Beruflich Dokumente
Kultur Dokumente
Process Systems Design and Control Laboratory, School of Chemical Engineering, Yeungnam University, Republic of Korea
Chemical Engineering Department, Universiti Teknologi PETRONAS, Bandar Sri Iskandar, 31750 Perak, Malaysia
Department of Chemical Engineering, COMSATS Institute of Information Technology, Lahore, Pakistan
A R T I C L E I N F O
Article history:
Received 30 April 2014
Received in revised form 28 May 2014
Accepted 29 May 2014
Available online 5 June 2014
Keywords:
Hollow ber
Experimental validation
Gas separation
Membrane process
Process simulation
A B S T R A C T
Conceptual process simulations and optimization are essential in the design, operation and
troubleshooting stages of a membrane-based gas separation system. Despite this, there are few
mathematical models/tools associated with a hollow ber membrane module available in a commercial
process simulator. A mathematical model dealing with the hollow ber module characteristics that can
be included within a commercial process simulator is needed to examine the performance and
economics of a gas separation system. In this study, a hollow ber membrane model was incorporated in
Aspen HYSYS as a user dened unit operation for the study of carbon dioxide separation from methane.
The hollow ber membrane model was validated experimentally. The study of a double stage membrane
module with a permeate recycle, which was proposed to be the optimal conguration in previous
studies, was extended to consider the effects of the module characteristics (such as the ber length,
radius of the ber bundle, diameter of the bers, and porosity) on the process performance and
economics. The gas processing cost (GPC) increased with increasing ber length and bundle radius, and
decreased with increasing outer diameter of the bers and porosity. At the same time, the separation
efciency (product quality) was also dependent on these module parameters. Therefore, the tradeoff for
the hollow ber membrane module characteristics needs to be determined based on the minimum GPC
with respect to the desired product purity.
2014 The Korean Society of Industrial and Engineering Chemistry. Published by Elsevier B.V. All rights
reserved.
Introduction
Membrane separation processes have been used successfully in
the industrial separation of gases because of their many advantages,
such as environmental friendly, favorable economics, ease of
operation, and low maintenance, over conventional separation
techniques [13]. In general, membrane separation units use spiralwound, capillary-ber and hollow-ber modules because of the
large membrane area packed in a small volume [4,5].
The hollow-ber membrane module, which has wide applications, consists of a large number of membrane bers packed in
module shell housing. Typically, the feed gas at higher pressure
enters the shell side of the module at one end, and the more
permeable components are withdrawn from the inside of the bers
through openings on the ber tube sheet [6,7].
In many areas of chemical process design and operation, the
application of mathematical models for process synthesis, optimization and control studies has signicant advantages [8]. Similarly,
the mathematical modeling of membrane separation systems is an
essential and integral aspect for better understanding the effects of
various process parameters and ow patterns on the membrane
performance [9]. When evaluating the performance of hollow ber
membrane modules for gas separation, mathematical modeling
offers the advantages of cost effectiveness, safety and exibility to
extensive parametric studies compared to pilot plant testing.
Weller and Steiner developed the rst mathematical model
for membrane gas separation dealing with binary component
gas mixtures using a non-porous membrane [10]. Pan presented
a model for the practical representation of gas separation using
a high ux, asymmetric hollow ber membrane [11]. The
http://dx.doi.org/10.1016/j.jiec.2014.05.041
1226-086X/ 2014 The Korean Society of Industrial and Engineering Chemistry. Published by Elsevier B.V. All rights reserved.
Nomenclature
v
2
A
Am
Af
BPC
CC
CH4LS
CMC
Wcp
x1
x2
y1
y2
1247
Greek symbols
selectivity of the membrane
pressure ratio
porosity of the membrane (%)
packing density (%)
molar permeation into an element (in.3(STP)/s)
radial increments
axial increments
membrane thickness (in.)
compressor efciency (%)
?cp
a
b
e
K
DQ
Dr
Dz
d
1248
[(Fig._1)TD$IG]
Pi
ph xi pl yi
(1)
where Ji is the ux of the gas component, ph and pl are the feed and
permeate side pressures, respectively, xi and yi are the fractions of
component i on the feed and permeate sides, respectively, and d is
the membrane thickness.
