Membrane All Type
Membrane All Type
Gas–Gas Separation
by Membranes
Thomas Harlacher and Matthias Wessling
Aachener Verfahrenstechnik, Chemical Process Engineering, RWTH Aachen University, Aachen, Germany
Contents
Nomenclature 558
Indices 558
1. Introduction 559
2. Membrane Modules for Gas Separation 560
2.1 Module Construction 560
2.1.1 Hollow Fibre Modules 560
2.1.2 Spiral Wound Modules 561
2.1.3 Plate and Frame Modules 562
2.2 Membrane Material 563
2.3 Mass Transfer through Polymer Membranes 564
2.3.1 Influence of Gas Phase Pressure 566
2.3.2 Influence of Gas Phase Temperature 566
2.3.3 Concentration Polarization Effects 567
2.4 Concepts of Flow Control 567
3. Process Design 568
3.1 Pressure Ratio and Stage Cut 568
3.2 Driving Force Generation 569
3.3 Feed Pretreatment 570
3.4 Module Interconnection 571
3.5 Economic Parameters for Process Design 573
4. Applications of Gas Permeation Processes 574
4.1 Oxygen/Nitrogen Separation 574
4.1.1 Nitrogen Enrichment 574
4.1.2 Oxygen Enrichment 578
4.2 Biogas Treatment 580
References 584
GLOSSARY
Selectivity Characteristics of a membrane to differentiate among various components.
Stage cut Parameter which defines the fraction of the feed stream which permeates through
the membrane.
Pressure ratio Dimensionless parameter which describes the ratio of the total feed pressure
to the total permeate pressure.
Retentate The part of the feed stream, which is rejected by the membrane, and leaves the
module without passing the membrane on the high pressure side.
Permeate The part of the feed stream, which passes the membrane and leaves the module
on the low pressure side.
Recovery The ratio of the amount of a product, enriched in a useful product stream, to the
total amount of this component in the feed stream.
Permeability Parameter for the trans-membrane flux per unit driving force per unit mem-
brane thickness.
Permeance The trans-membrane flux per unit driving force.
NOMENCLATURE
D Diffusivity m2/s
Ea Activation energy J/kmol
n_ 00 Specific mole flow kmol/m2/s
P Permeability Barrer
S Solubility kmol/m3/bar
T Temperature K
x Mole fraction e
y Mole fraction e
d Membrane thickness mm
DHL Heat of solution J/kmol
P Permeance GPU
F Pressure ratio e
Q Stage cut e
INDICES
i, j, k Component i, j, k
0 Standard conditions
F Feed
P Permeate
Gas–Gas Separation by Membranes 559
1. INTRODUCTION
In recent decades, next to conventional technologies, such as physical
adsorption, chemical or physical absorption and cryogenic fractioning, and
membrane processes have become an industrial alternative for gas separation.
For separation tasks with moderate flow rates and moderate purities, gas
permeation is the state-of-the-art technology in many cases. Compared to
conventional techniques, gas permeation processes are characterized by a
simple, flexible, and compact design. Table 1 shows a detailed collection
of advantages and drawbacks of gas permeation processes.
This chapter gives a condensed overview of the basics of gas permeation
processes. Both the design and operation of membrane modules as well as
approaches for process design are described. Examples of industrial mem-
brane gas separation processes illustrate the application. Detailed information
about gas separation processes can be found in Baker (2004), Drioli (2010,
2011), Favre (2011), Melin 2007), Merkel 2010), Mulder (1996), Nunes
(2006), Ohlrogge (2006), and Yampolskii (2010). Yampolskii reports partic-
ularly on membrane materials for gas permeation Yampolskii (2006).
Table 2 Key characteristics of the three major types of modules used for industrial
applications of gas separation processes (Drioli 2010)
Hollow fibre Spiral wound Envelope
2 3
Packing density (m /m ) 500e10,000 200e1000 30e500
Approximate are per module (m2) 300e600 20e40 5e20
Pretreatment requirement High Fair Minimal
Resistance to fouling Poor Fair Good
Pressure drop High Fair Low
Flow distribution Good Moderate Fair
Manufacturing costs Low Medium High
Table 3 Membrane suppliers and their membrane module characteristics for gas
permeation processes (Scholz 2013)
Supplier Module type Polymer
The capillaries are bedded in resin for sealing against the inlet and outlet
sides (Figure 1).
