Robinson, D. B., Ind. Eng. Chem. Fundam., 15, 59 (1976). Piroomov, R. S.,Beiswenger, G. A., Proc. Am. Pet. Inst., 10(2), 52 (1929). Soave, G., Chem. Eng. Sci., 27, 1197 (1972). Starling, K. S.,Han, M. S., Hydrocarbon Process., 51(6), 107 (1972). Winn, F. W., Pet. Refiner, 36(2), 157 (1957). White, R. R., Brown, G. G., Ind. Eng. Chem., 34, 1162 (1942).
Received for review June 4, 1979 Accepted F e b r u a r y 4, 1980 F i n a n c i a l s u p p o r t of this work was p r o v i d e d by the Department o f R e f i n i n g o f t h e American P e t r o l e u m I n s t i t u t e .
Simulation of Low-Temperature Water-Gas Shift Reactor Chandra P. P. Singh and Deokl N. Saraf" Department of Chemical Engineering, Indian Institute of Technology, Kanpur-2080 16, India
A rate?equation for water-gas shift reaction over a low-temperature catalyst, similar to that over a high-temperature catalyst, has been used. This rate equation takes into account the effects of temperature, pressure, and age of the catalyst on the catalyst activity. It also considers the reduction in reaction rate due to diffusional resistances. Subsequently, this rate equation has been used in a mathematical model developed for design and simulation calculation of the reactor. Agreement between plant data and calculated values is generally very good.
The reformed gas in an ammonia plant contains a high percentage of carbon monoxide in addition to hydrogen and carbon dioxide. A high-temperature shift reactor (HT), using an iron oxide catalyst and operating at 35G450 "C, is used to convert most of the carbon monoxide to hydrogen and COz. Since high temperature favors high CO content even at equilibrium, the H T effluent still contains about 3% CO. This is usually treated in a lowtemperature shift reactor (LT) which uses a CuO-ZnO catalyst and operates in the temperature range 180-250 "C. The L T exit gases contain less than 0.3% CO, which is eventually removed in the methanator. The total amount of conversion that takes place in an L T reactor is small compared to that in an HT reactor. However, a slightly higher carbon monoxide content at the L T outlet can result i n heavy purge loss or/and high inert content in synthesis gas, both resulting in reduced production of ammonia. This makes the study of the behavior of an L T reactor important from the industrial point of view.
Reaction Rates The first commercial application of an LT catalyst dates back to 1930 (Larson, 1931). The catalyst, however, did not find significant commercial use due to its relatively poor life of about 6 months. The catalyst has subsequently been improved and has found widespread commercial application since the early 1960's. Therefore, the studies pertaining to the preparative (Habermehl and Atwood, 1964; Lombard, 1969; Saleta et al. 1970; Ahmed et al. 1971) and kinetic (Cherednik et al., 1969; Kasaoka et al., 1970; Tsuchimoto et al. 1970; Yureva et al., 1969) aspects of L T catalysts are relatively recent and these are not as ex0196-4305/80/1119-0393$01 .OO/O
haustive as that over the H T catalyst (Ruthven, 1969; Singh and Saraf, 1977). In general, a rate equation similar to that for reaction over an H T catalyst has been found to be suitable for L T catalysts (Ahmed et al. 1971,1972; Cherednik et al., 1969; Kasaoka et al.,1970; Mahapatra et al. 1971). In the present work, the following equation has been used to represent the rate of the shift reaction over the catalyst pellets
r = EffX 2.955
X
1013exp(-20960/Rgn
X
Agf X
Pf(XC0
-
x*co) (1)
where Effis the effectiveness factor which accounts for intrapellet diffusional resistance (Ahmed et al., 1972). Wheeler's (1955) method has been used to calculate Effin the same way as done for the H T catalyst (Singh and Saraf, 1977). External mass and heat transfer resistances have been neglected (Ahmed et al., 1971; Kasaoka et al., 1970; Tsuchimoto et al., 1970; Yureva et al., 1969). A , is an aging factor which accounts for loss in activity of the catalyst with usage. This has been correlated with temperature and age from the data reported for the catalyst (Mahapatra et al., 1971) as log A,f = (4.66 X - 1.6 X 10-6T) X r
