Arab J Sci Eng (2018) 43:2237–2245

https://doi.org/10.1007/s13369-017-2645-5

RESEARCH ARTICLE - CHEMICAL ENGINEERING

CFD Simulation of the Distribution of Pressure and Shear Rate

on the Surface of Rotating Membrane Equipped with Vanes
for the Ultrafiltration of Dairy Effluent
Soufyane Ladeg1 · Zhenzhou Zhu2 · Nadji Moulai-Mostefa1 ·
Luhui Ding2 · Michel Y. Jaffrin3

Received: 14 May 2016 / Accepted: 7 June 2017 / Published online: 17 June 2017
© King Fahd University of Petroleum & Minerals 2017

Abstract This paper reports computational fluid dynamics Keywords Ultrafiltration · Dairy effluent · Rotating disk ·
(CFD) studies on the performance of flow inside a rotating Pressure distribution · Shear stress · CFD
disk filtration module. This module consists of a disk rotat-
ing at speeds between 1000 and 3000 rpm inside a cylindrical
housing equipped with a stationary circular flat membrane. 1 Introduction
In order to investigate the influence of surface geometry, a
smooth disk and a disk equipped with vanes were used. Com- In recent years, filtration in rotating disk modules (RDM) has
mercial CFD software was utilized for the characterization of received attention by researchers and engineers. Dynamic
the flow inside the membrane module. The predicted results ultrafiltration by RDM can be applied to a wide variety of
were compared at different values of rotating and inlet veloc- applications such as chemical industry, food technology and
ities with previous works and also with experimental ones water treatment [1,2].
obtained in the filtration of dairy effluent in the same con- The main problem of membrane filtration is fouling,
ditions. The influence of the geometry of the disk was also caused by particles and macromolecules contained in the feed
evaluated. The obtained results show that an increase in rotat- solution, which blocks the pores or build up a cake over the
ing velocity leads to an increase in the pressure and shear membrane, reducing it’s permeability [3,4].
stress at the membrane surface. The highest pressures were Fouling can be considered to be completely external if
obtained when using disks equipped with vanes, this was due no particles are expected to penetrate the pores [5]. To
to the turbulent area created between the vanes visualized in avoid these phenomena, several methods were proposed by
animated views. It was also found that the presence of vanes researchers, namely the use of static turbulence promoters
on the surface of disk has a great effect on the shear stress; by placing inserts in the flow channel, or creating pulsa-
it contributes to the reduction in fouling and improvement in tions of feed flow [6,7]. Also, the flux can be enhanced by
filtration flux. application of RDM or vibrating membranes [8,9]. As conse-
quence, the membrane fouling and cake layer thickness can
be reduced by increasing the shear rate at the membrane sur-
face. It was demonstrated in ultrafiltration with RDM that the
flux is mainly governed by the maximum shear rate and can be
increased to considerably high levels by increasing the rota-
B Soufyane Ladeg tion speed or by equipping the disk with radial vanes [10–15].
[email protected]
It was observed that the presence of vanes on the surface disk
1 Materials and Environmental Laboratory, University permits a complete elimination of fouling and produces high
of Medea, Ain D’Heb 26001, Medea, Algeria permeate flux. The high performance of dynamic ultrafiltra-
2 EA 4297 TIMR, Technological University of Compiegne, tion was confirmed by Brou et al. [16]. They demonstrated
60205 Compiegne Cedex, France that the addition of vanes to the membrane disk improves
3 UMR 7338, Technological University of Compiegne, the permeate flux and reduces the consumed energy. They
60205 Compiegne Cedex, France observed that the increase in permeate flux with the 6 mm

