Separation Science and Technology

ISSN: 0149-6395 (Print) 1520-5754 (Online) Journal homepage: https://www.tandfonline.com/loi/lsst20

DESIGN METHODOLOGY FOR THE OPTIMIZATION

OF MEMBRANE SEPARATION PROPERTIES FOR
HYBRID VAPOR PERMEATION-DISTILLATION
PROCESSES

Alaa Fahmy , Dieter Mewes & Katrin Ebert

To cite this article: Alaa Fahmy , Dieter Mewes & Katrin Ebert (2001) DESIGN METHODOLOGY
FOR THE OPTIMIZATION OF MEMBRANE SEPARATION PROPERTIES FOR HYBRID VAPOR
PERMEATION-DISTILLATION PROCESSES, Separation Science and Technology, 36:15,
3287-3304

To link to this article: https://doi.org/10.1081/SS-100107903

Published online: 15 Feb 2007.

Submit your article to this journal

Article views: 111

Citing articles: 10 View citing articles

Full Terms & Conditions of access and use can be found at

https://www.tandfonline.com/action/journalInformation?journalCode=lsst20
SEPARATION SCIENCE AND TECHNOLOGY, 36(15), 3287–3304 (2001)

DESIGN METHODOLOGY FOR THE

OPTIMIZATION OF MEMBRANE
SEPARATION PROPERTIES FOR HYBRID
VAPOR PERMEATION-DISTILLATION
PROCESSES

Alaa Fahmy,1 Dieter Mewes,1,* and Katrin Ebert2

1
Institut für Verfahrenstechnik, Universität Hannover,
Callinstr. 36, D-30167 Hannover, Germany
2
GKSS Forschungszentrum, Institut für Chemie,
Max-Planck-Str., D-21502 Geesthacht, Germany

ABSTRACT

Hybrid processes of distillation and vapor permeation are consid-

ered promising alternatives to azeotropic and extractive distilla-
tion. The research is directed toward developing membranes with
improved separation properties and longer life times.We show that
membranes with very high selectivities are not desirable when a
very high purity of the retentate is required. A design methodology
was developed to calculate the optimum separation properties for
the membranes within hybrid processes. The results of the simula-
tion encourage the use of low-selectivity membranes, especially if
the membrane step is coupled to a distillation step. The method is
applied to an ethanol dehydration process. However, it can be ap-
plied to other membrane separation processes as well.

*Corresponding author. Fax: 49/511/7623031; E-mail: dms@c36.uni-hannover.de

3287

Copyright © 2001 by Marcel Dekker, Inc. www.dekker.com

 ORDER REPRINTS

3288 FAHMY, MEWES, AND EBERT

Key Words: Vapor permeation; Hybrid; Distillation; Process de-

sign; Optimization

INTRODUCTION

Hybrid processes of distillation and membrane separation processes have

gained much attention in recent years. The results of this coupling benefit from the
specific advantages of each separation step: the low cost of distillation and the in-
dependence of the membrane process on the boiling curve of the phase diagram.
Typical implementation areas of hybrid processes are closely boiling and
azeotropic separations. These hybrid processes can offer cost savings when com-
pared to conventional azeotropic, extraction, or 2-pressure distillation processes.
Pressly and Ng (1) presented a review of the economical aspects of hybrid
processes. They suggested a break-even analysis for investigating the feasibility
of various types of distillation-membrane hybrids. Lipnizki, Field, and Ten (2)
presented a review of process design and the economics of pervaporation-based
hybrid processes. Stephan, Noble, and Koval (3), Pettersen et al. (4), and Moganti,
Noble, and Koval (5) analyzed olefin purification by a distillation-membrane hy-
brid process. They studied different process configurations for determining the
optimal operating conditions for each configuration. Their optimization analysis
was restricted to the calculation of the membrane area and the number of theoret-
ical column trays. Pettersen and Lien (6) presented parametric studies illustrating
some of the trade-offs in a hybrid distillation-membrane process for ethanol de-
hydration. Their calculations were based on an algebraic model for describing
mass transport through the membrane. Rautenbach et al. (7) optimized a hybrid
process for separating methanol from dimethylcarbonate by calculating the pro-
cessing cost. The feed and retentate stream composition were the variables ma-
nipulated for this membrane study.
The membrane process may be a pervaporation or a vapor permeation unit.
Both processes are closely related, differing only in the state of the feed. For per-
vaporation, the feed is liquid and the permeate evaporates through the membrane.
The heat supply to the feed stream is usually accomplished by intermediate heat
exchangers that are installed between a number of pervaporation modules in se-
ries. For vapor permeation, the feed is vapor, which does not change phase while
crossing the membrane surface making the process less complex. For both pro-
cesses the driving force for material transport through the dense membrane is the
difference in the chemical potential of the permeating components between the
feed and permeate side. This driving force is usually realized through lowered
pressure on the permeate side. Vapor permeation is especially suitable for the pu-
rification of top streams of fractionation columns that can be used as feed directly.
Although vapor permeation is less sensitive to concentration polarization at the
feed side of the membrane, the mass transport is sensitive to the degree of super-
 ORDER REPRINTS

