Separation and Purification Technology 171 (2016) 289–296

Contents lists available at ScienceDirect

Separation and Purification Technology

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

Influence of operating parameters and membrane materials on fouling

of ceramic hollow fibre membranes
F. Arndt a,⇑, F. Ehlen b, S. Schütz b, H. Anlauf a, H. Nirschl a
a
Karlsruhe Institute of Technology (KIT), Germany
b
MANN+HUMMEL GmbH, Ludwigsburg, Germany

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

Article history: Fouling is one of the critical issues in membrane filtration, especially in the field of wastewater treatment.
Received 19 November 2015 In this study sodium alginate was used as model compound for extracellular polymeric substances (EPS)
Received in revised form 1 July 2016 in order to investigate the filtration and fouling characteristics of two newly developed ceramic hollow
Accepted 29 July 2016
fibre membranes. The membranes studied have an asymmetrical structure consisting of an Al2O3 support
Available online 30 July 2016
layer, which is covered by an active layer. Two different active layers (Al2O3 and SiC) were studied under
various operating conditions (dead-end and cross-flow filtration) by filtering sodium alginate solutions in
Keywords:
the presence and absence of calcium ions. Fouling mechanisms taking place in dead-end mode were eval-
Ceramic hollow fibre membrane
Alginate
uated according to the cake filtration model. Dead-end filtration experiments revealed a change in fouling
Fouling mechanisms from pore blocking and cake filtration in the absence of calcium ions to cake filtration only
Operational parameters in the presence of calcium ions. Furthermore, cross-flow filtration experiments showed a big influence of
Cross-flow velocity calcium ions on the removal of the resulting alginate layer from the membrane surface due to shear
forces induced by cross-flow. In the absence of calcium ions, the filtrate flux was found to depend
strongly on transmembrane pressure (TMP) and cross-flow velocity, whereas in the presence of calcium
ions, the chosen cross-flow velocities (1–3 m/s) were not sufficient to control alginate fouling. Cross-flow
experiments without calcium addition revealed a higher flux for SiC membranes compared to Al2O3
membranes. For dead-end filtration experiments, no significant difference was found.
Ó 2016 Elsevier B.V. All rights reserved.

1. Introduction in MBR due to a promotion of molecular binding [7,10]. The Ca2+

Membrane fouling is one of the critical issues, as it affects
productivity, plant operation and maintenance costs in membrane
filtration processes [1]. This is especially true for membrane appli-
cations in membrane bioreactors (MBR) and other treatment
processes of water with high organic loads. Membrane fouling is
the unwanted accumulation of materials on the membrane surface,
leading to an enhanced filtration resistance [2]. It was reported
that one of the main causes of membrane fouling in MBR are the
extracellular polymeric substances (EPS) [1,3,4]. Among proteins
and humic substances, polysaccharides have been identified to be
main constituents of EPS [5]. In membrane filtration research,
sodium alginate often serves as a model compound for EPS
[3,4,6–9]. Sodium alginate is a polysaccharide, which is negatively
charged near the neutral pH and very sensitive to changes in ionic
strength. Alginates form gels in the presence of divalent cations,
such as Ca2+, which are also discussed to affect membrane fouling
⇑ Corresponding author.
E-mail address: Felicitas.Arndt@kit.edu (F. Arndt).
ions preferably bind to the carboxylic groups of alginates, leading
to a highly organised network which can be explained by the
‘‘egg-box” model [4,6]. This phenomenon is well-known in litera-
ture, but leads to different conclusions in terms of permeate flux
decline [6]. Some studies show that the addition of Ca2+ cause an
increase in fouling behaviour of the system [11], while other
groups found that the addition of Ca2+ improved permeate flux
and reduced fouling resistances [6]. It has to be taken into account
that the fouling behaviour of a system can be affected by operating
and hydrodynamic conditions, membrane material and its pore
size distribution, as well as by the feed composition. Hence, these
contradictory results may be due to different feed and operation
conditions.
The aim of this work is to improve the understanding of ultra-
filtration fouling phenomena using alginate as model compound
for EPS. In particular, the influence of hydrodynamic parameters
(TMP and cross-flow velocity), feed solution characteristics in
terms of Ca2+ concentration, as well as the influence of different
membrane materials (Al2O3-coated and SiC-coated membranes)
on membrane fouling characteristics is studied. We use ceramic

http://dx.doi.org/10.1016/j.seppur.2016.07.046
1383-5866/Ó 2016 Elsevier B.V. All rights reserved.
290 F. Arndt et al. / Separation and Purification Technology 171 (2016) 289–296

