Download as pdf or txt
Download as pdf or txt
You are on page 1of 12

Ind. Eng. Chem. Res.

1998, 37, 1917-1928 1917

SEPARATIONS

Measurement of Residue Curve Maps and Heterogeneous Kinetics in


Methyl Acetate Synthesis
Wei Song,† Ganesh Venimadhavan, Jason M. Manning, Michael F. Malone, and
Michael F. Doherty*
Department of Chemical Engineering, University of Massachusetts, Amherst, Massachusetts 01003

Kinetically controlled reactive distillation is a bridge between the two limiting cases of nonreactive
and equilibrium reactive distillation. In this paper, we study the influence of heterogeneous
catalysis on the transition from the nonreactive to the equilibrium reactive limits for the
esterification of acetic acid with methanol. A Langmuir-Hinshelwood/Hougen-Watson rate
model was developed to represent the reaction kinetics. Since reactive distillation is carried
out at the boiling temperature of the liquid, the pressure of the system plays an important role
in determining the product composition. Higher operating pressures imply higher temperatures
and faster rates of reaction but could also lower equilibrium conversions (for exothermic systems)
and trigger unwanted side reactions. To study the importance of side reactions and the effect
of pressure on product selectivity, we also include the reaction kinetics for methanol dehydration.
Isothermal batch kinetic experiments were performed using a heterogeneous (Amberlyst 15W)
catalyst at various temperatures and catalyst concentrations. Independent binary adsorption
experiments were also performed to estimate the adsorption equilibrium constants. As a test
of the kinetic model, independent equilibrium reactive open-evaporation experiments were
performed and the model prediction was found to be in good agreement with the experimental
residue curve measurements. The model was also used to predict the behavior of the system in
the kinetically controlled regime. Residue curves for the kinetically controlled cases are
qualitatively similar to that of the equilibrium case, showing that the production of methyl acetate
can be carried out using reactive distillation in both the equilibrium and the kinetically controlled
regimes.

Introduction stand the transition from the nonreactive to the equi-


librium reactive distillation system, we need to know
Reactive distillation is an important hybrid separation the kinetics of the reaction (Venimadhavan et al., 1994;
process (e.g., Agreda and Partin, 1984; Agreda et al., Rev, 1994; Rehfinger and Hoffmann, 1990; Sundmacher
1990; DeGarmo et al., 1992; Doherty and Buzad, 1992; and Hoffmann, 1994). In this paper, we study the
Smith, 1981, 1984, 1990). A useful tool for the synthesis influence of heterogeneous catalysis on the transition
and analysis of both nonreactive and reactive distillation from the nonreactive to the equilibrium reactive limits.
systems is the simple distillation residue curve map We report the related simple distillation experiments
(RCM) (e.g., Fidkowski et al., 1993; Foucher et al., 1991; for the reactive mixture.
Wahnschafft et al., 1992; Venimadhavan et al., 1994).
The structure of the residue curve map is determined Methyl acetate can be made by the liquid-phase
by the number of singular points in the mixture together reaction of acetic acid (HOAc) and methanol (MeOH)
with their temperature and composition. For a non- catalyzed by sulfuric acid or a sulfonic acid ion-exchange
reactive mixture, these singular points are the pure resin in the temperature range of 310-325 K and at a
components and azeotropes; however, their number, pressure of 1 atm. The reaction is
temperature, and composition may change with the
introduction of chemical reaction. Reactive distillation acetic acid + methanol S methyl acetate + water
systems are bounded by two limiting cases: (1) the (HOAc) (MeOH) (MeOAc) (H2O)
equilibrium reactive case where there is simultaneous (1)
vapor-liquid and reaction equilibrium (Barbosa and
Doherty, 1988; Ung and Doherty, 1995a-d) and (2) the This is a classic example in reactive distillation (Agreda
nonreactive case where no reaction occurs. To under- et al., 1984, 1990; Siirola, 1995).
The simple distillation experiment is an easy way of
* To whom correspondence should be addressed. studying the feasibility of continuous and batch distil-
† Present address: Pennzoil Products Company, 1520 Lake lation processes (Bernot et al., 1991; Fidkowski et al.,
Front Circle (77380), P.O. Box 7569, The Woodlands, TX 1993; Wahnschafft et al., 1992). For reactive systems,
77387. the limiting case of simultaneous vapor-liquid and
S0888-5885(97)00879-8 CCC: $15.00 © 1998 American Chemical Society
Published on Web 04/01/1998
1918 Ind. Eng. Chem. Res., Vol. 37, No. 5, 1998

