Desalination 287 (2012) 200208

Contents lists available at SciVerse ScienceDirect

Desalination
journal homepage: www.elsevier.com/locate/desal

CFD analysis of the initial stages of particle deposition in spiral-wound

membrane modules
Yu-Ling Li a,, Kuo-Lun Tung a,, Yu-Shao Chen b, Kuo-Jen Hwang b
a
R&D Center for Membrane Technology and Dept. of Chemical Engineering, Chung Yuan University, Chung-Li, Taoyuan 320, Taiwan
b
Department of Chemical and Materials Engineering, Tamkang University, Danshui, New Taipei City 251, Taiwan

a r t i c l e i n f o a b s t r a c t

Article history: Particle deposition in a spiral-wound membrane module was simulated using computational uid dynamics
Received 7 July 2011 (CFD). A scheme similar to the EulerianLagrangian numerical method was adopted for the two-phase ow
Received in revised form 30 September 2011 simulation. The effect of curvature on particle transport in four spacer-lled channel congurations with per-
Accepted 3 October 2011
meable membrane surfaces was analyzed by considering the uid drag, body force and lift force exerted on
Available online 26 October 2011
the particles. The numerical results showed that there are inherent changes in the particle deposition prole
Keywords:
in the spacer-lled channel due to variations in curvature. Comparing the particle deposition proles and de-
Membrane module position ratios for submerged, zigzag, i-cavity and o-cavity spacer-lled channels showed that, for a given
Curvature feed velocity and permeation rate, the zigzag-type spacer is best at decreasing the inuence of curvature
CFD and preventing particle fouling on the membrane. A microscopic understanding derived from the CFD anal-
Spacer ysis could improve module design and enhance membrane module performance.
Particle deposition 2011 Elsevier B.V. All rights reserved.

1. Introduction A three-dimensional CFD study of spacer geometries in spiral-

Computational uid dynamics (CFD) has been widely used to sim-
ulate the hydrodynamic behavior of membrane separation processes
using membrane modules [15]. A substantial body of literature ex-
ists for the spiral-wound membrane module because of its complex
structure, which includes a feed spacer, membrane, permeable spacer
and collection-tube. A spiral-wound membrane module has three
types of ow paths: axial ow (along the feed spacer-lled channel,
i.e., the z-direction), spiral ow (parallel to the permeate ow path,
i.e., the -direction) and radial ow (toward the membranes, i.e.,
the r-direction).
A two-dimensional model has been used to analyze transport phe-
nomena in the spiral-wound membrane module owing to simplify
the simulation system. Cao et al. [6] used CFD to qualitatively observe
the local wall shear stresses associated with different spacer congu-
rations that have laments adjacent to the wall. Schwinge et al. [7, 8]
revealed the effect of three different spacer arrangements, including
cavity, zigzag and submerged spacers, on the hydrodynamics and
mass transfer. Their results showed that the zigzag spacer was the
most efcient spacer type for the spacer-lled channel. Later, several
theoretical studies used CFD to examine mass transfer and concentration
polarization [812], particulate deposition on a at channel with a per-
meable membrane surface [13] and unsteady ow phenomena [12,14-
15] in a spiral-wound membrane module.
Corresponding author. Tel.: + 886 3 2654129; fax: + 886 3 2654199.
Corresponding author. Tel.: +886 3 2654195; fax: +886 32654198.
E-mail addresses: yuling@cycu.org.tw (Y.-L. Li), kuolun@cycu.edu.tw (K.-L. Tung).
wound membrane modules was performed by Karode and Kumar
[16]. They used a laminar model to calculate the ow eld in the
spacer-lled channel, and the ow pattern and shear stress they
obtained was used to qualitatively compare the performance of differ-
ent spacers. To enhance the performance of spiral-wound membrane
modules, new spacer designs have been studied [1725]. Additionally,
the appropriate cell types and periodic boundary conditions have
been suggested based on a three-dimensional CFD analysis of spacer-
lled membrane module designs with various spacer arrangements
[26]. A new three-dimensional computational model that couples
uid dynamics, solute transport and biofouling via biolm formation
in spacer-lled membrane modules has also been described [2728].
The numerical simulations showed that the biomass accumulation
was due to attachment and biolm growth over time. Biomass accumu-
lation strongly affected the feed channel pressure drop, liquid velocity
distribution and residence time distribution.
Fluid ow in a spacer-lled, curved channel has recently been in-
vestigated [2930]. However, the authors focused on the ow path in
the spiral-direction. Li and Tung [31] pointed out that although the
outward appearance was similar to that of a spiral channel, the uid
in the spacer-lled channel displayed the axial ow. Their results
showed that the curvature of spacer-lled channels affects the ow
eld in spiral-wound membrane modules. Using a two-dimensional
numerical scheme, they also found that in a curved, spacer-lled
channel, the shear stress on the inner wall is greater than that on the
outer wall. In addition, three-dimensional CFD and an experimental
set-up with a curved channel lled with a two-layer lament spacer
were used to analyze the uid ow in the channel [32]. Efforts to

