See

discussions, stats, and author profiles for this publication at: http://www.researchgate.net/publication/231370452

N-Butyl Acetate Synthesis via Reactive

Distillation: Thermodynamic Aspects, Reaction
Kinetics, Pilot-Plant Experiments, and
Simulation Studies

ARTICLE in INDUSTRIAL & ENGINEERING CHEMISTRY RESEARCH · OCTOBER 2002

Impact Factor: 2.59 · DOI: 10.1021/ie020179h

CITATIONS READS

81 615

2 AUTHORS:

Sven Steinigeweg Juergen Gmehling

7 PUBLICATIONS 308 CITATIONS Carl von Ossietzky Universität Oldenburg
371 PUBLICATIONS 11,073 CITATIONS
SEE PROFILE

SEE PROFILE

Available from: Juergen Gmehling

Retrieved on: 16 December 2015
 Ind. Eng. Chem. Res. 2002, 41, 5483-5490 5483

n-Butyl Acetate Synthesis via Reactive Distillation: Thermodynamic

Aspects, Reaction Kinetics, Pilot-Plant Experiments, and Simulation
Studies
1 rgen Gmehling*
Sven Steinigeweg and Ju
Industrial Chemisty, Carl von Ossietzky University of Oldenburg, P.O. Box 2503,
D-26111 Oldenburg, Germany

The development of a heterogeneously catalyzed reactive distillation process for the production
of n-butyl acetate by the esterification of n-butanol with acetic acid is presented. Thermodynamic
aspects of the considered system are discussed, and UNIQUAC interaction parameters are given.
The reaction was catalyzed heterogeneously by a strongly acidic ion-exchange resin (Amberlyst-
15). The reaction kinetics was investigated, and the kinetic constants for a pseudohomogeneous
kinetic model are presented. Reactive distillation experiments were performed using the
structured catalytic packing Katapak-S. Several operation conditions were varied (reboiler duty,
reactant ratio, total feed rate), and two different setups were realized experimentally. The
experimental results are presented in comparison with simulation results. An equilibrium stage
model is capable of describing the experiments quantitatively. n-Butanol conversions of 98.5%
accompanied with n-butyl acetate purities of 96.9% were achieved using an equimolar total
column feed. The simulation was used subsequently to determine the influence of important
design factors such as feed location, use of a prereactor, and numbers of reactive and nonreactive
stages, resulting in an optimized process.

Introduction able to replace the toxic and teratogenic ethoxy ethyl

acetate that is often used as a solvent.5
Reactive distillation has received increasing attention Investigations of different process alternatives for the
over the past several years as a promising alternative production of n-butyl acetate have been performed by
to conventional processes. Especially for equilibrium- Hartig and Regner6 and Block and Hegner.7 These
limited and consecutive reactions, reactive distillation authors considered the homogeneously catalyzed reac-
offers distinct advantages through the direct removal tion. Theoretical investigations of the n-butyl acetate
of the reaction products by distillation. Therefore, system were presented by Löning et al.8 These authors
conversions far beyond equilibrium conversions and focus their attention on n-butyl acetate hydrolysis and
higher selectivities can be obtained, leading to signifi- analyzed reactive residue curve maps to gain insights
cantly lower investment and operating costs. Although into the different kinds of singular points present in the
invented in 1921,1 the industrial application of reactive system. Also, data on the reaction kinetics of n-butyl
distillation did not take place before the 1980s.2 acetate hydrolysis and the chemical equilibrium are
Especially interesting equilibrium reactions suitable provided, but no information about a process for the
for reactive distillation are esterifications, ester hy- production of n-butyl acetate is presented. Recently,
drolysis reactions, etherifications, and transesterifica- Hanika et al.9 investigated the esterification of n-
tions. In recent years, attention has been paid to methyl butanol with acetic acid by reactive distillation. These
acetate synthesis and hydrolysis, which serves as a authors included only a few data about pilot-plant
model system for reactive distillation processes.3,4 experiments, without supplying data on reaction kinet-
In contrast to methyl acetate synthesis, information ics. Many design strategies assume that chemical equi-
about n-butyl acetate synthesis by esterification of librium is reached instantaneously on each reactive
n-butanol with acetic acid (eq 1) in a reactive distillation tray. It was shown recently by Pöpken et al.3 that this
column can hardly be found in the literature. simplification is critical for methyl acetate synthesis,
and one can expect that the reaction kinetics is much
acetic acid (HOAc) + n-butanol (BuOH) h more important for n-butyl acetate synthesis.
n-butyl acetate (BuOAc) + water (H2O) (1) This paper presents a systematic approach to the
development of a heterogeneously catalyzed reactive
distillation process for the production of n-butyl acetate
n-Butyl acetate is an important solvent in the chemical including reliable thermodynamic and kinetic data that
industry. Primarily it is used in paint and coating cannot be found in the literature. Heterogeneously
manufacture and in the lacquer industry. Because of catalyzed reactive distillation offers advantages com-
its lower impact on the environment, n-butyl acetate is pared to homogeneous catalysis, e.g., by sulfuric acid.
The size and location of the reactive section can be
* To whom correspondence should be addressed. Tel.: chosen without regard for thermodynamic constraints,
+49-441-7983831. Fax: +49-441-7983330. E-mail: gmehling@ and corrosion problems can be minimized. Immobiliza-
tech.chem.uni-oldenburg.de. Internet: http://www.uni- tion of the heterogeneous catalyst inside the reactive
oldenburg.de/tchemie/. section of the column can be achieved by structured
10.1021/ie020179h CCC: $22.00 © 2002 American Chemical Society
Published on Web 10/03/2002
5484 Ind. Eng. Chem. Res., Vol. 41, No. 22, 2002