reaction equilibrium is now well understood theoreti-


cally (Barbosa and Doherty, 1988; Ung and Doherty,
1995a-d, Espinosa et al., 1995). However, in many
applications, conditions of reaction equilibrium may not
be approached because of limited rates of reaction and
residence times. Furthermore, in systems where the
main product is an intermediate that reacts further with
the reactants or with other products, it is undesirable
to let the system reach reaction equilibrium. The
behavior of a system near reaction equilibrium may be
quite different from conditions where the reaction is
kinetically controlled, and we must have a good under-
standing of the reaction kinetics to examine the behav-
ior.
Kinetic rate expressions used in reactive distillation
models are best written in terms of activities as opposed
to the more familiar concentration-based rate expres-
sions (e.g., Venimadhavan et al., 1994; Rehfinger and
Hoffmann, 1990; Sundmacher and Hoffmann, 1994). For
the methyl acetate system the rate expression in terms Figure 1. Experimental setup for the measurement of residue
of activities is strongly preferred because the high curve maps.
polarity of water and methanol compared to methyl tion reaction at different temperatures. This reaction
acetate leads to strongly nonideal solution behavior. is
Earlier works on the kinetics of this system have
reported preexponentials and the activation energies for 2MeOH S DME + H2O (2)
the concentration-based rate expressions for the esteri-
fication kinetics of different alcohols (Smith, 1939) or We developed a LHHW model for this reaction also. In
have not addressed the range of species and catalyst the next section, we describe the details of the ap-
concentrations of interest for reactive distillation space paratus, the experimental conditions, and the data
(Neumann and Sasson, 1984; Xu and Chuang, 1996). acquisition for the kinetic and adsorption experiments.
In this paper we report the results of batch kinetic
experiments for the methyl acetate system over a range Experiments
of molar feed ratios more typical of reactive distillation
conditions. Materials. Methanol (purity > 99.9 wt %), acetic
acid (purity > 99.7 wt %), methyl acetate (purity > 99.5
A macroreticular ion-exchange resin such as Am- wt %), and ethylene glycol dimethyl ether (anhydrous,
berlyst 15W is an ensemble of microspheres. More than water < 0.005 wt %) were purchased from Aldrich. The
95% of all functional groups (the protons) are inside the catalyst was a strongly acidic macroporous resin (Am-
microspheres and are only accessible to substances berlyst 15 wet, moisture content 53 wt %) purchased
which are able to penetrate the network of cross-linked from Rohm & Haas Co. The concentration of acid sites,
polymer chains. Consequently, adsorption effects must the average pore diameter, and the surface area of the
be taken into account in order to describe reactions dry catalyst were 4.9 mmol/g, 250 Å, and 45 m2/g,
catalyzed by ion-exchange resins, e.g., Oost and Hoff- respectively, as reported by the manufacturer.
mann (1996). Therefore, we developed a heterogeneous Apparatus. A 500-mL, three-neck glass flask was
Langmuir-Hinshelwood/Hougen-Watson (LHHW) type used for kinetic and adsorption experiments. One neck
rate expression. The LHHW model has three types of was connected to a thermometer and another to a
parameters, the forward reaction rate constant, the condenser with a pressure release on top, and the third
reaction equilibrium constant, and the individual com- was sealed with a septum cap. The flask was immersed
ponent adsorption equilibrium constants. To decouple in a stainless steel tank filled with a 1:1 mixture of
the reaction and adsorption phenomena, we performed ethylene glycol and water as the heating fluid. The
independent isothermal binary adsorption experiments contents were stirred by a magnetic stirrer (Thermolyne
for the nonreactive binary pairs. SP18400) positioned underneath the bath. A desired
Since reactive distillation is carried out at the boiling constant temperature in the bath was obtained by a
temperature of the liquid, the pressure of the system digital circulator (HAAKE, DC3) which ensures a tem-
plays an important role in determining the product perature constancy of (0.1 °C. The cooling water
composition. Higher operating pressures imply higher circulation for the condenser was achieved by a circula-
temperatures and faster rates of reaction. Faster reac- tor (NESLAB RTE-111).
tion rates give higher conversions and/or smaller resi- Figure 1 shows a schematic of the experimental setup
dence times and hence smaller equipment. However, for the residue curve measurements. A 500-mL, three-
higher temperatures could also lead to triggering of neck glass flask was connected to a thermometer, a
unwanted side reactions with higher activation energies custom-made distillation head, and a septum cap. The
and, in the case of exothermic reactions, lower equilib- heating was controlled by a voltage controller (Ace Glass
rium conversions. Higher temperatures also require 12088) and heating mantles. A magnetic stirrer was
more expensive materials of construction since acetic used to mix the reactants. The distillation head and
acid/water mixtures are corrosive at elevated temper- the reaction flask were insulated with glass wool to
atures. To study the importance of side reactions and minimize heat loss to the surroundings and to prevent
the effect of pressure on product selectivity, we also condensation of vapor and reflux back to the liquid
include the reaction kinetics of the methanol dehydra- mixture.
Ind. Eng. Chem. Res., Vol. 37, No. 5, 1998 1919

Composition Analysis. An HP 5890 series II gas


chromatograph (thermal conductivity detector) equipped
with 7673 Automatic Injector and 3365 Chemstation
Software was used to analyze the composition of samples
from the liquid phase of the reactive mixture. The
sample size for GC was 1 µL. Both injection port and
detector temperatures were set to be 230 °C. A 30 m ×
530 µm × 1 µm HP INNOWax column (HP 19095N-
123) was used to separate methanol, acetic acid, methyl
acetate, water, and ethylene glycol dimethyl ether
(solvent used to dilute the sample). The column tem-
perature was programmed to rise from an initial value
of 50 to 80 °C at 30 °C/min, followed by a 60 °C/min
ramp up to 130 °C, and then held constant at 130 °C
for an additional 1.5 min. High-purity helium gas
(Merriam Graves 1071-300) with a minimum purity of
99.997% was used as a carrier gas. The flow rate of
the carrier gas was 17 mL/min, and a complete GC run
took 3.33 min.
All four components were calibrated for their indi-
vidual GC responses, and the calibrations were verified
by mixtures of the components.
Procedure. For kinetic studies, the catalyst was air-
dried for roughly 2 weeks until the weight remained
constant (within 0.5 wt %). For the kinetic experiments,
two 500-mL flasks, one charged with a measured
amount of methanol and the other charged with meas-
ured amounts of acetic acid and the air-dried catalyst,
were first kept in the constant-temperature bath for
about 1 h to reach the desired temperature. Then the
reactants and catalyst were mixed together, and timing
was started. The total volume of liquid was 306 mL.
Periodically, a 0.1-mL sample was taken from the liquid
phase using a 1-mL syringe and immediately diluted
with 20 mL of anhydrous ethylene glycol dimethyl ether Figure 2. Adsorption equilibrium diagrams for the three non-
in a sample vial. Subsequently, 1 µL of the diluted reactive pairs at 45 °C.
sample was injected into the GC for analysis. At the Adsorption Experiments. Binary adsorption pa-
beginning of the experiment samples were taken at ∼10- rameters are needed for the kinetic models (described
min intervals; the sampling intervals were increased as in the next section). Among the four components in the
the experiment progressed. A kinetic experiment was methyl acetate reaction, it is possible to carry out three
stopped when the compositions leveled off, which was independent binary adsorption experiments. Of the four
judged to happen when the difference between three nonreactive binary pairs, we investigated (1) acetic acid
consecutive measurements taken at 1-h intervals were with water, (2) methanol with methyl acetate, and (3)
within instrument error (5%). Typically, about 20 methanol with water. These pairs were picked to
samples were taken for one run. maximize the expected adsorption strength difference
For residue curve measurements, the flask was in order to increase experimental accuracy.
charged with 306 mL of liquid mixture of desired initial The experimental setup for the adsorption experi-
compositions along with a measured amount of air-dried ments was similar to that for the kinetic experiments.
catalyst. While valve A of the distillation head was open The flask was immersed in a constant-temperature bath
and valve B was closed (see Figure 1), a constant heat maintained at 45 °C. The Amberlyst 15 catalyst was
input (117 W) was supplied and the liquid was totally dried under vacuum (500-635 mmHg) at 80 °C for 8 h
refluxed until temperatures of both liquid and vapor and then under vacuum at room temperature for
phases of the mixture were stabilized. Then valve A another 12 h. Thirty grams of the vacuum-dried
was closed and valve B was opened, and the vapor phase catalyst were used in all adsorption experiments.
was removed continuously and condensed into a sepa- A binary liquid mixture was prepared with a total
rate collection flask. During the distillation process, volume of ∼100 mL and a composition of 10 mol % of
liquid samples were taken periodically and measured the stronger adsorbing species. The mixture was heated
for their composition. Eventually, the amount of liquid in the constant-temperature bath for 40 min before the
residue in the reaction flask was small enough to vacuum-dried Amberlyst 15 catalyst was added. In
preclude sampling, and the experiment was stopped. A preliminary experiments, the composition of the mixture
typical experimental run took around 1 h to stabilize became constant within ca. 10 min after the addition of
and could be run for 2 and 3 h (depending on the initial the catalyst. To ensure a close approach to adsorption
conditions) to collect the equilibrium RCM data before equilibrium, a liquid sample was taken 20 min after the
the still ran dry. Subsequent runs, beginning at com- addition of the catalyst. The composition of the sample
positions as close as possible to the measured end point was then determined by GC.
of the previous experiment, were necessary in most The relative adsorption strengths for the components
cases to get a significant portion of each residue curve. were determined from literature (Mazzotti et al., 1997).
1920 Ind. Eng. Chem. Res., Vol. 37, No. 5, 1998

