Communications

Reaction Kinetics and Reactive acetate are investigated. Equilibrium conversion starting from
Distillation ± On the Transfer of an equimolar mixture of acetic acid and methanol is 72.2 % at
50 C (calculated using activities by the UNIQUAC method,
Kinetic Data from a Batch Reactor giving Ka = 26.3; Kx = 6.9).
to a Trickle-bed Reactor In most of the publications [e.g. 2], it is assumed that the
reaction reaches chemical equilibrium instantaneously. This
By Tim Pöpken, Robert Geisler, Lars Götze, Axel Brehm, assumption is often not justified. The esterification of acetic
Peter Moritz, and Jürgen Gmehling* acid with methanol reaches chemical equilibrium in the order
of hours, even at higher temperatures and high catalyst
Dedicated to Professor Dr. Gerhard Emig on the occasion concentrations (using 20 wt.-% catalyst, the mixture reaches
of his 60th birthday 95 % of equilibrium conversion after about 4 h at 40 C and
The design of reactive distillation processes requires besides after about 1 h at 60 C). When designing reactive distillation
the knowledge of phase equilibrium data reliable kinetic data. processes for the production or hydrolysis of methyl acetate,
This paper presents the results of kinetic measurements of reaction kinetics therefore has to be taken into account.
acetic acid esterification with methanol together with results The ratio of the reaction rate to product removal determines
for the backward reaction, the hydrolysis of methyl acetate, the existence of stationary points, so-called kinetic or reactive
both catalyzed by the macroreticular, strongly acidic ion- azeotropes, which cannot be overcome by reactive distillation
exchange resin Amberlyst 15. Microkinetic parameters are [1]. These are singular points in residue curve maps. The
determined from batch experiments in a thermostated stirred knowledge of residue curve maps is required for the design of
tank reactor. The applicability of the derived kinetics is reactive distillation processes. To calculate residue curve maps
checked against experimental results obtained in a trickle-bed for the reactive system, a reliable model for the reaction rate is
reactor. While the trickle-bed reactor can be modeled needed.
qualitatively using the microkinetic parameters, some macro- Microkinetic parameters can be determined easily by
kinetic phenomena occur (likely selective swelling of the monitoring the concentration over time in a stirred tank
polymeric matrix of the catalyst). To improve the applicability reactor operated batchwise. It has to be ensured, however,
of the kinetic model, swelling phenomena are currently being that the kinetics is also valid under column conditions, i.e. if
investigated. macrokinetic phenomena, such as diffusional resistances or
heat transfer limitations occur. A good choice for an
experimental apparatus to investigate these phenomena is a
1 Introduction trickle-bed reactor, since trickle-bed reactors show similar
hydrodynamics of the liquid phase when using the same
In recent years the combination of chemical reaction and structured packing as in the reactive distillation column.
separation by distillation in one unit has been investigated, Catalytically active structured packings are an interesting
theoretically [1,2] as well as experimentally [3,4]. Integrating choice for column internals for reactive distillation processes.
these processes, commonly known as reactive distillation, They combine good separation efficiency with an excellent
offers advantages due to lower capital investment and lower contact between liquid phase and catalyst. In this work, the
energy costs when compared to the conventional approach [5]. laboratory version of Katapak-S (Katapak-S is a registered
This is achieved by exploiting the synergetic effects resulting trademark of Sulzer Chemtech AG, Switzerland) filled with
from the interaction of chemical reaction and distillation, the macroreticular, strongly acidic ion exchange resin
which allow one to suppress unwanted side reactions or to Amberlyst 15 by Rohm & Haas as catalyst was employed.
achieve higher than chemical equilibrium conversions. Equi-
librium limited reactions, such as etherifications, esterifica-
2 Experimental Apparatus and Procedure
tions, transesterifications, or hydrations, have been identified
to be suitable for this process alternative [6].
2.1 Batch Reactor for Microkinetic Measurements
In the present paper the esterification of acetic acid with
methanol, and the reverse reaction, the hydrolysis of methyl
The measurements were carried out batchwise in a
± thermostated glass reactor of 500 cm3 volume equipped with
[*] Dipl.-Chem. T. Pöpken; Dipl.-Chem. R. Geisler; Dipl.-Chem. L. Götze; stirrer, baffle and reflux condenser to ensure good mixing and
Dr. A. Brehm; Prof. Dr. J. Gmehling, Carl von Ossietzky Universität prevent loss of the volatile compounds. In preliminary
Oldenburg, Technische Chemie, Postfach 25 03, D-26111 Oldenburg;
Dipl.-Ing P. Moritz; Sulzer Chemtech AG, Katalysatortechnik, Postfach experiments, the stirrer speed was varied to find sufficient
65, CH-8404 Winterthur. conditions ensuring the absence of external mass transfer

