Research Article

Received: 22 December 2014 Revised: 26 January 2015 Accepted article published: 5 February 2015 Published online in Wiley Online Library:

(wileyonlinelibrary.com) DOI 10.1002/jctb.4661

Overcoming any configuration limitation: an

alternative operating mode for pervaporation
and vapour permeation
Giuseppe Genduso,a* Patricia Luisb and Bart Van der Bruggena

Abstract
BACKGROUND: Membranes are the core and the main limitation of pervaporation. For example, in certain cases (more often for
organic–organic mixtures) a pure product is not attainable as a permeate of a single stage and a series of membrane stages
spaced by a condenser is required, making the process economically unacceptable.

RESULTS: This work discusses the use of an alternative approach by exploring the multi-stage-batch-pervaporation (MSBP) unit
operation. Here, the permeate obtained after each batch-stage is recycled to the feed tank to increase the permeate product
purity in a following stage. The separation of methanol–methyl acetate mixtures was chosen as a case study.

Simulations demonstrate how a multi-stage-batch-pervaporation unit is able to meet product purity requirements, by varying
the stage-termination condition and the number of stages, employing a single membrane-module and a single condenser.

Moreover, a new way to visualize pervaporation separation performance of different membranes is proposed to replace the
pervaporation separation index (PSI) analysis.

CONCLUSION: Being very flexible, the multi-stage-batch-pervaporation unit operation could be of benefit for all small batch
productions, even when using medium-low performance membranes. Parts of continuous productions could also be suitable
for multi-stage-batch-pervaporation systems.
© 2015 Society of Chemical Industry

Keywords: batch pervaporation; configuration limitation; pervaporation membrane limitation

LIST OF SYMBOLS AND ABBREVIATIONS t process time

BPTI batch pervaporation time index 𝛽 separation factor
MM20
BPTI98 batch pervaporation time index employed for BPTIo98 standardized version of the batch pervaporation
separations in which each stage ends once a time index for separations that start from an initial
constant purity of the component enriched in the feed concentration of 50 mole% and end when the
retentate is reached purity of the component enriched in the retentate
MM20
BPTIrec,98 batch pervaporation time index employed for reaches a value of 98 mole%.
separations in which each stage ends once a
constant value of stage-recovery of the component
enriched in the permeate is reached
INTRODUCTION
MM20 industrial stream made of ∼30 wt% methyl acetate The process of pervaporative separation consists of two steps:
in methanol selective dissolution and diffusion of the components of a liquid
F0 initial amount in the feed tank charge through a dense membrane and its consequent partial
x 1,0 initial concentration in the feed tank evaporation. The driving force for permeation is the difference
in chemical potential of components at the two sides of the
F instantaneous amount in the feed tank
j1 instantaneous flux of the component enriched in
the permeate ∗ Correspondence to: Giuseppe Genduso, Department of Chemical Engineering,
x1 instantaneous molar concentration in the feed tank
Process Engineering for Sustainable Systems (ProcESS), KU Leuven, W. de Croy-
y1 instantaneous molar concentration in the permeate laan 46, B-3001 Leuven, Belgium. E-mail: [email protected]
tank
A membrane area a Department of Chemical Engineering, Process Engineering for Sustainable
Systems (ProcESS), KU Leuven, W. de Croylaan 46, B-3001 Leuven, Belgium
MSBP multi-stage-batch-pervaporation
PSI pervaporation separation index b Materials & Process Engineering (iMMC-IMAP), Université catholique de Lou-
PVDF polyvinylidene fluoride vain, Place Sainte Barbe 2, 1348 Louvain-la-Neuve, Belgium

J Chem Technol Biotechnol (2015) www.soci.org © 2015 Society of Chemical Industry