Nomenclature

A active filtration area of the membrane [m2] Rm membrane hydraulic resistance [1/m]
a parameter in Eq. (9) [–] Rt total resistance [1/m]
b parameter in Eq. (9) [–] V filtrate volume [ml]
cb bulk concentration of organic foulant [kg/m3] vCF crossflow velocity [m/s]
J permeate flux [l/(m2 h)]
Nf resistance number [–] Greek letters
NS shear stress number [–] a specific cake resistance [m/kg]
TMP transmembrane pressure [kPa] q density of the feed solution [kg/m3]
PWP Pure water permeability [l/(m2 h bar)] g dynamic viscosity [Pa s]
Rf fouling resistance [1/m] DP transtubular pressure drop [kPa]

hollow fibre membranes, because these membranes are charac- resistance, and cb is the bulk concentration of organic foulant
terised by high chemical and thermal stability as well as by a large (e.g. alginate). If a linear relationship is the result of plotting
specific membrane filter surface area. t/V = f (V), this means that cake formation is the only fouling mech-
anism taking place.
2. Materials and methods
2.2. Ceramic hollow fibre membranes
2.1. Resistance determination
Ceramic hollow fibre membranes combine the advantages of
A number of mechanisms may lead to flux decline in the cross-
both inorganic membrane material and hollow fibre geometry
flow and dead-end modes. Filtrate flux can be calculated using
[13]. The main advantages resulting from this combination are
Darcy’s law for constant TMP according to the resistance in series
high chemical and thermal stability associated with a high specific
model [5,12]:
packing density [13]. Further advantages of ceramic membranes
1 dV TMP TMP are their abrasion resistance as well as the possibility to backflush
J¼  ¼ ¼ ð1Þ
A dt gðRm þ Rf Þ g  Rt the membranes. Major drawbacks which have limited the wide-
spread application of ceramic membranes are high capital costs
where J is the permeate flux, TMP is the transmembrane pressure, g as well as the brittleness of the membranes [14].
is the dynamic viscosity, Rm is the membrane resistance, Rf is the The ceramic hollow fibre membranes studied have an asym-
fouling layer resistance, and Rt the total resistance at the end of metrical two-layer structure. The a-Al2O3 microfiltration support
the filtration experiment. layer has open pores and determines the stability of the fibre.
TMP was calculated from the pressure data taking the pressure The support layer can be covered by different ceramic coating lay-
drop along the hollow fibre feed channel into account: ers for ultrafiltration (active layers). Fig. 1 shows a cross-section of
 
DP a ceramic hollow fibre membrane and a detailed representation of
TMP ¼ PFeed   PPerm ð2Þ the active layer on the inner lumen of the membrane. In this study
2
ceramic hollow fibre membranes with two different inorganic
where PFeed is the pressure of the feed stream, PPerm the pressure of active layers (Al2O3 and SiC) are used. The membranes are manu-
the permeate stream (ambient pressure) and DP is the pressure factured by a simultaneous spinning and phase inversion process.
drop along the hollow fibre membrane. After the spinning process the fibre is washed and sintered at high
Membrane resistance Rm was determined by filtering ultrapure temperatures. During this sintering process the polymer, which
water for 30 min at a TMP of 100 kPa in the dead-end mode was besides ceramic powder, a number of additives and a corre-
through the unused membrane before every fouling experiment. sponding solvent system, one compound of the spinning mass, is
Rm was calculated from the clean water fluxes J0 according to burned completely out of the fibre structure. After sintering, the
following equation: support layer is covered by the functional ultrafiltration ceramic
TMP coating layer and sintered a second time. Further information on
Rm ¼ ð3Þ the manufacturing process of the membranes used in this study
g  J0
is given by Ebrahimi et al. [13].
The total resistance Rt was calculated from the resulting filtra- Fig. 2 shows SEM images of the Al2O3 and SiC active layers
tion flux at the end of the filtration experiment: investigated. The active layer is located on the feed side of the
TMP membrane. Depending on the filtration mode, the coating is
Rt ¼ ð4Þ applied on the inside of the hollow fibre membrane (in-out mode)
g  J120
or on the outer surface of the hollow fibre (out-in mode).
J120 is the flux after 120 min of dead-end filtration. Knowing Rt Membranes operated in the in-out filtration mode allow an opera-
and Rm, the fouling layer resistance Rf can be calculated. tion in the cross-flow mode. Membranes coated on the outside of
Fouling mechanisms for dead-end filtration can be visualised by the hollow fibre are used for the treatment of higher concentrated
plotting the flux data in an appropriate linearised form according feed solutions, which are difficult to pump through the inner
to the following model [3]: lumen of the hollow fibre membrane and preferably operated in
  dead-end mode.
t gRm gacb
¼ þ V ð5Þ Different combinations of the Al2O3 support layer with the two
V TMP  A 2  TMP  A2 different active layers result in different pure water permeabilities
Here, Rm is membrane resistance, A is the active filtration area of (PWP) due to differences in the particle size distributions of the
the membrane, V is the filtrate volume at time t, a is specific cake active layer starting material. Pore size distributions are measured
 F. Arndt et al. / Separation and Purification Technology 171 (2016) 289–296 291

Fig. 1. SEM of (a) a detail of the cross-section of an Al2O3 membrane with the active layer on the inside and (b) a detail of the active layer and transition to the support layer
[18].