0011-9164/$ see front matter 2011 Elsevier B.V. All rights reserved.
doi:10.1016/j.desal.2011.10.001
 Y.-L. Li et al. / Desalination 287 (2012) 200208 201

Fig. 1. The simulation systems: (a) the ow channel with permeable membrane surfaces, (b) the transverse lament congurations for the zigzag, submerged and cavity (o-cavity
and i-cavity types) spacers.

mitigate the curvature effect in the spacer-lled channel of a spiral-
wound membrane module have used a spacer with unequal lament
diameters between the inner and outer layers. This type of spacer can
be used to reduce the shear stress imbalance between the inner and
outer walls and extend the service life of the membrane module.
In this work, the effect of curvature on particle transport in four
spacer-lled channel congurations with permeable membrane sur-
faces was investigated using CFD. A scheme similar to the Eulerian
Lagrangian numerical method was used to model the uid drag,
body force and lift force exerted on the particles in a two-phase
ow simulation of a spiral-wound membrane module.
2. Theoretical study
2.1. The simulated system
The simulation system for the spiral-wound membrane module is
illustrated in Fig. 1(a). The feed channel with permeable membrane
surfaces and various spacers, modeled by a exible cylindrical ber
of circular cross-section, was assembled with two coaxial cylindrical
tubes. Four different spacer congurations, submerged, zigzag, i-
cavity and o-cavity, were used in this study, as shown in Fig. 1(b).
The i-cavity and o-cavity types in the curved channel were named

Fig. 2. Schematic diagrams of the zigzag spacer-lled channels with two dimensionless radii of curvature: (a) = 0 and (b) = 0.18.
202 Y.-L. Li et al. / Desalination 287 (2012) 200208

Table 1
The geometric parameters and values of the spacer-lled channels with permeable membranes for various spacer congurations.

df lf hm hch Lch Lm =0 = 0.18

Ri Ro Ri Ro

Empty channel
Submerged type
Zigzag type 0.5 4 0.01 1 40 24 na na 5 6
o-Cavity type
i-Cavity type
na: not available. Unit: mm

asymmetric spacers, and the zigzag and submerged types were (II) Momentum equations
named symmetric spacers. Six laments were installed in each
channel and were positioned perpendicular to the feed ow direction. 1 1 P
r uz uz r u u 4
In this axisymmetric, two-dimensional (zr plane) simulation of the r z r r 
r z z


spacer-lled channel, the lateral movements of depositing particles 1 u 2 uz ur ur
r 2 z
and the membrane surface effects in the -direction were neglected. r z z 3 z r r
" !!#
Schwinge et al. [7] noted that the channel entrance distance needs 1 uz ur
r :
to be at least 10 times the cylindrical lament diameter to achieve r r r z
fully developed ow. In addition, the exit distance needs to be at 

1 1 P 1 ur uz
least twice the entrance distance to avoid any channel-exit effects r uz ur r ur ur r 5
on eddy formation behind the cylindrical laments. r z r r r r z z r



