See discussions, stats, and author profiles for this publication at: https://www.researchgate.

net/publication/309624011

Techno-economical evaluation of membrane based biogas upgrading system:

A comparison between polymeric membrane and carbon membrane
technology

Article  in  Green Energy & Environment · October 2016

DOI: 10.1016/j.gee.2016.10.003

CITATIONS READS

73 4,985

3 authors, including:

Shamim Haider May-Britt Hägg

Aker Solutions Norwegian University of Science and Technology
9 PUBLICATIONS   210 CITATIONS    114 PUBLICATIONS   5,431 CITATIONS   

SEE PROFILE SEE PROFILE

Some of the authors of this publication are also working on these related projects:

Carbon Membranes for CO2 Removal from High Pressure Natural gas in Subsea Process (CO2Hing) View project

CLEANWASTE - Clean and environmentally friendly animal waste technologies or fertilizer and energy production View project

All content following this page was uploaded by Shamim Haider on 27 April 2017.

The user has requested enhancement of the downloaded file.

 Available online at www.sciencedirect.com

ScienceDirect
Green Energy & Environment 1 (2016) 222e234
www.keaipublishing.com/gee

Research paper

Techno-economical evaluation of membrane based biogas upgrading

system: A comparison between polymeric membrane and carbon membrane
technology
Shamim Haider, Arne Lindbråthen, May-Britt H€agg*
Norwegian University of Science and Technology, NTNU, Department of Chemical Engineering, 7491 Trondheim, Norway
Received 1 September 2016; revised 21 October 2016; accepted 22 October 2016
Available online 31 October 2016

Abstract

A shift to renewable energy sources will reduce emissions of greenhouse gases and secure future energy supplies. In this context, utilization
of biogas will play a prominent role. Focus of this work is upgrading of biogas to fuel quality by membrane separation using a carbon hollow
fibre (CHF) membrane and compare with a commercially available polymeric membrane (polyimide) through economical assessment. CHF
membrane modules were prepared for pilot plant testing and performance measured using CO2, O2, N2. The CHF membrane was modified
through oxidation, chemical vapour deposition (CVD) and reduction process thus tailoring pores for separation and increased performance. The
post oxidized and reduced carbon hollow fibres (PORCHFs) significantly exceeded CHF performance showing higher CO2 permeance
(0.021 m3(STP)/m2 h bar) and CO2/CH4 selectivity of 246 (5 bar feed vs 50 mbar permeate pressure). The highest performance recorded through
experiments (CHF and PORCHF) was used as simulation basis. A membrane simulation model was used and interfaced to 8.6 V Aspen HYSYS.
A 300 Nm3/h mixture of CO2/CH4 containing 30e50% CO2 at feed pressures 6, 8 and 10 bar, was simulated and process designed to recover
99.5% CH4 with 97.5% purity. Net present value (NPV) was calculated for base case and optimal pressure (50 bar for CHF and PORCHF). The
results indicated that recycle ratio (recycle/feed) ranged from 0.2 to 10, specific energy from 0.15 to 0.8 (kW=Nm3 feed ) and specific membrane
area from 45 to 4700 (m2 =Nm3 feed ). The high recycle ratio can create problems during start-up, as it would take long to adjust volumetric flow
ratio towards 10. The best membrane separation system employs a three-stage system with polyimide at 10 bar, and a two-stage membrane
system with PORCHF membranes at 50 bar with recycle. Considering biomethane price of 0.78 $/Nm3 and a lifetime of 15 years, the techno-
economic analysis showed that payback time for the best cascade is 1.6 months.
© 2016, Institute of Process Engineering, Chinese Academy of Sciences. Publishing services by Elsevier B.V. on behalf of KeAi Communications
Co., Ltd. This is an open access article under the CC BY-NC-ND license (http://creativecommons.org/licenses/by-nc-nd/4.0/).

Keywords: Carbon membrane; Biogas upgrading; Techno-economical analysis; NPV calculations

1. Introduction [2e4]. Development and use of renewable energy have

become of major importance for long-term sustainability. By
Combustion of fossil fuel to meet the ever-increasing en- the year 2020, it is predicted in a Swedish case study that
ergy demand has resulted in depletion of natural resources almost 20% of energy will be produced from renewable
[1]. At the same time, greenhouse gas emissions, especially sources [5].
CO2 produced by combustion of fossil fuel, is a major source Biogas produced by microbial digestion of farm waste or
of the contribution to global warming and climate change sewage waste contains high concentrations of methane, which
could be combusted to meet the energy and power demands.
The biogas produced by microbial digestion of waste consists
* Corresponding author. of several gases among which CH4 and CO2 account for most
E-mail address: [email protected] (M.-B. H€agg).

http://dx.doi.org/10.1016/j.gee.2016.10.003
2468-0257/© 2016, Institute of Process Engineering, Chinese Academy of Sciences. Publishing services by Elsevier B.V. on behalf of KeAi Communications Co.,
Ltd. This is an open access article under the CC BY-NC-ND license (http://creativecommons.org/licenses/by-nc-nd/4.0/).
 S. Haider et al. / Green Energy & Environment 1 (2016) 222e234 223