Fig. 2. SEM images of clean membrane surfaces with different active layers. (a) Al2O3; (b) SiC.

with a capillary flow porometer (PMI, Model CFP AEX 1500) using
Galwick (c = 15.9 mN/m) as wetting liquid. Pore size distributions
are calculated from the pressure required to displace the wetting
liquid from the most constricted part of the pore and the resulting
gas flow according to Young-Laplace’s equation:

4c cos h
D¼ ð6Þ
P
where c is the surface tension coefficient of the liquid, h is the con-
tact angle of the liquid on the pore wall, and P is the critical gas
pressure to reopen the pore [15]. The resulting mean flow pore
diameters are 32 nm for Al2O3- and 56 nm for SiC-coated mem-
branes. The standard deviation is ±7% for each type of membrane.
Zeta potential measurements of the active layer starting
material displayed for SiC a negative zeta potential and for Al2O3
a positive zeta potential near neutral pH. This is in good agreement Fig. 3. Flow diagram of filtration test rig. (1) Pressure vessel (feed), (2) membrane
with literature values for the iso-electric point of SiC, which ranges module, (3) balance, (4) needle valve, and (5) pressure vessel (backwash).

between 2.5 and 3.5 [16,17] and for Al2O3 particles or surfaces of
8–9.4 [16,17]. Hereinafter, it will be assumed that these values
can be transferred to the membranes.
Different filtration modes can be applied for ultra- and microfil-
tration processes. In the dead-end mode the feed medium is
pressed completely through the membrane. This leads to a growth
of the resulting alginate layer and causes rapid permeability decay.
In the cross-flow mode shear forces are generated and lead to a
removal of the cake layer. Experiments in the dead-end mode as
well as in the cross-flow mode have been conducted for this study.
2.3. Filtration set-up
Sodium alginate fouling is studied experimentally in a lab-scale
single-fibre module apparatus shown in Fig. 3. It is possible to
carry out inside-out filtration in the cross-flow and dead-end
modes as well as outside-in filtration in the dead-end mode.
Depending on the filtration mode (outside-in or inside-out), two
different membrane housings are used. The membranes used in
this study have an internal diameter of 1.8 mm, an outer diameter
of 3 mm, and a length of 185 mm. The membrane module is
connected to a feed vessel, which contains up to 18 l sodium algi-
nate solution, with compressed air being the pressure source.
Experiments are conducted at constant TMP.
The transmembrane pressure (TMP) is calculated from data of
one pressure sensor (P) in the feedline and a differential pressure
sensor (PD) connected to the feed and retentate line directly
upstream and downstream of the membrane module to measure
transtubular pressure drop along the membrane axis. The filtrate
pressure is equal to ambient pressure. The cross-flow velocity
can be set with a needle valve on the basis of data from a flow
meter (F) in the retentate line. Filtrate mass is recorded gravimet-
rically. LabVIEW is used for data acquisition and valve control.
292 F. Arndt et al. / Separation and Purification Technology 171 (2016) 289–296
Table 1
Pure water permeabilities of ceramic hollow fibres.
Active layer Type of filtration PWP [l/(m2 h bar)]
Al2O3 In/out 479 ± 86
Al2O3 Out/in 696 ± 80
SiC In/out 1130 ± 249
SiC Out/in 1011 ± 246
A new membrane is used for each experiment. Before conduct-
ing fouling experiments, the pure water permeability (PWP) of
each of the membranes is measured in order to ensure integrity
of the membrane. For this purpose ultra pure water was filtered
30 min at 270 kPa through the membrane in order to saturate all
pores with water and ensure the integrity of the membrane. After-
wards pure water permeabilities are measured in the dead-end
mode at 100 kPa transmembrane pressure for 30 min. The pure
water permeabilities measured are listed in Table 1.
2.4. Materials
Sodium alginate (Sigma Aldrich; ‘‘medium viscosity”) from
brown algae with a reported molecular weight of 80–120 kDa is
used in all experiments. For each experiment, fresh stock solutions
of 1 g/l sodium alginate are prepared in ultrapure water. The feed
solutions are then produced by dilution of the stock solution with
the appropriate amount of ultrapure water to adjust the desired
concentration. The sodium alginate concentration in this study is
50 mg/l except otherwise noted. The pH of the feed solution is
6.2. For experiments with addition of Ca2+, CaCl2 is added from a
stock solution to the feed solution prior to the filtration step in
order to prevent overdosing.
3. Results
The experiments are conducted with a sodium alginate concen-
tration of 50 mg/l in the presence (cCa2+ = 2 mmol/l) or absence of
Ca2+.
3.1. Dead-end filtration