Chem. Eng. Technol. 21 (1999) 5, Ó WILEY-VCH Verlag GmbH, D-69469 Weinheim, 1998 0930-7516/99/0505-00401
0930-7516/99/0505-00401$$17.50+.50/0
17.50+.50/0 401
 Communications

limitations. Catalyst and one reactant were placed in the

reactor and heated. By adding the preheated second reactant
the experiment was started. Samples of about 1 cm3 volume
were drawn in regular intervals and analyzed by potentio-
metric titration. Several experiments under variation of
temperature (40 up to 60 C) and mass of catalyst (0 up to
56 g, corresponding to 0 up to 28 wt.-%) were carried out. The
forward reaction (esterification) as well as the backward
reaction (ester hydrolysis) were investigated. We used
different sieve fractions of the catalyst to check for internal
mass transfer limitations. Furthermore, the influence of inerts
on the reaction kinetics and the chemical equilibrium was
investigated. These experiments will be presented in detail
later [7]. Figure 1. Experimental (symbols) and calculated (from the second order model,
lines) concentrations of acetic acid for three batch reactor experiments. All
three experiments were carried out under the following conditions: catalyst
mass mcat » 17.5 g (dry weight), initial composition of the liquid phase: xHOAc =
xMeOH » 0.4; xH2O » 0.2.
2.2 Trickle-bed Reactor

The functional internals of the trickle-bed reactor are six equilibrium data, heat of mixing data and activity coefficients
KatapakÒ-S elements with 200 mm height each, of which the at infinite dilution were included in the fitting procedure. By
topmost is filled with glass beads to ensure a good liquid taking the real behavior (gi) into account, the description of
distribution. The other elements contain 60 g (dry weight) the experimental data was improved in all cases. Thus the
AmberlystÒ 15 (Æ > 0,6 mm) each. The catalytic packing has a average relative error was reduced by 1 to 2 % (e. g. from 8.3 %
diameter of 70 mm and a total height of 1 m. The reactor was to 7.3 % in the case of the second order rate law, see below).
fed from the top with liquid loads between 3 and 18 m3×m±2×h±1. Different kinetic models were tested against the experi-
The feed temperature was varied between 40 and 60 C. The mental data, including adsorption-based models. A better
temperature in the reactor bed was measured at four different agreement with the experimental data was achieved with
levels. After steady state was reached which was indicated by a pseudohomogeneous models, which is in good agreement with
constant temperature profile for a sufficient period of time, the theory that the catalyst contains solvated protons in its
samples of feed and product streams were withdrawn and pores, which are responsible for the catalytic activity [8]. A
analyzed using gas chromatography. power-law model
To simulate the trickle-bed reactor, the residence time
1 1 dni
behavior has to be known additionally. The residence time r r1 r 1
mcat vi dt
distribution was measured as function of the liquid load using
1 N hydrochloric acid as a tracer. Furthermore, the liquid k1 aHOAc aMeOH k aMeOAc aH (1)
1 2O
holdup of the catalytic bed was measured for different liquid
loads.
gave the best description (mean relative deviation: 6.7 %) of
the experimental data. The temperature dependence of the
3 Results rate constants ki was described by the Arrhenius-law.

