Chemical Engineering and Processing 43 (2004) 791–801

Investigation of different column configurations for the

ethyl acetate synthesis via reactive distillation夽
M. Klöker a , E.Y. Kenig a,∗ , A. Górak a , A.P. Markusse b , G. Kwant b , P. Moritz c
a Department of Chemical Engineering, University of Dortmund, Emil-Figge-Str. 70, D-44227 Dortmund, Germany
b DSM Research, P.O. Box 18, NL-6160 MD Geleen, Netherlands
c Sulzer Chemtech Ltd., Separation and Reaction Technology, P.O. Box 65, CH-8404 Winterthur, Switzerland

Received 21 January 2003; received in revised form 14 April 2003; accepted 15 April 2003

Abstract

The ethyl acetate synthesis via reactive distillation is studied theoretically and experimentally using different catalytic packings. Experiments
are carried out at laboratory scale in a 50 mm diameter column with a packing height of 3 m, and at semi-industrial scale in a 162 mm diameter
column with a packing height of 12 m. The experimental set-up is similar for both cases. The commercially available packings studied are
KATAPAK® -S and two different variants of MULTIPAK® . Modelling is performed with a rate-based stage model. The simulation environment
ASPEN Custom ModelerTM is used for the implementation and solution of the model equations. The results of the rate-based simulations agree
well with the corresponding experimental results. In addition, suitable operating conditions and the influence of the selected catalytic internal
on conversion and product purity are investigated. The developed model enables the scale-up from laboratory to industrial size columns, based
on the respective packing characteristics.
© 2003 Elsevier B.V. All rights reserved.

Keywords: Ethyl acetate synthesis; Reactive distillation; Catalytic internals; Column configuration; Scale-up

1. Introduction fore, the choice of catalytic column internals as well as the

Reactive distillation has been successfully used in several
important industrial processes and its field of application is
expanding. However, the design of reactive distillation pro-
cesses still remains a challenge, since it is difficult to obtain
process models capable of reliably describing several com-
plex and interrelated phenomena including simultaneous
reaction and separation. As a result, safe scale-up from lab-
oratory experiments to industrial plants is quite challenging.
The use of solid catalysts in reactive separation processes
has led to a new design of column internals on the basis
of trays, random and structured packings. The internals for
reactive separations have to enhance both separation and
reaction and maintain a sound balance between them. There-
placement of catalytic and non-catalytic zones in the col-
umn are crucial for the overall performance of the column.
This study deals with the heterogeneously catalysed es-
terification of ethanol and acetic acid to ethyl acetate and
water by reactive distillation. Experiments are performed
at two different scales, in a 50 and 162 mm diameter col-
umn. For this system, rate-based modelling is applied and
the simulations are compared with experimental results.
The influence of operating conditions and of the selected
catalytic column internals on the overall performance is
investigated. Moreover, the scale-up ability of reactive dis-
tillation columns equipped with structured catalytic column
internals is considered.

2. Catalytic column internals for reactive distillation

夽 Extended version of the lecture presentation at the International
Conference on Distillation & Absorption, September 30–October 2, 2002,
Baden-Baden, Germany.
∗ Corresponding author. Tel.: +49-231-755-2357;
fax: +49-231-755-3035.
E-mail address: e.kenig@ct.uni-dortmund.de (E.Y. Kenig).
The basic concept of reactive distillation has already been
introduced in the 1920s [1]. However, until recent times, this
process has not found significant application. The progress
in modelling and simulation, as well as impressive examples

0255-2701/$ – see front matter © 2003 Elsevier B.V. All rights reserved.
doi:10.1016/S0255-2701(03)00084-9
792 M. Klöker et al. / Chemical Engineering and Processing 43 (2004) 791–801

