Sie sind auf Seite 1von 8

2224

I n d . Eng. Chem. Res. 1988,27, 2224-2231 for Methanol Conversion to Olefins". Znd. Eng. Chem. Process Des. Dev. 1983, 22, 532. Mihail, R.; Straja, S.; Maria, G.; Musca, G.; Pop, G . "Reply to Comments on "Kinetic Model for Methanol Conversion to Olefins" with Respect to Methane Formation at Low Conversion". Znd. Eng. Chem. Res. 1987,26,637. Mole, T.; Whiteside, J. A. "Conversion of Methanol to Ethylene over ZSM-5 Zeolite in the Presence of Deuterated Water". J. Catal. 1982, 75, 284. NegoitH, C. V.; Ralescu, D. A. "Applications of Fuzzy Sets t o Systems Analysis"; Birkhauser: Basel, 1975. Ono, Y.; Mori, T. "Mechanism of Methanol Conversion into Hydrocarbons over ZSM-5 Zeolite". J . Chem. SOC., Faraday Trans. 1 1981, 77, 2209. Peereboom, M. "Approximate Lumping Applied to the Isomerization of Methylcyclohexenes". Znd. Eng. Chem. Res. 1987, 26, 1663. Perot, G.; Carmerais, F. X.; Guisnet, M. "Carbon-13 Tracer Study of the Conversion of Dimethyl Ether into Hydrocarbons on Silica-Alumina and HZSM-5 Zeolite". J.Molec. Catal. 1982,17,255. Too, J. R.; Nassar, R.; Fan, L. T. "Simulation of the Performances of a Flow Chemical Reactor by Markov Chains". In Residence Time Distribution Theory in Chemical Engineering; Verlag Chemie: Weinheim, 1982. Van den Berg, J. P.; Wolthuizen, J. P.; Van Hooff, J. H. C. "The Conversion of Dimethyl Ether to Hydrocarbons on Zeolite HZSM-5-The Reaction Mechanism for Formation of Primary Olefins". In Proc. 5th Znt. Conf. on Zeolites; Rees, L. V. C., Ed.; Heyden: London, 1980; pp 649-660. Wei, J.; Kuo, J. C. W. "A Lumping Analysis in Monomolecular Reaction Systems". Znd. Eng. Chem. Fundam. 1969,8, 114. Zhu, K.; Chen, M.; Yan, W. "An Engineering Model of a Network of Multicomponent, Reversible Reactions". Znt. Chem. Eng. 1985, 25, 542. Received for reuiew March 1, 1988 Revised manuscript received July 18, 1988 Accepted July 30, 1988

Fan, L. T.; Shenoi, S.; Gharpuray, M. M.; Lai, F. S. "Chemical and Process System Engineering. Applications of Fuzzy Set Theory: Tutorial Note". Paper presented a t the Summer School on Advances in Chemical Engineering Mathematics, Hannover, West Germany, 1985. Frenklach, M. "Computer Model of Infinite Reaction Sequences: a Chemical Lumping". Chem. Eng. Sci. 1986, 40, 1843. Haag, W. 0.;Lago, R. M.; Rodewald, P. G . "Aromatic Light Olefins and Mechanistic Pathways with ZSM-5 Zeolite Catalyst". J . Molec. Catal. 1982, 17, 161. Hoerl, A. E.; Kennard, R. W. "Ridge Regression: Biased Estimation for Nonorthogonal Problems". Technomet 1970, 12, 55. Hoffmann, U.; Hofman, H. "Reaction Engineering. 8. Kinetics of Multicomponent Multireaction Systems and Simplification of Their Description". Znt. Chem. Eng. 1977, 17, 414. Hosten, L. H.; Froment, G. F. "Kinetic Modelling of Complex Reactions". In Recent Advances in the Engineering Analysis of Chemically Reacting Systems; Doraiswamy, L. K., Ed.; Wiley: New York, 1984. Iordache, 0. "Polystochastic Models in Chemical Engineering"; VNU-Science Press: Utrecht, Netherlands, 1987. Corbu, S. "A Stochastic Model of Lumping". Chem. Iordache, 0.; Eng. Sci. 1987, 42, 125. Jardine, N.; Sibson, K. Mathematical Taxonomy; Wiley: New York, 1971. Li, G. "A Lumping Analysis in Mono- or/and Bimolecular Reaction Systems". Chem. Eng. Sci. 1984, 29, 1261. Liu, L.; Tobias, R. G.; McLaughlin, K.; Anthony, R. G. "Conversion of Methanol to Low-Molecular-Weight Olefins with Heterogeneous Catalysts". In Catal. Conuers. Synth. Gas. Alcohols Chem. Herman, R. G., Ed.; Plenum: New York, 1984. Luss, D.; Golikeri, S. V. "Grouping of Many Species Each Consumed by Two Parallel First-Order Reactions". AZChE J. 1975,21,865. Maria, G.; Muntean, 0. 'Model Reduction and Kinetic Parameters Indentification for the Methanol Conversion to Olefins". Chem. Eng. Sci. 1987, 42, 1451. Mihail, R.; Straja, S.; Maria, G.; Musca, G.; Pop, G. "Kinetic Model

Inhibition by Product in the Liquid-Phase Hydration of Isobutene to tert -Butyl Alcohol: Kinetics and Equilibrium Studies
Enric Velo, Luis Puigjaner, and Francesc Recasens* Department of Chemical Engineering, ETS Enginyers Industrials de Barcelona, Universitat Politecnica de
Catalunya, Diagonal 647, 08028 Barcelona, S p a i n