 www.soci.org G Genduso, P Luis, B Van der Bruggen

membrane.1 A common dense membrane consists of a dense layer

oriented towards the feed/retentate side in the pervaporation cell
on top of a porous supporting layer oriented towards the perme-
ate side in which vaporization occurs.2 The selectivity of perva-
porative separation depends on the different transport rates of
compounds, which in turn depends on their diffusivity and solubil- 5
ity in a dense membrane layer. Thus, the separation is not affected 3 2 4
by the presence of an azeotrope.3 – 6 Moreover, pervaporation has
the advantage of being a low energy consumption technology,2,7
economical and eco-friendly.8
Pervaporation membranes with very high performance can
be found mainly in the pervaporative dehydration of organic 1
solvents, because water is small and diffuses faster than organic
compounds.9 – 13 However, as reported by Smitha et al.,8 suc- 6
cessful separations have also been reported more recently
in the pervaporation of organic–organic mixtures involving:
Figure 1. Baker et al.15 batch pervaporation multi-tank apparatus:
aromatic–alicyclic compounds, e.g. benzene–cyclohexane; (1) = pervaporation membrane module; (2) = ‘first process’ tank;
aromatic–aliphatic compounds, e.g. benzene–n–hexane; iso- (3) = ‘holding tank’; (4) = ‘second process tank’; (5) = ‘concentrate tank’;
mers; and the separation of organic polar–apolar compounds, (6) = condenser.
e.g. methanol/toluene–benzene or ethanol–ethyl tert-butyl ether
mixtures. The separation of methanol–organic compounds is of a high purity permeate are still strictly linked to the separation
an interesting sub-case. As an organic compound, methanol is performance of the membrane.
very similar to water, which may lead to the same high level of In 1993 Baker et al.15 published a patent describing a batch per-
separation when the right membrane is employed. For example, vaporation system, composed of different service tanks connected
the organic–organic separation of methanol from larger apolar in a network and referring to the same membrane module (Fig. 1).
molecules, e.g. methyl tert-butyl ether or toluene has already been The separation of methyl acetate–methanol mixtures is an
proven to have high separation factors.8,14 interesting case study to test a batch approach similar to the one
However in general, for medium–low separation factor mem- suggested by Baker et al. Such a stream can be found in the context
branes, as is the case during organic–organic pervaporation of of the transesterification of methyl acetate into n-butyl acetate.1
similar molecules, to obtain high purity products, a pervaporation This economically favorable conversion, starts from a waste
unit should be arranged in a series/parallel configuration of mem- stream, the so-called MM20 stream, made of ∼30 wt% methyl
brane modules. In particular, when the objective is the depletion acetate in methanol.18,19 This conversion process is in an advanced
of a product from a retentate stream, a multi-step pervapora- stage of research but still not implemented in industry18 – 22 and is
tion configuration, i.e. a series of membranes modules using the shown in Fig. 2, where the first block represents the concentration
retentate of the previous module as the feed, is employed.15 A step of the MM20 to higher concentrations of methyl acetate. This
historical case, reported by Rautenbach and Albrecht16 is the GFT™ step is needed to obtain high conversions into n-butyl acetate.
pervaporation unit employed during continuous dehydration of The azeotropic nature of this mixture (33 mole% methanol in the
ethanol where the retentate is depleted from water employing feed at 1 atm and about 54 ∘ C), (Genduso G and Farrokhzad H et
a series of PVA-PAN membrane steps and including re-heating al., accepted for publication) suggests the use of advanced distil-
after each step. In contrast, when the objective is to obtain a pure lation methods, e.g. extractive distillation23 with water or ethylene
permeate, a multi-stage configuration can be designed, where the glycol, to obtain pure methyl acetate.18 Pervaporation may also
permeate product exiting a membrane module is first condensed be considered for the same purpose. However, to date the liter-
and then further purified in a subsequent module.15 However, ature concerning pervaporative separation of methanol–methyl
the need for a condenser after each stage makes a long series of acetate mixtures is still limited. Generally, since methanol is a
pervaporation stages unpractical and uneconomic. small and fast diffusing molecule, methanol-selective membranes
From a theoretical point of view pervaporation provides a are thought realistic.21,24 – 27 Recently, polyvinylidene fluoride
promising partial replacement of distillation as the separation (PVDF) membranes were suggested as membranes with methyl
based on the vapor liquid equilibrium can be enhanced by adding acetate selectivity, however, even with very high fluxes, up to
the contribute of the membrane-component affinity;17 however, 35 kg m−2 h−1 , the membrane had quite low separation factors.23
from a practical point of view since pervaporation does not allow A review of permeate/feed concentration data, pervaporation
the design of a cascade of contacting stages, typical of a distilla- McCabe–Thiele diagram, and total flux data of literature mem-
tion column, or the design of a long series of pervaporation stages, branes employed for pervaporation of methanol-methyl acetate
when high purity products are required, it is necessary to develop mixtures is shown in Fig. 3(a)–(c).
and employ highly selective membranes with high separation What is presented in the following parts of this paper is a recon-
factors. This means that pervaporation suffers from configura- sideration of Baker et al.’s work, using today’s state-of-the-art
tion limitations and the separation is strongly dependent on the membranes. In the first part, a multi-stage-batch-pervaporation
separation performance of the membrane. (MSBP) unit is presented and simulated for a methyl
Operation of pervaporation in a batch-mode has one extra acetate–methanol separation case study employing all mem-
degree of freedom over continuous-mode operation due the fact branes reported in Fig. 3. In particular two new MSBP operating
that time is a parameter in the separation. In particular the purity modes are proposed showing MSBP unit application potential and
of the retentate can be improved by varying the batch operation limits. In addition a new way to index batch pervaporation mem-
time; nevertheless, component recoveries and the achievement brane separation effectiveness is also presented and discussed.

wileyonlinelibrary.com/jctb © 2015 Society of Chemical Industry J Chem Technol Biotechnol (2015)

Overcoming any configuration limitation www.soci.org

Figure 2. Block scheme of process to treat MM20 waste stream to obtain n-butyl aceate product stream.

(a)
Total flux [kg*m-2*h-1]

14
12 Pervap 2255-30 (40 °C)
10 Pervap 2255-40 (45 °C)
Pervap 2255-50 (45 °C)
8
Pervap 2255-60 (45 °C)
6
Cuprophan (45 °C)
4 Nylon-6 (40 °C)
2 PolyAl Typ M1 (44 °C)
0
0.0 0.2 0.4 0.6 0.8 1.0
Feed molar concentration in methanol [mol/mol]

(b) 40
Total flux [kg*m-2*h-1]

35 Polyvinylidene fluoride (44 °C)

30
25
20
15
10
5
0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9
Feed molar concentration in methanol [mol/mol]

(c) 1.0
in the permeate/vapor, [mol/mol]

0.8
Pervap 2255-30 (40 °C)
Methanol concentration

Pervap 2255-40 (45 °C)

