You are on page 1of 6

Ind. Eng. Chem. Res.

1992,31, 1035-1040
Kautz, K.; Kirsch, H.; Laufhiitte, D. W. Spurenelementgehalte in Steinkohlen und den daraus entstehenden Reingassauben. VGB Kraftwerkstech. 1975,55 (lo), 672-6. Knbzinger, H. Benetzung im festen Zustand-Ein neuer Weg zur Herstellung uon oxidischen TrEigerkatalysatoren; Dechema: Frankfurt, June 1,1990. Linnros, B. The Crystal Structure of LiMo02Asz0,. Acta Chem. Scand. 1970,24, 3711-22. Pertlik, F. Structure Refinement of Cubic Asz03(Arsenolithe) with Single-Crystal Data. Czech. J. Phys. 1978, B B , 170-6.

1035

Rademacher, J.; Borgmann, D.; Hopfengiirtner, D.; Wedler, G.; Hums, E.; Spitznagel, G. W. X-Ray Photoelectron Spectroscopic (XPS) Study of DeNO, Catalysts after Exposure to Slag Tap Furnace Flue Gas. Appl. Catal. 1992, in press. Russell, A. S.; Stokes, Jr., J. J. Surface Area in Dehydrocyclization Catalysis. Ind. Eng. Chem. 1946, 38, 1071-4.

Received for review May 13, 1991 Revised manuscript received August 1, 1991 Accepted October 14,1991

Intrinsic and Global Reaction Rate of Methanol Dehydration over 7-A1203Pellets


Gorazd BerEiEt and Janez Levec*J
Department of Catalysis and Chemical Reaction Engineering, Boris KidriE Institute of Chemistry, and Department of Chemical Engineering, University of Ljubljana, 61 000 Ljubljana, Slovenia, Yugoslavia

Dehydration of methanol on y-Al,O, was studied in a differential fixed-bed reactor at a pressure of 146 kPa in a temperature range of 290-360 "C. A kinetic equation which describes a Langmuir-Hinshelwood surface controlled reaction with dissociative adsorption of methanol was found to fit the experimental results quite well. Coefficients in the equation follow the Arrhenius and the van't Hoff relation. The calculated value for the activation energy was found to be 143.7 kJ/mol, while calculated values for the heat of adsorption of methanol and water were 70.5 and 42.1 kJ/mol, respectively. The measured global reaction rates for 3-mm catalyst particles were compared to those calculated by means of intrinsic kinetics and transport processes within the particles. A reasonable agreement was found when the effective diffusion coefficients for reaction components were calculated using a parallel-pore model assuming that only Knudsen diffusion is important.

Introduction Catalytic dehydration of methanol over an acidic catalyst (e.g. y-A1203) offers a potential process for dimethyl ether (DME) production, which is used as an alternative to freon spray propellants. In the MTG process, as has been described by Chang et al. (1978),the first reactor performs such a reaction. The open literature provides no information on kinetic equations which can be used successfully in designing a commercial reactor. From the patent literature (Woodhouse, 1935; Brake, 1986) it can be concluded that reaction takes place on pure y-alumina and on y-alumina slightly modified with phosphates or titanates, in a temperature range of 250-400 "C and pressures up to 1043 kPa. The kinetics of methanol dehydration on acidic catalysts has been studied extensively resulting in different kinetic equations. A summary of the published equations is presented in Table I. Most of the equations, i.e. eqs 4-9, have been derived from the experiments conducted in conditions not found in an industrial reactor. The experiments were mainly performed with mixtures of methanol, water, and nitrogen at low vapor pressures. Since water produced during the reaction considerably retards the reaction rate, the derived rate equations have, more or less, a semiempirical character and are not suitable for the industrial reactor design, where reaction takes place at high conversion levels. The outlet component concentrations correspond to the equilibrium values. However, the rate equations (1)-(3) in Table I, which were derived for an acidic ion exchange resin as a catalyst and are based on the Langmuir-Hinshelwood (L-H) or the Eley-Rideal (E-R) mechanism, can be used for design purposes after a reversible term is introduced into the driving-force term. The aim of this work was to determine an intrinsic rate equation which can be used to model the global reaction
t Boris