Intrinsic rates of isobutene hydration t o tert-butyl alcohol on Amberlyst-15 particles were measured to establish a rate equation in a solvent-free, liquid-phase system. T h e ranges of temperature and concentration are those likely to be found in a multiphase reactor, such as trickle bed or slurry type. Although inhibition by water may also be present, the effect of TBA is far more significant. Thus, the alcohol is found to inhibit the rate more than expected from the values of the equilibrium constant for the hydration reaction, determined in separate experiments. Various rate expressions, including that derived from the accepted hydration mechanism, are tested t o account for product inhibition. Extreme care has been put in ascertaining the effects of internal and external mass-transfer resistances, by use of suitable derived criteria. Introduction tert-Butyl alcohol ( T B A ) is an important oxygenated octane enhancer that is used to replace toxic lead additives in gasoline (O'Sullivan, 1985). For other oxygenates, such as methyl tert-butyl ether, the technology is already well established. For TBA, however, processes have been filed (Franz et al., 1975; Matsuzawa et al., 1973; Moy and Rakow, 1976), but only a few seem to be in commercial operation (O'Sullivan, 1985; Huls, 1983). A high-yield, catalytic route to TBA is based on the direct hydration of isobutene (iB) contained in refinery C4 streams. The synthesis reaction
0888-5885/88/2627-2224$01.50/0
CH3
(1)

H3C\ ,C=CHz H3C

4-

H20

I e CH3-C-OH I

CH3

catalyzed by acid, is exothermic and reversible at low temperatures (50-90 "C) and highly selective toward the desired product. Reaction 1 involves a multiphase mixture of a hydrocarbon, an aqueous phase, and a solid catalyst. Successful processes depend on how the problem of the limited miscibility of the components is overcome. Existing pro-

0 1988 American Chemical Society

Ind. Eng. Chem. Res., Vol. 27, No. 12, 1988 2225 cesses use either a cosolvent (Delion et al., 1986) or excess TBA (Huls, 1983) to achieve miscibility. Three-phase catalytic contactors, such as trickle-bed reactors (Satterfield, 1975; Herskowitz and Smith, 1983) operating with liquid water and gaseous butene, offer an interesting alternative not yet fully studied. Cosolvent would not be necessary, and relatively low operating pressure could be used. Besides, excess water would shift the equilibrium favorably, and low dissolved iB would reduce byproduct formation by competing reactions of higher order (Levenspiel, 1972). On the other hand, the low solubilities of gaseous iB in aqueous mixtures (Leung et al., 1987a) and the interphase mass-transfer resistances associated with trickle-flow operation (Leung et al., 1987b) may retard the global rate substantially. Furthermore, early rate measurements on ion-exchange catalysts show that both intrinsic kinetics and pore diffusion affect the global rate (Gupta and Douglas, 1967). Previous Work. Amberlyst-15 (A-15), a preferred catalyst for the process, is a sulfonic acid macroporous ion-exchange resin. In preliminary studies (Leung et al., 1986), the hydration reaction was found to be linear in dissolved iB, with significant internal diffusion limitations even for a small bead size (0.45 mm) and low temperatures (30 "C). Using two pellet sizes allowed us to separate the intrinsic first-order rate constant and the effective diffusivity of iB within the particles over a wide temperature range (Leung et al., 1986). Also, the measurement of point rates in a laboratory trickle-bed reactor showed that the wetting efficiency of the catalyst by the liquid flow, as well as the liquid-to-particle mass-transfer coefficient, affected the rate (Leung et al., 1987b). In these prior studies, the reaction was first order since little iB was present kmol/m3). TBA is known to reduce the intrinsic rate since reaction 1 is strongly reversible (Delion et al., 1986). But in contrast, the presence of TBA in the liquid enhances the solubility of gaseous iB exponentially (Leung et., 1987a). This is most important when considering the optimal operation of a multiphase reactor, where a liquid mixture consisting of water and TBA is contacted with the gas. In this regard, Clceres et al. (1988) studied the autocatalytic effect brought about by the enhanced solubility of iB on the optimal operation of a trickle-bed reactor for the process. It is known (Levenspiel, 1972) that reactor-size minimization by product recycle is best for autocatalytic reactions. In the presence of high TBA concentrations, the direct rate would be increased exponentially due the presence of more dissolved iB, whereas the effect of TBA in accelerating the reverse reaction is only linear. Depending on the relative importance of kinetics on the global rate in a trickle-bed reactor, an optimal compromise between liquid recycle and temperature can be found that minimizes the bed-size requirement for a given production (CBceres et al., 1988). The Catalyst. In our studies we use A-15 beads as a catalyst whose pores bad been previously filled with water. The catalyst is a styrene-divinylbenzene copolymer with sulfonic acid functional groups. The particles are composed of agglomerates of gel-type microparticles surrounded by a macroporous matrix (Pitochelli, 1975). Cumulative pore-volume distribution, average pore diameter, and surface area as well as other pertinent characteristics are available in the literature (Kun and Kunin, 1967; Dooley et al., 1982; Leung et al., 1986). We reasoned (Leung et al., 1986) that because of the small size of the microparticles, with Thiele modulus of about the reaction is not retarded much by diffusion within the

L
Figure 1. Apparatus: 1, reactor; 2, reservoir; 3, condenser; 4, thermostated bath; 5 and 6, pumps; 7, wet gas meter; 8, rotameter; 9 and 10,sampling septa; 11, throttling valve; 12, bypass valve; 13, back-pressure valve.