Pervap 2255-50 (45 °C)
0.6 Pervap 2255-60 (45 °C)
Cuprophan (45 °C)
PolyAl Typ M1 (44 °C)
Polyvinylidene fluoride (44 °C)
0.4 Nylon-6 (40 °C)

VLE (1 atm)
0.2 diagonal

0.0
0.0 0.2 0.4 0.6 0.8 1.0
Methanol concentration in the feed/liquid, [mol/mol]

Figure 3. (a) and (b), total flux data (weight-basis) variation with feed methanol concentration of various commercial and in-house synthesized membranes
employed during methanol–methyl acetate pervaporation (working temperature in the range 40–45 ∘ C).21,24 – 26,23,27 (c) Permeate/feed concentration
literature data of all these membranes. The same graph also reports the vapour–liquid-equilibrium curve obtained using the UNIQUAC property method.

J Chem Technol Biotechnol (2015) © 2015 Society of Chemical Industry wileyonlinelibrary.com/jctb

 www.soci.org G Genduso, P Luis, B Van der Bruggen

( )
Feed tank d F × x1
− = j1 × A (2)
Pervaporation dt
stack
where A is the membrane area, F is the instantaneous amount
in the feed tank, j1 is the instantaneous flux of the component
enriched in the permeate, x 1 is the instantaneous molar concentra-
Condenser
tion in the feed tank of the component enriched in the permeate,
Service line y1 the instantaneous molar concentration in the permeate of the
After each batch stage component enriched in the permeate.
Combining (1) and (2) leads to (3), which is similar to the Rayleigh
equation for batch distillation;28 here, however, the thermody-
After each batch stage
namic equilibrium relation between the vapour and liquid phases
is replaced by the separation characteristic of the membrane.
Permeate tank
Retentate (1th product)
dF dx1
storage tank = (3)
(2nd product) F y1 − x1

Figure 4. Scheme of the MSBP unit studied in this paper. All membrane separation data were derived from experimentally
obtained transport and separation data, i.e. methanol molar flux
and variation of permeate concentration with methanol feed
METHODS concentration, were obtained from the literature (see Fig. 3).
A multi-stage-batch pervaporation unit: terminology Fixing appropriate boundary conditions, i.e. x 1 and the initial
and model equations concentration in the feed tank x 1,0 , (3) results in (4) and (5).
The MSBP unit (Fig. 4) used in this work is comprised of three
x1,0
service tanks, in place of the four in the general version of Baker F0 dx1
ln = (4)
et al.’s unit (Fig. 1); in fact, according to the new operating modes F ∫x1 y1 − x1
proposed in this paper, a 4th tank is not needed.
Hence, in order to increase the purity of the permeate to be
F = F0 × exp (−z) (5)
obtained as the product, at the end of each stage what is left
in the feed tank is sent to the storage tank, and what has been where,
permeated, condensed and collected in the permeate tank, is x1,0
dx1
returned to the feed tank for further purification. Therefore, the z= (6)
∫ x1 y1 − x1
retentate product (hereinafter always reported referring to the
concentration of the component not selective for the membrane By combining (1) and (4) the time dependency can be obtained
[e.g. methyl acetate concentrations, if the membrane is methanol (Equation (7)), more relationships derived from this model can be
selective]) is the product accumulated in the storage tank; the found elsewhere.29
permeate product (hereinafter always reported in concentration
unit of component selective for the membrane) is the product F0 x1,0 y1 exp (−z)
t= dx1 (7)
obtained at the end of the entire MSBP process and contained in A ∫x1 y1 − x1 j1
the permeate tank; the stage-termination condition is the condition
that terminates an MSBP stage; the process-termination condition
MATLAB simulations
is the condition that ends an MSBP process; the operating mode
To simulate the MSBP-modus operandi, MATLAB® was used
indicates the way in which the MSPB unit is run and depends on
assuming that: (i) the storage and permeate tanks are perfectly
the stage-termination condition; a component stage-recovery and
mixed; (ii) the membrane adapts infinitely fast to the new concen-
total-recovery are the ratios between the amount of component
trations in the feed; and (iii) the time required for liquid circulation
at the end (in the respective tanks) over the amount of the same
is negligible compared with the processing time. All the integra-
component (in the feed tank) at the beginning of the stage or at
tions were performed employing numeric integration on very
the very beginning of the process, respectively.
short intervals of feed concentration, i.e. lower than 0.01 mole%
Considering that the energy of permeation comes from the
methanol, to minimize the integration error.
retentate side and that during condensation the permeate rejects
heat to the cooling utility, during each stage both retentate and
permeate have to be heated employing a heat exchanger system. Two MSBP termination conditions
However this step is not studied and discussed here as it is outside As already described, medium–low performance membranes may
the scope of this study. lead to the design of expensive pervaporation units made of a
The model equations of this process unit are the same unsteady long series of membrane modules, discouraging the use of this
state mass balance equations of a batch-pervaporation unit (Fig. 4), technology. The MSBP unit described earlier may allow this prob-
with the sole difference that in the case of MSBP-mode, for each lem to be overcome. Here, after each batch-stage it is possible to
stage the initial conditions result from the outcomes of the previ- increase the purity of the permeate product by recycling it to the
ous stage and the time is cumulative. feed tank using a single condenser system. Figure 5(a), (b) shows
two proposed (basic) operating modes, represented schematically,
dF j to operate an MSBP unit (Table 1 includes all stage-termination
− = 1 ×A (1) conditions and process-termination conditions for both these
dt y1

wileyonlinelibrary.com/jctb © 2015 Society of Chemical Industry J Chem Technol Biotechnol (2015)

Overcoming any configuration limitation www.soci.org

(a) (b)

Figure 5. Two schemes for variation of compositions of permeate vs. retentate with the number of stages: (a) multi-stage-batch-pervaporation mode
in which each stage is terminated after reaching a certain value of stage-recovery of the component enriched in the permeate; and (b) each stage is
terminated when a constant purity value of the retentate product is reached.

