w a t e r r e s e a r c h 4 6 ( 2 0 1 2 ) 2 5 3 5 e2 5 4 4

Available online at www.sciencedirect.com

journal homepage: www.elsevier.com/locate/watres

A comparison of cake properties in traditional and turbulence

promoter assisted microfiltration of particulate suspensions

Yuanfa Liu, Gaohong He*, Baojun Li, Zhengwen Hu, Jia Ju

State Key Laboratory of Fine Chemicals, R&D Center of Membrane Science and Technology, School of Chemical Engineering, Dalian University
of Technology, 2 Linggong Road, Dalian 116024, China

article info abstract

Article history: The use of turbulence promoter can effectively enhance the permeate flux in crossflow
Received 13 October 2011 microfiltration (CFMF) of particulate suspensions. Flux enhancement which is generally
Received in revised form attributed to the reduction in cake thickness, however, has still not been clearly under-
29 January 2012 stood. In this study, the effects of turbulence promoter on cake properties (thickness,
Accepted 1 February 2012 porosity and particle size) were investigated during CFMF of calcium carbonate suspension.
Available online 9 February 2012 It indicates that turbulence promoter has important effects on cake properties that directly
affect the cake resistance. The significant reduction in thickness and slight increase in
Keywords: porosity are positive to reduce the cake resistance. The remarkable decrease in particle size
Membrane fouling is the negative effect due to the increased specific resistance of a cake. As a whole, the
Flux enhancement overall cake resistance is still diminished by turbulence promoter and therefore permeate
Turbulence promoter flux is improved. The theoretical calculation of cake resistance shows a good consistence
Cake properties with the experimental result. The cake properties in both cases (using a turbulence
promoter or not) almost exhibit the similar trends under various operating conditions.
Differently, the use of turbulence promoter can greatly alleviate the effects of trans-
membrane pressure or feed concentration on growth of cake layer and intensify the effects
of inlet velocity on diminishing the particle deposition.
ª 2012 Elsevier Ltd. All rights reserved.

1. Introduction crossflow microfiltration (CFMF) of particulate suspensions

Microfiltration (MF) is a pressure-driven membrane filtration
process to retain colloidal particles including silica, iron
oxides, calcite, and clays (Walker et al., 2006) during drinking
water (Carroll et al., 2003; Lee et al., 2008) and wastewater (Al-
Malack and Anderson, 1997; Geissen and Xi, 2001) treatment.
However, the applications of MF are seriously hindered by the
membrane fouling (Kimura et al., 2007; Li et al., 2011; Zhang
et al., 2010) that causes an undesired decline of permeate
flux (Lin et al., 2009) with time. The membrane fouling in
refers to the deposition of rejected particles on the membrane
surface, leading to blockage of membrane pores and forma-
tion of a cake layer. Since surface deposition of cake particles
is achieved quickly (about a few minutes) due to the high
filtration flow during the initial filtration period, cake forma-
tion is regarded as the predominant mechanism responsible
for the flux decline.
The cake permeability is generally much lower than
membrane, and thus the cake layer plays an important role in
the permeate flux of membrane. It is of great significance to

* Corresponding author. Tel.: þ86 411 84707892; fax: þ86 411 84007700.
E-mail address: hgaohong@dlut.edu.cn (G. He).
0043-1354/$ e see front matter ª 2012 Elsevier Ltd. All rights reserved.
doi:10.1016/j.watres.2012.02.002
2536 w a t e r r e s e a r c h 4 6 ( 2 0 1 2 ) 2 5 3 5 e2 5 4 4