123
2238 Arab J Sci Eng (2018) 43:2237–2245

vanes is higher than the power augmentation. Li et al. [17] secondly, to find the distributions of pressure and shear stress
found for ultrafiltration of emulsions that a disk equipped at the membrane and disk surface. Therefore, different condi-
with 6 mm vanes produces higher shear rate and reduces the tions of rotation speed, inlet velocity and membrane surface
concentration polarization at the membrane surface. They nature were selected for this study. In order to increase the
also demonstrated that during tests of concentration, the spe- shear rate in the module, the disk was equipped with radial
cific energy consumption was less for a disk with vanes than vanes. This geometry leads to an increase in the fluid angu-
with a smooth disk. lar velocity in the gap between the membrane and disk [28].
Computational fluid dynamics (CFD) are a powerful tool The effects of internal hydrodynamics of the filtration system
to investigate processes involving fluid flow [18]. CFD and on its performances were evaluated using CFD. Two types
models of flow through RDM and tubular membranes were of disks (with vanes and without vanes) were applied for
investigated by several researchers. Torras et al. [19] used the ultrafiltration of dairy effluent. This fluid was chosen in
numerical simulation to show the characteristics of the flow order to compare the obtained results with those predicted by
inside a dynamic membrane module with a smooth rotating a theoretical model proposed by Bouzerar et al. [5] and, also
disk. They determined the correlation relating the disk rota- with experimental values and numerical results obtained by
tion speed with the surface-averaged pressure and the shear Torras et al. [19] using CFD.
stress on the membrane. Parvareh et al. [20] studied the role
of fluid flow pattern and pressure on the permeate flux. They
found that the change in dynamic pressure upon the mem-
2 Materials and Methods
brane surface has a direct effect on the permeate flux. Rainer
et al. [21] examined the 3D-flow regime of a crossflow filter
2.1 Experimental Device
with horizontal rotating disks. They used a commercial CFD
program for evaluating the influence of scrapers on velocity,
A rotating disk membrane (RDM) was used for the ultrafil-
turbulence and dissipation. Li et al. [22] studied the influ-
tration of dairy effluent (viscosity 9.8 10−4 Pa s and density
ence of the geometry of the membrane on the flow rate using
of 924 kg/m3 ). The RDM is fed by an annular inlet and the
CFD. They found that a variation in structural configura-
fluid exits through the retentate on the axis. The flat mem-
tion causes an inherent change in hydrodynamic behaviour.
brane was attached on the cover of the cylindrical housing
Rahimi et al. [23] have used CFD for predicting the shear
in front of a disk. Two types of disks were used, one with
stress upon a polyvinylidene difluoride (PVDF) membrane.
vanes and another without vanes (smooth disk). Therefore,
They found by comparing the experimental observations with
the simulations were performed on the same physical model
CFD modelling that a high shear stress reduces fooling. Liu
(Fig. 1).
et al. [24] simulated the fluid flow in baffle-filled membrane
The control volume has an annular inlet and an exit tube;
tubes. Their simulation results showed an increase in shear
this tube has a length equal to 40 mm (two diameters). This
stress and average velocity, caused by the presence of baf-
length was taken in order to avoid the influence of the exit
fles, and therefore an increase in permeate flux. Ghidossi et
[17].
al. [25] used a commercial CFD package to determine the
Three inlet fluid velocities (0.017, 0.0011 and 0.009 m/s)
pressure drop in a hollow fibre membrane at different oper-
were considered. For the first velocity (0.017 m/s), three
ating conditions. They compared their experimental results
different rotating velocities of the disk (1000, 2000 and
with predicted results to check the numerical model. Liu et al.
3000 rpm) were selected. For the second and last velocity,
[26] studied the performance of screw inserts on microfiltra-
only a rotating velocity of 2000 rpm was used.
tion using CFD and experimental study. Their experimental
The radius of the disk for both models is 72.5 mm, the gap
results showed that turbulence promoters can enhance the
between the disk and membrane is 10.7 mm. Vanes height
permeate flux of membrane, consequently improve permeate
is 6 mm; their length is 65.5 mm and have three different
flux. Therefore, they used CFD to confirm the efficiency of
thicknesses 1.2, 2 and 4 mm.
screw inserts for improving permeate flux. Recently, Hwang
and Wu [27] studied the effect of shape and number of vanes
equipping membranes on the efficiency of microalgae micro- 2.2 Software CFD Description
filtration using CFD. They found that the disk structure and
rotation speed were the most crucial factors affecting the Commercial software (Ansys fluent version 14.5) was used
filtration performance. A rotating disk equipped with more in this study. A geometry and mesh in Ansys interface were
vanes can generate higher shear stress on the membrane sur- created. The following boundary conditions were considered:
face for a given rotation speed. (i) the inlet velocity was imposed in inlet annular section,
The aims of this study were, first to determine the pressure and outflow at the retentate exit, (ii) the prescribed pressure
drop in a module depending on the operating conditions and in the outflow was maintained at zero, (iii) the velocity on