Table 1. Stage-termination conditions and process termination conditions of the two operating modes used in this work during MSBP simulations

MSBP unit operating mode End-stage condition End-process condition

Constant purity of the component fixed purity (i.e. 98 mole%) of the fixed purity (i.e. 98 mole%) of the
enriched in the retentate component enriched in the retentate component enriched in the permeate
Constant stage-recovery of the component fixed stage-recovery of the component fixed purity (i.e. 98 mole%) of the
enriched in the permeate enriched in the permeate component enriched in the permeate

operating modes). Figure 5(a) and (b) were produced simulating an influences x 1 , the choice of the stage-termination condition,
imaginary membrane separation; in particular, Fig. 5(a) shows the which may also influence x 1 , and the initial concentration of the
case in which each stage is terminated once a fixed stage-recovery feed mixture at t = 0, i.e. x 1,0 :
of the component enriched in the permeate is reached, and
x1,0
Fig. 5(b) the case in which each stage is terminated when reach- A y1 exp (−z)
BPTI = t = dx1 (8)
ing a desired retentate product purity. Starting from the feed point F0 ∫x1 y1 − x1 j1
and following the time direction it is possible to see the dynam-
ics of the purity of the permeate and retentate products for each Three versions of the batch pervaporation time index are used:
MM20
stage. When terminating each stage at the desired retentate prod- (i) BPTIrec,98 for separations in which each stage ends once a
uct purity (Fig. 5(b)), at the end of the entire process it is possible constant value of stage-recovery of the component enriched in the
MM20
to obtain both products, i.e. storage and permeate tank products, permeate is reached (Table 1); (ii) BPTI98 for separations in which
with the desired purities. On the other hand, when terminating each stage ends once a constant purity of the component enriched
each stage after reaching a certain value of stage-recovery of the in the retentate is reached; and (iii) a standardized version of the
component enriched in the permeate, only the permeate prod- batch pervaporation time index (BPTIo98 ), for separations that start
uct can be extracted pure as the membrane limits the retentate from an initial feed concentration of 50 mole% and end when the
product purity. However, in this case the retentate product accu- purity of the component enriched in the retentate reaches a value
mulated in the storage tank can be recycled to the feed tank for of 98 mole%.
further processing or be treated by other separation systems, e.g. Furthermore, the diagram BPTIo98 - yo the batch pervaporation-
pervaporation, distillation, etc. membrane performance graph, i.e. the normalized
These two ways of operating the MSBP unit are basic examples. permeate-product concentration, (y1 − x 1,0 )/(1 − x 1,0 ), in this work
More elaborate combinations may be designed and employed is compared with the results of a pervaporation separation index
depending on the particular separation. However, it is not the (PSI) analysis. PSI, is defined by Huang30,31 as the product of sep-
purpose of the present work to explore them. aration factor and total flux values for each feed concentration.32
Here a modified version of this index widely accepted in the
literature is used:33,34
The batch pervaporation time index
In the ‘Results and discussion’ section all time data-points are PSI = J ∗ (𝛽 − 1) (9)
reported in a normalized form.
The so-called batch pervaporation time index (BPTI), (Equation (8)) Employing this version of the PSI-index, when 𝛽 is equal to 1 no
depends on: (i) the membrane separation performance which separation occurs and PSI = 0.

J Chem Technol Biotechnol (2015) © 2015 Society of Chemical Industry wileyonlinelibrary.com/jctb