Pf accounts for the effect of pressure on the rate of reaction as follows pf = p(0.5- P / 2 5 0 ) In the absence of actual data available on L T catalysts and the similarity with HT, the above expression valid for H T catalysts (Singh and Saraf, 1977) has been used here also: R, = universal gas constant and T is absolute tem0 1980 American
Chemical Society
394
Ind. Eng. Chem. Process Des. Dev., Vol. 19, No. 3, 1980
Table I. Experimental Data and Calculated Results I case inlet dry gas flow rate, Nm3/h 13980 inlet steam to gas ratio, S/G 0.812 pressure, kg/cmZ 16.0 volume of catalyst, m s 18.4 age of catalyst, days 98 outlet
I1
I11
78813 0.546 20.7 56.0 107
92873 0.653 29.3 50.2 215
outlet
outlet
composition of the gas, % (dry basis)
inlet
calcd
exptl
inlet
calcd
exptl
inlet
calcd
exptl
H,
60.6 1.9 17.1 0.2 20.2 193
61.30 0.10 18.56 0.20 19.84 192.64
61.0 0.2 18.6 0.2 19.9 193
56.6 3.1 20.4 0.3 19.7 200
57.80 0.21 22.60 0.24 19.15 218.66
57.4 0.20 22.8 0.3 19.3 220
59.8 1.8 17.5 0.3 20.3 201
60.41 0.22 18.82 0.28 20.27 209.36
60.3 0.2 18.9 0.3 20.3 213
Table 11. Experimental Data and Calculated Results case inlet dry gas flow rate, Nm3/h inlet steam to gas ratio, S/G pressure, kg/cmZ volume of catalyst, m 3 age of catalyst, days
IV
V
VI
94520 0.524 22.4 56.0 120
90280 0.605 21.8 53.0 63 1
88457 0.46 28.6 50.2 196
outlet
outlet
outlet composition of the gas, % (dry basis)
inlet
calcd
exptl
inlet
calcd
exptl
inlet
calcd
exptl
3)
56.5 3.1 20.4 0.2 19.7 202
57.74 0.25 22.60 0.23 19.18 220.04
57.7 0.2 22.6 0.2 19.3 221
56.3 3.2 20.51 0.2 19.7 220
57.56 0.26 22.80 0.22 19.16 217.53
57.2 0.3 23.0 0.3 19.2 217
59.4 3.1 16.7 0.3 20.5 200
60.51 0.25 18.99 0.29 19.96 217.61
60.4 0.2 19.1 0.3 20.0 218
CO,
CH,
N, + Ar temperature, C
perature; xco = mole fraction of CO; x*co = mole fraction of CO in equilibrium conditions = ( x * ~. X * C ~ , ) / ( X * ~ & , ) ; and Keq= equilibrium constant = exp[(9998.22/T - 10.213 + 2.7465 X 10-3T - 0.453 X 10-6P - 0.201 X In T)/R,]. The values 2.955 X 1013and 20960 are the preexponential factor and activation energy reported for the LT catalyst over which the model has been tested (Ahmed et al., 1971) having the following characteristics. Catalyst Specification. Chemical composition: CuO, 33%; ZnO, 66%; real density, 5.09 g/cm3; density of the pellet, 2.575 g/cm3; specific surface area, 27.6 m2/g; porosity, C200 A, 0.208 cm3/g; total porosity 0.243 cm3/g. Description of the Mathematical Model. The low temperature shift reactors are single bed adiabatic reactors for which the following mathematical model has been developed. A differential cross section of the catalyst bed is considered throughout which temperature and composition are assumed constant. Axial diffusion of mass and heat has been neglected. The material and energy balance over such a differential section subject to the above assumptions yield the following equations which describe the composition and temperature of the reaction system along it dE -- -r' _
du G dT--AH r' Xdv C, G x . - x.0
t=-
1
1
0' = 1, , . ., 4)
(4)
"i
The L T reactors generally operate a t moderately high pressures (15 to 30 atm) at which the reacting gaseous system cannot be considered ideal. Calculation procedure similar to that used for H T simulation (Singh and Saraf, 1977) was adopted for solution of the model which yielded
Table 111. Temperature at Different Volume Fractions of LT reactor bed volume (fractional)
calcd
exptl
calcd
0.00 0.25 0.50 0.75 1.00
183.00 188.48 191.11 192.21 192.64
183 189 190 191 193
200.00 211.04 216.05 217.86 218.46
case I
case I1
case I11
exptl calcd
exptl
200 211 215 218 220
201 206 207 209 213
201.00 205.09 207.45 208.71 209.36
Table IV. Temperature at Different Volume Fractions of LT reactor bed volume (fractional)
calcd
exptl
calcd
exptl
calcd
exptl
0.00 0.25 0.50 0.75 1.00
202.00 212.03 217.11 219.24 220.04
202 213 217 219 221
200.00 209.01 214.13 216.50 217.53
200 209 214 216 217
