You are on page 1of 69

9

Reactive Separations in Fluid Systems

E.Y. Kenig and A. Górak


University of Dortmund, Dortmund, Germany

H.-J. Bart
University of Kaiserslautern, Kaiserslautern, Germany

1. INTRODUCTION: AN OVERVIEW OF
REACTIVE SEPARATIONS
Chemical manufacturing companies produce materials based on chemical reactions
between selected feed stocks. In many cases the completion of the chemical reac-
tions is limited by the equilibrium between feed and product. The process must then
include the separation of this equilibrium mixture and recycling of the reactants.
The fundamental process steps of bringing material together, causing them to react,
and then separating products from reactants are common to many processes.
Conventionally, each unit operation—whether mixing or absorption, distil-
lation, evaporation, crystallization, in fact, any of the heat-, mass-, and momentum-
transfer operations so familiar to chemical engineers—is typically performed in
individual items of equipment, which, when arranged together in sequence, make
up the complete process plant. As reaction and separation stages are carried out
in discrete equipment units, equipment and energy costs are added up from these
major steps. However, this historical view of plant design is now being challenged
by the combination of two or more unit operations into one plant unit. The poten-
tial for capital cost savings is obvious, but there are often many other process
advantages that accrue from such combinations (1).

Copyright © 2004 by Marcel Dekker, Inc. All Rights Reserved.


In recent decades, a combination of separation and reaction inside a single
unit has become more and more popular. This combination has been recognized
by chemical process industries as having favorable economics for carrying out
reaction simultaneously with separation for certain classes of reacting systems,
and many new processes (called reactive separations) have been invented based
on this technology (2–9). Reactive separation units may also be treated as a kind
of multifunctional reactor in which the functionalities of several processes are
combined to generate the new reactor concept (Figure 1).
The most important examples of reactive separation processes (RSPs) are
reactive distillation (RD), reactive absorption (RA), and reactive extraction (RE). In
RD, reaction and distillation take place within the same zone of a distillation column.
Reactants are converted to products, with simultaneous separation of the products
and recycling of unused reactants. The RD process can be efficient in both size and
cost of capital equipment and in energy used to achieve a complete conversion of
reactants. Since reactor costs are often less than 10% of the capital investment, the
combination of a relatively cheap reactor with a distillation column offers great
potential for overall savings. Among suitable RD processes are etherifications, nitra-
tions, esterifications, transesterifications, condensations, and alcylations (2).

FIGURE 1 Reactive separation units as multifunctional reactors. (Inspired by


Ref. 4.)

Copyright © 2004 by Marcel Dekker, Inc. All Rights Reserved.


Similarly, in RA, reactions occur simultaneously with the component trans-
port and absorptive separation, in the same column zone. These processes are
used predominantly for the production of basic chemicals, e.g., sulphuric or nitric
acids, and for the removal of components from gas and liquid streams. This can be
either the cleanup of process gas streams or the removal of toxic or harmful sub-
stances in flue gases. Absorbers or scrubbers where RA is performed are often
considered gas–liquid reactors (10). If more attention is paid to the mass trans-
port, these apparatuses are instead treated as absorption units.
Reactive extraction uses liquid ion exchangers that promote a selective
reaction or separation. The solutes are very often ionic species (metal ions or
organic/inorganic acids) or intermediates (furfural phenols, etc.), and the extrac-
tion chemistry is discussed elsewhere (11–13). Reactive extraction can be used for
separation/ purification or enrichment or conversion of salts (14). A 2001 review
on reactive phase equilibria, kinetics, and mass transfer and apparative techniques
is given in Ref. 8. Reactive extraction equipment is discussed in detail in Ref. 15,
and recent advances are given in Ref. 16.
Reactive absorption, distillation, and extraction have much in common.
First of all, they involve at least one liquid phase, and therefore the properties of
the liquid state become significant. Second, they occur in moving systems; thus
the process hydrodynamics plays an important part. Third, these processes are
based on the contact of at least two phases, and therefore the interfacial transport
phenomena have to be considered. Further common features are multicomponent
interactions of mixture components, a tricky interplay of mass transport and
chemical reactions, and complex process chemistry and thermodynamics.
On the other hand, RD, RA, and RE have a number of specific features that
should be considered with care and described by different approaches. Before
going into detail, it is worthy to note that the operating window of reactive separ-
ations may be somewhat limited, since these operations are feasible only if they
allow for both separation and reaction within the same range of temperature and
pressure and, on the other hand, for the safe operation from the constructional
point of view (Figure 2).

1.1. Reactive Absorption


The main purposes of absorption processes are the removal of one or more com-
ponents from the gas phase, production of particular substances in the liquid
phase, and gas mixture separation (3). Industrial absorption operations are usually
realized by combining absorption and desorption units.
The example given in Figure 3 illustrates this combination of two processes.
In an absorber, one or several gas components are absorbed by a lean solvent,
either physically or chemically. A rich solvent, after preheating in heat exchan-
gers H1 and H3, is transported to the top of a desorption unit, which usually operates

Copyright © 2004 by Marcel Dekker, Inc. All Rights Reserved.


FIGURE 2 Feasibility of reactive separation, depending on mechanical
design, chemical reaction, and separation performance.

under a pressure lower than that in the absorber. Part of the gas absorbed by the
rich solvent is desorbed due to flashing and heating. The other part has to be de-
sorbed in the stripper via countercurrent contact of liquid with the inert gas or
steam. The lean solvent then flows through heat exchanger H1 to recover heat nec-
essary for heating the reach solvent, passes through heat exchanger H2 to cool
down to a desired temperature, and finally enters the absorber (3).
Usually a small amount of fresh solvent should be added to the column in
order to equalize the solvent loss due to evaporation in the desorber or to irre-
versible chemical reactions occurring in the whole system (3).
Reactive absorption represents a process in which a selective solution of
gaseous species by a liquid solvent phase is combined with chemical reactions.
As compared to purely physical absorption, RA does not necessarily require ele-
vated pressure and high solubility of absorbed components; because of the chem-
ical reaction, the equilibrium state can be shifted favorably, resulting in enhanced
solution capacity (17). Most RA processes involve reactions in the liquid phase
only; in some of them, both liquid and gas reactions occur (18,19).
Usually the effect of chemical reactions in RA processes is advantageous
only in the region of low gas-phase concentrations, due to limitations stemming
from the reaction stoichiometry or equilibrium (20). Further difficulties of RA
applications may be caused by the reaction heat through exothermic reactions and
by relatively difficult solvent regeneration (21,22). Most RA processes are

Copyright © 2004 by Marcel Dekker, Inc. All Rights Reserved.


steady-state operations, either homogeneously catalyzed or auto-catalyzed. Some
important industrial applications of RA are given in Table 1.
Reactive absorption can be realized in a variety of equipment types, e.g., in
film absorbers, plate columns, packed units, or bubble columns. This process is char-
acterized by independent flow of both phases, which is different from distillation and
permits both cocurrent (downflow and upflow) and countercurrent regimes.
Reactive absorption is essentially an old process, known since the founda-
tion of modern industry. This is a very important process, too, being the basic
operation in many technological chains. More recently, the role of RA as a key
environmental protection process has grown up significantly.
Despite the clear importance of RA, its behavior is still not properly under-
stood. This can be attributed to a very complex combination of process thermody-
namics and kinetics, with intricate reaction schemes including ionic species, reaction
rates varying over a wide range, and complex mass transfer and reaction coupling.
As compared to distillation, RA is a fully rate-controlled process, and it definitely
occurs far from the equilibrium state. Therefore, practitioners and theoreticians are
highly interested in establishing a proper rate-based description of this process.

1.2. Reactive Distillation


Reactive distillation is a combination of chemical reactions and distillation
(Figure 4b). This operation provides promising process alternatives to traditional

FIGURE 3 Scheme of an absorber–desorber link. (Adapted from Ref. 3.)

Copyright © 2004 by Marcel Dekker, Inc. All Rights Reserved.


TABLE 1 Applications of RA Processes
Aim of the process Example Application area Refs.

Removal of harmful Coke oven Gas purification 23–25


substances gas purification,
amine washing
Retrieval/regeneration Solvent Gas separation 26
of valuable substances regeneration
or nonreacted reactants
Production/preparation Manufacture of Chemical 27–29
of particular products sulphuric acid, synthesis
formaldehyde
preparation,
manufacture of
soda ash
Manufacture of Fertilizer 30
nitrogenous industry
fertilizers
Water removal Water removal Gas drying 31,32
from natural
gas, air drying
Conditioning of Synthesis gas Gas separation/ 26
gas streams conditioning gas purification

sequential operations, shown in Figure 4a. Among potential advantages of


RD are:
New, less expensive products
Higher efficiency because of overcoming thermodynamic and kinetic
limitations
Better selectivity due to suppressing of undesired reactions
Higher raw material conversion
Avoiding of hot spots
Savings due to smaller equipment
Less environmental pollution
One can distinguish between homogeneously and heterogeneously cata-
lyzed RD; the latter is often called catalytic distillation (CD).
The applicability of the RD process is highly dependent on the properties
of the chemical system at hand. A classical example for which RD is recom-
mended may be the reaction in which the products are generated by a reversible
reaction, e.g., in the production of methyl acetate. This system is very complex
because of the occurrence of several azeotropes between reactants and products.

Copyright © 2004 by Marcel Dekker, Inc. All Rights Reserved.


FIGURE 4 Reactive distillation (b) as alternative to the sequential operation (a).

A usual solution in this case is a sequence of a reactor and several separation units
(Figure 5). Another way—an integrated RD process such as shown in Figure 6—
allows for simultaneous formation of methyl acetate in the reaction zone, extract-
ive distillation and product enrichment in the upper part of the column, and
methanol separation in the stripping zone. The production of esters such as
methyl acetate, ethyl acetate, and butyl acetate has for years been an interesting
RD application.
The most important application of RD today seems to be the production of
ethers such as methyl tertiary butyl ether (MTBE), ethyl tertiary butyl ether
(ETBE), and tertiary amyl methyl ether (TAME), which are widely used as mod-
ern gasoline components. Figure 7, upper part, shows a traditional process for
MTBE production, which is a strongly exothermic reaction. The disadvantages of
that process can be avoided if the reaction and separation take place within the
same zone of the reactor (Figure 7, lower part).
Table 2 gives a short overview of possible RD applications.
The design of RD is currently based on expensive and time-consuming
sequences of laboratory and pilot-plant experiments, since there is no commer-
cially available software adequately describing all relevant features of reactions
(catalyst, kinetics, holdup) and distillation (VLE, thermodynamics, plate and pack-
ing behavior) as well as their combination in RD. There is also a need to improve
catalysts and column internals for RD applications (1,51). Figures 8 and 9 show
some examples of catalytic internals, applied for reactive distillation.

1.3. Reactive Extraction


Liquid–liquid extraction is based on partial miscibility of liquids. In the simplest
extraction system, two compounds have to be separated. This can be done by
extracting with a carefully selected solvent, in which one compound (solute) easily
dissolves whereas the other (nonsolute) does not. The solvent has to be recovered

Copyright © 2004 by Marcel Dekker, Inc. All Rights Reserved.


FIGURE 5 Methyl acetate synthesis: conventional scheme. (From Ref. 33.)

Copyright © 2004 by Marcel Dekker, Inc. All Rights Reserved.


FIGURE 6 Methyl acetate synthesis: reactive distillation scheme. (From Ref. 33.)

FIGURE 7 MTBE synthesis: conventional scheme (above) and reactive distil-


lation scheme (below).

Copyright © 2004 by Marcel Dekker, Inc. All Rights Reserved.


from the extract for recycling. In countercurrent extraction processes, there is a
light phase and a heavy phase, with one phase dispersed in the other. Which phase
has to be dispersed is an important question in the design of the process.
In reactive extraction, the use of liquid ion exchangers is recommended in
order to extract ionic solutes. These exchangers can be applied to manifold extrac-
tion processes in the chemical industry (e.g., extraction of furfural, organic, and inor-
ganic acids), biochemical and pharmaceutical productions (e.g., penicillin, amino
acids), hydrometallurgy (e.g., mining of metals), and all related environmental appli-
cations. These last are especially attractive, since liquid ion exchangers react very
selectively and have an advantageous performance at very low feed concentrations.
For practical purposes, an ion exchanger is usually diluted in a nonaromatic,
high-boiling diluent (boiling point about 500 K) that is immiscible with water.
This prevents solvent losses and toxic problems and gives the organic phase the
required physical properties (high interfacial tension, low viscosity, low density),
since most liquid ion exchangers are highly viscous or even solid. In some cases
a modifier, usually a long-chain alcohol, is added to help in the solubilization of
the solute–ion exchanger complex. At very high solute loadings, a split of the
organic phase in a solvent-rich and a solvent-poor fraction may occur, especially
when using aliphatic diluents. The organic phase in RE is thus not a single substance,
as in physical extraction systems in which such three-phase liquid systems are not
encountered. Re-extraction is usually performed with chemicals, for instance,
with strong mineral acids.
All liquid ion exchangers can be mixed together in order to generate syner-
gistic effects. As a special case, an equimolar mixture of cation and anion
exchangers gives a “mixed” extraction system, which can extract salts or acids. In
this case the re-extraction occurs by shift of either temperature, aqueous ionic
strength, or acidity/basicidity.
Equilibrium and selectivity constitute important aspects of reactive and
nonreactive extraction processes. Another important factor is the reaction kine-
tics, which has to be reasonably fast. Most RE processes are close to equilibrium
in less than five minutes. Many ion exchangers need reaction times of less than
one minute, and thus diffusion of the solute complex in the organic phase is the
rate-determining step.
The cation and anion exchangers are amphiphilic substances that are ad-
sorbed at the interface. The latter is then rigid and independent of the droplet
diameter, since friction forces are shielded. This is similar to physical extraction
systems, in which an analogous behavior is caused by surfactants in the aqueous
feed accumulated at the interface.
The problems concerning reaction equilibrium and kinetics description
based on chemical potentials rather than on concentrations are extensively dis-
cussed in Refs. 54 and 55, using the zinc system. The latter is recommended as a
reactive liquid–liquid reference extraction test system by the European Federation of

Copyright © 2004 by Marcel Dekker, Inc. All Rights Reserved.


TABLE 2 Applications of RD Processes
Reaction type Synthesis Catalysta Refs.

Esterification Methyl acetate from hom. 33


methanol and acetic acid
Methyl acetate from het. 34,35
methanol and acetic acid
Ethyl acetate from no data 36
ethanol and acetic acid
Butyl acetate from hom. 37
butanol and acetic acid
Transesterification Ethyl acetate from hom. 38
ethanol and butyl acetate
Diethyl carbonate from het. 39
ethanol and dimethyl
carbonate
Hydrolysis Acetic acid and het. 40
methanol from methyl
acetate and water
Etherification MTBE from isobutene het. 41,42
and methanol
ETBE from isobutene het. 43
and ethanol
TAME from het. 44
isoamylene and methanol
Alcylation Cumene from het. 45
propylene and benzene
Condensation Diacetone alcohol het. 46
from acetone
Bisphenol-A from no data 47
phenol and acetone
Dismutation Monosilane from het. 48
trichlorsilane
Hydration Mono ethylene glycol hom. 49
from ethylene oxide
and water
Nitration 4-Nitrochlorobenzene hom. 50
from chlorobenzene
and nitric acid
a hom.: homogeneously catalyzed, het.: heterogeneously catalyzed.

Copyright © 2004 by Marcel Dekker, Inc. All Rights Reserved.


FIGURE 8 (a) Schematic of an RD column filled with catalytic internals CD TECH
(1—catalytic balls, 2—feed, 3—distillate, 4—bottom product, 5—sieve tray) and
(b) catalytic structured packing Sulzer Katapak-S. (Part a from Ref. 52.)

Chemical Engineering (EFCE) and is thus well documented (http://www.icheme.uk/


learning/ or http://www.dechema.de/extraktion).
The selection of the right solvent is the key to successful separation by non-
reactive and reactive liquid–liquid extraction. In this respect, different criteria
should be taken into account, e.g.,
Selectivity
Capacity
Recoverability of solvent
Density
Viscosity and melting point
Insolubility of solvent
Interfacial tension
Toxicity and flammability
Corrosivity
Thermal and chemical stability
Availability and costs
Environmental impact.