microspheres (Ihm et al., 1988). Therefore, the measured values of the effective diffusivity corresponded to the macropore region. Presently liquid-phase kinetic work on macroporous ion-exchange particles is scarce. Thus, Gupta and Douglas (1967) separated the first-order kinetic constant and the diffusivity of iB on particles of Dowex 50, a gel-type ion-exchange resin. For the reverse reaction (dehydration of TBA), Heath and Gates (1972) and Gates and Rodriguez (1973) studied the catalysis for both geltype and macroporous resins. They found that with excess water the reaction is catalyzed by the hydrated protons freely present about the macropores rather than by localized HS03 groups. Recent studies postulate a micropore-macropore model with different diffusion rates in each region (Ihm et al., 1988). Using a series of macroporous ion-exchange catalysts, Dooley et al. (1982) found that the rates of certain gas-phase reversible reactions (reesterifications) could well be described by a model accounting for diffusion in the macropores and Langmuir adsorption followed by surface reaction in the swollen microparticles. For the case of liquid-phase hydration of linear olefins, Petrus et al. (1984, 1986) studied kinetics and equilibria over macroporous ion-exchange catalysts similar to A-15. For the purpose of evaluating a trickle-bed process for TBA, it is first necessary to have an intrinsic kinetic expression for the hydration reaction in a solvent-free liquid system. The aim of this work is to provide such kinetic information for the range of temperatures and TBA concentrations likely to be found in a trickle-bed reactor. Furthermore, in order to use thermodynamically consistent rate expressions, equilibrium measurements, over the appropriate temperature range, are also necessary. These are also provided here.

Experimental Methods Apparatus. Figure 1 shows the apparatus where rate and chemical equilibrium determinations were made. The reactor, 1, was an AIS1 316 stainless steel tube, 1.21 cm in diameter, that contained the catalyst packing between mesh screens. The reactor was operated liquid-full with B , was recycled recirculation. The liquid, resaturated with i upflow through the catalyst bed. The reactor temperature was controlled by means of water from a thermostat bath, 4, available through pump 6. Temperature readings were made with a Pt-100 resistor inserted into the catalyst. The

2226 Ind. Eng. Chem. Res., Vol. 27, No. 12, 1988
Table I. ODerating Conditions catalyst particle size," mm catalyst mass (dry), kg liquid flow rate, m3/s temperature, K pressure, kPa total liquid vol, m3 iB flow to reservoir, m3/s TBA concn,b kmol/m3 iB concn, kmol/m3 per pass conversion of iB, %
= 1.2-2.2 kmol/m3 in equilibrium runs.

0.124 (4.0-6.5) X (0.56-5.6) X lo4 303-333 iO0-250 3 x 10-3 8.3 X lO* 0-3.2 (1.2-38) X 10-20

"Effective wet size of a dry-sieved fraction, 100-115 mesh.

Table 11. Properties of A m b e r l y ~ t - 1 5 ~ ~ ~ grade 15 dry strongly acidic, macroporous type functionality HSOC physical form beads 393 max operating temp, K moisture content, % <1 50 x 103 surface area, m2/kg 0.36 porosity 4.7 capacity, equiv/kg 24 av pore diameter, nm pellet size, 85% within, mesh 16-30

"Rohm and Haas, Philadelphia, PA. bSee also pore volume distribution (Leung et al., 1986).

recirculation of process liquid is made through a jacketed reservoir, 2 (AIS1 316, 0.003 m3), where iB is bubbled abundantly to ensure saturation. The reservoir was operated at the same temperature of the reactor. During a reaction run, the dissolved iB concentration could be held constant by proper adjustment of the gas flow to the sparger. The reservoir exit was connected to the atmosphere through a reflux condenser, 3, and a back-pressure valve, 13. Exit flow of gas was measured with a wet gas meter, 7. By use of chilled water (2 "C) in the condenser, the carryover of TBA was prevented. Process liquid was circulated with a pump, 5. The line was equipped with a calibrated rotameter, 8, to control liquid flow. This was adjusted by a throttling valve, 11, and a bypass valve, 12. During a run, component concentrations were measured by analysis of samples taken at locations 9 and 10. Precise measurements of liquid flow were made with a stop watch. The liquid flow rates were selected as to obtain 10-20% per pass conversions of iB. Because per pass conversions and operating times were small, TBA accumulation due to reaction could be neglected. Measured operating conditions are given in Table I. In equilibrium runs, the reservoir was fdled with aqueous solutions of known TBA concentration. The reservoir was first saturated with iB, and then the iB flow was stopped and the system was isolated from the atmosphere. Liquid recirculation through the catalyst was then started until constant iB concentration a t all points of the apparatus was reached. At this time, equilibrium concentrations were measured. Analytical Procedures. Rate Measurements. The concentrations of iB were about a thousandth of those of TBA (see Table I). Consequently, the rate of TBA accumulation was very small; thus, it could not be used to measure reaction rate. In contrast, iB concentrations at inlet and outlet of the catalyst bed could be accurately measured by gas chromatography (GC). Usually 10-12 replicate analyses of dissolved iB in the reactor streams were made accurately ( f l % ) . For conversions below 10-20%, the reactor behaves differentially (Massaldi and Maym6, 1969; Vatcha and Dadyburjor, 19861, and the reaction rate can be calculated simply as (2) r = QL(CB; - C,,)/m Gas chromatographic measurements were made with a Shimadzu GC-8A apparatus with flame-ionization detector. For analysis of TBA, isopropyl alcohol was used as the internal standard. For the analysis of iB, an external standard was employed. This consisted of distilled water saturated with iB at 303 K and atmospheric pressure, whose concentration is known precisely (Kazanskii et al., 1959; Leung et al., 1987a). Samples for analysis were directly withdrawn at locations 9 and 10 (see Figure 1)with 10-pL gas-tight syringes. Chemicals. Isobutene with a stated purity of 99% from Linde A.G. (West Germany) was used. tert-Butyl and