Li and Tung [31] have proposed that the dimensionless radius of 1 ur 2 uz ur ur
r 2
curvature, , be dened by r r r 3 z r r


ur 2 uz ur ur uz 2
2 2 :
r 3 r z r r r
Ro Ri
1
Rave To study the uid ow through the permeable membrane sur-
faces, the membrane permeation effect was investigated. The driving
force was the transmembrane pressure (TMP) gradient, and the per-
where Rave is the arithmetic mean of the outer radius,Ro, and inner meate ux, J, was dened by Darcy's equation:
radius Ri:
TMP
J 6
Rm Rc
Ro Ri where Rm is the resistance of the clean membrane. The initial stage was
Rave : 2
2 used as the simulation condition in this study, so the resistance of the
cake, Rc, was neglected. The following simulated operating conditions
were used: a feed velocity of 0.23 m/s, an Rm of 1.01012 1/m with refer-
The degree of curvature of spacer-lled channels increases as the ring the range order of 1012 [34] and a TMP of 500 kPa. For the dispersed-
value of increases. An value of 0 corresponds to a at channel. In phase simulation, the particle motion was governed by Newton's 2nd law
this study, two dimensionless radii of curvature were chosen for the using the Lagrangian approach. Because the volume fraction of particles
analysis: = 0 and = 0.18 (Fig. 2). The geometries of the empty was less than 10%, the existence of a dispersed phase in the continuous
and spacer-lled at and curved channels are shown in Table 1. The phase did not noticeably affect the uid physical properties. Because
channel height, hch, was 1 mm (r = 0 to r = 1 10 3 m) and the chan- the Brownian motion was small relative to other particle transport
nel length was 40 mm (z = 0 to z = 4 10 2 m). The channel Reyn-
olds number [31] in the spacer-lled system was about 360 to
assure that the laminar ow model was valid [7].

2.2. Governing equations for the uid ow

The ow eld and particle transport in the at and curved chan-

nels with permeable membrane surfaces were calculated using the
discrete phase model in FLUENT. The scheme is similar to the Euler-
ianLagrangian numerical method [13, 33] and was adopted for the
two-phase ow simulations. This method limits that a fairly low vol-
ume fraction of the discrete phase usually less than 1012%. For the
continuous-phase simulations, the ow eld was calculated based
on the uid-phase continuity and momentum equations. The conti-
nuity and momentum equations for the two-dimensional, axisym-
metric simulation were as follows:
(I) Continuity equation

u
ur uz r 0: 3
r z r Fig. 3. The simulation procedure using the discrete phase model.
 Y.-L. Li et al. / Desalination 287 (2012) 200208 203

Fig. 4. The deposition ratios at different positions on the inner and outer membranes in the empty and curved channels.

Fig. 5. The deposition ratios at different positions on the inner and outer membranes in the submerged spacer-lled at and curved channels.
204 Y.-L. Li et al. / Desalination 287 (2012) 200208

Table 2
The effect of various spacer congurations on the deposition ratio and the variation in the deposition ratio for the inner and outer membranes in at and curved channels.

Empty channel Submerged type Zigzag type o-Cavity type i-Cavity type

o (%) 9.2a/9.5b 9.02a/9.18b 6.58a/6.7b 4.67a/5.34b 9.07a/9.13b

i (%) 9.2a/8.9b 9.04a/8.80b 6.58a/6.4b 9.07a/9.02b 4.67a/4.06b
Et (%) 0a/6.04b 0.2a/4.2b 0a/3.6b 64.05a/51.31b 64.05a/76.79b
a
Based on = 0.
b
Based on = 0.18.