MEMBRANE SEPARATION PROPERTIES 3289

heating and to friction losses in the feed side. On one hand, little superheating low-
ers the transmembrane flux in polymeric membranes as a result of decreased sorp-
tion and swelling. On the other hand, heat losses due to friction can result in the
formation of an undesired condensate film that partially covers the membrane on
the feed side. Brüschke and Schneider (8) invented a process that combines both
vapor permeation and pervaporation in which the feed is introduced as a liquid-
vapor mixture that flows up a vertically mounted membrane module. This process
results in a complete mixing of the 2 phases, and a stationary liquid film does not
form on the membrane surface. The mass transport could be improved by insur-
ing saturation conditions throughout the feed side of the membrane. Detailed in-
vestigations on vapor permeation and its comparison to pervaporation are reported
in the literature (9–17).
Various types of membranes have been investigated for pervaporation and
vapor permeation (8–20). Poly(vinyl alcohol) (PVA) membranes are considered
standard dehydration membranes because of their high selectivities and their rel-
atively high fluxes in many aqueous organic solutions. Research is directed to-
ward producing membranes of higher fluxes with reasonable selectivities. Our
aim in the present work was to illustrate a design method for the determination of
the optimum selectivity of the membranes required for a certain separation pro-
cess. This goal was achieved through the minimization of the annual cost of the
complete membrane process as a function of the membrane separation properties.
The method was applied to the dehydration of ethanol in which 99.9% product pu-
rity was required.

Simulation Models

The mass transfer through the membrane is based on sorption-diffusion-

desorption steps. Rigorous modeling requires the simulation of these steps. The
prediction of the mixture solubilities using the UNIQUAC model, which gives
very good results in the case of polar liquid mixtures and hydrophilic membranes
(21), can be made. However, no promising models for the prediction of the mix-
ture diffusivities have been found. Great effort has been directed toward model-
ing the mass transport through molecular modeling (22–24). Reasonable results of
a detailed molecular-dynamics simulation have been illustrated, but they are un-
fortunately limited to the feed-polymer interface. More extensive results are still
expected from this simulation methodology.
We present a general model for material balance and membrane area calcu-
lations. It is valid for binary mixtures. The main assumptions in this model are
1. negligible pressure drop along either side of the membrane surface,
2. plug flow along the feed side of the membrane, and
3. cross flow along the permeate side of the membrane.
 ORDER REPRINTS

3290 FAHMY, MEWES, AND EBERT

The vapor permeation module with the main design variables is shown
schematically in Fig. 1. All the mass fractions refer to the faster permeating com-
ponent. A mass balance based on a differential element of the membrane results
in the following equation:
dmF dx F
    (1)
mF x P–x F
where m and x are the local mass flow rate and the local mass fraction respectively.
The superscripts refer to the location at the feed (F) or permeate (P) side. The in-
tegration of Eq. (1) along the concentration interval from feed to retentate gives
the ratio of permeate flow rate mP to feed flow rate mF.

 
mp xR
dx F
    1  exp P  (2)
mF xF x  xF
The ratio  is called the “module cut rate” and introduces additional mass balance
information to the overall and component mass balance equations. The subscripts
F, P, and R refer to the overall feed, permeate, and retentate streams.
The change in flow rate along the feed side with respect to the membrane
area is expressed by the total permeation flux JP.
dmF
  JP (3)
dA
By rearranging and integrating Eq. (3), the membrane area A can be calculated.
The total permeation flux JP must be expressed as a function of the local mass flow
at the feed side mF.

 mR
dmF
A  (4)
mF JP(mF )
An approximate calculation (25) of the local transmembrane flux for each perme-
ating component i based on the solution-diffusion model may be carried out by the

Figure 1. Schematic diagram of a vapor permeation module showing the material bal-
ance variables and flow assumptions.
 ORDER REPRINTS