isopropyl alcohols, for GC calibrations, were reagent grade from Panreac (Barcelona, Spain). Distilled plus deionized water (0.10 pS/cm2), from laboratory facilities, was used. Catalyst. Reaction rate and equilibrium studies were made with almost perfectly spherical particles of Amberlyst-15 whose properties are given in Table 11. A fine particle size was used in our studies (see Table I). This was obtained by sieving between 100 and 115 mesh. Prior to use, the catalyst was boiled overnight with distilled water and then vacuum filtered to obtain a moist cake. Before filtration, water content was determined and average particle diameter measured by optical microscopy. At the same time, a weighted portion of the moist cake was packed into the reactor and pretreated with H2S04. This was done by passing 1.5 bed volumes of a 10% aqueous H,S04 at a space velocity of 0.001 s-l and then rinsing with distilled water.

Reaction Mechanism The mechanism of olefii hydration catalyzed by mineral acids was studied by Taft and co-workers in early papers (Purlee et al., 1955; Boyd et al., 1960; Nowlan and Tidwell, 1977). For the hydration of linear butenes catalyzed by macroporous ion exchangers, Petrus e t al. (1984, 1986) assumed that the same mechanism applied except that the protons were free around the liquid-filled pores (Gates and Rodriguez, 1973). In the presence of excess H20, the rate-determining step is the formation of the tert-butyl cation.
H3C C ,\ , = C H 2 H3C

k<

H + 3 C , C H 3 ,,

H30+
k I-,

H20

(3a)

H3C

This can revert to iB or it can react reversibly with water to form TBA and regenerate the proton, according to the reaction

For sufficiently low temperatures, the competing formation of the sec-butyl cation by a reaction analogous to eq 3a can be disregarded. Other parallel reactions, such as the acid-catalyzed formation of di-tert-butyl ether as well as dimerization to diisobutenes, do not take place to an appreciable extent. Consequently, the reaction toward the desired product, TBA, is highly selective. Petrus et al. (1984) used the above mechanism in combination with Helfferich's theory on catalysis by ion-exchange resins (Helfferich, 1962). According to this theory, for a very slow reaction, the concentrations of reactants and products within the resin matrix are assumed to be at equilibrium with those in the bulk fluid due to partition equilibria. For the case of propylene hydration, Petrus et al. (1984) even

Ind. Eng. Chem. Res., Vol. 27, No. 12, 1988 2227
Table 111. Weisz-Prater Modulus" for Linear Reversible Kinetics ( d . = 0.124 mm) T,K 103CBmt,kmol/m3 lo3&, kmol/m3 *Lb
303 313 323 333 33 28.4 26.5 20.3 0.71 1.23 2.16 3.64 0.032 0.055 0.077 0.141

Table IV. Minimum Superficial Velocities in Liquid-Full Reactor (dD= 0.124 mm)
T, K
303 313 323 333 1O5k1,m3/kg 4.07 9.61 21.5 50.4 10l0De? m2/s 3.7 5.1
8.0
T~

11.0

0.97 0.96 0.94 0.91

106uL, m/s 4.8 14.0 38.6 114.0

"CA = 3 kmol/m3, C B calculated ~ from K, (Taft et al., 1955), CB,Mt from Leung et al. (1987a). bCalculated from eq 7 with De from Leung et al. (1986) given in Table IV, and r o b corresponding to C A = 3 kmol/m3. pp = 930 kg/m3.

" k and De from Leung et al. (1986). bCalculated from Thiele modulus, #, of our particles.

30 "C

provide values of the adsorption coefficients, Xi,determined in independent experiments. In our case, Xi is defined as
X i = (Ci/CJeq

i = A, W, B

(4)

By use of the steady-state hypothesis for the carbenium ion, the mechanism of eq 3a and 3b leads to the following rate expression in terms of bulk fluid concentrations:
r= k(cBcw - CA/K,) 1 + KwCw

(5)
Taf t -Riesz

where K , is the chemical equilibrium constant based on molar concentrations

K, =

A)e CBCW

(-)(

k-,'k-;

T) XBXW

(6)
3.0
3.2 I O ~ / T . K- '

In the derivation of eq 5, it is assumed that partition of components is linear and reversible, eq 4. Also the concentration of H30+within the pores is constant and zero in the bulk fluid. In eq 5 the observed kinetic coefficient, 12, is related to other constants, as
k = CHsOk-2'KcXA (7)

3.4

Figure 2. Effect of temperature on the chemical equilibrium constant, K,. Comparison with literature values.

