Novel Batch Distillation Processes For The Separation of Systems With and Without Chemical Reaction

Lehrstuhl für Fluidverfahrenstechnik
Technische Universität München
Novel Batch Distillation Processes for

the Separation of Systems with and
without Chemical Reaction
Daniel Demicoli
Vollständiger Abdruck der von der Fakultät für Maschinenwesen der

Technischen Universität München zur Erlangung des akademischen Grades eines
Doktor-Ingenieurs
genehmigten Dissertation.
Vorsitzender: Univ.-Prof. Dr.-Ing. D. Weuster-Botz
Prüfer der Dissertation:

1. Univ.-Prof. Dr.-Ing., Dr.-Ing. habil. J. Stichlmair
2. Univ.-Prof. Dr.-Ing., Dr.-Ing. E.h. E. Blaß, em.
Die Dissertation wurde am 14.09.2005 bei der Technischen Universität München

eingereicht und durch die Fakultät für Maschinenwesen am 18.11.2005 angenommen.
Contents
1. Introduction 1
2. Literature review 3
2.1. Batch distillation processes . . . . . . . . . . . . . . . . . . . . . . . . 3
2.2. Integration of batch reaction with batch distillation . . . . . . . . . . 7
2.2.1. Coupling of batch reactor with a distillation unit . . . . . . . 8
2.2.2. Batch reactive distillation . . . . . . . . . . . . . . . . . . . . 11
2.3. Modelling and simulation . . . . . . . . . . . . . . . . . . . . . . . . . 12
3. Aim and outlook of the thesis 13
4. Modelling and simulation 15

4.1. Simulation software . . . . . . . . . . . . . . . . . . . . . . . . . . . . 15
4.2. Modelling of batch distillation column . . . . . . . . . . . . . . . . . 16
5. Batch Distillation in a Column with Side Withdrawal 19

5.1. Feasibility . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 20
5.2. Description of the process . . . . . . . . . . . . . . . . . . . . . . . . 20
5.3. Parametric studies . . . . . . . . . . . . . . . . . . . . . . . . . . . . 26
5.3.1. Geometric parameters . . . . . . . . . . . . . . . . . . . . . . 27
5.3.2. Operational parameters . . . . . . . . . . . . . . . . . . . . . 28
5.4. Comparison with the middle vessel batch distillation column . . . . . 31
5.5. Experimental validation of the process . . . . . . . . . . . . . . . . . 34
5.5.1. Apparatus . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 34
5.5.2. Experimental procedure . . . . . . . . . . . . . . . . . . . . . 34
5.5.3. Temperature and concentration profiles . . . . . . . . . . . . . 36
5.5.4. Conclusion . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 38
6. Integration of batch reaction with batch distillation - Equilibrium limited

reactions 39
6.1. Decomposition reaction . . . . . . . . . . . . . . . . . . . . . . . . . . 40
6.1.1. Minimum energy demand . . . . . . . . . . . . . . . . . . . . 41
6.1.2. Combining a reactor with a rectifying batch distillation column 44
6.1.3. Combining a reactor with a batch distillation in a middle vessel
column . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 48
6.1.4. Closed processes . . . . . . . . . . . . . . . . . . . . . . . . . 50
i
Contents
6.1.5. Conclusion . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 54
6.2. Production of a middle-boiling component . . . . . . . . . . . . . . . 56
6.2.1. Semi-batch reactor coupled with a distillation column . . . . . 56
6.2.2. Semi-batch reactive distillation . . . . . . . . . . . . . . . . . 61
6.2.3. Semi-batch reactive distillation in a column with a middle vessel 64
6.2.4. Comparison . . . . . . . . . . . . . . . . . . . . . . . . . . . . 67
7. Integration of batch reaction with batch distillation - Kinetically con-

trolled reactions 70
7.1. Combining a reactor with a regular batch distillation column . . . . . 71
7.2. Combining a batch reactor with a middle vessel column . . . . . . . . 74
7.3. Closed Processes . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 75
7.4. Comparison . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 80
8. Integration of batch reaction with batch distillation - Sequential reactions 81

8.1. Batch reactor . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 81
8.2. Combination of batch reactor with distillation processes . . . . . . . . 83
8.2.1. Light-boiling key educt . . . . . . . . . . . . . . . . . . . . . . 84
8.2.2. Heavy-boiling key educt . . . . . . . . . . . . . . . . . . . . . 91
9. Summary 94
References 97
A. Batch reactor coupled with a middle vessel column 101
ii
Nomenclature
a, b, c, d Components ordered in increasing boiling point temperature
B Bottom’s fraction, mol
Ḃ Bottom’s flow rate, mol/s
B1, B2, B3 Vessels
D Distillate fraction, mol
Ḋ Distillate flow rate, mol/s
F Amount of charge, mol
Ḟ Feed flow rate, mol/s
FC Flow controller
h Molar enthalpy, J/mol
HU Liquid molar holdup, mol
k Reaction rate constant, s−1
K Vapour-liquid equilibrium constant
Keq Chemical equilibrium constant
L̇ Molar liquid flow rate, mol/s
LC Level controller
L̇/V̇ Internal reflux ratio
M Middle vessel fraction, mol
N Number of moles, mol
nth Equilibrium stages
p Pressure, Pa
P Pump
p0i Saturated vapour pressure of component i, Pa
Q Energy demand, J
Q̇ Energy stream, J/s
r Latent heat of vaporisation, J/mol
r Reaction rate, s−1
RB Reboil ratio
RL Reflux ratio
S Selectivity
S1, S2, S3 Column sections
t Time, s
T Temperature, K
TC Temperature controller
T1, ..., T12 Temperature profile along column from top to bottom, ◦ C
iii
Nomenclature
V̇ Vapour flow rate , mol/s

V1, V2, V3 Valves
Ẇ Flow rate of withdrawal stream, mol/s
W1, W2, W3 Heat exchangers
x Molar fraction in the liquid phase, mol/mol
X Conversion
x̄ Average molar fraction in the liquid phase, mol/mol
y Molar fraction in the vapour phase, mol/mol
Y Yield
z Molar fraction in the feed stream, mol/mol
αij Relative volatility between components i, j
γ Activity coefficient
∆D Infinitesimal change due to distillation
∆NR Infinitesimal change of the number of moles due to the reaction, mol
∆R Infinitesimal change due to reaction
∆t Infinitesimal time interval, s
θo Boiling point temperature, ◦ C
ν Stoichiometric coefficient
πR Reaction pole
σ Recovery
τ Relative time
Subscripts
a, b, c, d Components ordered in increasing boiling point temperature

B Bottom vessel
D Distillate vessel
f Forward reaction
i General component
k Reaction counter
L Lower column section
M Middle vessel
max Maximum
min Minimum
n Equilibrium stage counter, numbered from top to bottom
p Reaction products
of f Off-cut
r Reactants
r Reverse reaction
tot total
U Upper column section
α Beginning of the process
iv
Nomenclature
ω End of the process
Superscripts
F Feed stream
N Number of components
L Liquid phase
SP Setpoint
V Vapour phase
W Withdrawal stream
v
1. Introduction
Speciality chemicals and pharmaceutical agents are produced in small amounts that
are usually required in very high purities. These chemicals have high market prices,
and either short life cycles, or a fluctuating demand. They are therefore produced
in multi-purpose discontinuous processes. These processes usually consist of one or
more reaction steps, followed by separation processes. Here, the desired products
are purified from solvents, by-products, and unreacted educts. Due to the short life
cycles of the products, both the reaction as well as the separation processes should be
capable to process a variety of mixtures, and to deliver a variety of product purities.
In batch synthesis, the reaction and separation tasks are in general sequentially
ordered [Ahmad and Barton 1995]. The reactors and separators are therefore consid-
ered as independent unit operations, between which, the process fluids are transferred
in batches. In this case, the nature of the mixture leaving the reactor usually deter-
mines the separation process, and therefore the separation has no influence on the
reaction. If, on the other hand, the reaction and separation are integrated in one
unit operation, the two phenomena reciprocally influence each other. In this way, the
yield, selectivity, and conversion of the reaction can be improved, and the separation
processes can be simplified.
Batch distillation is a classical unit operation for the separation of small amounts
of liquid mixtures. It is very simple and flexible, and furthermore it can deliver
very high product purities. Due to its importance on the industrial level, batch
distillation has been extensively studied, and several modifications of the process
have been developed. The operation of a batch distillation column has also been
comprehensively investigated, with several operation policies being developed and
applied on the industrial scale. The aim of most of the academic studies was to
optimise the operation, to postulate alternative processes, to develop novel separation
methods for azeotropic mixtures, and to increase the flexibility of the unit operation.
In this optic, a novel batch distillation process, for the purification of a middle-boiling
component from heavy and light-boiling impurities, is described and analysed in the
first part of this work. The operation policy of this process is explained with the aid of
simulation and experimental results. Furthermore, the advantages and disadvantages
of the novel distillation process are shown.
In the pharmaceutical industry, it is common practice to couple a batch reactor
with a rectifying column. In this way, the light-boiling reaction products can be
removed from the reaction system upon their formation. However, only few research
groups studied the integration of reaction with batch distillation, and in most cases,
the numerous modifications of the batch distillation process have not been considered.
1
1. Introduction
The main focus of this work is the synthesis of novel batch reactive distillation
processes. Several novel processes are developed and analysed for a variety of reaction
systems. For each case, the operation and basic control of the process is discussed, and
the effect of the most important parameters is studied. Furthermore, the advantages
of competing processes are outlined. In this way, new understanding of a bright
spectrum of processes is gained, providing the process engineer with a large palette
of alternative processes. The tools and criteria to select the best processes are also
illustrated.
2
2. Literature review
The synthesis of pharmaceutical active agents and of speciality chemicals is usually
carried out in multi-purpose plants that are operated discontinuously. Usually, a
single synthesis is characterised by several reaction and separation steps which are se-
quentially ordered. Due to its flexibility, simple operation, and low capital costs, batch
distillation has established itself as a standard unit operation for the separation of
small amounts of liquid mixtures. This unit operation has therefore been extensively
studied. In Section 2.1, the operation of the conventional batch distillation column
is described, and its disadvantages are outlined. Hence, some alternative operation
modes and some novel processes are introduced. In the second part, Section 2.2, the
published research concerning the integration of batch reaction with distillation is
reviewed.
2.1. Batch distillation processes

Batch distillation is a well-established process for the separation of small quantities
of liquid mixtures required to be in high purity. The most common process for
batch distillation is shown in Figure 2.1. Here, a charge is loaded into the still
placed at the bottom of a rectifying column. The products are obtained in order of
RL
a b1 ... bn
Charge
Bw
Figure 2.1.: Schematic representation of the regular batch distillation process.
3
increasing boiling point from the top of the column. During the separation process,
the heavy boiler is enriched in the column’s sump, from which a heavy-boiling fraction
is recovered at the end of the process. The key success of batch distillation has been
shown to be due to its simplicity and its flexibility. In fact, several mixtures can
be processed in different batches in one single separation unit. Nevertheless, batch
distillation presents some very important disadvantages that limit its applicability.
The energy demand of batch distillation is generally much higher than that of the
continuous process, and the control and operation are usually very complex due to
the continuously changing concentrations in the column. The residence time of the
mixture in the column’s sump is very long. In the sump there is also the highest
temperature, which, due to the enrichment in the heavy-boiling components, increase
with time. Furthermore, the most volatile components come into contact with hot
surfaces of the reboiler [Nitsche 1982]. These factors constitute a major disadvantage
for temperature sensitive mixtures, which might degrade under these conditions.
For the reasons mentioned above, and with the help of the new computer based
solving routines developed in the second half of the 1980’s, alternative processes to the
regular batch distillation have been extensively investigated. Two of these processes
are illustrated in Figure 2.2.
.
D
Charge Charge
Dw Mw
. .
B B
(i) (ii)
Figure 2.2.: Batch distillation: (i) inverted, and (ii) middle-vessel column configura-
tions.
4
In the inverted batch distillation, the charge is loaded into a vessel placed at the
head of the column, as shown in Figure 2.2 (i). The products are obtained in order
of decreasing boiling point from the column’s bottom, and a light-boiling fraction
can be obtained at the end of the process from the feed vessel. In this case, the
mixture in the feed vessel is subject to the lowest temperatures in the column, and
this temperature decreases with time. Furthermore, Sørensen and Skogestad [1996]
prove that if the mixture is rich in the heavy-boiling component, then the inverted
column delivers time and energy savings with respect to the regular column.
In the middle vessel column, shown in Figure 2.2 (ii), the charge is loaded into the
still placed in the middle of the column, between a stripping and a rectifying column
section. The light and heavy-boiling products are simultaneously obtained from the
column’s head and sump, respectively. As discussed by Warter [2001], Warter et al.
[2004] and other authors, this can have several advantages. The temperature in the
middle vessel can be kept low throughout the process, reducing the thermal exposure
of the components in the charge. Furthermore, the time required by the process can
be reduced with respect to the regular column, and, in some cases, the separation
procedure can be simplified. Warter and Stichlmair [2000], Skouras and Skogestad
[2004a] and Rodriguez-Donis et al. [2001] show that the middle vessel column can be
used for the separation of azeotropic mixtures, which would otherwise be impossible
or very complex to perform in a regular batch distillation column.
Operation modes for batch distillation processes

The three processes described in the previous section are characterised by an open
operating procedure, where the initial mixture is loaded into a charge vessel, and
the products are withdrawn from the column’s head and sump. The reflux and/or
reboil ratios can either be kept constant, or be used to regulate the composition of
the product [Stichlmair and Fair 1998]. Alternatively, they can be used to optimise
the process [Mujtaba and Macchietto 1992].
Batch distillation processes can be operated in a closed mode. In this case, there
is no net product withdrawal from the system [Skogestad et al. 1997, Warter and
Stichlmair 2002, Warter et al. 2004, Skouras and Skogestad 2004b]. The closed op-
eration of a middle vessel column is shown in Figure 2.3. The charge is initially
distributed between the middle, top and bottom vessels, which are connected to the
column. During the closed operation, the temperatures in the rectifying and in the
stripping column sections are controlled by setting the reflux rates from the top and
the middle vessels, respectively. At the end of the process, the concentrations in the
vessels reach steady state, therefore the temperature profile stabilises, and the column
is operated under total reflux conditions. Warter and Stichlmair [2002] and Warter
et al. [2004] experimentally validate the closed operation of a middle vessel column
under level control in the top and middle vessels. Here, the distillation column is
operated under total reflux, and therefore the best conditions for separations exist.
Furthermore, the operation and control of this process are very simple; nevertheless,
5
the composition of the charge has to be accurately known at the beginning of the pro-
cess. This might constitute a major disadvantage for the complex mixtures usually
processed in batch distillation units, since the exact composition of the charge might
fluctuate and be difficult to obtain.
Charge
Dw TC
Charge
Mw TC
Charge
Bw
Figure 2.3.: Closed operation of a batch distillation in middle-vessel column under

temperature control.
Minimum energy demand

Distillation is an energy-intensive separation process, since heat has to be supplied
to the system in order to create the second phase required for separation. Both
during the design and operation phases, the energy required by the process plays
a fundamental role. The minimum energy required for a given separation task is
that needed by a distillation column having an infinite number of stages. This is
therefore a theoretical minimum, which can be used during the design phase as a pre-
selection tool to restrict the choice among the available batch distillation processes.
6
2.2. Integration of batch reaction with batch distillation
On the other hand, during operation, the minimum energy demand can be used as
a measure of how far the process is from its thermodynamics optimal, and how wide
the optimisation margins of the process are.
The most popular method for the calculation of the minimum energy demand is the
one published by Underwood [1948]. This is an iterative method, which , under the
assumption of constant molar overflow, allows the calculation of the minimum reflux
ratio for ideal mixtures. Offers et al. [1995] describe a direct method to calculate the
minimum reflux ratio for a given separation both for ideal and real systems. This
method is applicable to multi-component batch distillation operated with constant
product composition. The calculation of the minimum reflux and reboil ratios is re-
peated for every concentration in the still. The still’s concentration can be determined
via Rayleigh’s equation, as described in Stichlmair and Fair [1998]. This calculation
requires only the knowledge of the relative volatilities of the mixture in the still.
2.2. Integration of batch reaction with batch

distillation
In batch synthesis, the products leaving a reactor are usually mixed to other products,
to reactants and to solvents. These mixtures need to be separated into the single
components, which are generally required in high purities. It is common industrial
practice to carry out the reaction and separation processes sequentially in separate
unit operations. In case of equilibrium limited reactions, this practice requires recycles
of the reactants to the reactor. On the other hand, in case of sequential or parallel
reactions, undesired side products might be formed at the expense of the main reaction
products. Integrating reaction with separation in a single unit operation would remove
the reaction products from the reaction region upon their formation, hence shifting
the chemical equilibrium to the products’ side, suppressing undesired side reactions
and overcoming distillation barriers.
The integration of a reactor with a distillation unit is commonly called either reac-
tive or catalytic distillation. In the case of batch processes, such integrated processes
might be classified in two families. On the one hand, a batch or a semi-batch reactor
can be coupled with a distillation column. Here, the reaction and separation pro-
cesses are only coupled in time, but they are spatially separated, the reaction being
limited to the reactor. On the other hand, the reaction can be carried out within the
distillation column. Here, the reaction and separation are integrated both in time
and space. In this case, one speaks of batch reactive distillation. In the following
paragraphs, a literature review of the two processes is given.
7
2.2.1. Coupling of batch reactor with a distillation unit

When exothermic reactions are carried out in the liquid phase in stirred tank reactors,
the temperature control of the system could be simplified by allowing the reaction
mixture to react at its boiling point temperature. A total reflux condenser is placed
on top of the reactor, so that setting the operating pressure of the condenser deter-
mines the operating temperature in the reactor. Another common practice in the
pharmaceutical industry is to couple batch reactors with rectifying columns. In this
case, the light-boiling reaction products are immediately removed from the reaction
mixture. For equilibrium limited reactions, this shifts the conversion to the products’
side, and improves the selectivity of the desired products by suppressing side reac-
tions. Notwithstanding its popularity on industrial level and its early appearance in
the literature [Ruby et al. 1978], this process has received limited attention by the
academic world.
Ruby et al. [1978] were the first authors to consider a batch reactor coupled with a
rectifying column. Here, a non-specified esterification reaction, with the formation of
light-boiling products, is considered. However, the main focus of this publication is
the solution of the mathematical problem. This was a main issue, since stable solvers
for Differential Algebraic Equations (DAE) were developed only in later years as for
example by Cuille and Reklaitis [1986]. In a second publication, Egly and Ruby [1980]
compare experimental and simulated results of the optimised process. Furthermore,
they experimentally show that the conventional control system for batch distillation
is able to reject disturbances to the combined process.
Mayer and Wörz [1980] compare continuous reactive distillation, a batch reactor
coupled with a rectification column, and a continuous stirred tank reactor (CSTR)
coupled with a rectifying column. A fast equilibrium reaction is here taken into
consideration. The boiling points of the reactants lie between those of the reaction
products, which are therefore obtained as top and bottom fractions. The three con-
figurations are studied for varying values of the reaction equilibrium constant, and
for different volatilities of the components. The authors compare the conversion of
one of the reactants for a given energy input to the process. For most cases, the
continuous reactive distillation process shows the best results. However, for high val-
ues of the chemical equilibrium constant, and for increasing volatility of the heavy
boiler, the difference between the continuous and the batch process are considerably
reduced. This later result is however of minor importance, since in the continuous
reactive distillation the reactive part of the column was limited to the stages below
the feed point. The CSTR coupled with a rectifying column always shows the worst
performance.
Reuter et al. [1989] treat in more detail the dynamic simulation of the batch distil-
lation column. An equilibrium limited transesterification reaction, with the formation
of light-boiling products, takes place in the column’s sump. The simulated and exper-
imental distillate concentrations are compared for an optimised process. The results
show good agreement.
8
Control of a batch reactor coupled with a rectifying column

The control issues are more extensively treated in Sørensen and Skogestad [1994].
The authors show that the open-loop reflux policies, i.e. constant reflux, or constant
distillate flow rate policies, should not be applied to the batch reactor coupled with
a distillation column. Furthermore, the authors compare three alternative control
strategies for this process:
a) One-point bottom control, with direct control of the temperature in the reac-
tor. Effective control is only obtained if the set-point temperature is set below to a
maximum value, which is difficult to estimate a priori. The set-point temperature is,
therefore, usually chosen to be too low, slowing down the process.
b) Two-point control, where both the distillate composition and the reactor temper-
ature are controlled. This results in two control loops, which show strong interactions.
c) One-point column control. The temperature on a tray in the column is controlled
by changes to the distillate stream. This is the usual control strategy adopted in batch
distillation. The heat duty to the reboiler is either set to a constant value, or used to
control the pressure drop in the column. In the latter case, the control loop is slow,
and it is used to maximise the throughput of the column. The authors state that this
control strategy is the simplest to implement and it delivers good control, provided
that the temperature measurement point is properly selected. This control strategy
is subsequently applied to an optimised process [Sørensen et al. 1996].
Coupling a batch reactor with different column configurations