In order to investigate the influence of active layer material,
dead-end filtration experiments are conducted for 120 min at a
constant TMP of 50 kPa. The resulting filtration data are shown
in Fig. 4.
During dead-end filtration experiments, even at relatively low
pressures of 50 kPa, a significant reduction of flux can be observed
in the first filtration stage. This indicates severe membrane fouling
and is in good agreement with results of other sodium alginate
fouling studies [4]. As it is shown in Fig. 4, the flux decline rate is
affected by the presence of Ca2+ in the feed solution. In the early
filtration stage, flux decline decreases when Ca2+ is added to the
feed solution. This effect was reported previously [4]. The decrease
in flux decline in the presence of Ca2+ can be attributed to the
formation of aggregates, which increases the effective size of
the alginate molecules. The aggregates become too large to enter
the pore of the membrane and as a result a stable alginate gel layer
is formed on the membrane surface. In the absence of Ca2+, by con-
trast, some alginate molecules are able to enter the pores and lead
to a rapid flux decrease due to pore blocking phenomena.
Fig. 4 shows that for the chosen dead-end filtration conditions
flux decline is not significantly affected by the active layer materi-
als. But it can be seen that for both coating materials, Ca2+ addition
influences flux development. For the filtration conditions without
addition of Ca2+ the flux after 120 min was significantly higher
Fig. 4. Filtration flux as a function of time for dead-end filtration of sodium alginate
solutions (50 mg/l) (a) with 2 mmol/l Ca2+ ions (d – Al2O3 and r – SiC) and without
addition of Ca2+ (s – Al2O3 and } – SiC) for different membrane active layers at a
transmembrane pressure of 50 kPa.
compared to the flux with Ca2+ addition. This indicates a higher
resistance of the alginate gel layer in case of Ca2+ which might be
attributed to the formation of a stable gel network under these
conditions. If Ca2+ is absent, the structure of the alginate layer
seems more of a concentration polarization type and exhibits
therefore a lower resistance [18]. This might lead to the conclusion
that for dead-end filtration experiments, effects caused by a
change of feed composition have a larger influence on fouling
mechanisms than membrane material. In order to verify the
assumptions made above, fouling mechanisms are evaluated in
the following chapter.
3.1.1. Fouling mechanisms
Fig. 5 shows filtration data for the two different coatings Al2O3
and SiC. The filtration data from experiments in the presence and
absence of Ca2+ are replotted according to the cake filtration model
(Eq. (5)) in order to identify the fouling mechanisms. For filtration
experiments with addition of 2 mmol/l Ca2+, Fig. 5 displays a
straight line for both active layer materials. In this case, the cake
filtration model describes the filtration mechanisms adequately
during the whole filtration period. This also is in good agreement
with the slower flux decline in Fig. 4(a) for filtration experiments
with addition of Ca2+ and the assumptions made above. The addi-
tion of Ca2+ leads to a formation of a gel network on the membrane
surface and cake filtration is the only mechanism taking place.
In the absence of Ca2+, the cake filtration model cannot be
applied. Especially in the early filtration stage deviations from
the cake filtration model are high. The higher slope at the begin-
ning of filtration supports the assumption of pore constriction
and pore blocking being mechanisms occurring in both active
layers in the absence of Ca2+ addition. The results obtained with
and without addition of Ca2+ are in line with previous findings
regarding sodium alginate filtration with a PES membrane
obtained by Katsoufidou et al. [4].
It has to be noted, however, that the filtration data of the SiC
membrane in Fig. 5(b) show a smaller slope at the beginning
(grey line) compared to the data of Al2O3 membranes depicted in
Fig. 5(a).
This smaller slope implies a lower resistance at the beginning of
the filtration experiment. Two mechanisms might be the cause of
this. On the one hand, the larger pores of the SiC membrane are
not directly blocked by the alginate molecules; therefore the resis-
tance at the start of the filtration is lower. One the other hand, the
negative zeta potential of both the alginate and SiC layer might
lead to repulsive interactions at the beginning of the filtration
 F. Arndt et al. / Separation and Purification Technology 171 (2016) 289–296
Fig. 5. Data from dead-end (TMP 50 kPa) filtration experiments without backwashing replotted according to cake filtration theory. 50 mg/l sodium alginate solution was
filtered with (a) an Al2O3 and (b) a SiC membrane. without Ca2+ addition and with Ca2+ addition.