Table 1. Summary of the Kinetic Studies for Methyl Model and Data Analysis
Acetate
The LHHW Model. The batch kinetic data were in
mol of H+ initial initial total
run temp catalyst (from catalyst/ moles of moles of time the form of data sets for mole fraction of the components
no. (°C) (g) mol of mixture HOAc MeOH (h) as a function of time. Since the reactions were carried
out in the presence of a heterogeneous catalyst, we used
1 40 24 0.009 86 1.87 3.74 23.16
2 40 48 0.019 72 1.87 3.74 23.00 a heterogeneous LHHW model. The assumptions are
3 40 48 0.029 58 1.87 1.87 23.50 as follows: (1) the adsorption sites are uniformly
4 45 48 0.019 72 1.87 3.74 23.00 energetic; (2) there is a monolayer coverage (typical of
5 45 24 0.009 86 1.87 3.74 23.50 chemisorption); (3) a molecule on site 1 does not influ-
6 45 48 0.029 58 1.87 1.87 23.00 ence what attaches onto a nearby site 2; (4) a dual-site
7 50 24 0.009 86 1.87 3.74 5.50 mechanism is used for the reaction; and (5) the surface
8 50 48 0.019 72 1.87 3.74 22.00
9 50 48 0.029 58 1.87 1.87 21.00 reaction is the rate-controlling step. A possible mech-
10 50 12 0.004 93 1.87 3.74 23.92 anism for the esterification reaction is as follows (Fro-
11 50 48 0.019 72 3.74 1.87 11.00 ment and Bischoff, 1990):
12 50 24 0.014 79 1.87 1.87 5.50
13 50 24 0.009 86 1.87 3.74 23.00 HOAc + S S HOAc-S
14 50 48 0.019 77 2.80 2.80 12.00
MeOH + S S MeOH-S
Table 2. Experimental Data for the Methyl Acetate
System Run No. 1: Measured Liquid-Phase Mole HOAc-S + MeOH-S S MeOAc-S + H2O-S
Fractions vs Time
time MeOAc-S S MeOAc + S
(min) MeOAc MeOH H2O HOAc
1.5 0.0053 0.6005 0.1063 0.2879 H2O-S S H2O + S (3)
20 0.0551 0.5402 0.1550 0.2497
40 0.0917 0.5074 0.1884 0.2125 The resulting LHHW model is
60 0.1163 0.4746 0.2258 0.1833
90 0.1477 0.4559 0.2371 0.1593 rMeOAc )

( )
120 0.1682 0.4356 0.2638 0.1323
150 0.1826 0.4189 0.2809 0.1176 aMeOAcaH2O
180 0.1958 0.4096 0.2889 0.1057 kS,1KHOAcKMeOH aHOAcaMeOH -
Keq1
240 0.2143 0.3893 0.3117 0.0848
300 0.2276 0.3794 0.3213 0.0716 (1 + KHOAcaHOAc + KMeOHaMeOH + KMeOAcaMeOAc + KH2OaH2O)2
360 0.2391 0.3749 0.3230 0.0631
420 0.2453 0.3680 0.3316 0.0551 (4)
450 0.2462 0.3647 0.3323 0.0568
540 0.2513 0.3583 0.3405 0.0499 where S represents an active site and kS,1 is the forward
600 0.2525 0.3534 0.3480 0.0461 reaction rate constant for the surface reaction defined
660 0.2523 0.3499 0.3512 0.0466 as follows:
720 0.2560 0.3513 0.3477 0.0450
810 0.2578 0.3489 0.3516 0.0418 kS,1 ) kS0,1WeE/RT (5)
900 0.2563 0.3484 0.3531 0.0421
1390 0.2572 0.3441 0.3582 0.0404
W is the catalyst concentration (mol of H+ ions/mol of
Water is adsorbed most strongly, followed in decreasing mixture) and E is the activation energy for the surface
order of adsorption strength by methanol, acetic acid, reaction (kcal/mol). Ki are adsorption equilibrium con-
and methyl acetate. The next data point was obtained stants, ai are the liquid-phase activities and Keq1 is the
by adding more of the stronger adsorbing component thermodynamic equilibrium constant for reaction 1.
to the previous mixture in order to study the adsorption Lumping some of the constants in the numerator, we
over the entire range of liquid compositions. To limit get
the accumulation of error, measurements were stopped rMeOAc )

( )
when the total volume of the liquid in the flask reached
aMeOAcaH2O
∼300 mL. A subsequent run was then started at a k′S,1 aHOAcaMeOH -
composition close to the last data point of the previous Keq1
run, and this process was continued until the entire
(1 + KHOAcaHOAc + KMeOHaMeOH + KMeOAcaMeOAc + KH2OaH2O)2
composition range was investigated. Figure 2 shows the
adsorption curves obtained for the three binary pairs. (6)

where k′S,1 can be written as


Results
Fourteen sets of kinetic experiments were performed k′S,1 ) k′S0,1WeEAPP/RT (7)
at different molar ratios of reactants, catalyst concen-
trations, and temperatures; Table 1 gives the conditions. k′S0,1 is the preexponential factor, and EAPP is the
A typical data set for one of the kinetic experiments is apparent activation energy in kcal/mol. If we assume
given in Table 2. These data were regressed as de- that the adsorption equilibrium constants are independ-
scribed below. The model fit for the data in Table 2 is ent of temperature (i.e., that their heats of adsorption
given in Figure 5. The remaining data sets for the are small), then the apparent activation energy, EAPP,
methyl acetate kinetics are provided in the Supporting will be equal to the activation energy of the reaction,
Information (Tables S1-01-S1-13). The experimental E. The model has seven parameters: the preexponen-
data and the model fits for the adsorption experiments tial factor, the activation energy, the thermodynamic
are given in Figure 2. equilibrium constant, and the four adsorption equilib-
Ind. Eng. Chem. Res., Vol. 37, No. 5, 1998 1921

rium constants. Regression of all the coefficients in the adsorbed on the surface must be independent of surface
LHHW model from the reaction kinetics experiments composition. Therefore,
alone is problematic due to the correlation of param-
eters, giving rise to convergence difficulties (e.g., Parra nS1 + nS2 ) nS (14)
et al., 1994). As an alternative, we need independent
binary nonreactive adsorption experiments to obtain the where nS is the constant total number of moles which
adsorption equilibrium constants relative to a reference can be accommodated in the adsorbed phase by unit
component (methyl acetate). The next section describes mass of solid. Since xS1 ) nS1 /nS, eq 13 can be rewritten
the development of the required adsorption model. as
Modeling the Adsorption Experiments. For a
nonreactive binary adsorption experiment with two K2,1nSa1
components 1 and 2, an overall material balance gives nS1 ) (15)
the composite isotherm (Kipling, 1965; p 28) K2,1a1 + a2