Table 1. Uniquac ri and qi Values Table 2. UNIQUAC Interaction Parameters Used for the
Calculation of Activity Coefficients
component ri qi
component 1 component 2 i j aij (K) bij cij (K-1)
acetic acid 2.2024 2.0720
n-butanol 3.4543 3.0520 n-butanol water 1 2 1783.6 -10.037 1.33 × 10-2
n-butyl acetate 4.8274 4.1960 2 1 -3842.1 23.860 -3.34 × 10-2
water 0.9200 1.4000 acetic acid n-butanol 1 2 -198.40 1.563 0.0
2 1 162.28 -1.106 0.0
n-butyl acetate water 1 2 3512.6 -16.964 2.33 × 10-2
packings such as Katapak-S (Sulzer Chemtech), which 2 1 -775.4 4.448 -4.53 × 10-3
is used in this investigation for pilot-plant experiments. acetic acid n-butyl acetate 1 2 -61.31 -0.137 0.0
2 1 162.09 0.279 0.0
Katapak-S is made of corrugated wire-mesh sheets. n-butanol n-butyl acetate 1 2 -48.26 0.200 -4.54 × 10-4
Catalyst particles of 0.5 to about 2 mm can be fixed 2 1 260.06 -0.499 2.12 × 10-4
between the sheets of this packing. In this investigation, water acetic acid 1 2 -98.12 -0.294 -7.67 × 10-5
2 1 422.38 -0.051 -2.40 × 10-4
a strongly acidic ion-exchange resin (Amberlyst-15,
Rohm & Haas) was used as the catalyst. Amberlyst-15 to describe the temperature dependency of the activity
has already been applied successfully for methyl acetate coefficients.14 The resulting parameters are presented
synthesis and hydrolysis.3,4 This catalyst was also em- in Table 2, including the parameters recently published
ployed by Löning et al.8 for n-butyl acetate hydrolysis. by Horstmann et al.14 for the system acetic acid-water.
The procedure for the development of reactive distil- To account for the vapor-phase nonidealities, which are
lation processes applied here can be divided into four mainly caused by the dimerization of the carboxylic acid,
steps. In the first step, the thermodynamic properties the Nothnagel equation,15 which is based on chemical
of the system are investigated to obtain a reliable theory,16 was used.
description of liquid-phase nonidealities that can be
used for a preliminary feasibility analysis.10,11 Then, the Experimental Section
reaction kinetics is investigated at conditions that are
expected for the reactive distillation column. These Chemicals. The chemicals used for the reaction
results can be combined, and a simulation environment kinetics experiments were of analytical grade (99.8%,
based on an equilibrium stage model can be employed. Scharlau). Water was bidistilled. For the titrimetric
Experiments in a reactive distillation column on a pilot- analysis, Titrisol (1.0 and 0.1 N sodium hydroxide,
plant scale are carried out subsequently to verify the Merck) was used. The organic chemicals were dried over
simulation. Finally, simulation studies can be used to a molecular sieve prior to use. For the reactive distil-
identify the roles of important design parameters (e.g., lation experiments, the chemicals were of reaction grade
numbers of reactive and nonreactive stages, number and (99.5%) and were used without further purification.
location of feed positions, role of a prereactor) and to Acetic acid was supplied by Celanese, and n-butanol was
propose an optimal setup. supplied by BASF. The purity of all chemicals was
verified by gas chromatography.
Thermodynamic Aspects Analytics. All samples for reactive distillation ex-
periments were analyzed by gas chromatography (HP
The activity coefficients γi are necessary for the liquid- 6890 instrument with TCD; He as the carrier gas at 2.4
phase nonidealities to be taken into account. In this cm3 min-1; HP-Innowax 30 m × 0.032 mm column; split
work, the UNIQUAC equation12 was employed for the 10:1; temperature program ) 338.15 K hold for 5 min,
calculation of γi values to be used not only for the followed by heating at a rate of 80 K min-1 to 453.15 K;
determination of phase and chemical equilibria but also accuracy of mole fractions ) (0.001).
for the satisfactory description of the reaction kinetics. Reaction Kinetics. Löning et al.8 presented kinetic
The temperature dependence of the UNIQUAC interac- data for the heterogeneously catalyzed hydrolysis reac-
tion parameters was represented by polynomials of the tion. In this article, emphasis is placed on an accurate
form description of the reaction kinetics of the heteroge-
neously catalyzed synthesis reaction. Therefore, mainly
∆uij ) aij + bijT + cijT 2 (2) the synthesis reaction was studied. Two additional
kinetic experiments were performed for the hydrolysis
Van der Waals properties r and q (Table 1) were taken of n-butyl acetate to ensure that the kinetic constants
from the Dortmund Data Bank (DDB), version 2002, obtained are also valid for the backward reaction.
which was kindly placed at our disposal by DDBST Experiments. The experiments were conducted in
GmbH Oldenburg, Germany.13 UNIQUAC interaction a thermostated glass reactor with a volume of 500 cm3.
parameters were fitted simultaneously to the different The temperature of the heating jacket was kept constant
types of phase equilibrium data [vapor-liquid equilib- within (0.1 K. The stirrer was plate-type, and the speed
rium (VLE), liquid-liquid equilibrium (LLE), and azeo- was variable between 100 and 800 rpm. To improve
tropic data; activity coefficients at infinite dilution (γ∞); mixing, a baffle was installed. Furthermore, a reflux
heats of mixing (hE)]. Because of the complex phase condenser was installed to avoid any loss of volatile
equilibrium behavior with numerous azeotropes and components. Prior to use, the Amberlyst-15 catalyst was
large areas in the composition space where two liquid washed several times with water until the supernatant
phases are present, only reliable and thermodynami- liquid was colorless. Before each kinetic experiment was
cally consistent data can be used for the determination started, both reactants were brought to reaction tem-
of the UNIQUAC parameters. For the reactive systems perature in separate vessels. When the desired temper-
(n-butanol-acetic acid, n-butyl acetate-water), ad- ature was reached, the reactor was filled with both
ditional VLE and hE measurements were made by our reactants and the catalyst, and time measurement was
group. These data will be published elsewhere soon. started. Liquid samples of about 1 cm3 volume were
VLE measurements were performed in a static ap- taken using a syringe, cooled to 270 K to avoid any
paratus at low temperature. Heats of mixing were used further reaction, and weighed (accuracy of the balance
 Ind. Eng. Chem. Res., Vol. 41, No. 22, 2002 5485

