Ind, Eng. Chem. Process Des. Dev. 1982, 27, 188-192
188
Effect of Catalyst Loading. The catalyst loading was varied from 0.484 to 1.21 g/L at 69 "C, keeping the initial concentration of o-nitroaniline and hydrogen partial pressure at 0.701 M and 961.1 kN/m2, respectively (Figure 3). As expected, the reaction rate was found to increase linearly with the catalyst loading. Acknowledgment One of the authors (P.B.K) would like to thank University Grants Commission for award of a research fellowship.
I 5
6
7 Catalyst
3
loaalng x gm mi-'
9 10'
1
0
1
1
!2
4
Figure 3. Effect of catalyst loading on reaction rate.
expression (Table 111). A few experiments were also done varying the initial concentration of o-nitroaniline and the values of the specific reaction rate were observed to be essentially constant. Thus it appears that the kinetic data could be correlated by an empirical rate expression: r = kCN Effect of Temperature. The temperature of the reaction was varied from 47 to 74 "C. The rate of the reaction increased significantly with the increase in temperature. The apparent activation energy was found to be 13.0 kcal/mol (Figure 2).
Nomenclature CN = concentration of o-nitroaniline, mol/mL k = rate constant, s-l r = rate of reaction, mol/mL s I = fraction of o-nitroaniline converted into o-phenylenediamine Literature Cited Acres, G. J. K.; Cooper, B. J. J. Appl. Chem. Biotechnol. 1972,22, 769. Cihalik, J. Chem. Llsty 1955, 49, 1167; Chem. Abstr. 1955, 49, 14561. Condit, P. C. Ind. Eng. Chem. 1949, 4 7 , 1704. Dovell, F. S.; Ferguson, W. E.; (Leenfleld, H. Ind. Eng. Chem. Rod. Res. Dew. 1970,9, 224. Greenfield, H. J. Org. Chem. 1983,28, 2434. Hlnkel, L. E.; Ayllng. E. E.; Walters. 1.M. J. Chem. SOC. 1939,403. Meschke, R. W.; Hertung. W. H. J. Org. Chem. 1960,25, 137. Mozingo. R. "Organic Synthesis"; Wlley: New York, 1946; Vol. 26. Satterfleld. C. N. "Mass Transfer in Heterogeneous Catalysis"; Cambridge University Press, CambrMge, Mass.. 1970. Tucker, S. H. J. Chem. Educ. 1950, 27, 489. Yao, H.-C.; Emmett. P. H. J. Am. Chem. SOC.1959,87, 4125. Yao, H.C.; Emmett, P. H. J . Am. Chem. Soc. 1981a,8 3 , 796. Yao, H.C.; Emmett. P. H. J. Am. Chem. Soc. 196lb,83. 799. Yao. H.-C.; Emmett. P. H. J. Am. Chem. Soc.1962,8 4 , 1066.
Receiued f o r reuiew February 9, 1981 Reuised manuscript received September 9, 1981 Accepted September 30, 1981
COMMUNICATIONS Dilution of Catalyst in a Fluidized Bed. Dehydration of Ethanol' The dehydration of ethanol over alumina is studied in an integral (plug flow) and undiluted and diluted fluid bed reactors. At any given conversion level the Selectivity of the intermediate (ether) is always higher for the plug flow reactor than for the fluid bed reactor. Catalyst dilution in the fluid bed keeping residence time unaltered is seen to result in an increase in the formation of intermediate ether without a majar fall in overall conversion. Thus dilution of catalyst results in an improvement in the selectivity of intermediate without affecting the other useful features of fluid bed operation. An optimum catalyst dilution R' (mass of catalyst in the undiluted bed/mass of catalyst in the diluted bed) = 3.67 is shown to exist for the parameter values employed, such that intermediate formation at the bed exit is maximum.
