Beruflich Dokumente
Kultur Dokumente
a r t i c l e i n f o
a b s t r a c t
Article history:
Received 20 September 2012
Received in revised form
9 November 2012
Accepted 25 November 2012
Available online 7 December 2012
Conventional hollow ber models in process simulators usually assume constant membrane permeance
i.e., independent of pressure and temperature. In this work, hollow ber membrane model has been
proposed to cater the effects of temperature and pressure on membrane permeance. The proposed model is
incorporated with Aspen HYSYS as a user dened unit operation in order to study the performance of gas
separation system. The simulated model is validated by experimental and published data. The temperature
drop due to Joule Thomson effect and its contribution to the change in membrane permeance has also been
investigated. Similarly, the effect of pressure on membrane permeance has been studied. The inuence of
these effects on the separation performance and process economics has been investigated for the
separation of CO2 from natural gas. The proposed hollow ber membrane model has potential to be
applied for design, optimization and scale up of wide range of gas separation systems.
& 2012 Elsevier B.V. All rights reserved.
Keywords:
Gas separation
Process simulation
Hollow ber membrane
Joule thomson effect
Membrane permeance
1. Introduction
Membrane based gas separation is an important unit operation
in many industrial separations such as pharmaceutical, biotechnology, petrochemical and gas processing [1,2]. Hollow ber
membrane module is attracting wide range of applications due
to its convincing capabilities, e.g., high efciency/volume, lower
energy requirement, chemical-free operation, etc [3].
The modeling of hollow ber membrane separation has been
studied by number of investigators since the development of rst
mathematical model for membrane gas separation by Weller and
Steiner [4]. Thorman et al. [5] incorporated the effect of pressure
drop in a study on the binary mixtures separations employing
silicone rubber capillaries. A model for practical representation of
gas separation using high ux, asymmetric hollow ber membrane has been presented by Pan [6].
A new approach has been presented and analyzed by Thundyil
and Koros [7] in order to solve the mass transfer problem posed
by the permeation process in hollow ber membrane separator.
Rautenbach et al. [8] studied a variation in ber properties that
affects the performance of defect free hollow ber membrane
modules for air separation. Similarly, Lemanski and Lipscomb [9]
presented a theoretical and experimental study of the effects of
variable ber properties on countercurrent hollow ber gas
separation module performance.
2. Mathematical method
2.1. Mathematical model
The radial cross ow model (shell side feed) for hollow ber
membrane module is schematically represented in Fig. 1. The
bundle of bers is sealed at one end using epoxy The other end of
ber bundle is kept open to allow the ow of gases. The ber
bundle is housed in the middle of shell. Feed gas is introduced in
the system from the shell side that ows radially inward perpendicular to the bers toward the centre. The permeate into the
bers ows axially along to the permeate collector. As a result,
45
Shell side
Feed
Retentate
Permeate
j=1
j=M
i=1
i=N
Tube sheet
III
II
IV
Retentate End
Sealed end
sheet
Permeate
End
II
IV
III
aij
Pi
Pj
For a binary gas mixture, the local permeation rate at any point
in the stage over a differential membrane area, dAm is as follows
ydV
P1
ph xpl y dAm
1ydV
P2
ph 1xpl 1y dAm
1y
1x pl =ph 1y
46
fraction of the cross section area of all bers over the cross
section area of the module.
nf
Af
do
nf
Am
dm
15
Q si,j Q si1, j DQ
16
Q T i,j Q T i1, j DQ
17
A
41 e
do
V
The model will consider the specic bundle area, and divide it
up into a predetermined number of elements (index i in the
radial direction and index j in the axial direction). The nite
element is assumed to exist at radius r from the centre of the
bundle having radius of Dr and length of Dz. The membrane
separation area of each nite element is determined by it volume
(2prDrDz) and specic area of membrane bundle (membrane
bundle area/unit volume of membrane bundle) as follows.
A 2pDrDz1=do
In the case of radial cross ow, there are four types of elements
depending upon their radial and axial location in the tube side
(ber bundle) of the module as shown in Fig. 1a The computation
proceeds from the Type 1 to Type 3 elements till the end of
bundle and then starts again from Type 2 elements through the
succession of Type 4 elements [7].