estimate cake properties and investigate their influences on

the membrane filtration performance. Vyas et al. (2000b) 2. Experiment and method
developed a novel method to measure cake thickness which
was conducted by reconstruction of cake layer, and investi- 2.1. Ceramic tubular membrane
gated the effects of operating conditions on cake properties in
CFMF of lactalbumin suspension. It was found that the cake The ceramic tubular membrane (Nanjing University of Tech-
thickness increased with an increase in transmembrane nology, P.R. China) used in this study was 200 mm in length,
pressure (TMP) and with a decrease in crossflow velocity, but with inner diameter of 15 mm and outer diameter of 18 mm.
there was no significant effect of TMP or crossflow velocity on The pore size distribution of membrane is shown in Fig. 1,
cake porosity. Ould-Dris et al. (2000) demonstrated that the with average pore size of 0.91 mm.
cake thickness, porosity and particle size distribution had
significant effects on permeate flux in CFMF of calcium 2.2. Turbulence promoter
carbonate suspensions. Hwang and Hsueh (2003) proposed
a dynamic analysis method to estimate cake properties in The helical screw insert which is made of the stainless steel is
microfiltration of soft colloid suspension, and found that the presented in Fig. 2. The outer diameter of helical screw insert
decay of permeate flux was closely related to a rapid increase is 13 mm, and the diameter of central rod is 6 mm. The
in cake resistance and a decrease in cake porosity due to cake thickness of helix ridge is 2 mm, and the depth and width of
compression. Cabassud et al. (2001) suggested flux enhance- helical groove are 3.5 and 12 mm, respectively. The helical
ment by air sparging had a close relationship with cake screw insert is inserted centrally into the tubular membrane
properties that were clearly modified by two-phase flow in with the help of two supporters at each end of membrane
CFMF of clay suspension. module.
To alleviate the adverse effects of membrane fouling,
different types of turbulence promoters such as static mixer 2.3. Filtration unit
(Vatai et al., 2007), static rod (Chiu and James, 2006), twisted
wire-rod (Yeh and Chen, 2006), disc or doughnut shape insert A laboratory-scale filtration unit was arranged, which consists
(Liu et al., 2009) and helical screw insert (Ahmad and of a feed reservoir (10 L) thermostated at 30  C, a peristaltic
Mariadas, 2004; Bellhouse et al., 2001; Millward et al., 1995) pump and measuring equipments (pressure gauge, flow
have been widely used during various CFMF processes. Among meter). In addition, a stirrer was utilized to ensure the
these approaches, helical screw insert attracts much interest complete mixing of feed concentration.
owing to its excellent streamline shape which guarantees the The test fluid is the suspension of calcium carbonate
relatively lower energy consumption. For the same reason, particles with the median particle size (D50) of 6.18 mm and
a helical screw insert was used to improve the permeate flux density of 2700 kg/m3. During a run, both the permeate and
of membrane in this study. the retentate were recycled back to the feed reservoir to
High wall shear stress due to the presence of turbulence maintain the constant feed concentration. After filtration, the
promoter can effectively reduce the thickness of cake layer membrane was backwashed with pure water to remove the
deposited, resulting in high permeate flux of membrane. It cakes, and then was rinsed with 0.1 M hydrochloric acid (HCl)
therefore has been widely accepted that flux enhancement by solution to dissolve the particles blocked in the pores. At last,
turbulence promoter is attributed to the diminished cake the membrane was rinsed with pure water to remove the
thickness. Jokic et al. (2010) found the insertion of turbulence residual HCl solution.
promoter changed the flow pattern in the channel and In order to carry out a comparative study, filtration
increased scouring of membrane surface. They proved the experiments with and without turbulence promoter were
flux improvement was largely related to the reduction in cake
thickness during CFMF of baker’s yeast suspension. Pal et al.
(2008a,b) demonstrated that the significant enhancement of
permeate flux by incorporating a turbulence promoter was
attributed to the reduction in cake thickness during the
membrane filtration of fruit juice. However, the previous
studies ignored the effects of turbulence promoter on the
porosity and particle size of cakes, which are two important
factors affecting the cake resistance. It indicates that the
mechanism for flux enhancement by turbulence promoter
has still not been fully understood.
In this study, a helical screw insert was used as turbulence
promoter in CFMF of calcium carbonate (CaCO3) suspension.
Cake thickness, porosity and particle size were measured in
both cases with and without a turbulence promoter under the
same operating conditions to investigate the effects of
turbulence promoter on cake properties. In addition, the
effects of operating conditions on cake properties in both
cases were also studied. Fig. 1 e Pore size distribution of ceramic membrane.
 w a t e r r e s e a r c h 4 6 ( 2 0 1 2 ) 2 5 3 5 e2 5 4 4 2537

drying process mentioned above, the rest of dry cake was

weighed ðm0d Þ.
When the total mass of dry cake ðmt ¼ md þ m0d Þ and cake
porosity (3 ) were known, derived from Eq. (1), the cake thick-
ness (dc) can be estimated approximately by the following
equation (Benkahla et al., 1995).

mt mt
Fig. 2 e Configuration of a helical screw insert (unit: mm). dc ¼ ¼ (4)
rs ð1  3 ÞA rs ð1  3 ÞpdL

where A is the area of membrane, d and L are the inner

conducted under the same operating conditions including
transmembrane pressure (TMP), inlet velocity (uin) and feed
concentration (C ). TMP varies in the range of 20e70 kPa, feed
concentration in the range of 0.5e10 g/L and inlet velocity in
the range of 0.25e0.75 m/s (Re w 3700e11,200). Due to the
insertion of turbulence promoter, the crossflow velocity near
the membrane wall was greatly increased. Accordingly, the
average value of velocities was in the range of 0.49e1.68 m/s,
and corresponding value of turbulence intensities was in the
range of 8.1e15.4%.
2.4. Measurement of cake properties
Filtration experiments were conducted under certain
a desired operating condition and filter cakes were allowed to
form on the membrane surface. After a period of filtration
time, the membrane module was isolated from the filtration
unit, standing vertically for less than 2 min to drain off slurry.
Some of wet cake deposited on the tube wall was collected (in
four different directions) as quickly as possible to control the
loss of free water to a least extent. The wet cake was weighed
(mw) using a high-precision electronic balance (0.0001 g) and
then dried to a constant weight by an electric heater (>170  C,
3 h). After being cooled down to room temperature in
a vacuum desiccator, the dry cake was weighed (md).
Once the mass of the wet cake (mw) and dry cake (md) was
obtained for porosity measurement, the following equations
can be established in terms of mass balance.
Vð1  3 Þrp ¼ md
diameter and length of membrane tube.
In addition, the paralleling filtration experiments were
conducted to collect the sample of wet cake for the
measurement of particle size distribution by a centrifugal
particle size analyzer (Shimadzu, SA-CP3).
2.5. Validation experiment
An additional experiment was conducted to validate the
calculated value of cake thickness in terms of Eq. (4). The
membrane surface (inner tube wall) was covered by a piece of
filter paper (50  10 mm, 0.19 mm in thickness) before filtra-
tion. After filtration, some samples (5  10 mm) were cut off
from the filter paper that was covered by a cake layer. The
thickness of sample was measured using an optical micro-
scope (XSP-BM).
2.6. Theoretical analysis of cake resistance
According to Darcy’s law, the steady-state permeate flux (Jf)
can be calculated in terms of the membrane resistance (Rm),
cake resistance (Rc) and pore blockage resistance (Rb) (Tarleton
and Wakeman, 1993).
TMP TMP
Jf ¼ ¼ (5)
ms Rt ms ðRm þ Rc þ Rb Þ
where TMP is the transmembrane pressure, ms is the viscosity
of feed suspension, and Rt is the total hydraulic resistance
(Rt ¼ Rm þ Rc þ Rb).
In general, the cake resistance can be expressed in terms of
(1) the specific resistance (r) and cake thickness (dc).