cycle. After a short period of time, the slope increases significantly,
which may be attributed to the effect of internal pore blocking and
constriction after a sufficient amount of alginate molecules has
reached the membrane. This effect is observed for SiC-coated
membranes as well as for Al2O3-coated membranes during the fil-
tration of alginate solutions in the absence of Ca2+ ions. This also
explains the stronger flux decline at the beginning of the filtration
experiment compared to experiments with Ca2+ addition. After
these short initial differences, the filtration mechanisms change
for both materials towards cake filtration. But the resistance in this
region is much lower compared to the resistance for the gel layer
build up in presence of Ca2+. This also supports the assumption
of the alginate layer being more of a concentration polarization
type for conditions without Ca2+.
3.2. Cross-flow filtration
The following sections refer to the influence of varying opera-
tional parameters (TMP and vCF) on the alginate fouling.
3.2.1. Influence of alginate concentration
Fig. 6 illustrates the strong effect of the presence of Ca2+ on
cross-flow filtration of sodium alginate solutions. The development
Fig. 6. Variation of flux over time with 2 mmol/l Ca2+ and without Ca2+ addition for
three sodium alginate concentrations at a constant transmembrane pressure of
100 kPa and cross-flow velocities of 1 m/s. Without Ca2+: N – 10 mg/l, r – 25 mg/l,
and d – 50 mg/l and with Ca2+; 4 – 10 mg/l, } – 25 mg/l, and s – 50 mg/l.
293
of filtrate flux J is depicted over time for a constant transmembrane
pressure of 100 kPa and a cross-flow velocity of 1 m/s. While stable
filtrate fluxes can be achieved for the filtration of alginate solutions
in the absence of calcium, the filtrate flux drops rapidly in the pres-
ence of Ca2+. Variation of the sodium alginate concentration has a
minor influence compared to the variation of Ca2+ concentration.
At a cross-flow velocity of 1 m/s and in the absence of Ca2+, the
increase of the alginate concentration is small. Shear forces
are sufficient to reach stable fluxes of about 90% of the initial pure
water flux.
As outlined previously, adding of Ca2+ promotes the aggregation
of the alginate molecules [19]. This leads to the development of a
gel layer on the membrane surface, although cross-flow velocities
and, hence, shear forces are induced. The resulting shear forces at
a cross-flow velocity of 1 m/s are not sufficient to remove the gel
layer. While it is possible to achieve a stable filtrate flux of about
90% of the initial flux for the filtration of alginate solutions in the
absence of calcium, the filtrate flux drops to 10% of the initial flux
in the presence of Ca2+. On the one hand, Ca2+ may promote
interactions between the membrane surface and the alginate
molecules, which might lead to a stronger attachment of the algi-
nate layer to the membrane surface. On the other hand, severe
fouling effects might be caused by the enhanced interaction
between alginate molecules, which also promotes the formation
of a stable gel layer on the membrane surface [10,19]. Furthermore
for conditions with Ca2+ the filtration flux decreases especially in
the first filtration period significantly with increasing alginate
concentration. This is because more alginate molecules are
transported towards the membrane surface, leading to a faster
formation of the gel layer in the beginning of the filtration process.
It has to be taken into account that monolayer coating of the
membrane surface caused by foulant-membrane interactions takes
place in the initial filtration period. After this period, the rate and
extent of organic fouling are determined mostly by the
foulant-foulant interactions [20,21].
3.2.2. Influence of TMP and cross-flow velocity
The influence of varying transmembrane pressures and cross-
flow velocities is summarised in Fig. 7 for the filtration of sodium
alginate solutions with and without Ca2+ addition. For this purpose,
the steady-state filtration flux is plotted versus the applied trans-
membrane pressure. Experiments are conducted with a Al2O3-
coated membrane. The active layer is located on the inner surface
of the hollow fibre.
For filtration experiments in the absence of Ca2+, variation of
operational parameters has a significant influence on the resulting
294 F. Arndt et al. / Separation and Purification Technology 171 (2016) 289–296
Fig. 7. Permeate fluxes for filtration of 50 mg/l alginate solutions as a function of
transmembrane pressure at different cross-flow velocities (N – 1 m/s, r – 2 m/s,
d – 3 m/s) without the addition of Ca2+ and (4 – 1 m/s, } – 2 m/s, s – 3 m/s) with
Ca2+ addition. The dash-dotted straight line represents pure water flux.