Copyright © 2004 by Marcel Dekker, Inc. All Rights Reserved.


Some of these criteria are crucial, while others are desirable properties improving
separation and/or making it more economical. Solvent selectivity, recoverability, and
a large density difference in respect to the raffinate are essential. Some of the require-
ments on the solvent can be in conflict, and thus a compromise may be necessary.
Because aromatic diluents are more expensive and more toxic than alipha-
tic ones, the latter are preferably used in industrial practice (see earlier). Aromatic
diluents, with equivalent molecular weights comparable to those of aliphatic
ones, are more polar and thus more water soluble. The degradation of the diluent
is usually negligible in comparison with that of the ion exchanger. The latter one
can be chemically and thermally degrading and also can be poisoned by an irre-
versibly extracted compound.
Reactive extraction is closely related to the droplet phenomena, and thus
most theoretical models are based on droplet consideration. Their experimental
evaluation can be done using either a rising (falling) droplet apparatus (Figure 10a)
for short residence times or a Venturi tube for long contact times (Figure 10b) (56).

FIGURE 9 (a) Catalytic structured packing Montz Multipak and (b) an exam-
ple of reactive trays. (Part b from Ref. 53.)

Copyright © 2004 by Marcel Dekker, Inc. All Rights Reserved.


FIGURE 10 (a) Rising-droplet apparatus and (b) Venturi tube for droplet mass
transfer experiments. (1—feed storage, 2—metering pump, 3—double-flow
valve, 4, 5—pumps, 6—heat exchanger, 7—collecting funnel, 8—stream to
analysis, 9—valve).

Monodispersed droplets can be produced and in the latter case captured by the
counterflowing continuous phase in the conus of the Venturi tube (see Figure 10b).
The RE process proceeds in three major types of equipment: mixer-settler
systems, column extractors, and centrifugal extractors. Countercurrent column
extractors can be further subdivided into nonagitated nonproprietary columns and
agitated proprietary extractors. Agitating the liquid–liquid system breaks up
droplets and increases the interfacial area to improve the mass transfer and
column efficiency. Various forms of energy input are used, e.g., rotation of pro-
pellers, impellers, and discs; pulsation, vibration, and ultrasonic devices; and cen-
trifugal devices.
Some examples of mechanically agitated contactors are the rotating-disk
contactor (RDC), Karr, Oldshue–Rushton, Scheibel, and Kühni columns shown
in Figure 11.
There are three types of nonproprietary nonagitated types of extraction
columns (see Figure 12). The spray columns are the simplest type of extractors,
containing only distributors for the feed (often through perforated pipes). This

Copyright © 2004 by Marcel Dekker, Inc. All Rights Reserved.


FIGURE 11 Agitated extractors. Left to right: RDC, Karr, Oldshue–Rushton,
Scheibel, Kühni extraction columns.

makes them cheap; however, they are limited in use due to significant axial mix-
ing in the column and the fact that the phases are not coalesced and redistributed.
This often results in low efficiencies, which are comparable to one or two
theoretical equilibrium stages. Packed columns are much more efficient since the
packing reduces back-mixing and enhances drop reformation. The packing types
that can be used are the same as those for normal distillation operations (e.g.,
rings, saddles, or slightly modified structured packings of corrugated metal
sheets). Compared to packed beds, structured packings need a reduced cross-
sectional area for liquid flow, resulting in smaller column diameters. Sieve-tray
columns resemble the distillation column design, except that there is no weir. In

FIGURE 12 Nonagitated extractors. Left to right: spray, packed, sieve-tray


(light), sieve-tray (heavy) column.

Copyright © 2004 by Marcel Dekker, Inc. All Rights Reserved.


an extraction process, either phase, the light or the heavy one, can be dispersed.
This means that there are also two sieve-tray designs in respect to downcomer and
top- or bottom-settler/distributor design.
Generally, the selection of a specific RE contactor is complicated due to the
large number of types available and the number of design parameters. The prac-
tical handling and design of a reactive solvent extraction processes can be found
elsewhere (see, e.g., Refs. 12 and 13).

2. FUNDAMENTALS OF PROCESS MODELING


2.1. General
As already mentioned, all three considered RSPs reveal significant similarity, and
hence their modeling methods are based largely on the same framework.
Because of their multicomponent nature, RSPs are affected by a complex
thermodynamic and diffusional coupling, which, in turn, is accompanied by simul-
taneous chemical reactions (57–59). To describe such phenomena adequately,
specially developed mathematical models capable of taking into consideration col-
umn hydrodynamics, mass transfer resistances, and reaction kinetics are required.
Homogeneously catalyzed RD, with a liquid catalyst acting as a mixture
component, and auto-catalyzed RD present essentially a combination of transport
phenomena and reactions taking place in a two-phase system with an interface. In
this respect they are very similar to RA and RE, and, generally, reaction has to be
considered both in the bulk and in the film region. For slow reactions, a reaction
account exclusively in the bulk phase is usually sufficient.
For heterogeneous systems (CD), it is generally necessary to consider add-
itionally the phenomena in the solid catalyst phase. In this case, very detailed
models using intrinsic kinetics and covering mass transport inside the porous cata-
lyst arise (see, e.g., Refs. 60–62). However, it is often assumed that all internal
(inside the porous medium) and external mass transfer resistances can be lumped
together (35,63,64). In this case each catalytically active site is in contact with the
liquid bulk, i.e., the catalyst surface is totally exposed to the liquid bulk phase and
can be completely described by the bulk variables (9,64). This results in the so-
called pseudo-homogeneous models. If the reaction (either homogeneous or het-
erogeneous) is very fast, it does not depend on the reaction kinetics and thus can
be described using the data on chemical equilibrium only.
Modeling of hydrodynamics in gas/vapor/liquid–liquid contactors includes
an appropriate description of axial dispersion, liquid holdup, and pressure drop. The
correlations giving such a description have been published in numerous papers and
are collected in several reviews and textbooks (e.g., Refs. 65 and 66). Nevertheless,
there is still a need for a better description of the hydrodynamics in catalytic col-
umn internals; this is being reflected by research activities in progress (67).

Copyright © 2004 by Marcel Dekker, Inc. All Rights Reserved.


The description of thermodynamics and chemical properties of the RSP is
very process specific, and hence its general detailed discussion would constitute
a separate issue. Therefore, we will give only a brief discussion of these topics in
the context of the following case studies (Section 3). Further related details can
be found in Refs. 68–74.
In order to model large industrial reactive separation units, a proper sub-
division of a column apparatus into smaller elements is usually necessary. These
elements (the so-called stages) are identified with real trays or segments of a
packed column. They can be described using different theoretical concepts, with
a wide range of physicochemical assumptions and accuracy.

2.2. Equilibrium-Stage Model


In recent decades, the modeling and design of RSPs has usually been based on
the equilibrium-stage model. Since 1893, when the first equilibrium-stage model
was published by Sorel (75), numerous publications discussing various aspects of
model development, application, and solution have appeared in the literature (76).
The equilibrium-stage model assumes that each gas/vapor/liquid stream leaving a
tray is in thermodynamic equilibrium with the corresponding liquid stream leav-
ing the same tray. For the packed columns, the idea of the height equivalent to the
theoretical stage (HETS) is used. In case of RSPs, the chemical reaction has to be
additionally taken into account, either via reaction equilibrium equations or via
rate expressions integrated into the mass and energy balances.
In this respect, much depends on the relation between the mass transfer and
reaction rates in a particular RSP. The definition of the Hatta number represent-
ing the reaction rate in reference to that of the mass transfer helps to discriminate
between very fast, fast, average, and slow chemical reactions (68,77).
If a fast reaction system is considered, the RSP can be satisfactory
described assuming a reaction equilibrium. Here, a proper modeling approach is
based on the nonreactive equilibrium-stage model, extended by simultaneously
using the chemical equilibrium relationship.
Such descriptions can be appropriate enough for instantaneous reactions
and those close to them. In contrast, if the chemical reaction is slow, the reaction
rate dominates the whole process, and therefore, a reaction kinetics expression
has to be integrated into the mass and energy balances. This concept has been
used in a number of studies, for RA (e.g., Refs. 78 and 79), RD (e.g., Refs. 80
and 81), and RE (e.g., Refs. 8 and 12) process simulations.
In practice, RSPs rarely operate at thermodynamic equilibrium. Therefore,
some correlation parameters, such as tray efficiencies or HETS values, have been
introduced to adjust the equilibrium-based theoretical description to reality. For
multicomponent mixtures, however, this concept often fails, since diffusion interac-
tions of several components result in unusual phenomena such as osmotic or reverse

Copyright © 2004 by Marcel Dekker, Inc. All Rights Reserved.


diffusion and mass transfer barrier (82,83). These effects cause a strange behavior of
the efficiency factors, which are different for each component, vary along the column
height, and show a strong dependency on the component concentration (57,83,84).
The acceleration of mass transfer due to chemical reactions in the inter-
facial region is often accounted for via the so-called enhancement factors
(27,68,69). They are either obtained by fitting experimental results or derived theo-
retically on the grounds of simplified model assumptions. It is not possible to
derive the enhancement factors properly from binary experiments, and significant
problems arise if reversible, parallel, or consecutive reactions take place.
The equilibrium-stage model seems to be suitable for esterification reaction
in CD processes (see Refs. 35 and 74). However, it cannot be recommended for
all reaction types, especially those with higher reaction rates.

2.3. Rate-Based Approach


A more physically consistent way to describe a column stage is known as the rate-
based approach (57,85,86). This approach implies that actual rates of multi-
component mass and heat transfer and chemical reactions are taken into account
directly.
Considering homogeneous RSPs, mass transfer at the gas/vapor/liquid–
liquid interface can be described using different theoretical concepts (57,59). Most
often the two-film model (87) or the penetration/surface renewal model (27,88) is
used, in which the model parameters are estimated via experimental correlations.
In this respect the two-film model is advantageous since there is a broad spectrum
of correlations available in the literature, for all types of internals and systems.
For the penetration/surface renewal model, such a choice is limited.
In the two-film model (Figure 13), it is assumed that all of the resistance to
mass transfer is concentrated in thin stagnant films adjacent to the phase interface
and that transfer occurs within these films by steady-state molecular diffusion
alone. Outside the films, in the bulk fluid phases, the level of mixing is so high
that there is no composition gradient at all. This means that in the film region,
only one-dimensional diffusion transport normal to the interface takes place.
Multicomponent diffusion in the films is described by the Maxwell–Stefan
equations, which can be derived from the kinetic theory of gases (89). The
Maxwell–Stefan equations connect diffusion fluxes of the components with the
gradients of their chemical potential. With some modification these equations
take a generalized form in which they can be used for the description of real gases
and liquids (57):
n
xi N Lj  x j N Li
di  ∑
j1
c Lt Dij
i  1, . . . , n (1)

Copyright © 2004 by Marcel Dekker, Inc. All Rights Reserved.


FIGURE 13 Two-film model.

where di is the generalized driving force:


xi di
di  i  1, . . . , n (2)
ℜT dz

Similar equations can be also written for the gas/vapor phase.


Thus the gas/vapor/liquid–liquid mass transfer is modeled as a combination
of the two-film model and the Maxwell–Stefan diffusion description. In this stage
model, the equilibrium state exists only at the interface.
The film thickness represents a model parameter that can be estimated
using mass transfer coefficient correlations. These correlations reflect the mass
transport dependence on physical properties and process hydrodynamics and are
available from the literature (see, e.g., Refs. 57, 68 and 90).
The two-film model representation can serve as a basis for more compli-
cated models used to describe heterogeneously catalyzed RSPs or systems con-
taining suspended solids. In these processes a third solid phase is present, and
thus the two-film model is combined with the description of this third phase. This
can be done using different levels of model complexity, from quasi-homogeneous
description up to the four-film presentations that provide a very detailed descrip-
tion of both vapor/gas/liquid–liquid and solid/liquid interfaces (see, e.g., Refs. 62,
68 and 91). A comparative study of the modeling complexity is given in Ref. 64
for fuel ether synthesis of MTBE and TAME by CD.

2.4. Computational Fluid Dynamics


Every separation unit operation is governed by the continuum conservation laws,
and thus, in principle, everything meaningful to know in the continuum for any
process can be determined with computational fluid dynamics (CFD) (92). In
recent years there have been significant academic and industrial efforts to enable

Copyright © 2004 by Marcel Dekker, Inc. All Rights Reserved.


the use of CFD for the design, scale-up, and optimal operation of a variety of
chemical process equipment.
Special attention has been given to the CFD modeling of two-phase flows. The
most frequently encountered computational techniques for calculating multiphase
flows are Euler–Lagrange and Euler–Euler methods. Euler–Lagrange models are
applicable to dispersed flows (93). In these models the flow of the so-called “carrier
phase” is simulated by solving continuum-flow equations. The motion of individual
particles (or group of particles) of the dispersed phase is tracked through the flow
domain using the calculated carrier-phase flow field as input; afterwards, mass,
momentum, and energy transfer between the two phases are computed and applied to
the carrier-phase flow field prediction. This procedure requires several iterations (94).
Euler–Euler models assume interpenetrating continua to derive averaged
continuum equations for both phases. The probability that a phase exists at a cer-
tain position at a certain time is given by a phase indicator function, which, for
steady-state processes, is equivalent to the volume of fraction of the correspond-
ent phase (volume-of-fluid technique). The phase-averaging process introduces
further unknowns into the basic conservation equations; their description requires
empirical and problem-dependent input (94). In principal, Euler–Euler models
are applicable to all multiphase flows. Advantages and disadvantages of both
methods are compared, e.g., in Refs. 95 and 96.
The volume-of-fluid technique can be used for a priori determination of the
morphology and rise characteristics of single bubbles rising in a liquid (97,98).
Considerable progress has been made in CFD modeling of bubbling gas–solid
fluidized beds by adoption of the Eulerian framework for both the dilute (bubbles)
and dense (emulsion) phases (99–102). The use of CFD models for gas–liquid
bubble columns has also aroused significant interest in recent years, and both
Euler–Euler and Euler–Lagrange methods have been employed for the description
of the gas and liquid phases (94–96,103–113).
There are also some attempts available in the literature to model tray hydro-
dynamics using CFD (114–119).
Despite considerable success in some fields of application, the CFD simu-
lations are still not fully mastered, especially where the considered processes
reveal clearly nonhomogeneous, segregated fluid flow patterns. The latter are usu-
ally the basic phenomenon in packed or filmlike units used in reactive and non-
reactive separations.
One of the important issues with RSPs is the development of efficient col-
umn internals. Such internals have to enhance both separation and reaction and
maintain a sound balance between them. This is valid for both homogeneously
and heterogeneously catalyzed processes, being especially important for CD. An
understanding of the complex, multiphase flow on the internals interrelated with the
mass transport and chemical reaction constitutes a very important challenge for the
future. Some first steps in this respect have been done concerning the performance

Copyright © 2004 by Marcel Dekker, Inc. All Rights Reserved.


of SULZER packings KATAPAK-S® and SULZER BX and OPTIFLOW
(67,120–122) as well as trays on which chemical reaction occurs (119).
Recently, a substantial effort has been made to optimize column internals
for reactive separations and to reduce the number of expensive hydrodynamics
experiments via the CFD simulations (67,119,122). Such simulations can be
regarded as virtual experiments carried out in order to predict the performance of
the internals by varying geometrical and structural parameters, thus reducing the
optimization time.
Necessary new hydrodynamic models have to be formulated and tested for
three-dimensional description of two-phase flow through the internals. Since
accurate resolution of the trickle-flow scale is not feasible at the moment, such
flow details have to be simplified and are subject to the subgrid modeling sup-
ported by experimental investigations of small-scale phenomena.
The CFD simulations should be linked with the rate-based process simula-
tor, providing important information on the process hydrodynamics in the form of
correlations for mass transfer coefficients, specific contact area, liquid holdup,
residence time distribution, and pressure drop. An ability to obtain these correlation
via the purely theoretical way rather than by the traditional experimental one should
be considered a significant advantage, because this brings a principal opportunity to
virtually prototyping of new optimized internals for reactive separations.
The local aspects of liquid–liquid two-phase flow in RE has been the focus
of CFD analysis by different research groups (123–126). In principle, all aspects
concerning single-phase flow phenomena (residence time distribution, impeller
discharge flow rate, etc.) can be tackled, even with complex geometries. However,
the two-phase CFD is still a challenge, and the droplet interactions (breakup and
coalescence) and mass transfer are not implemented in commercially available
codes. Thus these issues constitute an open area for further research and devel-
opment (127).