Rc ¼ rdc (6)
V3 rl ¼ mw  md (2)
r can be described by Kozeny-Carman’s equation (Chandler
and Zydney, 2006).
Eq. (1) represents the solid content of cake and Eq. (2) indi-
cates the moisture content of cake. V is the volume of some K ð1  3 Þ2
wet cake; 3 is the cake porosity, approximately, equal to the r¼
2 (7)
jdp 33
moisture volume fraction of cake; rp is the particle density; rl
is the water density. Through solving the above equations, where K is the factor of Kozeny-Carman, j is the volumetric
cake porosity can be calculated as follows. shape factor and dp is the cake particle diameter.
When combined with Eq. (7), Eq. (6) is rewritten as follows.
ðmw  md Þrs
3 ¼ (3)
md rl þ ðmw  md Þrs K ð1  3 Þ2
Rc ¼
2 dc (8)
jdp 33
To measure the cake thickness, the total mass of dry cake
(mt) should be obtained. Therefore, the rest of wet cake
deposited on the membrane surface was carefully collected 2.7. Calculation of various hydraulic resistances
into a beaker by back-flushing with pure water. After being
settled overnight, the supernatant liquid was removed and The method to measure various hydraulic resistances was
the precipitated cake particles were collected. Repeating the widely reported elsewhere (Chandler and Zydney, 2006;
2538 w a t e r r e s e a r c h 4 6 ( 2 0 1 2 ) 2 5 3 5 e2 5 4 4

Hwang et al., 2006). According to Eq. (5), the total hydraulic Fv ¼ 3:165psw d2p (14)
resistance (Rt) can be calculated as follows.
where sw is the wall shear stress (Pa).
TMP The drag force of Ff, driving the particle toward the
Rt ¼ Rm þ Rc þ Rb ¼ (9)
ms J f membrane, can be calculated as follows.
The membrane resistance (Rm) can be calculated from the
Ff ¼ 3pms dp Jf (15)
pure water flux before filtration.
where Jf is the permeate flux (m/s).
TMP
Rm ¼ (10) The gravitational force (Fg) can be calculated as follows.
m0 J 0
1  
where J0 is the pure water flux and m0 is the viscosity of pure Fg ¼ pd3p rp  rl g (16)
6
water. When a filtration was terminated, the pure water flux
was measured after the membrane was backwashed with where rp is the particle density (kg/m3), rl is the feed density
pure water to remove the cake layer. Then Rb can be obtained (kg/m3) and g is the gravitational acceleration (9.81 m/s2).
from the difference of pure water flux between after
membrane backwashing and before filtration.

TMP TMP TMP 3. Results and discussion

Rb ¼  ¼  Rm (11)
m0 J00 m0 J 0 m0 J00
0 3.1. Mechanism of flux enhancement by turbulence
where J0 is the pure water flux after membrane backwashing.
promoter
Then Rc can be calculated from the following formula.

Rc ¼ Rt  Rm  Rb (12) 3.1.1. Effect of turbulence promoter on permeate flux

Filtration experiments in both cases with a turbulence
promoter (TP) or none (NTP) were conducted in CFMF of 1.0 g/L
2.8. Analysis of forces applied on the particle
CaCO3 suspension at the same inlet velocity (uin) of 0.5 m/s
and TMP of 50 kPa. Fig. 4 shows the variation of permeate flux
There are different forces (Altmann and Ripperger, 1997; Vyas
with filtration time. The permeate fluxes in both cases decline
et al., 2001) applied on the particle, including the back-
with filtration time, which is largely attributed to the particle
diffusion force due to the shear-induced diffusion (Fb), the
deposition on the membrane surface, increasing the hydraulic
drag force due to crossflow velocity (Fv), the drag force due to
resistance to filtration flow. The use of turbulence promoter
filtration flow (Ff), and the gravitational force due to the weight
can improve the permeate flux by an increase of 28% at
of the particle (Fg), as illustrated in Fig. 3. The back-diffusion
t ¼ 90 min. In general, flux enhancement is attributed to the
force (Fb), driving the particle away from the membrane, can
reduction in cake thickness, since the presence of turbulence
be calculated as follows.
promoter in the membrane module yields a high wall shear
0:15pms uav d3p stress, which can effectively reduce the particle deposition on
Fb ¼ (13) the membrane surface (Jokic et al., 2010; Pal et al., 2008a,b).
a2
where ms is the feed viscosity (Pa s), uav is the average cross-
3.1.2. Effect of turbulence promoter on cake thickness
flow velocity (m/s) and a is the radius of tube (m).
Fig. 5 shows the variation of cake thickness with filtration
The drag force of Fv, driving the particle away from the
time. The cake thickness in both cases increases with filtra-
membrane, can be calculated as follows.
tion time, which corresponds to the decline of permeate flux

