See

discussions, stats, and author profiles for this publication at: https://www.researchgate.net/publication/231366278

Catalytic Dehydration of Methanol to Dimethyl

Ether. Kinetic Investigation and Reactor
Simulation

ARTICLE in INDUSTRIAL & ENGINEERING CHEMISTRY RESEARCH · NOVEMBER 1993

Impact Factor: 2.59 · DOI: 10.1021/ie00023a006

CITATIONS READS

55 416

2 AUTHORS:

Gorazd Bercic Janez Levec

National Institute of Chemistry University of Ljubljana
28 PUBLICATIONS 682 CITATIONS 116 PUBLICATIONS 3,931 CITATIONS

SEE PROFILE SEE PROFILE

All in-text references underlined in blue are linked to publications on ResearchGate, Available from: Gorazd Bercic
letting you access and read them immediately. Retrieved on: 01 February 2016
2478 Ind. Eng. Chem. Res. 1993,32, 2478-2484

Catalytic Dehydration of Methanol to Dimethyl Ether. Kinetic

Investigation and Reactor Simulation

Gorazd Ber6iEt and Janez Levee'**

Laboratory of Catalysis and Chemical Reaction Engineering, National Institute of Chemistry, and
Department of Chemical Engineering, University of Ljubljana, P.O.Box 537, 61001 Ljubljana, Slovenia

One-dimensional heterogeneous and pseudohomogeneous plug flow models were employed to model
an adiabatic fixed bed reactor for the catalytic dehydration of methanol to dimethyl ether.
Longitudinal temperature and methanol conversion profiles predicted by these models were compared
to those experimentally measured in a pilot reactor. The reactor was packed with 3-mm yAlzOs
pellets and operated in a temperature range of 29O-36OoCand a t a pressure of 2.1 bar. Intraparticle
mass transport was found to be the rate-controlling step.

Introduction some properties of the reactor and catalyst are listed in

Table I. The reactor consisted of four stainless steel
Catalyticdehydrationof methanol over an acidic catalyst segments connected by flanges. The radial temperature
offers a potential method for production of dimethyl ether gradient in the isolation shell was measured by thermo-
(DME),a new spray propellant. From the patent literature couples located at different radial positions in the shell
(Woodhouse,1935; Brake, 1986),it can be concluded that and connected to the microprocessor-based thermoreg-
reaction takes place on pure y-alumina and on y-alumina ulator. By means of two independently powered heating
slightly modified with phosphates or titanates, in a tapes in each element, one placed on the outside surface
temperature range of 250-40O0C and pressures up to 10 of the reactor wall and the other placedwithin the isolation
bar. The kinetics of methanol dehydration on an acidic shell, it was possible to maintain almost zero temperature
catalyst has been studied extensively,resulting in different gradient in the shell, and therefore adiabatic conditione
kinetic equations. Most of the rate equations have been existed within the catalyst bed. The radial and axial
derived from the experiments conducted in conditionsnot temperatures in the catalyst bed were measured using a
found in an industrial reactor. The experimenta were HP 3421A data acquisitionlcontrolunit and a HP 150PC.
mainly performed with mixtures of methanol, water, and Thermocouples were located in the bed axis at different
nitrogen at low water vapor pressures. Since water axial positions, namely, at z = 0, 5, 10, 15, 25,36,45,55,
produced during the dehydration considerably retards the and 65 cm, respectively. At the axial positions of 10,25,
reaction rate, the derived rate equations are not suitable and 45 cm, thermocouples were also placed radial at F =
for the commercial reactor design,where the reaction takes 0, i = 0.5R, and i = 0.95R. Sampling tubes were located
place at highconversions. BerEiE and Levec (1992) recently within the catalyst bed at 7; = 0. Gas samples, taken at
reported an intrinsic rate equation in the form different axial positions (z = 10,25, and 45 cm) and from
the reactor outlet, were led through a four-way Valco SD
flow path valve to a GC sampling valve.
Analysis of the gas samples was performed by means of
a HP 5890 GC equipped with a TCD. The SS column
which represents the kinetic behavior of the dehydration (230cm X 118)waspackedwithPorapakT (100/120mesh),
reaction over 7-alumina more realistically than equations while helium was used as the carrier gas. The oven
published earlier. Equation 1thus offers a powerful tool temperature was kept at llO°C for the first 17 min and
that can be used to simulate the dehydration reactor then continuously increased to 130 OC (at a rate of 4OoC/
performance. In the open literature, no information on min). A gas sample (0.25 mL) was injected by means of
the reactor simulation is given in spite of the fact that the the sampling valve kept at 110OC. A calibration curve
methanol dehydration reaction is considered one of the prepared for each component (methanol,water, and DME)
basic reactions in the so-called C-1chemistry (e. g., MTG was used to determine the compositionof each gas sample.
Process; Chang et al.,1983). Duration of the analysis was about 25 min.
The aim of this work is to present the experimental and Experiments with 3-mm particles were also carried out
simulated results of an adiabatic fixed bed dehydration in a differential fixed bed reactor in a temperature range
reactor operated at conditions typically employed in a of 290-360 OC and the constant pressure of 146 kPa. The
large-scale reactor. The global rates as well as the reactor was operated free of the external mass- and heat-
temperature and concentration profiles predicted by transfer resistances. The inlet concentrationsof reactants
different models are compared with the experimental were varied between 15 and 90 mol 7% for methanol and
results obtained in a pilot reactor having the production between 0 and 50 mol % for water. Even though DME
capacity of about 150 kg of DMElday. was not present in the feed,we simulatedan integral reactor
that is made up of series of differential reactors. The
Experimental Section objective was to determine the rate in these differential
The pilot-scale reactor system employed in this study reactors (BerEiE and Levec, 1992).
is illustrated in Figure 1. Operating conditions as well as
Reactor Model
* Author to whom correspondence should be addressed.
+ Laboratory of Catalysisand Chemical ReactionEngineering. A compromise between an excessive complexity (e.g., a
* Department of Chemical Engineering. two-dimensional heterogeneous model) and oversimpli-
Q888-5885/93/2632-2478$~4.00/0 1993 American Chemical Society
 Ind. Eng. Chem. Res., Vol. 32, No. 11,1993 2479