n0∆x with a similar expression for nS2 . Substituting these


) nS1 x2 - nS2 x1 (8) expressions in eq 8 gives the composite isotherm
m
where n0 is the total initial number of moles in the liquid n0∆x nS(K2,1a1x2 - a2x1)
phase, ∆x is the change in the mole fraction in the liquid ) (16)
m K2,1a1 + a2
phase, m is the mass of the catalyst, and nS1 and nS2 are
the number of moles of 1 and 2 transferred onto the There are two parameters in this model, K2,1 and nS,
surface of unit mass of catalyst. The quantities n0, x1, which are found by nonlinear regression of the experi-
x2, ∆x, and m can all be measured, leaving two un- mental adsorption data.
knowns in eq 8, nS1 and nS2 . There are four pairs of components for which adsorp-
To get an expression relating the number of moles of tion experiments could be performed (the other pairs
each species adsorbed (nSi ) and the activities of the are reactive). Only three of these pairs are independent
components in the two phases, we consider the chemi- (the fourth ratio can be used as a consistency check).
sorption process as a reaction of the following form: Hence, we have three ratios of adsorption equilibrium
constants, and we can choose one of the Ki’s as a
1(l) + 2(s) S 1(s) + 2(l) (9) reference component and relate all the others to it. We
chose the adsorption equilibrium constant of methyl
We make the following assumptions about the nature acetate as the reference and related the other adsorption
of the adsorbed layer: (1) The adsorbed layer is confined coefficients to it through these experimentally deter-
to a single molecular layer, (2) the adsorption sites are mined ratios:
all equivalent, (3) the two components of the binary
mixture have the same molecular area in the adsorbed KHOAc ) K1,3KMeOAc
phase, and (4) the fraction of sites occupied is constant
for all binary pairs at all compositions. If we also KMeOH ) K2,3KMeOAc
assume that the liquid phase is nonideal and the solid
phase is ideal, the equilibrium constant for reaction 9 KH2O ) K4,3KMeOAc (17)
is

xS1 a2 The adsorption experiments were performed at 45 °C.


) K2,1 (10) The denominator of the LHHW rate expression can be
xS2 a1 written in terms of the reference component, and then
the kinetic model has only two adjustable parameters,
K2,1 can be thought of as a ratio of the equilibrium namely, the forward reaction rate constant k′S,1 and the
constants for two independent adsorption processes of adsorption equilibrium constant for the reference com-
the form ponent KMeOAc. The ratios of the adsorption equilibrium
constants obtained from the adsorption experiments and
1(l) + s S 1(s) K1 (xS1 ) K1a1) (11a) the value of KMeOAc obtained from the kinetic data fitting
at 45 °C were assumed to be independent of tempera-
ture.
2(l) + s S 2(s) K2 (xS2 ) K2a2) (11b)
The Equilibrium Constant. The reaction equilib-
rium constant Keq1 was obtained as follows. The kinetic
Equation 9 can be obtained by subtracting eq 11b from experiments were run to long times until the composi-
eq 11a and, hence, K2,1 is the ratio of K2 to K1 tion (mole fraction) vs time curves were essentially flat.
K2,1 ) K2/K1 (12) To determine whether the reaction had truly reached
equilibrium, we performed the following test. The
relative error of the GC was less than 5%. We per-
Since xS2 ) 1 - xS1 , it follows that formed an error propagation analysis to get the error
in the measured mole fractions, and at long times we
K2,1a1
xS1 ) (13) checked the change in mole fractions between three
K2,1a1 + a2 successive pairs of measurements. If the change in the
mole fraction was less than the error in the measure-
Because the total number of occupied sites is a constant ment for these three successive pairs, we considered all
and all molecules occupy the same number of sites (have points after the third pair to be at equilibrium. These
the same molecular area), the total number of moles data points were then used to get a value of Kx
1922 Ind. Eng. Chem. Res., Vol. 37, No. 5, 1998

Table 3. Thermodynamic Data for the Methyl Acetate/DME System


Antoine Coefficients
component A B C Vi (m3/mol)
acetic acid (1) 22.1001 -3654.62 -45.392 57.54
methanol (2) 22.4999 -3643.3136 -33.434 44.44
methyl acetate (3) 21.1520 -2662.78 -53.460 79.84
water (4) 23.2256 -3835.18 -45.343 18.07
DME (5) 21.2303 -2164.85 -25.344 69.07
Wilson Parameters
A11 ) 0.0 A12 ) 2535.2019 A13 ) 1123.1444 A14 ) 237.5248 A15 ) -96.6698
A21 ) -547.5248 A22 ) 0.0 A23 ) 813.1843 A24 ) 107.3832 A25 ) 900.9358
A31 ) -696.5031 A32 ) -31.1932 A33 ) 0.0 A34 ) 645.7225 A35 ) -17.2412
A41 ) 658.0266 A42 ) 469.5509 A43 ) 1918.232 A44 ) 0.0 A45 ) 703.3566
A51 ) 96.7797 A52 ) -418.6490 A53 ) -21.2317 A54 ) 522.2653 A55 ) 0.0

B
ln Psat ) A + Psat [Pa], T [K]
T+C

( )
C C xkΛki
ln γi ) 1 - ln( ∑ xjΛij) - ∑ C

∑x Λ
j)1 k)1
j kj
j)1

where

Λij )
Vj
Vi
exp ( )
-Aij
RT
Vj[m3/mol], Aij [cal/mol]

Aij ) 0 implies ideality


The dimerization constant in the vapor phase for acetic acid is
3166.0
log(KD) ) -12.5454 + KD [1/Pa], T [K]
T