Li et al. [1998a] couple a semi-batch reactor with a rectifying column for a trans-
esterification reaction in which a light-boiling alcohol, and a heavy-boiling ester are
formed. The reactants are both charged to the reactor at the beginning of the process.
However, in order to boost the reaction rate, excess reactant alcohol is continuously
fed to the reactor. This maintains a relatively high alcohol concentration in the re-
actor, therefore improving the reaction rate. The product ester is then purified from
the excess alcohol in a second distillation step. The authors solve a time optimisation
problem, obtaining optimised profiles both for the feed flow rate, and the reflux ratio
policy.
Mujtaba and Macchietto [1992] were the first to consider the combination of a
batch reactor with an inverted and with a middle vessel column. Several chemical
equilibrium limited reactions are taken into consideration, and the suitable column
configurations for each reaction are identified. Furthermore, the authors state that
in the case that the boiling point of the reaction products lies between that of the
reactants, none of the discussed column configurations would be suitable. Neverthe-
less, the authors neither take into consideration semi-batch processes, nor reactive
columns. In a later publication [Mujtaba and Macchietto 1994], the authors compare
three column configurations for a given reactive mixture, where light and heavy-
boiling components are formed.
9
Overcoming distillation boundaries

Guo et al. [2003] address the feasibility of the combined batch reaction-distillation
processes for ternary azeotropic mixtures. They develop simple criteria to graphically
determine the feasible products. These require only the knowledge of the residue curve
maps and the chemical equilibrium. In their work, semi-batch processes and reactive
columns are however not considered. The authors show that distillation boundary
lines could be crossed, and pure products could be obtained, depending on the position
of the chemical equilibrium line and the column configuration. The results of the
graphical analysis and the feasibility criteria are proved solving simplified, pseudo-
steady-state material balances developed by Venimadhavan et al. [1999].
Espinosa [2002] presents a novel process, in which a batch reactor is coupled with
a batch extractive distillation in a middle vessel column. This process is illustrated
for the esterification of ethanol and acetic acid. This mixture presents a highly non-
ideal vapour-liquid equilibrium, since three minimum binary azeotropes and a ternary
azeotrope are formed. In this process, acetic acid is continuously fed to the upper part
of the extractive column section, thus functioning both as an extractive solvent (in
extractive distillation) and as reactant (in the middle vessel). An operating procedure
is postulated, and the influence of the operating and geometrical parameters of the
column are analysed.
Suppressing side reactions

Gadewar et al. [2000] illustrate the application of a batch reactor coupled with a
rectifying column for consecutive reactions, where the desired product is produced by
the first reaction, and consumed in the second. Furthermore, this is the light-boiling
component. They compare the selectivity and yield of the combined process with
that of a plug flow reactor (PFR) as a function of the Damköhler numbers of the two
reactions. The Damköhler number is a dimensionless number expressing the ratio
of the characteristic liquid residence time to the characteristic reaction time. They
show that the combined process always outperforms the PFR over the whole range
of the Damköhler numbers. The study is repeated for a parallel reaction in which the
desired product reacts with one of the reactants. Since, the desired product is the
heavy-boiling component of the system, they suggest coupling the batch reactor with
a stripping column. This is also compared to a PFR. Even though the two process
alternatives show similar behaviour for various Damköhler numbers, the batch reactor
coupled with the batch stripper shows the best yield and selectivity to the desired
product. The simplified model developed by Venimadhavan et al. [1999] is applied
also in this study.
10
2.2.2. Batch reactive distillation

The spatial integration of reaction with distillation requires special hardware, which
enables the contact of the reactants to the catalysts and provides the necessary volume
for the reaction, without diminishing the interfacial area between the liquid and the
vapour phases [Krishna 2003]. Such hardware makes batch reactive distillation a very
specific process, increases the investment costs, and reduces the flexibility of the batch
unit. Furthermore, the chemical reaction has an important effect on the vapour-liquid
equilibrium [Barbosa and Doherty 1988a;b], so that the so-called reactive distillation
lines result [Frey and Stichlmair 1999]. These might give rise to reactive azeotropes,
which are otherwise not present in the non-reactive case, hence creating new barriers
to the separation process. For these reasons, batch reactive distillation has met
little interest in the literature. Furthermore, only one real industrial application
of the process has been published [Osorio et al. 2005], namely wine distillation for the
production of Pisco.
Cuille and Reklaitis [1986] and Albet et al. [1994] address the problems associated
with the simulation of a dynamic reactive distillation column. Albet et al. [1994] pay
particular attention to the start-up procedure and to the initialisation of the model
of the batch reactive distillation column. Wajge and Reklaitis [1999] introduce a sim-
ulation and optimisation framework for batch distillation with and without chemical
reaction. Xu and Dudukovic [1999] present a model for the photo-catalysed chlorina-
tion of toluene. The reaction takes place both in the liquid and in the vapour phases.
All the above-mentioned models are based on the assumption that vapour-liquid and
reaction equilibria are reached on each tray. A detailed non-equilibrium stage model
including the mass transfer between the phases for a heterogeneously catalysed batch
reactive distillation column is solved in Schneider et al. [2001].
The esterification of methanol and acetic acid in a semi-batch reactive distillation
column is studied by Fernholz et al. [2000] and Schneider et al. [2001]. The light-
boiling reactant, methanol, is charged to the sump, and the heavy-boiling reactant,
acetic acid, is continuously fed at the top of the reactive column section. Hence, the
two reactants are subject to a counter current flow in the catalytic section. Methyl
acetate, the light-boiling reaction product is therefore recovered as distillate fraction.
The control of batch and semi-batch reactive distillation columns has been discussed
in Monroy-Loperena and Alvarez-Ramirez [2000], Balasubramhanya and Doyle [2000]
and Engell and Fernholz [2003].
11
2.3. Modelling and simulation
2.3. Modelling and simulation

Several authors have published the modelling of batch distillation processes. This is
based on dynamic mass and energy balances, on vapour-liquid and chemical equilib-
rium relationships, and on the hydrodynamics of the column. The models presented
in the literature differ on the grade of detail considered. The highest degree of detail
is provided by the so-called non-equilibrium or rate-based models. Here, the mass
transfer between the phases is described in detail, this requires empiric mass transfer
correlations. In the greatest number of publications, the batch distillation processes,
both non-reactive as well as reactive, are described by equilibrium stage models. It
is assumed that vapour-liquid equilibrium is reached on every stage. Even though
such relationships are simpler than those required for the rate-based models, several
authors, as for example Warter et al. [2004] and Wittgens and Skogestad [2000], have
shown that the simulated and experimental results are in sufficiently good agreement.
Furthermore, in most models, it is assumed that the hydrodynamics of the column, i.e.
pressure drop and holdup, have a negligible effect on the vapour-liquid equilibrium,
and are therefore not included in the model.
The mathematical description of a batch distillation process results in a system of
Differential-Algebraic-Equations (DAE). The current solvers are able to solve prob-
lems that have an index of 1. These are DAE systems which can be transformed
in Ordinary-Differential-Equations (ODE) merely with algebraic manipulations [Pan-
telides et al. 1988]. However, if the hydrodynamics of the column, and in particular
the pressure drop in the column, were not included in the modelling, a DAE system
of index 2 would be required to describe the process. This means that it is not pos-
sible to convert the DAE into an ODE system with algebraic manipulation of the
equations. Pantelides et al. [1988] suggest two ways to circumvent this problem. A
simplified pressure drop relationship, relating the pressure drop on every equilibrium
stage to the square of the vapour flow rate, can be included in the model description.
The second possibility is to assume that the dynamics of the enthalpy changes on
the stages are very slow compared to those of mass transfer between the phases. A
pseudo-steady state energy balance can therefore be written.
The various batch distillation processes treated in this thesis have been described
with the equilibrium stage model. The mathematical models involved and the simu-
lation software used in this work are illustrated in Chapter 4.
12
3. Aim and outlook of the thesis
As has been outlined in the literature review, numerous authors studied different
aspects of batch distillation and proved the flexibility of this well established unit
operation. However, only few works treated the integration of batch reaction with
batch distillation, and to date, there is no work that extensively treats the synthesis
of integrated batch processes. The aim of this work is to provide the process engineer
with novel tools for the design of batch distillation processes with particular attention
paid to batch reactive distillation processes. This is achieved by describing the oper-
ation of novel processes, underlining their advantages, and marking their application
fields.
A novel process for the purification of a middle-boiling component from light and
heavy-boiling impurities is introduced in Chapter 5. An operation procedure and a
basic control structure are postulated. These are subsequently verified by process
simulations. The most important process parameters are identified, and a parametric
study is performed. The novel process is compared to the batch distillation in a
middle vessel column. This comparison illustrates the importance of the novel batch
distillation process and marks its application field. Finally, the feasibility and the
practical implementation of the process are demonstrated via experiments in a pilot-
plant.
In each of the three following chapters, various batch reactive distillation processes
are investigated and compared for a given reaction class. In Chapter 6, different
processes for the treatment of equilibrium limited reactions are introduced for two
different reaction structures. In Section 6.1, the operation procedure and basic con-
trol of several process alternatives are illustrated for a decomposition reaction where
heavy and light-boiling products are formed. The combination of batch reactors with
a distillation column, as well as batch reactive distillation columns are taken into
consideration. In Section 6.2, an addition reaction with the formation of a middle-
boiling component is considered. For this reaction type, several semi-batch processes
are postulated and analysed. The reaction is carried out either in a semi-batch reactor
coupled with a batch distillation column, or in the liquid phase of a batch reactive
distillation column. These processes are compared at the end of the section.
Integrated processes for kinetically controlled reactions are discussed in Chapter 7.
Only reactions with the formation of either light or heavy-boiling products are consid-
ered. Due to the long residence times required by the reaction, only reactors coupled
with distillation columns play an important role in process synthesis for this class of
reactions. Several operating procedures are postulated and verified by simulations.
The simulation results are then used to compare these processes.
13
3. Aim and outlook of the thesis
Finally, the importance of batch reactive distillation processes for the suppression
of side reactions is illustrated in Chapter 8. The volatility of the main product
and that of the key educt play an extremely important role in process synthesis.
This is illustrated for two systems with parallel reactions. In both cases, the main
product, formed in the first reaction, is a middle boiler. This then reacts with one
of the educts, forming an undesired product. In one case, the key educt is a light-
boiling component, and in the other, it is a heavy boiler. The reaction is carried
out in integrated semi-batch processes. Both the combination of a semi-batch reactor
with a batch distillation process, as well as batch reactive distillation processes are
synthesised and compared.
14
4. Modelling and simulation
Batch processes, as opposed to continuous operation, are characterised by a non-
steady state behaviour. Their mathematical description is therefore based on time
dependent mass and energy balances, resulting in systems of equations which include
both differential and algebraic equations. In the past decades, several solvers able to
solve such complex systems have been developed and made commercially available.
Furthermore, in recent years, tools for the optimisation of time dependent processes
have also appeared, and the first works treating the optimisation of batch distillation
processes have been published. However, this remains a young field requiring further
research.
This chapter provides an overview of the basics of the thermodynamic principles
and of the mathematical models necessary to describe the batch distillation processes
developed in the course of the thesis.
4.1. Simulation software

The batch distillation processes investigated in this thesis are modelled using the soft-
ware package ”general PROcess Modelling System” (gPROMS) developed by Process
Systems Enterprise Ltd., London. gPROMS is a powerful general purpose modelling
and optimisation environment, used to enhance the design and operation of continuous
and dynamic processes [gPROMS 2003]. It is an equation-based simulation environ-
ment, where the user can write his own mathematical models to express the physical
and chemical relationships, as well as the operational task sequences describing the
processes and their operation. gPROMS analyses the relationships governing the
process and performs dynamic or steady-state simulation, optimisation, or parameter
estimation.
In this modelling environment, it is possible to divide the complex process into sub-
models that describe basic units. These sub-models can be called by higher hierarchy
models, with which they exchange information. A gPROMS model consists of a
system of models each with a different hierarchy. The models are connected together
via data and material streams. The use of such a hierarchical modelling approach
allows a complex process to be reduced to its component parts. This simplifies the
model writing process, reduces model complexity, improves model readability, and
facilitates debugging. It is also possible to use a model that has been written for a
particular unit operation in other processes or even repeatedly in the same process.
The thermodynamic properties of the liquid and vapour phases are written in those
15
4.2. Modelling of batch distillation column
models having the lowest hierarchy. These models are then used in the next level,
where the single process units (such as vessels, equilibrium stages, and heat exchang-
ers) are described. The single process units are then connected together via material
streams in the model having the highest hierarchy. This model represents the flow-
sheet of the process. Moreover, the values of the constants are assigned in this model.
These are then automatically passed to the models having lower hierarchy.
In the ”PROCESS” section, the boundary conditions of the differential variables
are assigned. The detailed operation of the process is also described in this sec-
tion. gPROMS is able to treat discontinuities, thereby allowing for the simulation of
complex operating procedures consisting of several phases.

A complex batch distillation column is modelled. This column is used as a super-
structure for modelling the various processes described later in the thesis. However, in
order to reduce the computational time and improve the stability of the mathematical
model, the redundant parts of the complex column are removed.
The general complex column is illustrated in Figure 4.1 (i). The dotted lines repre-
sent the balance volumes. These are described with mass and energy balances as well
as with equilibrium relationships. The equilibrium stage model is used for the de-
scription of the column. This is based on the assumption of perfect mass and energy
transfer between the phases. Furthermore, the streams leaving an equilibrium stage
are assumed to be in thermodynamic and mechanical equilibrium. An equilibrium
stage, represented in Figure 4.1 (ii), is therefore described by the so-called MESH
equations, which, for the time dependent case, are written as described below.
The component mass balance results from the principle of conservation of mass.
This principle states that for every component, the number of moles accumulated in
the control volume is equal to the difference between the number of moles entering
and that leaving the control volume, added to the moles generated by the chemical
reaction.
d(HUn · xn, i )
= L̇n−1 · xn−1, i + V̇n+1 · yn+1, i − V̇n · yn, i − L̇n · xn, i +
dt X
+ Ḟn · zn, i − Ẇn · xn, i + {rn, k · νk, i } · HUn · xn, i (4.1)
k
The vapour-liquid equilibrium relates the molar fraction of each component in the
vapour phase to that of the same component in the liquid phase. Under the assump-
tion of an ideal gas phase, this is expressed by:
yn, i · pn = γn, i · p0n, i · xn, i (4.2)
In Chapter 5 a real mixture is modelled. The saturated vapour pressures (p0 ) and
the activity coefficients (γ) are therefore calculated using empirical models. In the
16
.
. QC
VU
. .
LU D
. .
. Ln-1 Vn Condenser
F
. .Tray
Ln Vn+1 . .
L n-1 Vn
Upper x n-1 yn
. section
LW .
F
. Lower zn
LL section
Vessel . .
. . Ln V n+1
L.k-1 V.k xn y n+1
Lk Vk+1
.
VL
.
QR
.
B
Reboiler
(i) (ii)
Figure 4.1.: (i) Flowsheet of the model of a complex batch distillation column, and
(ii) scheme of an equilibrium stage.
rest of the work, the liquid is assumed to behave ideally, i.e. γi = 1; and the relative
volatilities (αi, j ) of the components are assumed to be independent of the tempera-
ture. The vapour-liquid equilibrium is therefore expressed in terms of the saturated
vapour pressure of the heaviest boiling component and the relative volatilities with
respect to this component:
yn, i · pn = αi, d · p0n, d · xn, i (4.3)
From the definition of mole fraction the following summations result:
C
X C
X
xn,i = yn,i = 1 (4.4)
i=1 i=1
Furthermore, the principle of conservation of energy gives the following energy
17
balance:
d(HUn · hLn )
= L̇n−1 · hLn−1 + V̇n+1 · hVn+1 − V̇n · hVn − L̇n · hLn +
dt
+ Ḟn · hFn − Ẇn · hW
n + Q̇n (4.5)
As already discussed in the literature review in Section 2.3, if the hydrodynamics

of the column are ignored, the resulting systems of equations cannot be solved by the
current solvers. In order to isolate the thermodynamic behaviour of the system from
its hydrodynamics, the models used do not take into consideration the hydrodynamics
of the column. However, it is necessary to reduce the index of the system to 1. This is
achieved by assuming that the dynamics of energy transfer are much slower than the
mass transfer dynamics. This assumption translates to writing the energy balance as
a steady state balance:
0 = L̇n−1 · hLn−1 + V̇n+1 · hVn+1 − V̇n · hVn − L̇n · hLn +

+ Ḟn · hFn − Ẇn · hW
n (4.6)
In order to prove the validity of this later model, three models have been written:
one including the detailed description of the fluid dynamics of a batch distillation
column; one relating the pressure drop to the square of the vapour flow rate through
the column; and the third one with a pseudo steady-state energy balance given by
Equation 4.6. Under the assumption of no heat generation and no heat transfer on
the stages, the three models deliver similar concentration and temperature profiles. It
was therefore concluded that, as long as no heat is generated, and there is no external
heat transfer on the stages, the simplified models have the advantage of requiring
less geometric data of the distillation equipment, without reducing the quality of the
thermodynamic results.
18
5. Separation of Zeotropic Ternary
Mixtures via Batch Distillation in a
Column with Side Withdrawal
As already pointed out in the literature review, several batch distillation processes
are available for the separation of zeotropic mixtures. In this chapter, a novel process
will be introduced. This is shown in Figure 5.1. It consists of a distillation column
equipped with sump and distillate vessels, into which the charge is loaded at the
beginning of the process. The main characteristic of this process is the withdrawal
stream placed in the middle of the separation column. Here, a middle-boiling mixture
having the required concentration can be continuously withdrawn during the process.
The light-boiling fraction would, therefore, be recovered at the end of the process in
the top vessel (Dω ); the middle-boiling fraction at the side withdrawal stream (Ẇ );
and the heavy-boiling fraction in the reboiler vessel (Bω ).
Charge
Stripping
Dw section
.
W
Rectifying
section
Charge
Bw
Figure 5.1.: Batch distillation column with side withdrawal.
19
5.1. Feasibility
5.1. Feasibility
To analyse the feasibility of the process, it is useful to conceptually divide the dis-
tillation column into two separate sections placed on top of each other, as shown in
Figure 5.1. The lower column section can then be seen as a rectifying column, and
the upper column section as a stripping column. The vapour leaving the rectifying
section is directly fed to the bottom of the stripping section. Analogously, the liquid
leaving the bottom of the stripping section provides both the reflux to the rectifying
section and the product stream (Ẇ ).
This conceptual decomposition of the column allows for the application of feasibility
studies developed for stripping and rectifying columns, provided that the concentra-
tion profiles of the two columns meet at the withdrawal stage. Therefore, it is possible
to obtain a pure intermediate-boiling product from a column with an infinite number
of stages operated with total reflux ratios, only if the top and bottom vessels do not
contain the heavy and the light-boiling components, respectively.
Figure 5.2 shows the simulated concentration path of the liquid phase in the top
vessel (xD ), on the withdrawal tray (xW ), and in the bottom vessel (xB ) of a column
having a very high number of stages and being operated at close to total reflux
conditions. Here, the separation of an equimolar mixture of ethanol, 1-propanol and
1-butanol is depicted.
It is clearly visible that, as long as the top and bottom vessels contain heavy and
light boiler respectively, the concentration of the intermediate boiler on the with-
drawal tray is smaller than one. It follows that, to obtain a high purity intermediate-
boiling component on the withdrawal stage, a two-step operating procedure is re-
quired. In the first operation step, the light boiler should be transferred from the
bottom to the top vessel. Simultaneously, the heavy boiler is transferred from the top
to the bottom vessel. During this process step, the concentration of the intermediate
boiler in the column will be lower than the required purity. No product is therefore
withdrawn from the column. This step is termed closed operation, and is equivalent
to the process introduced by Sørensen and Prenzler [1997]. When the concentration
of the intermediate boiler on the withdrawal stage has reached the required value,
an intermediate-boiling fraction can be continuously withdrawn at the withdrawal
stream. The procedure will be discussed in detail in the next section.
5.2. Description of the process

The process will be illustrated for the purification of 1-propanol (θbo (1 bar) = 97.8 ◦ C)
from a mixture containing 0.1 mol/mol ethanol (θao (1 bar) = 78.4 ◦ C) and 0.1 mol/mol
1-butanol (θco (1 bar) = 117 ◦ C). The relative volatilities of the components do not
vary much over the temperature range of the column, and their average values are:
αab = 2.0 and αbc = 2.2. The total number of equilibrium stages in the column is
nth = 21, and the withdrawal stage is placed in the middle of the column. Further
20
1.0
ethanol
mol
mol
distillate vessel xD
Concentration in
0.6
0.4
0.2
0.0
1.0
mol
mol
withdrawal stream xW
1-propanol
0.6
Concentration of
0.4
0.2
0.0
1.0
1-butanol
mol
mol
Concentration in
sump vessel xB
0.6
0.4
0.2
0.0
0.0 0.2 0.8 1.0
Relative time t/t tot
Figure 5.2.: Concentration path in distillate vessel, on the withdrawal stage and in
the sump vessel at very high number of stages and close to total reflux
conditions.
21
constants assigned in the simulation are summarised in Table 5.1.
Table 5.1.: Simulation parameters for the process described in Section 5.2.
Molar fractions in the charge (xa , xb , xc ) (0.1, 0.8, 0.1)