realised in industrial practice, such as methyl acetate pro-

duction via reactive distillation [2], have led to increasing
interest in such processes.
The integration of both reaction and distillation in one
apparatus can result in significant capital and operational
cost savings, especially in the case of equilibrium limited
reactions, such as esterifications and etherifications. Due to
the continuous removal of products from the reaction zone,
both conversion and selectivity are increased. Another ben-
efit is that, for exothermal reactions, the heat of reaction
is directly used for the evaporation of liquid components
and thus contributing to energy savings. Furthermore, in Fig. 1. KATAPAK® -S laboratory packing (left), KATAPAK® -S 250.Y
some cases reactive distillation can be applied for the sep- (right).
aration of azeotropic or close boiling mixtures, e.g. the
separation of meta- and para-xylene [3]. However, reactive Whereas for KATAPAK® -S the granular catalyst is
distillation is not suitable for extreme reaction conditions, embedded between two wire gauze sheets (Fig. 1),
such as reaction pressure far above 10 bar or tempera- MULTIPAK® consists of corrugated wire gauze sheets
ture far above 400 ◦ C, because of increased operating and and catalyst bags made of the same material assembled in
equipment costs. Design and control of such integrated pro- alternating sequence (Fig. 2).
cesses still remain—due to their complexity—challenging Compared with conventional non-reactive structured
tasks. packings, the specific surface area is moderate. Pressure
In the past decade much of the research in this field has drop is low for all four types of packings. Flooding occurs
focused on the use of solid catalysts for reactive distilla- at F-factors of about 2 for moderate liquid loads. Details
tion. This process is also known as catalytic distillation. No on the applied commercially available packings are shown
separation of the catalyst is necessary, if the solids are im- in Table 1. The catalyst fraction of MULTIPAK® -2 is sig-
mobilised in packings, on trays or in form of a catalytic nificantly higher as compared with MULTIPAK® -1 and
coating, which is advantageous as compared with homoge- the laboratory scale KATAPAK® -S. However, at laboratory
neously catalysed reactive distillation. Some investigations scale, specific surface area and separation efficiency are
have also been made on solid catalysts distributed as slurry similar for all three constructions.
on trays [4]. The construction of MULTIPAK® enables high column
The use of immobilised catalyst facilitates the exact defi- loads via open channels for the vapour flow and sufficient
nition of reaction and separation zones. However, one of the mass transfer efficiency, by a segmentation of the catalyst
major disadvantages of catalytic distillation is the catalyst bags and numerous contact spots with the wire gauze sheets.
exchange in case of deactivation, since the catalyst support The construction of laboratory-scale and technical-scale
has to be replaced or refilled with new catalyst. The place- packings is very uniform which helps to ensure similar
ment of catalyst containers in the downcomers and on trays hydrodynamic behaviour. Experimental investigation of the
has also been investigated [5,6]. methyl acetate synthesis has proven the packing’s applica-
On the basis of structured and random packings, two dif- bility for catalytic distillation [10].
ferent techniques have been established. The first technique Similar to MULTIPAK® , the KATAPAK® -S construction
is the catalytic coating of packings [7,8] or even the man- benefits from large catalyst contents and high column loads
ufacturing from catalytically active material. The second due to the open channels for the vapour flow. The separation
method is the immobilisation of granular catalyst in bags as efficiency for the laboratory packing is considerably high,
part of these internals. For the latter technique, conventional with number of theoretical stages per meter (NTSM) values
granular catalyst can be used by the packing manufacturer,
while for the coating procedure, which is in many cases
rather complex, the support of the catalyst supplier becomes
necessary.
Among the constructions with immobilised granular cat-
alyst are catalytic bales [9] and structured packings such as
KATAPAK® and MULTIPAK® .
In this work, two different types of packings have been se-
lected for investigation, KATAPAK® -S and MULTIPAK® ,
each of them in two variants. Both packing types combine
the benefits of modern structured packings, such as low
pressure drop and high throughput and offer the advantages
of heterogeneous catalysis. Fig. 2. Catalytic packing MULTIPAK® .
 M. Klöker et al. / Chemical Engineering and Processing 43 (2004) 791–801
Table 1
Geometric data of the applied catalytic packings
KATAPAK® -S Laboratory
Diameter (mm) 50
Height (mm) 100
Hydraulic diameter (mm) 6.4
Material
Catalyst fraction (vol.%) 15.8
Specific surface area (m2 /m3 ) 270
NTSM (ca.) (1/m) 3–3.5
of about 3, further increasing near the loading region [11].
However, separation efficiency is lower for the industrial
type KATAPAK® -S 250.Y, because of the smaller specific
surface area. Several authors have experimentally studied
reactive distillation processes with KATAPAK® -S, among
them the synthesis and hydrolysis of methyl acetate [12]
and the formation of 2-methyl-propyl acetate [13]. In these
experiments, the packing showed good suitability.
3. Modelling basis
Because of the complex interactions of both separation
and reaction phenomena, the prediction of the overall per-