 www.soci.org G Genduso, P Luis, B Van der Bruggen

final purity as a function of the stage-recovery of the component

Table 2. Number of stages, BPTI and percentage of permeate prod-
uct over initial feed amount for all simulated separations at different enriched in the permeate.
stage-recoveries of the component enriched in the permeate From a theoretical point of view all the membranes included in
Fig. 3 would be able to yield a permeate with the specified concen-
Percentage tration. However, during intermediate stages all the membranes
of total initial that lead to very low separation performances consume almost all
feed in permeate Permeate- the feed in order to satisfy the stage-termination condition and the
No. tank at end of product stage purity requirement for the permeate product (process-termination
Membrane MM20
stages BPTIrec,98 the process recovery condition). This results in very low permeate product amounts.
Thus in this case a non-feasible application is obtained.
PERVAP 2255-40 4 1.80 0.70 0.3
This is the case for the PolyAl Typ M1 and the PVDF mem-
5 5.73 6.65 0.6
branes, which were not able to reach the required specification,
5 9.48 20.44 0.75
i.e. stage-recovery of the component enriched in the permeate
7 19.39 41.17 0.9
of 0.3, 0.6, 0.75 or 0.9 together with 98 mole% purity for the per-
PERVAP 2255-50 4 4.29 0.71 0.3
meate product; hence, their simulation results are not reported in
5 13.60 6.71 0.6
Table 2 and Fig. 6. The same situation was found for the Pervap
5 22.41 20.47 0.75
2255–40, Pervap 2255–50 and Cuprophan membranes when set-
7 45.51 41.17 0.9
ting a stage-recovery of the component enriched in the permeate
PERVAP 2255-60 3 6.74 2.34 0.3
of 0.3, as can be noticed looking at the final permeate/initial feed
4 20.47 11.11 0.6
amount ratio in Table 2. These data are also not reported in Fig. 6.
5 35.22 20.29 0.75
Table 2 and Fig. 6(a) show how high performance membranes,
6 65.37 45.69 0.9
such as Nylon-6, ensure a high permeate product purity (>98
Nylon-6 1 0.74 25.39 0.3
mole%) together with a high final permeate/initial feed amount
1 1.57 50.73 0.6
ratio.
1 2.05 63.42 0.75
In contrast, membranes with a medium separation perfor-
1 2.58 76.09 0.9
mance, i.e. Pervap 2255–40, Pervap 2255–50, Pervap 2255–60,
Cuprophan 4 4.01 0.00 0.3
and Cuprophan membranes allow a smaller amount of permeate
7 12.68 2.43 0.6
product with a high purity and a higher retentate product amount
9 22.84 6.48 0.75
with a lower purity to be obtained.
13 51.14 21.90 0.9
Figure 6(b) indicates that for low stage-recoveries, only mem-
branes with high separation factors are able to ensure retentate
products with sufficient purity. Moreover it also indicates that
for very high stage-recoveries it is also possible to obtain pure
RESULTS AND DISCUSSION retentate products. There is a power-like trend of the curve of
Stage-termination at a fixed stage-recovery of the the concentration of the component enriched in the retentate vs.
component enriched in the permeate stage-recovery of the component enriched in the permeate.
MM20
Table 2 reports the number of stages, the BPTIrec,98 and the final
amount of permeate as a fraction of the initial feed, obtained Stage-termination at fixed retentate product purity
by varying the stage-recovery of the component enriched in the To obtain both products at the desired purity each stage can be
permeate, i.e. stage-termination condition, for the separation ended once a constant purity for the retentate product is reached,
of an MM20 methanol–methyl acetate stream. Moreover, Fig. 6 i.e. stage-termination mode described in Fig. 5(b). In this case,
shows: (a) permeate product final purity; and (b) retentate product Fig. 7(a) and (b) show, respectively, the variation of permeate

(a) (b)
1.000 0.7
Permeate tank product final purity, [mol/mol]

Storage tank product final purity, [mol/mol]

Pervap 2255-40
0.975 0.6 Pervap 2255-50
Pervap 2255-60
Cuprophan
0.950 0.5 Nylon-6

0.850 0.4

0.800 0.3

0.750 0.2

0.700 0.1
0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1.0 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1.0
Permeate tank product stage-recovery Permeate tank product stage-recovery

Figure 6. Purity of permeate (a) and retentate (b) products against variation in stage-recovery of the component selective for membrane (stop condition
for each MSBP simulation). The data for PVDF and PolyAl Typ M1 are not represented since the two membranes were not able to give a real separation.

wileyonlinelibrary.com/jctb © 2015 Society of Chemical Industry J Chem Technol Biotechnol (2015)

Overcoming any configuration limitation www.soci.org

(a) (b)
1.00 1.0

of the component enriched in the retentate

Permeate-tank-product purity, [mol/mol]
0.9

0.95 0.8
0.7

Total recovery
0.90 0.6
0.5
0.4
0.85
0.3

0.60 0.2
0.40
0.1
0.20
0.00 0.0
1 10 100 1000 1 10 100 1000
Number of batch pervaporation stages (log10 scale) Number of batch pervaporation stages (log10 scale)

(c) 10000
BPTI98 (log10 scale), [h*m2/kmol]

1000

Pervap 2255-40
Pervap 2255-50
100 Pervap 2255-60
Nylon-6
Cuprophane
PolyAl Typ M1
10 PVDF

1
1 10 100 1000
Number of batch pervaporation stages (log10 scale)

Figure 7. Variation of purity of the permeate product (a), total recovery of the component enriched in the retentate (b) and pervaporation cumulative
time (c) with variation of the number of batch-stages for all the membranes treated in this work (with the exception of the Pervap 2255–30 membrane)
for a temperature of 40–45 ∘ C (depending on the membrane).

MM20 and product recoveries esti-

Theoretically, with an MSBP working at constant retentate prod-
Table 3. Number of stages, BPTI98 uct purity in the stage-termination condition, all membranes
mated when both products reach a purity of 98 mole%
would be able to give both products at high purity; however, prac-
Retentate tically the PolyAl Typ M1, PVDF and Pervap 2255–40 membranes
product Permeate were not able to ensure the specification for the permeate prod-
recovery product uct (i.e. 98 mole% purity) even after 1000 batch stages, indicated
No. of (methyl recovery in Table 3 by the infinity [∞] symbol.
Membrane stages MM20
BPTI98 acetate basis) (methanol basis) It has to be pointed out that the time depends on the membrane
area and the initial amount of mixture in the feed tank (as seen
Pervap 2255-40 >1000 ∞ - -
from Equations (7) and (8)). The higher the membrane area for
Pervap 2255-50 40 379 0.90 ∼1.00
Pervap 2255-60 40 575 0.93 ∼1.00
fixed initial feed amounts, the shorter the stage-time. Thus, mem-
Nylon-6 2 5 0.94 ∼1.00 branes requiring a large number of stages to obtain the required
Cuprophan 140 845 0.89 ∼1.00 purities may need a reasonable time to complete the separation
PolyAl Typ M1 >1000 ∞ - - when an appropriate membrane area is used with an automatic
PVDF >1000 ∞ - - system of valves and pumps (needed to pass from one stage to
the following one).