Fig. 3 e Different forces applied on the particle. Fig. 4 e Variation of permeate flux with filtration time.
 w a t e r r e s e a r c h 4 6 ( 2 0 1 2 ) 2 5 3 5 e2 5 4 4
Fig. 5 e Variation of cake thickness with filtration time.
(Fig. 4). Differently, the thickness of NTP increases much
faster than that of TP with time. As a result, the thickness of
NTP (0.59 mm) is about six times of that of TP (0.1 mm) at
t ¼ 90 min. In order to testify the cake thickness, validation
experiment was conducted for NTP case under the same
operating condition. As one can observe from Fig. 6 that the
cake thickness measured by an optical microscope is 0.54 mm,
which is consistent with the calculated value of 0.59 mm with
a relative error of less than 7%.
The thickness of TP is greatly reduced due to the increased
wall shear stress, which seems to support the mechanism of
flux enhancement mentioned above. However, if the flux
enhancement is just attributed to the reduction in cake
thickness, according to Eqs. (5) and (6), permeate flux of TP is
supposed to be much higher than that of NTP (Fig. 4). In fact,
flux enhancement is only 28%, which is far less than expec-
tation, suggesting that the mechanism of flux enhancement
has still not been clearly understood.
Fig. 6 e Micrograph of cake thickness (unit: mm).
2539
3.1.3. Effect of turbulence promoter on cake porosity
To better understand the mechanism of flux enhancement,
various hydraulic resistances to permeation flow (at
t ¼ 90 min) were measured for both cases, as shown in Table 1.
It can be calculated that the ratio of Rc between TP and NTP is
0.58. While the ratio of dc between TP and NTP of 0.17 (Fig. 5)
suggests that the corresponding ratio of r is 3.41 (according to
Eq. (6)). The different r value for TP and NTP cases indicates
that turbulence promoter has effects on the specific resistance
of cake. The porosity and particle size are two important
factors affecting the cake specific resistance (see Eq. (7)), so
turbulence promoter should have effects on cake porosity or
particle size.
The variation of cake porosity with filtration time is pre-
sented in Fig. 7. As one can see, the cake porosity of TP is
higher than that of NTP, which is attributed to the agitating
action caused by turbulence promoter during cake formation.
The slight decay of cake porosity of NTP with time is attributed
to the cake consolidation/compaction/compression (Vyas
et al., 2000a; Wakeman, 1994) exerted by operating pressure,
which is consistent with the studies of Vyas et al. (2000a,b).
The nearly constant porosity of TP implies that the agitating
action can counteract the action of cake compression. The
presence of turbulence promoter greatly increases the turbu-
lence level of fluid flow in the membrane module under the
same operating condition. Under the circumstances, the
particles suspended in the feed are not readily deposited on
the membrane surface, forming an incompact cake layer with
relatively high porosity. The cake layer of TP is too thin to
measure the cake porosity during the initial period of filtration
(t  30 min).
3.1.4. Effect of turbulence promoter on cake particle size
Particle size distributions of cakes (t ¼ 90 min) in both cases
are shown in Fig. 8. The cake particles of NTP are normally
distributed in a wide size range of 0.5e30 mm, with D50 of
5.15 mm. The cake particles of TP are mainly distributed in
a relatively narrow size range of 0.5e3.0 mm, with D50 of
1.99 mm. The wide distribution of particle size for NTP case
implies the penetration of smaller particles through the cake
layer (Mikulasek et al., 1998; Vyas et al., 2000a) and filling of
the inter-particle voids due to the driving forces of filtration
flow. As a result, the rearrangement of cake particles leads to
the decrease in cake porosity, which can partially explain why
cake porosity of NTP is smaller than that of TP. The cake of TP
is comprised of progressively smaller particles, which can be
explained in terms of particle size classification effect
(Tarleton and Wakeman, 1993; Vyas et al., 2000a; Wakeman,
1994). During the CFMF process, the drag force of Fv (Eq.
Table 1 e Various hydraulic resistances in both cases at
t [ 90 min.
Case Membrane Cake Pore
resistance resistance blockage resistance
(Rm/1010 m1) (Rc/1010 m1) (Rb/1010 m1)
TP 6.88 6.29 1.11
NTP 6.88 10.93 0.47
2540 w a t e r r e s e a r c h 4 6 ( 2 0 1 2 ) 2 5 3 5 e2 5 4 4

Table 2 e Various cake properties in both cases at t

[ 90 min.
Case Cake thickness Cake Particle size
(dc/mm) porosity (D50/mm)
(3 )