Table 3. Kinetic Parameters for the Pseudohomogeneous

Kinetic Model
reaction i k0i (mol g-1 s-1) EA,i (kJ mol-1)
esterification 1 6.1084 × 104 56.67
hydrolysis -1 9.8420 × 104 67.66

) (0.001 g). Measurements were performed in the

temperature range between 331.15 and 363.15 K. In
addition to the temperature of the experiments, the
amount of catalyst and the initial reactant ratio were
varied. Most of the experiments were continued until
chemical equilibrium was reached. In all cases, the
water content of the catalyst was determined after the
kinetic experiment by drying the catalyst at 353 K
under vacuum until the mass remained constant (usu-
ally after 2 days). The ion-exchange capacity of Am-
berlyst-15 was previously determined.17 Figure 1. Arrhenius diagram of the rate constants for the
To determine the influence of external mass transfer esterification reaction k1 (b) and the hydrolysis reaction k-1 (O)
on the reaction rate, the stirrer speed was varied of the heterogeneously catalyzed reaction. (Lines: overall fit.)
between 100 and 600 rpm in additional runs. No
influence of the stirrer speed on the reaction rate was
detected above 200 rpm. Therefore, further experiments
were conducted at a stirrer speed of 400 rpm. Pöpken
et al.17 already showed the absence of internal mass
transfer resistance for methyl acetate synthesis. Also
Xu and Chuang18 stated that internal diffusion is
insignificant for the esterification of acetic acid with
methanol catalyzed by Amberlyst-15. Because Am-
berlyst-15 is composed of very small gel-type micro-
spheres with large macropores,19 internal mass transfer
can also be excluded for n-butyl acetate synthesis.
Data Analysis. Both pseudohomogeneous and ad-
sorption-based models have been used to describe es-
terification reactions that are heterogeneously catalyzed
by ion-exchange resins. For methyl acetate synthesis,
a pseudohomogeneous model and an adsorption-based
model were previously discussed,3 and it was shown that Figure 2. Rate constants for the esterification reaction k1 (b) at
a pseudohomogeneous model is sufficient for the de- 333 K versus catalyst mass. (Line: overall fit.)
scription of the column profiles if small to medium
amounts of water are present in the column. Esterifi- Reactive Distillation Experiments. n-Butyl ac-
cation reactions are known to be reversible reactions of etate, as the high-boiling product, is obtained as the
second order. Therefore, the pseudohomogeneous model bottom product. The n-butanol-water-n-butyl acetate,
can be written as n-butanol-water, and n-butyl acetate-water azeotropes
have similar boiling points,20 and therefore, n-butanol,
1 1 dni n-butyl acetate, and water are the main components
r) ) k1aHOAcaBuOH - k-1aBuOAcaH2O (3) present in the distillate. At higher reboiler duties, acetic
mcat νi dt
acid can also be found in the distillate.
Setup. To achieve countercurrent flow in the reactive
Activities are used instead of concentrations or mole section of the column, the low-boiling reactant is usually
fractions. This leads to a more consistent and accurate fed into the column below the reactive section, and the
description.17 The temperature dependency of the rate high-boiling reactant is fed above the reactive section
constant is expressed by Arrhenius’ law of the column. The normal boiling point of acetic acid