3. CASE STUDIES
3.1. Absorption of NOx
3.1.1. Chemical System
The reactive system considered is a basic one in the production of nitric acid as
well as in some other industrial processes (19). It consists of 10 components,
including air (N2, O2), water (H2O), oxyacids of nitrogen (HNO2, HNO3), and
nitrogen oxides (NO, NO2, N2O, N2O3, N2O4). The components are involved in
simultaneous, parallel, and consecutive reactions occurring in both phases. The
reactions are of high orders and most of them are exothermic.
Reaction kinetics is described by the scheme suggested in Ref. 128 and
modified in Ref. 129. This scheme involves eight reactions and can be regarded

Copyright © 2004 by Marcel Dekker, Inc. All Rights Reserved.


as the most extensive reaction system so far. The gas-phase reactions are gov-
erned by the following equations:

2NO  O 2 → 2 NO 2 H R0 114 kJ/mol (R1)

NO  NO 2 ↔ N 2 O 3 H R0 39.9 kJ/mol (R2)

2NO 2 ↔ N 2 O 4 H R0 57.2 kJ/mol (R3)

3NO 2  H 2 O ↔ 2HNO 3  NO H R0 35.4 kJ/mol (R4)

whereas the corresponding equations for the liquid phase are

2 NO 2  H 2 O → HNO 2  HNO 3 H R0 10.72 kJ/mol (R5)

N 2 O 3  H 2 O → 2HNO 2 H R0 3.99 kJ/mol (R6)

N 2 O 4  H 2 O → HNO 2  HNO 3 H R0 5.03 kJ/mol (R7)

3HNO 2 ↔ HNO 3  H 2 O  2NO H R0 7.17 kJ/mol (R8)

The liquid-phase reactions are valid for nitric acid concentrations below
34 wt %. In the case of higher nitric acid concentrations, Reactions (R5) to (R7)
become reversible. The oxidation of NO (Reaction (R1)) is the slowest reaction in
this system. Therefore, the total gas-phase holdup in absorbers can be determined
using the kinetic data for this reaction (130). The other gas-phase reactions are
instantaneous equilibrium reactions.
3.1.2. Process Setup
Measurements of an industrial NOx absorption process, schematically shown in
Figure 14, were described in Ref. 131. The absorption plant constitutes a sequence
of four units used for the removal of nitrogen oxides from the waste gas of an adipin
acid factory. Each unit is separated by a metal plate into two sections. In fact there
are eight columns joined together as a countercurrent absorption plant. This plant
is operated at atmospheric pressure. Columns 1–7 have a pump around for cooling
of the liquid. The diameter of each column is 2.2 m; the height is 7 m. The packing
height is 3.2 m. The packing consists of 35 mm INTALOX ceramic saddles.

Copyright © 2004 by Marcel Dekker, Inc. All Rights Reserved.


FIGURE 14 Absorption plant consisting of four units (eight columns).

The liquid feeds entering columns 7 and 8 are low-concentration nitric


acids. The liquid product has a HNO3 concentration of about 35 wt %. The gas
feed has a concentration of NOx of about 60,000 vppm. A quarter of NOx is NO;
the rest is NO2.
3.1.3. Results and Discussion
The sensitivity analysis performed in Ref. 129 shows that the suggested model
provides concentration profiles that are qualitatively correct. For the simulation of
the industrial absorption process shown in Figure 14, the following correlations
ensuring the most reliable results are selected:
The rate constant of Reaction (R1) (the slowest and hence the most import-
ant reaction in the system) according to Ref. 132
The liquid-side mass transfer coefficient according to Ref. 133
The gas-side mass transfer coefficient according to Wehmeier (see Ref. 134)
Figures 15 and 16 give an illustration of the model quality. Figure 15 shows
a comparison of the simulated and measured gas-phase concentrations of NO and
NO2 throughout the whole absorption plant, whereas in Figure 16, experimental
and simulated liquid-phase concentrations of HNO3 and HNO2 are demonstrated.

Copyright © 2004 by Marcel Dekker, Inc. All Rights Reserved.


FIGURE 15 Experimental and simulated gas-phase concentrations of NO and
NO2 throughout the absorption plant.

The zigzag form of the simulated concentration profiles results from switching
different sections of each single column (see Ref. 135). Good agreement between
experimental and simulation results can be readily observed, except for the first
two columns. Here the larger deviations between experiments and simulated
results can be attributed to the fact that at high concentration of HNO3 Reactions
(R5) to (R7), assumed to be irreversible reactions, convert to reversible ones; the
data on their rate constants are lacking.

3.2. Coke Gas Purification


3.2.1. Chemical System
Coke oven gas consists mainly of a mixture of carbon monoxide, hydrogen,
methane, and carbon dioxide. It is contaminated with a variety of organic and
inorganic compounds that have to be separated in absorption columns before its
further use as a synthesis gas. The selective absorption of coke plant gas contam-
ination results from a complex system of parallel liquid-phase reactions.
Instantaneous reversible reactions:

NH 3  H 2 O ↔ NH4  OH ( R9)

H 2 S  H 2 O ↔ HS  H 3O ( R10)

Copyright © 2004 by Marcel Dekker, Inc. All Rights Reserved.


HCN  H 2 O ↔ CN   H 3O ( R11)

HCO3  H 2 O ↔ CO 32  H 3O ( R12)

H 3O  OH ↔ 2 H 2 O ( R13)

Finite-rate reversible reactions:

CO 2  OH ↔ HCO3 ( R14)

CO 2  2H 2 O ↔ HCO3  H 3O ( R15)

CO 2  NH 3  H 2 O ↔ H 2 NCOO  H 3O ( R16)

The reactions including CO2 obey first- and second-order kinetics, where-
as the other reversible reactions are based on simple proton transfers and are
therefore regarded as instantaneous by the corresponding mass action law equa-
tions. The formation of bicarbonate ions (HCO3) takes place via two different

FIGURE 16 Experimental and simulated liquid-phase concentrations of HNO3


and HNO2 throughout the absorption plant.

Copyright © 2004 by Marcel Dekker, Inc. All Rights Reserved.


mechanisms. The rate of the direct reaction between carbon dioxide and hydroxyl
ions (the most important step) is taken from Ref. 28.
Usually the reaction between CO2 and water is very slow and hardly con-
tributes to the total rate of reaction of carbon dioxide. Nevertheless, in this work
it was considered of the first order with respect to CO2, since the reaction kine-
tics depends on the carbonation ratio (136).
The absorption rate of carbon dioxide increases in the presence of amines
or ammonia. Therefore, the reaction kinetics of NH3 and CO2 has been consid-
ered in the model equations, too. The rate constant as a function of the temperature
has been determined according to Ref. 136. The coefficients for the calculation of
the chemical equilibrium constants in this system of volatile weak electrolytes are
taken from Ref. 137.
The CO2 absorption is hindered by a slow chemical reaction by which the
dissolved carbon dioxide molecules are converted into the more reactive ionic
species. Therefore, when gases containing H2S, NH3, and CO2 contact water, the
H2S and ammonia are absorbed much more rapidly than CO2, and this selectivity
can be accentuated by optimizing the operating conditions (23). Nevertheless, all
chemical reactions are coupled by hydronium ions, and additional CO2 absorption
leads to the desorption of hydrogen sulfide and decreases the scrubber efficiency.
3.2.2. Process Setup
Today’s coke plant gas purification processes are mostly carried out under atmos-
pheric pressure, employing a circulated ammonia-based absorbent. The con-
sumption of the external solvent is reduced via the use of ammonia available in
the coke gas (138). An example of innovative purification processes is the ammo-
nia hydrogen sulfide circulation scrubbing (ASCS) (Figure 17), in which the
ammonia contained in the raw gas dissolves in the NH3 absorber and then the
absorbent saturated with the ammonia passes through the H2S absorber to selec-
tively absorb the H2S and HCN components from the coke gas. The next step is the
thermal regeneration of the absorbent with the steam in a two-step desorption plant,
whereas a part of the deacidified water is fed back into the H2S absorber (25).
Pilot-plant experiments have been carried out at real process conditions in
the coke plant “August Thyssen” (Duisburg, Germany). The DN 100 pilot column
(Figure 17) was made of stainless steel and equipped with about 4 m of structured
packing (Sulzer MELLAPAK® 350Y), three liquid distributors, and a digital con-
trol system. Several steady-state experiments have been compared with the sim-
ulation results and supported the design optimization of the coke gas purification
process (25).
3.2.3. Results and Discussion
A number of steady-state simulations have been performed with the aim of ana-
lyzing the influence of numerical and physicochemical parameters, beginning

Copyright © 2004 by Marcel Dekker, Inc. All Rights Reserved.


FIGURE 17 Ammonia hydrogen sulfide circulation scrubbing process for the coke oven gas purification
(right) and H2S absorber (left).

Copyright © 2004 by Marcel Dekker, Inc. All Rights Reserved.


with a single stage and ending with a simulation of a column. Different film and
packing section discretizations, several mass transfer and hydrodynamic corre-
lations, and different driving forces and diffusion models have been thoroughly
tested (Figure 18).
The most sensitive components appeared to be those involved in finite-rate
reactions, especially CO2. Furthermore, the impact of electrical forces enhances
the absorption of the strong electrolytes H2S and HCN by 3–5%, while the CO2
absorption rate is dominated by the reaction in the film (139,140). Significant
changes in the concentration profiles and the component absorption rates due to
the film reaction have been established (141,142).
Single-stage simulations reveal that intermolecular friction forces do not
lead to reverse diffusion effects, and thus the molar fluxes calculated with the
effective diffusion approach differ only slightly from those obtained via the
Maxwell–Stefan equations without the consideration of generalized driving
forces. This result is as expected for dilute solutions and allows one to reduce
model complexity for the process studied (143).
As a further model simplification, a linearization of the film concentration
profiles has been studied. This causes no significant changes in the simulation
results and at the same time reduces the total number of equations by half and sta-
bilizes the numerical solution (142). The assumption of chemical equilibrium in

FIGURE 18 Absorption rates calculated with different model assumptions


concerning reaction consideration.

Copyright © 2004 by Marcel Dekker, Inc. All Rights Reserved.


FIGURE 19 Liquid-phase axial concentration profiles for the H2S scrubber:
comparison between experimental and simulation results based on different
model approaches.

the liquid bulk phase does not change the absorption rates significantly, which
indicates fast conversion. Therefore, neglecting the film reaction unrealistically
reduces the absorption rates. On the other hand, neglecting the reaction kinetics
within the film results in completely different orders of magnitude for the calcu-
lated absorption degree. As a consequence, the reactions of carbon dioxide should
not be regarded as instantaneous, although the corresponding Hatta number of
about 7 characterizes the reaction as very fast (3).
The model optimized with respect to the numerical parameters and physico-
chemical properties has been validated against experimental data, whereas the
axial concentration and temperature profiles for both phases demonstrated good
agreement (Figure 19). It has also been found that the simulations of the scrubber
based on the equilibrium-stage model extended by the chemical reaction kinetics
yield results completely inconsistent with the experimental studies; namely, the
selectivity toward H2S and HCN absorption cannot be reflected (Figure 19). In
this case, the film reaction represents an essential element of the rate-based
approach that has to be considered in the model. As a result, the only feasible sim-
plification is represented by a linearization of the film concentration profiles,
including the implementation of the average reaction kinetics in the liquid film
region (143).

Copyright © 2004 by Marcel Dekker, Inc. All Rights Reserved.


3.2.4. Dynamic Modeling
Steady-state modeling is not sufficient if one faces various disturbances in RA oper-
ations (e.g., feed variation) or tries to optimize the startup and shutdown phases of
the process. In this case, a knowledge of dynamic process behavior is necessary.
Further areas where the dynamic information is crucial are the process control as
well as safety issues and training. Dynamic modeling can also be considered as the
next step toward the deep process analysis that follows the steady-state modeling
and is based on its results.
The dynamic formulation of the model equations requires a careful analy-
sis of the whole system in order to prevent high-index problems during the
numerical solution (144). As a consequence, a consistent set of initial conditions
for the dynamic simulations and suitable descriptions of the hydrodynamics have
to be introduced. For instance, pressure drop and liquid holdup must be corre-
lated with the gas and liquid flows.
The model optimized based on steady-state analysis allows for a dynamic
real-time simulation of the entire absorption process. Because dynamic behavior is
determined mainly by process hydraulics, it is necessary to consider those elements
of the column periphery that lead to larger time constants than the column itself.
Therefore, major elements of the column periphery, such as distributors, stirred
tanks, and pipelines, have been additionally implemented into the dynamic model.
The dynamic behavior of the coke gas purification process has been inves-
tigated systematically (139,140,145). For instance, local perturbations of the gas
load and its composition have been analyzed. A significant dynamic parameter is
represented by the liquid holdup. Figure 20 demonstrates the changes of the sol-
vent composition after a decrease of the gas-flow rate from 67 m3/h to 36.4 m3/h
and a simultaneous small increase in the liquid-flow rate.
The liquid holdup of the packing section decreases, which leads to a lower
conversion of the kinetically controlled reactions of CO2 and a reduction in the
CO2 absorption rate. As a consequence, the solvent mole fractions of HCO3 and
carbamate decreases whereas the relative fraction of HS increases. The select-
ivity of the absorption process toward the H2S and HCN reduction is enhanced by
minimizing the liquid holdup of the column. At the same time, a larger interfacial
area improves the performance of the plant. Therefore, modern industrial sour gas
scrubbers should be equipped with structured packings.
Figure 21 illustrates the system response after a sudden increase in the gas
flow by 20% and its H2S load by 100%. As expected, the H2S load increases
everywhere along the column height in the gas phase. The change is more signifi-
cant in the lower part of the absorber than at the top because some additional
hydrogen sulfide is absorbed. The new steady state is already achieved after
30 minutes, which justifies the implementation of dynamic models for the column
periphery.

Copyright © 2004 by Marcel Dekker, Inc. All Rights Reserved.


FIGURE 20 Dynamic change of solvent composition after a sudden significant decrease in the gas-flow
rate and a simultaneous small increase in the liquid-flow rate.

Copyright © 2004 by Marcel Dekker, Inc. All Rights Reserved.


FIGURE 21 Dynamic change of the H2S gas-phase concentration along the
column after a sudden increase in the gas flow and its H2S load.

3.3. Methyl Acetate Synthesis, Batch Distillation


3.3.1. Chemical System
The synthesis of methyl acetate from methanol and acetic acid is a slightly
exothermic equilibrium-limited liquid-phase reaction:

CH 3OH  (CH 3 )COOH ↔ (CH 3 )COO(CH 3 )  H 2 O


H R0 4.2 kJ/mol (R17)

The low equilibrium constant and the strongly nonideal behavior that causes the
forming of the binary azeotropes methyl acetate/methanol and methyl acetate/
water make this reaction system interesting as a possible RD application (33).
Therefore, methyl acetate synthesis has been chosen as a test system and investi-
gated in a semibatch RD column. Since the process is carried out under atmo-
spheric pressure, no side reactions in the liquid phase occur (146).
3.3.2. Process Setup and Operation
The catalytic packing MULTIPAK® (147) applied in this case study consists of
corrugated wire gauze sheets and catalyst bags of the same material assembled
in alternate sequence. Sufficient mass transfer between gas and liquid phase is

Copyright © 2004 by Marcel Dekker, Inc. All Rights Reserved.


guaranteed by segmentation of the catalyst bags and numerous contact spots with
the wire gauze sheets. The packing was equipped with the acid ion exchange resin
known as an effective catalyst for esterification processes (34,148).
A batch distillation column with a diameter of 100 mm and a reactive pack-
ing height of 2 m (MULTIPAK I®) in the bottom section and an additional meter
of conventional packing (ROMBOPAK 6M®) in the top section was used. The
flow sheet of the column is shown in Figure 22.
At first, the distillation still was charged with methanol—the low-boiling
reactant—and heated under total reflux until steady-state conditions were achieved.
At that moment, acetic acid—the high-boiling reactant—was fed above the reaction
zone to the second distributor. After 30 min the reflux ratio was changed from
infinity to 2 and the product withdrawal at the top of the column began. During the

FIGURE 22 Reactive distillation column, batch operation.