Table 4. Minimum-Error Estimate of the Equilibrium Figure 3 shows a plot of the logarithm of the value of
Constant for the Methyl Acetate Reaction the thermodynamic equilibrium constants of all the
temp (°C) Keq temp (°C) Keq experimentally determined data points (at the three
40 30.2 50 24.0 different temperatures of 40, 45, and 50 °C (Table S3))
45 27.4 and one point at 154 °C from Menschutkin (1879) as a
function of reciprocal temperature. From the plot, the
following expression was obtained for the thermody-
()xMeOAcxH2O/xHOAcxMeOH). The liquid-phase activity co- namic equilibrium constant as a function of temperature
efficients for each point were obtained using the Wilson (T in K)
model with the parameters given in Table 3. Thether-
modynamic equilibrium constant for that point is found ln(Keq1) ) 0.839 83 + 782.98/T (19)
from
aMeOAcaH2O From this expression, we estimate the free energy of
Keq1 ) ) KxKγ (18) the reaction (∆G°) to be -1.55 kcal/mol. In the tem-
aHOAcaMeOH perature range of interest (40-120 °C), the value of Keq1
ranges from 28.2 to 17.0, which is slightly higher than
The thermodynamic equilibrium constant is only a the value quoted by earlier workers (Agreda et al., 1990;
function of temperature. The value of Keq at a particular Barbosa and Doherty, 1988). They used a value of Kx
temperature was obtained by adjusting Keq to minimize ) 5.2 at 154 °C from Menschutkin (1879). Then using
the squared error between calculated and measured a UNIFAC estimate for the liquid-phase activity coef-
equilibrium compositions of all the data points (with ficients γι and the fact that Keq ) KxKγ, they estimated
different molar ratios and catalyst concentrations). A the value for Keq1 to be 14.7.
detailed listing of the point values of Kx, Kγ, and Keq for If the Menschutkin data point is removed from Figure
the equilibrium composition data points and the mini- 3, the slope of the curve increases and the new expres-
mum-error estimate of the equilibrium constant at the sion for Keq1 is ln(Keq1) ) -0.8226 + 1309.8/T. In this
three experimental temperatures is given in the Sup- case, ∆G° ) -2.6 kcal/mol, and the value of Keq1 varies
porting Information (Table S3). Table 4 gives the from 28.8 to 12.3 over the temperature range of 40-
minimum-error estimate of the equilibrium constant at 120 °C. This does not affect our results significantly
the three different temperatures. since the change in Keq1 over the 40-50 °C range (over
An independent method for estimating Keq is from free which the kinetic experiments were performed) is minor.
energies of formation (i.e., Keq ) e∆G°/RT). However, the The value of Keq1 from these two correlations is signifi-
free energies of formation are large and their difference cantly different at the higher temperature end, but we
(∆G°) is small. The result, therefore, has a large believe that inclusion of the Menchutkin data point (at
uncertainty, and we did not find this method reliable. 154 °C) would give a better estimate of the equilibrium
Ind. Eng. Chem. Res., Vol. 37, No. 5, 1998 1923

Table 5. Results of the Regression for the Methyl


Acetate/DME System
Reaction Equilibrium Constant for Methyl Acetate
Keq1 ) 2.32e782.98/T
Adsorption Equilibrium Constants
KH2O ) 10.50
KMeOH ) 4.95
KHOAc ) 3.18
KMeOAc ) 0.82
Forward Reaction Rate Constants
k′S,1 ) e(24.64-(6287.7/T))
k′S,2 ) e(27.4-(10654.0/T))

Figure 3. Temperature dependence of the reaction equilibrium


constant for the methyl acetate reaction: ln(Keq) vs 1/T. Open
symbols represent the individual data points; filled symbols are
the minimum-error estimates.

constant at higher temperatures in comparison to a


correlation that is purely based on equilibrium constant
values at much lower temperatures.
Regression of Kinetic Data. A summary of the
kinetic experiments performed is given in Table 1. At
45 °C, the LHHW model has two parameters that need
to be estimated from the kinetic data fit, namely, the
forward reaction rate constant k′S,1 and the adsorption
equilibrium constant of the reference component (meth-
yl acetate). These were obtained by minimizing the Figure 4. Logarithm of the forward reaction rate constant for
following objective function: the MeOAc reaction as a function of reciprocal temperature.

6287.7
total time 4
ln(k′S,1) ) 24.64 - (21)
min f(x) ) ∑ ∑
time)0 j)1
(xj,expt - xj,model)2 (20) T (K)

Figure 5 shows a typical fit of the mole fraction for


where xj’s represent the mole fractions of the four each component as a function of time using the LHHW
species, namely, acetic acid, methanol, methyl acetate, model. As can be seen, the model does a very good job
and water. The minimization was done using a modi- of predicting the mole fraction profiles as a function of
fied Levenberg-Marquardt algorithm (the IMSL rou- time.
tine DBCLSF (IMSL MATH/LIBRARY, 1987)) embed- The DME Reaction. One of the side reactions in
ded within the LSODES integrator (Hindmarsh, 1983). the methyl acetate system is the dehydration of metha-
Once the adsorption equilibrium constants are deter- nol to dimethyl ether and water (eq 2). An increase in
mined at 45 °C, only one parameter, namely, the pressure leads to an increase in the boiling points of
forward reaction rate constant k′S,1, remains to be the components and hence the operating temperature
fitted using the kinetic experiments at the other two in reactive distillation. This leads to higher rates of
temperatures. This is because we have assumed that reaction and smaller equipment, leading to a savings
the value of the adsorption equilibrium constant for of investment costs. On the other hand, an increase in
methyl acetate and the ratios of the adsorption equi- temperature leads to a decrease in the equilibrium
librium constants for the other components are inde- constant (and hence the equilibrium conversion) and
pendent of temperature. The rate constants were could potentially trigger unwanted side reactions such
obtained by minimizing the objective function in eq 20. as the methanol dehydration reaction. Eleven sets of
The results are given in Table 5. Figure 4 shows the high-pressure experiments were carried out with an
logarithm of the forward reaction rate constant k′S,1 as initial charge of pure methanol both with and without
a function of reciprocal temperature for the methyl the catalyst to measure the extent of the dehydration
acetate reaction. The slope of the curve gives an reaction. A second apparatus similar to the one de-
apparent activation energy of 12.49 kcal/mol, and the scribed in Figure 1 was built out of stainless steel using
intercept gives a preexponential factor of 5.02 × 1010 a Parr reactor (model 4521 constructed from T316
(mol of mixture) (mol of H+ ions)-1 min-1. From the stainless steel with a volume of 1000 mL) with a
plot, we get the following Arrhenius expression for the maximum working pressure of 700 psig. The details of
forward reaction rate constant as a function of temper- the high-pressure equipment are given in Venimadha-
ature: van (1998).
1924 Ind. Eng. Chem. Res., Vol. 37, No. 5, 1998

Table 6. Summary of the Kinetic Studies for the DME


Reaction
mol of H+ initial initial total
run temp catalyst (from catalyst)/ moles of moles of time
no. (°C) (g) mol of mixture MeOH H2O (h)
1 100 25.0 0.009 65 9.88 0.280 12.0
2 100 49.0 0.018 60 9.88 0.659 12.5
3 100 71.0 0.028 31 9.88 0.610 12.0
4 100 0.0 0.000 00 9.84 0.140 12.0
5 100 72.0 0.028 29 9.88 0.688 12.0
6 100 72.0 0.028 55 9.88 0.590 12.0
7 100 130.0 0.015 26 9.88 9.880 12.0
8 100 82.4 0.027 21 9.88 1.190 12.6
9 100 72.9 0.028 27 9.88 0.660 12.0
10 115 71.0 0.028 61 9.88 0.570 12.0
11 85 72.0 0.028 56 9.88 0.590 12.0