formance of internals in reactive distillation columns is
difficult. Therefore, in this work, simulations have been per-
formed, in which the characteristics of the applied internals
and their influence on mass transfer, hydrodynamics and
chemical reactions are directly accounted for. As a basis
for the simulations the relevant mathematical models have
been developed and validated.
For the investigation of complex integrated separation
processes like the synthesis of ethyl acetate via reactive
distillation, reliable models are essential in order to re-
duce the development time and expensive experimental
investigations.
Traditional equilibrium stage models and efficiency ap-
proaches are often inadequate for reactive separation pro-
cesses. In multicomponent mixtures diffusion interactions
can lead to unusual phenomena and it is even possible to ob-
serve mass transport of a component in the direction oppo-
site to its own driving force—the so-called reverse diffusion
[14]. For multicomponent systems, the stage efficiencies are
different for different components and may range from −∞
to +∞. To avoid possible significant errors in the parame-
ter estimation, it is necessary to take into account the actual
mass and heat transfer rates and hydrodynamics [14,15].
In this work, the so-called rate-based approach [16] for
the description of a column segment is applied by which a
direct account of process kinetics is realised. The essence
of this approach is that actual rates of multicomponent mass
transport, heat transport and chemical reactions are con-
sidered immediately in the equations governing the stage
phenomena. Mass transfer at the vapour–liquid interface is
described via the well-known two-film model.
793
MULTIPAK® -1 MULTIPAK® -2 KATAPAK® -S 250.Y
50 50 162
166 166 210
9.5 9.8 22.5
Wire gauze
16.0 27.7 25.0
317 262 125
4 3–3.5 1
Multicomponent diffusion in the films is covered by the
Maxwell–Stefan equations which can be derived from the
kinetic theory of gases [17]. The influence of the process hy-
drodynamics is taken into account by applying correlations
for mass transfer coefficients, specific contact area, liquid
hold-up, residence time distribution and pressure drop. The
two-film model concept has been extended by considering
chemical reactions in the liquid bulk phase. A sketch of the
applied non-equilibrium stage model is shown in Fig. 3. The
necessary hydrodynamic and mass-transfer correlations for
the catalytic packings have been obtained experimentally
in the context of the European research project INTelligent
Column INTernals for Reactive Separations (INTINT).
The models described above are implemented into
the simulation environment ASPEN Custom ModelerTM
(ACM).
Supplementary models have been developed for periph-
erals, such as reboiler, decanter, pre-reactor, heater and con-
denser. These models are linked together in the simulation
flowsheet of ACM. ASPEN physical properties database
Properties PlusTM is connected with the models and allows
comfortable changes from one chemical system to another
as well as the implementation of measured properties data.
Three different modelling depths have been implemented
into ACM, namely, an extended equilibrium stage model
(including the actual rates of the chemical reactions in the
bulk phase), a rate-based model with effective diffusivities
and a rate-based model with the Maxwell–Stefan equations.
The results presented in this paper have been obtained with
the last model.
Fig. 3. Sketch of an extended two-film model.
794 M. Klöker et al. / Chemical Engineering and Processing 43 (2004) 791–801
4. Chemical system
Ethyl acetate is a commodity chemical substance pro-
duced and used worldwide. It is primarily used as a sol-
vent for paints and coatings (80 ktons/a), printing inks
(65 ktons/a) and adhesives (36 ktons/a), as a process solvent
(38 ktons/a), a chemical feedstock (15 ktons/a), and for
cosmetics and food (12 ktons/a) (all numbers for Western
Europe 1997) [18]. Mainly two production processes for
ethyl acetate are currently used [18,19]:
– Esterification of ethanol and acetic acid in the presence
of an acid catalyst, usually sulphuric acid (the main pro-
duction method in Western Europe).
– Tishchenko reaction of acetaldehyde in the presence of an
aluminium ethoxide catalyst (the main production method
in Japan).
Recently, for synthesis of ethyl acetate heterogeneously
catalysed reactive distillation has been developed. Kolena
et al. [20] and Wu and Lin [21] both suggested the com-
bination of a pre-reactor and a reactive distillation column
to carry out the reaction. The only difference between the
processes is the location of the pre-reacted feed to the reac-
tive distillation column. The process of Kolena et al. [20] is
nowadays commercialised by Sulzer Chemtech Ltd [22].
The boiling points of the pure components at atmo-
spheric pressure are ranged as follows: ethyl acetate
(ETAC) 77.2 ◦ C; ethanol (ETOH) 78.3 ◦ C; water (H2 O)
100.0 ◦ C; acetic acid (HAC) 118.0 ◦ C. There are three bi-
nary azeotropes and one ternary azeotrope shown in Table 2,
with respective boiling points at atmospheric pressure. The
normal boiling points for the pure components as well as
the compositions of the azeotropes have been obtained from
ASPEN Properties PlusTM and show satisfactory agreement
with the data available elsewhere [23]. Whereas the binary
azeotropes are matched very well, the composition of the
ternary azeotrope differs slightly.
For a feasibility study, reactive distillation lines have been
analysed by Kenig et al. [23]. Due to the presence of min-
imum boiling azeotropes, it is not possible to obtain high