filtrate flux. The higher the applied pressure is, the higher cross-
flow velocities are needed for the removal of the resulting alginate
layer.
As shown above for dead-end filtration as well as cross-flow
filtration with varying alginate concentrations, Ca2+ ions have a
significant influence on the fouling behaviour of the system. The
resulting fluxes are very low compared to the resulting fluxes
without Ca2+ addition. Neither variation of transmembrane pres-
sure (between 50 kPa and 250 kPa) nor variation of cross-flow
velocity has a significant impact on steady-state fluxes. In all cases,
the shear forces resulting from the cross-flow velocities applied are
not sufficient to completely remove the alginate gel layer
from the membrane surface. Only for conditions of low TMP
(25 kPa) the Ca2+ alginate gel layer might be controlled. For higher
TMP in the presence of Ca2+, the region controlled by mass transfer
seems to be reached. In the absence of Ca2+, the increase of TMP as
well as of cross flow velocity leads to a significant increase in
filtrate flux. Especially at low TMP, the filtrate flux is in the range
of pure water flux. This can also be seen in Fig. 8, which shows
the resulting fouling resistances calculated according to Eq. (1).
The higher the applied transmembrane pressure is, the higher
are the resulting fouling resistances. The main observation from
Fig. 8 is the dependency of fouling resistances on cross-flow veloc-
ity. This was not evident from flux data plotted in Fig. 7 and is due
to higher shear forces generated by higher cross-flow velocities.
Furthermore, the resistance of the fouling layer increases signifi-
cantly with the addition of Ca2+. For alginate filtration experiments
at a TMP of 50 kPa without Ca2+ addition, increase in cross-flow
Fig. 8. Fouling resistances for cross-flow filtration experiments (a) without (N – 1 m/s, r – 2 m/s, d – 3 m/s) and (b) with (4 – 1 m/s, } – 2 m/s, s – 3 m/s) addition of Ca2+ as
a function of vCF for different TMP.
Fig. 9. Shear stress number Ns vs. resistance number Nf (N – 1 m/s, r – 2 m/s,
d – 3 m/s) without addition of Ca2+ and (4 – 1 m/s, } – 2 m/s, s – 3 m/s) with Ca2+
addition.
velocity has no influence on the fouling resistance Rf. This is due
to the fact that shear forces are high enough to control the devel-
opment of the alginate layer under these conditions. The
normalised filtrate flux J/J0 for these conditions is 0.91 ± 0.01. In
order to investigate the correlation between TMP and vCF, the
filtration data under steady-state conditions can be expressed
using two dimensionless numbers: the shear stress number NS
and the resistance number Nf [12,22–24].
NS compares the shear stress acting on the membrane wall to
the driving pressure (TMP) [24]:
q  v 2CF
NS ¼ ð7Þ
TMP
where q is the density of the feed solution. Nf compares convective
cross-flow transport in the tubular membrane to the permeate flux
through a layer, whose resistance is the sum of all resistances
induced by different processes which limit mass transport [24]. Nf
is obtained from Eq. (8):
g  Rf  v CF
Nf ¼ ð8Þ
TMP
In Fig. 9 the filtration data of Figs. 7 and 8 are recalculated
according to Eqs. (7) and (8).
For cross-flow filtration experiments with Ca2+ addition, the
plot of NS against Nf leads to linear relationships:
 F. Arndt et al. / Separation and Purification Technology 171 (2016) 289–296
Nf ¼ a þ b  NS
The slope ‘‘b” is negative under these filtration conditions and
indicates that flux is a monotonously increasing function of vCF,
which means that fouling can be controlled or eliminated when
cross-flow velocities are high enough. NS-values might exist, for
which Rf is 0 and, hence, Nf is 0 [12].
For filtration experiments without Ca2+ addition, Nf and Rf
values are one order of magnitude lower than the corresponding
values obtained when Ca2+ is added. Under these operating condi-
tions, fouling seems to be much more reversible and controllable.
This is in line with the previous findings and also supports the
previous observation that the addition of Ca2+ leads to the develop-
ment of a gel layer on the membrane, whereas the concentration
polarization layer without Ca2+ addition is more fluid-like and,
hence, easier to remove.
3.2.3. Influence of active layer material
In order to investigate the effect of different active layers under
cross-flow conditions, experiments are carried out at 100 kPa and a
cross-flow velocity of 1 m/s. Alginate concentration is set to
10 mg/l. Fig. 10 shows the results for experiments with and
without addition of Ca2+.
In filtration experiments without Ca2+ addition, it is possible to
achieve clearly higher filtrate fluxes (690 l m2 h1) for the SiC
active layer, which are almost twice as high as for the membrane