200.00 209.67 214.75 216.95 217.81
200 210 216 216 218
case IV
case V
case VI
composition and temperature throughout the reactor. The size of the integration step has been kept at of the catalyst volume since any reduction in its size did not improve the accuracy significantly. Calculation Program. On the basis of the mathematical model described above, a calculation program in Fortran IV has been prepared for the IBM 7044. The calculation time for checking the performance of a reactor is 12.66 s.
Results and Discussion The calculation results for different reactors operating under different conditions of temperature (456-493 K), pressure (16 to 26.8 atm), composition of the reacting system (CO, 1.9 to 3.1%),catalyst age (98 to 631 days), and steam to CO ratio (15 to 43) are presented along with the plant data in Tables I to IV and Figures 1 to 3.
Ind. Eng. Chem. Process Des. Dev., Vol. 19,
28
t\
I
24 \
3-00
OiL 015 016 017 Reactor bod volume (tractional)
d8
d9
1'0
Figure 1. Carbon monoxide concentration profile in an LT reactor (case IV).
Calculated profile
A
Plant d a t a
106
No. 3, 1980 395
For example, in a 600 tons/day capacity ammonia plant, a 0.1 increase (in absolute terms) in CO concentration could lead to more than 1%increase in methane concentration, purge rate remaining constant, or around 25% increase in purge loss, inerts concentration remaining constant. However, for all the cases considered in the present study including case I, no appreciable difference in measured and calculated values of methane concentration and purge rate was observed in the ammonia synthesis loop (Singh, 1978). It can therefore be concluded that the seemingly large discrepencies in exit CO concentration are due to measurement and round off errors which are expected to be high in the low range of concentration encountered here. The agreement between the measured and calculated composition and temperature at the outlet for all the cases considered does indicate the validity of the L T model. However, this alone is not sufficient to validate this model, particularly in a situation where the equilibrium is closely approached at the exit. It is desirable to compare the measured and calculated values of composition and temperature throughout the length of the reactor for this purpose. Since the compositions at intermediate points are not available, only temperatures could be compared. However, it should be noted that for an adiabatic reactor it is always possible to predict, from thermodynamic considerations, the conversion a t any section if the corresponding temperature is known. Hence, comparing the temperature profile can be taken as equivalent to comparing the compositions as well. Tables I11 and IV show the measured and calculated values of temperatures at various points of the reactors for all the cases considered. Figures 2 and 3 show the calculated temperature profile along with the measured values for case I and 11, respectively. The measured and calculated temperatures agree very well a t all the points except one, Le., at the exit in case 111. This can be taken as an exception which could be the result of measurement error in view of the fact that for all other points the error is within 5%. In addition, these small differences are randomly distributed, showing complete agreement between the measured and calculated temperature profiles. The performance of the L T model and its importance in predicting the overall performance of the ammonia plant may be realized from the effect any change in the L T catalyst volume may have on the overall plant production. For example, if in case IV, the volume of LT catalyst be 30% less than the actual one, the increase in CO concentration is less than 0.16% (Figure l),but the loss in production would amount to around 30 tonslday. The L T model presented here forms a part of the overall ammonia plant simulation which has been used successfully to predict the plant performance. This goes to show that the LT model presented here is adequate representation of the unit.