purity ethyl acetate at the top of the column, but rather
a ternary mixture with water as a by-product and some
non-converted ethanol. A liquid–liquid phase separation at
the top of the column allows a further enrichment of ethyl
acetate. The corresponding ternary phase diagram with cal-
culated liquid–liquid equilibrium at 25 ◦ C is shown in Fig. 4.
Table 2
Calculated azeotropic data of the system
Azeotrope ETAC ETOH H2 O
(wt.%) (wt.%) (wt.%)
ETOH–H2 O – 95.5 4.5
ETAC–ETOH 70.2 29.8 – 71.8
ETAC–H2 O 91.3 – 8.7
ETAC–ETOH–H2 O 82.0 9.8 8.2
Fig. 4. Ternary phase diagram: LLE for ethanol/ethyl acetate/water
(mass-fraction basis, calculated at 25 ◦ C).
Phase separation is only possible at low ethanol concen-
trations. A higher ethanol content prevents phase splitting.
Thus, it is important for the process that a sufficient ethanol
conversion is provided. If necessary, this can be achieved by
an excess of acetic acid in the column feed.
However, after the decanter (liquid–liquid phase separa-
tor) further separation steps are necessary in order to obtain
highly pure ethyl acetate. The solubility of water in ethyl
acetate at 25 ◦ C is about 3.0 wt.%; the solubility of ethyl
acetate in water at 25 ◦ C is 9.9 wt.% [24].
Within the developed model, the vapour phase is described
by the Hayden O’Connell equation of state accounting for
the non-idealities of this phase due to the dimerisation of
acetic acid. The UNIQUAC method is used for the calcula-
tion of the activity coefficients.
5. Reactive distillation experiments
Experiments were carried out in two different columns.
The first one is a laboratory scale column with 50 mm diam-
eter and a total packing height of 3 m, whereas the second
one in a semi-industrial scale column with 162 mm diame-
ter and a packing height of about 12 m. Both columns are
operated at atmospheric pressure and are equipped with a
liquid–liquid separator for the distillate stream.
Both column set-ups have a separation section below and
above the catalytic section. The bottom section is used to
separate acetic acid from the other components and the top
T section is mainly used to keep acetic acid in the reaction
(◦ C) zone.
78.1 The laboratory scale experiments were performed in a
50 mm inner diameter glass column, with a silver lined
70.9 vacuum jacket. The rectifying section consists of 1.0 m
70.4
Sulzer DX packing, with a re-distributor in the middle.
 M. Klöker et al. / Chemical Engineering and Processing 43 (2004) 791–801
Fig. 5. Column set-up of the laboratory column (50 mm diameter).
The catalytic section is equipped with 1.0 m packing, ei-
ther KATAPAK® -S or MULTIPAK® -1 or -2, filled with a
commercial cation exchange resin. The stripping section
consists of 1.0 m Sulzer BX packing. The basic set-up is
shown in Fig. 5.
The temperatures in the column were measured using
six glass jacketed Pt100 thermocouples. Besides, tempera-
tures are measured in the condenser, reboiler, flow path from
the condenser, and reflux flow path. The product from the
condenser is introduced into the decanter (a separation fun-
nel). In case of liquid–liquid phase separation, the aque-
ous phase is withdrawn completely. A part of the organic
phase is fed back to the column as reflux. The acetic acid
feed (in some experiments enriched with reaction products)
is located between the rectification and catalytic sections.
The ethanol feed is located between the catalytic and strip-
ping sections. The reactants are fed to the column at room
temperature. As part of the experimental programme, differ-
ent catalytic column internals and different operating condi-
tions are studied. The total feed rate, the molar feed ratio as
well as the distillate-to-feed and the reflux ratio have been
varied.
For an optimal column operation it is desirable to with-
draw only non-converted excess acid at the bottom in order
to avoid an additional purification step. The stream can be
recycled to the acetic acid feed. Therefore, in most experi-
ments, an excess of acetic acid was fed to the column.
The compositions were determined using a HP 6890 gas
chromatograph. The GC column was CP wax 52CB, ID
795
Fig. 6. Column set-up of the pilot scale column (162 mm diameter).
0.25 mm, length 25 m. Methyl-iso-butylketone was used as
internal standard, and acetone was added to prevent phase
separation. The temperature program of the GC was: 4 min at
60 ◦ C, then a constant increase of 20◦ /min until 150 ◦ C was
reached. The water content of the samples was determined
by Karl Fischer titration.
The 162 mm diameter semi-industrial column set-up,
shown in Fig. 6, is similar to that of the laboratory column.
The rectifying section consists of 4.3 m and the stripping
section of 3.1 m MELLAPAK® 500.Y. The catalytic sec-
tion is equipped with 5.3 m KATAPAK® -S 250.Y, filled
with the same commercial cation exchange resin. As in
the case of the laboratory column, acetic acid is fed above
and ethanol below the reaction zone. Both liquid feeds
are preheated to temperatures slightly below their boiling
points. The semi-industrial column has about seven sample
locations along the column height where either liquid or
vapour samples can be withdrawn. Again, different oper-
ating conditions have been investigated. The entire pilot
plant installation at Sulzer Chemtech Ltd. is illustrated in
Fig. 7.
The characteristics of the both column configurations are
summarised in Table 3.
In addition to the laboratory and pilot scale experiments,
the relevant hydrodynamic and mass transfer correlations
796 M. Klöker et al. / Chemical Engineering and Processing 43 (2004) 791–801