Table I. Summary of the Published Rate Equations ref eauation

Kallo and Knozinger, 1967

-rM

=k

CM1I2

+ k2Cw +
(5)'Vb

Sinicyna et al., 1986; kKM2CM2 Gates and -rM = (1 + KMCM K w C W ) ~ Johanson, 1971 Figueras et al., 1971 kKMCM'I2 -rM = 1 + KMCM1/' KwCw

~KMCM (7)' (1 + K M C M ) ~ Schmitz, 1978 -rM kl + kzCM Wb Rubio et al., 1980 -rM = klCM1lz kzCwl/z (9Y 'Acidic ion exchange resin as catalyst. bAlumina or silica-alumina as catalyst.
Than et al., 1972
-rM

rates in a pilot-plant reactor where 3-mm catalyst particles were used. In order to calculate the global reaction rate the effectiveness factor must be known. Since the intrinsic kinetic equation is highly nonlinear the effectiveness factor can be calculated only numerically.

* University of Ljubljana.

KidriE Institute of Chemistry.

Experimental Section Catalyst. A Bayer SAS 350 -pAl,O, catalyst support in the form of 3-mm spheres was employed as a catalyst. In order to avoid the intraparticle resistances, spheres were

o a a a - ~ a s ~ ~ ~ ~ ~ ~ s ~ ~ -0i 1992~ $ o ~ . oChemical Society o ~ American o / o

1036 Ind. Eng. Chem. Res., Vol. 31, No. 4, 1992

S u p e r f i c i a l velocity, c m / s

>
JZ

50.00 3.00 1 ' 1

1 ' 1 1 1 ' 1 1 1 ' 1 1 1 ' 1 ' " ' 1 ' " 1 ' 1 1 1 1 ' 1 " 1 ' 1 1

100.00

150.00

200.00

250.00 0.50

T=360"C

T=340C, dP=0.87mm 55wCHsOH

\ t

<

10.30

in the differential reactor,varied from 0 to 3 "C, dependent on the outlet conversion and inlet concentration of methanol. The inlet temperature was adjusted 90 that the mean temperature in the differential reactor was kept constant for all experiments. The maximum systematic error (Massaldi and Maymo, 1969) can be calculated when an adiabatic operation of the differential reactor is assumed. At measured conversions, the calculated systematic error was about 3%. Since the reactor was not operated adiabatically, the true error was even smaller.

1 ;
0.20 0.10

E
L

,g

8
LE

0.00

Particle size, mm Figure 1. Determination of experimental conditions where external


and internal transport resistances could be neglected.

0.1

Fo.00

crushed and sieved. Three different particle sizes were used 0.87,0.37, and 0.17 mm. Experiments showed that within the particles of 0.17 mm in size the intraparticle resistances are negligible. Reactor. The experiments were carried out in a differential reactor (8-mm i.d.) in a temperature range of 290-360 "C. The pressure was kept constant at 146 kPa. The reactor was operated free of interparticle heat and mass resistances (Figure 1). The inlet concentrations of reactants were varied between 15 and 90 mol % for methanol, and from 0 to 50 mol % for water. In this way we simulated the conditions which are found at high methanol conversions in a commercial reactor. In fact, the produced water strongly retarded the reaction rate. As an inert gas nitrogen was used. The total inlet volumetric flow rate was kept constant a t 65.3 cm3/s. Methanol or a methanol-water mixture was fed into the reactor by means of a Beckman 114M solvent delivery module; nitrogen flow was controlled by means of an MKS 246 flow controller. Methanol was evaporated in a specially designed evaporator that consisted of a 5-m long SS coiled tube (1/4-in. 0.d.). The temperatures of the evaporator and reactor oven were controlled by microprocessor-based controllers and were kept within *0.3 "C. In order to achieve the desired conversions, 2-15%, the mass of the catalyst was varied between 0.2 and 2.4 g. A database of more than 400 points was obtained. Analysis. The analysis was performed by means of a HP 5890 GC connected to the reactor outlet. The GC conditions were as follows: a 230-cm X l/s-in. column packed with Porapak T (100/120 mesh). The carrier gas He (the flow rate through the column was 18 mL/min and through the reference 25 mL/min). The detector was a thermal conductivity detector (TCD) set in the low position. The oven temperature was 110 "C for the first 17.5 min, increasing at a rate of 40 "C/min until 130 "C was reached. The duration of the analysis was 25 min. A gas sample (0.25 mL) was injected into the GC through a sampling valve which was kept at 110 "C. The reactant outlet-gas compositions were determined by the calibration curve for each component. The reaction rates were calculated on the basis of the measured conversions, inlet flow rates of methanol, and the mass of catalyst in the differential reactor. The estimated error for the rate was within &7% (BerEiE, 1990). The observed temperature rise, due the reaction progress