Figure 1. Schematic drawing of the experimental apparatus for catalytic dehydration of methanol: (1)gas cylinder (nitrogen), (2)methanol
reservoir, (3)mass flow controller, (4)metering pump, (5)evaporator, (6)ball valve, (7) preheater, (8) manometer, (9) needle valves, (10)tube
and shell heat exchanger, (11) phase separator, (12)coolers, (13)thermocouples, (14)calcium silicate insulation block, (15)glass wool, (16)
catalyst bed, (17)inert bed, (18)sampling valve.
Table I. Properties of Catalyst and Reactor and Operating Table 11. One-Dimensional Plug Flow Models of the Fixed
Conditions Bed Reactor Used in This Studv
catalyst Bayer SAS y-AhOs interfacial intraparticle model
diameter of catalyst particles (mm) 3 t w e of model eradients zradients desimation
catalyst bed height (mm) lo0 heterogeneous yes yes I
catalyst particle density (kp/m3) 1.47 heterogeneous no Yes I1
catalyst bed porosity (/) 0.40 heterogeneous yes nd I11
reactor diameter (mm) 78 pseudohomogeneous no nd IV
reactor height (mm) loo0
preheater packing height (mm) 250 a Intraparticle gradients are accounted for in the apparent rate
inlet temp (K) 551.15-56.15 coefficients (eq 12).
pressure (bar) 2.1
methanol feed rate (L/h) 4.34-6.74
V'JJM(T,C;) dV
'10 = (4)
fication (e.g., a one-dimensional pseudohomogeneous rM(P,CP)
model) is usually made when one wishes to reduce the where the reaction rateat the catalystactivesite, rM(T,Ci),
computational time without an appreciable loss of the is given by eq 1. Equations 2 and 3 are subject to the
accuracy of model prediction. An adiabatic reactor can initial conditions that specify the feed composition and
be well described by one-dimensional models. In this temperature:
study, we adopted heterogeneous models that account for
the intraparticle and/or interfacial gradients as well as a c,"=ck0 atr =O
pseudohomogeneous model. The models used are sum-
marized in Table 11. P=T$ atz=o (5)
In the heterogeneous model, the mass and heat balances
in the fluid phase with isobaric plug flow are governed by The objective is to solve eqs 2 and 3 for the temperature
the following equations: and concentration profiles in the gas phase along the
reactor longitudinal coordinate.
As indicated in eq 4, the overall effectiveness factor can
be obtained by integrating the rate of reaction over the
volume of the whole particle. The concentration and
temperature profiles within the catalyst particle are
obtained by the simultaneous solution of the governing
equations which result by writing a mass and heat balance
within the particle

where the overall effectivenessfactor, 70, accounts for the

intraparticle as well as interparticle mass- and heat-
transport limitations (model I). The overall effectiveness
factor is defined as a ratio of the actual (global)rate to the
rate based on bulk conditions, thus
2480 Ind. Eng. Chem. Res., Vol. 32, No. 11,1993
0.20.