For a binary system, the permeate composition, y1 (faster
permeating component) in the rst and second types (Type
1 and Type 2) of nite elements is given in terms of mole fraction
of the shell side, x1 (faster permeating component) as follows.
y1
x1 i,j
18
y1 i,j
19
For elements in contact with feed (Type 3), the sufxes (i 1, 1)
are replaced with feed conditions such as Qf and xf. Solving the
above set of equations, mass transport across the membrane for the
each element can be computed. Using same approach for all the
elements, computation proceeds from the epoxy sealed end of the
hollow ber tubes to the tube-sheet end of these bers [7].
In addition, viscosity of gas mixture is calculated by Wilkes
method while viscosity of pure components and their temperature dependence are determined using Lucas method [29]. The
assumptions that follow the suggested model making it applicable only for ideal conditions are:
1. It holds only for the binary gas mixture. Even though it is an
ideal assumption, yet it is a rst step to understand realistic
modeling and simulation of many important separations
[6,7,10,11,19,21,30,31] such as CO2-methane separation in
the current work.
2. The shell side pressure variations are negligible (due to
constant bulk ow in an axial direction) while the permeate
side pressure drop is determined by HagenPoiseuille
[6,7,10,17,19,21].
3. The system operates at isothermal conditions [6,7,10,14,17,
19,21,31]
4. Membrane permeability is assumed to be independent of the
temperature and pressure [3,6,7,10,11,14,19,21,30]
9
where a is the selectivity of membrane and b is the pressure ratio
of higher pressure side to lower pressure side. The ow rate into
the nite element of the permeate side is given by
DQ P 1 =d ph x1 i1, 1pl y1 i,1 P2 =d ph: x2 i1,1
pl y2 i,1:A
10
11
Q T i,1 DQ
12
13
For elements in contact with feed (Type 1), the sufxes (i1, 1)
are replaced with feed conditions such as Qf and xf. These elements
will not have any preceding elements in the radial direction [7].
For Type 3 and Type 4 elements, shell and tube ow rates and
compositions are known and mass transport is measured by
solving the following equations
DQ P1 =d ph x1 i1, jpl y1 i,j P2 =d ph: x2 i1,jpl y2 i,j:A
14
0
C PI Rg T T00 2
Cp
C m,p
M
M
21
Dnn bn cn kn rr kn rr bn e
18
X
Z 1 Brm rr
C nn
n 13
Z 0 BK 3
18
X
k
r n
58
X
23
C nn bn cn kn rr kn rr bn ecnrr kn
24
n 13
C nn
Z 00 B0 K 3
58
X
C nn0
26
n 13
27
where, rm is gas mixture molar density, rr is reduced density, B is
second virial coefcient, kn and bn are equation of state parameters and Cnn is temperaturecomposition dependent coefcient. Therefore, the nal analytical equation for JouleThomson
coefcient is given by
mJT
RT 2
M
CPI RT T00 20
0
P
n n
RTZ2 58
n 13 C n Dn rZ TZ0 Z 1
r
n 13
Z1 Z0
where Po is the pre-exponential factor independent of temperature and Ep is the activation energy for permeation.
Safari et al. [39] examined different forms of equations based
on partial immobilization and Arrhenius models and presented
the following form that includes pressure and temperature effect
simultaneously.
cexp d=Rg T
b
P aexp
32
Rg T
1 e=T p
25
n 13
18
X
47
RTZ2 rTZ 1
28
48
TPI TFI SC
$ 5/ft3
$ 8650 *(Wcp/Zcp)0.82
MC CC
1.12 * FC
0.20 * BPC
BPCPC
0.10 * VOM
0.05 * TFI
0.015 * TFI
$ 15/h
1.15 * DL
$ 3/ft2 of membrane
$ 0.07/kw h
CH4LS NGLS * NHV * NWP
NGLS 365 * OSF * Lf * yP(CH4) * xf (CH4)
3. Experimental method
Mathematical models have to be supported by experimental
data. Thus, the proposed model is veried by experimental
method through comparison of the simulated and experiments
results. The experimental set up mainly consists of hollow ber
membrane module and the gas separation testing unit in which
the module is installed to facilitate the separation mechanism.