Results and Discussion Intrinsic Rate Equation. The published rate equations for dehydration of methanol to DME over alumina and acidic ion exchange resins are listed in Table I. It is evident that in almost all rate equations for dehydration the reaction rate is proportional to the square root of the methanol concentration. This indicates that the dehydration reaction undergoes dissociative adsorption of methanol on the catalyst surface. In the derivation of rate equations for the dehydration of methanol, the L-H concept, which has been further developed by Yang and Hougen (1950), was applied. With the assumption that the surface reaction is a controlling step and that the dissociative adsorption of methanol on the surface of yA1203is taking place, the L-H model can be represented in the following form (BerEiE, 1990):
-rM = k&M2(CM2 - cWcE/K) (1+ 2(KMCM)l/' KWcW)4

(10)

The adsorption term for DME in the denominator of eq 10 was neglected since the DME adsorption constant was too small compared to the adsorption constants of methanol and water (Gates and Johanson, 1971). Since the rates of dehydration with pure methanol or methanol-water mixtures were measured in the differential reactor, the concentrations of produced dimethyl ether were low. The reversible term in eq 10 has negligible effect on the value of driving force; therefore this equation can be compared to those listed in Table I, which all assume the reaction is irreversible. The reversible term of the driving-force term (eq 10) consists of three factors: concentrations of water and DME, respectively, and the equilibrium constant which takes values from about 7 to 11 (at conditions employed here). The mole fraction of water was varied from 0 to 0.5, while the mole fraction of DME formed during the reaction course was always in the order of a few hundredths (since methanol conversion is low). It was estimated that at the experimental conditions the reversible term might change the value of the driving-force term not more than 0.5% in a worst case (typically less than 0.1%). Since the following inequality applies CM2>> CwCD/K, what is implied is that the driving-force term is mainly determined by the concentration of methanol in the feed and does not depend practically on the methanol conversion. On the other hand, this term is important when equations are used for the modeling of an integrally operated industrial or pilot reactor (where Cw = CDand at the reactor outlet Cw > CM). In that case the reversible term along the reactor length increases and consequently reduces the driving-force term. The following criteria were used to distinguish between the kinetic equations shown in Table I: the obtained constants should be positive numbers, the equation gives the smallest values for least squares residuals, the coefficients in the equation must follow the Arrhenius and van't Hoff relation, and the equation should be applicable for predicting behavior of an integral reactor.

Ind. Eng. Chem. Res., Vol. 31, No. 4,1992 1037


Table 11. Residuals Obtained with Nonlinear Regression of Dehydration Rate Equations with Our Experimental Results" residuals at given T ("C) 290 "C 340 "C 320 O C 360 "C eq 0.021b 1 0.7792 0.407 0.909 0.250b 1.307 1.499 2 0.639 0.558 0.026b 0.621 3 0.380 5.179 O.17gb 6.862 4 7.585 1.972 0.004b 1.816 1.452 5 3.670 0.013b 5.565 5.623 6 0.021* b 0.021* 0.037* 0.023* 7 0.019' 0.034* 0.024* 0.013* 8 0.002*b 0.0031 0.015* 9 0.006* 0.690 0.567 0.02tjb 10 0.386
"Values marked with an asterisk indicate that only data for an inlet mixture of MeOH-N2 were used. Experiments were carried out with mixtures of MeOH-N2 only.