Introduction The concept of catalyst dilution has previously been applied to exothermic reactions carried out in tubular reactors packed with catalyst (Caldwell and Calderbank, 1969). Thus a specific temperature profile can be established in a fixed bed by catalyst dilution. Temperature control poses no problems in a fluidized bed as near-isothermal conditions is one of the main features of this type of reactor. The drawback in a fluidized bed is the reduced
' NCL Communication No. 2660. 0196-4305/82/1121-0188$01.25/0
conversion for the same mass of catalyst employed in relation to a tubular reactor and also the lower selectivity for complex reactions carried out in this type of reactor. The concept of catalyst dilution has recently been applied (Irani et al., 1979) to simple and complex first-order reaction schemes in a fluidized bed reactor on the basis of the model of Kunii and Levenspiel (1968a,b). Dilution of catalyst is seen to result in an improvement in the efficiency, E, of fluidized contacting where 0 < E < 1. This improvement in E partially offsets the decrease in the reaction rate constant which now has to be based on total 0 1981 American Chemical Society
Ind. Eng. Chem. Process Des. Dev., Vol. 21, No. 1, 1982 189
a
I
CALIBRATED
7
CATALYST AN0 SUPPORT
13
PRODUCT O U T L E T
2
MICRO PUMP
8
OUTER HEATINQ JACKET
14
WATER CONDENSER
3
F E E 0 INLET
9
TUBULAR REACTOR
46
THREE WAY STOP COCK
4
SILICA GEL TRAP
10
FLUIDIZED HEATINO OED
16
VENT
6
CAPILLARY FLOW M E T E R
41
THERMOWELL
( 7 LOW TEMPERATURE TRAPS
8
PREHEATER
12
CYCLONE
18
SOAP BUBBLE FLOW
49
@A8 8AYPLCR
ETHANOL
F E E D BOTTLE
METER
Figure 1. Experimental setup (integral reactor assembly).
solids in the bed and thus results in the fluid bed being relatively insensitive (with respect to overall conversion) to catalyst dilution as compared to fixed bed operation. For a complex reaction the contacting efficiencies of the different reaction steps are altered to different extents by catalyst dilution, and hence the ratio of the effective rate constants for the fluid bed is altered with catalyst dilution. This leads to concentration profiles different from that observed in the undiluted catalytic fluid bed. With the total mass of solids in the bed remaining unchanged it has been shown on the basis of the above model that in certain cases a diluted fluid bed displays an improvement in performance for complex successive reaction schemes over the corresponding undiluted fluid bed. It is the object of the present work to experimentally verify the above conclusions. The reaction chosen was the dehydration of ethanol to ethylene with diethyl ether formed principally as an intermediate. Experimental runs were conducted in an integral reactor (free of pore diffusion, temperature, and external mass transfer effects) and in undiluted and diluted fluid bed reactors. At the temperature of operation (338 "C) diethyl ether formation was appreciable and the effect of dilution of catalyst, keeping the residence time in the bed unchanged, on the selectivity of formation of ether and overall conversion of ethanol in the fluid bed was obtained and compared with the corresponding integral reactor results. For the parameter values employed in the fluid bed an optimum catalyst dilution ratio R'was sought to be obtained so that the production of intermediate diethyl ether at the bed exit was maximized. It is important to note that diethyl ether by itself is not of any importance and the reaction system was chosen primarily to indicate the influence of catalyst dilution on the intermediate formation. Reaction System Catalytic dehydration of ethanol on an alumina catalyst, besides being of industrial importance and finding increasing application in fluidized bed operation, has the additional advantage of being a well-studied reaction with analytical procedures well-established. Kinetic studies on ethanol dehydration over alumina show that ether for-
mation occurs in the temperature range 200-350 "C. At higher temperatures ether formation is negligible. There is considerable confusion in the literature (Winter and Eng, 1976) with regard to the exact mechanism followed by this reaction. In the absence of a clear-cut network for the reactions it is obviously not possible to establish the correctness of the proposed model for catalyst dilution quantitatively. On the other hand, there seems to be little doubt that a consecutive step with ether as the intermediate product is one of the steps in the overall reaction scheme. Thus experimental monitoring of the concentration of intermediate as a function of catalyst dilution should provide adequate proof of the qualitative correctness of the model.