MEMBRANE SEPARATION PROPERTIES 3291

equation
SiDi
i
JPi   (piF  piP)  L i(piF  piP) (5)
ƒio
where Si is the solubility coefficient of the component in the membrane polymer,
and Di is the diffusion coefficient of the component inside the membrane of the
thickness . 
i is the average fugacity coefficient of the component at the separa-
tion conditions and can be taken as unity for low pressures. ƒio is the standard or
reference fugacity. Li is the individual permeability coefficient of component i. It
can be determined experimentally for each component of the mixture. The driv-
ing force is expressed in terms of the partial pressure difference of component i
between the feed and permeate sides.
The selectivity of the membrane can be expressed either by the selectivity
coefficient
La
L   (6)
Lb
or by the separation factor
x P/(1  x P)
x   (7)
x F/(1  x F)
Semi-empirical models have been suggested to relate the permeability co-
efficient to the process parameters (11,12). For a 2-component system, nine semi-
empirical parameters must be determined experimentally.
However, at the early design stage the necessary information (experimental
data) may not be available for the parameters to be determined with significant ac-
curacy. Petersen and Lien (6) presented an algebraic model that can be used to
solve the material balance around the membrane and calculate the required surface
area for a specified separation through an averaged separation factor and an aver-
aged flux of the membrane for the investigated concentration interval. This model
was first published in the work of Naylor and Backer (26), who derived a calcu-
lation approach for the gaseous-diffusion-stage similar to the methods McCabe
and Thiele used for distillation. The calculation procedure can be summarized in
the following algebraic equations:
1  xR
    
mP 1 xR
    1  exp  ln   x
ln   (8)
mF
 x  1 x F 1  xF
mF 
A   (9)
JP
where JP and x are the logarithmic mean values of the total flux and the separa-
tion factor of the membrane at the feed and retentate compositions.
In addition to the earlier stated general assumptions, this model is based on
the assumption that a linear relationship exists between JP and x F and between ln
 ORDER REPRINTS

3292 FAHMY, MEWES, AND EBERT

x and x F. This assumption is valid only if the fugacity or the partial pressure of
the permeate side is of a negligible value when compared to that of the feed. The
flux can then be considered proportional to the concentration or partial pressure of
the faster permeating component in the feed side. This is a reasonable assumption
for the dehydration processes if the required purity of the retentate is not very
high. A considerable amount of water will still be on the feed side, which keeps
the driving force high; that is, piF  piP. An application with a relatively high
driving force along the whole membrane is overcoming the azeotrope through the
dehydration of isopropanol. The required separation is 85–89% isopropanol. The
dehydration of ethanol, is an example in which a high product purity is desired;
the subazeotropic ethanol must be concentrated from 95.5% up to 99.5% or
higher. For this type and for similar dehydration problems, the local flux is not a
linear function of the local feed-side composition.
When these nonlinearities appear, the algebraic model will not provide suf-
ficient accuracy, and the design calculations must be carried out with Eqs. (1–5).

Illustrative Example

The following example demonstrates the importance of selecting a proper

membrane selectivity for a separation task. The calculations are carried out with
the differential model through the implementation of Eqs. (1–5). In the selected
process, 1000 kg/h of an aqueous ethanol must be dehydrated from 93% (wt) up
to 99.9%. For the first calculations, a PVA membrane is selected. Its water per-
meability is comparable to that of a Sulzer ChemTech (formerly GFT) standard
dehydration membrane (10,12,13) and to the modified PVA membrane of GKSS
(18). A selectivity coefficient of 2000 is assumed. Constant permeation coeffi-
cients for water and ethanol were assumed to be in line with the investigated con-
centration interval. This assumption could be accepted for narrow concentration
intervals, especially in this early design phase, where no extensive experimental
data are available for the membranes. More exact assumptions could be consid-
ered in a later phase of design and optimization in a second iteration after the sug-
gested membranes have been developed and tested. The feed pressure was 1 bar
and the permeate pressure was 20 mbar. Figure 2 represents the driving force for
water transport and the required membrane area as functions of the retentate com-
position. The driving force is represented as the partial pressure difference be-
tween feed and permeate. A great increase of the required membrane area was ob-
served after 99% concentration, where the driving force nearly vanished. To
increase the driving force for fixed feed conditions, either the permeate pressure
pp or the permeate water concentration xp should be lowered. Lowering the per-
meate water concentration can be realized through the use of a less selective mem-
brane.
 ORDER REPRINTS

MEMBRANE SEPARATION PROPERTIES 3293

Figure 2. Difference of the water partial pressure across the membrane and the required
membrane area as functions of the retentate composition.