with the Al2O3 active layer (370 l m2 h1).
In filtration experiments with Ca2+ addition, the flux also
decreases rapidly at the beginning for the SiC active layer and
reaches steady-state conditions after 40 min. After a filtration time
of 10 min, filtrate fluxes for both coating materials reach about
the same value around 175 l m2 h1. Within the next 30 min,
filtrate fluxes decrease only slightly and reach steady-state condi-
tions at 95 l m2 h1. Furthermore, it is worth noting that severe
alginate fouling under conditions of Ca2+ addition is independent
of the coating layer material. Therefore, it is assumed that the
strong fouling effect of Ca2+ is most likely caused by the promoted
bridging of the alginate molecules. Otherwise, a change in mem-
brane zeta potential would have a bigger impact on the fouling
behaviour of the system. Hence, repulsive forces between nega-
tively charged alginate molecules and the negatively charged SiC
Fig. 10. Variation of (a) flux J with time for Al2O3 (d/s)- and SiC (r/})-coated
membranes with (open data points) and without (closed data points) Ca2+ addition
for an alginate concentration of 10 mg/l. TMP: 100 kPa and vCF = 2 m/s.
295
ð9Þ
Fig. 11. Permeate fluxes for filtration of 50 mg/l alginate solutions as a function of
transmembrane pressure at different cross-flow velocities for SiC-coated mem-
branes (r – 2 m/s, d – 3 m/s) and Al2O3-coated membranes (} – 2 m/s, s – 3 m/s).
Black and grey dashed straight lines represent the related pure water fluxes.
membrane are not strong enough to overcome the convective flow
of material against the membrane.
To investigate the influence of operation conditions on filtrate
flux in the absence of Ca2+, further experiments with varying
TMP values and cross-flow velocities are conducted with the SiC
membranes. Results are compared with those obtained for
membranes with Al2O3 active layers and are shown in Fig. 11.
For low TMP, the membranes with SiC active layers exhibit
significantly higher fluxes compared to membranes with Al2O3
active layers. One explanation of this phenomenon could be the
larger pore size and therefore higher pure water permeability of
the SiC-coated membrane. For a TMP of 50 kPa, the flux of the
Al2O3-coated membrane lies in the pressure-controlled region.
Consequently, an increase in PWP might also lead to higher fluxes
for this membrane. For a TMP of 100 kPa and a cross-flow velocity
of 2 m/s, the PWP plays a minor role, but filtrate fluxes of the
SiC-coated membrane are higher than those of the Al2O3-coated
membrane. For this reason, it can be assumed that repulsive forces
between the negatively charged SiC membrane and the negatively
charged alginate molecule may positively affect fouling mitigation.
For higher TMP, by contrast, the difference between the steady-
state fluxes resulting for the two coating materials is much smaller.
This may be explained by the higher driving forces (TMP), which
will overcome the repulsive forces under certain operation
conditions.
4. Conclusions
The influence of operating conditions, membrane material, and
feed composition on the fouling behaviour of ceramic hollow fibre
membranes has been studied. Hollow fibre membranes with two
different active layers (SiC and Al2O3) were studied in a single-
fibre lab-scale apparatus. Dead-end filtration experiments revealed
that the addition of Ca2+ leads to the development of an alginate
gel layer on the membrane surface. This is due to the higher
intermolecular binding forces between the alginate molecules in
the presence of calcium ions. For this reason, the filtration mecha-
nism changes from a two-stage process (pore blocking and cake
filtration) in the absence of calcium addition to cake filtration only
in the presence of Ca2+. Membrane material has no significant
influence on flux development in dead-end filtration but might
affect cleanability of the membranes by removing the fouling layer
296 F. Arndt et al. / Separation and Purification Technology 171 (2016) 289–296
by backwashing. For this purpose backwashing experiments will
be performed in the next step in order to evaluate the influence
of the membrane material on the backwashing efficiency of the
membranes. For cross-flow filtration experiments with Ca2+ addi-
tion the membrane material has also no influence on the resulting
flux. For both active layer materials, the flux decreases rapidly at
the beginning of the filtration experiment and reaches the same
steady-state value. The behaviour is different for cross-flow filtra-
tion experiments in the absence of Ca2+. Membranes with SiC
active layers show higher fluxes under identical filtration condi-
tions than membranes with active layers of Al2O3. This may be
due to the larger pores and therefore higher clean water flux of