In eq 5 the constant Kw is
Kw = k~Xw/k-1'

(8)

Note that the water concentration, Cw is written explicitly in eq 5. This is because for the higher alcohol concentrations of this study (C, = 2-3 kmol/m3), CWis slightly less than its value in pure water (55.5 kmol/m3). On inspection, eq 5 predicts that, in the absence of alcohol and excess water, the reaction is linear in iB; hence no inhibition by reactant is expected. Also, increasing the TBA concentration accelerates the reverse reaction, but since Cw depends on CA, the acceleration is in principle nonlinear. Note also that eq 5 will remain thermodynamically consistent provided that the same value of K, used in it describes the observed chemical equilibrium as well. Mass-Transfer Limitations. In order to check an intrinsic kinetic expression such as eq 5, extreme care was taken to eliminate internal and external mass-transfer effects. In addition to eq 5, alternative semiempirical Langmuir-Hinshelwood rate expressions were also tried. Intraparticle diffusion limitations were assessed by use of the Weisz-Prater criterion for linear reversible kinetics. This is
(9)

calculated for the most stringent case (CA = 3 kmol/m3), where the moduli are largest. aL increases with increasing temperature as expected. The highest aLin Table 1 1 1is below the accepted limit, 0.15 (Levenspiel, 1979). In order to evaluate external mass-transfer limitations, we likewise use a criterion available for a linear irreversible reaction,
kLPa > 10klTl (10)

with k, evaluated from the pseudo-first-order intrinsic kinetic constant, available from prior studies (Leung et al., 1986). The effectiveness factor, ql, was calculated from the Thiele modulus, &, of our catalyst. We prove in the supplementary material that condition 10 is more stringent than the corresponding criterion derived for other types of rate expressions. Hence, it can be used as a sufficient condition for neglecting liquid-to-particle mass-transfer effects. Reaction runs performed by Leung (1986) show that liquid-to-particle mass-transfer coefficients, kLsusr could well be estimated by using the correlation of Dwivedi and Upadhyay (1977). Therefore, we used this correlation to estimate the minimum values of the liquid velocities that meet the requirement of eq 10. These are given in Table IV. In our experiments, the velocities actually used were well above those of Table IV.

We show in the supplementary material that the Weisz moduli for nonlinear rate expressions are smaller than the value of @L given by eq 9. Therefore, the condition, aL << 1, is sufficient to ensure that rates are not restricted by internal mass transfer. Table I11 shows the values of @L calculated for our experimental conditions. These were

Results and Discussion 1. Chemical Equilibrium. In order to make kinetic expressions thermodynamically consistent, we will make use of the measured values of K,. The values obtained are given in Figure 2 for two initial concentrations of TBA (CA = 1.2 and 2.2 kmol/m3). Figure 2 shows also the values of K, for the homogeneously catalyzed hydration (Eberz and Lucas, 1934; Taft et al., 1955). In the low-temperature range, our values are significantly lower than those for the

2228 Ind. Eng. Chem. Res., Vol. 27, No. 12, 1988
70
50

30

O C

I - -

Homo

Catalyst

2 ''

T h i s work, A-15
YX

C,=i.2 kmol

IT'

-3

Figure 3. Effect of cosolvents and catalyst type on the equilibrium constant, K,.

c 0X 1 0 ,

kmol m

-3

homogeneous reaction, with better agreement at high temperatures. In Figure 3, we depict the corresponding values of K,, compared with those determined by Delion et al. (1986) on A-15. They used different cosolvents to achieve an homogeneous liquid mixture. As is evident in Figure 3, for systems with cosolvent, K , can be an order of magnitude lower than in our case, where a dilute solvent-free aqueous solution is present. In the low-temperature range, our K , would be intermediate between those of the homogeneously catalyzed reaction and those measured in the presence of cosolvents. In our case, the excess TBA itself might act as a cosolvent. The widely different values of the equilibrium constant in liquid-phase reactions have been explained quantitatively by Delion et al. (1986) for the hydration of iB and by Colombo et al. (1983) for other reactions also catalyzed by A-15, (the etherification of iB with methanol). These authors point out that the observed value of K, w i l l depend strongly on the nonideality of the liquid solution, recommending that the liquid-phase activity coefficients of the components should be examined. The different values of K, would be the result of abnormally low values of K , in different cosolvents. The activity coefficients, of course, are expected to depend strongly on solute-solvent interactions. Using a predictive method (UNIFAC (Fredenslung et al., 1977)) to estimate yi, Delion et al. (1986) proved that the true thermodynamic constant, K = K , K,, is approximately independent of the cosolvent. In our case, using binary activity coefficients for the pair TBA-water (Hirata et al., 1975) together with the activity coefficients of iB in the binary system iB-water, available from measured Henry's law constants (Leung et d., 1987a), we reach the conclusion that K , is practically constant within our ranges of temperature and alcohol concentration (Velo et al., 1988). From a practical point of view, note that, in our dilute aqueous mixtures, equilibrium will be more favorably shifted to production of TBA than in the presence of cosolvents (Figure 3). As seen in Figures 2 and 3, the values of K, show about the same temperature dependence, regardless of TBA concentration. In practical kinetic expressions, we w i l l take

Figure 4. Plots of r versus CB in irreversible runs (C, = 0). Continuous lines represent first-order intrinsic kinetic constant from Leung et al. (1986).

values of K , or K , at an intermediate TBA concentration (C, = 1.2 kmol/m3) since this choice barely affects the values of the additional kinetic parameters appearing in the equations. The final expressions for the equilibrium constants are K , = exp(3160/T - 6.78) (114

K , = exp(3210/T

2.99)

(Ilb)