Copyright © 2004 by Marcel Dekker, Inc. All Rights Reserved.


column operations, the liquid-phase concentration profiles along the column and
the temperature profiles were measured. For the determination of the liquid-phase
composition, two methods were applied simultaneously. On the one hand, samples
were taken and analyzed by gas chromatography. On the other hand, an online
NIR spectrometer was used to determine the concentration without taking any
samples (149).
3.3.3. Results and Discussion
Figures 23 and 24 show the liquid-phase compositions for, respectively, the
reboiler and condenser as functions of time. After column startup, the concentra-
tion of methanol decreases continuously whereas the distillate mole fraction of
methyl acetate reaches about 90%. A comparison of the rate-based simulation
(with the Maxwell–Stefan diffusion equations) and experimental results for the
liquid-phase composition at the column top and in the column reboiler demonstrates
their satisfactory agreement (Figures 23 and 24). Figure 25 shows the simulation

FIGURE 23 Liquid mole fractions in the column reboiler: lines, simulations;


dots, experiments.

Copyright © 2004 by Marcel Dekker, Inc. All Rights Reserved.


FIGURE 24 Liquid mole fractions in the column condenser: lines, simula-
tions; dots, experiments.

results obtained with different modeling approaches, after an operation time of


10,000 s. The reference model employs the rate-based approach and the Maxwell–
Stefan diffusion equations. Another rate-based model assumes effective diffusion
coefficients instead of the Maxwell–Stefan equations. The third model used is an
equilibrium-based one. Both the reference model and effective-diffusion model
show similar results. The equilibrium-stage model is only able to describe the
process behavior qualitatively. This is in contrast to the reactive absorption
processes (see Sections 3.1 and 3.2) and can be explained by the low reaction rate,
which dominates the whole process kinetics.

3.4. Methyl Acetate Synthesis, Steady-State Distillation


3.4.1. Chemical System
The synthesis of methyl acetate from methanol and acetic acid analyzed in this
case study is the same as described by Reaction (R17):

CH 3OH  (CH 3 )COOH ↔ (CH 3 )COO(CH 3 )  H 2 O

Copyright © 2004 by Marcel Dekker, Inc. All Rights Reserved.


.

FIGURE 25 Axial concentration profiles for the semibatch column


(t  10,000 s ) .

3.4.2. Process Setup


The column (35) has an inner diameter of 50 mm and a total packing height of
4 m composed of a reactive section of 2 m and two nonreactive sections of 1 m
each, below and above the reactive part of the column (Figure 26). The acetic acid
feed is located above the catalytic packing, while methanol is fed to the column
below the reactive section. A similar column design was presented in Ref. 150.
3.4.3. Results and Discussion
A series of experiments have been performed with a stoichiometric feed ratio of
acetic acid and methanol. The reflux ratio was kept constant at a value of 2.0, the
feed flow rate at a value of 3.0 kg/h, while the heat duty to the reboiler was varied
over a wide range. A comparison of experimental results and model prediction for
the liquid-phase composition profiles along the column is given in Figure 27 for
different reboiler duties (151). The theoretical values are displayed with continu-
ous lines and empty symbols, whereas the experimental data measured along the

Copyright © 2004 by Marcel Dekker, Inc. All Rights Reserved.


column are shown by the relevant filled symbols. It can be seen that the process
behavior is reflected by the simulations with high accuracy. A maximum concen-
tration of methanol and acetic acid can be observed at the respective feed loca-
tions, while methyl acetate is enriched toward the top and water toward the
bottom of the column.

3.5. Synthesis of Methyl Tertiary Butyl Ether


3.5.1. Chemical System
The synthesis of methyl tertiary butyl ether (MTBE) is one of the most important
applications of RD. MTBE is produced via an acid-catalyzed reaction between
methanol and isobutylene:
CH 3OH  C(CH 3 ) 2 CH 2 ↔ C(CH 3 )3 OCH 3
H R0 37.7 kJ/mol (R18)
This reaction has been extensively studied by several authors, e.g., Refs. 152–154.
3.5.2. Process Setup
MTBE synthesis was investigated both theoretically and experimentally. Here,
some results for a pilot-scale RD column at Neste Oy Engineering, Finland, are

FIGURE 26 Reactive distillation column, steady-state operation.

Copyright © 2004 by Marcel Dekker, Inc. All Rights Reserved.


FIGURE 27 Liquid-phase composition profiles along the 50-mm-diameter
catalytic distillation column for methyl acetate synthesis at reflux ratio of 2.0
and different reboiler duties: (a) 295W (b) 873W (c) 1161W.

presented (155). The column (used as an example here) has a catalytic section in the
middle part. This catalytic section may consist either of a packed bed of catalytical-
ly active rings (91) or of structured catalytic packing (147). The rectifying and strip-
ping sections are filled with Intalox Metal Tower Packing. The methanol feed is
introduced just above the catalyst section of the column and the hydrocarbon feed
just below.
3.5.3. Results and Discussion
Figure 28 demonstrates the simulated and measured concentration profiles for the
pilot test made in the column, with the reactive section filled with catalytically

Copyright © 2004 by Marcel Dekker, Inc. All Rights Reserved.


active rings. In the simulations four components—methanol, isobutene, MTBE,
and 1-butene—were chosen to describe the system under consideration. Here,
segment 1 corresponds to the reboiler. A satisfactory agreement between calcu-
lated and measured values can be clearly observed. In Figure 29, the simulation
results for the column with different reactive internals, catalytic packing
MULTIPAK®, are shown. Here, 16 components were considered. Again, the
liquid bulk composition profiles demonstrated in Figure 29 agree well with the
experimental data.

3.6. Reactive Extraction of Zinc


3.6.1. Chemical System
The extraction of zinc with the cation exchanger di(2-ethylhexyl)phosphoric acid,
RH, is recommended by the EFCE as a test system for RE. Physical properties,
handling, equilibrium data, etc. are documented on the internet (http:// www.
dechema.de/Extraction, http://www.icheme.org/learning).
In brief, the ion exchanger is dimeric, R 2 H 2 , in aliphatic diluents (156),
and the overall reaction is then

aZn 2  b R 2 H 2 ↔ Zn a R 2 a (RH)2 b2 a  2 H (R19)


At low concentrations, polynuclear complexes do not exist (157); thus a  1:

Zn 2  b R 2 H 2 ↔ ZnR 2 (RH)2 b2  2 H (R20)

FIGURE 28 Calculated and experimental liquid compositions for experiments


with catalytically active rings.

Copyright © 2004 by Marcel Dekker, Inc. All Rights Reserved.


FIGURE 29 Calculated and experimental liquid compositions for experiments with catalytic
structured packing.

Copyright © 2004 by Marcel Dekker, Inc. All Rights Reserved.


The mass action law on the basis of concentrations then yields

[ ZnR 2 ( RH) 2 b2 ][H ]2 [H ]2


K eq   D (3)
[ Zn 2 ][ R 2 H 2 ]b
Zn
[ R 2 H 2 ]b

where DZn denotes the partition coefficient of zinc between the organic and the
aqueous phase (bar indicates organic species). In logarithmic form, the stoichio-
metry of the complex, b, is determined by slope analysis:

log DZn  b log([R 2 H 2 ])  2 pH  log K eq


(4)
As can be seen, the partition increases with increasing ion exchanger concentra-
eq
tion, R 2 H 2 , and with pH. The determination of the equilibrium constant, K , is
discussed, e.g., in Ref. 158.
3.6.2. Process Setup
Apart from the nuclear industry, the most frequently installed pieces of process
equipment are mixer-settler cascades. The advantage is the easy control of each
stage regarding the pH value, selection of the phase to be dispersed, etc. (12,15).
The disadvantages are the high investment costs and large solvent inventory.
Nowadays, modern designed extractants (with fast kinetics) allow a process
design with highly efficient extraction columns (159). Applications are found in
the chemical industry, e.g., with sulfonic acid extraction (160). The first applica-
tion on a big scale in hydrometallurgy was reported in 1997 at WMC Olympic
Dam, Australia, where 10 pulsed Batman columns (0.5 to 3 m in diameter and
35 m tall) were used for uranium recovery. An increased recovery (from 90% to
97%) was found after replacing the formerly used mixer-settler units.
The RE of zinc is reported in detail when using an RDC, including the dis-
cussion of the stripping process to regenerate the ion exchanger for cyclic reuse
(161–163).
3.6.3. Results and Discussion
A comparison of predicted and experimental mass transfer coefficients is given in
Figure 30. The simulated overall mass transfer coefficient originates from a
model that is a combination of the microkinetic reaction according to Eq. (B9)
and eddy diffusion according to Eq. (B11) (see Appendix B). Figure 30 shows
that the mass transfer coefficient at higher concentrations is generally underesti-
mated, thus including some safety value for the process design. As discussed in
detail in Ref. 56, combinations of microkinetics [Eq. (B9)] with other eddy dif-
fusion correlations instead of Eq. (B11) (e.g., Refs. 164 and 165) are also appro-
priate to describe the system. In contrast to this, the combination of microkinetics
with molecular diffusion concepts fails, and the same is true when the equilibrium

Copyright © 2004 by Marcel Dekker, Inc. All Rights Reserved.


FIGURE 30 Comparison of experimental and simulated overall mass transfer
coefficients at different initial zinc concentrations and droplet diameters
(1, 2, or 3 mm ) .

approach is used neglecting the kinetics rate law. The results of the column sim-
ulations are discussed in Refs. 162 and 163, and a special discussion on contam-
ination effects is given in Ref. 166.

4. SUMMARY AND OUTLOOK


This chapter concerns the most important reactive separation processes: reactive
absorption, reactive distillation, and reactive extraction. These operations combin-
ing the separation and reaction steps inside a single column are advantageous as
compared to traditional unit operations. The three considered processes are similar
and at the same time very different. Therefore, their common modeling basis is dis-
cussed and their peculiarities are illustrated with a number of industrially relevant
case studies. The theoretical description is supported by the results of laboratory-,
pilot-, and industrial-scale experimental investigations. Both steady-state and
dynamic issues are treated; in addition, the design of column internals is addressed.
Reactive absorption, reactive distillation, and reactive extraction occur in
multicomponent multiphase fluid systems, and thus a single modeling framework
for these processes is desirable. In this respect, different possible ways to build such
a framework are discussed, and it is advocated that the rate-based approach pro-
vides the most rigorous and appropriate way. By this approach, direct consideration

Copyright © 2004 by Marcel Dekker, Inc. All Rights Reserved.


of the diffusional and reaction kinetics is realized. Special attention is paid to the
application of CFD, which could become a powerful theoretical tool to predict the
flow behavior for different column units and internals geometries for engineering
applications. In particular, CFD can play an outstanding role in the development of
the column internals for reactive separations. Fundamental advances in the under-
standing of the underlying physicochemical phenomena when coupled with CFD
would go a long way toward support of reactive separation technology.
The modeling of RA is illustrated by the absorption of NOx and by the
coke gas purification process. The first case is modeled by using an analytical treat-
ment of the film phenomena, whereas the second one is solved by a purely numer-
ical technique. The simulation results are compared with the experimental data
obtained at an industrial absorption plant consisting of eight units with
pump around (NOx) and at a pilot column for the ammonia hydrogen sulfide
circulation scrubbing process (coke gas purification). For the latter case, both
steady-state and dynamic conditions are considered. The comparison results, on the
one hand, demonstrate a good agreement between the rate-based simulations and
experimental data, and, on the other hand, warn of using the equilibrium approach,
which appears completely inappropriate in the case of complex finite-rate reactions.
The modeling of RD processes is illustrated with the heterogenously catalyzed
synthesis of methyl acetate and MTBE. The complex character of reactive distilla-
tion processes requires a detailed mathematical description of the interaction of
mass transfer and chemical reaction and the dynamic column behavior. The most
detailed model is based on a rigorous dynamic rate-based approach that takes into
account diffusional interactions via the Maxwell–Stefan equations and overall reac-
tion kinetics for the determination of the total conversion. All major influences of the
column internals and the periphery can be considered by this approach.
As an application example, the dynamic model was used for the simulation
of the steady-state and semibatch production of methyl acetate, performed in a
packed column with a catalytic packing. For the model validation, several experi-
ments were carried out in a pilot-plant column. For the investigated operation
range, the simulation results are in good agreement with the experimental data.
The use of this model for model-based process control calls for suitable
model reductions without a significant decrease in the predictivity. For the methyl
acetate process, a simplified description of the mass transfer using effective dif-
fusion coefficients and neglecting diffusional interactions seems to be sufficient.
On the other hand, a detailed description of the reaction, including the specific
phenomena of the heterogeneous catalysis by an adequate consideration of the solid
phase, is required for the predictive simulation of even more complex systems,
including side and consecutive reactions. Optimal functioning of reactive distil-
lation depends on careful process design, with appropriately selected column
internals, feed locations, and catalyst placement. Greater understanding of the
general and particular features of the process behavior is equally essential.

Copyright © 2004 by Marcel Dekker, Inc. All Rights Reserved.


The modeling of RE processes is strongly related to the knowledge of the
reaction equilibrium and kinetics and mass transfer regime. The latter is decisive for
column simulations, whereas eddy diffusivity concepts often have to be used. The
parameters can easily be obtained in small-scale laboratory devices, with a minimum
of substance involved. A real challenge is the correct hydrodynamic description of
the two-phase flow. The assumption of plug flow gives acceptable results with col-
umn diameters smaller than 0.1 m. The commonly applied axial dispersion model
or back-mixing model uses one parameter to account for nonideal flow in each
phase. Here the dispersed phase is considered to be pseudo-continuous and
monodisperse. Droplet population models taking into account the dynamic pro-
cesses of coalescence and breakup of droplets should give a more realistic picture
and thus a more firm design of a process. The use of CFD calculations in liquid–
liquid dispersed-phase flow is limited to single-droplet flow or low column holdup.
The simulation of large industrial columns especially is not feasible nowadays.
Some important general aspects of rate-based modeling as well as further
peculiarities of the specific process applications and the different solution strat-
egies are given in Appendices A and B.
The key reactive separation topics to be addressed in the near future are a
proper hydrodynamic modeling for catalytic internals, including residence time
distribution account and scale-up methodology. Further studies on the hydrodynam-
ics of catalytic internals are essential for a better understanding of RSP behavior
and the availability of optimally designed catalytic column internals for them. In
this regard, the methods of computational fluid dynamics appear very helpful.
The development of new methodologies enabling the creation of intelligent,
tailor-made column internals and consequent RSP optimization constitutes one
of the burning present-day challenges. Such a development is already in progress
in some European research projects.
Despite the recent rapid development of computer technology and numer-
ical methods, the rate-based approach in its current realization still often requires
a significant computational effort, with related numerical difficulties. This is one
of the reasons the application of rate-based models to industrial tasks is rather
limited. Therefore, further work is required in order to bridge this gap and pro-
vide chemical engineers with reliable, consistent, robust, and comfortable simu-
lation tools for reactive separation processes.

ACKNOWLEDGMENTS
We would like to thank our colleagues at the Chair of Fluid Separation Processes,
Dortmund University, and all other project partners who have been involved
in the research activities. We are also grateful to the German Research Founda-
tion (DFG, Grants No. Schm 808/5-1, Ba 1569/2-1 2-2, Ba 1569/6-1), the
Volkswagen Foundation (Project No. I/70 875, 876, 877), the European

Copyright © 2004 by Marcel Dekker, Inc. All Rights Reserved.