3.1. Hollow ber membrane module
Hollow ber modules may have different congurations to
meet the needs of different applications. The current work used
cross ow (shell-side feeding) hollow ber module as shown in
Fig. 1. In this conguration, two tube sheets hold the ber ends in
place and separate the retentate from the permeate ow. One is a
plug-sealed tube sheet in which the openings of ber ends are
blocked by the epoxy resin; the other is an open-end tube sheet in
which the bores of hollow bers are exposed [27].
The hollow bers used for the experimental work are commercial (Alpha Membrane Hi-Tech Pte. Ltd, Singapore). The
material of membrane used is polyimide having the permeance
of 22 GPU for CO2 and 0.7 GPU for CH4 at 50 1C. The bers have
outer diameter of 400 mm and inner diameter of 180 mm. The
Fig. 2. Process ow diagram (PFD) of double stage system with permeate recycle in Aspen HYSYS.
49
T
F
Pressure guage
Flow meter
Flow meter
F
Thermocouple
Hollow Fiber
Membrane Module
Flow meter
Static mixer
Flow
controller
Feed Vessel
Natural gas
CH4
CO2
Infrared Analyzer
N2
Fig. 3. Flow sheet of gas separation testing unit for experimental validation.
Data Acquisition
System (Computer)
50
50
45
40
45
40
35
Experimental
30
25
20
40
60
40 % CO2
30
70 % CO2
25
20
15
10
5
20
0
10 % CO2
35
80
10
20
40
50
60
Fig. 6. Effect of stage cut on temperature drop (Feed temperature 323 K).
45
25
40
0.95
30
Stage Cut
0.9
0.85
0.8
0.75
0.7
0.4
Experimental data by
Pan [10]
Simulation results by
suggested model
20
35
30
15
25
20
10
15
Temperature drop
10
Permeance of CO2
0
0.45
0.5
5
0.55
0.6
Stage Cut
Fig. 5. Model Validation with published literature by Pan [10].
0.65
10
20
30
40
50
0
60
Stage Cut
Fig. 7. Effect of stage Cut on CO2 permeance of for 70% CO2 feed.
0.9
40
0.8
35
0.7
30
0.6
25
0.5
20
0.4
0.3
15
10
5
0
Temperature drop
0.2
Permeance of CH4
0.1
20
30
40
50
3.5
3
Non-ideal model
(variable permeance)
2.5
2
1.5
1
0.5
0
10
51
4.5
Retentate CO2
Composition (%)
45
60
100
Stage Cut
200
300
400
Fig. 8. Effect of stage cut on CH4 permeance of for 70% CO2 feed.
25
20
0.5
15
0.45
10
0.4
CO2
5
0.35
CH4
0.55
0.6
25
20
15
10
Non-ideal model
(variable permeance)
0
0
100
200
300
400
0.3
0
0
10
20
30
40
50
60
Pressure Ratio
0.35
0.3
Fig. 9. Effect of pressure ratio on membrane permeance of CO2 and CH4 for 70%
CO2 feed.
Stage Cut
[49,50]. Fig. 9 shows the effect of pressure ratio on the CO2 and
CH4 permeance of the membrane material for the higher CO2
concentration feed. It can be seen that permeance value increases
with the increase in pressure ratio and vice versa. Thus, pressure
ratio is an important factor to determine the non ideal permeance
of the separation system.
0.25
0.2
0.15
0.1
Non-ideal model
(variable permeance)
0.05
0
0
100
200
300
400
membrane separation- area and vice versa [7]. Fig. 10(b) presents the
methane loss as a function of ber length for both ideal and real
scenario. Methane loss increases with the increase in ber length as
expected but it is less pronounced in non ideal case (with maximum
difference of 6%) where permeance of both CO2 and CH4 decreases
with the temperature and pressure drop.
Fig. 10(c) shows the effect of ber length on stage cut. Stage
cut increases with the increase in ber length. For the same ber
length, non ideal case shows lower stage cut in comparison with
ideal case with maximum difference of 4%. It is due to the reason
that ow rate of permeate stream decreases due to decrease in
permeance of CO2 and CH4. As a result, stage cut being ratio of
permeate ow rate to feed ow rate, gets reduced for the non
ideal case. The results are consistent with those obtained by Safari
et al. [39] and Schloz et al. [25].