of the volume fraction. Other gases including H2S, N2, O2 and The Robeson plot in Fig. 1 shows that a membrane with high
water vapours coexist in traces. Depending upon the raw CO2 selectivity usually has low permeability [25,26].
material, digestion procedure and process condition, the CO2 The process to produce biomethane should be inexpensive
concentration in biogas can reach up to 50%. This high con- and simple to control. Commercially available membranes for
centration of CO2 significantly reduces the calorific value of CO2/CH4 separation are mostly polymeric membranes, and
biogas [6e9]. The produced biogas needs to be enriched in these membranes do not have high enough separation factor
methane (CH4 > 95%) by removing CO2 and other impurities (selectivity) to achieve a high recovery of CH4 in a single
from the gas [10,11]. stage. The amount of energy required for biogas upgrading is a
The CO2 can be removed from a gas stream by many key factor when selecting a technology for this purpose. Due
different techniques. Some of the most investigated techniques to low selectivity but high permeance of commercial mem-
for the upgrading of biogas involve CO2 capture by physical or branes, big recycle stream has to be treated if high recovery
chemical absorption in liquid, pressure swing adsorption, (99.5% of CH4) and pipeline spec (97.5% CH4) has to be
membrane technology or cryogenic separation [7,12e16]. reached in the two-stage system. This results in high costs due
Among all these techniques, membrane technology offers to the compressor price and compressor duty while in opera-
several advantages like the compact modular design, small tion. The process operating cost can be reduced by optimiza-
footprint, low capital and operational cost, simple operation tion of feed pressure, inter stage pressure and recycle flow. A
and easy maintenance [17e19]. Due to these advantages, three-stage membrane separation could also be helpful to
biogas upgrading by using membrane technology has gained a achieve high recovery and purity; however, operational
lot of attention. Much work has been done in the development complexity regarding intermediate pressure and biogas com-
of a competitive membrane material for different gas separa- ponents concentration, suggest that adjustment of these vari-
tion applications during the last two or three decades. The ables is the key towards optimization. Plasticization inhibits
most important factor in membrane separation is the mem- the polymeric membranes to a threshold pressure [27],
brane material. Different materials have been suggested, here whereas CHF membrane has shown a stable performance and
is only referred to a few representing both polymers and car- no plasticization up to 50 bar or even higher [28,29].
bon membranes [20e23]. Many researchers have been conducting Simulations and
A novel carbon hollow fibre (CHF) membrane was syn- modelling of multi-stage membrane systems to evaluate and
thesized at NTNU using cellulose acetate as a precursor, optimize membrane systems for CO2 capture [30e37]. Baker
which was de-acetylated to cellulose prior to the carbonization et al. [17] have provided a guideline for conducting simula-
process. The CHF membrane showed high performance and tions on commercially available membranes, which also shows
potential to become an economically viable solution for biogas the comparison between membrane system and amine ab-
purification [24]. Their membrane showed high CO2/CH4 sorption process. Baker suggested that membrane technology
selectivity and good CO2 permeance. These CHF membranes is suitable in small (less than 6000 Nm3/h) and medium scale
were developed and tested on a pilot scale for biogas by (6000e50,000 Nm3/h) processes.
MemfoACT AS a company which has now closed down. The This study intends to demonstrate biogas upgrading with
CHF membranes showed promising results and attractive membrane separation technology. Permeance and selectivity of
properties for biogas upgrading under real test conditions at a gases in polymeric membranes for biogas upgrading is
biogas plant in Southern Norway. The modules of 2 m2 area
were made to test the membrane performance at the pilot
plant. The plant was capable of processing 20 Nm3/h of
biogas. CHF membrane modules with feed on the shell side,
showed weak mechanical properties when tested at high
pressure, and many broken fibres inside the module were
observed. Secondly, the potting which should bind that many
fibres together at high pressure was not god enough and more
research was needed. Modules with a small number of fibres (a
few cm2 modules) tested in the laboratory showed good me-
chanical properties up to 70 bar. CHF membranes displayed a
CO2/N2 selectivity of 30 with a CO2 permeance of 5.5E-
3 m3(STP)/m2 h bar. A further modification was done by
applying oxidation, CVD, and reduction process on CHF
membranes, which dramatically increased the CHF membrane
performance (CO2/N2 selectivity of 82 and CO2 permeance of
2.1E-2 m3(STP)/m2 h bar).
The major separation in biogas upgrading process is CO2/
CH4 separation. The membranes for CO2/CH4 separation are
based on solution-diffusion mechanism, there is usually a
trade-off between CO2 separation selectivity and permeability. Fig. 1. Robeson upper bounds for CO2/CH4 membrane separation [31].
224 S. Haider et al. / Green Energy & Environment 1 (2016) 222e234

abundantly available in the literature. However, similar data for operating conditions and separation efficiency of CHF mem-
carbon membranes has seldom been reported. Various modular brane increased appreciably after oxidation and reduction
configurations containing membranes with different CO2 se- process (PORCHF). The first oxidation step will produce quite
lectivities, have been investigated in the current study, using large membrane pores with fairly low selectivity, then chem-
HYSYS simulations for optimal performance and minimum ical vapour deposition (CVD) was performed using propylene
energy consumption. Considering that the membrane separation to coat the membrane surface with a thin polymeric layer
system fulfils the German national standard for biogas as closing the pores again. The oxidation process was then
vehicle fuel, the evaluation focus to achieve 99.5% CH4 re- repeated to open (tailoring) the pores in the newly formed
covery and 97.5% CH4 purity. The important parameters such as layer to fit the molecular sieving of the gases in question (here
compressor duty, recycle ratio (recycle/feed) and membrane CO2eCH4), finally followed by a reduction process, which
area are discussed. In the end, techno-economical evaluation of reduced the aging of the membrane by stabilizing the pores.
the entire plant, including running costs and net present value
(NPV) have been calculated. The purpose of this study is to 2.4. Transport mechanism and membranes used in this
present an economically viable scheme to upgrade biogas by work
using membranes with high CH4 recovery. Biogas composition
depends on the source of the gas as shown in Table 1 [6]. The mass transport properties of CHF and PORCHF were
measured with the single pure gases CO2, O2, and N2 at different
2. Material and methods feed pressure and experiments were carried out without sweep
gas on the permeate side. He et al. has performed the mixed gas
2.1. Materials experiments on the same type of carbon membrane (same pro-
tocol) and results indicated that the membrane performance for
Acros Organics (Belgium) supplied cellulose acetate (CA) CO2 separation is the same or even higher in some cases for
and 1-Methyl-2-pyrrolidinone, while 99.5% (NMP) and PVP mixed gas as compare to single gas separation [39]. Due to fire
(Polyvinylpyrrolidone) were purchased from SigmaeAldrich hazard limitations, CH4 was not tested at membrane production
(Norway). Ionic exchanged water was used for coagulation. facility, only in the lab. Therefore, the values for CH4 gas are
estimated values (three times of N2 selectivity) in this work
2.2. Hollow fibre preparation based on work done by He et al (Fig. 6 of the article) [39]. The
performance of the membrane was evaluated by measuring the
Using a dry-wet spinning process at commercial scale plant, CO2 permeance in m3(STP)/(m2 h bar) and CO2/N2 selectivities
delivered by Philos Korea, cellulose acetate hollow fibres (a) using the Eqs. (1) and (2).
(CAHFs) were spun from a dope composed of CA/NMP/PVP.
qp;i qp yp;i Ji Pi
CAHF were soaked in water and glycerol solution respectively ¼ ¼ ¼ ð1Þ
after the spinning process. CAHF were deacetylated with Am ðph xi  pl yi Þ Am ðph xi  pl yi Þ ðph xi  pl yi Þ l
NaOH/isopropanol/water solution and then dried in a humidity-
Pi
controlled environment. Carbon hollow fibres (CHFs) were f ¼ ð2Þ
prepared by the carbonization of deacetylated CAHF in the Pj
presence of CO2 and N2. Details of this membrane preparation
where J (m3(STP)/m2 h) is the flux of gas component i, qp is
are given in the patent held by H€agg and Lie [24].
the volume of the permeating gas (i) (m3(STP)/h), Pi is the
permeability of gas component i ((m3(STP)/m2 h bar), Ph and
2.3. Modification
Pl are feed and permeate side pressures (bar), xi and yi are the
mole fractions of component i on the feed and permeate sides
The pore size of CHF membrane, already mounted into a
and Am (m2) is the membrane area [36].
module with stainless steel casing, was tailored to enhance the
A benchmarking polymeric membrane is considered for
membrane separation properties. The Fig. 2 illustrate the steps
comparison, as a future biogas upgrading membrane with a
followed in this work, as also reported in the patent held by
1 mm thick selective layer, having a permeability of 100 barrer
Soffer et al. [38]. The durability, mechanical stability at
and a selectivity of 100 for CO2/CH4, which is above Robeson
upper bond 2008. Gas permeation properties of CHF and
Table 1 PORCHF with other membranes used in the simulation of this
Biogas composition from various sources [6]. work, are shown in Table 2.
Component Farm plant Sewage digester Landfill
CH4 55e58 61e65 47e57 3. Process description and simulation method
CO2 37e38 34e38 37e41
N2 Trace Trace 1e17 3.1. Pre-treatment of biogas
O2 Trace Trace 0e2
H2S <1 <1 <1
H2O 4e7 4e7 4e7 As shown in Table 1, raw biogas contains several impu-
Aromatic hydrocarbon Trace Trace Trace rities, like water, dust, H2S, CO2, siloxanes, hydrocarbons,
 S. Haider et al. / Green Energy & Environment 1 (2016) 222e234 225