The relevant kinetic parameters are given in [27].

The equilibrium conversion and thus the maximum con-
version in conventional reactors predicted with these kinet-
ics is about 64% at 50 ◦ C, and it decreases to 54% at 100 ◦ C
if the reactants are fed in equimolar composition.

6. Experimental and theoretical results

6.1. Experimental results

A series of experiments has been performed at both col-

umn scales. The presented results focus on the laboratory
scale column set-up.
Selected experimental results for the laboratory scale
column are summarised in Table 4 together with the cor-
responding operating conditions. The conversion in the
column is limited by the catalyst amount installed, and
thus by the height of the catalytic section and the catalyst
volume fraction of the applied internals. Experiments with
KATAPAK® -S show conversions of up to 78% for ethanol
at feed ratios close to stoichiometric ratio. As expected,
conversion increases with increasing distillate-to-feed ratio.
This is illustrated by experiments 3, 7 and 10, in which
both molar feed ratio and total feed rate are compara-
Fig. 7. Pilot plant installation at Sulzer Chemtech Ltd. (162 mm diameter ble, while the reflux ratio is slightly higher in experiment
column). 3. However, for the selected operating conditions with
KATAPAK® -S, no liquid–liquid phase separation is ob-
for the applied catalytic packings have been determined served, because in these cases the ethanol content in the
within the INTINT project (see, e.g. [25]) or taken from distillate prevents phase splitting as illustrated in Fig. 4. The
open literature [10,26]. The kinetics of the esterification thermodynamic description of the investigated system pre-
reaction, determined in the same project, are adjusted to dicts the occurrence of miscibility gaps only at low ethanol
a pseudo-homogeneous approach showing an exponent 1.5 concentrations [24].
dependence on the acetic acid concentration [27]: The results of the experiments with MULTIPAK® -1 are
similar to those with KATAPAK® -S. For some experiments,
1,5
ri = vi (kfwd · xETOH · xHAC − kbwd · xH2 O · xETAC ) two liquid phases in the decanter have been observed,
whereas in experiment 15, a strong excess of acetic acid is
Table 3 fed resulting in an increased conversion of ethanol. More-
Pilot plant characteristics over, in this experiment about 2.9 wt.% water is already
Laboratory scale Pilot scale column present in the ethanol feed being advantageous for the phase
column splitting, while for all other laboratory scale experiments
Column diameter 50 mm 162 mm given in Table 4 ethanol with a purity of about 99.9 wt.%
Rectifying section 1.0 m (Sulzer DX) 4.3 m (MELLAPAK® is used. The product quality is increased by the additional
500.Y) separation step in the decanter.
Reactive section 1.0 m (KATAPAK® -S 5.3 m (KATAPAK® -S Experiments with MULTIPAK® -2 showed a significant
Lab) or 1.0 m 250.Y)
(MULTIPAK® -1) or increase in conversion and thus product purity, which is
1.0 m caused by the higher catalyst fraction as compared with that
(MULTIPAK® -2) of other internals under study at laboratory scale. In most
Stripping section 1.0 m (Sulzer BX) 3.1 m (MELLAPAK® of the experiments, ethanol conversion proved to be high
500.Y)
enough for the subsequent liquid–liquid phase separation.
Operating pressure Atmospheric Atmospheric
Insulation Silver lined vacuum Heated jacket Even for nearly equimolar feed composition, very high con-
jacket versions are obtained (up to 89%).
Condenser type Total condenser Total condenser Although the laboratory set-up is only equipped with 1 m
Feed rate (total) 0.5–1.5 kg/h 30–50 kg/h of catalytic packing, experiments for all three internals show
Sample points 1 (vapour) 4 (vapour), 3 (liquid)
that both conversion and product purity can be increased as
Decanter operated at 20 ◦ C 40 ◦ C
compared with conventional chemical reactors.
 M. Klöker et al. / Chemical Engineering and Processing 43 (2004) 791–801 797

Table 4
Selected experimental results for 50 mm diameter column at different operating conditions and with different internals
Experiment Applied catalytic Total feed Molar feed ratio reflux D/F-ratio LL-phase Ethanol Ethyl acetate
number internal rate (g/h) HAC/ETOH (−) ratio (kg/kg) separation conversion (org. distillate)
(kg/kg) (mol/mol) (wt.%)
1 KATAPAK-S 713 2.01 6.65 0.50 No 0.674 68.3
3 KATAPAK-S 693 1.02 6.24 0.57 No 0.362 49.1
7 KATAPAK-S 832 1.03 4.16 0.76 No 0.639 65.9
10 KATAPAK-S 688 1.02 4.41 0.88 No 0.777 72.2
12 MULTIPAK-1 679 1.05 5.08 0.76 No 0.650 67.4
15 MULTIPAK-1 663 2.03 4.18 0.58 Yes 0.860 87.2
27 MULTIPAK-2 541 1.42 3.55 0.75 Yes 0.917 93.1
28 MULTIPAK-2 855 1.01 3.25 0.93 Yes 0.885 87.4