Number of equilibrium stages nth 21
Position of withdrawal stream nW 10
Position of upper TC nTU C 8
Position of lower TC nTL C 12
Setpoint to upper TC TUSP 370.8 K
Setpoint to lower TC TLSP 372.4 K
Final concentration of the side product xW b 0.965 mol/mol
The charge to the process is divided in two equal fractions. A fraction is loaded into
the reboiler vessel and the other into the vessel connected to the head of the column,
as shown in Figure 5.3 (i). A constant heat duty is supplied to the reboiler, bringing
the liquid to its boiling point. The vapour thus produced rises in the column and is
condensed in the condenser at the column’s head. The condensate flows to the top
Charge
LC Dw TC . .
LU VU
.
W
TC . .
Charge LL VL
Bw
(i) Start-up (ii) Production step
Figure 5.3.: Process steps: (i) start-up (close), and (ii) production (open).
22
1.0
ethanol
mol
mol
Concentration in
0.6
0.4
0.2
0.0
1.0
mol 1-propanol
mol
0.6
Concentration of
0.4
0.2
0.0
1.0
1-butanol
mol
mol
Concentration in
sump vessel xB
0.6
0.4
0.2
0.0
0.0 0.2 0.4 0.6 0.8 1.0
Figure 5.4.: Concentration path in the distillate vessel, the withdrawal stream, and in
the sump vessel.
23
vessel, where it is mixed with the charge. The liquid outlet of this vessel thus provides
the reflux to the column. During the start-up (close) phase, no product is withdrawn
from the column; hence the level in the two vessels is kept constant. During the closed
operation phase, the light-boiling component depletes from the lower vessel, while the
heavy-boiling component depletes from the upper vessel. At the end of this process
step, the concentration of the light boiler in the reboiler vessel and that of the heavy
boiler in the top vessel reach very low values. This can be seen in Figure 5.4, which
shows the concentration path in the top and bottom vessels, and on the withdrawal
stage.
At this point, somewhere in the column, the concentration of the middle-boiling
component reaches a maximum, which is close to one. In order to obtain the maximum
possible concentration of the intermediate boiler, the withdrawal stream (Ẇ ) is placed
close to this stage. During this operation step, which is also referred to as production
or open phase, the liquid flow rates to the two column sections are used to control
the temperatures in the respective sections, Figure 5.3 (ii), hence determining the
concentration of the liquid in the withdrawal stream. Furthermore, the concentrations
of the light and heavy-boiling components increase in the top and bottom vessels,
respectively.
The concentration profile in the liquid phase at two different times are shown in
Figure 5.5. The equilibrium stages are here numbered from bottom to top. It can be
seen that ethanol (light-boiling component) is only present on the upper stages of the
column. Analogously, 1-butanol is only present in the lower column section. On the
other hand, the concentration of 1-propanol is high throughout the column, showing
1
1-propanol
Stage number nth
7
ethanol
t = 0.4
1-butanol t = 0.9
14
21
mol
0.0 0.2 0.4 0.6 1.0
mol
Concentration in
the liquid phase
Figure 5.5.: Liquid concentration profile along the column at τ = 0.4 (full lines) and
τ = 0.9 (dashed lines). The stages are numbered from top to bottom.
24
a flattened maximum in the middle of the column. The upper (stripping) column
section is therefore needed to purify the intermediate-boiling component from the
light boiler; and the separation between the intermediate and the heavy boiler occurs
in the lower (rectifying) column section. Hence, manipulating the flow rates of the
reflux streams to the upper and lower column sections controls the temperature profile
in the corresponding sections, and therefore the purity of the withdrawal stream. In
other words, the internal reflux ratio of the upper column section (L̇/V̇ )U determines
the concentration of light boiler in the product stream; and the internal reflux ratio
of the lower section (L̇/V̇ )L determines the concentration of the heavy boiler in the
product stream. The internal reflux ratios are related to the flow rate of the product
stream (Ẇ ) through the mass balance around the withdrawal stage:
Ẇ = L̇U − L̇L
! ! !
L̇ L̇
= V̇ · − (5.1)
V̇ U
V̇ L
The internal reflux ratios necessary to provide the required purity of the side prod-
uct are shown in Figure 5.6. During the closed operation phase, when no product is
withdrawn from the column (rel. time ≤ 0.38), the column is operated under total
reflux conditions: (L̇/V̇ )L = (L̇/V̇ )U = 1. During the production phase, the internal
reflux ratios are manipulated in the control loops of the two column sections. The
flow rate of the product stream is initially (0.38 ≤ rel. time ≤ 0.60) very high. This
is due to the high concentrations of the intermediate boiler in the two vessels. As
2.0
.
Internal reflux ratio (L /V (
1.6
((
L
. .
.
V U
1.2
0.8
.
(( L
V
.
L
0.4
0.0 0.2 0.4 0.6 0.8 1.0
Figure 5.6.: Internal reflux ratios for the upper (L̇/V̇ )U and lower (L̇/V̇ )L column
sections.
25
5.3. Parametric studies
these concentrations decrease, the separations in both column sections become more
difficult, and therefore the internal reflux ratios approach one. Hence, the flow rate of
the product stream decreases. Towards the end of the process, the concentrations of
1-propanol in the upper and lower parts of the column are very small. The difference
between the internal reflux ratios decreases, drastically reducing the flow rate of the
product stream. At the end of the process, the reflux ratios tend towards one and
the flow rate of the product stream towards zero:
lim (L̇/V̇ )L = lim (L̇/V̇ )U = 1 ⇒ lim Ẇ = 0 (5.2)

τ →1 τ →1 τ →1
A further purification of the light and heavy fractions can be achieved by taking an
off-cut. This could be added to the charge of the following batch.

The analysis of the process parameters is performed by systematically varying one
parameter at a time, and studying the changes on the process outputs. The results
of this analysis are presented and discussed in the following paragraphs. These are
obtained under the assumption that the liquid phase behaves ideally, and that the
relative volatilities are constant over the temperature range of the column. The
physical properties of the ideal liquid phase are chosen to represent average values for
the alcoholic mixture analysed in the previous section. The most important constant
parameters applied in the simulation are summarised in Table 5.2.
Table 5.2.: Simulation parameters for the process described in Section 5.3.
Molar fractions in the charge (xa , xb , xc ) (0.1, 0.8, 0.1)

Relative volatilities αab 2.0
αbc 2.0
Position of upper TC nTU C 6
Position of lower TC nTL C 10
Setpoint to upper TC TUSP 374 K
Setpoint to lower TC TLSP 379 K
26
5.3.1. Geometric parameters

The geometric parameters influencing the separation capability of the process are the
number of equilibrium stages and the position of the withdrawal stage. These are
usually chosen during the design stage of the column, and the operator does not have
the possibility to change them during operation. Other geometric parameters, such as
column internals and diameter, are not considered here, since these have no influence
on the thermodynamic study based on the equilibrium stage model.
Number of equilibrium stages

In this investigation, the total number of stages of the column is varied, while the
other parameters are kept constant. The withdrawal stage is placed in the middle
of the column so that the two column sections have the same number of equilibrium
stages. Furthermore, the column is operated in closed loop and the relative position
of the controlled temperatures with respect to the withdrawal stage is maintained
unvaried. The set-points to the controllers are also kept constant. In this way, the
concentration profile around the withdrawal stage is maintained constant throughout
the investigation, and the composition of the withdrawal stream is independent of the
total number of stages.
With increasing number of equilibrium stages, lower reflux ratios are required to
achieve the required purity of the intermediate-boiling component. Hence, with in-
creasing number of stages, the flow rate of the withdrawal stream increases, and in
turn, the batch time decreases. Similarly to conventional batch distillation, the re-
quired concentration of the intermediate boiler can be reached only if the number of
stages per column section is higher than the minimum number of stages. This, in
turn, depends on the concentrations of the intermediate boiler in the top and bottom
vessels. Therefore, the concentrations of the intermediate boiler in the top and bot-
tom vessels at which the process becomes infeasible, decrease with increasing number
of stages. The recovery of the intermediate boiler and the purities of the light and
heavy-boiling components both increase with increasing number of stages.
Position of withdrawal stage

The position of the withdrawal stage determines the relative size of the two column
sections. Hence, the optimal position of the withdrawal tray depends on the relative
volatility of the components being separated in the corresponding column section.
This means that the size of the upper column section is determined by the rela-
tive volatility between the light and the intermediate-boiling components (αab ), and
similarly, the size of the lower column section is determined by that between the inter-
mediate and heavy boilers (αbc ). If, for instance, the separation of the intermediate
boiler from the low boiler is more difficult than that from the high boiler, then a
longer upper column section would be required and the position of the withdrawal
27
tray would shift downwards. In the case that the two separations are similar, then
the best position for the withdrawal point would be the middle of the column.
5.3.2. Operational parameters

The column presents several operational parameters. Usually, the operator has access
to some or all of these parameters during the operation of the process. Therefore, the
operational parameters can be varied to reach the required separation. Among them,
we find the composition of the charge, the reflux ratios, and the termination of the
closed operation step.
Composition of the charge

To study the effect of the composition of the charge, equal amounts of feeds of dif-
ferent compositions are processed in the same column operated in closed loop. The
separation task, i.e. the required composition of the intermediate-boiling fraction, is
also unchanged throughout the investigation. The effect of the composition of the
intermediate-boiling component on the duration of the process, and therefore on the
energy demand, is shown in Figure 5.7. The duration of the process goes through a
maximum. On the left-hand side of this maximum, the recovery of the intermediate
boiler in the side product is very low. This is due to the fact that, for the given number
of stages, the separation becomes infeasible at lower product recoveries. Therefore,
the shorter duration of the process is not representative of reality.
100
entire process
%
80
Relative duration
60
of the process
open operation mode

40
20
close operation mode

0
mol
0.2 0.4 0.6 1.0
mol
Concentration of intermediate
boiling component in the charge
Figure 5.7.: Relative duration of the process versus the concentration of the interme-
diate boiler in the charge.
28
On the right-hand side of the maximum, the duration of the process decreases with
increasing concentration of the intermediate-boiling component in the feed. This can
be explained with the aid of Figure 5.8, which is a plot of the internal reflux ratios
against the relative process time, with the concentration of the intermediate boiler
in the charge (xb, α ) as parameter. For high concentrations of the intermediate boiler
in the charge, the internal reflux ratios differ largely from one. This corresponds
to a high production rate of the intermediate-boiling fraction. Therefore, at high
concentrations of intermediate-boiling component, it is possible to withdraw a larger
amount of product at higher flow rates. In their comparison of the inverted and
regular batch distillation columns, Sørensen and Skogestad [1996] find similar results.
They demonstrate that when the mixture to be separated is richer in the light-boiling
component, a regular batch distillation column is time-saving when compared to an
inverted batch distillation column.
2.0 x b,a =0.8
x b,a =0.6
x b,a =0.2
Internal reflux ratio (L /V (
1.6
. .
.
(( L
V
.
U
1.2
0.8 .
((
L
V
.
L
0.4
0.0 0.2 0.4 0.6 0.8 1.0
Figure 5.8.: Effect of the composition of the intermediate-boiling component in the

charge (xb, α ) on the internal reflux ratios in the upper and lower column
sections.
It can be further observed that the duration of the initial separation step also
decreases with increasing concentration of intermediate boiler in the charge. During
this phase the column is operated at total reflux. The duration of this step is therefore
determined only by the amount of components that has to be transferred between
the two vessels, and by the termination criterion.
Termination criteria for the first process step

Increasing the duration of the first process step reduces the concentration of the light-
boiling component in the sump, and that of the heavy boiler in the top vessel. Hence,
29
with increasing duration of this process step, the concentration of the intermediate-
boiling component at the withdrawal point increases. This leads to: lower internal
reflux ratios at the beginning of the second process step, higher feasible concentration
of the intermediate-boiling fraction, and higher recoveries of the light and heavy-
boiling fractions. Nevertheless, long times are required to obtain a pure intermediate-
boiling component on the withdrawal stage. Therefore, the best termination criterion
should be determined by process optimisation.
Our subsequent investigations are limited to the case that the feed is much richer
in the intermediate boiler than in light and heavy boilers. In such cases, the relative
content of the extreme boilers plays a minor role and influences mainly the duration
of the start-up of the process i.e. the closed operation mode.
Effect of the internal reflux ratios

In the first process step, the light and heavy-boiling components are separated under
total reflux conditions. The analysis of the effect of different reflux ratios can therefore
be limited to the second process step. During this study, it is assumed that the
initial concentration of the light boiler in the sump, and that of the heavy boiler in
the distillate vessel are both zero. The study is limited to the second process step.
During this phase, the internal reflux ratios are set to constant values. The simulation
is interrupted when one of the vessels is empty.
In Figure 5.9 , the concentration of the intermediate boiler in the withdrawal stream
(xW b ) and its molar recovery (σb ) are plotted against the relative distillation time for
1.0 1.0
mol x Wb mol
Molar fraction of intermediate
mol mol
Recovery of intermediate
Wb
0.6 0.6
boiler in side stream x
sb
0.4 0.4
. . . .
(L /V ( = 0.9;(L /V ( = 1.1
0.2 . .L . .U 0.2
boiler sb
(L /V ( = 0.6;(L /V ( = 1.4
. .L . .U
(L /V ( = 0.5;(L /V ( = 1.5
L U
0.0 0.0
0.0 0.2 0.4 0.6 0.8 1.0
Figure 5.9.: Concentration of the side product (full lines) and molar recovery (dashed
lines) plotted versus the relative distillation time, for different internal
reflux ratios.
30
5.4. Comparison with the middle vessel batch distillation column
different values of internal reflux ratios. This plot shows that to achieve high recovery
and high purity in the intermediate product, high reflux ratios are required. On the
other hand, increasing the reflux ratios increases the duration of the process. However,
at the end of the process, the applied constant internal reflux ratios are not able to
maintain the purity of the product.
Additionally, to study the interactions between the two column sections, the inter-
nal reflux ratio of the upper column section (L̇/V̇ )U is varied while the internal reflux
ratio in the lower section (L̇/V̇ )L is kept constant. It is observed that the concentra-
tion of high-boiling impurity in the side product increases as the internal reflux ratio
in the upper column section (L̇/V̇ )U decreases. The concentration of the low boiler
in the withdrawal stream changes very slightly. Furthermore, the concentration of
the heavy boiler in the upper column section considerably increases with decreasing
(L̇/V̇ )U , indicating a strong interaction between the upper and lower column sections.
5.4. Comparison with the middle vessel batch

distillation column
Among others, Warter et al. [2002] and Skouras and Skogestad [2004b], analysed the
batch distillation column with a middle vessel for the separation of zeotropic ternary
mixtures. They state that, in some cases, the middle vessel batch distillation column
might present important advantages over the regular batch distillation column. An
extremely important advantage of this column is that the components in the feed are
subject to a lower thermal stress. This is due to the possibility of keeping the charge
cooled in the middle vessel, hence allowing separations that would otherwise be im-
possible to carry out in a conventional batch distillation column. Further advantages
are the simpler process control and handling of liquid fractions; higher flexibility; and
lower energy demand [Warter 2001].
Here, the batch distillation column with side withdrawal is compared with the
middle vessel batch distillation column. The middle vessel batch distillation column,
shown in Figure 5.10, consists of two column sections connected by the middle vessel.
The upper column section functions as a regular batch column, while the lower column
section resembles an inverted column. The low-boiling component is separated from
the intermediate boiler in a rectifying section (upper section). The heavy-boiling
component is simultaneously separated from the intermediate boiler in the inverted
column (lower section). On the other hand, in Figure 5.1 it has been shown that
the batch distillation column with side withdrawal can be visualised as an inverted
batch distillation column placed on top of a regular batch distillation column. Here,
the separation between the intermediate and the light-boiling components is carried
out in an inverted column. The separation between the intermediate and the heavy-
boiling components is simultaneously carried out in a regular column. This structural
difference between the two columns is the base for their different behaviour.
31
.
D
Rectifying
section
Charge
Mw
Stripping
section
.
B
Figure 5.10.: Batch distillation column with a middle vessel.
The duration of the separation task carried out in the two columns is plotted in
Figure 5.11 as a function of the concentration of the intermediate boiler in the charge
and of the relative volatilities αab and αbc , under the condition that these are equal
(αab = αbc ). The total number of stages, the final purity of the intermediate-boiling
fraction, and the recovery of the middle-boiling component are the same in both
processes. Both processes are operated with a constant heat duty. Furthermore, both
the middle vessel and the side withdrawal stream are placed in the middle of the
column. The simulation data is summarized in Table 5.3.
Table 5.3.: Simulation parameters for Section 5.4.

Position of middle vessel nM 14
Concentration of middle boiler in product:
αab = αbc = 1.50 xW b ω 0.96
xM b ω 0.96
αab = αbc = 2.0 xW b ω 0.98
xM b ω 0.98
αab = αbc = 2.5 xW b ω 0.98
xM b ω 0.98
32
1.0
Relative duration of the process

0.8
middle vessel
0.6
0.4
0.2
side withdrawal
0.0
1.5
aa 2.0 mol 0.9
b ab 0.7 0.8
c 2.5 0.5 0.6 mol
of the
Concentration
iler
intermediate bo
in the char ge
Figure 5.11.: Comparison of the relative duration of a batch distillation process in a

middle vessel with that in a column with a side withdrawal.
For very difficult separations (low relative volatilities), the process with the side
withdrawal requires shorter time, and therefore it has a lower energy demand, than
that in the middle vessel column. On the other hand, for higher values of the relative
volatilities, this is true only for mixtures that are very rich in the intermediate-boiling
component.
Furthermore, in the middle vessel column, the two column sections are separated
from each other by a huge holdup, and therefore the two sections are dynamically
decoupled [Farschman and Diwekar 1998]. Unfortunately, this is not the case for
the column with side withdrawal. The vapour passes directly from the lower to the
upper column section, and the liquid phase flows directly from the upper to the lower
column section. This introduces strong interactions between the two column sections.
The design of the two control loops therefore has a higher degree of complexity.
Nevertheless, if the required high purity product is the intermediate boiler, the column
with side withdrawal allows for direct control of its purity, whereas in the middle vessel
column, it is only possible to control the purity of the light and heavy components
(impurities). The purity of the intermediate-boiling fraction therefore results from
the extent of the distillation.
33
5.5. Experimental validation of the process

The batch distillation process with a side withdrawal was experimentally investigated
in collaboration with Degussa AG. The aim of the experiments was to validate the
operating procedure described in Section 5.2.
5.5.1. Apparatus
A simplified flow sheet of the experimental apparatus is shown in Figure 5.12. This
is a packed distillation column having an internal diameter of 50 mm. The column is
divided into three packed sections (S1, S2 and S3), each with a height of 1 m. It is
possible to feed and withdraw a liquid stream above each column section. The vapour
stream cannot leave the column, it therefore flows directly from each section to the
one placed directly above. The lowest column section (S3) is placed above the reboiler
(B3). This is a 20 l vessel provided with a spiral heat exchanger (W3), heated via a
3 kW thermostat. The uppermost section is connected to a vertical spiral condenser
(W1). By operating valve V1, the condensate can either be fed back to the column,
or be withdrawn to the stirred tank (B1) placed close to the head of the column.
Liquid can be pumped (P1) back from this vessel to the column’s head. This liquid
stream passes through a heat exchanger (W2), where it is pre-heated before being fed
back to the column. This avoids wild condensation of the rising vapour.
The temperature is measured at eleven points (TR2-TR12) along the column, at
the condenser (TR1), in the reboiler (TR13), and in the liquid streams leaving and
entering the column. The flow rate of the liquid stream leaving vessel B1 and the mass
of liquid accumulated in the vessels B1 and B2 are also measured. Liquid samples are
taken from the liquid streams entering and leaving the column, and from the reboiler
vessel.
The individual column sections, the flanges connecting the various column parts
and the reboiler are all insulated against thermal losses. Furthermore, the column
can be ventilated with an inert gas, such as nitrogen. This gas leaves the top of
the column via a vacuum pump (not shown in the flow sheet), which allows for the
regulation of the operating pressure at the condenser.
5.5.2. Experimental procedure

The total charge has a mass of 10.4 kg, and a composition of 0.101 mol/mol ethanol,
0.798 mol/mol 1-propanol, and 0.101 mol/mol 1-butanol. The charge is divided in two
equal parts. One part is charged to the reboiler (B3), and the second one is loaded
into the distillate vessel (B1) connected to the head of the column. The liquid in the
reboiler is heated with a constant heating load of approximately 2.7 kW. When the
mixture in the reboiler reaches its boiling point, vapour starts to rise along the column.
The temperatures in the column therefore rise from the bottom upwards. The rising
vapour finally reaches the condenser (W1), where it is condensed. The condensate
34
W1
V1
TR
1
B1
TR
P1 W2 2
TR
S1 3
WIC FIC
1 1
TR
TIC 4
5
V2
TR
6
TR
S2 7
TIC TR
10 8
TR
B2 V3 9
S3 TR
WR 11
2
TR
12
TR
B3 13
W3
Figure 5.12.: Simplified flow sheet of the experimental apparatus.
35
is sent to the top vessel (B1). At this point, liquid is fed back from this vessel to
the column, keeping the mass of liquid in the vessel constant (WIC1). This feed is
pre-heated in W2 before entering the column. The liquid flowing down the column is
passed from each section to its underlying one, without any being withdrawn. Enough
time is allowed for the temperature profile along the column to reach steady state
conditions. An offline analysis of the liquid in the top vessel and in the reboiler shows
that the light boiler has depleted from the reboiler, and the heavy boiler is no longer
present in the top vessel.
At this point, the control loop of the upper column section is changed to tempera-
ture control mode, i.e. WIC1 is deactivated, and TIC5 is activated. Simultaneously,
the temperature control loop of the lower column section (TIC10) is closed. The
upper control loop regulates the temperature in the upper column section by manip-
ulating the flow rate of the liquid fed from the top vessel to the column. On the other
hand, the lower temperature control loop manipulates the reflux ratio in the lower
column section. This is achieved by operating valve V3, placed between the column
sections S2 and S3.
5.5.3. Temperature and concentration profiles