0.16

0.12

0.08

0.04

0.00
0.00 0.02 0.04 0.05 0.0s 0.00 0.04 0.08 0.12 0.f6 C !O

t O'JO T=613 K
+loa a I
0.40

4? 0.20 0.30

0.20

0.10

0.00 k- - 0.00
0.00 0.10 0.20 0.30 0.oo o.<o 0.20 0.30 0.40 0.50
Measured rate Measured rate
Figure 2. Comparison between experimentally measured global reaction rates for 3-mm catslyst pellets and those calculated by eqs 1 and
12. Area marked corresponds to the pilot reactor operating conditions.
and the following boundary conditions Kutta method while the surface conditions (C'and P )in
the heterogeneous model were determined by a Newton
iteration method. In these calculations, p. cp and AHr
were considered as a function of temperature. Nonlinear
equations (eqs 6 and 7), coupled through the concentration
dCi and temperature within the catalyst particle, had to be
ki(Cp - q)= -D.~;T; at r = d42 (9)
solved in each node of the computational grid used in the
integration of the mass and heat balances. Due to the
h ( T - f i =-A d- T
at r = d42 (10) complex rate equation, the solution is not so straightfor-
e dr ward. However, one canuseeither Bischoff 8 approximate
When the reaction rate is small compared to the external solutionwithageneralizedThiele modulus (Bischoff,1965)
heat- and mass-transfer rates, the surface concentration or a numerical solution. In this study, the numerical
and temperature both approach those in the bulk gas appr0acbwasused;the boundaryvalueproblem (eqs6-10)
phase, hence was first converted into an initial value problem by adding
an appropriate transient term to eqs 6 and I and then
q = C p and T= integrated according to the initial conditions to the steady
and consequently state (Riggs, 1988; Hindmarsh, 1986).
?I?,= 1 To avoid a time-consuming numerical calculation of the
effectivenessfactors, one can employ eq 1with the intrinsic
This leads to the one-dimensional heterogeneous model rate coefficients replaced by apparent coefficients that
where only the intraparticle resistances have to be account for the intraparticle mass- and heat-transport
considered (model 11). Equations 2 and 3 with initial limitations. For that purpose, the rate per catalyst particle
conditions (eq 5) still apply, but the overall effectiveness can be calculated from the relation
factor, qo. has to be replaced by the particle effectiveness
factor, 7, defined as

where the apparent coefficients(n.,i<~,&) are evaluated

from the experiments performed with particles employed
Results and Discussion in the pilot reactor. However, eq 12 also represents an
approximate solution of the integral in eqs 4 and 11,
Integration of the reactor mass and heat balances (eqs respectively, for the concentration range investigated.
2 and 3) was performed by using a fourth-order Runge- Employing eq 12 in the heterogeneous models, they can
 Ind. Eng. Chem. Res., Vol. 32, No. 11, 1993 2481
Table 111. Intrinsic and Apparent Coefficient Valuations
for 3-mm ?Ala08 Catalyst Pellets
parameter intrinsic (ea 1) amarent (ea l2)O
k, 5.35 X 1013exp(-17280/T) 6.60 X 108exp(-10800/T)
KM 5.39 X lo-’exp(8487/T) 0.72 X exp(830/T)
Kw 8.47 X le2exp(5070/T) 0.45 X 1(F2exp(ll3O/T)
Component concentrations are expressed by mole fraction.
Constanta I?, are dimensionless in this case.
1.00 , I
- 670

\0.80 - 650
Y
I
0 - 630
+
0.60 3 0.00 0.10 0.20 0.30
+
Y- - 610 E 700’
lA a,
VI
a, 0.40 - 590 Q
E
S
.->
+ - 570 I-
0)
.6 6 0 -
0 PI
a, 0.20
Y-
L 640-
+ S
- 550 +
w 0 620 -
L