3.1 Microkinetics ki k0i exp EA i RT (2)

Typical results are shown in Fig. 1. It has been observed that The good agreement between experiments and calculations
the reaction rate is proportional to the catalyst mass mcat (see with the power-law model can be explained with the high
Fig. 2). Thus, the reaction rate can be defined based on the number of parameters. A simple second order rate law (all
catalyst mass as is usually done in heterogeneous catalysis. The exponents a to d in Eq. (1) are equal to unity) gives also a good
particle size has a negligible influence on the reaction rate. description of the data (mean relative deviation: 7.3 %) with
Therefore, no significant internal mass transfer limitation only four parameters. The results of the fitting procedure to 48
occurs within the studied temperature interval. experiments for both models are presented in Tab. 1.
All calculations were performed both with and without
taking into account liquid phase activities. Activity coeffi-
cients were determined using the UNIQUAC model. The 3.2 Modeling of the Trickle-bed Reactor
UNIQUAC-interaction parameters were fitted to experimen-
tal data stored in the Dortmund Data Bank (DDB, 1998 The residence time distribution of the trickle-bed reactor
version, DDBST GmbH (Oldenburg, FRG)). Vapor-liquid was described using the tanks-in-series model. Analysis of

402 Ó WILEY-VCH Verlag GmbH, D-69469 Weinheim, 1999 0930-7516/99/0505-00402 $ 17.50+.50/0 Chem. Eng. Technol. 21 (1999) 5
 Communications

Table 1. Results of the fitting procedure for the pseudohomogeneous models. the trickle-bed than on the batch reactor. Currently, the phase
The values were determined by minimizing the objective function
equilibrium between polymer and liquid phase for the
all samples
F xHOAc calc xHOAc exp xHOAc exp 100% nsamples considered system is investigated.
using the Simplex-Nelder-Mead method. The reaction rate is calculated based
on dry catalyst mass.

k01 EA,1 k0±1 EA,-1 a b g d F

model [mol×g±1×s±1] [kJ×mol±1] [mol×g±1×s±1] [kJ×mol±1] [-] [-] [-] [-] [%]
power 3.55×106 61.3 58.2×106 80.6 1.22 1.36 1.20 0.93 6.7
law
second 1.59×106 61.5 1.40×106 70.5 1 1 1 1 7.3
order

experimental data gave numbers of tanks between 50 and 80.