Table 7. Experimental Data for the Kinetic Study of the


DME System (Run No. 10): Measured Liquid Mole
Fractions vs Time
time
(min) MeOH H2O DME
5 0.9361 0.0627 0.0012
120 0.8829 0.0856 0.0315
240 0.8213 0.1230 0.0557
360 0.7710 0.1534 0.0757
Figure 5. Methyl acetate system: isothermal batch kinetic data 480 0.7230 0.1739 0.1031
and model fit. The dots represent experimental data, and the solid 600 0.7020 0.1968 0.1012
line represents the heterogeneous (LHHW) model (T ) 40 °C, 720 0.6625 0.2272 0.1103
catalyst concentration ) 0.009 mol of H+/mol of mixture).
where ∆G° is in kcal/mol and R ) 0.001 987 kcal/mol‚K.
The reaction was not significant at low temperatures In the experimental temperature range (85-115 °C),
(pressures) even with high catalyst concentrations and the value for the thermodynamic equilibrium constant
large liquid holdups. However, at higher pressures and for the methanol dehydration reaction ranges from 68.4
catalyst concentrations (8.41-15.69 atm and 61 g of to 52.3. It was assumed that the amount of DME
catalyst for a 400-mL methanol charge), significant adsorbed was small and that the adsorption equilibrium
amounts of DME were formed. The LHHW model for constants for methanol and water had the same values
the methanol dehydration reaction is as those from the methyl acetate data fit. This leaves

( )
only one parameter, namely, the forward reaction rate
aDMEaH2O constant, to be estimated from the kinetic data for the
kS,2KMeOH2 aMeOH2 - DME reaction.
Keq2
rDME ) The kinetic data from the methanol dehydration
(1 + KMeOHaMeOH + KDMEaDME + KH2OaH2O)2 reaction were regressed in a manner similar to that for
(22) the methyl acetate reaction. Table 6 is a summary of
the kinetic experiments performed for the DME reac-
where kS,2 is the forward reaction rate constant for the tion. Table 7 gives the measured data from one of the
surface reaction, ai’s are the liquid-phase activities experimental runs (at 85 °C and a catalyst concentration
computed using the Wilson model with the parameters of 72 g of catalyst in ∼400 mL of methanol). The data
given in Table 3, Ki’s are the adsorption equilibrium for the remainder of the experimental runs are given
constants, and Keq2 is the thermodynamic equilibrium in the Supporting Information (Tables S2-01-S2-10).
constant for reaction 2. DME is the lightest boiling Figure 6 shows a plot of the logarithm of the forward
component in the mixture, with a boiling point of 44.68 reaction rate constant (k′S,2) as a function of the recip-
°C at 10 atm pressure. Isothermal batch kinetic experi- rocal temperature. From this graph, we find an appar-
ments were performed at various temperatures in the ent activation energy of 21.17 kcal/mol and a preexpo-
range 85-115 °C. If most of the DME is in the vapor nential factor of 7.9 × 1011 (mol of mixture) (mol of H+
phase and if we lump the constants in the numerator ions)-1 min-1, giving the following Arrhenius rate
together, eq 22 can be simplified to expression

rDME )
(
k′S,2 aMeOH2 -
aDMEaH2O
Keq2 ) (23)
ln(k′S,2) ) 27.40 -
10654.0
T (K)
Figure 7 shows a sample fit of the mole fraction for
(25)

(1 + KMeOHaMeOH + KH2OaH2O)2 each component as a function of time for the methanol


dehydration data given in Table 7. The solid line is the
The thermodynamic equilibrium constant for the computed value while the dots are the experimental
methanol dehydration reaction was taken from Nisoli data points. Our results indicated that the amount of
et al. (1997): DME formed was quite small at lower temperatures
(40-50 °C) but became significant at higher tempera-
∆G°
Keq2 ) exp -( RT
,) tures (85-115 °C).
Residue Curve Maps. As mentioned earlier, a
∆G° ) -2.4634 1.5167 × 10-3T (K) (24) useful tool for the synthesis and analysis of both
Ind. Eng. Chem. Res., Vol. 37, No. 5, 1998 1925

is the reference forward reaction rate constant (defined


at a convenient temperature such as the lowest boiling
temperature in the mixture), W is the normalized weight
of catalyst in moles of H+ ions per moles of mixture,
and V0 is the initial rate of the vapor leaving the still.
The model for a kinetically controlled reactive distil-
lation system is given below (for the detailed derivation
of the model; see Venimadhavan et al., 1994).

dxi H V0 N

) xi - yi + Da
H0 V j)1

(νi,j - νT,jxi)Ri

(i ) 1, ..., (R + P - 1)) (27)


where xi and yi are the liquid and vapor phase mole
fractions, respectively, dξ is a “warped” time variable
defined by dξ ) (V/H) dt, V is the instantaneous vapor
rate, H is the instantaneous liquid holdup in the still, t
is the clock time, νi,j is the stoichiometric coefficient of
species i in reaction j, νT,j is the sum of the stoichiometric
coefficients in reaction j, and

Ri ) rj/k′S,1,ref (j ) 1, ..., N) (28)


Figure 6. Logarithm of the forward reaction rate constant for
the DME reaction as a function of reciprocal temperature. where Ri is the normalized reaction vector, rj is the rate
of an individual reaction, and k′S,1,ref is the reference
surface reaction rate constant of the first reaction.
The i’s correspond to the five species, namely, acetic
acid, methanol, methyl acetate, water, and DME. The
rj’s are rMeOAc and rDME which were determined earlier
using the data from the kinetic and adsorption experi-
ments. The reference reaction rate constant is chosen
as the value for the methyl acetate reaction at the
boiling point of DME at 1 atm which is -24.72 °C (the
lowest point on the boiling surface). The matrix νi,j and
the vector νT,j in eq 27 are respectively

y) ( -1 -1 1 1 0
0 -2 0 1 1 ) (29)

and

yT ) ()
0
0
(30)

A limiting case that is well-understood is the case of


Figure 7. DME system: Isothermal batch kinetic data fit. The
equilibrium reactive distillation where simultaneous
dots represent experimental data, and the solid line represents
the heterogeneous (LHHW) model (T ) 85 °C, catalyst concentra- vapor-liquid and reaction equilibrium prevail. The
tion ) 0.028 mol of H+/mol of mixture). assumption of reaction equilibrium allows us to define
certain transformed composition variables which make
nonreactive and reactive distillation systems is the the analysis of the equilibrium reactive residue curve
residue curve map (e.g., Fidkowski et al., 1993; Foucher maps easier (Barbosa and Doherty, 1988; Ung and
et al., 1991; Wahnschafft et al., 1992; Venimadhavan Doherty, 1995a-d).
et al., 1994). An important parameter in the study of As a test of our model at the equilibrium limit, we
the kinetically controlled reactive distillation system is performed independent equilibrium open-evaporation
a Damkohler number defined by the ratio of a charac- experiments. One way of ensuring that the system
teristic residence time to a characteristic reaction time reaches equilibrium rapidly is to have a large amount
(Venimadhavan et al., 1994). The Damkohler number of catalyst present in the reactor. We ensured that
can be thought of as a dimensionless holdup in the there was a sufficiently large amount of catalyst present
system or a dimensionless rate of reaction. For a by performing several experiments starting with the
heterogeneous system, the Damkohler number can be same number of moles of acetic acid and methanol but
redefined to explicitly show its dependence on a normal- with increasing amounts of catalyst on successive runs.
ized catalyst concentration as follows: For each run, we plotted the data on a transformed mole
fraction diagram until a further increase in the amount
H0/V0 of catalyst did not result in any appreciable shift in the
Da ) (26) RCM (Figure 8). We then concluded that, under the
1/k′S,1,refW given conditions, the reaction could not proceed any
faster and thus we had achieved almost instantaneous
where H0 is the initial molar holdup in the still, k′S,1,ref reaction and phase equilibrium. Earlier calculations
1926 Ind. Eng. Chem. Res., Vol. 37, No. 5, 1998

Figure 8. Effect of the mass of catalyst for simultaneous reaction


and phase equilibrium.