( )
with 391.01 K is very close to that of n-butanol (390.53
-EA,i
ki ) k0i exp (4) K), and therefore, the optimal feed locations are not
RT obvious but an important design factor. Therefore, two
different setups were employed. Figure 3 shows the first
This means that four adjustable parameters (k01, k-1 0
, setup. For the second setup, the n-butanol and acetic
EA,1, EA,-1) have to be fitted for the pseudohomogeneous acid feed locations were reversed. Experiments on the
model. The values are given in Table 3. The Arrhenius pilot-plant scale were performed in a glass column
diagram (Figure 1) shows the results from the simul- supplied by QVF Engineering with an inner diameter
taneous fit of all data at a given temperature (k1, k-1) of 50 mm. A vessel with a low liquid holdup (approx-
and the result from the simultaneous fit of all experi- imately 1 dm3) was applied as a reboiler. In the reboiler,
mental data available (k01, k-1 0
, EA,1, EA,-1). The kinetic the liquid was heated electrically by rod-shaped quartz
model (eq 3) makes the assumption of a linear relation- heaters (Vogelsberger Quarzglastechnik). The reboiler
ship between the catalyst mass and the reaction rate. duty was controlled with a transformer and determined
That this is a valid assumption can be seen from Figure with a digital multimeter (Voltcraft M-3860M) to within
2, which shows the experimental data in comparison (1%. The reactive section consisted of Katapak-S ele-
with calculated reaction rates. ments filled with Amberlyst-15, whereas the nonreac-
5486 Ind. Eng. Chem. Res., Vol. 41, No. 22, 2002

Figure 3. Setups of the reactive distillation experiments.

Figure 4. Profile of liquid-phase composition for S1-6. Experi-
tive sections contained Sulzer-BX packings. Each col- mental data: BuOH, 4; BuOAc, 0; H2O, ×; HOAc, O. Simulation
umn section had a height of 1.2 m and an effective results: BuOH, ---; BuOAc, -‚-; H2O, - -; HOAc, s.
packing height of 1 m. Except for the flanges and the
reboiler, the column was insulated by a vacuum jacket.
The flanges and reboiler were insulated by mineral wool.
Nevertheless, the heat loss was measured previously,
and a linear relationship between the temperature
inside the column and the heat loss was found.3 The
heat loss was determined experimentally and can be
described by the following equations

Q̇loss,reboiler ) 3.8 W K-1∆T (5a)

Q̇loss,per section ) 0.39 W K-1∆T (5b)