0.00 I
0.00
1

0.10
I

0.20
I

0.30
‘530
0.40
PI
n600-

Reactor length, m p&OXl Exp. points To=551.15 K

AAAAA Exp. points (T.=561.15 K]
Figure 3. Particle and overall effectivenessfactors and temperature - Model I
as a function of the longitudinal reactor coordinate.
- - Model II
Model 111
__
.- - - - Model IV
be further simplified: model I accounts then for the 540‘ I I I I I
0.00 0.10 0.20 0.30 0.40 0.50
interfacial gradients only and is termed as model 111,while
model I1 is transformed to a pseudohomogeneous model Reactor length, m
(model IV). In this work, the apparent coefficients were Figure 4. Comparison between experimentally measured temper-
calculated by means of a nonlinear regression technique ature profiles and those predicted by the models at two different
methanol feed rates.
from the results obtained in the differential reactor
employing 3-mm particles. These experiments were
carried out in the absence of external mass- and heat- equilibrium is attained. At that length, the effectiveness
transfer limitations. The intrinsic and apparent coeffi- factors become equal.
cients are listed in Table 111. A parity plot (Figure 2) Figure 4 shows the experimentally measured and
shows the differences between the global rates calculated predicted temperature profiles along the reactor length at
numerically (based on eq 1) and by means of eq 12 to different inlet temperatures and feed flow rates. In Figure
those measured experimentally in the differential reactor. 5, the measured conversion profiles in the pilot reactor
The differences between predicted and measured rates are compared to those predicted by both models. When
increase when the temperature increases, most likely due eq12 was employed in the heterogeneous model, the
to the simplification made in the rate calculation. Nev- external mass- and heat-transfer limitations were ac-
ertheless, at the operating conditions of the pilot reactor counted for only (overall effectiveness factor was equal to
(shaded area) the agreement is certainly within a 10% a ratio of the rate evaluated at the surface conditions to
error in both cases. Therefore, the rate equation with the the rate evaluated at the bulk gas phase conditions), while
apparent coefficients (eq 12) may also be successfully in the homogeneous model it was simply set: 70 = 7 = 1.
employed in simulation studies. It is seen from Figures 4 and 5 that the predictions made
In Figure 3, the overall and particle effectiveness factors with the models neglecting interfacial gradients (models
as well as temperature are shown as a function of the reactor I1 and IV) agree better with the experimental values than
length. For operating conditions used in this work, the the predictions obtained by the models accounting for the
heat-transfer coefficient, h, between the bulk gas phase interfacial gradients (models I and 111). On the other hand,
and the catalyst surface was calculated by means of both models employing eq 12 (models I11 and IV) predict
Martin’s correlation (Martin, 1978). Values of the coef- almost the same profiles as those models using the
ficient lay between 0.1 and 0.14 W/m2.K. The mass- numerically calculated reaction rates (models I and 11).In
transfer coefficient, ki,was calculated by using the concept any heterogeneous model, a temperature (and concen-
of heat- and mass-transfer analogy, and it ranges between tration) difference between the gas phase and the particle
0.04 and 0.06 m/s. From this plot, it may be concluded surface is always established, no matter how large the heat-
that the external mass- and heat-transfer resistances are transfer coefficient is. Since the dehydration reaction is
negligible in comparison to those within the particles. A exothermic, the particle surface temperature is higher than
ratio of the overall effectiveness factor to the particle the temperature of the gas phase (Figure 3). The rate
effectiveness factor exceeds a value of 1 because the calculated by the heterogeneous models accounting for
reaction is exothermic. The differences between the the interfacial gradients is therefore higher when the same
surface and bulk concentrations and temperatures, re- inlet temperature of the gas phase is imposed to the models.
spectively, vanish at the bed depth where the reaction Consequently, the temperature profiles predicted with
2482 Ind. Eng. Chem. Res., Vol. 32, No. 11, 1993
1.00 I
1
"1"

To = 5 5 1 . 1 5 K

\ Oa80 t
P = 2.1 b a r
/ ,r7- J g/
0

!
E 0.60
I

.-
P
$ 0.40
S
0 03333 Experimental points
Qu=4.34 I/h - Model 111
0
0.20
0u=6.74 I/h _ - h' = 1.2 h
. _ _ h'__
_ _ h' == 0.8
5.0 * h
h
..-.
.... h' = 0 . 2 * h
0.00 0.40 I-"