A better understanding of the process and the operating procedure can be achieved
by following the temperature and the concentration profiles shown in Figure 5.13 and
Figure 5.14, respectively. The profiles are plotted starting from the point when the
first vapour reached the condenser.
During the initial phase of the process, the total reflux operation, the temperatures
in the reboiler vessel (T13) and in the lower column section (T10-T12) rise slightly, to
reach steady state conditions. As the concentration profile clearly shows, during this
120
°C
T9
T13
100
Temperature profile
T5
along the column
90
T1
80
70
0.0 0.2 0.4 0.6 0.8 1.0
Figure 5.13.: Experimental temperature profile along the distillation column with a
side withdrawal.
36
1.0
mol
mol
Concentration in
ethanol
0.6
0.4
0.2
0.0
1.0
mol
mol
1-propanol
0.6
Concentration of
0.4
0.2
0.0
1.0
mol
mol
1-butanol
Concentration in
sump vessel xB
0.6
0.4
0.2
0.0
0.0 0.6 0.8 1.0
Figure 5.14.: Experimental concentration path in the distillate vessel (top), of the
withdrawal stream (middle) and in the sump (bottom) for the distillation
column with a side withdrawal.
37
process phase, the light boiler is transferred from the sump to the top vessel, and the
heavy boiler follows the opposite path from the top to the sump vessel. By the end of
this process phase (τ = 0.66), the temperature profile reaches steady state conditions,
and the concentration of the light boiler in the sump and that of the heavy boiler in
the top vessel reach very low values. This time is located via a real time analysis of
the liquid phase in the reboiler and of that in the top vessel. The concentration of the
intermediate boiler at the withdrawal point has, by this time, reached the required
value. The control loop in the upper column section is therefore turned from mass to
temperature control, and the temperature control loop of the lower column section
is simultaneously closed. The reference temperatures of the two column sections are
directed by the controller towards their set points. The liquid holdup in the top vessel
and in the sump start decreasing, and the product is collected in the side vessel (B2).
The concentration of the intermediate boiler is successfully maintained to the desired
value. The concentration of the light boiler in the top vessel gradually increases as
that of the intermediate boiler decreases. Analogously, the concentration of the heavy
boiler in the sump increases. This is accompanied by a decrease in the temperature
at the column’s head, and by a temperature increase in the sump. Towards the end
of the process, the temperature profile in the column stabilises once more (τ = 0.90),
and the flow rate of the withdrawal stream (Ẇ ) tends to zero.
The final concentrations and recoveries are given in Table 5.4. At the end of the
process, the intermediate boiler has been recovered with the desired concentration
(xB2ω = 0.92 mol/mol) and with a high recovery (σb = 0.98). Nevertheless, the
concentration and recovery of the light and heavy-boiling fractions are low.
Table 5.4.: Summary of experimental results.
Molar fraction of main product Recovery of main product

Top Vessel (B1) 0.92 0.57
Side Vessel (B2) 0.92 0.98
Sump (B3) 0.72 0.62
5.5.4. Conclusion
The simulated and experimental concentration paths of the withdrawal stream as well
as those in the top and bottom vessels are in qualitative agreement. Therefore, the
experimental investigation confirms the feasibility of process, and furthermore, the
practical realisation of the batch distillation process with side withdrawal is demon-
strated. Moreover, both the theoretical and the experimental investigations show
that this batch distillation process is adequate for the purification of an intermediate
boiler from light and heavy-boiling contaminants.
38
6. Integration of batch reaction with
batch distillation - Equilibrium
limited reactions
In recent years, there has been extensive research on process integration. Most of this
research has focussed on attempts to integrate reaction with separation steps for com-
plex continuous processes. In such cases, it might be possible to reduce very complex
flowsheets to a couple of units where reaction and separation occur simultaneously.
As has already been pointed out in the literature review, process integration for batch
distillation has not received the same attention paid to continuous processes. This
might be due to the fact that batch processes are usually applied in multipurpose
plants where the same reactor is used for the production of various chemicals. For
instance, in the pharmaceutical industry, flexibility of the batch units, hygiene and
sterilisation are of primary importance. Nevertheless, batch reactors are very often
directly coupled with a condenser or with a batch distillation column, as shown in
Figure 6.1.
.
D
Rectifying
section
Charge Charge
Batch Batch
reactor reactor
Bw Bw
(i) (ii)
Figure 6.1.: Combination of a batch reactor with (i) a condenser, and (ii) a rectifying
column.
39
6.1. Decomposition reaction
To this date, little attention has been paid by the academic literature to the con-
ceptual design of processes where a batch reactor is coupled with a batch distillation
column. In order to fill this gap, in the following sections, some of the alternative
processes in which a batch reactor is combined with a distillation unit are analysed
via computer aided simulations. A few illustratory reaction families are considered.
The various process alternatives are discussed for each family, and, where possible,
the processes are compared in order to give to the process designer tools to select the
appropriate process.
Every chemical reaction can be considered as an equilibrium reaction, where the
concentrations of the reactants and the products are in equilibrium. If enough time is
allowed for the reaction to reach equilibrium, the final concentration in the reactor is
limited by the value of the equilibrium constant (Keq ) which relates the concentrations
of the products (xp ) to that of the reactants (xr ):
Y
xνpp
p
Keq = Y (6.1)
xνr r
r
where νi indicates the stoichiometric coefficient of component i. In some cases, the

equilibrium constant might be big enough for the reaction to reach almost complete
conversion. On the other hand, for moderate values of the equilibrium constant, the
conversion is limited by the equilibrium curve, so that non-reacted educts are still
present in the reaction mixture even after infinite time. In this case, a separation
step is required to recover the reactants; these are usually allowed to react once more.
Le Chatelier’s principle tells us that if one of the products is removed from the re-
action environment, the reaction system shifts to re-establish equilibrium conditions,
i.e. it tends to produce more of the removed product. Therefore, integrating reac-
tion with separation yields higher conversions than those allowed by the equilibrium
relationship.

From the point of view of a distillation process, the boiling point order of the compo-
nents present in the reaction mixture is of primary importance. As a first example,
the case where an intermediate-boiling component (b) decomposes into a light boiler
(a) and into a heavy boiler (c) is considered.
νb b ⇋ νa a + νc c (6.2)
For this reaction type, there are many process alternatives which the designer should
consider. Since both products are extreme boilers, any one can be removed by a
distillation process, and therefore the chemical equilibrium is shifted to the product’s
40
side. The batch reactor can be alternatively coupled with a rectifying, a stripping,
or a middle vessel column, as shown in Figure 6.2. In the first case, the light-boiling
product (a) is recovered at the head of the column, and as the reaction proceeds, the
heavy boiler is enriched in the reactor placed at the sump. For the combination of
a batch reactor with a stripping column, the heavy-boiling product (c) is recovered
from the sump. As the reaction proceeds, the light boiler is enriched in the reactor
placed at the column’s head. In a middle vessel column, the reactor takes the place of
the middle vessel. The two products are simultaneously recovered from the top and
bottom of the column. The concentration path in the middle vessel is bound by the
equilibrium curve, and depends on the flow rates of the two product streams.
. .
D, a D, a
Charge, b Charge, b Charge, b
Batch Batch Batch

reactor reactor reactor
Bw,c Dw,a Mw
. .
B, c B, c
(i) (ii) (iii)
Figure 6.2.: Combination of a batch reactor with (i) a rectifying, (ii) a stripping, and
(iii) a middle vessel column.
6.1.1. Minimum energy demand

One tool available to the process designer to perform a pre-selection among the three
processes introduced in the previous paragraph, is the minimum energy demand of
the process. The calculation of the minimum energy demand requires knowledge of
the concentration in the still. For fast equilibrium reactions, the concentration in the
reactor always lies on the chemical equilibrium curve. To facilitate the calculation, the
41
concentration change in the reactor can be constructed as a series of infinitesimally

small distillation and reaction steps. This graphical construction is here explained
with the aid of Figure 6.3 for the reaction
2b ⇋a+c (6.3)
carried out in a batch reactor coupled with a rectification column having an infinite
number of stages. An infinitesimal distillate withdrawal, with the light-boiling compo-
nent (a) having a concentration xDa = 1, shifts the concentration in the reboiler from
xB along a straight line joining xD and xB . This is depicted in Figure 6.3 with (∆D).
The new concentration lies away from the chemical equilibrium. The infinitesimal
reaction step (∆R) brings it back to the chemical equilibrium curve, moving along
a stoichiometric line. Therefore, the concentration in the reboiler moves along the
chemical equilibrium curve away from the concentration of the distillate product, and
towards the heavy boiler (c) which is obtained at the end of the process in pure
form in the reactor vessel. The concentration path in a reactor coupled either with a
stripping column, or with a middle vessel column, is constructed in a similar way.
Light boiler (a)

xD
Chemical
equilibrium
pR K eq = 1
a
{
x
xB
Concentration
{ DD path in reactor
{
DR
High boiler (c) Middle boiler (b)

xb
Figure 6.3.: Infinitesimal construction of the concentration path in the batch reactor
coupled with a rectifying column for the Reaction (6.3).
The dimensionless minimum energy demand for a batch reactor coupled with a
rectifying, a middle vessel and to a stripping batch distillation column, is shown in
Figure 6.4 for various values of the relative volatilities. In the three cases, the chemical
equilibrium constant has a value of Keq = 1.0. The dimensionless minimum energy
demand is expressed as:
Qmin /(r · F ) (6.4)
42
with Qmin being the minimum energy demand of the process, r the latent heat of
vaporisation of the charge, and F the amount of charge.
The process where a reactor is coupled with a middle vessel column has the lowest
minimum energy demand for the whole conversion range, if the relative volatility of
the light boiler (a) to the intermediate boiler (b) is equal to that of the intermediate
boiler to the heavy boiler (αab = αbc ), Figure 6.4 (ii). The combination of a batch
reactor with a rectifying column requires the lowest minimum energy demand over a
wide range of conversions if the relative volatility αab is bigger than αbc , Figure 6.4 (i).
Finally, the combination of a batch reactor with a stripping column could be advan-
tageous if the separation of the heavy boiler from the intermediate boiler is easier
than that between light and intermediate boiler (αab < αbc ).
12
aab=2.0; abc=1.2 aab=1.2; abc=1.2 aab=1.2; abc=2.0
8
Qmin/(r*F)
Rectifying Stripping
column column
4
Middle vessel
column
0
0.0 0.2 0.4 0.6 0.8 1.0 0.0 0.2 0.4 0.6 0.8 1.0 0.0 0.2 0.4 0.6 0.8 1.0
Conversion Conversion Conversion

(i) (ii) (iii)
Figure 6.4.: Minimum energy demand for the combination of a batch reactor with a
rectifying, a middle vessel and a stripping column, for the Reaction (6.3),
with Keq = 1 and (i) αab = 2.0, αbc = 1.2,
(ii) αab = αbc = 1.2, and (iii) αab = 1.2, αbc = 2.0.
The stoichiometry of the reaction also influences the energy demand of the three
processes. This is illustrated in Figure 6.5. Here, three equimolar reactions are taken
into consideration. The minimum energy demand of a batch reactor coupled with
a rectifying column decreases with increasing stoichiometric coefficient of the light-
boiling component, νa . This appears to be in contradiction with the fact that by
increasing the stoichiometric coefficient of the light boiler, the amount of distillate
to be recovered increases. On the other hand, the higher concentration of the light-
boiler in the reactor allows for lower reflux ratios to be applied, therefore decreasing
the energy demand. In the case that the stoichiometric coefficient of the light boiler is
bigger than that of the heavy boiler (νa > νc ), this process presents a lower minimum
energy demand than the combination of a batch reactor with a middle vessel column,
also for very high conversions, as shown in Figure 6.5 (i). Analogously, the minimum
43
12
3b 2a+c 2b a+c 3b a+ 2c
8
Qmin/(r*F)
Stripping
Rectifying Middle vessel column
column column
4
0
0.0 0.2 0.4 0.6 0.8 1.0 0.0 0.2 0.4 0.6 0.8 1.0 0.0 0.2 0.4 0.6 0.8 1.0
Conversion Conversion Conversion

(i) (ii) (iii)
Figure 6.5.: Minimum energy demand for the combination of a batch reactor with a
rectifying, a middle vessel and a stripping column, for the reactions
(i) 3b ⇋ 2a + c, (ii) 2b ⇋ a + c, and (iii) 3b ⇋ a + 2c, with Keq = 1,
αab = 2.0, and αbc = 1.2.
energy demand of a batch reactor coupled with a stripping column decreases with
increasing stoichiometric coefficient of the heavy boiler.
6.1.2. Combining a reactor with a rectifying batch distillation

column
In this section a reactor is coupled with a rectifying batch distillation column as shown
in Figure 6.6 (i). The decomposition Reaction (6.3):
2b ⇋a+c (6.5)
takes place in the liquid phase within the reactor. The rectifying column has 30 equi-
librium stages, and the concentration of the distillate product is maintained constant
at xDa = 0.98. The relative volatilities are αab = 2.0 and αbc = 1.5. The concen-
tration profile in the reactor is shown in Figure 6.6 (ii). This has been obtained via
rigorous simulation of the process, and compares very well with the concentration
path shown in Figure 6.3 obtained via a geometrical construction. Every symbol in
Figure 6.6 (ii) represents a time unit. At the beginning of the process, the fast de-
composition reaction moves the concentration in the reactor from its initial point xBα
along a stoichiometric line to the equilibrium curve. Since the light boiler is removed
as distillate product, the concentration in the reactor moves along the chemical equi-
librium curve away from the light-boiling component, and towards the heavy boiler
(c). At the end of the process, the reactor contains the heavy-boiling product at
44
Light boiler (a)

xD
.
D, a
Chemical
equilibrium
K eq = 1
. .
L V
a
x
Charge, b
Concentration
path in reactor
Batch
reactor x Bw
x Ba
Bw,c xb
(i) (ii)
Figure 6.6.: Batch reactor coupled with a rectifying column (i) process scheme, and
(ii) concentration path in the reactor for the Reaction (6.5).
the required purity. The reflux ratio (RL ), Equation (6.6a), necessary to maintain
the required distillate purity, and the recoveries of the two reaction products (σDa
and σBc ), defined in Equations (6.6b) and (6.6c), are plotted as a function of the
relative distillation time in Figure 6.7 (i). It is evident that the last 10% recovery of
the products consumes 80% of the total distillation time, and therefore of the total
energy demand. This is clearly due to the very high reflux ratios required at the low
concentrations of light boiler in the sump of the column during most of the process
time.
L̇
RL = (6.6a)
Ḋ
νb · Da
σDa = (6.6b)
νa · Bb, α
νb · Bc
σBc = (6.6c)
νc · Bb, α
It is therefore attempted to improve the process by imposing a maximum limit on

the reflux ratio. When the reflux ratio reaches its maximum value (RL, max = 50),
a low purity off-cut is taken at a lower reflux ratio (RL, of f = 10). The off-cut is
necessary to purify the heavy-boiling product from the non-reacted educt still present
in the sump. With two such simple moves, the process time is reduced by 56%,
but the recovery of the heavy-boiling product also decreases to σBc = 85%, since
some unreacted educt is lost with the off-cut. As shown in Figure 6.7 (ii), during
45
1.0 1500 1.0 50

sBc
sBc
mol 1200 mol 40
mol mol
Recovery s
Recovery s
Reflux ratio
Reflux ratio
0.6 900 0.6 30
sDa
sDa
0.4 600 0.4 Reflux ratio 20
Reflux ratio
0.2 300 0.2 10
0.0 0 0.0 0
0.0 0.2 0.4 0.6 0.8 1.0 0.0 0.2 0.4 0.6 0.8 1.0
Relative time t/t tot Relative time t/t tot
(i) (ii)
Figure 6.7.: Recovery of the light and heavy boilers, and reflux ratio for the Reac-
tion (6.5) carried out in a batch reactor coupled with a rectifying column:
(i) for the process without off-cut, and (ii) with off-cut.
the withdrawal of the off-cut, there is a decrease in the recovery of the heavy-boiling
component, because some is lost with the off-cut. This phase is nevertheless necessary
to reach the required high concentration of the sump product.
1.0 30
sBc
mol 24
mol
Recovery s
Reflux ratio
0.6 18
sDa
0.4 12
Reflux ratio
0.2 6
0.0 0
0.0 0.2 0.4 0.6 0.8 1.0
Figure 6.8.: Recovery of the light and heavy boilers, and reflux ratio for the Reac-
tion (6.5) carried out in a batch reactor coupled with a rectifying column:
optimised process.
46
In the two operating procedures described in the previous paragraphs, it is necessary

to continuously adjust the reflux ratio in order to maintain the necessary distillate
composition. Alternatively, the reflux ratio can be forced to follow a stepwise profile,
i.e. it is maintained constant for given time intervals. An example of this reflux policy
can be found in Li et al. [1998b]. They state that such a stepwise reflux profile is in
practice easier to implement than a continuously changing reflux. Figure 6.8 shows the
recoveries of the two reaction products for the modified process where a stepwise reflux
ratio policy is used. The duration of the process is minimised under the conditions
that the final purity of the distillate product is bigger than xDa, omega > 0.97 mol/mol,
and its recovery higher than σDa > 95 %. Furthermore, the reflux ratio is not allowed
to change before a minimum time interval of 240 s has elapsed. Here, the gOPT tool
of gPROMS [gPROMS 2003] is used.
The most relevant simulation data for the three processes presented in this section
is summarised in Table 6.1.
Table 6.1.: Simulation data for the processes presented in Section 6.1.2.
Molar fractions in the charge xBb, α 1.0 mol/mol

αbc 1.5
Process with off-cut
Maximum reflux ratio RL, max 50
Reflux ratio during off-cut RL, of f 10
Optimised process
Minimum final purity of distillate product xDa, omega > 0.97mol/mol
Minimum recovery of light boiler σDa, min 95 %
Minimum interval at constant reflux 240 s
Optimisation objective Qmin
Comparison
The simulation results for the three processes are summarised in Table 6.2. The
process without off-cut, where the reflux ratio is allowed to increase to arbitrary
values requires the longest time, and therefore has the highest energy demand. The
duration of the process with the off-cut and that of the optimised process are much
shorter. These two processes have a time and energy saving of 57% and 64% with
respect to the first process, respectively. However, the major time savings of these
processes are paid for by a slight decrease in either the purity or the recovery of
one of the components. In the optimised process, the recovery of the heavy boiler is
increased at the expense of its purity. This is due to the fact that some non-reacted
47
educt is still present in the sump of the column at the end of the process. Nevertheless,
the optimised process is easier to implement in the practical case due to the simpler
stepwise profile of the reflux ratio.
Table 6.2.: Comparison of the three processes.
Process without off-cut Process with off-cut Optimised process

τ 1.00 0.43 0.36
σDa 0.99 0.98 0.95
σBc 0.97 0.86 0.94
xDa, ω 0.98 0.98 0.98
xBc, ω 0.98 0.95 0.90
6.1.3. Combining a reactor with a batch distillation in a middle

vessel column
The two products of the decomposition Reaction (6.3):
2b ⇋a+c (6.7)
are the light and heavy boilers in the mixture. This reaction can be carried out in a
batch reactor coupled with a middle vessel column as shown in Figure 6.9 (i). The
reaction educt (b) is loaded into the reactor at the beginning of the process. The
light and heavy-boiling reaction products are simultaneously recovered as overhead
and bottom’s products, respectively. The concentration path in the reactor is shown
together with the concentration of the product streams and the equilibrium curve
in Figure 6.9 (ii). The relative volatilities of the components in the mixture are
αab = αbc = 1.5. The reactor is therefore placed in the middle of the column. In
this case, the concentration in the reactor vessel does not vary much during the
whole duration of the process. Thereby the temperature and concentration profiles
are almost stationary, thus considerably simplifying the design of the temperature
control. The reflux (RL = L̇U /Ḋ) and reboil (RB = V̇L /Ḃ) ratios, and the recoveries
of the two products (σDa and σBc ) are plotted against the relative distillation time
in Figure 6.10. The recoveries reach very high values at the end of the process, and
they move along two paths lying very close to each other. This is due to the fact that
the flow rates of the two product streams (Ḋ and Ḃ) are similar. Furthermore, the
reflux and reboil ratios are predominantly constant, and also take on similar values
throughout the process, confirming the near to steady state character of this batch
process.
The most relevant simulation data for the batch reactor coupled with a middle
vessel column is summarised in Table 6.3.
48
.
D, a
Light boiler (a)

xD
. .
LU VU
Chemical
equilibrium
Charge, b K eq = 1
a
x
Batch
reactor
Mw
. . Concentration
LL VL path in reactor
xB x Ma

xb
.
B, c
(i) (ii)
Figure 6.9.: Batch reactor coupled with a middle vessel column (i) process scheme,
and (ii) concentration path in the reactor for the Reaction (6.7).
1.0 25
mol 20
Reflux and reboil ratio
mol
sDa
Recovery s
0.6 Reboil ratio 15
0.4 10
sBc Reflux ratio

0.2 5
0.0 0
0.0 0.2 0.4 0.6 0.8 1.0
Figure 6.10.: Recovery of the light and heavy boilers; reflux and reboil ratios for the
Reaction (6.7) carried out in a batch reactor coupled with a middle vessel
column.
49
Table 6.3.: Simulation data for the processes presented in Section 6.1.2.
Molar fractions in the charge xM b, α 1.0 mol/mol