123
Arab J Sci Eng (2018) 43:2237–2245 2239

Fig. 1 Experimental and simulated RDM models. a Disk with vanes, b disk without vanes

the housing and inlet annular wall and on the membrane was the surface of the disk can be estimated by using Bernoulli’s
taken equal to zero. equation:
The fluid flow in this simulation was assumed to be steady-
state regime, and the standard κ-omega turbulence model was P(r ) = P0 + 0.5ρk 2 ω2 r 2 (1)
selected.
In order to reproduce the element shape, the structured where P0 is the pressure at the symmetry axis (r = 0), ρ
net model has 1520614 tetrahedron elements and 280865 the density of the fluid, ω the angular velocity, r the radial
nodes. This mesh was checked using the quality report option position from the centre of the disk and k the velocity factor
(Ansys, Fluent option). This configuration is different from which is equal to 0.42 for the smooth disk and 0.84 for disk
that used by Torras et al. [19]. The difference between with vanes [11].
meshes was in the shape of elements and their number. The wall shear stress is induced by the transverse veloc-
The results are similar for both hexagonal and tetrahedron ity gradient; it tends to reduce fouling during the filtration
meshes, especially in the peripheral pressure. However, tetra- step. Assuming Newtonian fluid, and a cylindrical system,
hedron meshes are easy to use and give the best orthogonal the tangential and radial components of the wall shear stress
quality in complicated geometry than hexagonal meshes. can be defined as [29]:
 
dVθ
τtg = μ (2)
2.3 Theoretical Background dz z=0
 
dVr
Several authors [5,10,11,28] have described the flow in the τrad = μ (3)
dz z=0
inviscid region (between disk and membrane) with an angular
velocity kω, where the constant k depends on the geometry where Vθ and Vr are, respectively, the tangential and radial
of the disk. They suppose that the flow is generally turbulent velocities, z the axial distance and μ the dynamic viscosity
at speeds higher than 500 rpm. The pressure distribution on of the fluid.

123
2240 Arab J Sci Eng (2018) 43:2237–2245

Fig. 2 Comparison of pressure distributions along the membrane radius deduced by CFD simulation with theoretical and experimental results
(ultrafiltration of milk at a rotating velocity of 2000 rpm)

The shear stress is imposed by the azimuthal velocity gra- when the radius is less than 0.05 m. The curve behaviour is
dient across the boundary layer which is a function of radial parabolic; it is similar to the curve representing the variations
position and rotational speed [28], and thus the shear stress of the theoretical pressure at k = 0.42. This agreement con-
at the stationary membrane in laminar regime is given by: firms that the selected model correctly represents the dynamic
  system of filtration. On the other hand, the experimental value
dVθ 3 1 of pressure agrees well with the predicted values for a smooth
τwl = μ = 0.77ρr (kω) 2 υ 2 . (4)
dz z=0 disk with CFD.
The pressure distribution is higher for a disk with vanes
where ν is the fluid kinematic viscosity. as the coefficient k is equal to 0.84 against 0.42 for a smooth
Bouzerar et al. [5] proposed in the turbulent regime a rela- disk. In order to fit exactly the simulation data, k must be
tion for the determination of the shear rate on the stationary close to 0.95 for disk with vanes, and for smooth disk, it
membrane: must be close to 0.65.
For the disk with vanes, the behaviour of the curve is
γwt (r ) = 0.0296 ν −0.8 (kω)1.8 r 1.6 (5) not parabolic implying that Eq. 1 may not govern the dis-
tribution of pressure at the surface of membrane in this
case. The results also show that the rotating disk filtra-
3 Results and Discussion tion system equipped with vanes generates higher shear rate
than the smooth disk. The presence of vanes on the disk
3.1 Pressure Distribution surface permits a better drive of fluid and consequently a
higher shear according to Eq. 3. As a result, the formation
In the first time, the numerical simulation was checked by of deposit or cake on the surface of membrane is limited.
comparing the obtained results using the smooth disk (with- These observations are in agreement with those obtained
out vanes) with those obtained by Torras et al. [19] in the same experimentally in the ultrafiltration of colloidal solutions
conditions. These results were also compared with exper- [14,15,17]. It was noticed that the permeate flux increases
imental values and those predicted by a theoretical model when the smooth disk is replaced by a disk equipped with
governed by Eq. 1. Fig. 2 plots pressure variations with radius 6 mm vanes.
for both disks with vanes and without at a rotating velocity of Figure 3 compares pressure variations as function of radius
2000 rpm and an inlet velocity of 0.017 m/s, respectively. It for disks with and without vanes. We noted that the pressure
must be noted that the obtained curves for the smooth disk are increases along the radius and the effect of vanes is clearly
close with a slight difference for a radius between 0.06 and observed as shown in Fig. 3a. On the other hand, the pressure
0.08 m. However, the pressure values are slightly different increases with the radial coordinate because of an increase in