Commission (BRITE-EURAM program, CEC Project No. BE95-1335), the
German Federal Ministry of Education and Research (BMBF, Project No.
03C0306), the Foundation “Rheinland-Pfalz für Innovation” (836-386261/193),
as well as BASF AG, Bayer AG, Axiva GmbH, Degussa.

NOMENCLATURE
aI specific gas–liquid interfacial area m2/m3
As column cross section m2
B liquid load m3/(m2s)
c molar concentration mol/m3
CIP adjustable parameter, Eq. (B10)
dC column diameter m
di generalized driving force for component i 1/m
dp droplet diameter m
D Maxwell–Stefan diffusion coefficient m2/s
Dax axial dispersion coefficient m2/s
Deff effective diffusion coefficient m2/s
DZn partition coefficient of zinc
E length-specific energy holdup J/m
E dimensionless residence time distribution
F Faraday’s constant 9.65  104 C/mol
FC gas capacity factor Pa0.5
G gas molar flow rate mol/s
h molar enthalpy J/mol
H R0 reaction enthalpy J/mol
ky overall mass transfer coefficients m/s
Ki distribution coefficient
Keq equilibrium constant
[K] reaction matrix [Eq. (B1)] 1/s
l axial coordinate m
L liquid molar flow rate mol/s
n number of components of mixture
Ni molar flux of component i mol/(m2s)
Q heat flux W/m2
R total component reaction rate mol/m3s
R column vector with elements Ri mol/m3s
 gas constant 8.3144 J /(mol K)
Re Reynolds number
Sc Schmidt number
Sh Sherwood number
t time s
T temperature K
uL liquid velocity m/s
U length-specific molar holdup mol/m

Copyright © 2004 by Marcel Dekker, Inc. All Rights Reserved.


w terminal velocity m/s
xi first fluid-phase (liquid) mole fraction mol/mol
of component i
x column vector with elements xi mol/mol
yi second fluid-phase (gas, vapor, or liquid) mol/mol
mole fraction of component i
z film coordinate m
zi ionic charge of component i

Greek Letters
 film thickness m
 dimensionless film coordinate
 forward-reaction constant m3/2/(mol1/2s)
r backward-reaction constant s1
 thermal conductivity W/(m K)
 chemical potential J/mol
c dynamic viscosity of continuous phase Pa  s
d dynamic viscosity of dispersed phase Pa  s
 volumetric holdup m3/m3
electrical potential V

Subscripts
G gas or second fluid phase
i, j component/reaction indices
L liquid phase
t mixture property

Superscripts
B bulk phase
I phase interface

Abbreviations
ADM axial dispersion model
CD catalytic distillation
PDE piston flow model with axial
dispersion and mass exchange
RA reactive absorption
RD reactive distillation
RE reactive extraction
RH di(2-ethylhexyl) phosphoric acid
RSP reactive separation process

Copyright © 2004 by Marcel Dekker, Inc. All Rights Reserved.


REFERENCES
1. Noble T. Multifunctional equipment—all for one, and one for all? Chem Eng Prog
2001; 97(11):10–13.
2. Doherty MF, Buzad G. Reactive distillation by design. Trans IChemE 1992; 70:
448–458.
3. Zarzycki R, Chacuk A. Absorption: Fundamentals and Applications. Oxford: Perga-
mon Press, 1993.
4. Agar DW. Multifunctional reactors: old preconceptions and new dimensions. Chem
Eng Sci 1999; 54:1299–1305.
5. Kelkar VV, Ng KM. Design of reactive crystallization systems incorporating kine-
tics and mass-transfer effects. AIChE J 1999; 45:69–81.
6. Stankiewicz AI, Moulijn JA. Process intensification: transforming chemical engi-
neering. Chem Eng Prog 2000; 96(1):22–34.
7. Fricke J, Schmidt-Traub H, Kawase M. Chromatographic reactors. In: Ullmann’s
Encyclopedia of Industrial Chemistry. Weinheim, Germany: Wiley-VCH, 2001.
8. Bart HJ. Reactive Extraction. Berlin: Springer, 2001.
9. Noeres C, Kenig EY, Górak A. Modelling of reactive separation processes: reactive
absorption and reactive distillation. Chem Eng Process 2003; 42:157–178.
10. Hatcher WJ. Reaction and mass transport in two-phase reactors. In: Cheremisinoff
NP, ed. Handbook of Heat and Mass Transfer. Vol. 2. Houston: Gulf, 1986:837–868.
11. Ritcey GM, Ashbrook AW. Solvent Extraction Principles and Applications to
Process Metallurgy. Vols. I, II. New York: Elsevier, 1979.
12. Lo TC, Baird MHI, Hanson C, eds. Handbook of Solvent Extraction. New York:
Wiley, 1983.
13. Thornton JD, ed. Science and Practice of Liquid–Liquid Extraction. Vols. I, II.
Oxford: Oxford University Press, 1992.
14. Marr R, Bart HJ. Metallsalzextraktion. Chem Ing Tech 1982; 54:119–129.
15. Godfrey JC, Slater MJ, eds. Liquid–Liquid Extraction Equipment. Chichester, UK:
Wiley, 1994.
16. Marcus Y, Sengupta AK, eds. Ion Exchange and Solvent Extraction. Vol. 15.
New York: Marcel Dekker, 2002.
17. Sherwood TK, Pigford RL, Wilke CR. Mass Transfer. New York: McGraw-Hill, 1975.
18. Sherwood TK, Pigford RL. Absorption and Extraction. New York: McGraw-Hill, 1952.
19. Counce RM, Perona JJ. Designing packed-tower wet scrubbers: emphasis on nitro-
gen oxides. In: Cheremisinoff NP, ed. Handbook of Heat and Mass Transfer. Vol. 2.
Houston: Gulf, 1986:953–966.
20. Sattler K. Thermische Trennverfahren. Weinheim, Germany: VCH, 1988.
21. Strelzoff S. Choosing the optimum CO2-removal system. Chem Eng 1975; 82:
115–120.
22. Falbe J. Chemierohstoffe aus Kohle. Stuttgart: Georg-Thieme-Verlag, 1977.
23. Kohl AL, Riesenfeld FC. Gas Purification. Houston: Gulf, 1985.
24. Yu WC, Astarita G. Selective absorption of hydrogen sulphide in tertiary amine
solutions. Chem Eng Sci 1987; 42:419–424.
25. Thielert H. Simulation und Optimierung der Kokereigaswäsche. Ph.D. dissertation,
Technical University Berlin, Berlin, 1997.

Copyright © 2004 by Marcel Dekker, Inc. All Rights Reserved.


26. Kobus A. Ein heuristisch-numerischer Ansatz zum systematischen Entwurf und
Design von Absorptionsverfahren. Düsseldorf: VDI, 1999.
27. Danckwerts PV. Gas–Liquid Reactions. New York: McGraw-Hill, 1970.
28. Astarita G, Savage DW, Bisio A. Gas Treating with Chemical Solvents. New York:
Wiley, 1983.
29. Pal SK, Sharma MM. Solvay-type process: absorption of CO2 in ammoniated aque-
ous and nonaqueous solutions of NaNO3, Ca(NO3)2, and KCl. Ind Eng Chem Proc
Des Dev 1983; 22:76–79.
30. Sauchel V. Chemistry and Technology of Fertilizers. New York: Reinhold, 1960.
31. Gandhidasan P. Heat and mass transfer in solar regenerators. In: Cheremisinoff
NP, ed. Handbook of Heat and Mass Transfer. Vol. 2. Houston: Gulf, 1986:
1475–1499.
32. Kenig EY, Kholpanov LP, Katysheva LI, Markish IH, Malyusov VA. Calculation of
two-phase nonisothermal absorption in a liquid film in downward cocurrent flow.
Theor Found Chem Eng 1985; 19:97–102.
33. Agreda VH, Partin LR, Heise WH. High-purity methyl acetate via reactive distilla-
tion. Chem Eng Prog 1990; 86(2):40–46.
34. Kreul LU, Górak A, Dittrich C, Barton PI. Dynamic catalytic distillation: advanced
simulation and experimental validation. Computers Chem Eng 1998; 22:371–378.
35. Górak A, Hoffmann A. Catalytic distillation in structured packings: methyl acetate
synthesis. AIChE J 2001; 47:1067–1076.
36. Komatsu H, Holland CD. A new method of convergence for solving reacting distil-
lation problems. J Chem Eng Japan 1977; 4:292–297.
37. Hartig H, Regner H. Verfahrenstechnische Auslegung einer Veresterungskolonne.
Chem Ing Tech 1971; 43:1001–1007.
38. Davies B, Jeffreys GV. The continuous trans-esterification of ethyl alcohol and
butyl acetate in a sieve-plate column. Part III: trans-esterification in a six-plate
sieve-tray column. Trans IChemE 1973; 51:275–280.
39. Luo HP, Xiao WD. A reactive distillation process for a cascade and azeotropic
reacting system: carbonylation of ethanol with dimethyl carbonate. Chem Eng Sci
2001; 56:403–410.
40. Fuchigami Y. Hydrolysis of methyl acetate in distillation column packed with reac-
tive packing of ion exchange resin. J Chem Eng Japan 1990; 23:354–358.
41. DeGarmo JL, Parulekar VN, Pinjala V. Consider reactive distillation. Chem Eng
Prog 1992; 88(3):42–50.
42. Sundmacher K, Hoffmann U. Activity evaluation of a catalytic distillation packing
for MTBE production. Chem Eng Technol 1993; 16:279–289.
43. Thiel C. Modellbildung, Simulation, Design und experimentelle Validierung von
heterogen katalysierten Reaktivdestillationsprozessen zur Synthese der Kraftstoffether
MTBE, ETBE und TAME. Ph.D. dissertation, Technical University of Clausthal,
Clausthal, Germany, 1997.
44. Oost C. Ein Beitrag zum Umweltschutz: Reaktionstechnische Untersuchungen zur
Flüssigphasesynthese des Antiklopfmittels TAME. Ph.D. dissertation, Technical
University of Clausthal, Clausthal, Germany, 1995.
45. Shoemaker JD, Jones EM. Cumene by catalytic distillation. Hydroc Proc 1987;
66:57–58.

Copyright © 2004 by Marcel Dekker, Inc. All Rights Reserved.


46. Podrebarac GG, Ng FTT, Rempel GL. The production of diacetone alcohol with
catalytic distillation. Part I: catalytic distillation experiments. Chem Eng Sci 1998;
53:1067–1075.
47. Ung S, Doherty MF. Vapor–liquid equilibrium in systems with multiple chemical
reactions. Chem Eng Sci 1995; 50:23–48.
48. Müller D, Schäfer JP, Leimkühler HJ. The economic potential of reactive distillation
processes exemplified by silane production. Proceedings of VDI-GVC, DECHEMA
and EFCE Meeting, Section “Thermal Separations,” Bamberg, Germany, 2001.
49. Ciric AR, Miao P. Steady-state multiplicities in an ethylene glycol reactive distilla-
tion column. Ind Eng Chem Res 1994; 33:2738–2748.
50. Belson DJ. A distillation method of aromatic nitration using azeotropic nitric acid.
Ind Eng Chem Res 1990; 29:1562–1565.
51. Fair JR. Design aspects for reactive distillation. Chem Eng 1998; 105(11):158–162.
52. Rock K, Gildert GR, McGuirk T. Catalytic distillation extends its reach. Chem Eng
1997; 104(7):78–80.
53. Marion MC, Viltard JC, Travers P, Harter I, Forestiere A. U.S. Patent 5,776,320,
Institute Français du Petrol, 1998.
54. Klocker H, Bart HJ, Marr R, Müller H. Mass transfer based on chemical potential
theory: ZnSO4/H2SO4/D2EHPA. AIChE J 1997; 43:2479–2487.
55. Bart HJ, Rousselle HP. Microkinetics and reaction equilibria in the system ZnSO4/
D2EHPA/isododecane. Hydrometallurgy 1999; 51:285–298.
56. Mörters M, Bart HJ. Mass transfer into droplets in reactive extraction. Chem Eng
Process, 2003, in press.
57. Taylor R, Krishna R. Multicomponent Mass Transfer. New York: Wiley, 1993.
58. Kenig EY, Górak A. A film model based approach for simulation of multicompo-
nent reactive separation. Chem Eng Process 1995; 34:97–103.
59. Kenig EY. Modeling of Multicomponent Mass Transfer in Separation of Fluid
Mixtures. Düsseldorf: VDI, 2000.
60. Sundmacher K, Hoffmann U. Multicomponent mass and energy transport on dif-
ferent length scales in a packed reactive distillation column for heterogeneously cata-
lyzed fuel ether production. Chem Eng Sci 1994; 49:4443–4464.
61. Krishna R, Wesselingh JA. The Maxwell–Stefan approach to mass transfer. Chem
Eng Sci 1997; 52:861–911.
62. Higler AP, Krishna R, Taylor R. Nonequilibrium modeling of reactive distillation:
a dusty fluid model for heterogeneously catalyzed processes. Ind Eng Chem Res
2000; 39:1596–1607.
63. Yuxiang Z, Xien X. Study on catalytic distillation processes. Part II. Simulation of
catalytic distillation processes—quasi-homogenous and rate-based model. Trans
IChemE 1992; 70:465–470.
64. Sundmacher K, Uhde G, Hoffmann U. Multiple reactions in catalytic distillation
processes for the production of fuel oxygenates MTBE and TAME: analysis by rig-
orous model and experimental validation. Chem Eng Sci 1999; 54:2839–2847.
65. Mackowiak J. Fluiddynamik in Kolonnen mit modernen Füllkörperpackungen und
Packungen für Gas/Flüssigkeitssysteme. Frankfurt am Main: Sauerländer, 2001.
66. Billet R. Packed Towers. Weinheim, Germany: VCH, 1995.

Copyright © 2004 by Marcel Dekker, Inc. All Rights Reserved.


67. Kenig EY, Klöker M, Egorov Y, Menter F, Górak A. Towards improvement of reac-
tive separation performance using computational fluid dynamics. Proceedings of
International Symposium on Multifunctional Reactors (ISMR-2), Nuremberg, 2001.
68. Doraiswamy LL, Sharma MM. Heterogeneous Reactions: Analysis, Examples and
Reactor Design. New York: Wiley, 1984.
69. Westerterp KR, van Swaaij WPM, Beenackers AACM. Chemical Reactor Design
and Operation. Chichester, UK: Wiley, 1984.
70. Chen CC, Evans LB. A local composition model for the excess Gibbs energy of
aqueous electrolyte systems. AIChE J 1986; 32:444–459.
71. Mock B, Evans LB, Chen CC. Thermodynamic representation of phase equilibria
of mixed-solvent electrolyte systems. AIChE J 1986; 32:1655–1664.
72. Reid R, Prausnitz JM, Poling BE. The Properties of Gases and Liquids. New York:
McGraw-Hill, 1987.
73. Froment GF, Bischoff KB. Chemical Reactor Analysis and Design. New York:
Wiley, 1990.
74. Taylor R, Krishna R. Modeling reactive distillation. Chem Eng Sci 2000; 55:
5183–5229.
75. Sorel E. La rectification de l’alcool. Paris: Gauthiers—Villais et Fils, 1893.
76. Henley EJ, Seader JD. Equilibrium-Stage Separation Operations in Chemical
Engineering. New York: Wiley, 1981.
77. van Krevelen DW, Hoftijzer PJ. Kinetics of gas–liquid reactions—Part I: General
theory. Recl Trav Chim Pays-Bas 1948; 67:563–586.
78. Holma H, Sohlo J. A mathematical model of an absorption tower of nitrogen oxides
in nitric aid production. Computers Chem Eng 1979; 3:135–141.
79. Carta G. Scrubbing of nitrogen oxides with nitric acid solutions. Chem Eng
Commun 1986; 42:157–170.
80. Suzuki I, Yagi H, Komatsu H, Hirata M. Calculation of multicomponent distillation
accompanied by a chemical reaction. J Chem Eng Japan 1971; 4:26–33.
81. Chang YA, Seader JD. Simulation of continuous reactive distillation by homotopy
continuation method. Computers Chem Eng 1988; 12:1243–1255.
82. Stewart WE, Prober R. Matrix calculation of multicomponent mass transfer in
isothermal systems. Ind Eng Chem Fundam 1964; 3:224–235.
83. Toor HL. Solution of the linearized equations of multicomponent mass transfer.
AIChE J 1964; 10:448–455, 460–465.
84. Górak A. Simulation thermischer Trennverfahren fluider Vielkomponentengemische.
In: Schuler G, ed. Prozesssimulation. Weinheim, Germany: VCH, 1995:349–408.
85. Seader JD. The rate-based approach for modeling staged separations. Chem Eng
Prog 1989; 85(10):41–49.
86. Katti SS. Gas–liquid–solid systems: an industrial perspective. Trans IChemE 1995;
73(part A):595–607.
87. Lewis WK, Whitman WG. Principles of gas absorption. Ind Eng Chem 1924;
16:1215–1220.
88. Higbie R. The rate of absorption of a pure gas into a still liquid during short peri-
ods of exposure. Trans Am Inst Chem Engrs 1935; 31:365–383.
89. Hirschfelder JO, Curtiss CF, Bird RB. Molecular Theory of Gases and Liquids.
New York: Wiley, 1964.