product purity and total recovery of the component enriched in Batch pervaporation membrane performance representation
the retentate with the number of batch-stages. Figure 7(c) shows (a new way)
the relation between the number of stages and time expressed in Generally, during selection of the right membrane for a batch
the normalized form. Table 3 shows the normalized time and the pervaporation separation, PSI analysis is often taken as a reference.
number of stages needed to obtain both products with a purity of For the particular mixture presented in this text the PSI analysis
98 mole%. is shown in Fig. 8(a), (b). If for example the feed mixture has a

J Chem Technol Biotechnol (2015) © 2015 Society of Chemical Industry wileyonlinelibrary.com/jctb

 www.soci.org G Genduso, P Luis, B Van der Bruggen

(a) (b)
2400 2400
PVDF
1600 Pervap 2255-30 (40 °C) 1600
800 Pervap 2255-40 (45 °C) 800
Pervap 2255-50 (45 °C)
Pervap 2255-60 (45 °C)
16 16

PSI, [kg/(m2*h)]
Cuprophan (45 °C)
PSI, [kg/(m2*h)]

14 PolyAl Typ M1 (44 °C) 14

Nylon-6 (40 °C)
12 12
10 10
8 8
6 6
4 4
2 2
0 0
0.0 0.2 0.4 0.6 0.8 1.0 0.2 0.4 0.6 0.8
Feed molar concentration in methanol, [mol/mol] Feed molar concentration in methanol, [mol/mol]

Figure 8. (a) Pervaporation separation indexes (PSI) of all the commercial membranes reported in the literature;21,24 – 26,23,27 (b) PSI analysis for the only
methyl acetate selective membrane (i.e. the PVDF membrane). Polynomial fitting curves are also shown.

composition of 50 mole% in the feed, the membrane ranking in 1.0

Pervap 2255-40
order of decreasing separation performances is: Pervap 2255-50
Pervap 2255-60
0.8 Nylon-6
Nylon-6 >> Pervap 2255–40 > PolyAl Typ M1≈ Pervap 2255-50 Cuprophan
hig PolyAn TypM1
he
> PVDF > Pervap 2255-60 ≈ Cuprophan (y1-x1,0)/(1-x1,0) rp
erf
orm
PVDF
0.6 an min. performance line
(Ranking 1) ce
sep
ara
tion
s
However, for higher or lower concentrations than 50 mole% in 0.4
the feed, this ranking changes completely. In particular from low
to medium feed concentrations in the component selective for the
membrane, which is the optimal range from an energetic point of 0.2
view, since the component selective for the membrane is present
in a lower amount, the PSI analysis suggests the selection of
0.0
Nylon-6 membrane, or PolyAl Typ M1 membranes. Instead, when 0 20 40 60 80
considering the medium–high feed concentration range, the best
BPTI98, [h*m2/kmol]
choices would be Nylon-6 or Pervap 2255-40 membranes.
A different result is obtained when representing the normalized
Figure 9. The batch pervaporation-membrane performance graph. In this
permeate product purity obtained after a single stage as a function graph the subscript ’1’ indicates the component selective for the membrane
of the BPTIo98 , i.e. normalized standard time obtained when the (e.g. for the PVDF membrane this is methyl acetate), that is the component
feed tank product reaches 98 mole% purity and starting from a enriched in the permeate.
feed concentration of 50 mole% in the component selective for
the membrane. This batch pervaporation-membrane performance
graph is shown in Fig. 9. In this case, when giving priority to the Thus, looking at Ranking 2 it can be noted that Nylon-6 and
separation ability of the membrane together with the contribution Pervap 2255–50 membranes have top positions and should be
of the driving force, the following ranking is obtained: selected for the separation. Conversely, PVDF, PolyAl Typ M1, and
Pervap 2255–40 membranes should be discarded. This outcome is
Nylon-6 > Pervap 2255-50 ≈ Cuprophan > Pervap 2255-60 perfectly consistent with the simulation data previously discussed,
where PVDF, PolyAl Typ M1, and Pervap 2255–40 membranes
> Poly AlTyp M1 > Pervap 2255 − 40 ≈ PVDF were not able to ensure any product specification even after a
(Ranking 2) very high number of batch-stages, or gave a very poor separation
performance and Nylon-6 and Pervap 2255–50 gave the best
Or when giving priority to the BPTI, i.e. the time to reach a desired performance.
separation: Briefly, in this section the inability of the PSI analysis to help
during membrane selection procedure for batch pervaporation
Nylon-6 ≈ PVDF > Poly AlTyp M1 > Pervap 2255-50
applications when medium–low separation factor membranes
> Pervap 2255-60 > Cuprophan > Pervap 2255-40 are compared is shown. In fact apart from the Nylon-6 membrane,
(Ranking 3) a very high performance membrane, PSI analysis favours the Per-
vap 2255–40 and PolyAl Typ M1 membranes which instead come
However, the latter ranking is not preferable since membranes out as the worst choice during batch pervaporation-membrane
that give very low permeate purity during each stage, tend to con- performance graph analysis and MSBP simulations.
sume all the feed to respect the imposed conditions on permeate In the batch pervaporation-membrane performance graph, it is
or component enriched in the retentate purity, making the sepa- possible roughly to trace a line that separates very low perfor-
ration unfeasible. mance pervaporation membranes from the rest (dashed line in

wileyonlinelibrary.com/jctb © 2015 Society of Chemical Industry J Chem Technol Biotechnol (2015)