(Figure 8) showed that, for all values of Da > 80, the


structure of the residue curve map was indistinguish-
able from that at simultaneous vapor-liquid and reac-
tion equilibrium (Da f ∞). The residue curves com-
puted from our model at Da ) 99 (chosen well above
the minimum required to model equilibrium behavior)
were compared to the experimentally obtained residue
curves (Figure 9a). Figure 9b gives the curves computed
by Barbosa and Doherty (1988). The qualitative agree-
ment between their diagram and the experimental data
is good considering the fact that they had obtained the
value of Kx ) 5.2 from Menschutkin (1879) and esti-
mated the value of Kγ ≈ 2.83 using the UNIFAC method
to get an overall value of Keq ≈ 14.7 (independent of
temperature). This value is somewhat lower than the
Figure 9. (a) Comparison of computed (solid line) and experi-
range of values that we obtained from the experimental mentally measured equilibrium reactive RCM for the methyl
data in this work (Keq in the range of 20-30). The solid acetate system at P ) 1 atm. (b) RCM for the methyl acetate
lines in Figure 9a show the computed curves, while the system at vapor-liquid and reaction equilibrium computed by
symbols show the measurements. The agreement is Barbosa and Doherty (1988).
good, especially considering that none of the parameters
in the model were adjusted to fit the measurements. The usually written in terms of the liquid-phase activities
good agreement appears to validate the modeling ap- of the components rather than the more common
proach and provides confidence in applying our kinetic concentration-based rate expressions used in reaction
and VLE models to reactive distillation systems of engineering. The methyl acetate synthesis is a classic
commercial interest. problem for the application of reactive distillation, but
We used the model to predict residue curve maps at there is a lack of kinetic rate data/expressions in the
lower values of Da in the kinetically controlled regime. range of interest for this system. We performed kinetic
Figure 10 shows the results at Da ) 1, 10, and 50. The experiments on a solid sulfonic acid resin catalyst
model predicts that the lightest boiler is always the (Amberlyst 15W) for the methyl acetate reaction as well
methyl acetate-methanol azeotrope, that the heaviest as for an unwanted side reactionsthe dehydration of
boiler is always pure acetic acid, and that methyl methanol to dimethyl ether and water. We developed
acetate, methanol, and water are always intermediate a heterogeneous (LHHW) model for the reaction kinetics
boilers for all values of Da. Since the structure of the for these two reactions. To decouple the reaction and
residue curve map is qualitatively the same for all the the adsorption phenomena, we performed independent
values of Da, we predict that methyl acetate can be nonreactive binary adsorption experiments. These ex-
successfully produced by reactive distillation in both the periments gave estimates of the three independent
equilibrium-controlled and the kinetically controlled ratios of the adsorption equilibrium constants. The
regimes. remaining parameters of the LHHW model were fitted
from the kinetic data. The resulting kinetic models fit
Conclusions the experimental data quite well. These rate expres-
sions are applicable over a wide range of catalyst
Analysis of kinetically controlled reactive distillation concentrations and molar ratios of reactants. To test
systems requires knowledge of the chemical kinetics. the equilibrium limit of the model, we performed
Rate expressions for reactive distillation systems are independent equilibrium reactive open-evaporation ex-
Ind. Eng. Chem. Res., Vol. 37, No. 5, 1998 1927

Acknowledgment

We are grateful for the financial support provided by


the National Science Foundation (CTS-9613489) and the
National Environmental Technology Institute. We are
also grateful to Ms. Pamela Stephan for preparing the
illustrations.

Supporting Information Available: Tables of


experimental data for kinetic study of the methyl
acetate and the DME systems for several runs and
equilibrium constants for the methyl acetate reaction
(26 pages). Ordering information is given on any
current masthead page.

Notation

ai ) activity of component i
Da ) Damkohler number ()(H0/V0)/(1/kS,1,refW))
E ) activation energy (kcal/mol)
EAPP ) apparent activation energy (kcal/mol)
∆G° ) change in free energy (kcal/mol)
H ) liquid holdup (mol)
H0 ) initial liquid holdup (mol)
kf ) forward reaction rate constant for a homogeneous
reaction (1/time)
kS,j ) forward reaction rate constant for surface reaction j
(1/time)
k′S,j ) apparent forward reaction rate constant for surface
reaction j (1/time)
k′S,j,ref ) reference apparent reaction rate constant for the
surface reaction (1/time)
Keqj ) thermodynamic equilibrium constant for reaction j
Ki ) adsorption equilibrium constant for component i
Ki,j ) ratio of the adsorption equilibrium constants for
components i and j
Kx ) reaction equilibrium constant based on mole fractions
Kγ ) ratio of the activity coefficients of the products to those
of the reactants
m ) mass of catalyst for the adsorption experiments
mol ) gmol
nSi ) number of moles of species i transferred onto the
surface of unit mass of catalyst
n0 ) total initial number of moles in the liquid phase
rj ) rate of reaction for reaction j (1/time)
R ) universal gas constant (1.987 cal/mol‚K)
Rj ) normalized reaction vector defined as rj/k′S,1,ref
S ) active site on the catalyst
t ) time (min)
T ) temperature (K)
V ) vapor rate (mol/time)
V0 ) initial vapor rate (mol/time)
Figure 10. RCMs at Da ) 1, 10, and 50 on transformed W ) normalized catalyst concentration (mol of H+ ions/
coordinates for the methyl acetate reaction. mol of mixture)
∆x ) change in the liquid-phase mole fraction for the
periments. The experimental residue curves were adsorption experiments
compared to the predictions generated by the model, and xi ) liquid-phase mole fraction for component i
the agreement was found to be very good considering xSi ) mole fraction on the catalyst for component i
that none of the parameters in the model were adjusted yi ) vapor phase mole fraction for component i
to fit the simple distillation data. The good agreement
Greek Symbols
appears to validate our vapor-liquid equilibrium and
kinetic models and gives us confidence in applying these ξ ) “warped” time
models to the reactive distillation of systems of com- νi,j ) stoichiometric coefficient of component i in reaction j
mercial interest. νT,j ) sum of the stoichiometric coefficients for reaction j
1928 Ind. Eng. Chem. Res., Vol. 37, No. 5, 1998