6.2. Model validation experiments no phase splitting is observed experimentally,

To ensure the predictability of the developed model, both
laboratory and pilot scale experiments are used for the model
validation. The agreement between simulations and experi-
ments is in general very good and deviations regarding con-
version and product purity are typically smaller than 5%.
In Fig. 8, the calculated temperature and concentration
profiles are compared with the corresponding experimen-
tal data, for the operating conditions in experiment 10 (see
Table 4). In this experiment, KATAPAK® -S is applied. Con-
centration profiles show a very high concentration of acetic
acid in the reactive section in the middle, which is favourable
for the processes, since the reaction rate shows an exponent
1.5 dependence of the acid concentration. The conversion
is about 78% for ethanol for a nearly equimolar feed. The
agreement between experimental and theoretical results is
very good, not only for both distillate and bottoms concen-
trations but also for the temperature profile along the col-
umn. At the bottom of the column, highly pure acetic acid is
obtained, which can be recycled to the feed, while the ethyl
acetate fraction in the distillate is about 72 wt.%. In this
nor is it predicted by the simulations.
Model validation for the pilot scale is performed as well.
In the first experiment S1, again no liquid–liquid phase sep-
aration is observed, since in this experiment the operating
conditions are not optimal. The total feed rate in this exper-
iment is 37 kg/h, the molar feed ratio acetic acid to ethanol
is 1.38, the reflux ratio is 4.1 and the distillate-to-feed ratio
is 0.64. The calculated concentration profiles and the corre-
sponding experimental data are presented in Fig. 9. The con-
centration profiles for this example differ significantly from
those in Fig. 8 for a laboratory scale experiment. While the
acetic acid concentration in the catalytic section of the col-
umn is rather low (around 20 wt.%), the ethyl acetate con-
tent is very high. Ethanol accumulates at the bottom end of
the catalytic section around the correspondent feed location.
Moreover, the intermediate boiler water is withdrawn not
only at the top, but also at the bottom of the column. Since it
is desirable to withdraw nearly pure acetic acid at the bottom
and to have high concentrations of acetic acid and low con-
centrations of ethyl acetate in the catalytic section, it can be
summarised, that the operating conditions in this experiment

Fig. 8. Liquid phase concentrations and vapour phase temperature in laboratory column: lines with empty symbols, calculations; solid symbols, experiment.
798 M. Klöker et al. / Chemical Engineering and Processing 43 (2004) 791–801

Fig. 9. Liquid phase concentrations in pilot scale column: lines with empty symbols, calculations; solid symbols, experiment.

are suboptimal. However, the conversion of ethanol is about
77% in the experiment, while the simulated value is slightly
lower. Again, the agreement between simulation and exper-
iment is very satisfactory as illustrated in Fig. 9.
For this case, an increase of the distillate-to-feed ratio
proved to be successful in theory and practise for increasing
both conversion and product purity as well as the purity of
the bottom product. The concentration profiles become very
similar to those at laboratory scale (cf. Fig. 8).
6.3. Influence of the operating conditions
The influence of the predominant operating conditions has
been studied in detail with the validated process simulator.
Fig. 10 shows the influence of the heat duty on the
conversion and distillate-to-feed ratio. Simulations have
been performed with the laboratory scale set-up and
MULTIPAK® -1 as the applied catalytic packing. The molar
feed ratio of acetic acid and ethanol is 1.03, whereas the
reflux ratio is 5.3 and the total feed rate is 0.7 kg/h. The
process behaviour is highly non-linear, and the investigated
heat duty region can be subdivided into three parts. In the
first regime (I), ethanol is mainly accumulated in the cataly-
tic section of the column, and, as expected, both distillate-
to-feed ratio and conversion increase with increasing
reboiler heat duty. In the second regime (II), water, the
intermediate boiler, significantly accumulates in the cat-
alytic section of the column which results in a steep fall
down of both the conversion and distillate-to-feed ratio,
because less ethyl acetate (low boiling component) is
formed. In the third regime (III), acetic acid accumulates
in the catalytic section, and conversion increases more