123
Arab J Sci Eng (2018) 43:2237–2245 2241

Fig. 3 Profiles of pressure distribution on the membrane surface at rotating velocity of 2000 rpm and inlet velocity of 0.017 m/s. a Disk with vanes,
b disk without vanes

Fig. 4 Influence of the presence of vanes and rotating velocity on the variation of pressure across the membrane radius. a Disk with vanes, b disk
without vanes

the rotating velocity and it reached maximum values near the vanes) at three rotating velocities (1000, 2000 and 3000 rpm).
disk tip. The scale shown on the left indicates that the pressure It was noted that an increase in rotating velocity has an impor-
varied from 0 to 1.078 bar when the disk with vanes is used; tant effect on the pressure distribution in comparison with the
nevertheless when the smooth disk is used, it ranged from 0 effect of inlet velocity which exerts a small effect on pressure
to 0.54 bars (Fig. 3b). In this case, the magnitudes of pres- across the radius. This is due to the centrifugal force gener-
sure were smaller than that generated by a disk with vanes. ated by the disk which acts on the fluid inside the module due
The pressure distribution on the membrane surface was sym- to the disk rotation. This force acting on the fluid between the
metrical in the form of rings in the smooth disk, whereas, in disk and membrane is proportional to the square of velocity
the disk with vanes, the ring areas were not symmetric. Lee and radial position [31].
and Clark [30] showed that an increasing of transmembrane
pressure resulted in a denser cake layer, which increases the Fv = ρ r ω2 (6)
specific cake resistance. Thus, we can infer in our case that
the cake resistance will be denser at the rim of membrane In addition, it was also observed for different rotating
than that in centre of membrane. velocities, that the presence of vanes leads to an increase
Figure 4 shows the variation of pressure with radius for in pressure for the whole used velocities. Jaffrin [32] eval-
both disks (A for a disk with vanes and B for a disk without uated the influence of different hydrodynamic parameters

123
2242 Arab J Sci Eng (2018) 43:2237–2245

Fig. 5 Vector velocity distribution on the control volume at an inlet velocity of 0.017 m/s. a Disk with vanes, b disk without vanes

on the permeate flux obtained by two different rotating disk as it reached a maximum of 940 Pa for the disk with vanes at
ultrafiltration systems. The comparison of the experimental 3000 rpm and decreased to 360 Pa at R = 0.08 m. The max-
results showed that the permeate flux could be augmented imum shear stress was 300 Pa at R = 0.076 m and 3000 rpm
to very high levels by increasing the rotation velocity or by for a disk without vanes.
equipping the disk with large vanes. Hence, the presence of This trend can be explained by the triangular area of high
vanes reduced the reversible fouling when the velocity was shear stress, formed on the surface of membrane in radii var-
increased. ied between 0.056 and 0.066 m due to the presence of vanes
In Fig. 5, we represent the tangential velocity vector dis- (Fig. 7a) and behind this area in the tip of membrane there
tribution on the control volume at a rotating velocity of is a ring area at less shear stress. In disk without vanes, the
2000 rpm and inlet velocity of 0.017 m/s for the two model ring area of high shear stress was observed in tip of mem-
disks. The highest vector velocity was observed at the sur- brane (Fig. 7b). These results were confirmed by Bouzerar
face for radius values ranging between 0.04 and 0.08 m, et al. [5]; they found that the local permeate flux is higher at
associated with the variation of rotating velocity. However, the periphery than in the centre part and they explained their
when approaching the centre of the membrane, the vector results by the high shear stress located at tip of membrane.
velocity decreases. In the case of disk with vanes, we can The effect of vanes is clearly observed in Fig. 7a, where
notice that the area of the highest velocity is wider than in eight areas in triangular forms at great shear stress (marked
the case of smooth disk. Most of the membrane surface is in yellow) appear. The scale shown in the left of figure indi-
coloured in blue in the absence of vanes (Fig. 5b), while in cates that the values of shear stress increase from the centre
the case of the disk with vanes the surface of membrane is to membrane edge. A comparison between Fig. 7a, b, indi-
coloured in yellow and green (Fig. 5a). Thus, the vanes may cates that the shear stress in the disk with vanes reached
create a difference in the tangential velocity by decreasing values ranged between 474.8 and 542.6 Pa (triangular areas
the gap between the disk and membrane. Hwang and Lin marked in yellow) for a range of radii between 0.056 and
[33] found that the shear stress can be increased by reduc- 0.066 m. But for a disk without vanes, the maximal shear
ing the gap width between the disk and membrane. On the stress is between 144.1 and 162.1 Pa (marked in sky blue)
other hand, Bouzerar et al. [28] confirmed that, for the best and it is located at the tip of membrane (0.07–0.078 m). We
performance, the rotating disk should be operated at high can note that the vanes had created a high shear stress that
speeds; they proposed an estimation of 20 m/s for azimuthal progresses from the tip of the membrane towards the centre.
velocity. To reach this azimuthal velocity, the rotating disk In the tip of membrane, there is also a high pressure which
should be operated at a rotating velocity between 2000 and compresses the cake layer and makes it less permeable and
2500 rpm. more resistant to erosion by shear stress [30]. These observa-
tions are in agreement with those obtained by Hwang and Wu
3.2 Shear Stress Distribution [27], and Hwang and Lin [33]. They found that an increase
in shear stress leads to a smaller cake mass because of the
Figure 6 shows the variation of shear stress with radius and significant sweeping effect occurring at the membrane sur-
rotation speed. The shear stress does not behave as pressure face.