TP 0.1 0.65 1.99

NTP 0.59 0.55 5.15

diminish the cake resistance. Eq. (7) shows that the specific
resistance is inversely proportional to the porosity and the
square of particle size. Therefore, increasing cake porosity will
decrease the specific resistance, while decreasing the particle
size will increase the specific resistance. Due to the remark-
able decrease in particle size, as a whole, the cake specific
Fig. 7 e Variation of cake porosity with filtration time. resistance is increased, which can be calculated as follows.
According to Eq. (7), the ratio of r between TP and NTP can
be calculated as
 2  3  2
(14)), which is basically responsible for driving particles away rTP 1  3 TP 3 NTP dp;NTP
¼   (17)
from the membrane, is proportional to the square of particle rNTP 1  3 NTP 3 TP dp;TP
size. It suggests that the larger the particle size, the stronger
Accordingly,
the drag force of Fv under the same operating condition. Due
to the insertion of turbulence promoter, the wall shear stress  2  3  2
rTP 1  0:65 0:55 5:15
and the drag force of Fv are greatly increased, which causes ¼   ¼ 2:45 (18)
rNTP 1  0:55 0:65 1:99
the larger particles to migrate away from the membrane
surface at a faster rate than the smaller ones. As a result, the According to Eq. (6), the ratio of Rc between TP and NTP can
smaller particles are selectively deposited on the membrane be calculated as
surface. Frankly, the smaller particle readily penetrates into    
Rc;TP rTP dc;TP
the membrane pores during the initial period of filtration, ¼  (19)
Rc;NTP rNTP dc;NTP
thereby increasing the pore blockage resistance (Rb), which
can be seen in Table 1. Accordingly,
 
Rc;TP 0:1
3.1.5. Theoretical analysis of cake resistance ¼ 2:45  ¼ 0:42 (20)
Rc;NTP 0:59
The cake properties in both cases at t ¼ 90 min are summa-
rized in Table 2. Due to the presence of turbulence promoter, Clearly, the calculated value of 0.42 is close to the experi-
the cake thickness is greatly reduced, cake porosity is slightly mental value of 0.58, but there still exists a deviation. There
increased and particle size is significantly decreased. Eq. (6) are two possible reasons accounting for this deviation. To
suggests that reducing the cake thickness can effectively estimate the thickness of cake layer, the assumption of
uniform cake was widely adopted (Benkahla et al., 1995; Vyas
et al., 2000b, 2001). Thus the average value of cake thickness
was calculated in terms of Eq. (4). Actually, cake grows from
the membrane outlet where the mass transfer coefficient is
unfavorable compared to that at membrane inlet (Bacchin,
2004; Bacchin et al., 2002), which can be explained in terms
of the concept of critical flux (Bacchin et al., 2006). And the
cake thickness is not uniform especially in the case of TP. It is
relatively thinner in the position where the helix ridge is
located and thicker in the position corresponding to the helix
groove. In addition, the constant porosity was assumed in
literatures (Benkahla et al., 1995; Vyas et al., 2000b, 2001),
whereas in reality the cake porosity may vary over its thick-
ness. In other words, the structure of a cake changes from the
portion near the membrane surface to the portion near the
cake surface which has a higher porosity.
From above analysis, it can be concluded that turbulence
promoter has both positive and negative effects on cake
resistance associated with cake properties. The positive effect
Fig. 8 e Particle size distributions of cakes. is to greatly diminish the cake thickness and to slightly
 w a t e r r e s e a r c h 4 6 ( 2 0 1 2 ) 2 5 3 5 e2 5 4 4 2541

increase the cake porosity. The negative one is the remarkable

decrease in particle size, resulting in an increase of specific
resistance. The significant reduction in cake thickness can
offset the increase in specific resistance, and consequently the
overall cake resistance is reduced and the permeate flux is
improved.