Fig. 10. Influence of the heat duty on conversion and distillate-to-feed-ratio (laboratory scale set-up, total feed rate: 0.7 kg/h, molar ratio HAC/ETOH
1.03, reflux ratio 5.3).
 M. Klöker et al. / Chemical Engineering and Processing 43 (2004) 791–801
Fig. 11. Influence of the selected column internal on conversion (a) and product purity (b) (laboratory scale set-up, total feed rate: 1.4 kg/h, molar ratio
HAC/ETOH 1.02, reflux ratio 4.41).
rapidly with the heat duty due to the exponent 1.5 depen-
dence of the acetic acid concentration in the kinetic expres-
sion. The last regime is undoubtedly the most suitable for
the column at high distillate-to-feed ratios.
This strong non-linear dependence of the conversion on
the heat duty is very similar for all internals and for both col-
umn scales. Experiments have been performed in all of the
three regimes. For example, the lab-scale experiment 3 (cf.
Table 4) has been performed in region II, while experiment
10 (cf. Fig. 8) is in region III. The pilot scale experiment S1
(cf. Fig. 9) corresponds to region I in which ethanol accu-
mulates in and below the catalytic section of the column.
6.4. Influence of the column configuration
Further, the influence of the applied catalytic internals for
the laboratory scale set-up is investigated.
In Fig. 11, the product purity and the conversion are
depicted versus the heat duty varied over a wide range.
799
Here, the molar feed ratio of acetic acid and ethanol is
1.02, the reflux ratio is 4.41, and the total feed rate is 1.4
kg/h. In this example, no liquid–liquid phase separation oc-
curs. Qualitative process behaviour is similar to that rep-
resented by Fig. 10, though the total feed rate is doubled.
Differences in maximum conversion and maximum prod-
uct purity are attributed to different catalyst contents (sig-
nificantly higher in MULTIPAK® -2, cf. Table 1). In some
regions, however, these differences decrease or die away.
It can be stated, that the behaviour of KATAPAK® -S and
MULTIPAK® -1 is similar. Simulations have been carried out
with the actually implemented catalyst mass which is 15%
higher for MULTIPAK® -1, while the volume fraction of the
catalyst bags is almost identical (cf. Table 1). Generally, at
lower feed rates, the influence of the catalyst content is less
significant.
For the pilot scale column set-up, the influence of the
height of the catalytic section on conversion and product
purity is studied in Fig. 12 at constant operating conditions.
800 M. Klöker et al. / Chemical Engineering and Processing 43 (2004) 791–801
Fig. 12. Influence of the height of the catalytic section (KATAPAK® -S 250.Y) in pilot scale set-up on conversion (X) and product purity (w) (total feed
rate: 30 kg/h, molar ratio HAC/ETOH 1.2, reflux ratio 3, distillate-to-feed ratio 0.9).
The total feed rate is 30 kg/h, the molar feed ratio
acetic acid to ethanol is 1.2, the reflux ratio is 3 and the
distillate-to-feed ratio is 0.9. For these operating condi-
tions, liquid–liquid phase splitting is predicted for all in-
vestigated catalytic section heights. It is shown that both
conversion and product purity strongly increase with a
catalytic section height changing between 1 and 3 m. At
section heights above 4 m, conversion and thus product
purity remain almost constant. In this case, the conver-
sion for the selected operation conditions is around 96%.
The organic distillate consists of 92 wt.% ethyl acetate,
about 5 wt.% water, whereas the rest is non-converted
ethanol and acetic acid. It can be concluded, that the col-
umn operates very well at high distillate-to-feed-ratios,
and even at moderate catalytic sections heights of about
4 m. Moreover, an additional increase of conversion is
possible by further adjustment of the operating condi-
tions.
6.5. Scale-up issues
Since the developed model has been successfully vali-
dated for both column scales, reliable predictions of the
scale-up of catalytic distillation columns are possible. To
a certain extend, this is due to the fact, that for labora-
tory and industrial scale, the necessary information on the
applied catalytic internals and the relevant correlations for
the rate-based process simulator are available. For the both
MULTIPAK® types, catalyst fraction, mass transfer and hy-
drodynamic characteristics are similar for laboratory and in-
dustrial scale. With respect to KATAPAK® -S, laboratory and
industrial scale packings differ quite significantly. However,
the properties of the KATAPAK® -S 250.Y are the same for
the investigated 162 mm diameter semi-industrial scale and
for larger industrial sizes. Thus, it is possible to perform
a computer-aided scale-up of reactive distillation columns
equipped both KATAPAK® -S and MULTIPAK® .
7. Conclusions
In this work, the synthesis of ethyl acetate from ethanol
and acetic acid by heterogeneously catalysed reactive distil-
lation is investigated. Since the choice of column internals
and arrangement of catalytic and non-catalytic zones have
significant influence on the overall performance, different
configurations are experimentally and theoretically studied.
Modelling of this process is accomplished with a rate-based
stage model.
Experiments are performed at two different scales, in a 50
mm and in a 162 mm diameter column, with four different
catalytic internals. The simulations are in a very satisfactory
agreement with the results obtained at both column scales.
The models are able to describe the process behaviour in
an entire range of operating conditions and contribute to a
better understanding of the process.
Theoretical investigations show that the performance of
KATAPAK® -S and MULTIPAK® -1 is very similar at labo-
ratory scale, while conversion and product purity are higher
with MULTIPAK® -2 because of the greater catalyst con-
tent. The simulations are further applied for investigations
of the scale-up from the laboratory to the semi-industrial
set-up.
The production of ethyl acetate via reactive distillation
can significantly increase conversion as compared with con-
ventional reactors. Almost total conversion of ethanol can
be obtained with moderate column heights and even without
using a pre-reactor. An excess of acetic acid is necessary to
achieve nearly complete conversion of ethanol, and there-
fore, nearly ethanol-free top product. Furthermore, a high
 M. Klöker et al. / Chemical Engineering and Processing 43 (2004) 791–801 801