mechanisms and could be neglected, a particle size of 1 m was selected where dp is the particle diameter, u is the uid velocity and up is the par-
for the simulation. In addition, the particles have little-to-no surface ticle velocity. In addition, Rep is the Reynolds number for the particle and
charge, so any electrical double-layer effects were ignored. In this is dened as
preliminary study, the momentum exchange from the particles to the  
uid was neglected. The net gravitational force, hydraulic drag force dp uup
and lift force were considered when calculating the suspended particle Rep 10

motion:

dup   where is the uid viscosity and CD is the drag coefcient. The drag coef-
FG FD uup FL 7 cient is evaluated using the following equation:
dt
a2 a
where FG, FD and FL are the net gravity force, particle drag force and CD a1 3 11
Rep Re2p
Saffman lift force, respectively. These forces were calculated per unit par-
ticle mass. The net gravitational force exerted on a deposited particle can where a1, a2 and a3 are constants associated with smooth, spherical parti-
be calculated by cles over a range of Rep [35]. The Saffman lift force is dened as
 
p 0:5 0:5  
FG g 8 3:1 du=dr
p FL uup 12
dp p
3
where (998 kg/m ) is the density of the uid (water). In addition, the
particle density, p, is referred to the polymethyl methacrylate particle where is the kinematic viscosity. The simulation procedure is shown in
about 1210 kg/m3 [32]. The drag force is dened as Fig. 3. The particles were uniformly released at the feed channel inlet from
r=0 to r=1103 m. It should be noted that the center of a particle was
3 CD Rep
FD uup 9 used as its position in the CFD simulation, and the real particle size was
4p d2p neglected by the CFD code [36]. To overcome this limitation, (UDF) facility

Fig. 6. The deposition ratios at different positions on the inner and outer membranes in the zigzag spacer-lled at and curved channels.
 Y.-L. Li et al. / Desalination 287 (2012) 200208 205

Fig. 7. The deposition ratios at different positions on the inner and outer membranes in the i-cavity spacer-lled at and curved channels.

provided by FLUENT [37] was used to consider particle radius when
detecting particle attachment to the membrane or lament surface.
3. Results and discussion
The empty and four spacer congurations in at and curved chan-
nels with permeable membrane surfaces were used in the analysis.
Based on the simulation results obtained for the membrane module,
the effect of curvature on the ow pattern and particle deposition
with various spacer-lled channels is discussed comprehensively in
the following sections.
positions were dened as the number of particles deposited on a
given section of membrane divided by the total number of particles
fed to the channel. In this work, the channel in the z-direction was
divided into 24 zones (1C, 1D, 2A2D, , 6A6D, 7A and 7B) for inter-
preting and comparing the particle deposition in at and curved chan-
nels. For uid ow in the two types of channels, the velocity decreased
slightly along the channel due to uid suction toward the membrane
surface, as depicted in the center of Fig. 4. The deposition ratio was al-
most identical for the relative positions at top and bottom membrane
surfaces in the at channel. The average deposition ratios for the mem-
brane surfaces, o and i, were dened by

3.1. The effect of curvature on the ow pattern and particle deposition for No
o 13
symmetric spacer-lled channels Nf

To conrm the feasibility of the simulation process, the effect of cur- and
vature on the ow pattern and particulate deposition ratio was studied
for at and curved channels with permeable membrane surfaces, and Ni
i 14
the results are presented in Fig. 4. The deposition ratios at the relative Nf

Fig. 8. The velocity contours near the lament in the i-cavity spacer-lled at and curved channels.
206 Y.-L. Li et al. / Desalination 287 (2012) 200208

Fig. 9. The deposition ratios at different positions on the inner and outer membranes in the o-cavity spacer-lled at and curved channels.