123
Arab J Sci Eng (2018) 43:2237–2245 2243

Fig. 6 Effect of the presence of vanes on the profiles of shear stress distribution across the membrane radius at different rotating velocities. a Disk
equipped with vanes, b disk without vanes

Fig. 7 Profiles of the shear stress distribution on the membrane surface at a rotating velocity of 2000 rpm. a Disk with vanes, b disk without vanes

3.3 Effect of the Vane Width on the Wall Shear Stress R = 0.08 m. The other thicknesses give similar shear stress
versus radius except for about 200 Pa between 0.036 and
In this section, we have varied the width of vanes to 0.05 m.
evaluate its effect on the distribution of pressure and We can also notice this difference in Fig. 9. For a disk
shear stress at the surface of membrane. For this pur- equipped with vanes of 4 mm width, the green region (Fig. 9a)
pose, three thicknesses (1.2, 2 and 4 mm) were selected takes slightly a rectangular form. However, when the width
where the first one represents the actual width of vane passes from 4 to 1.2 mm (Fig. 9b), it appeared in a triangular
and the second represents the initial width of vane when form. This means that the surface of green region created
it has been fabricated [28]. Figure 8 shows the increase between vanes has been slightly widened. Thus, the widen-
in shear stress due to higher vanes thickness. A maxi- ing of vanes from 1.2 to 4 mm widened slightly the surface
mum of shear stress of 450 Pa is reached at R = 0.06 m of shear stress on the surface of membrane, and as a result
for vanes of 4 mm thickness and dropped to 150 Pa at the flux will increase.

123
2244 Arab J Sci Eng (2018) 43:2237–2245

Fig. 8 Effect of width vanes on the wall shear stress along the membrane radius

Fig. 9 Profiles of the wall shear stress on the membrane surface. a Disk with vanes of 4 mm width, b disk with vanes of 1.2 mm width

4 Conclusion surface of membrane. However, the inlet velocity was found

to have a negligible effect.
In this study, the filtration of a dairy effluent through two In addition, it was noticed that the disk equipped with
models of rotating disks (with and without vanes) was simu- vanes increases the shear stress and pressure. It was observed
lated using a commercial CFD package (Ansys fluent). The that the presence of vanes had increased the shear stress value
validity of the used numerical models was checked by com- by a factor of three. We can infer that the vanes in the rotating
parison of the results of simulation with those obtained by disk module squeeze the fluid and increase the shear stress
other authors in the same conditions. When a smooth disk is and pressure. Also, the width of vanes can enhance the area
used, a good agreement between the results was obtained con- of the wall shear stress on the membrane surface.
firming that selected models represent correctly the dynamic The benefit of this study is to help in the design or
system of filtration. optimization of rotating disk membranes with complex
The results obtained by numerical simulations demon- geometries.
strated the effects of rotating and inlet velocities. Because
of the centrifugal force, the rotating velocity was the best Acknowledgements This research is supported by the Grant Tassili
(13MDU889) between French and Algerian Governments. We are very
factor that can enhance the shear stress and pressure on the