3.2. Effects of TMP on cake properties

TMP is the driving force for a pressure-driven membrane

filtration process. Therefore, TMP has important effects on the
cake properties during CFMF of particulate suspensions. An
increased TMP serves to expedite the particle deposition on
the membrane surface. According to the Stoke’s equation (Eq.
(15)), the drag force of filtration flow (Ff), driving the particles
in the feed toward the membrane surface, is proportional to
the permeate flux (Jf). The increased TMP yields high permeate
flux of membrane and thus increases the force of Ff, causing
the additional particle deposition on the membrane surface.
The cake thickness therefore increases with increasing TMP,
as illustrated in Fig. 9, which is consistent with the studies
reported by Wakeman (1994) and Vyas et al. (2000a,b). The
sharp increase in cake thickness of NTP indicates the strong
effect of TMP on the growth of cake layer, which diminishes
the effect of driving force on permeate flux. Fortunately, the
use of turbulence promoter can alleviate to a great extent the
effect of TMP on the particle deposition due to the increased
wall shear stress. As a result, the cake thickness of TP at TMP
of 70 kPa (0.21 mm) is even smaller than that of NTP at TMP of
20 kPa (0.25 mm).
The high TMP apparently tends to strengthen the action of
cake compression, leading to formation of tightly packed
cakes. The cake porosity therefore decreases with increasing
TMP, as shown in Fig. 10. However, the studies of Vyas et al.
(2000a,b) suggested TMP had no significant effect on the
cake porosity during the CFMF of lactalbumin suspension,
which was not consistent with the general rule. Fig. 10 shows
that cake porosity of TP is higher than that of NTP at the same
TMP, which is attributed to the agitating action on cake
formation caused by turbulence promoter. The agitating
action is so strong that it even can alleviate the effect of TMP
Fig. 9 e Effects of TMP on cake thickness.
Fig. 10 e Effects of TMP on cake porosity and particle size.
on cake compression. As a result, the cake porosity of TP at
TMP of 70 kPa is even larger than that of NTP at TMP of 30 kPa.
The particle deposition on the membrane surface is
controlled by the interaction between Fv and Ff under a certain
operating conditions. It also determines the particle size of
cakes deposited on the membrane. Generally, increasing the
force of Ff due to high TMP tends to increase the average
particle size of cakes, since the effect of particle size classifi-
cation is weakened. The particle size of cake therefore
increases with increasing TMP, as illustrated in Fig. 10, which
is consistent with the studies of Vyas et al. (2000a,b) The
particle size of TP is smaller than that of NTP at the same TMP
due to the insertion of turbulence promoter which can
increase the crossflow velocity and thus the effect of particle
size classification is enhanced.
3.3. Effects of inlet velocity on cake properties
To investigate the effect of inlet velocity on the cake proper-
ties, filtration experiments were conducted in the inlet
velocity range of 0.25e0.75 m/s. Increasing the inlet velocity
normally increases the wall shear stress and thus the force of
Fv, which drives additional particles away from the
membrane. The cake thickness therefore decreases with an
increase of inlet velocity. Our experimental results are
consistent with the general rule, as illustrated in Fig. 11. When
the inlet velocity increases from 0.25 to 0.75 m/s, the cake
thickness reduces by 26% and 66% for NTP and TP cases,
respectively. Obviously, the use of turbulence promoter can
augment the effect of inlet velocity on diminishing the cake
thickness. The thickness of TP at inlet velocity of 0.25 m/s is
even thinner than that of NTP at inlet velocity of 0.75 m/s. It
indicates that high flux enhancement can be achieved by
turbulence promoter at relatively low inlet velocity.
Increasing the crossflow velocity increases the shearing
action on the membrane surface. It serves to not only prevent
the particle deposition, but also alleviate the action of cake
compression (Mikulasek et al., 1998), which changes the cake
structure during cake formation, leading to the formation of
relatively incompact cakes. The cake porosity therefore
increases with increasing crossflow velocity, as shown in
2542 w a t e r r e s e a r c h 4 6 ( 2 0 1 2 ) 2 5 3 5 e2 5 4 4
Fig. 11 e Effects of inlet velocity on cake thickness.
Fig. 12. However, the studies of Vyas et al. (2000a,b) suggested
crossflow velocity had on significant effects on the cake
porosity and thickness, which were not consistent with the
general rule. It was attributed to the fact that they adopted
a dead-end filtration mode where no shearing action works
during the cake reconstruction. The cake porosity of TP is
higher than that of NTP at the same inlet velocity, which is
owing to the strong agitating action on the cake formation
caused by turbulence promoter.
Apparently, the effect of particle size classification is
augmented when the inlet velocity increases during CFMF
process. The particle size of cake therefore decreases with
increasing inlet velocity, as illustrated in Fig. 12, which is
consistent with the studies of Vyas et al. (2000a,b). The smaller
particles are prone to the blockage of membrane pores due to
the ready penetration, thereby increasing the filtration resis-
tance, which is the negative effect of inlet velocity to improve
the permeate flux. The pore blockage seems to be more severe
when using a turbulence promoter. The particle size of TP is
smaller than that of NTP under the same inlet velocity since
the effect of particle size classification is enhanced by turbu-
lence promoter.
Fig. 12 e Effects of inlet velocity on cake porosity and
particle size.
3.4. Effects of feed concentration on cake properties
Increasing the concentration of feed suspension normally
stimulates the particle deposition on membrane surface,
which is confirmed by the fact that cake thickness of NTP
sharply increases with increasing feed concentration, as
illustrated in Fig. 13. The use of turbulence promoter can
greatly alleviate the effect of feed concentration on particle
deposition. Consequently, the cake thickness of TP slightly
increases with feed concentration. The thickness of TP at feed
concentration of 10 g/L (0.22 mm) is even thinner than that of
NTP at feed concentration of 0.5 g/L (0.38 mm), indicating that
better flux enhancement can be achieved by turbulence
promoter at higher feed concentration.
The effects of feed concentration on cake porosity are
presented in Fig. 14. As one can see, the porosity increases
with feed concentration in NTP case, while it almost keeps
constant in TP case. The increasing porosity of NTP is attrib-
uted to the remarkable increase in cake thickness with feed
concentration (Fig. 13). For example, cake thickness of NTP is
1.91 mm at feed concentration of 10 g/L, about 5 times of that
at feed concentration of 0.5 g/L. The thicker cake layer can
weaken to some extent the effect of cake consolidation exer-
ted by operation pressure, resulting in an increase of porosity.
The constant porosity of TP is largely attributed to the strong
agitating action during cake formation. Although the cake
thickness of TP slightly increases with feed concentration
(Fig. 13), it is still too thin (0.22 mm, at 10 g/L) to affect the
action of cake consolidation.
Fig. 14 also shows a descending trend of particle size with
feed concentration. There are two possible reasons
accounting for it. On the one hand, the cross-section of
membrane tube is constricted owing to the cake particle
deposition (especially at high TMP), thereby increasing the
crossflow velocity, which enhances the effect of particle size
classification. On the other hand, the feed suspension of high
concentration provides a great amount of small particles that
tends to selectively deposit on the membrane surface. The use
of turbulence promoter can promote the effect of particle size
classification during CFMF of particulate suspensions under
the same operating condition. Therefore, the cake particle size
in TP case is much smaller than that in NTP case.
Fig. 13 e Effects of feed concentration on cake thickness.
 w a t e r r e s e a r c h 4 6 ( 2 0 1 2 ) 2 5 3 5 e2 5 4 4 2543