Overcoming any configuration limitation www.soci.org

Fig. 9) for a normalized permeate concentration of 0.2; thus, mem- Finally, treatment of small fractions of continuous productions
branes that lie below this line should not be used during a real could also be suitable for MSBP systems.
MSBP separation that require high purity products.
Finally, it has also to be stressed that in certain cases, i.e. for
concentrations higher than 50 mole% in the component selective ACKNOWLEDGEMENTS
for the membrane, a more accurate prediction can be obtained The authors wish to thank Dr. Richard W. Baker from Membrane
when estimating the batch pervaporation-membrane performance Technology and Research Inc., for his help and encouragement
graph starting from these actual feed concentrations, as shown in during the writing of this work.
the case of the MM20 stream concentration.

REFERENCES
CONCLUSIONS ̇
1 Bozek-Winkler E and Gmehling J, Transesterification of methyl acetate
In conclusion, when pursuing an MSBP separation it is possible and n-butanol catalyzed by Amberlyst 15. Ind Eng Chem Res
45:6648–6654 (2006).
to overcome the limitation due to the use of a condenser after 2 McCabe WL, Smith JC and Harriott P, Unit Operations of Chemical
each stage, necessary when enriching permeate streams by con- Engineering. McGraw-Hill Book Company, New York (2005).
ventional pervaporation. In MSBP, the product purity is increased 3 Vandi L-J, Hou M, Veidt M, Truss RM and Paton R, Interface diffusion and
by recycling the condensed permeate to the feed tank. This means morphology of aerospace grade epoxy co-cured with thermoplastic
that just one condenser is needed for the unit. Moreover, theoret- polymers, in 28th International Congress of the Aeronautical Science.
International Council of the Aeronautical Sciences (ICAS) (2012).
ically this unit operation is also able to overcome any membrane 4 Uragami T, Sumida I, Miyata T, Shiraiwa T, Tamura H and Yajima T,
limitation since the separation performance of the membrane is Pervaporation characteristics in removal of benzene from water
bypassed by increasing the number of MSBP stages, as any mem- through polystyrene-poly (dimethylsiloxane) IPN membranes. Mater
brane is able to give permeates and retentates with high purities. Sci Applic 2:169 (2011).
5 Anjali Devi D, Smitha B, Sridhar S and Aminabhavi TM, Pervaporation
However, from a practical point of view it was demonstrated that
separation of isopropanol/water mixtures through crosslinked chi-
membranes with very low pervaporation ability tend to consume tosan membranes. J Membr Sci 262:91–99 (2005).
all the feed in order to reach the required product purity. 6 El-Zaher NA and Osiris WG, Thermal and structural properties of
Medium performance membranes gave permeate and retentate poly(vinyl alcohol) doped with hydroxypropyl cellulose. J Appl Poly
products with high purity when terminating each stage at a con- Sci 96:1914–1923 (2005).
7 Lipnizki F, Field RW and Ten P-K, Pervaporation-based hybrid process:
stant value of retentate product purity. On the other hand, in this a review of process design, applications and economics. J Membr Sci
case the number of stages can be very high. This number can be 153:183–210 (1999).
decreased by one order of magnitude when working at constant 8 Smitha B, Suhanya D, Sridhar S and Ramakrishna M, Separation of
stage-recovery of the component enriched in the permeate. How- organic–organic mixtures by pervaporation – a review. J Membr Sci
241:1–21 (2004).
ever in this case, for stage-recoveries of the component enriched
9 Khajavi S, Jansen JC and Kapteijn F, Application of a sodalite membrane
in the permeate that are not very high, the purity of the retentate reactor in esterification – coupling reaction and separation. Catal
product is limited by the membrane separation ability. Today 156:132–139 (2010).
It is important to note that these two operational modes can be 10 Jafar JJ and Budd PM, Separation of alcohol/water mixtures by perva-
further elaborated to fit any specific product requirement. Further- poration through zeolite A membranes. Micropor Mater 12:305–311
(1997).
more, the storage tank can be connected to a secondary system of 11 Morigami Y, Kondo M, Abe J, Kita H and Okamoto K, The first large-scale
separation that helps reduce the concentration of the component pervaporation plant using tubular-type module with zeolite NaA
selective for the membrane, even during MSBP operation. In the membrane. Sep Purif Technol 25:251–260 (2001).
MSBP proposed in this work a static storage vessel was used. 12 Bolto B, Hoang M and Xie Z, A review of membrane selection for
In this paper it is also shown how, apart from consideration of the dehydration of aqueous ethanol by pervaporation. Chem Eng
Process: Process Intens 50:227–235 (2011).
very high performance membranes, PSI analysis gave an incor- 13 Chapman PD, Oliveira T, Livingston AG and Li K, Membranes for the
rect prediction of the membrane performance during the MSBP dehydration of solvents by pervaporation. J Membr Sci 318:5–37
process with two of the three best membrane choices actually (2008).
performing the worst in the simulations. Hence a new and more 14 Mandal S and Pangarkar VG, Separation of methanol-benzene and
methanol-toluene mixtures by pervaporation: effects of thermo-
accurate performance analysis was proposed, based on the batch dynamics and structural phenomenon. J Membr Sci 201:175–190
pervaporation-membrane performance graph. In this case the per- (2002).
meate product purity is shown as a function of a normalized time, 15 Baker RW, Kaschemekat J, Wijmans JG, Inventors, Membrane Technol-
the initial feed amount over membrane area, needed to obtain a ogy and Research, inc., assignee. Pervaporation apparatus and pro-
retentate product with certain specification by a single pervapo- cess. WO patent 1993012865 A1 (1993).
16 Rautenbach R and Albrecht R, Membrane Processes. John Wiley & Sons,
ration stage. In this way a complete description of the membrane New York (1989).
ability during batch-pervaporation is given. 17 Baker RW, Membrane Technology and Applications. John Wiley & Sons
What has been shown in this work may lead to some interesting (2012).
conclusions about the applicability of MSBP units. An MSBP could 18 Jiménez L and Costa-López J, The production of butyl acetate and
methanol via reactive and extractive distillation. II. Process mod-
be the natural substitute of all stand-alone batch distillation pro-
eling, dynamic simulation, and control strategy. Ind Eng Chem Res
cesses, in particular for azeotropic mixtures. Moreover, the removal 41:6735–6744 (2002).
of bottlenecks in batch pervaporation separations could also be 19 Fuchigami Y, Hydrolysis of methyl acetate in distillation column packed
another application. with reactive packing of ion exchange resin. J Chem Eng Japan
Production of pharmaceuticals and in general all small batch 23:354–359 (1990).
20 Jiménez L, Garvín A and Costa-López J, The production of butyl acetate
productions could benefit from the employment of an MSBP unit, and methanol via reactive and extractive distillation. I. Chemical
even with the use of medium-low performance pervaporation equilibrium, kinetics, and mass-transfer issues. Ind Eng Chem Res
membranes. 41:6663–6669 (2002).