The distillate stream was condensed and collected in a

decanter operated at about 359 K, which was installed
at the top of the column. The aqueous phase was
withdrawn, whereas the organic phase was completely
refluxed into the column. All feed streams were mea-
sured by determining the mass flow using balances with Figure 5. Profile of liquid-phase composition for S2-4. Experi-
an accuracy of (1%. The column was controlled by a mental data: BuOH, 4; BuOAc, 0; H2O, ×; HOAc, O. Simulation
process control system based on Opto boards (Optoware) results: BuOH, ---; BuOAc, -‚-; H2O, - -; HOAc, s.
that were connected to a PC (WinNT workstation).
Temperatures were measured using Pt-100 thermom- ment S1-2, n-butanol conversions of 98.5% in combina-
eters (accuracy of (0.1 K). Thermometers were installed tion with n-butyl acetate purities of 96.9% were achieved.
at the lower end of each column section (except at the Figures 6 and 7 show the conversion as a function of
feed positions), in the reboiler, and in the decanter. The the reboiler duty for setups 1 and 2, respectively.
top and bottom pressures as well as the pressure drop
were recorded by pressure transducers (Bosch, accuracy Simulation
(0.1%). At the lower end of each column section, in the
reboiler, and from the aqueous stream leaving the All simulations were carried out with the steady-state
decanter, liquid samples were withdrawn by the means model RADFRAC from the process simulator Aspen
of a syringe, immediately cooled to 270 K to avoid any Plus (version 10.2).21 The model is based on a rigorous
further reaction, and analyzed by gas chromatography. equilibrium stage model for solving the MESH equa-
Below the reactive section, the liquid load of the column tions. In addition to the UNIQUAC interaction param-
was measured by recording the time needed to collect a eters, data for the column heat loss and reaction kinetics
specified amount (40 cm3) of liquid in a graduated vessel are incorporated into the process simulator. The model-
inside the column. ing of the decanter was achieved by the DECANTER
Experimental Results. A typical composition profile model of Aspen Plus. For the prereactor (323.15 K),
for the first setup is shown in Figure 4 and for the chemical equilibrium was assumed. The equilibrium
second setup in Figure 5. Experimental details for the constant was taken from Löning et al.8
first setup are given in Table 4 and for the second setup The separation efficiency of Katapak-S3 was incorpo-
in Table 5. In this work, the aim was not only to rated into the simulation model without changes. For
determine the influence of relevant parameters experi- low to medium water contents, an NTSM value of 4 m-1
mentally, but also to show that it is possible to obtain was found for Katapak-S, whereas an NTSM value of 5
high conversions accompanied by high purities of n- m-1 was assumed for Sulzer-BX.3 For the small range
butyl acetate. As can be seen from Table 4, in experi- of liquid loads (responding to F factors of about 0.1-
 Ind. Eng. Chem. Res., Vol. 41, No. 22, 2002 5487

Table 4. Experimental Data for the First Setup

run number S1-1 S1-2 S1-3 S1-4 S1-5 S1-6 S1-7
P1 (mbar) 995.1 1038.45 1033.84 1032.0 1028.31 1026.46 1004.79
∆P (mbar) 4.22 20.18 4.83 2.42 3.05 7.71 7.4
ḞHOAc (mol/h) 34.0 38.0 35.0 35.0 26.3 23 32
ḞBuOH (mol/h) 35.0 36.0 36.0 36.0 26.5 52 40
Ḃ (mol/h) 42.0 35.0 39.0 52.0 31.9 51 44
stream leaving decanter 27.0 38.0 29.0 19.0 20.8 24 28
(mol/h)
molar ratio HOAc/BuOH 0.97 1.05 1.03 0.97 0.99 0.44 0.80
xdec (HOAc) 0.143 0.031 0.132 0.211 0.158 0.121 0.166
xdec (BuOH) 0.000 0.008 0.003 0.002 0.004 0.006 0.002
xdec (BuOAc) 0.008 0.003 0.004 0.007 0.004 0.002 0.004
xdec (H2O) 0.849 0.958 0.861 0.780 0.834 0.870 0.828
xB (HOAc) 0.171 0.003 0.120 0.289 0.166 0.001 0.136
xB (BuOH) 0.287 0.008 0.263 0.392 0.283 0.586 0.371
xB (BuOAc) 0.517 0.969 0.607 0.272 0.528 0.381 0.460
xB (H2O) 0.025 0.020 0.010 0.047 0.023 0.031 0.033
XBuOH (%) 64.58 98.5 69.42 41.56 66.33 38.09 56.47
Q̇reb (W) 1070 2200 1150 900 850 1290 1100
wL (m h-1) 9.16 14.66 8.15 4.07 6.11 9.78 9.78
ϑfeed,HOAc (°C) 18.38 20.44 20.13 20.25 20.88 21.19 21.38
ϑfeed,BuOH (°C) 19.31 21.94 21.44 21.38 22.19 22.56 22.69
ϑdecanter (°C) 82.75 84.69 79.69 75.56 74.06 81.88 79.31
ϑ14,L (°C) 93.13 118.06 95.63 96.13 95.31 117.38 94.94
ϑ28,V (°C) 119.0 124.94 120.25 118.0 120.00 115.63 118.25
ϑ28,L (°C) 119.75 126.13 121.06 118.0 120.88 116.63 119.13