where No and Ni are the sum of discharging positions deposited onto the
outer and inner membranes, respectively, and Nf is the sum of dischar-
ging positions for the feed particles. In addition, the variation in the av-
erage deposition ratio on the outer and inner membranes, Et, was
dened by
jo i j
Et : 15
o i =2
Table 2 also shows an average membrane surface deposition ratio
of 9.2%. Nearly 80% of the particles ow out of the empty channel with
two membrane surfaces. However, the deposition ratios at the posi-
tions on the outer membrane surface were greater than those at the
respective positions on the inner membrane surface, as shown in
Fig. 4. In addition, Table 2 shows a similar average deposition ratio
(9.5% > 8.9%), which was due to the shear stress on the inner mem-
brane surface being higher than that on the outer membrane surface
[31]. Therefore, the shear stress plays a major role in particle
deposition in the empty channel. Hence, the particulate deposition
location will be affected by the curvature of the channel, in addition
to the injection position (r-coordinate) in channel, the feed velocity
and the permeation ux.
The effect of curvature on the ow pattern and particle deposition
for the submerged spacer-lled channels is illustrated in Fig. 5. The par-
ticulate deposition ratios in zones B and C (3B and 3C, 4B and 4C, 4B and
4C) on the outer membrane surface were higher than those on the inner
membrane surface because of the lower velocity near the outer mem-
brane surface. It is notable that there was no obvious particle deposition
trend in zones D and A (2D and 3A, 3D and 4A, , 6D and 7A). The
velocity increases as the uid passes through the submerged spacer,
which increases the shear stress. However, the higher normal velocity
(ur) in front of each lament caused particles to deposit onto the mem-
brane. In other words, the particle deposition behavior was dependent
on both the shear stress and the normal velocity.
The effect of curvature on the ow pattern and particle deposition in
the zigzag spacer-lled channels is illustrated in Fig. 6. Recirculation

Fig. 10. The velocity contours near the lament in the o-cavity spacer-lled at and curved channels.
 Y.-L. Li et al. / Desalination 287 (2012) 200208 207