For a binary gas mixture, the local permeation rate at any point
in the stage over a differential membrane area, dAm, can be
expressed as
ydV
P1
ph x pl ydAm
1 ydV
P2
ph 1 x pl 1 ydAm
(2)
(3)
1 y 1 x pl = ph 1 y
(4)
ai j
P1
P2
(5)
[(Fig._2)TD$IG]
1249
Nf
1e
p=4d2o
(7)
A
1 e4
V
do
(8)
[(Fig._3)TD$IG]
QS (u-1, v)
(a) Dimensions of the bers bundle, such as the ber length, L, and
radius of ber bundle, R.
(b) Inner and outer diameter of the bers, di and do.
(c) Measure the packing density and porosity.
The packing density of the hollow ber membrane module is
dened as the fraction of the cross section area of all bers over the
cross section area of the module [38]:
nf
Af
do
nf
Am
dm
(9)
QT (u,v-1)
QS (u, v)
Element (u,V)
QT (u,v)
QS (u-1,v)
(6)
1250
the epoxy sealed end of the hollow ber tubes to the tube-sheet
end of these bers.
Type II) of the nite elements is given in terms of the mole fraction
of the shell side, x1 (faster permeating component), as follows [15]
2
y1
0:5
(10)
where a is the selectivity of the 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 the
following equation:
P1
P2
DQ
ph x1i1; j pl y1i; j
ph x2i1; j pl y2i; j A
(11)
where x2 = 1 x1 and y2 = 1 y1 (binary component gas
mixture).
Similarly, the shell side ow rate, Qs(u, 1), and permeate side
ow rates, QT(u, 1) contacting the next element are given by
Q s u; 1 Q s u 1; 1 DQ
(12)
Q T u; 1 DQ
(13)
P1
ph x1 u 1; v pl y1 u; v A
(14)
For elements in contact with the feed (Type I), the sufxes
(u 1, 1) are replaced with the feed conditions, such as Qf and xf.
These elements will not have any preceding elements in the radial
direction.
For Type III and Type IV elements, the shell and tube ow rates
and compositions are known and mass transport is measured by
solving the following equations [15]:
DQ Pd1 ph x1 u 1; v pl y1 u; v
(15)
Pd2 ph: x2 u 1; v pl y2 u; v A
DQ 1
(16)
Q s u; v Q s u 1; v DQ
(17)
Q T u; v Q T u 1; v DQ
(18)
x1 u; v
Q s u 1; vx1 u 1; v DQ 1
Q s u; v
(19)
y1 u; v
Q T u 1; vy1 u 1; v DQ 1
Q T u; v
(20)
W c p h p Rg T
2
X
Q p;n ln
n1
ph
pl;n
!
1:341
(21)
1251
Table 1
Economics parameters for gas processing cost [41].
Total plant investment (TPI):
Membrane module cost (MC)
Installed compressor cost (CC)
Fixed cost (FC)
Base plant cost (BPC)
Project contingency (PC)
Total facilities investment (TFI)
Start up cost (SC)
Annual variable operating and maintenance cost (VOM):
Contract & material maintenance cost (CMC)
Local taxes and insurance (LTI)
Direct labor cost (DL)
Labor overhead cost (LOC)
Membrane replacement costs (MRC)
Utility cost (UC)
Annual cost of CH4 lost in permeate (CH4LS):
Annual natural gas lost (NGLS)
Gas processing cost (GPC)
Annual capital related cost (CRC)
Membrane life (t)
Wellhead price of crude natural gas
Heating value of natural gas
On stream factor (OSF)
Compressor efciency (hcp)
TPI = TFI + SC
$5/ft3
$8650 * (Wcp/hcp)0.82
MC + CC
1.12 * FC
0.20 * BPC
BPC + PC
0.10 * VOM
VOM = CMC + LTI + DL + LOC + MRC + UC
0.05 * TFI
0.015 * TFI
$15/h
1.15 * DL
$3/ft2 of membrane
$0.07/kwh
CH4LS = NGLS * NHV * NWP
NGLS 365OSFL f yPCH4 x f CH4
GPC = (CRC + CH4LS + VOM)/[365 * OSF * Lf * (1 SCE) * 1000
0.2 * TPI
4 years
$2/MMBTU
1066.8 MMBTU/MMSCF
96%
0.8
Experimental method
Mathematical models must be supported and validated by
experimental data. Therefore, the proposed model was veried by
a comparison of the simulated and experimental results. The
experimental set up consisted of a hollow ber membrane module
and a gas separation testing unit, where the module is installed, to
evaluate the performance of the membrane.