As expected for an exothermic reaction, the equilibrium constant decreases with increasing temperature. However, the observed reaction enthalpies differ from case to case. While the standard enthalpy change for the reaction (evaluated for enthalpies of formation in liquid phase) is AHr,298 = -36.1 kJ/mol, the observed AH for the homogeneously catalyzed reaction is -44.3 kJ/mol (upper line of Figure 3). In different cosolvents, AH is seen to change from -27.3 to -33.5 kJ/mol (Delion et al., 1986). In our case, AH = -26.5 kJ/mol. 2. Irreversible Runs. In order to see the effect of iB concentration on the rate, irreversible runs, with CA = 0, were performed at temperatures up to 333 K. In these runs, the concentrations of iB were changed by increasing its partial pressure in the reservoir (see Figure 1). The plots of rate versus CB are given in Figure 4. As seen, the rates are proportional to iB concentrations for all temperatures. The apparent first-order kinetic coefficient, obtained by linear regression through the origin, almost coincides with the slope of the straight lines depicted in Figure 4. These were drawn using the values of the intrinsic kinetic constant obtained previously by Leung et al. (1986) from separation of k 1 and De from rate measurements on two catalyst sizes. Except for experimental error, the agreement is excellent. This result is important because it indicates that for our conditions, both internal and external mass-transfer effects are absent. This corroborates the validity of our preliminary calculations to assess mass-transfer limitations. In other words, the effectiveness factors of our particles would be close to unity (see Table 11).

Ind. Eng. Chem. Res., Vol. 27, No. 12, 1988 2229
Table V. Hydration of Isobutene. Summary of Kinetic Expressions temp, K paraea form metersn 303 313 323 333 k(CBCw - CA/K,) 10% 2.25 2.69 3.01 3.47 1 ' 1 KwCw K, -0.012 -0.015 -0.016 -0.017

1.01

2'

~(cBcW

- c~/K,)

107k

K,
3 '

7.34 15.3 7.34 0.214 7.35 0.099 NC NC NC 17.4 0.510 17.0 0.201 17.2 0.306 1.44

41.8

8 . 1
42.5 1.23 41.5 0.419 42.9 1.90 0.79

13.9 13.2 103.3 2.68


100.0
P
h

v1

va . 7 ~
I
0
0
Y

.
s
V

k(CBCw - CA/K,.) 1 + KACA

l O ' k
KA

0
i
v

303K

,
I

m
0

4/ b

~(CBCW - CA/K,) (1 + KACA)*


k(CBCw- CA/K,)
+ KACA)n

lO'k
KA

0.684 102.0 1.3 1.4

107k
KA

51 b

a See units in Nomenclature. K, taken from chemical equilibrium study, eq l l a . NC = No convergence.

0.25

The fact that the reaction is first order in iB suggests that the reactant does not appreciably adsorb on the catalyst for partial pressures up to 250 kPa. Consequently, if Langmuir-Hinshelwood rate forms are used to fit kinetic data, the inhibitory effect of iB can safely be neglected. Thus, the concentration of iB does not appear in the denominators of the equations of Table V. 3. Effect of TBA on the Reaction Rate. After checking the reaction order, we next study the extent of inhibition by product. For this purpose, reaction rates were measured in the presence of TBA, with concentrations up to 3 kmol/m3 and temperatures up to 333 K. By using the first-order kinetic constant obtained in irreversible runs (Figure 4) and the measured equilibrium constant, K , (Figure 2), it is possible to compare the observed rates with the simple reversible rates calculated by the product, k(CBCw- C A / K , ) ,at corresponding temperatures. The results are presented in Figure 5. Generally, it is seen that the rates are lower than the theoretical reversible values, with larger deviations a t higher temperature. This suggests that in our temperature range the rates are inhibited by TBA more than one would expect from the value of a simple, reverse reaction kinetic constant, k / K , . This means in fact that the equilibrium constant would be smaller than the measured one, this effect being larger a t lower temperatures. If one still wants to fit the data with a simple rate form such as eq 2 (Table V), values of k and K,, different from those found in prior experiments (Figures 4 and 2, respectively), are to be used. These are given in Table v. They were calculated by linear regression analysis of the data. It is then obvious that eq 2, with its new parameters, will predict rates acceptably well, but it will fail to predict equilibrium compositions. On these grounds, the simple reversible equation should be rejected. The use of eq 1 (Table V), derived from the mechanism summarized by eq 3, for fitting our rate data is discussed next. Rearranging eq 5 into a linear expression gives

0.0
0

cA ,

kmol

fi3

Figure 5. Comparison of observed rates and those predicted by simple reversible kinetics.
20

E'

9
10

*.

l5

I
313 K

O L
40 45
C

50 W '
kmol

55

m3

Figure 6. Check on the kinetic equation derived from reaction mechanism, eq 5.

CBCW-CA/Kc
robs

1 =-

+ KW -cw k

(12)

By use of the measured value for Kc,the left-hand side of eq 12 can be correlated linearly with Cw. From the corresponding straight line, k and Kw are readily calculated. In our measurements, Cw could not be changed independently of CA (as no cosolvent was used). However, Cw,

corresponding to each concentration of alcohol, can still be calculated with sufficient accuracy by means of a liquid-phase density correlation for TBA-water mixtures (Perry and Green, 1984). The results of the linear plots of eq 2 are given in Figure 6 and Table V. Negative values of K w that are statistically significant are obtained at all temperatures (see slopes of plot of Figure 6). Inspecting eq 8, which defines Kw, negative values are clearly impossible.