J Chem Technol Biotechnol (2015) © 2015 Society of Chemical Industry wileyonlinelibrary.com/jctb

 www.soci.org G Genduso, P Luis, B Van der Bruggen

21 Steinigeweg S and Gmehling J, Transesterification processes by combi- 28 Henley EJ, Seader JD and Roper DK, Separation Process Principles. Wiley
nation of reactive distillation and pervaporation. Chem Eng Process: (2011).
Process Intens 43:447–456 (2004). 29 Feng X and Huang RYM, Separation of isopropanol from water by
22 Luyben WL, Design and control of the butyl acetate process†. Ind Eng pervaporation using silicone-based membranes. J Membr Sci
Chem Res 50:1247–1263 (2010). 74:171–181 (1992).
23 Berg L and Yeh A-I, Separation of methyl acetate from methanol by 30 Wang Y-C, Li C-L, Huang J, Lin C, Lee K-R, Liaw D-J et al., Pervaporation of
extractive distillation. US Patent 4597834 (1986). benzene/cyclohexane mixtures through aromatic polyamide mem-
24 Gorri D, Ibanez R and Ortiz I, Comparative study of the separation of branes. J Membr Sci 185:193–200 (2001).
methanol-methyl acetate mixtures by pervaporation and vapor per- 31 Huang RYM and Yeom CK, Pervaporation separation of aqueous mix-
meation using a commercial membrane. J Membr Sci 280:582–593 tures using crosslinked polyvinyl alcohol membranes. III. Permeation
(2006). of acetic acid-water mixtures. J Membr Sci 58:33–47 (1991).
25 Sain S, Dinçer S and Savaşçy𝚤 ÖT, Pervaporation of methanol–methyl 32 Baker RW, Wijmans JG and Huang Y, Permeability, permeance and
acetate binary mixtures. Chem Eng Process: Process Intens selectivity: a preferred way of reporting pervaporation performance
37:203–206 (1998). data. J Membr Sci 348:346–352 (2010).
26 Abdallah H, El-Gendi A, El-Zanati E and Matsuura T, Pervaporation 33 Sampranpiboon P, Jiraratananon R, Uttapap D, Feng X and Huang RYM,
of methanol from methylacetate mixture using polyamide-6 mem- Pervaporation separation of ethyl butyrate and isopropanol with
brane. Desal Water Treatment 51:7807–7814 (2013). polyether block amide (PEBA) membranes. J Membr Sci 173:53–59
27 Genduso G, Amelio A, Colombini E, Luis P, Degrève J and Van der (2000).
Bruggen B, Retrofitting of extractive distillation columns with high 34 Kopeć R, Meller M, Kujawski W and Kujawa J, Polyamide-6 based
flux, low separation factor membranes: a way to reduce the energy pervaporation membranes for organic–organic separation. Sep Purif
demand? To be submitted for publication. Technol 110:63–73 (2013).

wileyonlinelibrary.com/jctb © 2015 Society of Chemical Industry J Chem Technol Biotechnol (2015)