1.56E-003

~~~~

1 .60EL003

'

1.64EL003

"

~~~*~~

1 .6SEL003

'

Equations 1-10 were compared to the experimental results for each temperature, applying the Marquardt nonlinear regression method (Duggleby, 1984). Since, in the beginning, all nonlinear iteration procedures require approximate values of parameters, some of the equations have to be transformed into linear forms. With the use of a linear and robust linear regression (Rousseeuw and Leroy, 1987),we obtained values of parameters which were subsequently used as starting values of parameters in the Marquardt minimization algorithm. Convergence was fast and independent of starting approximations of parameters in almost all equations; varying starting values of parameters for a few orders of magnitude have not influenced the results. Reaction rates were calculated with the mean concentrations of components in the differential reactor. Proportional weights were used. The equations which do not include the term of water concentration were tested only with the data obtained by the methanol-nitrogen inlet mixture. Results are summarized in Table 11. From this table it is evident that eqs 1,3, and 10 have the smallest and almost equal sums of least squares. Therefore, an additional test was necessary. From Figure 2 it can be

1 /T
Figure 2. Arrhenius plots for constants obtained by nonlinear regression for rate eqs 1, 3, and 10.

concluded that the constants obtained for eqs 3 and 10 agree reasonably well with the Arrhenius and van't Hoff equation and can therefore be considered as the most favorable rate expressions. With respect to the optimization algorithms, eqs 3 and 10 are nondistinguishable, since they represent the same target function; thus
m X I 2

Y=

-X2X,/K)

(1 + 2 ( B x 1 ) ' / 2

+ DX&4

(11)

In the case of eq 3, coefficient A is equal to the product of coefficients A and B obtained for eq 10. Coefficients B and D have the same values as is evident from Table III. Small differences are obtained by neglecting the reversible term in the case of eq 3. On the basis of statistical criteria alone, we cannot distinguish between the L-H mechanism and the E-R mechanism for dehydration of methanol on yA1203. Furthermore, when calculated values of the ac-

Table 111. Optimum Parameter Set for Tested Equations Obtained with Nonlinear Regression (NLR) and Starting Values of Parameters Obtained with Linear (LR) and Robust Regressions (RR) k, kmol(kg/h) KM,m3/kmol Kw, ma/kmol T,"C eq NLR LR RR NLR LR RR NLR LR RR 320 1 5.4 6.0 4.6 52.6 34.2 54.0 453.9 384.2 386.9 3 12 167 1 957 1 13415 1000.4 1077.2 1333.6 455.4 447.6 479.0 10 11.7 11.1 10.1 884.6 1086.2 1346.0 418.4 449.1 480.1 340 1 13.1 25.1 15.8 50.4 10.5 25.3 463.1 286.4 360.4 3 23 186 14 468 13371 722.2 425.4 359.2 414.1 305.8 298.9 . 10 31.9 33.9 37.1 551.3 429.9 362.7 355.4 307.0 299.9 360 1 33.6 61.7 30.1 25.3 7.5 25.5 301.6 224.6 283.0 3 31 753 23 193 23 263 422.9 290.3 290.5 269.4 220.2 221.5 10 73.5 79.6 78.3 358.3 293.7 305.5 243.2 221.0 226.0 Table IV. Summary of Published Values for Activation Energy and Heat of Adsorption for Methanol and Water on Various Catalysts E,, kJ/mol HM,kJ/mol Hw, kJ/mol T, "C catalyst ref 108.3" 160-195 alumina Kallo and Knininger, 1967 66.6" 289-418 silica-alumina Schmitz, 1978 120.P 300-350 alumina Rubio et al., 1980 138.9* 125.5 20.9 150-225 silica-alumina Bakshi and Gavalas, 1975 114.2b 127.6 111-150 ion exchange resin Than et al., 1972 123.0 Than et al., 1972
This Work 117.2c 290-360 alumina Figure 3 143.7b 70.5 41.1 320-360 alumina eq 10 75.1b 67.0 40.7 320-360 alumina eq 3 a Value obtained by linear regression of initial rate data. Value obtained by nonlinear regression for selected equation. Value obtained from slope In (-rM) v8 1/T.