Figure 3 shows how far the driving force could be increased in the last con-
centration region by the use of low selectivity membranes. Low selectivity is re-
alized during the simulation by increasing the ethanol permeation coefficient. The
water permeation coefficient is assumed to equal that of the high selectivity mem-
branes. This assumption about permeation is very conservative because low se-
lectivity membranes would offer a higher permeability for water. The pressure on

Figure 3. Driving force for water as a function of the retentate composition for different
selectivity values.
 ORDER REPRINTS

3294 FAHMY, MEWES, AND EBERT

Figure 4. The required membrane area as a function of the retentate composition for dif-
ferent selectivity values.

the permeate side is also held constant to limit the investigation to the effect of the
selectivity.
Figure 4 shows the decrease of the required membrane area achieved
through the use of 2 low selectivity membranes. In these cases, the ethanol con-
centration in the permeate stream was higher than when high selectivity mem-
branes were used. However, if the permeate stream were recycled to a distillation
column, no ethanol would be lost. In this hybrid combination, care should be taken
to ensure that ethanol is recycled back into the proper position in the column.
The increased mass flow of the permeate stream due to the presence of high
levels of ethanol is shown in Fig. 5. This increase in flow rate would elevate the
cost of the condensation and vacuum system of the permeate. Nevertheless, one
can expect cost savings by using the membrane of L  100 because the enormous
savings in the membrane area would not be abrogated by a 20% increase in the
permeate flow.
The results of mass balance calculations for the hybrid process are shown in
Fig. 6. The calculations are based on a column feed of 7% (wt) ethanol, which is
a typical composition of the product beer of a fermentation unit (27).
These results show that the mass flow of the recycled permeate stream is al-
ways lower than 1% of the column feed, so its recycle will not result in hydrody-
namic problems in the column.

Optimization Calculations

To find the optimum selectivity of the membranes for the above defined
separation problem, the whole membrane system must be predesigned, the major
 ORDER REPRINTS

MEMBRANE SEPARATION PROPERTIES 3295

Figure 5. Permeate mass flow rates as a function of the retentate composition for differ-
ent selectivity values.

equipment must be sized, and the annual cost should be minimized. For this pur-
pose a simulation program was developed with the program ASPEN Plus. The
main flow sheet and the main calculation loops are shown in Fig. 7. A developed
user model allows one to calculate the material balance around the membrane, the
required membrane area, and the membrane and module costs. The material bal-
ance is determined through the use of Eqs. (1–5) of the differential model previ-
ously described. A leakage air stream is inevitable in any vacuum system; how-
ever, the amount of leakage is important for subsequent calculations of the

Figure 6. Mass balance for the hybrid process of ethanol dehydration. A large selectiv-
ity range is covered. All the mass fractions given are for ethanol.
 ORDER REPRINTS

3296 FAHMY, MEWES, AND EBERT

Figure 7. Main loops of the optimization program.

condenser and the vacuum pump. An air stream of assumed volume is mixed with
the permeate stream and sent to the condenser. To decrease the load to the vacuum
pump, the permeate mixture is condensed and subcooled. Only a very small equi-
librium amount will remain with the leakage air in the gaseous phase. The con-
densation temperature is varied within an iteration loop. The results of creating
this simulation loop were the determination of heat transfer area and the condenser
volume, the design specifications of the refrigeration unit and the vacuum pump,
and the total annual cost of this condensation and vacuum system. The result of
using the inner loop was the determination of optimum subcooling temperature,
which gives the minimum total cost of this part of the system. From the calculated
condenser volume, the flow of the leakage air stream was corrected and the outer
loop of the leakage air was run until it converged to a constant air stream. The
global variables that could be manipulated through the whole program were mem-
brane selectivity (i.e., the ethanol permeation coefficient) and the permeate pres-
sure. The water permeation coefficient was held equal to that of the most selective
membrane.
A typical structure of the total annual cost calculation is presented in Fig. 8.
The membrane replacement costs held the highest share of the total annual cost be-
cause of the increased specific area of the membrane required as the required pu-
rity is exceedingly high. Throughout these cost calculations, the membrane mate-
rial is depreciated within two years, and all the other parts of the system within five
years. The cost of the membrane material was assumed to be 150 DM/m2, while
the fixed cost of the membrane modules was taken as 500 DM/m2 of the membrane
area (28). Other cost functions developed for the fixed and operating costs, as noted
from several manufacturer offers, were regressed. Electric power costs are the main
operating costs of the refrigeration system and of the vacuum pump. The total cost
 ORDER REPRINTS