Subscripts Oost, C.; Hoffmann, U. The Synthesis of Tertiary Amyl Methyl


Ether (TAME): Microkinetics of the Reactions. Chem. Eng. Sci.
i ) component i 1996, 51 (3), 329-340.
0 ) initial value Parra, D.; Tejero, J.; Cunill, F.; Iborra, M.; Izquierdo, J. F. Kinetic
Study of MTBE Liquid-Phase Synthesis Using C4 Olefinic Cut.
Literature Cited Chem. Eng. Sci. 1994, 49 (24A), 4563-4578.
Rehfinger, A.; Hoffmann, U. Kinetics of Methyl Tertiary Butyl
Agreda, V. H.; Partin, L. R. Reactive Distillation Process for the Ether Liquid-Phase Synthesis Catalyzed by Ion Exchange
Production of Methyl Acetate. U.S. Patent 4,435,595, Mar 6, Resin-I. Intrinsic Rate Expression in Liquid-Phase Activities.
1984. Chem. Eng. Sci. 1990, 45 (6), 1605-1617.
Agreda, V. H.; Partin, L. R.; Heise, W. H. High Purity Methyl Rev, E. Reactive Distillation and Kinetic Azeotropy. Ind. Eng.
Acetate via Reactive Distillation. Chem. Eng. Prog. 1990, 86 Chem. Res. 1994, 33 (9), 2174-2179.
(2), 40.
Siirola, J. J. An Industrial Perspective on Process Synthesis.
Barbosa, D. A. G.; Doherty, M. F. The Simple Distillation of
AIChE Symp. Ser. 304 1995, 304, 222-33.
Homogeneous Reactive Mixtures. Chem. Eng. Sci. 1988, 43, 541.
Bernot, C.; Doherty, M. F.; Malone, M. F. Feasibility and Separa- Smith, H. A. Kinetics of the Catalyzed Esterification of Normal
tion sequencing in Multicomponent Batch Distillation. Chem. Aliphatic Acids in Methyl Alcohol. J. Am. Chem. Soc. 1939, 61,
Eng. Sci. 1991, 46 (5/6), 1311-1326. 254.
DeGarmo, J. L.; Parulekar, V. N.; Pinjala, V. Consider Reactive Smith, L. A. Catalytic Distillation Process. U.S. Patent 4,307,254,
Distillation. Chem. Eng. Prog. 1992, 88 (3), 43-50. Dec 22, 1981.
Doherty, M. F.; Buzad, G. Reactive Distillation by Design. Trans. Smith, L. A. Catalytic Distillation Structure. U.S. Patent 4,443,559,
Inst. Chem. Eng. 1992, 70A, 448-458. April 17, 1984.
Espinosa, J. P.; Aguirre, P.; Perez, G. A. Product Composition Smith, L. A. Method for the preparation of methyl tertiary butyl
Regions of Single-Feed Reactive Distillation Columns: Mixtures ether. U.S. Patent 4,978,807, Dec 18, 1990.
Containing Inerts. Ind. Eng. Chem. Res. 1995, 34, 853. Sundmacher, K.; Hoffmann, U. Macrokinetic Analysis of MTBE-
Fidkowski, Z. T.; Doherty, M. F.; Malone, M. F. Feasibility of Synthesis in Chemical Potentials. Chem. Eng. Sci. 1994, 49 (18),
Separations for Distillation of Nonideal Ternary Mixtures. 3077-3089.
AIChE J. 1993, 39, 9 (8), 1303-1321. Ung, S.; Doherty, M. F. Necessary and Sufficient Conditions for
Foucher, E.; Malone, M. F.; Doherty, M. F. Automatic Screening the Reactive Azeotropes in Multireaction Mixtures. AIChE J.
of Entrainers in Homogeneous Azeotropic Distillation. Ind. Eng. 1995a, 41, 2383.
Chem. Res. 1991, 30, 760. Ung, S.; Doherty, M. F. Vapor-Liquid-Phase Equilibrium in
Froment, G. F.; Bischoff, K. B. Chemical Reactor Analysis and Systems with Multiple Chemical Reactions. Chem. Eng. Sci.
Design, 2nd ed.; Wiley Series in Chemical Engineering; Wiley, 1995b, 50, 23-48.
New York, 1990; pp 61-83. Ung, S.; Doherty, M. F. Theory of Phase Equilibria in Multireac-
Grosser, J. H.; Doherty, M. F.; Malone, M. F. Modeling of Reactive tion Systems. Chem. Eng. Sci. 1995c, 50, 3201.
Distillation Systems. Ind. Eng. Chem. Res. 1987, 26, 983. Ung, S.; Doherty, M. F. Calculation of Residue Curve Maps for
Hindmarsh, A. C. ODEPACK: A Systematized Collection of ODE Mixtures with Multiple Equilibrium Chemical Reactions. Ind.
Solvers in Scientific Computing; Stepleman, R. S., et al., Eds.; Eng. Chem. Res. 1995d, 34, 3195.
North-Holland: Amsterdam, The Netherlands, 1983; pp 55-
Venimadhavan, G. Ph.D. Dissertation, 1998.
64.
IMSL MATH/LIBRARY. FORTRAN Subroutines for Mathematical Venimadhavan, G.; Buzad, G.; Doherty, M. F.; Malone, M. F. Effect
ApplicationssIMSL Problem-Solving Software Systems, Version of Kinetics on Residue Curve Maps for Reactive Distillation.
1; IMSL: New York, 1987, pp 876-881. AIChE J. 1994, 40 (11), 1814-1824. Correction: AIChE J. 1995,
Kipling, J. J. Adsorption from Solutions of Non-Electrolytes; 41 (12), 2613.
Academic Press: New York, 1965; pp 23-69. Wahnschafft, O. M.; Koehler, J. W.; Blass, E.; Westerberg, A. W.
Mazzotti, M.; Neri, B.; Gelosa, D.; Kruglov, A.; Morbidelli, M. The Product Composition Regions for Single-Feed Azeotropic
Kinetics of Liquid-Phase Esterification Catalyzed by Acidic Distillation Columns. Ind. Eng. Chem. Res. 1992, 31, 2345.
Resins. Ind. Eng. Chem. Res. 1997, 36 (1), 3-10. Xu, Z. P.; Chuang, K. T. Kinetics of Acetic Acid Esterification over
Menschutkin, N. Ueber den einflufs der isomerie der alkohole und Ion Exchange Catalysts. Can. J. Chem. Eng. 1996, 74, 493-
der sauren auf die bildung zusammen-gesetzter aether. Justus 500.
Leibigs Ann. Chem. 1879, 195, 334-364.
Neumann, R.; Sasson, Y. Recovery of Dilute Acetic Acid by Received for review December 1, 1997
Esterification in a Packed Chemorectification Column. Ind. Eng. Revised manuscript received February 4, 1998
Chem. Process Des. Dev. 1984, 23 (4), 654-659. Accepted February 5, 1998
Nisoli, A.; Malone, M. F.; Doherty, M. F. Attainable Regions for
Reaction with Separation. AIChE J. 1997, 43, 3 (2), 374-387. IE9708790

You might also like