Oxidation CVD Oxidation Reduction

(Opening pores) (Surface coating) (Opening pores) (Stabalization)

Fig. 2. Steps followed for membrane pore tailoring.

97% initial values before the membrane was exposed to high

Feed Gas Stage. 1 Biomethane
Membrane humidity.
Biogas can contain up to 3000 ppm H2S, which is a very
Compressor
high amount to expose the membranes to. Different techniques
are used to remove sulphur from biogas: silica gel, activated
carbon, iron sponge and biological filtration. Polyimide
Permeate membranes, unlike CHF and PORCHF, have high H2O and
H2S permeability, which make it suitable for biogas upgrading
Fig. 3. Single stage membrane process.
process without special pre-treatment as these components
will permeate with CO2 [42]. However, biomethane as a
NH3, oxygen and several other components that must be vehicle fuel (German legislation) demands sulphur contents
removed in order to increase the membrane lifetime to avoid below 4 ppm and water dew point of 10  C at 200 bar, which
corrosion of the upgrading system and to comply for the require pre-treatment in both carbon and polyimide membrane
biogas being approved as biomethane for vehicle fuel. Some separation process as considered in this work. It can be ex-
membrane materials like polyimide have high permeability pected that biogas contains traces of organic components such
and can work in the presence of components which often are as alkanes, halogenated hydrocarbons, ketones, aromatic
harmful to the membrane, like H2S and H2O [41]. As shown in compounds, siloxane, alkyl sulphides and alcohols depending
Table 1, Impurities like H2S and aromatic hydrocarbons need on the substrate used for anaerobic digestion [42,43]. Toluene,
to be removed before biogas encounter the carbon membrane. an aromatic hydrocarbon, is usually detrimental to the mem-
Biogas is usually saturated with water, and the amount of brane, and it significantly decreased CO2 permeability and
water which needs to be removed depends on how much water CO2/N2 selectivity when tested at 100 ppm for CHF. The
is allowed to enter into the compressor and membrane system. PORCHF membrane showed a significant increase in perme-
This water can be removed before entering the compressor or ability of CO2, whereas selectivity of CO2/N2 decreased.
after membrane separation prior to high-pressure compression Heptane showed the similar effect as toluene when tested
for vehicle fuelling or pipeline requirements. In the case of 1000 ppm for both CHF and PORCHF. However, the presence
carbon membranes, less than 30% relative humidity (RH) is of methanol significantly lowered the permeability and CO2/
acceptable [29]. However, the presence of water may influence N2 selectivity in both CHF and PORCHF. The presence of 300
the separation of the other components, and it was documented ppm toluene on polyimide membranes decreased the CO2
that the permeability of N2 increased and CO2 decreased for permeability significantly and a slight decrease in CO2/CH4
CHF and PORCHF at 50% RH. When the fibres were again selectivity was observed by Wind et al. [44]. Many of these
dried, it was observed that the permeability of CO2, N2 and components have not yet been tested and will need further
selectivity of CO2/N2 slightly increased as compared to the investigations. With the presently available membranes, it was

97%
Feed Gas Stage. 1 Biomethane
Membrane

Compressor
Permeate 1
Stage. 2
Membrane
Retentate 2

Permeate 2

Recycle stream

Fig. 4. Two stage membrane process.

226 S. Haider et al. / Green Energy & Environment 1 (2016) 222e234

97%
Feed Gas Stage. 1 Retentate 1 Stage. 2 Biomethane
Membrane Membrane

Compressor
Permeate 2

Permeate 1
Stage. 3
Membrane Retentate 3
Mixer

Permeate 3

Recycle stream

Fig. 5. Three stage membrane process.