Number of stages in upper column section nth, U 15
αbc 1.5
Average concentration of distillate product x̄Da 0.99 mol/mol
Average concentration of bottom product x̄Bc 0.99 mol/mol
6.1.4. Closed processes

In a closed process, the reaction/distillation system can be considered to be a closed
system since no product stream leaves the process. Two such processes are described
in this section. In the first process, two batch reactors are placed at the top and
the bottom of a distillation column. This is shown in Figure 6.11 (i). In the second
process, a batch reactor replaces the middle vessel of a middle vessel column operated
in a closed mode. This process is shown in Figure 6.13 (i).
Charge, b
Light boiler (a)
x Dw
Batch Concentration
reactor path in upper
reactor
Dw,a
TC
a
x
. .
L V
Concentration
Charge, b path in bottom Chemical
reactor equilibrium
K eq = 1 x Da
Batch x Bw
x Ba
reactor
xb
Bw,c
(i) (ii)
Figure 6.11.: Closed operation of a multi-vessel batch distillation column with two re-
actors: (i) process scheme, and (ii) concentration path in both reactors.
50
Closed operation of a batch distillation column coupled with two reactors
In the process show in Figure 6.11 (i) two reactors are coupled with a batch distillation
column operated in closed mode. One is placed at the top and the other at the bottom
of the distillation column. The educt (b) is charged into both reactors, where it is
decomposed to the light (a) and heavy boiler (c). Component a leaves the lowest
reactor as it is formed, and accumulates in the top reactor. Analogously, component
c is enriched in the lowest reactor. In both reactors, the concentration path follows the
chemical equilibrium line. However, as shown in Figure 6.11 (ii), the concentration in
the top reactor moves towards the pure light boiler, and simultaneously, that in the
lower reactor migrates towards the pure heavy boiler.
The flow rate of the reflux stream from the top reactor (L̇) is used to control
the temperature in the distillation column. The stoichiometric coefficients determine
the shape of the equilibrium curve, and therefore the concentration path in the two
reactors. The value of the internal reflux ratio of the column, defined as L̇/V̇ , therefore
depends both on the relative volatilities of the components in the mixture, as well as
on the stoichiometric coefficients of the reaction. The reflux ratios for three different
stoichiometries are plotted in Figure 6.12 (i) as a function of the conversion of the
reaction educt. The relative volatilities of the components, the total number of stages,
and the separation task are constant for the three cases. The charge is equally divided
between the two reactors.
1.6 1.6
1.4 1.4
3b a+3c
3b a+3c
Internal reflux ratio
Internal reflux ratio
1.2 1.2
3b a+c 3b a+c
1.0 1.0
0.8 0.8 3b 3a+c

3b 3a+c
0.6 0.6
0.6 0.7 0.8 0.9 1.0 0.6 0.7 0.8 0.9 1.0
Conversion X b Conversion X b
(i) (ii)
Figure 6.12.: Internal reflux ratios required by closed operation of multi-vessel batch
distillation column with two reactors for three different stoichiometries
when the charge is (i) equally distributed between the reactors, and (ii)
distributed according to Equation (6.8).
51
If the stoichiometric coefficients of the light and heavy-boiling reaction products

are equal, νc = νa , then both are produced in the same amount in the two reactors.
The vapour and liquid streams in the column are therefore approximately equal,
i.e. the column operates close to total reflux conditions. In this case, operation of
the column is very simple, and the maximum separation capability of the column is
used. However, the situation changes with varying stoichiometry. If for instance the
stoichiometric coefficient of the heavy boiler is bigger than that of the light boiler (νc >
νa ), then in both reactors less light boiler is formed than heavy-boiling component.
The column’s task is mainly to transfer the heavy boiler to the bottom reactor, and it
therefore assumes the character of a stripping column. Hence, the internal reflux ratio
is initially larger than one, i.e. L̇ > V̇ . For higher conversions, the concentration of
the light boiler in the bottom reactor and that of the heavy boiler in the top reactor
become similar, so that the internal reflux ratio tends towards one. If, on the other
hand, the stoichiometric coefficient of the light boiler is bigger than that of the heavy
boiler (νc < νa ), then the internal reflux ratio is initially smaller than one, i.e. L̇ < V̇ .
By distributing the charge to the two reactors according to the ratio of the stoichio-
metric coefficients of the light and heavy-boiling products, as given by Equation (6.8),
the internal reflux ratio takes values which are very close to unity, as shown in Fig-
ure 6.12 (ii). This is of great advantage because the flow rate of the reflux stream is
almost constant, significantly simplifying the control of the process, and, furthermore,
Table 6.4.: Simulation data for the processes shown in Figure 6.11 (i).

αbc 1.5
2b ⇋a+c
Final concentration of distillate product xDa, ω 0.98 mol/mol
Final concentration of bottom product xBc, ω 0.98 mol/mol
3b⇋ a+3c
3b⇋ 3a+c
3b ⇋a+c
52
the maximum separation capability of the distillation column is used.

Dα νa
= (6.8)
Bα νc
For an equimolar reaction (i.e. if the sum of the stoichiometric coefficients of the
reactants is equal to that of the products) and an appropriate charge distribution, the
process can be operated with total internal reflux ratio. Nevertheless, both purities
and recoveries are high. These depend only on the duration of the separation process.
The main constants and the separation tasks for the process shown in Figure 6.11 (i)
are reported in Table 6.4.
Closed operation of a batch reactor coupled with a middle vessel column

The process shown in Figure 6.13 (i), consists of a reactor placed between a rectifying
and a stripping column section, replacing the middle vessel of a middle vessel column.
The intermediate boiling educt (b) is charged into the reactor at the beginning of the
process. The reaction products leave the reactor, accumulate and enrich in the top
vessel and in the column’s sump. The internal reflux ratio of the rectifying section
(L̇/V̇ )U determines the purity of the light-boiling product, and similarly the internal
reflux ratio of the stripping section (L̇/V̇ )L determines the purity of the heavy-boiling
product in the sump. The concentration in the reactor is determined by the chemical
equilibrium curve, by the rate of removal of the two reaction products from the reactor,
and by the position of the reaction pole πR . If the internal reflux ratios of the two
sections (hence the rates of removal of the reaction products from the reactor) are
adjusted according to Equation (6.9), then the concentration in the reactor vessel
would be constant and it would lie at the intersection of the chemical equilibrium
curve with the straight line joining the vertex b of the triangular diagram and the
reaction pole πR . Equation (6.9) is derived in Appendix A.
! !
νa L̇ νc L̇
· + · =1 (6.9)
νa + νc V̇ L νa + νc V̇ U
This is shown in Figure 6.13 (ii) for the decomposition Reaction (6.3). Here, the
concentration of the three components in the reactor, and the recoveries of the two
products are plotted as a function of the relative distillation time. The initial charge to
the reactor is pure educt. A minimum amount of charge consisting of pure products is
loaded into the top and bottom vessels. The process is operated with constant internal
reflux ratios, related by Equation (6.9). During most of the duration of the process,
the concentration in the reactor is constant. Only at the end of the process, when the
holdup in the reactor is very small, and most of the products have been recovered,
does the concentration deviate from its initial value. However, this has no influence on
the concentrations of the top and bottom products, which are very high throughout
53
Dw,a 1.0 1.0

TC sBc
. .
LU VU mol mol
mol mol
Molar recovery s
in the reactor xM
Charge, b sDa
Concentration
0.6 0.6
x Ma
Batch
0.4 0.4
reactor
0.2 0.2
TC x Mb
. . x Mc
LL VL
0.0 0.0
0.0 0.2 0.4 0.6 0.8 1.0
Bw,c
(i) (ii)
Figure 6.13.: Closed operation of a batch reactor coupled with a middle vessel column:
(i) process scheme, and (ii) concentration path in the reactor, together
with the recoveries of the light and heavy-boiling products.
the process: xDa, ω > 0.99 and xBc, ω > 0.99. Similarly to the corresponding open
process, the concentration and temperature profiles in the column have steady state
behaviour, significantly simplifying the design of the control system. The dynamic
behaviour of the column is limited to the changing holdups in the reactor and in the
two vessels. The simulation parameters are summarised in Table 6.5.
6.1.5. Conclusion
In this section, several combinations of a reactor with batch distillation processes
have been illustrated for an equilibrium limited reaction where a light and a heavy
boiler are formed by chemical reaction. For this class of reactions, the minimum
energy demand of the processes, which can be calculated with short cut methods,
depends both on the volatilities of the components in the mixture, as well as on the
54
Table 6.5.: Simulation data for the processes shown in Figure 6.13.

Number of equilibrium stages in upper column section nth, U 15
αbc 1.5
Internal reflux ratio of the upper column section (L̇/V̇ )U 0.9
Internal reflux ratio of the lower column section (L̇/V̇ )L 1.1
stoichiometric coefficients.
The concentration in a reactor coupled with a rectifying batch distillation column
varies during the duration of the process. The reflux ratio necessary to maintain a
constant distillate composition follows an exponentially increasing profile. Therefore,
the control of the process is very complex. Two modified reflux policies have been
illustrated. In both cases, the simplified operation of the process leads to a slight loss
in product quality. The combination of a reactor with a stripping batch distillation
column is expected to show a similar behaviour.
The closed operation of a batch distillation column with two reactors allows for a
simplification of the operation of the process. However, the charge has to be appro-
priately distributed between the two reactors. If the reactants have different relative
volatilities, they would be separated from each other. Therefore, this process is suit-
able for decomposition reactions, and for close boiling reactants.
The concentration in a reactor combined with a batch distillation column with a
middle vessel can be maintained constant. In this case, the reflux and reboil ratios
necessary to maintain constant distillate and bottom’s concentrations do not change
much during the process. Therefore, the control and operation are considerably sim-
plified. Moreover, the open and closed operation modes of this process show little
differences. These processes are, in general, suitable for reactions where the products
are the light and heavy boilers of the system.
55
6.2. Production of a middle-boiling component

The processes discussed in the previous section are applicable in the case that one of
the reaction products is the extreme boiler of the reaction mixture. In the case of an
association reaction, where the product is a middle-boiling component:
νa a + νc c ⇋ νb b (6.10)
a new class of processes should be taken into consideration. These processes are
characterised by the continuous feed of one of the reactants. The second reactant is,
once again, charged into the reactor or to the charge vessel of the column. These
semi-batch processes are divided in two categories. In the first case, the reaction is
carried out in a semi-batch reactor that is coupled with a distillation column; in the
second group, the reaction is carried out in a reactive distillation column.
6.2.1. Semi-batch reactor coupled with a distillation column

The addition Reaction (6.11) is taken into consideration.
a+c⇋2b (6.11)
A semi-batch reactor is coupled with a rectifying column, as illustrated in Fig-

ure 6.14 (i). In this process, the heavy boiler (c) is charged into the reactor, while
the light-boiling reactant (a) is continuously fed into the reactor. These react to form
the middle boiler (b), which, under appropriate conditions, is recovered as distillate
product from the top of the column.
The presence of the feed stream to the reactor introduces further degrees of freedom
to the process. The effect of the dimentionless flow rate of the feed stream (Ḟ · r/Q̇)
on the concentration of the distillate is shown in figure 6.14 (ii), where the distillate
concentration (xD ) and the concentration in the sump (xB ) are plotted for various
flow rates of the feed stream. The flow rate of the distillate stream (Ḋ) is maintained
constant. On the one hand, if the feed rate were too big, there would be non-reacted
light boiling educt in the sump of the column, which would be preferentially removed
as distillate product. On the other hand, if Ḟ were too small, there would be too little
intermediate-boiling product formed, therefore its concentration in the sump would
be very small. The recovery of the intermediate boiler would be more difficult, and
the distillate product would be contaminated with the heavy-boiling reactant.
The ratio Ḟ /Ḋ has an optimal value. This can be found by considering the process
as a sequence of three infinitesimal steps:
a) infinitesimal feeding of the light-boiling component to the reactor;
b) reaction to re-establish chemical equilibrium conditions;
c) distillation of the middle-boiling component.
In the following, the mass balances for the three infinitesimal steps are written for
the addition Reaction (6.10), with νa = 1. The initial charge to the reactor consists
56
1.0 . .
. F . r Q = 0.053
D, b
mol .
mol
.
F . r Q = 0.0425
0.6
in the distillate x Db
Concentration of b
0.4
Charge, c . .
F . r Q = 0.106
0.2
.
F, a
0.0
Semi-batch
reactor 0.0 0.2 0.4 0.6 0.8 1.0
(i) (ii)
Figure 6.14.: Semi-batch reactor coupled with a distillation column: (i) process
scheme; (ii) Effect of the flow rate of the feed on the concentration
of the reaction product b for the process operated under a constant dis-
tillate flow rate Ḋ · r/Q̇ = 0.096.
of almost pure heavy-boiling educt c, i.e. Na ≪ Nb −→ 0. Furthermore, the mixture

in the reactor is initially in chemical equilibrium. The initial amounts of the three
components Na ; Nb ; Nc are therefore related by the equilibrium relationship:
Nbνb
Keq = 1 (6.12)
Na · Ncνc
a) The infinitesimal feed of the light boiler shifts the system away from chemical
equilibrium. Due to the feed, the new amounts of the three components may be
written as: Na + Ḟ · ∆t; Nb ; and Nc .
b) Equilibrium is re-established in a reaction step, hence the amounts of the three
components may be written as: Na + Ḟ · ∆t − ∆NR ; Nb + νb · ∆NR ; and Nc − νc · ∆NR .
Here, ∆NR is the change of number of moles of component a due to the chemical
reaction.
Hence,
(Nb + νb · ∆NR )νb
Keq = (6.13)
(Na + Ḟ · ∆t − ∆NR ) · (Nc − νc · ∆NR )νc
for Na −→ 0; and ignoring the second power of infinitesimal differences:
57
Nbνb + νb2 · Nbνb −1 · ∆NR

Keq =
(Ḟ · ∆t − ∆NR ) · (Ncνc − νc · Ncν−1 · ∆NR )
Nbνb + νb2 · Nbνb −1 · ∆NR
= (6.14)
Ḟ · ∆t · Ncνc − Ncνc · ∆NR
re-arranging:
Ncνc · Keq · Ḟ · ∆t − Nbνb

∆NR =
Ncνc · Keq + νb2 · Nbνb −1
Ḟ · ∆t · Nbνb − Nbνb · Na
= 2
νb · Nbνb −1 · Na + Nbνb
Ḟ · ∆t − Na
=
1 + νb2 · Na /Nb
≈ Ḟ · ∆t (6.15)
c) The reaction step is followed by an infinitesimal distillation step, where the pure
reaction product is obtained at the distillate. The amounts of the three components
can therefore be written as: Na ; Nb + νb · ∆NR − Ḋ · ∆t = Nb + νb · Ḟ · ∆t − Ḋ · ∆t;
and Nc − Ḟ · ∆t. Since the amount of the reaction product in the reactor (Nb ) is
very small, the amount of distillate product should be smaller or equal to the amount
of reaction product, otherwise the heavy-boiling educt would pollute the distillate
product:
∆NR νb νb
Ḋ ≤ · = Ḟ · (6.16)
∆t νa νa
On the other hand, if too little distillate product were withdrawn, the amount of
reaction product in the reactor would tend to increase. This favours the reverse
reaction, leaving non-reacted light-boiling educt in the reactor, which preferentially
ends at the top of the column, polluting the distillate.
Hence, the optimal value of the ratio between the feed and the distillate stream is
equal to the ratio of the stoichiometric coefficient of the light-boiling educt (νa ), to
that of the reaction product (νb ):
Ḟ νa
= (6.17)
Ḋ νb
Figure 6.15 further illustrates the importance of the ratio between the flow rates
of the feed and distillate streams Ḟ /Ḋ. The flow rate of the distillate stream is kept
constant and that of the feed stream is varied stepwise around a fixed value given by
Equation (6.17). When Ḟ /Ḋ, drops below νa /νb = 0.5, the heavy-boiling reactant (c)
pollutes the distillate product. On the other hand, when Ḟ /Ḋ becomes bigger than
νa /νb = 0.5, the non-reacted educt a pollutes the distillate product.
58
0.25
mol
of the distillate xD
mol
Concentration
xDa
0.15
0.10
xDc
0.05
0.00
0.625
mol/s
0.500
.
D
Flow rate
0.375
0.250
.
F
0.125
0.000
0.0 0.2 0.4 0.6 0.8 1.0
Figure 6.15.: Effect of the ratio Ḟ /Ḋ on the concentration of the light and heavy-
boiling educts in the distillate stream.

αbc 1.5
Chemical equilibrium constant Keq 10
Final concentration of distillate product xDb, ω 0.98 mol/mol
Recovery of intermediate boiler σDb 0.95
59
Figure 6.16 (i) shows an improved process control scheme. The setpoint to the
feed of the light-boiling reactant is determined in a cascade by the distillate’s flow
rate, which is manipulated to control the temperature at the column’s head. In this
way, the ratio of the two flow rates is kept constant. The concentration profiles in the
reactor vessel and at the column’s head are shown in Figure 6.16 (ii). In this case, the
concentrations of the light boiler in the sump and at the column’s head are very small.
On the other hand, both the distillate concentration of the middle-boiling reaction
product and its recovery are very high: xDb = 0.98 and σDb = 0.95, respectively. The
simulation data is summarised in Table 6.6.
1.0
mol
mol x Db
concentration xD
Distillate 0.6
.
D, b 0.4
x Dc
TC 0.2
x Da
0.0
Charge, c
. 1.0
F, a
Semi-batch mol x Bc
FFC reactor mol
Concentration in
0.6
B
sump x
0.4
x Ba x Bb
0.2
0.0
0.0 0.2 0.4 0.6 0.8 1.0
(i) (ii)
Figure 6.16.: Combination of a semi-batch reactor with a distillation column: (i)

process; and (ii) distillate and bottom’s concentrations.
60
6.2.2. Semi-batch reactive distillation

A semi-batch reactive distillation column is depicted in Figure 6.17 (i). This con-
sists of a reactive column section (shown in grey) placed between the reboiler and
a rectifying column section. The light-boiling reaction educt (a) is charged into the
column’s sump, and the heavy-boiling educt (c) is continuously fed into the top of a
reactive column section. Due to their relative volatility, the two components tend to
flow in opposite directions along the reactive column section, where the middle boiler
(b) is formed according to Reaction (6.11):
a+c⇋2b
1.0
mol
mol
concentration xD x Db x Db
0.6
Distillate
0.4
. 0.2 x Dc x Da
D, b
0.0
.
F, c
TC
1.0
TC
x Bc
Charge, a mol
mol x Ba
Concentration in
0.6
B
x Bb
sump x
0.4
0.2
0.0
0.0 0.2 0.4 0.6 0.8 1.0
(i) (ii)
Figure 6.17.: Semi-batch reactive distillation column: (i) process scheme and,
(ii) concentration path of the distillate and in the column’s sump.
61
In the upper part of the column, above the reactive section, the reaction product
is purified from the heavy-boiling educt, and is recovered as distillate product. To
reduce the amount of non-reacted light boiler leaving the distillation column with
the distillate, an excess of the heavy boiler is continuously fed into the column. This
accumulates in the bottom vessel. When the light-boiling educt is entirely consumed,
and its concentration in the column’s sump reaches zero, the feed of the heavy boiler
is interrupted. The distillation process is continued in order to recover the reaction
product, and to purify the heavy-boiling educt for further use. During this process
step, depending on the value of the reaction equilibrium constant, a considerable
amount of the product might decompose forming the light-boiling educt, which would
pollute the distillate product.
Figure 6.17 (ii) shows a plot of the concentration path in the reboiler (xB ) and that
of the distillate (xB ) stream, for the process operated under temperature control.
The average distillate composition (x̄Bb ) is plotted as a dotted line. In the column’s
sump, the concentration of the reaction educt a steadily decreases. However, its
concentration in the distillate xDa is very low, meaning that it is consumed in the
reactive column section, where the reaction product b is formed. The intermediate-
boiling component is obtained in high purity as distillate product. Its concentration
in the column’s sump initially increases, to reach a maximum. It then gradually
decreases to reach very low values at the end of the process. At a relative distillation
time of τ = 0.65, the concentration of educt a in the sump reaches very low values,
indicating that it has been almost completely consumed by the reaction. Nevertheless,
at this time, there is a large amount of reaction product left in the sump. This is
recovered as distillate product in a subsequent distillation step, during which the feed
of educt c is interrupted. At the end of the process, the excess non-reacted educt c is
recovered in pure form in the sump of the column.
In Figure 6.18, the concentration profile in the column is plotted in a triangular
diagram for a relative distillation time of τ = 0.25. The concentration path of the
liquid phase in the sump moves along the dotted line, from the pure light boiler to
the pure heavy boiler. The concentration profile in the reactive column section (dark
grey circles), follows the chemical equilibrium curve starting from the concentration in
the reboiler and moving towards the heavy boiler. Above the feed stage, no reaction
occurs in the liquid phase. Therefore, the profile in the rectifying section (light grey
triangles), leaves the chemical equilibrium curve, and moves close to the b − c binary
edge. The concentration of the intermediate boiler at the top of this section is very
high.
From the above considerations, it follows that the reflux ratio of the rectifying
column section, as well as the number of stages of this section determine the con-
centration of component c in the distillate. On the other hand, the concentration of
component a in the distillate product is determined by both the number of stages
in the reactive column section and the flow rate of the feed Ḟ of the heavy-boiling
reactant.
In Figure 6.19 the molar holdup (B) in the sump is plotted for each component as
62
Light boiler (a)
Chemical
Concentration equilibrium
path in sump K eq = 10
a
x
Concentration profile
in reactive section Concentration profile
t = 0.25 in rectifying section
t = 0.25

xb
Figure 6.18.: Semi-batch reactive distillation column: concentration path in column

for τ = 0.25.
a function of the relative distillation time. The total molar holdup in the reboiler,
and the total amount of heavy boiling educt fed into the sump are also plotted in
this figure. The process requires a considerable excess of the reaction educt c. This
accumulates in the reboiler, and it can be recovered only at the end of the process.
Therefore, the temperature in the reboiler considerably increases. Since the maximum
6000
Amount of c fed
to the column
mol
Molar holdup
Total holdup
3600
Ba Bc
2400
1200
Bb
0
0.0 0.2 0.4 0.6 0.8 1.0
Figure 6.19.: Semi-batch reactive distillation column: molar holdups in the column’s
sump, and total amount of heavy-boiling educt c fed into the column.
63
Table 6.7.: Simulation data for the processes shown in Figure 6.17 (i).

