Journal of Membrane Science 489 (2015) 183–193

Contents lists available at ScienceDirect

Journal of Membrane Science

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

Leaf protein concentration of alfalfa juice by membrane technology

Wenxiang Zhang a, Nabil Grimi a, Michel Y. Jaffrin b, Luhui Ding a,n
a
EA 4297 TIMR, University of Technology of Compiegne, 60205 Compiegne Cedex, France
b
UMR 7338, Technological University of Compiegne, 60205 Compiegne Cedex, France

art ic l e i nf o a b s t r a c t

Article history: Membrane technology (microfiltration (MF) and ultrafiltration (UF)) of alfalfa juice was studied as an
Received 28 February 2015 alternative method to conventional leaf protein concentration. Three types of filtration modules (dead
Received in revised form end filtration using laboratory Amicon cell (DA), dynamic cross filtration using rotating disk module
26 March 2015
(CRDM) and dead end filtration using rotating disk module (DRDM)) were used to investigate
Accepted 27 March 2015
concentration efficiency of MF and UF with full recycling tests and concentration tests. Rotating speed
Available online 20 April 2015
and transmembrane pressure (TMP) improved flux behavior, but higher permeate flux caused by higher
Keywords: rotating speed reduced leaf protein rejection of MF. The strong rotating shear effect and open flow
Alfalfa juice channel structure of CRDM could control concentration polarization and membrane fouling, therefore, it
Leaf protein concentration
gave the best flux behavior, least flux decline, best clarification effect, smallest irreversible fouling and
Dead-end filtration
highest permeability recovery in membrane cleaning. However, the best leaf protein rejection was
Dynamic cross-flow filtration
Flux behavior obtained by DRDM, because of high shear rate and “secondary filtration” of fouling layer created by
closed flow channel structure. Besides, CRDM showed the highest productivity and best potential for
industrial application. These results from laboratory-scale tests can be very useful for concentrating leaf
protein from alfalfa juice and serve as a valuable guide for process design in industrial scale.
& 2015 Elsevier B.V. All rights reserved.

1. Introduction juice. According to different separation mechanisms, they can be

As a common perennial vegetable, alfalfa has been widely
cultivated as a forage crop in Europe and North America and
represents about 32 million hectares in the worldwide [1]. After
drying green crop, alfalfa is utilized as raw material for the
production of fodder pellets for cattle, due to its high feed value
and high crude proteins content (about 2600 kg/ha) [2]. During the
pellet production process, the green crop of alfalfa is chopped and
pressed before drying, while much alfalfa juice is generated. This
green extracted juice containing high proteins content has been
recognized as an effective source of high quality proteins for
animals and human consumption, because of abundant sources,
high nutritive value and absence of animal cholesterol [3]. Alfalfa
leaf protein includes about 50% hydrophilic proteins and 50%
lipophilic proteins [2]. Hydrophilic proteins have high digestibility
and a balanced aminogram, which possesses significant functional
properties, such as emulsifying, jellifying and foaming agents [4–6].
In order to recover protein in alfalfa juice, numerous protein
separations and concentration technologies have been used to
concentrate and produce leaf proteins for food industry of alfalfa
n
Corresponding author. Tel.: þ 33 3 4423 4634; fax: þ 33 3 4423 7942.
E-mail address: Luhui.ding@utc.fr (L. Ding).
divided into three categories: (1) difference of solubility [7–9]:
salting, organic solvent fractionation, chromatography, crystalliza-
tion, heating and centrifugation; (2) differences of molecular size
and shape [2,10]: molecular sieve chromatography and mem-
brane; (3) difference of charge: ion exchange [11]. However, most
conventional separation and concentration methods have various
intrinsic disadvantages, such as high energy cost, low separation
efficiency, damage of nutritive proteins, complex operation and
high investment, which limit their industrial applications.
As a promising separation method, membrane technology has
wide applications in food industry and water treatment, including
manufacturing of vegetal extracts and juices, meat and fish pro-
ducts, sugar, alcoholic and non-alcoholic beverages, dairy efflu-
ent treatment and dairy products [7]. A few previous studies
[7,10,12,13] used UF to separate and concentrate crude protein
from waste leaf extraction juice. However, during the concentra-
tion process, serious flux decline caused by membrane fouling and
concentration polarization occurred, increasing the operation cost
and restricting its sustainable operation and industrial application.
For the purpose of controlling flux decline, various strategies
have been utilized to eliminate membrane fouling and decrease
concentration polarization, such as the modification of feed chara-
cteristic and membrane surface [14–16], optimization of operation
[17], choice of filtration modules [18] and membrane cleaning [19].