123
Arab J Sci Eng (2018) 43:2237–2245 2245

grateful to Professor A. Ould Driss for very interesting and helpful 17. Li, L.; Ding, L.; Tua, Z.; Wan, Y.; Clausse, D.; Lanoisellé, J.-L.:
discussions. Recovery of linseed oil dispersed within an oil-in-water emulsion
using hydrophilic membrane by rotating disk filtration system. J.
Membr. Sci. 342, 70–79 (2009)
References 18. Wang, Z.J.; Fidkowski, K.; Abgrall, R.; Bassi, F.; Caraeni, D.; Cary,
A.; Deconinck, H.; Hartmann, R.; Hillewaert, K.; Huynh, H.T.;
1. Luo, J.; Ding, L.; Wan, Y.; Paullier, P.; Jaffrin, M.Y.: Application Kroll, N.; May, G.; Persson, P.O.; van Leer, B.; Visbal, M.: High-
of NF-RDM (nanofiltration rotating disk membrane) module under order CFD methods: current status and perspective. Int. J. Numer.
extreme hydraulic conditions for the treatment of dairy wastewater. Methods Fluid 75, 811–845 (2013)
Chem. Eng. J. 163, 307–316 (2010) 19. Torras, C.; Pallares, J.; Garcia-Valls, R.; Jaffrin, M.Y.: Numerical
2. Ding, L.; Jaffrin, M.Y.; Luo, J.: Dynamic filtration with rotating simulation of the flow in a rotating disk filtration module. Desali-
disks, and rotating or vibrating membranes, chapter 2. In: Steven nation 235, 122–138 (2009)
Tarleton, E. (ed.) Progress in Filtration and Separation, pp. 27–59. 20. Parvareh, A.; Rahimi, M.; Madaeniand, S.S.; Alsairafi, A.A.:
Elsevier B.V, The Netherlands (2015) Experimental and CFD study on the role of fluid flow pattern on
3. Cheryan, M.: Ultrafiltration and Microfiltration Handbook. CRC membrane permeate flux. Chin. J. Chem. Eng. 19, 18–25 (2011)
Press, Boca Raton (1998) 21. Rainer, M.; Hoflinger, W.; Koch, W.; Pongratz, E.; Oechsle, D.:
4. Mulder, M.: Basic Principles of Membrane Technology, 2nd edn. 3D-flow simulation and optimization of a cross flow filtration with
Kluwer Academic Publishers, Dordrecht (1996) rotating discs. Sep. Purif. Technol. 26, 121–131 (2002)
5. Bouzerar, R.; Ding, L.; Jaffrin, M.Y.: Local permeate flux–shear– 22. Li, Y.; Lin, P.; Tung, K.: CFD analysis of fluid flow through a
pressure relationships in a rotating disk microfiltration module: spacer-filled disk-type membrane module. Desalination 283, 140–
implications for global performance. J. Membr. Sci. 170, 127–141 147 (2011)
(2000a) 23. Rahimi, M.; Madaeni, S.S.; Abolhasani, M.; Alsairafi, A.A.: CFD
6. Hadzismajlovic, D.E.; Bertram, C.D.: Flux enhancement in lam- and experimental studies of fouling of a microfiltration membrane.
inar cross flow microfiltration using a collapsible-tube pulsation Chem. Eng. Process. 48, 1405–1413 (2009)
generator. J. Membr. Sci. 142, 173–189 (1998) 24. Liu, Y.; He, G.; Liu, X.; Xiao, G.; Li, B.: CFD simulations of
7. Cai, M.; Wang, S.; Liang, H.: Optimization of ultrasound-assisted turbulent flow in baffle-filled membrane tubes. Sep. Purif. Technol.
ultrafiltration of Radix astragalus extracts with hollow fiber mem- 67, 14–20 (2009)
brane using response surface methodology. Sep. Purif. Technol. 25. Ghidossi, R.; Daurelle, J.V.; Veyret, D.; Moulin, P.: Simplified CFD
100, 74–81 (2012) approach of a hollow fiber ultrafiltration system. Chem. Eng. J. 123,
8. Frappart, M.; Jaffrin, M.Y.; Ding, L.; Espina, V.: Effect of vibra- 117–125 (2006)