2230 Ind. Eng. Chem. Res., Vol. 27, No. 12, 1988

2or-4 3

./
3K

1 1

"

'

'

/I

'

//

m
1 0
r

35

1 . 03
OBSERVED

0.5
6

'3

R A T E x,10 , k m o l ( k g s)-

Figure 8. Predicted versus observed rates with kinetic eq 3 and 4


(Table V) with n = 1 and 2. Temperatures 303-333 K.
GO
I

50

LO

30T
I

---

Leung et al(198G) This work . _

$
1

10-5

0 1 n

Figure 7. Linear plots for eq 3 (Table V) with n = 1.

In summary, the reaction mechanism put forward by Petrus et al. (1984,1986) to account for the hydrations of linear olefin seems to fail in predicting the intrinsic rates in the case of iB hydration in the presence of large concentrations of product. 4. Accounting for Inhibition by Product. So far, the results indicate that the rate is retarded by TBA more than expected from the rate of the reverse reaction (Figure 5). In order to quantitatively study the product inhibitory effect, semiempirical Langmuir-Hinshelwood-type equations such as eq 3-5 (Table V) were tested. It was assumed that inhibition by TBA is accounted for by means of a simple constant, KA, in the denominators. By contrast, inhibition by reactant can be safely ignored. Furthermore, we force these equations to be consistent at equilibrium by using experimentally determined values of K , taken from eq lla. The rest of the parameters are then obtained by least-squares linear regression. Thus, in order to check eq 3 and 4 (Table V), we plot rate data according to

10-6

31 IO~~T K-' ,

33

Figure 9. Arrhenius plots for fitted parameters.

with n = 1 and 2, respectively. Alternatively, if eq 5 (Table V) is to be tried, there are three unknown parameters (n,K A ,and k ) . In this case, a modified Gauss-Newton searching algorithm (NAG, 1983), using as objective function the sum of squares of deviations, is used. Then, the fit of the rate data with eq 5 gives an exponent n close to, say, 1.5, whereas k is very close indeed to the values reported by Leung (1986). We use these results for arbitrarily setting n at either 1 or 2 in eq 3 and 4 (Table V), respectively. The plot of eq 13 with n = 1is given in Figure 7. From intercepts and slopes of these plots, the values of k and KA as a function of T can be calculated. A plot similar to that of Figure 7 with n = 2 (not shown) provides likewise the values of k and KA. In practice, because of the experimental error, it is not possible to discriminate between the two equations. However, the case n = 1 is obviously easier to handle than

n = 2 in reactor calculations. Figure 8 clearly shows that dispersion of data points does not allow an accurate selection of n. For all temperatures of the study, more than about 80% of the rates is well predicted within an error less than 10% with either equation. Interesting enough, the values found for the first-order constant are practically the same, irrespective of n, and very close to the values given by Leung et al. (1986) and to those obtained in irreversible runs (Figure 4). Figure 9 shows the Arrhenius plots of all fitted coefficients for cases n = 1 and 2 (Table V). This allows the calculation of the corresponding energies of activation. In summary, the hydration of iB to TBA, in the presence of TBA concentrations of 0-3 kmol/m3 with liquid saturated with gaseous iB a t partial pressures up to 250 kPa, can be represented by a rate equation such as k(CBCW - cA/Kc) r = (14) (1 + KACA)"
with n = 1 or 2. The value of K,, independent of n, is given by eq l l a , while k is also independent of n and given by h = exp(15.03 - 8844/T) (15) By contrast, KA depends on n:

Ind. Eng. Chem. Res., Vol. 27, No. 12, 1988 2231

for n = 1

KA = exp(26.6
for n = 2

- 8540/T)

(16) (17)