Position of feed stage nth, F 7
αbc 5.0
Final concentration in sump xBc, ω 0.99 mol/mol
holdup in the sump is 1.5 times larger than the initial charge, it is necessary to over-
dimension the sump. The simulation data for the process is summarised in Table 6.7.
6.2.3. Semi-batch reactive distillation in a column with a middle

vessel
A semi-batch reactive distillation in the middle vessel column is shown in Figure 6.20 (i).
The charge vessel, into which the light-boiling educt a is charged, is placed between
the stripping column section and the reactive column section. Similarly to the pro-
cess introduced in the previous section, the heavy-boiling educt c is continuously fed
at the top of the reactive column section, above which there is a rectifying section.
The concentration profile in the column is shown in Figure 6.20 (ii). In the rectifying
column section (light grey triangles), the intermediate boiler b is purified from the
heavy-boiling educt c, and is consequently obtained in high purity as distillate prod-
uct. In the reactive column section, the light-boiling reactant a, which tends to climb
up the column, reacts with the down-flowing educt c according to Reaction (6.11):
a+c⇋2b (6.18)
The concentration profile in the reactive column section (dark grey circles) is there-
fore anchored to the chemical equilibrium curve. In the stripping column section no
reaction occurs, the concentration profile (black squares) leaves the chemical equilib-
rium curve, and moves from the middle vessel’s concentration to the heavy-boiling
component. The heavy boiler is obtained in pure form at the bottom of the column.
Analogously to the process introduced in the previous section, the concentration of
component c in the distillate is determined by the reflux ratio of the rectifying column
section, as well as by the number of stages of this section. On the other hand, the
concentration of component a in the distillate product is determined by the number
of stages in the reactive column section, and by the flow rate of the feed Ḟ of the
heavy-boiling reactant. The reboil ratio of the stripping column section determines
the concentration of the bottom product.
64
.
D, b
Light boiler (a)
.
F, c
TC
Concentration Chemical
TC path in sump equilibrium
K eq = 10
Charge, a
Concentration profile
a
x
in reactive section
t = 0.25
Concentration profile Concentration profile

in stripping section in rectifying section
TC t = 0.25
t = 0.25

xb
.
B, c
(i) (ii)
Figure 6.20.: Semi-batch reactive distillation in a column with a middle vessel: (i)
process scheme, and (ii) concentration path in column for τ = 0.25.
The molar holdups in the middle vessel (M) and the flow rates of the feed stream
(Ḟ ) and that of the bottom product (Ḃ) are shown in Figure 6.21. The distillate
consists mostly of the intermediate boiler, which is recovered with a purity of xDb =
0.92 and a recovery of σDb = 0.87. The amount of educt a in the middle vessel
decreases steadily with time. This is mostly consumed by the chemical reaction.
At the same time, the molar holdup of the intermediate boiler in the middle vessel
initially increases to reach a maximum; it then decreases to zero at the end of the
process. The excess heavy-boiling educt c is separated from the other components
in the stripping column section, and it is recovered as a bottom product. Therefore,
the amount of heavy-boiling educt accumulated in the middle vessel is very low. The
amount of this component required by the process is plotted in Figure 6.21 as a dotted
line. An excess of 20% of educt c is required. This excess is mostly recovered in the
final phase of the process. During the initial phase, the flow rate of the feed stream
(Ḟ ) is much larger than that of the bottom product (Ḃ) . As the concentration of a
in the middle vessel decreases, and that of b increases, the required feed rate of the
heavy boiler decreases, and at the same time Ḃ increases. Therefore, some of the
non-reacted educt c is recovered during the process and is fed back to the top of the
reactive section. This reduces the total amount of c necessary for the process. After
65
mol
s
Flow rates
.
F
2
1
.
B
0
6000
mol Amount of c fed

to the column
Total holdup
3600
Molar holdup
Ma
2400 Mb
Mc
1200
0
0.0 0.2 0.4 0.6 0.8 1.0
Figure 6.21.: Semi-batch reactive distillation in a column with a middle vessel: flow
rates of feed stream Ḟ and of bottoms stream Ḃ (top); and molar holdups
in middle vessel (bottom).
a relative time of τ ≈ 0.75, the flow rate of the bottom product becomes bigger than
the feed rate, resulting in a net recovery of educt c.
The simulation data for the batch reactive distillation column with a middle vessel
is presented in Table 6.8.
66

Position of middle vessel nth, M 15
αbc 5.0
Concentration in sump xBc 0.98 mol/mol
6.2.4. Comparison
In this section, the three processes discussed in Sections 6.2.1, 6.2.2, and 6.2.3 are
compared. For the aim of this comparison, the total number of stages of the various
distillation columns, the relative volatility of the components, and the chemical equi-
librium constant are assigned equal values for each process. The termination criteria
for the two reactive distillation processes discussed in Sections 6.2.2 and 6.2.3 are also
the same. The simulation is interrupted when the molar holdup of b in the feed vessel
has dropped to zero. This is not possible for the process discussed in Section 6.2.1.
Table 6.9.: Simulation data for the comparison of the processes presented in Sec-
tion 6.2.

αbc 5.0
Final concentration in sump xBc 0.98 mol/mol
Semi-batch reactor coupled with a rectifying column
Termination criterion RL, max 60
Batch reactive distillation in a regular column
Termination criterion Bb, ω 1.0 mol
Batch reactive distillation in a column with a middle vessel
Position of middle vessel nth, M 14
Termination criterion Mb, ω 1.0 mol
67
Here, the process is interrupted when the reflux ratio (RL ) reaches a given maximum
value, indicating that steady state conditions have been reached. Both termination
criteria give, however, the maximum attainable recovery of the reaction product b.
The final average distillate concentrations are also similar for the three processes.
The main data is summarised in Table 6.9.
The conversions of the two reactants are plotted for the three processes in Fig-
ure 6.22. The conversion of reactant i is defined as the ratio of the number of moles
of component b collected at the distillate to the number of moles of reactant i con-
sumed in the process.
1.0
Educt a
0.8
Conversion to b
0.6
0.4
Educt c
0.2
0.0
Semi-batch reactor + Semi-batch reactive Semi-batch reactive
rectifying column distillation column distillation in a middle
vessel column
Figure 6.22.: Comparison of semi-batch modes for Reaction (6.11): conversion to the
intermediate-boiling reaction product (b).
The three processes show a high conversion of the light-boiling educt. Component
a can only leave the system with the distillate. However, for the three processes, it
is possible to keep the distillate concentration of the most volatile educt (xDa ) at low
values. The conversion of the heavy-boiling educt is different for the three processes.
This is extremely low for the semi-batch reactive distillation. Here, a large excess of
component c is required to keep a low concentration of a at the top of the reactive
column section. However, this excess is recovered only at the end of the process.
The maximum conversion of c is obtained for the semi-batch reactive distillation in
a middle vessel column. Even though an excess of reactant c is required, it is mostly
recovered as bottom product during the process, and is recycled to the top of the
reactive column section. This reduces the actual amount of educt required at the
beginning of the process.
The relative duration of the three processes, which is equivalent to the energy
demand, is plotted in Figure 6.23. The duration of the semi-batch reactor combined
68
with a rectifying column is taken as reference. This process has the longest duration.
The duration of the batch reactive distillation column is approximately 50% shorter
than that of the reference process. The batch reactive distillation column with a
middle vessel requires the shortest time. This process provides time savings of up to
60% for the studied system.
100
%
80
demand of the processes
60
Relative energy (time)
40
20
0
Semi-batch reactor + Semi-batch reactive Semi-batch reactive
rectifying column distillation column distillation in a middle
vessel column
Figure 6.23.: Comparison of semi-batch modes for the Reaction (6.11): energy de-
mand of the processes.
69
batch distillation - Kinetically
controlled reactions
Kinetically controlled reactions are defined as those reactions which require a rela-
tively long time to reach equilibrium conditions. In this case, the residence time of
the reactants in the reactor is of primary importance. In this chapter, only combina-
tions of batch reactors with distillation columns are taken into consideration. Various
process alternatives, which have already been introduced in the previous chapter, are
discussed for the decomposition of an intermediate-boiling component (b) to a light
boiler (a) and a heavy boiler (c):
kf
−⇀
2b↽−a+c (7.1)
kr
where kf and kr express the rate constants for the forward and reverse reactions,
respectively. Furthermore, under the assumption of a molecular reaction, the reaction
rate (r) can be written as:
r = kf · x2b − kr · xa · xc (7.2)
At chemical equilibrium, the reaction rate is zero, and therefore by combining

Equation (7.1) with Equation (6.1), a relationship between the chemical equilibrium
constant (Keq ) and the rate constants (kf and kr ) is obtained:
kf
Keq = (7.3)
kr
Figure 7.1 shows the concentration profile in a batch reactor for different values of
the reaction rate constants and with Keq = 1. The reactive system reaches chemical
equilibrium after different residence times in the reactor, depending on the value of the
rate constants. The light grey curve shows the concentration profile for a fast reaction
(kf = kr = 1 s−1 ), in which case chemical equilibrium is reached instantaneously. For
slow reactions, (kf = 10−4 s−1 ), chemical equilibrium is reached after much longer
times.
70
7.1. Combining a reactor with a regular batch distillation column
1.0
mol
mol
Concentration x
0.6
0.4
0.2
0.0
0 1800 3600 s 7200
Residence time in reactor
Figure 7.1.: Effect of reaction kinetics on the concentration path in a batch reactor
for Reaction (7.1).
7.1. Combining a reactor with a regular batch

distillation column
The combination of a batch reactor with a rectifying column has already been dis-
cussed in Section 6.1.2 for a fast reaction. For a slow reaction, the rate of formation
of the reaction products is much slower; hence, for the same heat duty, an extremely
high reflux ratio is required to obtain similar concentrations of the distillate product.
This is shown in Figure 7.2 (i). Here, the reflux ratios required to maintain a dis-
tillate concentration of xDa = 0.98 mol/mol, for a fast reaction (kf = 1 s−1 ) and for
two slow reactions (kf = 10−3 s−1 and kf = 10−4 s−1 ), are plotted with respect to the
recovery of the light-boiling product (σDa ). The heat duty to the reboiler is the same
for the three compared cases, and is constant for the whole duration of the process.
The main simulation data for this comparison is reported in Table 7.1. It can be
seen that the slower reactions require very high reflux ratios for the whole duration
Table 7.1.: Simulation data for the processes discussed in Section 7.1.
Molar fractions in the charge xBb, α 1.0 mol/mol

αbc 1.5
71
5 Light boiler (a)

10 xD
4
10 Chemical
equilibrium
Reflux ratio RL
3 K eq = 1
10
-4
k f = 10 s-1
a
x
2
10
-3
-3
k f = 10 s-1 -1
k f = 10 s-1
10 kf = 1 s
k f = 1 s-1 x Bw x Ba
1
0.0 0.2 0.4 0.6 0.8 1.0 High boiler (c) Middle boiler (b)
xb
-4
Recovery of light boiler sa k f = 10 s-1
(i) (ii)
Figure 7.2.: Combination of a batch reactor with a rectifying column: effect of re-
action kinetics (i) on the required reflux ratio (RL ) of the distillation
column; and (ii) on the concentration path in the reactor
of the process. This has a negative effect on the duration and, in turn, on the energy
demand of the process. Furthermore, the concentrations in the reactor vessel, shown
in Figure 7.2 (ii), do not approach the chemical equilibrium curve, and, as in the case
of the slowest reaction kinetics, the concentration of the light-boiling component is
very low. This indicates that, notwithstanding the very high reflux ratio, the light-
boiling component is removed from the vessel upon its formation. Hence, the reaction
kinetics dominate the dynamics of the process.
The extremely high reflux ratios needed for this process are difficult to implement
in practice. As an alternative, a cyclic reflux procedure is studied. In this case, phases
of total reflux and finite reflux operation cyclically follow each other. During the total
reflux phases, no distillate is withdrawn. Time is allowed for the formation of the
light-boiling product. Its concentration in the column and in the reactor therefore
increase. This is followed by a finite reflux phase, during which the reaction product
is withdrawn from the distillate with the required purity. After reaching a switching
criterion, the distillate withdrawal is interrupted and the column is again operated
with total reflux.
The selection of the switching criteria between the two phases of each cycle is of
primary importance. The switching from the finite reflux to the total reflux opera-
tion should be effected when the concentration of the distillate drops under a given
value. This could be inferred from an increase in temperature in the upper part of
72
the column. On the other hand, the switching from the total reflux to the finite reflux
phases should be effected when a sufficient amount of light-boiling product is accu-
mulated in the reactor. The implementation of this criterion might require a direct
measurement of the concentration in the reactor, since a temperature measurement
does not provide the required information.
The reflux ratio and the recovery of the two products (σDa , σBc ) are plotted in
Figure 7.3 (i) for a very slow decomposition reaction with kf = 10−4 s−1 . The reflux
ratio is initially increased to maintain a constant distillate concentration. When
the reflux ratio reaches a predefined value, in this example RL = 50, the operation is
turned to total reflux (i.e. Ḋ = 0), and the cyclic operation starts. As time passes, the
concentration of the reactant in the still (xBb ) decreases and that of the heavy-boiling
product (xBc ) increases. According to Equation (7.2), the reaction rate decreases,
and therefore the time required by the total reflux phase increases. However, for
this case study, the recoveries of both products reach 80% just after two cycles. The
concentration of the heavy boiler in the sump is, nevertheless, lower than 70%, hence
the process should either be carried on as in Figure 7.3 (i), or a distillate off-cut
should be taken at a low reflux ratio. In this way, the residue in the reactor vessel is
purified.
During the total reflux phase, the column functions as a condenser with a large
holdup, in which the light-boiling component is enriched. For extremely slow re-
actions, the total reflux phases might be very long, and it might be energetically
1.0 62.5 1.0

Reflux ratio Combined process
mol 50.0 mol
mol mol
sDa
Recovery of c in reactor sc
Reflux ratio RL
0.6 37.5 0.6 Batch reactor

Recovery s
sBc
0.4 25.0 0.4
0.2 12.5 0.2
0.0 0 0.0
0.0 0.2 0.4 0.6 0.8 1.0 0.0 0.2 0.4 0.6 0.8 1.0
Relative time t/t tot Relative time t/t tot
(i) (ii)
Figure 7.3.: (i) Cyclic reflux operation of a batch reactor coupled with a rectifying
column. Reflux ratio, recovery, and purity of the reaction products.
(ii) Recovery of the heavy-boiling reaction product for a batch reactor and
for the combined process.
73
7.2. Combining a batch reactor with a middle vessel column
convenient to carry out the reaction at lower heat loads. The heat duty can be in-
creased before the end of the total reflux phase, this provides the necessary vapour
and liquid loads to the column.
The main advantage of this process is that for slow reactions it avoids the high
reflux ratios required by the process discussed previously. This is achieved without any
negative effect on the duration of the process, on the recovery and on the concentration
of the products. Due to the continuous removal of one of the reaction products, the
concentration of the reactant in the still of the combined process is higher than in a
batch reactor. The rate of the forward reaction is also larger. Therefore, the combined
process is faster than a batch reactor. The recovery of the heavy-boiling component
for the reaction carried out in a batch reactor is plotted in Figure 7.3 (ii). The recovery
of this component is limited by the chemical equilibrium. The recovery of the heavy
boiler for the reactor combined with a batch rectifying column is also plotted in this
figure. In this case, the light boiler is removed from the reactor. Therefore, the
recovery is not limited by the chemical equilibrium, so that a much higher recovery
is obtained.
7.2. Combining a batch reactor with a middle vessel

column
The combination of a batch reactor with a middle vessel column is illustrated in
Section 6.1.3 for a fast reaction. The process scheme is shown in Figure 6.2 (iii). For
-3
10
s-1
Batch reactor +
-4 middle vessel
Reaction rate
10 column
Batch reactor +
rectifying
-5 column
10
-6
10
0.0 0.2 0.4 0.6 0.8 1.0
Figure 7.4.: Reaction rate for the decomposition Reaction (7.1) with kf = kr =
10−3 s−1 carried out in a batch reactor coupled with a rectifying and with
a middle vessel column.
74
7.3. Closed Processes
slow reaction kinetics, in order to provide the required residence time in the reactor,
very high reflux and reboil ratios are required. However, in this case, both light and
heavy-boiling reaction products are removed from the reaction mixture. Thus the
concentration of the educt in the reactor is higher than that in a reactor combined
with a rectifying column. Figure 7.4 shows a comparison of the reaction rate in a
batch reactor coupled with a rectifying and with a middle vessel column. The reaction
rate constants are kf = kr = 10−3 s−1 , and the two columns have the same number
of equilibrium stages. The reaction rate for the process in the middle vessel is much
higher, and therefore the duration of the process is shorter. For this case study, the
duration of the process in the middle vessel is 64% shorter than that in a batch reactor
coupled with a rectifying column.
The most relevant simulation data for the batch reactor coupled with a middle
vessel column is summarised in Table 7.2.
Table 7.2.: Simulation data for the batch reactor coupled with a middle vessel column
illustrated in Section 7.2.

αbc 1.5

Both processes described in Sections 7.1 and 7.2 require either high reflux ratios, or a
complex operating procedure with a cyclic reflux policy. Hence, a simplification of the
process is required. Closed processes, which are illustrated in Section 6.1.4 for a fast
equilibrium limited reaction, significantly simplify the operation of the process, since
no product is withdrawn from the system. The two available processes are described
in Section 6.1.4.
Closed operation of a batch distillation column coupled with two reactors

The process shown in Figure 7.5 (i) consists of two reactors placed at the bottom
and at the head of a distillation column. The charge is initially loaded into both
reactors so that the initial molar holdups are given by the ratios of the stoichiometric
75
coefficients of the two products, as given by Equation 6.8:

Dα νa
= (7.4)
Bα νc
The distillation column is therefore operated at close to total reflux conditions. This
has the double advantage of using the maximum separation capability of the distilla-
tion column, and more importantly, of drastically simplifying the operating procedure.
Similarly to the case of fast reactions, a total reflux operation delivers high purities
and high recoveries in both reactors only in the case of equimolar reactions.
The concentration path in the two reactors for Reaction (7.1)
kf
2b−
⇀
↽−a+c (7.5)
kr
and for two values of the reaction rate constants (kf and kr ) are plotted in a triangular
diagram in Figure 7.5 (ii). Since the case studied is an equimolar reaction, the charge
has been equally distributed between the reactors, and the column has been operated
under total reflux.
Both profiles start from the reactant vertex. The concentration path in the upper
reactor migrates from here towards the light boiler vertex. That in the bottom reactor
simultaneously migrates towards the heavy-boiling component. The path followed
Charge, b Light boiler (a)

x Dw
Chemical
Batch
equilibrium
reactor
K eq = 1
Path in top
Dw,a reactor
a
x
TC
. .
L V
-4
Charge, b k f = 10 s-1
-3
k f = 10 s-1
x Bw x Ba = x Da
Batch
reactor High boiler (c) Middle boiler (b)
xb
Path in bottom
Bw,c reactor
(i) (ii)
Figure 7.5.: Closed operation of a multi-vessel batch distillation column with two re-
actors: (i) process scheme, and (ii) concentration path in both reactors,
for two different reaction rate constants.
76
depends on the reaction rate constant. For fast reactions the concentration profile
jumps to the equilibrium curve and follows it, as shown in Figure 6.11 (ii). For very
slow reactions, the concentration path in the upper reactor moves along the binary
edge joining the intermediate to the light boiler. Similarly, the path in the lower
reactor moves along the binary edge joining the intermediate to the heavy boiler.
This is due to the fact that the reaction is much slower than the separation, and
therefore the light and heavy-boiling products are respectively removed from the
lower and upper reactors upon their formation.
The concentration paths in the reactors for a moderately slow reaction (kf = kr =
10−3 s−1 ) are also plotted in Figure 7.5 (ii). In this case, the concentration paths
lie between the binary edges and the chemical equilibrium curve. Here, the reaction
and the separation processes have similar kinetics. Hence, the two processes strongly
interact, and the dynamic behaviour of one influences the performance of the other.
The simulation data for the closed operation of a batch distillation column coupled
with two reactors is summarised in Table 7.3.
Table 7.3.: Simulation data for the closed operation of a batch distillation column
coupled with two reactors.
Molar fractions in the charge xDb, α 1.0 mol/mol

xBb, α 1.0 mol/mol
αbc 1.5
Closed operation of a reactor coupled with a middle vessel batch distillation

column
The process shown in Figure 7.6 (i) consists of a reactor coupled with a closed middle
vessel column. This process is illustrated for an equilibrium limited decomposition
reaction in detail in Section 6.1.4. Reaction (7.6)
kf
4b−
⇀
↽−a+3c (7.6)
kr
takes place only in the liquid phase in the middle vessel. The middle-boiling reaction
educt is charged to this vessel, where it undergoes the decomposition reaction. The
light and heavy-boiling products thus formed tend to move towards the upper and
lower parts of the distillation column respectively. The reflux ratios of the two column
77
sections determine the purity of the two fractions collected in the top and bottom
vessels. The concentration paths for two different values of the reaction rate constants
are plotted in a triangular diagram in Figure 7.6 (ii). The concentration path in the
reactor vessel is determined by the relative speeds of the reaction and separation
processes. For very slow reactions, the two products are removed immediately upon
formation, and the reactor contains almost exclusively the reaction educt b. On the
other hand, for moderately slow reactions, where the reaction rate is comparable to
the dynamics of the distillation column, the distillation process does not manage to
remove the two reaction products upon their formation. The concentration in the
reactor moves away from the educt vertex towards the chemical equilibrium curve.
The path followed depends on the position of the reaction pole (πR ), and on the
reflux ratios of the two column sections. If the ratio of the rates at which the two
reaction products are removed from the reactor vessel were equal to the ratio of the
Dw,a
TC Light boiler (a)
. . x Dw
LU VU
Charge, b Chemical
equilibrium
K eq = 1
a
x
Batch
reactor
-3
k f = 10 s-1
TC -4
k f = 10 s-1
. .
LL VL x MVa
x Bw

xb
Bw,c
(i) (ii)
Figure 7.6.: Closed operation of a reactor coupled with a middle vessel batch distilla-
tion column for Reaction (7.6): (i) process scheme, and (ii) concentra-
tion path in the reactors, for two different reaction rate constants.
78
stoichiometric coefficients, then the concentration in the reactor vessel would move
along the straight line joining the vertex b of the triangular diagram, to the reaction
pole πR . This condition is satisfied when the internal reflux ratios of the two column
sections, (L̇/V̇ )U and (L̇/V̇ )L , are related via Equation (6.9):
! !
νa L̇ νc L̇
· + · =1 (7.7)
νa + νc V̇ L νa + νc V̇ U
Furthermore, if the rate of removal of the reaction products were much slower than
the reaction rate, for instance under total reflux operation, the concentration in the
vessel would reach the chemical equilibrium line.
In this process, the flow rate of the reflux stream from the top vessel and that
from the reactor are used to control the concentrations at the top and bottom of
the column, respectively. The internal reflux ratios, (L̇/V̇ )U and (L̇/V̇ )L , relate very
closely to Equation (7.7). This is shown in Figure 7.7, where the left hand side of
Equation (7.7) is plotted as a function of the relative distillation time. The value of
this function oscillates around one.
! ! ! ! !
L̇ L̇ νa L̇ νc L̇
f , , νa ,νc = · + · (7.8)
V̇ L V̇ U νa + νc V̇ L νa + νc V̇ U
The simulation data for the closed operation of a reactor coupled with a middle
vessel batch distillation column is summarised in Table 7.4.
1.05
1.03
-3
k f = 10 s-1
, . ,n ,n )
a c
1.01
U V L
( (( (
0.99
L
.
-4
k f = 10 s-1
V
f( .
L
.
0.97
0.95
0.0 0.2 0.4 0.6 0.8 1.0
Figure 7.7.: Closed operation of a reactor coupled with a middle vessel batch distil-
lation column for Reaction (7.6): plot of the function given in Equa-
tion (7.8).
79
7.4. Comparison
Table 7.4.: Simulation data for the closed operation of a reactor coupled with a middle
vessel batch distillation column.

αbc 1.5
7.4. Comparison
For kinetically controlled reactions, the open processes might require very high reflux
and reboil ratios to keep a constant product concentration. Therefore, closed processes
are very advantageous for this type of reactions. Two closed processes were illustrated
in Section 7.3. The operation of both processes is greatly simplified with respect to
the open processes. The concentration path in the reactors of the two closed processes
greatly differ. On the one hand, in the closed operation of a batch distillation column
coupled with two reactors, the concentration of the reaction educt in both reactors
decreases during the process. Therefore, the rate of the chemical reaction decreases.
On the other hand, in the closed operation of a reactor coupled with a middle vessel
batch distillation column, the concentration of the educt remains high throughout the
process. Hence, the reaction rate varies in a narrower range. For the case studied, the
duration of this process is 70% shorter than that of the closed operation of a batch
distillation column coupled with two reactors.
80
batch distillation - Sequential
reactions
Side reactions, where the desired products further react to form undesired side prod-
ucts, are very common in reactive systems. Such reactions have the double drawback
of decreasing the yield of the desired product, and at the same time, due to the forma-
tion of undesired components, increasing the complexity of the downstream separation
process. In fact, both the reactants, as well as the desired products should be sepa-
rated from the side products. Suppressing the side reactions is therefore an important
task for the reaction engineer. It is common industrial practice to resort to selective
catalysts, which do not allow the products to further react. As an alternative, the
reaction products can be removed from the reaction zone upon their formation. In
the latter case, a reactor could be coupled with a distillation process. The reaction
products are therefore removed from the reactive mixture and are obtained either as
overhead or bottoms of the distillation process.
As already pointed out in the literature review, only one publication [Gadewar et al.
2000] considered the combination of a batch reactor with a distillation process applied
to the case of a sequential reaction. In this chapter, several sequential reactions are
considered, and some combined processes are analysed.
8.1. Batch reactor

In the sequential reaction
k1 2k
νc c −
→ νa a −
→ νb b (8.1)
the reactant c decomposes into the main product a which further reacts to the side
product b. The yield and selectivity are usually used to characterise these reactive
systems. The yield of the main product (Ya ) is defined as the ratio of the number
of moles of product formed to the maximum amount that could be formed by the
reaction. For Reaction (8.1) this may be written as:
νc Ma
Ya = · (8.2)
νa Mc, α
The selectivity (S) is defined as the ratio of the number of moles of the product
formed to the number of moles of reactant consumed. This ratio is normalised with
81
8.1. Batch reactor
the ratio of the stoichiometric coefficients:

νc Ma
Sa = · (8.3)
νa Mc, α − Mc
The yield and selectivity of the main product are plotted in Figure 8.1 as a function
of the conversion of the reactant (Xc = (Mc, α − Mc )/Mc, α ), for the reaction carried
out in a batch reactor. If the rate constants of the two reactions are equal (k1 = k2 ),
the yield of the main product goes through a maximum. By the time that all the
reactant c has been converted, the desired product a has completely reacted to the
side product b, so that its yield and selectivity drop to zero. The situation becomes
more drastic if the rate constant of the side reaction is larger than that of the main
reaction (k2 > k1 ). This is also shown in Figure 8.1 for the case that k1 /k2 = 0.1.
Here, the yield of the desired product also goes through a maximum, however, this is
much smaller than for the previous case. On the other hand, if the rate constant of
the first reaction is much larger than that of the second reaction (k1 /k2 = 10) both
the yield and the selectivity of the desired product are high. This phenomenon is
usually used in reaction engineering for the design of selective catalysts.
1.0
k1
mol = 10
mol Sa k2
Yield Ya ; selectivity S a
0.6 k1
=1
k2
0.4
k1
0.2 Ya = 0.1
k2
0.0
0.0 0.2 0.4 0.6 0.8 1.0
Conversion of educt c X c
Figure 8.1.: Selectivity (full lines) and yield (dotted lines) of the main component
a for the sequential Reaction (8.1) carried out in a batch reactor as a
function of the conversion Xc of the educt, for three values of the ra-
tio between the rate constant of the first to that of the second reaction
(k1 /k2 ).
A further class of reactions is given by the parallel Reactions (8.4), where one of
the reaction educts (a), also referred to as key educt, reacts with one of the desired
products (b) produced by the first Reaction (8.4a) to form an undesired product (d):
82
8.2. Combination of batch reactor with distillation processes
1k
νa a + νc c −
→ νb b (8.4a)
2k
νa a + νb b −
→ νd d (8.4b)
The yield and selectivity of the main product are referred to the key educt, since this
is the limiting reagent. The definitions are given by Equation (8.5):
νa Mb
Yb = · (8.5a)
νb Ma, α
νa Mb
Sb = · (8.5b)
νb Ma, α − Ma
The selectivity and the yield of the desired products are limited by the rate of
the parallel Reaction (8.4b). In a batch reactor, the rate of the second reaction can
be reduced by allowing a large excess of the educt which does not further react (c).
This is achieved in a semi-batch reactor, where the key educt (a) is continuously
fed to the reactor, maintaining a low concentration of this reactant. However, the
speed of the second reaction increases with the consumption of educt c and with the
formation of the product b. This product further reacts to the undesired product d,
so that, depending on the reaction rate constants, for a complete conversion of educt
c, the main product b might have been completely converted to the by-product. The
situation worsens as the side reaction becomes faster and if the first reaction were
equilibrium limited. In this case, the concentration of the main product in the batch
reactor might be very small throughout the process.
8.2. Combination of batch reactor with distillation

processes
The combined batch processes discussed in the previous chapters may also be applied
to reactive systems with sequential reactions. The selection of the appropriate process
depends on the volatilities of reactants and products, as well as on the type of reactive
system. The key product, i.e. the reaction product which participates in the second
reaction, should be removed from the reactor. The relative volatility of this compo-
nent is therefore of primary importance. To further suppress the secondary reaction,
the key reactant is continuously fed into the reactor, keeping its concentration very
low. The reactor is, in this case, operated in a semi-batch mode. To illustrate the
importance of the order of the relative volatilities of reactants and products, two cases
are studied in the following sections. In both cases, the main product is produced by
the first reaction. Furthermore, its boiling point lies between that of the two educts.
This product then reacts with one of the two educts. The case where the key educt
is the light boiler is considered in Section 8.2.1. In Section 8.2.2, the heavy boiler is
the key educt.
83
8.2.1. Light-boiling key educt

In this section, the parallel Reaction (8.6) is studied.
1 k
a+c−
→ 2b (8.6a)
2 k
a+b−
→ d (8.6b)
The main product is a middle-boiling component (b). However, this reacts in a

second Reaction (8.6b) with the light-boiling educt (a) to form an undesired heavy-
boiling product (d). The main reaction is similar to the case studied in Section 6.2.
Analogously to that case, two process alternatives are studied. In the first process, a
semi-batch reactor is coupled with a rectifying column. The process scheme is shown
in Figure 8.2. In the second process discussed in this section, the same reaction is
carried out in a batch reactive distillation column with a middle vessel, as shown in
Figure 8.6.
Coupling of a semi-batch reactor with a rectifying column

the process scheme of a semi-batch reactor coupled to a rectifying column is shown
in Figure 8.2. The key educt a is continuously fed to the reactor, so that its concen-
tration is maintained very low. In this way, the middle-boiling reaction product is
removed from the reactor and is obtained as overhead product. The concentrations
of both components taking part in the second Reaction (8.6b) are very low, the rate
of this reaction is therefore small. Hence, the by-product is formed only in small
amounts. However, it should be noted that having an extremely low concentration
of the intermediate boiler in the reactor has a negative effect on the duration of the
separation process. This is due to the fact that a higher reflux ratio has to be applied
in this case.
.
D, b
TC
Charge, c
.
F, a
Semi-batch
FFC reactor
Figure 8.2.: Semi-batch reactor coupled with a rectifying column.
84
2000 1.0
k1
Amount of a fed =1
Bc k2
mol to the reactor mol
mol
Yield Yb; selectivity Sb

Sb
Molar holdup
1200 0.6
800 0.4
Bb Bd Yb
400 0.2
Ba
0 0.0
0.0 0.2 0.4 0.6 0.8 1.0 0.0 0.2 0.4 0.6 0.8 1.0
Conversion of educt c X c Conversion of educt c X c
(i) (ii)
Figure 8.3.: Reaction (8.6) carried out in a semi-batch reactor coupled with a rec-
tifying column: (i) molar holdups in the reactor (B) and total amount
of educt a fed to the reactor; and (ii) Yield and selectivity of the main
product b with respect to the fed educt a, plotted against the conversion
of educt c.
The component molar holdups in the reboiler (B) together with the total amount of
the light-boiling educt fed to the reactor are shown in Figure 8.3 (i). The simulation
was carried out with a constant molar fraction of b in the distillate (xDb = 0.98).
Furthermore, the flow rate of the feed Ḟ to the semi-batch reactor was adjusted to
the distillate flow rate Ḋ according to Equation (6.17):
Ḟ νa
= (8.7)
Ḋ νb
The process was terminated when the reflux ratio reached 1000. The maximum
conversion of component c is Xc = 0.77. In this case, the yield and selectivity of the
main product are defined in Equations (8.8a) and (8.8b), respectively:
νa MDb
Yb = · (8.8a)
νb MBa, α + 1 Ḟ (τ ) dτ
R
0
νa MD b
Sb = · (8.8b)
νb MBa, α + 0τ Ḟ (τ ) dτ
R
Notwithstanding the low conversion of the heavy boiling educt, the yield and selectiv-
ity of the main product with respect to the limiting educt a, shown in Figure 8.3 (ii),
are very high Yb = Sb > 0.918. Furthermore, the reflux ratio necessary to maintain
85
1000
Reflux ratio
100
10
0.0 0.2 0.4 0.6 0.8 1.0
Conversion of educt c X c
Figure 8.4.: Reflux ratio required by the semi-batch reactor coupled with a rectifying
column for the case of the parallel Reaction (8.6).
the distillate concentration increases exponentially with time, as shown in Figure 8.4.
Towards the end of the process, the reflux ratio increases asymptotically, and there-
fore the separation process becomes infeasible, so that the maximum conversion of
the heavy-boiling educt is reached.
The effect of the feed ratio is shown in Figure 8.5. The process is operated with
constant distillate concentration. The ratio between the feed rate of the light-boiling
educt to the distillate flow rate (Ḋ/Ḟ ) is varied. Decreasing the ratio increases the
instantaneous amount of the reaction product formed in the reboiler, so that the
conversion of the high-boiling educt increases. On the one hand, this allows a lower
reflux ratio to be applied to the process. Therefore, the average productivity of the
process increases. On the other hand, the rate of the side Reaction (8.6b) increases,
hence a larger amount of undesired by-product is formed and the selectivity of the
process decreases. Furthermore, when the value of the ratio (Ḋ/Ḟ ) is increased above
that given by Equation (8.7), the amount of product formed results to be smaller
than the amount removed at the distillate. Therefore, the reflux ratio necessary to
maintain the required purity increases faster and the process becomes infeasible at
smaller conversions of the heavy-boiling educt.
Table 8.1 summarises the main data used for the simulation of the semi-batch
reactor coupled with a rectifying column.
86
1.0
rel. amount of by-product

0.8
0.6
rel. productivity
Conversion Xc
0.4
Yield of b
0.2
0.0
1.8 1.9 2.0 2.2
.
Flow rate of distillate D.
Flow rate of feed stream = F
Figure 8.5.: Effect of the distillate to feed ratio Ḋ/Ḟ on the semi-batch reactor coupled
with a rectifying column for the case of the parallel Reaction (8.6).
Table 8.1.: Simulation data.

αbc 2.0
αcd 2.0
Reaction rate constants k1 1 s−1
k2 1 s−1
Concentration of distillate product xDb 0.98 mol/mol
Distillate to feed ratio Ḋ/Ḟ νb /νa
Batch reactive distillation column with a middle vessel

The Reaction (8.6):
1 k
a+c−
→ 2b (8.9a)
2 k
a+b−
→ d (8.9b)
is carried out in a batch reactive distillation column with a middle vessel. The process
scheme is shown in Figure 8.6. The light-boiling educt is initially charged to the
middle vessel, and the heavy-boiling educt is continuously fed close to the top of
the upper column. The reaction is carried out exclusively in the reactive section
87
.
D, b
.
F, c
TC
TC
Charge, a
TC
.
B, c-d
Figure 8.6.: Batch reactive distillation column with a middle vessel.
of the column. This section is included between the feed and the middle vessel.
The equilibrium stages above the feed are necessary to purify the top product from
the heavy-boiling educt. A heavy-boiling fraction is obtained as bottom’s product.
The process is operated in closed loop with three temperature controls, as depicted in
Figure 8.6. The upper temperature control loop maintains a constant molar fraction of
the main product in the distillate xDb = 0.93. The molar holdups of the distillate and
bottom’s products, and in the middle vessel are plotted versus the relative distillation
time in Figure 8.7. At the end of the process, the light-boiling product has been
completely converted to the main and side-products. However, the molar holdup of
the main product b, which is recoverred exclusively with the distillate, is much lower
than its theoretical maximum. The yield of this product is about 30%. The heavy-
boiling side product is completely collected in the bottom’s product together with
the excess educt fed to the column. The yield of this product is approximately 70%.
This means that the process does not suppress the side reaction. This is due to the
relatively high concentrations of the key educt a and of the main product b in the
lower part of the reactive column section. In this part of the column, the reaction
rates of the main (r1 ) and of the side (r2 ) reactions, shown in Figure 8.8, are both
very high. Furthermore, the reaction rate of the side reaction has values which are
very close to that of the main reaction. This means that a considerable fraction of
the main product reacts further in the second reaction.
88
3000
mol
distillate product D
Molar holdup in
1800
D
b
1200
600
D
c
0
3000
mol
M M
a
middle vessel M
c
Molar holdup in
1800
1200
M
M b
d
600
3000
mol
bottom’s product B
B
c
Molar holdup in
1800
B
d
1200
600
B
a Bb
0
0.0 0.2 0.4 0.6 0.8 1.0
Figure 8.7.: Molar holdup of the distillate and bottoms products and in the middle ves-
sel, for Reaction (8.9) carried-out in a batch reactive distillation column
in a middle vessel.
89
0.1
r (14)
2
r (14)
1
s -1
Reaction rate r
r (12)
1
0.01
r (12)
2
0.001
0.0 0.2 0.4 0.6 0.8 1.0
Conversion of educt a X a
Figure 8.8.: Reaction rates of the main (r1 ) and of the side (r2 ) reactions in the
reactive column section one (nth = 14) and two (nth = 12) equilibrium
stages above the middle vessel.
Increasing the number of stages in the rectifying section allows lower reflux ratios
to be used. Hence, the residence time of the main product in the reactive zone
of the column is reduced, and the yield of this product is increased. Nevertheless,
the concentration profile in the reactive zone changes only slightly. That is, the
concentration of the light-boiling educt and that of the main product remain high in
the lower part of the reactive section. The selectivity to the main product therefore
increases only very slightly.
The data used for the simulation of the batch reactive distillation column with a
middle vessel is reported in Table 8.2.
Table 8.2.: Simulation data: batch reactive distillation column with a middle vessel.
Total Number of equilibrium stages 19

Number of equilibrium stages above feed 9
Number of equilibrium stages in the reactive column section 5
αbc 2.0
αcd 2.0
k2 1 s−1
Concentration of distillate product xDb 0.93 mol/mol
90
8.2.2. Heavy-boiling key educt

A very different situation results if the key educt is the heavy boiler, as represented
by Reaction (8.10).
k
1
νa a + νd d −
→ νb b (8.10a)
k
2
νd d + νb b −
→ νc c (8.10b)
The first reaction producing the main product b remains unchanged. However, this
time, the heavy-boiling educt takes part in both reaction steps. If this reaction
were carried out in a semi-batch reactor coupled with a rectifying column, as in the
previous case, there would be an excess of the heavy-boiling educt in the reactor. The
product b would react with the excess educt d, according to Reaction (8.10b). This
would consume the desired product. This reaction should therefore be carried out in
a semi-batch reactor under an excess of the light-boiling educt (a) and a continuous
feed of d. The middle-boiling product (b) thus formed is separated from the reaction
mixture by coupling the reactor with a stripping column, as shown in Figure 8.9 (i).
This column configuration is discussed in the next paragraph. Furthermore, a batch
reactive distillation column, similar to the one shown in Figure 6.17 (i) has been
studied.
Coupling of a semi-batch reactor with a stripping column

The process scheme of a semi-batch reactor coupled with an inverted column is shown
in Figure 8.9 (i). The light-boiling educt is charged into the reactor at the beginning of
the process. The heavy-boiling educt is continuously fed into the semi-batch reactor.
The ratio between the flow rates of the feed stream (Ḟ ) and of the bottom product
stream (Ḃ) is related to the stoichiometric coefficient of the heavy-boiling educt (νd )
and that of the main product (νb ), according to Equation (8.11):
Ḟ νd
= (8.11)
Ḃ νb
In this way, the same amount of main product produced by the reaction is removed
from the reactor via distillation. Hence, the main product does not accumulate in the
reactor, and therefore the side reaction is largely suppressed.
The concentration path in the reactor (xD ) and the average bottom’s concentration
(x̄B ) are shown in Figure 8.9 (ii). In this case, the stripping column, having 15
equilibrium stages, is operated with a constant reboil ratio (RB = 33) until the light-
boiling educt depletes from the reactor. The initial charge to the reactor contains
some heavy-boiling educt, which immediately reacted to form the desired product. In
this way, the initial concentration of the main product in the reactor is xDb, α = 0.1.
However, some of the product reacts with the heavy-boiling educt in the second
Reaction (8.10b). The side product c thus formed contaminates the bottom product.
91
Due to the chosen feed policy, the molar fraction of b in the reactor remains constant
throughout the process. In this case, the light-boiling educt is completely converted
to the products. The selectivity to the desired component is approximately 86%.
The main constant parameters used to simulate this process are summarised in
Table 8.3
1.0
mol
mol x
Da
0.6
Concentration
in reactor xD
0.4
Charge, a
x
Db
0.2
x x
Feed, d Dc Dd
0.0
Semi-batch
reactor 1.0
x
Bb
mol
mol
Average bottom’s
concentration x B
0.6
.
B, b
0.4
x
Bc
0.2
x x
Ba Bd
0.0
0.0 0.2 0.4 0.6 0.8 1.0
(i) (ii)
Figure 8.9.: Semi-batch reactor coupled with a stripping column for the production
of a middle-boiling product according to the Reaction (8.10): (i) process
scheme; and (ii) concentration path in the reactor, and average bottom’s
concentration.
92
Table 8.3.: Simulation data of a semi-batch reactor coupled with a stripping column