MEMBRANE SEPARATION PROPERTIES 3297

Figure 8. Typical annual cost structure of the membrane unit: mF  1000 kg/h; xF 
0.93; xR  0.999; pF  1 bar; pp  8 mbar; L  40.

as the sum of the mentioned costs in Fig. 8 is illustrated in Fig. 9 as the function of
the manipulated variables, the selectivity, and permeate pressures. Some of the
curves are broken. If they had not been broken the data would show that the tem-
perature of the heat transfer surface in the condenser was lower than the freezing
point of the permeate mixture. This could result in the formation of ice crystals on
the heat transfer surface, which could block the operation. The curves show opti-
mum selectivities between 20 and 40 and optimum operating pressures between 4
and 6 mbar. These low selectivities may be a surprising result; however this result
is specific for this separation problem in which a very high purity was required.

Figure 9. The membrane unit total cost as function of the membrane selectivity and the
permeate pressure.
 ORDER REPRINTS

3298 FAHMY, MEWES, AND EBERT

Table 1. Optimization Results for Three Different Process Schemes

The same methodology was used to compare three different process

schemes, which are shown in Table 1. The first is a conventional scheme with one
type of membrane equipped with a condensation and vacuum system. The second
utilizes two different types of membranes in series integrated in the same system.
In the third scheme the two membranes have different condensation and vacuum
systems. The three schemes were studied for two different separation tasks, one is
to increase dehydration from 93.0 to 99.9%, and the other is to increase it from
83.0 to 99.9%. The composition between the two membranes is assumed to be
99% (wt) ethanol for the last two cases. The results are shown in Table 1. A heuris-
tic rule was derived: A higher selective membrane is needed at high concentra-
tions of the faster permeating component, and a lower selectivity is favored for the
regions of lower concentration. Based on the assumption of constant water per-
meability, the use of two different membranes in series will bring savings of up to
6% of the total annual cost for the first separation and up to 14% in the second sep-
aration in which the concentration interval is larger.

DISCUSSION

The general assumptions of negligible pressure drop, constant temperature,

plug flow along the feed side, and cross flow along the permeate side are discussed
in detail in the paper of Pettersen and Lien (6) and of Naylor and Backer (26). The
assumption that the less selective membranes have the same water permeation co-
efficient as the most selective membranes is a very conservative assumption con-
 ORDER REPRINTS

MEMBRANE SEPARATION PROPERTIES 3299

cerning the calculation of the membrane area and the membrane cost. Usually the
less selective membranes allow higher fluxes (9). Therefore, the cost savings of the
less selective membranes demonstrated in the separation example were considered
to be safe calculations. Much more cost savings would be expected if the increase
of the water flux as a result of a decreased selectivity were a part of the calculation.
For separations through which the water concentration in the retentate is rel-
atively high, such as when the azeotrope of the isopropanol water system is bro-
ken, the low selectivity membranes would not appear to provide any cost savings
if the calculations were carried out on the basis of the previously mentioned as-
sumption of constant flux. However, if the flux increases observed when the se-
lectivity is lowered were integrated in the calculations, use of low selectivity
membrane would show cost savings.
The solid curve of Fig. 10 may represent the real behavior of the flux against
selectivity of different membranes with different selectivities for the same sepa-
ration application. During the analysis, the horizontal dotted line was assumed as
the flux was held constant despite lowered selectivity of the membrane. As a first
approximation, one could assume an inverse proportionality between flux and se-
lectivity as shown by the straight dashed line of the same figure. These assump-
tions allow for the optimization calculations to be carried out for membranes of
the same separation index as defined by SI  J
. This calculation is analogous
to the concept of the pervaporation separation index introduced by Huang (29).1
The most exact relation between the flux and the selectivity of the considered

Figure 10. Relationship between flux and selectivity shown schematically for 1 mem-
brane type.