http://dx.doi.org/10.1016/j.memsci.2015.03.092
0376-7388/& 2015 Elsevier B.V. All rights reserved.
184 W. Zhang et al. / Journal of Membrane Science 489 (2015) 183–193
At the same time, different strategies may also affect separation
efficiency, production quality and membrane cleaning and reuse
[17,20]. Therefore, a suitable flux decline control strategy cannot
only sustain the flux behavior of whole filtration process at a high
level, but also improve separation efficiency and permeability rec-
overy in subsequent membrane cleaning.
The configuration of membrane modules (i.e., dead-end, cross-
flow and shear-enhanced) strongly influences the flux behavior
and fouling evolution [21]. The shear rates of dead-end and cross-
flow modules are produced, respectively, by a stirring effect and an
increase the tangential fluid velocity along the membrane. Shear-
enhanced modules can create a high shear rate on the membrane
surface by a moving part such as a rotating membrane, or a disk
rotating near a fixed circular membrane or by vibrating the mem-
brane either longitudinally or torsionally around a perpendicular
axis, inducing dispersion of solutes on the membrane surface and
elimination of membrane fouling [22]. Therefore, it outweighs the
conventional cross-flow membrane filtration process in excellent
effluent quality, stable permeate flux, low concentration polariza-
tion and high retention and has been successfully implemented in
many fields of research and engineering such as wastewater
treatment [18], medical engineering [22] and food engineering
[23] as well as biotechnological separations [24].
The objective of the present work is to investigate the applica-
tion of membrane technology (MF and UF) to separate and
concentrate leaf protein from alfalfa juice. In order to control flux
decline, three types of filtration modules (DA, DRDM and CRDM)
were chosen and compared. The focus of this work is to (1) discuss
the effect of TMP and rotating speed on flux behavior and
Table 1
Main characteristics of alfalfa juice.
Index Alfalfa juice
Crude protein (g L  1) 21
Chlorophyll a (mg L  1) 12.38
Chlorophyll b (mg L  1) 20.82
Dry matter (g L  1) 86
Ash (g L  1) 21
Turbidity (NTU) 600
Conductivity (ms cm  1) 9.29
pH 5.8
Soluble matter (1Brix) 8.1
Density, ρ (g ml  1) 1.20
Protein purity (%) 24.4
Fig. 1. (a) Photo of dead end filtration cell and (b) schematic representation of DA.
separation efficiency in full recycling tests; (2) investigate con-
centration efficiency, flux behavior and membrane cleaning in
concentration tests. The experiments should be useful for evaluat-
ing the membrane performance of concentrating leaf protein from
alfalfa juice and understanding the process efficiency, separation
performance, flux behavior, and membrane fouling in various
filtration modules.
2. Materials and methods
2.1. Test fluid
Alfalfa juice provided by Luzéal, Pauvres, France, was pre-
filtered by a mesh of 0.4 mm pore size and mixed, then stored at
the temperature of  20 1C until further use. In order to prevent
serious membrane fouling, before experiment the juice was
centrifuged at 4000 rpm for 10 min using a Sigma 3-16P device
for separating the insoluble materials. The main characteristics of
alfalfa juice are shown in Table 1.
2.2. Filtration modules
2.2.1. Dead end filtration using Amicon cell (DA)
The dead-end filtration Amicon 8200 cell (Millipore, Billaica,
USA) was used for alfalfa juice filtration. As shown in Fig. 1, the
internal diameter of the cell is 6.35 cm and maximum volume is
180 mL. The membrane was located at the bottom of the cell. The
effective membrane area is 3.17  10  3 m2. A constant pressure
was provided by filling the cell with nitrogen gas and maximal
pressure could reach 6 bar, while permeate was collected in a
tube placed on an electronic scale in order to calculate the
permeate flux.
2.2.2. Dynamic cross filtration using rotating disk module (CRDM)
A rotating disk module (RDM), shown in Fig. 2, was used for
alfalfa juice filtration. A flat membrane, with an effective area of
176 cm2 (outer radius R1 ¼7.72 cm, and inner radius R2 ¼1.88 cm),
was fixed on the cover of the cylindrical housing in front of the
disk. The disk equipped with 6 mm-high vanes, which can gen-
erate very high shear rates on the membrane, at rotation speeds
up to 2500 rpm. The module was fed from a thermostatic and
stirred tank containing 12 L of fluid by a volumetric diaphragm
pump (Hydra-cell, Wanner, USA). The peripheral pressure (Pc) was
adjusted by a valve on outlet tubing and monitored at the top of
the cylindrical housing by a pressure sensor (DP 15–40, Validyne,
 W. Zhang et al. / Journal of Membrane Science 489 (2015) 183–193
Fig. 2. Schematic diagrams of (a) RDM and (b) CRDM experiment set-up.
USA), and the data was collected automatically by computer.
Permeate was collected in a beaker placed on an electronic scale
(B3100P, Sartorius, Germany) connected to a computer in order to
measure the permeate flux.
2.2.3. Dead end filtration using rotating disk module (DRDM)
The RDM, described in Section 2.2.2, was fed from a 1 L
reservoir connected to compressed air (with a maximum pressure
of 6 bar). As shown in Fig. 3, permeate was collected in a beaker
placed on an electronic scale (B3100P, Sartorius, Germany) con-
nected to a computer in order to measure the permeate flux.
2.3. UF and MF membranes
UF and MF membranes fabricated by MICRODYN-NADIR GmbH
were tested in the present study. According to manufacturer's
information, their properties are summarized in Table 2.
185
2.4. Experimental procedure
A new membrane was used for each series of experiments to
ensure the same initial membrane conditions for the entire study.
The membranes were soaked in deionized water for at least 24 h
before use, and pre-pressured with deionized water for 1 h under
a pressure of 2 bar. After stabilization, the pure water flux of
membranes was measured to calculate water permeability (Lp).
Before the experiments started, the feed was heated to 35 1C, and
was fully recycled in the system at zero TMP, and this process
lasted about 10 min for each test. Then experiments were per-
formed in two modes: full recycling tests and concentration tests.
Series 1: full recycling tests: in order to estimate the membrane
performance rapidly in alfalfa juice filtration, these tests were
performed with permeate and retentate recycling to limit the
change of feed concentration to less than 10%. Feed volumes were
100 ml, 0.6 L and 3 L for DA, DRDM and CRDM, respectively. A pre-
filtration was carried out for 10 min at lowest tested TMP and a
rotating speed of 200 rpm, to ensure membranes stabilization.
186 W. Zhang et al. / Journal of Membrane Science 489 (2015) 183–193