αbc 2.0
αcd 2.0
k2 1 s−1
Reboil ratio RB 33
Distillate to feed ratio Ḋ/Ḟ νb /νa
Batch reactive distillation column

In Reaction (8.10), the educts have the highest and lowest volatilities. Furthermore,
the volatility of the main product is higher than that of the undesired by-product. The
process could therefore be carried out in the batch reactive distillation column shown
in Figure 8.2. The light-boiling educt is charged to the reactor, while the heavy boiler
is continuously fed to the top of the reactive column section. The two components
tend to flow in opposite directions along the reactive section, they therefore react,
forming component b. This rises along the column. However, the concentration of
the heavy-boiling educt d on the upper stages of the reactive section is very high, so
that most of component b further reacts to form the undesired side product c. The
yield and selectivity of the main product are very low. This process is therefore not
suitable for the Reaction (8.10).
93
9. Summary
The aim of this work is to provide the process engineer with criteria for the synthesis of
batch distillation processes with particular attention paid to batch reactive distillation
processes. To achieve this a novel batch distillation process together with several
batch reactive distillation processes are developed and analysed. These processes are
investigated, their advantages are outlined, and their fields of application are marked.
A novel batch distillation process for the purification of an intermediate boiler
from light and heavy-boiling impurities is illustrated in Chapter 5. Based on feasi-
bility studies, an operation procedure is postulated and is validated via simulations
of the process. The effect of the most important design parameters is studied. This
shows that the duration of the process decreases as the concentration of the interme-
diate boiler in the charge increases. The process is finally compared with the batch
distillation in a middle vessel column. It is demonstrated that the duration of the
novel process is shorter than that of the middle vessel column for charges rich in the
intermediate boiler, and for difficult separations. However, the two column sections of
the novel batch distillation process strongly interact, and therefore the control of this
column configuration is more difficult than that of a middle vessel column. Moreover,
an experimental investigation of the process is described. The simulated and experi-
mental concentration and temperature profiles qualitatively agree. The feasibility of
the novel process is demonstrated, and its practical implementation is explained.
The integration of batch reaction with batch distillation processes is discussed in
the second part of this thesis. The investigation treats equilibrium limited reactions,
kinetically controlled reactions, and sequential reactions. Several processes are de-
veloped and analysed for each reaction group. It is shown that the combination of a
batch reactor with a batch distillation column is a very versatile process alternative,
in which several reactions can be carried out.
Equilibrium limited reactions, where extreme boiling products are formed, can be
carried out in a variety of integrated batch processes. Only the coupling of batch
reactors with distillation columns is investigated. Both open and closed operation
modes of these processes are analysed. In each case, complete conversion of the educt
as well as very high product purities are obtained. This constitutes a very important
advantage over the conventional sequential process. In this way, the handling of the
liquid fractions between the reaction and separation is eliminated, the separation is
simplified, and no recycles are necessary. Furthermore, if the separation is carried
out in open processes, the minimum energy demand can be determined via short-
cut methods. This reflects the minimum duration of the integrated batch processes.
It is shown that the minimum energy demand greatly depends on the volatilities of
94
9. Summary
the components as well as on the stoichiometry of the reaction. This is therefore an

important tool for an initial screening of process alternatives.
If the reaction products are intermediate boilers, integrated semi-batch processes
are used. Two groups of processes have been identified. In one case, a semi-batch
reactor is coupled with a rectifying column. The heavy-boiling educt is charged into
the reactor at the beginning of the process. The light-boiling educt is continuously fed
into the reactor, so that its concentration is maintained low throughout the process. In
this way, the reaction product with the lowest boiling point is withdrawn as distillate
product. In the second group of processes, reactive distillation columns are used.
Here, the heavy boiling educt is continuously fed at the top of the reactive column
section. It has been shown that the reactive batch distillation in a column with a
middle vessel requires a much shorter time than the semi-batch reactor combined with
a rectifying column. In the case studied, the duration of the process was reduced by
about 60%.
The kinetics of the reaction have a very important influence on the separation
process. For slow reaction kinetics, at a constant heat load (i.e. at a constant gas
load), the reflux ratio required for the process drastically increases. For very slow
reactions, the range of reflux ratios required cannot be applied in practice. In this
case, a cyclic reflux policy can be applied. However, this complicates the operation
of the process. Closed processes are therefore used. Here, no products are withdrawn
from the column, thus allowing high reflux and reboil ratios to be easily applied. Two
such processes have been analysed. On the one hand, a batch distillation column
coupled with two reactors can be operated close to total reflux conditions, thus using
the maximum separation capability of the column, and simplifying the control of the
process. However, this process can only be applied when a decomposition reaction
takes place in the liquid phase. On the other hand, a batch reactor can be coupled
with a middle vessel batch distillation column operated in closed mode. In this
case, both reaction products are simultaneously removed from the reactor, thus the
concentration of the reactants is maintained high throughout the process. The rate
of the chemical reaction is therefore maximised, and therefore the duration of the
process is reduced. In the case studied, time savings of up to 70% are obtained.
In Chapter 8, sequential reactions are investigated. If the desired product has the
possibility to react further, its yield in a batch reactor might be very low. This
complicates the separation process, and the required amount of educts is increased.
Therefore, suppressing the side reactions is a very important task in reaction engineer-
ing. Combining a batch (or a semi-batch) reactor with a batch distillation process
allows for the removal of the desired product from the reaction environment. The
volatilities of the components in the mixture play a very important role in process
synthesis. It is shown that if the desired product is an intermediate boiler, the com-
bination of a semi-batch reactor with a batch distillation column delivers high yield
and selectivity. On the contrary, batch reactive distillation columns are not able to
suppress the secondary reactions, since the educts and the products are in contact
within the reactive column section.
95
9. Summary
This thesis has shown that the integration of reaction with batch distillation is very
advantageous for the separation of zeotropic mixtures. However, several technical
mixtures exhibit a non-ideal behaviour, with the formation of separation barriers.
The separation of azeotropic mixtures via batch distillation requires the design of
complex separation processes. When reaction is combined with batch distillation, the
concentration path in the reactor is determined by both phenomena, and therefore
distillation boundaries might be overcome. Hence, further investigation on process
synthesis and analysis of reaction integrated with batch distillation for azeotropic
mixtures is required.
96
References
B.S. Ahmad and P.I. Barton. Solvent recovery targeting for pollution prevention
in pharmaceutical and speciality chemical manufacturing. In AIChE Symposium
Series, volume 90, pages 59–73, 1995.
J. Albet, J.M. Le Lann, X. Joulia, and B. Koehret. Operational policies for the start-
up of batch reactive distillation column. In IChemE Symposium Series, volume
133, pages 63–70. Pergamon Press, 1994.
L.S. Balasubramhanya and F.J. Doyle. Nonlinear model-based control of a batch

reactive distillation column. Journal of Process Control, 10(2-3):209–218, 2000.
D. Barbosa and M.F. Doherty. The influence of equilibrium chemical reactions on

vapor-liquid phase diagrams. Chemical Engineering Science, 43(3):529–540, 1988a.
D. Barbosa and M.F. Doherty. The simple distillation of homogeneous reactive mix-
tures. Chemical Engineering Science, 43(3):541–550, 1988b.
P.E. Cuille and G.V. Reklaitis. Dynamic simulation of multicomponent batch rec-
tification with chemical reactions. Computers and Chemical Engineering, 10(4):
389–398, 1986.
H. Egly and V. Ruby. Betrieb und Regelung energieoptimaler Batch Prozesse.

vt”verfahrenstechnik”, 14(5):351–355, 1980.
S. Engell and G. Fernholz. Control of a reactive separation process. Chemical Engi-

neering and Processing, 42(3):201–210, 2003.
J. Espinosa. Integration of reaction and separation in a batch extractive distillation

column with a middle vessel. In J. Grievink and J. van Schijndel, editors, European
symposium on computer aided prcess engineering-12, pages 187–192, Amsterdam,
2002. Elsevier.
C.A. Farschman and U. Diwekar. Dual composition control in a novel batch distilla-
tion column. Industrial and Engineering Chemistry Research, 37(1):89–96, 1998.
G. Fernholz, S. Engell, L.U. Kreul, and A. Górak. Optimal operation of a semi-

batch reactive distillation column. Computers and Chemical Engineering, 24(2-7):
1569–1575, 2000.
97
REFERENCES
T. Frey and J. Stichlmair. Thermodynamic fundamentals of reactive distillation.

Chemical Engineering and Technology, 22(1):11–18, 1999.
S.B. Gadewar, M.F. Malone, and M.F. Doherty. Selectivity targets for batch reactive
distillation. Industrial and Engineering Chemistry Research, 39(6):1565–1575, 2000.
gPROMS. gPROMS Advanced User Guide. Process Systems Enterprise Ltd., London,
2003.
Z. Guo, M. Ghufran, and J.W. Lee. Feasible products in batch reactive distillation.
AIChE Journal, 49(12):3161–3172, 2003.
R. Krishna. Hardware selection and design aspects for reactive distillation columns.
In K. Sundmacher and A. Kienle, editors, Reactive Distillation - Status and Future
Directions, pages 217–240, Weinheim, 2003. Wiley-VCH.
P. Li, H.A. Garcia, G. Wozny, and E. Reuter. Optimization of a semibatch distillation

process with model validation on the industrial site. Industrial and Engineering
Chemistry Research, 37(4):1341–1350, 1998a.
P. Li, H.P. Hoo, and G. Wozny. Efficient simulation of batch distillation processes
by using orthogonal collocation. Chemical Engineering and Technology, 21(11):
853–861, 1998b.
H.H Mayer and O. Wörz. Reaktorauswahl bei Reaktionen mit überlagerter Destilla-
tion. Chemie Ingenieur Technik, 52(2):174–175, 1980.
R. Monroy-Loperena and J. Alvarez-Ramirez. Output-feedback control of reactive

batch distillation columns. Industrial and Engineering Chemistry Research, 39(2):
378–386, 2000.
I.M. Mujtaba and S. Macchietto. Optimal operation of reactive batch distillation.

AIChE 1992 Annual Meeting, 1992.
I.M. Mujtaba and S. Macchietto. Optimal operation of multicomponent batch distil-

lation - a comparative study using conventional and unconventional columns. Proc.
IFAC Sypmosium ADCHEM’94, Kyoto, pages 401–406, 1994.
M. Nitsche. Probleme der Blasendestillation. Chemie-Technik, 11(3):223–226, 1982.
H. Offers, R. Düssel, and J. Stichlmair. Minimum energy requirement of distillation

processes. Computers and Chemical Engineering, 19(Supplement 1):247–252, 1995.
D. Osorio, R. Perez-Correa, A. Belancic, and E. Agosin. Rigorous dynamic modeling

and simulation of wine distillations. Food Control, 15(7):515–521, 2005.
98
REFERENCES
C.C. Pantelides, D. Gritsis, K.R. Morison, and R.W.H. Sargent. The mathematical
modelling of transient systems using differential-algebraic equations. Computers
and Chemical Engineering, 12(5):449–454, 1988.
E. Reuter, G. Wozny, and L. Jeromin. Modeling of multicomponent batch distilla-

tion processes with chemical reaction and their control systems. Computers and
Chemical Engineering, 13(4-5):499–510, 1989.
I. Rodriguez-Donis, V. Gerbaud, and X. Joulia. Middle vessel heterogeneous batch dis-

tillation of an azeotropic mixture. In R. Gani and S.B. Jørgensen, editors, European
2001. Elsevier.
V. Ruby, H. Egly, and B. Seid. Zeit-und energieminimale Betriebsweise diskontinuier-

licher Prozesse. vt”verfahrenstechnik”, 12(7):421–425, 1978.
R. Schneider, C. Noeres, L.U. Kreul, and A. Górak. Dynamic modeling and simulation
of reactive batch distillation. Computers and Chemical Engineering, 25(1):169–176,
2001.
S. Skogestad, B. Wittgens, R. Litto, and E. Sørensen. Multivessel batch distillation.

AIChE Journal, 43(4):971–978, 1997.
S. Skouras and S. Skogestad. Separation of ternary heteroazeotropic mixtures in a

closed multivessel batch distillation-decanter hybrid. Chemical Engineering and
Processing, 43(3):291–304, 2004a.
S. Skouras and S. Skogestad. Time (energy) requirements in closed batch distillation

arrangements. Computers and Chemical Engineering, 28(5):829–837, 2004b.
E. Sørensen, S. Macchietto, G. Stuart, and S. Skogestad. Optimal control and on-line

operation of reactive batch distillation. Computers and Chemical Engineering, 20
(12):1491–1498, 1996.
E. Sørensen and M. Prenzler. A cyclic operating policy for batch distillation - the-
ory and practice. Computers and Chemical Engineering, 21(Supplement 1):S1215–
S1220, 1997.
E. Sørensen and S. Skogestad. Control strategies for reactive batch distillation. Jour-
nal of Process Control, 4(4):205–217, 1994.
E. Sørensen and S. Skogestad. Comparison of regular and inverted batch distillation.

Chemical Engineering Science, 51(22):4949–4962, 1996.
J.G. Stichlmair and J.R. Fair. Distillation - Principles and Practice. Wiley and Sons,
New York, 1998.
99
REFERENCES
A.J.V. Underwood. Fractional distillation of multicomponent mixtures. Chemical

Engineering Progress, 44:603–614, 1948.
G. Venimadhavan, M.F. Malone, and M.F. Doherty. A novel distillate policy for batch
reactive distillation with application to the production of butyl acetate. Industrial
and Engineering Chemistry Research, 38(3):714–722, 1999.
R.M. Wajge and G.V. Reklaitis. RBDOPT: a general-purpose object-oriented mod-

ule for distributed campaign optimization of reactive batch distillation. Chemical
Engineering Journal, 75(1):57–68, 1999.
M. Warter. Batch-Rektifikation mit Mittelbehälter. PhD thesis, Technische Universität

München, 2001.
M. Warter, D. Demicoli, and J. Stichlmair. Batch distillation of zeotropic mixtures in

a column with a middle vessel. In J. Grievink and J. van Schijndel, editors, European
2002. Elsevier.
M. Warter, D. Demicoli, and J. Stichlmair. Operation of a batch distillation col-

umn with a middle vessel: experimental results for the separation of zeotropic and
azeotropic mixtures. Chemical Engineering and Processing, 43(3):263–272, 2004.
M. Warter and J. Stichlmair. Batch distillation of azeotropic mixtures in a column

with a middle vessel. In S. Pierucci, editor, European symposium on computer aided
prcess engineering-10, pages 691–696, Amsterdam, 2000. Elsevier.
M. Warter and J. Stichlmair. Batch-Rektifikation mit Mittelbehälter. Chemie Inge-

nieur Technik, 74(9):1195–1206, 2002.
B. Wittgens and S. Skogestad. Closed operation of multivessel batch distillation:

Experimental verification. AIChE Journal, 46(6):1209–1217, 2000.
Z. Xu and M.P. Dudukovic. Modeling and simulation of semi-batch photo reactive

distillation. Chemical Engineering Science, 54(10):1397– 1403, 1999.
100
A. Batch reactor coupled with a
middle vessel column
The closed operation of a reactor coulped with a middle vessel batch distillation
column is discussed in Section 6.1.4, and is shown Figure 6.13 (i). In this process
the concentration in the middle vessel (xM ) can be maitained constant if the internal
reflux ratio of the column and the stoichiometric coefficients are related according to
Equation (6.9) . The relationship between the reflux ratios can be derived by mass
balances around the distillate, middle, and bottom vessels.
Mass balance around the distillate vessel:
! !
dD L̇
= V̇U − L̇U = V̇U · 1 − (A.1)
dt V̇ U
Mass balance around the bottom vessel:

! !
dB L̇
= L̇L − V̇L = V̇L · −1 (A.2)
dt V̇ L
Component balances around the middle vessel:

dM d(D · xD ) d(B · xB )
xM · =− − +ν ·r·M (A.3)
dt dt dt
Under the condition that the concentrations in the distillate and bottom vessels are
constant, i.e. dxD /dt = 0 and dxB /dt = 0, Equation (A.3) can be re-written as:
dM dD dB
xM · = −xD · − xB · +ν ·r·M (A.4)
dt dt dt
With xDc = xBa = 0:
dM dD
xM a · = −xDa · + νa · r · M (A.5)
dt dt
dM dB
xM c ·= −xBc · + νc · r · M (A.6)
dt dt
And dividing Equation (A.5) by Equation (A.6)
xM a νa −xDa · dD/dt + νa · r · M
= = (A.7)
xM c νc −xBc · dB/dt + νc · r · M
101
A. Batch reactor coupled with a middle vessel column

dB dD
νa · −xBc · + νc · r · M = νc · −xDa · + νa · r · M (A.8)
dt dt
In combination with Equations (A.1) and (A.2), and under the assumptions that
xDa = xBc , and VU = VL :
! ! ! !
L̇ L̇
νa · − 1 = νc · 1 − (A.9)
V̇ L V̇ U
! !
L̇ L̇
νa · + νc · = νa + νc (A.10)
V̇ L V̇ U
Therefore Equation (6.9) results:

! !
νa L̇ νc L̇
· + · =1 (A.11)
νa + νc V̇ νa + νc V̇
L U
102

Novel Batch Distillation Processes For The Separation of Systems With and Without Chemical Reaction

Uploaded by

Copyright:

Available Formats

Novel Batch Distillation Processes For The Separation of Systems With and Without Chemical Reaction

Uploaded by

Document Information

Original Title

Copyright

Available Formats

Share this document

Share or Embed Document

Sharing Options

Did you find this document useful?

Is this content inappropriate?

Copyright:

Available Formats

Novel Batch Distillation Processes For The Separation of Systems With and Without Chemical Reaction

Uploaded by

Copyright:

Available Formats

Lehrstuhl für Fluidverfahrenstechnik

Technische Universität München

Novel Batch Distillation Processes for

Vollständiger Abdruck der von der Fakultät für Maschinenwesen der

Vorsitzender: Univ.-Prof. Dr.-Ing. D. Weuster-Botz

Prüfer der Dissertation:

Die Dissertation wurde am 14.09.2005 bei der Technischen Universität München

3. Aim and outlook of the thesis 13

4. Modelling and simulation 15

5. Batch Distillation in a Column with Side Withdrawal 19

6. Integration of batch reaction with batch distillation - Equilibrium limited

7. Integration of batch reaction with batch distillation - Kinetically con-

8. Integration of batch reaction with batch distillation - Sequential reactions 81

A. Batch reactor coupled with a middle vessel column 101

V̇ Vapour flow rate , mol/s

a, b, c, d Components ordered in increasing boiling point temperature

ω End of the process

2.1. Batch distillation processes

Figure 2.1.: Schematic representation of the regular batch distillation process.

Operation modes for batch distillation processes

Figure 2.3.: Closed operation of a batch distillation in middle-vessel column under

Minimum energy demand

2.2. Integration of batch reaction with batch

2.2.1. Coupling of batch reactor with a distillation unit

Control of a batch reactor coupled with a rectifying column

Coupling a batch reactor with different column configurations

Overcoming distillation boundaries

Suppressing side reactions

2.2.2. Batch reactive distillation

2.3. Modelling and simulation

4.1. Simulation software

4.2. Modelling of batch distillation column

yn, i · pn = γn, i · p0n, i · xn, i (4.2)

yn, i · pn = αi, d · p0n, d · xn, i (4.3)

From the definition of mole fraction the following summations result:

Furthermore, the principle of conservation of energy gives the following energy

As already discussed in the literature review in Section 2.3, if the hydrodynamics

0 = L̇n−1 · hLn−1 + V̇n+1 · hVn+1 − V̇n · hVn − L̇n · hLn +

Figure 5.1.: Batch distillation column with side withdrawal.

5.2. Description of the process

constants assigned in the simulation are summarised in Table 5.1.

Molar fractions in the charge (xa , xb , xc ) (0.1, 0.8, 0.1)

(i) Start-up (ii) Production step

lim (L̇/V̇ )L = lim (L̇/V̇ )U = 1 ⇒ lim Ẇ = 0 (5.2)

5.3. Parametric studies

Molar fractions in the charge (xa , xb , xc ) (0.1, 0.8, 0.1)

5.3.1. Geometric parameters

Number of equilibrium stages

Position of withdrawal stage

5.3.2. Operational parameters

Composition of the charge

open operation mode

close operation mode

Figure 5.8.: Effect of the composition of the intermediate-boiling component in the

Termination criteria for the first process step

Effect of the internal reflux ratios

5.4. Comparison with the middle vessel batch

Figure 5.10.: Batch distillation column with a middle vessel.

Table 5.3.: Simulation parameters for Section 5.4.