Appendix A

Symbols
A effective area of membrane, m2
a radius of membrane tube, m
dp particle diameter, m
Fb back-diffusion force due to the shear-induced
diffusion, N
Ff drag force due to the filtration flow, N
Fg gravitational force, N
Fv drag force due to the crossflow velocity, N
J0 pure water flux, L/(m2 h)
0

Fig. 14 e Effects of feed concentration on cake porosity and J0 pure water flux after membrane backwashing, L/(m2 h)
particle size. Jf permeate flux of membrane, L/(m2 h)
K factor of Kozeny-Carman
md mass of dry cake, kg
mw mass of wet cake, kg
mt total mass of dry cake, kg
Rb pore blockage resistance, m1
4. Conclusions
Rc cake resistance, m1
Rm membrane resistance, m1
Turbulence promoter has important effects on cake properties
Rt total hydraulic resistance, m1
that directly affect the cake resistance. The cake thickness is
r cake specific resistance, m2
greatly reduced due to the increased wall shear stress caused
TMP transmembrane pressure, Pa
by turbulence promoter, which serves to diminish the cake
uav average crossflow velocity, m/s
resistance. The slight increase in cake porosity, due to the
V volume of wet cake, m3
strong agitating action during cake formation, is also positive
to diminish the cake resistance. Nevertheless, the particle size Greek letters
of cake is significantly diminished due to the effect of particle dc cake thickness, mm
size classification, leading to the increase of cake resistance. 3 cake porosity
As a whole, the overall cake resistance can still be reduced by ms viscosity of feed suspension, Pa s
turbulence promoter and therefore permeate flux is improved. m0 viscosity of pure water, Pa s
The theoretical analysis of cake resistances is consistent with rp particle density, kg/m3
experimental results. rl water density, kg/m3
Cake properties in both cases almost exhibit the similar sw wall shear stress, Pa
trends under various operating conditions. The thickness j volumetric shape factor
increases, porosity decreases and particle size of cake
increases with increasing TMP in both cases. Differently, the
use of turbulence promoter can greatly alleviate the effects of references
TMP on particle deposition and cake consolidation. The
thickness decreases, porosity increases and particle size of
cake decreases with increasing inlet velocity in both cases. Ahmad, A.L., Mariadas, A., 2004. Baffled microfiltration
Differently, the use of turbulence promoter can intensify the membrane and its fouling control for feed water of
effects of inlet velocity on particle deposition and the effect of desalination. Desalination 168, 223e230.
particle size classification. The thickness increases and Al-Malack, M.H., Anderson, G.K., 1997. Use of crossflow
microfiltration in wastewater treatment. Water Research 31
particle size of cake decreases with increasing feed concen-
(12), 3064e3072.
tration in both cases. Differently, the use of turbulence
Altmann, J., Ripperger, S., 1997. Particle deposition and layer
promoter can greatly weaken the effects of feed concentration formation at the crossflow microfiltration. Journal of
on particle deposition. Membrane Science 124 (1), 119e128.
Bacchin, P., 2004. A possible link between critical and limiting flux
for colloidal systems: consideration of critical deposit
formation along a membrane. Journal of Membrane Science
Acknowledgments 228 (2), 237e241.
Bacchin, P., Aimar, P., Field, R.W., 2006. Critical and sustainable
fluxes: theory, experiments and applications. Journal of
The authors appreciate the financial support of National
Membrane Science 281 (1e2), 42e69.
Science Fund for Distinguished Young Scholars of China Bacchin, P., Si-Hassen, D., Starov, V., Clifton, M.J., Aimar, P., 2002.
(Grant No. 21125628) and Open Fund of Key Laboratory in A unifying model for concentration polarization, gel-layer
Beijing University of Technology. formation and particle deposition in cross-flow membrane
2544 w a t e r r e s e a r c h 4 6 ( 2 0 1 2 ) 2 5 3 5 e2 5 4 4
filtration of colloidal suspensions. Chemical Engineering
Science 57 (1), 77e91.
Bellhouse, B.J., Costigan, G., Abhinava, K., Merry, A., 2001. The
performance of helical screw-thread inserts in tubular
membranes. Separation and Purification Technology 22e23
(1e3), 89e113.
Benkahla, Y.K., Oulddris, A., Jaffrin, M.Y., Hassen, D.S., 1995. Cake
growth-mechanism in cross-flow microfiltration of mineral
suspensions. Journal of Membrane Science 98 (1e2), 107e117.
Cabassud, C., Laborie, S., Durand-Bourlier, L., Laine, J.M., 2001. Air
sparging in ultrafiltration hollow fibers: relationship between
flux enhancement, cake characteristics and hydrodynamic
parameters. Journal of Membrane Science 181 (1), 57e69.
Carroll, T., Meier-Haack, J., Booker, N.A., 2003. A permeability-
controlled microfiltration membrane for reduced fouling in
drinking water treatment. Water Research 37 (3), 585e588.
Chandler, M., Zydney, A., 2006. Effects of membrane pore