The most important advantage of hollow fibre modules is their high pack-
ing density which is up to 10,000 m2/m. However, the modules are prone to
impurities and pressure losses, due to the small fibre diameter. Feed pressures
up to 15 bar lumen-sided and 70 bar shell-sided can be applied.
Feed Retentate
Permeate
spacer in-between. The envelope is sealed on three sides and the fourth,
open side is connected to the collecting pipe. The module is placed inside
a tubular pressure vessel. The feed gas enters the module at the end wall
and passes axially through the module. Meanwhile, the permeate streams
spirally through the spacer to the collecting pipe (Figure 2).
The permeate flow path to the collecting pipe might be very long which
causes significant pressure losses. These pressure losses can be reduced by
using multiple short envelopes.
Top view on a
membrane envelope
Permeate Feed
Retentate
collec ng pipe
Permeate Permeate
Permeate
Feed
Retentate
Permeate
envelopes are pushed onto a collecting pipe. For flow conduction, deflec-
tion sheaves are added, whereby the feed flow velocity can be kept in a
defined range by the selection of various envelope numbers per compart-
ment (Figure 3). The modular placing of the single membrane envelopes en-
ables the exchange of individual envelopes, but in contrast to the hollow
fibre and spiral wound modules, the envelope modules have lower packing
densities and more complex sealing is required.
commercially available. An overview of the used polymers and their pure gas
permeabilities is given in Table 4.
Membranes for gas separation generally feature an asymmetric structure.
The active layer, responsible for the separation of gas mixtures, is manufac-
tured as thinly as possible to minimize the mass transfer resistance. To ensure
mechanical stability, the active layer is combined with a porous structure.
The thickness generally is around 0.05–0.5 mm for the active layer and
30–200 mm for the porous layer, respectively.
Asymmetric membranes can be differentiated as integrally asymmetric
and composite asymmetric. Integrally asymmetric membranes are produced
by phase inversion and consist of one material, so that the active layer and
the porous support structure only differ by the pore size. During the
manufacturing of composite membranes, a thin polymer layer is applied
on a porous structure, where different polymers for the active layer and
the porous layer can be selected, although a separate optimization of the
two layers with regard to stability and separation performance is possible.
During the permeation process, the gas molecules absorb into the mem-
brane polymer on the feed side, diffuse through the membrane, and desorb
on the permeate side. According to this solution-diffusion model, the gas
transfer through the membrane can be calculated:
n_00k ¼ Pk $ xk pF yk pP (1)
The flux of a component k through the membrane is proportional to the
partial pressure difference. The permeance Pk is the permeability Pk related
to the membrane thickness d. The permeance is a gas specific and module
specific parameter and has to be determined experimentally. The permeance
is often indicated in GPUs (gas permeation units) whereby one GPU is equal
to 2:7$103 m3N ðm2 h barÞ1 .
The permeance for component k is proportional to the diffusivity Dk and
the solubility Sk of component k in the membrane and inversely propor-
tional to the membrane thickness d:
Dk $Sk
Pk w (2)
d
This correlation illustrates the importance of a preferably thin active
layer. Diffusivity and solubility depend on the respective gas but also on
the polymer.
The diffusion coefficients of gas molecules characterize their mobility in
the polymer and diffusion coefficients generally increase as gas molecule sizes
decrease.
The solubility describes the number of gas molecules sorbed in a polymer
and solubility depends on the boiling point of the respective gas. The solu-
bility rises with a rising boiling point, so that condensable gases generally
have higher solubility than permanent gases.
In view of the influence of the polymer on diffusivity and solubility, the
polymers can be divided into two groups characterized by their glass transi-
tion temperature: Polymers with a glass transition temperature lower than
the operating temperature are in a rubbery state, ones with a glass transition
temperature higher than the operating temperature are glassy. Diffusion ef-
fects dominate mass transfer through glassy polymers. Therefore, smaller
molecules preferentially pass the membrane even if their solubility is rela-
tively low. By contrast, for rubbery polymer membranes the mass transfer
mainly depends on solubility.
Membranes in a glassy state generally have high permeabilities for water va-
por, helium and hydrogen, but are less permeable for nitrogen, methane and
566 Thomas Harlacher and Matthias Wessling
N2 0.20
O2 0.26
CH4 0.42
CO2 1.05
C3H6 1.95
C3H8 2.03
hollow fibres or on the shell side in hollow fibre modules. Furthermore, the
feed and permeate can flow in co-current, counter-current, cross flow, and
ideal-mixed compartments.