Module manufacture
The early designs of hollow ber modules were derived from the
patents of Dow [48] and DuPont [49]. In design from Dow, modules
were formed by potting the ends of the individual bers or a ber tow
in the form of a plug. Finally, the plugs were placed in the holes
(arranged in a regular array) of two opposite metal plates. In the
DuPont design, loops of bers were formed by winding the individual
bers or ber tows around a rotating wheel. Tubesheets were formed
by placing a mold over the end of the ber that possessed connections
to introduce a tubesheet forming material, such as epoxy resin [50].
Hollow ber modules might have different congurations to
meet the needs of the different applications. The present work used
shell-side feeding hollow ber module, as shown in Fig. 1. In this
conguration, two tube sheets hold the ber ends in place and
separate the retentate from the permeate ow. One is a plug-sealed
[(Fig._4)TD$IG]
Fig. 4. Process ow diagram (PFD) in ASPEN HYSYS for a design conguration with a double stage with permeate recycle (DSPR).
1252
tube sheet, in which the openings of the ber ends are blocked by
the epoxy resin; the other is an open-end tube sheet, in which the
bores of the hollow bers are exposed [38].
The hollow bers used for the experimental work are
commercial (Alpha Membrane Hi-Tech Pte. Ltd, Singapore). The
material of the membrane used is polyimide with a permeance of
10 GPU for CO2 and 0.25 GPU for CH4 at 50 8C. The bers have an
outer diameter of 400 mm (0.016 in.) and an inner diameter of
180 mm (0.007 in.). The required number of bers and the ber
length were calculated based on the diameter of the hollow bers
and the length of the module assuming a 4050% packing density.
For the test in the current study, the bers were cut into
approximately 28 cm lengths. Five different cases, 5, 15, 20, 30 and
50 bers in the bundle, were investigated.
Using the same procedure for bundle preparation, as explained
in the literature [38,51], the bers were cut to the desired length
while defective bers were removed. The remaining bers were
placed in parallel order as a ber bundle. A piece of barrier lm
(Paralm M1) was taken, stretched and wrapped on one end of the
ber bundle. The end of the ber bundle became denser due to
shrinkage of the lm. The wrapped side was cut with a sharp razor
blade to yield a smooth end.
Shells made from stainless steel (SS 316) with an outer
diameter of 1/2 in. (1.27 cm) and 1 in. (2.54 cm) were used. With
the help of string, the ber bundle was housed in the shell. The void
space between the bers and the internal wall of the shell was
potted, i.e., lled with epoxy glue (Loctite1 E-30CL Hysol1
adhesive). The purpose was to separate the permeate stream
from the retentate stream. Unlike the open side, the other side of
the bers was sealed completely by the epoxy glue to form a dead
end. The glue hardened in several minutes but reached its
[(Fig._5)TD$IG]
Fig. 5. Flow sheet of the gas separation testing unit for experimental validation.
[(Fig._6)TD$IG]
1253
Fig. 6. Experimental validation of the hollow ber ow model. (a) Effect of the CO2
content in the feed on stage cut. (b) Effect of the number of bers on the stage cut.
(c) Effect of stage cut on permeate CO2 mole fraction using data by Pan [11].
[(Fig._7)TD$IG]
Stage Cut
0.1205
0.12
0.1195
0.119
0.1185
0.118
0.1175
0.117
0.1165
0.116
0
100
200
300
400
Number of divisions
Fig. 7. Error analysis.