acetic acid concentration in the catalytic section of the col- [5] R.J. Carland, Fractionation tray for catalytic distillation, US Patent,
umn increases the reaction rate. No. 5308451, 1994.
[6] J.M. van Baten, J. Ellenberger, R. Krishna, Hydrodynamics of re-
active distillation tray column with catalyst containing envelopes:
experiments vs. CFD simulations, Catal. Today 66 (2001) 233–240.
Acknowledgements [7] U. Kunz, U. Hoffmann, Perparation of catalytic polymer/ceramic
ionexchange packings for reactive distillation columns, in: G. Ponclet
This work has been accomplished in the context of the (Ed.), Preparation of Catalysts VI, Elsevier, 1995, pp. 299–308.
[8] P.J.M. Lebens, F. Kapteijn, S.T. Sie, J.A. Moulijn, Potentials of
European research project INTINT (Intelligent Column internally finned monoliths as a packing for multifunctional reactors,
Internals for Reactive Separations) (Project No. GRD1 Chem. Eng. Sci. 54 (1999) 1359–1365.
CT1999 10596). The financial support of the European [9] K.H. Johnson, A.B. Dallas, Catalytic distillation structure, US Patent
Commission (Contract No. G1RD CT1999 00048), of the No. 5348710, 1994.
German Ministry for School, Science and Research of the [10] A. Górak, A. Hoffmann, Catalytic distillation in structured packings:
methyl acetate synthesis, AIChE J. 47 (2001) 1067–1076.
Land North Rhine Westphalia (decree: 5130041) and of [11] P. Moritz, H. Hasse, Fluid dynamics in reactive distillation packing
the Swiss Federal Office for Education and Science (decree: Katapak-S, Chem. Eng. Sci. 54 (1999) 1367–1374.
99.0724) is highly appreciated. [12] P. Moritz, S. Blagov, H. Hasse, Reactive distillation process design
and scale-up. Proceedings Separations Technology Topical Confer-
ence, AIChe Annual Meeting 2001, 2, 2001, pp. 906–913.
Appendix A. Nomenclature [13] J. Hanika, Q. Smejkal, J. Kolena, 2-Methylpropylacetate synthesis
via catalytic distillation, Catal. Today 66 (2001) 219–223.
[14] R. Taylor, R. Krishna, Multicomponent Mass Transfer, Wiley, New
D/F-ratio distillate to feed ratio (kg/kg) York, 1993.
k reaction rate constant (mol/kg s) [15] E.Y. Kenig, A. Górak, A Film model-based approach for simulation of
NTSM number of theoretical stages per meter (1/m) multicomponent reactive separation, Chem. Eng. Process. 34 (1995)
T temperature (◦ C) 97–103.
ri reaction rate of component i (mol/kg s) [16] J.D. Seader, The rate-based approach for modeling staged separations,
Chem. Eng. Prog. 85 (1989) 41–49.
w mass fraction (kg/kg) [17] J.O. Hirschfelder, C.F. Curtiss, R.B. Bird, Molecular Theory of Gases
x mole fraction (mol/mol) and Liquids, Wiley, New York, 1964.
X conversion [18] SRI, Chemical Economics Handbook, SRI International, 1999.
vi stoichiometric coefficient of component i [19] Wiley-VCH (Ed.), Ullmann’s Encyclopedia of Industrial Chemistry,
Electronic Release, sixth ed., Wiley-VCH, Weinheim, 2001.
[20] J. Kolena, J. Lederer, P. Moravek, J. Hanika, Q. Smejkal, D. Skala,
Subscripts Zpusob vyroby etylacetatu a zarizeni k provadeni tohoto zpusobu
(Process for the Production of Ethyl Acetate and Apparatus for
ETOH ethanol
Performing the Process), Czech Patent No. Cz PV 3635-99, 1999.
ETAC ethyl acetate [21] K.C. Wu, C.-T. Lin, Catalytic Processes for the Preparation of Acetic
HAC acetic acid Esters, US Patent No. 5,998,658, 1999.
H2 O water [22] P. Moritz, Product Information “Ethyl acetate production by reactive
bwd backward reaction distillation”, in: Brochure of Sulzer Chemtech Ltd., 2002.
[23] E.Y. Kenig, H. Bäder, A. Górak, B. Beßling, T. Adrian, H. Schoen-
eff effective
makers, Investigation of ethyl acetate reactive distillation process,
fwd forward reaction Chem. Eng. Sci. 56 (2001) 6185–6193.
sim simulation [24] A. Arce, L. Alsonso, I. Vidal, Liquid–liquid equilibria of the sys-
tems ethyl acetate+ethanol+water, butyl acetate+ethanol+water, and
ethyl acetate+butyl acetate+water, J. Chem. Eng. Jpn. 32 (1999)
440–444.
References [25] A. Kolodziej, M. Jaroszynski, I. Bylica, Diffusional and hydraulic
characteristics of KATAPAK-S, in: Proceedings of the International
[1] A.A. Backhaus, Continuous processes for the manufacture of esters, Conference on Distillation and Absorption, Baden Baden, Germany,
US Patent No. 1400849, 1921. 2002.
[2] V.H. Agreda, P.H. Partin, W.H. Heise, High purity methyl acetate [26] A. Kolodziej, M. Jaroszynski, A. Hoffmann, A. Górak, Determination
via reactive distillation, Chem. Eng. Prog. 86 (1990) 40–46. of catalytic packing characteristics for reactive distillation, Catal.
[3] S. Saito, T. Michishita, S. Maeda, Separation of meta- and para- Today 69 (2001) 75–85.
xylene mixture by distillation accompanied by chemical reactions, [27] G. Hangx, G. Kwant, H. Maessen, A.P. Markusse, M.I. Urseanu, Re-
J. Chem. Eng. Jpn. 4 (1971) 37–43. action kinetics of the esterification of ethanol and acetic acid towards
[4] W. Linang, Z. Wang, Y. Jin, Z. Yu, S. Yang, Performance of a three- ethyl acetate, deliverable 22, intelligent column internals for reactive
phase fluidized bed as a reactive distillation device, Chem. Eng. separations (INTINT), Technical Report to the European Commis-
Technol. 19 (1996) 456–461. sion, http://www.cpi.umist.ac.uk/intint/NonConf Doc.asp, 2001.