Fig. 3. Scheme of DRDM experiment set-up.

Table 2 2.6. Calculated parameters

Properties of membranes tested.
The permeate flux (J) is calculated as follows:
Membrane Manufacturer Surface Water permeability
material (L m  2 h  1 bar  1) 1 dV
J¼ ð1Þ
A dt
P020F (UF 20 K) Nadir PH-PES 43
MV020T (MF 0.2 μm) Nadir PVDF 710 where A is the effective membrane area (m2), V is the total volume
of permeate (m3), and t is the filtration time (h).
Volume reduction ratio (VRR) is defined as follows:
V0
VVR ¼ ð2Þ
Afterward, the TMP was increased in steps at 500 rpm. Then for VR
DRDM and CRDM, TMP was fixed, while rotating speed was
where V0 and VR are initial feed volume and retentate volume,
decreased in steps from 2500 rpm to 500 rpm. The filtration was
respectively.
periodically stopped by suddenly releasing the pressure (0.2 bar)
The mean TMP is obtained by integrating the local pressure pc
in order to mimic back flushing and minimize the fouling
(Pa) over the membrane area as follow:
accumulation from the last TMP step. Samples were collected in
permeate 5 min after the beginning of each TMP increment or 1
TMP ¼ pc  ρκ2 ωR2 ð3Þ
each rotating speed decrement in order to obtain stabilized flux 4
and transmission conditions. where ρ is the density of the fluid (g L  1), κ is the velocity factor
Series 2: concentration tests: in this series, permeate was not (0.89) for this RDM system and R is the housing inner
returned to feed tank. Feed volumes of 180 ml, 1.2 L and 6 L were diameter (m).
concentrated to 30 ml, 0.2 L and 1 L for DA, DRDM and CRDM at Protein purity (PP, %), which can be used to estimate clarifica-
constant conditions, respectively. All volume reduction ratios tion effect, is calculated as follows:
(VRR) were 6. Samples were collected in permeate at every half
Cp
unit of VRR. PP ¼ ð4Þ
After each series of tests, at a rotating speed of 500 rpm, the Cd
filtration system was flushed by deionized water for 10 min at where Cp (g/L) and Cd (g/L) are, respectively, crude protein and dry
1000 rpm. Then alkaline cleaning was carried out by using a P3- matter concentrations.
ultrasil 10 (Ecolab, cleaning USA) detergent to remove protein and Productivity (Pr, L h  1 m  2 bar  1) is defined as follows:
colloid residues, at 0.25% concentration and 1000 rpm, and Lp was
Vc
measured to determine the permeability recovery. Pr ¼ ð5Þ
T o  A  TMP
where Vc is concentrated volume (L) and To is operation time (h).
2.5. Analytical methods

Turbidities of permeate were measured with a Ratio Turbidi-
meter (Hach, USA). Conductivity was measured with a Multi-
Range Conductivity Meter (HI 9033, Hanna, Italy) and pH was
measured with a pH Meter (MP 125, Mettler Toledo, Switzerland).
Dry matter was determined by measuring the weight loss after
drying samples at 10572 1C for 5 h in an oven. Soluble matter
measurements (1Brix) were done, at room temperature, by means
of a digital refractometer PR-32α (ATAGO Co., LTD, Japan). The
crude protein concentration in solution was determined using the
Kjeldahl method. To convert organic nitrogen, except ammonium,
the factor 6.25 was used [7]. Chlorophyll A and B were measured
by a spectrophotometric method (Biochrom Ltd., Cambridge
Science Park, Cambridge, Angleterre).
3. Results and discussion
3.1. Full recycling tests
Full recycling tests were performed to investigate the effect of
TMP and rotating speed on separation performance and flux
behavior for different types of filtration modules.
3.1.1. Effect of TMP on separation performance and flux behavior
Fig. 4 shows the effect of TMP on performance flux and
separation performance (crude protein concentration in permeate
and 1Brix of permeate) for DA, DRDM and CRDM at rotating speed
of 500 rpm. As expected, higher TMP offered larger driving force
 W. Zhang et al. / Journal of Membrane Science 489 (2015) 183–193 187

Fig. 4. Flux behavior and separation performance at various TMP for MF and UF.