Besides the non-ideal flow, the incomplete wetting of the
catalyst at low liquid loads was incorporated in the reactor
model. This was done by using a correlation with the liquid
holdup [9]. Experimental values for feed composition and
temperature served as model input. A comparison between
calculated and experimental results for product composition Figure 2. Reaction rates of the esterification reaction at different temperatures
and temperature allows conclusions about the applicability of versus dry catalyst mass for a given reaction volume. The dependence is linear
within experimental error. The slopes of the curves show the forward reaction
the kinetic model and the occurrence of macrokinetic rate constant k1 at the corresponding temperature.
phenomena.
The ion-exchange capacity of the catalyst used in the trickle-
bed experiments and fresh catalysts were measured and
3.3 Application of the Microkinetic Model for the Trickle-bed compared. Even after five months of intensive use, no
Reactor deactivation could be detected. The measured capacities of
4.7 meq×g±1 are comparable to the specifications of the catalyst
The comparison between experimental and calculated manufacturer.
(using microkinetic parameters from the batch reactor
experiments) conversion shows that the experimental con-
version exceeds the predicted value by about 50 % for the 4 Discussion
forward reaction and 10 % for the backward reaction. A
possible explanation is the increasing catalyst temperature Reliable kinetic data are necessary for the design of all kinds
due to the reaction enthalpy. However, this is unlikely because of reactors. This is especially true for reactive distillation
of the small reaction enthalpy of 3.4 kJ×mol±1. To clarify this, processes, where reaction and separation interact and
experiments have been carried out in the trickle-bed reactor stationary points in reactive residue curve maps may occur.
where the catalyst in the Katapak-S elements was replaced by For practical and economical reasons, the use of hetero-
a mixture of 50 wt.-% catalyst and glass beads. The reduction geneous catalysts has been shown to be advantageous in
of the catalyst mass had no effect on the observed reaction reactive distillation. With a large number of acid or base
rate, therefore a heat transfer limitation is very unlikely. catalyzed reactions, ion-exchange resins can be used as
A more likely reason for the higher conversion is the catalyst. On one hand, the catalyst needs not to be removed
different catalyst weight fraction in batch and trickle-bed from a product stream, as it would be in the case of
experiments. In the batch reactor a maximum of about 28 wt.- homogeneous catalysis, but on the other hand the complexity
% catalyst can be used, while in the trickle-bed reactor the of the system becomes higher. The composition of the reaction
weight fraction of the catalyst is between 50 and 70 wt.-%, mixture at the reaction locus, the strongly polar polymeric
depending on the liquid load. The extrapolation of the straight resin, may differ significantly from the liquid phase composi-
lines in Fig. 2 to higher catalyst concentrations can lead to tion. In the investigated system, the swelling of the resin is
erroneous results. Furthermore, in acetic acid esterifications qualitatively known to be selective for polar compounds.
with ethanol catalyzed by Amberlyst 15 it was observed that Without quantitative knowledge of the swelling properties
the catalyst acts as a selective adsorbent [10]. The affinity of of the catalyst, the usage of kinetic parameters derived from
the resin is highest for water and lowest for the ester. Since the batch reactor experiments in a trickle-bed reactor may result
reaction takes place in the polymer phase, the activities of the in conversion differences up to 50 % between experiments and
reactants inside the swollen polymer phase have to be used in calculations. Since conditions in the trickle-bed reactor and
the kinetic model. Because of the higher catalyst weight the reaction column are rather similar, an error of a similar
fraction, the swelling effect should have a stronger impact on magnitude can be expected when using kinetic data from a

Chem. Eng. Technol. 21 (1999) 5, Ó WILEY-VCH Verlag GmbH, D-69469 Weinheim, 1998 0930-7516/99/0505-00403 $ 17.50+.50/0 403
 Communications

batch reactor directly for the design of reactive distillation References

processes.
[1] Venimadhavan, G.; Buzad, G.; Doherty, M.F.; Malone, M. F., AIChE J. 40
(1994) 11, pp. 1814±1824.
[2] Bessling, B.; Schembecker, G.; Simmrock, K. H., Ind. Eng. Chem. Res. 36
5 Outlook (1997) pp.3032±3042.
[3] Agreda, V.H.; Partin, L.R.; Heise, W. H., Chem. Eng. Prog. 86 (1990) 2, pp.
40±46.
For a full understanding of reactive distillation processes [4] Krafczyk, J.; Gmehling, J., Chem.-Ing.-Tech. 66 (1994) 10, pp. 1372±1375.
using polymeric catalysts, it is necessary to understand the [5] Gmehling, J.; Brehm, A., Grundoperationen, G. Thieme Verlag, Stuttgart
swelling behavior of these catalysts. This is currently being 1996.
[6] Podrebarac, G. G.; Ng, F. T. T.; Rempel, G. L., Chemtech 27 (1997) 5, pp.
investigated in our research group. Further measurements of 37±45.
the reaction rate in the trickle-bed reactor, including data [7] Götze, L.; Geisler, R.; Pöpken, T.; Brehm, A.; Gmehling, J., in preparation.
[8] Pitochelli, A. R., Ion Exchange Catalysis and Matrix Effects, Rohm &
under boiling conditions, and measurements in a mini plant
Haas Co., Philadelphia 1988.
reactive distillation column will follow. The procedure should [9] Satterfield, C. N., AIChE J. 21 (1975) 2, pp. 209±228.
also be applied to other reversible reactions (esterifications, [10] Mazzotti, M.; Neri, B.; Gelosa, D.; Krugov, A.; Morbidelli, M., Ind. Eng.
Chem. Res. 36 (1997) pp. 3±10.
hydrolyses, ... ). Simultaneously, the development of software
tools for the simulation and design of kinetically controlled This paper was also published in German in Chem. Ing. Tech. 71 (1999) No.1+2.
reactive distillation processes is planned. _______________________