Copyright © 2004 by Marcel Dekker, Inc. All Rights Reserved.


90. Pelkonen S, Kaesemann R, Górak A. Distillation lines for multicomponent separa-
tion in packed columns—theory and comparison with experiment. Ind Eng Chem
Res 1997; 36:5392–5398.
91. Sundmacher K, Hoffmann U. Development of a new catalytic distillation process for
fuel ethers via a detailed nonequilibrium model. Chem Eng Sci 1996; 51:2359–2368.
92. Davidson DL. The enterprise-wide application of computational fluid dynamics
in the chemicals industry. Proceedings of the 6th World Congress of Chemical
Engineering, Melbourne, 2001.
93. Sommerfeld M. Modellierung und numerische Berechnung von partikelbeladenen
turbulenten Strömungen mit Hilfe des Euler/Lagrange-Verfahrens. Aachen,
Germany: Shaker Verlag, 1996.
94. Scheuerer G. Solution algorithms and implementations strategies for Eulerian–
Eulerian multiphase-flow models. Proceedings of ACFD 2000 International
Conference on Applied Computational Fluid Dynamics, Beijing, 2000.
95. Kuipers JAM, van Swaaij WPM. Application of computational fluid dynamics to
chemical reactor engineering. Rev Chem Eng 1997; 13:1–118.
96. Sokolichin A, Eigenberger G, Lapin A, Lübbert A. Direct numerical simulation of
gas–liquid two-phase flows. Euler/Euler versus Euler/Lagrange. Chem Eng Sci
1997; 52:611–626.
97. Krishna R, van Baten JM. Simulating the motion of gas bubbles in a liquid. Nature
1999; 398:208.
98. Krishna R, Urseanu MI, van Baten JM, Ellenberger J. Wall effects on the rise
of single gas bubbles in liquids. Int Commun Heat Mass Transfer 1999; 26:
781–790.
99. Ding J, Gidaspow D. A bubbling fluidization model using the kinetic theory of
granular flow. AIChE J 1990; 36:523–538.
100. Fan LS, Zhu C. Principles of Gas–Solid Flows. Cambridge: Cambridge University
Press, 1998.
101. Marschall KJ, Mleczko L. CFD modeling of an internally circulating fluidized-bed
reactor. Chem Eng Sci 1999; 54:2085–2093.
102. van Wachem BGM, Schouten JC, Krishna R, van den Bleek CM. Validation of the
Eulerian simulated dynamic behavior of gas–solid fluidized beds. Chem Eng Sci
1999; 54:2141–2149.
103. Grienberger J, Hofmann H. Investigations and modelling of bubble columns. Chem
Eng Sci 1992; 47:2215–2220.
104. Lapin A, Lübbert A. Numerical simulation of the dynamics of two-phase gas–
liquid flows in bubble columns. Chem Eng Sci 1994; 49:3661–3674.
105. Sokolichin A, Eigenberger G. Gas–liquid flow in bubble columns and loop reactors.
Part I. Detailed modelling and numerical simulation. Chem Eng Sci 1994; 49:
5735–5746.
106. Grevskott S, Sannaes BH, Dudukovic MP, Hjarbo KW, Svendsen HF. Liquid circu-
lation, bubble-size distributions, and solids movement in two- and three-phase
columns. Chem Eng Sci 1996; 51:1703–1713.
107. Lin TJ, Reese J, Hong T, Fan LS. Quantitative analysis and computation of two-
dimensional bubble columns. AIChE J 1996; 42:301–318.

Copyright © 2004 by Marcel Dekker, Inc. All Rights Reserved.


108. Delnoij E, Lammers FS, Kuipers JAM, van Swaaij WPM. Dynamic simulation of
dispersed gas–liquid two-phase flow using a discrete bubble model. Chem Eng Sci
1997; 52:1429–1458.
109. Jakobsen HA, Sannaes BH, Grevskott S, Svendsen HF. Modeling of bubble-driven
vertical flows. Ind Eng Chem Res 1997; 36:4052–4074.
110. Sanyal J, Vasquez S, Roy S, Dudukovic MP. Numerical simulation of gas–liquid
dynamics in cylindrical bubble-column reactors. Chem Eng Sci 1999; 54: 5071–5083.
111. Thakre SS, Joshi JB. CFD simulation of bubble-column reactors: importance of
drag-force formulation. Chem Eng Sci 1999; 54:5055–5060.
112. Krishna R, Urseanu MI, van Baten JM, Ellenberger J. Influence of scale on the
hydrodynamics of bubble columns operating in the churn-turbulent regime: exper-
iments vs. Eulerian simulations. Chem Eng Sci 1999; 54:4903–4911.
113. Pfleger D, Becker S. Modelling and simulation of the dynamic flow behavior in a
bubble column. Chem Eng Sci 2001; 56:1737–1747.
114. Fischer CH, Quarini GL. Three-dimensional heterogeneous modeling of distillation
tray hydraulics. Proceedings of AIChE Annual Meeting, Miami Beach, 1998.
115. Metha B, Chuang KT, Nandakumar K. Model for liquid-phase flow on sieve trays.
Trans IChemE 1998; 76:843–848.
116. Krishna R, van Baten JM, Ellenberger J, Higler AP, Taylor R. CFD simulations of
sieve-tray hydrodynamics. Trans IChemE 1999; 77:639–646.
117. Lju CJ, Yuan XG, Yu KT, Zhu XJ. A fluid-dynamic model for flow pattern on a dis-
tillation tray. Chem Eng Sci 55:2287–2294, 2000.
118. van Baten JM, Krishna R. Modelling sieve-tray hydraulics using computational
fluid dynamics. Chem Eng J 2000; 77:143–151.
119. van Baten JM, Ellenberger J, Krishna R. Hydrodynamics of reactive distillation tray
column with catalyst-containing envelopes: experiments vs. CFD simulations.
Catal Today 2001; 66:233–240.
120. van Gulijk C. Using computational fluid dynamics to calculate transversal disper-
sion in a structured packed bed. Computers Chem Eng 1998; 22:767–770.
121. Higler AP, Krishna R, Ellenberger J, Taylor R. Countercurrent operation of a struc-
tured catalytically packed-bed reactor: liquid-phase mixing and mass transfer.
Chem Eng Sci 1999; 54:5145–5152.
122. van Baten JM, Ellenberger J, Krishna R. Radial and axial dispersion of the liquid
phase within KATAPAK-S structure: experiments vs. CFD simulations. Chem Eng
Sci 2001; 56:813–821.
123. Rieger R, Weiss C, Wigley G, Bart HJ, Marr R. Investigating the process of liquid–
liquid extraction by means of CFD. Computers Chem Eng 1996; 20:1467–1475.
124. Zahiarieva A, Masberat O, Aoun Nabli M, Giraud P, Gourdon C. Experimental and
numerical 2D turbulent field in a pulsed extraction column. In: Shallcross DC
et al., eds. Value Adding Through Solvent Extraction. Vol. 2. Melbourne: University
of Melbourne, 1996:1233–1238.
125. Fei WY, Wang YD, Wan YK. Physical modelling and numerical simulation of
velocity fields in rotating-disc contactor via CFD simulation and LDV measure-
ment. Chem Eng J 2000; 78:131–139.
126. Modes G, Bart HJ. CFD simulations of nonideal dispersed-phase flow in stirred
extraction columns. Chem Eng Technol 2002; 24:1242–1245.

Copyright © 2004 by Marcel Dekker, Inc. All Rights Reserved.


127. Bart HJ. From single droplet to column design. Keynote lecture at 6th World
Congress on Chemical Engineering, Melbourne, 2001.
128. Suchak NJ, Jethani KR, Joshi JB. Modeling and simulation of NOx absorption in
pilot-scale packed columns. AIChE J 1991; 37:323–339.
129. Wiesner U, Wittig M, Górak A. Design and optimization of a nitric acid recovery
plant from nitrous waste gases. Computers Chem Eng 1996; 20:S1425–S1430.
130. Ulrichs K, Laue W, Renovanz HD. Salpetersäure und salpetrige Säure, Stickst-
offoxide, Nitrate, Nitrite. In: Ullmanns Encyclopädie der technischen Chemie.
Vol. 20. Weinheim, Germany: VCH, 1981:305–362.
131. Wiesner U, Górak A, Kenig EY, Steude H, Börger GG. Modelling of absorption of
nitrous waste gases. IChemE Symp Ser 1997; 142(1):323–333.
132. Kafarov VV, Shestopalov EA, Novikov EA, Belkov VP. Mathematical model of
absorption of NOx in production of weak nitric acid. Soviet Chem Ind 1975; 7:
1224–1226.
133. Kolev N. Wirkungsweise von Füllkörperschüttungen. Chem Ing Tech 1976; 48:
1105–1112.
134. Brauer H. Stoffaustausch einschliesslich chemischer Reaktionen. Aarau: Sauerländer,
1971.
135. Kenig EY, Wiesner U, Górak A. Modeling of reactive absorption using the
Maxwell–Stefan equations. Ind Eng Chem Res 1997; 36:4425–4434.
136. Danckwerts PV, Sharma MM. The absorption of carbon dioxide into solutions of
alkalis and amines (with some notes on hydrogen sulfide and carbonyl sulfide). The
Chemical Engineer 1966; (July–Aug):244–280.
137. Maurer G. On the solubility of volatile weak electrolytes in aqueous solutions. ACS
Symp Ser 1980; 133:139–172.
138. Tippmer K. Progress in the desulphurization of coke oven gas and utilization of
H2S-Still Otto’s SOLOX process. Coke Making Int 1994; 6:32–43.
139. Schneider R, Kenig EY, Górak A. Dynamic modelling of reactive absorption with
the Maxwell–Stefan approach. Trans IchemE 1999; 77(part A):633–638.
140. Kenig EY, Schneider R, Górak A. Rigorous dynamic modelling of complex reactive
absorption processes. Chem Eng Sci 1999; 54:5195–5203.
141. Schneider R, Górak A. Model optimization for the dynamic simulation of reactive
absorption processes. Chem Eng Technol 2001; 24:979–989.
142. Schneider R, Kenig EY, Górak A. Complex reactive absorption processes: model
optimization and dynamic column simulation. Proceedings of ESCAPE 11 European
Symposium on Computer-Aided Process Engineering, Kolding, Denmark, 2001.
143. Kenig EY, Schneider R, Górak A. Reactive absorption: optimal process design via
optimal modelling. Chem Eng Sci 2001; 56:343–350.
144. Pantelides CC. The consistent initialization of differential-algebraic systems. SIAM
J Sci Stat Comp 1988; 9:213–231.
145. Schneider R, Kenig EY, Górak A. Dynamische Simulation reaktiver Absorption-
sprozesse am Beispiel einer Sauergaswäsche: Modellentwicklung, -analyse und
-optimierung. Chem Ing Tech 2000; 72:1224–1229.
146. Song W, Venimadhavan G, Manning JM, Malone MF, Doherty MF. Measurement
of residue curve maps and heterogeneous kinetics in methyl acetate synthesis. Ind
Eng Chem Res 1998; 37:1917–1928.

Copyright © 2004 by Marcel Dekker, Inc. All Rights Reserved.


147. Górak A, Kreul LU, Skowronski M. Strukturierte Mehrzweckpackung. German
patent 19701045 A1, 1998.
148. Schneider R, Noeres C, Kreul LU, Górak A. Dynamic modeling and simulation of
reactive batch distillation. Computers Chem Eng 1999; 23:S423–S426.
149. Schneider R, Noeres C, Kreul LU, Górak A. Dynamic modeling and simulation of
reactive batch distillation. Computers Chem Eng 2001; 25:169–176.
150. Krafczyk J, Gmehling J. Einsatz von Katalysatorpackungen für die Herstellung von
Methylacetat durch reaktive Rektifikation. Chem Ing Tech 1994; 66:1372–1375.
151. Hoffmann A, Górak A. Scale-up of reactive distillation columns with catalytic
packing. Chem Eng Process (submitted).
152. Gicquel A, Torck B. Synthesis of methyl tertiary butyl ether catalyzed by ion-
exchange resin. Influence of methanol concentration and temperature. J Catal 1983;
83:9–18.
153. Rehfinger A, Hoffmann U. Kinetics of methyl tertiary butyl ether liquid-phase syn-
thesis catalyzed by ion exchange resin. I. Intrinsic rate expression in liquid-phase
activities. Chem Eng Sci 1990; 45:1605–1617.
154. Parra D, Tejero J, Cunill F, Iborra M, Izquierdo JF. Kinetic study of MTBE liquid-
phase synthesis using C4 olefin cut. Chem Eng Sci 1994; 49:4563–4578.
155. Kenig EY, Jakobsson K, Banik P, Aittamaa J, Górak A, Koskinen M, Wettmann P.
An integrated tool for synthesis and design of reactive distillation. Chem Eng Sci
1999; 54:1347–1352.
156. Kolarik Z. Critical evaluation of some equilibrium constants involving acidic
organophosphorous extractants. Pure Appl Chem 1982; 54:2593–2674.
157. Kolarik Z, Grimm R. Acidic organophosphorous extractants XXIV: the polymer-
ization behavior of Cu(II) Cd(II), Zn(II) and Co(II) complexes of D2EHPA in fully
loaded organic phases. J Inorg Nucl Chem 1976; 38:1721–1727.
158. Mörters M, Bart HJ. Extraction equilibria of zinc with bis(2-ethyl-hexyl) phos-
phoric acid. J Chem Eng Data 2000; 45:82–85.
159. Bart HJ. Reactive extraction of acids and metals—the state of the art in column
design. Chem Eng Sci 2002; 57:1633–1637.
160. Schäfer JP, Sluyts D. Aufarbeitung eines Prozessabwassers mittels Reaktivextraktion.
In: VDI-Jahrbuch. Düsseldorf: VDI, 1996:285–295.
161. Veglio F, Slater MJ. Design of liquid–liquid extraction columns for the possible test
system Zn/D2EHPA in n-dodecane. Hydrometallurgy 1996; 42:177–195.
162. Mansur MB, Slater MJ, Biscaia EC. Kinetic analysis of the reactive liquid–liquid
test system ZnSO4/D2EHPA/n-heptane. Hydrometallurgy 2002; 63:107–116.
163. Mansur MB, Slater MJ, Biscaia EC. Equilibrium analysis of the reactive liquid–
liquid test system ZnSO4/D2EHPA/n-heptane. Hydrometallurgy 2002; 63:117–126.
164. Steiner L. Mass-transfer rates from single drops and drop swarms. Chem Eng Sci
1986; 41:1979–1986.
165. Slater M. A combined model of mass transfer coefficient for contaminated drop
liquid– liquid systems. Can J Chem Eng 1995; 73:462–469.
166. Brodkorb M, Slater MJ. Multicomponent and contamination effects on mass trans-
fer in a liquid–liquid extraction rotating disk contactor. Trans IChemE 2001;
79(part A):335–346.

Copyright © 2004 by Marcel Dekker, Inc. All Rights Reserved.