For pressures higher than 15 bar, the feed has to flow on the shell side
because of the pressure stability of the hollow fibres. Otherwise, the feed
flow through the hollow fibre lumen is advantageous. In the lumen a defined
flow is realized, whereas channeling often occurs on the shell side, which re-
duces the module separation performance. Furthermore, the influence of
pressure losses is generally lower for applications with a lumen-side feed.
Counter-current flow is to predominate flow conduction in hollow
fibre modules. Since not only the concentrations but also the pressure influ-
ences the driving force, counter current is not generally the superior flow
conduction. If the influence of pressure losses on the driving force is higher
than the influence of concentration changes, the parallel flow leads to a
favorable driving force along the membrane.
In spiral wound modules the feed and permeate stream are led in cross
flow whereupon the channels for feed and permeate are constructional pre-
determined. In envelope modules (plate and frame modules) the flow con-
struction is a combination of co-current and counter-current flows.
3. PROCESS DESIGN
3.1 Pressure Ratio and Stage Cut
Next to selectivity as a membrane property, two operating parameters
determine the separation performance of a gas separation system: the partial
pressure ratio and the stage cut.
According to the mass transfer equation, a flux of a component i through
the membrane only occurs if the partial pressure of component i on the feed
side is higher than its partial pressure on the permeate side:
x i pF > yi pP : (5)
Hence, the maximum separation which can be achieved by the mem-
brane is:
yi pF
< ¼4 (6)
xi pP
The enrichment in the permeate relative to the feed is always smaller
than the feed-to-permeate pressure ratio, even at high membrane
Gas–Gas Separation by Membranes 569
selectivities. Even though the permeate purity rises with increasing pressure
ratios, maximum industrial applied pressure ratios are limited and depend on
the individual process. Since large pressure ratios require expensive compres-
sors or vacuum equipment and significant energy demands are required, the
applied pressure ratio is usually between five and 20.
The stage cut is the fraction of the feed stream which permeates through
the membrane:
permeate flow rate
q¼ (7)
feed flow rate
The stage cut has an influence on the trade-off between a high purified
retentate stream and a high concentrated permeate stream. With a rela-
tively low membrane area for a defined feed stream, low stage cuts are real-
ized and the permeate stream is highly concentrated. However, only a
small fraction of the feed stream is separated and the resulting retentate
stream resembles the feed stream in flow rate and composition. In contrast,
high retentate purity is achieved at high stage cuts. In this case, a high frac-
tion of the feed stream permeates through a relatively high membrane area
and the enrichment of fast permeating components in the permeate is
insignificant.
These parameters illustrate the limit of a single module separation perfor-
mance. A retentate-sided product can be concentrated to any purity by a sin-
gle membrane module with any pressure ratio higher than one and
selectivity. Yet, the achievable product recovery declines with increasing
product purity. In contrast, the achievable purity of a permeate product in
a single membrane module is limited by the membrane selectivity and the
pressure ratio.
Compressor
Vacuum
pump
Acidic sulfur-containing gases (H2S, SO2) easily react with the mem-
brane components and cause destruction of the membrane. Water vapor
can destroy the selective active layer as well. Moreover, water vapor can
induce plasticization which significantly reduces the membrane selectivity.
Contamination of the membrane surface and support layer with liquids,
such as water or condensed hydrocarbons, reduces the mass transfer rate
through the membrane by inducing additional resistance. Particles can phys-
ically damage the active layer or block the feed or permeate channels, in
particular the thin hollow fibres have to be protected from particles. The
operating temperature significantly influences the module separation perfor-
mance and high temperatures even cause a thermal degradation of the mem-
brane material.
To reduce the impact of the treated gas on the membrane module, a feed
pretreatment has to be implemented. Applicable units in a pretreatment pro-
cess are as follows:
• a coalescing filter, knockout drum or mist eliminator vessel for liquid
elimination;
• an adsorbent guard bed for the removal of trace contaminants, such as
hydrocarbons, H2S, and SO2;
• a particle filter for dust removal after the adsorbent bed; and
• a heat exchanger to set the desired feed temperature.
Retentate
Retentate
Compressor
Feed Feed
Compressor
Compressor
Permeate
Permeate
modules. The retentate of the second stage is recycled to the first module
inlet. A higher permeate purity compared to the one-stage process is realized,
but an additional compressor and higher membrane area are required. Even
several membrane stages can be realized theoretically, economically the num-
ber of stages is limited to two, exceptionally to three in industrial applications.