Fig. 6. A three-stage membrane system integrated into Aspen HYSYS.

concluded that pre-treatment is needed before the membrane is recovery and purity has been simulated. Polyimide membrane
exposed to biogas. is a commercially available membrane with modest selectivity,
and in order to judge the separation performance of this
3.2. Membrane configurations membrane, using only one stage simulation at 10 bar feed
pressure is evaluated (Fig. 3).
3.2.1. Three cases
Membrane processes may vary with respect to operational 3.2.1.2. Two-stage membrane process. To maximize the re-
units, their arrangement and applied process conditions. Three covery of biomethane from biogas, a two-stage system has been
different cases were evaluated in this study. simulated using all four membrane types with different selec-
tivities. The permeate from the first stage enters into the second
3.2.1.1. Single stage membrane process. A membrane system stage membrane to recover more CH4. Retentate from second
using only one stage to separate biogas for required methane stage mix with feed in the form of recycle stream prior to the

Table 2
Gas permeation properties used in this work.
Membrane type Permeance, (m3(STP)/(m2 h bar)) Single gas selectivity Ref.
[GPU], (m3(STP)/(m2 h bar)) 1 GPU ¼ 2.736E-3
CO2 CH4 N2 CO2/CH4 CO2/N2
Polyimide hollow fibre 5.6E-2 [20.5] 1.7E-3 [0.62] 6.0E-4 [0.22] 33 31 [40]
CHF 5.5E-3 [2.02] 6.1E-5 [0.0224] 1.8E-4 [0.0673] 90 30 This work
PORCHF 2.1E-2 [7.75] 8.6E-5 [0.0315] 2.6E-4 [0.0943] 246 82 This work
Benchmarking polymeric 1.4E-2 [5.01] 1.4E-4 [0.0501] 4.1E-4 [0.15] 100 33 See text
membrane
 S. Haider et al. / Green Energy & Environment 1 (2016) 222e234 227

compressor for better CH4 recovery (Fig. 4). There is no 3.4. Process conditions
compression between the stages and the pressure of permeate 1
is adjusted with flow valve at Retentate 2. The pressure at A water-saturated biogas stream of 300 Nm3/h with
permeate 1 was kept at a constant value for the specific feed 3000 ppm of H2S is considered as a base-case. Biogas enters
pressure, obtained with formula as shown in Eq. (3). The basis of into biological H2S remover for bulk removal down to be-
the formula is to maximize the pressure ratio (hence maximum tween 50 and 100 ppm and passes further through activated
perm purity) on both stages simultaneously by setting the charcoal, where H2S is taken down to below 1 ppm. The
interstage retentate. It was observed that the intermediate refrigeration process followed by zeolite molecular sieve is
pressure value acquired, gave optimized membrane area for used to remove water from biogas in order to achieve a dew
required purity and recovery of CH4. point of 10  C at 200 bar. Biogas is compressed (single stage
qffiffiffiffiffiffiffiffiffiffiffiffi
  compression is used in simulations but multi stage compres-
Pinterstage ¼ Pfeed $Ppermeate2 ð3Þ sion with inter stage cooling is considered in cost calculation)
to required feed pressure (6, 8, 10 bars and all pressure values
are absolute in this work) and then filtered to remove dust and
3.2.1.3. Three-stage membrane process. A three-stage mem- oil droplets before entering the membranes. Biogas is fed to
brane system may give better separation and reduce energy the membrane and the resulting biomethane is compressed up
demand [45]. A three-stage configuration is shown in Fig. 5. to 250 bars before it is stored for further usage as a vehicle fuel
Evonik Fibres Gmbh has applied for the patent of this (see Table 3). A three-stage simulation using ChemBrane
configuration, and according to the patent, no one but Evonik model in Aspen Hysys is shown in Fig. 6.
can use a membrane with a CO2/CH4 selectivity of 30 or
higher on the first stage [46]. Considering this patent is 3.4.1. Effect of N2
accepted, the energy demand of the process may increase by Using biological desulfurization may result in the addition
0.027 kWh/Nm3 for other membrane providers. The three- of air components in the biogas stream. CH4 and N2 have close
stage system is simulated and economically evaluated here selectivity value for commercially available membranes,
by using polyimide membranes. which makes it difficult to remove N2 from CH4 to achieve
required CH4 purity. CHF membranes and PORCHF mem-
branes have shown CH4/N2 selectivity of 3, making it less
3.3. Simulation basis
vulnerable to N2 in the gas stream. Only PORCHF membrane
with the two-stage system has been simulated including N2.
 NTNU has an in-house membrane simulation model
The intermediate pressure needed some extra optimization to
(Chembrane) which can easily be integrated into Aspen
get required results. N2 concentration in feed and operating
HYSYS. This model uses fourth-order RungeeKutta
conditions is shown in Table 4.
method to calculate the flux along membrane length, and
then iteration over permeate values to converge to a so-
3.5. Cost estimation and economic parameters
lution. ChemBrane model is integrated into 8.6 V Aspen
HYSYS for all simulations in this work.
Accurate economic assessment of any process depends on
 Countercurrent gas transportation without sweep on
available design details, a method of analysis used for
permeate side has been used in all hollow fibre membrane
calculation and accuracy of available cost data. Therefore,
modules. Literature data shows that counter current flow
the economic calculations may differ considerably, as they
exhibits the superior separation and uses lowest membrane
are justified by the data available and also cost model. An
area in hollow fibre module design. This module design is
economic evaluation was performed to assess the different
very efficient and has a high packing density (can be up to
membranes and their configurations, by taking capital cost,
30,000 m2/m3 for certain designs) [47].
 The Peng Robinson sour fluid package was used.
 In order to run the simulations smoothly, only separation of
Table 3
the main components CH4 (50e70%) and CO2 (50e30%) is Process conditions and feed composition used in this work.
considered in the gas stream entering the membrane system.
Process conditions used in simulation
But techno-economic evaluation includes H2S removal,
water removal and dust removal system. Feed composition 30e50% CO2, balance CH4
Feed flow rate (Nm3/h) 300
 Intermediate pressure (inlet pressure for the stage 2 CH4 purity in product (%) 97.5
membranes in two-stage configuration and inlet pressure CH4 recovery (%) 99.5
for the stage 3 membranes in three-stage configuration) is S total <4 ppm
kept same for different membranes to balance the Water dew point 10  C at 200 bar
complexity of the system. Effect of intermediate pressure Feed pressure (bar) 6, 8, 10
Intermediate pressure in two stage (bar) 2.45, 2.82, 3.16
and sweet gas has already been studied by Deng et al. [31]. Intermediate pressure in three stage (bar) 2.45, 2.82, 3.16
 The adiabatic efficiency of the compressors is modelled as Flow pattern in membrane module Countercurrent
75%. Biogas delivery pressure (bar) 250
228 S. Haider et al. / Green Energy & Environment 1 (2016) 222e234