and led to higher permeate flux, however, when TMP exceeded
2 bar for MF and 3 bar for UF, the flux reached a plateau. This could
be explained by the threshold flux regime [21]: below a certain
value, fouling rate remained at a low and nearly constant level and
flux increased linearly with increasing TMP. But above it, a fouling
layer of higher thickness or density on membrane surface was
formed, and fouling rate was high and flux dependent, so the
growth rate of flux decreased. The flux increased in the following
order: CRDM 4DRDM4 DA. Comparing with DA, because of the
disk equipped with vanes, higher shear rate at the membrane
surface for CRDM and DRDM could eliminate concentration
polarization and prevent membrane fouling effectively, thus
improving flux behavior greatly. Besides, the open flow channel
structure of CRDM enhanced the mobility of feed on membrane
surface, which improved shear rate and decreased concentration
polarization, so CRDM had the best flux behavior. Due to bigger
pore size and higher permeability, MF gave a higher permeate flux
than UF, despite its lower TMP.
In a filtration process, the permeate flux and fouling layer are
two main factors that affect separation efficiency [25,26]. When flux
is low, its increase caused more solvent to pass through membrane
and reduced solute transmission by “dilution effect” [27]. But at
high fluxes, a further increase improved concentration polarization
and solute concentration on membrane surface increased, then
enhancing the concentration gradient and diffusive effect transfer
through the membrane, thus solute transmission improved [27].
188 W. Zhang et al. / Journal of Membrane Science 489 (2015) 183–193

The fouling layer dominated by solutes accumulation on membrane
surface narrowed membrane pores, played a “secondary filtration”
role and improved solute rejection [28]. As shown in Fig. 4(c)–(f),
with the elevation of TMP, permeate flux increases, more solutes
accumulate on membrane surface and the “diffusion effect”
improves, so solute rejection reduces. Due to more serious concen-
tration polarization, higher TMP enhanced fouling layer on mem-
brane surface, which may reduce solute rejection. But this effect
was less significant than permeate flux during a TMP-increasing
process, because at rotating speed of 500 rpm, concentration
polarization could be controlled at a low level. Furthermore, solute
rejection was in the sequence DRDM4CRDM4DA. Comparing
with DA, CRDM and DRDM had much higher Reynolds number Re
than DA at the same rotating speed due to their significantly
different structures, bigger agitator diameter and higher shear rate
[23], producing more intense hydrodynamics and greatly decreas-
ing concentration polarization. Therefore, for CRDM and DRDM, the
effect of permeate flux on separation efficiency was significantly
reduced and their main factor was fouling layer, but for DA, the
effect of permeate flux was dominant. Moreover, because of the
open flow channel structure, CRDM had a thinner fouling layer, thus
its “secondary filtration” effect was lower than for DRDM. As for DA,
its shear rate created by the stirring effect on membrane surface
was least, coupling with its closed flow channel structure [23], thus
DA had a most serious concentration polarization and “diffusion
effect”. In addition, due to larger membrane pores and higher