'8L'v------lSZO
01
02
:3
Reactor bed
OL
05
06
07
Os
09
10
volume ( l r o c t , o n a l 1
Figure 2. Temperature profile in an LT reactor (case I).
0
01
02
04 05 06 07 08 Reactor bed volume ( t r a c t ~ o n a l )
:3
09
10
Figure 3. Temperature profile in an LT reactor (case IV).
The difference in calculated and measured composition (in absolute terms) a.nd temperature at the reactor exit is insignificant for all the cases considered (Table I and 11), but the difference for CO concentration would seem to be very high (maximum 50% for case I) in terms of the measured values. A more realistic way to compare the model results with the observation would be to consider the percent conversion of CO. Looking from this angle, for the above referred case (case I), the plant measurements show that the inlet concentration of 1.9% CO is reduced to 0.2% giving a conversion of 89.5%. The model predicts a CO conversion of 94.7 % which is quite close to observed value. For all other cases if the comparison is made in this fashion, the agreement is within 2.5%. A small change in the value of CO concentration at the LT is reflected in relatively larger change in concentration of methane in the synthesis loop if the purge rate is maintained constant. Alternatively, it may result in large purge losses if the level of inerts is maintained constant.
Conclusions A rate equation has been successfully developed for water-gas shift reaction over low-temperature CuO-ZnO catalyst, using the similarity of the same reaction over high-temperature catalyst (Fe2O3-Cr2O3). This rate equation takes into consideration the effects of pressure, diffusional resistances, and catalyst age, and has been used in a model developed to simulate the performance of LT reactors. The application of the LT model for several reactors gave very good results showing its validity and generality. However, all the reactors on which the model has been tested had the same catalyst and therefore it would require some more work to obtain a truly general
396
Ind. Eng. Chem. Process Des. Dev. 1980, 19, 396-401
rate equation which could be valid for all the L T reactors irrespective of the catalyst make. Nomenclature
C, = specific heat capacity of reacting gaseous mixture,
cal/g-mol K G = volume flow rate, Nm3/h AH = heat of reaction, cal/g-mol of CO reacted j = number used to represent different components P = total pressure, atm r = rate of reaction, cm3 of H2/h g of catalyst r' = rate of reaction, cm3 of H2/h cm3 of catalyst R = gas constant, cal/g-mol K ?k= temperature, K z = mole fraction cy = stoichiometric coefficient (negative value is used for reactants and positive for products of reaction) 7 = age of the catalyst, days Subscripts j = component j CO = component carbon monoxide Superscripts
* = equilibrium
condition
0 = inlet condition
Literature Cited Ahmed, S., Sengupta, A., Bhattacharya, N. B., Sen, S. P., Technology(lndia),
8,218 (1971). Ahmed, S., Sengupta, A., Sen, B., Bhattacharya, N. B., Technology(lndia),9,
301 (1972). Cherednik, E. M., Morozov, N. M., Temkin, M. I., Kinet. Katal., 10, 603 (1969). Habermehl, R., Atwood, K., Am. Chem. Soc.,Div. FuelChem., Prepr., 8(1), 10 (1964). Kasaoka, S., Takeda, S., Sasaoka, E., Maruo, H., Kogyo Kogaku Zasshi, 73(1),
133 (1970). Larson, A. T., US. Patent 1797426 (1931). Lombard, J. F., Hydrocarbon Process., 48(8),1 1 (1969). Mahapatra, H., Ray, N., Sen, B., Technology(lndia),8, 211 (1971). Ruthven, D. M.. Can. J . Chem. Eng., 47,327 (1969). Saleta, L., Such, M., Vagnerova, V., DehtuRopy, 9, 99 (1969);Chem. Abstr.,
72, 160545 (1970). Singh, C. P. P., Ph.D. Thesis, I.I.T. Kanpur, 1978. Singh, C. P. P., Saraf, D. N., Ind. Eng. Chem. Process Des. Dev., 18, 313
(1977). Tsuchimoto, K., Oda, Y., Morita, Y., Kogyo Kogaku Zasshi, 73, 137 (1970). Yureva, T. M., Boreskov, G. K., Gruver, V. Sh., Kinet. Katal., I O , 862 (1969). Wheeler, A., in "Catalysis", Vol. 11, Chapter 2, p 111, P. H. Emmett, Ed., Reinhold, New York, 1955.