Table 4 the feed at the applied operating conditions, and methane re-
N2 concentration in feed and operating pressure. covery decreases with increase in CO2 content in the feed as
N2 in feed Feed pressure Intermediate pressure shown in Fig. 7. The single stage system, as expected gives the
% bar bar lowest recovery; therefore, multiple stage system with recycle
0 10 4.24 is discussed in the further results.
0.5 10 4.24 The feed pressure has a big effect on required membrane
1 10 3.7 area for different CO2 concentrations in the feed as shown in
1.5 10 3.55 Fig. 9. The results show that higher the CO2 present in the
2 10 2.95
2.1 10 2.95
feed, less area is required in this case to reach the targeted
product (CH4) specifications. Area needed at low feed pressure
(here 6 bar) is more sensitive to the feed concentration. Fig. 9
operating cost, pre-treatment cost and high-pressure demonstrates the required area for a two-stage separation
compression (CBG) cost into account. A high recovery of system of polyimide membrane with varying CO2 loadings
biomethane is achieved, resulting in a very small fraction of under a set of applied pressure. Results indicate that required
CH4 loss with permeate stream (CO2). Thus CO2 obtained area per Nm3 of feed gas under different concentrations of
on permeate side is 99% pure and could be used for other CO2 in the feed is three times higher when operating feed
applications. The price for the CO2 is not considered in this pressure is decreased from 10 bar to 6 bar. The effect of CO2
economical assessment. Table 5 is showing process param- concentration in the feed gas is more sensitive at low pres-
eters for economic assessment of a biogas upgrading suredthis is as expected in this range when considering the
process. basic equations.; here it shows the decline in required mem-
brane area from 175 m2/Nm3 to 125 m2/Nm3 when increasing
4. Results and discussion from 30- to 50% CO2, whereas, at 8 and 10 bar pressure, the
area is almost constant. The product purity is affected by
4.1. Membrane configurations selectivity limited region at 6 bar (Separation factor vs
permeate purity plot of WellereSteiner equation [53]). When
A constant pressure of 10 bar was applied to the feed gas the pressure is 8 bar or higher, the product purity is in the
with different CO2 concentration to see the influence on re- pressure ratio limited region. Two-stage membrane separation
covery and membrane area. The result (Fig. 8) shows that cascade using a polymeric membrane with CO2/CH4 selec-
methane loss is high up to 17% when 30% CO2 is present in tivity of 100 is presented in Fig. 10.

Table 5
Process parameters for economic assessment of biogas upgrading plant [51,52] ($ used in this work is US).
Total plant investment (TPI)
Polymeric membrane cost (PMC) $ 20/m2 [31,33,48,49]
Carbon membrane cost (CMCo) $ 100/m2
Installed compressors cost (CC) $ 8700  (HP/h)0.82
High pressure compressor cost (CBGC) Ccomp,ins ¼ 912$ (Wcomp)0.9315$ fm$ fi$ finst [50]
Fixed cost (FC) PMC/CMCo þ CC þ PTC þ CBGC
Base plant cost (BPC) 1.12  FC
Project contingency (PC) 0.2  BPC
Total facility investment (TFI) BPC þ PC
Start-up cost (SC) 0.10  VOM
TPI TFI þ SC
Annual variable operating and maintenance cost (VOM)
Contract and material maintenance cost (CMC) 0.05  TFI
Local taxes and insurance (LTI) 0.015  TFI
Direct labour DL, cost based on 8 h/day $ 15/h
Labour overhead cost (LOC) 1.15  DL
Utility cost (UC) ($/kWh) 0.07/kWh
VOM CMC þ LTI þ DL þ LOC þ MRC þ UC
Other assumptions
Membrane life for polyimide (t) 7.5 years
Membrane life for carbon (t) 5 years
Biomethane sales price ($) $ 0.8/Nm3
Nominal interest rate (%) 6%
Depreciation (t) 15 years
LCC/LCI factor (ordinary annuity factor) 9.7122
Plant availability (%) 96%
CO2/CH4 in feed (%) 40/60
 S. Haider et al. / Green Energy & Environment 1 (2016) 222e234 229

Biogas Pretreatment Biogas Upgrading Process Compressed Biomethane

Biomethane

Biological H2S

H2O Removal
Membrane High Pressure

Activated
Removal

System
Carbon
Biogas Feed Stage. 1 Compressor
Compressor

Membrane
Stage. 2

CO2

Fig. 7. Schematic diagram of biogas upgrading system.

CH Recovery 10 bar
83.5
Membrane specific Area 8.0 Membrane specific Area (m /Nm feed)
8 bar
6 bar
83.0

Membrane specific Area (m /Nm feed)

900
7.8
82.5
CH Recovery (%)

82.0 7.6

81.5 600
7.4
81.0

7.2
80.5

30 35 40 45 50 300
CO in feed (%)
200
30 35 40 45 50
Fig. 8. A single stage separation with polyimide membrane (10 bar, 23  C). CO in feed (%)

200 Fig. 10. Two stage membrane separation cascade (selectivity 100, T: 23  C).
10 bar
8 bar
6 bar 2500
10 bar
Membrane specific Area (m /Nm feed)

8 bar
150 6 bar
Membrane specific Area (m /Nm feed)

2000

100
1500

50 1000

500
30 35 40 45 50
CO in feed (%)

30 35 40 45 50
Fig. 9. Two-stage membrane separation cascade (Polyimide membrane, T: CO in feed (%)
23  C).
Fig. 11. Two stage membrane separation cascade (CHF membrane, T: 23  C).

Required area per Nm3 of feed is lowest when feed pressure and a linear decline in required membrane area can be
is 10 bar and 50% CO2 is present in the feeddthis is ac- observed for different CO2 loadings. Maximum area is
cording to theory for solution-diffusion separation. Fig. 11 required when feed pressure is 6 bar and the feed concentra-
shows the results for two-stage CHF membrane separation tion of CO2 is 30%. A similar trend is observed in the case of
230 S. Haider et al. / Green Energy & Environment 1 (2016) 222e234