0.oo 0.1 0 0.20 0.30 0.00 0.10 0.20 0.30 0.40 0.50

c
1.00 I 1 680 I I
To = 5 6 1 . 1 5 K Qu=6.74 I/h
P = 2.1 b a r 660 P = 2.1 bar

t
To=551.15 K
Y
\Oaso ,640 -
2)
.-$0.60 t
3 620 -
v)
L E -
>
a 0.40 Q)
a
600
S -
0 Ooxo Experimental points
Exp. points - Model 111
0 Exp. points - - k' = 1.2 k
0.20 Model I . _ _ k' __ = 0.8 k
Model II 560 __ k' = 5.0 k
Model 111 ........ k ' = 0.2 k
Model IV
0.00 5401 I I I I
0.00 0.10 0.20 0.30 0.40 0.00 0.10 0.20 0.30 0.40 0.50
Reactor length, m Reactor length, m
Figure 5. Comparison between experimentallymeasured conversion Figure 7. Effect of the heat- and mass-transfer Coefficients on
profiles and those predicted by the models at two different inlet temperature profile predicted by the heterogeneous model 111.
temperatures.
illustrated in Figure 7, where the temperature profiles are
plotted for different values of the mass and heat transport
0 coefficients, respectively. These two physically relevant
parameters were varied one at a time to assess their effect
on the model features. Results in Figure 7 show that
increasing the transport coefficients even by a factor of 5
does not significantly change the loci of the temperature
profiles. When the heat transport coefficient is decreased
by a factor of 5, the locus is moved far above the
experimental points. As a direct consequence of a lower
heat-transfer rate, both the catalyst surface temperature
-
,- f