All these results are repeated for medium CO2 feed concentration
of 40% as shown in Fig. 11 and higher concentration of 70% as shown
in Fig. 12. It can be seen that the performances at the higher feed
concentration is an amplication of the results obtained at the lower
feed concentration. For example, the maximum difference between
ideal and non ideal cases for retentate/product CO2 composition is
52
Retentate CO2
Composition (%)
15
10
100
200
300
400
35
20
25
Non-ideal model
(variable permeance)
20
15
10
5
0
30
100
200
300
400
30
50
25
20
15
10
35
100
200
300
30
20
Non-ideal model
(variable permeance)
10
0
0
40
400
0
100
200
300
400
0.25
0.2
0.15
0.1
100
200
300
400
Stage Cut
0.45
0.3
0.4
0.35
0.3
Non-ideal model
(variable permeance)
0.25
0.2
100
200
300
400
increased to 10% (for 40% CO2 concentration feed) and 15% (for 70%
CO2 concentration feed). More concentration of CO2 in feed causes
more temperature drop due to JT effect that leads to decrease in
membrane permeance. As a result, retentate CO2 composition tends
to increase while stage cut and methane loss tend to decrease
further for non ideal model.
4.4. Process economics study
In order to evaluate the process economics of the system, feed
gas composition of CO2 is xed at 40% while methane purity at
96% (4% CO2 in retentate stream) and the effect of ber length on
the compressor power requirement and gas processing cost has
been investigated both for the ideal and non ideal case of
membrane permeance. The simulations conditions are same as
mentioned in the Section 2.4 (Feed pressure59.6 bar, permeate
pressure1.8 bar, feed temperature323 K and feed ow
rate1.3 MMSCF). The compressor efciency of 80% is used as
shown in Table 1.
The effect of membrane module characteristics on the compressor power requirement has been investigated for the proposed design conguration (double stage with permeate recycle)
Fig. 12. Comparison of (a) retentate CO2 composition (b) Methane loss (c) stage
cut by ideal model (constant permeance) with non ideal model (temperature and
pressure dependent permeance) for 70% CO2 feed.
ph
Q p, n ln
n1:341
pl,n
n1
2
X
33
300
350
250
300
200
150
100
50
250
200
100
200
300
400
150
100
50
0
0
500
100
200
300
400
500
Fiber length(cm)
Fig. 13. Comparison of compressor power by ideal model (constant permeance)
with non ideal model (temperature and pressure dependent permeance) at feed
CO2 composition 40% and methane purity 96%.
0.07
0.08
0.06
53
0.05
0.04
0.03
0.02
0.01
0
0
100
200
300
400
500
0.07
0.06
0.05
0.04
0.03
0.02
0.01
0
0
Fig. 14. Comparison of gas processing cost by ideal model (constant permeance)
with non ideal model (temperature and pressure dependent permeance) at feed
CO2 composition 40% and methane purity 96%.
non ideal case at the same ber length is 6%. It is explained by the
reason that compressor power mainly depends on the permeate
ow rate and thus stage cut. With the decrease in stage cut due to
decrease of membrane permeance for gases, there will be low
requirement of compressor power in non-ideal case where
membrane permeance is dependent on temperature and pressure.
The effect of ber length on gas processing cost (GPC) for both
ideal and non ideal cases is shown in Fig. 14. It can be noted that
larger ber lengths yield higher GPC. It is due to the reason that GPC
mainly depends on membrane module cost (MC), compressor cost
(CC) and annual cost of CH4 lost in permeate (CH4LS) as shown in
Table 1. With the increase in membrane area (or ber length), the
above mentioned costs increase resulting in the increase of GPC.
It can also be observed that non-ideal case show less GPC in
comparison with ideal case (maximum difference of 8%) for the
same ber length. As stage cut and a compressor power requirement
is decreasing with the decrease in membrane permeance, it leads to
reduce the GPC of the membrane separation system.
In order to see the effect of higher concentration CO2 feed on
process economics, the above procedure is repeated for 70% CO2
in feed gas as shown in Fig. 15 (for compressor power) and Fig. 16
(for gas processing cost). It can be observed that the difference
between performance of ideal and non ideal models are more
pronounced in the case of higher CO2 composition feed. For
example, the maximum difference between ideal and non ideal
cases for gas processing cost is increased from 8% (for 40% CO2
concentration feed) to 15% (for 70% CO2 concentration feed). It is
due to the reason that higher CO2 in feed leads to increase in
temperature drop. As a result, membrane permeance reduces
while contributing to further decrease in compressor power and
gas processing cost.