Table 5. Experimental Data for the Second Setup

run number S2-1 S2-2 S2-3 S2-4 S2-5
P1 (mbar) 996.95 996.03 994.65 994.64 1002.02
∆P (mbar) 9.16 4.59 1.56 6.44 7.72
ḞHOAc (mol/h) 37.0 35.0 35.0 24.0 49.0
ḞBuOH (mol/h) 36.0 36.0 36.0 47.0 24.0
Ḃ (mol/h) 37.0 42.0 54.0 46.0 49.0
stream leaving decanter 35.0 29.0 17.0 25.0 25.0
(mol/h)
molar ratio HOAc/BuOH 1.03 0.97 0.97 0.51 2.04
xdec (HOAc) 0.000 0.000 0.000 0.000 0.000
xdec (BuOH) 0.009 0.027 0.001 0.041 0.018
xdec (BuOAc) 0.001 0.008 0.000 0.009 0.001
xdec (H2O) 0.990 0.965 0.990 0.951 0.981
xB (HOAc) 0.050 0.130 0.332 0.003 0.475 Figure 6. Conversion versus reboiler duty for setup 1. Experi-
xB (BuOH) 0.001 0.162 0.348 0.511 0.002 mental data, O; simulation results, s.
xB (BuOAc) 0.930 0.629 0.307 0.476 0.474
xB (H2O) 0.020 0.079 0.014 0.009 0.050
XBuOH (%) 98.89 77.67 48.05 45.33 98.00
Q̇reb (W) 1340 1200 930 1200 1150
wL (m h-1) 7.33 6.11 4.07 7.33 8.14
ϑfeed,HOAc (°C) 19.13 20.25 20.31 20.25 20.25
ϑfeed,BuOH (°C) 17.69 19.56 19.75 18.06 18.19
ϑdecanter (°C) 84.63 79.65 64.13 72.63 79.75
ϑ14,L (°C) 125.06 118.44 94.00 113.31 120.75
ϑ28,V (°C) 123.19 123.38 111.56 115.63 119.50
ϑ28,L (°C) 124.44 124.63 113.56 116.63 120.38

1.5 Pa0.5) used during this investigation, the separation

efficiency can be considered independent of the liquid
load.
Stages are numbered from the top to the bottom, with
stage 1 as the condenser and stage N as the reboiler.
This results in 28 theoretical stages for the pilot-plant Figure 7. Conversion versus reboiler duty for setup 2. Experi-
column equipped with Katapak-S (Figure 3). mental data, O; simulation results, s.

Process Development stages. After determining the influence of these param-

eters, a final process can be proposed. This process can
It can be seen from Figures 4-7 that the experimental be used for a scale-up and optimization with respect to
data are in good agreement with the simulation results. economic issues. All further calculations were conducted
The deviations are within experimental error for both with the assumption of zero heat loss.
setups. To ensure the most effective reactive distillation Pressure. An option to increase conversion in reac-
process, the influence of important design factors on tive distillation processes is to increase the column
conversion has to be evaluated. The major design pressure and, thereby, to increase the temperature in
parameters are the column pressure, reboiler duty, feed the reactive section leading to an enhancement of the
flow rates, number and location of feed positions, role reaction rate. This is not possible using Amberlyst-15
of a prereactor, and numbers of reactive and nonreactive for this reaction because the increase in the column
5488 Ind. Eng. Chem. Res., Vol. 41, No. 22, 2002

Figure 9. Conversion versus the feed location of the prereactor

stream (reactive stage above which the stream is fed into the
column) (equimolar feed, total feed rate of 0.14 kmol h-1, 10
reactive stages, five stages each above and below reactive section).

Figure 8. Conversion versus reboiler duty for different column

setups. Both reactants fed above the first reactive stage (s). Acetic
acid fed above the first reactive stage, n-butanol into the middle
of the reactive section (-‚-). n-Butanol fed above the first reactive
stage, acetic acid into the middle of the reactive section (---).

pressure would result in temperatures inside the reac-

tive section of the column that are significantly above
398 K, which is the maximum operating temperature
for Amberlyst-15 recommended by the manufacturer.19
Reboiler Duty. It can be concluded from Figures 6
and 7 that conversions close to 100% can be obtained
at high reboiler duties. The second setup experimentally
investigated is more favorable because conversions close
Figure 10. Conversion versus number of reactive stages (equimo-
to 100% can be achieved at lower reboiler duties. Better lar feed, total feed rate of 0.07 kmol h-1, five stages each above
acetic acid partitioning is achieved at lower reboiler and below reactive section). Overall conversion with (s) and
duties in the case of the second setup. without (- -) a prereactor.
Feed Location. The influence of the feed location on
the conversion and reboiler duty is presented in Figure
8. It can be seen that the most effective approach is to
feed both reactants into the column above the first
reactive stage because high conversions can be expected
at low reboiler duties. Feeding n-butanol at the top and
acetic acid in the middle of the reactive section or vice
versa results in higher reboiler duties necessary to
obtain nearly total conversion. Feeding both reactants
above the first reactive stage results in low water
contents and a good distribution of the reactants in the
reactive section.
Role of a Prereactor. For the prereactor, chemical
equilibrium was assumed. The application of a prere-
actor is a favorable process alternative for minor dif- Figure 11. Conversion versus number of reactive stages (equimo-
ferences in the boiling points of the reactants. Other- lar feed, total feed rate of 0.14 kmol h-1, five stages each above
wise, the use of a prereactor will have a detrimental and below reactive section). Overall conversion with (s) and
effect because both reactants are fed into the column without (- -) a prereactor.
at the same stage and countercurrent flow of the Number of Reactive Stages. Figures 10 and 11
reactants in the reactive section cannot be realized. show the influence of the number of reactive stages on
Calculations were performed with a sufficient reboiler conversion. It can be seen that the process including the
duty that high conversions could be expected. (Usually, prereactor is more effective because higher conversions
it was assumed that 30% of the total feed is withdrawn can be obtained for a lower number of reactive stages.
as bottoms.) From Figure 9, it can be concluded that For total feed rates up to 0.14 kmol h-1, a minimum
the optimal feed location of the stream leaving the number of 10 reactive stages is needed. An increase in
prereactor is above the first reactive stage. This results the total feed rate would demand a higher number of
in an optimal distribution of the reactants in the reactive stages. The difference between the process with
reactive section. Feeding the prereactor stream into the and without a prereactor increases with increasing total
column below the first reactive stage leads to a higher feed flow.
amount of water in the sections above the feed position Number of Nonreactive Stages. Further calcula-
and, hence, to lower conversions in these sections. tions were performed to evaluate the number of nonre-
 Ind. Eng. Chem. Res., Vol. 41, No. 22, 2002 5489