1038 Ind. Eng. Chem. Res., Vol. 31, No. 4,1992

\ 'I r \ I
I u -

0 , -

&
A\ \
8

Concentration of CHIOH In Inlet


(volr)

mixture CHIOH-N2

0 CHIOH 1% ( u L u 3 M CHIOH Onnno 45% CHIOH


&&Ab

UIW 60% CHIOH 75% CHIOH

-90%

CHIOH

Table V. Sensitivity of Numerically Calculated Effectiveness Factor on Parameter Variation variation of produced change parameter parameter, % in calculated 7, % Ae +goo, -90 -0.1; +1.5 De f20 f8 f10 F8 d, =i4 f10 -rM
PP

*lo

74

10 - I 1 S7E-003

Table VI. Physical Properties of Catalyst Particles (3-mm Granules of Bayer SAS 350 yA1208) surface area (BET) 247 m2/g surface area (a > 100 8, Hg porosimetry) 2.24 m2/g 1.2846 g/cm3 bulk density (Hg porosimetry) particle density (He picnometer) 3.268 g/cm3 particle porosity 0.607

+m ' ' ' ' ' '

1.65E-003

'

'

'

''' ''

1.72E-003

'

' ' 1 '7'

1.80E-003

1/T

,1/K

Figure 3. Determination of apparent activation energy from initial rate measurements.

ferential equations describing mass and heat transfer within a catalyst particle must be solved. M s and heat as transport within a spherical particle are governed by

dr2 + - r dr
with boundary conditions
c

d2C;

2 dCi

= ppyirv

(14)

atr=O
0.1 :

.-

U Q) +

0 J u 0

0 0.01

4%
I

0.01

Measured r a t e

0.1

Figure 4. Comparison between experimentally measured and calculated intrinsic reaction rates with eq 10.

tivation energies are compared to the data found in the literature (Table IV) and to the results of the initial rate measurements (Figure 3), eq 10 seems to give more realistic predictions than eq 3. The comparison between the experimentally determined reaction rates and those calculated for the same reaction conditions by eq 10 is illustrated in Figure 4. Global Reaction Rate. The catalyst particles which are used in commerical fixed-bed reactors are usually greater than those used in the kinetic rate equation determining experiments. These particles are less reactive, as can be seen from Figure 1, because they exhibit intraparticle mass- and heat-transfer resistances. The reaction rate measured for such particles is called the global reaction rate. The global reaction rate is calculated from the intrinsic rate equation and effectiveness factor by the following equation -RM = q(-rM) (12) where the effectiveness factor is defined as (l/WJ(-rd
7 =
(-rM) IS

dV (13)

In fact, to calculate the effectiveness factor a set of dif-

atr=R T = Ts; Ci = Cis (17) A n analytical solution of the above system is possible only when iosthermal conditions within a particle are maintained. For the power-law rate expressions, analytical solutions are given in terms of the Thiele modules (Froment and Bischoff, 1979). For more complex rate equations, a generalized approach or a numerical computation can be applied. In the case of the nonisothermal conditions within a catalyst particle, the effectiveness factor must be calculated numerically. To solve the set of eqs 14-17, the algorithm proposed by Riggs (1988) was used. The approach taken in that algorithm converts a boundary value problem with ordinary differential equations (ODES)into an initial value problem with partial differential equations (PDEs) by adding an appropriate transient term to the differential equations describing the material and energy balances. The obtained set of PDEs is then integrated, according to initial conditions, using a mathematical package LSODE (Hindmarsh, 1986), to the steady state. The solution obtained at that point is also a solution to the original problem. In eqs 14 and 15, two parameters appear which describe i) the transport of heat (Aeff) and mass (Deff for a particular component in a catalyst particle. From the results of the sensitivity analysis presented in Table V, it is obvious that the effective diffusion coefficients for components should be predicted more accurately than the coefficient of effective heat conductivity. A value of the effective thermal conductivity coefficient was predicted on the basis of literature data and correlations and was found to be 2.7 X kJ/(s m K). From Table VI, where some data of the measured physical properties of the catalyst particles are summarized, it can be concluded that the majority of the total surface arises from the pores with a diameter under 100 A. Thus, it an be assumed that for the reaction under the investigated conditions the Knudsen diffusion prevails