Received: May 20, 1998 [K 2428]

Microwave High Pressure Thermo-
Symbols used
chemical Conversion of Sewage Sludge
as an Alternative to Incineration
ai [±] activity of component i
EA,i [J×mol±1] activation energy of reaction i By Janine T. Bohlmann, Christian M. Lorth, Anja Drews, and
ki [mol×g±1×s±1] reaction rate constant Rainer Buchholz
k0i [mol×g±1×s±1] frequency factor
Ka [±] thermodynamic chemical The disposal of moist organic wastes poses a severe problem.
equilibrium constant Various thermal treatment methods have been established.
Kx [±] chemical equilibrium constant, These, however, require the energy-consuming preliminary
calculated from xi drying of organic wastes, which typically contain 70±80 %
mcat [g] catalyst mass moisture. High-pressure thermochemical conversion pro-
ni [mol] number of moles of component i cesses follow a different route with conversion taking place
r [mol×g±1×s±1] reaction rate in the presence of liquid water. Since moist substances can be
R 8,31441 J×mol±1×K±1 universal gas constant easily heated by microwave power, this study examines if
t [s] time reclamation of raw materials can be accomplished more
T [K] absolute temperature economically in a novel microwave high-pressure (MHP)
wL [m3×m±2×h±1] liquid load reactor than in a conventional autoclave. Thermochemical
xi [±] mole fraction of component i conversion at high pressure yields a product mixture
composed of oil, solid residue, reaction gas and an aqueous
phase. Maximum oil yields of 30.7 % were achieved. On
Greek symbols average, the higher heating value of produced oils was Hh =
36.4 MJ/kg. The specific heat content of the mechanically
a, b, g, d [±] exponents of power-law model dewatered product mixture (50 % dry matter content)
ni [±] stoichiometric coefficient for amounts to e = 5.69 MJ/kg as opposed to that of sewage
component i sludge at 19.9 % dm of e = 1.96 MJ/kg.

±
Subscripts [*] Poster presented at the GVC Annual Meeting in Freiburg, Sept. 29±
Oct. 1, 1998.
[**] Dr. Ing. J. T. Bohlmann, Dipl. Ing. C. M. Lorth, Dipl. Ing. A. Drews,
1 forward reaction (esterification)
Technische Universität Berlin, Institut für Biotechnologie, Fachgebiet
±1 backward reaction (hydrolysis) Bioverfahrenstechnik, Sekr. GG2, Seestr. 13, D-13353 Berlin;
HOAc acetic acid Prof. Dr. rer. nat. R. Buchholz, Technische Universität Berlin, Institut für
MeOH methanol Biotechnologie, Fachgebiet Bioverfahrenstechnik, Sekr. GG2, Seestr. 13,
D-13353 Berlin; Forschungsinstitut für Bioverfahrenstechnik der Ver-
MeOAc methyl acetate suchs- und Lehranstalt für Spiritusfabrikation und Fermentation (VLSF),
H2O water Seestr. 13, D-13353 Berlin, Germany

404 Ó WILEY-VCH Verlag GmbH, D-69469 Weinheim, 1999 0930-7516/99/0505-00404 $ 17.50+.50/0

0930-7516/99/0505-00404 $ 17.50+.50/0 Chem. Eng. Technol. 21 (1999) 5