Experimental Section Integral Reactor. The experimental assembly for the integral reactor and fluid bed reactor experiments is shown schematically in Figures 1and 2. The stainless steel tube containing the catalyst in the integral reactor was of 3/4 in. i.d. and 13 in. length. The catalyst employed was commercial Flyk alumina (-70 to +80 mesh) supplied by Associated Cement Companies Ltd. and reported to have a surface area of 240-260 m2/g and minimum A1203content of 99.5%. The catalyst was pretreated at 400 "C for 2 h in a nitrogen atmosphere before use. After attaining the required temperature of operation, the catalyst was kept in a nitrogen atmosphere for 'Iz h. The first part of the reaction products was discarded while the system was reaching a steady state. Runs were carried out by employing different flow rates of the 95.6% w/w ethanol and 4.4% w f w water feed. The condensable reaction products were condensed in traps cooled in a calcium chloride-ice mixture and an MK-70 cooling unit in series. The products were analyzed using an NCL-AIMIL chromatograph. The separation of the condensable products was effected in. diameter, 6 f t length 10% on the following column: 'I4 Carbowax 1500 on Chromosorb W. The hydrogen carrier gas flow rate was 50 cm3/min and column temperature was 85 "C. The olefin composition was investigated on a in. diameter, 12 f t length column of silver nitrate in
190
Ind. Eng. Chem. Process Des. Dev., Vol. 21, No. 1, 1982
I
CALIBRATED
2
PUMP
ETHANOL F E E 0 VESSEL
8
SUPERHEATER THERMOWELL
I5
CYCLONE
9
BYPASS COWDEWSER
16
THERMOCOUPLE IN MOVEAOLE
3
PREHEATER
IO
PERFORATED @RID PLATE
17
PULLEY
4
5
VAPORIZER
(I
ALUMIWA CATALYST BED
( 8 'CONDEWSER
THERMOWELL
12
4'$
6
6AS ROTAYETER FOR
N, FEED
13
DISEMSAQIMO SECTION
7
SUPERHEATER
PACKING
I4
PRESSURE TAP
WITH
1.1.
FLUID 8ED REACTOR
19
LOW TEMPERATURE TRAPS
90
UNCONDEWSEO 6 A S SAMPLER
RS. SHEATH
Figure 2. Experimental setup (fluidized bed reactor assembly).
ethylene glycol, after the condensable products were frozen out. The uncondensed ethylene gas was also measured by means of soap bubble flow meter. Preliminary runs without catalyst exhibited negligible conversion and thus the possibility of homogeneous reaction was precluded. Runs with three different catalyst sizes (-70 to +80, -80 to +loo, -100 to +120 mesh) under identical conditions yielded only marginal differences in overall conversion, and thus it was established that for the particle size (-70 to +80 mesh) employed in the kinetic runspore diffusion effecta were not prevalent. The method of Ford and Perlmutter (1963) was employed to check the partial pressure gradient AP/P, and establish that for the runs in the present study AP/P, was not appreciable, thus showing the absence of external mass transfer effects. Fluidized Bed Reactor. For the fluidized bed experiments Flyk alumina catalyst of the size -70 to +120 mesh was employed, since at the operating feed rates elutriation occurred to a significant extent with finer mesh size of particles. Initial runs were carried out to adjust the ethanol feed rate so that appreciable elutriation did not occur and the quality of fluidization was good. A feed rate of 1.9 L/h was fixed upon and kept constant throughout the runs so that the hydrodynamics of the bed remained unaltered in tenns of the initial feed rate. A catalyst mass of 2.2 kg was employed in the undiluted bed. A variety of solid inert diluenb were tested for bulk density and fluidization characteristics similar to the alumina catalyst employed, as well as inertness to the reaction. These included pure silicon, sea sand, pumice, silica gel, glass, etc. Whereas some of the materials (e.g., pumice, silicon) showed slight activity and were not totally inert at the temperature of operation, others, such as silica gel, possessed a bulk density at great variance with that of the catalyst. Washed and dried glass powder (-120 to +200 mesh] was found to be relatively inert and also have a bulk density (p, = 1.10 g/cm3) in closer agreement with alumina (pa = 0.94 g/cm3) than some of the other inerts. In order to eliminate segregation of catalyst and inert in the bed, glass powder of a finer size (-120 + 200 mesh) was employed as inert along with the alumina catalyst in the fluid bed. The small difference in bulk densities was compensated for by
I
I
TEMPERATURE : 338.C
I
I
15
30 W/FA,
-
I
45
Figure 3. Conversion of ethanol to ethylene and ether in the integral reactor.