regions were found behind each lament; they were found in zones 2A 4. Conclusions
to 2C, 4A to 4C and 6A to 6C of the inner membrane surface and zones
3A to 3C, 5A to 5C and 7A to 7C of the outer membrane surface. The Computational uid dynamics was used to investigate the effect of
recirculation created higher shear stress and prevented particle deposi- curvature on particle deposition in a spacer-lled channel with a perme-
tion. For the at channel, it was found that the particle deposition ratio able membrane surface in a spiral-wound membrane module. A scheme
was higher in zone D than in zone C. For the outer membrane surface in similar to the EulerianLagrangian numerical method was adopted for
the curved channel, however, the particulate deposition ratios in zones the two-phase ow simulation. The forces acting on the particles included
4D and 6D were lower than those in zones 4C and 6C. The velocities be- uid drag, body force and lift force. Four different spacer congurations,
tween the outer membrane surfaces and laments in zones 4A and 6A submerged, zigzag, i-cavity and o-cavity, were modeled in this study.
were reduced for the curved channel, thereby increasing recirculation The simulation results showed that recirculation behind the lament
before the lament. The increased recirculation increased the shear and adjacent to membrane surface caused high shear stress and pre-
stress and prevented particle deposition. vented particle deposition behind each lament. The recirculation size
and the positions on the inner and outer membranes caused different
deposition mechanisms. The particle deposition on the membranes in
3.2. The effect of curvature on the ow pattern and particle deposition for the empty channel was dependent on the shear stress; however, the
asymmetric spacer-lled channels ow patterns were also important in the curved channel. The zigzag
and submerged spacers decreased the effect of curvature on particle
In a at channel, the conguration of an i-cavity spacer is the same deposition. The i-cavity and o-cavity spacers enhanced the effect of curva-
as that of an o-cavity spacer, which results in a similar ow eld. ture and led to unequal deposition of the colloidal material on the inner
However, the two spacers produce different ow patterns in curved and outer membranes and to a reduction in the service life of the mem-
channels [31]. The effect of curvature on the ow pattern and particle brane module. Based on this preliminary study, it is concluded that a
deposition for the i-cavity spacer-lled channels is illustrated in Fig. 7. multi-phase ow simulation can provide a microscopic understanding
The particulate deposition ratios for the outer membrane surface po- of the fouling mechanisms in a spiral-wound module and is a powerful
sitions with the curved channel were generally larger than the tool for use in membrane module design.
respective ratios in the at channel due to the lower shear stress.
There was very little particle deposition in each A zone on the inner Nomenclature
membrane surface because of a recirculation region that existed be- ai constant for i, with i (1, 2, 3),
hind the lament adjacent to the inner membrane surface. Neal et CD drag coefcient,
al. [38] have also noticed phenomenon by direct observation through df lament diameter, mm
the membrane. It was notable that the particle deposition tendency in Et the variation in particle deposition ratio between the inner
each D zone was different for the i-cavity spacer than for the zigzag and outer membranes,
spacer. Recirculation zones were generated between the adjacent FD drag force per unit particle mass, m/s 2
laments for the i-cavity spacer, while recirculation zones were gen- FL Saffman's lift force per unit particle mass, m/s 2
erated behind and in front of the laments for the zigzag spacer. The FG net gravity force per unit particle mass, m/s 2
limit of recirculation in the curved channel for the i-cavity spacer was g gravitational acceleration (m/s 2)
far from the lament, as shown in Fig. 8. Because the curved channel hch channel height, mm
makes the uid ow toward the inner membrane surface, it will sup- hm membrane thickness, mm
press the recirculation size and decrease the normal velocity at the J permeate ux, m 3/m 2s
inner membrane surface. It will also decrease the probability of parti- Lch channel length, mm
cle deposition. Lm membrane length, mm
The effect of curvature on the ow pattern and particle deposition lf distance between laments, mm
for the o-cavity spacer-lled channels is shown in Fig. 9. The particu- Nf the sum of discharging positions for feed particles,
late deposition ratios for the positions on the inner membrane surface Ni the sum of discharging positions deposited onto the inner
in the curved channel were generally less than the respective ratios in membrane,
the at channel because of higher shear stress. It is notable that the No the sum of discharging positions deposited onto the outer
magnitude of the recirculation between the adjacent laments in membrane,
the curved channel was greater than that in the at channel, as P pressure, Pa
shown in Fig. 10. Because the curved channel makes uid ow toward r r-coordinate,
the inner membrane surface, it will enhance the recirculation close to Rep Reynolds number based on the particle,
the lament and increase the normal velocity at the outer membrane Rc resistance of the cake, 1/m
surface. This effect will increase the probability of particle deposition Rave arithmetic mean of the outer Ro and inner Ri radii, mm
on the outer membrane surface. Ri inner channel radius, mm
To estimate the effect of curvature on the performance of the spiral- Rm resistance of the clean membrane, 1/m
wound membrane module, the changes in the particle deposition ratio Ro outer channel radius, mm
for the inner and outer membranes are presented in Table 2. For the t time, s
curved channel without spacers, it is interesting that there were similar u uid velocity, m/s
results for Et and for the variation in the shear stress between the inner up particle velocity, m/s
and outer walls [31]. The particle deposition on membranes without ur uid velocity in the r-direction, m/s
spacers depends on the shear stress. Using data for the average deposition uz uid velocity in the z-direction, m/s
ratios on the membrane, we found that channels lled with submerged or z z-coordinate,
zigzag spacers decreased particle deposition and that the curvature inu-
enced the results. There were higher Et values for the o-cavity and i-cavity Greek letters
spacers. However, the non-uniform performance between the membrane dimensionless radius,
leaves in the module can accelerate the degradation of the membrane i average deposition ratio of the inner membrane surface,
module [16]. Hence, a membrane module lled with a suitable spacer o average deposition ratio of the outer membrane surface,
will have enhanced operational life and performance. uid density, kg/m 3
208 Y.-L. Li et al. / Desalination 287 (2012) 200208