500
600
[(Fig._10)TD$IG]
[(Fig._8)TD$IG]
1254
50
Lower concentraon feed
(10 % CO2)
Medium concentraon
feed (40% CO2)
Higher concentraon feed
(70 % CO2)
50
40
30
45
40
Methane losses (%)
60
20
10
35
30
25
20
15
10
5
0
0
100
200
300
400
500
0.005
0.01
0.015
0.02
0.025
0.03
[(Fig._9)TD$IG]
28
Lower concentraon
feed (10 % CO2)
Medium concentraon
feed (40% CO2)
Higher concentraon
feed (70 % CO2)
24
Methane losses (%)
Methane loss
Methane (CH4) loss can be described as the percentage of
methane lost in the permeate stream to the methane present in the
feed stream. The loss increases with increasing membrane
separation-area and vice versa [15].
Figs. 8 and 9 show the effects of the ber length and radius of
the ber bundle, respectively, on CH4 loss for different CO2
concentration feed. The CH4 loss increased with increasing ber
length and radius of the ber bundle. An increase in ber length
and the radius of the ber bundle basically enhances the
membrane separation area. As a result, there will be higher
permeation through the membrane, leading to higher methane
loss.
Figs. 10 and 11 show the effects of the outer diameter of the
bers and porosity on the level of methane loss. The level of
methane loss decreased in both cases but the extent of the decrease
was more pronounced with a decrease in the outer diameter of
bers than the porosity of the membrane module. This is because
the increase in outer diameter or porosity of the ber at a constant
ber length and radius of the ber bundle results in a decrease in
membrane area. The decrease in membrane area leads to lower
permeation and less methane loss.
Higher feed concentrations (40% CO2 and 70% CO2) lead to
higher methane loss compared to lower feed concentrations (10%
CO2).
Fig. 10. Effect of the outer diameter of the ber on methane losses.
[(Fig._1)TD$IG]
20
16
12
8
4
40
30
25
[(Fig._12)TD$IG]
0.6
20
15
65
Compressor power
The effect of the membrane module characteristics on the
compressor power requirement was investigated for the proposed
Stage cut
35
60
0.7
Lower feed concentraon
(10 % CO2)
Medium feed
concentraon (40% CO2)
Higher feed concentraon
(70 % CO2)
55
0.8
40
50
Porosity (%)
50
45
45
0.5
0.4
0.3
0.2
10
0.1
5
0
0
0
10
15
20
25
30
35
100
200
300
Length of Fibers (cm)
400
500
[(Fig._13)TD$IG]
[(Fig._16)TD$IG]
400
0.8
Lower feed concentraon (10
% CO2)
Medium feed concentraon
(40% CO2)
Higher feed concentraon
(70 % CO2)
Stage cut
0.6
0.5
350
Compressor power (hp)
0.7
0.4
0.3
0.2
1255
300
250
200
150
100
50
0.1
10
15
20
25
30
100
200
400
500
[(Fig._17)TD$IG]
Fig. 13. Effect of the ber bundle radius on the stage cut.
[(Fig._14)TD$IG]
300
35
400
Lower feed concentraon (10
% CO2)
Medium feed concentraon
(40% CO2)
Higher feed concentraon (70
% CO2)
0.8
Lower feed concentraon
(10 % CO2)
Medium feed
concentraon (40% CO2)
Higher feed concentraon
(70 % CO2)
Stage cut
0.6
0.5
0.7
350
0.4
0.3
0.2
300
250
200
150
100
50
0.1
0
0
0.005
0.01
0.015
0.02
0.025
0.03
10
15
20
Radius
of ber bundle (cm)
25
30
35
Fig. 17. Effect of the ber bundle radius on the compressor power.
[(Fig._15)TD$IG]
[(Fig._18)TD$IG]
0.7
0.5
Medium feed
concentraon (40% CO2)
0.4
300
Compressor power (hp)
0.6
Stage cut
350
0.3
0.2
250
200
150
100
50
0.1
0
40
45
50
55
Porosity (%)
Fig. 15. Effect of the porosity on the stage cut.
60
65
0.005
0.01
0.015
0.02
0.025
0.03
[(Fig._19)TD$IG]
[(Fig._21)TD$IG]
220
200
180
160
140
120
100
80
100
0.07
90
0.06
0.04
70
Methane Purity (%)
0.03
60
0.02
GPC (USD/MSCF of
product)
50
0.01
0
40
60
40
45
50
55
60
0.05
0.045
0.04
0.035
0.03
0.025
97.5
0.02
97
0.015
GPC (USD/MSCF of
product)
96.5
0.01
96
100
99.5
98
300
400
500
[(Fig._2)TD$IG]
0.005
0.08
100
98.5
200
Fig. 21. Comparison of the methane purity and the gas processing cost for a medium
concentration feed (40% CO2) at different ber lengths.