influence of the total molar feed on the number of the

reactive stages was studied. It was shown that 10
reactive stages are necessary for total molar feed rates
of 0.14 kmol h-1. The numbers of nonreactive stages
above and below the reactive section are of minor
importance for the performance of the overall process.
Finally, a process was proposed that can serve as the
basis for the scale-up of the process and further opti-
mization with respect to economic issues.

Acknowledgment
We thank the “Fonds der Chemischen Industrie” for
providing a scholarship to S.St. and Celanese and BASF
for providing the chemicals required for the measure-
Figure 12. Scheme of the process developed (equimolar feed, total ments. Also, we express our thanks to Sulzer Chemtech
feed rate of 0.07 kmol h-1). for the packings. Furthermore, we thank S. Laue for
performing some of the kinetic measurements.
active stages necessary. The calculations performed
using a prereactor and 8 reactive stages indicate that 2 Nomenclature
nonreactive stages above the reactive section and 4
nonreactive stages below the reactive section are neces- Greek Letters
sary to achieve high conversions. Only when high-purity γi ) activity coefficient of component i
n-butyl acetate is required a higher number of nonre- γ∞i ) activity coefficient at infinite dilution of component i
active stages below the reactive section of the column ϑ ) temperature (°C)
is needed. The separation efficiency offered by the
Katapak-S elements is sufficient for the purification of Latin Letters
n-butyl acetate in the lower column sections. At the top, BuOAc ) n-butyl acetate
no further separation stages are required because the BuOH ) n-butanol
decanter ensures a sufficient separation of water from EA ) apparent activation energy
the organic compounds because of the low solubilities hE ) heat of mixing
of n-butanol and n-butyl acetate in water. H2O ) water
Final Process. The previous evaluation permits us HOAc ) acetic acid
to present a process in Figure 12 that combines a ki ) kinetic constant
minimal number of stages with high conversions and, k0i ) preexponential factor
therefore, represents a very effective manner of produc- LLE ) liquid-liquid equilibrium
ing n-butyl acetate. A scale-up of this process to a mcat ) mass of catalyst
technical plant and a further optimization with regard P ) pressure
to economic issues is promising. T ) temperature
∆uij ) UNIQUAC interaction parameter between compo-
Conclusion nents i and j
VLE ) vapor-liquid equilibrium
A procedure for the development of reactive distilla- wL ) liquid load
tion processes was applied. The thermodynamic aspects xi ) mole fraction
were discussed, and UNIQUAC interaction parameters X ) conversion
were determined with the help of the Dortmund Data
Subscripts
Bank (DDB). The reaction kinetics of n-butyl acetate
synthesis was investigated, and kinetic constants for the 1 ) esterification reaction (in kinetic expressions; other-
pseudohomogeneous model based on activities were wise, stage 1)
derived. The results were incorporated into the process -1 ) ester hydrolysis reaction
simulator Aspen Plus (Aspen RADFRAC). Several reac- B ) bottom stream
tive distillation experiments in a pilot-plant column dec ) stream leaving the decanter
were performed with variations in several operating i ) stage number or component
conditions, including the reboiler duty, total feed flow, L ) liquid
and reactant ratio. Furthermore, two different setups V ) vapor
were applied. n-Butanol conversions of 98.5% accom-
panied by n-butyl acetate purities of 96.9% were achieved Literature Cited
experimentally for an equimolar reactant ratio. A
(1) Backhaus, A. A. Continuous Process for the Manufacture
comparison of the experimental data with simulation of Esters. U.S. Patent 1,400,849, 1921.
results indicate that an equilibrium stage model is (2) Agreda, V. H.; Partin, L. R.; Heise, W. H. High-Purity
capable of describing the column profiles quantitatively. Methyl Acetate via Reactive Distillation. Chem. Eng. Prog. 1990,
With the help of the reliable simulation, the influence 86, 40.
of several important design factors was investigated. It (3) Pöpken, T.; Steinigeweg, S.; Gmehling, J. Synthesis and
was shown that the use of a prereactor is advantageous Hydrolysis of Methyl Acetate by Reactive Distillation Using
Structured Catalytic Packings: Experiments and Simulation. Ind.
and that the stream leaving the prereactor should be Eng. Chem. Res. 2001, 40, 1566.
fed into the column directly above the reaction section. (4) Krafczyk, J.; Gmehling, J. Einsatz von Katalysatorpackun-
The number of reactive stages necessary to achieve gen für die Herstellung von Methylacetat durch reaktive Rekti-
conversions close to 100% was determined, and the fikation. Chem.-Ing.-Tech. 1994, 66, 1372.
5490 Ind. Eng. Chem. Res., Vol. 41, No. 22, 2002