Ind. Eng. Chem. Res., Vol. 31, No. 4,1992 1039


Table VII. Comparison between the Experimentally Determined and Numerically Calculated Effectiveness Factors inlet composition" conversion VulC b eq 1 eq 3 eq 10 mol % CH30H mol % H20 T,"C dp = 0.17 mm dp = 3 mm qexpt 12.1 0.370 0.413 0.394 0.394 320 7.2 45 10 0.294 0.292 0.275 0.275 340 9.9 11.8 45 10 0.196 0.194 0.236 0.206 10.1 360 10.6 45 10 0.205 0.216 0.218 6.8 0.217 340 7.9 45 0 0.356 0.327 0.323 9.1 0.349 340 6.5 45 20 0.379 0.372 0.415 0.415 340 4.3 7.1 45 30 0.469 0.430 0.421 5.7 0.449 340 3.2 45 40 0.243 0.256 0.236 0.229 340 15.0 14.2 15 10 0.258 0.255 0.277 0.277 340 11.0 12.8 30 10 0.290 0.290 0.282 0.303 340 4.8 10.6 60 10 0.302 0.304 0.268 0.312 340 4.5 9.6 75 10
"Mole percent of CH30H and H20 in N2.bqexpt= [-R,(3 mm)/-r~(0.17mm)llT,~l.

6' T = 30C 4' T = 30C T = 3200C T = 20C 9'

690

Y 665OPERATING CONDITIONS TI = 551.15 K c - P = 2.1 bar 0" = 5.32 kg/h dl = 0.078 rn Q) 615-

640:

E :

2 590565 -

;:

$
5401, 0.00
, I ,

, I

I , ,

I ,

, I

-.__
I

0.10

0.20

0.30

0.40

F 0.00

0.01

0.1

Measured rate
Figure 5. Comparison between experimentally measured global reaction rates for 3-mm catalyst particles and those calculated numerically.

Axial coordinate, m Figure 6. Comparison between experimentally measured temperature and concentration profiles in an adiabatic pilot reactor and those predicted by a pseudohomogeneous model.

in the catalyst particles. For materials like yAl,O, effective diffusivity can be estimated using a parallel pre model: Deffi= dit/^ (18) According to Satterfield (1970), effective diffusivity, within a factor of 2, can be predicted if it is assumed that a parallel pore model with an experimentally measured porosity and a value of 4 for the tortuosity factor is used. Assuming different values of this factor, calculations on a trial and error basis were performed until the value which provided a good agreement between the measured and calculated reaction rates for 3-mm catalyst particles was obtained. When a value of 3 was taken for the tortuosity factor in calculating the effective Knudsen diffusion coefficient and the numerical procedure for calculating the effectiveness factor, the agreement between the measured and calculated reaction rates was fairly good. A comparison of the calculated and experimental measured global rates in a differential reactor is shown in Figure 5. In Table VI1 some values are given for the calculated end experimentally determined effectiveness factor at different temperatures and compositions of reaction mixtures entering the differential reactor. As one can conclude from the results in Table W, either one of eq 1,3, or 10 predicts the effectiveness factor quite reasonable. In Figure 6, comparison between experimentally measured temperature and concentration profiles in an integral adiabatic fixed-bed pilot reactor is shown (BerEiE, 1990). For operating conditions as specified in Figure 6, it was shown (BerEiE and Levec, 1991) that a one-dimensional

pseudohomogeneous model can be successfully used for modeling of an industrial reactor. From that figure it is obvious that neglecting the reversible term in a rate equation leads to the wrong predictions, as demonstrated by the use of eq 1 or 3.