99
100
65
Porosity (%)
0.05
80
0.07
90
0.06
80
0.05
0.04
70
Methane Purity (%)
0.03
60
0.02
GPC (USD/MSCF of
product)
50
240
1256
0.01
0
40
0
100
200
300
400
500
Conclusions
A cross ow model for hollow ber membrane separation was
included in the process simulation (Aspen HYSYS) as a user dened
unit operation along with the other available unit operations to
investigate the membrane separation system for CO2 removal from
natural gas. The simulated model was validated experimentally
and with a comparison with published data. The simulation
showed good agreement with the experimental and published
results. The design sensitivity was investigated by changing the
membrane module characteristics, such as the ber length, radius
of the ber bundle, diameter of the bers, and porosity, for the
proposed design conguration. The level of methane loss, stage cut
and compressor power increased with increasing ber length or
radius of ber bundle. On the other hand, an increase in the outer
diameter of the ber or porosity decreases the methane loss, stage
cut and compressor power. The GPC increased with increasing
membrane area (ber length) but the separation efciency was
also dependent on membrane area. Therefore, the membrane
module characteristics need to be optimized based on the
minimum GPC with respect to the required separation efciency.
95.5
0
100
200
300
400
500
Acknowledgment
We are grateful for the nancial and technical support of
Universiti Teknologi PETRONAS and COMSATS Institute of
Information Technology. We also thank Basic Science Research
hmix
6
7
n
n
X
X
6
Xi 7
7
hi 6
1
F
i
j
6
X j7
4
5
i1
j1
j 6 i
(22)
and
2
1
M i 1=2 4
Fi j p 1
1
Mj
2 2
hi
hj
!1=2
Mi
Mj
1=2
32
5
(23)
(24)
where
1=6
TC
1=2 2=3
M
PC
(25)
1257
[6] T. Katoh, M. Tokumura, H. Yoshikawa, Y. Kawase, Sep. Purif. Technol. 76 (2011) 362.
[7] M. Saidi, S. Heidarinejad, H.R. Rahimpour, M.R. Talaghat, M.R. Rahimpour, J. Nat.
Gas Sci. Eng. 18 (2014) 274.
[8] J.I. Marriott, E. Srensen, I.D.L. Bogle, Comput. Chem. Eng. 25 (2001) 693.
[9] A.S. Kovvali, S. Vemury, K.R. Krovvidi, A.A. Khan, J. Membr. Sci. 73 (1992) 1.
[10] S. Weller, W.A. Steiner, Chem. Eng. Prog. 46 (1950) 585.
[11] C.Y. Pan, AIChE J. 32 (1986) 2020.
[12] J.M. Thorman, H. Rhim, S.T. Hwang, Chem. Eng. Sci. 30 (1975) 751.
[13] C.R. Antonson, R.J. Gardner, C.F. King, D.Y. Ko, Ind. Eng. Chem. Process Des. Dev. 16
(1977) 463.
[14] R.T. Chern, W.J. Koros, P.S. Fedkiw, Ind. Eng. Chem. Process Des. Dev. 24 (1985)
1015.
[15] M.J. Thundyil, W.J. Koros, J. Membr. Sci. 125 (1997) 275.
[16] D.T. Coker, B.D. Freeman, G.K. Fleming, AIChE J. 44 (1998) 1289.
[17] S. Zhao, Z. Li, Y. Liu, L.e. Wang, Desalination 233 (2008) 310.
[18] J. Lemanski, G.G. Lipscomb, J. Membr. Sci. 167 (2000) 241.
[19] R. Khalilpour, A. Abbas, Z. Lai, I. Pinnau, Chem. Eng. Res. Des. 91 (2) (2013) 332.
[20] M.R. Sohrabi, A. Marjani, S. Moradi, M. Davallo, S. Shirazian, Appl. Math. Modell.
35 (2011) 174.
[21] A. Ebadi Amooghin, P. Moradi Shehni, A. Ghadimi, M. Sadrzadeh, T. Mohammadi, J.