1
Another common separation index is the Rony’s extent of separation (30), which was re-
fined by Sirkar (31) for single entry barrier separation processes. This index has been ex-
tensively used for the characterization of gas separation stages (32,33) and pervaporation
(34).
 ORDER REPRINTS

3300 FAHMY, MEWES, AND EBERT

membranes should be derived experimentally. Therefore, a number of different

membranes with different separation properties must be developed for the same
application. These membranes must be tested experimentally for one to get an idea
about flux variation with respect to selectivity for the membrane. Moreover, rela-
tionships between the permeation coefficient and the process parameters should
be determined for each membrane. This determination requires that a targeted set
of experiments be completed to increase the accuracy of the optimization proce-
dure. Nevertheless, though our optimization may not give the most accurate quan-
titative result, it gives a qualitative conclusion about using membranes of low se-
lectivity in separations that require high retentate purity. However, increased
reliability of rigorous or molecular modeling would save a great deal of experi-
mental work that must be done when optimizing on an empirical basis.
Although the use of low selectivity membranes may offer cost savings, the
presented calculations were completed based on the assumption that the presence
of the membrane step was within a hybrid process. That is, the purity of the per-
meate stream was not assumed to be a constraint throughout the calculations. Per-
meate streams containing considerable amounts of the slow permeating component
would be recycled to a distillation step without disturbing the distillation operation
due to its relatively small quantities. Increased amounts of the organic components
in the permeate stream would decrease the mass flow of the product retentate. How-
ever, the cost calculations are based on a unit of weight of the retentate.
For stand-alone membrane processes (i.e., those not coupled to a distillation
column), a 2-stage membrane process is suggested for membrane area savings. A
scheme of such a process is demonstrated in Fig. 11 for ethanol dehydration. How-

Figure 11. A stand-alone 2-stage membrane process for ethanol dehydration in which
very high product purity is necessary.
 ORDER REPRINTS

MEMBRANE SEPARATION PROPERTIES 3301

ever, it is applicable for any dehydration of organic solvents. The first stage is the
dehydration stage with a low selectivity membrane. Its permeate stream of in-
creased organic content is concentrated by a small membrane unit up to the con-
centration of the first feed. The second unit has a high selectivity membrane to
produce high purity wastewater as permeate. This unit is relatively small because
its retentate purity is lower than that of the first one so that the water content of the
retentate side is sufficient to create a reasonable driving force at the end of the sep-
aration. Moreover, its feed stream is much smaller than the main feed. Simulation
and optimization of this 2-stage process results in an optimum selectivity for the
first membrane between 40 and 50 with an optimum permeate pressure of 8 mbar
and a selectivity of 1000 for the second permeate-purification membrane with an
optimum permeate pressure of 110 mbar. The cost of the second stage comprises
10.5% of the total cost of the process.

CONCLUSIONS

A method for tailoring membrane separation properties for vapor perme-

ation within hybrid processes is presented and applied to some applications of
ethanol dehydration. Membranes of a broad range of different selectivities were
required, each for a specific separation range. Simulating and optimizing the
whole process in an early phase offered a great deal of helpful information for de-
termining the appropriate membrane properties required.
The results of the optimization show that the use of low selectivity mem-
branes should be encouraged when the goal of the separation is to obtain a very
pure retentate.
Running the simulations on an empirical basis requires iterative interactions
with membrane development to increase the accuracy of the optimization and pro-
duce more tailored membranes for each type of application.

NOMENCLATURE

A membrane area (m2)

D diffusion coefficient (m2/s)
Jp total permeation flux across the membrane (kg/m2h)
JP average total permeation flux (kg/m2h)
L permeability or permeation coefficient (kg/m2hbar)
m mass flow rate (kg/h)
pi partial pressure of component i
S sorption coefficient (kg/m3)
x mass fraction of the most permeable component (–)
 ORDER REPRINTS

3302 FAHMY, MEWES, AND EBERT

xF local mass fraction of the most permeable component at the feed side (–)
xP local mass fraction of the most permeable component at the permeate side
(–)
L membrane selectivity (–)
x separation factor (–)

x average separation factor (–)
 membrane thickness (m)
ƒo standard (reference) fugacity (bar)
 module cut rate (–)

Subscripts

F feed stream
P permeate stream
R retentate stream
a most permeable component
b less permeable component
i component i

Superscripts

F locally on the feed side

P locally on the permeate side

ACKNOWLEDGMENTS

Alaa Fahmy expresses his gratitude for the support provided by the Institut
für Chemie of the GKSS Forschungszentrum. Dipl.-Ing. Olaf Stange is especially
acknowledged for providing experimental data.