keeping the residence time in the bed unchanged. Thus, in the case of the undiluted bed
In the case of the diluted bed, a weighted mean of the bulk densities pa and p, was considered in the denominator of eq 1 and thus the total weight of solids in the bed was taken such that the residence time 7 was unaltered with dilution of catalyst in the bed. The catalyst dilution ratio R' was defined as mass of catalyst in the undiluted bed R'= mass of catalyst in the diluted bed (2) and runs were taken employing different catalyst dilution ratios and maintaining the temperature in the fluidized bed at 338 O C . Results and Discussion The experimental results obtained in the integral reactor are presented in Figure 3 in terms of the moles of the
Ind. Eng. Chem. Process Des. Dev., Vol. 21, No. 1, 1982 101
Table I. Results of the Undiluted and Diluted Fluid Bed Runsa mol converted catalyst to etherlmol of reading dilution ethanol fed to no. ratio, R ' the reactor
0.07 0.09 0.10 0.12 0.13 0.15 0.16 0.15
1.0 1.40 1.83 2.20 2.51 3.03 3.67 4.19
1 2 3 4 5 6 7 8
mol of ethylene formed/mol of ethanol fed t o the reactor
0.61 0.53 0.46 0.40 0.35 0.27 0.22 0.20
a Catalyst size, -70 to +120 mesh; catalyst mass for undiluted bed, 2200 g; temperature, 338 "C;feed, 95.6% w/w ethanol + 4.4% w/w water; feed rate, 1.9 L/h.
v) Y
d 0
,
I
I
2
a TEMPERATURE
.35
:
/
I
I
0.50
0.65
OVERALL
CONVERSION
catalyst dilution has no effect on the ratio R/S. At the operating temperature the possibility of the reaction proceeding exclusively by parallel steps is thus precluded and the reaction proceeds either by the consecutive or simultaneous consecutive-parallel pathways. Even if a reaction scheme such as
R'- i.0
338.C
OF E T H A N O L
, I
OPTIMUM
0 ----)
Figure 4. Selectivity of ethanol conversionto ether-overall ethanol conversion plots for the integral reador, undiluted and diluted fluid beds.
principal products diethyl ether and ethylene formed per mole of ethanol fed to the reactor, as a function of the time factor W / F A p The results for the undiluted and diluted fluid beds at the same temperature are presented in Table
n
\
S
were assumed, catalyst dilution should have no effect on product distribution. In view of the definite influence of catalyst dilution, the existence of a consecutive step cannot be ignored. The most general scheme would therefore appear to be
I. The experimental results obtained in the integral, undiluted, and diluted fluid bed reactors are presented in Figure 4, which shows a modified comparative selectivity-conversion plot for formation of the intermediate diethyl ether. The integral reactor corresponds to plug flow conditions and hence at any given level of conversion the corresponding selectivity for ether will always be higher than for the fluid bed. Here selectivity for ether is represented in terms of moles of ethanol converted to ether per mole of ethanol converted overall. As the moles of ether formed have been experimentally obtained for different levels of conversion, from the stoichiometry of the
ZR\ A-S
This, however is speculative, and it is not the object of the present communication to establish the precise network for this complex scheme. The related problem of obtaining the optimum catalyst dilution ratio for intermediate formation for a given throughflow and residence time is considered in Figure 5. Here ether formation at the bed exit is plotted for various catalyst dilution ratios with the other hydrodynamic conditions and bed temperature remaining unchanged. The
192
Ind. Eng. Chem. Process Des. Dev. 1982, 21, 192-195
product of selectivity and corresponding conversion, as obtained from Figure 4, also corresponds to ether formation a t the bed exit. From the figure it is seen that a catalyst dilution ratio of 3.67 is the optimum for formation of intermediate product ether at the bed exit and at this dilution ratio ether formation is increased more than twofold (7% to 16%) over that obtained from the undiluted bed, with the advantage of lesser catalyst being employed. Thus, when the intermediate is the desired product, as is the case in most complex reactions, catalyst dilution of the fluid bed can help to overcome the main drawback of a fluid bed, which is lower selectivity for a given level of conversion, without affecting the other useful features of fluidized bed operation. Nomenclature E = contacting efficiency in a fluid bed, dimensionless FAo = molar feed rate of reactant A, g-mol/h P = partial pressure of component in the bulk phase, atm d’= partial pressure gradient of component between bulk phase and catalyst fluid interface, atm R‘ = mass of catalyst in the undiluted fluid bed/mass of catalyst in the diluted bed, dimensionless