Ind. Eng. Chem. 19 (2013) 870.
[22] A. Hysys, Aspen HYSYS Customization Guide, Aspen Technology Inc., Burlington,
USA, 2010.
[23] S.S.M. Lock, K.K. Lau, A.M. Shariff, J. Ind. Eng. Chem. DOI: 10.1016/
j.jiec.2014.03.017.
[24] J.N. Sahu, V.S.R.K. Chava, S. Hussain, A.V. Patwardhan, B.C. Meikap, J. Ind. Eng.
Chem. 16 (2010) 577.
[25] R. Rautenbach, R. Knauf, A. Struck, J. Vier, Chem. Eng. Technol. 19 (1996) 391.
[26] S. Tessendorf, R. Gani, M.L. Michelsen, Comput. Chem. Eng. 20 (Suppl. 1) (1996)
S653.
[27] R.A. Davis, Chem. Eng. Technol. 25 (2002) 717.
[28] A. Arpornwichanop, K. Koomsup, S. Assabumrungrat, J. Ind. Eng. Chem. 14 (2008)
796.
[29] A. Hussain, M.-B. Hagg, J. Membr. Sci. 359 (2010) 140.
[30] F. Nosratinia, M. Ghadiri, H. Ghahremani, J. Ind. Eng. Chem. DOI: 10.1016/
j.jiec.2013.10.065.
[31] M. Scholz, T. Harlacher, T. Melin, M. Wessling, Ind. Eng. Chem. Res. 52 (2012)
1079.
[32] F. Ahmad, K.K. Lau, A.M. Shariff, G. Murshid, Comput. Chem. Eng. 36 (2012) 119.
[33] F. Ahmad, K.K. Lau, A.M. Shariff, Y. Fong Yeong, J. Membr. Sci. 430 (2013) 44.
[34] H. Jin, S. Han, Y. Lee, Y. Yeo, Kor. J. Chem. Eng. 28 (2011) 41.
[35] S.P. Kaldis, G.C. Kapantaidakis, T.I. Papadopoulos, G.P. Sakellaropoulos, J. Membr.
Sci. 142 (1998) 43.
[36] R. Qi, M.A. Henson, J. Membr. Sci. 148 (1998) 71.
[37] R. Rautenbach, in: M.C. Porter (Ed.), Process Design and Optimization, William
Andrew, Norwich, New York, 1990.
[38] D. Li, R. Wang, T.-S. Chung, Sep. Purif. Technol. 40 (2004) 15.
[39] R.C. Reid, J.M. Prausnitz, T.K. Sherwood, The Properties of Gases and Liquids,
McGraw-Hill, New York, 1977.
[40] M.S. Peters, K.D. Timmerhaus, Plant Design and Economics for Chemical
Engineers, McGraw-Hill, New York, 1991.
[41] J. Hao, P.A. Rice, S.A. Stern, J. Membr. Sci. 209 (2002) 177.
[42] R.W. Spillman, M.G. Barrett, T.E. Cooley, Gas Membrane Process Optimization,
AIChE National Meeting, New Orleans, 1988.
[43] R.E. Babcock, R.W. Spillman, C.S. Goddin, T.E. Cooley, Energy Prog. 8 (1988) 135.
[44] B.D. Bhide, S.A. Stern, J. Membr. Sci. 81 (1993) 209.
[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, J. Nat. Gas Chem. 21 (2012) 7.
[47] N.H. Darman, A.R.B. Harun, Technical Challenges and Solutions on Natural Gas
Development in Malaysia, Beijing, 2006.
[48] H.I. Mahon, Google Patents (1966).
[49] R.R. Donald, M.W. Everett, M.J. Murdock, Google Patents (1967).
[50] N.C. Mat, Y. Lou, G.G. Lipscomb, Curr. Opin. Chem. Eng. 4 (2014) 18.
[51] H. Huang, K. Schwab, J.G. Jacangelo, Membr. Water Treat. 2 (2011) 121.
[52] S. Saxena, Viscosity of Multicomponent Mixtures of Gases, 1973, p. 100.
[53] D.S. Viswanath, Viscosity of Liquids: Theory, Estimation, Experiment, and Data,
Springer, New York, 2007.