Combinations of a multistep and multistage design are possible, but they
are only rarely used due to the higher complexity of the process and higher
costs. Generally, several processes with module interconnection can be
designed. The optimum process design and economically acceptable process
complexity depend on the separation task.
Within a certain module interconnection, there are several parameters
which influence the module separation performance and have to be taken
into account. Next to specifying pressures and temperatures, the membrane
material and membrane area of each module have to be determined. The use
of various membrane materials in the different steps or stages might improve
process separation performance. The partition of the membrane area in
several identical constructed parallel lines potentially allows a dynamic
adapting of the membrane area to a change of the feed flow by switching
on and off the parallel lines.
Combining a membrane process with a conventional separation technol-
ogy in a hybrid process may be attractive in using the advantages of the
different technologies. The membrane process may outperform conven-
tional technologies for bulk removal and enrichment, but for further sepa-
ration to high purities (parts per million range) the conventional
technologies may perform better.
Gas–Gas Separation by Membranes 573
3
Process costs [€ m ]
2
-3
Total
1 Driving force
Membrane
0
5 10 15 20 25
Pressure ra o [-]
Figure 6 Example of the main costs for gas permeation processes as a function of the
pressure difference.
574 Thomas Harlacher and Matthias Wessling
Table 6 Industrial application areas for gas permeation processes (Melin 2007)
Gas
components Field of application Remarks and technical challenges
by using a partial stream of the retentate to increase the driving force for
oxygen. In this example, only compression as the driving force generator
is investigated.
1.0 0.5
Φ = 11 99 % N2 PI
0.8 PI 0.4
N2 recovery [-]
N2 recovery [-]
0.4 0.2
PDMS
0.2 0.1
PDMS
0.0 0.0
0.90 0.95 1.00 5 10 15
N2 purity [-] Pressure ra o [-]
100
10 PDMS
-3
-3
2
PI
10 1 PPO
PPO PI
PDMS
1 0.1
5 10 15 5 10 15
Pressure ra o [-] Pressure ra o [-]
Figure 8 Specific energy demand and required membrane area for a fixed product pu-
rity of 99% N2.
Nitrogen product
99 % N2
92.9 % N2
Air 81.9 % N2
79 % N2
88.0 % N2
Nitrogen product
99 % N2
95.9 % N2
90.5 % N2
shows the product recovery for the different process designs as a function of
the product purity and the pressure ratio. For multistep processes, the product
recovery is significantly increased compared to the one-step process. This ef-
fect is particularly important, when the feed is a valuable product and is not
gratis, as the air in this example. Next to the membrane material and pressure
ratio selection, the membrane area ratio between the respective process steps
can be adapted for economic process optimization.
1.0 0.7
Φ = 11 99 % N2
0.6 Three-Step
0.8 Three-Step
0.5
Two-Step
N2 recovery [-]
N2 recovery [-]
0.6 0.4
Two-Step
0.4 0.3 One-Step
One-Step 0.2
0.2
0.1
0.0 0.0
0.90 0.95 1.00 5 10 15
N2 purity [-] Pressure ra o [-]
Air
21% O2
Air
21% O2
0.7
0.6 PI
O2 purity [-]
0.5
PPO
0.4
0 2 4 6 8
3
Feed flow [m /h]
Figure 12 O2 purity in a one-step membrane process.
the feed permeates through the membrane. The maximum pressure differ-
ence is limited, so that the required membrane area is comparatively high
(Baker 2002).
The separation performance of a single-stage membrane process using PI
and PPO membranes is shown in Figure 12. By increasing the feed flow, the
O2-depletion in the feed stream of the membrane module is reduced. Thus,
the driving force for O2 is kept higher resulting in higher O2 purity in the
permeate. For a minimum O2 depletion a maximum O2 purity is achieved.
For the PPO membrane with a N2/O2 selectivity of 4.4, the maximum O2
purity is around 50%, for the PI membrane with a N2/O2 selectivity of 6.3
the maximum purity is around 60%.
For higher O2 purities, a hybrid process combining a membrane step
with an additional purification step can be applied. For small units in which
flow rates of up to 200 tons a day are treated, vacuum swing adsorption is
used; cryogenic processes are the most economic processes for product
flow rates of more than 200 tons a day. By air pretreatment using mem-
brane units, the second stage unit is smaller and cheaper compared to the
adsorption or cryogenic stand-alone units fed with air (Baker 2002).