p particle density, kg/m 3 [17] F. Li, G.W. Meindersma, A.B. de Haan, T. Reith, Optimization of non-woven spacers
by CFD and validation by experiments, Desalination 146 (2002) 209212.
uid viscosity, kg/m-s [18] J. Schwinge, D.E. Wiley, A.G. Fane, Novel spacer design improves observed ux, J.
kinematics viscosity, m 2/s Membr. Sci. 229 (2004) 5361.
[19] D. Dendukuri, S.K. Karode, A. Kumar, Flow visualization through spacer lled
channels by computational uid dynamics-II: improved feed spacer designs, J.
Acknowledgements Membr. Sci. 249 (2005) 4149.
[20] C.P. Koutsou, S.G. Yiantsios, A.J. Karabelas, Direct numerical simulation of ow in
We thank the Taiwan Water Corporation (TWC), Ministry of spacer-lled channels: effect of spacer geometrical characteristics, J. Membr. Sci.
291 (2007) 5369.
Economic Affairs (MOEA), R.O.C. for the nancial supports under grant [21] M. Shakaib, S.M.F. Hasani, M. Mahmood, Study on the effects of spacer geometry
100TWC10. in membrane feed channels using three-dimensional computational ow model-
ing, J. Membr. Sci. 297 (2007) 7489.
[22] J.L.C. Santos, V. Geraldes, S. Velizarov, J.G. Crespo, Investigation of ow patterns
References
and mass transfer in membrane module channels lled with ow-aligned spacers
using computational uid dynamics (CFD), J. Membr. Sci. 305 (2007) 103117.
[1] J. Schwinge, P.R. Neal, D.E. Wiley, D.F. Fletcher, A.G. Fane, Spiral wound modules
[23] P. Koutsoua, S.G. Yiantsiosa, A.J. Karabelas, A numerical and experimental study of
and spacers. Review and analysis, J. Membr. Sci. 242 (2004) 129153.
mass transfer in spacer-lled: effects of spacer geometrical characteristics and
[2] G.A. Fimbres-Weihs, D.E. Wiley, Review of 3D CFD modeling of ow and mass
Schmidt number, J. Membr. Sci. 326 (2009) 234252.
transfer in narrow spacer-lled channels in membrane modules, Chem. Eng. Pro-
[24] M. Shakaib, S.M.F. Hasani, M. Mahmood, CFD modeling for ow and mass transfer in
cess. 49 (2010) 759781.
spacer-obstructed membrane feed channels, J. Membr. Sci. 326 (2009) 270284.
[3] R. Ghidossi, D. Veyret, P. Moulin, Computational uid dynamics applied to mem-
[25] K.K. Lau, M.Z. Abu Bakar, A.L. Ahmad, T. Murugesan, Feed spacer mesh angle: 3D
branes: state of the art and opportunities, Chem. Eng. Process. 45 (2006) 437454.
modeling, simulation and optimization based on unsteady hydrodynamic in spi-
[4] Y.L. Li, P.J. Lin, K.L. Tung, CFD analysis of uid ow through a spacer-lled disk-
ral wound membrane channel, J. Membr. Sci. 343 (2009) 1633.
type membrane module, Desalination (2011), doi:10.1016/j.desal.2011.03.091.
[26] Y.L. Li, K.L. Tung, CFD simulation of uid ow through spacer-lled membrane
[5] K.L. Tung, Y.L. Li, C.C. Hu, Y.S. Chen, Power-law polymer solution ow in a
module: selecting suitable cell types for periodic boundary conditions, Desalina-
converging annular spinneret: analytical approximation and numerical computa-
tion 233 (2008) 351358.
tion, AICHE J. (2011), doi:10.1002/aic.12561.
[27] C. Picioreanua, J.S. Vrouwenvelder, M.C.M. van Loosdrecht, Three-dimensional
[6] Z. Cao, D.E. Wiley, A.G. Fane, CFD simulations of net-type turbulence promoters in
modeling of biofouling and uid dynamics in feed spacer channels of membrane
a narrow channel, J. Membr. Sci. 185 (2001) 157176.
devices, J. Membr. Sci. 345 (2009) 340354.
[7] J. Schwinge, D.E. Wiley, D.F. Fletcher, Simulation of the ow around spacer laments
[28] J.S. Vrouwenvelder, C. Picioreanu, J.C. Kruithof, M.C.M. van Loosdrecht, Biofouling