Vo = volumetric flow rate, cm3/s W = mass of solids in bed, g Greek Letters = bulk density of alumina, g/cm3
pa
pg = T
bulk density of glass, g/cm3
= residence time in the bed, s
Literature Cited Caldwell, A. D.; Caklerbank, P . H. Br. Chem. Eng. 1969, 74, 1199. Ford, F. E.;Perlmutter, D. D. AIChE J . 1963,9 ,371. Irani, R. K.; Kulkarni, B. D.; Doraiswamy, L. K. I n d . Eng. Chem. Process D e s . D e v . 1979, 18, 848. Kunil, D.; Levenspiel, 0. Ind. Eng. Chem. Fundam. Ifi68a. 7, 446. Kunll, D.: Levensplel, 0. Ind. Eng. Chem. Process Des. Dev. 1968b, 7, 481. Winter, O.,Eng, M. T. Hydrocarbon Process. 1976, 55, 125.
R. K. Irani B. D. Kulkarni L. K. Doraiswamy* S. Z. Hussain
National Chemical Laboratory Poona 411008, India Indian Institute of Technology Bombay, India
Received for review September 4 , 1980 Revised manuscript received July 6, 1981 Accepted July 24, 1981
Prediction of the Liquid Viscosity for Petroleum Fractions The transport properties of petroleum fractions can be predicted via a modified corresponding state method proposed by Ely and Hanley of the National Bureau of Standards. The basic idea behind the transport procedure is that the properties of a mixture can be equated to those of a hypothetical pure substance, whose properties are, in turn, determined via corresponding states from a reference fluid. The NBS model, after extensive comparisons with experimental data for pure hydrocarbons treated as undefined hydrocarbon mixtures, was extended to the prediction of the viscosity of petroleum fractions. A large variety of experimental data for typical American crude oils and various crude oils from the major oil-producing areas including Arabia, the Persian Gulf, and North Africa were used to test the method. The agreement between the theory and experiment for the viscosity of the fractions studied was less than 7 % .
Introduction Very recently, Ely and Hanley of the National Bureau of Standards (1981) reported a corresponding states, conformal solution procedure to predict the transport coefficients of nonpolar fluids and mixtures. The approach has several attractive features; namely, it is predictive, it can be applied to a wide variation of chemical types over a wide range of thermodynamic states from the dilute gas to the dense liquid, and the number of mixture components is unrestricted, in principle. The only input data required are the critical parameters, molecular weight, Pitzer’s acentric factor, and (for thermal conductivity) the ideal gas specific heat. The purpose of this note is to point out that the method seems very appropriate to estimate the properties of petroleum fractions if we regard a fraction as a hypothetical pure substance and estimate the necessary input parameters from the usual characteristics-the specific gravity at 60 OF and an average boiling point. The viscosity (7) of the fraction, designated by subscript x at a density ( p ) and temperature ( T ) is given in terms of the viscosity of a reference fluid, designated by subscript 0, evaluated at the corresponding density and temperature. The equation is (Hanley, 1976; Mo and Gubbins, 1976) 0196-4305/82/1121-0192$01.25/0
where
with A4 the molecular weight. The viscosity thus involves scaling ratios f and h which are, in general
vx*,%)
fx,o = ( T X CTO‘)~,,O( / Tx*,
(3) (4)
Here superscript c denotes the critical value and superscript * denotes reduction of the variable by the critical value: V is the volume (= l / p ) . The functions 0 and @ are the shape factors expressed in terms of the Pitzer acentric factor, w , via functions of the form ~x,o(~x,Vx,ux) = 1 + (ax- W O ) F ( T ~ , V , ) (5)
@ x , o ( T x , V x , ~= ~ ) 1+
(ax-
OO)G(T~,V,) (6)
where F and G are universal functions reported, for ex0 1981 American Chemical Society