10 bar
4.2. Recycle stream and compression duty
2000 8 bar
6 bar
Operating cost of a biogas upgrading process depends
Membrane specific Area (m /Nm feed)

largely on the compressor duty, and the recycle ratio (recycle/

1600
feed) higher than 1 can increase this compression energy
requirement to a higher level. Results obtained by the simu-
1200
lation of different membranes with two and three stage con-
figurations are plotted in Figs. 14 and 15. The Fig. 14 (a),
demonstrates the recycle ratio at 6 bar feed pressure, which is
800 seven for the two stage polyimide membrane system and it
would result in high operating cost; in the form of compres-
sion energy and also, a compressor with high capacity is
400
required to treat the total volume of the gas which would in-
crease capital cost as well. Whereas, the recycle ratio is below
30 35 40 45 50
CO in feed (%) one (Fig. 14 (a)) for PORCHF membrane system and the
compression duty required for this system is one fourth of the
Fig. 12. Two stage membrane separation cascade (PORCHF membrane, T: amount required for two stage polyimide system as shown in
23  C).
Fig. 15 (a). The efficiency of a membrane system increases
10 bar with high selectivity as it can be seen from the recycle ratio
50
8 bar and specific duty plots. The data shows that PORCHF having
6 bar
45
highest selectivity gives lowest recycle ratio and the required
Membrane specific Area (m /Nm feed)

specific duty values. It was observed that the recycle ratio

40 decreases in CHF and PORCHF membranes unlike polyimide
membranes with high CO2 present in the feed. It is very
35 important to choose an optimal point where capital investment
and running cost are low and the system is efficient at the same
30 time. The increasing of feed pressure to 8 bar results in lower
recycle ratio and energy demand. However, for two stage
25 polyimide membranes, recycle ratio is still quite high espe-
cially when 50% CO2 is present in the feed as shown in Fig. 14
20
(b). The Fig. 15 (b) indicates that the specific duty required is
still four times higher for the polyimide membrane system as
15
30 35 40 45 50 compared to PORCHF system. Plasticization effect inhibits
CO in feed (%) polyimide membrane to go to high pressures [27], so
Fig. 13. Three stage membrane separation cascade (Polyimide membrane, T: maximum pressure tested for polyimide membrane systems is
23  C). 10 bar in this section of work, which shows high recycle ratio
in two-stage configuration as shown in Fig. 14 (c). The three-
two-stage PORCHF membrane cascade (Fig. 12). Results stage system with polyimide shows recycle ratio about 1 for
presented in Figs. 9e12 show that change in required mem- 8 bar and 10 bar simulations, and the specific energy demand
brane area under a set of CO2 loadings is significant when feed for three stage polyimide is double as compare to PORCHF
pressure is low (6 bar). However, when pressure is high membrane system.
(10 bar) the trend line looks almost straight for Fig. 9 and the
variation in the area (Figs. 10e12) is not very high. Fig. 13 4.3. Effect of N2
shows the results for a three stage membrane separation
using polymeric membranes. In membranes, the performance It seems impossible to get fuel quality with this poly-
is a trade-off between selectivity and permeability, and pro- imide membrane if up to 2.5% N2 is present in the feed
ductivity is a function of material property and thickness of the biogas, as infinite membrane area would be required to
membrane [54]. Assuming a selective wall thickness of 1 mm separate out 100% CO2, whereas PORCHF membrane can
for polyimide membrane and 20 mm for carbon membranes, it tolerate more N2 due to high CH4/N2 selectivity as compared
was observed that carbon membrane requires >13 times larger to polyimide membrane as shown in Fig. 16. The curve of
area when same operational conditions are applied as for the area and duty in Fig. 16 is expected to be asymptotical,
polymeric membrane. A three-stage polyimide system re- which leads to infinite area and compression duty require-
quires the lowest area, which is three times less than the two- ment in the presence of high N2 percentage in feed biogas.
stage polyimide system. The area against different CO2 The presence of more N2 would result in increased recycle
loadings shows a linear decline for CHF and PORCHF ratio to achieve required fuel (CH4) purity and recovery,
membranes when CO2 increases in the feed, whereas the curve which again leads to high membrane specific area and
is more visible in case of three stage polyimide membrane. compression duty.
 S. Haider et al. / Green Energy & Environment 1 (2016) 222e234 231

feed)
feed)

CO2 in feed (%) CO2 in feed (%)

(c)
feed)

CHF membrane
Polyimide two stage
Polyimide three stage
PORCHF membrane
Polymer with 100 selectivity

CO2 in feed (%)

Fig. 14. Recycle ratio at different CO2 loadings in feed at 23  C, (a) at 6 bar, (b) at 8 bar, (c) at 10 bar.

4.4. Cost calculation membranes and $ 100/m2 for carbon membrane, results show
that membrane area can increase 30%e80% of total invest-
4.4.1. Processing cost ment if optimal pressure is not applied on the upgrading
Processing cost has been considered as the sum of capital plant.
and operating cost to calculate the price of compression duty
and area for membrane systems over 15 years. The Fig. 17 4.4.2. Net Present Value (NPV) calculation
shows the processing cost of compression energy and area The NPV calculation includes pre-treatment, upgrading
per Nm3 of upgraded biomethane for a two-stage polyimide part and high-pressure compression on biomethane to the fuel
membrane system. The effect of big recycle ratio can be seen standard. Considering that different material has specific ad-
in the form of high processing cost here for two-stage poly- vantages, an optimal pressure value has been applied on both
imide system. Even though the membrane cost is quite low, carbon and polymeric materials to calculate the NPV in cost
the cost of energy in capital investment and costs during estimation. Fig. 21 shows NPV at 10 bar pressure for poly-
running time is very high. The required processing cost is imide two-stage and three-stage system. Carbon membranes
reduced to one-fourth by using a three-stage system for pol- have shown a stable performance under different CO2 load-
yimide membrane as shown in Fig. 18. The effect of mem- ings, with no plasticization up to 50 bar [28]. It can be seen
brane efficiency is more prominent in Figs. 19 and 20 with from the NPV results that optimal pressure for carbon mem-
CHF and PORCHF membranes, showing considerably low brane is 50 bar or higher. For NPV calculations, 50% CO2
processing cost related to energy consumption, whereas the concentration in the feed is considered.
effective membrane area required is five-fold higher than for A three-stage polyimide membrane system has been
polymeric membranes. Considering $ 20/m2 for polymeric calculated to have a maximum NPV of $ 9.3 M at 10 bar,
232 S. Haider et al. / Green Energy & Environment 1 (2016) 222e234

(a) 1 (b)
0.6

0.5
0.6

Specific duty (kW/Nm 3 feed)

Specific duty (kW/Nm 3 feed)

0.4

0.4
0.3

0.2 0.2

0.1 0.125
30 35 40 45 50 30 35 40 45 50
CO2 in feed (%) CO2 in feed (%)

(c)
0.5

0.4
Specific duty (kW/Nm 3 feed)

0.3

CHF membrane 0.2

Polyimide two stage
Polyimide three stage
PORCHF membrane
Polymer with 100 selectivity
0.125
30 35 40 45 50
CO2 in feed (%)

Fig. 15. Specific Compression duty at different CO2 loadings in feed at 23  C, (a) at 6 bar, (b) at 8 bar, (c) at 10 bar.

whereas PORCHF membrane system looks competitive with membranes. However, this price can be reduced by optimizing
an NPV of $ 7.4 M. The resulting lower NPV value for the membrane production process (The production process
PORCHF is due to high membrane cost and estimated for a was not fully optimized for the pilot scale production of CHF
lifetime of 5 years in comparison with 7.5 years of polyimide and PORCHF at the company MemfoACT).