geometry on fouling behavior during yeast cell microfiltration.
Journal of Membrane Science 285 (1e2), 334e342.
Chiu, T.Y., James, A.E., 2006. Effects of axial baffles in non-circular
multi-channel ceramic membranes using organic feed.
Separation and Purification Technology 51 (3), 233e239.
Geissen, S.U., Xi, W.M., 2001. Separation of titanium dioxide from
photocatalytically treated water by cross-flow microfiltration.
Water Research 35 (5), 1256e1262.
Hwang, K.J., Hsueh, C.L., 2003. Dynamic analysis of cake
properties in microfiltration of soft colloids. Journal of
Membrane Science 214 (2), 259e273.
Hwang, K.J., Chou, F.Y., Tung, K.L., 2006. Effects of operating
conditions on the performance of cross-flow microfiltration of
fine particle/protein binary suspension. Journal of Membrane
Science 274 (1e2), 183e191.
Jokic, A., Zavargo, Z., Seres, Z., Tekic, M., 2010. The effect of
turbulence promoter on cross-flow microfiltration of yeast
suspensions: a response surface methodology approach.
Journal of Membrane Science 350 (1e2), 269e278.
Kimura, K., Yamamura, H., Chae, S., Watanabe, Y., 2007.
Transition in fouling mechanism in microfiltration of
a surface water. Water Research 41 (17), 3812e3822.
Lee, C.H., Lee, S.H., Park, P.K., Kim, J.H., Yeon, K.M., 2008.
Analysis of filtration characteristics in submerged
microfiltration for drinking water treatment. Water Research
42 (12), 3109e3121.
Li, Q.L., Wang, S., Liu, C., 2011. Fouling of microfiltration
membranes by organic polymer coagulants and flocculants:
controlling factors and mechanisms. Water Research 45 (1),
357e365.
Lin, C.F., Lin, A.Y.C., Chandana, P.S., Tsai, C.Y., 2009. Effects of
mass retention of dissolved organic matter and membrane
pore size on membrane fouling and flux decline. Water
Research 43 (2), 389e394.
Liu, Y.F., He, G.H., Liu, X.D., Xiao, G.K., Li, B.J., 2009. CFD
simulations of turbulent flow in baffle-filled membrane tubes.
Separation and Purification Technology 67 (1), 14e20.
Mikulasek, P., Wakeman, R.J., Marchant, J.Q., 1998. Crossflow
microfiltration of shear-thinning aqueous titanium dioxide
dispersions. Chemical Engineering Journal 69 (1), 53e61.
Millward, H.R., Bellhouse, B.J., Walker, G., 1995. Screw-thread flow
promoters e an experimental study of ultrafiltration and
microfiltration performance. Journal of Membrane Science 106
(3), 269e279.
Ould-Dris, A., Jaffrin, M.Y., Si-Hassen, D., Neggaz, Y., 2000. Effect
of cake thickness and particle polydispersity on prediction of
permeate flux in microfiltration of particulate suspensions by
a hydrodynamic diffusion model. Journal of Membrane
Science 164 (1e2), 211e227.
Pal, S., Ambastha, S., Ghosh, T.B., De, S., DasGupta, S., 2008a.
Optical evaluation of deposition thickness and measurement
of permeate flux enhancement of simulated fruit juice in
presence of turbulence promoters. Journal of Membrane
Science 315 (1e2), 58e66.
Pal, S., Bharihoke, R., Chakraborty, S., Ghatak, S.K., De, S.,
DasGupta, S., 2008b. An experimental and theoretical analysis
of turbulence promoter assisted ultrafiltration of synthetic fruit
juice. Separation and Purification Technology 62 (3), 659e667.
Tarleton, E.S., Wakeman, R.J., 1993. Understanding flux decline in
cross-flow microfiltration. 1. Effects of particle and pore-size.
Chemical Engineering Research and Design 71 (A4), 399e410.
Vatai, G.N., Krstic, D.M., Hoflinger, W., Koris, A.K., Tekic, M.N.,
2007. Combining air sparging and the use of a static mixer in
cross-flow ultrafiltration of oil/water emulsion. Desalination
204 (1e3), 255e264.
Vyas, H.K., Bennett, R.J., Marshall, A.D., 2000a. Influence of
operating conditions on membrane fouling in crossflow
microfiltration of particulate suspensions. International Dairy
Journal 10 (7), 477e487.
Vyas, H.K., Bennett, R.J., Marshall, A.D., 2001. Cake resistance and
force balance mechanism in the crossflow microfiltration of
lactalbumin particles. Journal of Membrane Science 192 (1e2),
165e176.
Vyas, H.K., Mawson, A.J., Bennett, R.J., Marshall, A.D., 2000b. A
new method for estimating cake height and porosity during
crossflow filtration of particulate suspensions. Journal of
Membrane Science 176 (1), 113e119.
Wakeman, R.J., 1994. Visualization of cake formation in cross-
flow microfiltration. Chemical Engineering Research and
Design 72 (A4), 530e540.
Walker, H.W., Chen, D., Weavers, L.K., 2006. Ultrasonic control of
ceramic membrane fouling: effect of particle characteristics.
Water Research 40 (4), 840e850.
Yeh, H.M., Chen, Y., 2006. Momentum balance analysis of
permeate flux for ultrafiltration in tubular membranes with
gradually increasing incidental angles of a wired-rod insert.
Journal of Membrane Science 278 (1e2), 205e211.
Zhang, J.Z., Zhou, J.T., Fane, A.G., Liu, Y., 2010. A comparison of
membrane fouling under constant and variable organic
loadings in submerge membrane bioreactors. Water Research
44 (18), 5407e5413.