Fig. 5. Flux behavior and separation performance at various rotating speed for MF and UF.
 W. Zhang et al. / Journal of Membrane Science 489 (2015) 183–193
permeate flux, MF had a higher transmission for crude proteins and
soluble organic matters and less separation performance than UF.
3.1.2. Effect of rotating speed on separation performance and flux
behavior
Fig. 5 shows the effect of rotating speed on permeate flux and
separation performance for MF and UF of alfalfa juice in full recycling
tests for CRDM and DRDM. The TMPs were set to 3 and 4 bar for MF
and UF and permeate flux, crude protein concentration and 1Brix
were measured at rotating speeds of 500, 1000, 1500, 2000 and
2500 rpm, respectively. It is obvious that increment of rotating speed
improved flux for both MF and UF, due to lower concentration
polarization and membrane fouling at higher shear rate. However,
MF and UF had different growth tendencies: for MF, the flux of
CRDM and DRDM increased linearly with rotating speed, implying
that elevating rotating speed could decrease concentration polariza-
tion effectively. But for UF, when rotating speed reached 1500 rpm,
the flux reached a steady-state and improving rotating speed could
not elevate flux significantly, because the concentration polarization
was already minimized for UF at a TMP of 4 bar. Permeate flux of MF
was much higher than that of UF, resulting in the different flux
behavior with variation of rotating speed, which was reasonable.
According to the studies of Bacchin et al. [29] and Luo et al. [17],
there were two anti-fouling mechanisms: Brownian diffusion (back
diffusion of rejected particles into the bulk, due to concentration
gradient) and shear-enhanced back transport. The shear-induced
back diffusion is proportional to the shear rate and square of the
particle size, while the Brownian diffusion coefficient is inversely
proportional to particle size and independent of shear rate [30]. The
elevation of rotating speed could improve shear-enhanced back
transport, and also facilitate the recombination of some macro-
particles, including leaf proteins and colloids, especially for UF with
higher solute rejection [23], thus decreasing Brownian diffusion. The
compromise of Brownian diffusion elimination and shear-enhanced
back transport enhancement probably made the flux improvement
less significant when rotating speed rose from 1500 to 2500 rpm.
Furthermore, CRDM and DRDM had similar variation tendencies,
because they had the same rotating equipment and similar hydro-
dynamics on membrane surface. In addition, because of the open
flow channel structure, CRDM had a higher flux.
Fig. 5(c)–(f) describes the effect of rotating speed on crude protein
and 1Brix in permeate. With increase of rotating speed, the crude
protein concentration and 1Brix of MF in permeate increased, whereas
these of UF decreased. These different mechanisms of separation
efficiency could be explained as follow: because of much higher
permeate flux, the main factor of MF for separation efficiency was
permeate flux, so, as mentioned in Section 3.1.1, the increase of
Fig. 6. Flux behaviors vs VRR during alfalfa juice concentration process by (a) MF at 3 bar and (b) UF at 4 bar.
189
permeate flux improved solute transmission; for UF, fouling layer was
dominant, higher permeate flux led to thicker fouling layer, greater
“secondary filtration” effect and better separation performance.
3.2. Concentration tests
Based on previous tests in full recycling mode, moderate
operation conditions (rotating speed: 1000 rpm for CRDM and
DRDM and 500 rpm for DA, TMP: 3 bar for MF and 4 bar for UF)
were selected for concentration tests.
3.2.1. Flux behavior
As shown in Fig. 6, the flux decreases rapidly when VRR varies
from 1 to 2, during which leaf protein foulants deposited and
adsorbed at membrane promptly and foulant–cleaning membrane
interaction was the main fouling mechanism. Then the flux
reduced slightly with increasing VRR in semi-log coordinates and
fluctuated a little when VRR increased from 2 to 6, due to mass
transfer limited regime [17]. With the “self-cleaning” effect of
shear rate, membrane fouling did not accelerate and fouling
mechanism was foulant–deposited foulant interaction. It could
be observed that compared with DA, CRDM and DRDM not only
had much higher permeate flux, but also presented much less flux
decline. This implied that with respect to stirring effect of DA, RDM
could control solute accumulation at membrane surface [23] and
total filtration resistance, and maintain permeate flux at a high
level during the whole concentration process, although solution
concentration and concentration polarization layer kept increasing
with VRR. Due to closed flow channel structure and smaller shear
rate on membrane surface, DA had the least flux and biggest flux
decline, even close to 95% for MF. Besides, because of its open flow
channel structure, CRDM had a slightly higher flux and smaller
flux decline than DRDM. Compared with UF, permeate flux of MF
was much higher, but its flux decline was also higher, which
corresponded to a relationship between membrane pore and
foulant [31]: as the main foulants, leaf proteins had a size similar
to MF pores and caused pore blocking easily, therefore, MF had a
more pore blocking and greater flux decline than UF.
3.2.2. Separation and concentration performance
The variation of crude protein concentration and 1Brix in
permeate at various VRR is illustrated in Fig. 7. The crude protein
concentration and 1Brix in permeate follow a similar trend: from
VRR 1 to 2, they kept at a stable value, however, when VRR
exceeded 2, they kept increasing with VRR. This phenomenon
corresponded to the variation of flux behavior and membrane
fouling mechanisms [32]. As mentioned in Section 3.2.1, when VRR
190 W. Zhang et al. / Journal of Membrane Science 489 (2015) 183–193

Fig. 7. Separation performances vs VRR during concentration of alfalfa juice by (a) crude protein in permeate and (b) 1Brix of permeate.

Fig. 8. Characteristics (crude protein and 1Brix) of permeate and retentate for MF and UF.

increased from 1 to 2, a large number of leaf proteins deposited,
were absorbed at membrane and formed foulant–cleaning mem-
brane interaction fouling, thus feed leaf proteins and soluble
matters remained stable and did not affect the separation perfor-
mance. When VRR varied from 2 to 6, the main fouling mechan-
ism was foulant–deposited foulant interaction and concentration
polarization increased, thus the greater concentration gradient
made more crude proteins and soluble matters pass through the
membrane.
The concentration efficiency of crude protein, 1Brix and chlor-
ophyll (A and B) for various types of filtration modules are shown
in Figs. 7–9. As shown in Fig. 7, with increase of VRR, more solutes
were rejected and feed concentration increased, causing higher
solute transmission, while flux decline strengthened this process.
However, fouling layer also aggravated and higher “secondary
filtration” formed, decreasing the solute transmission. At the
beginning of concentration test (VRR from 1 to 2), there was a
tradeoff between these two mentioned effects, so crude protein
concentration and 1Brix in permeate almost kept stable; while in
the latter stage, fouling layer was stable, but permeate flux was
decreasing and the feed concentration was increasing, so the
solute transmission increased slightly. Fig. 8 illustrates that CRDM
 W. Zhang et al. / Journal of Membrane Science 489 (2015) 183–193 191

Fig. 9. Chlorophyll A and B in permeate and retentate for MF and UF.