the SiC-coated membrane, but also to the repulsive forces of this
negatively charged membrane and the negatively charged alginate
molecules.
Acknowledgement
The authors gratefully thank Kompentenznetzwerk Verfahren-
stechnik Pro3 for project funding.
References
[1] P. Le-Clech, V. Chen, T.A.G. Fane, Fouling in membrane bioreactors used in
wastewater treatment, J. Membr. Sci. 284 (2006) 17–53.
[2] P. Le-Clech, Membrane bioreactors and their uses in wastewater treatments,
Appl. Microbiol. Biotechnol. 88 (2010) 1253–1260.
[3] Y. Ye, P. Le Clech, V. Chen, A.G. Fane, B. Jefferson, Fouling mechanisms of
alginate solutions as model extracellular polymeric substances, Desalination
175 (2005) 7–20.
[4] K. Katsoufidou, S.G. Yiantsios, A.J. Karabelas, Experimental study of
ultrafiltration membrane fouling by sodium alginate and flux recovery by
backwashing, J. Membr. Sci. 300 (2007) 137–146.
[5] S. Nataraj, R. Schomäcker, M. Kraume, I.M. Mishra, A. Drews, Analyses of
polysaccharide fouling mechanisms during crossflow membrane filtration,
J. Membr. Sci. 308 (2008) 152–161.
[6] K. Listiarini, W. Chun, D.D. Sun, J.O. Leckie, Fouling mechanism and resistance
analyses of systems containing sodium alginate, calcium, alum and their
combination in dead-end fouling of nanofiltration membranes, J. Membr. Sci.
344 (2009) 244–251.
[7] K. Katsoufidou, S.G. Yiantsios, A.J. Karabelas, An experimental study of UF
membrane fouling by humic acid and sodium alginate solutions: the effect of
backwashing on flux recovery, Desalination 220 (2008) 214–227.
[8] M. Hashino, T. Katagiri, N. Kubota, Y. Ohmukai, T. Maruyama, H. Matsuyama,
Effect of membrane surface morphology on membrane fouling with sodium
alginate, J. Membr. Sci. 366 (2011) 258–265.
[9] Y. Ye, V. Chen, A.G. Fane, Modeling long-term subcritical filtration of model EPS
solutions, Desalination 191 (2006) 318–327.
[10] P. van den Brink, A. Zwijnenburg, G. Smith, H. Temmink, M. van Loosdrecht,
Effect of free calcium concentration and ionic strength on alginate fouling in
cross-flow membrane filtration, J. Membr. Sci. 345 (2009) 207–216.
[11] K.S. Katsoufidou, D.C. Sioutopoulos, S.G. Yiantsios, A.J. Karabelas, UF
membrane fouling by mixtures of humic acids and sodium alginate: fouling
mechanisms and reversibility, Desalination 264 (2010) 220–227.
[12] C. Vincent-Vela, B. Cuartas-Uribe, S. Álvarez-Blanco, J. Lora-García, E.
Bergantiños-Rodríguez, Analysis of ultrafiltration processes with dilatant
macromolecular solutions by means of dimensionless numbers and
hydrodynamic parameters, Sep. Purif. Technol. 75 (2010) 332–339.
[13] M. Ebrahimi, S. Kerker, S. Daume, M. Geile, F. Ehlen, I. Unger, S. Schütz, P.
Czermak, Innovative ceramic hollow fiber membranes for recycling/reuse of
oilfield produced water, Desalination Water Treat. (2014) 1–14.
[14] A.Y. Tamime (Ed.), Membrane Processing Dairy and Beverage Applications,
Wiley-Blackwell, Chichester, West Sussex, 2013.
[15] J. Yu, X. Hu, Y. Huang, A modification of the bubble-point method to determine
the pore-mouth size distribution of porous materials, Sep. Purif. Technol. 70
(2010) 314–319.
[16] X.H. Wang, Y. Hirata, Colloidal processing and mechanical properties of SiC
with Al2O3 and Y2O3, J. Ceram. Soc. Jpn. 112 (2004) 22–28.
[17] B. Hofs, J. Ogier, D. Vries, E.F. Beerendonk, E.R. Cornelissen, Comparison of
ceramic and polymeric membrane permeability and fouling using surface
water, Sep. Purif. Technol. 79 (2011) 365–374.
[18] F. Arndt, U. Roth, H. Nirschl, S. Schütz, G. Guthausen, New insights into sodium
alginate fouling of ceramic hollow fiber membranes by NMR imaging, AIChE J.
62 (2016) 2459–2467.
[19] W.J.C. van de Ven, K.v.t. Sant, I.G.M. Pünt, A. Zwijnenburg, A.J.B. Kemperman,
W.G.J. van der Meer, M. Wessling, Hollow fiber dead-end ultrafiltration:
Influence of ionic environment on filtration of alginates, J. Membr. Sci. 308
(2008) 218–229.
[20] S. Lee, M. Elimelech, Relating organic fouling of reverse osmosis membranes to
intermolecular adhesion forces, Environ. Sci. Technol. 40 (2006) 980–987.
[21] Q. Li, M. Elimelech, Organic fouling and chemical cleaning of nanofiltration
membranes: measurements and mechanisms, Environ. Sci. Technol. 38 (2004)
4683–4693.
[22] M. Abbasi, A. Salahi, M. Mirfendereski, T. Mohammadi, A. Pak, Dimensional
analysis of permeation flux for microfiltration of oily wastewaters using
mullite ceramic membranes, Desalination 252 (2010) 113–119.
[23] M. Asaadi, D.A. White, A model for determining the steady state flux of
inorganic microfiltration membranes, Chem. Eng. J. 48 (1992) 11–16.
[24] S. Elmaleh, L. Vera, R. Villarroel-López, L. Abdelmoumni, N. Ghaffor, S. Delgado,
Dimensional analysis of steady state flux for microfiltration and ultrafiltration
membranes, J. Membr. Sci. 139 (1998) 37–45.