167. Kenig EY, Butzmann F, Kucka L, Górak A. Comparison of numerical and analyti-
cal solutions of a multicomponent reaction–mass transfer problem in terms of the
film model. Chem Eng Sci 2000; 55:1483–1496.
168. Noeres C, Benvenuti C, Hoffmann A, Górak A. Reactive distillation: nonideal-flow
behavior of the liquid phase in structured catalytic packings. Proceedings of Inter-
national Symposium on Multifunctional Reactors (ISMR-2), Nuremberg, 2001.
169. Higler AP, Krishna R, Taylor R. Nonequilibrium-cell model for multicomponent
(reactive) separation processes. AIChE J 1999; 45:2357–2370.
170. Danckwerts PV. Continuous flow systems—distribution of residence times. Chem
Eng Sci 1953; 2:1–13.
171. van Swaaij WPM, Harpentier JC, Villermaux J. Residence time distribution in the
liquid phase of trickle flow in packed columns. Chem Eng Sci 24:1083–1095, 1969.
172. Noeres C, Górak A. Zur dynamischen Modellierung der Reaktivrektifikation unter
Berücksichtigung des Verweilzeitverhaltens katalytischer Packungen. Proceedings
of GVC—Fachausschuss “Thermische Zerlegung,” Bingen, Germany, 2002.
173. Rod V. Calculating mass transfer with longitudinal mixing. Br Chem Eng 1966;
6:483–488.
174. Casamatta G. Compartment de la population des gouttes dans une colonne d’ex-
traction: transport, rupture, coalescence, transfer de matiere. Ph.D. dissertation,
Institute National Polytechnique de Toulouse, Toulouse, France, 1981.
175. Kronberger T, Ortner A, Zulehner W, Bart HJ. Numerical simulation of extraction
columns using a drop population model. Computers Chem Eng 1995; 19:639–644.
176. Ortner A, Kronberger T, Zulehner W, Bart HJ. Tropfen-Populanz-Bilanzmodell am
Beispiel einer gerührten Extraktionskolonne. Chem Ing Tech 1995; 67:984–988.
177. Gerstlauer A. Herleitung und Reduktion populationsdynamischer Modelle am
Beispiel der Flüssig–Flüssig Extraktion. Düsseldorf: VDI, 1999.
178. Modes G. Grundsätzliche Studie zur Populationsdynamik einer Extraktionskolonne
auf Basis von Einzeltropfenuntersuchungen. Aachen, Germany: Shaker, 2000.
179. Attarakih M, Bart HJ, Faqir NM. An approximate optimal moving-grid technique
for the solution of discretized population balances in batch systems. Proceedings of
ESCAPE 12 European Symposium on Computer-Aided Process Engineering, The
Hague, 2002.
180. Simon M, Bart HJ. Experimental studies of coalescence in liquid/liquid systems.
Chem Eng Technol 2002; 25:481–484.
181. Wei J, Prater CD. The structure and analysis of complex reaction systems. Adv
Catal 1962; 13:203–392.
182. Hikita H, Asai S. Gas absorption with (m, n)-th order irreversible chemical reac-
tions. Int Chem Eng 1964; 4:332–340.
183. Toor HL. Dual diffusion-reaction coupling in first order multicomponent systems.
Chem Eng Sci 1965; 20:941–951.
184. DeLancey GB. Multicomponent film-penetration theory with linearized kinetics—
I. Linearization theory and flux expression. Chem Eng Sci 1974; 29:2315–2323.
185. Astarita G, Sandler SI. Kinetic and Thermodynamic Lumping of Multicomponent
Mixtures. Amsterdam: Elsevier, 1991.
186. Kenig EY, Kholpanov LP. Analysis of formulation and solution of multicomponent
reaction-diffusion problems. Theor Found Chem Eng 1992; 26:510–521.

Copyright © 2004 by Marcel Dekker, Inc. All Rights Reserved.


187. Onda K, Takeuchi H, Okumoto Y. Mass transfer coefficients between gas and liq-
uid phases in packed columns. J Chem Eng Japan 1968; 1:56–62.
188. Fuller EN, Schettler PD, Giddings JC. A new method for prediction of binary gas-
phase diffusion coefficients. Ind Eng Chem 1966; 58:19–27.
189. Siddiqi MA, Lucas K. Correlations for prediction of diffusion in liquids. Can J
Chem Eng 1986; 64:839–843.
190. Mika V. Model of packed absorption column. I. Physical absorption. Coll Czech
Chem Commun 1967; 32:2933–2943.
191. Schwartz SE, White WH. Solubility equilibria of the nitrogen oxides and oxyacids in
dilute aqueous solutions. In: Pfafflin JR, Zeigler EN, eds. Advances in Environmental
Science and Engineering. Vol. 4. New York: Gordon and Breach, 1981:1–45.
192. Engels H. Anwendung des Modells der lokalen Zusammensetzung auf Elektro-
lytlösungen. Ph.D. dissertation, RWTH Aachen, Aachen, Germany, 1985.
193. Horvath AL. Handbook of Aqueous Electrolyte Solutions. Chichester, UK: Ellis
Horwood, 1985.
194. Rocha JA, Bravo JL, Fair JR. Distillation columns containing structured packings—
2. Mass transfer model. Ind Eng Chem Res 1996; 35:1660–1667.
195. Ranzi E, Rovaglio M, Faravelli T, Biardi G. Role of energy balances in dynamic
simulation of multicomponent distillation columns. Computers Chem Eng 1988;
12:783–786.
196. Ronge G. Überprüfung unterschiedlicher Modelle für den Stoffaustausch bei der
Rektifikation in Packungskolonnen. Düsseldorf: VDI, 1995.
197. Xu ZP, Chuang KT. Kinetics of acetic acid esterification over ion exchange cata-
lysts. Can J Chem Eng 1996; 74:493–500.
198. Powers MF, Vickery DJ, Arehole A, Taylor R. A nonequilibrium-stage model of
multicomponent separation processes—V. Computational methods for solving the
model equations. Computers Chem Eng 1988; 12:1229–1241.
199. Ellenberger J, Krishna R. Countercurrent operation of structured catalytically
packed distillation columns: pressure drop, holdup, and mixing. Chem Eng Sci
1999; 54:1339–1345.
200. Moritz P, Hasse H. Fluid dynamics in reactive distillation packing Katapak-S.
Chem Eng Sci 1999; 54:1367–1374.
201. Stichlmair JG, Fair JR. Distillation: Principles and Practice. New York: Wiley-
VCH, 1998.
202. Sundmacher K, Hoffmann U. Macrokinetic analysis of MTBE synthesis in chemi-
cal potentials. Chem Eng Sci 1994; 49:3077–3089.
203. Mörters M, Bart HJ. Fluorescence-indicated mass transfer in reactive extraction.
Chem Eng Technol 2000; 23:1–7.
204. Mack C. Untersuchungen zum Stofftransport und Stoffaustauschverhalten bei der
Extraktion von Chrom (III) und Zink mittels D2EHPA. Ph.D. dissertation,
Technical University of Darmstadt, Darmstadt, Germany, 2000.
205. Marangoni CGM. Über die Ausbreitung der Tropfen einer Flüssigkeit auf der
Oberfläche einer anderen. Ann Phys Chem (Poggendorf) 1871; 143:337–354.
206. Thomson W (Lord Kelvin). The influence of wind on waves in water supposed fric-
tionless. Phil Mag 1871; 10:330–333.

Copyright © 2004 by Marcel Dekker, Inc. All Rights Reserved.


207. Pitzer KS. Theory: ion interaction approach. In: Pytkowicz RM, ed. Activity
Coefficients in Electrolyte Solutions. Vol. 1. Boca Raton, FL: CRC Press, 1979:
157–208.
208. Hildebrand JM, Scott RL. The Solubility of Nonelectrolytes. New York: Reinhold,
1950.
209. Henschke M, Pfennig A. Mass transfer enhancement in single-drop extraction
experiments. AIChE J 1999; 45:2079–2086.
210. Handlos AE, Baron T. Mass and heat transfer from drops in liquid–liquid extrac-
tion. AIChE J 1957; 3:127–136.

APPENDIX A. A DETAILED DESCRIPTION OF


RATE-BASED MODELING
A.1. Balance Equations
The mass balance equations of the traditional multicomponent rate-based model
(see, e.g., Refs. 57 and 58) are written separately for each phase. In order to give
a common description to all three considered RSPs (where it is possible, of
course) we will use the notion of two contacting fluid phases. The first one is
always the liquid phase, whereas the second fluid phase represents the gas phase
for RA, the vapor phase for RD and the liquid phase for RE. Considering homo-
geneous chemical reactions taking place in the fluid phases, the steady-state bal-
ance equations should include the reaction source terms:

d
0  ( LxiB )  ( N LiB a I  RLiB  L ) As i  1, . . . , n (A1)
dl

d
0 (GyiB )  ( NGiB a I  RGiB G ) As i  1, . . . , n (A2)
dl

If chemical reactions take place in the (first) liquid phase only (this is
valid for most of RD processes), the phase balances for the second fluid phase
simplify to

d
0 (GyiB )  NGiB a I As i  1, . . . , n (A3)
dl

The bulk-phase balances are completed by the summation equation for the
liquid and second fluid bulk mole fractions:
n

∑x
i1
i
B
1
(A4)

Copyright © 2004 by Marcel Dekker, Inc. All Rights Reserved.


n

∑ y 1
i1
i
B
(A5)

The volumetric liquid holdup, L, depends on the gas/vapor and liquid
flows and is calculated via empirical correlations (e.g., Ref. 65). For the determi-
nation of axial temperature profiles, differential energy balances are formulated,
including the product of the liquid molar holdup and the specific enthalpy as
energy capacity. The energy balances written for continuous systems are as
follows:
d
0  ( LhLB )  (QLB a I  RLB L H RL
0
) As (A6)
dl

d
0 (GhGB )  (QGB a I  RGBG H RG
0
) As (A7)
dl
In the dynamic rate-based stage model, molar holdup terms have to be con-
sidered in the mass balance equations, whereas the changes in both the specific
molar component holdup and the total molar holdup are taken into account. For
the liquid phase, these equations are as follows:

∂ ∂
U Li  ( LxiB )  ( N LiB a I  RLiB  L ) As i  1, . . . , n (A8)
∂t ∂l

U Li  xiBU Lt  xiB ( L c Lt As ) i  1, . . . , n (A9)

The gas/vapor holdup can often be neglected due to the low gas-phase
density, and the component balance equation reduces to Eq. (A2) (see also
Ref. 139).

A.2. Mass Transfer and Reaction Coupling in Fluid Films


The component fluxes NiB entering into Eqs. (A1)–(A3) are determined based on
the mass transport in the film region. Because the key assumptions of the film
model result in the one-dimensional mass transport normal to the interface, the
differential component balance equations including simultaneous mass transfer
and reaction in the film are as follows:

dNLi
 RLi  0 i  1, . . . , n (A10)
dz
Equations (A10), which are generally valid for both liquid and second fluid
phases, represent nothing but differential mass balances for the film region, with the

Copyright © 2004 by Marcel Dekker, Inc. All Rights Reserved.


account of the source term due to the reaction. To link these balances to process
variables like component concentrations, some additional relationships, often called
constitutive relations (see Ref. 57), are necessary. For the component fluxes Ni,
these constitutive relations result from the multicomponent diffusion description
[Eqs. (1) and (2)] and, for the source terms, from the reaction kinetics description.
The latter strongly depends on the specific reaction mechanism, the stoi-
chiometry, and the presence or absence of parallel reaction schemes (69). The rate
expressions for Ri usually represent nonlinear dependences on the mixture com-
position and temperature. Specifically for the coupled reaction–mass transfer
problems, such as Eqs. (A10), it is always essential as to whether or not the reac-
tion rate is comparable to that of diffusion (68,77). Equations (A10) should be
completed by the boundary conditions relevant to the film model. These conditions
specify the values of the mixture composition at both film boundaries. For exam-
ple, for the liquid phase:
xi ( z  0)  xiI , xi ( z  L )  xiB i  1, . . . , n (A11)
Combining Eqs. (A10) with the boundary conditions (A11) written in vec-
tor form and using constitutive relations such as Eqs. (1) and (2), we obtain a
vector-type boundary-value problem, which permits the component concentration
profiles to be obtained as functions of the film coordinate. These concentration
profiles, in turn, allow one to determine the component fluxes. Thus the boundary-
value problem describing the film phenomena has to be solved in conjunction
with all other model equations.
The composition boundary values entering into Eqs. (A11) represent exter-
nal values for Eqs. (A10). With some further assumptions concerning the diffu-
sion and reaction terms, this allows an analytical solution of the boundary-value
problem [Eqs. (A10) and (A11)] in a closed matrix form (see Refs. 58 and 135).
On the other hand, the boundary values need to be determined from the total sys-
tem of equations describing the process. The bulk values in both phases are found
from the balance relations, Eqs. (A1) and (A2). The interfacial liquid-phase con-
centrations xiI are related to the relevant concentrations of the second fluid phase,
yiI , by the thermodynamic equilibrium relationships and by the continuity condi-
tion for the molar fluxes at the interface (57,135).
Due to the chemical conversion in the liquid film, the molar fluxes at the
interface and at the boundary between the film and the bulk of the phase differ.
The system of equations is completed by the conservation equations for the mass
and energy fluxes at the phase interface and the necessary linking conditions
between the bulk and film phases (see Refs. 57, 59, and 84).
Generally all these considerations are also valid for the second fluid film
phase, provided that reactions occur there (135). Both analytical and numerical
solutions of the coupled diffusion-reaction film problem are analyzed at full
length in Ref. 167; their particular applications are considered in Section 3.

Copyright © 2004 by Marcel Dekker, Inc. All Rights Reserved.


A.3. Nonideal Flow Behavior in Catalytic Column Internals
The mass balances [Eqs. (A1) and (A2)] assume plug-flow behavior for both the
gas/vapor and liquid phases. However, real flow behavior is much more complex
and constitutes a fundamental issue in multiphase reactor design. It has a strong
influence on the reactor performance, for example, due to back-mixing of both
phases, which is responsible for significant effects on the reaction rates and prod-
uct selectivity. Possible development of stagnant zones results in secondary un-
desired reactions. To ensure an optimum model development for CD processes,
experimental studies on the nonideal flow behavior in the catalytic packing
MULTIPAK® are performed (168).
The experimental results confirm that the fluid flow in MULTIPAK® devi-
ates from plug-flow behavior (Figure 31). Calculated axial dispersion coefficients
are about 104–102 m2/s, which are several orders of magnitude larger than that
for molecular diffusion (Figure 32). Therefore, in the investigated operating
range, nonideal mixing effects are caused by hydrodynamic rather than molecu-
lar diffusion effects. Calculated Bodenstein numbers are one order of magnitude

FIGURE 31 Comparison between the experimental RTD curve for the cata-
lytic packing MULTIPAK® (dC  0.1 m), the ADM model, and the PDE model.

Copyright © 2004 by Marcel Dekker, Inc. All Rights Reserved.


FIGURE 32 Axial dispersion coefficients for the catalytic packing MULTIPAK®
(dC  0.1 m ) , calculated based on the ADM model.

smaller than those for fixed-bed reactors, which may be caused by two effects: the
occurrence of stagnant zones in the catalyst layer, and liquid bypassing due to the
hybrid structure of the catalytic packing (168).
The rate-based models suggested up to now do not take liquid back-mixing
into consideration. The only exception is the nonequilibrium-cell model for
multicomponent reactive distillation in tray columns presented in Ref. 169. In this
work a single distillation tray is treated by a series of cells along the vapor and
liquid flow paths, whereas each cell is described by the two-film model (see
Section 2.3). Using different numbers of cells in both flow paths allows one to
describe various flow patterns. However, a consistent experimental determination
of necessary model parameters (e.g., cell film thickness) appears difficult, where-
as the complex iterative character of the calculation procedure in the dynamic
case limits the applicability of the nonequilibrium cell model.
A far more promising approach is represented by the so-called differential
models, such as the axial dispersion model (ADM) (170) as well as the piston-
flow model with axial dispersion and mass exchange (PDE) (171). Experimental
studies (168) show that the ADM gives an appropriate description of the nonideal
flow behavior of the liquid phase in catalytic packings (see Figure 31). Considering

Copyright © 2004 by Marcel Dekker, Inc. All Rights Reserved.


the nonideal flow behavior via the ADM, the dynamic mass balances [cf. Eqs.
(A8)] take the following form:

∂ D ∂2 ∂
ULi  ax 2 ( LxiB )  ( LxiB )  ( N LiB a I  RLiB  L ) As i  1, . . . , n (A12)
∂t u L ∂l ∂l

A thorough investigation of the influence of the flow nonideality in catalytic


packings on the dynamic process behavior of specific CD processes is an objec-
tive of some current studies (172).
Equation (A12) is widely used in RE, but it does not account for the spe-
cific interactions of the dispersed phase. In this respect current research is focused
on drop population balance models, which account for the different rising velo-
cities of the different-size droplets and their interactions, such as droplet breakup
and coalescence (173–180).