(5) Kroschwitz, J. I., Ed. Kirk-Othmer Encyclopedia of Chemical (15) Nothnagel, K. H.; Abrams, D. S.; Prausnitz, J. M. General-
Technology. 4. Bearing Materials To Carbon, 4th ed.; Wiley ized Correlation for Fugacity Coefficients in Mixtures at Moderate
Interscience: New York, 1992; p 696. Pressures. Ind. Eng. Process Des. Dev. 1973, 12, 25.
(6) Hartig, H.; Regner, H. Verfahrenstechnische Auslegung (16) Marek, J.; Standart, G. Vapor-Liquid Equilibria in Mix-
einer Veresterungskolonne. Chem. Ing. Technol. 1971, 43, 1001. tures Containing an Associating Substance. I. Equilbrium Rela-
(7) Block, U.; Hegner, B. Verfahrensvarianten eines Verester- tionships for Systems with an Association Component. Collect.
ungsverfahrens. Verfahrenstechnik 1977, 11, 157. Czech. Chem. Commun. 1954, 19, 1074.
(8) Löning, S.; Horst, C.; Hoffmann, U. Theoretical Investiga-
tions on the Quaternary System n-Butanol, Butyl Acetate, Acetic (17) Pöpken, T.; Götze, L.; Gmehling, J. Reaction Kinetics and
Acid and Water. Chem. Eng. Technol. 2000, 23, 789. Chemical Equilibrium of Homogeneously and Heterogeneously
(9) Hanika, J.; Kolena, J.; Smejkal, Q. Butyl Acetate via Catalyzed Acetic Acid Esterification with Methanol and Methyl
Reactive DistillationsModelling and Experiment. Chem. Eng. Sci. Acetate Hydrolysis. Ind. Eng. Chem. Res. 2000, 39, 2601.
1999, 54, 5205. (18) Xu, Z. P.; Chuang, K. T. Effect of Internal Diffusion on
(10) Frey, T.; Stichlmair, J. Thermodynamische Grundlagen der Heterogeneous Catalytic Esterification of Acetic Acid. Ind. Eng.
Reaktivdestillation Chem. Ing. Technol. 1998, 70, 1373. Chem. Res. 1997, 52, 33011.
(11) Bessling, B.; Löning, J.; Ohligschlaeger, A.; Schembecker, (19) Pitochelli, A. R. Ion Exchange Catalysis and Matrix Effects;
G.; Sundmacher, K. Investigations on the Synthesis of Methyl Rohm and Haas Co.: Philadelphia, PA, 1980.
Acetate in a Heterogeneous Reactive Distillation Process. Chem.
Eng. Technol. 1998, 21, 393. (20) Gmehling, J.; Menke, J.; Krafczyk, J.; Fischer, K. Azeo-
(12) Abrams, D. S.; Prausnitz, J. M. Statistical Thermodynam- tropic Data; VCH: Weinheim, Germany, 1994.
ics of Liquid Mixtures: A New Expression for the Excess Gibbs (21) Physical Property Methods and Models. Aspen Plus, ver-
Energy of Partly or Complete Miscible Systems. AIChE J. 1975, sion 10; Aspen Technologies Inc.: Cambridge, MA, 1999.
21, 116.
(13) Dortmund Data Bank; DDBST GmbH: Oldenburg, Ger- Received for review March 7, 2002
many, 2002 (www.ddbst.de). Revised manuscript received August 22, 2002
(14) Horstmann, S.; Pöpken, T.; Gmehling, J. Phase Equilibria Accepted August 28, 2002
and Excess Properties for Binary Systems in Reactive Distillation
Processes Part I. Methyl Acetate Synthesis. Fluid Phase Equilib.
2001, 180, 221. IE020179H