REFERENCES

1. Pressly, T.M.; Ng, K.M. A Break-even Analysis of Distillation-Membrane

Hybrids. AIChE J. 1998, 44, 93–105.
2. Lipnizki, F.; Field, R.W.; Ten, P.K. Pervaporation-Based Hybrid Process: A
Review of Process Design, Applications and Economics. J. Memb. Sci.
1999, 153, 183–210.
3. Stephan, W.; Noble, R.D.; Koval, C.A. Design Methodology for a Mem-
brane/Distillation Column Hybrid Process. J. Memb. Sci. 1995, 99, 259–272.
 ORDER REPRINTS

MEMBRANE SEPARATION PROPERTIES 3303

4. Pettersen, T.; Argo, A.; Noble, R.D.; Koval, C.A. Design of Combined
Membrane and Distillation Processes. Separ. Technol. 1996, 6, 175–187.
5. Moganti, S.; Noble, R.D.; Koval, C.A. Analysis of a Membrane/Distillation
Column Hybrid Process. J. Memb. Sci. 1994, 93, 31–44.
6. Pettersen, T.; Lien, K. Design of Hybrid Distillation and Vapor Permeation
Processes. J. Memb. Sci. 1995, 102, 21–30.
7. Rautenbach, R.; Knauf, R.; Struck, A.; Vier, J. Simulation and Design of
Membrane Plants with AspenPlus. Chem. Eng. Tech. 1996, 19, 391–397.
8. Brüschke, H.E.A.; Schneider, W.H. Membranverfahren zur Trennung Flu-
ider Gemische, German Patent, DE 44-10-243-C1, 1995. (in German)
9. Jansen, A.E.; Versteeg, W.F.; van Englenburg, B.; Hanemaaijer, J.H.; ter
Meulen, B.P. Methods to Improve Flux During Alcohol/Water Azeotrope
Separation by Vapor Permeation. J. Memb. Sci. 1992, 68, 229–239.
10. Will, B.; Lichtenthaler, R.N. Comparison of the Separation of Mixtures by
Vapor Permeation and by Pervaporation using PVA Composite Membranes.
Part 1. Binary Alcohol-Water Systems. J. Memb. Sci. 1992, 68, 119–125.
11. Rautenbach, R.; Meyer-Blumenroth, U. Module and Process Design for Va-
por Permeation. Desalination 1990, 77, 295–322.
12. Helmus, F.P. Dampfpermeation: Trennvermögen, Prozessentwicklung und
Einsatzmöglichkeiten, Ph.D. thesis, RWTH Aachen, 1994. (in German)
13. Sulzer Chemtech GmbH. Technical information, 1999.
14. Schehlmann, M.S.; Wiedemann, E.; Lichtenthaler, R.N. Pervaporation and
Vapor Permeation at the Azeotropic Point or in the Vicinity of the LLE Bound-
ary Phases of Organic/Aqueous Mixtures. J. Memb. Sci. 1995, 107, 277–282.
15. Hamada, T.; Taya, M.; Tone, S.; Nakatsuka, S. Pervaporation and Vapor
Permeation Behavior of Water and 2-Propanol in Water-Selective Mem-
branes. J. Chem. Eng. Jpn. 1997, 30, 600–608.
16. Okamoto, K.; Tanihara, N.; Watanabe, H.; Tanaka, K.; Kita, H.; Nakamura,
A.; Kusuki, Y.; Nakagawa, K. Vapor Permeation and Pervaporation Sepa-
ration of Water-Ethanol Mixtures Through Polyimide Membranes. J.
Memb. Sci. 1992, 68, 53–63.
17. Stange, O.; Ebert, K.; Wenzlaff, A.; Wind, I.; Ohlrogge, K.; Mewes, D. The
Influence of the Degree of Saturation of the Feed Vapor on the Permeability
in Hybrid Processes. Proceedings of ICOM 99, Toronto, June 1999.
18. Ebert, K.; Fritsch, D.; Stange, O.; Wenzlaff, A. Hydrophile Kompositmem-
bran zur Entwässerung Organischer Lösungen, German Patent, Amtl. Ak-
tenzeichen 199-25-475.3; (in press).
19. Okuno, H.; Renzo, K.; Uragami, T. Sorption and Permeation of Water and
Ethanol Vapors in Poly(vinylchloride) Membrane. J. Memb. Sci. 1995, 103,
31–38.
20. Scharnagl, N.; Peinemann, K.V.; Wenzlaff, A.; Schwarz, H.H.; Behling,
R.D. Dehydration of Organic Compounds with SYMPLEX Composite
Membranes. J. Memb. Sci. 1996, 113, 1–5.
 ORDER REPRINTS