Table 3 Table 4
Protein purity for various types of filtration modules. Operation time, concentrated volume and productivity for various types of
filtration modules.
VRR ¼ 6 Protein purity in permeate (%) Protein purity in retentate (%)
VRR ¼ 6 Operation time Concentrated Productivity
MF DA 14 44 (h) volume (L) (L m  2 h  1 bar  1)
DRDM 4.6 73
CRDM 5.5 93 MF DA 6.5 0.03 0.48
UF DA 9.3 39 DRDM 0.81 0.2 4.73
DRDM 4.8 41 CRDM 3.11 1 6.06
CRDM 5.3 45 UF DA 5.42 0.03 0.43
DRDM 1.95 0.2 1.42
CRDM 8.33 1 1.70
and DRDM had a much higher crude protein concentration in
retentate than DA, implying that the RDM with a high shear rate
could enhance the concentration capacity of leaf protein in alfalfa higher rejection, but its permeate flux and protein purity were
juice significantly. Compared with CRDM, DRDM had better con- obviously lower than MF. This implied than UF gave higher
centration efficiency, which was probably due to its better “sec- concentration capacity but lower efficiency than MF, which was
ondary filtration” effect caused by closed flow channel structure. contradictory. In Section 3.2.3, a new productivity concept is first
As for 1Brix and chlorophyll, as shown in Figs. 8(c) and (d), and 9, proposed to evaluate operation efficiency.
DA and DRDM had the highest concentration in retentate for 1Brix
and chlorophyll, respectively. Table 3 presents that retentates for
all these filtration modules had much higher protein purities than 3.2.3. Production efficiency
feed alfalfa juice, indicating that membrane technology could not In membrane filtration, there are many factors [26,33], such as
only concentrate leaf protein, but also improve protein purity, flux behavior, separation performance, concentration efficiency,
because a membrane separation process could reject large solutes, membrane fouling and membrane cleaning, affecting its operation
such as leaf protein, while many small solutes passed through and industrial application. However, some factors are interacting or
membrane and solute matter concentration reduced. In compar- conflicting [34], for example flux behavior and separation perfor-
ison with DA and DRDM, CRDM had higher protein purity and best mance are related in membrane filtration, and membrane fouling
clarification effect, due to its higher shear rate and lower “sec- has an influence on concentration efficiency and membrane clean-
ondary filtration” effect. Besides, UF had an obvious superiority for ing during concentration process. In order to study operation
concentration capacity than MF, owing to its small pore size and efficiency and evaluate its potential for industrial application,
192 W. Zhang et al. / Journal of Membrane Science 489 (2015) 183–193