100
200
300
400
500
Fig. 16. Comparison of gas processing cost by ideal model (constant permeance)
with non ideal model (temperature and pressure dependent permeance) at feed
CO2 composition 70% and methane purity 96%.
5. Conclusions
The temperature and pressure dependence of membrane permeance has been investigated for the case involving CO2 separation
from methane. The effect of variable permeance is included in the
cross ow model for hollow ber membrane separation. The model
is then included in the process simulation (Aspen HYSYS) as a user
dened unit operation along with other available unit operations in
order to investigate the membrane performance and process economics. The simulated model is validated with experimental published data where the simulated data exhibit good agreement with
the experimental and published results. The temperature drop due
to Joule Thomson (JT) cooling and its effect on membrane
permeance of both CO2 and CH4 has been studied. Similarly, the
effect of pressure ratio on membrane permeance has been reported.
The inuence of variable permeance (temperature and pressure
dependent) is studied by comparing separation performance (in
terms of retentate/product CO2 composition, methane loss and stage
cut) and process economics (gas processing cost) with the ideal
model where membrane permeance is assumed independent of
temperature and pressure. It has been shown that non ideal model
shows higher CO2 retentate composition, lower stage cut and
methane loss in comparison to ideal model for the same ber
length. It leads to lower compressor power requirements and gas
processing cost for the non ideal hollow ber membrane model. For
highly non ideal conditions (e.g., higher CO2 concentration feed),
non ideal effects related to module operation would be more
signicant in affecting the product quality (CO2 retentate/product
composition), methane loss, stage cut, compressor power and gas
processing cost of hollow ber separation systems. Thus, it is crucial
54
TPI
UC
UCP
VOM
List of symbols
Wcp
x1
x2
y1
y2
Z
Greek symbols
aij
b
e
^
DQ
Dr
Dz
rm
rr
Zcp
mJT
Acknowledgements
This work was done with the nancial and technical support
from CO2 Management (MOR) research group, Universiti Teknologi PETRONAS.
References
[1] R.W. Baker, Membrane Technology and Applications, Second ed., John Wiley
& Sons, Chichester, 2004.
[2] E. Drioli, L. Giorno, Membrane Operations, Wiley-VCH Verlag GmbH & Co.
KGaA, Weinheim, Germany, 2009.
[3] T. Katoh, M. Tokumura, H. Yoshikawa, Y. Kawase, Dynamic simulation of
multicomponent gas separation by hollow-ber membrane module: nonideal
mixing ows in permeate and residue sides using the tanks-in-series model,
Sep. Purif. Technol. 76 (2011) 362372.
[4] S. Weller, W.A. Steiner, Engineering aspects of separation of gases. Fractional
permeation through membranes, Chem, Eng. Prog. 46 (1950) 585590.
[5] J.M. Thorman, H. Rhim, S.T. Hwang, Gas separation by diffusion through
silicone rubber capillaries, Chem. Eng. Sci. 30 (1975) 751754.
[6] C.Y. Pan, Gas separation by high-ux, asymmetric hollow-ber membrane,
AlChE J. 32 (1986) 20202027.
[7] M.J. Thundyil, W.J. Koros, Mathematical modeling of gas separation
permeatorsfor radial crossow, countercurrent, and cocurrent hollow ber
membrane modules, J. Membr. Sci. 125 (1997) 275291.
[8] R. Rautenbach, A. Struck, M.F.M. Roks, A variation in ber properties affects
the performance of defect-free hollow ber membrane modules for air
separation, J. Membr. Sci. 150 (1998) 3141.
[9] J. Lemanski, G.G. Lipscomb, Effect of ber variation on the performance of
countercurrent hollow ber gas separation modules, J. Membr. Sci. 167
(2000) 241252.
[10] S. Zhao, Z. Li, Y. Liu, L.e. Wang, Simulation of binary gas separation in hollow
ber membrane-acetylene dehydration, Desalination 233 (2008) 310318.