0.26 16000 50k

Membrane specific Area Duty in Capital Investment/Nm of biomethane
1100 Specific duty
Area in Capital investment/Nm of biomethane
Duty in Running cost/Nm of biomethane
0.24 14000 Area in Running cost/Nm of biomethane
1000
Membrane specific Area (m /Nm feed)

10k
Duty/Nm of biomethane ($)

Area/Nm of biomethane ($)

Specific duty (kW/Nm feed)

900 0.22 12000

800
0.20 10000
1k
700
0.18 8000
600

0.16 6000
500 100

400 0.14 4000

-.5 0 1 2 0 10 20 30 40 50
N in feed (%) Feed pressure (bar)

Fig. 16. Effect of N2 concentration in feed gas (10 bar, 23  C). Fig. 17. $ for duty & area in processing cost, (two stage polyimide).
 S. Haider et al. / Green Energy & Environment 1 (2016) 222e234 233

3000
10k
PORCHF(50bar)
2850
CHF(50bar)

Area/Nm of biomethane ($)

1k
Duty/Nm of biomethane ($)

2700
Selectivity 100 ( as polymer and at 10bar)

2550
PORCHF(10bar)
100

2400 CHF(10bar)

Three stage Polyimide(10bar)

2250 10

0 10 20 30 40 50 NPV Two stage polyimide (10bar)

Feed pressure (bar)
-96.0M -88.0M -80.0M -72.0M -64.0M -56.0M -48.0M -40.0M -32.0M -24.0M -16.0M -8.0M 0.0 8.0M
Fig. 18. $ for duty & area in processing cost, (three stage polyimide).
NPV (US$ for base case [300 Nm /hr])

Fig. 21. NPV calculated for optimal pressures.

2800
Duty in Capital Investment/Nm of biomethane
Area in Capital investment/Nm of biomethane
2700 Duty in Running cost/Nm of biomethane
1M
NPV
Area in Running cost/Nm of biomethane

2600
Duty/Nm of biomethane ($)

Area/Nm of biomethane ($)

2500 8.0M
NPV (US$ for base case [300Nm /hr])

100k
2400
6.0M
2300

2200
10k 4.0M

2100

2.0M
2000 2.3k
0 4 10 20 30 40 50
Feed pressure (bar)
0.0
Fig. 19. $ for duty & area in processing cost, (two stage CHF).
Membrane cascade (if PORCHF price reduce to 60 $/m )

Fig. 22. NPV comparison of three stage polyimide and PORCHF membrane.
4.4.3. Sensitivity analysis
Assuming an optimized process producing PORCHF at a
price of $ 60/m2 instead of $ 100/m2 and a membrane lifetime applying 70 bar pressure, then it can increase NPV for
of 7.5 years will give NPV for PORCHF of $ 8.8 M. and PORCHF over $ 9 M as shown in Fig. 22.

1M 5. Conclusions
2200
In this study, it was found that the two-stage cascade pro-
cess with recycle using a polyimide membrane was not
2100
100k economically viable for biogas upgrading due to high recycle
Duty/Nm of biomethane ($)

Area/Nm of biomethane ($)

ratio, and thus resulting in high operating cost, whereas the

2000
three stage polyimide membrane system is quite feasible in
order to obtain fuel quality of biomethane. Carbon hollow
1900 fibre membrane and modified carbon hollow fibres produced
10k
on a pilot plant were tested to obtain the same fuel quality in
1800 two stage cascade, and these membranes consumed 22% less
energy as compared to three-stage polyimide system. The
1700 1k drawback of these membranes is, however, the production
0 4 10 20 30 40 50 cost, which is 5-fold higher than the assumed costs of a pol-
Feed pressure (bar)
yimide membrane. The optimization in the production process
Fig. 20. $ for duty & area in processing cost, (two stage PORCHF). and choosing an optimal operating pressure can reduce the
234 S. Haider et al. / Green Energy & Environment 1 (2016) 222e234