(a) CH4
(b) CH4
Raw gas
Raw gas
CO2 CO2
Raw gas
Raw gas
CO2
CO2
minimal maintenance and operator attention allows the operating of the unit
without detailed technical know-how.
Several process designs are discussed in the literature for membrane sup-
ported biogas treatment. The selection of the most suitable process strongly
depends on the feed composition and the required methane purities.
Figure 14 shows a selection of the discussed process designs.
Since single-stage gas permeation processes are inefficient due to signif-
icant CH4 losses, multistage gas permeation processes are applied to obtain
high product purity and simultaneously increase the CH4 recovery of the
upgrading system.
Figure 14 shows different membrane stand-alone processes. In process
(a), the first module is in charge of the CO2 bulk removal. A relatively
high CO2 fraction is received in the permeate, but the product stream leav-
ing this module does not comply with the gas grid requirements. In order to
purify the CH4 stream, a second membrane module is applied which con-
trols the product purity. In this module, the CH4 losses into the permeate are
quite high. By recycling this stream in front of the compressor, the CH4
recovery of the process is increased.
In process (b) the required product composition is obtained in one step.
Next to the CO2 and the trace components, a huge amount of CH4 perme-
ates through the membrane. Here, a second stage is added to reduce the
CH4 losses. The driving force for this module is generated by a second
Gas–Gas Separation by Membranes 583
Raw gas
Adsorp on
Absorp on
Adsorp on
Absorp on
Raw gas
CO 2
CO 2 CO 2
REFERENCES
Baker, R.W., 2002. Future directions of membrane gas separation technology. Ind. Eng.
Chem. Res. 41 (6), 1393–1411.
Baker, R.W., 2004. Membrane Technology and Applications, second ed. John Wiley and
Sons.
Calabro, V., Basile, A., 2011. Economic analysis of membrane use in industrial application.
In: Basile, A., Nunes, S.P. (Eds.), Advanced Membrane Science and Technology for
Sustainable Energy and Enviromental Applications. Woodhead Publishing.
Drioli, E., Giorno, L., 2010. Comprehensive Membrane Science and Engineering. Elsevier.
Drioli, E., Barbieri, G., 2011. Membrane Engineering for the Treatment of Gases. Royal
Society of Chemistry.
Favre, E., 2011. Polymeric membranes for gas separation. In: Drioli, E., Barbieri, G. (Eds.),
Membrane Engineering for the Treatment of Gases. Royal Society of Chemistry.
Follmann, P.M., 2010. Membrane gas separation processes for post combustion CO2 capture
from coal fired power plants, dissertation.
Freeman, B.D., 1999. Basis of permeability/selectivity tradeoff relations in polymeric gas sep-
aration membranes. Macromolecules 32, 375–380.
Melin, T., Rautenbach, R., 2007. Membranverfahren – Grundlagen der Modul- und Anla-
genauslegung. aktualisierte und erweiterte Auflage, third ed,. Springer.
Merkel, T., Lin, H., Wei, X., Baker, R.W., 2010. Power plant post-combustion carbon
dioxide capture: an opportunity for membranes. J. Membr. Sci. 359, 126–139.
Mourgues, A., Sanchez, J., 2005. Theoretical analysis of concentration polarisation in
membrane modules for gas separation with feed inside hollow fibres. J. Membr. Sci.
252, 133–144.
Mulder, J., 1996. Basic Principles of Membrane Technology, second ed. Springer.
Nunes, S.P., Peinemann, K.V., 2006. Membrane Technology in the Chemical Industry.
revised and extended edition, second ed. WILEY-VCH.
Ohlrogge, K., Ebert, K., 2006. Membranen. WILEY-VCH.
Robeson, L.M., 1991. Correlation of separation factor versus permeability for polymeric
membranes. J. Membr. Sci. 62, 165–185.
Scholz, M., Wessling, M., 2013. Transforming biogas into biomethane using membrane
technology. Renew. Sustain. Energy Rev. 17, 199–212.
Yampolskii, Y., Pinnau, I., Freeman, B., 2006. Materials Science of Membranes for Gas and
Vapor Separation. WILEY-VCH.
Yampolskii, Y., Freeman, B., 2010. Membrane Gas Separation. WILEY-VCH.