tion frequency and membrane shear rate on nanofiltration of diluted 26. Liu, Y.; He, G.; Ding, L.; Dou, H.; Ju, J.; Li, B.: Experimental and
milk, using a vibratory dynamic filtration system. Sep. Purif. Tech- CFD studies on the performance of microfiltration enhanced by a
nol. 62, 212–221 (2008) turbulence promoter. Chin. J. Chem. Eng. 20, 617–624 (2012)
9. Moulai-Mostefa, N.; Frappart, M.; Akoum, O.; Ding, L.; Jaf- 27. Hwang, K.-J.; Wu, S.-E.: Disk structure on the performance of a
frin, M.Y.: Separation of water from metal working emulsions by rotating-disk dynamic filter: a case study on microalgae microfil-
ultrafiltration using vibratory membranes. J. Hazard. Mater. 177, tration. Chem. Eng. Res. Des. 94, 44–51 (2015)
978–982 (2010) 28. Bouzerar, R.; Jaffrin, M.Y.; Ding, L.; Paullier, P.: Influence of
10. Bouzerar, R.; Jaffrin, M.Y.; Lefevre, A.; Paullier, P.: Concentra- geometry and angular velocity on performance of a rotating disk
tion of ferric hydroxide suspensions in saline medium by dynamic filter. AICHE J. 46, 257–265 (2000c)
cross-flow filtration. J. Membr. Sci. 165, 111–123 (2000b) 29. Aubert, M.-C.; Elluard, M.-P.; Barnier, H.: Shear stress induced
11. Brou, A.; Ding, L.; Boulnois, P.; Jaffrin, M.Y.: Dynamic micro- erosion of filtration cake studied by a flat rotating disk method.
filtration of yeast suspensions using rotating disks equipped with Determination of the critical shear stress of erosion. J. Membr. Sci.
vanes. J. Membr. Sci. 197, 269–282 (2002) 84, 229–240 (1993)
12. Jaffrin, M.Y.; Ding, L.; Akoum, O.; Brou, A.: A hydrodynamic 30. Lee, Y.; Clark, M.M.: Modeling of flux decline during crossflow
comparison between rotating disk and vibratory dynamic filtration ultrafiltration of colloidal suspensions. J. Membr. Sci. 149, 181–
systems. J. Membr. Sci. 242, 155–167 (2004) 202 (1998)
13. Moulai-Mostefa, N.; Akoum, O.; Nedjihoui, M.; Ding, L.; Jaffrin, 31. Serra, C.A.; Wiesner, M.R.: A comparison of rotating and station-
M.Y.: Comparison between rotating disk and vibratory membranes ary membrane disk filters using computational fluid dynamics. J.
in the ultrafiltration of oil-in-water emulsions. Desalination 206, Membr. Sci. 165, 19–29 (2000)
494–498 (2007) 32. Jaffrin, M.Y.: Dynamic shear-enhanced membrane filtration: a
14. Luo, J.; Cao, W.; Ding, L.; Zhu, Z.; Wan, Y.; Jaffrin, M.Y.: Treat- review of rotating disks, rotating membranes and vibrating sys-
ment of dairy effluent by shear-enhanced membrane filtration: the tems. J. Membr. Sci. 324, 7–25 (2008)
role of foulants. Sep. Purif. Technol. 96, 194–203 (2012) 33. Hwang, K.-J.; Lin, S.-J.: Filtration flux-shear stress-cake mass
15. Zhu, Z.; Luo, J.; Ding, L.; Bals, O.; Jaffrin, M.Y.; Vorobiev, E.: relationships in microalgae rotating-disk dynamic microfiltration.
Chicory juice clarification by membrane filtration using rotating Chem. Eng. J. 244, 429–437 (2014)
disk module. J. Food Eng. 115, 264–271 (2013)
16. Brou, A.; Jaffrin, M.Y.; Ding, L.H.; Courtois, J.: Microfiltration and
ultrafiltration of polysaccharides produced by fermentation using
a rotating disk dynamic filtration system. Biotechnol. Bioeng. 82,
429–437 (2003)

123