capital cost for CHF and PORCHF membranes, whereas the [18] Z. Dai, R.D. Noble, D.L. Gin, X. Zhang, L. Deng, J. Membr. Sci. 497
operating cost can be reduced by increasing the membrane life (2016) 1e20.
[19] A.L. Kohl, R. Nielsen, Gas Purif, fifth ed., Gulf Pub. Co, Houston, Texas,
through regeneration of carbon fibres by applying CVD pro- 1997.
cess if pore clogging is the problem. In the case of fibre [20] J.A. Lie, M.B. H€agg, J. Membr. Sci. 284 (2006) 79e86.
breakage, the broken fibres may be plugged in the module. [21] J. Ahmad, M.B. H€agg, J. Membr. Sci. 427 (2013) 73e84.
Instant boosting with electrical regeneration applying low [22] S. Kim, S.H. Han, Y.M. Lee, J. Membr. Sci. 403e404 (2012) 169e178.
voltage and direct current (DC) has been documented as [23] Y. Zhang, J. Sunarso, S. Liu, R. Wang, Int. J. Greenh. Gas. Control. 12
(2013) 84e107.
successful, but the effect of this regeneration procedure on the [24] J.A. Lie, M.B. Hagg. (US20100162887A1, 2010), chap. US 2010/
aging of the membrane is not sure. 0162887 A1.
[25] L.M. Robeson, J. Membr. Sci. 320 (2008) 390e400.
Conflict of interest [26] L.M. Robeson, J. Membr. Sci. 62 (1991) 165e185.
[27] S. Kanehashi, T. Nakagawa, K. Nagai, X. Duthie, S. Kentish, G. Stevens,
J. Membr. Sci. 298 (2007) 147e155.
There is no conflict of interest in the reported work; not [28] N. Tanihara, H. Shimazaki, Y. Hirayama, S. Nakanishi, T. Yoshinaga,
between the acknowledged researchers, authors or mentioned Y. Kusuki, J. Membr. Sci. 160 (1999) 179e186.
companies. [29] D.Q. Vu, W.J. Koros, S.J. Miller, Ind. Eng. Chem. Res. 41 (2002)
367e380.
[30] A. Makaruk, M. Miltner, M. Harasek, Sep. Purif. Technol. 74 (2010)
Acknowledgements 83e92.
[31] L. Deng, M.B. H€agg, Int. J. Greenh. Gas. Control. 4 (2010) 638e646.
The authors are very grateful to Ms Ingerid Caroline [32] C. Micale, Energy Proced. 82 (2015) 971e977.
[33] J.A. Lie, T. Vassbotn, M.B. Hagg, D. Grainger, T.J. Kim, T. Mejdell, Int.
Tvenning Andersen for excellent laboratory work on the car-
J. Greenh. Gas. Control. 1 (2007) 309e317.
bon hollow fiber membranes and Dr. Muhammad Saeed for [34] D. Grainger, M.-B. H€agg, Fuel 87 (2008) 14e24.
fruitful discussion. The authors would also like to thank The [35] X. He, J. Arvid Lie, E. Sheridan, M.-B. H€agg, Energy Proced. 1 (2009)
Department of Chemical Engineering at NTNU for providing 261e268.
the possibility to work with this article. [36] A. Hussain, M.-B. H€agg, J. Membr. Sci. 359 (2010) 140e148.
[37] B. Belaissaoui, Y. Le Moullec, D. Willson, E. Favre, J. Membr. Sci.
415e416 (2012) 424e434.
References [38] A. Soffer, H. Cohen, US005914434A vol. US005914434A, Carbon
Membr. Ltd., Arava, Israel, 1999.
[1] Z. Song, C. Zhang, G. Yang, Y. Feng, G. Ren, X. Han, Renew. Sustain. [39] X. He, J.A. Lie, E. Sheridan, M.B. H€agg, Ind. Eng. Chem. Res. 50 (2011)
Energy Rev. 33 (2014) 204e213. 2080e2087.
[2] R.H. Williams, Toward zero emissions from coal in China, China Cl. [40] F. Falbo, F. Tasselli, A. Brunetti, E. Drioli, G. Barbieri, Braz. J. Chem.
Energy Forum, Beijing, 2001. Eng. 31 (2014) 1023e1034.
[3] M. SoltaniehS, K. Thambimuthu, J.C. Abanades, Special Report on CO2 [41] M. Scholz, T. Melin, M. Wessling, Renew. Sustain. Energy Rev. 17
Capture and Storage, (IPCC), 2005. (2013) 199e212.
[4] H. Yang, Z. Xu, M. Fan, R. Gupta, R.B. Slimane, A.E. Bland, I. Wright, [42] M. Harasimowicz, P. Orluk, G. Zakrzewska-Trznadel,
J. Environ. Sci. 20 (2008) 14e27. A.G. Chmielewski, J. Hazard. Mater. 144 (2007) 698e702.
[5] Anon, Swedish Case Studies, Swedish Gas Association, 2008. [43] P. Weiland, Appl. Microbiol. Biot. 85 (2010) 849e860.
[6] S. Rasi, A. Veijanen, J. Rintala, Energy 32 (2007) 1375e1380. [44] J.D. Wind, D.R. Paul, W.J. Koros, J. Membr. Sci. 228 (2004) 227e236.
[7] J.B. Holm-Nielsen, T. Al Seadi, P. Oleskowicz-Popiel, Biores. Tech. 100 [45] S.P. Kaldis, G. Skodras, G.P. Sakellaropoulos, Fuel Process. Technol. 85
(2009) 5478e5484. (2004) 337e346.
[8] S.S. Hosseini, N. Peng, T.S. Chung, J. Membr. Sci. 349 (2010) 156e166. [46] M. Ungerank, G. Baumgarten, M. Priske, H. Roegl, US20130098242A1
[9] L. Yingjian, Q. Qi, H. Xiangzhu, L. Jiezhi, Sustain. Energy Technol. vol. US 20130098242A1, Evonik Fibres Gmbh, 2013.
Assess. 6 (2014) 25e33. [47] M. Mulder, Basic Principles of Membrane Technology, Springer,
[10] M. Scholz, M. Alders, T. Lohaus, M. Wessling, J. Membr. Sci. 474 Netherlands, 1996.
(2015) 1e10. [48] W.J. Koros, Membrane opportunities and challenges for large capacity
[11] A. Molino, M. Miglion, B. Bikson, G. Giordano, G. Braccio, Fuel 107 gas andvapour feeds, in: European Membrane Society's 20th Summer
(2013) 585e592. School, NTNU, Trondheim, 2003.
[12] L. Yang, X. Ge, C. Wan, F. Yu, Y. Li, Renew. Sustain. Energy Rev. 40 [49] X. He, C. Fu, M.-B. H€agg, Chem. Eng. J. 268 (2015) 1e9.
(2014) 1133e1152. [50] A. Lindbråthen, D.R. Grainger, M.B. H€agg, Sep. Sci. Technol. 42 (2007)
[13] E. Ryckebosch, M. Drouillon, H. Vervaeren, Biomass Bioenerg. 35 3049e3070.
(2011) 1633e1645. [51] M.S. Peters, K.D. Timmerhaus, Plant Design and Economics for
[14] A. Wellinger, A. Lindberg, Biog. Upgrad. Utilisat., 2005 [Internet] IEA Chemical Engineers, McGraw-Hill, New York, 1991.
Bioenerg. Task 24. [52] W.D. Baasel, Preliminary Chemical Engineering Plant Design, Van
[15] D.A. Ken Krich, J.P. Batmale, J. Benemann, B. Rutledge, D. Salour, Nostrand Reinhold, New York, 1990.
Natural Gas in California, Clear Concepts, California, 2005, pp. 47e69. [53] D.J. Stookey, C.J. Patton, G.L. Malcolm, Effects of Separation Factor and
[16] B. Rutledge, White Paper, California Biog. Ind. Assess, vol. 38, West- Pressure Ratio on Permeate Purity (Membranes Separate Gases Selec-
Start-Calstart, Pasadena, USA, 2005. tivity), Chemical Engineering Progress, 1986.
[17] R. Baker, Membr. Tech. (2001) 5e10. [54] W.J. Koros, R. Mahajan, J. Membr. Sci. 175 (2000) 181e196.

View publication stats