3304 FAHMY, MEWES, AND EBERT

21. Heinz, A.; Stephan, W. A Generalized Solution-Diffusion Model of the Per-

vaporation Process Through Composite Membranes. Part 1. Prediction of
Mixture Solubilities in the Dense Active Layer Using the UNIQUAC
Model. J. Memb. Sci. 1994, 89, 143–151.
22. Müller-Plathe, F. Diffusion of Water in Swollen Poly(vinyl alcohol) Mem-
branes Studied by Molecular Dynamics Simulation. J. Memb. Sci. 1998,
141, 147–154.
23. Hofmann, D.; Fritz, L.; Paul, D. Molecular Modeling of Pervaporation Sep-
aration of Binary Mixtures with Polymeric Membranes. J. Memb. Sci. 1998,
144, 145–159.
24. Hofmann, D.; Fritz, L.; Ulbrich, J.; Paul, D. Molecular Modeling of Amor-
phous Membrane Polymers. Polymer 1997, 38, 6145–6155.
25. Rautenbach, R. Membranverfahren–Grundlagen der Modul-und Mem-
branauslegung; Springer: Berlin, Heidelberg, Germany, 1997. (in German)
26. Naylor, R.W.; Backer, P.O. Enrichment Calculations in Gaseous Diffusion:
Large Separation Factor. AIChE J. 1955, 1, 95–99.
27. Collura, M.A.; Luyben, W.L. Energy-Saving Distillation Designs in Ethanol
Production. Ind. Eng. Chem. Res. 1988, 27, 1686–1696.
28. Personal information, GKSS Forschungszentrum, Institut für Chemie,
Geesthacht, Germany.
29. Huang, R.Y.M.; Rhim, J.W. Separation Characteristics of Pervaporation
Membrane Separation Processes. In Pervaporation Membrane Separation
Processes; Huang, R.Y.M., Ed.; Elsevier: Amsterdam, 1991; 111–180.
30. Rony, P.R. The Extent of Separation. In AIChE Symposium; New York,
1972; Ser. 120, Vol. 68, 89–104.
31. Sirkar, K.K. On the Composite Nature of the Extent of Separation. Separ.
Sci. 1977, 12, 211–229.
32. McCandless, F.P. Stage Extent of Separation in Ideal Countercurrent Recy-
cle Membrane Cascades. J. Memb. Sci. 1999, 154, 15–23.
33. Kao, Y.K.; Qiu, M.M.; Hwang, S.T. Critical Evaluation of Two Membrane
Gas Permeator Designs: Continuous Membrane Column and Two Strippers
in Series. Ind. Eng. Chem. Res. 1989, 28, 1514–1520.
34. Baudot, A.; Marin, M. Improved Recovery of an Ester Compound by Per-
vaporation Coupled with a Flash Condensation. Ind. Eng. Chem. Res. 1999,
38, 4458–4469.

Received July 2000

Revised January 2001
 Request Permission or Order Reprints Instantly!

Interested in copying and sharing this article? In most cases, U.S. Copyright
Law requires that you get permission from the article’s rightsholder before
using copyrighted content.

All information and materials found in this article, including but not limited
to text, trademarks, patents, logos, graphics and images (the "Materials"), are
the copyrighted works and other forms of intellectual property of Marcel
Dekker, Inc., or its licensors. All rights not expressly granted are reserved.

Get permission to lawfully reproduce and distribute the Materials or order

reprints quickly and painlessly. Simply click on the "Request
Permission/Reprints Here" link below and follow the instructions. Visit the
U.S. Copyright Office for information on Fair Use limitations of U.S.
copyright law. Please refer to The Association of American Publishers’
(AAP) website for guidelines on Fair Use in the Classroom.

The Materials are for your personal use only and cannot be reformatted,
reposted, resold or distributed by electronic means or otherwise without
permission from Marcel Dekker, Inc. Marcel Dekker, Inc. grants you the
limited right to display the Materials only on your personal computer or
personal wireless device, and to copy and download single copies of such
Materials provided that any copyright, trademark or other notice appearing
on such Materials is also retained by, displayed, copied or downloaded as
part of the Materials and is not removed or obscured, and provided you do
not edit, modify, alter or enhance the Materials. Please refer to our Website
User Agreement for more details.

Order now!

Reprints of this article can also be ordered at

http://www.dekker.com/servlet/product/DOI/101081SS100107903