90
Fouled membrane
80
After cleaning
Permeability recovery (%)

70
60
50
40
30
20
10
0
DA DRDM CRDM DA DRDM CRDM
Fig. 11. Schematic of filtration behavior for various types of filtration modules.
MF UF
Fig. 10. Membrane cleaning for concentration tests.
by leaf protein and irreversible fouling and enhances permeate
recovery in membrane cleaning, permitting membrane sustainable
utilization. Besides, by calculating productivities, Table 4 shows that
productivity was first proposed to estimate the production efficien- CRDM has the best production efficiency. Due to its open flow
cies of various types of filtration modules. As shown in Table 4, at channel structure, the CRDM has a better mobility for feed flow and
VRR¼6, CRDM uses 3.11 h to generate 1 L of concentrated alfalfa reduces concentration polarization. In fact, dynamic filtration sys-
juice, and its flux per bar is 6.06 L h  1 m  2 bar  1, which is highest tems can operate at low feed flow rates, which are just slightly
among all filtration modules. Compared with UF, MF needs less greater than permeate flow rate (about 10–13% for nanofiltration
operating time to generate the same concentrated volume and has and reverse osmosis and 3–5% for UF and MF) [22,24], thus they do
higher productivity. Therefore, the CRDM with MF was the most not need powerful and large pumps as in traditional dead-end and
efficient for industrial application. cross-flow filtration and have lower energy consumption for feed
pump. In summary, compared with other filtration modules, as
shown in Fig. 11, the CRDM owns many advantages and is most
3.2.4. Membrane fouling and cleaning
suitable for the industrial application of alfalfa juice concentration.
Membrane cleaning using P3-ultrasil 10 (Ecolab, USA) deter-
gent was utilized to investigate the permeability recovery and
efficiency of membrane cleaning. As shown in Fig. 10, permeability
recovery defined as the ratio of pure water flux before concentra-
4. Conclusions
tion and being fouled or after cleaning [24], is used to estimate
membrane fouling and cleaning efficiency for different types of
DA, DRDM and CRDM were tested to concentrate leaf protein
filtration modules. MF had a larger membrane fouling and lower
from alfalfa juice. MF and UF permitted filtration to obtain better
permeability recovery than UF, because the more serious pore
separation and concentration efficiency. Full recycling tests
blocking of MF produced greater irreversible fouling. It is evident
demonstrated that rotating speed reduced concentration polariza-
that for both MF and UF, the optimal permeability recovery
tion and enhanced higher flux behavior, but limited the separation
increases in the order as follows: CRDM 4DRDM 4DA. This
effect in MF. TMP improved permeate flux significantly and
obviously indicated that CRDM had the lowest irreversible fouling
reduced leaf protein rejection slightly. MF had a better flux
and recovered more easily its membrane permeability, due to its
behavior and less leaf protein rejection. Then concentrations tests
excellent “self-cleaning” capacity induced by high shear-effect [35]
indicated that due to shear effect and open flow channel structure,
in the concentration process. The high recovery of membrane
CRDM had the best flux behavior, least flux decline, smallest
permeability confirms a high potential application for alfalfa juice
irreversible fouling and highest permeability recovery in mem-
concentration.
brane cleaning, while its lowest fouling layer led to high protein
purity in retentate. Because of high shear rate and “secondary
3.3. Discussion of various types of filtration modules filtration” effect of fouling layer created by closed flow channel
structure, DRDM had highest crude protein concentration in
When alfalfa juice is concentrated by MF and UF, flux behavior, retentate. Through comprehensive calculation and consideration,
separation performance and membrane cleaning are three main CRDM with MF had the highest productivity for leaf protein
indexes to estimate the process efficiency for various types of concentration from alfalfa juice and best potential for industrial
membrane filtration modules. In this study, three types of filtration application. This investigation could serve as valuable information
modules (DA, DRDM and CRDM) were utilized to concentrate leaf for industry by extrapolating this process to industrial production.
protein from alfalfa juice. DA used a conventional and simple
stirring effect to produce shear rate on membrane surface and
control concentration polarization and membrane fouling. CRDM
and DRDM with a rotating-disk equipped with 6 mm-high vanes Acknowledgments
can create a much higher shear rate than DA for improving the
elimination of serious flux decline (seen in Fig. 6). At the same time, The authors would like to acknowledge Luzéal company
the high shear rate of CRDM and DRDM also enhances leaf protein (Pauvres, France) for the kind supply of alfalfa juice and also
rejection and decreases impurities in concentration process (seen in would like to acknowledge the financial support of China Scholar-
Figs. 7–9). Furthermore, it reduces membrane pore blocking caused ship Council for Wenxiang Zhang's thesis fellowship.
 W. Zhang et al. / Journal of Membrane Science 489 (2015) 183–193
References
[1] J. Volenec, S. Cunningham, D. Haagenson, W. Berg, B. Joern, D. Wiersma,
Physiological genetics of alfalfa improvement: past failures, future prospects,
Field Crop. Res. 75 (2002) 97–110.
[2] L. Firdaous, P. Dhulster, J. Amiot, A. Gaudreau, D. Lecouturier, R. Kapel, et al.,
Concentration and selective separation of bioactive peptides from an alfalfa
white protein hydrolysate by electrodialysis with ultrafiltration membranes,
J. Membr. Sci. 329 (2009) 60–67.
[3] Z. Xie, J. Huang, X. Xu, Z. Jin, Antioxidant activity of peptides isolated from
alfalfa leaf protein hydrolysate, Food Chem. 111 (2008) 370–376.
[4] B. Lamsal, R. Koegel, S. Gunasekaran, Some physicochemical and functional