K A = exp(19.49 - 6602/T)
Acknowledgment

Supplementary Material Available: Formal proofs for the criteria used to assess pore diffusion limitations and liquid-to-particle mass-transfer resistance in a packed bed with nonlinear kinetics (4 pages). Ordering information is given on any current masthead page. Literature Cited
Boyd, R. H.; Taft, R. W., Jr.; Wold, A. P.; Christman, D. R. J . Am. Chem. SOC. 1960,82, 4729. Ciceres, E.; Puigjaner, L.; Recasens, F. Chem. Eng. J. 1988,37,43. Colombo, F.; Corl, L.; Dalloro, L.; Delogu, P. Znd. Eng. Chem. Fund. Am. 1983,22, 219. Delion, A.; Torck, B.; Hellin, M. Ind. Eng. Chem. Process Des. Dev. 1986, 25, 889. Dooley, K. M.; Williams, J. A.; Gates, B. C.; Albright, R. L. J. Catal. 1982, 74, 361. Dwivedi, P. N.; Upadhyay, _ _ S. N. Znd. Enp. Chem. Process Des. Dev. 1977, 16, 157. Eberz. W. F.: Lucas. H. J. J . Am. Chem. SOC. 1934. 56. 1230. Franz; F.; Volkamer; K.; Nestler, G.; Schubert, E. German Patent 2538036 (to BASF), 1975. Fredenslung, A.; Gmehling, J.; Rasmussen, P. Vapor-Liquid Equilibria Using UNZFAC; Elsevier: Amsterdam, 1977. Gates, B. C.; Rodriguez, W. J. Catal. 1973, 31, 27. Gupta, V. P.; Douglas, W. J. AZChE J . 1967,13, 883. Heath, H. W.; Gates, B. C. AIChE J. 1972, 18, 321. Helfferich, I. Zon-Exchange Resins; McGraw-Hill: New York, 1962. Herskowitz, M.; Smith, J. M. AZChE J. 1983, 29, 1. Hirata, M.; Ohe, S.; Nagahama, K. Computer Aided Data Book of Vapor-Liquid Equilibria; Kodanska-Elsevier: Tokyo, 1975. Huls, C. W. Chem. Eng. 1983, Dec 12, 60. Ihm, S.-K.; Chung, M.-J.; Park, K.-Y. Znd. Eng. Chem. Res. 1988,27, 41. Kazanskii, V. S.; Entelis, S.G.; Chirkov, N. M. Z.Fiz. Khim. 1959, 33, 1409. Kun, K. A.; Kunin, R. J. Polym. Sci. Part C 1967, 16, 1457. Leung, P. C. Hydration of Isobutene in Trickle-bed Reactors. Ph.D. Dissertation, The University of California at Davis, 1986. Leung, P. C.; Zorrilla, C.; Recasens, F.; Smith, J. M. AZChE J. 1986, 32, 1839. Leung, P. C.; Zorrilla, C.; Puigjaner, L.; Recasens, F. J. Chem. Eng. Data 1987a, 32, 169. Leung, P. C.; Recasens, F.; Smith, J. M. AZChE J. 198713, 33, 996. Levenspiel, 0. Chemical Reaction Engineering; Wiley: New York, 1972. Levenspiel, 0. The Chemical Reactor Omnibook; Oregon State University Press: Corvalis, 1979. Massaldi, H. A.; MaymB, J. A. J . Catal. 1969, 14, 16. Matsuzawa, H.; Ikeda, M.; Sugismoto, Y.; Uchida, S. US Patent 4 011 272 (to Mitsubichi Rayon), 1973. Moy, D.; Rakow, M. S. US Patent 4096 194 (to Cities Services), 1976. NAG FORTRAN Library Manual; Numerical Algorithms Group Ltd.: Oxford (UK), 1983. Nowlan, V. J.; Tidwell, T. T. Acc. Chem. Res. 1977, 10, 252. OSullivan, D. A. Chem. Eng. News 1985, March 18, 10. Perry, R. H.; Green, D. Perrys Chemical Engineers Handbook, 6th ed.; McGraw-Hill: New York, 1984. Petrus, L.; De Roo, R. W.; Stamhuis, E. J.; Joosten, G . E. H. Chem. Eng. Sci. 1984, 39, 433. Petrus, L.; De Roo, R. W.; Stamhuis, E. J.; Joosten, G. E. H. Chem. Eng. Sci. 1986, 41, 217. Pitochelli, A. R. Zon-exchange Catalysis and Matrix Ejjects; Rohm and Haas: Philadelphia, 1975. 1955, Purlee, E. L.; Taft, R. W.; DeFazzio, C. A. J.Am. Chem. SOC. 77, 837. Satterfield, C. N. AIChE J. 1975, 21, 209. Taft, R. W.; Purlee, E. L.; Riesz, P. J. Am. Chem. SOC. 1955, 77,899. Vatcha, S. R.; Dadyburjor, D. B. Znd. Eng. Chem. Process Des. Dev. 1986, 25, 229. Velo, E.; Puigjaner, L.; Recasens, F. Proceed. XI Simposio Iberoamericano de Catllisis, June 12-17, 1988, Guanajuato, Mgxico, Paper 103.
Received for review February 16, 1988 Accepted June 22, 1988

Partial financial support to this project provided from CAICYT, Spain (Project PA86-0191), is thankfully acknowledged. A personal grant from the Ministry of Education and Science, Spain, to one of the authors (Enric Velo) is appreciated. Nomenclature A, TBA = alcohol, tert-butyl alcohol a, = liquid-solid mass-transfer area, m2/kg of catalyst B, iB = butene, isobutene Ci= molar concentration of species i (=A, B, W) in bulk liquid, kmol/m3 Ci= molar concentration in liquid-filled pores, kmol/m3 De = effective diffusivity of iB, m2/s d, = diameter of a spherical catalyst, m g(C,) = kinetic function, g(C,) = r/k, kmol/m3 AH = enthalpy change of reaction, kJ/kmol KA = inhibition constant for TBA, m3/kmol K , = chemical equilibrium constant, eq 6, m3/kmol K , = constant in terms of mole fractions K = constant in terms of activities Kw = inhibition constant for water, m3/kmol K , = ratio of activity coefficients, K , = y A / y B y W k, = pseudo-first-order kinetic constant, m3/ (kgs) k = second-order kinetic constant, m6/ (kmol-kgs) kh = liquid-to-particle mass-transfer coefficient in a packed bed, m/s kl, k2/ = second-order kinetic constants for forward reactions in (3a) and (3b), m3/(kmol.s) k+ k-2 = second-order kinetic constants in reverse reactions, m3/(kmol.s) m = mass of catalyst, kg n = exponent of denominator, eq 5, Table V QL = liquid flow rate, m3/s r = reaction rate, kmol/(kg of catalyst-s) uL = superficial velocity of liquid, m/s W = water Greek Symbols
yi = activity coefficient of component i 7 = effectiveness factor Xi = partition coefficient of component i (=A, B, W) defined
p,

by eq 3 = catalyst density, kg of catayst/m3 4 = Thiele modulus = (d,/6)(p,k1/De)/2 9 = Weisz-Prater modulus, eq A1 of supplementary material a,, = Weisz - Prater modulus for a linear reversible kinetics, eq 9 Subscripts B = isobutene A = tert-butyl alcohol i = in 0 = out obs = observed e = equilibrium s = external surface of catalyst sat = saturated W = water Registry No. A-15, 9037-24-5; (H3C)&=CH2, 115-11-7;
(H&)&OH, 75-65-0.

Das könnte Ihnen auch gefallen