/
' EExperimental
- - T-' .
Model Ill
. " = T." + 2 K
points
and the reaction rate increase. Due to the increase of the
reaction rate, the temperature profile becomes steeper.
rcn-
To' = To - 2 K
The same decrease of the mass transport coefficient results
in a shift of the temperature profile in the opposite
540' I I I I I direction. It should be noted that both transport coef-
0.00 0.10 0.20 0.30 0.40 0.50
ficients (k, h) change proportionally to the Reynolds
Reactor length, m number; in our case of exothermic reaction, their influence
Figure 6. Effect of ,the inlet temperature on temperature profile on the predicted temperature profile is almost counter-
predicted by the heterogeneous model 111. balanced.
The catalyst particles can be considered isothermal
these models are found above those predicted by the (BerEiE, 1990);thus, only the intraparticle mass transport
models with neglected interfacial gradients. Figure 6 shows may retard the reaction. Figure 8 shows the temperature
the temperature profiles for different inlet temperature profiles predicted by the heterogeneous model 11, where
of the gas phase. As can be seen, only a slight difference a value of the effective diffusion coefficient was changed
in the inlet temperature causes quite a bit of change in the for f20%. The effective diffusion coefficient was esti-
profile location. However, by decreasing the inlet tem- mated from the relation
perature by only 2 K, certainly within the error of
temperature measurement, one can match the predicted De,i= 9.636 X 1 0 4 m (cm2/s) (13)
and experimental profiles quite well. This further alludes
that the dehydration rate is not controlled by the external which is based on the parallel pore model for Knudsen-
transport processes. Confirmation of this hypothesis is type diffusion (BerEiE, 1990). Influence of the particle size
 Ind. Eng. Chem. Res., Vol. 32, No. 11, 1993 2483
--- I Conclusions
The intrinsic rate of dehydration reaction is realistically
expressed by eq 1. Using this equation and an appropriate
numerical procedure, one can reasonably well predict the
global reaction rates that account for the intraparticle heat-
and mass-transfer limitations. The axial temperature and
conversion profiles in an adiabatic fixed bed reactor for
the DME production are well predicted by models
neglecting interfacial gradients where the source term is
either in the form of eq 1with the intrinsic rate parameters
(model 11)or in the form of eq 12 with the apparent rate
Experimental parameters (model IV). The apparent rate parameters
Model II can be easily determined by experiments in a differential
,,I = 0.8 reactor employing particles of the same size that are going
5401 1 I I I
to be used in a commercial reactor.
0.00 0.10 0.20 0.30 0.40 0.50
Reactor length, m Acknowledgment
We acknowledge support from the Ministry of Science
Figure 8. Effect of the effective diffusivity on temperature profile and Technology of Slovenia under Grant No. C2-0541-
predicted by the heterogeneous model 11. 104 and from Nafta-Lendava for making it possible to run
680 I I
a pilot reactor at their facilities for more than 2 months.
We also thank Bayer, A. G., Germany, for the catalyst
supply.
Nomenclature
Ci = concentration in fluid phase, kmol/m3
cp = specific heat of fluid, kJ/(kg-K)
D,i = effective diffusion coefficient, m2/s
d, = particle diameter, m
AHr = heat of reaction, kJ/kmol
h = heat-transfer coefficient, W/(m2.K)
K = thermodynamic equilibrium constant
Fi = adsorption constant, m3/kmol
. _ _P'_ _3.8 = P Ki = apparent adsorption constant
_ _ d,' = 2.0 * d,
- -...- -. d,' = 0.5 d, ki = mass-transfer coefficient, m/s
540
0.00
' 0.10
I
0.20
I
0.30
I
0.40
I I
0.50
k, = reaction rate constant, kmol/(kg-h)
k, = apparent reaction rate constant, kmol/(kg.h)
Reactor length, m Mi = molecular weight, kg/kmol
P = pressure, bar
Figure 9. Effect of the particle size and operating pressure on R = reactor radius, m
temperature profile predicted by the heterogeneous model 11. r = particle radial coordinate, m
0.10
F = reactor radial coordinate, m
rM = rate of methanol disappearance, kmol/(kg.h)
T = temperature, K
0.08 0 u = superficial velocity, m/s
\
0 ' 8 V = particle volume, m3
0 0 z = reactor longitudinal coordinate, m
\
- 0.06
0 8
0
Xi = mole fraction
Greek Letters
E 0.04 Mixture composition
MeOH:DME:H,O:N2 4~ = methanol feed rate, L/h
n 0003035 : 9 : 9 : 47 7 = effectiveness factor
E 00000 41 : 13: 13: 33
0.02 k X ? 0 Q 5 0 : 9 : 9 : 32 70 = global effectiveness factor
I A, = particle effective thermal conductivity, kJ/(ms.K)
W
0.00 ui = stoichiometric coefficient
0 1 2 3 4 5 6 7 8 9 10 p = gas-phase density, kg/m3
Pressure, b a r p~ = catalyst bed density, kg/m3
pp = catalyst particle density, kg/m3
Figure 10. Effect of the operating pressure on global reaction rate
for 3 mm y-Al2O3 catalyst pellets. Subscripts and Superscripts
b = bulk conditions
and the operating pressure on the longitudinal temperature E = dimethyl ether
profile is depicted in Figure 9. It is obvious that the effect e = effective
of particle diameter is appreciable. It is somewhat i = ith component (methanol, DME, water)
surprising that the total pressure has almost no effect on M = methanol
0 = inlet conditions
the predicted temperature profile. One can speculate that s = surface conditions
the reduction of the intrinsic reaction rate due to the W = water
pressure rise is compensated by the increase of the
effectiveness factor. The net result is, however, almost Literature Cited
the same global reaction rate. This conclusionis confirmed BerEiE, G. Dehydration of Methanolover y-Al203. Kinetica of Reaction
with the experimental results obtained in a CSTR system and Mathematical Model of an Industrial Reactor. Ph.D. Thesis,
using the same catalyst pellets (Figure 10). The University of Ljubljana, 1990.
2484 Ind. Eng. Chem.Res., Vol. 32, No. 11, 1993

BerEiE, G.; Levec, J. Intrinsic and Global Reaction Rata of Methanol Riggs, J. M. An Introduction to Numerical Methods for Chemical
Dehydration over Y-AlaOa Pellets. Ind. Eng. Chem. Res. 1992,31, Engineers; Texas Tech University Press: Lubbock, TX, 1986;p
1036-1040. 406.
Bischoff, K. B. Effectivenew Factors for General Reaction Rate Woodhouse, J. C. U.S.Patent 2,014,408,1935.
Forms. AIChE J. 1961,11,361-366.
Brake, L. D. U.S. Patent 4,695,786,1986.
Chang, C. D. Hydrocarbons from Methanol; Marcel Decker: New Received for review March 1, 1993
York, 1983;p 76. Revised manuscript received July 1, 1993
Hindmarsh, A. C. Solving Ordinary Differential Equations on an Accepted July 27, 19930
IBM-PC Using LSODE. LLNL Tentacle Magazine; LLNL:
Livermore, CA, April 1986;Vol. 6, No. 4.
Martin, H. Low Peclet Number Particle-*Fluid Heat and Mass @ Abstract published in Advance ACS Abstracts, October 1,
Transfer in Packed Beds. Chem Eng. Sci. 1978,33,913-919. 1993.