properties of alfalfa soluble leaf proteins, LWT – Food Sci. Technol. 40 (2007)
1520–1526.
[5] W.E. Barbeau, J.E. Kinsella, Ribulose bisphosphate carboxylase/oxygenase
(rubisco) from green leaves‐potential as a food protein, Food Rev. Int. 4
(1988) 93–127.
[6] R. Douillard, O. De Mathan, Leaf protein for food use: potential of Rubisco, New
and Developing Sources of Food Proteins, Springer, United States (1994)
307–342.
[7] W. Koschuh, G. Povoden, V.H. Thang, S. Kromus, K.D. Kulbe, S. Novalin, et al.,
Production of leaf protein concentrate from ryegrass (Lolium perenne x multi-
florum) and alfalfa (Medicago sauva subsp. sativa). Comparison between heat
coagulation/centrifugation and ultrafiltration, Desalination 163 (2004)
253–259.
[8] C.C. Ibarra-Herrera, O. Aguilar, M. Rito-Palomares, Application of an aqueous
two-phase systems strategy for the potential recovery of a recombinant
protein from alfalfa (Medicago sativa), Sep. Purif. Technol. 77 (2011) 94–98.
[9] L. Firdaous, P. Dhulster, J. Amiot, A. Doyen, F. Lutin, L.-P. Vézina, et al.,
Investigation of the large-scale bioseparation of an antihypertensive peptide
from alfalfa white protein hydrolysate by an electromembrane process,
J. Membr. Sci. 355 (2010) 175–181.
[10] B. Bals, B.E. Dale, Economic comparison of multiple techniques for recovering
leaf protein in biomass processing, Biotechnol. Bioeng. 108 (2011) 530–537.
[11] C. Arulvasu, S.K.S. Shakthi, G. Babu, N. Radhakrishnan, Purification and
identification of bioactive protein from leaves of Datura inoxia P. mil, Biomed.
Prev. Nutr. 4 (2014) 143–149.
[12] S.S. Fernández, C. Menéndez, S. Mucciarelli, A.P. Padilla, Saltbush (Atriplex
lampa) leaf protein concentrate by ultrafiltration for use in balanced animal
feed formulations, J. Sci. Food Agric. 87 (2007) 1850–1857.
[13] S. Fernández, C. Menéndez, S. Mucciarelli, A. Pérez Padilla, Concentration and
desalination of zampa (Atriplex lampa) extract by membrane technology,
Desalination 159 (2003) 153–160.
[14] L.Y. Ng, A.W. Mohammad, C.Y. Ng, A review on nanofiltration membrane
fabrication and modification using polyelectrolytes: effective ways to develop
membrane selective barriers and rejection capability, Adv. Colloid Interfac 197
(2013) 85–107.
[15] S. Meng, Y. Liu, Alginate block fractions and their effects on membrane fouling,
Water Res. 47 (2013) 6618–6627.
[16] S. Meng, M. Rzechowicz, H. Winters, A.G. Fane, Y. Liu, Transparent exopolymer
particles (TEP) and their potential effect on membrane biofouling, Appl.
Microbiol. Biotechnol. 97 (2013) 5705–5710.
193
[17] J. Luo, L. Ding, Y. Wan, M.Y. Jaffrin, Threshold flux for shear-enhanced
nanofiltration: experimental observation in dairy wastewater treatment,
J. Membr. Sci. 409 (2012) 276–284.
[18] W. Zhang, L. Ding, Investigation of membrane fouling mechanisms using
blocking models in the case of Shear-enhanced ultrafiltration, Sep. Purif.
Technol. (2014).
[19] Z. Wang, J. Ma, C.Y. Tang, K. Kimura, Q. Wang, X. Han, Membrane cleaning in
membrane bioreactors: a review, J. Membr. Sci. 468 (2014) 276–307.
[20] J. Luo, S.T. Morthensen, A.S. Meyer, M. Pinelo, Filtration behavior of casein
glycomacropeptide (CGMP) in an enzymatic membrane reactor: fouling
control by membrane selection and threshold flux operation, J. Membr. Sci.
469 (2014) 127–139.
[21] W. Zhang, J. Luo, L. Ding, M.Y. Jaffrin, A review on flux decline control
strategies in pressure-driven membrane processes, Ind. Eng. Chem. Res.
(2015).
[22] M.Y. Jaffrin, Dynamic shear-enhanced membrane filtration: a review of
rotating disks, rotating membranes and vibrating systems, J. Membr. Sci.
324 (2008) 7–25.
[23] Z. Zhu, H. Mhemdi, L. Ding, O. Bals, M.Y. Jaffrin, N. Grimi, et al., Dead-end
dynamic ultrafiltration of juice expressed from electroporated sugar beets,
Food Bioprocess Technol. (2014) 1–8.
[24] Z. Zhu, J. Luo, L. Ding, O. Bals, M.Y. Jaffrin, E. Vorobiev, Chicory juice
clarification by membrane filtration using rotating disk module, J. Food Eng.
115 (2013) 264–271.
[25] H. Lin, M. Zhang, F. Wang, F. Meng, B.-Q. Liao, H. Hong, et al., A critical review
of extracellular polymeric substances (EPSs) in membrane bioreactors: char-
acteristics, roles in membrane fouling and control strategies, J. Membr. Sci.
460 (2014) 110–125.
[26] A.S. Al-Amoudi, Factors affecting natural organic matter (NOM) and scaling
fouling in NF membranes: a review, Desalination 259 (2010) 1–10.
[27] J. Luo, Y. Wan, Effects of pH and salt on nanofiltration—a critical review,
J. Membr. Sci. 438 (2013) 18–28.
[28] W. Guo, H.-H. Ngo, J. Li, A mini-review on membrane fouling, Bioresour.
Technol. 122 (2012) 27–34.
[29] P. Bacchin, P. Aimar, R.W. Field, Critical and sustainable fluxes: theory,
experiments and applications, J. Membr. Sci. 281 (2006) 42–69.
[30] G. Belfort, R.H. Davis, A.L. Zydney, The behavior of suspensions and macro-
molecular solutions in crossflow microfiltration, J. Membr. Sci. 96 (1994) 1–58.
[31] P. van der Marel, A. Zwijnenburg, A. Kemperman, M. Wessling, H. Temmink,
W. van der Meer, Influence of membrane properties on fouling in submerged
membrane bioreactors, J. Membr. Sci. 348 (2010) 66–74.
[32] J. Luo, W. Cao, L. Ding, Z. Zhu, Y. Wan, M.Y. Jaffrin, Treatment of dairy effluent
by shear-enhanced membrane filtration: the role of foulants, Sep. Purif.
Technol. 96 (2012) 194–203.
[33] W. Gao, H. Liang, J. Ma, M. Han, Z.-l. Chen, Z.-s. Han, et al., Membrane fouling
control in ultrafiltration technology for drinking water production: a review,
Desalination 272 (2011) 1–8.
[34] H. Lin, W. Peng, M. Zhang, J. Chen, H. Hong, Y. Zhang, A review on anaerobic
membrane bioreactors: applications, membrane fouling and future perspec-
tives, Desalination 314 (2013) 169–188.
[35] W. Zhang, Z. Zhu, M.Y. Jaffrin, L. Ding, Effects of hydraulic conditions on
effluent quality, flux behavior, and energy consumption in a shear-enhanced
membrane filtration using box-behnken response surface methodology, Ind.
Eng. Chem. Res. 53 (2014) 7176–7185.