[11] H. Jin, S. Han, Y. Lee, Y. Yeo, Modeling and control of CO<sub>2</
sub> separation process with hollow ber membrane modules, Korean J.
Chem. Eng. 28 (2011) 4148.
[12] R. Khalilpour, A. Abbas, Z. Lai, I. Pinnau, Analysis of Hollow Fibre Membrane
Systems for Multicomponent Gas Separation, Chemical Engineering Research
and Design.
[13] V. Soni, J. Abildskov, G. Jonsson, R. Gani, A general model for membranebased separation processes, Computers &, Chem. Eng. 33 (2009)
644659.
55
[35] R. Wang, S.S. Chan, Y. Liu, T.S. Chung, Gas transport properties of poly(1,5naphthalene-2,20 -bis(3,4-phthalic) hexauoropropane) diimide (6FDA-1,5-NDA)
dense membranes, J. Membr. Sci. 199 (2002) 191202.
[36] T.-S. Chung, C. Cao, R. Wang, Pressure and temperature dependence of the
gas-transport properties of dense poly[2,6-toluene-2,2-bis(3,4-dicarboxylphenyl)hexauoropropane diimide] membranes, J. Polym. Sci., Part B: Polym.
Phys. 42 (2004) 354364.
[37] D.R. Paul, W.J. Koros, Effect of partially immobilizing sorption on permeability and the diffusion time lag, J. Polym. Sci., Part B: Polym. Phys. 14 (1976)
675685.
[38] J.P. Wauquier, PETROLEUM REFINING V.2: Separation Processes, Editions
Technip, 2000.
[39] M. Safari, A. Ghanizadeh, M.M. Montazer-Rahmati, Optimization of
membrane-based CO2-removal from natural gas using simple models
considering both pressure and temperature effects, Int. J. Greenhouse Gas
Control 3 (2009) 310.
[40] J. Hao, P.A. Rice, S.A. Stern, Upgrading low-quality natural gas with H2S- and
CO2-selective polymer membranes: Part I. Process design and economics of
membrane stages without recycle streams, J. Membr. Sci. 209 (2002)
177206.
[41] B.D. Bhide, S.A. Stern, Membrane processes for the removal of acid gases from
natural gas. I. Process congurations and optimization of operating conditions, J. Membr. Sci. 81 (1993) 209237.
[42] R.E. Babcock, R.W. Spillman, C.S. Goddin, T.E. Cooley, Natural gas cleanup: a
comparison of membrane and amine treatment processes, Energy Prog. 8
(1988) 135142.
[43] R.W. Spillman, M.G. Barrett, T.E. Cooley, Gas Membrane Process Optimization, In AIChE National Meeting, New Orleans, 1988.
[44] K. Haraya, K. Obata, T. Hakuta, H. Yoshitome, The permeation of gases
through a new type polyimide membrane, Maku (Memb.) 11 (1986) 4852.
[45] IEA, CO2 Capture and Storage, A Key Carbon Abatement Option, International
Energy Agency (IEA), Organization for Economic Co-operation and Development, OECD, France, 2008.
[46] L.S. Tan, K.K. Lau, M.A. Bustam, A.M. Shariff, Removal of high concentration
CO2 from natural gas at elevated pressure via absorption process in packed
column, J. Nat. Gas Chem. 21 (2012) 710.
[47] N.H. Darman, A.R.B. Harun, Technical challenges and solutions on natural gas
development in Malaysia. in: The petroleum Policy and Management (PPM)
Project 4th Workshop of the ChinaSichuan Basin Case Study, Beijing, 2006.
[48] H. Huang, K. Schwab, J.G. Jacangelo, Development of a robust bench-scale
testing unit for low-pressure membranes used in water treatment,
Membr.Water Treat. 2 (2011) 121136.
[49] T. Visser, N. Masetto, M. Wessling, Materials dependence of mixed gas
plasticization behavior in asymmetric membranes, J. Membr. Sci. 306
(2007) 1628.
[50] T. Visser, G.H. Koops, M. Wessling, On the subtle balance between competitive sorption and plasticization effects in asymmetric hollow ber gas
separation membranes, J. Membr. Sci. 252 (2005) 265277.
[51] M.S. Peters, K.D. Timmerhaus, Plant Design and Economics for Chemical
Engineers, McGraw-Hill, 1991.