Conclusions From the results discussed above it is obvious that there exist no significant differences among eqs 1,3, and 10 when they are used for interpretation of the results obtained in a differential reactor. Equations 3 and 10 provide the temperature dependency of the activation energy and heat of adsorption according to the Arrhenius and van't Hoff relation. On the basis of the criteria applied, it is further concluded that eq 10 represents the kinetic behavior of the dehydration reaction more realistically and is appropriate for modeling purposes. It is also evident that the global reaction rate can be successfully predicted with the proposed intrinsic rate equation and the numerical procedure for the effectiveness factor calculation. The effective diffusivities for components can be effectively predicted using a parallel pore model and assuming the Knudsen diffusion mechanism. Acknowledgment
We acknowledge support from the Research Council of Slovenia under Grant No. C2-0541-104 and Nafta Lendava for making it possible to run a pilot reactor at their facilities for more than 2 months.

Nomenclature c i = pore diameter, 1\

1040

Ind. Eng. Chem. Res. 1992,31, 1040-1045


BerEiE, G. Dehydration of Methanol over yA1208. Kinetics of Reaction and Mathematical Model of an Industrial Reactor. Ph.D. Dissertation, The University of Ljubljana, 1990. BerEiE, G.; Levec, J. Reactor Model for the Catalytic Gas-Phase Dehydration of Methanol to Dimethyl Ether (DME). Vestn. Slou. Kem. Drus. 1991,38,253-270. Brake, L. D. U.S. Patent 4,595,785,1986. Chang, C. D.; Kuo, J. C. W.; Lang, W. H.; Jacob, S. M. Process Studies on the Conversion of Methanol to Gasoline. Ind. Eng. Chem. Process Des. Deu. 1978,17,255-260. Duggleby, R. G. Regression Analysis of Nonlinear Arrhenius Plots: An Empirical Model and a Computer Program. Comput. Biol. Med. 1984,14,447-455. Figueras, F.; Nohl, A.; Mourgues, L.; Trambouze, Y. Dehydration of Methanol and tert-Butyl Alcohol on Silica-Alumina. Trans. Faraday SOC. 1971,67,1155-1163. Froment, G. F.; Bischoff, K. B. Chemical Reactor Analysis and Design; John Wiley & Sons: New York, 1979;p 185. Gates, B. C.; Johanson, L. N. Lungmuir-Hinshelwood Kinetics of the Dehydration of Methanol Catalyzed by Cation Exchange Resin. AIChE J . 1971,17,981-983. Hindmarsh, A. C. Solving Ordinary Differential Equations on an IBM-PC Using LSODE. LLNL Tentacle Magazine; LLNL Livermore, CA, April 1986;Vol. 6,No.4 . Kallo, D.; Knozinger, H. Zur Dehydratisierung von Alkoholen an Aliminiumoksid. Chem.-Ing.-Tech. 1967,39,676-680. KlusaEek, K.; Schneider, P. Stationary Catalytic Kinetics via Surface Concentrations from Transient Data. Methanol Dehydration. Chem. Eng. Sci. 1982,37,1523-1528. Massaldi, H. A.; Maymo, J. A. Error in Handling Finite Conversion Reactor Data by the Differential Method. J. Catal. 1969, 14, 61-68. Riggs, J. M. An Introduction to Numerical Methods for Chemical Engineers; Texas Tech University Press: Lubbock, TX, 1988;p 406. Rousseeuw, P. J.; Leroy, A. M. Robust Regression and Outlier Detection; John Wiley & Sons: New York, 1987. Rubio, F. C.; Diaz, S. D.; Castillo, D. D.; Trujillo, J. D.; Alvarez, R. A. Deshidratacion Catalitica de Metanol en Fase Vapor. Ing. Quim. (Madrid) 1980,12,113-119. Satterfield, C. N. Mass Transfer in Heterogeneous Catalysis; M.I.T. Press: Cambridge, MA, 1970; p 42. Schmitz, G. Deshydration du Methanol Sur Silice-Alumine. Chim. Phys. 1978,746504355, Sinicyna, 0. A.; Cumakova, V. N.; Moskovskaja, I. F. Kinetika Degidratacii Metanola do Dimetilovogo Efira na SVK Ceolite. Kinet. Katal. 1986,27,1160-1162. Than, L. N.; Setinek, K.; Beranek, L. Kinetics and Adsorption on Acid Catalysts. IV. Kinetics of Gas-Phase Dehydration of Methanol on a Sulphonated Ion Exchanger. Collect. Czech. Commun. 1972,37,3878-3884. Woodhouse, J. C. U.S. Patent 2,014,408, 1935. Yang, K. H.; Hougen, 0. H. Determination of Mechanism of Catalyzed Gaseous Reactions. Chem. Eng. B o g . 1960,46, 146-157.