Received for review June 8, 1979 Accepted March 12, 1980
Coal Liquefaction Catalysis by Zinc Chloride Melts in Combination with Organic Solvents Edward A. Grens 11," Frank Hershkowitz, Ronald R. Holten, John H. Shinn, and Theodore Vermeulen Energy and Environment Division, Lawrence Berkeley Laboratory, and Department of Chemical Engineering, University of California, Berkeley, California 94720
High conversions of subbituminous coal to cyclohexane- and pyridine-soluble materials have been achieved, at temperatures below the threshold for coal pyrolysis, by treatment in reaction media composed of a zinc chloride melt mixed with a suitable organic solvent. At 250 O C and 35 atm hydrogen, for example, addition of a tetralin phase increased the conversion of Wyodak coal to 68% pyridine solubles, compared to 28% obtained with use of zinc chloride alone. The conversion is accompanied by marked reduction of oxygen content in the treated coal. The hydrogen required is provided partly in the coal itself as well as by hydrogen donors and molecular hydrogen.
Zinc chloride and other Lewis acids have been widely investigated as catalysts for hydrocracking of coal and coal extracts. These catalysts are either used as melts in massive quantities (Zielke et al., 1966, 1979; Struck et al., 1969), or in small amounts impregnated into the coal (Wood and Wiser, 1976). In either case they have been employed at temperatures above the pyrolysis temperature of the coal, that is, above about 325 "C. Under these conditions, thermal decomposition of the coal is the initial step in the conversion process, just as it is in most coal liquefaction technology. At lower temperatures where coal pyrolysis does not occur, high conversions of coal to soluble products have not been obtained by treatment under hydrogen with zinc chloride melts alone (Derencsenyi and Vermeulen, 1975). In the course of investigations, a t Lawrence Berkeley Laboratory, of coal liquefaction with homogeneous catalysts below coal pyrolysis temperature, coal has been treated with zinc chloride melts in combination with organic solvents, a t 250 to 300 "C. This procedure was found to give much greater conversions of coal to extractable 0196-4305/80/1119-0396$01 .OO/O
products than did the use of the melt or solvent separately (Holten and Vermeulen, 1977). This result is in contrast to previous work by Conoco Coal Development Co. that had shown no significant improvement in conversion of coal to liquids when tetralin was added to zinc chloride processing media operated a t 350 "C or higher (Gorin et al., 1968). The use of organic solvents with zinc chloride melts a t lower temperatures has many advantages. These solvents provide hydrogen-donor activity, improve penetration of the melt into the pore structure of the coal, act as a transport vehicle for hydrogen from the gas phase, and in some cases serve as alkylating agents. In addition, under proper conditions, they can extract reaction products from the coal to leave the coal more accessible, the dissolved products more isolated from catalysis of (undesired) subsequent reactions, and the melt less contaminated with reaction residues. This paper presents the results of treatment of subbituminous coals with "massive" or excess quantities of zinc chloride melts in combination with a number of pure organic solvents. Subbituminous coals were studied because 0 1980 American
Chemical Society