between narrow channel walls. 1. Hydrodynamics, Ind. Eng. Chem. Res. 41 (2002)
in spiral wound membrane systems: Three-dimensional CFD model based evalu-
29772987.
ation of experimental data, J. Membr. Sci. 346 (2010) 7185.
[8] J. Schwinge, D.E. Wiley, D.F. Fletcher, Simulation of the ow around spacer la-
[29] V.V. Ranade, A. Kumar, Comparison of ow structures in spacer-lled at and an-
ments between narrowchannelwalls. 2. Mass-transfer enhancement, Ind. Eng.
nular channels, Desalination 191 (2006) 236244.
Chem. Res. 41 (2002) 48794888.
[30] S. Wardeh, H.P. Morvan, Detailed numerical simulations of ow mechanics and
[9] A.L. Ahmad, K.K. Lau, M.Z. AbuBakar, Impact of different spacer lament geome-
membrane performance in spacer-lled channels, at and curved, Desalin.
tries on concentration polarization control in narrow membrane channel, J.
Water Treat. 1 (2009) 277288.
Membr. Sci. 262 (2005) 138152.
[31] Y.L. Li, K.L. Tung, The effect of curvature of a spacer-lled channel on uid ow in
[10] A. Subramani, S. Kim, E.M.V. Hoek, Pressure, ow, and concentration proles in
spiral-wound membrane modules, J. Membr. Sci. 319 (2008) 286297.
open and spacer-lled membrane channels, J. Membr. Sci. 277 (2006) 717.
[32] Y.L. Li, K.L. Tung, M.Y. Lu, S.H. Huang, Mitigating the curvature effect of the spacer-lled
[11] S. Wardeh, H.P. Morvanb, CFD simulations of ow and concentration polarization
channel in a spiral-wound membrane module, J. Membr. Sci. 329 (2009) 106118.
in spacer-lled channels for application to water desalination, Chem. Eng. Res.
[33] K.L. Tung, Y.L. Li, K.T. Lu, W.M. Lu, Effect of calendering of lter cloth on transient
Des. 86 (2008) 11071116.
characteristics of cake ltration, Sep. Purif. Technol. 48 (2006) 115.
[12] A.L. Ahmad, K.K. Lau, Impact of different spacer laments geometries on 2D un-
[34] S.F.E. Boerlage, M.D. Kennedy, M.R. Dickson, D.E.Y. El-Hodali, J.C. Schippers, The modi-
steady hydrodynamics and concentration polarization in spiral wound mem-
ed fouling index using ultraltration membranes (MFI-UF): characterisation, ltration
brane channel, J. Membr. Sci. 286 (2006) 7792.
mechanisms and proposed reference membrane, J. Membr. Sci. 197 (2002) 121.
[13] Y.L. Li, T.H. Chang, C.Y. Wu, C.J. Chuang, K.L. Tung, CFD analysis of particle deposi-
[35] S.A. Morsi, A.J. Alexander, An investigation of particle trajectories in two-phase
tion in the spacer-lled membrane module, J. Water Supply Res. Technol. Aqua 55
ow systems, J. Fluid Mech. 55 (1976) 193208.
(78) (2006) 589601.
[36] M.J. Lehmann, Improved CFD modeling of particle-wall collision by including in-
[14] G.A. Fimbres-Weihs, D.E. Wiley, D.F. Fletcher, Unsteady ows with mass transfer
terception, Filtr. Solut. 6 (2006) 159162.
in narrowzigzag spacer-lled channels: a numerical study, Ind. Eng. Chem. Res.
[37] FLUENT V 6.2.16 Users Manual, FLUENT Inc., 2005.
45 (2006) 65946603.
[38] P.R. Neal, H. Li, A.G. Fane, D.E. Wiley, The effect of lament orientation on critical
[15] A. Alexiadis, D.E. Wiley, D.F. Fletcher, J. Bao, Laminar ow transitions in a 2D
ux and particle deposition in spacer-lled channels, J. Membr. Sci. 214 (2003)
channel with circular spacers, Ind. Eng. Chem. Res. 46 (2007) 53875396.
165178.
[16] S.K. Karode, A. Kumar, Flow visualization through spacer lled channels by com-
putational uid dynamics I. Pressure drop and shear rate calculations for at
sheet geometry, J. Membr. Sci. 193 (2001) 6984.