A , B, D = constants in eq 11, dimensionless c = specific heat of component i, kJ/(mol K) = concentration of component i, kmol/m3 dp = particle diameter, m dT = reactor diameter (Figure 6), m Di = diffusivity of component i, m2/s Deff = effective diffusivity of component i, mz/s E, = activation energy, kJ/mol AH = reaction enthalpy, kJ/mol ks = rate constant of surface reaction, mol/(g,,,/h) K = equilibrium constant, dimensionless Ki = adsorption constant of component i, m3/kmol P = pressure, bar r = radial coordinate, m R = particle radius, m -rM = reaction rate, mol/(g,,/h) -RM = global reaction rate, mol/(g,./h) Hi= heat of adsorption of component i , kJ/mol T = temperature, K TI= reactor inlet temperature (Figure 6), K V = particle volume, m3 x i = mole fraction of component i X = conversion, dimensionless

Greek Letters
c = catalyst particle porosity, dimensionless CB = catalyst bed porosity (Figure 6), dimensionless 9 = effectiveness factor, dimensionless

A r = effective heat conductivity of catalyst particle, kJ/(s ~


ui = stoichiometric coefficient of pp = particle density, g/cm3
T

m K)

component i, dimensionless

@JM =

= tortuosity, dimensionless methanol mass flow rate (Figure 6), kg/h

Subscripts

D = dimethyl ether i = component i (H20,(CH3)20,CH30H, N,) M = methanol S = conditions at catalysts surface v = per volume W = water 0 = initial conditions
Registry No. MeOH, 67-56-1; -pAlz03, 1344-28-1; DME, 115-10-6.

Literature Cited
Bakshi, K. R.; Gavalas, G. R. Effects of Nonseparable Kinetics in Alcohol Dehydration over Poisoned Silica-Alumina. AIChE J. 1975,21,494-500.

Received for review May 7, 1991 Accepted November 1, 1991

Kinetics of the Catalytic Hydrochlorination of Methanol to Methyl Chloride


Albert0 M. Becerra, Adolfo E. Castro Luna, Daniel E. Ardissone, and Marta I. Ponzi*
Facultad de Ingenieria y Administracion, Uniuersidad Nacional de San Luis, INTEQUI-CONICET, Au. 25 de Mayo 384, 5730 Villa Mercedes, San Luis, Argentina

The intrinsic kinetics of the catalytic hydrochlorination of methanol to methyl chloride on yAl,O, was determined from experiments in a tubular reactor in the temperature range of 513-593 K and at atmospheric pressure, after a catalyst screening and a study of operative conditions. A large number of detailed reaction mechanisms was considered. A strategy of model discrimination and parameter estimation led to a Hougen-Watson type model with statistically significant and thermodynamically consistent parameters. The two main technologies used commercially for obtaining methyl chloride are hydrochlorination of methanol

and chlorination of methane. Methyl chloride is used as an intermediate in the obtainment of chlorinated bypro-

0888-5885/92/2631-1040$03.00/0 1992 American Chemical Society 0

You might also like