APPENDIX B. MODELING PECULIARITIES AND


MODEL PARAMETERS FOR THE CASE STUDIES
B.1. Absorption of NOx
In terms of the concentration vector, Eq. (A10) is a nonlinear differential equa-
tion of the second order. The boundary-value problem [Eqs. (A10) and (A11)] is
usually solved numerically. However, it is also possible to linearize the reaction
term using the method suggested in Ref. 181:
R ≅ [ K ] x (B1)
Equation (B1) provides a satisfactory representation for many processes over the
entire reaction range and is a good linear approximation for most systems in a suf-
ficiently small range (see, e.g., Refs. 68 and 182–184). Equation (B1) has gained
widespread acceptance in various chemical and reactor engineering areas (185) and
is recommended for use in the modeling of reactive separation operations (59,184).
The approximation of Eq. (B1) allows one to reduce Eqs. (A10) and (A11)
to a linearized boundary-value problem (183,184,186). The latter can then be
solved analytically and yields a compact matrix-form solution for the concentra-
tion profiles in the film region [58]. Such a solution gives simple analytical
expressions for the component fluxes with regard to the homogeneous reaction in
the fluid films (see Ref. 135), which can be of particular value when large indus-
trial reactive separation units are considered and designed.
The methods of determination of the reaction matrix [K] are considered in
Refs. 167, 181, 183, 184 and 186. Another important matrix parameter entering
into the linearized film mass transport equation is the multicomponent diffusion
matrix [D]. The latter results from the transformation of the Maxwell–Stefan Eqs. (1)
to the form of the generalized Fick’s law (83). Matrix [D] is generally a function of

Copyright © 2004 by Marcel Dekker, Inc. All Rights Reserved.


the mixture composition and it is assumed constant along the diffusion path (83).
The direct expressions for the elements of the diffusion matrix [D] can be found,
for example, in Ref. 57.
The linearization of the initial film mass transport equation and its analytical
solution were applied to simulate the industrial NOx absorption process considered.
In order to calculate the multicomponent diffusion matrices [D], the binary
diffusivities in both phases should be known. The film thickness representing an
important model parameter is estimated via the mass transfer coefficients (57,83).
The binary diffusivities and mass transfer coefficients were calculated from the
correlations summarized in Table 3.
The correlations of Billet (66) and Onda et al. (187) are valid for various
mixtures and packings and cover both absorption and distillation processes. The
correlation of Kolev (133) is obtained for RA and certain random packings. In
general, the mass transfer coefficient correlations need to be compared to one
another and validated using experimental data. This shows, in particular, the way
the mass transfer correlations influence the concentration profiles of the compo-
nents and other relevant process characteristics.
Nitric acid is a strong electrolyte. Therefore, the solubilities of nitrogen
oxides in water given in Ref. 191 and based on Henry’s law are utilized and fur-
ther corrected by using the method of van Krevelen and Hoftijzer (77) for elec-
trolyte solutions. The chemical equilibrium is calculated in terms of liquid-phase
activities. The local composition model of Engels (192), based on the UNIQUAC
model, is used for the calculation of vapor pressures and activity coefficients of
water and nitric acid. Multicomponent diffusion coefficients in the liquid phase
are corrected for the nonideality, as suggested in Ref. 57.

TABLE 3 Binary Diffusion Coefficients and Mass Transfer


Coefficients

Mass transfer
Binary diffusion coefficient
Phase coefficient correlation

Gas Ref. 188 Ref. 187


Ref. 66
Wehmeier (see
Ref. 134)
Liquid Ref. 189 Ref. 187
Ref. 66
Ref. 133
Ref. 190

Copyright © 2004 by Marcel Dekker, Inc. All Rights Reserved.


B.2. Coke Gas Purification
In Ref. 139 a purely numerical approach to the solution of the considered com-
plex RA problem was suggested. The liquid film is treated as an additional bal-
ance region, in which reaction and mass transfer occur simultaneously. Therefore,
the reactions are considered both in the liquid-bulk-phase mass balances, Eq. (A1),
and in the differential balances for the liquid film, Eq. (A10).
To be able to describe the presence of electrolytes in the system, the elec-
trical driving force also needs to be taken into account (57). Therefore, the gradi-
ent of the electrical potential  is introduced into the generalized driving force
di [cf. Eq. (2)]:

xi 1 ∂i F 1 d
di   x i zi i  1, . . . , n (B2)
ℜT L ∂ ℜT L d

In dilute electrolyte systems, the diffusional interactions can usually be


neglected, and the generalized Maxwell–Stefan equations are reduced to the
Nernst–Planck equations (B3):

c Lt DLi, eff  dxi F d 


N Li    x i zi   xi N Ln i  1, . . . , n  1 (B3)
L  d ℜT d 

where n is the solvent index. The consideration of the electrical potential requires
an additional condition, the electroneutrality, which has to be met in each point
of the liquid phase:
n

∑x z 0
i1
i i (B4)

Thermodynamic nonidealities are considered both in the transport equations


(A10) and in the equilibrium relationships at the phase interface. Because elec-
trolytes are present in the system, the liquid-phase diffusion coefficients should be
corrected to account for the specific transport properties of electrolyte solutions.
The thermodynamic equilibrium at the gas–liquid interface is described as
follows:
yiI  Ki xiI i 1, . . . , n (B5)
where the distribution coefficient Ki comprises fugacities in both phases and
activity coefficients in the liquid phase. For the system considered, the values of
Ki, Eq. (B5), are determined from the electrolyte NRTL method (70,71).
The liquid-phase diffusion coefficients are found with the Nernst–Hartley
equation (193), which describes the transport properties in weak electrolyte
systems. The gas-phase diffusion coefficients are estimated according to the

Copyright © 2004 by Marcel Dekker, Inc. All Rights Reserved.


Chapman–Enskog–Wilke–Lee model (72). The correlations for the mass transfer
coefficients are taken from Ref. 194.

B.3. Methyl Acetate System, Batch Distillation


The rate-based models usually use the two-film theory and comprise the materi-
al and energy balances of a differential element of the two-phase volume in the
packing (148). The classical two-film model shown in Figure 13 is extended here
to consider the catalyst phase (Figure 33). A pseudo-homogeneous approach is
chosen for the catalyzed reaction (see also Section 2.1), and the corresponding
overall reaction kinetics is determined by fixed-bed experiments (34). This
macroscopic kinetics includes the influence of the liquid distribution and mass
transfer resistances at the liquid–solid interface as well as diffusional transport
phenomena inside the porous catalyst.
For the determination of conversion corresponding to the average residence
time, the reaction kinetics is integrated into the mass balances, and the liquid

FIGURE 33 Film model for a differential packing segment with heteroge-


neous catalyst.

Copyright © 2004 by Marcel Dekker, Inc. All Rights Reserved.


holdup, as the accumulation term, is accounted for simultaneously, as in Eqs. (A8)
and (A9). Because of low vapor-phase density, the vapor holdup is neglected, and
the vapor-phase-component balance equation reduces to Eq. (A2).
Ranzi et al. [195] found that the full energy balances, including the accu-
mulation term, have to be considered in order to predict correct dynamic process
behavior. Therefore, the differential dynamic energy balance for the liquid phase
is applied as follows:
∂ ∂
EL  ( LhLB )  (QLB a I  RLB L H RL
0
) As
∂t ∂l
2 (B6)

∑N h

Q  L (TLB  T I ) 
B
i  1, . . . , n
L
L Li Li
i1

where

EL  hLB ( L c Lt As ) (B7)
Similar to the mass balance equation, the vapor-phase energy balance simplifies
to Eq. (A7).
Experimental studies were carried out to derive correlations for mass trans-
fer coefficients, reaction kinetics, liquid holdup, and pressure drop for the pack-
ing MULTIPAK® (35). Suitable correlations for ROMBOPAK 6M® are taken from
Refs. 90 and 196. The nonideal thermodynamic behavior of the investigated mul-
ticomponent system was described by the NRTL model for activity coefficients
concerning nonidealities caused by the dimerisation (see Ref. 72).
Binary diffusion coefficients for the vapor phase and for the liquid phase
were estimated via the method proposed by Fuller et al. and Tyn and Calus,
respectively (see Ref. 72). Physical properties such as densities, viscosities, and
thermal conductivities were calculated from the methods given in Ref. 72. Heat
losses through the column wall were measured at pilot scale.

B.4. Methyl Acetate System, Steady-State Distillation


The model is based on the film theory and comprises the material and energy bal-
ances of a differential element of the two-phase volume in the packing. Each element
consists of an ideally mixed vapor and liquid bulk phase and a vapor film region
adjacent to the interface, as shown in Figure 33. A first guess of the bulk phase
compositions and temperatures was provided by the solution of an equilibrium-
stage model without reactions, as suggested in Ref. 198. The catalyzed reaction
is described by the quasi-homogeneous approach of Ref. 197, since the concen-
tration of acid sites has been determined as aCat  4.7 molH/gCat for dry Lewatit
K2621, which is close to the data of Ref. 197 given for Amberlyst 15.

Copyright © 2004 by Marcel Dekker, Inc. All Rights Reserved.


In order to determine the model parameters, several experiments were per-
formed at laboratory scale. Pressure drop experiments were carried out in glass
columns, with a total packing height of 1 m at ambient pressure. Air/water was used
as a test system, with a circulating liquid phase set at a constant temperature of 20C.
The experimental data cover a wide range of possible column loads. The gas
load for the column with 100-mm diameter was restricted to 1.7 Pa0.5. Therefore,
the liquid load was increased to higher values to reach the flooding region of the
catalytic packing. Two different flow regimes similar to those of conventional
structured packings can be observed. Flooding of the packing can be observed at
a pressure drop above 103 Pa/m. The possible column loads for MULTIPAK® are
very similar to those reported in Refs. 199 and 200 for KATAPAK-S.
The number of theoretical stages per meter of the catalytic packing was
determined as a function of the gas capacity factor. For the whole range of column
loads, the separation efficiency is at least four theoretical stages per meter. Moritz
and Hasse (200) determined an NTSM value of 3 for the laboratory-scale
KATAPAK-S. The separation efficiency remains constant for a wide loading
range of the packing. For lower column loads, the NTSM value increases to 6, a
phenomenon already reported in Ref. 90 for the conventional structured packing
Montzpak A3-500. A simple transfer-unit concept assuming all mass transfer
resistance in the vapor phase was used to determine the vapor-side mass transfer
coefficients (201). The mass transfer correlation

Sh G  0.009  Re G0.92  Sc1/3 (B8)


represents all experimental data with an accuracy of 13%. A comparison with
experimental data is shown in Figure 34.

B.5. Synthesis of Methyl Tertiary Butyl Ether


The mathematical description considered in Section 2.3 and Appendix A was
used as a modeling basis for the specially developed completely rate-based
simulator DESIGNER (155). This tool consists of several blocks, including
model libraries for physical properties, mass and heat transfer, reaction kinetics,
and equilibrium, as well as a specific hybrid solver and thermodynamic package.
DESIGNER also contains different hydrodynamic models (e.g., completely
mixed liquid–completely mixed vapor, completely mixed liquid–vapor plug flow,
mixed pool model, eddy diffusion model) and a model library of hydrodynamic
correlations for the mass transfer coefficients, interfacial area, pressure drop,
holdup, weeping, and entrainment that cover a number of different column inter-
nals and flow conditions.
In DESIGNER, different ways of taking account of heterogeneous reaction
kinetics are available, depending on the reaction rate and character. One further pos-
sibility is to use a detailed model for the heterogeneous catalyst mass transfer

Copyright © 2004 by Marcel Dekker, Inc. All Rights Reserved.


efficiency based on the approach of Ref. 91. When applying this type of kinetic
model, the intrinsic kinetics data are needed. Another way is the pseudo-homoge-
neous approach, with effective kinetic expressions, by which the kinetics description
is introduced as source terms into the balance equations [cf. Eqs. (A1) and (A2)].
For the system considered here, the reaction is slow as compared to the
mass transfer rate. For this reason the pseudo-homogeneous approach is used, the
reaction being accounted for in the liquid bulk only.
Basically, DESIGNER can use different physical property packages that are
easy to interchange with commercial flowsheet simulators. For the case consid-
ered, the vapor–liquid equilibrium description is based on the UNIQUAC model.
The liquid-phase binary diffusivities are determined using the method of Tyn and
Calus (see Ref. 72) for the diluted mixtures, corrected by the Vignes equation
(57), to account for finite concentrations. The vapor-phase diffusion coefficients
are assumed constant. The reaction kinetics parameters taken from Ref. 202 are
implemented directly in the DESIGNER code.

B.6. Reactive Extraction of Zinc


In conventional RE processes, the diffusive resistance is concentrated mainly inside
the droplet, whereas the aqueous-side resistance can be neglected. This has been
proven in Ref. 203 using the laser-induced-fluorescence (LIF) technique. Usually
the organic phase is more viscous and the diffusion coefficients of the organic com-
plexes are larger than those at the aqueous side, which supports this finding.
The mass transfer within a rigid droplet is determined by the Maxwell–
Stefan diffusion. The appropriate diffusion coefficients experimentally determined

FIGURE 34 Sherwood number correlation for MULTIPAK®.

Copyright © 2004 by Marcel Dekker, Inc. All Rights Reserved.


TABLE 4 Ternary Fick Diffusion Coefficients for the System
ZnR2 ( RH ) (1), RH (2), and Diluent (3) at 298.15 K .

Isododecane (3) Toluene (3)

D11 1.01  0.30  1010 m2/s 3.60  1.10  1010 m2/s


D12 0.11  0.10  1010 m2/s 0.35  0.42  1010 m2/s
D21 4.11  0.80  1010 m2/s 0.16  3.50  1010 m2/s
D22 2.79  0.30  1010 m2/s 7.68  0.21  1010 m2/s

for this zinc extraction system in Ref. 204 are presented in Table 4. With nonrigid
droplets, a mass transfer enhancement by internal convection has to be consid-
ered. However, with industrial feed solutions there are always impurities present
that may dampen the mass transfer (8). In contrast, there also might be a mass
transfer increase due to Marangoni effects (205,206). Therefore, for a final design
of a column, mass transfer measurements are recommended.
The macrokinetics of zinc extraction is discussed in detail in Ref. 8. It is a
combination of a reaction kinetics term (55) with the Maxwell–Stefan (54) or
eddy diffusion (56). The rate law is as follows:
2

d[ Zn 2 ]   [ R2 H2 ]1.5  [ Zn 2 ]  r  [ H ]2  [ ZnR2 ( RH )]  [ R2 H2 ]  (B9)


  v  
dt [ R2 H2 ]1.5  C1  [ H ]2  C  [ R H ] 
 2 2 2 

where C1, C2, , and r, are the estimated kinetics parameter (see EFCE test
systems discussed earlier).
The rate constant for the backward reaction, r, can be replaced by the ther-
modynamic equilibrium constant:
v
K eq  (B10)
r

The species concentrations are formulated in activities using the Pitzer


model (207) for the aqueous phase and the Hildebrand–Scott solubility parame-
ter (208) for the organic phase.
The effective diffusion coefficient is calculated according to the model of
Ref. 209, which accounts for interfacial instabilities. This model includes a
Handlos–Baron-like correlation (210) and one adjustable parameter, CIP :
w  dp
Deff  (B11)
  
2048  CIP 1  d 
 c 

Copyright © 2004 by Marcel Dekker, Inc. All Rights Reserved.

You might also like